African Archaeological Review, Vol 15, No. 3, 1998

A Review of Swahili Archaeology Felix A. Chami1

The Swahili people have been viewed as of Persian/Arabic or Cushitic-speaking origin. Scholars have used historical and archaeological data to support this hypothesis. However, linguistic and recent archaeological data suggest that the Swahili culture had its origin in the early first centuries AD. It was the early farming people who settled on the coast in the last centuries BC who first adopted iron technology and sailing techniques and founded the coastal settlements. The culture of the iron-using people spread to the rest of the coast of East Africa, its center changing from one place to another. Involvement in transoceanic trade from the early centuries AD contributed to the prosperity of the coastal communities as evidenced by coastal monuments. More than 1500 years of cultural continuity was offset by the arrival of European and Arab colonizers in the seventeenth and nineteenth centuries AD. Le peuple Swahili a souvent ete considere comme un peuple dont la langue avait pour origine le Perse/Arabe ou le Cushite. Les chercheurs ont utilise des donees historiques et archeologiques afin de supporter cette hypothese. Cependant I'etude linguistique de cette langue, ainsi que de nouvelles decouvertes archeologiques suggerent que la culture Swahili trouve son origine au debut de l'ere chretienne. Ils furent les premiers fermiers a s'installer le long du littoral, fondant des villages cotiers, vers les premiers siecles de notre ere, les premiers aussi a adopter les techniques du fer et les techniques de navigation. La culture du fer s'etendit rapidement au reste des cotes d'Afrique de l'Est, son centre se deplacant d'un endroit a un autre. Leur implication dans le commerce oceanique contrbua a la prosperite de leur communautes cotieres, mise en evidence notamment par les monuments le long du littoral. Plus de 1500 ans de continuite culturelle pris fin a l'arrive des colonisateurs Europeens et Arabes de dixseptieme et dixhuitieme siecles.

1Faculty

of Arts and Social Sciences, University of Dar es Salaam, P.O. Box 35051, Dar es Salaam, Tanzania.

199 0263-0338/98/0900-0199$15.00/0 C Plenum Publishing Corporation

200

Chami

KEY WORDS: East Africa; Swahili coast; early ironworking cultural complex; pottery; contextual seriation; Arabo/Persian empire; Bantu vs Cushitic speakers.

INTRODUCTION

The dimensions of the Swahili territory have been conceptualized by previous scholars to be the narrow coastal littoral and the off-shore islands extending from southern Somalia to northern Mozambique and the Comores (Fig. 1) (Chittick, 1975; Horton, 1984, p. 299; Sutton, 1990, p. 57). While it is the case that this conception holds true for a certain period in the history of the Swahili coast, i.e., the eleventh to the fifteenth centuries AD, recent research reveals that the territory of the coastal culture extended some distance, in some cases up to 100 km, into the hinterland. This was probably more so in the seventh and ninth centuries AD and after the seventeenth century, when the interior of Africa was influenced by Swahili traders as far as the Congo Republic (Chami, 1994). It is with this understanding that in reviewing the archaeology of the coast of Swahili, I present a picture that covers a wider area of the coastal cultural influence and a longer chronology than has been previously considered. The time period covered ranges from the early farming period in the last centuries BC to the sixteenth century, which is the beginning of the historic period—a duration of some 1500 years. Before my work (Chami, 1994) there were scanty archaeological data with which to examine the Swahili coast before the ninth century AD. Data are now available for scholars to discuss cultural configuration and processes that took place on the Swahili coast for the first 15 centuries of our era. Pottery recovered from archaeological contexts from several sites on the central coast of Tanzania is arranged chronologically to provide the first contextual pottery seriation (Fig. 2) for the coast of East Africa that outlines a new perspective of the first 1500 years of the Swahili coastal history. Since this work concerns 1500 years of cultural development that took place on the coast of East Africa, it is first necessary to introduce the concepts which are used to represent different cultural periods. Some terminologies, e.g., Early Iron Working (EIW) and Triangular Incised Ware (TIW), have been described before (Chami, 1994) and have been used in other previous publications (Chami, 1994-1995; Chami and Msemwa, 1997a, 1997b). In my work, the EIW has replaced the Early Iron Age (EIA), which has been imposed on the archaeology of Sub-Saharan Africa. It is questionable whether the eastern and southern region of Africa had an Iron Age similar to that which occurred in western Asia and Europe. In many parts of Africa iron smelting was practiced after the last few centuries BC. Other metals, e.g., copper and lead, were utilized up to 300

A Review of Swahili Archaeology

201

Fig. 1. The Swahili coast extending from Barawa south of Somalia in central Mozambique.

years later (Phillipson, 1993; Chami, 1994). The EIW concept also fits better with the early iron-using communities than the early farming communities, which are now seen to comprise stone-using, food producers (Robertshaw, 1990, p. 4; Phillipson, 1993, pp. 5, 113-147). TIW represents a pottery tradition which succeeded that of the EIW along the coast from the sixth century to the tenth century AD. It is the same cultural tradition that is found in the early occupations of stone town sites, i.e., Kilwa and Manda (Chittick 1974, 1984). A decorative element of triangular panels has been used to characterize the tradition because the frequency of this element comprises over 20% of the 42 identified ele-

202

Chami

Fig. 2. Contextual pottery seriation for the Swahili Coast. Group A—EIW Limbo phase, ca. 1-200 AD; Group B-EIW Kwale phase, ca. 200-550 AD; Group C-EIW Mwangia phase, ca. 500-600 AD; Group D-TIW tradition, ca. 600-1000 AD; Group E-Plain Ware (PW) tradition, ca. 1000-1250 AD; Group F-Swahili Ware (SW) tradition, ca. 1250-1500 AD.

ments (Chami, 1994, p. 72). The concept replaces the Tana tradition because the sites of Tana/Wenje (Phillipson, 1979) on the north coast of Kenya were not the first of the TIW tradition to be found on the coast of East Africa. Kilwa, Unguja Ukuu, Amboni, and Usambara, all on the Tanzanian coast, had been known by 1970 (Chami, 1994). The labeling of this coastalhinterland tradition as the Tana tradition also implied, according to Horton (1984, 1987) and Abungu (1989), that the origin of the tradition was on

A Review of Swahili Archaeology

203

the north coast of Kenya, an idea which is no longer tenable (Chami, 1994; Horton, 1996). To get rid of the biases contained in the concept it was necessary to adopt a new technological concept—TIW. Other terminologies used are also quite new. Plain Ware (PW) is used for a pottery and, hereafter, a cultural tradition that emerged on the coast of Tanzania from about the tenth to the thirteenth century AD. PW pottery is mostly plain, and in only a few cases have sherds been found with a line of punctates around the shoulder. Neck Punctating (NP) stands for a pottery and cultural tradition that emerged from the thirteenth century to the end of the fifteenth century AD. NP potsherds form the bulk of excavated materials from the ruined stone town sites. In the early fourteenth century Ibn Battuta (Gibb, 1939, p. 112) identified the settlements of this period as Swahili.

THE IDENTITY OF THE SWAHILI PEOPLE

From the beginning of this century a discussion has existed about the identity of the prehistoric Swahili people (Allen, 1982, 1993). Two models exist. The first and older model was presented by geographers and travelers from the Middle East and Europe who visited the coast of East Africa between 800 and 1700 AD (Freeman-Grenville, 1962). This model saw the Swahili as the indigenous, negroid people, and probably Bantu speakers. The model, which is well discussed in the works of Chittick (1975) and Mathew (1963), was championed by Mathew (1963), Freeman-Grenville (1963), Nicholls (1971), and Allen (1977) after 1960. One example of this view is the report given by Ibn Battuta when he visited Mombassa and Kilwa in about 1331 AD. He described Kilwa as a large town on the "Sawahil country" where the majority of inhabitants were "Zanj, jet-black in color, and with tattoo-marks on their faces" (Gibb, 1939, p. 112). In support, Mathew (1963, pp. 114-115) argued that "trade had caused gradual development of a town life among the negroid population and their rulers began to build in masonry." It has also been argued that the overall picture of Swahili ... is of more characteristically African society than has hitherto appeared: it is continental- or mainland-based, urban-based but with a very clear urban-rural continuum, and economically much more dependent upon agriculture and rather less upon trade. (Allen 1980, p. 363)

The second model was developed during the colonial time. The Swahilis were viewed as descendants of intermarriages between African women and Persians and/or Arabs who came to trade and settle on the coast of East Africa. For instance, Coupland (1956, p. 11) defined the Swahili as "mixed people of the maritime belt, the Swahili, a mixture of mixtures,

204

Chami

arising . . . from the impingement of Asiatic immigrants on the Bantu . . . ." This definition went hand in hand with another theory that Persians and Arabs had flocked to the coast of East Africa to establish settlements which they built with coral rags and lime. According to this theory, the coast of East Africa was an Arab-Persian empire. Thousands of the ruined settlements dotting the coast from southern Somalia to Mozambique were attributed to these people. Thus, Hollingsworth (1951, p. 39) argued the following while discussing the alleged "Zanj Empire." The civilization which sprang up during the tenth, eleventh and twelfth centuries was therefore Shirazian or ancient Persian in character. The Persian settlers also left their mark upon the natives of the region, for there was a certain amount of intermarrying.

As already noted, this model was rejected in the 1960s in favor of a more Africanist model. As shown in the next section, Chittick (1984) was the only person who carried the Asiatic model into the 1980s (Allen, 1982). In the 1980s, a new version of the Africanist model was launched. This was based on archaeological excavations on the north coast of Kenya. It was argued that Swahili people were the descendants of Pastoral-Cushitic people of northern Kenya and the Rift Valley (Horton, 1984, 1990; Abungu, 1989, 1994-1995). These people were thought to have founded early coastal and hinterland settlements before the coming of the EIW people. This theory was popularized (Sutton, 1990, pp. 59-60) even though the Swahili language was basically that of Bantu speakers (Nurse and Spear, 1985). The notion of a Cushitic origin for the Swahili people was challenged in the mid-1990s (Chami, 1994, 1994-1995). A view based on new archaeological finds and interpretation was advanced which attributed a Swahili origin to the preceding TIW and EIW Bantu speakers. The settlements of the first millennium AD were now thought to have belonged to this group of people. In the eighth century AD the influence of the newly found TIW tradition spread from the central Tanzanian coast, where it was probably first conceived, to the northern coast of Kenya and south to Mozambique (Chami, 1994, 1994-1995; Haaland, 1994-1995; Schmidt, 1994-1995). As shown below, the origin of the Swahili culture was on the Kenyan coast in the eleventh century and was realized all over the coast of East Africa in the beginning of the fourteenth century AD.

COASTAL ARCHAEOLOGY BEFORE 1990

The archaeology of the coast of East Africa was first studied by Kirkman in the 1940s and 1950s. He excavated at Kilepwa and Gedi in Kenya and illustrated archaeological and house remains from many parts of the

A Review of Swahili Archaeology

205

coast. In his 1964 work, Men and Monuments of the East African Coast, he used this archaeological and architectural data to castigate the long-existing idea that there had been a Persian and Arabic empire in Eastern Africa. However, he argued that "the historical monuments of East Africa belong not to the Africans but to the Arabs and Arabized Persians, mixed in blood with the African but in culture utterly apart from the Africans who surrounded them" (Kirkman, 1964, p. 22). An even more active archaeologist after Kirkman was Neville Chittick. The most prominent of his works included the excavation of Kisimani Mafia and Kilwa in Tanzania (Chittick, 1961, 1974) and Manda on the Kenyan coast (Chittick, 1984). He demonstrated archaeological sequences of the ruined stone town sites. He sorted and illustrated various cultural materials including pottery, beads, and glass objects—work which has remained relevant today. The main criticism of Chittick's work has been the emphasis he put on imported ceramics to date various cultural horizons (Chami, 1994, p. 90). By that method, he fixed the date for the beginning of the early coastal settlements in the ninth century AD, the accepted date for the making of early Islamic ware. He has also been criticized for thinking that post-ninth century cultural horizons were founded by massive waves of Arab/Persian immigrants (Chittick, 1975, p. 40; Allen, 1980, p. 361). He also did very little work to analyze the bulk of locally made pottery that formed "the greatest part by volume even in the early periods" (Chittick, 1984, p. 217). To explain how this pottery came to dominate in the settlements of immigrants from the Middle East, he suggested the existence, in the vicinity of the stone-built towns, of African settlements that would have produced the pottery and traded it to the town dwellers (Chittick, 1984, p. 217). It should be noted that there had been some archaeological research conducted on the immediate hinterland before the 1980s which was unfortunately not compared with that of Chittick because the coastal settlements were thought to be orientated toward the Indian Ocean (Allen, 1980). The hinterland work included the excavation of the Kwale EIW site near Mombassa in 1966 (Soper, 1967a) and the survey of east Usambara (Soper, 1967b) where EIW sites had been found. Also of importance in the Usambara research was the occurrence of TIW sites similar to those of Chittick's early Kilwa and Manda cultural horizons. Phillipson (1979) excavated another TIW site in the hinterland of the Tana promontory. He affiliated the pottery with the EIW pottery tradition, a suggestion objected to by later coastal archaeologists (Horton, 1987). Following the discovery of EIW sites on the immediate coastal hinterland, a theory was also formulated asserting that in the third century AD, EIW Bantu speakers had entered the Swahili coast through the northern

206

Chami

and eastern highlands of Tanzania to the Kwale area southeast of Kenya. This theory implied that a second movement ensued from a nucleus area around Kwale north toward the south coast of Somalia and south to Mozambique and Southern Africa. As a corollary, it was hypothesized that EIW settlements on the coast of East Africa and Mozambique were naturally of Kwale tradition (Soper, 1971; Cruz e Silva, 1977; Sinclair, 1987; Phillipson, 1993). This theory contradicted an earlier linguistic one that had existed since the 1950s—that the EIW people had entered the coast of East Africa from central Africa through an area south of Tanzania and north of Mozambique (Oliver, 1966). After reaching the coast their second movement was north and south along the coast from a nucleus area located between the Rufiji and the Ruvuma rivers. Of interest at this juncture is how the coastal archaeologists of the 1980s ignored the above two theories on the peopling of the coast of East Africa by the EIW Bantu speakers. Given the occurrence of TIW sites in many other parts of the coast, i.e., Mozambique (Sinclair, 1982), the Comores (Wright, 1984), and Shanga in the Lamu archipelago (Horton, 1984), there grew more interest in the archaeology of early Swahili settlements. Linguists were offering an even more interesting contribution when they asserted that the language of Swahili was related to that of Bantu speakers (Nurse and Spear, 1985) rather than Arabic (Chittick, 1984). The major work of Mark Horton in the excavation at Shanga has influenced many scholars in the interpretation of early Swahili culture. Horton (1984, 1987) asserted that the foundation of the ruined settlements was not Arabic but African, with an origin in the north coast of Kenya. TIW pottery had been produced in the hinterland of the Tana area and was of Cushitic culture/influence. The early coastal culture was alleged to have resembled that of the Neolithic pastoralists of the Rift Valley of Kenya (Abungu, 1994-1995). Further evidence that the founders of the coastal ruins were mainly Cushitic was the occurrence of cattle and camel bones. Cattle enclosures were also found. Initially, Horton thought that the coastal settlements would have acted as market centers. The settlers may have learned more about trade from incoming foreign traders, intermarrying with them, forming a trading group, and sailing southward to the rest of the East African coast. Horton (1984) conceived the beginning of these developments to be in the eighth century AD, a hundred years earlier than had been suggested by Chittick (1984). It was recognized that the ceramics, then known as Sassanian-Islamic, which were found in the early occupation layers of the TIW sites in Kilwa, Manda and Shanga, had been recovered at Siraf (Persia) from the eighth century archaeological context. Since, according to Horton

A Review of Swahili Archaeology

207

(1984, 1987), the north coast of Kenya had been the origin of the early Swahili culture, the rest of the settlements on the coast of East Africa were founded or influenced by immigrants from the north coast of Kenya. Recent archaeological surveys and the excavation of several sites on the central coast of Tanzania have provided data for an alternative archaeological theory regarding the issues of Swahili origin. The new work has been able to link the coast with the hinterland and the Swahili sites with the EIW culture. This archaeological effort began in 1990 and continues to date. Sites have been found by both individual and collective efforts of the members of the University of Dar-es-Salaam Archaeology Unit and the National Museum. The sites comprise those of stone-using communities, EIW, TIW, PW, and NP traditions. A discussion of sites found and excavated before 1994 is presented elsewhere (Chami, 1994; Schmidt et al., 1992; Haaland, 1994-1995). Post-1994 research is provided in several other reports (Chami and Kessy, 1995; Chami, 1996, 1998; Chami and Msemwa, 1997a, 1997b; Chami and Mapunda, 1998). It is from the archaeological work carried out since 1990 that a new perspective on the configuration and cultural sequence of the Swahili coast is advanced in this paper.

1500 YEARS OF CULTURAL CONFIGURATION OF THE SWAHILI COAST

The Swahili coast is now known to have been occupied since at least the Upper Paleolithic. Late stone-using sites have been found in several parts of the coast of Tanzania and Kenya (Chittick 1975, p. 17; Karoma, personal communication). Microlithic artifacts have also been recovered from sites such as Mutesa near Mombassa (Figs. 3a and b), Kiwangwa near Bagamoyo (Chami, 1996), Ziweziwe (Chami and Kessy, 1995), and Kibiti (Figs. 3c-e). Indications now exist which suggest that the stone-using people on the Swahili coast produced pottery probably from the eleventh century BC onwards (Chami, 1996). The occurrence of a microlithic industry stratified below the earliest known EIW horizons suggests that the iron-using agriculturists succeeded the former through natural culture change (Chami, 1994) or replacement through conquering (Phillipson, 1993). It has been noted above that up until very recently scholars of coastal archaeology had very little knowledge about the people who occupied the coast before the ninth century AD. Now we know that it was the EIW, agriculturists, who succeeded the stone-using people. These agriculturists occupied the Swahili coast from probably the last centuries BC. The first known EIW site on the coast is the Kwale site located on the Shimba hills southwest of Mombassa which was excavated in 1966 by Soper (1967a). As

208

Chami

Fig. 3. Late Stone Age artifacts: a and b are scrapers from Nutesa near Mombassa; c-e are flakes from Kibiti near the Rufiji Delta.

indicated earlier, since 1986 many more EIW sites have been found on the coast of Tanzania. They include Limbo (Chami, 1992) and Mtyeke (Schmidt et al., 1992), both about 75 km south of Dar-es-Salaam. From 1994 to 1997 about 20 more sites were found between Dar-es-Salaam and Rufiji (Chami, 1998; Chami and Kessy, 1994; Chami and Mapunda, 1998; Chami and Msemwa, 1997a). The sites are found on the islands, including the Mafia Island, on the littoral near the shore, and in the immediate hinterland as far as 20 km in to the interior. The study of the EIW tradition from both surface and excavated cultural materials suggests that three phases of this tradition can now be recognized on the Swahili coast: the Limbo, Kwale, and Mwangia. The earlier idea that the EIW tradition of the coast of Swahili consisted of Kwale ware (Soper, 1971) should now be modified to allow room for these three phases. Using C14 dates (Chami, 1998), the Limbo phase is dated from the last centuries BC to about the third century AD. The second, Kwale phase, is dated to between the third and the fifth centuries AD and the Mwangia phase to the sixth century AD.

A Review of Swahili Archaeology

209

The three phases have also been distinguished by studying the pottery. Regarding shape, all three traditions have necked pots and bowls. The Limbo assemblage has bowls with upturned rims and the inside of the lip curving gradually to meet the body. The Kwale bowls have an obtuse angle which separates the up-turned rim from the body (Fig. 2, Gr. A, row 1, fourth sherd; row 2, first sherd; Fig. 2, Gr. B, row 2, third and fifth sherd) [also compare up-turned bowls of Limbo in Chami (1994, pp. 70-71, Fig. 15, b, c, e, i, j) and those of Soper (1967a, pp. 7-10)]. The necked vessels of the two phases were similar except for an introduction in the Kwale phase of a vessel type with a longer and more flared rim (Fig. 2; Gr. B, row 1, third sherd). There was no difference in shape between the Kwale and the Mwangia traditions except for a decrease in beveling. Rims continue to be thickened (Fig. 2, Gr. C). There was also not much fundamental change in the decoration style of the Limbo and Kwale traditions. A key factor separating the two traditions was the introduction of false relief chevrons in the Kwale phase, which became one of the dominant motifs (Fig. 2, Gr. B, row 2, first and second sherd). This motif has not been found in the Limbo and Mwangia phases. Also, a zigzag line of incision, which was very rare in the Limbo phase, became dominant in the Kwale phase (Fig. 2, Gr. A, row 1, third sherd; Gr. B, row 1, third sherd). This decorative element is pronounced and bold in the Mwangia tradition (Fig. 2, Gr. C, fourth sherd) and continued as such in the later traditions (Fig. 2, Gr. D). There were some other motifs that appeared for the first time in the Kwale phase and continued in use to the very later traditions of the coast, including triangles (Fig. 2, Gr. B, last sherd), comb-stamping and punctations (Gr. B, C, and D, second column from right), and vertical and slightly oblique bolded incisions (Gr. B, C, and D, third column from left). There was also a gradual and continuous tendency toward bold motifs from the Limbo base to the Mwangia phase. By the time of the Mwangia phase all the decorative motifs are no longer "neat"; they are all bold and somehow roughly made, a tendency that continues in the TIW tradition (Fig. 2). The tradition that succeeded the EIW around the sixth century AD has been termed Triangular Incised Ware (TIW) (Chami, 1994). As already noted, this tradition was previously thought to have dated to after the eighth century AD. However, the excavation of several sites on the Tanzanian coast, including Mpiji and Kaole north of Dar-es-Salaam, Kiwangwa and Masuguru in the hinterland of Bagamoyo, and Misasa south of Dar-es Salaam, suggests that this coastal tradition flourished from the sixth until the tenth century AD (Chami, 1994, 1994-1995). Ongoing and completed research in the site of Dakawa (Haaland, 1994-1995), Unguja Ukuu (Juma,

210

Chami

1996), and Pemba (LaViolette and Fleisher, 1995) and near the Rufiji delta (see above) supports this cultural periodization. As noted above, by 1990 the dominant idea was that the founders of the TIW settlements had originated from the northern coast in the eighth century AD, a date that coincided with the coming of Islam to East Africa. The study of TIW cultural materials recovered from the newly found sites now suggests that there were early and later phases of the TIW tradition. The early phase of this culture was pre-Islamic, that is, before Islam was introduced into East Africa in the eighth century AD. For instance, some potters in the early phase of the TIW tradition continued to thicken, and occasionally bevel, the vessel rims (Fig. 2, Gr. D, row 2, last bottom sherd). Bowls of up-turned rims continued to appear with vestigial elements of thickening and beveling (Fig. 2, Gr. D, row 2, second sherd). The necked vessels were more open with the maximum body diameter only slightly large than the neck diameter (compare Fig. 2, Gr. A, row 1, first sherd, and Gr. D, row 1, last sherd, and row 2, fourth sherd). This TIW necked-vessel shape has been labeled elsewhere as "bag-like" (Chittick, 1974, p. 320). The decorative motifs of the early phase TIW assemblage had become quite different from those of Limbo and Kwale assemblages, but maintained a similarity to that of Mwangia. Thus, several vestigial elements of EIW pottery tradition can still be observed in the early-phase TIW pottery assemblage (see above; Chami, 1994, pp. 72-73). The above analysis suggests that the people of the TIW tradition were the descendants of the EIW people, who exploited the same environment utilized by their ancestors. Indeed, TIW archaeological horizons overlying those of EIW have been observed in two excavated sites near the Rufiji Delta (Chami and Mapunda, 1998; Chami, 1998). The coexistence of people of the early phase TIW tradition and those of the last phase of the EIW (Mwangia) tradition can be shown to have taken place in the sixth century AD. In this early phase of the TIW tradition, iron-smelting continued unabated, with copper, lead, and bronze now used. Many grooved coral, sandstone, and pottery objects are found which indicate bead making or the sharpening of tools (Chami, 1994, p. 63; Chami and Mapunda, 1998). The remains of the imported materials in this early phase of the TIW tradition are non-Islamic and are mainly of early alkaline glass, glass beads, and green and blue glazed pottery with the white/milky paste of the Sassanid period (previously known as Sassanian Islamic) (Chami, 1994, p. 66). The later-phase TIW tradition was of the Islamic period. This is firmly dated to between the eighth and the tenth centuries, agreeing with the date attributed to the northern Kenya settlements. This later phase TIW

A Review of Swahili Archaeology

211

tradition lost most of its EIW cultural background. Rims of vessels were rarely thickened. Vestigial up-turned bowls were rare and this category of vessels was dominated by globular restricted and open bowls which were all undecorated. These were similar to those of its succeeding PW tradition (Fig. 2, Gr. E, second and third sherds). The later-phase TIW pottery assemblage also had a declining number of decorative motifs and a declining frequency of decorating. Triangles (Fig. 2, Gr. D, last column), and zigzag incised lines, both single and double (Gr. D, middle column), were dominant. Many technological features of the early-phase TIW tradition are carried forward; grooved objects, slag, and metals of different types are found in the later-phase TIW archaeological contexts. Islamic products, including white-glaze and color-splashed pottery and glass ware are dominant. Beads were of the "trade wind" type and were mainly from India (Chami, 1994). Recent rethinking (Horton, 1996, p. 408) and reports (Wilson and Omar, 1996) from the northern Kenyan coast suggest that the TIW sites there fit more closely in the later phase of the TIW tradition. This would imply that the northern TIW tradition flourished later, apparently with influence from the coast of Tanzania, as "the grouping around the coast of mainland Tanzania and Zanzibar are producing consistently earlier dates" (Horton, 1996, p. 408). If this is the case, as it seems now, it would confirm the idea that cultural forces leading to the formation of the TIW culture were first realized on the southern coast. We now know that the southern coast, especially that of Tanzania, is seen as the core of the ancient coastal EIW tradition (Chami, 1994). However, there is a sign that calls for caution from the hinterland of the northern Kenya coast, in the upper Tana where EIW sites are now being found, associated in one area with those of the TIW tradition. Although proper study of these sites is yet to be conducted, it has been suggested that the TIW sites are later than those of the EIW (Kiriama et al., 1996). No cultural connection has been suggested yet, but there is an indication that the two type of sites derived their influences from the southern coast. A similar picture of development between the EIW and the TIW cultures is also conceived for the Mozambican coast (Sinclair et al., 1993). Archaeological research on the northern and southern coast now suggests that in the eleventh century the two coastal areas went into two separate cultural developments. This phenomenon lasted for about two centuries before a common East African pan-Swahili culture was readopted. While the northern coast carried on, heavily based on the earlier TIW tradition, the southern coast entered into a new culture which is archaeologically represented by the PW tradition. Here the pottery has long, flared, and thin rims, and in many cases the neck is wider than the maximum body diameter (Fig.

212

Chami

2, Gr. E; see also Chami, 1994, p. 80, a). This tradition has been labeled Plain Ware because its makers in the tenth to the thirteenth centuries did not decorate their vessels except for a rare line of punctation around the shoulder or some marks on the lip (Fig. 2, Gr. E). Chittick (1974) has presented this tradition as "period two pottery," dating it from the twelfth to the thirteenth century AD. This same pottery has recently been excavated from Kaole, Changwehela, Kwale Island, and Kivinja, sites all located between Bagamoyo and the Rufiji delta. Radiocarbon dates from these sites firmly put the tradition between the tenth and the thirteenth centuries AD (Chami, 1994, p. 91; Chami and Msemwa, 1997b). The pottery has also been found in many other sites in the immediate hinterland of the southern coast both on the surface and in excavation but is yet to obtain widespread recognition (Chami, 1992; Schmidt et al., 1992; Chami and Kessy, 1995; Chami and Mapunda, 1998). Apart from a few imports, e.g., sgraffiato ware, which is found associated with this pottery in Kilwa (Chittick, 1974), Kaole (Chami, 1994), and Kwale Island (Chami and Msemwa, 1998), many sites in the PW tradition, especially those on the littoral and immediate hinterland, lack remains of imports, suggesting a self-sustaining society or one that is cut off from external trade. The wealth during this period seems to have been concentrated on the islands, where settlements like Kilwa in Tanzania and Mwali Mjini in the Comores grew economically (Chittick, 1974). Kilwa and Shanga started to mint coins, an economic activity maintained at Kilwa into the sixteenth century (see Chittick, 1974; Horton, 1996). It is also during this period that most of the island settlements, both in the north and in the south, started experimenting with coral and lime in monument constructions. As noted above, the northern coast did not adopt the PW tradition. The area north of Tanga probably developed with most of the elements of the TIW tradition, but also introduced new innovative traits. This continued TIW tradition is known there as Tana tradition Phases B and C (Horton, 1996) or Periods II and III (Wilson and Omar, 1996). Newly introduced traits include a vessel shape with a vestigial neck, so that the vessel looked more like a bowl than a necked vessel (Fig. 2, Gr. F). There were frequent cases of raised vessel shoulders which was a tendency toward carination (Fig. 2, Gr. F, first sherd; Horton 1996, Fig. 171, c, and Fig. 178 e). There was also a tendency to decorate the vestigial neck with a line of punctation that was bolder and more irregular than that found in the earlier traditions (Fig. 2, Gr. F, third sherd). Carinated vessels had the area between the raised shoulder and the neck decorated with cross or oblique hatches (Fig. 2, Gr. F, first two sherds). Occasionally, triangles appeared on the neck (Wilson and Omar, 1996, p. 549). This tradition has been termed Neck-

A Review of Swahili Archaeology

213

Punctating (NP) because of the high frequency of this feature (see also Radimilahy, 1998). The economic and cultural dichotomy that emerges between the northern and the southern coast, and between the islands and the mainland of the southern coast, calls for explanation. I have already suggested that the demise of the early phase TIW settlements followed by the rise of later phase TIW settlements located mainly on the islands and on the northern coast is an indication of upheaval (Chami, 1994, pp. 99-100). I suggested that this upheaval is reminiscent of a similar change caused by Islamic incursions in the Middle East and elsewhere in North Africa and Asia. Islamic traders and missionaries would have come to East Africa, creating pockets of Muslim settlements. The new settlements, including African converts, would then have been used as places from which to raid resistant non-Muslim communities. Al-Masud found such a newly Islamicized community on an island known as Qanbalu between 916 and 917 AD; the island is thought to be either Pemba or Zanzibar (Freeman-Grenville, 1962). In the ninth century, Arab travelers also recorded communities that still observed African religion. These had elected kings, and their troops made war on one another. They also had preachers (holy men) who delivered sermons in their own language (Chittick, 1975, p. 23). It is logical that, as time went on, some of the newly created Muslim communities, located on the islands and on the near shore, could have become antagonistic toward the hinterland indigenous African communities. The former would control trade links to the north Indian Ocean and could therefore suffocate their enemies economically. War captives could then have been sold into slavery. It is also apparent that the northern coast, which would have adopted Islam earlier, was distinguishable from the lessIslamic southern coast. This division between the north and the south of the Swahili coast between 1000 and 1300 AD is clearly reflected in Arabic documents. The north is called Bilad az-Zanj and the south Bilad as-Sufala (Trimingham, 1975, p. 120), which fits clearly with our new archaeological observations. The southern coast becomes economically stronger, more artistic, and more aesthetic at the end of the thirteenth century. It is at this time that the whole cultural package, including pottery type, that developed in the north is adopted along the southern coast as far as Madagascar (Radimilahy, 1998). The type of pottery representing this pan-East African coastal cultural tradition was recognized by Chittick (1974) as Early Kitchen Ware type 3-5. This is the NP pottery tradition described above. This cultural tradition, which was first realized on the northern coast around the eleventh century and matured there around the beginning of the fourteenth century, is hereafter viewed as Swahili for the following reasons.

214

Chami

The spread of this tradition all over the coast of Eastern Africa at the end of the thirteenth century is a sign of a spread of a larger cultural package known as the Swahili culture. Linguists agree that the language of this culture first developed on the northern coast between the Pangani River and Lamu Archipelago (Fig. 1). The people of the northern coast, as already suggested, adopted Islam earlier than those in the south due to their proximity to the Arab world. When the protracted struggle of Muslim converts replacing African ideology with Islam was almost completed around 1300 AD, reamalgamation of the north and south became possible. In about 1331 AD, the Arab traveler Ibn Battuta found the Muslims of Kilwa still fighting because their country was "contiguous to the heathen" (Gibb, 1939, p. 112). It seems, therefore, that Islam became a ruling idea only after 1300 AD. The Swahili language was then effectively adopted as a lingua-franca for the whole coast of East Africa. It is at this time that the word "Swahili" appears in the Arabic documents, when Ibn Battuta refers to the Zanj territory as "Sawahil country" (Gibb, 1939, p. 112), meaning the "coastal land" ("Sahel" means "coast", pl. Sawahil, adj. Sawahli or Sawahili). This Swahili cultural period, 1300 to 1500 AD, was of enormous prosperity, with Kilwa on the southern coast becoming the most important settlement of the whole coast of East Africa. It seems that the first century of this period, 1300-1400 AD, was formative in terms of culture consolidation, trade expansion, and capital accumulation. The archaeology of this period exhibits a good amount of imports from the Middle East, China, and India, including glazed and unglazed ware, glasses, and beads. Coral house floors and lime concrete are found which suggests an improvement in housing conditions (Chittick, 1974; Chami, 1994). Ibn Battuta is again our first-hand informer as he visited Mombassa and Kilwa in this formative period. He described two "substantial towns" with all their buildings made of wood (Gibb, 1931, p. 112). In the fifteenth century, the communities are very affluent. Hundreds of monuments of large size, now in ruins, can be seen dotting the littoral and islands from the north coast of Kenya to the Comores and northern Mozambique. They include mosques, palaces, defense walls, and tombs, all built in coral and mortar with porite door and window frames that have beautiful curved panels (for conspectus see Chittick, 1974; Sutton, 1990). This rich cultural resource is the one that has enticed many archaeologists and historians to the coast of East Africa. Major archaeological works have been conducted in many of these monumental sites, including those of Kirkman (1964), Chittick (1974, 1984), Wilson (1982), Duarte (1993), and Horton (1996). Various trade goods have been found in the sites, which suggests far-reaching trade connections. Gold from Zimbabwe is thought

A Review of Swahili Archaeology

215

to have propelled this prosperity. Kilwa coins of this period have been found in many archaeological contexts in the region, including Great Zimbabwe (Sutton, 1990). Post-1500 AD Portuguese documents provide information about what the Swahili civilization looked like at the end of the fifteenth century AD. Different from what Ibn Battuta saw a century earlier, Vasco da Gama found Kilwa to be a very impressive city. The city was large and, as well as containing wooden houses, had "good buildings of stone and mortar with terraces" (Freeman-Grenville, 1962, p. 66).

CONCLUSION

Recent archaeological finds from the Swahili coast suggest that the ancient coast of East Africa was occupied by EIW Bantu speakers during the first five centuries AD. The formally adopted idea that the Bantu speakers had not reached the coast in those early centuries (Horton, 1990, Sutton, 1994-1995, p. 231) should not, therefore, be sustained. EIW sites have been occupied by Bantu speakers (Phillipson, 1993; Soper, 1971) and finds on islands and the littoral of the East African coast should confirm this. Indeed, even the main opponent of this idea is slowly changing his mind in its favor (Horton, 1996, pp. 410-411). The people of the TIW tradition, who lived on the Swahili coast between the sixth and the tenth centuries AD, were of Bantu speaking (Chami, 1994), rather than Cushitic-speaking (Horton, 1990), origin. From the study of TIW cultural materials, it can be seen that their cultural elements resemble those of the EIW people (Fig. 2). The date of the formation of this tradition and the environment in which it first evolved together suggest that it is the same EIW Bantu speakers who transformed their culture into a new one. Explanations for this cultural change have been given elsewhere (Chami, 1994), where causal factors such as new trade opportunities, a growing population, and an adoption of new technologies have been suggested. This tradition probably flourished on the northern Kenyan coast at the beginning of the eighth century AD. The emergence of an economically and aesthetically poor PW tradition on the southern coast between the eleventh and the thirteen centuries AD, at variance with the continued TIW tradition on the northern coast, suggests a cultural crisis on the general coast of East Africa. I have suggested here and elsewhere (Chami, 1994) that the coming of Islam to East Africa after the eighth century AD created antagonism between the Muslim converts and the African orthodoxies. It is no wonder that the latter group, being non-Muslim, was targeted for slavery, as attested to in the Middle

216

Chami

East in the tenth century AD (Hodges and Whitehouse, 1983, p. 151). Military conquests, as described by Ibn Battuta, were necessary to gain a larger territory and slaves for trade. Strategic settlements of the newly converted Muslims grew on the isolated islands, i.e., the Lamu Archipelago and the Mafia Archipelago and Kilwa. This is reflected archaeologically in the relative wealth found in such settlements as opposed to those on the littoral and the hinterland. It is also at this period that the north appeared more prosperous and artistic than the south. The pan-Eastern African Swahili speaking and Islamic culture matured at the end of the thirteenth century AD. Local pottery recovered from the monumental and other sites is the same all over the coast of East Africa, as far as northern Madagascar (Radimilahy, 1998). As already suggested by Mathew (1963) and implied by Ibn Battuta (Gibb, 1939, p. 112), the burgeoning economy reflects the exploitation by local people of various opportunities, including new Islamic ideology, trade, and expanded cultural contacts. It was a long protracted struggle that amassed experience and wealth. The minting of coins at Kilwa, which lasted for about 500 years, from 1100 to 1600 AD, confirms gradual economic growth that culminated in the great prosperity of 1400 to 1500 AD. Wealth was now concentrated around Kilwa, a complex of settlements controlling the gold trade route to Zimbabwe. This was the culture found by Europeans when they penetrated the coast of East Africa in the sixteenth century AD. Their first-hand narratives describe the settlers of these settlements as "moors" or "Suaili," as opposed to "Arab," "for those whose home was, or had been, Arabia" (Allen, 1993, p. 3).

REFERENCES Abungu, G. (1989). Communities on the River Tana, Kenya: An Archaeological Study of Relations Between the Delta and River Basin 700-1800 AD, Ph.D. thesis, Cambridge University, Cambridge. Abungu, G. (1994-1995). Agriculture and settlement formation along the East African coast. Azania 29-30: 248-256. Allen, J. de V (1977). Settlement patterns on the East African coast, c. AD 800-1900. In Leakey, R. E. and Ogot, B. A. (eds.), Proceedings of the 8th Panafrican Congress of Prehistory and Quaternary Studies Nairobi, The International Louis Leakey Memorial Institute for African Prehistory, Nairobi, pp. 361-363. Allen, J. de V (1982). The "Shirazi" problem in East African coastal history. Paideuma 28: 9-28. Allen, J. de V (1993). Swahili Origins, James Curry, London. Chami, F. A. (1992). Limbo: Early iron-working in south-eastern Tanzania. Azania 27: 45-52. Chami, F. A. (1994). The Tanzanian Coast in the First Millennium AD, Uppsaliensis SAA 7, Uppsala. Chami, F. A. (1994-1995). The first millennium AD on the East Coast: A new look at the cultural sequence and interactions. Azania 29-30: 232-37.

A Review of Swahili Archaeology

217

Chami, F. A. (1996). The excavation of Kiwangwa Last Stone Age site. In Pwiti, G., and Soper, R. (eds.), Aspects of African Archaeology, University of Zimbabwe, Harare, pp. 307-316. Chami, F. A. (1998). The ancient cultural sequences for the central coast of East Africa: New perspective. In Sinclair, P., Mutoro, H., and Abungu, G. (eds.), Aspects of Urbanism, Papers from the World Archaeological Congress on "Urban Origins in East Africa" held in Mombassa, 1993 (in press). Chami, F., and Kessy, E. (1995). Archaeological work at Kisiju, Tanzania, 1994. Nyame Akuma 43: 38-45. Chami, F., and Mapunda, B. (1998). The 1996 archaeological reconnaissance north of the Rufiji Delta. Nyame Akuma 49 (in press). Chami, F., and Msemwa, P. (1997a). A new look at culture and trade on the Azanian coast. Current Anthropology 38(4): 673-677. Chami, F. A., and Msemwa, P. (1997b). The excavation and Kwale island, south of Dar-esSalaam. Nyame Akuma 48: 45-56. Chittick, N. (1961). Kisimani Mafia: Excavations at an Island settlement on the East African coast. Antiquities Division Occasional Paper 1, Dar-es-Salaam. Chittick, N. (1974). Kilwa: An Islamic Trading City on the East African coast (2 vols.), The Finds, British Institute in Eastern Africa, Nairobi. Chittick, N. (1975). The peopling of the East Africa coast. In Chick, N., and Rotberg, R. (eds.), East Africa and the Orient: Cultural Syntheses in Pre-Colonial Times, Africana, New York, pp. 16-43. Chittick, N. (1984). Manda: Excavations at an Island Port on the Kenya Coast (Memoir 9), British Institute in Eastern Africa, Nairobi. Coupland, R. (1956). East Africa and Its Invaders, Clarendon Press, Oxford. Cruz e Silva, T. (1977). First indication of Early Iron Age in southern Mozambique: MatolaIV 1/68. In Leakey, R. E., and Ogot, B. A. (eds.), Proceedings of the 8th Panafrican Congress of Prehistory and Quaternary Studies, International Louis Leakey Memorial Institute for African Prehistory, Nairobi, p. 349. Duarte, R. T. (1993). Northern Mozambique in the Swahili World, Edwardo Mondlane University, Maputo. Freeman-Grenville, G. (1962). The East African Coast: Select Documents from the First to the Earlier Nineteenth Century, Clarendon Press, Oxford. Freeman-Grenville, G. (1963). The coast, 1498-1840. In Oliver, R., and Mathew, G. (eds.), History of East Africa, Clarendon Press, Oxford, pp. 129-168. Gibb, A. (1939). Ibn Battula Travels in Asia and Africa, George Routledge and Sons, London. Haaland, R. (1994-1995). Dakawa: An early iron age site in the Tanzanian hinterland. Azania 29-30: 238-247. Hodges, R., and Whitehouse, D. (1983). Mohammed, Charlemagne and the Origins of Europe, Cornell University Press, New York. Horton, M. (1984). The Early Settlements of the Northern Swahili Coast, Ph.D. thesis, Cambridge University, Cambridge. Horton, M. (1987). Early Muslim trading settlements on the East African coast: New evidence from Shanga. Antiquaries Journal 67: 290-322. Horton, M. (1990). The Periplus and East Africa. Azania 25: 95-9. Horton, M. (1996). Shanga: The Archaeology of a Muslim Trading Community on the coast of East Africa, British Institute in Eastern Africa, London. Hollingsworth, L. (1951). A Short History of the East Coast of Africa, Macmillan, London. Juma, A. (1996). The Swahili and the Mediterranean worlds: Pottery of the late Roman period from Zanzibar. Antiquity 70: 148-154. Kiriama, H., Mutoro, H., and Ngari, L. (1996). Iron working in the upper Tana valley, Kenya. In Pwiti, G., and Soper, R. (eds.), Aspects of African Archaeology , University of Zimbabwe, Harare, pp. 505-507. Kirkman, F. (1964). Men and Monuments on the East African Coast, Lutterworth Press, London.

218

Chami

LaViolette, A., and Fleisher, J. (1995). Reconnaissance of sites bearing Triangular Incised (Tana Tradition) Ware on Pemba island, Tanzania. Nyame Akuma 44: 59-65. Mathew, G. (1963), The East African coast until the coming of the Portuguese. In Oliver, R., and Mathew, G. (eds.), History of East Africa, Clarendon Press, Oxford, pp. 94-128. Nicholls, C. (1971). The Swahili Coast , George Allen and Unwin, London. Nurse, D., and Spear, T. (1985). The Swahili, University of Pennsylvania Press, Philadelphia. Oliver, R. (1966). The problem of the Bantu expansion. Journal of African History 7: 361-76. Phillipson, D. (1979). Some iron age sites in the lower Tana valley. Azania 14: 155-60. Phillipson, D. (1993). African Archaeology, Cambridge University Press, Cambridge. Radimilahy, S. (1998). Mahilaka: An Archaeological Investigation of an Early Town in NorthWestern Madagascar, Uppsaliensis, Uppsala (in press). Robertshaw, P. (1990). Early Pastoralists of South-western Kenya, British Institute in Eastern Africa, Nairobi. Schmidt, P. (1994-1995). The agricultural hinterland and settlement in Tanzania. Azania 29-30: 261-262. Schmidt, P., Karoma, J., LaViolette, A., Fawcett, A., Mabulla, L., Rutabanzibwa, L., and Saanane, C. M. (1992). Archaeological Investigations in the Vicinity of Mkiu, Kisarawe District, Tanzania, University of Dar-es-Salaam, Dar-es-Salaam. Sinclair, P. (1982). Chibuene: An early trading site in southern Mozambique. Paideuma 28: 149-64. Sinclair, P. (1987). Space, Time, Social Formation: A Territorial Approach to the Archaeology and Anthropology of Zimbabwe and Mozambique, ca. 0-1700 AD, Uppsaliensis, Uppsala. Sinclair, P. J., Morais, J., Adamowicz, L., and Duarte, L. (1993). A perspective on archaeological research in Mozambique. In Shaw, T., Sinclair, P., Andah, B., and Okpoko, A (eds.), The Archaeology of Africa: Food, Metals and Towns, Routledge, London, pp. 1-31. Soper, R. (1967a). Kwale: An Early Iron Age site in south-eastern Kenya. Azania 2: 1-17. Soper, R. (1967b). Iron Age in north-eastern Tanzania. Azania 2: 19-37. Soper, R. (1971). A general review of Early Iron Age of the southern half of Africa. Azania 5: 5-37. Sutton, J. (1990). A Thousand Years of East Africa, British Institute in Eastern Africa, Nairobi. Sutton, J. (1994-195). East Africa: Interior and coast. Azania 29-30: 227-2-31. Trimingham, J. (1975). The Arab geographers and the East African coast. In Chittick, N., and Rotberg, R. (eds.), East Africa and the Orient, Africana, New York, pp. 115-146. Wilson, T. (1982). Spatial analysis and settlement patterns on the East African Coast. Paideuma 28: 201-219. Wilson, T., and Omar, L. (1996). Excavation at Pate on the East African coast. In Pwiti, G., and Soper R. (eds.), Aspects of African Archaeology, University of Zimbabwe, Harare, pp. 543-554. Wright, H. (1984). Early seafarers of the Comore islands: The Dembeni phase of the ixth-xth centuries AD. Azania 19: 13-60.