Saturday, 4 April 2009


by Martin Walsh

[This is the corrected text of an article that appeared in March 1992 in the East Africa Natural History Society Bulletin, 22 (1): 2-9. A pdf version (with minor diacritical marks that can't be shown here) can be downloaded from the site of the Ethno-ornithology Research and Study Group.]

The following notes set out to solve a simple linguistic puzzle: why do the Mijikenda of the East African coast have more than one word meaning “bird”? Why is the generic term for avifauna in some dialects the name of a particular species in others? And why do some Mijikenda dialects have no special term at all, but refer to birds in general by circumlocutions such as “little creatures”, “flying animals” and “animals with wings”? The answer to these questions leads far beyond an account of the principles or accidents of local taxonomy. Perhaps surprisingly, it reveals something of a forgotten episode in coastal history; while in general it provides an introduction to the relatively unexplored field of ethno-ornithology, in this case as exemplified by the practices and beliefs of the Mijikenda.

First, a few remarks about the people and language in question. The Mijikenda inhabit a large part of the East African coast and its immediate hinterland, between the Tana River in the north and the Usambara Mountains in the south. Until relatively recently they were known as the Nyika or “people of the wilderness” in implied (and unkind) contrast to the urbane Swahili of the littoral. Their modern name is a reference to the original “nine villages” or kayas in the lowland rainforest said to have been occupied by the different ethnic groups which comprise the Mijikenda today: the Giriama, Kauma, Chonyi, Dzihana, Kambe, Rihe, Raßai, Duruma and Digo. All of these people speak mutually intelligible dialects of the same language, also called Mijikenda. Its closest linguistic relatives are Pokomo, Comorian, Swahili and Elwana, which together with Mijikenda form the Sabaki group of the North East Coast Bantu languages. The dialects of Mijikenda, meanwhile, can be divided on phonological grounds into two groups, northern and southern. Southern Mijikenda consists of Raßai, Duruma, and at least two varieties of Digo: the remaining dialects, the most widely spoken of which is Giriama, belong to the northern group. Available evidence suggests that these two groups began to diverge less than 500 years ago and that the emergence of the modern Mijikenda dialects thus postdates the first appearance of the Portuguese on the East African coast.

The Mijikenda dialects share most items of basic vocabulary. The word for “bird”, however, is not shared in this way: instead a number of different terms are in use. To make matters more complicated, it seems that their application varies within as well as between dialects. Moreover, it is evident that this usage is in a continuing process of change, and has been since records of the language were first made by the missionaries Krapf and Rebmann in the middle of the last century.

Basically three terms are involved: nyuni, ts’ongo and, variously qualified, nyama. Nyuni is recorded as the generic term for birds in Dzihana, in 19th century Raßai, and among some, but by no means all, speakers of the northern and southern varieties of Digo. In Duruma, however, it refers solely to woodpeckers, one and probably more species of the family Picidae. In Kauma and northern Digo nyuni is also recorded as referring to a particular bird species, though whether these are woodpeckers too remains to be clarified. The second term, ts’ongo, has a much wider currency as a generic term: it has been recorded with this meaning in all of the dialects except Dzihana (on which lexical information is too sparse for its use to be ruled out) and Duruma. In Duruma, though, ts’ongo is the name of a bird, otherwise unidentified, which feeds upon sorghum and other grain crops. Early records indicate that ts’ongo had the same restricted meaning in 19th century Raßai and suggest that it has only displaced nyuni as the generic term for birds in the relatively recent past. The third word, nyama, is the common Mijikenda term for animals (and their meat) and appears in a number of different forms and expressions. In Duruma nyama ya kuburuka, literally “flying animal”, is the generic expression for a bird. Similar phrases, meaning “animal with wings” are recorded for northern Digo (chinyama cha maßa, mnyama wa maßa) as well as Giriama (nyama wa mahaha). Giriama also use the diminutive form kanyama, “little animal”. The overall result, as can be seen, is an extremely irregular pattern of usage and distribution which does not, for example, conform neatly to the division of Mijikenda into northern and southern groups nor the further subdivision of these into individual dialects.

Given the relative youth and closeness of the Mijikenda dialects, it is unusual to find their speakers at variance over the term for such a basic feature of everyday life. The coastal mosaic of forest and shrub hosts a rich avifauna which the Mijikenda are much less indifferent to than a quick glance at their generic terminology would suggest. Birds play a significant practical role in the lives of rural Mijikenda, and not just as crop pests or shooting practice for children. They provide an occasional source of meat as well as other specialized products like the feathers (especially of vultures) for fletching arrows. Perhaps more importantly, birds act as sensitive indicators of changes in the weather and local environment, and they help to identify, for example, the most poisonous Acokanthera trees or, in the case of the Black-throated Honeyguide, the sources of wild honey. Their economic importance is underlined by the fact that the Mijikenda have a large vocabulary to describe the different bird species they know (but none, for example, to describe different kinds of butterfly). Thus the published dictionaries, although they are far from complete, include more than one hundred different names of birds, and there are no doubt many waiting to be recorded, not to mention identified. Why then so some Mijikenda call birds “flying animals”? And why are there such basic differences in terminology that one speaker’s bird is another’s particular species and vice versa?

The answer to these questions turns upon the role which birds play as omens of the future, and the way in which Mijikenda beliefs in this respect have developed over time. To begin with, let us look at the general form of these ideas. Like their Swahili neighbours, not to mention many other peoples in the region, the Mijikenda believe that certain birds, most notably owls, are harbingers of misfortune. On top of this, however, the Mijikenda have a much more specific set of beliefs about the role of birds as omens. We owe the most detailed description of these beliefs, or at least one variant of them, to J.B. Griffiths, a Methodist missionary who lived among the Duruma of Mazeras for some 35 years. His account, which was published in 1935 (see the bibliographic note at the end of this article), is reported in full below:

“They [the Duruma] are in the habit of taking the auspices of two birds which are called Jelele and Kokota. Jelele has a deep blue coat and a red beak, and makes its home in a hole it burrows in the bank of a watercourse or a pit. It is never heard except when it screams in flying from tree to tree. Kokota has a brown coat with regular rows of white circular spots and a reddish crest, and makes its home in a hole it digs in the bole of a dead tree. It is frequently heard to call “Nje-nye-nje” after a spell of tree-tapping, the taps being so rapid that one can hardly take count of them. They are shy birds and are not often seen near dwellings.
The nature of the omen is determined by the position of the birds at the time one hears them. If the birds are in front of one, it is a warning not to proceed; and if they are behind one, it forebodes trouble in one’s absence. If they are on one’s right, it foretells of good health with scarcity of food; and if they are on one’s left, it foreshadows plenty of food and poor health. But if the one is on one’s right and the other is on one’s left, or if a Jelele or a Kokota is on one’s right and another Jelele or Kokota is on one’s left, the omen is that of good health and success.
The second, fourth and eighth of the first and second decades of the moon and the second and fourth of the last three eight days of the moon are called the “Days of the Birds.” They are the auspicious days of the people. Travelling, removing, building, cultivating, planting, harvesting, sacrificing: in fact, everything of importance, domestic or tribal, is commenced on one of these days.
It was well known to me that they were in the habit of listening to these birds, but it was a long time before I understood why they did so. One morning, when I was on a journey, a Jelele flew screaming across our path, and I heard the man who was in front of me mutter to himself: “I wonder whose shade that is.” That gave me my first cue. I have toiled many years among these people since then, and I have no doubt that they regard these birds as the mouthpiece of the shades, and that is why they consult them. They listen to the birds to find out the disposition of the family shades.”

This passage provides a fascinating picture of the way in which a people’s ideas about birds can permeate their everyday lives. Unfortunately nothing more is known (at least to this author) about the jelele, its identity (could it be a kingfisher?) and its role as a bird of omen. The kokota can be positively identified as a woodpecker, if not the Nubian Woodpecker, Campethera nubica, then one of the related species with a red crown or nape – or, equally probable, all these together. Kokota is also recorded as a term for woodpeckers among the Giriama and Raßai. Meanwhile, contemporary Duruma from the Taru area contradict Griffiths by saying that their name for this bird is nyuni and that it is the Digo who call woodpeckers kokota.

Whatever the case, there are a number of indications that the ideas and practices described by Griffiths did not originate with the Mijikenda, but were borrowed from elsewhere. Both jelele and kokota appear (because of the presence of the non-inherited phonemes /j/ and /t/) to be loanwords in Mijikenda, and though the latter is clearly related to Swahili gogota, also a woodpecker, the immediate origin of both words is unclear. More significant, perhaps, is the fact that analogous beliefs do not exist among, or at least have not been described for, the Mijikenda’s closest relatives, the Pokomo, Swahili and other speakers of Sabaki languages. If this specific set of ideas about birds of omen did not originate with the Mijikenda themselves, then where did it come from?

A first clue comes from the following passage, taken from Gerhard Lindblom’s classic 1920 monograph on The Akamba. After mentioning the role of owls as birds of ill-omen, Lindblom goes on:

“The most important and best known of all prophesying animals is also a bird, the ŋgomakomi, a red-headed species of woodpecker, to which the natives listened, especially in former times, before marching out on plundering expeditions. It is considered to be a good or bad omen according to the side on which one hears his pecking. The interpretation varies to some extent in different parts of Ukamba; the following detailed account is from Kikumbuliu, the south-east part of the country.
If the bird is heard straight in front, one will “see blood”, i.e. get scratched in the thickets, be gored by a rhinoceros or wounded in fighting, etc.; which of these things is most probable depends on the object of the expedition or the environment one is in or is going to be in. To hear the bird in front in an oblique direction and high up is also a bad sign, whereas if it is low in the same direction it only means that the listener will return without having effected his object. The left side is, on the other hand, the good side (in other districts the bad one), and if the bird is heard on that side, one has prospects of acquiring women, cattle and other wealth. Finally, if it is heard from behind, it denotes that the listener will carry a burden, so that if he is going out hunting he will probably shoot something, if he is about to cut the honeycombs from the beehives, he may be pretty sure of a good result, and similarly with those who are going to steal cattle, etc.
This woodpecker is looked upon as a messenger from the ancestral spirits; it is not killed, and its flesh may not be eaten by men. This prohibition does not apply to women, probably because as a rule they do not know of this bird, as they seldom have cause to go out into the desert, where the bird principally stays. In the immediate neighbourhood of Machakos, where trees are very rare and the bird is consequently not found, only a very few people seem to know of it. The Akamba who live there also carried out most of their campaigns on the steppe, where they probably had no opportunity of observing it.
The natives state that even certain animals, such as the giraffe, wild boar, etc. are so shrewd that they listen to and understand the ŋgomakomi’s call.”

Lindblom continues his account by giving the special Kamba names for the different directions from which the woodpecker is heard, before concluding with an anecdote about a travelling party which turned back home after hearing the birds call on the second day of their journey. The beliefs he describes are sufficiently similar to those of the Mijikenda to enable us to posit some connection, but different enough to render it unlikely that they were borrowed directly by them. Similar ideas are also held by at least two East Nilotic peoples, the Turkana and (as Lindblom also noted) the Maasai. Again, some kind of connection can be posited, though it is not clear between whom or in which direction, and further elucidation of this wider pattern of diffusion remains beyond the scope of the present article.

The connection between the Kamba and the Mijikenda is a tangible one. Most Kamba live far to the west of the Mijikenda, the two peoples separated by the vast semi-arid expanse of the Tsavo. For the past 200 years or more, however, Kamba have traded with the coast, while by the mid-19th century significant numbers of them were settled permanently in close proximity to the Raßai and Duruma. The Kamba also speak a Bantu language, though it is not very closely related to Mijikenda nor even a member of the North East Coast group to which the Sabaki languages belong. Instead, Kamba speech is classed together with that of the Kikuyu, Meru and others living in the vicinity of Mount Kenya as belonging to the Central Kenya Bantu, or what is otherwise known as the Thagicu group of languages. Most of the cultural and linguistic ties between the Mijikenda and Kamba arose not as a result of their direct contact but through the historical influence of another member of this group, the Segeju. And it is to these people that we must look to find the immediate source of Mijikenda ideas about birds of omen.

Who are or were the Segeju? It is tempting to say that if Segeju history had not happened it would have been impossible to invent; although this is precisely what earlier historians, misled by the Segeju’s own traditions, tried to do. A reliable picture, informed by linguistic research, has only begun to emerge in the past decade, and many of its details have yet to be filled out. The following is a brief outline.

Contrary to earlier speculation, the linguistic evidence makes it clear that the Segeju originate from the same area as other Thagicu speakers, most probably somewhere on the upper reaches of the Tana River. Indeed the different names by which the Segeju are known are all variants of the name Thagicu, which was given to the language group because of its widespread occurrence in the historical traditions of different members: “Segeju” itself is derived from the Swahili version (wasegeju, whereas the Mijikenda call them asagidzu, the Sagidzu). Sometime in the 16th century, possibly before, the ancestors of the modern Segeju migrated down the Tana River and settled near to the coast. There they became involved in shifting patterns of conflict and alliance with their neighbours, including the different Swahili communities in the area. This brought them to the attention of Portuguese visitors to Malindi, where the first reference to the Segeju dates from 1569. Before the end of the century they had achieved considerable fame for their military exploits in support of the ruler of Malindi. The Portuguese portrayed them as barbarous pastoralists who subsisted on fresh blood and milk, fearless warriors who kept the genitals of their victims as trophies of war. This notoriety was, however, short-lived. The Segeju were pushed south by another, much larger, group of pastoralists, also responsible for destroying many Swahili settlements along the coast: the Galla or Oromo. By the mid-17th century a number of Segeju had settled in and around Bwiti on the north-east edge of the Usambara Mountains. Their modern descendants are mixed farmers who call themselves Daisũ (cognate with Thagicu) and still speak a Thagicu language. Meanwhile, still in the 17th century, a significant section of them moved down to the coast where they acted as mercenaries for the Swahili of Vumba Kuu (near Vanga), subsequently settling in the territory they had helped to conquer. Today most of these Segeju live on the coast between Tanga and the Kenya-Tanzania border, where they speak a dialect of Swahili and/or a variety of Digo as their mother tongue.

The history books are largely silent about past relations between the Segeju and Mijikenda. Likewise the Segeju themselves, although they do recall more recent relations with their Digo neighbours. By contrast, recorded Mijikenda traditions have a lot to say about the Segeju, especially their shared conflict with the Galla and subsequent flight south. Some northern Mijikenda – Chonyi – even go so far as to claim that they were once Segeju. This is not as far-fetched as it may seem, if it is taken to mean that some Segeju were absorbed by the Mijikenda during this early period. Indeed, it becomes a very real possibility when the linguistic and other cultural evidence is considered. This evidence indicates a very close interaction between the Mijikenda and the Segeju in the past. The Mijikenda lexicon, for example, is replete with loanwords from a Thagicu language which can be conclusively identified as that once spoken by the Segeju, the direct ancestor of modern Daisũ. Similarly, and in conjunction with the linguistic evidence, many cultural practices of the Mijikenda can be traced to the Segeju. The picture which emerges confirms, corrects, and at the same time is much richer than that bequeathed to us by the Portugese. It shows, for example, that the Segeju were much more than just livestock herders and efficient fighters: they were also traders and left an important legacy of political organization and ritual practice.

Mijikenda ideas about birds of omen are part of this legacy. The existence of parallel beliefs among the Kamba, close relatives of the Segeju, is only one indication of this. Much stronger evidence, however, comes from the terminology employed by Mijikenda to refer to omens and the practices associated with them. Many of these terms were borrowed directly from the Segeju, as can be surmised from their phonological characteristics and the fact that cognates can be found in other Thagicu languages. This includes the word for a portent or ill-omen, mudhana (as recorded in Giriama and 19th century Raßai). Likewise verbs reported to mean “to seek an omen from the birds” (given as kudhecha or kudheja) and “to meet with a good omen” (recorded as kudhenja). Also belonging to the same set is another expression recorded from Raßai, kuera nyuni, “to take the bearing of birds when seeking an omen from them”, recalling that it is the direction in which they are heard which determines the interpretation of the omen. The wider vocabulary of prediction and prophecy is similarly permeated with Segeju loanwords. One of these is the name of a bird, called madhio or mario. These are described as large birds which are rarely seen except in flocks circling in the sky, and whose appearance is taken as a sign that the rains are imminent.

This brings us back to our original puzzle: why do the Mijikenda have different generic terms for birds? The answer lies with the Segeju and the distinctive set of ideas about birds of omen which they introduced. Before their intensive contact with the Segeju it seems that the Mijikenda had a far less developed notion of birds as omens, something akin perhaps to their current belief in the misfortune presaged by owls. Like other Sabaki speakers they had inherited a single generic term for bird, nyuni, which may already have had the secondary and extended meaning “omen”, as it does in Swahili and the modern Thagicu languages. The Segeju changed all that. By introducing a new and more pervasive complex of ideas about birds of omen they unwittingly set in motion the processes of semantic change and innovation which have given this particular segment of Mijikenda ethno-taxonomy the heterogeneous shape it has today. In the speech of many Mijikenda the secondary meaning of nyuni as an “omen”, including omens drawn from sources other than the birds, became its primary connotation. Meanwhile ts’ongo, the name of a common and gregarious species, began to take its place as the generic term for birds. The current distribution of these terms suggests that this process began in one of the northern dialects, probably Giriama, and has since been spreading south. In 19th century Raßai nyuni was still the generic term for both birds and omens: today, however, ts’ongo is used for birds. The latter name has also spread further south to the Digo: in this case, though, it has not succeeded in erasing earlier usage or preventing the adoption of other alternatives.

These alternatives stem from a different, but parallel, process, initiated by the speakers of another southern Mijikenda dialect, Duruma. Among the Duruma nyuni became restricted in meaning to the bird species which provides most of their omens: the woodpecker described by Griffiths and called kokota by other Mijikenda. In place of nyuni as the generic term for birds the Duruma adopted a neologism: nyama ya kuburuka, “flying animal”. Variations upon this neologism have since spread to other Mijikenda, including the Digo (“animal with wings”) and Giriama (“little creature”), perhaps in response to the confusion caused by the conflicting connotations of nyuni and ts’ongo. The result is today’s complex and cross-cutting pattern of usage. To all intents and purposes, this is still in a state of flux and it may only be resolved in the long run by adoption of the Standard Swahili term ndege, which is already gaining some currency among young and educated Mijikenda speakers.

If there is a lesson to be drawn from this, it is surely that there is often much more to a name or a classification than even its users might suspect. Ethno-taxonomy is typically treated as little more than a guide to identification, a diversion from more serious pursuits, and at best a source of material for parentheses and footnotes. As a serious topic of study ethno-ornithology is virtually non-existent, and it is no accident that most of the names of birds cited above await proper identification. However, the wealth of information contained in these names, not to mention the associated practices and beliefs, suggests that the collection and study of data of this kind is far from worthless, and is ignored at the cost of a fuller understanding of the human and natural environment in which we live.


In writing this article I have drawn upon a large number of sources in addition to my own research among the Mijikenda. Many of these sources are listed in Thomas Spear’s standard, though now outdated, study The Kaya Complex: A History of the Mijikenda People of the Kenya Coast to 1900 (Nairobi, Kenya Literature Bureau, 1978).

Linguistic data are taken from L. Krapf and J. Rebmann (ed. T.H. Sparshott) A Nika-English Dictionary (London, Society for Promoting Christian Knowledge, 1887) and W.E. Taylor Giriama Vocabulary and Collections (London, S.P.C.K., 1891), as well as more recent work by Thomas Hinnebusch, Philip Sedlak, Wilhelm Möhlig, Derek Nurse and myself. Among the sources on the Segeju special mention should be made of Derek Nurse’s paper “Segeju and Daisũ: A Case Study of Evidence from Oral Tradition and Comparative Linguistics”, History in Africa 9, 175-208 (1982).

The two passages reproduced in the text are taken from pp.276-277 of J.B. Griffiths “Glimpses of a Nyika Tribe (Waduruma)”, Journal of the Royal Anthropological Institute 65, 267-296 (1935) and pp.293-294 of Gerhard Lindblom The Akamba in British East Africa: An Ethnological Monograph (Uppsala, Appelbergs Boktryckeri Aktiebolag, 1920).

Otherwise Mijikenda, and in particular Kauma, beliefs about birds are set in fascinating context in Maurice Kambishera Mumba’s novel The Wrath of Koma (Nairobi, Heinemann Kenya, 1987). Finally, and for an introduction to the ethno-ornithology of another Kenyan people, readers are recommended to turn to Anthony Barrett’s Akiyar A Ngiturkana: Turkana Way of Life (Nairobi, New World Printers, 1988), “a book of poems, stories and pictures of birds found in Turkana country”.

Friday, 3 April 2009

THROWING AWAY THE DEAD: Communal Sites for the Disposal of Corpses in Pre-colonial South-west Tanzania

by Martin Walsh

[This is a corrected version of an article originally published in 1998 in Mvita: Bulletin of the Regional Centre for the Study of Archaeology in Eastern and Southern Africa, 7: 1-4. For some time it was available on the National Museums of Kenya website, together with a French translation by Edouard Bugingo (‘Jeter les Cadavres: Sites Communs où sont Déposés les Cadavres au Sud-Ouest Tanzanien dans la Periode Pre-Coloniale’). This led to its incorporation in The Rough Guide to Tanzania, which since 2003 has been suggesting to travellers that they try seeking out one of several “ritual sites next to ravines from where the Sangu tribe threw their dead (and, sometimes, the living, if they had been convicted of certain crimes, including philandery).” I’d be interested to hear from anyone who has tried!]


The ethnographic record indicates that disposal of the dead by leaving their corpses in the bush was once a widespread feature of mortuary practice in East Africa, practised by many Bantu as well as non-Bantu speakers. The following notes describe an interesting variation upon this practice, recorded (and still remembered) among different groups of Bantu speakers in what is now south-west Tanzania. Whereas most peoples known to have disposed of their dead in this way appear to have done so in any convenient spot away from habitation, the Sangu and others referred to below “threw away” their dead in particular sites designated for this purpose. These sites for communal disposal are recognisable by their common names and, though no longer used in quite the same way, are still feared through their association with death.

Sangu Practice
I have most information on the practice of the Sangu (avasango), among whom I conducted anthropological research between 1980 and 1982. The Sangu are the indigenous inhabitants of the Usangu Plains, which lie to the north of the mountain ranges which rise up from the northern shores of Lake Malawi. They speak an Eastern Bantu language which is classified in the Southern Highlands group together with Hehe, Bena, Wanji, Kinga, Pangwa, Kisi and Manda (Nurse 1988: 59). The development of trade routes from the East African coast in the early nineteenth century appears to have stimulated the early unification of the Sangu into a single chiefdom. The Hehe followed suit, and in the late 1870s forced the Sangu out of Usangu. The Sangu chief, Tovelamahamba Merere, established a new capital in the hills of Usafwa, to the west. The Sangu remained in exile there until after the turn of the twentieth century, when they were restored to Usangu (under Tovelamahamba’s son and successor, Mgandilwa Merere) by the newly established German administration (see Walsh 1984 for a detailed survey of Sangu history).

The contemporary Sangu bury all of their dead, and appear to have done so since the first decade or so of the colonial period. Earlier European travellers and missionaries, however, noted otherwise. When Elton and his party visited Tovelamahamba Merere in 1877 at Mfumbi, on the southern border of Usangu, they found large numbers of decomposing corpses, the victims of Hehe assaults, piled up outside of the stockade. At first it seemed that this may have been a temporary exigency of warfare, but after observing a dead woman being thrown into the bush Elton concluded that the Sangu did not in any event bury their dead (1879: 350-351, 358, 361-362). The Moravian missionaries who founded a mission station in 1895 outside of the Sangu capital-in-exile, Utengule-Usongwe (Kwa Mwalyego in Usafwa), were shocked to learn that corpses were being tossed into a nearby ravine. They appealed to Mgandilwa Merere to stop this practice, which he did, at least in this particular place (PA 1896: 288).

Comparative evidence suggests that the Sangu, like many of their neighbours, once disposed of all of their commoners in this way. Indeed contemporary linguistic usage has preserved the memory of this practice: the phrase kitaga umunu, ‘to throw away a person’, has been retained in ishisango, the Sangu language, as a polite euphemism for burial, and is used in preference to the verb kisiila, which means ‘to bury’ pure and simple. However, chiefs and certain other special categories of person, including twins and their parents, are said to have always been buried. The burial of the Sangu chief was a particularly elaborate affair, and when Tovelamahamba Merere died at the end of 1893 he was consigned to the grave together with a number of retainers (Heese 1913: 141), as well, it is said, with a large number of elephant tusks and other worldly goods to help him on his way.

Contemporary accounts also make it clear that the Sangu possessed designated sites for the disposal of corpses. The ravine near the Moravian mission station at Utengule-Usongwe was presumably one of these. One site, called Pitago (sometimes Kwitago), literally ‘the place for throwing away’, is located at the north-east of the twentieth century Sangu capital, Utengule-Usangu. It is possible that this site dates back to the nineteenth century, because Tovelamahamba Merere’s pre-1877 capital also stood close by. Informants cannot recall Pitago ever having been used for the general disposal of the dead, though it is described as a sacred site on which sacrificial offerings can be made for the whole population. According to one middle aged male informant it was used in the past (in the days of his grandfather) to dispose of transgressors, such as men who had unwisely seduced wives of the chief. In such a case the tribal elders would approach the accused and ask him to dress in his best clothes in order to accompany them to Pitago to sacrifice a bull. Once there they would slaughter the bull, and shortly thereafter the unwitting victim, whose corpse they would simply throw down on the spot. My informant thought that the victim’s decapitated head might have been brought back to the settlement for display, but he was not sure on this point. He added that in the daytime corpses left at Pitago would be consumed by vultures, and at night by hyenas (compare Elton: “...and now over a heap of skeletons, scattered leather aprons and beads, hovered flocks of vultures and gigantic storks, which, gorged with their loathsome feast, had scarcely power to flap away into the lower branches of the magnificent forest trees which adorn the once peaceful Usango valley”, 1879: 350-351).

This account recalls that of the Moravian missionaries, who discovered the Sangu practice of throwing away the dead following an incident in 1896 when Mgandilwa Merere had ordered the execution of two of his wives and a man for adultery. We can hypothesise that at this time Sangu mortuary practice was in a stage of transition, as burial became increasingly fashionable among commoners, perhaps initially for those who held political or military office and their close relatives. Disposal by exposure was subsequently restricted to the corpses of criminals and people in lower social categories, such as non-Sangu (as suggested by another informant in discussing the uses of Pitago). Thereafter burial became the universal practice: this was certainly the case in the south of Usangu around Brandt Lutheran Mission when Heese (1913) wrote on local customs. At the same time the chief / commoner distinction remained marked by different forms of burial: chiefs and the special ritual categories of persons treated analogously to them were and still are buried in a sitting position, whereas ordinary people are buried lying down.

Pitago in Utengule-Usangu still has strong associations with death for the local population. Broken pots and other remains are said to be found there, and the site is thought to be especially fearful during and after the rainy season when the grass has grown long. In 1981 Pitago had been set aside as a site for a future village cemetery, an appropriate transformation of its original purpose. I was unable to establish whether other, similar sites, are recognised elsewhere in Usangu: it may be that communal exposure was only practised in the vicinity of the Sangu capital(s), where the density of population and numbers of people dying over time meant that disposal of corpses in isolated spots in the bush was not as feasible an option as it was in less densely settled and cultivated areas. Certainly all of the known sites of collective disposal by the Sangu are located close to past and present capital settlements both in and outside of Usangu (Mfumbi, Utengule-Usongwe, and Utengule-Usangu), though this may be a function of observer bias.

The River Patagu

One possible exception is the River Patagu, which forms the boundary between the territories of the Sangu and the Poroto in the south west corner of Usangu. The Poroto are often classed as a sub group of the Safwa, though they have retained something of a separate identity, and are definitely thought of as separate by the Sangu. Like the Safwa they speak a language which is not particularly closely related to that of the Sangu, but is classified in the Nyika sub group and Corridor group of Eastern Bantu languages (other languages of the Nyika sub group include Lambya, Malila, Nyiha, and Tambo: Nurse 1988: 20). Throughout the German colonial period and for some years afterwards a section of Poroto territory across the Patagu was under the nominal control of a Sangu chief. This was Kahemere, a brother of Mgandilwa Merere who had refused to recognise the latter’s accession to the Sangu stool and subsequently also refused to follow him back into Usangu (though he relented during the British period, long after the death of Mgandilwa, when he was offered a sub-chiefship in south-east Usangu).

The Sangu nickname for the Poroto is avaxawuxa, which literally means ‘the dried up ones’. This is a reference to what the Sangu think of as the deep and throaty voices of the Poroto, which are alleged to be a consequence of their habit of drinking water from the River Patagu. No one else, it is said, will drink from this river. The implication of this, recalling also the name of the river (which is cognate with the Sangu Pitago), is that corpses were once cast into it and have therefore contaminated the water, rendering it unsuitable for human consumption. I have no information, however, on who might have been responsible for this (the Sangu or even the Poroto themselves), or whether the disposal of bodies in the river was a singular occurrence or a regular event.

Nyakyusa and Kukwe Practice

The existence of communal exposure sites can be more readily documented among the Nyakyusa and related peoples whose territory begins further to the south west of Usangu, beyond that of the Poroto. The Nyakyusa (once known, together with the closely related Ngonde, as the Konde) live on the plains and in the mountains at the northern tip of Lake Malawi. Their language, like that of the Poroto, belongs to the Corridor group of Eastern Bantu, but is classified, together with that of the Ndali, in a separate sub-group (Nurse 1988: 59). Like the Sangu, the Nyakyusa now bury their dead, but there is good evidence that this was not always the case:

“Here and there throughout Kondeland are places called Itago, so named from the verb kutaga, to cast away. In the long past, when a man was dying, and all hope of recovery had been abandoned, he was carried to the Itago, placed in a sitting position, and left to die. After death the flesh was devoured by birds or beasts. Nowhere, as far as I have discovered, is this repulsive practice now followed.” (MacKenzie 1925: 296)

While working in Usangu I spoke to Kukwe informants who confirmed that ‘throwing away the dead’ (no mention was made of the dying) was once practised at such communal sites. The Kukwe live in the north-west of Unyakyusa, on the western side of Mount Rungwe, and maintain a separate identity, despite their adoption of common Nyakyusa culture. One Kukwe woman I spoke to knew of a particular cliff called Itago, over which corpses had been cast in the days before burial had become widespread. This place was also called Ipanga after its steep cliff. She opined that corpses treated in this way would simply decompose, given the absence of either vultures or hyenas in this part of Kukwe territory. Another Kukwe informant, a man, recalled the existence of a similar site called Itagano, east of the main road to Tukuyu, from which corpses were also once cast. Modern maps of the Rungwe area show a number of possibly related place names, including Itaga (a place and a river), c.14 km north of Itumba, and Itagata, c.10 km south of Mount Rungwe. These might well be worth further investigation.


While there are clear parallels between what is known of Sangu practice and that of the Nyakyusa / Kukwe, it is difficult to determine whether these might be the result of independent developments (out of a common culture of throwing away the dead) or of contact between the two peoples. The Sangu and the Nyakyusa / Kukwe are not closely related and were not immediate neighbours in the pre-colonial period, although various contacts can be traced between them in addition to a pattern of raiding by the Sangu which was particularly intense during their period of exile in Usafwa. The Sangu are the only members of the Southern Highlands group who are known to have regularly exposed the dead at communal sites. It is possible, however, that further research among the Corridor peoples will reveal that this practice had a wider distribution than I have been able to document here. The apparent absence of the practice among other Southern Highlands speakers and the geographical proximity of the Sangu to Corridor speakers suggest that the Sangu may well have borrowed it from the south and west.

Otherwise it is conceivable that the Sangu and Nyakyusa / Kukwe practices represent independent developments. In the Sangu case the exposure of corpses on designated sites outside of the royal capital might be interpreted as a solution to the problem of disposal in a context of high population density. Whether or not high population density also favoured the development of a similar practice among the Nyakyusa / Kukwe is more difficult to determine, though it seems quite possible given reports of dense settlement in some chiefdoms in the immediate pre-colonial period. The Kukwe predilection for casting their dead off the tops of cliffs suggests that the nature of local topography may also have played a role in the evolution of this practice, although MacKenzie’s report indicates that the dead (or rather dying) in Unyakyusa were not everywhere thrown away in such a dramatic fashion. The cold climate and absence of animal scavengers in Ukukwe may also have made it a more attractive option for the living to place some vertical distance between themselves and the slowly rotting dead. The Usangu Plains are completely lacking in mountainous terrain, but are (or at least once were) replete with suitable scavengers to complete the work of disposal. Once they had moved up into Usafwa, however, the Sangu were quick to make use of local ravines for the same grim purpose.

If anything, it is evident that further research is required to determine the distribution of communal sites for the disposal of corpses and the possible reasons for this apparently unusual variation upon the practice of throwing away the dead. The preservation of these sites in name and memory, as well as their contemporary use as sacred (and possibly burial) sites, suggests that ethnographic enquiry on this subject may still bear fruit, despite the passage of time since they were last used for their original purpose. The practice of throwing away the dead (and abandoning the dying), whether randomly or in such designated places has obvious implications for our interpretation of the archaeological record. The prehistory and history of mortuary practices in south-west Tanzania, not to mention elsewhere in the region, remains to be described in detail. Exposure of the dead was clearly an important component of this history, at least in the immediate pre-colonial period. At present, however, we can only guess at what future research might reveal about the past of this and other mortuary practices.

My research in Tanzania in 1980-82 was funded by the then Social Science Research Council (U.K.), with additional support from the Smuts Fund and Wolfson College in the University of Cambridge. I am very grateful to my hosts in Utengule-Usangu and all those who provided me with information on the topic of this article, in particular Ngwila Simuhongole, Eliuter Shinangonele, Betitha Mwakalinga, and Jackson Mwakabalile.


Elton, J. F. 1879. Travels and Researches among the Lakes and Mountains of Eastern and Central Africa (edited by H. B. Cotterill). London: John Murray.

Heese, P. 1913. “Sitte und Brauch der Sango”, Archiv für Anthropologie, 40 (n.s.12): 134 146.

MacKenzie, D. R. 1925. The Spirit ridden Konde. London: Seeley, Service & Co.

Nurse, D. 1988. “The Diachronic Background to the Language Communities of Southwestern Tanzania”, Sprache und Geschichte in Afrika, 9: 15 115.

[PA] = Periodical Accounts Relating to the Foreign Missions of the Church of the United Brethren (London).

Walsh, M. T. 1984. The Misinterpretation of Chiefly Power in Usangu, South-west Tanzania, unpublished Ph.D. dissertation, University of Cambridge.

JETER LES CADAVRES: Sites Communs où sont Déposés les Cadavres au Sud-Ouest Tanzanien dans la Periode Pre-Coloniale

by Martin Walsh, translated by Edouard Bugingo

[This is a translation of 'Throwing away the dead: communal sites for the disposal of corpses in pre-colonial south-west Tanzania', originally published in Mvita: Bulletin of the Regional Centre for the Study of Archaeology in Eastern and Southern Africa, 7: 1-4 (1998). It was posted on the National Museums of Kenya website together with the text of the English original, and the following is the copy that I made when it was still online. I didn't participate in the translation and haven't attempted to correct it.]


Le rapport ethnographique indique que le système de se débarrasser des morts en jetant leurs cadavres dans la brousse était à un certain moment une pratique funèbre très répandue en Afrique orientale chez la plupart des peuples bantu et même parmi d’autres qui ne parlent pas la langue bantu. Les notes qui suivent décrivent une variété intéressante de cette pratique enregistrée (et toujours en cours) chez les différents groupes parlant le Bantu dans ce qui est de nos jours le sud-ouest Tanzanien. Alors que la plupart de gens connus pour s’être débarrassé des morts de cette manière semblent l’avoir fait en un endroit convenable à cet effet loin des habitations, les Sangu et bien d’autres abandonnaient leurs morts dans des sites particuliers désignés pour cette cause. Ces sites où les corps étaient jetés en commun se reconnaissent par leurs mêmes dénominations et bien qu'ils ne soient plus utilisés de la même façon, ils inspirent toujours la peur car ils sont associés à la mort.

La pratique des Sangu

La grande information dont je dispose est sur la pratique des Sangu (Avasango), parmi lesquels j’ai dirigé une recherche anthropologique entre 1980 et 1982. Les Sangu sont des habitants indigènes des Plaines Usangu qui s’étend au nord des chaînes montagneuses s’èlevant des côtes septentrionales du lac Malawi. Ils parlent la langue bantu orientale classifiée dans le groupe Hauts Plateaux du Sud avec les langues Hehe, Bena, Wanji, Kinga, Pangwa, Kisi et Manda (Nurse 1988 p.59). Le développement des routes commerciales à partir de la côtes Est Africaine au dèbut du 19e siècle semble avoir stimulé plus tôt l’unification des Sangu sous l’autorié d’un seul chef. Les Hehe intentèrent un procès et vers les années 1870, forcèrent les Sangu à quitter l’Usangu. Tovelamahamba Merere, le chef Sangu, fixa la nouvelle capitale dans les collines de Usafwa vers l’Ouest.

Les Sangu restèrent là-bas en exil jusqu’ après le tournant du 20e siècle, lorsqu’ils furent ramenés à Usangu (sous le règne de Mgandilwa Merere, fils et successeur de Tovelamahamba) par l’administration allemande récemment mise en place (voir Walsh 1984 pour une étude détaillée de l’histoire Sangu).

Les Sangu de l’èpoque actuelle enterrent leurs morts et semblent l’avoir fait ainsi depuis la première décennie ou environ de la période coloniale. Et pourtant un peu plus tôt, des voyageurs et des missionnaires européens avaient remarqué le contraire. Quand J. Elton et son équipé visitèrent Tovelamahamba Merere en 1877 à Mfumbi sur la frontière sud de l’Usangu, ils trouvèrent un grand nombre de cadavres en décomposition, c’ètaient des victimes des assaillants Hehe entassés par pile hors de l’enclos. De prime abord, il semblait que cela pouvait être dû à un impératif temporaire de la guerre, mais après avoir observé une femme morte et ensuite jetée dans le buisson, Elton a conclu que les Sangu en aucun cas n’enterrent pas leurs morts (Elton 1879 p. 350-351, 358, 361-362). Les missionnaires de Moravie qui, en 1895 ont fondé une mission hors de la capitale en exil de Sangu appelée Utengule-Usongwe (Kwa Mwalyego en Usafwa), furent choqués d’apprendre que les cadavres sont jetés dans un ravin tout proche. Ils lancèrent un appel à Mgandilwa Merere pur mettre fin à cette pratique, et il y a mis fin du moins pour cet endroit spécifique (PA 1896 p. 288).

Un témoignage comparé fait une suggestion comme quoi, à l’instar de leurs voisins, les Sangu avaient l’habitude de se débarrasser de leurs morts de cette façon. En effet, l’usage de la linguistique contemporaine a préservé la mémoire de cette pratique, l’expression “Kitaga Umunu” (jeter une personne) a été retenue en langue Sangu, Ishisango comme un euphémisme de politesse pour dire enterrement et on préfère l’employer á la place du verbe “Kisiila” qui signifie enterrer purement et simplement. Par ailleurs, on dit que des chefs et certaines autres catégories de personnes spéciales notamment les jumeaux et leurs parents devaient toujours être enterrés. L’enterrement d’un chef Sangu, était une affaire particulièrement minutieuse et lorsque Tovelamahamba Merere décéda à la fin de 1893, on le descendit dans le tombeau avec un certain nombre de ses sujets (Heese 193 p.141) ainsi que, dit on avec un grand nombre de défenses d’élephants et d’autres objets matériels pour l’aider au cours de son chemin.

De nos jours, l’opinion générale montre clairement que les Sangu avaient choisi des sites où ils mettaient leurs morts. Le ravin situé près de la mission moravienne à Utengule-Usongwe était sans doute un de ces sites. Il existe un site appelé Pitago (des fois on dit Kwitago) littéralement “le lieu où l’on jette”, il se trouve au nord-est de Utengule-Usangu, la capitale de Sangu au 20e siècle. C’est possible que ce site existe depuis le 19e siècle car la capitale de Tovelamahamba Merere, antérieure à 1877 s’ y tient debout à proximité. Des informateurs ne peuvent pas se rappeler si Pitago a servi comme un lieu public où l’on dépose les morts bien qu’il est décrit comme un site sacré où l’on peut sacrifier des offrandes pour tout le peuple. D’après un informateur d’ âge moyen, ce site servait du temps de son grand père de lieu final pour les malfaiteurs, tels les hommes qui se permettaient de séduire les femmes du chef. Dans un cas pareil, les chefs de tribu appelaient l’accusé et lui demandaient de mettre ses meilleurs vêtements pour qu’il les accompagne à Pitago pour le sacrifice d’un taureau. Une fois arrivés là, ils abattaient le taureau et directement après, ils tuaient cette victime dont le corps était simplement jeté sur place. Mon informateur pensait que la tête coupée du victime pouvait être ramenée à domicile pour y être exposée mais il n’était pas sûr sur ce point. Il ajouta que pendant la journée les cadavres laissés à Pitago était dévorés par les vautours et la nuit par les hyènes. (Comparez Elton “... et maintenant un tas de squelettes, des morceaux de chair et de perles éparpillés, une ruée de vautours et de cigognes géants planant au -dessus des cîmes d’arbres qui avoisinent la vallée Using, alors connue pour sa tranquillité, Elton 1879, p. 350-351).

Ce rapport rappele celui des missionnaires de Moravie, qui ont découvert la pratique des Sangu de jeter les morts d’après un incident survenu en 1896 lorsque Mgandilwa Merere avait ordonné l’exécution des 2 de ses femmes et un homme pour adultère. Nous pouvons faire une hypothèse que à cette époque, la pratique funèbre des Sangu se trouvait dans une étape de transition, puisque l’enterrement devenait de plus en plus commode parmi les habitants, sans doute en commençant par ceux qui détenaient un poste politique ou militaire ainsi que pour leur proche parent. Laisser les cadavres en plein air, c’était réservé aux cadavres des criminels et aux gens de basses catégories sociales, tels les non Sangu (Ceci m’a été dit par un autre informateur quand on discutait à propos de Pitago) Alors, l’enterrement devint une pratique universelle: Ce fut le cas certainement dans le sud de Usangu autour de la mission Brandt Lutheran quand Heese écrivit en 1913 sur les coutumes locales. En même temps, la distinction chef - paysan demeura marquée par les différentes formes d’enterrement: les chefs et les catégories spéciales et rituelles de personnes étaient traités selon leur rang, ils étaient et même de nos jours enterrés en position assise, alors que les gens ordinaires le sont en position couchée.

A Utengule-Usangu, Pitago est toujours fortement associé avec la mort par la population locale. Des pots cassés et d’autres restes se trouveraient sur les lieux et on pense que le site inspire plus de peur spécialement pendant et après la saison pluvieuse quand l’herbe a beaucoup poussé. En 1981, Pitago avait été choisi pour servir de site pour une cimetière de village plus tard, ce qui est une transformation appropriée de son rôle original. Je n’ai pas pu prouver que d’autres sites similaires sont reconnus ailleurs en Usangu: Il est possible que déposer les morts dans un lieu commun était seulement pratiqué dans les environs de la capitale de Sangu, où la densité de la population et le nombre de gens qui meurent, des fois, signifiait que déposer les corps dans des lieux isolés dans la brousse n’était pas aussi faisable comme c’était le cas dans des localités moins habitées et dans des régions cultivées. Sûrement, tous des sites communs connus par les Sangu se trouvent près des anciens et des nouveaux emplacements de la capitale, que ce soit à l’intérieur ou à l’extérieur de Usangu (Mfumbi, Utengule-Usongwe et Utengule-Usangu), à moins que cela puisse relever d’une erreur d’observation.

Le Fleuve Patagu

Une exception possible c’est le fleuve Patagu qui forme la limite entre les territoires de Sangu et de Poroto dans le coin sud-ouest de Usangu. Les Poroto sont souvent classés dans le sous-groupe des Safwa, bien qu’ils aient gardé quelque chose d’une identité séparée et effectivement, on enseigne qu’ils sont séparés d’avec les Sangu. Comme les Sakwa, ils parlent une langue qui n’est pas particulièrement apparentée à celle des Sangu mais qui est classée dans le sous-groupe Nyika et dans le groupe du corridor des langues bantu de l’est (d’autres langues du sous-groupe Nyika comprennent Lambya, Malila, Nyika et Tambo: Nurse 1988 p.20). Durant toute la période coloniale Allemande et pendant quelques années après, une portion du territoire des Poroto au-delà du fleuve Patagu se trouvait sous le contrôle nominal d’un chef Sangu. Il s’agissait de Kahemere, un frère de Mgandilwa Merere qui avait refusé de reconnaître l’ascension de ce dernier sur le trône Sangu et par conséquent avait également refusé de le suivre de retour à Usangu (en tout cas, il changea d’avis durant la colonisation britannique longtemps après´s la mort de Mgandilwa Merere, lorsqu’il fut nommé sous-chef pour Usangu du sud-est).

Le surnom que les Poroto donnent aux Sangu est celui de Avaxawuxa, ce qui se traduit littéralement par “les assoiffés”. Ceci est en référence à ce que les Sangu pensent à propos des voix basses et caverneuses des Poroto, lesquelles seraient une conséquence de leurs habitudes de boire l’eau du fleuve Patagu. On dit que désormais, personne d’autre ne boira plus l’eau de ce fleuve. Ceci implique, en rappelant le nom du fleuve (qui sonne comme le Sangu Pitago) que dans le temps, on y a déposé des cadavre qui ont ensuite contaminé l’eau la rendant impropre à la consommation humaine. Je ne possède pourtant aucune information sur la personne qui en aurait été responsable (les Sangu ou bien les Poroto eux-mêmes) ou bien si déposer les morts dans le fleuve était un fait particulier ou un événement régulier.

La pratique des Nyakyusa et des Kukwe

L’existence des sites où les morts étaient déposés en commun peut trouver une plus large documentation parmi les Nyakyusa et d’autres peuples dont le territoire commence plus loin au sud-ouest de l’Usangu au-dela de celui des Poroto. Les Nyakyusa (pour un temps connu être apparentés aux Ngonde, ou Konde) vivent dans des plaines et dans des montagnes qui touchent la partie nord du lac Malawi. Leur langue comme celle des Poroto appartient au groupe du corridor Bantu de l’est mais elle est classifiée avec celle des Ndali dans un sous-groupe à part (Nurse 1988 p.59). Comme les Sangu, pour le moment les Nyakyusa enterrent leurs morts mais il existe une preuve évidente que cela n’a pas toujours été le cas:

"Ici et là à travers le pays Kondela, existent des endroits appelés Itago, ainsi nommés à partir du verbe Kutaga, jeter. Dans un passé lointain, lorsqu’ une personne mourait et que tout espoir de récupération avait été abandonné, on la transportait à Itago, la plaçait en position assise et on l’ y laissait pour mourir. Après la mort, la chair été dévorée par des oiseaux ou des fauves. Je n’ai trouve nulle part où l’on imite cette pratique repoussante." (MacKenzie 1925 p. 296)

Alors que j’étais à l’oeuvre à Usangu, j’ai parlé avec des informateurs Kukwe qui m’ont confirmé que le fait de “jeter les morts” ( sans aucune mention de l’état du mourant) était une pratique courante dans ces sites communs. Les Kukwe habitent au nord-ouest de Unyakyusa, du côté occidental du Mont Rungwe et ils gardent leur identité propre bien qu’ils aient adopté la même culture que les Nyakyusa. Un jour j’ai dialogué avec une femme qui connait une falaise particulière appelée Itago, là où les corps étaient abandonnés avant que l’enterrement devienne chose commune. Cette place s’appelait aussi Ipanga à son endroit le plus raid. Elle fut d’avis que les corps ainsi abandonnés se décomposaient tout simplement étant donné l’absence des vautours ou des hyènes dans cette région du territoire Kukwe. Un autre informateur Kukwe, de sexe masculin, s’est rappelé de l’existence d’un site similaire appelé Itagano, à l’est de la grand-route qui même à Tukuyu, une place où les corps étaient également jetés. Les cartes actuelles de la région de Rungwe montrent un nombre de lieux ayant ces noms de même relation, notamment Itaga (un lieu et un fleuve), c:14 km au nord d’Itumba et Itagata, c:10km au sud du mont Rungwe. Cela nécessiterait des recherches plus approfondies.


Ce pendent qu’il existe des liens bien parallèles entre ce qui’est connu des pratiques Sangu et celles des Nyakyusa/Kukwe, il est difficile de décider si cela est dû au fait d’actions indépendantes (de la culture commune de jeter des morts) ou d’un contact entre les deux peuples. Les Sangu et les Nyakyusa/Kukwe ne sont pas directement apparentés et m’étaient même pas des voisins immédiats durant la période pré-coloniale, bien que des divers contacts entre eux peuvent être trouvés en plus que les Sangu avaient l’habitude de faire des raids particulièrement violents durant leur période d’exil à Usafwa. Les Sangu sont les seuls membres du groupe Hauts Plateaux sud, connus pour avoir laissé leurs morts dans des sites communs de façon régulière. Par ailleurs, il est possible que des recherches ultérieures parmi les gens du corridor vont révéler que cette pratique avait une plus large distribution que celle dont je dispose pour ma documentation. L’absence apparente de cette pratique chez les Bantu des Hauts Plateaux sud et la proximité géographique des Sangu aux Bantu du corridor suggère que les Sangu pouraient bien l’avoir empruntée du sud et de l’ouest.

Autrement, il est envisageable que les pratiques des Sangu et celles des Nyakyusa / Kukwe se sont développées indépendamment. Dans le cas des Sangu, le fait d’exposer les cadavres dans des sites bien désignés en dehors de la capitale royale, pourait être interprété comme une solution au problème de se débarrasser des morts dans un contexte d’une population à haute densité. Que ce soit une population à haute densité ou pas, qui aurait également favorisée le développent d’une pratique similaire parmi les Nyakyusa / Kukwe, cela est plus difficile de le dire, quoiqu’il semble tout à fait possible étant donné de rapports de peuplement dense dans certaines chefferies juste après la période pré-coloniale. La préférence des Kukwe de jeter leurs morts à partir des hauts des falaises montre que la topographie locale de la nature a joué un rôle dans l’évolution de cette pratique, bien que MacKenzie dans son rapport indique que les morts (ou les mourants) en Unyakyusa n’étaient pas partout jetés d’une manière aussi dramatique. Le climat froid et l’absence de fauves en Ukukwe pourait avoir rendu cette option plus attrayante pour les vivants d’établir une certaine distance verticale entre eux et les corps lentement en décomposition. Les plaines de l’Usangu ne connaissent pas du tout de terrain montagneux, mais elles ont (ont eu dans le passé) des fauves pour manger les cadavres complétant ainsi le travail d’abandon des morts une fois qu’ils sont montés vers Usafwa, les Sangu avaient pourtant vite appris à se servir des ravins pour leur sinistre travail.

Il est évident qu’une recherche ultérieure est requise pour déterminer la distribution des sites communs pour l’abandon des cadavre et des raisons possibles pour cette variation apparemment inhabituelle relative à la pratique de jeter des morts. La présentation de ces sites pour leur nom et mémoire, aussi bien que leur usage actuel comme site sacré (même pour l’enterrement) suggère que la recherche ethnographique sur ce sujet peut toujours porter des fruits en dépit du temps passé depuis qu’on les à dernièrement utilisés pour leur finalité originale. La pratique de jeter les morts (et d’abandonne les mourants) dans certains endroits ou dans des lieux bien désignés, à des implications évidentes pour notre interprétation du rapport archaeologique. La préhistoire et l’histoire des pratiques funéraires au sud-ouest Tanzanien, pour ne pas mentionner ailleurs dans la région, reste à décrire en détail. L’exposition des morts était visiblement un élément important de cette histoire, au moins juste après la période pré-coloniale. Pourtant, pour le moment, nous ne pouvons que deviner quelle recherche ultérieure pourrait éclairar à propos de son passé et d’autres pratiques funéraires.

Mon travail de recherche en Tanzanie en 1980-82 était financé par le conseil de la Recherche en Sciences Sociales (Royaume-Uni) avec le supplément de l’aide par la Fondation Smuts et le Collège Wolfson à L’Université de Cambridge. Je suis très reconnaissant à mes hôtes à Utengule - Usangu et à tous ceux ou celles qui m’ont fourni des informations sur le sujet de cet article en particulier Ngwila Simuhongole, Eliuter Shinangonele, Betitha Mwakalinga et Jackson Mwakabalile.


Elton, J. F. 1879. Travels and Researches among the Lakes and Mountains of Eastern and Central Africa (edited by H. B. Cotterill). John Murray, London.

Heese, P. 1913. "Sitte und Brauch der Sango", Archiv für Anthropologie, 40 (n.s.12):134-146.

MacKenzie, D. R. 1925. The Spirit-ridden Konde. Seeley, Service & Co., London.

Nurse, D. 1988. "The diachronic background to the language communities of south-western Tanzania." In Sprache und Geschichte in Afrika, 9:15-115.

[PA] = Periodical Accounts Relating to the Foreign Missions of the Church of the United Brethren (London).

Walsh, M. T. 1984. "The misinterpretation of chiefly power in Usangu, south-west Tanzania." unpublished PhD dissertation, University of Cambridge.

MERERE THE ARAB: Legitimacy Upside-Down and Inside-Out

by Martin Walsh

[This paper was presented to a symposium on the ‘Cultures of Southwestern Tanzania’ at the XI International Congress of the Anthropological and Ethnological Sciences in Vancouver, on 25 August 1983. Some of its arguments were developed further in my Ph.D. dissertation, The Misinterpretation of Chiefly Power in Usangu, South-west Tanzania (University of Cambridge, 1984).]


In Mbeya District Book there is “A History of Usangu. Related by the Wazee [elders] of Utengule”, written in English and dated 6 January 1930.(1) Half of the text comprises an account of the foundation of the Sangu royal line. Briefly: a stranger comes to Usangu via Uwanji and makes the daughter of a local chief pregnant. The product of this union, with the help of war medicine and iron spears provided by his father, proves a worthy successor to the rule of this petty chiefdom. His own son and successor, Mwahavanga, conquers and so unifies the whole of Usangu for the first time, some years before the coming of the Ngoni (whose main clash with the Sangu can be dated to the 1850s).

This is not untypical of histories of its kind. What is perhaps its most incongruous detail is introduced when the stranger, replying to a question from his Sangu mistress, declares: “ ‘The country where I came from is Ubarawa’ ”, which a footnote locates in Somaliland, the coastal town now known as Brava. Later the hero explains to both mistress and son that he is a Somali, before returning home for good after seeing his son assume power. Although in this last instance the claim is incorporated in the main body of the text, it is tempting to interpret these Somalian glosses as the interpolations of an unsigned administrative officer, someone with a handy geographical knowledge and imbued with the spirit of “the Hamitic hypothesis”.(2) At the very least an identification teased out by a leading question. If I add that this is the only Sangu history I have seen or heard which makes this connection, then you might wonder whether it really matters how or why this apparently idiosyncratic embellishment came into being. In this paper I want to argue that it does. Not that I hold out any hope of resolving this particular textual anomaly. It is, though, only part of a more general interpretive problem.

Similar themes recur in modern versions of this history. The stranger is, almost without exception, called Mbalawe and said, as the name implies, to have come from Ubalawe;(3) though no one can say exactly where this is, except that it is somewhere to the north or east. Descriptions of his appearance - “He was tall and slender and his complexion was fair” - and that of his son - “a fine youth; intelligent, handsome, fair and slim while his nose was high-bridged like that of an Arab” - echo these words from the District Book. The most frequent gloss is, indeed, that Mbalawe was or was like “an Arab” and Ubalawe “in Arabia” or thereabouts.(4)

There are only two other printed versions of this early history, collected by a Dar es Salaam University student in 1968. They are particularly explicit in drawing the Arabic link. One begins: “The Arab who is claimed to have come from Barawa was one of the Arabs who had come to East Africa for trading”, and leaves us in no doubt by naming his half-caste son “Saidi” and stating that he also left behind a copy of the Koran, further implying that this has remained an object of veneration through to the present day.(5) The other account speaks of Arab traders accompanying a group of missionaries travelling from Mombasa to Nyasaland: “one of these Arabs, whose native town was Barawa, left the missionaries going to Nyasaland at Uwanji and came down to [the] Usangu Basin for trade…”, before consorting with the daughter of a local chief.(6) If these histories are coupled with what is known of nineteenth century Sangu history, then it is not difficult to conjure up an explanation. There is, moreover, a parallel case to hand from north-eastern Tanzania. Steven Feierman, noting analogous claims about the provenance of Mbegha, founding hero of the Shambaa, writes: “Quite probably the Arabic connection was created in the second half of the nineteenth century as a fictional genealogical link to be referred to in dealings with coastal traders” (1974: 82). It is easy to extend this argument to the Sangu case. Their “rise” and aggressive expansion northwards in the 1830s has been explained as a response to the extension of Arab trading interests towards Lake Tanganyika in this period (Shorter 1972: 246-247); while the permanent presence of coastal traders at the Sangu court from the mid-1870s onwards is well documented (for reference to the most famous of these, see below). This prompted the first Moravian missionaries, arriving from the south in 1891, to write of “the powerful Arab chief Merere from the north” (PA, I, 8, Dec.1891, 415-417). Feierman also notes that two of his three sources on this topic appeared to possess aspirations or interests matching those Arabic ones they ascribed to their hero. The authors of the last two Sangu histories quoted share an even clearer affiliation, embodied in their commitment to Islam. While there may no longer be any need to legitimate coastal connections, Muslims, contemporary witnesses to such an orientation, can be trusted to preserve and elaborate upon earlier charters.

This is a plausible, but incomplete, explanation. There are, for example, important differences between the Sangu and Shambaa cases. Where the story of Mbegha is common knowledge (Winans 1962: 79), the history of the stranger from Ubalawe is not. While the few ascriptions of Arabic attributes to Mbegha can be glossed as “historical material woven into the myth, of a kind easily changed without altering the myth’s structure” (Feierman 1974: 83), the same details strike at the structural core of the Sangu version. If Sangu can remember anything at all of this history it will likely include some intimation that their former rulers are putatively descended from Arabs, or a peoples of similar origin. This goes for both Muslim and non-Muslim informants. Now, in seeking an explanation for this, we can proceed to the outline of a much richer account.

Legitimacy upside-down

Detailed histories of Mbalawe’s coming to Usangu are few and far between.(7) There is little agreement between different versions. Names, places, political setting, motives, pursuits, episodes and genealogical affiliations all appear in radically dissimilar permutations. These do not even preclude the identification of Mbalawe as a woman. Not surprisingly the fullest, and in many ways most compelling, interpretation I recorded begins by distancing itself from other accounts. I will quote at length.

“Sangu have told different stories about Mbalawe. In fact he came here like his elders, his masters, in the time when they were looking for elephant tusks, for ivory… Mbalawe was only a youth. People are mistaken if they say that he came here on some business of his own. He was only a young man [accompanying the traders].
Now the Sangu of long ago, the people of Mwana Mgawa and Mhami, when they saw this youth they had a yearning for his colour. ‘If only we could get his seed, his offspring!’ His masters, for their part, just wanted to trade, for ivory for example; and they had no scruples about leaving him behind because he was only a child and incapable of doing anything for himself. And [the Sangu] wanted his seed. They would let him sleep with one of their daughters. The one they chose was compliant, and they called her Shihwago… by likening her mild manner with that of a cow which will give milk to a child if it asks.(8) Whenever a gentle child is born into the royal family [these days] she is called Shihwago. And this is what the Sangu of long ago meant.
So, having decided this, they gave the young man [Mbalawe] girls as he pleased, including the one who was to [bear and] bring up his child. The reason they did this was to acquire the appearance which they desired – the colour white.”(9)

Cutting a long story short, the stranger, Mbalawe, fell ill and died while Shihwago became pregnant. When she gave birth to a son the Sangu were delighted by his appearance. Quoting again: “…he was taken along… when we went to war in search of wealth: looking at him, the white one, the people saw that he led them like a Jemadar”, in other words like his father’s people. This was how the line of Mwahavanga (as the descendants of Mbalawe’s son were later known) was founded.

This history was told by a Sangu man from Msangaji and now living in Chimala, Ali Mashaka Ndelele. One of the most interesting features of his account is its emphasis upon Mbalawe’s colour. Mashaka later implied that the name Mbalawe can, in fact, be etymologically derived from the verb radical –vala (s), ‘to be white, shining, pure’.(10) If they would not volunteer, other informants at least confirmed the possibility of such a derivation. Considered in isolation these are tentative grounds on which to suggest that the interpretation of the hero’s name as an ethnic label might be turned on its head, and viewed rather as a reinterpretation of an originary sense.

The later manifestations of such themes and colour symbolism are, though, widely known. A comparative survey would have to take in not only the whole of south-western Tanzania, but large areas of East and Central Africa. Nyakyusa ideas about the white colour of their chiefs and the blackness of common folk are, for example, incorporated in much the same way into their histories of origin; much as they surface in the reports, still to be heard, that the arrival of white men was prophesised well before the Europeans eventually came. There is plenty of material here for reconsideration.(11)

Burton was told of Mwahavanga in 1858 that he was:

“a man of venerable aspect, tall, burly, and light-coloured, with large ears, and a hooked nose like a ‘maghrabi’. His sons… all resembled him, their comeliness contrasting strongly with the common clansmen, who are considered by their chiefs as slaves. A tradition derives the origin of this royal race from Madagascar or one of its adjoining islets.” (1859: 304)

Over a decade later Livingstone heard similarly that the Sangu were “a fair people, like Portuguese, and very friendly to strangers” (1874: I, 212), and learned that Mwahavanga’s descendants were “very light coloured, and have straight noses” (1874: II, 88). These statements mark a neat conjuncture between prevailing European notions, recalling the world of Rider Haggard’s “She” (1976 [1887]). This is to say nothing of any ideas the Arab conveyors of this knowledge may have had, and conceptions which have persisted in Sangu thinking through to the present day. They have a truth which spills over from ideology into practice. There is evidence to suggest that these ideas are played out in marriage strategies, notably in the selection of fair-skinned wives for chiefs; something I have observed myself.(12)

I can do no more here than point to the possibility of pursuing a line of enquiry which would parallel Marshall Sahlins’ recent work on the early history of the Sandwich Islands and the apotheosis of Captain Cook (esp. Sahlins 1981).(13) Interestingly, Feierman, suggesting that the myth of Mbegha may be a revision of the story of an earlier hero (Sheuta) triggered by political change, anticipates the analysis of what Sahlins refers to as “structures of the long run” and the dialectical engagement of structure and event (Feierman 1974: 66-69). The insight is lost, though, in the distinction Feierman draws between history and myth: the structuralist paradigm challenged by Sahlins. The same division surfaces and is developed further by Roy Willis in his latest book, and its price can be measured in his unconvincing treatment of Fipa ideas about the Twa (1981: esp. 29-35): ideas not so far removed, I might add, from those which Sangu entertain about their own rulers, avatwa (s), in the past. Although I will not engage Willis directly here, it should also become clear that I distance myself from his interpretation of settler and stranger as fixed values in the seemingly timeless constellations of Fipa cosmology.

Legitimacy inside-out

The thrust of Sahlins’ argument is to show not only that cultural structures can be reproduced in events, but also to demonstrate how their engagement may result in both cultural and practical transformation. The skeleton of such an account is, I think, contained in Mashaka’s history. It opens by describing the aboriginal Sangu polity in terms of the interrelation of three (kindred) groups, known in short by the names Mgawa, Mhami and Mswaya. The first of these, the primary group, was settled at Ilamba and in the remote area north of the river Ruaha known as Unyamande, where they had lived from time immemorial, “multiplying like cattle”. The second, Mhami, was an offshoot of the first sent out to settle in the well-watered area of Madundani and Utengule, a staging point on the way to Uwanji: while the third, Mswaya, had been sent to Makondo in the west, on the path to Usafwa. Among themselves they would choose someone to look after their ritual affairs, filling the office of ‘Njali’. This title is said to be cognate with the word shali (s), ‘baby’ or ‘small child’, signalling the requirement that an incumbent must be wholly provided for by the people and so exempt from subsistence activity of any kind. Quoting Mashaka: “Had he worked he would have been in no position to look after our customs. Likewise in later times we have an elder (shehe) at the mosque and a padre in the church: they had their own equivalent of a padre or Muslim elder.” So the Sangu began to travel beyond the boundaries of Usangu to ensure Njali’s provision. This pattern was changed radically upon the accession of Mbalawe’s son, an event which offered a new blueprint for legitimation. The friction which was generated between old practice and new exploded upon Mwahavanga’s death (Mwahavanga being, in Mashaka’s version, alternatively Mbalawe’s son, or son’s son). When Mwahavanga’s favoured heir, a son, died before he could take office, his other sons found themselves in something of a quandary. To resolve the problem of the succession they turned to the precedent set by the Mbalawe episode. They appointed Tovelamahamba, the son of their sister, another Shihwago, and of an outsider, Nshilyama. Before long, however, jealous of Tovelamahamba’s position, they began to accuse him of failing to respect the distinction between sacred and secular power. Tovelamahamba responded by attacking his uncles, forcing them into exile and killing one, Gambali. In Mashaka’s words:

“At this point his mother [Shihwago] protested: ‘A-a-a – you’re killing my kinsmen: stop it! Why have you done this?’’ And he replied: ‘Haa – avija shi? Shene vaxamelele va vene!’ [s], ‘Ha – what did they say? Didn’t they give it [the chiefship] to me themselves!’… And in this way the name Merere began.”(14)

Reiterating this transformation in the use of chiefly power, the history continues: “Merere began to assert his authority, and his name became widely known. Now he could go to Uhehe and make war there”. And so onto the most familiar episodes in Sangu history.

This is a very compelling account. I should emphasise, though, that there are motives at work in this narrative other than the demands of consistency and coherence. These include both a sociological and a more personal interest on Mashaka’s part. If followed upon discussion and explanation of a system of titles (used, for example, in greetings), some of which are ascribed on the basis of affinal links of the kind which play an important part in this history: in particular through marriage to a woman of either royal line (Merere’s or Mwahavanga’s), to a woman with the title ‘Shihwago’. In certain respects the history was an appendix to this explanation; an extended illustration. Secondly, it functioned as an elaborate charter for the name Mgawa, posited as belonging to the primary and original Sangu group. In later history, as Mashaka made clear, all but one of the Sangu chiefs has been a descendant of another bearer of this name, Twanuxa Sinkunja, a wife of the first Merere. It is through his mother, also a Sinkunja, that Mashaka inherits the names Mhami and Mgawa. No other history I heard made similar claims about the early history of either name, though the Vaswaya are often accorded aboriginal status. And, as if to underline the provisionality of his version, no sooner had my interview with Mashaka come to an end than I was hailed by a man who volunteered an entirely different account. This, moreover, incorporated elements which Mashaka had just branded as betraying an insidious Hehe influence; clearest in tales that had Mbalawe coming down from Uwanji to Usangu as a hunter.(15)

Mashaka’s history does, nonetheless, highlight the issue of legitimacy in a way which is relevant to an understanding of the more generalised contemporary articulation of the origin story. Here it may be best to begin by sketching a more conventional historical account.

It seems certain that relations with coastal traders in the nineteenth century were essential in maintaining the power of the Sangu chiefs (if not instrumental in their creation). External trade provided the means by which capital, including land and labour, could be accumulated: whether by force or by dint of repute. These outside links were by no means constant or assured. They were, for example, subject to competition between chiefs and others in search of power. Severe difficulties of this kind appear to have been generated by Merere’s accession. The Arab alliance blew hot and cold. Relations reached a low point when the interests of the Bungu chief Kilanga and the dispossessed descendants of Mwahavanga coalesced, and did not improve until after the war which saw the deaths of both Kilanga and the trader Amran Masudi (in 1873: Shorter 1974: 10-14). Thereafter Merere was able to retain external support, despite constant Hehe pressure. Among his allies was the Baluchi Jemadar who joined Merere during his exile in Kiwele (c.1875), and stayed at the Sangu court for the next twenty years or so. He was given a daughter of the chief in marriage, and incorporated in a way which helps to explain Mashaka’s allusion (see above) and the fact that he is the only “Arab” from this period whose name, rendered “Nyamadali”, is still remembered by the Sangu.(16) In one sense, at least, his was an isolated case, not to be repeated.

With the rapid growth of European interests, the Arab connections were doomed to decline. At first Merere and his successor used their new allies to some advantage. But, as the Germans extended and consolidated their power, the Sangu rulers’ capacity to exercise similar authority slipped from their grasp.(17) Relations between successive chiefs and the colonial administrations, German then British, degenerated into mutual mistrust. They acquired an air of fatal necessity. The Sangu chiefs paid for legitimation by losing their earlier potential for domination. As swiftly as people had once been recruited to Merere’s cause, so they could now just as easily be hived off. And so they were, through a combination of internal and external pressures. By the early 1950s and the accession of Alfeo, great-grandson of the first Merere, chiefly power was limited to an ever-diminishing area, focusing upon the district around Utengule.

The dilemma which faced many Tanganyikan chiefs after the second world war, as nationalism grew and momentum gathered towards independence, is well documented.(18) The chief of Usangu, poised between conflicting sources of legitimation, was no exception to the general rule. In a way this was their permanent condition, with a history extending well before and after this one critical historical juncture. When his father, Mxanuwoga Merere, died at the end of 1950, Alfeo was still at school at Malangali. The Mbeya authorities seized upon this golden opportunity to groom a more compliant chief and packed Alfeo off for further schooling first at Tabora, then in Mwanza (ARPC 1953: 123). His father’s brother, Myotishuma, was appointed regent. When Alfeo returned to Utengule in 1953 to take up the royal stool the odds were already stacked heavily against him. His commitment to tradition was questioned by the elder generation, while his peers resented his sudden elevation to power. A group of Muslim youths from Mswiswi, a few miles south-west of Utengule, were particularly vocal in their opposition. One instance of their defiance can be singled out here. Much as they would not dismount from their bicycles and stoop to him in greeting, they refused to greet Alfeo with his proper title: “Aje njali!” (s). Instead they treated him to the Muslim salutation: “Sabalkheri!” (“Good morning!”): a calculated insult. There was sweet irony in this, and not only because Alfeo had just returned from Mwanza with a Muslim wife. The history of Mbalawe was conjured up with a vengeance, and the taunt, Merere the Arab, has stuck.

Many of the traditional practices and rituals of the Sangu chiefship fell into abeyance in the 1950s. Had Alfeo opted to project the traditional props of his position, this would have done him little good in the long run. His Muslim critics subsequently became prominent in the local organisation of TANU. And when TANU eventually came to power and the office of chief was erased from the ordinances in 1962-63, the traditional role was the only one which the new politics would allow for chiefs who were not recruited into the new government. To have accepted this definition would have been tantamount to political castration.

The history of Alfeo’s relations with government and people since independence has been correspondingly complex, impossible to summarise at all adequately here. The spread of the Christian, particularly Roman Catholic, church in the last two decades has completed Mashaka’s triple identification: chief, Muslim elder and priest, and adding a new focus for opposition. In 1981 the most powerful village politician in Utengule, the CCM (formerly TANU) Branch Secretary, was a Roman Catholic whose mother had led the royal rituals at Alfeo’s before her conversion. For the Christians in the community the taunt of the Muslims can be used more directly, echoing the words of the first German missionaries when they dubbed Merere an Arab.

With few exceptions the history, Mbalawe’s, which underpins these statements is not elaborated. Mashaka, as we have seen, provides one example. Others include the two printed accounts discussed at the beginning of the paper. Both are by members of split-off Muslim branches of Merere’s family; one of which provided Utengule with its elected Village Chairman through to early 1982. In general, though, the origin history is condensed into a simple binary formula. The insinuation that Merere is descended from an Arab, or other “white” outsider, is complemented by the claim that Sangu, real Sangu, are “black” and can trace their origin to the area known as Unyamande, in the remotest part of the plains. Its inaccessibility and isolation from the centres of modern administration, as well as from Alfeo’s retreat (1981) at Luhanga, just north of Utengule, provides another opportunity to emphasise the identification of both traditional and modern forms of government as external and foreign. Thus is Mashaka’s intricate reconstruction of the early history of Mgawa, Mhami and Mswaya distilled. This has its own reflex in practice, playing an important part in determining the orientation of the dead, many of whom are now buried looking towards Unyamande, as the putative origin of anyone with a good claim to be Sangu. Relatively few people, though, choose to trace (or care to recall) these connections in any great detail. For a very good reason. Many, after all, have genealogies which prove them to be as much outsiders as they would like to think Alfeo. And many are the descendants of men, women and children recruited by his predecessors more than seventy-five years ago.


Legitimacy twists and turns, blurring the distinction between myth and history, questioning a divide which values oral discourse over writing. Let me conclude, then. with an appropriate thought and by bringing discussion full circle and back to the origin of Mbalawe. I should perhaps have said earlier that there are reverberations of this detail further afield. In August 1981 Juma Mwamlima, a Nyiha chief, was claiming that “his grandfather told him that the most remote ancestors [of his line] had come from Balawi in Somalia”.(19) Chiefs can also, it seems, do research in District Books. Nothing surprises. It is Feierman who tells us that in about 1836 some Zigula, southern neighbours of the Shambaa, sold themselves as slaves to escape famine and “were taken to Somalia by Barawa traders” (1974: 137). Perhaps there is room for literal interpretation yet.(20)

Eighteen months’ fieldwork in Mbeya region, Tanzania, 1980-82, was funded by the SSRC, London, and the Smuts Fund, Cambridge, and conducted under the auspices of the Tanzania National Scientific Research Council, Dar es Salaam.


(1) This history was recorded in keeping with general guidelines for the compilation of District Books and in conjunction with a temporary thaw in relations between the District authorities in Mbeya and the then Sangu paramount, Mtenjela Merere (ARPC 1929: 56). I am grateful to John Iliffe for first introducing and making a copy of this history available to me, which I have since been able to check against microfilms in Rhodes House, Oxford, and in the Tanzania National Archives, Dar es Salaam.

(2) For a useful discussion of “The ‘Hamitic Myth’ and its Legacy” see Miller 1976: 4-11.

(3) I have adopted a transcription which is largely consistent with that employed by Bilodeau (1979), except that long vowels are generally not marked. Sangu, shisango, words and strings are marked ‘s’ in the text: Swahili terms are in italics only.

(4) ‘Arab’ is used in the customary (for the East African interior) loose way throughout this paper.

(5) Shaibu Sixapombe Merere interviewed on 18 April 1968, in Ndikwege 1968: 5/68/1/3/3.

(6) Salehe Mhanginonya Merere, undated interview in Ndikwege 1968: 5/68/1/3/10, 1-2.

(7) In addition to those I heard in the field, Alison Redmayne has kindly allowed me to read unpublished field notes containing a number of valuable accounts recorded by her on field trips to Usangu in 1966 and 1968.

(8) The name Shihwago is a derivative of the verb stem –hwaga (s), ‘to drive cattle’.

(9) Extract and translation from a tape-recorded interview in Swahili: Utengule, 3 October 1981.

(10) Compare mbalafu (s), ‘person with white (fair) skin’. Normally (when the name is pronounced in isolation) the stress falls upon the penultimate syllable of ‘Mbalawe’: one elderly informant differed by stressing the final syllable.

(11) See Monica Wilson 1959: esp. 1, 12-13, 153, 157; and for the related Ngonde tradition which makes an Arabic connection, Godfrey Wilson 1939: 10. For a version of the Nyakyusa prophecy, Thurnwald 1935: 328-329. Of elsewhere in the region, Musso 1968: esp. 1-24, on Hehe origins is the most fascinating document from this perspective. For reflections of the same in Zambia: Cunnison 1961: 62 (n.2). Note that I have omitted discussion of the third colour in Victor Turner’s (1966) classificatory triad: red. This is not for lack of evidence. Analysis of the interplay of red and white as the ascriptive colour of Sangu chiefs might begin with a reading of some of the tales recorded by Bilodeau (1979).

(12) For a reference to this practice among the Bena: Makwetta 1968: 4/68/1/1,6, information of Mzee Tatayila.

(13) For an even more direct parallel see the story of the apotheosis of Jumah Mfumbi, first Arab to enter Ugogo (Burton 1860: I, 302-303).

(14) i.e. –melele (s), from the verb stem –pela (s), ‘to give’.

(15) William Garland tells me that members of the Wanji chiefly line of Nyambo (at Mpangala) are reputed to be tall and fair and claim an origin from the east of Lake Victoria; while in general Wanji dismiss the Sangu story of origin as mistaken. I reserve detailed consideration of the Wanji element in many (but not all) versions of Mbalawe’s history for a future occasion. For the Hehe story see Redmayne 1964: 107-112.

(16) For the Jemadar see, for example, Elton 1879: 337, 344; Merensky 1892: 95-99; Abakari 1901: 91; Kootz-Kretschmer 1929: 279-283; and Ndikwege 1968: 3, 6-7.

(17) The best summary of these events can be found in Wright 1968.

(18) See, for example, Abrahams 1981: 28-33.

(19) Mwamlima was interviewed by John Gay (in the course of a RIDEP survey) on 20 August 1981. I am very grateful to him for the loan of notes from which the above statement is taken. Another Nyiha chief, Gilbert Nzowa, has claimed a Kenyan origin for both his and Mwamlima’s line (Slater 1976: 50-51). Brock (1968: 65-66) only notes that both lines independently claim to have come from (not via) Ugogo: while Knight (1974: 32-33) was, like John Gay, told of a Mwamlima – Merere link. Ali Mashaka rejected this last and similar claims as genealogical fictions.

(20) Derek Nurse adds the comment that there are two places on the East African coast with names which can be construed as “Barawa”: the Somali town of Brava (some of whose inhabitants are among the northernmost speakers of Swahili) and a village on the Kenyan coast presently settled by Boni. On the Tanzanian coast, particularly around Dar es Salaam, people claiming Shirazi descent specify a “Barawa” origin more frequently than they do other old Swahili settlements to the north.


Abakari, Selim bin, 1901. ‘Safari yangu ya Nyassa waqati bwana Major von Wissman alipopeleka stima yake katika juto la Nyassa’, pp.50-105 in Safari za Waswahili, ed. C. Velten. Gottingen: Vandenhoeck and Ruprecht.

Abrahams, R. G. 1981. The Nyamwezi Today: A Tanzanian People in the 1970s. Cambridge: Cambridge University Press.

Bilodeau, J. 1979. Sept Contes Sangu dans leur contexte culturel et linguistique. Elements de Phonologie du Sangu, langue bantoue de Tanzanie. Textes des contes avec traduction et notes, Thèse de Doctorat de troisième cycle, Université de la Sorbonne Nouvelle, Paris.

Brock, B. 1968. ‘The Nyiha (of Mbozi)’, pp.59-81 in Tanzania before 1900, ed. Andrew Roberts. Dar es Salaam: East African Publishing House.

Burton, R. F. 1859. ‘The Lake Regions of Central Equatorial Africa’, pp.1-454 in Journal of the Royal Geographical Society, XXIX.

Burton, R. F. 1860. The Lake Regions of Central Africa: A Picture of Exploration (2 vols.). London: Longman, Green, Longman and Roberts.

Cunnison, Ian, 1961. ‘Kazembe and the Portugese 1798-1832’, pp.61-76 in Journal of African History, II, 1.

Elton, J. F. 1879. Travels and Researches among the Lakes and Mountains of Eastern and Central Africa, ed. H. B. Cotterill. London: John Murray.

Feierman, Steven, 1974. The Shambaa Kingdom: A History. Wisconsin: The University of Wisconsin Press.

Knight, C. Gregory, 1974. Ecology and Change: Rural Modernization in an African Community. New York / London: Academic Press.

Kootz-Kretschmer, E. 1929. Die Safwa: Ein Ostafrikanischer Volksstamm in seinem Leben und Denken (Vol. II). Berlin: Dietrich Reimer.

Livingstone, David, 1874. The Last Journals of David Livingstone, in Central Africa, from 1865 to his death (2 vols.), ed. Horace Waller. London: John Murray.

Makwetta, J. M. 1968. ‘Maji Maji in Ubena’, Maji Maji Research Project No. 4/68/1/1, in Maji Maji Research Project 1968 Collected Papers, ed. John Iliffe, Department of History, University College, Dar es Salaam, March 1969.

Merensky, A. 1892. ‘Von den Kondebergen nach Usanga’, pp.91-102 in Mitteilungen der Geographischen Gesellschaft (für Thüringen) zu Jena, XI.

Miller, Joseph C. 1976. Kings and Kinsmen: Early Mbundu States in Angola. Oxford: Clarendon Press.

Musso, Michael, 1968. Mukwava na kabila lake. Dar es Salaam / Arusha: Eastern Africa Publications.

Ndikwege, M. S. 1968. ‘Chief Merere and the Maji Maji Uprising’, Maji Maji Research Project No. 5/68/1/1, in Maji Maji Research Project 1968 Collected Papers, ed. John Iliffe, Department of History, University College, Dar es Salaam, March 1969.

Redmayne, Alison H. 1964. The Wahehe People of Tanganyika, D.Phil. thesis, Oxford University.

Rider Haggard, H. 1976 [1887]. She. London: Coronet Books / Hodder and Stoughton.

Sahlins, Marshall, 1981. Historical Metaphors and Mythical Realities: Structure in the Early History of the Sandwich Islands Kingdom. Ann Arbor: The University of Michigan Press.

Shorter, Aylward, 1972. Chiefship in Western Tanzania: A Political History of the Kimbu. Oxford: Clarendon Press.

Shorter, Aylward, 1974. ‘The Rise and Decline of Bungu Power: A Forgotten Episode in the History of Nineteenth Century Tanzania’, pp.1-18 in Tanzania Notes and Records, No.73.

Slater, Mariam K. 1976. African Odyssey: An Anthropological Adventure. New York: Anchor Press / Doubleday.

Thurnwald, Richard C. 1935. Black and White in East Africa: The Fabric of a New Civilization. A Study in Social Contact and Adaptation of Life in East Africa. London: George Routledge and Sons.

Turner, Victor W. 1966. ‘Colour Classification in Ndembu Ritual: A Problem in Primitive Classification’, pp.47-84 in Anthropological Approaches to the Study of Religion, ed. Michael Banton. London: Tavistock Publications.

Willis, Roy G. 1981. A State in the Making: Myth, History and Social Transformation in Pre-colonial Ufipa. Bloomington: Indiana University Press.

Wilson, Godfrey, 1939. The Constitution of Ngonde (The Rhodes Livingstone Papers No.3). Livingstone: The Rhodes-Livingstone Institute.

Wilson, Monica, 1959. Communal Rituals of the Nyakyusa. London: Oxford University Press / International African Institute.

Winans, Edgar V. 1962. Shambala: The Constitution of a Traditional State. London: Routledge and Kegan Paul.

Wright, Marcia, 1968. ‘Chief Merere and the Germans’, pp.41-49 in Tanzania Notes and Records, No.69.

Serial Publications

ARPC: Tanganyika Territory. Annual Reports of the Provincial Commissioners on Native Administration. Dar es Salaam: Government Printer.

PA: Periodical Accounts Relating to the Foreign Missions of the Church of the United Brethren Among the Heathen. London.