Introduction

1Comprising the areas to the east and northeast of Lake Chad, Kanem was birthplace and homeland of the Kanem-Borno Kingdom/Sultanate between the 8th and 14th centuries (Lange & Barkindo 1988: 445-447). Large part of the region consists of a widespread sand dune landscape with a Sahelian savanna vegetation, at places interspersed by large natron-rich clay depressions liable to seasonal flooding. Archaeologically, Kanem is still a major research blank for even the most basic information is missing. In this sense, we are not only completely ignorant concerning the early stages of the human occupation in the area (Later Stone Age or earlier); we are even entirely unfamiliar with the kind of material remains associated with the period of foundation and early development of Kanem-Borno in the 8th to 14th centuries AD.

2Up to a few months ago, nearly all archaeological data concerning past human activities in the region regarded the tangible but ambiguous evidence around sixteen sites constructed by means of fired bricks (Lebeuf 1962; Bivar & Shinnie 1962; Zeltner 1980; Gonzemai 2002). Drawing on histor­ical knowledge, those places were very early considered as of especial importance. They were variously regarded as elite locations linked either to the Kanem-Borno dynastic line of the Sayfuwa, mid-11th to late 14th c., or to their Bulala successors, late 14th to late 16th c. As pre-colonial fired-brick architecture around Lake Chad is demonstrably related to traditional political power (Seidensticker 1981; Seignobos 1981; Gronenborn 2001; Magnavita et al. 2009), it is indeed safe to assume that the Kanem locations were likewise aristocratic rather than commoner’s settlements. Though the socio-political status of their founders and inhabitants was early clear, it long remained doubtful which fired-brick sites were respectively constructed by the Sayfuwa or the Bulala.

3Meanwhile, twenty-nine fired-brick localities are known in Kanem (Fig. 1). Through reconnaissance surveys and test excavations accomplished in early 2019, three of those sites (Tié [KA-2017-01], Tié Kalaté S1 [KA 2019-07] and Tié Dardanga S1 [KA 2019-01]) were thoroughly described and confidently dated to the period of Sayfuwa hegemony in the region (Magnavita forthcoming). The three locations themselves are part of a cluster of altogether twelve sites in central Kanem, which were presumably constructed and inhabited sometime between the 11th and 14th centuries (Fig. 1). While it can be provisionally suggested that a number of the remaining seventeen sites in the surrounding areas equally date to that period, this still has to be demonstrated by work on the ground.

© C. Magnavita

4From all fired-brick sites yet known in the region, Tié (KA-2017-01) is especially noteworthy owing to a series of unique settlement-related features (see below). This paper presents key results of the first archaeological excavations conducted at the perhaps most significant of those features, the so-called Mound 1. After briefly considering the site’s distinctive physical characteristics and introducing the particularities of Mound 1, we refer to the results of explorative surveys conducted there in early 2019 as well as to observations made by previous researchers regarding the mound. Next, we deal with the general objective, strategy and methodology of the excavations accomplished at the spot in November 2019, subsequently addressing the major results achieved. The paper concludes by tackling questions of chronology and briefly discussing other important issues, most notably the function of the building and the status of Tié in relation to the other sites of the cluster and beyond.

Tié: site plan, major features, state of preservation

5The fired-brick site Tié is meanwhile confidently dated to the 12th-14th centuries (Magnavita forthcoming). It consists of a main and nearly rectangular fired-brick enclosure (ca. 176x146 m) connected at its eastern face with two secondary enclosures (Fig. 2). With a total extent of 3.2 hectares, the site is not only larger than the other eleven fired-brick localities belonging to the above-mentioned site cluster. Local informants acquainted with the ruins also consider it to be the largest Kanem site purely built of fired bricks. Though the latter statement still has to be archaeologically validated, the point is that presently Tié comes closest to what could be labelled a central place. In fact, the site not only sits near the spatial midpoint of the mentioned site cluster (Fig. 1), but also displays a number of distinguishing structural features not discernible at any of the other archaeological localities in the surroundings.

6First and as the results of a geophysical survey conducted at the site show (Fig. 3), Tié is characterized by 2.5 m wide and hundreds of meters long walled passageways that link different site sectors with each other. Those walled passageways are integral part of a number of extensive linear partition walls that subdivide the enclosure into a number of secluded spaces or courtyards. Second, some of the courtyards encompass what is provisionally interpreted as the ruins of comparatively small round buildings (houses or huts), what in turn might indicate the residential nature of some spaces (Fig. 3c). Third, Tié displays two sizeable debris mounds in its northeastern sector. Mainly covered with potsherds, the larger northern mound or Mound 2 (ca. 2x50x70 m) is clearly a refuse heap. This interpretation is substantiated by the results of the geophysical survey conducted at the site, which does not display magnetic anomalies hinting at buried structural remains at the spot (compare Figs. 3a and 3b). In contrast to that, the surface of the southern mound or ‘Mound 1’ (ca. 2.1x40x40 m) is littered with a moderate number of both broken and entire fired bricks as well as fragments of what appears to be white plaster but no potsherds. Mound 1 thus unmistakably conceals the ruins of some kind of construction, a fact unequivocally supported by the results of the geophysical survey (Fig. 3a, b). The latter shows that the sub-surface extent of Mound 1 encloses an intricate and shapeless area of high-amplitude magnetic anomalies, no doubt indicating the presence of both collapsed and still standing ruins of one or several fired-brick buildings. While the geophysical survey itself is not explicit concerning number, shape and nature of buried structures under Mound 1, a glance at the satellite and aerial images of the spot appears to suggest the presence of a single and near-rectangular building oriented toward northeast (Fig. 2a, b; Magnavita forthcoming).

7Though there are today no architectural remains promptly visible on the surface of Mound 1, this has not been so in the past. At least two archaeologists that visited Tié some decades ago appear to have seen the remains of still standing structures on Mound 1 or elsewhere on the site. The first of these was Adrian D.H. Bivar who surveyed the place in March 1959. In regard to Mound 1 he sketched and described the existence of the “remains of a chamber with baked brick walls” locally called “the mosque” (Bivar & Shinnie 1962: fig. 3). The second to visit and provide an idea of the state of preservation of architectural remains at Tié was Alfred Gonzemai (2002: 46-48) who visited the site in May 2000. Alongside brief descriptions of what he saw in situ, a sketch plan and four photographs of the location emphatically illustrate both the enormous amount of fired bricks scattered all over the place as well as fired-brick walls still surviving at a height of two to three courses. Though it is unclear whether the wall remains shown were photographed on Mound 1 or elsewhere on the site, A. Gonzemai’s pictures demonstrate that the preservation of structural vestiges about twenty years ago was significantly better than it is today. In fact, Tié bears nowadays at surface level neither the massive number of fired bricks nor any standing walls as shown by the author. Though the latter author was throughout laconic in his text regarding preserved structural remains on Tié, his site sketch plan (Gonzemai 2002: fig. 11) reveals at the spot of Mound 1 the rectangular outline of what he calls the “palais” or “palais royal”. While the orientation of the drawn structure does not comply with reality (see below), the fact that it was recorded on the site plan possibly hint at a better preservation and visibility of ruins in early 2000 than today the case.

8Having in mind the state of preservation of the site as reported by A. Gonzemai (2002: 46-48), it can be thus concluded that Tié must have been heavily plundered for its bricks within the past two decades. Whilst that process certainly started much earlier as already witnessed by A.D.H. Bivar in 1959 (Bivar & Shinnie 1962: 8-9), the fact is that most of the original site constructions only partially survive today. As in the case of other sites in the region, the fired-brick structures at Tié are indeed generally leveled to the ground, if at all visible at surface level. Preliminary research involving surveys and test excavations at the site in early 2019 indicates a rather variable state of preservation of surviving structures (Magnavita forthcoming). Two discrete instances illustrate this situation. The first regards the results of two test-pits respectively dug at the western and northern segments of the enclosure’s perimeter wall. While foundations of the western wall showed up at a depth of only 0.4 m below surface, excavations up to a depth of 0.8 m along the northern wall did not come across its understructure. This implies a much better preservation of the northern segment ruins. In the second case, geophysical surveys hint at a partial plundering of fired-brick walls of the central walled passageway, since some portions of the same structure do not show up in the magnetic plot.

9In short and for reasons that are not entirely clear, some structures at Tié have obviously better survived the ravages of time than others have. Amongst the first is the construction concealed under Mound 1. In the following sections, we present and discuss the results of the archaeological excavations carried out at the spot.

Excavations on Mound 1

Objective, strategy, methodology

10Mound 1 is no doubt amongst the features of greatest archaeological interest and potential at Tié. This is the case not only because of the expected good preservation of the structural remains buried under more than two meters of sediments. As discussed later, it was also early thought that Mound 1 and especially the alleged ruins therein might furnish answers to a series of questions related to settlement growth, chronology, function and status. The primary objective of research was therefore to collect fundamental information about the buried structure such as size, shape, construction details, state of preservation, and, insofar as possible, to search for clues to its construction date and function(s). It is against this background and within the scope of the second field season of the project ‘The Lake Chad region as a crossroads’ (Magnavita et al. 2019) that the authors conducted a series of archaeological cuttings at the spot between the 2nd and 14th November 2019.

11The strategy employed to investigate Mound 1 was to assemble as much information as required by removing as little sediment as possible. The rationale behind that was twofold. First, restricted and selective excavation would allow a slow and careful exposition of the archaeological evidence in each cutting by simultaneously providing sufficient time for monitoring ongoing activities and documenting results. Second, restricted and selective excavation would permit to conserve as much cultural deposits for future and more sophisticated excavations as possible. That ‘just sampling’ strategy indeed proved successful, since we were not only able to retrieve the demanded data but, through the little invasive approach, kept most of the mound’s cultural strata safe.

12In order to uncover the evidence as stated above, a 26x26 m large excavation grid approximately pointing north was positioned on Mound 1 (Fig. 3). It comprised 2x2 m excavation units separated by one-meter wide stripes or balks (Fig. 4). Since position and orientation of the presumed buried building were only approximately known, that so-called box grid excavation proved extremely effective as the aligned squares could be dug or not according to whether sections of the structure were expected therein. Considering the overall objective and investigation strategy, only thirteen out of eighty-one 2x2 m grid squares were indeed dug to some depth and structures therein exposed. Common to almost all excavated units is that they followed the outline of the building, thus mainly revealing its perimeter or outer walls. As for stratigraphic control, a cemented datum point was fixed on the uppermost spot of Mound 1 at a height of 2.1 m above the site’s own datum point. The excavation units were dug in arbitrary levels of 0.2 m until standing structures became unambiguously visible in the plan. As will be seen below, the Mound 1 datum point was of particular importance for reconstructing the overall state of preservation of the building.

Primary results

13Except for excavation unit J6 (Fig. 4; see below), the 2x2 squares were dug in average to a depth of only 0.6 to 1.55 m under datum, thus at best exposing the topmost mound deposits. The excavated units revealed that the upper portions of walls belonging to the building were buried at depths of less than 0.1 to 0.3 m below mound surface. The sediment enclosing the uncovered walls consisted of a very homogeneous, white, light and powdery mixture of sand and natron-rich clay but also lime. Amid that matrix a large amount of collapsed fired bricks and fragments of white plaster were found scattered on and around the exposed walls. All this indicate that the excavated mound sediment no doubt consists of rubble formed by progressive structural collapse. Except for intact and broken bricks that were handpicked and stockpiled near the exposed areas, the excavated mound sediment was systematically screened using 5 mm meshed sieves. An extremely significant result of that work was that, except for isolated finds of iron nails (Fig. 5) and a single glass bead, practically no cultural materials other than fired bricks, white plaster and other building fragments were retrieved from the upper mound deposits. That finding is not only in line with the observations previously made regarding the kind of artefacts found on the surface of Mound 1. It is also a significant piece of information regarding the process of mound formation, as discussed later.

14In addition to the artefactual evidence (see below), some of the most appealing results of the excavations conducted on Mound 1 no doubt relate to the ruins uncovered. Though relatively small and superficial when compared to the overall area and the depth of deposits, the cuttings collectively provide both a general impression of the kind of construction unearthed and valuable insights into a number of architectural details. They show that Mound 1 encompasses the ruins of a rectangular building with a width of 16.8 and a length of 23.4 m, oriented toward northeast. The investigated areas reveal portions of the outer and of a number of partition walls. The exposed segments of the outer wall have a consistent width of about 1.4 m, hinting at an overall extremely solid and uniform construction. In contrast, the uncovered partition walls display variable widths ranging from 0.2, 0.4 to 0.7 m. Even though the complete ground plan remains yet unknown, they clearly suggest the building to have been at least in part subdivided into a number of rooms. As most of the partition walls, the substantially broader outer wall was constructed using rectangular fired bricks joined with natron-rich clay mortar.

Bricks and bricklaying pattern

15The bricks employed in the construction of the building’s outer and partition walls differ in a number of ways from those noted at other parts of the site. Most have a standard size of about 40x20x8 cm (Fig. 6) and were almost twice as large as those used in the walls and foundations of the site’s main enclosure. Bricks with sizes of about 28x14x6 cm and 20x10x5 cm were however also employed in the construction of the building (Fig. 6). Other than in the case of the enclosure wall and foundations, a large portion of the bricks from the building under Mound 1 was not hand-made but rather shaped in wooden moulds. The result is not only particularly regular bricks with square edges and flat sides. They also feature relatively smooth and nearly or fully flat upper faces, markedly diverging from the often convex and finger-scored upper face of hand-made bricks seen both at Tié and other fired-brick sites in the area. While there are noticeable differences between size, shape and quality of the building materials used at Tié constructions, the bricklaying pattern is identical. Just as in the case of the site’s enclosure wall, the outer wall of the building under Mound 1 was mainly laid using the so-called ‘English bond’, in which courses of stretchers alternate with courses of headers (Fig. 7). Occasionally, however, stretchers replaced headers or vice versa. This appears to have been also the pattern used in the erection of the 0.2 m wide partition wall 2 (Fig. 4).

Plaster and probable ceiling/roof fragments

17In addition to the interior wall coating, most excavated units but especially units I4, I6 (between 0.8 and 1.2 m under datum) and J6 (between 0.8 and 2 m under datum) exposed large and heavy fragments of what may be the remains of the former ceiling/roof of the building (Fig. 8b). Just as in the case of the wall plaster, the fragments that similarly display a smoothed face occur amid the mound sediment both within and outside the building. They are up to about fifteen centimeters thick and consist of a similarly white but slightly coarser material than that used as wall coating. Other than in the case of the latter, the material of the alleged ceiling/roof fragments is mixed with tiny pieces of gray grit and is extremely tough, having altogether a similar hardness to that of concrete. Since the fragments occur amid mound sediment from inside and outside the building and at depths far above what was once the structure’s ground level (see below), it can be ruled out that they are remains of a ground floor. Along with those, the lowest 0.2 m of sediment excavated at unit I6 (between about 1.15 and 1.35 m under datum) concealed well-preserved fragments of wood identified as Ziziphus sp. (Alexa Höhn, personal communication). Whereas, it is very probable that the building’s ceiling/roof construction included the use of wood beams, it is unclear whether the fragments recovered belonged to that or another wooden element of the building.

Special architectural features

18Even though the excavations on Mound 1 did not expose the entire building, they reveal some architectural features that shed some light both on the planned character of the structure and on modifications made during its use. Two of these are visible along the northeast building wall and are each related to the two rooms separated by partition wall 2 (Figs. 4, 9). The first of the two features is a recessed wall niche measuring 1.0x0.75 m at the northernmost room. The niche sits exactly at the midpoint of the building’s northeast wall. Initially, it was thought that the recess might be the prayer niche or mihrab of a mosque. After the discovery of partition wall 2, however, that interpretation became less probable (see Discussion and concluding remarks). In this context, it seems presently more likely that the recess merely marks the position of a former window, the sill of which must have been however at a minimum height of about two meters above ground floor (see below). As in the former case, the second architectural feature, this time associated with the southern room is likewise a wall niche. Though erosion destroyed much of the respective outer wall segment in the past, the recess is still easily discernible in the cutting. With a depth of about 0.65 m and a width of at least one meter, the recess can be provisionally equally regarded as a window niche. As in the case of most interior walls, both niches are coated with fine white plaster. The third architectural feature worth mentioning is situated at the northwestern outer wall of the building. Unlike the previous two, it probably consists of an at least 1.25 m wide gap or access point into the building, which was subsequently sealed by a wall of coursed clay. The latter was apparently constructed there where the fired-brick reveal of a probable doorframe once stood. An interesting detail is that the interior face of the clay wall similarly bore a layer of fine white plaster, probably in an effort to effectively conceal the existence of the former aperture.

Excavation unit J6

19In contrast to the remaining twelve excavation units, unit J6 at the southeastern outer corner of the building was dug to a substantial depth. For instance, while unit I4 reached a maximum depth of 1.55 m under datum, unit J6 was excavated to a depth of 3.8 m under datum. Central aim of that work was indeed to reveal the preserved height of the outer wall and its foundation, thus providing basic data on the general state of preservation of the construction. The results of the excavation at J6 are manifold.

20As in the other units too, the upper excavation levels until a maximum depth of 2.0 m under datum virtually only revealed collapsed building materials, above all bricks, plaster and possible ceiling/roof fragments as well as some iron nails (Figs. 5, 10). In the middle levels, below 2.0 until about 2.8 m under datum, there were much less rubble but instead a substantial number of other cultural remains (see below). A 0.1-0.5 m thick layer of a whitish sediment lies on the bottom of the middle excavation levels. That sediment consists of mixed deposits of clay varying from fine and compact to crumbly loosen material, virtually devoid of finds. The lowest excavation levels consisted of light red dune sand only sporadically encompassing potsherds but no other finds. Sterile dune sand appeared between 3.7 and 3.8 m under datum.

22Alongside revealing details related to the stratigraphy of Mound 1 and the construction therein, unit J6 also furnished an assemblage of artefacts as well as some animal bones and charred plant remains. Those materials only began appearing in the middle excavation levels from a depth of 2.0 m under datum. From that level on and for a better control of find provenance, excavation at J6 was done in spits of 0.1 m. The artefacts recovered mainly consist of pottery, iron objects such as nails and metal fittings (Fig. 5), a number of glass beads as well as a small piece of gold foil.

23A last point to be addressed is the overall state of preservation of the building. Based on the yet known depth of the plinth’s top face, the height of still standing walls can be indeed easily calculated for the entire construction. For instance, the outer wall segment in excavation unit E2 survives to a height of up to about 2.2 m, that in I3 to nearly 2 m and those in H1 and C9 to approximately 1.9 m. These figures are significant, as they show the extent to which the original building is preserved and, considering the depth of excavations units, reveal how much collapsed material remains untouched by our excavations. As regards the latter, the maximum depth reached by excavations, for instance, at unit I4 (1.0 m under datum) indicates that we only removed about 20% of the rubble fill in the room to the north of partition wall 2 and that approximately 1.6 m (or 80%) of cultural deposits remain intact. To judge from the approximate inner dimensions of the building (about 14x20.6 m), a minimum volume of about 580 m³ of cultural strata would need to be removed for exposing the entire interior of the construction. Whereas a challenging job (see Concluding remarks below), future excavations within and around the structure will no doubt reveal substantially more details about the building under Mound 1, particularly concerning its layout, overall context and, feasibly, the motivation behind its construction.

Absolute dates

Table 1. Thermoluminescence date related to the building under Mound 1. Year of analysis is 2020

Site (site code)	Lab no.	Sample code	Sample type	Depth (m)	Context	De (Gy), Std. Dev. (1-σ)	Age years ago, std. dev. (1-σ)	Calendar years (mean), std. dev. (1-σ)	Calendar years (1-σ; 68.2%)	Calendar years (2-σ; 95.4%)
Tié (KA-2017-01)	MAL-10467	Tie-GR1-J6-AF	Fired brick	3.7	Mound 1, building foundation	0.90± 0.06	910± 80	1110± 80	AD 1030–1190	AD 950–1270

Table 2. Radiocarbon dates related to the building under Mound 1

Site (site code)	Lab no.	Sample code	Sample type	Depth (m)	Context	Age BP, Std. Dev.	Cal. calendar years (1-σ; 68.2%)	Cal. calendar years (2-σ; 95.4%)
Tié (KA-2017-01)	Beta-552899	Tie-GR1-J6-270	Charcoal	2.7	Unit J6, lower layer 2a	580± 30	AD 1389–1408 (22%) AD 1316–1354 (46.2%)	AD 1380–1418 (31.8%) AD 1300–1369 (63.6%)
Tié (KA-2017-01)	Beta-567841	Tie-GR1-J6-150	Charcoal	1.5	Unit J6, lower layer 1, ceiling/roof fragment	850± 30	AD 1163–1220	AD 1152–1260 (90.4%) AD 1052–1080 (5.0%)
Tié (KA-2017-01)	Beta-562335	Tie-GR1-I6-135	Wood	1.35	Unit I6, SE-room, preserved wood fragment	1040± 30	AD 986–1021	AD 950–1032 (90.3%) AD 900–921 (5.1%)

25While the initial phase of site occupation appears to be relatively well established, a question that early kept us busy regarded the possible asynchronous construction of the two secondary enclosures to the northeast of the main one. In fact, and as figures 2 and 3 illustrate, the location and orientation of the two smaller enclosures early prompted us to think that they are perhaps later appendages to the larger structure. That initial impression had to be tested and we hoped to find a preliminary solution to the problem by examining Mound 1, since it seemed to be physically related to the construction of the secondary enclosures. Sitting adjacent to the northeast segment of the peripheral wall, the location of the mound indeed gives the impression as if it was added to the complex at a different point in time. The question was when that happened.

26In order to address the issue, three distinct archaeological samples from the excavation unit J6 and one from unit I6 were submitted to chronometric analysis (Tables 1, 2; Fig. 10). The samples from unit J6 comprise: (1) a fired brick taken from the deepest course of the building foundation at a depth of 3.7 m under datum; (2) charcoal from the neighborhood of the plinth’s topmost brick course at a depth of 2.7 m under datum (lower layer 2a); (3) charcoal trapped within a collapsed fragment of the building’s ceiling/roof at a depth of 1.5 m under datum (lower layer 1). The single sample from unit I6 consisted of a well-preserved piece of wood found amid collapsed building rubble at a depth of 1.35 m under datum. Below, we discuss the results of the chronometric analysis from those samples and draw conclusions regarding the probable age of the building.

27The fired brick (Tie-GR1-J6-AF) was dated by standard thermoluminescence analysis to 910 ± 80 years before sample taking in early 2020 (MAL-10467). That means a calendar date of AD 1110 ±80 with one-sigma (68.2%) and two-sigma (95.4%) age intervals respectively ranging from AD 1030 to 1190 and AD 950 to 1270 (Table 1). While the date is problematic in view of its relatively wide two-sigma age range, it is not worthless. In fact, and though not allowing to state whether the building was erected earlier, at around the same time or after the main enclosure, it implies construction to have occurred prior to the 14th century.

29Contrary to the lower, the upper charcoal sample (Tie-GR1-J6-150) from unit J6 furnishes direct evidence of the period of building construction. As above stated, the sample was retrieved from charcoal trapped in a collapsed fragment of the building’s ceiling/roof, found in lower layer 1 (Fig. 10). Radiocarbon analysis dated the sample to 850 ±30 BP (Beta-567841) with a calibrated age range stretching between AD 1052 to 1080 (5%) and AD 1152 to 1260 (90.4%) at a two-sigma confidence level (Table 2). This means that, with a probability of more than 90%, the construction of the ceiling/roof, and thus of the building itself, was accomplished sometime in the mid-12th to the mid-13th century AD. That date is for two reasons highly significant. First, it falls within the upper range of the two-sigma age interval of the TL sample that dates the foundation of the building (MAL-10467, AD 950–1270). Second, it is fully consistent with the two-sigma age ranges of both the TL and radiocarbon samples dating the construction of Tié’s main enclosure (MAL-10457, AD 1100–1260; Beta-521691, AD 1169–1270). We infer from all this that the analyzed charcoal sample of the ceiling/roof not only confidently dates the building under Mound 1. The sample also allows stating that construction must have taken place only a couple of years or decades before or after the construction of the main enclosure.

30The wood sample (Tie-GR1-I6-135) retrieved from excavation unit I6 at a depth of 1.35 m under datum came from amid the rubble that fills the building’s southeast room (Fig. 4). Identified as probable Ziziphus sp., the piece submitted to radiocarbon analysis was the splinter of a still buried larger wood fragment. It remains unclear whether we are dealing here with construction timber (beam, rafter) or furniture remains. The sample was radiocarbon dated to 1040 ±30 BP (Beta-562335), implying an age range stretching between AD 900 to 921 (5.1%) and AD 950 to 1032 (90.3%) at a two-sigma confidence level (Table 2). With a probability of over 90%, the wood splinter thus dates to the mid-10th to the first third of the 11th centuries. Particularly when compared to the dated charcoal trapped in the ceiling/roof fragment (Beta-567841), one is drawn to the conclusion that the wood splinter may be old wood.

31A number of conclusions can be now drawn from the results of the chronometric analysis and inferences presented above. First, the absolute dates related to the building under Mound 1 indicate that the structure was constructed and first used prior to the 14th century and thus within the period of Sayfuwa rule over Kanem. Second, whilst an earlier construction date for the building (10th-11th centuries) cannot be definitely ruled out, based on the intersection of most absolute dates presently available for the site, we consider the mid-12th to mid-13th centuries as the most likely construction period. Third, the absolute dates respectively related to the main enclosure and the building under Mound 1 do not allow ascertaining which one of those two structures was first constructed. Fourth and in view of the spatial relation between building and surrounding secondary enclosures, the latter may similarly date to the 12th-13th centuries. If correct, that would imply that Tié achieved its maximum area (3.2 hectares) at that period. Even though further evidence is needed for validating the latter conclusion (see below), that explanation better matches the chronological data now available on the site. Fifth and as the radiocarbon date Beta-552899 points out, the building was in use until at least sometime in the early 14th to early 15th centuries i.e., the late period of Sayfuwa and the early period of Bulala rule over Kanem.

Discussion

Stratigraphy and mound formation processes

32Considered in chronological order, the characteristics of the three major archaeological layers, equivalent to three major excavation levels (Fig. 10), tell us how the mound raised to its present height and appearance. The lowest layer 3 (about 3.8 to 3.0 m under datum), consisting of light red dune sand, represents the natural sediment into which a construction trench was once dug for accommodating the building foundation. The only sporadic find of potsherds in this layer, most of which date to an earlier (Iron Age) occupation of the dune, indicates a relatively fast construction of the foundation and subsequent trench closure. The superimposing middle layers consist of two deposit types: a lower white clay layer (2b) and an upper sand layer also containing clay (2a). The origin of the white clay layer adjacent to the plinth and practically devoid of artefacts is uncertain. Though it may be a byproduct of the building’s construction process, we have to grant the possibility that it was deliberately deposited to serve as a kind of pavement. The sand-clay layer above (2a, about 2.8 to 2.0 m under datum) contains the bulk of finds not related to construction (pottery, glass beads, gold foil, charcoal). It unmistakably formed during the time the building was in use. The charcoal sample Beta-552899 suggests that that layer started accumulating sometime between the early 14th to early 15th centuries. However, as the building was most likely built at latest sometime between the mid-12th to mid-13th centuries, it is unclear why earlier cultural deposits are apparently missing. At least two explanations can be put forward. Either previously accumulated sediment was removed from the spot or the dated charcoal sample is intrusive. The accumulation of the middle layers were followed by the abandonment of the site and the gradual collapse of the building. Consisting of a light and powdery mixture of sand and natron-rich clay as well as an enormous amount of collapsed bricks, wall plaster, probable ceiling/roof fragments and some iron nails, the upper layer (above 2.0 m under datum) clearly implies a long process of building decay. Except perhaps for a single iron arrowhead found on the bottom of the upper layer (1.9 m under datum), there are actually no sign of a fire or any violent and premeditated destruction neither of the building nor the site as a whole.

Building function and settlement status

34Notwithstanding the improbability of its use as a religious edifice, the building under Mound 1 is a remarkable and unique construction in the entire Lake Chad region, having even no parallel in the remaining Central Bilad-al-Sudan at the time. It undeniably testifies either the transfer of foreign building skills to local masons or an import of foreign and specialized work force into the region. The whole structure is indeed so well built, its angles and alignment in such an extent precise and the interior walls so finely finished, that it can hardly be doubted that experienced architects or master builders and masons were part of the crew that once accomplished it. In addition, the construction materials used such as the high-quality fired bricks and the fine wall plaster clearly hint at the enormous costs and efforts involved in production. The conclusion to be drawn out of all this is that the building was a high-status construction that once very probably fulfilled exceptional functions both within the settlement and in relation to the surrounding and wider Kanem territory. Against this background, it is conceivable that it might have served representative and/or administrative purposes both at local and regional scales, perhaps even as seat of government.

Time and historical context of building construction

36A last question that deserves some attention relates to the time and historical background of construction and use of the building under Mound 1. As demonstrated above, there is now reasonable chronological evidence suggesting that the building was erected sometime between the mid-12th and mid-13th centuries and potentially used at least sometime until the early 14th to early 15th centuries. That data is in fact largely substantiated by other chronometric analysis at samples from Tié, firmly implying an occupation of the site starting at some point in the 12th to 13th centuries (Magnavita forthcoming). From a historical viewpoint, the main significance of the plastered building under Mound 1 is twofold. First, its construction took place approximately three quarters of a century to two centuries after the Islamized Sayfuwa dynasty took control over Kanem (Lange 1977: 68). The high-status building itself is not only a further and unprecedented material evidence supporting the notion of sustained and close cultural contacts between the Sayfuwa sultans and the outside world, especially perhaps Islamic North Africa. It is also an important potential element bearing witness of the self-conception of the Sayfuwa as part and major representatives of the Dar al-Islam south of the Sahara. Second, the construction of the building occurred either shortly before or during a time of unparalleled achievements in the history of the Sultanate. In that respect, external Arabic sources acknowledge Sultan Dibalami (AD 1203-1242 or AD 1210-1248), son of the above-mentioned Sultan Salmama, as to have temporarily expanded Kanem political influence as far as Borno, Fazzan and perhaps even Darfur (Cuoq 1975: 209-211). In view of those achievements, especially the 13th century is acknowledged as the zenith of Kanem-Borno political and economic power (Barkindo 1985: 237-238). An appealing result of the excavations on Mound 1 regards the uniqueness of the archaeological evidence. While fired bricks are recurrent architectural features of 11th to 19th century Kanem-Borno elite constructions (Lavers 1971; Connah 1981: 229-235; Magnavita et al. 2009; Magnavita forthcoming), the innovative use of resilient plaster appears confined to high-status buildings of the 12th to 13th centuries. In fact, and except for Tié, none of the meanwhile thirty-nine known elite sites in Kanem and Borno, including the 15th-19th century capital Birni Gazargamo in Nigeria, is associated with such a kind of decorative wall coating. Taking all that into account, we not only consider the use of lime plaster in the building under Mound 1 as a reflection of the increased political and economic strength of the Sayfuwa in the course of the 12th to 13th centuries. We similarly regard it as a feature related to particularly important Sayfuwa high-status buildings and settlements in Kanem.

Summary and concluding remarks

37Overall, the archaeological excavations on Mound 1 provided valuable information on the nature of the perhaps most conspicuous archaeological feature at Tié. We now know that the formation of the mound in part resulted from the gradual decay of a massive multiroomed building made of fired bricks mainly laid in ‘English bond pattern’. We also now know that the building had an up to 1.4 m thick outer wall now only surviving to heights of up to 2.2 m and resting on about 1.05 m deep foundations. We are similarly aware that the interior walls of the structure were once very probably entirely covered with a very fine white lime plaster, but that the façade did not bear any coating. It is almost certain that the ceiling/roof consisted of a construction involving wooden beams and a white and fine-grained smoothed material resembling the plaster on the inner walls, but substantially more resilient than the latter. The building also displays some remarkable architectural elements such as potential window niches and changes in layout.

38Just as the straightforward evidence related to architecture, the question of chronology regarding the building is now satisfactorily settled. Each of the four analyzed TL and radiocarbon samples, directly or indirectly associated with the structure, points to its probable construction prior to the 14th century AD and thus during the period of Sayfuwa hegemony in Kanem. Even though an earlier construction cannot be excluded, it is most likely that the structure was erected sometime between the mid-12th to mid-13th centuries and used until at least sometime in the early 14th to early 15th centuries. The mid-12th to mid-13th centuries is not only the interval within which the main site enclosure probably came into existence. It is also the time the secondary enclosures are expected to have been built. In order to test the later suggestion, we are planning archaeological excavations at the secondary enclosures within the course of the next two fieldwork seasons in early or late 2021.

39A very last point to be made refers again to the building under Mound 1. As our excavations revealed, the entire structure is in a relatively good state of preservation. This is not only shown by the height of standing walls but particularly by the surviving plaster covering the building’s interior. As the excavations also revealed, the surviving architectural remains, be they walls or the plaster on them, are extremely fragile. In this sense, we recommend that any future archaeological investigations going beyond the extent of our own work should adopt especial provisions to avoid the collapse of standing walls or the damage of the exceptional wall plaster through deep and prolonged exposures. While we are aware of the challenges involved in measures such as the use of timber shoring and plywood bracing as well as the consultation or participation of restaurateurs, those actions seems to be unavoidable as to safeguard the structure. In fact, it must be eventually remembered that the ruins survived concealed under Mound 1 for at least six centuries. It would be therefore regrettable, if inappropriate actions within the scope of future investigations irrevocably damage or even destroy that superb piece of architectural evidence.

We thank the German Research Foundation (DFG) for the substantial funds that allowed generating the research results presented above (research grant MA-6948/5-1). We also thank the Chadian Ministère de l’Administration du territoire and the Ministère de l’Enseignement supérieur, de la Recherche et de l’Innovation for the respective travel and research permits that enabled fieldwork. His Excellency Dr. Hassan Terap, the present Governor of the Kanem Province, was so gentle as to furnish the project with logistical and security support. His Highness Mr. Mustapha Ali Zezerti, the Sultan of Kanem, and Mr. Abba Alifa Zezerti, the former’s representative, are sincerely thanked for backing our research activities in the region. We express our deep gratitude to the Chadian historian, project partner and friend Dr. Zakinet Dangbet for the active encouragement of the archaeological work in Kanem. Similarly, we thank Daniel Ishaya from the National Museum Maiduguri and the PhD students Djimet Guemona and Abakar Abanga, Université de N’Djamena, for assistance in the field.