1The difference in conceptualizing a geographic region as a hinterland rather than as a motherland is like the dissimilarity in being remembered as bystanders to an event rather than as people who made history. Today, along the central-Tanzanian coast and well into its interior live people who speak languages belonging to the Ruvu subgroup of languages descended from proto-Northeast-Coastal Bantu. Most of those people also speak Swahili. Swahili is a language that belongs to the Sabaki subgroup of proto-Northeast Coastal Bantu. An important distinction between Swahili and the Ruvu languages is that Swahili is not indigenous to the area. Though they share a common language ancestor in proto-Northeast-Coastal Bantu, Swahili evolved into a distinct language from its proto-Sabaki ancestor along Kenya's southern coast. Proto-Ruvu arose in central-east Tanzania.

2The Swahili language took root in the central-east-Tanzania landscape as a result of mixed events over the long run of history that began just before the start of the second millennium CE. That was many generations after early Ruvu speech communities had lived in the coastal and interior regions of Tanzania. Over time, Ruvu-speaking people played a part in the nascent Swahili culture that developed along the East African coast. Similar developments likely transpired among communities of people who lived all along the long narrow coastlines that are commonly termed the "Swahili Coast" or "Swahili Corridor." Yet minimal scholarly attention has been given to the histories of communities who were in those regions long before and long after Swahili speakers emerged. Those people and their homelands have been largely termed the "hinterlands." This chapter offers an alternative approach to understanding the complexities of "hinterland" histories.

3In this chapter I comb the findings of archaeology and comparative-historical linguistics to build a narrative account of Bantu immigration and settlement in central-east Tanzania. My goal is to bring new attention to the early diversity of the Bantu populations who converged in central-east Tanzania after they departed at various times and along independent trajectories from the Southern Nyanza Basin during the last millennium BCE and following the BCE/CE changeover. Each of those communities has left material and, in some cases, linguistic evidence of their having been in those zones. Eventually, however, the last arriving of those early Southern-Nyanza-Basin Bantu groups, those who are now referred to as proto-Northeast-Coastal-Bantu-speaking populations, came to predominate in the central-east Tanzania region. They and their later descendant-language speakers formed what I term the Northeast-Coastal Culture Zone.1 The zone comprised an area whose borders were continually reconfigured by Northeast-Coastal Bantu descendants who migrated, created new settlements, diverged into new language communities, and intermingled with distinct populations.2 In the last 1,400 years Ruvu language speakers have been the most southern-dwelling language descendants of PNECB. 3 And it is to them and their history that the latter half of the chapter and the remainder of the book give emphasis.

^top

Mashariki-Speaking Communities Move Into Tanzania

4The last centuries BCE and the early centuries of the first millennium CE ushered in a time of considerable population movement, characterized by both distant migrations and in situ reconfigurations in eastern, central, and southern Africa. Into those regions, Mashariki-Bantu language descendants belonging to the Kaskazi and Kusi language subgroups made their way from Africa's Great Lakes regions toward the far eastern and southern shores of the African continent. But they did not all migrate in the same way or all at once. They followed trajectories and timelines unique to each of their emerging speech communities. For instance, early Kusi societies initially made their homes in the middle areas of Lake Tanganyika's western shores. They later moved southward along its coastline, spreading their languages and communities into what is Zambia today. Later, some of their language descendants moved into the entire southeastern region of Africa enfolded between the northern areas of Mozambique and South Africa. Kaskazi-descended language speakers, on the other hand, likely lived in the areas north of their Kusi neighbors, but they also remained in the Great Lakes' western territories. They initiated movements toward Lake Tanganyika's eastern provinces via the avenue flanked by Lakes Tanganyika and Nyanza (Victoria).

5In the last few centuries of the first millennium BCE, southward-bound Kusi and Kaskazi-speaking people made their homes within the borders of what is now Tanzania. It seems that Tanzania's ecological diversity may have been a magnet for people looking for new places to root themselves. Its landscape reflects the physical geography resulting from core level continental shelf shifts that have created an array of topographical features whose varying elevations and correlated ecosystems run the gamut of possibilities. Among some of the most prominent environs within its boundaries include significant portions of the Western and Eastern Rift Valleys, which resulted from a subterranean tug-of-war that over millennia have thinned the continental floor, creating extensive valleys as well as formidable rift walls that continue beyond the borders of Tanzania to both the north and south. Beyond the rift valleys, but fundamentally connected to the underlying earth movements that create them, are the area's striking number of such mountains as Kilombero, Uluguru, Usambara, Pare, and the well-known Mount Kilimanjaro along Tanzania's northern border. Also of significance are its sprawling escarpments, which give way to generous plains that lead to the East African coast. Additionally, Tanzania has advantageous lake and basin environs shaped by Lake Nyanza on its northern border, Lake Tanganyika on its western border, and Lake Nyasa (Malawi) at its southwestern end, as well as many additional lesser lakes interspersed throughout. Connected to these are countless river and stream systems.

6With such diversity, Tanzania presumably offered environmental niches that appealed to peoples of myriad subsistence practices. They permitted them to carve out settlement areas suited to their needs. Like Eastern Sahelian and East- and West-Rift Southern Cushitic speakers as well as people of Khoisan language origins who had long made their homes in its diverse climes, various Kaskazi- and Kusi-language speech communities filled into the interstitial geographic arenas suited to their agricultural and other subsistence methods within Tanzania. They did this in little more than half a millennium.4 In all likelihood myriad circumstances motivated them. Some may have been prompted by problems associated with increased population densities in the regions farmed successfully by their ancestors as well as, possibly, soil degradation that can be intensive agriculture's unfortunate effect. But others may have moved for the chance to pursue more prosperous opportunities. Of course, their particular reasons evade recovery.

7During the last centuries of the last millennium BCE, clusters of southern Kaskazi communities settled into the northwestern regions of Tanzania, south and southwest of Lake Nyanza (Southern Nyanza Basin). At various times during the succeeding centuries several populations among them initiated migrations farther afield. Those who stayed in the vicinity, undergoing language divergence in situ, included the Lakes language communities, who moved around the Western Rift and southwestern Lake Nyanza areas.5 Some groups departed the area altogether. For instance, some populations established themselves in southern regions of Tanzania. They eventually comprised the Rungwe, Rufiji-Ruvuma, and Kilombero language subgroups.

8Over time, other Southern-Kaskazi populations embarked on easterly movements away from the Southern Nyanza Basin. They formed the ancestral Kati, pre-Langi, and Upland speech communities.6 In the centuries leading up to the end of the millennium, eventual Upland speakers moved out of the southern Nyanza areas in an easterly direction across north-central Tanzania. They settled into regions along the highland and coastal interior areas of eastern Kenya and northeastern Tanzania. About the same time, ancestral Langi-speaking people moved into the middle areas of north-central Tanzania, where they intermingled extensively with West- and East-Rift Southern Cushitic speakers and, to a lesser degree, Eastern Sahelian speakers in subsequent centuries.7

9In contrast, the Kati community remained in the vicinity of the Southern Nyanza Basin for a period after their Langi and Upland relatives departed. Eventually, however, it diverged into three speech communities. One Kati-language descendant group called Takama remained in the southern Nyanza area. Meanwhile, Kati's emergent Njombe and Northeast-Coastal speakers left toward lands far removed. Early Njombe communities initiated a southerly trajectory toward the Tanzanian southern highlands. About the same time, soon-to-be proto-Northeast-Coastal Bantu speakers moved toward Tanzania's central-east hinter-coast. It is to those regions that we now turn our attention.

^top

Azania's (Tanzania's) Early First Millennium CE Hinter-Coastland Milieu

10In the middle of the first century CE, a Greek-speaking Egyptian man from the Red Sea port city of Berenike authored a merchant handbook, the Periplus Maris Erythraei.8 It is thought that the writer was a former merchant who had extensive personal experience moving along most, if not all, the well-trodden sea routes bordering Roman Egypt, southern Arabia, India, and East Africa. In the handbook the author provided information that Indian Ocean merchants would have found valuable. For instance, he covered the types of goods that were in demand and made available at each port, as well as the names of local leaders at several of the harbors and in the inland environs. Potential readers were also told which posts had hospitable hosts and which were less welcoming. Seemingly of secondary importance, albeit worthy of inclusion, was sailing information concerning trade routes.9 Whatever the author's motive was for writing it, the document is now a rare gem because it is the oldest-known recorded source of information about the early commercial activities that took shape along East Africa's coast in the early centuries of the first millennium CE.

11Concerning East Africa, the Periplus provided detailed information about the routes between Roman Egypt's Red Sea port of Myos Hormos and Rhapta. Rhapta was described as a bustling metropolitan area located somewhere just inland from the central coast of mainland Azania, perhaps in the vicinity of modern-day Bagamoyo and Dar es Salaam.10 It also indicated that Rhapta was the most southerly stopping point along the East African commercial route, and that merchants interested in making a round trip to Rhapta from one of Roman Egypt's Red Sea ports had to be ready to depart in July because that was when the climate cycle necessary to make the voyage began. After making various port stops, whether to conduct business or to wait for optimum monsoon wind conditions, vessels typically arrived in Rhapta in November or December. Once they disembarked at Rhapta, merchants were obliged to remain a minimum of eight months before departing because it was in August that an incipient wind pattern allowed for measured movement north toward Kenya's coast. Soon thereafter a favorable climate cycle set in, permitting vessels to make the journey back to the Red Sea ports. They typically landed at their final destination in November or December of the subsequent year, a year and a half after their initial July departure. This meant that if merchants wanted to make back-to-back journeys, they would have had six or seven months to ready themselves for the necessary July departure. Although it was a long journey, the Periplus author noted that merchants had the security of traveling under relatively mild and reliable conditions.11

12The Periplus author described Azania's Rhapta as an emporium where Arabian merchants holding grants to do business met with local populations. Its indigenous inhabitants were described as "very big bodied" farmers who "behave, each in his own place, just like chiefs."12 He reported that "Arab skippers and agents who, through continual intercourse and intermarriage, are familiar with the area and its language" are found in the region. Finally, he added that a good deal of wine and grain were commonly brought, not as trade items but as gifts offered in "good will" to the locals.13 Such phrases convey the impression that Rhapta was an area where the intermixing of Arab populations with the locals was normative and fairly longstanding.

13Also among the details in the Periplus was information about the business aspect of their transactions. It noted the products desired by parties on both ends of the operation. Rhapta inhabitants sought imported iron spears, axes, knives, small awls, and glass stones. And in return they provided their associates with ivory, rhinoceros horn, tortoise shell, and, to a lesser degree, nautilus shell.14 Such details, from Rhapta's people to the products traded, suggest that Rhapta was a metropolis of commercial activity and sociocultural intermingling. Such practices likely resulted in the formation of ethnically mixed families and settlements. One of the questions that has intrigued scholars has been, who were the people the Periplus tells us were Rhapta's locals? In order to address this question we must turn our attention inward to the central-east-Tanzanian coast and its interior history.

^top

Hinter-Coastal Tanzania at the Turn of the First Millennium CE: Archaeology and Linguistics

14The Periplus has inspired Africanist scholars to search for additional evidentiary data about the Rhapta commercial era. In the past five decades archaeologists have worked carefully to recover physical evidence that might provide details about the societies and people who participated in those events. Early efforts proved disappointing to a certain extent because research carried out in what were expected to be the most promising sites for early coastal settlement did not turn up results that date earlier than the ninth century. This led at least one scholar to speculate that the evidence may have been lost to environmental degradation.15 Other scholars have suggested that Rhapta in particular and trading posts in general may have been transitory over centuries, and may thus prove elusive to pinpoint.16 But the lack of evidence for the early periods did not curb the desire to continue searching. Since the mid-1980s a new generation of Tanzania archaeologists has persisted in attempts to recover evidence. Some of their initial projects centered on areas where they anticipated inland farming settlements could be located, but since the 1990s most have prioritized sites along Tanzania's coast or immediately inland from it for excavation, with the majority of them having been centered on regions bordered by Bagamoyo and the Rufiji delta. Their finds have proven groundbreaking. Simply stated, they have unearthed evidence that hinter-coastal Bantu societies lived in the region before the start of the first millennium CE.17

15Archaeologist Felix Chami is among the most resolute proponent of what he argues is evidence that Bantu-speaking farming communities continuously lived along the central-Tanzanian hinter-coast before the start of the first millennium CE. The evidentiary basis of his position stemmed initially from research his team completed at Limbo, a site that lay 24 km (12.5 miles) from the central-Tanzanian coast and 75 km (46.6 miles) south of Dar es Salaam. Limbo yielded "sherds of the Early Iron Working (EIW) tradition, several tonnes of slag, and tuyere fragments."18 The recovered EIW/EIA pottery generated dates from as early as the second century BCE to as late as the third century CE. The general agreement has been that a date close to the turn of the era is likely most accurate. Chami concluded that the evidence of smelting materials coupled with the EIA pottery found there indicates that Limbo had been a Bantu industrial iron producing and farming site since that time. Although imported materials were not found at Limbo, based on subsequent findings he and Msemwa suggested that the region was probably home to a Bantu community engaged in iron production, farming, and local trading by the turn of the era, and that it was those Bantu people who likely "initiated the transoceanic trade" that the Periplus reported.19

16Chami's claims that Bantu farmers and iron producers were involved in the events that unfolded at the hinter-coast in those early periods are right on target, but the way he has delivered and reasoned the evidence continues to stimulate at least two lines of debate. One involves the way he has changed more than once the classification of EIA pottery found at Limbo. The second concerns the way he has used EIA evidence to challenge what he claims historians have said about the possibility of Bantu people having reached the East-African coast by the end of the last millennia BCE. Over the course of more than ten years the debates have become muddled to the point that people considering early East-African-coastal history often gloss over the details. But this has diverted us from the true value of Chami's hinter-coastal EIA evidence. Because the history suggested by the items he recovered at Limbo and subsequent EIA sites bring forth important data for discerning that period's history, it is necessary to spend some time teasing apart the debate's details.

17In his 1988 thesis, Chami identified the EIA pottery found at Limbo as fitting closely with Lelesu and Urewe assemblages.20 That same year, an article appeared in Nyame Akuma, wherein he maintained that the Limbo site yielded "an EIA [Early Iron Age] ware with affinities to Urewe ware of East and Central Africa."21 Prior to Chami's claim that the Limbo pottery relics were of Urewe/Lelesu type, archaeologists had recovered Urewe ware in the Lakes region between eastern Zaire and the northeastern side of Lake Victoria, northward to at least the Victoria Nile in the Murchinson Falls area. And identified Lelesu ware types had been previously recovered in the Sandawe and Lake Eyasi environs of north-central Tanzania.22

18Four years later, Chami retracted his Urewe/Lelesu claims in a 1992 Azania article. His new position was that "the Early Iron Age pottery seems to belong within the Kwale group." However, he noted that the "Limbo assemblage is dominated by horizontal and oblique hatched bands, as opposed to comb-stamping and criss-crossed bands found frequently with Kwale and Urewe respectively."23 His reservation about the classification is not unusual since there are typological resemblances among these wares that have continued to complicate archaeologists' ability to firmly classify them. For instance, in Soper's view, "the overall relationship was very close . . . Lelesu was intermediate, typologically as well as geographically, between Urewe and Kwale."24 One distinction among them, however, is that Kwale ware has had a much wider known distribution in hinter-coastal zones than either Urewe or Lelesu ware. For example, Kwale ware has been recovered in the North Pare mountains, the Usambara mountains, and the Ngulu hills in northeastern Tanzania, as well as along the Kenyan coast, north of the Rufiji delta, and along offshore islands in central Tanzania.25 If what Limbo revealed was in fact Urewe/Lelesu pottery, as Chami initially suggested, then Limbo represented an outlier within its distribution pattern known up to that point, but it would be an outlier that revealed new understanding about an early Bantu presence in the region. If it were a Kwale variant instead, then it would have been the earliest-known Kwale ware, as previously dated Kwale pottery dated to the third century CE.

19After advocating for a Kwale affiliation, in a 1997 Nyame Akuma article, "The Excavation at Kwale Island, South of Dar es Salaam, Tanzania," Chami withdrew the Kwale claims for the EIA stratum of pottery at Limbo. In it he offered a reason for the change and a new interpretation. He suggested that in the 1992 Azania article he was influenced by opinions held among African archaeologists of the time, namely, that known Kwale ware distributions suggested that it spread from north to east to south. Moving on, he explained that the inconsistent comments made in 1988 and 1992 could now be fully resolved because of the pottery he had since recovered at Kwale Island. In his amended view, "Limbo pottery must certainly be of Lelesu-Urewe connection." His conclusion rested on the fact that pottery found at Kwale Island "looks exactly like the Kwale variant," that is, the Kwale ware style Soper recovered in Kenya. Similar to Soper's Kwale, which shows a false relief chevron, Kwale Island too turned up the same motif, which the EIA pottery at Limbo did not. Secondly, Kwale Island pottery showed the "up turned rims," which commonly typified Kwale variants rather than styles found in the EIA Limbo stratum.26

20 Figure 2: Potsherd Illustrations from Sites of the Central Tanzania Coast Based on the Kwale Island data, Chami returned to his original opinion that the Limbo site offered evidence that Bantu-speaking Lelesu/Urewe pottery producers settled there in the first few centuries CE. He added that the Kwale variants found in later stratum at both Limbo and Kwale Island, which yielded third century CE dates, likely reflected a transition in style that correlated with the Kwale variants distributed more generally. In his view, "the original EIW [EIA] settlers on the coast were of older Urewe tradition, who after centuries of experience adopted cultural elements identified as Kwale and later on as TIW [Triangular Incised Ware] tradition."27 He agreed that, as archaeologist Soper and Phillipson intimated previously, Kwale variants have been the result of "cultural diffusion from the East African to the southern region in the third to the sixth centuries AD," presumably from Kwale ware pottery manufacturers.28 The illustration page included here reflects Chami's "transition" thesis.29

21In 1998 Chami and Mapunda published "The 1996 Archaeological Reconnaissance of North Rufiji Delta." That article contained an additional modification to the way Chami had previously named the EIA Limbo pottery. In their discussion of two survey sites, Nkukutu and Mwangia, which lay approximately 30 to 40 km south of Limbo, they noted that there were stylistic similarities shared between their wares and Limbo's EIA wares. The change was that they referenced the Limbo wares not as Urewe, Lelesu, or Kwale, but as the "Limbo tradition." Though the Nkukutu and Mwangia yields were at that point undated, they noted that the "Limbo tradition" appeared to be followed by a Kwale phase that gave way to a Mwangia period.30 In 1999, after the Kwale-like ware were shown to have dates between 200–400 CE, Chami confirmed that it indeed was suggestive of a "Limbo-Kwale interface" period about the third century CE when, as he had intimated earlier, Limbo ware had transitioned to reflect Kwale ware motifs.31

22Chami's classification shifts have raised questions, but they have not diminished the importance of the physical evidence for the historical period. A few things can be said at this point about the archaeological evidence. First, it was probably not essential to make absolute distinctions within the EIA pottery traditions found at Limbo. For the most part scholars have accepted, at least for now, a view of EIA as a mix of styles that derived from what has been termed the Chifumbaze tradition made by early Mashariki speakers up to and inclusive of the Urewe style, Lelesu, and early Kwale variants.32 Spear, for example, moves beyond the details of the classification arguments by stating that what Chami calls Early Iron Ware others call Kwale ware. But doing this obscures the weight of Chami's findings. Archaeologist Mark Horton and anthropologist John Middleton maintain that Chami's reclassification of the Early Iron Age ware "into three phases . . . is controversial and that not all archaeologists have accepted them." But they concede, as have others, and rightly so, that Chami's recoveries are important because of what they contribute to our understanding of Bantu history on the East African coast.33 At the same time, just what the items reveal has not been closely addressed. In order to understand their full implications, we must look at the physical evidence in conjunction with the linguistic data. Doing this requires that we cover briefly some misunderstandings that have transpired that involve linguistic evidence.

23In 2001, the International Journal of African Historical Studies published Chami's "A Response to Christopher Ehret's 'Bantu Expansions,'" a comment on an article by Ehret that appeared earlier that year. Within it Chami also commented on Ehret's 1998 book, An East African Classical Age. Chami opened with praise for Ehret's long-held position, and one shared by others, that the spread of Bantu languages happened over a long period of time and that it resulted in a complicated mosaic of descendant Bantu languages spoken over much of the southern third of Africa. But when it came to addressing the history of Bantu-speaking people in East Africa, Chami took exception to the way he believes Ehret characterized their historical presence there. Resting his position on the fact that he had recovered Bantu-affiliated settlements and wares more than ten years earlier, Chami claimed that "the idea advanced by Ehret that Bantu speakers had not reached the coast of East Africa by the BC/AD changeover, or it was the Cushites who found and lived in Rhapta—the metropolis of Azania—does not therefore hold water."34 In fact, Ehret did not make the claims Chami alleges. Ehret holds nearly the opposite view with regard to the possibility of a turn-of-the-era Bantu-speaking presence in coastal East Africa, and says nothing about Cushitic-speaking people playing a direct part in Rhapta's founding.

24If one turns to Ehret's position on the likelihood of a turn-of-the-era Bantu presence in East Africa, he proposes that Upland communities—who spoke Bantu-descended languages—settled, "perhaps as early as the beginning of the first millennium A.D., in wooded highland areas in northeastern Tanzania, most probably in the North Pare Mountains." Based on his awareness of the correlations in the archeological and linguistic data he added that Upland-speaking people were the likely makers of Kwale pottery. From North Pare, Ehret projected that some Upland people spread, along with their Kwale pottery, eastward to the Usambara Mountains and then made their way as far east as the warm, wooded southern "Kenya coastlands before the third century."35 In all references to the Upland communities Ehret left open the possibility that they may have reached the coast earlier than the third century.

25Beyond that, in reference to the point about "Cushites" being the creators of Rhapta, Ehret argues that Southern Cushitic ancestral Ma'a speakers belonging to the Mbuguan Southern Cushitic group "probably resided in the early first millennium A.D. in the central Maasai Steppe and in or near the inland parts of the Ruvu and Wami River watersheds of central eastern Tanzania."36 He does not suggest a direct connection between Cushitic speakers and the people who may or may not have lived at or created Rhapta.37 Ehret's position is that "the overall evidence for Kaskazi expansion favors the conclusion that the majority of these local people, at least through Rhapta's halcyon days from the first to the fourth centuries, would probably have been of Northeast-Coastal Bantu affiliations in language and in culture."38 Those "affiliates" would have comprised people speaking languages of close genetic relationship to NECB, and they well could have preceded NECB people in the region. Aware that one cannot claim outright that Bantu linguistic descendants preceded NECB in central-east Tanzania without solid evidence, Ehret hypothesized this based on his knowledge of the known patterns of Kaskazi-Bantu divergence.

26When it came to the question of which people were at Rhapta, Ehret took a stance that lends support to Chami's EIA Bantu hypothesis. Had Chami called on the linguistic record, he might have argued that the relics he yielded at such EIA sites as Limbo and Nkukutu provided some physical evidence, namely, EIA pottery and iron (Limbo only) that added weighty support to Ehret's prediction that ethno-linguistically distinct but closely related Bantu-speaking people preceded the emergence of proto-Northeast-Coastal Bantu speakers in Tanzania. To understand how we can arrive at this conclusion, the correlations between pottery, iron production, and language relationships need to be laid out.

27The relationship between Urewe, Lelesu, and Kwale pottery and language has been a topic of discussion for more than twenty years. As noted above, archaeologists early on recognized that they were stylistically variable over time and across regions, but that they also shared related histories, perhaps ultimately connected to Chifumbaze ware.39 Archaeologists recognized that Urewe ware was an earlier emergent pottery tradition than either Lelesu or Kwale are known to be. In Buhaya, which lay adjacent to southwest Lake Nyanza, Urewe ware dates from 800 BCE to 800 CE.40 Along Lake Nyanza's southern shores and southwest at Uvinza, Urewe ware is also found to yield dates between 800 BCE and 800 CE. East of Lake Nyanza it has been unearthed at Gogo Falls in western Kenya, where its dates, though not chronometrically verified, are believed to be of the third century CE.41 Based on known distributions, dates, and typologies, Soper proposed, "we may therefore provisionally accept Urewe as the origin of the Early Iron Age of the rest of East Africa and . . . Buhaya as the nearest thing we have to the "cradle" of Urewe."42 About ten centuries after Urewe originated, by the third century CE, Lelesu ware was crafted in north-central Tanzania and Kwale ware in Kenya, Tanzania, and Mozambique.

28In all cases the three types of pottery are found within settlements that give every indication they belonged to Bantu farmers, who in all likelihood spoke a Bantu language. For instance, Urewe's earliest dates correlate well with the middle-Mashariki period, when Kaskazi linguistic communities began taking shape in the Lakes region.43 Although we do not know precisely the linguistic identity of Urewe-ware makers in the Lakes region between 500 BCE and about 1200 CE, in Schoenbrun's view, Urewe ware's distribution with Lakes Bantu-speaking settlements is compelling enough to postulate a correlation.44 Beyond Urewe ware, Lelesu ware's distribution in north-central Tanzania's Sandawe and Lake Eyasi areas, when coupled with the linguistic evidence for the region, suggests that ancestral Langi speakers, who were the descendants of still-earlier communities living in the Southern Nyanza area c. 400–200 BCE, were likely its makers. And Kwale ware distribution is likely associated with a distinct Southern-Nyanza linguistic community, Upland Bantu, whose ancestral language was spoken in or near southern Lake Nyanza, also about 400–200 BCE.45 The most informative way to envision the Southern Nyanza Basin between 400–200 BCE is as having been composed of clusters of languages and communities that included, at least "one Southern Kaskazi community, two Lakes Bantu communities, and the Upland and pre-Langi peoples," who lived adjacent to one another in Lake Nyanza's southwest zones.46

29The historical foundations illuminated here suggest that in the final centuries of the last millennium BCE, Southern-Kaskazi Bantu communities living in the Southern Nyanza Basin crafted early Urewe/Lelesu/pre-Kwale EIA pottery styles. Over time, their descendants expanded to disparate areas, transporting their pottery knowledge and styles with them. Working on the assumption that the Limbo items reflect Urewe/Lelesu/pre-Kwale ware affinities, the most persuasive way to interpret their origins then is to posit that Limbo's EIA settlers were among the descendant societies whose ancestors had early Southern-Nyanza-area origins in the period of 400 to 200 BCE. Those populations, as we noted earlier in this chapter, initiated movements toward the east and southeast. What is most compelling about this interpretation is that the areas abutting Lake Nyanza were home to communities having the technology from which Limbo ancestors could have acquired not just their knowledge of pottery having Urewe affinities, but also the iron-making traditions in evidence at Limbo.47

30 Map 1: Azania Bantu, c. 200 BCE-1 CE By the turn of the era their descendants, whom we can call the Azania Bantu, had moved into the Limbo area and the broader environs of central and southern Tanzania's coast. There, Azania Bantu made their variant of EIA-derived pottery (Urewe, Lelesu, pre-Kwale), produced iron, farmed the land, fished the oceans, and established domestic economies that supported their communities. They also, as Chami maintains, likely participated in the flourishing transoceanic trade networks the Periplus reported.48

31 Map 2: Azania and Upland Bantu, c. 1-200 CE In subsequent eras, when eventual Upland-Bantu speakers spread the familiar third-century Kwale ware south along the Tanzanian coast, perhaps as participants in their own right in the by-then long-established transoceanic trade, they may have brought with them the distinctive post-third century Kwale pottery variants Chami recovered at Limbo, Nkukutu, and Kwale Island. In this scenario, Upland communities joined Azania Bantu populations after the first two centuries CE. If it was indeed Upland people who migrated into these Azania regions, and not just their pottery, it suggests that these distinct but closely related Southern Nyanza Bantu descendants comprised the Bantu hinter-coastal populations living in the area by the fourth century CE.

32More archaeological research is clearly needed to shed additional light on these hypotheses. It is indeed promising that the Azania Bantu communities who moved to the hinter-coastal areas in the last centuries BCE likely left in their tracks the material evidence of iron producing, of EIA (Chifumbaze-derived) pottery, and of farming that await discovery. If the scenario above is on point, then, as Ehret proposed, Azania Bantu people spoke a language closely affiliated with the southern-Nyanza areal cluster of Kaskazi languages.49

33What has been gained by untangling the threads of debate involving the hinter-coastal zones of Tanzania in these early times is evidence that the linguistic and archaeological data are mutually supportive of there having been EIA-Azania Bantu settlements in central-east Tanzania during Azania's Rhapta period. At the same time, their settlement marked only the start of ongoing Bantu migrations into the region over the next several centuries. Subsequent to Azania Bantu settlement, and the later arrival of Upland immigrants, additional populations of Bantu-speaking newcomers entered Azania's hinter-coastal region. They were the ancestral Northeast-Coastal Bantu speakers. It was their language descendants who eventually predominated in Azania (Tanzania) for the long run of history.

^top

The Nascent Northeast-Coastal-Bantu Period

34 Map 3: Proto-Northeast Coastal Bantu, c. 300 CE The linguistic data indicate that a population that eventually spoke proto-Northeast-Coastal Bantu settled Tanzania's hinter-coastal zones in the early centuries of the first millennium CE. Ehret and linguists Thomas Hinnebusch and Derek Nurse have been attentive to questions centering on PNECB speakers' emergence and divergence in central-east Tanzania. They share fairly comparable opinions about the timing and likely settlement locations of proto-Northeast Coastal Bantu communities. In Swahili and Sabaki: A Linguistic History, Nurse and Hinnebusch propose that the PNECB homeland was likely within the area bordered by "Mombasa, the Taita Hills, the Pare Mountains, Dodoma, and the Rufiji River."50 Ehret's view is that PNECB people settled in the "immediate hinterland of the central Tanzanian coast," and that the core of their communities likely took root in the area between the Wami and Ruvu Rivers, but that many of their speakers likely expanded quickly to areas as far away as the Rufiji delta to the south and to the Pangani River to the north.51 All three scholars propose that the PNECB communities were likely a nascent presence in northeastern Tanzania by 300 CE. This dating means that ancestral PNECB-speaking people would have moved in among the already-established Azania and Upland-Bantu people who preceded them in the region.52

35 As earlier Azania Bantu populations had, those early Northeast-Coastal Bantu-speaking communities settled into hinter-coastal settings that were characterized by well-watered, temperate niches that were similar to the Southern-Nyanza-Basin environs their ancestors had departed. In early years, along the inland watershed regions of the Ruvu and Wami Rivers, some PNECB speakers lived in close proximity to communities of Mbuguan (ancestral Ma'a) Southern Cushitic speakers whose economies were principally centered on livestock, but they also would have done a minimal level of farming and for that reason can be designated as agro-pastoral in subsistence practices. In contrast, farming was PNECB peoples' primary means of subsistence.53

36The language evidence indicates that those intermingling societies shared more than space, they shared some language. This is reflected in vocabulary transferred into proto-Northeast-Coastal Bantu that NECB language descendants still retain. Some of those exchanges are noted, for instance, in the way they named their seasons. Like their ancestors' previous Southern Nyanza homeland, seasonal conditions in the northern and central-east Tanzania include a long rainy season, *-tíkà, from March through May, that was prime for their varied agriculture. But the Tanzanian coastal seasons differed in that they are characterized by an additional season of short rain from October to December. Some PNECB speakers designated those light rain months *-limo, a word early Mashariki-Bantu speakers had long used to name those months in their ancestral homelands. However, the language evidence suggests other PNECB adopted a word from an Mbuguan (Ma'a) Southern-Cushitic-speaking neighbor, *-bUli, for the short rainy season.54 It is possible that PNECB speakers used *-limo and *-bUli synonymously in their societies.

37Additional PNECB loan words of Mbuguan origin that point to interactions among these populations include *-gome for "tree bark," *-gwasi meaning "bush pig," and *-bulu naming "dry dung."55 However, the overall limited numbers of identifiable Mbuguan loanwords suggest that their community interactions were for the most part the interactions of societies living in adjacent territories rather than competing for the same lands. Although it is still possible that some incorporation of Southern Cushitic speakers into PNECB-speaking communities may have occurred, they would not have involved sizable populations in that period.56 On the whole, the linguistic evidence for the early centuries CE illuminates a picture in which contemporaneous settlements of Southern- Cushitic- and Bantu-speaking populations lived as neighbors in different parts of the Wami and Ruvu River areas.57

38 Map 4: Northeast Coastal Bantu Language Groups, c. 500-600 CE The proto-Northeast-Coastal Bantu period was short-lived. By about 500–600 CE PNECB had diverged into three daughter languages: pre-Asu, proto-Sabaki, and proto-Wami. The proto-Wami community heartland likely remained in situ, in the areas close to the Wami, Ruvu, and Rufiji Rivers, adjacent to the Indian Ocean coast. But the pre-Asu moved north to establish their communities near the Pangani River along the South Pare Mountains. Within a century or two, Asu speakers likely had interactions with Chaga-Dabida Upland-derived communities who lived in the adjacent North Pare Mountain zones. To the northwest of the proto-Wami communities, proto-Sabaki speakers set up communities inland from the coast between the lower Pangani and Sabaki Rivers.58

39During the short-lived proto-Wami period loanword transfers suggest proto-Wami speakers continued to have relationships with Southern Cushitic Ma'a speakers. Among terms they incorporated in that period was one used to identify female youth, notably *-dele for "young girl/woman."59 To name a "wise, older person," they adopted *mbala. And to general group ceremonies they applied the word *lusona. *-Dala was used to describe something "hard (firm)."60 While such words are definitive markers of exchange, as in the case of their PNECB ancestors' prior interactions with early Ma'a speakers, the linguistic evidence is ambiguous as to whether Ma'a speakers were an influential force in proto-Wami-speaking communities. However, the fact that many transferred words relate to identifying social categories or events could be suggestive of the types of relationships that developed between the two groups during those times. These are issues taken up further in later chapters.

40 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE Perhaps no more than a century or two later, while Asu and early Sabaki speech communities strengthened their footing in their new homelands, the proto-Wami language began to diverge into proto-Seuta and proto-Ruvu. Ancestral Seuta speakers settled in the regions near but predominantly on the Pangani River's southern side, an area that would have been located within contemporary Tanzania's Handeni district. In that same period, Sabaki and Seuta language communities likely interacted with easterly descendants of then-extant Upland Bantu speakers. This is reflected in distinct word transfers into Seuta and Sabaki, among which include *-goma for an "adult female cow," *-moli for "heifer," and *-rika for "age-set."61

41South of proto-Seuta speech areas, proto-Ruvu speakers rooted some communities astride the Ruvu River in the hinter-coast zones, continuing, like their proto-Wami ancestors, to live nearest the PNECB heartland. Contemporaneously, west of the burgeoning Ruvu communities and adjacent to Ma'a speakers, Rift-Southern-Cushitic-speaking people made their way into the regions up the Wami River and to the south of the river into zones bordered loosely by the contemporary cities of Morogoro, Iringa, and Dodoma, where in later centuries they would meet up with newcomers who comprised a linguistic subgroup of proto-Ruvu.

42By the end of the sixth century CE, NECB language communities likely extended from the Sabaki River in the north, south to the Rufiji River. These early NECB developments formed the background and linguistic contexts that led to proto-Ruvu emergence in central-east Tanzania. With such diversity in communities, coupled with the ongoing commercial and sociocultural exchanges in this zone, it is not surprising that archaeological recoveries continue to turn up abundant material evidence of their having lived in those regions.

^top

Archaeology of Central-East Tanzania c. 300 – 700 CE

43 Figure 2: Potsherd Illustrations from Sites of the Central Tanzania Coast The archaeological evidence available for the few centuries leading up to 700 CE show that, during the latter part of the first half of the first millennium CE, pottery typologically distinct from EIA Urewe/Lelesu/pre-Kwale ware began to be made in Tanzania's hinter-coast regions and nearby offshore islands. Tana Ware and Triangular Incised Ware (TIW) are the names commonly given to the new ceramic style. As in the case of EIA pottery, these wares, too, have been referenced by other names in scholarly works. The term Triangular Incised Ware is used because of the triangular decorative features found among some, but not all of them. The name Tana ware is applied because of the early discoveries of these types of pottery along Kenya's Tana River.62 Tana ware's prevalence in the region did not preclude the occurrence of EIA Kwale ware. In fact, Kwale ware continued to be made until as late as the tenth century in areas of the NECB hinter-coast. However, after 500 CE Tana ware began to predominate in the NECB zone. At the heart of these histories lies an important question. Did a new ethno-linguistic community introduce Tana ware to the region, or did extant EIA Kwale ware makers innovate a new style through a series of transitions that resulted in Tana ware?

44At least three responses to the question have been previously offered. In one view the evidence of "considerable borrowing of forms, elements and techniques" among culturally distinct groups in East Africa indicates that there is "at present insufficient analysis" to know if Tana ware derives from EIA potteries.63 Another perspective is that that Kwale ware's correlation with Upland-Bantu settlement is too precise to be incidental, particularly when the timing of Tana ware's appearance and its known distribution give reasonable indication that it corresponds with the development and expansion of NECB languages.64 A contrasting position to this is that the differences in style between Kwale and Tana are the result of transitions in motif over time by a persistent Bantu population.65

45While all three interpretations are plausible, the seamless correlation among linguistic and archaeological data best supports the conclusion that ethno-linguistically distinct Bantu people originally manufactured Urewe/Lelesu/pre-Kwale, Kwale, and Tana wares. The linguistic evidence suggests that eventual PNECB populations took up residence adjacent to and among Upland-Bantu-speaking communities who already lived in the hinter-coast region. The archaeological record reveals a parallel picture, with Urewe/Lelesu/pre-Kwale wares initially present from the last centuries BCE, followed by an overlapping presence of both Kwale and Tana/TIW wares in the middle centuries of the first millennium CE. Eventually, as NECB communities came to predominate in the hinter-coast region, both linguistically and culturally, so did their Tana style pottery.

46The earliest evidence for Tana ware in the region thus far recovered and dated is reported for Misasa. Misasa lay on the mainland adjacent to the coast, about 85 km south of Dar es Salaam. Early Misasa Triangular Incised ware dates to the fourth century, 335±45 CE.66

47 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE At nearby Kivinja, which lay about 20 km (12.4 miles) north of the Rufiji River (71 km south of Dar es Salaam), successive horizons of distinct pottery were recovered. The older, lower level held evidence of "Kwale Early Iron Working pottery" of the period CE 431± 70. The second horizon, dated to CE 598± 70, showed evidence of Tana ware. The Kwale layer contained iron slag, iron hooks, and lead objects. And both horizons contained evidence of imported ceramics and glassware. What is important to note is that Kivinja's Kwale ware horizon was extant for at least a century while Tana ware prevailed at Misasa, which lay approximately 14 km north of Kivinja.67 This pattern implies exactly the history suggested by the linguistic evidence, in which the earliest NECB settlements, as represented by Misasa, intruded in among already present Bantu populations with Limbo/Kwale archaeology, and thus Upland-Bantu language, affiliations. The replacement of Kwale ware at Kivinja by Tana/TIW wares fits the linguistic evidence for a history in which the NECB communities gradually absorbed the earlier Bantu communities. The dating of the shift to Tana/TIW pottery at Kivinja fits well with the conclusion suggested by the linguistic evidence, that it was probably proto-Ruvu populations who introduced Tana ware to the area by the sixth century CE.

48Map 6: Northeast Coastal Bantu Language Groups, c. 800-900 CE If we return to the Limbo and Nkukutu sites introduced earlier to examine the strata that postdate the EIA-Urewe/Lelesu/pre-Kwale eras there is a pattern similar to that found at Kivinja. A Tana ware stratum follows Kwale ware. Though the dates yielded were not taken as conclusive, Chami suggests that they probably parallel similar contexts found at such nearby sites as Unguja Ukuu, Kilwa, and Lamu, where Tana ware dates between 500–900 CE.68 Again, this area and the dates correlate with the expected settlement distributions of Ruvu-speaking populations. Yet Kwale ware's continued early dominance at these sites suggests that earlier-settled Upland-Bantu populations likely remained in the region along with NECB-speaking populations for some time.

49 Chami has argued in his assessment of these sites that there is no reason to suspect that their inhabitants were different or distinct makers because the settlements seem to be those of "Bantu farmers" and the wares were likely transitional. As previously mentioned, Chami suggests that Kwale ware began to take on early Tana elements then fully "matured" into Tana in the seventh century. In his view, that is why early period Tana ware "preserves several elements deriving from that of the Early Iron Working tradition." 69 While his hypothesis, again, is tenable, it is not the most plausible in light of the language evidence. As we have noted, by the fourth century the linguistic evidence suggests that proto-Northeast-Coastal-Bantu-speaking people had settled in this hinter-coastal zone. Taking into consideration the marked typological distinctions in Tana ware motifs, even when shared aesthetic influences may be noted, it is more probable that recently arrived NECB speakers manufactured the Tana ware. This interpretation also more seamlessly explains why Kwale ware continued to be manufactured in other settlement areas in those same early periods. That was because early NECB communities were centered closer to the coast and only later when, as we shall see, NECB diverged, did similar "transitional" motifs turn up and eventually take Tana-ware influences north and south along the East-African coast and into the Tanzanian interior. The Northeast-Coastal-Bantu-Tana/TIW ware styles, it must be argued, spread as their language communities diverged.

Map 3: Proto-Northeast Coastal Bantu, c. 300 CE Figure 2: Potsherd Illustrations from Sites of the Central Tanzania Coast

50Chami is fundamentally correct in asserting that Kwale and Tana ware derived from a common Bantu past. However, as we have noted, their past has roots in the Southern Nyanza Basin of the middle of the last millennium BCE, not in central-east Tanzania. Bantu-derived populations who lived in Azania during the Rhapta period and later did not comprise a continual population of Bantu people who shared a common ancestry that dated to the late last millennium or early first millennium CE. They comprised at minimum three distinct Bantu populations and their descendants, who moved in, through, and around the region over the long-run of history.

51If, as Chami and the cumulative evidence strongly suggests, we accept the premise that Dar es Salaam, the Rufiji Delta, and the surrounding areas comprised a dynamic commercial zone in the time leading up to the middle of the first millennium CE, and I agree we that we should, then there is every reason to assume that the region was a heavily trafficked zone. In such a scenario, people on the ground, regardless of their language or means of subsistence, made choices about the material items they desired, manufactured, or acquired. At any given time, consumers likely had preferences for particular ceramic styles over others. The fact that one pottery style prevailed at one site does not preclude another from being contemporaneous or transitional. It is therefore possible that any number of diverse people lived at Nkukutu, Kivinja, and the surrounding Rufiji Delta areas.

52What the sites do reveal is what the people who dwelled there manufactured, used, or traded at the time. For all we know, Arab merchants and their wives and families who spent eight months at a time in Azania awaiting seasonal monsoons to return them home preferred certain ceramics over others. Perhaps they had a preference for the Tana ware over the Kwale type. After all, both styles of pottery persisted in the larger central east Africa region from the first century CE until the ninth or tenth centuries.70 Equally possible is that any group of speakers—perhaps even a heterogeneous one—merely transported both wares into the area a century later but that only one was preferred by their eventual owners, along with iron and lead items. Could not the people have negotiated a better deal on them as local traders passed through with their wares a century later? The point is that we cannot know these details. But such plausible scenarios do consider the possibility that mobile historical agents with preferences lived all over the region. In order to truly fill out the historical possibilities about the people who lived in the region we need much more comparative and correlative evidence within the entire Tanzanian hinter-coastal zone.

53 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE For now, we can add to the preponderance of the evidence suggested by Misasa, Kivinja, Limbo, and Nkukutu, by looking within the NECB interior regions where the evidence reveals an intermediate period in which Kwale ware and Tana ware coexisted. The coexistence of the two styles at different nearby sites gives added weight to the theory of their having been made by distinct populations. Archaeological materials recovered in the Usambara Mountains indicate that Kwale ware of the second and third centuries CE were overlaid in the fifth and sixth centuries by varieties of Tana-styled pottery. Those areas and dates coincide with the locales in which Upland speakers lived and that NECB Seuta and Sabaki speakers moved into during the fifth and sixth centuries. It was in that region that the word transfers mentioned earlier in the chapter transpired among those populations.71 Based on this correlation it has been proposed that the Tana ware retrieved corresponded with Sabaki and Seuta speakers' arrival in the Usambara Mountains. Importantly, additional evidence supports this hypothesis. This evidence involves the distinct way Upland and NECB people laid out their communities. For instance, Usambara Kwale-laden settlements were positioned on mountain ridges, and ridge-villages were an old Kaskazi tradition still found among Upland-descended Chaga and Thagiicu speakers. In contrast, incoming NECB communities created small single street villages. The introduction of Tana ware pottery thus correlated with the emergence of NECB-style settlements as well as with the timing of linguistic Upland-speaker contact in evidence in Seuta and Sabaki vocabularies.72

54In Kwale/Tana scenarios we should consider the extant archaeology, the patterns of artifact distribution in known areas, and the linguistic evidence, and then ask which historical explanation best fits the available evidence for the period before 700 CE. First we can acknowledge that the linguistic evidence does suggest that NECB language communities were spreading into the Rufiji Delta areas by the middle of the first millennium CE, which correlates with the C14 Tana ware dates of the sixth century at Kivinja and nearby islands. And at parallel times NECB communities were also moving into the Usambara Mountains, where Upland people had settled as early as the first century. Additionally, the wider distribution of Tana ware at sites that correlate with greater NECB language divergence suggests that these likely moved together. Kwale ware, in contrast, does still appear, even as late as the ninth and tenth centuries in some areas. At the same time, Kwale ware was not as abundant in these same zones as Tana ware after the middle first millennium CE.73 Taken together, these circumstances suggest that Northeast-Coastal Bantu likely introduced the Tana ware that supplanted, at least in prevalence, the Upland-style Kwale ware in the middle of the first millennium.74

55While there remain gaps in the archaeological record, it does seem that the first half of the first millennium likely featured a cosmopolitan environment in central-east Tanzania. Independent yet contemporaneous data are suggestive of the presence of a number of distinct groups of people in the historical arena. The historical picture that emerges suggests that Bantu farmers, perhaps Upland and NECB-speaking, Southern Cushitic Ma'a agro-pastoralists, and Khoisan speakers were doing far more than subsisting during their shared centuries in the hinter-coast of northeastern and central-east Tanzania. They moved widely within the region, spreading their languages and material cultures, intermingling, and, perhaps, influencing people to prefer one style of pottery to another, among other possibilities. But more than that, those who were so inclined likely created some of the earliest far-reaching trade networks in an effort to make the most of coastal/transoceanic opportunities. From the imports to the exports suggested by the Periplus, these populations were among those likely involved in procuring, crafting, and cultivating products seafaring and other merchants demanded.

^top

Ruvu Languages and Neighbors to 1000 CE

56Having critically reevaluated the issues surrounding the initial periods of the settlement of Bantu speakers in the hinter-coast of northeastern Tanzania, we are now ready to focus on the Ruvu descendants of the proto-NECB, who comprise the heart of the study. However, we continue to pay attention to their NECB relatives and other neighbors, giving particular attention to the point when their histories intersect. In this way, we do justice to the extent to which the Ruvu continued to participate in an interconnected geographic and cultural milieu comprising the Northeast-Coastal Culture Zone. During the second half of the first millennium, NECB language communities set about on distinct paths. At the same time, as descendants of a single ancestral parent language community, they carried with them many commonly inherited aspects of their cultures. As it suited them, they also blended with and borrowed ideas from neighboring communities. At times they also innovated entirely new ways of identifying and doing things. But in addition to the other NECB communities, Ruvu speakers also maintained enduring contacts with Southern-Cushitic- and Eastern-Sahelian-speaking populations and, presumably, any other number of newcomers.

57By the end of the first millennium, Seuta remained one language spoken by communities who made their homes in the areas of the modern Handeni district, mostly south of the lower Pangani River. During the same period, Asu communities remained in the South Pare Mountains, where they eventually interacted with the Upland-descended Chaga-Dabida speakers, who lived in the North Pare Mountains. Some time between 600 and 900 CE, emergent Sabaki language communities entrenched themselves around the hinter-coastal areas of southeastern Kenya, forming a formidable sociocultural zone in which they stood between the Indian Ocean and inland peoples. And by 800 CE ancestral Swahili speakers began to emerge along the coast.75

58 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE At about 600 CE, proto-Ruvu communities continued to interact with early Ma'a speakers as well as a contingent of Eastern Sahelian newcomers. Turning to the early lexical evidence from that period, we can identify words transferred among them. From Eastern Sahelian speakers they adopted the verb, *-boboda, meaning "to rub or scratch on the skin's surface."76 Across Ruvu languages the term described the way people applied clay or paste-like substances to their bodies during certain of their religio-ritual initiation customs involving young women. Additionally, there were at least two words related to topographic features derived from Southern Cushitic language sources. They include *-gema for "riverbank" and *-bululu identifying "homestead" areas.77 These words, indicative as they are of interactions, are not suggestive of significant exchanges.

59 Map 6: Northeast Coastal Bantu Language Groups, c. 800-900 CE Some time during the seventh century CE, proto-Ruvu divergence into distinct language communities began. By the end of the first millennium, the proto-Ruvu communities of speakers had diverged into three or four distinct communities. Proto-West-Ruvu-speaking communities positioned themselves between the Wami and Great Ruaha Rivers, perhaps as far as 150 miles west of the proto-Ruvu homeland. There they met and intermingled with southerly dwelling Rift Southern Cushitic speech communities who bordered them on the west. And they also met up with Kati-Bantu communities who spoke the ancestral Njombe language. A second proto-Ruvu offshoot, eventual Kaguru speakers, moved the core of their communities along the Wami River approximately 100 miles inland from the coast within the varied ecological niches characterizing the Ukaguru Mountains. The third proto-Ruvu language derivative, proto-East Ruvu, settled the areas south of the Wami, in the areas surrounding the present-day cities of Bagamoyo and Dar es Salaam. In addition to these three groups, a fourth proto-Ruvu daughter language, proto-South Ruvu, was likely spoken in the Rufiji River regions near the coast. Though it is now an extinct Ruvu subgroup, this set of people is suggested because languages of the Rufiji-Ruvuma branch of Kaskazi Bantu languages spoken in and south of the Rufiji areas contain borrowed Ruvu words. For instance, a word finga, a uniquely proto-Ruvu term for "egg," is attested in Matumbi, Ndengeleko, and Ruihi. These two words, among others discussed in coming chapters, suggest that extinct South-Ruvu speakers may have introduced them.78 The fact that these are spoken outside of this area only in the Ruvu languages and that in the Rufiji-Ruvuma languages they often do not have regular sound correspondence indicates that they are borrowed words. Ruvu-language speakers were the most probable donors.79 Though this community was later absorbed or died out, it was potentially an important meeting point for those people conducting business or living astride the Rufiji Delta regions before 1000 CE.

^top

Archaeology in Ruvu's Easterly Zones c. 700 – 1000 CE

60 Map 5: Northeast Coastal Bantu Language Groups, c. 600 CE Most of the archaeology carried out in Tanzania has focused attention on areas near or at the Indian Ocean coast. Because of this we are usually limited to these areas when it comes to evaluating congruity between language evidence and archaeology. To begin, there are a few important correlations to point out among easterly Ruvu communities. Recovery efforts made at Kiwangwa, a town about 45 km (28 miles) north of Dar es Salaam and inland from Bagamoyo, yielded Tana ware stylized pottery and evidence of "imported glass and green/blue ware" that dated to the seventh century.80 These settlements in all probability reflect the settlement areas of proto-Ruvu communities or their emergent proto-East-Ruvu descendants.81 The evidence of imports dated to the seventh century may represent the trickle of persistent exchange economies in the region that had originated in the Rhapta era but soon ceased. In later times, a new era of transoceanic trade emerged with the Swahili, a matter we will discuss later in the chapter.82

61Further widespread evidence of Tana ware from the eighth century in what might have been proto-West Ruvu or Kagulu settlement areas have also been recovered. The site is Dakawa, located about 200 km inland from Dar es Salaam, along the Wami River. The Dakawa site turned up furnaces, tuyeres, iron slag, bones, and the remains of hearths, indicating that communities of farmers who kept some small livestock and who smelted iron lived in the region. What is interesting at this site is the indication that Kwale and Tana ware producers may have lived amongst each other in the region. At Dakawa, Kwale ware was not found below Tana; they were contemporaneous. The pottery evidence also indicated some mutual decorative influences.83 Analyses done on the clay samples show that the ceramics were produced from the same local clay source. The presence of iron-smelting sites at Dakawa may thus represent the effects of Kwale-ware makers who had settled the region in an earlier period. After which, by the eighth century, ancestral West-Ruvu or Kaguru-speaking Tana-ware makers joined them when they moved inland along the Wami River.

62In the vicinity of downtown Dar es Salaam, recovery efforts turned up evidence of coastal settlements between the eighth and tenth centuries. Adjacent to the construction site of what became the New Africa Hotel in the city center, LaViolette's archaeology team found evidence suggestive of a village or small town characterized by wattle and daub houses in a prime waterfront location. Other material items found included "marine shells, a limestone sharpener or 'bead-grinder,' and the tip of a small husk, probably warthog because of its size, but possibly elephant." The ceramic materials recovered were similar to Tana ware pottery. In her estimation, the ceramic sherds likely belonged to "open-mouthed globular pots" embedded with incised triangles, and "cross-hatching, diagonal lines, and rows of round punctuates." These features make them consistent with the other ceramic styles found "at virtually all sites of this period." Evidence of imported items were not found, though LaViolette advises that ruling out the potential for importing ventures occurring in the region would be premature.84

63At Kaole, located about fifty kilometers north of modern-day Dar es Salaam, archeological findings with dates of between the eighth and tenth centuries were uncovered. They included Tana ware as well as "Islamic white-glazed and early sgraffiato pottery."85 The sgraffiato ceramics are likely remnants of early trade initiated by emergent Swahili merchants. A point that Chami highlights about Kaole's Tana ware is that by this period it no longer showed EIA elements. Instead, triangles and double zigzag incisions dominated Tana motifs.86 This point is important to Chami because the fact that it no longer showed EIA elements rounds out his argument that Kwale ware had fully transitioned into Tana ware. However, I maintain that because the Dar es Salaam and Kaole pottery dates coincide with the period when proto-East-Ruvu-speaking populations, and their later descendants, likely lived in the region, and when Kwale ware presence generally diminished throughout the area, East Ruvu populations were the likely makers of the Tana ware recovered in those places. This conclusion parallels other linguistic and Tana ware correlations previously discussed for Kivinja, Masasi, Limbo, Nkukutu, Kiwangwa, and Dakawa.

Map 6: Northeast Coastal Bantu Language Groups, c. 800-900 CE Figure 2: Potsherd Illustrations from Sites of the Central Tanzania Coast

64The dating of the coastal-oriented Tana ware, along with that from inland Dakawa, in combination with the proposed timing of NECB Ruvu language divergence, brings into sharp relief Horton's supposition that the "coastal population was one expression of a large area of eastern Africa from the fifth to sixth century onwards." In his view there is no reason to consider the possibility that the coast was a "trading post" zone as it was in the later Swahili period. While it is true that NECB descendants may have lived at the coast and traded with populations external to the continent in these early times, as we have seen from the available evidence of import materials, for the most part these communities were probably more involved with mainland population contacts and settlement. 87 Among the immediate coastal dwellers, their focus may have included, for example, a mixed marine and agricultural economy that may have well included relationships with offshore islands. The archaeology, in fact, suggests this, as does the record of linguistic divergence. By 1000 CE there were ancestral Swahili and other Sabaki language descendants in the southern Kenya coastal zones, and Seuta and Ruvu communities along the zones of northeastern and central Tanzania and well into the interior.

^top

Proto-West Ruvu Societies and Their Neighbors to 1000 CE

65For some time there have been strong indications, mainly in the form of published oral traditions, of reported upheavals and the incorporation of sizable immigrant populations into Ruvu's Gogo, Sagala, Kaguru, and Luguru communities.88 An examination of the history of these groups suggests that these historical processes began in periods that anteceded their emergence as distinct language communities, that is, in the last three centuries of the first millennium CE. The earliest indication of these processes involved the proto-West Ruvu language speakers of about 600 to 900 CE, who migrated westward from their proto-Ruvu homeland. Additionally, East Rift (Iringa) Southern Cushitic speakers, who arrived via the north into the southern Tanzania highlands in the second half of the first millennium CE, and Njombe-speaking communities, who had been in the region since the first half of the millennium, are part of this history. There are indications as well of exchanges, though to a lesser degree, with an Eastern Sahelian speech community.

66In order to understand those relationships it is helpful to examine the language evidence in distinct layers of word transfers. Each is important for understanding the mosaic of sociolinguistic complexity that characterized these societies over the long haul. One stratum of interface involved Iringa Southern Cushitic influences on proto-Njombe lexis. By the middle part of the first millennium CE, eleven words known to have been transferred to Njombe speakers include those for warthog, wild dog, tumbili (a type of monkey), calf, cow's hump, sheep dung, plain, thousand, arrow, bone, and egg.89 Particularly important in Ruvu history are the last two because they probably made their way into Ruvu languages as a secondary stratum of transfer from Njombe speakers.

67Concurrently, proto-Njombe words entered into proto-West Ruvu vocabulary.90 Among them were three Njombe core vocabulary words that still attest across West Ruvu languages. They include *malenga for "water," *-tumbula for "heart," and *-ganka for "egg."91 The first two of these are distinct Njombe words, and as we can see already in this early period, the Iringa Southern Cushitic word for egg made it, via Njombe speakers, into proto-West Ruvu before 1000 CE.

68Outside of Njombe evidence, a couple of additional words suggest that in the proto-West Ruvu period there may have been interactions with an Eastern Sahelian speech community. A word for potting clay, *-widi, in Sagala (West Ruvu) and Lugulu (East Ruvu), hints of an influence from an Eastern-Sahelian-speaking community.92 Its presence is explained by the adoption of an Eastern Sahelian root *wèr or *wèd "mud," whose meanings include both "clay" and "dung" in various Eastern Sahelian languages.93 The second Eastern Sahelian word known was a type of bracelet or bangle, *-pogo, which attests in Sagala and Gogo.94

69As telling as the suggestion of such word transfers are of early contacts, they likely signaled only one aspect of ongoing and important language and sociocultural relationships that involved speakers of West-Ruvu descendant languages, which later weighed in on the histories of emergent East-Ruvu descendant language speakers who made their way into the Uluguru Mountains in the second half of the second millennium and had contact with them. To highlight this point, we should take a moment to consider immediate examples of three words that were incorporated into East Ruvu's Luguru language after 1000 CE. As a result of ongoing relationships among Sagala and Luguru speakers, who were neighbors since the start of the second millennium, an avenue developed through which *malenga, "water," *-ganka, "egg," and *-widi, "clay" were adopted into the Luguru language.95 Because the first two words reflect core vocabulary, that is, a category of word theoretically more resistant to borrowing, it suggests that their relationships involved significant exchanges, perhaps involving the incorporation of good numbers of Sagala speakers into Lugulu communities.

70The presence of the words for potting clay and bracelet are potentially telling from a socio-cultural standpoint, though more investigation needs to be done to flesh out what associations may have been tied to the incorporation of the words and material items. Until that time, we might consider whether adoption of such words may be indicative of shared values about desired resources or one's social status. In the case of clay, it may have identified a particularly important type or source of clay. Other possible scenarios might include Eastern Sahelian people having provided clay to proto-West Ruvu communities or proto-West Ruvu people having entered Eastern Sahelian areas to procure supplies of potting clay. As has been suggested by archaeological recovery, good potting clay, at least in the first millennium, was likely an important source of potential wealth. Additionally, the use of bracelets has been seen as an important marker of clan or lineage affiliation among many Ruvu communities.96 Alternatively, they are known to be medicine, often referred to as "charms" or "fetishes" in other contexts.97 Any of these scenarios would intimate interesting relationships, but we shall return to these possibilities in subsequent chapters.

^top

Ruvu Language Descendants c. 1000 to 1800 CE

71Returning to the course of language divergence among Ruvu speech communities, subsequent to the original three- or four-way proto-Ruvu language splintering, the proto-West and proto-East Ruvu in turn diverged in the period between 1000 and 1400 CE, while Kaguru remained a distinct language. The proto-West Ruvu diverged early into two communities. Eventual Gogo speakers moved toward the far western stretches of the Wami River, near modern-day Dodoma. [edThe proto-Sagala-Vidunda founded communities southwest and south of Gogo areas, soon thereafter beginning to diverge into two incipient societies.] The Sagala live in what today are the southern Kilosa District and the southwestern areas of Mpwapwa District, as well as the northeastern Iringa and northern Ulanga Districts. Their lands, much like those of the Kaguru to their north, comprised a variety of lowlands, mountains, and plateaus. The Vidunda, whose populations were probably significantly smaller in comparison with the Sagala and Gogo, resided largely in the far southern areas of Kilosa District where the topography has striking contrasts between low-lying plains and high mountainous zones.98 This period of language divergence resulted in the development of the modern West Ruvu communities by 1500.

72 Map 7: Ruvu Peoples and Neighbors, c. 1400 CE But nearer the coast the proto-East Ruvu daughter language underwent a more complex set of events. Before the end of the eighteenth century, proto-East Ruvu gave rise to six divergent dialects, but it occurred in three successive stages. The first level of differentiation into three communities probably transpired between 1000–1400 CE. The nascent Lugulu and Doe speech communities splintered and moved to distinct areas. The Doe may have made just a slight move northward, settling into the areas inland from modern-day Bagamoyo, while the Luguru made their way south into the mountainous zones of the Uluguru. Meanwhile, a community of proto-Central-East Ruvu speakers remained mostly in situ.

73 Around the midpoint of the second millennium CE, two additional stages of divergence broke up the proto-Central-East Ruvu society, with incipient Kwere, Kami, and proto-Southeast Ruvu language communities emerging between 1400 and 1500. The Kami and Kwere communities turned west from the coast and settled in the interstices between southern-lying Doe and northern-positioned Lugulu speakers. In the regions of Bagamoyo/Dar es Salaam, the proto-Southeast group of East Ruvu speakers remained for perhaps a century or longer an intact community. But by the middle of the eighteenth century, it had begun to evolve into two dialect communities, the Kutu and Zalamo. The Kutu moved westward nearer to the Luguru, while Zalamo speakers remained in the central hinter-coastal zones.99

Map 8: Ruvu Peoples and Neighbors, c. 1500 CE Map 9: Ruvu Peoples and Neighbors, c. 1700 CE

74As we have seen, not until the second half of the second millennium did Ruvu communities of speakers complete the spread of NECB languages over the central-east Tanzanian hinter-coast. Contemporaneously, the proto-Sabaki and proto-Seuta communities experienced their own divergence. In the Ruvu case, the divergences created a chain of related linguistic communities that played a role in connecting up the interior central Tanzania region with the coast throughout the second millennium. Together they formed the effective but fluid limit of the southern boundary of the Northeast-Coastal Culture Zone. To gain perspective on how the contiguous presence of Ruvu languages across this area intermingled in this period, an examination of the archaeology and linguistic evidence is useful.

^top

Culture and Contact Zones After 1000 CE in Central-East Tanzania

75By the ninth and tenth centuries transoceanic seaborne trade was increasingly an aspect of coastal life in the Indian Ocean and along the southeastern African coast. This period represented the first escalation in significant transoceanic trade since the Rhapta period. Early transoceanic trade began waning in the fourth century and then disappeared altogether by the early seventh century. But in the eighth to tenth centuries the rise of the competing interests of the various major independent Middle Eastern successor states that arose out of the breakup of the Umayyad Caliphate, most notably Egypt and the Abbasid Caliphate, instigated the establishment of new, expansive Indian Ocean trade networks. By the second half of the eighth century, Swahili populations, descendants of proto-Sabaki speakers, began involvement in those commercial activities along the East-African coast.100 Swahili participation in transoceanic commercial activities marked the beginning of their eventual settling of the "Swahili Corridor," a 1,600-mile stretch of coast and offshore islands that extended from Somalia to Mozambique by the sixteenth century.101 Principal among the distinctions that separated the Swahili from the societies of the "hinterland" was the development of a socioeconomic class of merchants, who were predominantly Muslim and interethnic in origin.102

76Thus, all along the coast of central-east Tanzania, Swahili people were increasingly an aspect of the hinter-coast demographic. And as longstanding residents of this region, Ruvu communities likely became involved with many aspects of overseas trade. Their participation in trade ventures probably manifested in direct and indirect ways. Some likely oriented themselves to the coast, becoming part of the complex trading system as traders, masons, fisherman, and other specialists needed within the immediate coastal towns. Others may have engaged in the acquisition of goods from interior regions, particularly much sought-after African ivory. Some may have organized cargo transport to the coast or brokered the exchange of goods that were traded between the interior and the coast.

77But their participation did not end with the movement of things; a cultural exchange also developed. Although historians have often focused on the creation of the Swahili culture, the cultural processes of the ensuing eras did not involve only Swahili populations. They included multifarious cultural exchanges among people that moved in many directions simultaneously. In other words, what made coastal-dwelling Ruvu and Swahili speakers unique were not just their orientation toward the Indian Ocean, but also their orientation to the interior as well as their participation in local marine communities. Swahili's eventual diversity resulted from the many opportunities they had to move materials and culture among a mix of people.

78And yet, as we have seen, the East Ruvu communities of speakers held steadfastly to their languages. Not only that, in the middle of the second millennium they continued to undertake expansions of their settlements, which led to the emergence of new language communities, even in the face of a continual Swahili presence along the Swahili Corridor. The question of why Ruvu people initiated such movements is an interesting one. There are, of course, many possibilities we can consider, including population growth, land degradation, conflicts among populations, and the like. But there is the possibility, at least at the time of proto-Central-East Ruvu's divergence, that societal break-up may have been connected in part to something entirely different. One possibility is that it may have been related to the bubonic plague pandemic that devastated the Asian, then European, continents in the 1340s. Horton and Middleton, for example, have suggested that there is evidence of settlement abandonment at Tumbatu and at Kilwa that may have been due to the effects of the bubonic plague.103 Considering the very approximate nature of linguistically proposed dates, an interesting topic for historians to explore would be the possibility that the initial breakup, sometime close to 1400, of the proto-Central-East Ruvu populations, who had resided adjacent to the coast, might possibly have been set in motion by the arrival of the plague in the mid-fourteenth century.

79Archaeological discoveries, or rather the lack of site discoveries, made during a coastal survey between Bagamoyo and Dar es Salaam by LaViolette and her team, are in keeping with this idea, although other explanations are equally possible. LaViolette explains:

Although the results of this survey, limited in scope, need not be taken as definitive for the settlement history of the coastal stretch, the chronological pattern that emerges is suggestive. Abandonment of the coastline between Bagamoyo and Dar es Salaam during the 13th–15th centuries, a period at the end of which Portuguese colonial intervention took place is one possible explanation. Another is that coastal peoples moved their settlements inland at least 0.5 km during this period, causing us to miss them in the survey, or that the coastline in this area retreated during those centuries and the vestiges of settlement were destroyed.104

In this case, her findings and the hypotheses I propose are not necessarily mutually exclusive. Like the archaeology, the linguistic history, too, is "suggestive." What is particularly striking is that the earliest part of the period correlated with the "abandonment" and with the era of the bubonic plague pandemic. Following that came an economic recession experienced among the Swahili towns in the early decades of the fifteenth century.105 Finally, as LaViolette suggests, the arrival of Portuguese interlopers in the late fifteenth century in all probability compounded an already economically stressed coastal milieu.106 The Portuguese were eager to get their hands involved in any number of nascent conquest-era enterprises; namely, they were seeking to get a direct hand in the trade in spices or find sources of gold bullion. Although the Portuguese presence was not negligible, Swahili merchants managed to maintain a footing in interior and transoceanic trade networks.

80While the uncertainty ushered in by economic downturn alone may be a reason people moved their communities, when we consider the additional possibilities of an epidemic threat, followed by Portuguese intrusion, then there were clearly any number of scenarios that may have compelled people to move away from the coast. Surely coastal residents knew that there were potential consequences related to staying close to the waterfront under less than ideal conditions. In any of these instances, they may thus have made decisions to retreat.

^top

West Ruvu, East Ruvu, and Njombe: An Areal Culture Zone

81While the Swahili period ushered in changes that were far reaching, during a parallel time frame important cultural exchanges were ongoing among interior-lying Ruvu communities. An examination of the language evidence sheds light on the broad exchanges that transpired. These data suggest that Njombe speakers had increasing contact with various Ruvu descendant language communities in the second millennium. As previously noted, their relationships had begun in the proto-West Ruvu era in the first millennium, but in the longer run of history their later interactions had more significant impacts on Ruvu languages and communities. In fact, the Njombe appear to have had far-reaching influence on all the region's speakers who bordered their communities. They, for example, left their impression on the languages of the Kilombero, Mwika, and Rungwe groups of southern Tanzania.107 And perhaps more than any other group Njombe populations deeply factored into Gogo history. But it is not only those on their immediate borders who experienced their weight. Through language intermediaries Njombe influences reached Ruvu communities geographically removed from them. At the same time, those contacts did not undermine Ruvu cohesiveness. Instead, it seems that many Njombe-speaking descendants (after 1000 CE its descendants included Hehe, Bena, Sangu, Kinga, and Kisi) became Ruvu-identified in language and culture.

82As in the proto-Ruvu era, Njombe loanwords continued in the second millennium to imprint on Ruvu languages right down to core vocabularies. An examination of the words for "bone" and "feather" sheds light not only on the exchange between Luguru and other West Ruvu languages but also on their timing. It is suspected that in early Lugulu history, probably at the start of the second millennium CE, they adopted *-dzege, the word for "bone," from Njombe speakers, who earlier borrowed it from Iringa Southern Cushitic speakers.108 After that, it was transferred into Kutu and Zaramo. Based on *-dzege's distribution in Kutu and Zalamo, however, the most likely scenario is that during the proto-Southeast Ruvu period of about 1500 to 1700, Njombe communities had a substantial exchange with Lugulu speakers that resulted in its acquisition. The occurrence of this significant contact coincides with our suspicion that the coastal-dwelling proto-Southeast Ruvu people moved toward the interior during that time frame. At the same time, they were conceivably incorporating eastward-oriented Luguru speakers into their society.

83A second core vocabulary item reveals the significant relationships Lugulu had with their Kami neighbors. Here the borrowing of the word *lugala, "feather," resulted from post-1500 Njombe contacts with Kami speakers. These forgoing exchange scenarios are suggestive of ongoing substantial areal influence between Lugulu and Kami.109 The overall distribution of proto-Njombe-derived loanwords (which are covered in upcoming chapters) among these groups reflects an important areal contact zone—involving Njombe and East Ruvu language descendants—that has characterized the Uluguru region over the last five hundred years.

84For more recent centuries this is not new information, as we are aware of the disruptions brought on by the colonial incursions. But what this new information opens up is the knowledge that this region has been engaged in more than 1,200 years of contact exchanges that have included, among others, the Njombe, Iringa Southern Cushitic, and East Ruvu societies. Throughout all of this contact, the Lugulu speech communities seem to comprise a pivotal social group that brokered exchanges through the central zone in this broad historical context. Interestingly, Lugulu people tell fittingly complex histories about their origins.

85Lugulu clan oral traditions are as varied as the linguistic evidence would seem to suggest, yet they seem uniform in that they maintain that they are solely of Bantu origin. The depth and layers of their histories may be somewhat revealed in that they consider both Mgeta and Bunduki their oldest homeland areas.110 Additionally, Young and Fosbrooke found that "some [Luguru] state that they came from the Bena tribe [Njombe], lying to the south, and others, from the coastal region or from the north." Seemingly with some uniformity, "lineages may be traced for from thirteen to fifteen generations, from which they conclude that the Bantu ancestors came into the mountains some three hundred years ago."111

86It seems clear that some among the Lugulu were quite aware of the Njombe and coastal elements of their long history. And the antiquity of their earliest origins, i.e., in the pre-Swahili era, may support their Bantu-only origins stance. After all, there was no Swahili presence in the region at the time they began to diverge from their proto-East Ruvu ancestor and moved toward the Uluguru Mountains, so there was not any of the cultural prestige attached to being part of the Swahili culture, something commonly touted as desirable among narrators of more recent hinter-coastal histories. Yet their acknowledgement of Bunduki and Mgeta origins, may be suggestive of both their early settlement in Mgeta, deeper in the Uluguru areas, with Bunduki transpiring as a homeland in later times, perhaps in the parallel timeline that corresponds with their retelling of their clan histories that take them back only about three hundred years. Ultimately, as the linguistic evidence suggests, it is most beneficial to view this region as an exceptionally tight-knit and interactive sociocultural space.

87Similarly, we see ongoing exchanges among the Njombe and Vidunda and Gogo speakers. In Gogo retellings of their history, we know that population and cultural exchanges between them and Njombe, Iringa Southern Cushitic, and Eastern Sahelian groups have been a mainstay of their recollected experiences in recent history. These are reflected in significant sociocultural influences between their societies. But what is new about this from a linguistic standpoint is the strong evidence for their contacts having been initiated at an early period in Gogo history. Judging from their word transfers, exchanges with Njombe began, as we witnessed, with their proto-West Ruvu ancestor, but were carried on in early Gogo and Vidunda speech communities. Supporting this are three core vocabulary terms. Gogo carries Njombe ifugamilo for "knee," and mdela for the "root (of plant or tree)," while Gogo and Vidunda have Njombe ham(h)a for "leaf." These cases suggest that a significant component of proto-West Ruvu and their descendants' ancestry involved the incorporation of Njombe speakers. While other culture-related vocabularies were also adopted from Njombe, those of core lexis are especially highlighted because such borrowings suggest that speakers of the donor language formed an important population contingent absorbed into early West Ruvu societies. But paralleling such influences are usually additional cultural exchanges, which are taken up in the following chapters.

^top

Conclusion

88The foregoing serves as a way to conceptualize the interactive and fluid milieu that characterized the Northeast-Coastal Culture Zone. The language evidence indicates that Ruvu-speaking populations comprised a significant constituent of a cultural zone that came into being with the arrival of a particular subset of Bantu descendants in the hinter-coastal regions of Tanzania as long as 1,700 years ago. They are called the proto-Northeast-Coastal Bantu. When those PNECB settled into the area, they did so among people who already visited or lived in the region. At minimum this included both Southern Cushitic agro-pastoralists and Khoisan gatherer-hunter specialists. The Southern Cushitic speakers had been there for perhaps a couple of centuries, while the Khoisan-speaking populations were in all probability the descendants of ancestral language communities that had taken root there many millennia before.

89Moreover, the archaeological evidence and linguistic patterns of Kaskazi languages suggest that PNECB was not the first Bantu language spoken in the region. Though we cannot absolutely distinguish the linguistic identities of the very earliest groups late in the first millennium BCE, their settlements and material culture provide strong indicators that early Bantu-speaking offshoots from the Southern Nyanza Basin who produced a variety of EIA ceramics, farmed, and produced iron moved into the area just before the end of the last millennia BCE. We have designated them the Azania-Bantu populations. Subsequent to that, by 200 CE, populations of Southern-Nyanza-descended Upland-Bantu-speaking communities, who produced Kwale-styled wares, likely joined them. Close to that time, by about the early fourth century CE, ancestral Northeast-Coastal Bantu-speaking populations, who were also descended from the earlier cluster of Kaskazi communities in the Southern Nyanza Basin arrived in the region. Each of these layers of Bantu populations brought an additional dimension of cultural and economic change to an area that became home to diverse language and cultural communities.

90There is every indication that various of these populations were, until the seventh century CE, involved in transoceanic commercial exchanges. Both imported materials found near the Rufiji Delta and written information left in the Periplus Maris Erythraei show that merchants external to the continent likely made regular visits to the region. Such visits were regulated by monsoon patterns that required visitors to remain in Tanzania for months at a time before departing. Some visitors may have decided to establish permanent homes there, sometimes forming unions with local people and communities. While the research on this early period is still nascent, there is little doubt that Tanzanian hinter-coastal zones comprised a hub for interethnic and multilingual exchanges since the early centuries CE. Yet many more developments followed.

91Some time about the fourth century CE, proto-NECB-speaking communities came to live in the zone, triggering opportunities for new speech communities to evolve. By 1800 CE, the PNECB linguistic descendants accounted for more than twenty languages. Their speakers moved into areas that stretched as far north as Somalia to Mozambique in the south. But the majority of their language communities remained in the regions of the modern-day countries of Tanzania and Kenya. Over time, their languages and cultural expressions predominated in the region though interspersed among them remained communities of Southern Cushitic and Eastern Sahelian backgrounds. It would appear, however, that no Khoisan-derived language speakers persisted in these niches. They were either incorporated among the extant communities or departed the region altogether.

92As budding transoceanic trade networks took firm hold along the Eastern African coast late in the first millennium CE, one community of NECB descendant speakers emerged as key agents in that trade. They were the ancestral Swahili. Over the course of the early part of the second millennium Swahili speakers spread their linguistic and cultural marks along the coast, covering more than 1,600 miles. Their history is the history people are most familiar with when it comes to East Africa.112 But in tandem with Swahili achievements unfolded the histories of other NECB language descendants. For all intents and purposes they formed the backbone and backdrop of Swahili successes. They made up the diverse communities with whom Swahili speakers necessarily interacted and relied upon to procure their primary products of trade. Together they formed the Northeast-Coastal Culture Zone.

93Our knowledge of the intricacies of the history that unfolded in the region remains scanty. However, archeologists, linguists, and historians continue to develop an evidentiary base of data that bolsters our understanding of those motherland communities. It is the opportune moments when these efforts coincide that we make our biggest strides toward recovery. Indeed post-1985 archaeological efforts have shed much-needed light on, primarily, early coastal communities between Bagamoyo and the Rufiji River. They corroborate what people have long suspected, based on such early written sources as Periplus of Maris Erythraei; that is, that East African inhabitants have engaged in transoceanic exchanges since at least, and most likely before, the start of the first millennium CE. With full awareness that such achievements are worth noting, however, linguistic-based discoveries of complex societal divergence and intermixing serve to remind us that it is equally important not to ignore interior histories. Without the "hinterland," we present synchronic histories that serve to celebrate a few and overlook the majority.

94This brings us to the point that if Swahili history and civilization were spawned and supported in large part by the combined effects of the dynamic histories of the East African interior and the Indian Ocean, then we need to engage with the historical foundations laid by early hinter-coastal communities. An approach that recasts the larger hinter-coastal Swahili zones as the Northeast-Coastal Culture Zone would help this endeavor. Doing this serves to acknowledge in an inclusive way that there were more than Swahili ancestors on the historical scene. And it demands that far more than coastal-dwelling communities be recognized as fundamental to that history.

95But beyond that, the culture zone does not demand that there be one continuous community of any type: farming, gatherer-hunter, agro-pastoral, marine, merchant, or any combination thereof. Instead, it assumes that all developments hinged on the integration of many communities over more than two millennia. This approach situates the Swahili and all other Northeast-Coastal Bantu speakers into a more encompassing historical context. At the same time, it gives full attention to the fact that at some point in the early centuries CE, an important and landscape-altering community of speakers, the proto-Northeast-Coastal Bantu and their descendants, eventually predominated in the regions. In the words of Hinnebusch and Nurse:

Having ultimately all emerged from PNEC [proto-Northeast Coastal],  . . . [the languages of this zone] would have been very similar at all linguistic levels, lexical, phonological, morphological, and syntactic. On the one hand, divergence would have occurred as communities sought to define themselves linguistically by institutionalizing incipient differences. On the other hand, convergence would have been promoted by their inherited structural similarity, which would have contributed to the interchange of linguistic features.113

Their emphasis on divergence, convergence, and interchange underscores the understanding of an area that was a fluid landscape in both linguistic and sociocultural terms. And this is where the Ruvu language descendants are important.

96Over 1,500 years the cultural heirs of the speakers of the proto-Ruvu language diverged into at least ten languages in central-east Tanzania. Their linguistic developments suggest that in that time frame, Ruvu communities increasingly inhabited the interstitial areas surrounding their ancestors' original Wami/Ruvu homeland. They were predominantly farmers, though their locations near the coast and rivers likely encouraged varied levels of engagement with maritime matters, whether subsistence or commercial in nature. It is likely that because land was still available in the interior zones, there was no reason to insist on accommodation in areas that for various reasons became unusable, uncomfortable, or undesirable. Resulting from their movements and those of others, local culture areas comprised portions of larger culture zones. The immediate hinter-coastal zone between Bagamoyo and the Rufiji was one such area, but so too were the areas encompassing the Uluguru Mountains and the Tanzanian Southern Highlands. And along and south of the Rufiji River, now-extinct South Ruvu speakers influenced some of the neighboring languages of the Rufiji-Ruvuma subgroup. Additionally, in the regions of Dodoma and the Ukaguru Mountains, Ruvu interactions with Southern Cushitic and Eastern Sahelian communities were an important factor. Not least, from the late first millennium CE onward, the Swahili presence made itself felt all along the coast and some way into the interior. And as hosts to the many communities of Swahili immigrants they encountered over time, Ruvu speakers incorporated, intermingled, and engaged with the Swahili in ongoing dialectical exchanges. Yet Ruvu speakers also remained rooted in their dynamic, interconnected inland communities. This viewpoint intimates that Ruvu language descendants, among many others, were active agents in the motherland that conceived Swahili civilization. It is from these intersecting points of understanding that Ruvu histories are imparted in the following chapters.

Notes

Note 1: Steven Feierman, Peasant Intellectuals: Anthropology and History in Tanzania (Madison: University of Wisconsin Press, 1990), 35–6. The decision to use an enlarged framework, in both time and place, to address Ruvu history was inspired by Feierman's discussion of the artificiality of carving up Africa by ethnic zones. While differences among groups were acknowledged among them, people also interacted within larger geographic areas. And it is in such a context that I see descendant NECB-speaking communities as well as all others who lived within the cultural milieus they created. In this way it is possible to convey that what is local is often part of a much broader network. back

Note 2: Comaroff and Comaroff, Ethnography and the Historical Imagination, 98. back

Note 3: Ehret, An African Classical Age, 36; Nurse and Hinnebusch, Swahili and Sabaki, 19–23. back

Note 4: Ehret, African Classical Age, 53. back

Note 5: For an early history of this region, see Schoenbrun, A Green Place, A Good Place. back

Note 6: Ehret, African Classical Age, 184. back

Note 7: Ibid., 189. back

Note 8: L. P. Kirwan, "Rhapta, Metropolis of Azania," Azania 21 (1986): 99. back

Note 9: Lionel Casson, The Periplus Maris Erythraei: Text with Introduction, Translation, and Commentary (Princeton: Princeton University Press, 1989), 7–9, 34–35. back

Note 10: Ibid., x–9; Ehret, African Classical Age; Kirwan, "Rhapta, Metropolis of Azania," 92. back

Note 11: Casson, Periplus Maris Erythraei, 285–9. The contemporaneous route to India was filled with more turbulent weather conditions that could be injurious, if not life-threatening. back

Note 12: My use of "farmer" is an interpretation of Casson's use of "tillers of the soil," which is his translation of oratoi, a word that has been previously translated as "pirates." See Jan Vansina, "Slender Evidence, Weighty Consequences: On One Word in the 'Periplus Maris Erythraei,'" History in Africa 24 (1997): 393–397. back

Note 13: Casson, Periplus Maris Erythraei, 142, 61. back

Note 14: Ibid., 61. back

Note 15: Felix A. Chami, "Roman Beads from the Rufiji Delta, Tanzania: First Incontrovertible Archaeological Link with the Periplus," Current Anthropology 40, no. 2 (1999): 238; Neville Chittick, "Reconnaissance in Coastal Tanzania," Nyame Akuma, no. 20 (1982): 57–58. back

Note 16: Mark Horton and John Middleton, The Swahili: The Social Landscape of a Mercantile Society (Malden Mass.: Blackwell Publishers, 2000), 36. back

Note 17: Felix A. Chami, "Limbo: Early Iron-Working in South-Eastern Tanzania," Azania 27 (1992): 45–52; Chami, "Roman Beads," 237–41. back

Note 18: Chami usually uses the less conventional EIW (Early Iron Working) acronym. Hereafter, I use the more common EIA (Early Iron Age) convention. back

Note 19: Felix A. Chami and Paul J. Msemwa, "A New Look at Culture and Trade on the Azanian Coast," Current Anthropology 38, no. 4 (1997): 674–5. back

Note 20: Felix Chami, "The Excavation at Kwale Island, South of Dar es Salaam, Tanzania," Nyame Akuma 48 (1997): 50. back

Note 21: Felix Chami, "Excavations of a Coastal Iron Age Site in Kisarawe District, Tanzania," Nyame Akuma 30 (1988): 35. back

Note 22: Robert Soper, "Bantu Expansion into East Africa," in The Archaeological and Linguistic Reconstruction of African History, ed. Christopher Ehret and Merrick Posnansky (Los Angeles: University of California Press, 1982), 225. back

Note 23: Chami, "Limbo: Early Iron-Working in South-Eastern Tanzania," 47. back

Note 24: Soper, "Bantu Expansion into East Africa," 227. back

Note 25: For a discussion of these histories, see Felix A. Chami and Bertram Mapunda, "The 1996 Archaeological Reconnaissance North of the Rufiji Delta," Nyame Akuma 49 (1998); Chami and Msemwa, "A New Look"; Ehret, African Classical Age, 187–8; Robert Soper, "Kwale: An Early Iron Age Site in South-Eastern Kenya," Azania 2 (1967); Spear, "Early Swahili History Reconsidered." back

Note 26: Chami, "Excavation at Kwale," 50–2; Soper, "Kwale: An Early Iron." back

Note 27: TIW pottery has been alternatively referenced as "Kitchen" ware. Neville Chittick, "The Peopling of the East African Coast," in East Africa and the Orient: Cultural Synthesis in Pre-Colonial Times, ed. H. Neville Chittick and Robert I. Rotberg (New York: Africana Publishing Company, 1975). back

Note 28: Chami, "Excavation at Kwale," 52. back

Note 29:   I thank Felix Chami and Azania journal for granting permission to duplicate this page.  It originally appeared in Felix A. Chami, "The First Millennium AD on the East Coast: A New Look at the Cultural Sequence and Interactions," Azania 29–30 (1994–95): 235. back

Note 30: Felix A. Chami and Bertram Mapunda, "The 1996 Archaeological Reconnaissance North of the Rufiji Delta," 62–78. In a subsequent publication he alternates between Limbo and pre-Kwale as a reference to EIW in Limbo and Mkukutu (spelled Nkukutu in the 1998 article) before the third century. back

Note 31: Chami, "Roman Beads from the Rufiji Delta, Tanzania," 240. back

Note 32: Ehret, African Classical Age, 34, 94, 97, 108; Marie-Claude Van Grunderbeek and Emile Roche Hugues Doutrelepont, "L'age du fer Ancien au Rwanda et au Burundi Archeologie et Environment," Journal des Africanistes 52, no. 1–2 (1982): 24. back

Note 33: Horton and Middleton, The Swahili: The Social Landscape, 38, 216 fn.36; Spear, "Early Swahili History Reconsidered," 267, fn.34; For Chami's periodization, see Chami, "Roman Beads," 240; See also Randall L. Pouwells, "East African Coastal History," Journal of African History, no. 40 (1999). back

Note 34: Felix A. Chami, "A Response to Christopher Ehret's 'Bantu Expansions,'" The International Journal of African Historical Studies 34, no. 3 (2001): 649. back

Note 35: Ehret, African Classical Age, 188–9. back

Note 36: Ibid., 190. back

Note 37: Chami's critical response to the possibility that there was a Cushitic presence in the region seems to be a reaction to what some scholars early on suggested about a possible Cushitic influence in East African history. In some hypothetical scenarios posited, it was suggested that cattle-keeping Cushitic people might have entered into relationships with Bantu populations and that over time they formed an elite class that ruled over Bantu populations. A point that appears to concern Chami is that the Cushitic ruling class hypothesis was anchored in Western notions of progress and racism that valued pastoralist economies over agriculture economies, and that often spoke in privileging terms about the non-African origins of those cattle-keeping populations. See, for instance, Chittick, "The Peopling of the East African Coast," 28–9; Mark Horton, "The Periplus and East Africa," Azania 25 (1990): 96–99; Roland Oliver, "The Problem of Bantu Expansion," The Journal of African History 7, no. 3 (1966): 375–376. back

Note 38: Ehret, African Classical Age, 275. back

Note 39: Soper, "Bantu Expansion into East Africa," 224. Urewe ware was earlier termed 'dimple-based' pottery. See Peter Robertshaw, "Gogo Falls: Excavations at a complex archaeological site east of Lake Victoria," Azania 26 (1991): 25–26. back

Note 40: Schoenbrun, A Green Place, A Good Place, 33. back

Note 41: Robertshaw, "Gogo Falls," 25–26; Schoenbrun, A Green Place, A Good Place, 32–3. back

Note 42: David W. Phillipson, African Archaeology, 3rd ed. (New York: Cambridge University Press, 2005), 264; Soper, "Bantu Expansion into East Africa," 228. back

Note 43: Ehret, African Classical Age, 34; Schoenbrun, A Green Place, A Good Place, 46–48. back

Note 44: Schoenbrun, A Green Place, A Good Place, 35–6, 48. back

Note 45: Ehret, African Classical Age, 188–9. back

Note 46: Ibid., 34. back

Note 47: Schoenbrun, A Green Place, A Good Place, 69. back

Note 48: Chami, "Roman Beads," 239–40. An EIA horizon at Nkukutu dating between 200–400 CE yielded four imported Roman-era beads. It was the first physical evidence found in a protected horizon that supported transoceanic links during the Rhapta period. back

Note 49: Ehret, African Classical Age, 114. back

Note 50: Nurse and Hinnebusch, Swahili and Sabaki, 33; G Philippson and D. Nurse, "The Northeastern Bantu Languages of Tanzania and Kenya," 1–28. back

Note 51: Ehret, African Classical Age, 190–2. back

Note 52: This range of dates reflects the estimated time frame in which these early communities settled into the hinter-coastal region. back

Note 53: Ehret, African Classical Age, 173. back

Note 54: Ibid., 183, 330. back

Note 55: Ibid., 330–1. back

Note 56: More influential relationships among populations commonly are inferred when core vocabularies (e.g., one hundred-word list) are adopted from incoming communities of distinct language speakers. See chapter 1 for further discussion of this theory. Ehret, African Classical Age, 328, 330–1. back

Note 57: Stanley H. Ambrose, "Archaeology and Linguistic Reconstructions of History in East Africa," 138–146; Ehret, African Classical Age, 183; Horton and Middleton, The Swahili: The Social Landscape, 26. In this region, and in Tanzania more generally, there exists the likelihood that gatherer-hunter populations, who spoke Khoisan-derived languages, may have been contemporaneously present in the region, though evidence for their intermingling is not indicated in NECB lexis. See A. T. Culwick, "Ritual Use of Rock Painting at Bahi, Tanganyika Territory," Man 31, March (1931): 33–36; Tom Güldemann and Rainer Vossen, "Khoisan," 104–5; Fidelis T. Masao, The Rock Art of Kondoa and Singida: A Comparative Description (Dar es Salaam, National Museums of Tanzania, 1982). back

Note 58: Ehret, African Classical Age, 192; Nurse and Hinnebusch, Swahili and Sabaki, 33. Hinnebusch and Nurse differ from the view taken here that there was a short proto-Wami period. Their view is that PNECB diverged into four subgroups comprising Seuta, Ruvu, Pare/Taveta (pre-Asu), and Sabaki. back

Note 59: Bondei mndele "girl"; Zigua mdele "girl"; Shambaa mndee, wa- "young woman"; Southern Cushitic Dahalo –dééla "girl not yet of child-bearing age" final position *-a > *-e > *-dele is a regular Ma'a sound shift; proto-Wami *-dele "girl/young woman"; Christopher Ehret, The Historical Reconstruction of Southern Cushitic Phonology and Vocabulary, ed. Oswin Kohler, vol. 5, Kolner Beitrage Zur Afrikanistik (Berlin: Dietrich Reimer Verlag, 1980), 164. back

Note 60: Southern Cushitic: early Ma'a -dára "to coagulate, curdle, harden" > proto-Wami *-dala "hard." Southern Cushitic: *loos- "to assemble, to congregate" > Rift Southern Cushitic noun *losona > proto-Wami *lusona "group assembled for ceremony." Southern Cushitic: Dahalo бar- "to know" > proto-Wami *mbala "wise person," Zigua -bala "acute, cunning, clever, shrewd." Ehret, Historical Reconstruction of Southern Cushitic, 135, 163, 206. back

Note 61: Ehret, African Classical Age, 186, 329–30. back

Note 62: For examples of the diversity in naming, see Horton, "Periplus," 95–99; D. W. Phillipson, "Some Iron Age Sites in the Lower Tana Valley" Azania 14 (1979): 155–60; Paul Sinclair, "Chibuene—An Early Trading Site in Southern Mozambique," Paideuma 28 (1982): 149–164. back

Note 63: Mark Horton, Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa (London: The British Institute in East Africa, 1996), 410. This opinion differed from an earlier one in which he stated, "it is certainly clear that Tana tradition pottery cannot be derived from Kwale or other known Early Iron Age ceramic groups," but that they were likely of Pastoral Neolithic derivation. See Horton, "Periplus," 97, 99. back

Note 64: Ehret, African Classical Age,188, 191–92. For a map of Tana ware's (TIW) distribution, see Felix A. Chami, "The First Millennium AD on the East Coast," 233. back

Note 65: Chami, "Response to Christopher Ehret's," 650; Chami, "First Millennium AD," 235. See also Horton and Middleton, The Swahili: The Social Landscape, 42. back

Note 66: Chami, The Tanzanian Coast in the First Millennium AD, 95. back

Note 67: Chami and Msemwa, "A New Look," 674. back

Note 68: Chami, "Roman Beads from the Rufiji Delta," 240–41. back

Note 69: Chami, Tanzanian Coast in the First Millennium, 95. back

Note 70: Kwale ware extends from a little north of Mombasa through the highlands of northeastern Tanzania from Kilimanjaro to the Usambaras, and at least as far south as the Ngulu Hills. Soper, based on "typological comparisons," made a prediction that such wares would be found southward through the Uluguru Mountains, into Southeastern Tanzania and into Mozambique, which is what Chami's data confirms. The dates associated with these Kwale wares range from as early as the first century to as late as the ninth or tenth. Soper calls Urewe and Lelesu variants Kwale. Soper, "Bantu Expansion into East Africa," 225–233. back

Note 71: Ehret, African Classical Age, 187–8, 207. back

Note 72: Ibid., 187. back

Note 73: Horton, Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa, 407. back

Note 74: Ehret, African Classical Age, 187–8. back

Note 75: Nurse and Hinnebusch, Swahili and Sabaki, 22. back

Note 76: Christopher Ehret, A Historical-Comparative Reconstruction of Nilo-Saharan (Köln: Rüdiger Köppe Verlag, 2001), 271. Eastern Sahelian (Nilo-Saharan): *ɗ ̣or ɗ ̣ "to scratch (skin surface)" > proto-Ruvu *-boboda "to rub on the skin." The reduplication of the first syllable is a common way of implying repetitive action, as "to rub or scratch" implies. back

Note 77: Ehret, The Historical Reconstruction of Southern Cushitic. Southern Cushitic: Ma'a igemo "slope" > proto-Ruvu *-gema "riverbank"; Southern Cushitic: *buru "dust" > proto-Ruvu *-bululu "homestead." back

Note 78: Lyndon Harries, A Grammar of Mwera (Johannesburg: Witwatersrand Press, 1950). For more information on the Rufiji-Ruvuma history of this particular region, see Ehret, African Classical Age, 200–1. back

Note 79: The Mwera language changed proto-Wami and proto-Ruvu *b to /p/ in this word because Mwera has no /b/ and /p/ was their closest resemblant sound. back

Note 80: Chami and Msemwa, "A New Look," 674. back

Note 81: Today it is home to Doe and Kwere speakers, but in the seventh century those languages were not yet spoken because proto-East Ruvu had not yet diverged. back

Note 82: Horton, Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa, 409–410. back

Note 83: Randi Haaland, "Dakawa: An Early Iron Age Site in the Tanzanian Hinterland," Azania 29–30 (1994–95): 238–47. back

Note 84: Adria LaViolette et al., "Survey and Excavations between Dar es Salaam and Bagamoyo: University of Dar es Salaam Field School, 1988 (Part 2)," Nyame Akuma 52 (1999): 74–78. back

Note 85: Chami, "First Millennium AD," 234. back

Note 86: Ibid. back

Note 87: Horton, Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa, 409–10. back

Note 88: T. O. Beidelman, "A Note on Luguru Descent Groups," Anthropos (1960): 882–85; Thomas O. Beidelman, "Myth, Legend, and Oral History: A Kaguru Traditional Text," Anthropos 65 (1970): 74–97; W. J. Carnell, "Sympathetic Magic among the Gogo of Mpwapwa District," Tanzania Notes and Records 39 (1955): 25–38; Heinrich Claus, Die Wagogo Ethnographische Skizze eines Ostafrikanischen Bantustammes (Berlin: Druck und Verlag von B. G. Teubner, 1911); Henry Cole, "Notes on the Wagogo of German East Africa," The Journal of the Anthropological Institute of Great Britain and Ireland 32 (1902): 303–38; Mathias E. Mnyampala, The Gogo: History, Customs, and Traditions, trans. Gregory Maddox (London: M. E. Sharpe, 1995), 43–46; Peter Rigby, Cattle and Kinship among the Gogo: A Semi-Pastoral Society of Central Tanzania (Ithaca: Cornell University Press, 1969); Die Wagogo, "Baessler-Archiv," (1911): 66–72; A. N. Wood, "Itinerating in Usagara, 1888," The Church Missionary Intelligencer and Record 14 (1889): 24–32. back

Note 89: Ehret, African Classical Age, 331. back

Note 90: Ibid., 192–3, 200. back

Note 91: The Njombe form would be *-kanka, but the *g in the West Ruvu form is explained by Dahl's Law. Dahl's Law has had wide effects in the Ruvu languages. According to this formerly productive phonological rule, a word containing two voiceless stops in the first and second syllable position, like the two /k/ in Njombe *-kanka, will add voicing to the first of those voiceless stops, which would then give *-ganka in this case. These cases provide the first examples of the upwardly skewed cognation rates between the Sagala and Luguru languages. See Table 1: Ruvu Shared Percentages. back

Note 92: Here, as in cases previously noted, the presence of intervocalic *d, as in the prior *-dele, *-dara, and *-bobod- examples, is diagnostic of a loanword because these *d are not a regular Ruvu outcomes. back

Note 93: Ehret, Historical-Comparative Reconstruction, 570. back

Note 94: Acholi pogo "ring." This word also attests in Ganda as a type of anklet. In each of its meanings it is, then, a circular piece of material used for adornment purposes. This distribution is suspected not to be of an early Kaskazi-era borrowing, but an independent borrowing of the word. back

Note 95: Because of the evidence of *-widi in languages of two of the three primary branches of proto-Ruvu, it might be expected that it was borrowed in the period before they diverged; however, the skewed figures found in cognate retention rates between Sagala with Lugulu suggest it was far more likely the result of later exchange between them. back

Note 96: Author interviews, Tanzania, 1998–1999. For example, in Gogo communities ivory arm-rings are important in marital negotiations. Rigby, Cattle and Kinship, 213. back

Note 97: James L. Brain, "Symbolic Rebirth: The 'Mwali' Rite among the Luguru of Eastern Tanzania," Africa: Journal of the International African Institute 48, no. 2 (1978): 176–88. back

Note 98: T. O. Beidelman, The Matrilineal Peoples of Eastern Tanzania (London: International African Institute, 1967), 36, 51, 53. back

Note 99: European explorers report encounters with Kutu in 1879. See, for instance, comments in Roland Young and Henry Fosbrooke, Smoke in the Hills: Political Tension in the Morogoro District of Tanganyika (Evanston, Ill.: Northwestern University Press, 1960), 23. back

Note 100: Ehret, The Civilizations of Africa: A History to 1800, 248–9. back

Note 101: Horton and Middleton, The Swahili: The Social Landscape, 5. back

Note 102: Horton, Shanga: The Archaeology of a Muslim Trading Community on the Coast of East Africa, 412–3. back

Note 103: Horton and Middleton, The Swahili: The Social Landscape, 82. A nineteenth-century case of abandonment in the threat of bubonic plague is noted in Juhani Koponen, "War, Famine, and Pestilence in Late Precolonial Tanzania: A Case for Heightened Mortality," The International Journal of African Historical Studies 21, no. 4 (1988): 660. back

Note 104: LaViolette et al., "Survey and Excavations between Dar es Salaam and Bagamoyo: University of Dar es Salaam Field School, 1988 (Part 2)," 77. back

Note 105: The period of economic decline is thought to be linked to the collapse of the Zimbabwean state, which was a supplier of gold that Swahili merchants moved northward. See Ehret, Civilizations of Africa, 377–8. back

Note 106: LaViolette et al., "Survey and Excavations between Dar es Salaam and Bagamoyo: University of Dar es Salaam Field School, 1988 (Part 2)," 77. back

Note 107: Ehret, African Classical Age, 66. back

Note 108: Ibid., 331. back

Note 109: This is indicated by the upwardly skewed cognation shared between the two languages. back

Note 110: Beidelman, Matrilineal Peoples, 26; Young and Fosbrooke, Smoke in the Hills, 49; Brain, "Symbolic Rebirth," 81. back

Note 111: Young and Fosbrooke, Smoke in the Hills, 21–2. In an unpublished manuscript accessed by Vuorela, oral histories collected by Father Vermunt state that until the late eighteenth century, Lugulu subclans frequently split from clans. See Ulla Vuorela, The Women's Question and the Modes of Human Reproduction: An Analysis of a Tanzanian Village (Helsinki: Lomakepainatus, 1987), 87. back

Note 112: Spear, "Early Swahili History Reconsidered," 257. In 1990 Spear reported that the preceding fifteen years had netted 270 books, theses, and articles on Swahili history. back

Note 113: Nurse and Hinnebusch, Swahili and Sabaki, 34. back