KINGS AND

KINSMEN
EARLY MBUNDU STATES
IN ANGOLA

JO S E P H C. M IL L E R

CLARENDON PRESS • OXFORD

1976
 Oxford University Press, Ely House, London W. 1
GLASGOW NEW YORK TORONTO MELBOURNE W ELLINGTON
C A P E T O W N IBADAN NAIROBI DAR ES SALAAM LUSAKA ADDIS ADADA
DELHI BOMBAY CALCUTTA MADRAS KARACHI LAHORE DACCA
KUALA LUM PUR SINGAPORE HONG KONG TOKYO

isbn 0 19 822704 3
© Oxford University Press 1976

A ll rights reserved. No part o f this publication may be reproduced,

stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording or otherwise, without
the prior permission o f Oxford University Press

WITHDRAWN
Unlvirilty Q'i

Printed in Great Britain

by Cox & Wyman Ltd
London, Fakenham and Reading
In memory of my father, John W. Miller
1903-1974
 Preface

A l t h o u g h historians are accustomed to work with imperfectly

recorded information, imperfect at least for their individual purposes,
the data which support this study have special qualities which
require a preliminary statement of the ways in which they were
collected and the techniques used to analyse them. Professor Jan
Vansina’s pioneering studies of the historical meaning of non-
literary evidence necessarily inform any study which purports to
combine diverse data drawn from ethnographic, linguistic, docu­
mentary, and orally transmitted materials. Since our knowledge of
early state-formation among the Mbundu depends on all these
types of source, and especially since some of the data were recorded
long before any of these disciplines reached their present level of
sophistication, the problem of methodology assumes more import­
ance than it ordinarily might.
The data collected during my 1969-70 fieldwork among the
Imbangala of Angola require the most extensive discussion. These
consist of approximately thirty hours of tape-recorded interviews in
Kimbundu and Portuguese; in addition, there is a somewhat smaller
amount of written English translation of the Kimbundu texts. A
good many more hours of interviews were taken down in rough
hand-written notes and expanded in the form of typed working
notes. All these materials are in my possession for the present and
are, of course, available to anyone wishing to use them for scholarly
purposes. I hope that in the near future copies may be deposited in
suitable places in Africa and in the oral archives at Indiana University.
I tape-recorded the initial interviews in full with the intent of
preparing transcriptions and translations according to the standards
outlined by Professor Philip D. Curtin,1 but local conditions soon
made it more useful to abandon the tape recorder and continue with
written fieldnotes. No translators capable of producing accurate—or
even coherent—Portuguese translations of the Kimbundu texts
were available to me. This forced me to interview as much as possible
in Portuguese. Although a number of interviews were recorded in
Portuguese, the problems of telling Imbangala histories in a foreign
1 Philip D . Curtin (1968b).
viii PREFACE
language in a totally artificial context almost eliminated the subtle
qualities which the tape recorder is best adapted to capture.
The tape-recorded interview format also became inefficient owing
to a shift in the content covered during later conversations with
informants. Most later testimony consisted of genealogies and
informal discussions of linguistic and ethnographic information
which could be preserved as accurately and efficiently through the
use of written notes as with a tape recorder. Although a few narrative
compositions worth recording appeared in the course of later
interviews (attempts to repeat these for taping generally failed),
relatively little such material was left unrecorded. The use of written
notes, on the other hand, had the advantage of allowing more
penetrating investigation than tapes would have permitted of several
points felt to be too sensitive for discussion on tape.
Limitations on time and money, as well as certain restrictions on
research activities in Portugal and Angola, restricted my opportunity
to execute the theoretically optimal research design. It is obvious
that historians of Africa will fully utilize available source-materials
only when they approach their data with a facile command of both
the languages and the cultures involved. At that time it was difficult
for a non-resident historian to acquire the necessary sophistication in
these fields. The African languages of Angola have not been ade­
quately studied, and although one could gain a working knowledge
of a form of Kimbundu in Lisbon, a truly useful familiarity with the
language demands study in the field. Angolan ethnography suffers
from a similar neglect and imposes similar constraints on the
researcher. The historian ideally should precede his historical
studies with lengthy language training and ethnographic research,
but the unavailability of relevant materials outside of Angola makes
this impossible. Conditions in Angola made it impossible to plan
an orderly research project extending over several months or years.
I therefore found it more efficient to gather as much information as
possible in a short time and elected to bypass preliminary rigorous
ethnographic and linguistic groundwork in favour of recording the
most accessible information as quickly as was feasible. This strategy
guided my research over the five months I spent living near the
Imbangala in the District of Malanje.
I first located the individuals generally regarded as most likely to
provide comprehensive and accurate information on the Imbangala
past. It turned out—fortunately under the prevailing limitations on
the research design—that most surviving Imbangala historical
 PREFACE IX

traditions were known to only a handful of individuals, all of whom

spoke some Portuguese. These men, the ndala kandumbu (official
court historian of the former state of Kasanje) and the balca a
musendo (unofficial but professional ‘historians’) became the primary
informants for this study. It became clear that most other potential
sources of information, who may be termed secondary informants,
could add little to the data obtainable from the primary informants.
The presence of a secondary informant at an interview, however,
often stimulated a primary informant (ostensibly acting as inter­
preter) to recall information which he did not otherwise bring readily
to mind.
Most interviews opened with a voluntary statement by the infor­
mant in which he gave his personal version of the history of his title
and/or lineage. In the case of secondary informants, these statements,
often expressed with no particular skill or artistry, tended to be very
brief and incomplete. Some secondary informants chose to omit
this phase of the interview. The primary informant, who usually
accompanied me to interviews with secondary informants, followed
up the initial statement with questions designed to prod the secondary
informant to elaborate or to resolve internal contradictions. The
accompanying primary informant then offered his own version of the
same history under the guise of prompting the secondary informant.
Most interviews concluded with my inquiries about confusing points,
contradictions which I had noticed, and new concepts and terms
that had arisen in the course of the interview.
The initial interviews with primary informants followed approx­
imately the same format, but these led to a series of subsequent
meetings which ranged much more widely according to no particular
pattern. I generally opened the later sessions with a historical point
drawn from a previous interview and asked the informant to repeat
or to expand on it. The discussion usually moved quickly through
more questions and answers into general ethnographic and linguistic
problems as the informant attempted to clarify obscure points. The
often repetitive question-and-answer format of these later interviews
made the tape recorder superfluous for the most part.
Although ethnographic inquiries might better have been based on
the participant-observer method developed by anthropologists,
obstacles facing white foreign researchers in Angola made it impos­
sible to live in a village. Some ethnographic information in this study
comes from direct questioning on points which seemed relevant in
the light of the formal historical material. A second method of
X PREFACE
investigation was to ask questions based on a list of Kimbundu terms
taken from written sources on the Imbangala dating back to the
sixteenth century. I merely presented each word to the informants
and asked if they knew it and, if so, what it meant. This technique
opened up several new and fruitful lines of inquiry, often in quite
unexpected places. Other conclusions about Imbangala social
structure and cosmology emerged' from later analysis of both the
formal histories and other data. The hjstprian must beg the patience
of his anthropologist colleagues for the lack of systematic ethno­
graphic investigation. The basic linguistic research consisted of an
attempt to collect 200 word vocabularies in all eastern Kimbundu
dialects based on the list which linguists have used in connection
with their studies on glottochronology.2 Linguistic insights gained
from the general ethnographic investigation and the several available
dictionaries on the Bantu languages of Angola supplemented these
lists.
Written documentation on sixteenth- and seventeenth-century
Angola is now scattered over at least three continents. By far the
most important collection is housed in the various archives in
Lisbon, much of this in the manuscript collections of the Biblioteca
Nacional, the Arquivo Nacional da Torre do Tombo, and the
Biblioteca da Ajuda. Foreign scholars concerned with African topics
have on occasion found access to some of these documents restricted.3
I requested but never received authorization to consult the collection
in the Biblioteca Nacional, while reorganizations reported to be
under way in the Biblioteca da Ajuda and the Torre do Tombo
prevented me from seeing more than a few of the relevant materials
there. I have therefore depended on published versions of most
documents in Lisbon except those located in the Arquivo Histórico
Ultramarino. Fortunately, all known documents from the sixteenth
century and before, as well as a great deal of documentation from the
seventeenth century, have been published, the more important ones
often in several places. Of the many documents scattered through
Italy, Spain, England, France, and Brazil, to name only the most
important repositories, it was possible to consult personally only
those in the British Museum. The barely explored riches of the
Arquivo Histórico de Angola, which were readily accessible, relate
entirely to later periods.
Given the limitations created by the conditions just outlined, I
2 D . H. Hymes (1960), p. 6.
3 For the experiences o f another foreigner, V. Salvadorini (1969), pp. 16-19.
 PREFACE xi

have attempted to steer a course between the Scylla of acknowledging

the imperfections of the data to the extent of saying nothing at all
and the Charybdis of striving for a coherent reconstruction of events
that pushes the data beyond their inherent limitations. This dilemma
has a special relevance to the oral materials, since they are less well
known than the written ones and since the analysis depends on them
at several points. I have therefore devoted a section of Chapter I
to a description of the formal and informal histories of the Imbangala
and to an analysis of their meaning for Western historians. At this
point, I need only explain the considerations which led me to use the
traditions as I did. These traditions may be susceptible to study on
several other levels—notably, the intriguing possibility of a formal
literary criticism now being developed by Professor Harold Scheub
of the University of Wisconsin—but the constraints operative in this
case made it necessary to work out a relatively low-level ‘common
denominator’ which made the traditions I collected in 1969 compar­
able to other variants written down in the seventeenth, nineteenth,
and early twentieth centuries.
The method chosen, given the atrophying condition of the tradi­
tions, could not depend on comparing a great number of variants
which no longer exist. Nor could it demand a degree of linguistic
facility not obtainable in this case. It was also necessary to recognize
that all published traditions exhibited drastic changes from what
must have been their original oral form. The versions of these same
traditions which I collected suffer from similar (but less extensive)
mutilation since I was unable to secure accurate transcriptions or
translations of the Kimbundu texts. Even the versions recorded in
Portuguese must have undergone considerable modification as the
informant translated them for my benefit. As Chapter I, I hope,
makes clear, even the condition of reciting these histories for an
alien investigator inevitably influenced the way they were told.
No analysis dependent on a word-for-word explication of the text
could be employed; the words themselves had either not been
recorded or could not be understood sufficiently to justify such an
approach.
The research which led to the present study began as a rather
unfocused inquiry into the history of Kasanje, the Imbangala king­
dom founded at the close of the period finally chosen as the subject
of this work. The shift in period and subject resulted from the fact
that existing studies provided no useful background on which to base
a history of Kasanje. The literature generally confused the Imbangala
xii PR E F A C E
with the so-called ‘Jaga’, famous and fearsome warriors who prob­
ably never existed in the way in which they have been described.
Further, no systematic studies on the ethnography of the Mbundu
generally, and much less that of the Imbangala, had been published.
In short, it was impossible to describe the milieu in which the
Imbangala founded ICasanje or even to know who the Imbangala
had been in the seventeenth century. Without some understanding
of the conditions which the Imbangala encountered when they
reached their present home and fairly certain knowledge of the
tendencies they brought with them, it seemed to me, one could not
expect to write a meaningful history of the kingdom.
The present focus on processes of state-formation emerged from
my realization that little of what I had read about Jhe emergence of
states spoke to what I felt to be the central issue in the minds of the
Imbangala with whom I had spoken: the tension, between their
loyalties to j d n and their respect for Icings. An initial survey of the
literature on the subject convinced me that the experience of the
Mbundu might highlight an aspect of African political and social
history which had gone relatively unnoticed. It was with considerable
pleasure that I discovered, in the course of writing the study, that
historians and anthropologists in Nigeria, such astAbdullahi Smith,
Robin Horton) and E. J. Alagoa, had laid out both theory and facts
in a maimer which' at once clarified my own thinking and explained
certain lingering uncertainties which obscured my view of the
Mbundu past. I hope that they will agree that Mbundu history is
susceptible to the sort of analysis which they have pioneered.

In the absence of an ofiicially standardized orthography for Kim-

bundu, I have attempted to employ a version of the ‘practical ortho­
graphy of African languages’ simplified to conform to the capabilities
of a standard English alphabet.4 The major divergence from the
system suggested by the International Institute of African Languages
and Cultures and the one used here is my substitution of the French
‘j ’ to indicate the sound represented by -s- in the English word
pleasure. This should lead to no ambiguity as the English ‘j ’ does
not occur in ICimbundu. The spellings are phonemic rather than
phonetic and so, for example, the Mbangala/Colcwe/Lunda pro­
nunciation cl- appears here in the general Kimbundu form ki-.
The references in the footnotes are given in abbreviated form, but
the bibliography contains the customary complete citations that will,
4 International Institute o f African Languages and Cultures (1930).
 PREFACE xiii
I hope, clarify some of the obscure references found in many older
publications on the subject. Many of the documents used have been
published in various places by different editors, and I have tried to
include all locations known to me in these cases. In alphabetizing
Portuguese names, I have followed the U.S. Library of Congress
rules, although most readers will be aware that individual libraries
(especially those outside the United States) may follow different
conventions. Citations of a ‘testimony’ of an individual refer in
every case to interviews conducted during my fieldwork in Angola
(see Bibliography for a list of informants).
I have followed the accepted English convention in using no Bantu
prefixes before the names of ethno-linguistic groups, with the
exception of the name Ovimbundu (properly, the Mbundu) to
distinguish the inhabitants of the southern highlands from the
Mbundu who live north of the Kwanza. The plural of most Bantu
words appears in parentheses following its first occurrence in the
text, and a glossary of Bantu terms located at the end of the study
should facilitate the identification of terms unfamiliar to the reader.

It is a pleasure to thank, all too briefly, some of the persons and

institutions who have contributed to the completion of this study.
Dr. David Birmingham and his family made us welcome in London
and contributed to both the pleasure and the profit of our stay there.
Mme Marie-Louise Bastin Ramos introduced me to some of the
materials to be found in the Musée Royal de l’Afrique Centrale in
Tervuren, Belgium. In Lisbon, I add my gratitude to the thanks of
many others who have benefited from the friendly advice and
willing assistance of Professor Doutor Antônio da Silva Rego. I
also wish to express my appreciation to Dr. Alberto Iria, director
of the Arquivo Histórico Ultramarino, and to the staff of that
archive and the Biblioteca da Sociedade de Geografia de Lisboa who
assisted me in examining the rich documentary collections in Lisbon.
Dr. Dauril Alden more than generously shared the fruits of his
explorations in both archives and byways with me and certainly
contributed to the education of a novice historian in a new country.
Mrs. Asta Rose J. Alcaide, Mr. and Mrs. Robert Bently, and Mr.
and Mrs. Grayson Tennison also contributed to the pursuance of
my research while in Lisbon.
His Excellency Dr. Carlos Garcia de Azevedo, governor of the
District of Malanje in Angola, and Sr. Administrador José Manuel
Fernandes de Mota Torres granted the necessary permission to
xiv PREFACE
perform fieldwork in the Concelho do Quela. Sr. Engenheiro Agrón-
imo Manuel António Correia de Pinho, director of the Instituto de
Algodão de Angola kindly made available housing facilities owned
by the Instituto, without which I could not have worked successfully
among the Imbangala. In addition to the professional debt which I
owe to the padres and lay brothers of the Missão Católica dos
Bângalas for the work completed at the mission, my family and I
can never repay the personal kindnesses and cordial hospitality of
Padre Alfredo Beltrán de Otalora, Padre José Luis Rodrigues Sáez,
Esteban Arribas, Carmelo Ortega, and José Luis Martinez. Others
who graciously contributed to the success of my field research
include Sr. Adjunto e Sra. Vitor António dos Santos, Sr. Admini­
strador e Sra. Amándio Eduardo Correia Ramos, Sr. Administrador
e Sra. Adelino Correia da Silva, Sr. Administrador e Sra. Frederico
de Mello Garcés, Sr. Administrador Sigurd von Viller Salazar, Sr.
Alberto Manuel Pires, Padre Pedro Uria, Mr. and Mrs. Lloyd
Schaad, and many citizens of Quela. Of course, my work owes its
greatest debt to the Imbangala whose names appear in footnotes
throughout the book and whose expert historical knowledge provides
the basis on which every other aspect of this study has been built.
A special expression of gratitude goes to Sr. Sousa Calunga and Sr.
Apolo de Matos who came, I hope, to accept me as a fellow pro­
fessional.
In Luanda, the late Sr. Engenheiro Agrónimo Virgílio Cannas
Martins graciously put the facilities of the Instituto de Investigação
Científica de Angola at my disposal. Dra. Maria Angelina Teixeira
Coelho patiently helped me pick my way through the enormous
documentation of the Arquivo Histórico de Angola and provided
guidance and assistance in many other ways; her generous help and
advice opened many new avenues of research. Sr. Arquiteto Fer­
nando Batalha and Dr. José Redinha graciously shared the fruits
of their unequalled familiarity with the Angolan past. The late Dr.
Mário Milheiros gave freely of his time to help me become acquainted
with the archives of Luanda. I wish also to thank Padre António
Custódio Gonçalves, director of the Arquivo da Câmara Eclesiástica,
and Vice-presidente Ramos do Amaral of the Câmara Municipal for
permission to consult the collections of the Biblioteca e Arquivo da
Câmara Eclesiástica and the Biblioteca da Câmara Municipal.
Coronel Altino A. P. de Magalhães introduced me to the Arquivo
Histórico do Quartel Geral. I owe a special debt to the personnel
all these archives, too numerous to name here, who willingly and
 PREFACE xv

efficiently located the many volumes of documents which I requested.

Adjunto Sá Pereira of the Missão de Inquéritos Agricolas de Angola
and Eng.ro Mendes da Costa of the Serviços Geográficos e Cadastrais
permitted me to examine useful documents under their care. Sr.
Acácio Videira of the Companhia de Diamantes de Angola intro­
duced me to informants whose knowledge contributed to the comple­
tion of my studies. I am also grateful for many reasons to Michael
Chapman, Dr. e Sra. Luis Polonah, Mr. and Mrs. Peter de Vos,
Mr. Richard Williams, and Mr. and Mrs. Lester Glad.
None who have studied under Professor Jan Vansina can fail to
benefit from the example and encouragement he provides his
students and colleagues. His vast experience has informed the re­
search and writing of the study from its inception; it is to him that I
owe the original choice of subject. Inevitably, one feels that—in the
end—the study amounts to little more than elaboration on an idea
he once mentioned in the context of something more grand. Profes­
sors Philip D. Curtin, John Smail, and Harold Scheub of the Univer­
sity of Wisconsin read earlier drafts and offered invaluable criticism.
Throughout my years in Madison, Professor Curtin, especially,
exerted a steady and disciplined influence on my thinking and
writing. A year’s visiting appointment at the University of Wisconsin
afforded the opportunity to tap the insight and sharp analytical
abilities of Professor Steven Feierman, who guided me toward
much of the theoretical material on state-formation. If others find
something of themselves in my work, my failure to acknowledge
their contribution explicitly does not stem from lack of gratitude.
None bears any discredit for whatever weaknesses may remain, and
I accept full responsibility for such as there may be.
The research could not have been undertaken at all without a
generous grant from the Foreign Area Fellowship Program of New
York which enabled me to travel to Europe and Africa and to spend
more time in the preparation of the original dissertation (University
of Wisconsin, 1972) than would otherwise have been possible. The
Program, however, has no connection with the conclusions expressed.
I am also grateful to the University of Virginia for a Wilson Gee
Institute research grant and additional funds which supported the
preparation of the revised manuscript. Glyn Hewson, Beth Roberts,
and Paul Zeigler each allowed me the benefit of their special skills.
I have appreciated the sharp eyes and good judgement of the
editors of the Clarendon Press in preparing the manuscript for
publication. Although the footnotes no longer conform to the
XVI PREFACE
style preferred in historical writing, I have accepted the more eco­
nomical form suggested by the editors in recognition of the present
realities of the publishing industry.
My greatest personal gratitude goes, of course, to Janet, my wife,
and to my children who gladly followed the sometimes winding
paths which led to the present form of this work. Janet has somehow
found time to give valuable editorial advice in addition to keeping
the family functioning, and she must receive a considerable share of
whatever credit the study may merit.
J.C.M.
London, February 1975
 C ontents

List of Maps, Figures, Tables xix

List of Abbreviations XX

IN T R O D U C T IO N 1
The ‘Hamitic Myth’ and its Legacy 4
The Methodology 11
The Perspective 28
THE SETTIN G 31
Physical Environment 31
Ethno-linguistic Subdivisions in the Sixteenth Century 37
Selective Review of Mbundu Social Structure 42
IN D IG E N O U S B E G IN N IN G S 55
The Coming of the Lunga 59
The Ngola as a Lineage Symbol 63
Incipient States based on the Lunga 70
Kingdoms based on the Ngola 73
Conclusions 86

NEW ID EA S FR O M THE SOUTH 89

The Kulembe 89
Expansion of Libolo 90
The Mbondo after the Decline of Libolo 106

TH E PROBLEM OF STA TE-FO R M A TIO N

AM ONG THE SEG M EN TA RY LIN EA G ES TO
THE EAST 112
Early Political History of the Songo 112
Growth of Centralized Institutions among the Lunda 114
The Spread of Lunda Political Titles to the West 128
Cokwe States based on the kinguri 137
The Decline of Descent Groups among the kinguri’s
Lunda 141
 CONTENTS
THE IM B A N G A LA K ILO M B O - A RA D IC A L
SO LU TIO N 151
Songo States based on the Lunda Titles 151
Origins of the Kilombo 161
Formation of the Imbangala 167
THE IM B A N G A LA AND TH E PO R TU G U ESE 176
First Contacts—Establishing the Pattern 177
Kulashingo’s Imbangala in Angola 194
The Mature Portuguese-Imbangala Alliance 201
The Imbangala South of the Kwanza 210
Conclusions 221

IN S T IT U T IO N A L IZ IN G PO LIT IC A L
IN N O V A TIO N 224
Kin and Non-Kin 225
The Kilombo as a Military Machine 232
Men and Non-Men 242
Disappearance of the Kilombo 251
CO N CLU SIO N S 265
Defining a ‘State’ 266
How and Why Mbundu States Were Formed 270
Conclusions about Mbundu History 279

Bibliography 285
Glossary of African Language Terms Used in the Text 300
Index 303
 List of Maps

i. Geography of the Mbundu and their Neighbors (Modem) 33

ii. Lunga-Authorities in the Baixa de Cassanje (Sixteenth
Century and Earlier) 71
m. Pre-Seventeenth Century Mbundu Subgroups and Political
Institutions 77
IV. Expansion of the Ngola a Kiluanje (Before 1560) 80
v. Kulembe and Libolo (c. XVth-XVIth?) 91
vi. Songo and Mbondo Early History 107
vn. Expansion of the Mbondo Kingdom of the Ndala ICisua and
Subsidiary States (c. Sixteenth Century?) I ll
viii. Dispersal of Lunda Political Institutions (before c. 1600) 141
IX. The Imbangala and the Portuguese (c. 1600-1650) 196
x. Imbangala South of the Kwana (Seventeenth Century) 214

L ist of Figures

i. Representative Musendo Genealogy 17

ii. Diagram of how Genealogies Combine to Describe a Kingdom 20
hi. Schematic Diagram of the Structure and Relationships of the
Mbundu Historical Genealogies 22
iv. Relationships of the Ngola-titles 85
v. Genealogy illustrating the Alleged Origins of Ancient Lunda
Political Titles 121

List o f Tables

I. Chronology of the Ngola a Kiluanje 85

 Abbreviations Used in Footnotes
and Bibliography

AA Arquivos cle Angola (Luanda)

A.C.E. Arquivo da Câmara Episcopal (Luanda)
ACU Annaes do Concelho Ultramarino (parte não official)
A.G.S. Arquivo Geral, Salamancas
A.H.A. Arquivo Histórico de Angola (Luanda)
A.H.C.M.L. Arquivo Histórico da Câmara Municipal de Luanda
A.H.U. Arquivo Histórico Ultramarino (Lisbon)
Ajuda Biblioteca da Ajuda (Lisbon)
A.N.T.T. Arquivo Nacional da Torre do Tombo (Lisbon)
AMC Annaes Marítimos e Coloniaes (Lisbon)
ARSOM Académie royale des sciences d’outre-mer, Mémoires,
N.S. (Histoire)
A. Q.G. Arquivo do Quartel Geral (Luanda)
BIICA Boletim do Instituto de Investigação Científica de Angola
B. M. British Museum (London)
B.N.L. Biblioteca Nacional de Lisboa
B.N.M. Biblioteca Nacional, Madrid
B.N.R.J. Biblioteca Nacional do Rio de Janeiro
B.S.G.L. Biblioteca da Sociedade de Geografia de Lisboa
BSGL Boletim da Sociedade de Geografia de Lisboa
CA Current Anthropology
CEA Cahiers d‘études africaines
CSSH Comparative Studies in Society and History
EHA Études d'histoire africaine
IRCB Institut royal colonial beige, Mémoires in 8°
I.S.C.S.P.U. Instituto Superior das Ciências Sociais e Política Ultra­
marina (Lisbon)
JAH Journal of African History
JHSN Journal of the Historical Society o f Nigeria
MA Mensário Administrativo (Luanda)
MRAC Musée royal de VAfrique centrale (Tervuren), Annales,
sér. in 8°
M.R.A.C. Musée royal de l’Afrique centrale
PA Portugal em Africa
RLJ Rhodes-Livingstone Journal
SWJA Southwestern Journal o f Anthropology
 CH APTER I

Introduction

A s m o d e r n African nations drove toward political independence

during the late 1950s and early 1960s, historians did their part by
searching the African past for precedents which justified the capacity
and right of Africans to enter Kwame Nkrumah’s long-awaited
‘political kingdom’.1 They found abundant evidence in the Sudanic
empires, Zulu, interlacustrine states, and kingdoms of the savanna
which became the benchmarks punctuating the teaching and study
of African history up to the present time. All of these stood in
apparent stark contrast to vaguely understood ‘stateless’ areas in
the interstices between the states, however, and scholars found
themselves asking why and how so-called ‘stateless societies’ had
once made the transition to the various forms of polity which pre­
occupied academics and politicians alike. Their efforts, which this
introduction summarizes in rather schematic form, drew eclectically
on a number of intellectual currents present in Western academic
circles but tended to cluster about one of the most hoary myths of
the African past—the ‘Hamitic’ hypothesis—and derived from it a
variety of conclusions which this study questions through analysis
of the process of state-formation among the Mbundu people of
Angola.
The first generation of professional historians concerned with
Africa accepted the basically dichotomous concepts of ‘states’
and ‘stateless’ societies as their framework for studying African
political history.2 In doing so, they tended to ignore the implications
of several anthropological studies which not only de-emphasized
the contrast between centralized territorial ‘states’ and acephalous
‘stateless’ societies organized in terms of descent groups, age-
grades, secret societies, and the like, but also showed that many, if
not all, African ‘states’ incorporated a variety of strong lineages
1 If one may accept textbooks as barometers of the basic outlook o f a genera­
tion of historians, those of Robert I. Rotberg (1965) and Basil Davidson (e.g.
1959 and 1961, both o f which deal with kingdoms in spite o f their titles)
provide apt examples.
2 First stated formally in the classic study edited by Meyer Fortes and
E. E. Evans-Pritchard (1940), introduction.
2 IN T R O D U C T IO N
and other non-political institutions as basic elements of their
structure.3
These historians, more concerned with change than some of their
anthropologist colleagues, found that the available literature offered
little inspiration to help them handle development and process in
the kingdoms they studied. Most contented themselves with account­
ing for the ‘origins’ of a state (to use the somewhat oversimplified
terminology which has figured most prominently in the literature)
and then failed to describe any further change in the institutions
thus established. The apparent contrast between the exceedingly
complex states known to historians and the superficial ‘simplicity’
of societies lacking political institutions of a type familiar to western­
ers presented analysts with a dilemma in trying to explain how one
form of society changed into the other. Most early explanations fell
back on essentially cataclysmic processes of state-formation in
order to bridge the gap: migrating conquerors, secessions, defensive
reactions, and so on filled the literature as scholar after scholar
joined in the search for a way to link acephalous ‘stateless’ societies
with the centralized social and political controls observed in the
classic East and West African states known at the time.4
A new generation of West African historians has recently directed
the attention of scholars in directions which eliminate the old con­
trast between ‘state’ and ‘stateless’ and correspondingly reduce the
need for cataclysmic theories of state origins. They recognize that
‘state-like institutions’ (a term I would prefer not to define for the
moment) exist in societies formerly regarded as ‘stateless’ and that
3 E. E. Evans-Pritchard (1940, 1949) and A. W. Southall (1953) had earlier
provided monographs demonstrating not only the subtlety of the contrast
between ‘state’ and ‘stateless’ but had also described systems in the process of
transition from the latter to the former. Audrey Richards’ contribution to
Fortes and Evans-Pritchard (1940) mentioned but did not stress'the role of
lineages in the Bemba polity. Peter C. Lloyd (1954) pointed out that ‘the posi­
tion of the lineages in the more highly-developed kingdoms has rarely been
adequately stressed’. Jan Vansina (1962a) showed the great variety o f ways in
which already documented political structures related to their component
parts. But explicit anthropological attention turned to the issue o f states and
lineages only with Meyer Fortes (1969). Among French anthropologists,
Georges Balandier (1970) was sceptical about the dichotomy prevailing among
his British colleagues, but extended debate on the issue appeared in France
only in 1973 when Claude Tardits (1973) and Jean-Claude Galey (1973)
challenged the prevailing Levi-Straussian association o f ‘elementary structures’
(roughly equivalent to lineages) with ‘stateless societies’ and the identification
o f ‘complex structures’ with states.
4 A representative summary of theories current in the early 1960s appears
in Herbert Lewis (1966).
 THE METHODOLOGY 3
the process of state-formation often consists simply in bringing one
of these indigenous institutions to a position of overriding influence
in the society. The ‘stateless’ society may thus be converted to a
'‘state’ through a series of nearly imperceptible steps which cumu­
latively alter the relative significance of the great many structures,
centralized and decentralized, territorial and kinship, social and
political, found in most African cultures.5
The present study extends their ideas, which I see as a major new
emphasis in the study of state-formation in Africa, by reviewing the
data on the Mbundu of north-western Angola, people who lived in
a part of the continent as yet hardly touched by these stirrings of
innovation. It indicates the ways in which some of the Mbundu
built varying political structures during the centuries up to and
including the establishment of the small Portuguese state of Angola
after the turn of the seventeenth century. As a historical study, it
demonstrates, or suggests, how people organized states in a particular
social and intellectual framework, but it has little to do with devising
taxonomies of political structure or constructing generalized rules
about ‘process’. The latter tasks more properly belong in the capable
hands of anthropologists. As a historical study of non-literate
Africans, however, it necessarily borrows eclectically, and hopefully
wisely as well, from the ethnographer’s conceptual tool-kit in order
to explain the thought and behaviour of Mbundu state-builders. If
it is successful, it may describe some of the diverse forms of ‘state­
like institutions’ of the sixteenth and seventeenth centuries in terms
which make the Mbundu experience comparable to that of others
elsewhere in Africa. It may also suggest ways in which each institu­
tion changed, and simultaneously was changed by, the political and
social environment in which it flourished. The basic question, for
comparative purposes, might be phrased: how have institutions
resembling the conventional notion of a ‘state’ been formed in the
context of strongly autonomous descent groups in the case of the
Mbundu of Angola? This formulation of the problem is intended to
postpone the need for a precise a priori definition of ‘state’ since the
entire study represents in one sense a search for an empirical identi­
fication of Mbundu ‘political structures’ based.on their historical
experience. Perhaps the Mbundu states will ultimately contribute

5 1 refer particularly to recent work by E. J. Alagoa (1970, 19.71a, 1971b),

Abdullahi Smith (1970), and Robin Horton (1969, 1971). Steven Feierman
(1974) has introduced different but equally important approaches to the early
history o f eastern African states.
4 IN TR O D U C TIO N
to someone else’s universal definition of the ‘state’, but the consider­
able extent to which the Mbundu states were conditioned by the
society in which they existed stands as a warning against comprehen­
sive definitions which fail to consider the so-called ‘non-political’
aspects of society.

The ‘Hamitic M yth’ and its Legacy

The first analysts of state-formation in Africa relied excessively on
theoretical constructs drawn from simplistic European assumptions
about what was then regarded as ‘traditional Africa’. Whatever the
practical difficulties of recovering comprehensive historical data on
the remote periods when the most familiar African states had
coalesced, most early historians allowed the European intellectual
background to dominate both known fact and sometimes common
sense as they substituted often grotesque ‘elephants for want of
towns’6 at the basis of the early kingdoms. Since anthropologists of
various persuasions had been among the first European scholars to
take the civilizations of sub-Saharan Africa at all seriously, their
underlying premisses had the greatest initial influence on the question
of how states had begun in Africa. The basic assumptions of two
major schools of inter-war European ethnography,7 the culture-
historical or diffusionist Kulturkreise school represented in Africa
by Seligman and Baumann and Westermann,8 and British function­
alist social anthropology descended from Malinowski and Rad-
cliffe-Brown,9 had the greatest effect. The diffusionists’ search for
the external origins of elements in societies all over the world,
coupled with their predilection for migrations to account for the
spread of these elements, led historians to accept cataclysmic theories
of state origins, usually in the form of a ‘conquest theory’. The
functionalists’ emphasis on the integration and harmonious opera­
tion of societies left little room for change and prompted historians
to assume that an essentially ahistorical period of stability succeeded
6 Jonathan Swift, On Poetry: A Rhapsody (London, 1733), line 177.
7 American cultural evolutionists have affected African historical writing
only very recently. Robert F. Stevenson (1968) explicitly rejected the static
assumptions of early functionalism and Conrad P. Kottak (1972) and T. N .
Huffman (1972) have dealt in this vein with the problem o f state origins in
regard to two classic African cases. Some indirect influence (e.g. R. L. Cam-
eiro’s on the relation between population pressure and state-formation (1968))
may be detected in Horton (1971).
8 Charles G. Seligman (1957), Hermann Baumann and D . Westermann
(1962). For the Mbundu, see most notably H. C. de Decker (1939).
9 Most pre-1970 studies in English, including Fortes and Evans-Pritchard
(1940), represent variations on the basic theories o f this school.
 T H E ‘H A M IT IC M Y T H ’ A N D ITS LEG A CY 5
the foundation of most states and lasted until change (‘decline’)
resumed during the European conquests at the end of the nineteenth
century.10
The general image of African state-formation thus came to depend
on the arrival of skilled outsiders who imposed fully-developed state
institutions on less skilled peasants with little subsequent alteration
in the basic political structures established at the ‘conquest’. Spdrar
theory received no small reinforcement from the superficial'content
of many African oral traditions which held that ‘the world as it was
first created is the same as the world as it is now’.11 As a result, few
early historians sought more complex theories of state-formation
or looked in any systematic way for evolution of state structures
through time.12
The now infamous ‘Hamitic myth’ both anticipated and epitomized
these trains of thought and in a certain sense provided the basis on
which most early histories of African states were constructed.13
It is certainly no longer necessary to refute the racist overtones of
this theory, which have been exposed from a variety of angles,14
but it may still be useful to consider the ‘Hamitic myth’ as the proto­
type of several later and more subtle versions of the same logic of
state-formation. In its original form, the ‘Hamitic myth’ tended to
equate all ‘civilizations’ in Africa with large centralized states and
argue that
[a]part from relatively late Semitic influences . . . the civilization of
Africa are the civilizations of the Hamites, its history the record of these
10 Neo-Marxist evolutionists have had relatively little to say about state-
formation in Africa owing to the obvious inapplicability of the ‘Asiatic mode
o f production’ to conditions in Africa; see, for example, Jean Suret-Canale
(1964). Recent revisions suggested by Catherine Coquery-Vidrovitch (1969)
to reduce the dichotomy between ‘states’ and ‘stateless societies’ seek local
pressures that might cause state structures to coalesce. Emmanuel Terray
(1974), though still tending to underrate the political and economic potential
o f lineage-based societies, has begun to integrate descent groups into the
Marxist understanding of African states.
11 Jan Vansina (1965), p. 105.
12 Lloyd A. Fallers (1965) nicely illustrates the prevailing assumption that
‘change’, or at least discernible change, began when European influence
penetrated.
13 Seligman (1957) gave the classic formulation to the idea, o f long standing
in the cultural tradition o f Europe, that a vaguely racial/cultural/linguistic
group known as ‘Hamites’ spread themselves and much o f their culture over
sub-saharan Africa.
14 J. Greenberg (1949), St. Clair Drake (1959), R. G. Armstrong (1960),
B. A. Ogot (1964), and Edith Sanders (1969), to name only some of the most
systematic critiques. .
6 IN T R O D U C T I O N
peoples and of their interaction with the two other African stocks, the
Negro and the Bushman, whether this influence was exerted by highly
civilized Egyptians or by such wider pastoralists as are represented
at the present day by the Beja and the Somali. . . . the incoming
Hamites were pastoral ‘Europeans’—arriving wave after wave—
better armed as well as quicker witted than the dark agricultural
Negroes. .. ,15
For the history of state origins, Seligman’s vivid image of spear-
brandishing ‘Hamitic’ herdsmen descending upon placid and dull
farming populations encouraged historians to accept the ‘conquest
state’ hypothesis that most African states had resulted from the
imposition of alien institutions over indigenous agriculturalists. The
search for non-African origins for the states of the western Sudan
had already led to the purported discovery of European-like Jews
in the ancestry of the kings of Ghana; the derivation of all later
western Sudanic states (Mali, Songhai, etc.) from Ghanaian prece­
dents spread the original alien inspiration widely throughout West
Africa.16 Historians of the East African lake regions, with only
marginally greater justification based on the clear association of
nineteenth-century rulers there with cattle, ascribed ‘Hamitic’ origins
to Bunyoro and all the other states allegedly derived from it.17
‘Hamites’ have popped up in the history of other kingdoms as far
south as Rhodesia, but it is sufficient to mention here only a single
example to demonstrate their influence on Mbundu historiography.
A leading ethnographer and student of the Mbundu as recently as
1962 connected the origins of the most powerful sixteenth-century
Mbundu state with the south-westerly sweep of conquering invaders
whose ‘civilizing’ influence, even without the cattle of the classic
‘Hamites’, strongly resembled that of Seligman’s prototypes.18 All
of these invading state-founders, whether ‘Hamitic’, ‘Semitic’, or
‘Cushitic’,19 played the essentially identical roles of bringing fully-
formed political institutions to establish the African states which
emerged as those best known to nineteenth- and twentieth-century
European observers. The accuracy of this image was limited to the
15 Seligman (1957), pp. 85, 140-1.
16 Maurice Delafosse (1912), ii, 22-5, 38; E. W. Bovill (1968), p. 50; early
editions of John D . Fage (1961), p. 18, repeated this notion. Similarly alien
founders were given credit for founding the Hausa city-states; Smith (1970),
pp. 329ff.
17 See authorities cited in Ogot (1964), pp. 284-5.
18 José Redinha (1962) and Antonio Miranda Magalhães (1934), pp. 540-1.
The so-called ‘Jaga’ in Angola have also been identified explicitly as ‘Hamites’;
see Conde de Ficalho (1947), pp. 49-51, and my critique (1973a).
15 A modification introduced by George P. Murdock (1959), p. 44.
 T H E ‘H A M IT IC M Y T H ’ A N D ITS LEGACY 7
extent to which it reflected turn-of-the-century European myths
about the African past.
Shorn of its most objectionable racist qualities, the old ‘Hamitic
myth’ achieved renewed respectability at the end of the 1950s in
the form of a revamped ‘Sudanic state hypothesis’ propounded in
the first modem synthetic history of Africa.20 According to this
argument, a ‘civilization’ of vaguely Egypto-south-west Asian
provenance overspread much of sub-Saharan Africa at an early
date and brought with it the first large-scale centralized kingdoms
to appear in that region:
E ssentially the ‘Sudanic’ state was a parasitic grow th, fastening itself
u p on the econ om ic base o f pre-existing agricultural societies. T o these
societies it contributed certain new ideas o f political organization, and
certain new techniques, notably in the field o f m ining, m etallurgy, and
trade. Its earliest propagators seem to have m oved south-w est from the
N ile V alley, and to have established them selves, probably w ith the aid
o f the horse and o f cavalry warfare, am on g the agricultural peoples
im m ediately to the sou th o f the Sahara. . . . [later] the first miners and
ivory-traders reached the L ake regions, the K atanga and R h od esia,
using their superior techniques to establish political pow er according
to their ow n traditional patterns. . . .21

This hypothesis retained the central premise of the old ‘Hamitic

myth’—that outsiders had brought statecraft to sub-Saharan Africa
—but substituted ‘Sudanic’ for ‘Hamitic’ in conformity with the
evident effect of the new hypothesis, that of shifting the value
judgement implicit in the contrast between mounted foreigners and
local farmers. The ‘Sudanic state’ theory clearly favoured the
indigenous agriculturalists who, it could now be seen, had had to
suffer the ‘parasitic growth’ of an essentially alien form of political
structure, imposed in toto through the more complex technology of
the invaders, rather than their racial superiority. The old ‘Hamitic
myth’ of the late 1800s had achieved mid-twentieth-century respect­
ability clothed in the less pejorative terminology of technology
rather than race.22
The first generation of historical monographs devoted to African
states quickly demonstrated the inadequacy of the assumptions
20 Roland Oliver and J. D . Fage (1962), pp. 44-52.
21 Oliver and Fage 1962, pp. 51-2.
22 An observation also made by Robin Horton (1971), p. 110, n. 47. The
parallels between the ‘Sudanic hypothesis’ and once-current theories about the
imposition of European colonial rule in Africa are too obvious to require
extensive comment.
8 IN T R O D U C T IO N
underlying both the ‘Hamitic myth’ and the ‘Sudanic state’ hypoth­
esis. To name all the studies which refuted the ideas that a single
‘origin’ accounted for all or most African states or that a single
political structure lay beneath the variegated detail would require a
list of most state histories written since the early 1960s. In general,
they first attacked the notion of a single origin by identifying ‘clusters’
of states, each group of kingdoms presumably with an independent
source of innovation and a subsequent outward diffusion of institu­
tions within a restricted radius. Emphasis then turned to the unique­
ness of each individual state and tended to discount the importance
of any external influence.2324But most of these studies still retained
the basic diffusionist idea in the guise of local outsiders who founded
most African states and the functionalist notion that fully-developed
and integrated political institutions survived more or less unchanged
from the time of the state’s foundation until ‘decline’ set in during
the nineteenth century. The ‘conquest theory’ thus remained funda­
mentally intact, since revisionists had really changed little other
than to substitute local African conquerors for the ‘Hamites’ and
‘Sudanic states’ of their predecessors.
The next major innovation in historians’ approaches to African
state-formation explored the possibility of showing structural change
in pre-colonial political institutions. These studies began to cast
doubt on the lingering idea that ‘state-formation’ occurred in a single
cataclysmic revolution. It became evident that the mature institutions
of the late nineteenth century, which most historians had projected
back into the past in largely unchanged form, represented the end
result of a lengthy process of gradual accretion, and that states
known only in their latest phases might have had entirely different
shapes in less familiar times. M. G. Smith’s study on the Zazzau
state of northern Nigeria made the crucial breakthrough for West
Africa.2,1 Almost simultaneously, Yansina produced histories of
the ICuba and Rwanda kingdoms in Central Africa which convinc­
ingly demonstrated the fundamental changes which had occurred
before these states attained their late-nineteenth-century form.
Another study, based less on ethnographic data than the previous
23 Jan Vanslna (1966a) was the most advanced o f these studies and introduced
the notion o f at least three separate origins for the clusters of states that he
identified in the southern savanna. The volume o f studies edited by Daryll
Forde and P. M. Kaberry (1967) brought home the point o f unique origins for
parts of West Africa. A good statement o f these, and other, issues appears in
Peter C. Lloyd’s review article (1968).
24 M. G. Smith (1960).
 T H E ‘H A M IT IC M Y T H ’ A N D ITS LEG A CY 9
two, proved with documentary sources that the Kongo kingdom
had undergone revolutionary changes during the sixteenth and
seventeenth centuries.25 In East Africa, it became apparent that Bug-
anda had altered the balance between two types of office-holders in
the course of the nineteenth century, ending the period in a much
more centralized and bureaucratic form than it had begun.26 With
the appearance of these books, the historian’s concern with change
had begun to overcome the static tendencies inherent in earlier
studies based on the functionalist school of anthropology.
The last assumption of the ‘Hamitic myth’ to fall by the wayside
was the reliance on outsiders to explain state-formation. Anthro­
pologists, inspired by the power of structural analysis to reveal
unsuspected dimensions of oral traditions, began the assault against
the predominant tendency to interpret traditional materials literally
by showing that primarily ideological reasons might motivate
African rulers and subjects alike to ascribe foreign origins to found­
ing Icings whatever the real circumstances.27 The alien origins of the
ubiquitous heros civilisateur, they argued, gave him and his successors
a legitimacy denied to simple residents of the country, who seemed
doomed to remain prophets without honour among their own kind.
Further, identification of the Icing as an outsider divorced him from
any connection with local interest groups and rendered him theoretic­
ally impartial in dispensing justice to the people of his kingdom.
Historians have apparently hesitated to recognize the doubt which
this hypothesis casts on explanations of state-formation which
depend on outside conquerors, perhaps because of the ahistorical
implications of some structural anthropology; as a result, invading
foreigners continue to perform their customary function in some of
the most recent studies dealing with state-formation in Africa.28
Even so schematic a review of state-formation theories as this29
indicates the continuing pervasive influence of the ‘Hamitic myth’
25 Jan Vansina (1964a, 1964b) made available in French conclusions first
reached in (1963). See also Vanisina (1962b and, for Kongo, 1966, pp. 130-4,
138-42, 147-52).
20 An argument first advanced in Martin Southwold (1961).
27 As examples o f structualist studies in Africa one may cite J. S. Boston
(1964), T. O. Beidelman (1970), and Luc de Heusch (1972).
28 Citing only a few recent examples from East, West, and Equatorial
Africa: I. A. Akinjogbin (1971), pp. 307-11; Israel K. Katolce (1971), pp.
512-14; Jean-Luc Vellut (1972), pp. 62-3. B. Crine-Mavar (1973) preserves
‘massive migrations’ in his peculiarly Lunda-centric view o f the early history
o f southern Zaire.
251 have been influenced in my interpretation by Jack Goody (1968), John
D . Fage (1965), Daryll Forde (1967), and Roland Oliver (1968).
10 IN T R O D U C T IO N
after more than a decade of modem historical studies on Africa.
First propounded in overtly racist terms during the late nineteenth
century, it was dismissed on principle after the Second World War
but retained its influence in the form of its underlying assumptions.
Of the four major elements of the ‘Hamitic myth’ present to varying
degrees in most later approaches—state-formation by conquest,
identification of the conquerors as outsiders, attribution of some type
of superiority to the conquerors, and total and instantaneous state-
formation with little later modification of political structures—all
survived in only slightly modified form in the ‘Sudanic state hypothe­
sis’. Despite the attempts of critics to arrive at a more sophisticated
understanding of ‘migrations’, ‘conquest’, and ‘origins’,30 versions of
the same ideas continued to influence historians even after they had
abandoned the idea of a single outside source for all African king­
doms. Historians are now extending their quest for evidence of
change in African states as they develop more refined techniques for
the recovery and analysis of data. Most recently, they have begun
to examine more carefully the historical significance of the omni­
present migrating outsiders who appear to have founded many
African kindoms.31
The history of early states among the Mbundu provides an oppor­
tunity to test further the applicability of new approaches currently
under development by historians of other parts of Africa. The
Mbundu clearly had large and sophisticated political structures at the
time the Portuguese arrived, but had these in fact been founded by
the classic sort of migrating conquerors from the interior which
literal interpretations of some of the traditions seemed to indicate?32
Did Mbundu political history amount to no more than the establish­
ment of an already fully developed state which expanded geo­
graphically but showed no internal evolution? Is there evidence of
technological superiority associated with state founders among the
30 Notably, Vansina (1966a), pp. 14-18, and Lewis (1966). The warnings
expressed in these studies do not appear to have been universally heeded.
Also see ‘Introduction’ in Vansina, Mauny, and Thomas (1964), pp. 86-90.
31Feierman (1974); Jan Vansina (1971b); John D . Fage (1974); and Rex
S. O’Fahey (1970) call attention to other dimensions of growth and change in
African state-formation and development.
32 The standard Portuguese accounts, which focus on the activities of
Europeans among the Mbundu are: Alfredo de Albuquerque Felner (1933) and
Ralph Delgado (1948-55). David Birmingham (1966), esp. pp. 17-20, 26-41,
has given a sound critical treatment on the written sources but in integrating
published traditional materials has retained the migration hypothesis, based
largely on G. L. Haveaux (1954). Vansina (1966a) does not deal with the
question.
 TH E M E T H O D O L O G Y 11

Mbundu? The data will show, in conformity with studies proceeding

on other fronts, that state-formation in north-western Angola may
have been a much more complex process than previous analyses have
suggested.

The Methodology33
One of the developments which have forced historians to abandon
the ‘Hamitic myth’ has been a shift away from largely ‘conjectural
history’ of the Seligman variety, which was based almost exclusively
on European preconceptions rather than data, towards a more truly
historical approach founded on more and better factual evidence.
In many ways, increasing sophistication in the recovery of informa­
tion about the African past has provided the catalyst which has
forced historians to discard the theories of the 1950s, since these
simply did not fit the facts emerging in research published during
the 1960s. The historian’s reliance on evidence, however, obliges
him to exercise careful control over the methods by which he analyses
his information, especially if he attempts to develop new ways of
elucidating the historical content of oral tradition, linguistic evidence,
and ethnographic materials. Since anthropologists, linguists, and
historians have by no means reached universal agreement on criteria
for interpreting these sources, especially when they deal with the
more remote past, the historian ought to set forth the nature of his
data as explicitly as possible, specify the means by which he collected
them, and outline the logic which supports his interpretation of
them. This section deals with the first and last of these responsibilities.
Anyone purporting to write about Mbundu political history before
the arrival of the Portuguese in the late sixteenth century incurs a
special obligation to explain why such a reconstruction, based
preponderantly on non-literary sources, can pretend to any signifi­
cant degree of accuracy about events which occurred more than
four hundred years ago.34 One way to justify the approach is to spell
out unambiguously the inherent limitations of the study, since the
familiar rule that no historian can hope to write the entire history of
an era applies with special emphasis in this case. It is the nature of
the evidence which limits knowledge of early Mbundu history to a
331 use the term ‘methodology’ to refer to the logic governing my inter­
pretation of the sources for Mbundu history; the Preface contains a brief
review o f the techniques used to collect the data.
34 Other recent studies which explore the technical limits on data on the
distant African past include Christopher Ehret (1971) and David W. Cohen
(1972).
12 IN TR O D U C TIO N
rather small fraction of the obviously much more complex totality
of Mbundu life in the sixteenth and seventeenth centuries. Although
the traditions provide a fairly coherent picture of the development
of Mbundu political institutions, they have almost nothing to say
about individuals at this period. This point may bear some additional
emphasis, since most previous writing on the area has tended to
interpret the traditions literally as dealing with individual human
protagonists; in fact, the recoverable history of the Mbundu (before
Portuguese documents highlight a few individuals) deals almost
exclusively with dynasties (rather than Icings), offices (not officials),
and emblems of authority (rather than the holders of authority).35
Except as contemporary documents yield a modest amount of
information on actual patterns of human behaviour, moreover, the
knowable history of the Mbundu deals only with idealized versions
of reality instead of with the presumably less regular vagaries of
actual historical events. This distinction—between normative and
statistical perceptions of reality, which has long been a basic concept
to anthropoligists, who distinguish between what is, what is believed
to be (or reported to be), and what ought to be—affects all history,
documentary36 or otherwise. But the gap between the idealized past
and actual behaviour becomes somewhat wider than usual in this
case owing to the highly normative quality of the traditions. What
may be known about the early political history of the Mbundu,
therefore, is limited to a rather idealized perspective on selected
aspects of the institutional development of early states.
Mbundu oral traditions are histories of groups (general, local, and
familial, to employ the terminology proposed by Vansina37) and
strongly reflect the state of Mbundu social and political structures
at the time they are recited.38 The traditions therefore never deal with
individuals. They further refer selectively to only certain parts of the
past, those which have evident analogues in the present.39 Imbangala

35 Cf. Jack Goody and Ian Watt (1963), p. 308, where the authors point out
that the characters in the genealogies in Genesis refer to groups rather than to
individuals. I do not cite the growing volume o f recent work in this field.
38 Cf. this argument applied to statistics dealing with the Atlantic slave trade;
Philip D . Curtin (1969) suggests that slavers regularly reported as fact what
they felt exports ought to be rather than what they actually were. Such estimates
greatly exceeded figures verifiable from statistical sources.
37 Vansina (1971b), p. 451.
38 Aspects o f the Malinowskian concept o f a ‘mythical charter’ have obvious
relevance to the way in which the Imbangala use information about the past.
39 In addition to the sort o f histories discussed here, the Imbangala tell a
wealth of personal recollections which cover the period back to about 1870.
 THE METHODOLOGY 13
historians,40 from whose testimonies come most of the traditions
used here, view past centuries through the prism of social and political
conditions of their own time, seeking the origins of descent groups,
political titles, and structural relationships which have importance
in the present.41 This means that the Imbangala tend to preserve as
‘history’ (i.e. that which occurred in the period before living memory)
only those events which established social or political precedents
influencing contemporary behaviour patterns (e.g. lineages A and B
regard each other as enemies; the modern holder of an ancient
political title plays a specified role at the Icing’s court; lineage C
occupies the lands of lineage D subject to specified conditions, and
so on). Since the Imbangala explicitly view their history as concerned
with the present, their versions of the ancient past acquire a syn­
chronic timeless quality in the eyes of literate historians trained to
view the world in a sharply diachronic perspective. The Imbangala
word which most closely approximates to the English word ‘history’,
musendo,42 has the predominantly synchronic sense of ‘connection’
rather than ‘origin’.43 The Imbangala in many ways see their past as
little more than a slightly refracted mirage of the present and en­
visage the past as events removed somewhat from the perceptions
of living people but still present in the form of their consquences.
They draw an analogy between the near-congruence of past (history)
and present and the resemblance of potsherds to the formerly whole
pot; alternatively, they sometimes point out that history is like an
ancestor spirit (nzambi, plural jinzumbi) in relation to the ancestor
when he was alive.44 In so far as the Imbangala visualize the past
40 The Imbangala are a subgroup o f the Mbundu who lived on the eastern
edge o f the Kimbundu-spealdng area; they are generally acknowledged to have
the most vital traditions still to be found in northern Angola.
41 Cf. Goody and Watt (1963) who note (p. 310) that the non-literate indi­
vidual typically ‘has little perception of the past except in terms o f the present’.
12 The term musendo may assume a plural form, misendo, in other contexts,
but the Imbangala seem to use only the singular to approximate to ‘history*
in the literate sense; see Chatelain (1894), p. 21.
43 The simultaneously diachronic and synchronic character o f Imbangala
thought about the past finds parallels generally in non-literate African socie­
ties; Horton (1967), esp. pp. 176-8. Also E. E. Evans-Pritchard (1939), esp.
pp. 212-14. The crucial influence of the absence of writing, as recognized by
Horton, receives extensive treatment in Goody and Watt (1963). See also
Vansina (1965), pp. 104-5.
44 Testimony of ICiluanje lcya Ngonga.

Since these have entirely different characteristics from the historical traditions,
and since they do not concern the period under study, I exclude them from
my analysis.
14 IN TR O D U C TIO N
as little more than an aspect of the present, their outlook on the
past does not depart significantly from that of many non-literate
societies.
The intimate association which the Mbundu see between history
and modern society and politics obviously affects the way in which
literate historians may interpret these traditions. Since the modern
traditions tend to include only those past events which have visible
manifestations in the present, they do not provide a coherent or
integrated series of past events related to one another in any causal
or chronological sense. They instead refer to/a set of unrelated past
/happenings which modem Imbangala historians embed in an arti­
ficially contrived narrative framework if they wish to make intuitive
sense of their materials for their audience.45* Therefore, implied
narrative connections between the episodes of an Imbangala
historical performance, apparently antecedent and consequent,
rarely correspond to historical cause and effect, since decades may
have separated the events that the historian links for dramatic or
didactic effect.45 Interpretations of these traditions cannot depend
on the literal content of the narrative to supply information about the
motivations of actors or the conditions determining a given action.
Two events may appear in sequence in a tradition not because one
followed the other chronologically in the past but because some other
logic (geographical, structural, etc.) causes the modern Imbangala
historian to relate them in that order. The Imbangala view of their
past, as expressed in the traditions, consists ofia series of historically
unrelated points drawn from the past; there is no developmental
chain of related events set on a time-based continuum.47 This
45 The recitation o f history is very much a public performance among the
Imbangala, and historians obviously try to tailor their performances to the
tastes of their listeners. I owe my sensitivity to this dimension o f Imbangala
history to Professor Harold Scheub o f the University o f W isconsin; cf. reference
by Vansina (1971b), p. 446 and n. 8.
16 See, for example, the reigns o f three early Imbangala longs whose reigns
were said to have spanned only a few days but who in fact ruled for nearly
fifty years according to documentary sources; Joseph C. Miller (1972a, and
forthcoming^)).
47 The analogous structure o f the Imbangala view of the past with notions of
time-reckoning observed in other societies is apparent; cf. Evans-Pritchard’s
point that the Nuer rely on reference points rather than an abstract continuum
to express time (1939). The same notion has been phrased as ‘eventual time’
by D . F. Pocock (1964). The Imbangala, o f course, have a variety o f other
time-reckoning systems conforming to the purposes at hand. I am specifically
concerned here with perceptions o f time and not with time in the philosophical
sense.
 THE M ETHODOLOGY 15
precludes any absolute chronological calculations based exclusively
on the content of the traditions.48
The influence of contemporary social and political conditions on
the Imbangala historical traditions compensates for the loss of exact
chronology by causing them to retain records of some very ancient
events. Wherever institutions have survived for a long time, they
have tended to preserve the concomitant oral evidence of these very
early (though of indeterminable calendrical date) social and political
forms. Some officials now found in Mbundu lineages, as well as
many of the descent groups themselves, have clearly existed for
hundreds of years.49 Traditions accounting for the origin of these
titles and groups may be assumed to have come from similarly
remote periods in so far as the institutions have not undergone sub­
stantial structural modification in the intervening years. Many
Mbundu institutions have, of course, suffered substantial changes in
the past, and their accompanying traditions will have shifted cor­
respondingly from their original form, but in general enough elements
appear to have survived for the modern traditions to provide a
partial but reliable picture of Mbundu social and political structures
dating back to well before the sixteenth century. Documentary
sources show that the dominant institutions of the late nineteenth
century (states, lineages, etc.) had become established at least by
the mid-1500s. The correspondingly complete modem traditions
therefore afford a relatively good guide to events since that
time.
The close association between the historical traditions and the
institutional structure of Mbundu society further introduces great
stability in the traditions which survive. Since social and political
changes tend to cause old versions of traditions to disappear, those
remaining would seem to describe historical events connected with
the establishment of their affiliated structure with a high degree of
accuracy. Thus, interpretations of traditions describing the origin
of kings’ titles which still exist today (or existed at the end of the
nineteenth century) carry a relatively high probability of veracity.
Documentary proof of this stability, deduced up to this point solely
from the inherent logic of Imbangala historical traditions, comes

481 have developed this point at some length with reference to one aspect o f
Imbangala traditions in (forthcoming(c)).
49 Based on documentary evidence (see below). Beatrix Heintze (1970)
traces a similar stability among distantly related groups south of the Kwanza
from before 1600 to around 1900.
16 IN TR O D U C TIO N
from comparing modern traditions with a mid-seventeenth-century
tradition30 which corresponds to them in nearly every significant
aspect. Even where major political changes at the centre of a state
have modified the main line of transmission, parallel traditions
dealing with the same titles may retain their earlier forms outside
the area in which modifications have occurred; such archaic tradi­
tions often provide good data on early events obscured by later
developments in the core area. The inherent stability of the traditions
may therefore allow recovery of data bearing on the formative stages
even of states with the most turbulent histories.51
Two broad categories of Imbangala historical performance com­
prehend most of the materials used in this study: genealogies, or the
musendo proper, and narrative episodes called malunda (singular,
lunda).52 Other forms of Imbangala oral art contain materials useful
for historical reconstruction; these include several such genres as
proverbs (jisabu, singular sabu), praise names (kumbu, singular and
plural), songs, and other more purely aesthetic and didactic per­
formances. The time available for field research did not allow me to
collect sufficient material to permit the sort of sophisticated criticism
necessary to make historical sense out of such sources. In partial
justification of my decision to limit analysis to the two primary
modes of historical performance, I conform to a firm distinction
which the Imbangala draw between the musendo and the malunda
on one hand and all other categories of oral performance on the
other.53 The following brief descriptions of the Imbangala genealogies
and historical narratives illustrate how they embody aspects of
Imbangala historical thought as outlined above.
The historical genealogies, Or musendo, consist of sets of personal
names linked tb each other by the conventional relationships of
descent and affinity: fathers have sons, husbands have wives,
brothers, daughters, nephews, and so on, and all figure in the
genealogical trees which Imbangala historians recite in classic
Biblical form.
50 P° João António Cavazzi de Montecúccolo (1965). Detailed analysis
follows in Chapter VI.
51 The argument that significant historical evidence may be preserved in the
form o f archaisms outside the central area o f development is analogous to that
used to reconstruct early Rwanda history in Vansina (1962b), or to the tendency
o f archaic linguistic forms to appear in peripheral regions. On the latter
tendency, see Joseph Greenberg (1972), pp. 193-4.
52 The term is verified in Chatelain (1894), p. 21; also Sigurd von Wilier
Salazar (n.d,, c. 1965?), ii. 160.
53 Testimony of Domingos Vaz.
 THE METHODOLOGY 17
N g o la a K iluanje [a m ale political title] cam e from K o n g o dya M bulu
[an ethnic group show n here as a fem ale] and begot N dam bi a N g o la
and M wiji m w a N g o la , K angunzu ka N g o la (w ho is o f N egage [a
locality in north-w estern A n gola]), and M bande a N g o la [all subordinate
m ale political titles]. M bande a N g o la , n o w king in M arim ba [Portu­
guese adm inistrative post near the K am b o river], begot K am bala ka
M bande and K in gon go lcya M bande. K in gon go kya M bande begot
M bande a K in gongo. M bande a K in gon go w ent to ngana K abari ka
N zungani [i.e. to o k a w ife from this descent group] and begot N g o n g a
a M bande, F ula dya M bande, K am ana k a M bande, N g o la a M bande,
and N jinje a M bande. N g o la a M bande married M b om b o ya N d u m b u
[a w om an o f an unidentified descent group]. . . . K abila ka N g o la begot
K akunga k a K abila, M uhi w a K abila, N zu n g i ya K abila, N g o la a
K abila . . . the ones I have just n am ed are the present sobas [M bundu
political-title-holders recognized by the P ortuguese governm ent] near
M ucari [former adm inistrative p o st east o f M alanje].*54

The testimony quoted, which is typical of the form of the Imbangala

musendo, may be represented as a genealogical tree shown in Fig. I.
Although such ‘family trees’ appear to portray a process of bio­
logical generation, with marriages and descendants, etc., they in fact
refer exclusively to political titles (the male figures) and to descent

K ongo d y a M b u lu (1)

i
N dam bi a
N g o la a K ilu a n je (m)

i
M w iji m w a
1K a n g ui n z u ;k a p . ,r ,
M b a n d e a __ (ie m aie n o t
N gola N g o la N gola N go la ~j~ m e n tio n ed )

K am b a la k a K in g o n g o k y a _ (fem ale n o t
M bande M bande J m e n tio n e d )

(u n n a m e d w o m a n fro m lin e a g e _ M bande a

o f K ab ari k a N z u n g a n i) y K in g o n g o

M bom bo ya N g o la a N gonga a F u la d y a K am ana k a N jin je a

N d u m b u (1) ^ M bande M bande M bande M bande M bande

(in fo rm a n t failed
to s p e c ify th is lin k )

K abila k a __ (fem ale n o t

N g o la ~j~ m e n tio n e d )_______

K ak u n g a k a M uhi w a N zungi ya N gola a

K abila K abila K abila K abila

F i g . i . R epresentative M usendo G enealogy

54 Testimony o f Domingos Vaz.

18 IN TR O D U C TIO N
groups (the females), all of them institutions which survive into the
present. This lends a synchronic quality to the genealogy, which
may be confirmed, as the historian did at the end of this passage,
by referring to living individuals who hold the titles mentioned.
The genealogy does have significant historical content, of however
specific and limited a sort. Relationships between political titles
linked as ‘father’ and ‘son’ in the genealogy indicate in the con­
temporary sphere that the holder of the ‘son’-title must treat the
incumbent in the ‘father’-title (e.g. Ndambi a Ngola to Ngola a
Kiluanje, with respect analogous to that owed by human offspring
to their biological/social father. The holder of the junior title is
politically subordinate to the holder of the senior title in this sense.
Historically, these relationships also mean that an occupant of the
‘father’-title created the ‘son’-title, awarding it at some time in the
past to an unmentioned member of a descent group (possibly but
not necessarily also a biological son of the senior title-holder at the
time). The genealogy identifies the descent group either by its own
name (e.g. Kabari lea Nzungani) or by the name of a historical
woman of that lineage (e.g. Mbombo ya Ndumbu). The ‘marriages’
shown in the genealogy thus link (mostly) ‘male’ political titles as
‘sons’ to the ‘female’ lineages as ‘mothers’; the maternal lineage
holds the right to nominate incumbents to the ‘son’ position and
thus possesses the title in the sense that Imbangala matrilineages
control their own members. The elders or ‘uncles’ of the lineage
(makota, singular koto) act as guardians and advisers to the title-
holder.
Although most titles are affiliated with a single ‘maternal’ descent
group, such major positions as that of the ngola a kiluanje, to which
many lineages may nominate occupants in rotation, significantly
have no specific lineage (or female name) associated with them (as a
‘wife’). In this case, the ngola a kiluanje descends from Kongo dya
Mbulu, a name vaguely representing all the northern Mbundu
lineages. Historically, therefore, a genealogy like the one quoted
above may be read as a diachronic record of the spread of political
authority derived from a major Mbundu king, the ngola a kiluanje.
It simultaneously names the descent groups incorported into the
kingdom in association with the titles entrusted to their control.
Synchronically, the same genealogy amounts to an organization
chart of the state, since it specifies the hierarchical relationships
between titles and, by extension,between the affiliated lineages.
A genealogy composed of titles reveals nothing about the absolute
 THE M ETHODOLOGY 19
chronology of the state-building process. One cannot calculate on
the basis of assumed human life spans, for example, the minimum
or maximum time elapsed between the creation of the ngola a kiluanje
title and the multi wa kabila position. An unknown number of human
incumbents could have occupied both of these positions, as well as
all the intervening titles, and any one of the occupants, first, third,
or tenth, could have awarded the subordinate position listed as a
‘son’ in the genealogy. It is permissible to draw only the limited
conclusion that the creation of a ‘son’ title necessarily succeeds the
origin of its ‘father’ title chronologically, but theoretically this
interval may vary from a year or less to several centuries. Nor does
it follow that ‘sibling’ titles (e.g. Ndambi a Ngola, Mwiji mwa
Ngola, Kangunzu lea Ngola, and Mbande a Ngola) must have been
created more or less contemporaneously, since succeeding occupants
of the ‘father’-title may grant subordinate positions of equal genea­
logical rank. In general, positions in the senior levels of the genealogy
tend to be older than those near the bottom, for historical reasons,
but enough exceptions are known to make historical inferences on
this basis highly untrustworthy.
Presenting all these names as a single complex genealogy distorts
slightly the Imbangala conception of their musendo, since they
recognize a number of distinct genealogical trees, each corresponding
to a major social or political institution such as a state or an ethnic
group. The genealogy reproduced above begins with a name taken
from a distinct semi-mythical aetiological genealogy (Kongo dya
Mbulu)55 and then traces the development of a portion of the
sixteenth-century ngola a kiluanje state through the main line of
‘male’ (i.e. political) descent. The female figures in the genealogy all
come from other independent genealogies describing structural
relationships of the descent groups ruled by the ngola a kiluanje
and its associated subordinate titles. Because my data come primarily
from the eastern fringe of the Mbundu ethno-linguistic area, I
551 use the term ‘mythical’ in its technical sense o f referring to events
believed to lie in a period outside o f history, timeless, composed o f unconnected
episodes, the period of ‘God on earth’, as described, for example, by M. I.
Finley (1965). Although all Imbangala ‘history’ exhibits certain ‘mythical’
characteristics by literate standards (magic, and so forth), the Mbundu clearly
distinguish between their aetiological tradition and the historical musendo
genealogies. The names in the latter ‘exist’ in the sense that they have living
incumbents; those o f the former do not and never have. If the Imbangala have
an elaborate cosmology phrased in mythological terms, it did not come up in
the context of discussions of history; for the Mbundu generally, see Chatelain
(1894).
20 IN T R O D U C T IO N
cannot give an exhaustive list of all Mbundu lineage genealogies,
but the regular structure of these traditions suggests that there must
be a six or seven sets of names in all, one for each of the names in
the aetiological genealogy (of which Kongo dya Mbulu is only
one).
The internal shape of these lineage genealogies may vary con­
siderably. Of the three for which I have adequate information, one
exhibits certain features of a classic segmentary genealogy in which
the component descent groups are articulated into a single compre­
hensive genealogical tree, very deep (twelve or more ‘generations’)
and pyramidal in shape. The other two show explicit links only
between closely related lineages, rarely exceed two or three genera­
tions in depth, and jump directly from these levels to an apical
putative ancestor; this structure resembles that of a clan, with many
lineages of approximately equal rank, none of which can trace its
exact descent from the common founder.56 The variation in the
internal structures of these lineage genealogies makes no difference
for their historical interpretation, however, since the same sort of
‘marriage’ relates them all to the political genealogies.57

A e tio lo g ic a l g e n e a lo g y

(Ilc tiv e (ilc tiv e

a n c e s tra l lin k s) a n c e s tra l lin k s)

Fio. II. Diagram of how Genealogies Combine to Describe a Kingdom

Separate political genealogies composed of titles rather than line­

ages exist for each political entity recognized by the Mbundu, one
for the ngola a kiluanje, another for the Mbondo state of the ndala
kisua, another for the Lunda empire, yet another for the Kasanje
56 These groupings o f lineages are only incipiently clans, however, since they
do not descend from discrete eponymous ancestors (several groupings with
different names claim to come from the same ancestor). They lack ‘totems’
and impose no significant political obligations on their members.
57 Fig. II shows the links in diagrammatic form. Boston (1964), p. 112,
describes a parallel symbolic ‘marriage’ among the Igala.
 THE METHODOLOGY 21

state of the Imbangala, and so on.58 A few isolated titles not con­
nected with any of these coherent genealogies survive as remnants
from ancient states no longer in existence. Their separation from the
fixed genealogical structures of the main musendo frees them to
move about in the other genealogical fields according to the whim
or design of individual historians; some of these may be dated to
before the middle of the sixteenth century by documentary sources.
The Mbundu musendo may therefore be thought of as a number of
distinct genealogical sets which fall into two basic types: lineage
genealogies, which show structural relationships between existing
lineages and at the same time reveal aspects of the historical processes
of lineage fission which have led to the present distribution of
descent groups, and political genealogies simultaneously showing
the composition of Mbundu states and giving part of the historical
development of these kingdoms. The individual historian, like Levi-
Strauss’s bricoleur,59 constructs composite genealogies, like that
reproduced in Fig. I, to link individual descent groups to one or
another of the state structures, portraying these links as ‘marriages’
between male political titles and female descent groups. An over­
riding aetiological genealogy ties recognized Mbundu subgroups to
each other and relates the Mbundu as a whole to some of their
neighbours (see Fig. III).
The mahmda narrative episodes, the second form in which the
Imbangala recite their history, are appendages to the names of both
the political titles and the descent groups given in the genealogies.60
The oral historian may, after reciting a genealogy, tell in his own
words as few or as many as he chooses of a fairly standardized set of
narrative episodes connected with each name in the musendo. He
draws on a relatively small set of prose expositions to make a defined
number of points about the origins, rights, or responsibilities of the
titles or lineages involved. Each lunda accounts for a recognized
duty or privilege, and the finiteness and standardization of the set of
malunda associated with each title derive from the limited number of
formal relationships which most positions maintain with lineages or
with other titles.
58 Fig. I l l shows the main political structures recognized by the Mbundu
as circles in the lower half o f the diagram.
55 Claude Lévi-Strauss (1966), jpp. 16-36.
60 The historical malunda constitute a subset o f a much larger body of non-
historical but similarly structured prose compositions (animal tales, stories
dealing with domestic themes, etc.). Chatelain (1894) has published a number
of the Mbundu malunda in Kimbundu with English translations.
 22

A e tio lo g ic a l G e n e a lo g y
IN TR O D U C TIO N
 THE M ETHODOLOGY 23
Historians apparently use similar narrative episodes consistently
to make the same points about titles and descent groups over long
periods of time. This conclusion stems from comparing variants
collected from the seventeenth to the twentieth centuries of some of
the best-known malunda. Individual historians may, however,
employ strikingly diverse images, metaphors, cliches, or plots as
they apply their creative skills to make their performances works of
aesthetic as well as historical value. The resulting flexibility of detail
in variant malunda contrasts markedly with the apparent stability
of the central point of each episode which, in cases which can be
verified, has remained constant through decades and, in some cases,
centuries. In all known cases where documentary or eye-witness
recollections of an event may be compared with later narrative
episodes dealing with the same occurrence, the malunda accurately
preserve a nucleus of historical fact even while the historians who
recite the episode surround it with fictional artistic elaboration.61
Since the context of an authentic historical performance62 usually
establishes the historian’s purpose in selecting given episodes to
include in his narrative (e.g. to vindicate lineage rights to a title, to
establish a precedent, to instruct, to honour a title-holder present,
and so on), and since the historical nucleus of the chosen narrative
episode usually pertains directly to the historian’s aims, analysis of
the performance allows the determination (at least at the level of a
workable first approximation) of the historical significance of most
narrative episodes. The probabilities of accuracy become fairly high
when a number of variants of the same lunda may be compared
to identify the stable part which recurs in all instances.
The metaphors, the symbols, even the plot which historians choose
to illuminate the significance of each lunda to their listeners belong
011 refer here to the seventeenth-century near-eye-witness account recorded
in Cavazzi (1965) and to nineteenth- and twentieth-century malunda describing
the same events; additional confirmation comes from an analysis o f personal
recollections and written eyewitness accounts o f mid-nineteenth-century wars
which were beginning to be recounted in the form o f malunda in the 1960s.
The stable nucleus o f historical fact in an Mbundu lunda may be compared to
the ‘core cliches’ o f Xhosa intsomi analysed by Harold Scheub (1975). The
origin o f the idea o f a stable nucleus goes back to Lévi-Strauss; cf. J. S. Boston
(1969), p. 36.
52 An ‘authentic historical performance’ by definition arises only from the
normal activities o f people in the society, usually in connection with legal
disputes or other occasions calling into question the formal relationships
between titles and/or descent groups. Such circumstances occur with sharply
diminished frequency in modem Angola, and this circumstance accounts for
the atrophied condition o f the modem traditions.
24 IN T R O D U C T IO N
to the rich and complex fund of Mbundu artistic and intellectual
themes rather than to history.63 The artistic and intellectual level of
the malunda may provide a fertile field for the sort of structuralist
analysis which has proved valuable in penetrating the intellectual
processes and cosmologies of non-literate peoples the world over,
but the fact that Imbangala historical narratives exhibit this dimen­
sion at one level does not eliminate their value as historical sources
at another.64 Comparison of the malunda with documentary sources
demonstrates empirically that historical content is present, and the
internal logic of a historical performance, which relates malunda to
specific titles and lineages and to a resticted range of historical points,
makes it possible to distinguish the less obviously historical content,
which the individual historian may add on his own initiative, from
the stable historical skeleton of the narrative episode.65 The Western
critic, by distinguishing carefully between the historical nucleus and
the artistic embroidery, may confidently use the malunda as sources
for the limited range of topics with which this study deals.66
The malunda may be further sub-divided into political or lineage
episodes according to the type of name or title with which they are
associated. Those attached to the political genealogies generally
describe circumstances surrounding the creation of the titles and
establish the right of their descent group ‘owners’ to control them.
Other malunda connected with the same title may relate events
believed to account for its formal relationships to other political
positions. They may also deal with the origin of authority emblems
connected with the title, spell out its magical powers, or explain ritual
proscriptions affecting its occupants. Narrative episodes appended to
63 For an analysis o f similarly complex symbolic systems o f the related
Ndembu o f north-western Zambia, see Victor Turner (1967).
641 concur with the analogy offered by Vansina (1971b), p. 455, where he
argues that the highly symbolic overtones o f conventional American school-
text versions of the story of the Mayflower do not destroy its historical content;
the symbolism merely disguises the history. But for more sceptical opinions,
see Beidelman (1970) and Wyatt McGaffey (1970); both o f the latter, it seems
to me, demand that ‘history’ should attain a higher degree o f probability
(approaching virtual certainty) about the past than many modem historians
would consider necessary.
65 The relatively limited fieldwork on which this study is based did not per­
mit me to study systematically the important subject o f Mbundu symbolism.
Analysis of the historical significance o f the artistic dimension o f the malunda
remains a unexplored field o f great potential importance, hence my qualifica­
tion on the meaning of the artistic aspects o f the malunda as ‘less obviously’
historical.
66 Feierman (1974) contains the most sophisticated attempt to elaborate the
historical content o f similar materials known to me.
 THE METHODOLOGY 25
lineage genealogies almost always justify the fission of a new lineage
from its parent descent group, trace the route which members of the
‘nephew’ lineage followed to reach their current home, and relate
the conditions under which the group received rights to occupy its
present lands.67 Other malunda spell out the descent group’s relations
with its neighbours, defend its claim to control political titles and so
on. The obvious functional importance of the narrative episodes in
legitimizing present lineage or title rights does not obviate the fact
that most, if not all, current relationships rest on historical prece­
dents which may—with care—be identified through analysis of the
malunda. The present, as the Imbangala say, is like the bones of an
ancestor, and the analogy between historical reconstruction based
on Imbangala narrative episodes and palaeo-zoologists’ techniques
of physiological reconstruction may be apt.
It is important to recognize that each narrative episode may be
told entirely independently of all others. All properly performed
malunda have their own complete plot, a beginning and an ending,
and do not depend on other episodes for their meaning or their
artistic integrity. The Imbangala historian may recite in sequence
any number of malunda bearing on a given title, however, and if he
is skilful he may succeed in weaving together plots, themes, and
imagery which transcend the individual episodes and link the dis­
crete elements of his performance into a much longer, integrated
historical and aesthetic composition. But the Western historian would
be ill-advised to mistake the plot line constructed by the performer
for evidence of consistent historical development running through the
entire string of narrative episodes. The malunda selected for any
performance depend in part on its context rather than on the logic
of the historical events, and no two performances are likely to
include the same set of narrative episodes. It follows that these
longer performances are not subject to analysis which depends in
any way on connecting the implications of one lunda to those of
another in any direct sense; rather, the historical content of each
episode must derive from its own internal logic. Clearly, there can
be no chronology, even relative, based on the order in which episodes
may be told. It is often possible, on the other hand, to find indirect
li7 The structure o f these traditions resembles that o f lineage traditions
discussed—and dismissed as history—by McGaffey (1970), passim but esp. pp.
18IT. The stability o f Mbundu lineage genealogies and the greater coherence of
M bundu descent groups render Imbangala malunda less stereotyped and give
them considerably greater value as history than the Kongo traditions described
by McGaffey.
26 IN T R O D U C T IO N
indiqations of ths historical sequences of the events described by the
episodes. If, for example, the malunda detail the stages in a physical
movement from one geographical location to another, placing the
episodes along a straight line connecting the two points may approxi­
mate fairly well the order in which the actual events occurred.
The interpretation of Imbangala historical malunda obviously
requires that the historian should use all available external sources
which may help him to distinguish fact from fiction. Such sources
include words, especially proper names and toponyms, terms for
authority symbols connected with the titles and lineages of the
genealogies, and other technical terms which may indicate historical
processes of diffusion or population movements across known langu­
age or dialect barriers. Analysis of ethnographic evidence, especially
authority symbols or practices closely identified with distinct groups
of people, can provide similar assistance. Since both these types of
evidence exist only in the present and are often difficult to establish
directly for the past (except for archaeological recovery of material
objects), their application to the traditions dealing with the sixteenth
or seventeenth century imposes an obligation to determine that
language boundaries have not shifted in the mean time or that social
and belief systems have not changed significantly.68 The usual rule
that ‘the absence of evidence of change allows the historian to assume
past stability’ seems less and less acceptable in the light of growing
evidence, to which the Mbundu contribute, that major shifts can
and do take place even in those aspects of African life once thought
to be most resistant to change. I have tried to rest my analysis on such
stability of social structure and language only when positive evidence
indicates the probability of no significant change.
Under the circumstances, documentary sources become crucial
adjuncts to the use of the non-written sources described in the
preceding paragraphs. It is the availability of sufficient written
material for sixteenth- and seventeenth-century Angola which, in
the final analysis, makes it possible to attempt to reconstruct Mbundu
political history for the period. I have already justified use of Imban­
gala musendo and malunda on the grounds that early documentary
materials confirm both the absence of significant change in their
content over three centuries or more and the close correspondence
of the traditions to events described by eye-witnesses. I have made
the use of ethnographic and linguistic evidence dependent on the
68 Useful introductions include Daniel F. McCall (1964) and Vansina (1968
and 1970).
 TH E M E T H O D O L O G Y 27
possibility of establishing, largely by documentary methods, that
the necessary condition of stability holds. Written records for the
Mbundu are relatively abundant and accessible owing to the six­
teenth- and seventeenth-century activities of the Portuguese govern­
ment, Dutch trading companies, and missionaries from several
European countries (primarily Italy) in the area.69 Fortunately for
modern knowledge about Mbundu history, a few of these Europeans
took an active interest in things African and wrote accounts giving
their impressions of Mbundu traditions as they existed in the
seventeenth century. It is important to emphasize the distinction
between Mbundu histories as they were then told and seventeenth-
century European perceptions of them, since few writers understood
much of what they heard. Comparison of the written records with
modern oral and ethnographic evidence shows that the documentary
sources are only slightly less encrusted with the personality of their
authors than are the malunda with the artistry of individual Mbundu
historians. The documents are also comparable to the oral traditions
in that they provide nearly as selective a view of seventeenth-century
reality as do the genealogies.70 Written and unwritten sources over­
lap often enough to establish a complementary basis for mutual
criticism, but they also treat completely different aspects of events
sufficiently frequently to illuminate a relatively broad range of
Mbundu political history. The documents, for example, typically
describe battles against Mbundu title-holders whose origins and
significance in African terms may be deduced from the traditions.
This outline of the forms and characteristics of the sources for
early Mbundu political history provides the background necessary
for an explicit statement of the methodology which lies behind the
following historical reconstruction. The major technical difficulties
hinge on finding a rationale for (a) projecting facts observed in the
nineteenth and twentieth centuries (word lists, ethnographic data,
but also oral traditions) back over three centuries in time, (b)
projecting eastward in space the seventeenth-century documentary
evidence which deals primarily with the westernmost parts of the

09 In addition to Cavazzi (1965), the account o f António de Oliveira de

Cadomega (1940-2) provides valuable information. Nearly all known docu­
ments for the years before 1600 and many o f those between 1600 and 1655
are published in António Brásio (1952-71).
70 Cf. G. I. Jones (1963), p. 391, who states that European documents for the
Niger Delta region ‘may be more legendary in character than the African and
subject to just the same processes o f compression and the same dependence on
“structural time” ’.
28 IN T R O D U C T IO N
Mbundu region, and (c) generalizing to the entire Mbundu area
from field data collected mainly among the Imbangala subgroup of
the Mbundu. Recognizing thal^ill history is a matter of probabilities
rather than of certainty and acknowledging that the probabilities
in this case may not approach the level attainable in other times and
places, the qualities of the available data seem to vindicate drawing
conclusions about the institutional history of the Mbundu since (a)
the major Mbundu social, political, and intellectual systems cover
the entire region, (b) like all such structures, they are tightly inte­
grated with one another so that knowledge of the existence of one
aspect of the culture in one place allows us to infer a great deal
about other related aspects of the culture in other places, and (c)
changes in one part of the culture may be expected to induce changes
in others. If, therefore, an institution is known in some detail for the
nineteenth or twentieth century and parts of it are visible near the
coast in the seventeenth-century documents, the historian may pro­
ceed as if the entire structure existed elsewhere during the earlier
period.71 1 should not exceed the limitations of this method if I
deal only with the. history of political and social institutions based
on oral traditions, incorporating other aspects of Mbundu history,
the behaviour of individuals, and the like, when documentary
sources make them accessible.

The Perspective
An initial statement of the perspective adopted for this study may
make its argument more readily intelligible. I am attempting to
assume the point of view of a literate outside observer standing
somewhere near the centre of Mbundu territory, watching on the
one hand political developments which occur without visible external
influence among the people living nearby, and observing on the
other a series of changes stimulated by contacts between the Mbundu
and other people who live around them. The study thus describes
changes through time in a single relatively small region and shows
its political contacts with neighbouring areas in order to assess how
external influences affect and are in turn affected by a basically
continuous process of local historical evolution. It rejects, therefore,
the approach of following migrating conquerors across the landscape
and holds that relatively little large-scale demographic change has
taken place within the discernible past, that the Mbundu of today
71 Arguments referred to as those o f ‘inference’ and ‘extrapolation’ in
Vansina (1968), pp. 106-8.
 T H E PE R S P E C T IV E 29
are for the most part biological descendants of sixteenth-century
Mbundu, and that this sort of demographic continuity extends
back to the time of the introduction of agriculture and perhaps
before. State-formation seems to result more from the diffusion of
ideas, institutions, authority symbols, and the like than from mass
movements of people. This approach certainly conforms to our
present understanding of more recent periods in African history
where the examples of long-distance movements of large numbers of
people are well known only because of their rarity.72 No obvious
reasons suggest that such migrations should have been less rare in
the nineteenth century than in the sixteenth or the fourteenth
centuries.
The evidence for the Mbundu shows an institutional history of the
origin and diffusion not of ‘states’ but of principles of political
organization. These may be traced via the genealogies describing
the relationships between titles based on them and through surviving
distributions of the emblems connected with each form of authority.
The formation of an Mbundu state thus represents less a movement
of people than the reorganization of an existing population into new
groupings, established according to new sets of symbols of group
unity, and related to one another on the basis of the pattern followed
by the new authority emblem as it diffuses. The major historical
problem involves the construction of a reliable sequence for the
multiplicity of authority symbols now present among the Mbundu
as evidence of their highly complex political history. Diffusion
processes are handled as individual items moving over a limited
area and thus avoid the worst pitfalls of the Kulturlcreise school and
its world-wide diffusions of ‘complexes’ of culture traits. My hand­
ling of the ethnographic evidence tries to avoid immutable ‘survivals’
72 Only the mfecane comes to mind as a documented example of the kind
of mass migration customarily posited for early African state-formation; more
subtle and gradual population movements have occurred among such groups
as the West African Jula, the Fang o f Gabon, the Cokwe in Angola, and others,
all without the formation o f strong centralized states. Jan Vansina (1969), pp.
20-44, outlines such a diffusion o f an authority symbol (the nkobi), again
without attendant large-scale migration. More extensive population movements
can and do take place where technology allows people to fill an empty ecological
niche; the movement o f herders into areas occupied only by agriculturalists
constitutes the most important instance o f this phenomenon in Africa. One
documented migration explicable in these terms is that o f the Cokwe; Joseph
C. Miller (1970). Clearly, the relative permanence o f population stocks and the
relative ease with which ideas and institutions spread has been more common.
The Alur and Anualc examples (p. 2, n. 3 above) may be typical, allowing for
minor movements of individuals (not groups) in all times and places.
30 IN TR O D U C TIO N
in favour of ‘retentions’73 which may have undergone considerable
modification in the course of the centuries. Modern Mbundu culture
may, in short, be read as a congeries of independent (though inte­
grated) symbols, words, traditions, institutions, and practices which
come from varying times in the past. In so far as documents and logic
can reveal the sequence of their appearance as parts of an evolving
Mbundu culture, the historian may suggest hypotheses which
explain how and why people adopted them. Such must pass for the
early political history of the Mbundu.
73 A term suggested by Jan Vansina; personal communication.
 C H A P T E R II

T he Setting

H i s t o r i a n s have recently begun to recognize the importance of the

environment—in both the geographical and social senses—for
understanding the process of state-formation in Africa. The early
political history of the Mbundu becomes fully understandable only
in terms of its geographical setting of mountains, rivers, economic
resources, crops, and climate, and in terms of the general social and
intellectual background of their culture. Since few aspects of
the physical geography of Mbundu territory have changed since the
sixteenth century, no significant technical problems cloud the
following outline of the main ecological influences on Mbundu
political history. The human geography of the sixteenth-century
Mbundu is somewhat less certain, not only because the boundaries
of the main Mbundu ethno-linguistic subgroupings have changed
since then but also because even modern ethnic distinctions within
the Mbundu region are not well understood. Rather less probabilistic,
for reasons of documentation discussed in Chapter I, is the review
of selected aspects of early Mbundu social structure which concludes
this chapter and completes the review of background materials
necessary to evaluate the earliest known Mbundu states.

Physical Environment
The general contours of Angolan geography conform to the broad
pattern of the southern half of the African continent which geog­
raphers liken to a great inverted saucer: a low and narrow strip of
sandy land separates the Atlantic in the west from ranges of hills
which rise in terraces towards a high interior plateau in the east.
This basic relief pattern is less distinct in the north near the mouth
of the Congo (or Zaire) river, where the interior elevations are lower
but it becomes very well marked towards the south where the high­
lands attain altitudes well in excess of 6,000 feet, sometimes rising
in abrupt escarpments above the coastal plains. Numerous rivers
run generally from east to west, draining the mountain slopes in
shallow, rocky beds which do not widen to navigable dimensions
until they approach very near the coast. Even the largest rivers—the
32 T H E SE T T IN G
Congo, the Kwanza, and the ICunene—allow ocean navigation to
penetrate for only 100 miles or less, and all but the Congo have
only short reaches of open water in the interior. The rivers of the
northern interior tend to flow east and north towards the main
tributaries of the Congo, while those in the south feed the upper
Kwanza, the Kubango-Okavango, and the Zambezi systems. The
Kwango is the main northward-flowing tributary of the Congo in
Mbundu territory.
Rainfall decreases from north to south, ranging from generally
reliable, though seasonal, rains near the Congo river to desert
conditions as one approaches the lower Kunene. It also moderates
from east to west owing to the prevailing easterly wind patterns.
Most precipitation in the interior comes from the east during the
summer months, from September to April in the north but with a
markedly shorter rainy period towards the south. The entire coastal
lowland is significantly drier than the highlands behind since prevail­
ing onshore winds blow cool air from the cold Benguela current
from southern Africa almost as far north as the mouth of the Congo.
This air warms and dries as it passes over the land and drops almost
no precipitation except where the windward slopes of the mountains
force these winds aloft. A generally cloudless and pleasant dry
season in the interior highlands contrasts with a cool, humid, and
cloudy (but rainless) winter known as cacimbo along most of the
coast.1
Sixteenth-century Mbundu demography corresponded roughly
to the hydrography of north-western Angola. Ethno-linguistic
boundaries tended to follow major watersheds except in the west
where the mountains east of Luanda island provided a natural
boundary separating the Mbundu in the highlands from Kongo in
the meteorologically and geographically distinct lowlands.2 Other­
wise, the Mbundu were generally confined to the region drained by
the Kwanza river. The most important northern tributary of the
Kwanza, the Lukala, flows down through a plateau sloping upward
from elevations of around 3,000 feet on its western tongue along the
middle reaches of the river to over 4,000 feet in what was in 1969
eastern Malanje district. The northern boundaries of the Mbundu,
which divide them from the Kongo, follow the hilly highlands which
1 See F. Mouta and H. O’Donnell (1933), D . S. Whittlesley (1924), and
Domingos H. G. Gouveia (1956).
2 This corrects the general but apparently erroneous impression that Mbundu
territories extended to the ocean; Joseph C. Miller (1972b) and remarks
following.
 P H Y S IC A L E N V IR O N M E N T
33

M ap i . Geography of the Mbundu and their Neighbors (Modem)

34 T H E S E T T IN G
rim this plateau around the headwaters of the Nzenza (Bengo) and
Dande rivers (the area known later as the ‘Dembos’) up to their
crest near the modern Portuguese towns of Carmona and Negage.
The southern tributaries of the Kwanza flow from the so-called
Ovimbundu highlands, or the Benguela plateau, through territory
inhabited by the Mbundu at least as far south as the Longa, the first
major river south of the Kwanza. The highland people who lived on
the watershed to the south and south-west were known later as
Ovimbundu3 and differed somewhat more from the Mbundu in
language and culture than did the Kongo. The inhabitants of the
dry coastal lowlands immediately south of the Kwanza—called
Kisama as far as the lower Longa, Sumbe between the Longa and
the Kuvo, and Seles, Mundombe, etc., beyond—appear to have
had more in common with the Ovimbundu than with their Kongo
or Mbundu neighbours to the north. The occupants of the so-called
Luanda Plateau, the area drained by the Kwije and Luhando rivers
east of the upper Kwanza, also belonged to the Mbundu; no definite
boundary demarcated their south-eastern limit and the south-eastern
Mbundu shaded gradually into Cokwe and Ngangela.
The major exception to the generally highland environment
occupied by the Mbundu occurred in the far north-east where people
sharing the same ethno-linguistic characteristics lived in the rela­
tively low (1,200 to 2,000 feet) and wide basin of the Kambo, Lui,
and Kwango rivers. A nearly vertical escarpment, varying from a
few hundred to over 2,000 feet in height, runs south-west from the
headwaters of the Kwale to beyond the Kwango and separates
these lowlands, known more recently as the Baixa de Cassanje,
from the much higher elevations just to the west. The Mbundu of
the northern parts of the Baixa de Cassanje seem to have merged
gradually into Kongo, while those along the Kwango tended to
resemble the Cokwe/Lwena and Lunda who inhabited the high
savannas which extend east from that river for nearly 1,000 miles.
Other than those of the Baixa de Cassanje, the Mbundu generally
lived only in the higher elevations of the plateaus surrounding the
Kwanza east of the mountains which separate its interior basin from
the lowlands near the sea.
The dominant vegetation pattern on the Mbundu highlands
consists of open grassland interspersed with occasional wooded
3 For the purposes of this study, the Mbundu are the people under examina­
tion generally, and they speak the Kimbundu language. The Ovimbundu live
on the Benguela plateau and speak the Umbundu language.
 PH Y SIC A L E N V IR O N M E N T 35
savanna, now restricted to a few areas but probably more extensive
in the sixteenth century before increased hunting with fire in recent
years destroyed many trees. Other exceptions to the prevalent grass­
land environment include rain and cloud condensation forests
covering the mountains along the western edge of the Luanda
plateau, patches of gallery forests (mishito, singular mushito) along
the lower courses of the larger rivers, and patches of forest on the
escarpment of the Baixa de Cassanje.4 Generally moderate amounts
of rain, varying from 36 inches per year in the west up to 55 inches
annually near the Kwango, prevent the growth of heavier vegetation
elsewhere. The rains come in an extremely uneven seasonal pattern—
light and irregular from September or October to December, then a
short and highly variable dry spell, with the heaviest precipitation
following in February and March—and consequently had an
important influence on human economic activity.5
Although little research has illuminated sixteenth-century Mbundu
agricultural techniques, it is clear that most people farmed and it is
probable that they mainly grew varieties of millets and sorghum.
In the absence of a drought-resistant staple crop, such as manioc,6
successful agriculture depended on taking maximum advantage of
the rainy portion of the year to grow sufficient food to last through
the dry months from May to September, when no rain falls in most
years (although one or two storms occasionally punctuate the dry
season), and the first part of the ensuing growing season. The
Mbundu supplemented these dietary staples with wild vegetables
and fruits, and they valued especially the plants found in the mushito
forests along the rivers. For these reasons, and because the unavail­
ability of a dry land crop such as manioc restricted the amount of
land suitable for growing food to somewhat less than the Mbundu
farm today, they tended to settle in the lower and moister areas,
especially river bottoms and mountain valleys where ground moisture
evaporated less quickly than it did along the ridges.7 To these
4 John Gossweiler (1939); E. K. Airy Shaw (1947) has summarized Goss-
weiler in English.
5 Angola, Serviços Meteorológicos (1955) contains maps showing recent
rainfall and temperature patterns for the region; paleoclimatological studies
do not exist.
6 The Mbundu north o f the Kwanza today depend on manioc as their main
agricultural ataple, but all varieties descend from plants (mainly manihot
utilissima) imported from Brazil in the seventeenth century; José Redinha
(1968), pp. 96-7.
7 Even with manioc, farming techniques still are based on a strategy o f
maximal use o f ground moisture; reports in the archive o f the Missão de
36 T H E S E T T IN G
vegetable foods, the Mbundu added wild game, which individuals
pursued with bows, arrows, and traps throughout the year and which
large groups of men hunted with fire towards the end of each dry
season. The dry months also provided opportunities for fishing
while lowered water levels in the rivers confined fish to a few acces­
sible pools. There is no reason not to assume that most Mbundu
tended chickens, goats, and pehaps some sheep; cattle probably
grazed only in the high elevations south of the Kwanza since tsetse
fly made herding unreliable elsewhere.
The Mbundu had a form of iron-age technology based mainly on
local supplies of ore. Ore deposits were found both north of the
Kwanza in the Nzongeji river valley and south of the river in the
hills which ascended toward the Benguela plateau. Some iron may
have reached the eastern Mbundu from the Cokwe/Lwena east of
the Kwango,8 and the southern Mbundu doubtless had access to
supplies of iron lying in Ovimbundu territory near Andulo and the
modem town of Teixeira da Silva.9 Salt, which seems to have been
second only to iron in its importance for iron-age African societies,10
came both from the sea and from saline marshes scattered through
the interior. One source of maritime salt, the Kakwako lagoons just
north-east of Luanda island, was well developed by the sixteenth
century, but it lay within the area of Kongo suzerainty and available
evidence does not show whether its production went north to Kongo
or east up the Bengo/Nzenza to the Mbundu.11 The Mbundu had a
more likely supply of salt south of the Kwanza in Kisama; the first
Portuguese to travel there reported that the Mbundu traded this salt
throughout much of the interior.12 Salt also came from marshes
8 Testimony o f Mwa Ndonje.
9 Near the sources of the Ituvo river (the modern town o f Teixeira da Silva)
and along the left-bank tributaries o f the upper Kwanza near Andulo; David
M. Abshire and Michael A. Samuels (1969), p. 300.
10 Brian M. Fagan (1969) emphasizes the early importance of salt and iron
in Rhodesia. It seems fair to generalize his hypothesis to the Mbundu in the
light o f the data presented in the succeeding chapters o f this study.
11 The appearance of a trade route running up the Bengo based on slave
exports by the second half o f the sixteenth century suggests that the Kongo
on the coast may have had earlier contacts with the Mbundu who lived up­
river; Miller (1971).
12 Antonio Mendes to Padre Geral, 9 May 1563 (B.N.R.J., 1-5, 2, 38);
Brdsio (1952-71), ii, 495-512; also AA, ser. 2, xvii, nos. 67-70 (1960),

Inquéritos Agricolos de Angola in Luanda. Testimony of Sousa Calunga, 27

July 1969, for the Baixa de Cassange area. José Redinha (1958), p. 228, con­
firms that early villages in Angola had been concentrated along the river
valleys since neolithic times.
 E T H N O - L I N G U I S T I C SU BD IV ISIO N S 37
located somewhat farther east in Libolo, and the Mbundu were using
this salt as a medium of exchange on both sides of the Kwanza as
late as the end of the eighteenth century.13 The Baixa de Cassanje
contained two important centres of salt production. The Lutoa, a
tributary of the middle Lui, supplied salt to many of the eastern
Mbundu and to the Colcwe/Lwena to the south-east; the Kihongwa
river, an affluent of the Luhanda, fed extensive salt marshes along
its lower course, and residents of that area exported their production
to the northern Mbundu and even to the eastern Kongo in the
sixteenth century.14 These sources of salt and iron probably formed
the nodes of a complex set of regional trading networks which
brought the Mbundu into contact with one another and with their
neighbours on a regular basis. Certainly by the seventeenth century,
copper from Katanga and palm cloth from the forested regions to
the north were reaching the Mbundu, and there is no reason to doubt
that other similarly wide-ranging economic contacts had developed
long before the arrival of European traders on the coast.15

Ethno-linguistic Subdivisions in the Sixteenth Century

The outlines of sixteenth-century Mbundu ethno-linguistic sub­
groupings are even less distinct than the nature of early economic
activities in the Kwanza valley. Murdock’s judgement, delivered in
the late 1950s, that the Mbundu were ‘among the least adequately
described [people] in the entire African continent’16 remains true
today for the present characteristics of the people but applies with
special force to earlier centuries as well, since the intimate contacts
of some Mbundu with literate observers since before 1600 generally
failed to provide meaningful ethnographic data. Linguistic studies on
their language, Kimbundu, research on Mbundu material culture,
and information on their social and political institutions remains
at a lamentably low level despite the intermittent efforts of amateur
and professional Portuguese ethnographers over the last three
centuries.17 As a result, neither the external boundaries of the
13 R. J. da Cunha Mattos (1963), p. 317.
14 Miller (1973a) provides evidence for this trade route.
15 In general, see David Birmingham (1970) and Jan Vansina (1962c).
16 Murdoch (1959), p. 292.
17 The first recorded ethnographic data on the Mbundu appeared in the
correspondence of the Jesuit missionaries who went to Angola in the 1560s;

14-27. António Leite de Magalhães (1924), map and p. 5, gave the exact
location as near Ndemba, some 40 kilometres south-west of Muxima by
road.
38 T H E S E T T IN G
Mbundu nor variations within the Mbundu area are well known. My
field research has clarified some aspects of Mbundu ethnography,
especially that of the people living in the southern Baixa de Cassanje,
and has suggested the need for major revisions in conventional atti­
tudes toward much of the remainder.
Most ethnographic classification schemes distinguish the entire
Mbundu group of peoples only from such equally gross categories
as the Kongo to the north, the Colcwe/Lwena in the east, and various
Ovimbundu groups to the south. These distinctions rest primarily on
linguistic differences since linguists have been able to identify
Kikongo, Kicokwe, and Umbundu as separate languages from
Kimbundu, although Kikongo and Kimbundu seem more closely
related than any of the others.18Languages exhibiting features of both
Kikongo and Kimbundu (such as Hungu, Ndembu, and Soso)
blur the hypothetical dividing line between these two groups.19
Kimbundu dialects of a similarly transitional nature (Shinje and
Minungo) may bridge the gap between the eastern Mbundu and the
neighbouring Colcwe/Lwena in the same way.20 An almost total
absence of precise information on the Kimbundu dialects spoken
south of the Kwanza obscures the nature of the linguistic border
there, but there are some indications that it lies near the Longa
river since the Libolo variety of Kimbundu becomes unintelligible
south of that line.21 According to modern linguistic evidence, then,
the western Kimbundu-speakers live roughly between the Longa
18 Malcolm Guthrie (1967); Kikongo is his Zone H, Group 1; Kimbundu
is Zone H, Group 2. The Umbundu languages are in Zone R, Group 1;
Colcwe/Lwena falls in Zone L. In his later Comparative Bantu (1967-72), iii,
Guthrie indicates his uncertainties about the classification he has adopted for
the subgroups o f the Kimbundu language group. His caution appears well-
advised in the light of my data.
19 Guy Atldns (1954, 1955).
20 Based on comparison o f 200-word basic vocabularies (D. H. Hymes
(I960)) of Mbangala, Shinje, Minungo, Cokwe (western), Songo, and Mbondo.
211 know of no Kisama word-list recorded before the twentieth century; see
Mattenklodt (1944), pp. 106-7, for the 1920s. A short 32-word vocabulary
(including 10 numbers) is the only known information on the Libolo languages
(‘Libolo, Seles, N ovo Redondo, Benguela Velha, Amboim, Quibala, and
Gango’); Leite de Magalhães (1924), pp. 55-7 (reprinted in José Ribeiro da
Cruz (1940), pp. 166-7). For the linguistic boundary along the upper Longa,
Paes Brandão (1904), p. 226, and António Miranda Magalhães (1922), p. 11.
Redinha did not specify the sources on which he based his conclusions.

see Gastão Sousa Dias (1934). Modern ethnographic descriptions still rely
basically on such nineteenth-century compilations as José Joaquim Lopes dc
Lima (1846) and José de Oliveira Ferreira Diniz (1918).
 E T H N O - L I N G U I S T I C SU B D IV ISIO N S 39
river in the south and the Bengo/Nzenza in the north; to the east,
their limits run from approximately the Luhando river in the south
to the lower Kambo in the north.
These external boundaries appear to have changed significantly
since the sixteenth century only in the west, where the Kimbundu
language area now extends to the Atlantic ocean south of the Bengo
near the city of Luanda and to the region of Kisama beyond the
Kwanza river.22 A variety of evidence suggests that this was a rela­
tively recent development since speakers of Kilcongo lived near
Luanda during the sixteenth century; the language of the region
would have changed after Europeans brought large numbers of
K imbundu-spealcing slaves to the coastal plain during the seventeenth
century and after.23 The inhabitants of Kisama still spoke a variety
of Umbundu as recently as the latter part of the eighteenth century24
before epidemics of sleeping sickness at the beginning of the present
century sharply reduced the original Kisama population25*and prob­
ably contributed to linguistic change through repopulation by
speakers of Kimbundu. The southern boundaries of the Mbundu,
although greatly disturbed by political upheavals during the seven­
teenth century, do not seem to have moved. Changes in the north
brought by southward expansion of the Kongo kingdom seem to have
been largely complete by the sixteenth century and probably con­
tributed to the formation of such currently transitional groups as the
Ndembu and the Hungu.25
More extensive changes seem to have affected the internal sub­
divisions of the Mbundu, since their aetiological tradition, which may
be presumed to reflect historic rather than modem divisions within
the group, does not correspond well to the present distribution of
dialects and lineages.27 In general, wider gulfs divided Mbundu
from one another than is the case today. In terms of language, for
example, two dialects of recent origin, an eastern and a western one,
have incorporated a number of Portuguese words and are tending
22 See José Redinha (1961).
23 Miller (1972b).
24 Bernardo Maria de Cannecattim (1854), p. xv.
25 Heintze (1970), p. 170. Birmingham (1966), p. 145, implicitly confirms that
this linguistic change probably began somewhat earlier, since he notes that
Kisama had become a refuge for (presumably Kimbundu-spealdng) slaves
escaping from Luanda by the late eighteenth century.
2GFor the Kongo kingdom, with which I do not intend to deal here, see
Vansina (1966a), pp. 38-40, and Georges Balandier (1969), chapter I.
27 Analysed in my unpublished dissertation ‘Kings and Kinsmen’ (1972),
chapter III.
40 T H E SE T T IN G
to replace some of the older dialects. The western variant, centered
in Luanda, has resulted from the modern congregation of Mbundu
and other Africans from all parts of Angola in the city. They have
begun to identify themselves as ‘Alcwaluanda’ or ‘Ambundu’ to
distinguish themselves and their language from their rural relatives.28
The eastern pan-Kimbundu dialect, known as Ambakista, originated
in the growth of a Luso-Mbundu community of traders near the
Portuguese presidio of Ambaca on the middle Lukala during the
seventeenth and eighteenth centuries; their Portuguese-influenced
dialect spread as a trade language throughout much of the eastern
Mbundu area during the nineteenth century. Both these dialects
now serve as Kimbundu linguae francae and a pan-Mbundu life­
style with a pseudo-Portuguese component had in 1969-70 begun to
emerge as a result of shared experiences in the cities of Luanda and
Malanje, the conscription of young men from every corner of the
Mbundu region into the armed forces, and the spread of literate
education. The sense of a common ‘Mbundu’ identity, publicized
by the writings of a small community of Luso-Mbundu intellectuals
in Luanda,29 is of relatively recent provenance and should not be
allowed to contribute to the false assumption that a similar unity
prevailed in the past.30
Even the recent strong influences working towards the creation of
a single homogenized Mbundu language and culture have not erased
the local diversity surviving as evidence of earlier distinctions which
the Mbundu observed among themselves. Outlying Mbundu groups
have at least as much in common with their nearest non-Mbundu
neighbours as they share with distant Kimbundu-speakers.31 Despite
the paucity of solid ethnographic data, available evidence shows that
some of the most basic features of Mbundu culture and society have
always occurred in non-congruent distributions which overlap with
other people on all sides. These facts make it difficult to identify
any deep-running ethnic divisions in western Angola.
28 By far the most thorough sociological study of the African population of
Luanda is Ramiro Ladeiro Monteiro (1973).
29 Douglas Wheeler, in Wheeler and Rend Pelissier (1971), chapter IV, has
described the intellectual aspect o f this development.
30 It is well known that the first documentary mentions o f the Mbundu as
a named group come from the Kongo. This suggests that the term originated
as a collective referent used by outsiders.
311 have been unable to locate, for example, western Kimbundu-speakers
who can understand the Mbundu dialect o f eastern Kimbundu; Cokwe-
speakers find Mbangala almost!' as easy to discern as Mbundu bom and raised
in Luanda.
 E T H N O - L I N G U I S T I C SU BD IV ISIO N S 41
The Mbundu aetiological tradition outlines the historic distinc­
tions which sixteenth-century Mbundu drew among themselves.
The major groups included the Lenge, the Ndongo (a subgrouping
and not simply an alternative collective name for all Mbundu as
most authorities have claimed), the Songo, the Mbondo, the Pende,
the Hungu, and the Libolo. According to the professional Imbangala
historians, the westernmost Mbundu, the Lenge, lived in the mount­
ains between the Nzenza (or Bengo) and Kwanza rivers. Their
lands included the iron-workings located in the Nzongeji river
valley. The Ndongo seem originally to have occupied only the
highlands which form the upper drainage basin of the Lulcala river
and the Wamba. The northernmost Mbundu,' the Hungu, lived on
the south bank of the Kwale river as far east as the Kwango. The
so-called Pende had one of the largest territorial distributions of
any of the ancient Mbundu subgroups, extending all over the
northern Luanda plateau east from the Lenge to the entire Baixa de
Cassanje. The Songo lived beyond the Kwije river on the southern
part of the plateau, reaching from the rim of the Baixa de Cassanje
to the sources of the Luhando. The Libolo included a variety of
little-known people on the south side of the Kwanza and may per­
haps have been divided into the Libolo proper (west of the Luhinga
river) and the Halco (east of the Luhinga).
Without anticipating the political history to follow, it may be
worth while briefly to trace subsequent changes in this very early
ethno-linguistic pattern, probably dating to some time before the
fifteenth century. The Hungu absorbed infusions of Kongo culture
as the Kongo kingdom expanded east and south and as such Kongo-
related groups as the Sulcu formed east across the Kwango.32
Influences from Libolo penetrated north across the Kwanza and
helped to create a new Mbondo ethno-linguistic subgroup from Pende
who lived on the high plateau just west of the Baixa de Cassanje.
Then a series of Ndongo ideas and institutions from the north
moved toward the Kwanza, absorbing some of the original Pende
population along the lower Lulcala, causing the Lenge to lose their
distinctive identity almost completely, and overspreading some
32 For the Suku, etc., see Vansina (1966a), pp. 203-4. This would have taken
place before the sixteenth century, since the mani Kongo was already dis­
tinguishing himself as ‘lord’ of the Suku (among others) by the 1530s. This
title made an implicit contrast with the position o f ‘king’ which he claimed
within the integral Kongo provinces that he ruled directly. See Letter from
Rei do Congo, 12 February 1539 (A.N.T.T., Corpo Cronológico, 1-64-25);
Brásio (1952-71), ii. 70-2.
42 T H E S E T T IN G
northern Libolo areas as well. All these changes occurred before
the arrival of Europeans.
The seventeenth and eighteenth centuries witnessed further shifts
in the identity and location of Mbundu ethnic subgroupings. These
may be mentioned here briefly only to connect the earlier patterns
to the present distribution of the Mbundu peoples. The western
Lenge/Ndongo came under direct Portuguese administration during
the 1620s, thus beginning the process of homogenization and
assimilation which culminated in the emergence of the Ambaldstas
in the nineteenth century. The remaining Ndongo, who lived mainly
beyond the upper Lulcala, had achieved a degree of political unity
centred on the successor state to the Matamba kingdom ruled by the
famous seventeenth-century Queen Nzinga (by then located in the
ICambo river valley). They acquired the name of ‘Jingas’ from the
Portuguese, who tended to call many Africans after the titles of their
rulers (in this case, Nzinga). The Hungu broke out of their original
homeland during the later years of the eighteenth century and,
thoroughly Kongo in language and culture by that time, expanded
westward as far as the headwaters of the Dande river.33 The eastern
Pende fragmented into several new groupings, mainly Holo and
Imbangala gathered into a pair of strong kingdoms in the Baixa de
Cassanje, and a variety of small populations living along the lower
Lui in the interstices between the states. The latter acquired such
ethnonyms as ICari and Palca. Some of the original Pende may have
abandoned the Kwango area and moved to their present homes in
the Bandundu province of Zaire.34 The incipient split between Libolo
and Halco became more pronounced, and only the Songo remained
relatively unaffected by change. Great numbers of people from all
parts of the Mbundu territories were taken to Luanda where they
became slaves and employees of a growing Portuguese community;
there they coalesced into the group now known as ‘Ambundu’ or
‘Luandas’.35

Selective Review o f Mbundu Social Structure

One useful way to describe Mbundu society is in terms of the many
and diverse institutions which it contained. Since nearly every
Mbundu took care to participate in the activities of as many of

33 This movement began in the 1760s and continued into the nineteenth
century; Birmingham (1966), pp. 150-2.
34 For the Pende, Vansina (1966a), pp. 95-7, and sources cited.
35 These names appear in Redinha (1961).
 R E V IE W O F M B U N D U SO CIA L ST R U C T U R E 43
these institutions as possible, and since people sometimes found
that their functions overlapped and conflicted with one another, the
plurality of organizations afforded most people an opportunity to
play one institution against another in order to advance their own
interests. Although nothing about this aspect of Mbundu society
distinguished it from any other social system in the world, I want to
introduce the concept explicitly at the outset because the potential
for conflict inherent in the presence of multiple institutions played a
key role in Mbundu political history. Understanding the pattern of
Mbundu political development in these terms does not require a
comprehensive ethnography, which would in any case be impossible
to write owing to the insufficiency of data. But it is both possible and
necessary to outline enough of Mbundu social structure to indicate
the characteristic tension found in many matrilineal societies
between the principle of reckoning descent and inheritance through
women but leaving most forms of authority in the hands of males
and to suggest the importance of ‘cross-cutting institutions’36
which united Mbundu across the social boundaries set up by the
dominant matrilineal descent groups.
The Mbundu by the sixteenth century were farmers who had lived
in compact village settlements at least since the introduction of
agriculture at an unknown time in the dim past. The village geo­
graphically expressed the residential dimension of the Mbundu
lineage structure since each settlement ideally centred on a group of
adult males belonging to a single descent group, or ngimdu (plural,
jingundu). Because the Mbundu were matrilineal, the adult male core
of a mature village tended to be composed of a senior group of full
brothers and a middle generation of their nephews, that is, children
of their sisters. Adult females belonging to other lineages lived as
wives with the nephews of the village, and the youngest generation
usually included the children of these marriages, who were all
members of their alien mothers’ jingundu and therefore not part of
the village lineage core. Widowed or divorced sisters of the men in
the oldest generation often returned to reside with their brothers.
This residence pattern had the effect of reinforcing the lineage
identity of the village by gathering its most senior living members
together in a single place. The junior members of the Mbundu
lineages, young married women and their children, lived dispersed
in neighbouring villages with the kinsmen of their husbands and
36 The idea and the term have both received systematic treatment in the
stimulating chapter by Horton (1971).
44 T H E SE T T IN G
fathers. Since they rarely held positions of lineage responsibility
and did not share directly in inheritance, their absence from the
village did not detract from the concentration of lineage activities in
the village and the strong solidarity of the lineage group.
The Mbundu lineage-villages each had their own lands (ishi) in
which the members of the ngundu collectively controlled access to the
soil for farming, to the streams for fishing, and to the meadows and
woods for hunting and for gathering fruits, berries, and nuts which
grew wild there. As crops exhausted the fertility of the soil and it
became necessary to open new fields elsewhere, individual farmers
moved their plots every few years from one part of the lineage lands
to another in a system of shifting cultivation. A lineage might invite
outsiders to live with them and to share their resources, but such
men always remained guests and could not pass on their personal
privileges to their nephews. The lineage authorized and directed the
communal fire hunts which took place at the end of each dry season.
The major units of economic co-operation therefore tended to coin­
cide with the Mbundu descent groups, just as did the residential
units. The collective jingundu remained stable through time, each
associated with its own fixed ishi.
Individual Mbundu moved from place to place, however, in a
pattern common to matrilineal societies where the nieces of a lineage
spent their child-bearing years away from their own kinsmen,
residing in the lineage-villages of their husbands and raising the
lineage’s children there. Girls were usually born in their father’s
lineage village, remained there until marriage, and then went to live
with the kinsmen of their husband. Frequent visits to the village of
their mother’s brother kept up their contacts with their own ngundu
and prepared the way for their move ‘home’ when they ceased to
bear children or became divorced or widowed. Most women never
lived in their own lineage village until they reached old age. Boys,;,
on the other hand, returned-to their, own ngundu's . village much
earlier in life than their sisters. They also grew up among their
father’s kinsmen but soon after puberty tended to return directly
to their own relatives at the village of their mother’s brothers, where
they remained for the rest of their fives as part of the core of matri-
lineally related males. Their wives and children, all members of other
jingundu, lived with them during their mature years but tended to
drift away as the women grew old, the daughters married, and the
sons moved away to join their own uncles. The constant movement of
people in these patterns, from the perspective of the descent groups,
 R E V IE W O F M B U N D U SO CIA L S T R U C T U R E 45
meant that each lineage’s members were born and grew up away
from their own lineage village but returned to it as they grew older,
the males somewhat earlier in their lives than the females.
Although the Mburtdu descent groups collectively tended towards
the sort of structure just outlined, individual ngundu moved through
a fairly predictable series of stages. A mature lineage typically had a
single elderly man in its oldest generation, with a number of fully
mature nephews (behwa, singular mwehwa, ‘sister’s sons’) generally
in charge of running lineage affairs. If more than one of these
nephews aspired to head his own ngundu, the death of the old man
caused the descent group to split as each mwehwa, or each set of
full brothers within the group of nephews, felt himself no longer
restrained by the unity imposed by the old man’s presence and left
with his wives, young children, and sister’s sons to establish a new
and independent ngundu. The new groups might divide the lands
formerly held in common, or some of the new jingundu might go
elsewhere to settle as guests on the lands of related lineages. This
sort of lineage fission constantly generated new descent groups but
seldom eliminated the old lineages as corporate groups, since one of
the nephews customarily took over the position of the dead uncle
and preserved the identity of the old ngundu. Lineages could thus
survive indefinitely, although misfortune might reduce their living
membership so drastically that the few remaining survivors would
scatter to live with other relatives. In such a case, the formal identity
of the ngundu might still remain, kept alive by related groups as a
memory even though it had no living members.
The ability of the Mbundu to preserve this sort of abstract ngundu
without living members had an explanation in their cosmological
system. The Mbundu, in common with the Cokwe, the Lunda, the
Ndembu, the Bemba, and others who lived to the east, visualized
their society as a set of named roles, personalized social statuses
associated with defined rights and obligations which living individuals
might temporarily assume to the exclusion of any other person.37
The Mbundu customarily described the relationship of each such
named position to other named positions in the language of kinship,
‘father-son’, ‘uncle-nephew’, ‘brother-brother’, and so on, and
regarded all such connections between existing roles as immutable.
This aspect of the resulting network of named roles has been termed
‘perpetual kinship’ in Central Africa, from the ldnship idiom used
37 Ian Cunnison (1956). Although these ideas are known elsewhere primarily
as political techniques, they permeate every area o f Mbundu social structure.
46 T H E S E T T IN G
to describe the social structure and the permanence of the relation­
ships between its elements. Individuals thus might take possession
of one or more of these permanent named roles, or ‘enter’ (Icuhinga)
the title, as the Mbundu said, hold it for a time, and then bequeath
it to a successor. Each position had a number of different incumbents
over time, all of whom took the name of the position as their own,
treated the occupants of ‘brother’ and other related positions as if
they were biological brothers and relatives, and exercised the rights
and duties attendant to the position. Hence, the term ‘positional
succession’ used to indicate the temporal dimension of the system
in which heirs succeeded to the named roles of their predecessors.
Descent groups in this system of perpetual kinship and positional
succession abstractly consisted not of individuals but of named
positions which existed independently of their living occupants.
Perpetual genealogies described the formal relationships between the
names in the lineages, with a set of closely related positions constitut­
ing an ngundu. The matrilineal descendants of the occupants of the
set of names remained responsible for providing living incumbents
in perpetuity. The Mbundu thus distinguished between the formal
structure of an ngundu, or the names, and the people who temporarily
filled the positions, and it was in this sense that related descent
groups might retain ‘empty’ ngundu positions when no living kinsmen
were available to occupy them. Lineages often had a few of their
names unfilled at any point in time, or an entire set of ngundu names
might become vacant if the descent line responsible for filling them
died out.
Certain names within each ngundu endowed their holders with
special responsibilities for the welfare of the group. These duties
attached to the senior positions in the ngundu genealogy which the
Mbundu referred to simply as the ‘uncles’ (,malemba, singular lemba)
of the lineage. Lineage members credited the first holder of one of
these senior titles with having separated the lineage from related
descent groups and usually believed that he had led their ancestors
to their present lands. The ngundu took its collective name from this
title, the lemba dya ngundu, and looked to its holder to perform many
of the rites which, they believed, assured bounteous harvests, guar­
anteed the fertility of the lineage women so that they might produce
future occupants for the lineage names, brought copious rains, and
attracted plentiful supplies of game. The lemba dya ngundu stood as
an intermediary between the living lineage members and dead occu­
pants of the same positions, the ancestors who collectively repre-
 R EV IEW O F M B U N D U SO CIA L ST R U C T U R E 47
sented the spiritual dimension, of each title. He also mediated
between the ngundu and the spirits of the lands and waters which
they owned. His was the position on which the social and ritual life
of the lineage turned. The occupants of a varying number of other
senior positions acted as councillors to the lemba dya ngundu and bore
the designation of malcota (singular, kota or dikota, literally ‘elder’).
If their overt responsibilities were somewhat less than those of the
lemba dya ngundu, their subtle political and social functions as
advisers and arbiters were perhaps greater.
This brief description of the structure and officials of the Mbundu
lineages leads to a fuller explanation of the lineage genealogies
discussed as historical sources in Chapter I. The names in these
genealogies were the names of the malemba of each ngundu. In the
context of the genealogies, the lemba represented all the subordinate
names related to him by separate internal ngundu genealogies (which
were also different from the relatively unimportant biological gen­
ealogies of lineage members which showed only biological, not
social, links). The lemba and makota granted kinsmen the right to
leave their ancestral ngundu to establish new independent groups;
when they did so, they awarded a new name to the relative at its head.
This name or title consisted of a distinctive first name followed by
the title of the senior lemba as a surname, and the new lineage would
thus be known through the name of its headman as descended from
the old ngundu: if, for example, the ngundu of Mahashi na Pakasa
split, it might lead to the creation of a new descent group called
Nzenza ya Mahashi, and each lineage would have headmen of the
same names. Lineage fission combined with the preservation of old
lemba-titles through the mechanism of positional succession to
generate Mbundu lineage genealogies relating all existing jingundu
through ties of perpetual kinship.
These descent groups probably constitute some of the oldest sur­
viving institutions of Mbundu social organization and may date
from the introduction of agriculture and the settlement of the ances­
tors of the Mbundu in permanent lineage-villages. In the absence of
thoroughgoing archaeological studies, this hypothesis is subject to
verification only through such arguments as the ‘age-area hypothesis’
which posits that ‘wider distribution generally means longer time
depth’.38 The widespread distribution of lineages like those of the
Mbundu, and of their associated symbols, would accordingly speak
38 Vansina (1970); the author adds that ‘this is not always the case; thus the
test is only partial’ (p. 169).
48 T H E SE T T IN G
for their great antiquity. The Mbundu traditions emphasize the
breadth of the single system of descent groups of which they are a
part by including genealogies linking them to people who lived as
far away as the Lunda of Katanga. Very similar lineage structures
seem to exist south of the equatorial forest as far east as the great
African lakes. The Mbundu lemba dya ngimdu, like his counterparts
all over the southern savannas, for example, used a sacred white
powder called pemba to insure the fertility of the women of his
lineage. He dispensed a red powder called talcula, of similarly wide­
spread occurrence, to the men of the ngundu.39 His authority was
closely associated with a mulemba tree which the Mbundu always
planted in front of his dwelling in the village.40 The exact symbolic
meanings of the mulemba in the Mbundu context remain uncertain,41
but the practical facts that branches from this tree rooted themselves
when placed in the ground and grew rapidly (aside from their possible
symbolic meanings) assumed some significance for shifting agricul­
turalists who moved their villages every few years. The lemba dya
ngundu and the makota valued the broad shade which it provided
for their deliberations.42
The jingundu commanded an intense loyalty from most Mbundu.
The crucial functions of the lineage in providing access to land, since
all arable territories lay in the domain of one lineage or another, in
coercing the rains to fall on time and in sufficient quantity, in mediat­
ing between the living and the dead, and in defining the individual’s
place in Mbundu society made it in many ways the fundamental
institution in their lives. The Mbundu accepted as fellow human
beings only people who had positions in their lineage system, either
as holders of one of the ngundu names or as formal affiliates (‘slaves’,
pawns, etc.) of some descent group.43 Not to belong to an ngundu
in theory excluded a person from the right to call on ldnsmen for
support, prevented him from marrying or growing food, denied him
spiritual solace, and in practice often amounted to a choice between
certain death and abject subordination to the will of a patron with a
place in the lineage structure. For all these reasons, the Mbundu
39 Powdered wood of the pterocarpus tinctorius according to W. D . Hambly
(1934), p. 117.
40 Ficus psilopoga; Chatelain (1894), p. 267, n. 171.
41 The same tree figures prominently in the complex symbolic system of the
Ndembu o f Zambia for example; Turner (1967), passim.
42 Testimony o f Sousa Calunga, 1 Oct. 1969; cf. Otto Schiitt (1881), pp.
84-5. The distribution of the mulemba as a symbol o f lineage authority is also
very broad; Jose Redinha (1963), p. 72.
43 Joseph C. Miller (forthcoming(b)).
 R E V IE W O F M B U N D U SO CIA L ST R U C T U R E 49
placed great importance on their lineages and evidently succeeded in
preserving their basic structure without great change over several
centuries.44
If an individual could survive in Mbundu society only within the
protective shield afforded by his ngundu, the lineage as a group could
survive only through close co-operation with its neighbours. The
rule of lineage exogamy forced lineage members to marry outside
the coniines of their descent group, and it may be assumed that a
variety of marrige alliances linked particular Mbundu lineages to
one another. I have almost no data to suggest the type or scope of
these arrangements, certainly not for the sixteenth century, but
those I have indicate that preferential marriage rules capable of
producing permanent wife-exchanging pairs of jingundu tended to
occur more commonly among those Mbundu who lacked the deep
segmentary-type genealogies of the Songo. The comprehensive Songo
lineage genealogy seems to have ordered inter-lineage relations there
without necessitating the additional links provided by clearly
developed preferred or prescribed marriage rules. It is therefore
impossible to say much about affinal alliances as means of structuring
inter-descent group relations other than that their importance may
have varied inversely with the comprehensiveness of the different
sets of lineage genealogies.45
From the perspective of an individual Mbundu, the formal
relationships between the jingundu as collectivities, whether genea­
logical ties resulting from lineage fission or affinal bonds created by
marriage, did not always provide a viable framework through which
he could pursue opportunities arising in areas where his network of
kinsmen and affines did not extend. The ties of personal kinship
derived from his group relationships could be manipulated within
limits to produce varying emphases within the structure, but they
did not necessarily reflect the personal interests which a man might
share with non-lcinsmen engaged, for example, in the same economic
activities as he. Mbundu society contained several other sorts of
institution which responded to such practical needs by uniting people

44 This is, o f course, the crucial assumption for the entire analysis to follow.
In the absence o f complete data for the seventeenth century, it must rest on
the fragmentary data we have to suggest that nothing important has changed
and on such arguments as the ‘age-area’ hypothesis.
45 Limitations on both time and the scope o f my research in Africa prevented
the collection of reliable information on Mbundu social practice, as contrasted
with Mbundu social theory. These practices, in any case, have clearly altered
enormously under the impact o f war and urbanization in modern Angola.
50 T H E S E T T IN G
across the social boundaries set up by the fierce loyalty of most
Mbundu to their descent groups. In structural terms, such institu­
tions cut across the dominant lineage structure of the society. Classed
according to duration and specialization of purpose, these cross-
cutting institutions ranged from quite informal and ad hoc agree­
ments made between strangers thrown together in common pursuit
of a limited goal to comprehensive and institutionalized secret
societies of several types. It is, of course, impossible to reconstruct
any of the more ephemeral institutions of the distant past, but
comparisons of modern Mbundu institutions with similar organiza­
tions among related peoples give a good idea of the general char­
acteristics of those which must have been present long ago.
Skilled professionals of various sorts—especially the diviners and
healers (nganga, plural jinganga, or kimbanda, plural yimbanda)—
maintained informal but intensive relations with one another regard­
less of their lineage affiliations. This sort of professional solidarity
enabled successful practitioners of these arts to travel far beyond
their spheres of kinsmen and to receive a respectful welcome from
local colleagues and their relatives wherever they went. Even an
informal network of mutual professional interests provided individual
Mbundu with a supra-descent group network of contacts and served
to transmit skills and knowledge widely throughout the society.
Diviners seem to have favoured their sons over their nephews as
heirs to their professional status, and to that extent they set them­
selves apart from the dominant emphasis on matrilineal group
membership.46
More structured, but also more ephemeral, were curing cults and
witchcraft eradication movements.47 These provided institutional
vehicles through which people could temporarily abandon their
primary loyalty to their ngundu in favour of ties to non-kinsmen
based either on common affliction with a given disease or on a
common effort to eliminate witches from their midst. Among the
Mbundu, these rituals characteristically involved techniques of
spirit possession, a notion very closely related to the theory of
positional succession, in which the occupant of a name sought
ritually to identify himself with the spiritual essence of the position
and through it to communicate with some or all of its previous

46 Cavazzi (1965), passim but esp. i. 193-200, gives a number o f details

suggestive o f this conclusion for the 1650s.
47 Here I draw primarily on the terminology and concepts of Victor Turner
(1968) and Jan Vansina (1971a).
 R EV IEW O F M B U N D U SO CIA L ST R U C T U R E 51
incumbents. These operated across the lineage boundaries and had
quite different consequences for Mbundu social structure from other
spirit mediumship techniques (the Jcushingilisa, see below) which
worked exclusively within the lineages and tended to reinforce the
solidarity of the individual descent groups.
Mbundu neighbourhoods held regular circumcision camps in
which the young men of a locality joined together regardless of their
positions in the descent group genealogies. They formed a kind of
incipient age-set, united temporarily by their shared experiences in
the circumcision camp, and retained limited ties throughout their
lives. Aside from the educational function of the ceremony, and its
certification of Mbundu boys as young Mbundu men, it also provided
cross-cutting ties among the young males residing locally with their
fathers. When these men dispersed after the initiation ceremonies
to join their mothers’ brothers in other, perhaps distant, villages,
they created a web of association which ultimately stretched over a
wide area. The graduates of the circumcision camp carried special
circumcision names which indicated their ties to each other in a
manner analogous to the permanent names and genealogies of the
jingundu.
An association of master hunters (yibinda, singular kibinda)
provided a good example of how these associations created personal
ties which united people outside the structures of kinship. The
permanence and close connection of the kibinda society with some
later forms of Mbundu political authority make it worthy of more
extensive description than some of the others. The kibinda was a
hunter skilled not only in the use of bow, arrow, and spear, which
he used to kill hippopotamus, lion, leopard, wild boar, large ante­
lope, red buffalo, and crocodile, but also a specialist believed to
know the magical arts of making himself invisible to his prey, flying
through the air, or wielding charmed weapons which never missed
their mark. Whatever the methods by which the kibinda stalked
and killed his quarry, he performed several functions essential to
the welfare of the ngundu. His was the responsibility of entering the
forest in. search of certain animals believed necessary to divine the
intentions of lineage spirits or to augur the success of some con­
templated enterprise. If the kibinda hunted well, lineage members
could rest assured that their relations with the spiritual world were
harmonious, but if no game appeared they took his bad luck as a sign
that they should discover the source of the apparent supernatural
displeasure. In this and other ways, the activities of the yibinda
52 T H E S E T T IN G
master hunters reinforced the integrity of the Mbundu descent
groups.
A different aspect of the kibinda association, however, united all
Mbundu professional hunters to one another regardless of their
lineage status. An aspirant hunter (mona a yanga, or ‘child of the
kibinda’, who was also known as a yanga) could apprentice himself
to any master for training in the professional hunter’s arts. The
relationship between mona and master was regarded as analogous
to that of son to father, that is, cutting across the matrilateral ties of
uncle to mwehwa. On the occasion of the death of a famous kibinda,
yibinda and bana a yanga from far and near assembled to participate
in ceremonies which culminated in the extraction of a tooth from the
dead hunter’s jaw and a communal hunt. The bonds uniting yibinda,
who recognized each other by secret signs, extended far beyond the
limits of kin and even ethnicity to include Songo, Pende, Imbangala,
Cokwe, and Lunda alike. To meet anothe kibinda in the forest was
said to be like encountering a kinsman, and the status of master
hunter obliged a kibinda to extend to his colleagues all the benefits
he derived from his own ngundu.46 These strong ties facilitated the
movements of hunters who often penetrated unknown regions in
pursuit of large game and provided a potent mechanism for inte­
grating Mbundu society beyond the framework afforded by the
lineage structure.
Finally, some indication of Mbundu ideas about the nature of
authority may help to prepare the ground for the following examina­
tion of early Mbundu political history. The Mbundu distinguished
clearly between the locus of authority and the wielder of authority,
much as they made a distinction between the abstract named roles
of the social structure and the living incumbents. Authority rested
on the ability to invoke supernatural sanctions and inhered not in
human beings but in the authority emblems associated with the
titles. Sometimes the Mbundu expressed this idea by emphasizing
the intimate connection between a living incumbent, powerless in
himself, and the dead predecessors in his title who collectively ‘were’
the authority of his position. Names carried varying measures of
authority of this sort, and the most influential names in the lives of
most people before the appearance of states were the titles charged
with responsibility for the welfare of the descent groups, the malemba
dya ngiindu and the makota. Renowned professional hunters, diviners,
48 Testimonies of Alexandre Vaz and Ngonga a Mbande, 23 Sept. 1969;
Sousa Calunga, 2 Oct. 1969.
 R EV IEW O F M B U N D U SO CIA L ST R U C T U R E 53
mediums, and the like primarily exercised their authority not over
people but over animals, oracles, or spiritual forces.
The holder of a powerful name usually gained access to the spiritual
forces behind it through possession of some object believed to
mediate between the visible world of the living and the invisible world
of the supernatural. The lemba dya ngundu, for example, was helpless
without the mulemba tree or his pemba, just as diviners worked
through a variety of physical objects which they endowed with
special powers, and the kibinda master hunter achieved his success
through possession of his master’s tooth or through the manipulation
of carved figures, animal horns, or plants. All of these mediatory
objects also served as visible badges of office, emblems of the special
status accorded to the possessor of the name or title to which they
belonged.
The notion of authority as access to spiritual forces obtained
through possession of special objects meant that any Mbundu in
control of an authority emblem could delegate a portion of his power
simply by awarding a part of his object to another. This kind of
logic lay behind the ceremonies which marked the fission of the
lineages; the founder of the new ngundu usually took a branch from
the mulemba tree of the parent lineage and planted it, along with the
spiritual essence of the descent group as well as the authority of the
lemba, wherever he settled in his new village. Hunters initiated
aspirants into kibinda status by giving them charms which presum­
ably gave them access to the same magical powers which had given
the master his own success. Diviners transferred their skills not only
through instruction but also through the presentation or sale of
objects which gave their heirs and customers more or less limited
access to the same secrets as they possessed. On a more ephemeral
level, the lemba dya ngundu could appoint agents empowered to act
in his stead simply by giving them a material symbol of his authority.
The formation of the first Mbundu states occurred in a social
context not unlike the one just described: strong and independent
matrilineal descent groups which tended to monopolize the loyalties
of most individuals. The cohesiveness and autonomy of the lineages
diminished the importance of institutions which cut across the
social boundaries set up by the rule of unilineal descent, but a
number of such institutions blurred the lines separating kinsmen
from non-kinsmen. Of these, the most durable and extensive seems
to have been the association of kibinda hunters. To the extent that
these institutions allowed men to circumvent the rigidity of descent
54 T H E SE T T IN G
groups based on positional succession and perpetual kinship, they
broadened the scale of interaction among the Mbundu and afforded
opportunities for people to co-operate towards ends not met by the
lineage structure. In so far as some of these institutions tended to
endure through time, they resembled ‘state-like institutions’ which
performed ‘political’ fbnctions. The remainder of the book examines
the ways in which the Mbundu developed such incipiently political
institutions as these through the distribution of authority symbols
and delegation of authority into several political structures which
closely resembled ‘states’ by the beginning of the seventeenth century.
 C H A PT E R III

Indigenous Beginnings

T he w o r l d , according to the Mbundu, began when Ngola Inene

arrived from lands far to the north-east and settled where the
Mbundu live today. Ngola sired a daughter, Samba, and Samba in
turn gave birth to ICurinje lcwa Samba and Kiluanje kya Samba.
Kurinje lcwa Samba (leaving aside momentarily the descendants of
Kiluanje kya Samba) was the parent of Mbulu wa Kurinje and
Mbulu wa Kurinje produced Zundu dya Mbulu, Kongo dya
Mbulu, Mumbanda a Mbulu, Matamba a Mbulu, Kajinga ka
Mbulu, Mbumba a Mbulu, and perhaps Kavunje ka Mbulu, the
founders of the Ndongo, Hungu, Pende, Lenge, Mbondo and
Imbangala, the Songo, and the Libolo. The world, also according to
the Mbundu, began when the ancestors of the same present-day
ethno-linguistic subgroups came with malunga from the sea and
stopped when they reached the hills and valleys where their descend­
ants are now to be found. Others add that the world may have
begun with Adam and Eve, two people who lived very far away and
who had many descendants. Among them was Cain, father of all the
black people.
For most Europeans, on the other hand, the origins of the world
are too remote to consider in connection with the formation of the
Mbundu, and it is easier to believe that several local stone- and iron­
working populations, known only from the artifacts they left
behind in long-abandoned campsites, have occupied north-western
Angola for many centuries. No less contradictorily than the con­
flicting stories of Ngola Inene and the malunga from the sea, the
Bantu language the Mbundu speak today links them to people who
now live far to the north-west in modern Nigeria. In the face of the
archaeological and linguistic facts, what, if anything, do the tradi­
tions about Ngola Inene and the malunga reveal about Mbundu
history, as opposed to mere legend? The most plausible answer is
that the traditions refer not to the origins of the people but, like the
Mbundu adaptation of the story of Cain and Abei, indicate the
appearance of new modes of political organization and are therefore
germane to the question of how Mbundu descent groups first
56 I N D I G E N O U S B E G IN N IN G S
organized themselves according to standards other than those of
lineage fission.
•The traditions of Ngola Inene and the malunga clearly belong to
‘historical time’ for the Mbundu rather than to the preceding ‘myth­
ical’ and ‘proto-historical’ periods found in the oral art of many
African peoples.1 Many non-literate societies envisage their remote
past as falling into three stages, moving from a mythical epoch when
inverted monsters roamed the world through a transitional age of
supermen into a fully historical period populated by humans not
unlike modern men.2 The Mbundu make no attempt to link whatever
philosophical beliefs they may hold about the creation of the world
to the relatively straightforward accounts of how their social and
political structures took form. They clearly regard both Ngola Inene
and his progeny and the people who brought the malunga from the
sea as very much like themselves. Even though some of these proto­
typical humans knew how to make strong magic, none of them
walked on their heads, as more mythical ancestors in other societies
are said to have done, or otherwise signalled their non-humanity
by grotesque or shocldng behaviour. Their essentially human quali­
ties suggest that they represent authentic historical events and stand
for more than a statement of present Mbundu cosmology in a
mythical idiom. The mythical and heroic ages of the past usually
purport to describe the formation of present social structures through
presenting a contrast between a former period of chaos and the
emergence of orderly modern social arrangements.3 The essential
humanity of Ngola Inene and the maliinga-bnngers, if this is so for
the Mbundu, thus signifies that the Mbundu believe that the outlines
of their present social institutions were already present at the time
their historical traditions begin.
Even though the shortage of information on the earliest phases of
Mbundu political history leaves an element of conjecture in any
interpretation of this period, certain continuities in the structure of
their society enable the historian at least to delineate the areas where
reliable reconstructions are possible and to acknowledge those
parts of the Mbundu past which remain shadowy or entirely un­
known. Using ethnographic data collected from the Mbundu and
11 employ the terminology o f E. V. Thomas and D . Sapir (1967).
2 Frank E. and Fritzie P. Manuel (1972), pp. 84-90, illustrate this sequence in
Greek mythology.
3 One can appreciate such analyses as John Middleton (1965), pp. 18-24,
or Beidelman (1970) without condemning all historical traditions to the never-
never land o f ‘ideological data’ (Beidelman, p. 96).
 IN D I G E N O U S B E G IN N IN G S 57
their neighbours during the last century, one can say with fair
certainty that one form of change in their society before 1600
consisted of frequent and small-scale innovations in social organiza­
tion, ideas constantly appearing and flickering briefly within the
limits of a single descent group or confined to a few lineages before
they died out and left the aggregate pan-Mbundu social institutions
largely unaffected. New religious beliefs, for example, spread in this
manner and then vanished; lineage headmen invented and tested new
magical techniques which they hoped might bring them copious
rains or numerous nephews. Men everywhere made formal but
temporary arrangements with their neighbours or acquaintances by
fabricating and displaying visible tokens of their intentions: an
amulet, hats, skins, bracelets, pieces of iron, carved wooden figures,
staffs, bells, gongs, and a variety of other emblems.
Since most such innovations failed, as the disparity between the
high hopes they raised and the inadequacy of their means bound them
to fail,4 the great majority had no lasting effect on Mbundu society
except for a minor residue in the form of an obscure charm in the
magical armoury of a lineage official or a new feather in the cap of
a title-holder. These remain lost to history forever. Yet some in­
novations remained and became fixtures of Mbundu social structure,
providing new links between people where none had existed before,
reinforcing the authority of certain officials at the expense of others,
and cumulatively propelling the Mbundu through a sequence of
historical developments which in retrospect seems to possess an
identifiable structure, to define a recurrent pattern. The beginnings of
Mbundu political history therefore consist of identifying and placing
in the proper chronological order the most important innovative
techniques of social organization, the ones which permanently
affected M bundu.social structure and have thus encouraged the
conservation of the corresponding physical symbols as bits of
evidence from the past. All else must remain unknowable.
The multiplicity of symbols now in use among the Mbundu shows
that successive generations of officials have eclectically incorporated
new emblems without abandoning the older authority symbols of
their ancestors even after their faith in the power of the antique
relics had begun to wane. The precise meaning and use of each
emblem undoubtedly has varied through time. Many of the present
minor symbols of authority, for example, linger as faded remnants
of objects whose reputations once inspired awe in the hearts of all
4 A point mentioned by Jack Goody (1971).
 58 IN D I G E N O U S B E G IN N IN G S
Mbundu. Title-holders accumulated whole sets of such symbols,
adding to their collections as they claimed new powers for themselves
in response to changing times much as Western law rests on a thick
sediment of statutes reflecting the circumstances of times past. The
outsider may thus read the panoply of objects associated with present
Mbundu title-holders as a record of techniques of governance tried
in the past.
The Mbundu oral traditions describing their ‘origins’ in terms of
Ngola Inene and the mahmga explain the provenance and meaning
of two important symbols of this type. They are thus the conscious
record of political development, couched in the personalized idiom
of human founders but meant to account for two of the numerous
authority emblems in the hands of present lineage headmen and
holders of political titles. It is because of this that lineages whose
headmen today care for an authority emblem called a lunga (plural,
malunga) recount the traditions about ancestors who came from the
sea. Histories of Ngola Inene explain the central sacred relic of
descent groups which attribute their headman’s powers to a small
piece of iron called an ngola (plural, jingola). And, true to the pattern
of explaining ‘origins’ in terms of whatever protective devices the
lineage may regard as most fundamental, Mbundu adaptations of
Biblical origin myths tend to appear in villages situated on mission
property or those counting a substantial proportion of Christians
among their members.
The superficial contradictions found in multiple myths about
Adam and Eve, malunga, Ngola Inene, and others are not to be
resolved by selecting a single Mbundu ‘origin’ as definitiye. They
i result from a ,long and' complicated history in which a number of
different principles of socialand political organization have had
different effects on the present Mbundu lineages. The available data,
complex as they are, still consist only o f a small part of the total
history: those authority emblems and associated traditions which
have survived the passage of the years to remain as respected symbols
at present. Presumably, these are the ones which achieved the widest
diffusion and provided sufficient advantages for the people who
adopted them to integrate these emblems at the heart of their lineage
structure. Although such symbols may have been the ‘most important’
in this sense, a great many other paired symbols and organizational
techniques failed to spread or achieved only temporary significance
before they were abandoned in favour of newer, more satisfying,
or more timely ideas. The processjof continual invention, diffusion,
 T H E C O M IN G O F L U N G A 59
alteration, and disappearance of religious movements among a
twentieth-century people with similar social institutions, described
for the Kuba of the Kasai province of Zaire in fascinating detail,5
cannot differ significantly from the attitude of social and religious
experimentation which must have animated the Mbundu many
hundreds of years ago and which still motivates them today despite
the stultifying effects of colonial rule on fluid African social institu­
tions. Movements of this type may hold the key to the appearance of
states among the Mbundu.
Since the available records show only a fraction of the social and
political institutions which the Mbundu have tried, the present
discussion is limited to two developments which the Mbundu
remember as the most fundamental since the coalescence of the basic
jingundu matrilineages and their acceptance of the mulemba trees
and the lineage headmen (the malemba dya ngundu) associated with
them. Both these emblems, th e . lunga and the ngola, at different
times seem to have supported two contrasting movements which
underlie Mbundu political history: tendencies toward particularism
and localism centred in the lineages, and opposing attempts to trans­
cend the level of lineage organization through the development of
larger-scale centralized political structures.

The Coming o f the Lunga

The Mbundu lineages 'living on the middle and northern reaches of
the Lui river revere a kind of lineage emblem called a lunga, a sacred
relic which assumes various physical shapes but usually has taken the
form of a human figure carved from wood. These malunga, the
Mbundu explain, came originally from ‘Kalunga’ which they
identify as the ‘great water’, without any clearmotion of where this
aquatic source may lie. Because almost all European observers
have interpreted the word as referring to the Atlantic Ocean, or to
the African ‘great lakes’, the Mbundu today also give this meaning
to the name ‘Kalunga’ and believe that their ancestors who long ago
brought the first malunga set out from the island of Luanda where
the Portuguese have had an administrative centre for nearly 400 .
years.6 There is obviously no basis in fact for this assertion. Malunga
today have a close association with rain and with water, ‘dwelling’
in rivers and lakes and helping their guardians call the rains. They
are linked also with the success of agriculture and hence with life
3 Vansina (1971a).
6 A published example o f this sort o f tradition appears in Haveaux (1954).
60 IN D I G E N O U S B E G IN N IN G S
itself in the Mbundu farmer’s view of the world. The ancestors who
first came with the malunga were ICajinga lea Mbulu, who founded
the Mbondo, Matamba a Mbulu, ancestress of the Lenge, Mum-
banda a Mbulu, mother of all the Pende, Kongo dya Mbulu who
gave birth to the Hungu, and Zundu dya Mbulu, great ancestress
of the Ndongo, and perhaps others now forgotten. An outsider
might also observe that the malunga had something to do with an
early organization of the Mbundu lineages into the ethno-linguistic
subgroups recognized today.
Several characteristics of the malunga suggest that they spread
among the Mbundu lineages at a very early period. Their close
association with water and with the land, since lineages with malunga
are also the land-owning descent groups, points to their connection
with the oldest significant stratum in the present Mbundu social and
political structure. The broad grouping of the Mbundu into Ndongo,
Lenge, Pende, and so on was already present at least as early as the
late sixteenth century when the first documentary sources picked
up some of these names. The Mbundu expressly concur in the
opinion that the landowners are the most ancient ldn groups in their
territories. The /z/nga-holding lineages, where they survive, govern the
use of the land, authorizing changes in residence, selecting the sites
for new villages, locating water by divining techniques based on
manipulation of the physical /nnga-object, as well as summoning the
rains'at the end of each dry season.7 The lunga has in some cases
become assimilated to the complex of symbols centred on the
mulemba tree and the lemba dya ngundu, since the human guardian
of the lunga distributes the lineage pemba to his kinswomen and
performs other duties normally associated with these officials. The
female identity of the ancestors who brought the malunga reinforces
their connection with the lineages since the Mbundu perceive their
descent groups as feminine in contradistinction to most extra­
lineage institutions which they see as ‘masculine’.
The extensive modern distribution of ancestors and state-founders
with names based on the root -lunga also gives the impression of
great antiquity. These figures appear in a wide band through the
southern savannas in association with very ancient forms of auth­
ority, from the Kuba where ‘Keloong’ (Kalunga) is said to come
from very ancient times,8 through the Luba where traditions portray
Kalala Ilunga as the founding ancestor who brought a new political
7 Testimony o f Kimbwete,
8 Jan Vansina, personal communication.
 T H E C O M IN G OF L U N G A 61
order in the form of the ‘second Luba empire’,9 to the southern end
of Lake Malawi where traditions have ‘Kalonga’ arriving near the
beginning of political history with new institutions brought from
Katanga.10 Closer to the Mbundu, the Yalta peoples, who live just
down the Kwango from the eastern Mbundu, include certain very
old Kalunga ‘clans’, and the Cokwe to the east of the Kwango
recognize a Kalunga as one of the founding ‘parents’ of their line­
ages.11 The malunga survive as, dominant lineage emblems within
Mbundu territory primarily among the remnant Pende populations,
the Hari and the Palca of the lower Lui who seem to have escaped
incorporation in any of the kingdoms which supplanted the lunga in
most other regions. This sort of evidence is too fragmentary to
suggest a point of origin for the lunga,12but the very vagueness of the
data testifies to its great age since subsequent political developments
have intervened to obscure its history in most areas.13
The modern characteristics of the lunga, where it has survived
without significant change, suggest something about the way ancient
Mbundu lineage headmen assimilated the lunga to their descent
group positions and modified their lineage alliances in terms of these
emblems of authority. In contrast to the Mbundu mulemba tree,
which represented a purely kinship-based lineage structure not unlike
the classic segmentary lineage system,14 the lunga brought a form of
territorially based, authority into the lives of the Mbundu, since the
holder of the lunga claimed authority over anyone who lived in a
territorially defined domain regardless of their relationship to him
or his lineage. According to Horton’s analysis of the ecological and
demographic conditions which favour specified types of social and
9 E. Verhulpen (1936), pp. 90ff.
10 Harry W. Langworthy (1971).
11 M. Plancquaert (1971). For the Cokwe, Redinha (1958).
12 Wauters (?) (n.d.), p. 1, found that the modern Pende remembered a
‘kingdom’ named Kalunga based on the lunga-e,mblem in their original home
among the Mbundu.
13 Confusion reigns in most secondary authorities over the meaning o f the
root -lunga and the so-called ‘high god’ Kalunga. Future research could
reinterpret this vague figure as an archaic form o f political authority and re­
evaluate Ms alleged ‘origins’ in the ocean as a simple association o f this form
of authority with water. ‘ICa-’ occurs commonly as a Bantu prefix (class 12)
denoting unusual size, either large or small, and turns up specifically as a
prefix on words for authority symbols which both personalizes them as
‘founding ancestors’ and stands for the abstract principles o f the authority
itself (e.g. Kalunga, ICanuma, and Kajinga are founding ancestors o f states
with the lunga, the numa, and the Icijinga as central authority emblems).
14 The mulemba tree may be taken as the symbol o f Horton’s ‘Type 1 seg­
mentary lineage system’ among the Mbundu (1971, pp. 84-93).
 62 I N D I G E N O U S B E G IN N IN G S
political organization, this sort of territoriality15 tends to appear
when alien groups block the path of the kin groups on the edges of
an expanding segmentary lineage system.16 It would be excessively
speculative to account for the appearance of hinga-based authority
in the basin of the Kwango by applying this hypothesis to vague
notions of population movements at so early a period of Mbundu
history,17 but the nature of the structure associated with the limga
seems clear. Each lunga ‘lived’ in a specified river or lake under the
care of an official who alone knew the secret of communicating with
the spiritual forces believed to inhere in it. This guardian of the lunga
exercised powers of granting access to land to stranger lineages who
.could find no unoccupied territory to settle. He controlled the
fertility of the fields and the rains which fell over the drainage basin
of whatever stream or river his lunga occupied, with the lunga-
guardians of major watercourses claiming a vague superiority over
the malunga and people living on tributary streams. The control of
the /wnga-guardians over land and rain gave them an obvious lever
with which to demand the respect, and perhaps the tribute, of local
farmers who depended on them to ensure adequate harvests. They
could command outright subservience from outsiders who owed them
their right to occupy land.
Aside from pointing out the potential of the lunga for establishing
an incipient hierarchy of lineages within a given territory, nothing
indicates how politics may actually have functioned under these
emblems. Presumably, control of the malunga passed matrilineally
within descent groups then as it does now. To that extent the malunga
reinforced the solidarity of the ngundu as the basic social institution
in Mbundu life. Initially, at least, the malunga need not have intro­
duced a significant degree of centralization beyond the formal
restructuring of parts of the lineage hierarchy on the basis of hydro­
graphy. They may thus have linked for the first time lineages not
connected by descent group genealogies generated by lineage fission.
But the essentially static qualities of the malunga, fixed in place by
their connections to specific bodies of water, gave the hierarchy a
. rigidity and artificiality which prevented lineages from forming new
alliances in response to changes in economic conditions or the rise
15 Horton’s ‘Type 2 dispersed territorially defined community’ (1971,
pp. 93-7).
16 Marshall D . Sahlins (1961).
17 Possibilities might include the interaction o f Kongo lineages expanding
to the south and Mbundu lineages moving northward from the centre o f the
Songo segmentary lineage system.
 T H E N G O L A AS A L IN E A G E SYMBOL 63
of an unusually able leader or confrontation with an external
military threat requiring reorganization and centralization for
defence. The malunga provided a partial answer to Mbundu needs
for institutions which cut across the basic lineage genealogies of their
society, but it did not satisfy the need for flexibility which ultimately
led many Mbundu to replace them with another method of linking
descent groups into new and larger aggregates.

The Ngola as a Lineage Symbol

The authority principle which provided for greater flexibility for
most Mbundu came with the spread of small pieces of iron known
as jingola. This emblem reached the Mbundu much more recently
than the lunga and is still revered by most lineages as their funda­
mental symbol. As a result, many more details have survived about
its origin and function in Mbundu political history. It was originally
an iron object of defined shape, a hammer, a bell, a hoe, or a knife.
Its significance has declined along with that of the malunga in recent
years as more recent states, including the present Portuguese ad­
ministration, have made its reputed powers seem less effective;
lineages now accept almost any misshapen or rusty piece of metal
as their ngola. Like the malunga, it was incorporated by the descent
groups at first and tended to reinforce their independence without
significant centralization of authority above the lineage level.
Essentially, the ngola provided yet another way of building non-
hierarchical links between Mbundu descent groups.
When an Mbundu lineage received an ngola, it appointed a
guardian for it in the belief that it, like their other symbols of
authority, gave him access to special spiritual forces useful for
regulating the affairs of men. They ascribed to the ngola some of the
same functions that obsolescent relics and emblems had once
performed and eventually adopted it as their most important
lineage emblem. It mediated between the living and dead members
of the descent group, as had other objects associated with the
mulemba tree. It helped its guardian to resolve disputes by divining
the justice of each party’s cause and assisted him in making decisions
affecting the welfare of his kinsmen. The Mbundu lineages raised
these guardians of the ngola to the status of important lineage
officials called malemba dya ngola (the ‘uncles’ of the ngola) and
gave them many of the duties formerly performed by the lemba
of the lineage. This left some malemba dya ngundu with only
the less worldly concerns of distributing pemba to the women of
64 I N D I G E N O U S B E G IN N IN G S
the kin group while the malemba dya ngola became the effective
lineage headmen and the predominant political officials of the
group.
The ngola solved the problems of structural rigidity and inflexi­
bility, which had remained even after the spread of the lunga, by
increasing the physical and social mobility of the Mbundu lineages.
It freed them from their ties to a single plot of ground and provided
a symbol which allowed lineages to move as a group without dis­
banding. Just as the midemba tree had symbolized the integrity of
the ldn group, its roots had represented the attachment of each
ngundii to the ground in which the tree grew. The association of
malunga with specific rivers and lakes had also tied the spiritual
well-being of each ngundu to the lands it occupied under the guidance
of the /wngc-guardian. These features of earlier political systems had
tended to restrict the mobility of the Mbundu lineages as groups.
The ngola, however, was portable, both physically and symbolically,
as the malunga and mulemba were not. Lineages with jingola could
move freely in pursuit of opportunities arising from trade, warfare,
or other circumstances. The ngola thus brought the potential for
revolutionizing the relatively static Mbundu world of /wngn-kings
and midemba trees by introducing the element of mobility which
previous structures had lacked.
Increased rates of lineage fission probably attended the increase
in physical mobility, since the techniques of dividing the jingola
provided easy ways of legitimizing the creation of new groups. In
accordance with established patterns of division of authority,
younger men who wished to escape domination by their elder kins­
men simply requested a new ngola from any lineage willing to sponsor
them in exchange for their agreement to assist the sponsor in war or
some other enterprise. Nephews could thus leave their uncles’
villages and establish new lineage segments under their own control
on the basis of the power of jingola derived from unrelated descent
groups. Lineages could break existing alliances based exclusively
on geographical contiguity to form new bonds reflecting economic
or other ties which did not find formal expression under the old
systems. Wealthy or influential lineages could offer subordinate
jingola as means of building lineage coalitions in which they held
the central position. The ngola thus brought a potential for
the rearrangement of lineages in more complex hierarchies than
before.
All evidence points to the highland regions near the headwaters
 T H E N G O L A AS A L IN E A G E SYMBOL 65
of the Lukala river as the proximate source of the ngola which
produced these changes in the nature of Mbundu social structure.
The people who brought the jingolci to the Mbundu are remembered
as ‘Samba’, obviously the ‘Samba a Ngola’ of the ngola aetiological
legend.18 Although the original Samba have now disappeared as an
identifiable ethnic group, their heirs still act as guardians of these
symbols in most Mbundu lineages. Matamba, the sixteenth-century
Kilcongo name for the ancient province lying in these same high­
lands identifies the source of both the Samba and the ngola since
the Portuguese form of the name, ‘Matamba’, was the same word
as the Kimbundu term ‘Samba’. Sound shifts which distinguish
Kimbundu from Kilcongo show that the original form of the name
had ts- as its initial consonant.19 Since Kimbundu had no initial
Is-,20 the Mbundu evidently dropped the t- from their pronunciation
of the word and say ‘Samba’ instead of ‘Tsamba’. The Portuguese
rendition of the same word dropped the -s- from the original
Kilcongo, since Portuguese also lacked an initial ts-, and -tamba
became the standard designation for this region in all written sources.
Addition of the Bantu plural prefix ma-, which commonly designates
ethic groups,21 completed the transformation from Tsamba to
Matamba and now establishes the identity of the two words.
The location of the chiefs holding the oldest Samba titles in the
Mbundu historical genealogies confirms their origin in the area now
known as Matamba. The oldest remembered Mbundu Samba posi­
tion, the one called Kiluanje kya Samba, belonged to lineages living
on the middle Lukala river just south of ancient Matamba.22 A
number of other political titles, less important in later periods but
of equal antiquity, preserved the name of Samba in the area of the
18 Although other authorities have not emphasized the role o f the Samba
in the history of the Mbundu, the name has appeared in published accounts of
other Mbundu traditions. See Wauters (?) (n.d.), pp. 4-5, for a note that the
Pende lineages (he called them ‘clans’) claimed descent from a ‘Gangila Samba’
(Kanjila (lea) Samba?, the ‘great bird of the Samba’). Atkins (1955), p. 344,
found that the H olo (formerly Pende near the middle Kwango) remembered
an ancient female known as Samba.
19 Current pronunciation in parts of Zaire has preserved the original form o f
the word; Plancquaert (1971), p. 13, where he gives the names Tsaam and
Tsamba.
20 See Atkins (1955), p. 328, who gave Tsotso as the Matamba form o f the
name that Kimbundu-speakers pronounce ‘Soso’.
21 e.g. Masongo, Maholo, Mahungu. These sound shifts explain why the
Kimbundu personal name ‘Matamba’, as in Matamba a Mbulu, has no con­
nection with the kingdom o f Matamba; the Kimbundu form of the name o f the
kingdom would be Masamba.
22 Lopes de Lima (1846), iii. 131-2.
66 IN D I G E N O U S B E G IN N IN G S
upper Lulcala as late as the nineteenth century.23 One of the import­
ant early divisions in the political genealogies of the Samba titles
divided them into two major groups at a place near Matamba called
Kambo lea Mana on the middle Kambo river.24 Traditions of the
Nlcanu and Soso, neighbours of Samba lineages now living in Zaire,
attribute the origin of these Samba to the sources of the Kwilu, a
river which flows north from the ancient province of Matamba
towards the Kwango.25
Documentary evidence from both Kongo and the Mbundu further
confirms the origins of the Samba in Matamba. Seventeenth-century
Mbundu descendants of the Samba called their original rulers by the
title of musuri, the Kimbundu word for blacksmith.26 Somewhat
earlier, in 1535, the Icing of Kongo, Afonso I, had identified himself
in his official correspondence as ‘Senhor dos Ambudos, d’Amgolla,
da Quisyma, e Musuru, de Matamba, e Muyllu, de Musucu, e does
Amzicos e da conquista de Pamzualubu . . .’27 Although the stylistic
irregularities of sixteenth-century written Portuguese would allow
this sequence of proper nouns to be combined in various ways
without violating the loose grammatical rules of the time, most
scholars have translated the elements in the series as parallel and
independent terms: ‘Lord of the Mbundu, of the Ngola (i.e. a Icing
of the Mbundu), of Kisama (the region south of the Kwanza), of
Musuru (not identified), of Matamba (the province), of Mwilu (not
identified), of Musuku (a region east of Kongo) . . . ’ and so on.
A different and more accurate interpretation of this title emerges
by regrouping its elements on the basis of the distinction implied
by the use of both ‘de’ (of) and ‘e’ (and) as connecting particles.
With modernized spellings, the passage would then read: ‘Lord of
the Mbundu of the ngola of the Samba,28 of Musuri (the king) of
Matamba, of Mwilu (king?) of the Sulcu, the Tyo . . . ’ The Kongo
king in fact had given both the names of the neighbouring peoples

23 See, for example, the map o f Chatelain (1894). The use o f Samba as a
surname indicated direct descent from the Samba people. These Samba titles
occupied subordinate positions in the (probably later) Ndembu states; see
Mattos (1963), p. 321.
24 Testimony o f Sousa Calunga, 29 September 1969.
25 Plancquaert (1971). This river should not be confused with another
(and better known) Kwilu which flows east o f the Kwango.
26 Cavazzi (1965), i. 253; cf. J. Pereira do Nascimento (1903), p. 51.
27 Letter from el-Rei do Congo to Paulo HI, 21 Feb. 1535 (A.N.T.T., C.C.,
1 -3-6 and 1-48-45); published in Brasio (1952-71), ii. 38-40.
28 This interpretation assumes that ‘Quisyma’ embodies an orthographic
eccentricity also found a century later in Cavazzi (1965), i. 253.
 T H E N G O L A AS A L IN E A G E SYMBOL 67
and the titles of their rulers in the order of their location along the
frontiers of their kingdom, beginning with the Mbundu on his
southern border and moving through the Suku in the east to the
Tyo in the north-east. A scribe at the Kongo court evidently
attempted to distinguish peoples from rulers through careful use of
‘de’ and ‘e’ but failed to convey the significance of the distinction
to most European readers. ‘Musuru’ of Matamba appears to be the
same musuri (or ‘Ngola-Musuri’) mentioned over 100 years later in
seventeenth-century Mbundu oral traditions as the idealized found­
ing smith-icing of the ngola. The coincidence provides a strong case
for identifying Matamba as the home of the Samba.
Seventeenth-century Mbundu traditions associated the Samba
with sophisticated iron technology and may thereby suggest one
reason why the ngola spread rapidly among their lineages. The
legendary founding Samba Icing, ‘Ngola-Musuri’, brought iron­
working skills which enabled the Mbundu to make axes, hatchets,
knives, and arrowheads for the first time.29 The traditions of modern
groups living in Zaire agree in portraying the Samba as excellent
blacksmiths who long ago introduced new methods of working iron
east of the Kwango river.30In confirmation of the connection between
the Samba and new iron-working techniques, the jingola which
they spread among the Mbundu always took the form of objects
fabricated from iron. This contrasted strikingly with earlier authority
emblems which included a variety of non-metallic objects: carved
wooden figures, head-dresses, feathers, mats, and so on. The emblems
brought by the Samba appropriately symbolized the skills which
made them welcome among the Mbundu.
The jingola probably arrived as a new form of lineage symbol
diffused peacefully from lineage to lineage without the waves of
migrating conquerors brandishing axes, knives, and arrowheads
who have dominated some European descriptions of these events.
A variety of local conditions could have predisposed many jingundu
to adopt it voluntarily. They could use the ngola to reassert their
autonomy in opposition to constraints imposed by such outsiders
as the /wngiz-kings of the major rivers. The powers attributed to the
ngola permitted each ngundu to disregard existing outside rulers by
substituting it for whatever subordinate authority emblem they had
at the time. The lineages preferred the ngola to other symbols because
their elders, rather than outsiders, nominated the incumbents who
29 Ibid.
30 Plancquaert (1971).
68 I N D I G E N O U S B E G I N N IN G S
filled the office of lemba dya ngola. These incumbents, by their
initiation into the position, theoretically acquired the status of a
‘Samba’ outsider, even though they might have been born locally,
and thus had prestige and neutrality in lineage affairs which enabled
them to act as arbiters in disputes between lineages; by assuming
this function, they replaced lineage councils, hmga-ldngs, headmen
of descent groups with senior positions in the perpetual genealogies,
and others. The powerful supernatural sanctions attributed to the
ngola allowed a lineage with this emblem to ignore, if they wished,
other descent groups and officials on whom they had previously
depended. The receptiveness of the lineages meant that the ngola
probably brought relatively few alien people with it. Instead, it
travelled by borrowing, converting local Mbundu who adopted it
and its associated technological benefits into ‘Samba’ by initiation.
Although the ngola, originally no more demanding of exclusive
lineage loyalties than any other authority symbol, may have co­
existed for a time with older emblems, its advantages gradually made
it the predominant form of political power among the Mbundu. In
the end, it produced a revolution in inter-lineage relations which
reached far beyond the internal changes it worked in the structure
of individual descent groups. As the ngola grew in prestige, it
regrouped Mbundu lineages into completely new units based on the
distribution of jingola rather than the old ties of lineage fission or
geographical contiguity. Each ngola carried a perpetual name which
simultaneously denoted the object itself, the spiritual forces it repre­
sented, and its lemba or guardian. Senior malemba dya ngola could
grant similar but subordinate titles to other lineages by awarding
them portions of their own jingola. A new permanent relationship,
expressed in kinship terms like all the others, linked the resulting
titles and united the lineages which held them in a new perpetual
genealogy. The lineages thus adapted the ancient institutions of
perpetual kinship and positional succession to build new title and
lineage groupings based on the ngola.
Certain other customs and beliefs may have reached the Mbundu
from the Samba along with this political symbol. One such custom
forbade any contact with the chameleon. The same avoidance
distinguishes the Zaire Samba groups now living on the right bank
of the Kwango from their non-Samba neighbours.31 Even though
the Mbundu still hestitate to touch a chameleon, they no longer
31 Ibid. The fact that the Kongo have this prohibition as well may confirm
the link between the Samba and the Kongo border region o f Matamba.
 T H E N G O L A AS A L I N E A G E SYMBOL 69
associate this prohibition explicitly with Samba descent. They seem
to have lost sight of the origins of the idea, since it originally served
to distinguish lineages possessing the ngola from those without one.
Once Mbundu lineages uniformly adopted the ngola, avoidance of
the chameleon lost its distinguishing function and the Mbundu
forgot its association with the Samba. All full members of the Mbundu
descent groups now consider themselves ‘Samba’, and the avoidance
persists only as a vestige which now differentiates locally born people
from slaves of non-Mbundu origin.32
No person eligible to hold an office connected with an ngola-
owning lineage could eat the flesh of a bushbuck.33 The ngola
eventually exerted so fundamental an influence on Mbundu beliefs
that they extended this prohibition to any titled position whatsoever
and made it a mark of full lineage membership. The Mbundu
explain this custom by noting a fancied resemblance of white spots
on this antelope’s back to leprosy sores and arguing that they may
contract leprosy from eating its flesh. Since they know that slaves
and other non-Mbundu, who do not claim Samba ancestry, eat
bushbuck without fear of infection, the prohibition evidently has a
primarily symbolic value in connecting them with the ngola and the
Samba.
This description of how the ngola solved problems at a period
when Mbundu social organization depended mainly on the lineages
and on the malunga does not exhaust the ways in which other symbols
still in use among various groupings of Mbundu lineages also affected
the patterns of authority and alliance. But these two illustrate some
of the principles which governed the origin and spread of such
symbols. Each came associated with certain powers, expressed in
terms of supernatural forces lurking somewhere behind the physical
object itself, and gave its possessors a form of authority which the
lineages assimilated without altering the basically descent-oriented
structure of their society. They assigned the care of these symbols
to permanent named roles, just as they had always given the malemba
similarly specialized power, and often simply collapsed the powers
of the /wKgfl-guardian and sometimes even the malemba dya ngola
into a single lineage position. The spread of malunga and jingola
32 Fear o f the chameleon is not uncommon elsewhere in Central Africa;
the Luba, among others, avoid it carefully. See, for example, Verhulpen (1936),
p. 70.
33 An antelope with white spots on its back called nguhmgu in Kimbundu;
John Charles Baron Statham (1922), p. 281, and Gladwyn M. Childs (1949),
p. 43, gave the species as tragelaphus scriptus.
70 I N D I G E N O U S B E G IN N IN G S
therefore did little to diminish the dominance of the lineages as the
basic institutions of Mbundu society. Although the lunga contained
an element of territoriality, it initially involved no significant central­
ization of power or increase in the scale of social organization. The
ngola represented a return from territorially defined authority to
links uniting descent groups regardless of the physical distance
which separated them. Both symbols stand as examples of cross­
cutting institutions adopted by a lineage-structured society to
elaborate the simple links provided by lineage fission. But neither in
itself marked more than a hesitant development in the direction of
incipient states.

Incipient States based on the Lunga

Although the adoption of the lunga at first led to no more than a
partial and non-hierarchical realignment of Mbundu lineage relation­
ships, certain /wnga-holders built their positions into something
which the Mbundu remember as very akin to ‘states’. The available
data indicate that these achieved their highest development among
lineages considered to have been Pende inhabiting the Luanda
plateau from the present town of Lucala eastwards to the escarp­
ment of the Baixa de Cassanje. They also occupied the lowlands of
the Baixa east of the ICambo river.34 The descendants of these
lineages have preserved a few malunga which once belonged to kings
of considerable note.
One of the most powerful of the Pende /wnga-kings was Butatu
a ICuhongo kwa Wutu wa Nyama. The traditions now recall only the
title of this king, probably because his authority declined long before
the formation of the present perpetual genealogies in the seventeenth
century. At the height of the influence of the kings who held this title,
‘Butatu’ ruled all the lands beyond the Kambo, probably on the basis
of a lunga living in the Luhanda or Lui river. A more concrete source
of power may have come from their control over salt pans along the
lower Luhanda. Ownership of these valuable economic resources
enabled the Icings who bore this title to extend their authority down
the trade routes which channelled salt from the Luhanda to much of
the northern Baixa de Cassanje. These kings went into decline as a
result of reverses suffered during the early 1500s at the hands of
34 Various testimonies of Sousa Calunga. The Pende testimony reproduced
in Haveaux (1954) has usually been interpreted to indicate that the Pende
lived near Luanda Bay in the late sixteenth century. Sufficient evidence to the
contrary has already been presented to obviate the need to offer a formal
refutation o f the point.
 I N C I P I E N T STATES BASED ON T H E L U N G A 71

M ap ii , Lunga-Authorities in the Baixa de Cassanje (Sixteenth Century

and Earlier)

rulers from the expanding state of Matamba.35 Although Butatu’s

title dwindled in influence after his kingdom had crumbled, it did
not disappear. The position passed into the lineage of a Pende title-
holder called Kanje who lived far to the south-east, and this lineage
preserved Butatu’s name as a subsidiary title of their own chief down
to the present day.36 The case of Butatu provides a good example of
how ancient titles and symbols survived in altered form to provide
evidence of the past in the present.
The figure of another early lunga-Idng, Kayongo ka Kupapa,
emerges from the mists of remnant Pende traditions as the ancient
ruler of Yongo, the region between the Lui and Kwango rivers north
of the present government post at Iongo. Since subordinate Mbundu
35 Cavazzi (1965), i. 22, mentioned that a ‘Batuta’ was the victim of the
expansion o f the Matamba kingdom. Printed versions of manuscripts from this
period commonly confused the closed letters ‘a’ and ‘o ’ with the open ‘u \
36 Testimonies o f Sokola; Kasanje ka Nzaje; Domingos Vaz; Sousa Calunga,
16 June 1969; maps of Schiitt (1881) and H. Capello and R. Ivens (1882), ii.
All agree that Kanje in the late nineteenth century lived on the west bank o f
the Lui river near Mt. Mbango (see Map II).
72 IN D I G E N O U S B E G IN N IN G S
political titles often took the proper name of their overlord as their
own surname, the alternative title of these Icings, Kayongo ka
Butatu,37 indicates that the men who held this position once paid
tribute to the Butatu a Kuhongo kwa Wutu wa Nyama. Kayongo ka
Butatu (or Kupapa) apparently broke away from Butatu a Kuhongo
kwa Wutu wa Nyama when the senior title’s kingdom went into
decline. The surname Kupapa could have been acquired from un­
known sources at that time, or it might have been a title belonging
to these rulers before they became subject to the great Butatu.38
Other Pende tegcz-lcings ruled the southern Baixa de Cassanje
under titles of a type known as yilamba (singular, kilamba).39 Title-
holders called Mahashi na Pakasa controlled the portion of the Lui
which turns eastward below the valley called the Baixa de Kafushi.
Their influence extended as far north as the salt marshes of the
Luhanda.40 The holders of Mahashi na Pakasa granted at least one
subordinate lunga-title to their southern neighbours, where Nzenze
ya Mahashi became the ruler of the flatlands along the upper Lui
river.41 Another title, Swaswa dya Swali, owned the heavily wooded
hills between the Lui and Mount Mbango.42 A lunga-king known as
Kikungo kya Njinje dominated the Pende of the Moa river basin on
the south side of the Lui. At an unknown time, probably after the
decline of Butatu, Kikungo kya Njinje briefly expanded his authority
north into Yongo.43
37 The modem form of the name occurs as Kayongo ka Kutatu; the surname
‘Kutatu’ probably represents a distortion o f the archaic title Butatu.
38 Kupapa and the titles o f most other Pende /««ga-kings seem to have been
praise names. The term kupapa, for example, may be an archaic Kimbundu
word having to do with making a craclding noise, as a fire crackles. António
de Assis Júnior (n.d.), p. 221, gives related verbal extensions, kupapajana,
kupapana, and kupapanesa, all having to do with crackling.
39 Imbangala informants described them as yilamba in the 1850s; Francisco
de Salles Ferreira (1854-8), p. 26. Dias de Carvalho (1898), p. 15, paraphrased
Salles Ferreira. The antiquity o f these titles was evident from the fact that
seventeenth-century Portuguese knew the Baixa de Cassanje as the ‘Kina kya
Kilamba’; kina designated any depression in the ground, as a cave or a grave;
see Assis Jr. (n.d.), p. 131. Later authorities attributed one or another extended
sense to the term: Chatelain (1894), p. 8 , Leite de Magalhães (1924), p. 73.
The term is adopted here for convenience in order to distinguish these rulers
from other sorts of chiefs. For the seventeenth century usage, p. 218, n. 120 below.
40 Testimony of Sousa Calunga, 22 July 1969.
41 Testimony of Alexandre Vaz, 31 July 1969; mahashi are spurts o f blood
and, by extension, various serius illnesses believed to be caused by excess
blood; Assis Jr (n.d.), p. 271.
42 The íiMwwa is an unidentified species o f tree; Assis Jr. (n.d.), p. 358.
43 Testimonies of Sousa Calunga, 24 July and 29 Sept. 1969. This title, unlike
the others, is composed o f personal names. The njinje is a small wild cat.
 K IN G D O M S BASED O N TH E N G O L A 73
The continual rise and fall of rulers which marked the history of
Yongo, the middle Lui, and the Luhanda river valleys did not occur
in the southern corner of the Baixa de Cassanje.44 There a static
array of lunga-kings ruled relatively unchanging territories. Kang-
ongo lea Pango controlled the higher elevations of the ICembo region
in the southern corner of the Baixa near the entry of the Kwango
river into the valley.45 Another, Ushi wa Nzumbi, ruled the rolling
woodlands which slope eastward from the Xa-Muteba region towards
the valley of the Kwango.46 IColco na Mumbi had control over the
flat plains of the Lutoa river basin and exploited the valuable salt
pans which were located there.47

Kingdoms based on the Ngola

Some Mbundu lineages turned the ngola from a lineage emblem
into the basis of a new type of political structure, just as others had
earlier performed the same transformation on the lunga. Since the
most powerful of the kings who based their authority on the ngola
still ruled the western Mbundu at the time the first Portuguese
reached Angola during the sixteenth century, more information
illuminates the history of these states than of those based on the
lunga. The most successful of the wgo/a-kingdoms attained a degree
of centralization, had a hierarchy of several levels, and expanded to
greater size than any of its predecessors. Armies capable of defeating
European military forces backed the will of its Icings and enabled
them to establish effective control over the affairs of many lineages,
perhaps for the first time. By creating hierarchical centralized political
institutions on the basis of the ngola, they had reversed the structural
significance of an emblem which had first come to the Mbundu as
no more than another in the long series of symbols and ideas
absorbed by the Mbundu lineage system.
44 The surviving fragmentary traditions describe only the political situation
at the moment of the Imbangala conquest o f Kasanje. They may therefore give
a false picture of stability in this part o f the Baixa de Cassanje.
15 The name o f this lunga-long apparently refers to an official staff given to an
emissary o f a chief; kangongo is a stick with designs carved on it, while pango
(or pangu) means ‘news’ according to Assis Jr. (n.d.), pp. 96, 332.
•|S According to Assis Jr. (n.d.), p. 374, ushi refers to a seller o f honey; the
significance in this context is unclear. An nzumbi is an ancestral spirit.
47 Kolco na Mumbi’s titles referred to his reputation as the only person who
knew the secret of climbing to the top o f Kasala, a thousand-foot-high vertical
cliff which lay above his lands and provided roosts for numerous storks. The
word koko refers to a species of climbing vine (dervis nobilis) and a mumbi is a
stork in Kimbundu; Assis Jr. (n.d.), p. 154; Rodrigues Neves (1854), p. 35;
testimony of Sousa Calunga 10 Sept. 1969.
74 IN D I G E N O U S B E G IN N IN G S
The evidence on how the earliest ngola-Icings transformed their
authority from essentially egalitarian lineage alliances to something
which was clearly a kingdom remains speculative, unless seventeenth-
century Mbundu oral traditions provided meaningful information
when they described the first great ‘Ngola’ as the bringer of ‘axes,
hatchets, knives, and arrows, the things which help the Black man
in hunting and in war’.48 The prevalence of founding smith-kings
in the oral traditions of states all over Africa makes highly suspect
the image of iron-wielding conquerors who create states by virtue of
their superior armament and skill. Scepticism seems especially
appropriate in view of the services which such iron-bearing con­
querors have performed in the ‘Hamitic myth’ and other dubious
conquest theories of state-formation. In the case of the ngola-ldngs,
no known ethnographic data suggest a massive influx of conquering
foreigners either during the preliminary diffusion of the ngola as a
lineage symbol or during the later expansion of «go/a-states. Rather,
one or another of the ngo/a-holding officials of the local Mbundu
descent groups must have extended his authority to neighbouring
lineages by virtue of subordinated new ngo/a-titles rather than
through the co-ordinate relationship of allied lineages.
The basic Mbundu aetiological genealogy of ‘Ngola Inene’ traces
the main lines of these new rulers’ development. The ‘Ngola Inene’,
or the ‘great ngola’, represents the abstract principle of political
organization based on the ngola and may be equated with the period
before the ngola became an important political symbol among the
Mbundu. ‘Samba a Ngola’, whom the genealogy depicts as the
‘daughter’ of ‘Ngola Inene’, stands for the Samba people of Matamba
whom the Mbundu associate with the ngola at that period. The rest
of this genealogy, like most Mbundu political genealogies, describes
the genesis of a hierarchy of political titles composed of successive
generations of ‘son’ titles created by incumbents in immediately
superior or ‘father’ positions. Thus, the first remembered ngola
political title among the Mbundu was the kiluanje kya samba
held by lineages who lived not far from the source of the ngola
in Matamba. The meaning of the title kiluanje, ‘conqueror’,49
seems to indicate that the Mbundu regard this title as the spear­
head of the advance of Samba political institutions into Mbundu
territory.
The ngola came into its own among the Mbundu with the rise of
48 Cavazzi (1965), i. 253.
40 Assis Jr. (n.d.).
 K IN G D O M S BASED ON T H E N G O L A 75
a second position called the ngola a ldlitanje. Its status in the gen­
ealogy as a ‘son’ of Kiluanje kya Samba implies that it originated
as a subordinate position in the original state of the kiluanje kya
samba. Kings in the ngola a kiluanje title then eclipsed the holders
of the senior title during the early sixteenth century, thus continuing
the southward movement which had begun in the shift from Mat-
amba to kiluanje kya samba. The ngola a kiluanje made their capitals
between the lower Lulcala and the Kwanza near the mountains which
bounded the Luanda plateau on the west.
A seventeenth-century narrative tradition described the rise of the
ngola a kiluanje in terms which parallel the sequence evident from
the modern genealogies. Despite some confusion on the part of the
missionary who recorded this tradition, it made clear that the great
‘Ngola Musuri’ (equivalent to Ngola Inene) in Matamba had yielded
to lineages belonging to the Mbundu subgroup which lived on the
upper Lulcala. According to the narrative, Musuri, the king in
Matamba, married an unnamed woman known only by a title,
ngana inene or great lord. This ‘wife’, evidently no more than a
distorted reference to some other honorific title belonging to Musuri,
bore three daughters, ‘Zunda dya Ngola’, ‘Tumba dya Ngola’, and
a third whose name the missionary neglected to record. A slave of
the great smith-ldng contrived his master’s death, and the royal title
fell to the daughter ‘Zunda dya Ngola’. The other daughter, ‘Tumba
dya Ngola’, married a man called Ngola a Kiluanje kya Samba
(‘Angola Chiluuangi Quiasamba’) and eventually came into conflict
with her sister. After an appropriate series of treacherous deeds,
‘Tumba’ and her husband defeated ‘Zunda’ on the strength of
superior armed might, and Ngola a Kiluanje went on to become the
first great conquering king of the Mbundu. He had ‘a great, great
number of descendants through his various concubines, who were
the chiefs of the most important families of the realm’.50
In conformity with the canons of modern Mbundu perpetual
genealogies, this tradition showed the pairs of linked lineages and
titles which dominated the ngo/cz-lcingdoms of the fifteenth and six­
teenth centuries. ‘Ngola Musuri’ represented the ngola-Icings in
Matamba. The name of ‘Zunda dya Ngola’ has been altered from
its proper form, which appears in the modern aetiological genealogies
as zundu dya mbulu, to make it conform to the rule that ‘daughter’
titles take the first name of the ‘father’ position (in this case, Ngola)
as their surname. Zundu dya Mbulu (Ngola) was the legendary
50 Cavazzi (1965), i. 253-5.
76 IN D I G E N O U S B E G IN N IN G S
ancestress of the Ndongo subgroup of the Mbundu, precisely the
lineages which lived on the middle Lulcala where the kilmnje kya
samba (mentioned in the tradition only by implication through the
appearance of his title as a surname for Ngola a Kiluanje kya
Samba). The tradition thus stated that the lineages of the upper
Lulcala (read ‘the kiluanje kya samba’) built a kingdom on the
powers they derived from the great ngola in Matamba.
Since the missionary who recorded the tradition confessed his
confusion about the correct names of the three ‘daughters’ of
‘Ngola-Musuri’, there is no reason to suppose that he recorded
‘Tumba dya Ngola’ correctly. The name ‘Tumba’ occurs in no mod­
ern variant of the Mbundu traditions and in this case was probably
introduced from the histories of peoples who lived east of the
Kwango where the name now figures as an ancestress of the Colcwe.
‘Tumba’ very likely took the place of Matamba a Mbulu, the aetio-
logical ancestor of the Lenge people of the lower Lukala. By the rule
that marriages in these genealogies signify the claiming of a title by a
lineage or group of lineages, the marriage of ‘Tumba a Ngola’ (read
‘Matamba a Mbulu’, or the Lenge people) to Ngola a Kiluanje and
their eventual victory over the Ndongo of the northern Lukala
describes the battles in which the southern Mbundu wrested leader­
ship from the kiluanje kya samba and built a new centre of ngola-
based political power on the lower Lukala.
The capitals of the ngola a kiluanje state lay in the Lenge province,
as the tradition suggested, and economic factors probably contri­
buted to the expansion of these kings’ power, just as the salt pans
of the Lui had formed one basis for some of the fengcz-kingdoms in
the Baixa de Cassanje. The Lenge region included the iron-workings
of the Nzongeji river valley, and the superior metallurgical tech­
niques known to the Samba apparently made these ore supplies a
valuable strategic prize. Not only did the ngola a kiluanje locate their
capitals very near the mines, but they later tenaciously resisted all
Portuguese advances towards this area while showing relatively little
concern about Portuguese penetration elsewhere in their domains.
The kings’ sensitivity to threatened loss of this region may have come
from their desire to protect their capital towns, as the Portuguese
attackers assumed, but their dependence on control of the ore
deposits could also have influenced their tactics. A secondary eco­
nomic axis of the kingdom turned on possession of the Ndemba salt
mines south of the Kwanza in Kisama. The ngola a kiluanje seem to
have established their hegemony over the local political authorities
 K IN G D O M S BASED ON T H E N G O L A 77
near these mines early in the history of their kingdom since by the
1560s they had imposed a tax on salt which the Kisama title-holders
sent to the court near the Nzongeji. Blocks of this salt circulated as a
kind of ^currency’ in the prestige sphere of exchange in the kingdom.,51
As late as 1798, successors of the Kisama rulers still sent tribute to the
heirs of the ancient ngola a kiluanje who by the latter date had moved
to the Wamba river and bore the Portuguese titles of the ‘reis
jingas’.52

Map m. Pre-Seventeenth Century Mbundu Subroups and Political

Institutions

Once firmly in command of the economic resources of the king­

dom, the kings who held the title of the ngola a kiluanje extended their
influence into the lineages living far away through the distribution
of subordinate noble titles derived from their ngola. In an elaboration
51 Antonio Mendes to Padre Geral, Lisboa, 9 May 1563; Brasio (1952-71),
ii. 495-512.
52 ‘Noticias do paiz , . . ’ (1844), pp. 123-4.
 78 IN D I G E N O U S B E G IN N IN G S
of the classic Mbundu technique of linking lineages by the award of
a new perpetual named position, they awarded ‘son’ titles surnamed
-a ngola to many descent groups. If possible, they took a wife from
the lineage and awarded the original title to a biological son born of
this marriage; they trusted him, and later his successors, to represent
royal interests to their kinsmen and related lineages. One of the
missionaries who visited the court of the ngola a kiluanje in the
1560s provided eye-witness confirmation for the seventeenth-century
tradition which stated that the king had ‘a great number of descend­
ants through his various concubines, who were the chiefs of the most
important families of the realm’; he reported that the ruler of the
1560s kept over 400 wives at his court.53 Allowing for the obvious
mistranslations of the Mbundu terms for ‘wife’ (as ‘concubine’) and
lineage (as ‘family’) in the written version of the seventeenth-century
tradition, the original tradition had clearly consisted of a classic
political genealogy showing the ties between subordinate political
titles (‘descendants’) and the lineages (‘concubines’) which held them.
Drawing on these economic bases and structural devices, the
ngola a kiluanje kings began about 1510 to extend their influence in
all directions from the lower Lukala. The evidence for the date of
these developments comes from Jesuits who visited them during the
1560s and calculated that an ‘Ngola Inene’ had begun to overwhelm
the many small states which had preceded the kingdom some fifty
years earlier.54 There is no reason not to equate the ‘Ngola Inene’
they mentioned with the first of the ngola a kiluanje Icings rather than
with the kiluanje kya samba or the ‘Ngola-Musuri’ in Matamba. The
missionaries confused the political principle of the ngola (‘Ngola
Inene’ in the usual phrase of the Mbundu) with the first incumbents
in the ngola a kiluanje position. The Mbundu custom of referring to
the major political title-holder of each Mbundu subgroup by the name
53 Antonio Mendes to Padre Geral, Lisboa, 9 May 1563; Brdsio, (1952-71),
ii. 495-512.
54 po pero Rodrigues et at., ‘Historia da residencia dos Padres da Companhia
de Jesus em Angola e cousas tocantes ao reino e conquista’, 1 May 1594,
Asquivo Romano da Companhia de Jesus, Lus,, 106, fols. 29-39; published
by P° Francisco Rodrigues (1936), and Brdsio (1952-71), iv. 546-81. Also
Antonio Mendes to Padre Geral, Lisboa, 9 May 1563; Brdsio (1952-71), ii,
512. Documentary allusions from the Kongo seem to support this date. As
late as 1512, the Kongo king (Afonso I) claimed authority over the ‘Ambudos’
to the south, but by 1520 the Portuguese were preparing an expedition to visit
the ‘rey d’Amgola’, obviously the ngola a kiluanje. The Portuguese emissary
who eventually reached the Mbundu king’s capital found a monarch powerful
, and arrogant enough to keep him prisoner until 1526. See Birmingham (1966),
pp. 28-30, and documents cited.
 K IN G D O M S BASED ON T H E N G O L A 79
of its symbolic ancestress explains the mistake. Kilamba kya Ndungu,
for example, one of the most powerful Pende title-holders, is often
called ‘Mumbanda a Mbulu’, the legendary founder of the Pende
people. In the case of the ngola a kiluanje in the sixteenth century,
the Mbundu would have praised the king to European visitors as
‘Ngola Inene’, founder of the Samba people. Such a convention
would merely have reflected the political dominance of the ngola a
kiluanje at the time the Jesuits collected the information.
Recognition that ‘Ngola Inene’ was an idea rather than a person
and that the names of his ‘descendants’, Kiluanje kya Samba and
Ngola a Kiluanje, represented perpetual titles rather than individual
rulers pushes the dates for the early history of the Samba states
back at least to the fifteenth century. The early sixteenth-century
date for the emergence of the ngola a kiluanje remains acceptable on
the basis of the near-contemporary record provided by the Jesuits in
the 1560s. The kiluanje kya samba probably preceded the ngola a
kiluanje by much more than the single biological generation implied
by an interpretation of the genealogy as a human lineage; no data
known to me suggest a more definite date for the rise of this position.
Since the ngola first spread as a lineage emblem at an even more
remote period, it could have reached the Mbundu as part of the
general southward spread of Kongo influence and political structures
as early as the thirteenth or fourteenth century.
Modern political genealogies preserve enough of the network of
titles created by the sixteenth-century ngola a kiluanje to outline the
expansion which extended their influence to all the Mbundu regions
and beyond. An Ndambi a Ngola and a Kangunzu ka Ngola, from
the appearance of ‘ngola’ as the second term in their names both
evidently titles awarded directly by incumbents in the ngola a kilu­
anje, became important positions in the Matamba highlands north
of the kingdom’s centre. The establishment of these titles, and doubt­
less others as well, may have followed the defeat of the Matamba
people in the wars between the northern Mbundu (‘Zundu dya
Ngola’) and the central lineages (‘Tumba dya Ngola’) described in
the seventeenth-century Mbundu traditions. Other titles, Kalunga
ka Ngola, Muhi wa Ngola, and Nzungi a Ngola among others,
brought the lineages of the central Mbundu plateau near the Mulcari,
Lushimbi and Tumba rivers under the sway of the ngola a kiluanje.55
The award of still other ngo/a-titles west of the upper Lui extended
55 Testimonies o f Sousa Calunga, 21 Aug. 1969 and 30 Sept. 1969; also
Domingos Vaz.
80 IN D I G E N O U S B E G IN N IN G S
the central kings’ influence eastward among the Pende lineages
living in the Baixa de Cassanje. The northern and eastern expansion
depended, for the most part, on the incorporation of unrelated
political titles, perhaps those of independent political structures
which had preceded the appearance of the ngola in those areas. A
marriage between the ngola a kiluanje and a lineage known as
Mbekesa a Lulcunga extended these kings’ authority to the north
bank of the middle Kwanza.56 Farther to the west, the ngola a kiluanje
absorbed a cluster of titles in the Lenge region surnamed -a keta.

M ap iv . Expansion of the Ngola a Kiluanje (Before 1560)

The existence of several positions with the same surname indicates

that a little-known but ancient kingdom had grown up in that area
before the expansion of the ngola a kiluanje. By the early seventeenth
century, all of these positions had become local representatives of
the ngola-kings, including the most powerful of them, holders of the
ngoleme a keta in the Uamba area west of the Lulcala river.57 The
56Testimony o f Sousa Calunga, 21 Aug. 1969.
57 Ngoleme a Keta later became an important enemy o f the Portuguese. He
was prominent when Paulo Dias de Novaes first arrived in the 1560s and was
still a major power when the second Portuguese expedition to Angola came in
1575. See letter from Paulo Dias de Novaes to el-Rei, 3 July 1582 (B.M., Add.
MSS. 20, 786, fols. 182—183v; Brdsio (1952-71), iv. 341-5. A later Ngoleme
still possessed sufficient strength in 1644 to defeat a Portuguese army, but he
finally succumbed in the late 1650s as the Europeans took reprisals for his
1644 victory; Cadomega (1940-2), i. 349-55 and ii. 141-9.
 K IN G D O M S BASED ON T H E N G O L A 81
original keta had probably based their state on control of the iron-
ore deposits located in the mountains there.38 These keta-kings
apparently never developed the ore deposits to their full potential,
perhaps because their limited iron-worldng technology prevented
them from making as effective military use of the ore as did the Samba
under the ngola a kiluanje. Their inability to fabricate weapons
would have left them vulnerable to the better-armed outsiders.
A variety of titles carried the authority of the ngola a kiluanje
south of the Kwanza at an early stage in the history of the kingdom,
but these nobles tended to break away from the control of the central
kings as soon as they penetrated beyond the Mbundu culture area.
According to the modern elders of the Kibala kingdom in the
Ovimbundu highlands beyond the Longa river, descendants of some
of the most far-flung ngola title-holders in this diaspora, several titles
had crossed the Kwanza at about the same time; these included such
non-ngola positions as Kiteke kya Bengela, Kafushi ka Mbari,
Mbumba a Mbundo, and Mbumba a Kavenge, in addition to the
more standard position of Ngama a Ngola. The holders of the
kafushi ka mbari moved south-west until the title eventually came to
rest among the non-Mbundu people at the Kisama salt deposits.
The others settled on both banks of the Ngango river, with Ngama
a Ngola and Mbumba a Kavenge moving along the left bank in
Halco and Kiteke kya Bengela going farthest south into the highlands
where they established themselves as rulers of the ‘Marimba’ people
south of the Longa.5859 A few isolated titles descended from the
ngola a kiluanje may have penetrated as far south as the Hanya
who lived in the mountains above the later Portuguese town of
Benguela; their modern descendants still ascribe their origins to the
‘great Ngola of the north’.60 All of these southern title-holders broke
58 Seventeenth-century documents usually spelt the name ‘Caita’ or ‘Gaeta’.
G. Weeckx (1937), p. 151, equated Keta with Musuri and Mbumba a Mbulu,
implicitly confirming the antiquity o f the title. Sousa Calunga, testimonies of
11 and 30 Sept. 1969 gave genealogies for the titles o f this state and added that
ngola a kiluanje later took wives from those lineages (i.e. incorporated them).
59 Brandão (1904), pp. 77, 407-8. Brandao’s note that the Kisama recognized
the kiluanje kya samba as their overlord rather than the ngola a kiluanje
suggests that this phase o f the expansion may have preceded the rise of the
latter ngola-title. Although it is probably impossible to arrive at an exact date
for these events, it is quite possible that these titles spread south long before
the arrival of the Portuguese, contrary to the naive chronological calculations
of Brandão, who argued that the positions were those o f ngc/cr-officials fleeing
before Portuguese military forces in 1582. Ngama a N gola’s presence in Libolo
was documented during the 1650s; Cavazzi (1965), i. 28.
60 Alfred Hauenstein (1967b), pp. 229ÍT, and (1960), p. 222. Hauenstein’s
82 IN D I G E N O U S B E G IN N IN G S
away from the ngola a kiluanje and became politically independent
as the people who controlled the positions adapted them to local
cultures which lacked the Mbundu sense for the subtleties of perpetual
kinship and positional succession.
A brief seventeeenth-century history listing the early holders of the
ngola a kiluanje title provides tantalizing glimpses of the internal
political history of the kingdom. Its basic structure, an alliance of
lineages linked by fictitious ‘marriages’ to a single dominant title,
meant that the descent groups in control of the most powerful
subordinate positions of the state fought among themselves for control
of the royal title. The written versions of this tradition, distorted as
usual, took the form of a kinglist:61
1. Ngola a Kiluanje;
2. Ndarnbi a Ngola;
3. Ngola a Kiluanje;
4. Jinga a Ngola a ICilombo kya Kasenda (a usurper);
5. Mbande a Ngola.
The missionary who recorded these ‘names’ confused the title of the
position, ngola a kiluanje (numbers 1 and 3), with the names of the
subordinate titles which rotated in control of the kingship (numbers
2, 4, and 5). Comparison of this kinglist with the modem political
genealogies of the ngola a kiluanje state shows that the succession
from (presumably) the original ngola a kiluanje to the ndambi a
ngola marked the passage of power from lineages in control of the
senior title to those holding the subordinate ndambi a ngola position.
Geographically, there was a movement from the founding lineages
on the lower Lulcala back to northern lineages living not far from the
Matamba area where the Samba had originated. Their accession
marked a resurgence of the northern Samba over their southern
relatives who had broken away to build the ngola a kiluanje into its
independent status. The takeover by the ndambi a ngola may have
occurred during the 1540s when the occupant of the southern ngola
a kiluanje position tried to contact the Portuguese king.62 His desire
01 Cavazzi (1965), i. 256-7. Contemporary sources customarily confused titles
o f rulers with the names o f the incumbents.
62Antonio Mendes to Padre Geral, Lisboa, 9 May 1563; Brasio (1952-71),
ii. 4 9 7 .1 concur in the analysis presented by Birmingham (1966), p. 34.

speculation that these ngola-chiefs moved south after the 1671 defeat o f the
puppet ngola a kiluanje in Pungo Andongo is not based on any evidence. It
seems much more likely that this title represents part o f the expansionary
period o f the state rather than its dying gasp.
 K IN G D O M S BASED O N THE N G O L A 83

to contact the Europeans could have arisen from pressures which

the southern Samba king was feeling from his northern relatives in
Matamba.
The hypothesis that this diplomatic initiative of the 1540s came
from a king who felt himself on the defensive against holders of a
subordinate ‘son’ position, the ndambi a ngola, explains the behaviour
of the succeeding northern king. He received in a hostile fashion the
missionaries whom the Portuguese crown finally dispatched to the
Mbundu twenty years later in the 1560s. The king whom these
Jesuits met presented himself as ‘Ndambi a Ngola’, a ‘son’ of the
ngola a kiluanje who had requested them two decades before, and
made it clear that he, unlike his ‘father’, regarded the Europeans as a
threat. His hostility to the foreigner could have stemmed from fears
that they might discover the change in the descent groups in control
of the title; the location of the capital among the southern lineages
far from his own kinsmen would have made him vulnerable to an
attempt at restoration of the former regime.63 At the same time, the
ndambi a ngola obviously hesitated to treat his guests too roughly.
His consequent inability to reconcile conflicting opinions at his court
as to the best way to handle their presence led to the vacillating
policies which kept the leader of the expedition, Paulo Dias de
Novaes, and his Jesuit companions captives at the capital for most
of the decade.64
During the 1570s, 1580s, or 1590s, power seems to have passed
from the lineages behind the ndambi a ngola to other descent groups
in control of another political title remembered in the traditions as a
‘usurper’. Since this was a period when the Portuguese were largely
confined to a beach-head they had established at Luanda in 1575
and to the lower Kwanza valley, written sources add almost nothing
to the internal history of the ngola a kiluanje kingdom. The intruder,
called ‘Jinga a Ngola a Kilombo lcya Kasenda’, may probably be
identified through the modern Mbundu genealogies. Pende lineages
of the Lui valley in one of the old limga-kingdoms, that of Swaswa

63 Alternatively, o f course, Mbundu attitudes towards the Europeans might

have changed as a result o f stories reaching them from Kongo where the
Portuguese residing at the maiti Kongo’s court were falling from the favour they
had enjoyed during previous years. Their ability to disrupt the internal politics
o f that state was becoming increasingly apparent.
04 The documentation on the visit of Dias de Novaes and the Jesuits to the
court o f the ngola a kiluanje during the 1560s has been published in the AA,
ser. 2, xvii, nos. 67-70 (1960), 8-32; these letters also appear in Brasio (1952-
71), ii and iv, and in Sousa Dias (1934).
84 IN D I G E N O U S B E G IN N IN G S
dya Swali, had created a small subsidiary state in which the central
title bore the name Kasenda lea Swaswa (dya Swali). ‘Jinga a Ngola
a Kilombo kya Kasenda’ may thus have been an ngola-title, the
jinga a ngola, which belonged to the Pende lineage of Kilombo kya
Kasenda (ka Swaswa dya Swali). Perpetual genealogies from the
middle Lui suggest that lineages related to Swaswa dya Swali were
among the most powerful descent groups in the Baixa only a few
years later during the 1620s to 1640s. Their prominence at the later
period may well have derived from a period of control over the
ngola a kiluanje during the late sixteenth century. They could have
originally won the «goto-title as a result of their proximity to the
Luhanda salt pans once dominated by the lunga-king Butatu. The
shift in the internal political balance of the ngola a kiluanje state to
the Baixa de Cassanje may have coincided with the spread of ngola-
titles to the Pende of the middle Lui.65
The accession of ‘Mbande a Ngola’, the next title named in the
seventeenth-century kinglist, represented the transfer of effective
political power to a third lineage or group of lineages. These descent
groups also lived in the Baixa de Cassanje, somewhat west of Swaswa
dya Swali in the basin of the Kambo. The later ngola a kiluanje
looked back on the mbande a ngola as a legitimate king in contrast
to the usurper ‘Jinga a Ngola a Kilombo kya Kasenda’, but the
apparent discontinuity in terms of legitimacy concealed an under­
lying continuity in terms of the lineages who controlled the position,
as indicated by the fact that power remained in the east, in the Baixa
de Cassanje. The holder of the mbande a ngola title who claimed the
ngola a kiluanje had been a ‘son’ of the usurper ‘Jinga a Ngola a
Kilombo kya Kasenda’. Since the sources for this tradition (recorded
in the 1650s) could have recalled these events from living memory
(not more than sixty to eighty years earlier), they probably referred
to a biological son, not a derivative title, of the intruding holder of
the jinga a ngola. This jinga a ngola had evidently married a woman
from a lineage in possession of the legitimate ngola-position of the
mbande a ngola and had then manoeuvred his real son into the title
to which he would have been fully entitled under the rules of matri-
lineal succession. The entry of the son, whose personal name was
evidently not recorded, kept the lineage of the father (Kilombo kya
Kasenda) in position to exercise a strong voice in the affairs of the
realm. Although no evidence dates the accession of the first mbande a
ngola, the ngola a kiluanje who died in 1617 was probably from
65 Cf. Muhi wa Ngola, etc., p. 79.
 K IN G D O M S BASED ON T H E N G O L A 85
this lineage, and the Icing who died in 1624 certainly bore this
title.66
Defeats by Portuguese armies interrupted the autonomous political
processes of the ngola a kiluanje state during the 1620s. The victorious
Portuguese transferred the central title to another group of lineages,
holders of the hari a kiluanje, a senior ngola position in a collateral
line, a ‘brother’ position to the ngola a kiluanje and a direct descend­
ant of the old kiluanje kya samba. Although the hari a kiluanje had
legitimacy or even seniority in a technical sense, holders of that
position had apparently never exercised much influence in the affairs
of the kindom, and the Mbundu never acknowledged them as heirs
to the mbande a ngola. The title does not appear at all in the modern
political genealogies, and the hari a kiluanje kingdom belongs to the
history of the Portuguese conquest rather than to the study of state-
formation in the context of the Mbundu lineages.

N g o la I n e n e / N g o l a - M u s u ri

Kiluanje kye samba

ngola a kiluanje hari a kiluanje

ndambi a ngola jinga a ngola mbande a ngola

nzinga a mbande (?)
(Q u e e n N zin g a )

F ig . iv . R elationship o f Ngola- Titles.

T able 1

Chronology o f the Ngola a Kiluanje

Ruleris) Location D ates
‘Ngola Inene’/‘Ngola-Musuri’ Matamba ?
kiluanje ky a samba Upper Lukala (fifteenth century)
ngola a kiluanje Middle Lukala c. 1510-1540s
ndambi a ngola Matamba c. 1550s-1560s
jin g a a ngola (usurper) Baixa de Cassanje c. 1570s-1590s?
mbande a ngola Baixa de Cassanje c. 1600s 7-1624
f hari a kiluanje (puppet) Pungo Andongo 1624-1671
\ Nzinga Baixa de Cassanje 1620s-present
* rulers of the separate successor states (‘Hari’ and ‘Matamba’).

Up to that point, politics in the ngola a kiluanje kingdom had

revolved around competition between lineages or coalitions of
66 Bishop D . Simao Mascarenhas to el-Rei, 2 Mar. 1624 (A.H.U., Angola,
cx. 1); Brdsio (1952-71). vii, 199-203.
86 IN D I G E N O U S B E G IN N IN G S
lineages based on the pre-ngola lineage groupings. Each had tried to
place their own kinsmen, holders of the senior state titles, in the
central position of the ngola a kiluanje. The central kings presumably
tried to break down the solidarity of these lineage groupings through
the placement of subordinate ngo/a-positions in the lineages most
likely to give their loyalty to the central kingship. Although the
holders of the ngola a kiluanje grew powerful through their control
of the iron-rich country near the Nzongeji river and seem to have
maintained their capitals in that region, the men who exercised this
authority came from a variety of lineages located elsewhere in the
kingdom. Generally, the lineages to the north and east seemed to
dominate through their possession of such senior state titles as the
ndambi a ngola and the mbande a ngola.
The holders of the prt-ngola titles in the south and west never
acquired much influence at the centre of the kingdom, although
some of them—notably Ngoleme a ICeta, Kafushi lea Mbari, and
Kitelce kya Bengela—became powerful provincial lords in their
own right. The far southern title-holders tended to break away,
especially when they had become established among non-Mbundu
populations where the ngola exercised a weaker attraction than it
did nearer its origins in Matamba. The retreat of the heirs of the
ngola a kiluanje to the Baixa de Cassanje after the Portuguese
defeated the main kingdom in the 1620s therefore represented a
withdrawal of the capital toward the lineages which had held effective
authority since the 1560s. This hypothesis accounts for the survival
of the successor ‘Jinga’ state in the Baixa de Cassanje during the
eighteenth and nineteenth centuries. The puppet hari a kiluanje who
remained on the plateau represented an entirely different set of
lineages and were regarded by the Mbundu as a separate state which
existed only at the pleasure of the Portuguese and had no legitimacy
in traditional terms.

Conclusions
The /wngiz-lcings of the Baixa de Cassanje and the kingdom of the
ngola a kiluanje were only the two best-known examples of a much
larger class of early Mbundu states which sprang from entirely local
roots. Although many of the other early state-like structures which
appeared among the Mbundu have, like the Iceta-kings of the Lenge,
almost entirely dropped from the historical record, all may be
analysed entirely in terms of indigenous Mbundu beliefs about
authority, the distribution of symbolic objects to confer authority
 C O N C L U S IO N S 87
over men, and beliefs in positional succession and perpetual kinship.
Mutually corroborative evidence from documentary sources and
oral traditions suggests that Mbundu state-building techniques were
a product of a fundamentally inflexible lineage structure, made
unusually rigid by the ability of perpetual kinship and positional
succession to preserve historic ties between kin groups. Since the
Mbundu did not have the freedom to alter their lineage genealogies
to reflect changing social and political conditions as did similarly
structured societies elsewhere,67 they had to provide other social
channels through which men might pursue personal ambition or
respond to circumstances not comprehended by the patterns of kin­
ship. Although the traditions do not indicate the exact conditions
which prompted certain lineages to adopt new symbols and to restruc­
ture their relationships to other descent groups, numerous descent
groups clearly made such adjustments repeatedly throughout their
history. The early malunga and jingola provide examples of non-
centralized and basically non-hierarchical reorganizations of this
type.
Once a new symbol of authority had spread among the Mbundu
lineages, however, individual holders of certain titles were repeatedly
able to expand their personal spheres of influence beyond the confines
of their lineage to claim a measure of authority over persons not
related to them. In the case of the malunga-kings of the Baixa de
Cassanje, control over an extensive salt trade radiating from salt
pans located within their territories seems to explain the rise of the
Butatu and later the Swaswa dya Swali dynasties. Factors of a
similarly economic nature seem to underlie the growth of the ngola
a kiluanje, since those rulers derived some of their strength from their
domination of the iron mines of the Nzongeji River. The expansion
of this kingdom clearly pointed toward the salt pans of Kisama, and
later internal political developments revealed the strength of the
north-eastern lineages nearest the Luhanda river salt deposits.
However important the economic causes of the growth and expan­
sion of the early Mbundu kingdoms may have been, these elements
inevitably attract the attention of historians since the evidence of
their presence survives for all to see in the form of salt pans and
abandoned iron tailings.-What history has lost, in all these cases,
is the role of individual human genius, the play of chance, and most
of the intricacies of political manoeuvring which must also have
contributed to the development of these states.
67 A classic case is the Tiv o f Nigeria; Paul and Laura Bohannon (1953).
88 IN D I G E N O U S B E G IN N IN G S
The generalized model of indigenous Mbundu state-formation
takes as its starting-point the assumption that new cross-cutting
institutions continually appeared among the Mbundu descent
groups. The strength of the kin groups was sufficient in many cases
to convert the symbols of these movements into emblems of the
lineage structure, as it did initially with both the mahinga and the
jingola. Among the countless anonymous holders of titles associated
with these symbols, a few talented or fortunate individuals managed
from time to time to convert their control of a valuable economic
resource into effective political power. Their states took the form of
more or less extensive lineage coalitions based on awards to other
descent groups of political titles subordinate to their own. Some of
these states were growing even as others were declining throughout
the entire period before 1600. The hinga-kings were well into their
history before the holders of ngola reached their peak towards the
middle of the sixteenth century. The ngola a kilucmje state eclipsed
all but the most remote of the /wnga-kings before it, too, bowed to
outsiders, the Imbangala and the Portuguese, who introduced new
political structures of a fundamentally different type in the opening
decades of the seventeenth century.
 C H A P T E R IV

N ew Ideas from the South

T he f o c u s of the preceding chapter on indigenous Mbundu methods

of building states leading up to the sixteenth-century expansion of the
ngola a Iciluanje excluded any mention of an important contempor­
aneous kingdom called Libolo. Libolo, which was centred among the
Mbundu south of the Kwanza river, requires separate treatment
because its kings used political institutions entirely different from the
perpetual titles and linked descent groups of the northern Mbundu
state to restruct the Mbundu lineages into a kingdom. Although the
origins of Libolo as yet remain even more obscure than those of the
lunga and the ngola, its major organizing technique, a title called a
vunga, has left clear evidence that it came from regions outside the
Mbundu area where positional succession and perpetual kinship
reigned supreme. The ?vunga Iembodied a conception of authority
which was structurally the opposite of the hereditary titles awarded
to the lineages by the /«nga-kings and the ngola a Iciluanje since it
introduced for the first time a type of position which lay outside the
control of the descent groups!

The Kulembe
Although the vunga reached the Mbundu through the intermediary
of the Libolo political system, these titles had originated in an earlier
state known only by the title, kulembe, of a shadowy line of kings
who claimed authority over portions of the Benguela plateau
several centuries before the present Ovimbundu kingdoms in that
region were formed. The capital of these rulers lay somewhere near
the sources of the three major rivers draining the north-western
part of these highlands, the Longa, the Kuvo, and the Ngango.1
Documentary and oral sources concur in dating the rise of the
1 Kings bearing the title o f Kulembe kwa Mbandi still preserved this name
in the 1850s when they lived in ‘Selles’, roughly the highland area just west
o f the sources of the ICuvo river. See László Magyar (1859), p. 379. An area
known as ‘Lulembe’ still existed somewhere on the highlands south of
the Kwanza in the late seventeenth century; Cadornega (1940-2), iii. 249.
The prefix to the -lembe root varied considerably in written sources o f
this period.
90 NEW ID E A S F R O M TH E SO U TH
kulembe to long before the mid-sixteenth century.2 It was thus one
of the earliest Benguela Plateau kingdoms now visible in the surviving
evidence. None the less, the significance of the kulembe state has
eluded most historians, who have described the area south of the
Kwanza almost exclusively in terms of the later Ovimbundu states.
Neglect of the kulembe has probably resulted from the fact that the
traditions of the later Ovimbundu kingdoms, like those of the
Mbundu, date from the formation of these political entities (roughly
the late seventeenth century) and have eliminated all but the faintest
traces of earlier historical periods.3 Outside the Ovimbundu area,
however, Mbundu perpetual genealogies have preserved memory
of the kulembe title as the progenitor of a series of derivative political
titles which ultimately led to the later dynasty of Mbondo kings.4
The social and political structures of these pre-Ovimbundu states
remain too poorly known to justify speculation beyond the fact that
the kulembe apparently ranked among the most important early
states in the Benguela highlands.5

Expansion o f Libolo
The distinctive political institutions which originally emanated from
the kulembe reached the Mbundu indirectly through an intermediary
dynasty of kings who bore the title of hango. These rulers built another
kingdom, now known as Libolo,6 located somewhat north of the
kulembe among Mbundu who lived on the Ngango affluent south
2Testimony o f Alexandre Vaz, 31 July 1969; the date is deduced from the
position of the kulembe title in a genealogy including later perpetual names
known through documents to have become powerful well before 1600. E. G.
Ravenstein (1901), p. 85 (where the spelling o f the name was ‘Elernbe’), and
Cavazzi (1965), i. 188-90.
3The best summary o f the history o f the later Ovimbundu states is Childs
(1949), pp. 164-90. New fieldwork on these kingdoms, and their successors, is
urgently needed. One obvious direction in which such studies might look is
towards the builders o f the numerous stone ruins which now dot the Benguela
highlands.
4This instance of remnant traditions surviving outside the area o f their
origin parallels other examples of traditions no longer extant among the
Imbangala but still alive among the Colcwe, Songo, and Ovimbundu; the
following chapters provide examples.
5Beyond the position o f the kulembe in the perpetual genealogies, most
documentary sources from the seventeenth century contain references to the
kulembe as a ‘great and powerful king’. The vague quality o f these references
confirms the impression that the kingdom had declined by that time.
6The ancient Mbundu probably used the name Libolo only for the regions
south o f the Kwanza where the hango kings had their capitals. Mbundu king­
doms usually took their names from the title o f their kings, in this case the
hango. Libolo, is however, the name used today by Mbundu historians.
 E X P A N S IO N O F LIBOLO 91

of the Kwanza. The hango expanded their control in part by granting

perpetual subordinate titles to the lineages as the ngola a kihtanje
had done, but they also relied on a type of temporary position which
retained a greater concentration of power in the hands of the central
kings. Although, once again, a nearly complete lack of historical
data on Libolo precludes detailed reconstruction of the rise of the
kingdom, it was one of the earliest large states to appear among
Kimbundu-spealcers, probably contemporary with the Pende lunga-
kings and certainly older than the ngola a kiluanje. It evidently
flourished well before the mid-sixteenth century, when documents
implicitly showed that it had already fallen into decline and had been
replaced by the ngola a kiluanje everywhere north of the Kwanza.
Mbondo perpetual genealogies identify Hango dya Kulembe, the
Libolo king, as a ‘son’ of the kulembe. While these traditions accu­
rately trace the derivation of the authority claimed by the hango,
 NEW ID E A S F R O M T H E SO U T H
they do not describe the historical process by which Libolo kings
grew powerful and eventually asserted their independence from the
older state to the south. The predecessors of the hango-kings might
have existed among the southern Mbundu for some time, probably
under another name, before they obtained a ‘son’ title from the
Icitlembe at whatever time the influence of the older lcingdom pene­
trated the upper Ngango region. Alternatively, the lculembe may have
first asserted their authority over the southern Mbundu lineages
together by imposing a provincial governor to rule them under the
title of hango. Whatever the origins of these kings, the history of
Libolo comes into clear focus only after the hango had prospered for
some time and spread their authority to the Mbundu living north of
the Kwanza. At its height, the kingdom reached as far north-east
as the Pende living in the highland sources of the Kambo river above
the Baixa de Cassanje. Its eastern provinces included the Songo
who dwelt north of the Luhando river. The southern Libolo bound­
ary with the kingdom of the lculembe remains undetermined for
want of evidence.
The locations of a number of seventeenth-century political titles
surnamed ‘Hango’ establish the former boundaries of Libolo along
and beyond the lower Lukala in the north-west. Officials subordinate
to the ngola a kiluanje entitled Kalculu lea Hango governed the Muselce
and Ilamba region along the lower Lukala river as late as 1592.7
Even though they had by that time become provincial governors of
the Kgo/a-kingdom, their titles revealed that they had originated as
outposts of Libolo and had survived the decline of Libolo to
be incorporated in typical Mbundu pattern as elements in later
political structures. Other /(««go-positions dotting this area provided
evidence of former Libolo control: Kiluanje lcya Hango lived in the
Lukamba area of the middle Lukala, and Ngungu ya Mbulcu wa
Hango held the north bank of the Kwanza just below the falls at
Kambambe.8
7 Cadornega (1940-2), iii. 235,240, placed Kakulu ka Hango on the southern
bank o f the Kwanza near Mushima in the late seventeenth century, where he
had probably fled from the advancing Portuguese; Sala a Hango remained
in Ilamba at that time.
8 Domingos de Abreu e Brito, ‘Rella?ao breve das cousas, que se contem
neste tratado dangola e Brazil’ (B.N.L., MS. 294); published in Alfredo de
Albuquerque Felner (1931), and AA, iii, nos 25-7 (1937), 249-90; also excerp­
ted in Brasio (1952-71), iv. 533-45. The Portuguese spellings were quaquluqui-
ambo Quiloange quiambo, and Gungu ambo cambo; comparison with
Cadornega (1940-2) leaves no doubt that these spellings all represented
‘Hango’.
 E X P A N S IO N O F LIBOLO 93
Elsewhere, only a few hereditary named titles of Libolo origin
appear in written sources or oral traditions. Lukunga, a position
identified as subordinate to the hango, ruled the area north of the
confluence of the ICwije and the Kwanza.9 Lungu dya Hango lived
in Ambaca during the latter half of the seventeenth century and,
although his presence was not recorded earlier, had probably been
there for some time.10 Another title-holder whose capital occupied
one of the islands in the upper Kwanza, Mbola na Kasashe, appar­
ently served as guardian of Libolo’s south-eastern frontiers.11
The north-eastern Libolo frontier, to which I shall return, ran along
the escarpment above the Baixa de Cassanje.
The central provinces of the Libolo kingdom during its most
prosperous periods lay in the valley of the Ngango river. The later
Portuguese name for this area, ‘Halco’, identifies it as the seat of the
hango-Icings since the European toponym probably represented a
European corruption of the Kimbundu title.12 The perpetual
genealogies show a concentration of hango-demed titles in the Halco
region, including the kaza ka hango, a position whose incumbents
became famous in the 1620s as allies of the Mbundu queen Nzinga in
her battles against the advance of Portuguese military control. The
exact location of these kings’ capitals has not been found nor, to my
knowledge, systematically sought.
The northern boundaries of Libolo contracted under pressure from
the expanding kingdom of the ngola a kiluanje during the sixteenth
century. Beyond loss of the Lenge titles such as Kalculu ka Hango,
which became subordinate officials of the Samba kings, other former
Libolo positions like the Lukunga sent women as wives to the
ngola a kiluanje and received in return subordinate ngo/n-positions
as agents of the central rulers. The southward expansion of such
ugo/a-title-holders as the Kitelce kya Bengela, the Ngama a Ngola,
and others drove the finngo-kings out of their home provinces
in Halco. Only a few hango-titles survived there as minor officials
at the courts of the new representatives of the great ngola in
9Testimony of Sousa Calunga, Kambo ka ICikasa; Lukunga was said to
have been a 'kata' of the hango, that is a lineage title belonging to a descent
group with the right to elect incumbents to the royal hango-title.
10Cadornega (1940-2), iii. 244.
11Various testimonies o f Sousa Calunga.
12Cf. p. 90, n. 6 above, on the tendency o f Mbundu to name regions after
political officials, e.g. Hango/Halco. The demonstrated irregularities o f seven­
teenth-century spellings o f Kimbundu words (‘Hango’ in particular) and
parallel sound shifts in other words establish the identity o f ‘Hango’ and
‘Halco’.
94 N EW ID E A S F R O M T H E SO U T H
Ndongo.13 By 1600, only the far south-western province of greater
Libolo, the area nearest ICisama, remained as the last refuge for its
once-powerful rulers. Hango-Icings remained as minor local rulers
in this remnant state on the southern bank of the Kwanza until at
least the end of the seventeenth century.14
The hango of Libolo managed to dominate a wider area than any
previous Mbundu Icings (before the ngola a Icihtanje) by awarding an
appointive political title called a vunga (plural, mavunga). Unlike the
perpetual hereditary titles of the Mbundu, the vunga carried only
temporary authority delegated by the Icing to his appointed nominee.
The Libolo Icings seem to have ruled through classic Mbundu
perpetual named positions only in their home province of Halco
and among the north-western Lenge lineages. Elsewhere, beyond the
limits of the relatively easily controlled home territory, the hango
ruled through provincial governors holding mavunga. The hango-
lcings awarded these titles on their own initiative, usually through
presentation of some symbol indicating the mandate of the recipient,
and these appointments lasted only until revocation or until the
death of the individual title-holder. The appointee’s emblem then
returned to his lord rather than remaining with his heirs.
Since vwwga-holders could not bequeath their positions to kinsmen,
the mavunga did not become heritable lineage-controlled titles as did
the Mbundu named perpetual positions. The tendency of individual
lineages to gain control over the perpetual variety of title had hind­
ered the lunga-Icings’ efforts to build large centralized states, and the
kingdom of the ngola a kiluanje suffered from continual strife among
its component lineages. But, since the hango-kings could appoint
men who owed their positions solely to the central ruler and would
give undivided loyalty to him, no similar instability limited the
expansion of a state based on mavunga. The vunga enabled the
hango to overcome an additional disadvantage which had beset
kings holding malunga. The rigid correspondence between the
/HHga-positions and the hydrography of the land made these titles
an inflexible means of expansion, while Libolo Icings, on the other
hand, could appoint as many vwwgo-holders as they found expedient
and could locate them wherever and whenever they wished.
The nature of the Mbundu oral traditions filtered out all direct

13 Brandão (1904), p. 137, noted a mukilci a hango in such a position at the

end o f the nineteenth century.
14 Cadornega (1940-2), iii. 240, recorded the presence o f title-holders o f this
name.
 E X P A N S IO N O F LIBOLO 95
information on how mavunga positions functioned in the sixteenth
century. The perpetual genealogies preserved a record of only the
hereditary lineage titles, and, because the mavunga did not fit into
any of the limited categories of preserved information, no recollec­
tion of these titles could survive. Still, indirect evidence derived
from mavunga awarded much later by the political heirs of the
hango, Imbangala and Mbondo kings, allows a sketch of the func­
tions and significance of the ancient Libolo titles.
Mbondo and Imbangala mavunga obliged their holders to perform
specialized duties in support of the Icing and his court.13*15 Some
vwngfl-holders, for example, sent supplies of food to the royal capital;
others maintained armed forces ready to defend the king against
outside invaders. Still others might keep the roads and paths running
through a specific territory in good repair. Each vwnga-holder
discharged his obligation to his king by extracting labour and wealth
from the people who lived in the area assigned to him. The demands
made by the vtOTga-holders created tensions which divided them from
the lineages and placed them in opposition to the lineages’ interest
in autonomy. They thus depended exclusively on the king for their
authority and had little chance of raising the local kinsmen in
rebellion against the hango. The mavunga titles thus produced a
relatively centralized state structure, one beyond the reach of the
restless Mbundu lineages, which helped the hango kings expand
north and south of the Kwanza.
The history of the Libolo province (and later the Mbondo king­
dom) of the ndala (Icisua) offers the best evidence to date on the nature
of hango overrule among the northern Mbundu. It provides detail
about the names and functions of some mavunga and shows how the
Libolo governors’ titles lost their independence from descent group
control and became little more than lineage titles when the power of
the central hango-kings declined. The lineages of the area which
became the ndala province of Libolo had once formed part of the
relatively undifferentiated Pende population which extended from
the Kwango west to the middle Lukala river. The Pende of the high­
lands east of the Lushindo river and north of the Kwije16 acquired
a distinctive identity as Mbondo when the hango-kings brought them
into the Libolo kingdom under the immediate authority of a vunga-
holder called the ndala, or mamba, a species of highly poisonous
13 Testimonies o f Alexandre Vaz, Domingos Vaz, 26 June 1969; Alexandre
Vaz, 30 July 1969; Alexandre Vaz, Domingos Vaz, Ngonga a Mbande.
16 Testimony of Kasanje lea Nzaje.
96 N EW ID E A S F R O M T H E SO U T H
snake found throughout southern Angola.17 The prevalence of similar
ndala positions south of the Kwanza connects the origin of the
Mbondo ndala with the Benguela highlands beyond Libolo, perhaps
in Kulembe itself. Elsewhere in the highlands, the Ovimbundu,
present-day descendants of the ancient people of the kulembe,
regard the ndala snake with great awe, describing it as a magical
serpent which dwelt high on the inaccessible slopes of mountains and
could fly mysteriously through the air. The modern Ovimbundu
imply that the ndala is one of the oldest representations of political
authority known to them, since they sometimes associate it with the
mulemba tree and recall that the great ndala fled (i.e. declined into
insignificance) at the arrival of the present political authorities, whose
power rests on other bases.18
The location of the Mbondo ndala1s capital and the shape of their
domain suggest, in the absence of better information, the role they
played in the larger Libolo kingdom. The ndala’s capital occupied a
nearly inaccessible and easily defended location on the crest of the
1,500-foot-high escarpment of ICatanya.19 The eastern and northern
limits of the Mbondo ndala’s authority lay just beyond the foot of
these cliffs and abutted the lands of Pende /wnga-kings living in the
Baixa de Cassanje. The bulk of the Mbondo kingdom extended
through thickly wooded grasslands south-west from the capital
towards Libolo. From their secure redoubt high above the surround­
ing lowlands, the ndala in effect guarded the north-eastern frontier
of Libolo.20 The identity of the threat which led Libolo kings to place
the ndala in that location is not known, but Butatu a ICuhonga lcwa
Wutu wa Nyama, the powerful Pende king of that period, is a likely
candidate. A very early trade route running from the Kwanza
through Mbondo to the salt marshes of the Luhanda river in the
Baixa de Cassanje might have drawn the attention of the hango-
kings in that direction.
A certain vagueness in the perpetual genealogies surrounds the
17 See Statham (1922), p. 280, who gave the species as Dendraspis anguisticep.
18 The Mbundu refer explicitly to the alien origins o f this title when they
say that the words comprising it mean ‘different language’ (which they do not
in any but a symbolic way); testimony of Kingwangwa leya Mbashi. For data
on the Ovimbundu ndala see A. Hauenstein (1960, pp. 224, 231; 1964, p. 930;
1967a, p. 921).
10 Testimonies of Sousa Calunga, 21 and 22 July 1969; confirmed for the late
nineteenth century by Capello and Ivens (1882), ii. 15.
20 Both the Mbundu and the Ovimbundu saw the magical variety of the
serpent as a supernatural guardian; P° Albino Alves (1951), i. 812; Cavazzi
(1965), i. 210.
 E X P A N S IO N O F LIBOLO 97
origin of the present Mbondo rulers’ title, now called the ndala
kisua. This uncertainty paradoxically identifies it as a vunga and
simultaneously confirms its great antiquity. Most modern Mbundu
genealogies metaphorically describe the foundation of the Mbondo
kingdom as the arrival of Kajinga ka Mbulu, a fictional founding
ancestress of the Mbondo lineages. Kajinga ka Mbulu, according to
the narrative accompanying the genealogies, settled in an area called
Lambo in the highlands near the sources of the ICambo river. There
she ‘married’ a male known by a great variety of names: Ndala a
Kikasa, ICikasa lcya Ndala, Kikasa lcya Kikululu lcya Hango, Kikasa
lcya Hango, or Ndala Kisua.21 The uncertainty about the name of
Kajinga ka Mbulu’s husband on the part of modem informants
indicates that his title originated at a time before the creation of the
present Mbondo state and the associated traditions. As a result,
Mbondo historians have no means of placing it securely in the
framework established by more recent genealogies and consequently
disagree about its ‘true’ position, giving it a variety of names which
reflect their feelings about its proper location. Its origin as an appoint­
ive vunga rather than a perpetual position would have produced the
same effect. If such arguments from the nature of the data establish
that the ndala came to the Mbondo at a very early time, the hypothe­
sis that it originated as a Libolo vunga would account for its presence
when the Mbondo later adopted Kajinga ka Mbulu as their fictional
progenitor.
The occurrence of the name ‘Hango’ in two variants of the name
of Kajinga ka Mbulu’s husband almost certainly establishes the
connection between the Mbondo ndala and the hango kings in
Libolo. At least four other related positions, whose names identify
them as derived from Libolo, have survived among the lineages at the
centre of the Mbondo kingdom: Kyango lcya Hango, Kongo dya
Hango, Kilcango lcya Hango, and Kabele ka Hango.22 These titles’

21 Kajinga ka Mbulu should not be confused with the similarly-named

Mbundu queen Nzinga, whom the Portuguese usually called ‘Jinga’. The name
o f the later state o f ‘Jinga’ derives from the queen ‘Jinga’ and likewise has no
association with Kajinga. For the names, see testimonies o f the Mbondo group;
Solcola; Alexandre Vaz, 30 July 1969; Sousa Calunga, 29 and 30 Sept. 1969;
Kingwangwa kya Mbashi; Kimbwete; Mahashi; Kabari ka Kajinga; Apoio
de Matos, 8 July 1969.
22 Testimonies of Sousa Calunga, 30 Sept. 1969; Mahashi; Kimbwete;
Kingwangwa kya Mbashi. The Mbondo elision o f the connecting particle
with the surname Kyango Kyango, Kongo Dyango, Kabele Kango, etc. may
result from combining a surname which has no meaning in Kimbundu (e.g.
hcmgo) with familiar Kimbundu first names, leading speakers to combine it
98 N EW ID E A S F R O M T H E SO U TH
positions in the perpetual genealogies reveal the same ambiguities
which surround that of the ndala, thus suggesting that Mbondo
historians have here, too, faltered on the incompatibility of the
vunga titles with genealogies based on hereditary Mbundu positions.
One set of Mbondo narrative episodes implicitly emphasizes the
connection of the Mbondo state with Libolo by claiming that
Kajinga ka Mbulu came from somewhere south of the Kwanza.
Her alleged southern origins assume additional significance in this
case because this theme contradicts the usual insistence, found
everywhere else among the Mbundu, that Kajinga ka Mbulu came
‘from the sea’ along with the other ancestresses of the major ethno-
linguistic subgroups. Interpreted with care, the tradition also
accounts for the presence of certain non-lineage Mbondo officials
attached to the ndala kisua's court. The titles of these officials may be
independently identified as having belonged originally to people
living south of the Kwanza. The literal content of the tradition
obviously has no historical significance since Kajinga ka Mbulu
was a purely metaphorical representation of a group of lineages and
could never have ‘come’ from anywhere in the manner that the
tradition describes her journey
Since the tradition requires detailed criticism, I will begin by
paraphrasing the recorded version of the narrative.23 Kajinga ka
Mbulu once lived near Luanda24 with Ngola a ICiluanje. When the
Portuguese arrived there, Kajinga ka Mbulu and Ngola a ICiluanje
at first fought together against the European invaders but were
eventually forced to flee. Ngola a Kiluanje went north-east where he
settled on the Wamba river. Kajinga ka Mbulu fled in the opposite
direction across the Kwanza river to ‘Bailundo’.25 Although the
Kwanza presented a serious obstacle to Kajinga’s flight, she managed
to cross with the aid of Katumbi ka Ngola a Nzenza, a chief who
knew magical charms capable of transporting people across the river.
He placed Kajinga in a great trunk or box which floated across the
river like a boat. Katumbi ka Ngola a Nzenza, however, had deceived
Kajinga, for he intended to capture and kill her when she landed on
23Testimony of Fernando Comba, reproduced in Salazar (n.d.), ii. 140-1.
24An anachronism: the later Portuguese administrative capital at the coast.
25An anachronism: the Mbundu now use this term to denote all the Ovim-
bundu ldngdoms of the Benguela plateau. N o state o f Mbailundo existed at
the time these events allegedly occurred.

with the preceding connective particle. If so, this would provide added evidence
in support o f the southern, even non-Kimbundu origins o f the hango-kings’
title.
 E X P A N S IO N O F LIBOLO 99
the other side. Once she was safely on the southern bank of the river,
he attempted to trap her inside the box by sitting on the lid of the
trunk. Kajinga, well equipped with strong magic of her own, managed
to escape from the box and slew Katumbi lea Ngola a Nzenza. She
cut up the body and made his skin into a rope, a drum, a marimba,
and a bowstring which gave her control of her enemy’s magical
powers.
Kajinga ka Mbulu then resumed her flight from the pursuing
Portuguese. Her sorcerer, Muta a Kalombo,26 flooded each river
after they had crossed it to prevent their enemies from capturing
them. She eventually fell in love with a man named Kima a Pata.
They married, but their marriage produced no children until, after a
number of years, Kajinga consulted a diviner about her sterility.
He attributed the problem to an unspecified transgression committed
during her flight from the Portuguese when she had crossed the river
Kazanga without permission.27 The diviner advised Kajinga to
expiate her guilt by throwing a charm made of eggs, palm oil, and
itshila28 into the Kori river.29 Kajinga did this and became pregnant
almost immediately; she bore five children in all, Kikato kya Kajinga,
Kisua kya Kajinga, Nyange a Kajinga, Yivo ya Kajinga, and
Mupolo wa Kajinga. She then travelled north with her husband and
children to Kabatukila where they settled near the cliff's surrounding
the Baixa de ICafushi, a valley at the edge of the Baixa de Cassanje.
The following interpretation of this narrative elucidates the ways
in which it repeatedly indicates southern origins for the modern
titles whose names appear as characters in the narrative. The first
two episodes, those describing Kajinga ka Mbulu’s sojourn with
Ngola a Kiluanje at Luanda and the retreat of Ngola a Kiluanje
towards the Wamba, are conventional beginnings to most recitations
of Mbundu history and have little more historical meaning than the
‘Once upon a time . . . ’ which opens many English stories. Although
no queen Kajinga ka Mbulu ever lived in Luanda, modern Mbondo
historians customarily place her there in order to establish her as an
20The meaning o f kalombo in Kimbundu is ‘sterility’; Assis Jr. (n.d.), p. 87.
27 The name ‘Kazanga’ almost certainly dates and locates this narrative,
since it was an archaic name for the highlands south o f Libolo which later
became known as Mbailundo; A. V. Rodrigues (1968), p. 183, implied that
the name probably predated the sixteenth-century (?) movement o f the kitelce
kya bengela title to Kibala.
28 Word not identified.
29Probably the ‘Guri’, a minor stream near the mouth o f the Luhando river;
see Map VI. Anton E. Lux (1880), map, showed it as a ‘town’. Cf. Petennanns
Geograpliische Mittheilungen, ii (1856), Tafel 17.
100 N EW ID E A S F R O M T H E SO U T H
equal of Ngola a Kiluanje, the touchstone of historical greatness
among the Mbundu. Reference to the arrival of the Portuguese
ordinarily explains both the present location of the ngola a kiluanje
on the Wamba river, clearly an interpolation which falls well beyond
the main line of the narrative, and ICajinga’s departure from her
‘ancestral home’. These episodes have no historical significance
whatsoever.
The next episode, detailing ICajinga’s encounter with ICatumbi ka
Ngola a Nzenza, includes the first historical data in the narrative
since it serves to account for certain authority symbols held by later
Mbondo kings, in particular a drum, a cord, and a bowstring said
to have been fashioned from human skin, as well as a marimba. The
image of a ldng facing difficulties in crossing a river is a cliche which
recurs throughout Mbundu narrative episodes to explain an innova­
tion in political or social structure. The river presents an obstacle
which the ruler must overcome by introducing some drastic change
in the rituals and symbols connected with his position. The Kwanza
in this case constituted an effective metaphor for specific historical
difficulties not recorded in the tradition, since all Mbundu were
familiar with the hindrances to travel presented by this large water­
course.
The alleged location of ICatumbi ka Ngola a Nzenza on the lower
Kwanza near Luanda derives from two sorts of factor, some historical
and others non-historical. The standard non-historical setting of the
opening narrative in Luanda forces the Mbondo historian to have
ICajinga cross that river in order to place her in the southern regions
where the rest of the story must take place. In one sense ICajinga’s
crossing of the Kwanza is no more than a fictional device employed
to connect two different parts of the historian’s plot, a location
determined more by the logic of the narrative than by historical fact.
As it happens, however, ICatumbi ka Ngola a Nzenza probably did
live just south of the Kwanza, since documentary sources locate a
so-called ‘Jaga’ Nzenza a Ngombe in Libolo as late as the early
seventeenth century.30 The authority symbols mentioned at this
point in the narrative, the cord, drum, bowstring, and marimba,
have elsewhere been associated with ‘Jaga’ Icings in Angola. The
presence of the name Nzenza in both titles and the coincidence of the
‘Jaga’ symbols probably connect the Katumbi ka Ngola a Nzenza
30 Bishop D . Simao Mascarenhas to el-Rei, 2 Mar. 1624; Brasio (1952-71),
vii. 199-203. See also Elias Alexandre da Silva Correa (1937), i. 238, and
Joao Carlos Feo Cardoso de Castello Branco e Torres (1825), p. 164.
 E X P A N S IO N O F LIBOLO 101

of this tradition with Nzenza a Ngombe of the documents. The

episode of ICatumbi ka Ngola a Nzenza represents the first historical
element in Kajinga ka Mbulu’s fictitious journey, and accounts for
Libolo emblems of power acquired by the ndala Icisua long before
the formation of the present traditions.31
The Mbondo historian has introduced Muta a Kalombo into the
narrative of Kajinga ka Mbulu’s journey because the prototypical
muta a kalombo title originated as a Libolo vunga entrusted with
control of certain supernatural powers. Appointive officials holding
this title today act as advisers and supernatural specialists at the
courts of major Mbondo title-holders.32The remainder of this episode
consists of a series of images found commonly in Mbundu narrative
episodes. The cliche describing charms which cause a river to rise
and thus impede the progress of pursuing enemies occurs often in
other traditions; such spells formed an essential skill attibuted to
some kings and were closely associated with rain-making techniques.
The apperance of Muta a Kalombo at this point in the narrative
explains how Kajinga might have escaped from the trailing Portu­
guese army, a feat regarded (for at least the last century in Mbondo)
as difficult if not impossible to achieve.
Kima a Pata,33 Kajinga’s ‘husband’ in the narrative, was a title
31It was the location o f the episode south o f the Kwanza which led the
Mbondo historian to include the story o f Katumbi ka Ngola a Nzenza in his
narrative. The narrative provides a good example of the way in which the
Mbondo reorganized much older fragments o f traditions to make them
conform to the framework of the new musendo developed in the seventeenth
century. A fictitious journey o f the symbolic founder in the new traditions,
Kajinga ka Mbulu, leads her in picaresque fashion through all the areas where
earlier forms of Mbondo authority originated; it therefore took her south of
the Kwanza to Libolo. Owing to the basic structure of the narrative as a journey
assumed (falsely) to have begun in Luanda and to have ended in Lambo,
geography rather than a sequence of actual events determines the order in
which Kajinga is envisaged as having acquired the symbols in question.
According to geography, for example, she must have reached Katumbi ka
Ngola a Nzenza before she came to the Kuri river. The narrative in fact
provides no basis for estimating the order in which the various symbols and
titles in fact reached the Mbondo.
32Testimony o f Sousa Calunga, 16 June 1969. Chatelain (1894), p. 11, had
Muta a Kalombo as a ‘demon’ in Mbundu mythology who was ‘king or
governor’ o f the forest, hence in control o f hunting and travellers. Mattos
(1963), p. 337, said he was the ‘god of hunting’. The ability to control rivers
was attributed especially to the class of kibinda professional hunters. Linguistic
analyses presented elsewhere show that the root -lombo may be uniquely
associated with Umbundu languages, thus reinforcing the hypothesis of
southern origins for this title.
33Kima appears in the genealogies with various surnames. The kima is a
species o f baboon closely associated with certain political positions.
102 N E W ID E A S F R O M T H E SO U T H
which belonged to the Swela lineage group which today inhabits
both banks of the Kwanza river above its confluence with the
Luhando. This area of origin, on the edge of the old Libolo central
province of Hako, tallies with the location of the Kori river men­
tioned in the narrative to identify the kima as a Libolo position.
Modern Mbondo cite the Swela merely as convenient substitutes for
the Libolo, the true originators of the title, since they have forgotten
the position’s ancient affiliation and invoke the Swela because they
are the only modern Mbundu lineages with titles of southern, i.e.
‘Bailundo’, ancestry. The Swela in fact acquired their present
Ovimbundu titles only in the eighteenth century, long after the
events described in this tradition. The reference to Kima a Pata
therefore comes from a much earlier period, when the Swela had
Libolo titles, and thus reiterates ICajinga’s connection with Libolo
even though modem Mbondo historians no longer recognize the
true significance of the title.
The ‘marriage’ between Kajinga and Kima a Pata described in the
following episode is an imaginative description of the union of the
kima a pata title and the lineages represented by the figure of Kajinga.
The image of a ‘marriage’ accords with the generalized model of
title-lineage pairings in the perpetual genealogies. It metaphorically
represents the incorporation of a Libolo title by the Mbondo
descent groups resident in Lambo.
These events probably took place in Mbondo rather than some­
where to the south, since such political symbols and ideas as the
kima travelled more easily by diffusion than lineages could move by
migration. The title of kima a pata still survives in Mbondo as a
remnant from the period of Libolo rule when holders of this title,
probably subordinate to the ndala, lived in the area of Kabatulcila.
ICajainga’s sterility metaphorically represents the fact the Mbundu
regarded earlier emblems of authority as somehow inadequate to
insure their prosperity. The historical circumstances might have
involved almost any misfortune: sickness, lack of rains, or the
inability to unite sufficient armed forces to repel an invader, but the
tradition typically gives no clues to the nature of the historical
difficulties. ICajinga’s inability to bear children thus indicates that
the Mbondo lineages welcomed the new political system brought
by the ndala and the kima a pata. This idea contrasts with other
episodes in this tradition and in other narratives which explain the
introduction of new political symbols in terms of conflict and
suggests the need to explain why the Mbondo did not resist the
 E X P A N S IO N O F LIBOLO 103
imposition of Libolo overrule. No data known to me solve this
riddle.
The image of a special charm involving offerings to water spirits34
alludes metaphorically to the supernatural powers which the Mbondo
attributed to their vunga-kings. They regarded these powers as
crucial to Mbondo prosperity and helpful in spreading Libolo
authority symbols to neighbouring descent groups. The union of the
kima with the Mbondo lineages proved ‘fertile’ in the sense that it led
to the creation of at least five (but probably more) subordinate titles,
the ‘children’ named in the narrative. These new positions almost
certainly once used ‘Kima’ as their surname to indicate the Libolo
sources of their legitimacy, but the later decline of the hango-kings
allowed the Mbondo to change the surnames to indicate a more
autonomous association with Kajinga. In a fashion similar to the
apparent surname changes which some of the Yongo hmga-titles
underwent, this tactic provided a local, if falsified, source of legiti­
macy for title-holders who had lost their foreign sponsor. Although
these positions may once have been important elements in the Libolo
state structure, they became obsolete when the Mbondo threw off
hango overrule and survived to the present with sharply diminished
functions. The narrative of Kajinga’s movement from Luanda to
Mbondo, interpreted according to the rules governing Mbundu
oral traditions, thus points repeatedly toward the south, that is
Libolo, as the source of the most ancient known Mbondo political
titles.
The titles of several Mbondo court positions confirm this inter­
pretation of the tradition since they show clear linguistic affinities
with titles known elsewhere only south of the Kwanza. The equivalent
positions at the capitals of most other Mbundu kings bear different
names. The three most important Mbondo court titles, all mavunga,
bear the names balanga (or palanga), kasanje, and Icitashi.35 At least
since the sixteenth century the northern Mbundu officials who per­
form the same functions have been called tandala, ngola a mbole,
and muzumbo.36 Of the uniquely Mbondo terms, balanga certainly
comes directly from sources in the Benguela plateau region; the
term does not exist in Kimbundu but in modern Umbundu
:,+ Malungal The tradition does not identify them, but new rulers from the
south would logically have propitiated the spirits o f the former Pende owners
of the land.
35Testimonies o f Sousa Calunga, 16 June 1969; Kasanje lea Nzaje; Kasanje
ka Nzaje, Kitubiko, and Nzaje. See also Salazar (n.d,), ii. 102.
36For tandala and ngola a mbole, see Rodrigues (1936).
104 N EW ID E A S F R O M T H E SO U T H
refers to a court official with the same duties as the Mbondo
position.37
Despite great confusion in European sources about the significance
of the term kasanje (it occurs as the name of the later Imbangala
kingdom in the Baixa de Cassanje, as the title of its Icings, and as the
name of the regions where they ruled), it clearly originated as an
appointive vungci position somewhere south of the Kwanza, probably
in the kulembe or Libolo kingdoms. Kings all over the Umbundu-
speaking area appointed vw«ga-olficials with this title, while the
Mbundu and their northern neighbours apparently had no such
term. The Ovimbundu kingdom of Wambu contained a ‘subchief’
called kasanje.38 The title turned up again on the south side of the
Kwanza when the Portuguese encountered a chief known as Kasanje
ka Yela near the Bay of Quieombo during the 1640s; the region was
at that time inhabited by speakers of Umbundu.39 The title usually
written as Mbola na Kasashe, the Libolo noble who occupied the
islands in the upper Kwanza, may have been mbole na kasanje before
Portuguese language documents distorted it through inaccurate
transcription.40 The term kasanje originally meant any sort of guard­
ian, with the implication that this official had no autonomous powers
but merely administered the forces inherent in objects belonging to
others. The Imbangala still used the word with this meaning in
the nineteenth century as the name of a diviner, the kasanje ka
mbambo. His name, translated literally, meant the ‘guardian of the
mbambo’ basket of divining objects.41 It also occurred as kasanje

37 Alves (1951), ii. 1045.

38 Gladwyn M. Childs (1964), p. 376. The reference would be to an appointed
official, not a ‘subchief’ properly speaking, and would represent a title o f pre-
Ovimbundu origin rather than evidence o f the later passage o f the Imbangala
as Childs suggested.
39 Letter from Antonio Teixeira de Mendonça, 14 Sept. 1645 (A.H.U.,
Angola, cx. 3, cap. 8).
40 The term mbole occurs in the context o f an Mbundu political official’s
title, the ngola a mbole. The Songo who lived around Mbola na ICasashe spoke
Kimbundu and would have used this term for the position. The name kasanje
came from Umbundu-speakers through Libolo domination o f the area. Because
the title kasanje meant nothing in Kimbundu (in fact, it had a derogative sense),
this chief acquired a double title that incorporated one element from each
language. Thus he became the Kimbundu (ngola a) mbole and (na) the
Umbundu kasanje.
41 Capello and Ivens (1882), i. 384, and picture facing. The Imbangala still
use the term to describe one o f their divining specialists; testimony of Apolo
de Matos, 6 Oct. 1969. The word kasanje also occurred in reference to other
sorts o f guardian. Schiitt (1881), map, observed accurately that the exact
meaning o f the word kasanje was ‘guardian’; he translated it as Ver waiter, a
 E X P A N S IO N O F LIBOLO 105
lea ngongo to denote the guardian of the Kasanje king’s double
gong.42
The word kitushi does not occur in standard Kimbundu but the
root -tusi in Umbundu (equivalent to -tushi in Kimbundu) means an
insult or injury in the specialized sense of an offence permitting the
injured party to claim redress through a chief’s court.43 The prefix
Id- in Kimbundu can indicate the person in charge of the object
denoted by the following lexical root.44 The Mbondo kitushi, in fact,
hears cases brought before noble title-holders; the Umbundu root
-tusi and the Kimbundu prefix lei- combine to indicate exactly this
function in the Kimbundu-ized form, kitushi. The Mbondo also have
an official called the lumbo, the same title as that of a seventeenth-
century vunga brought from Libolo or from the Jeniembe.45 These
Mbondo titles possess characteristic features of the original mavunga
since they are appointed and the men who hold them serve only at the
discretion of their king.
Two final details complete the chain of evidence linking the
Mbondo ndala Idsua with the ancient kingdom of Libolo. Mbondo
nobles, alone among their Mbundu neighbours, cannot eat the flesh
of an ox. Since, in Angola, only people living south of the Kwanza
draw this sort of connection between their nobility and cattle, the
extension of the custom to the Mbondo points once again to Libolo
and to the Iculembe (whose ‘wife’, Mbumba a Nyasi, sunned herself
on an oxhide, it was said) as the source of their political institutions.
Finally, the title ndala Idsua itself may support the hypothesis through
the linguistic evidence it offers. Since the word ndala means mamba
in both Kimbundu and Umbundu, this term does not help to
trace the origin of the title. But the word kisua provides a
historical etymology in Umbundu, but not in Kimbundu, which
seems to identify Libolo as the original overlord of the Mbondo
state. The form Idsua appears in neither modern language, but the
12Salles Ferreira (1854-8).
13Alves (1951), ii. 1576.
14Héli Chatelain (1888-9), pp. 120-1.
45 Testimony o f Kasanje ka Nzaje; cf. Cavazzi (1965), i. 192. Although
modern dictionaries provide little aid in locating the sources of other Mbondo
titles o f this type, those already described adequately demonstrate the connec­
tion of Mbondo to the south. Testimony of Kisua kya Njinje; Sousa Calunga,
16 June and 9 July 1969; Sokola; Kasanje ka Nzaje; Kasanje ka Nzaje,
Kitubiko, and Nzaje; also Salazar (n.d.), ii. 102, give tope, kikwiku, kishinga, a
mbambi, Iwamba, and ndala a makita.

German word used to denote various managers, stewards, trustees, etc., all
o f whom control property belonging to others.
106 N EW ID E A S F R O M T H E SO U T H
seventeenth-century reflex of the word took the form kisuba
or kisuva, 4B which turns out to be a western ICimbundu word
meaning ‘that which remains, or something left over’.47 The full title
of the Mbondo Icing, ndala kisua, therefore designated the ‘remaining
ndala’ or the ‘ndala left over’ or behind among seventeenth-century
Mbundu who remembered the decline of Libolo.
Elsewhere, the hango kings’ domination of the Songo extended only
to the northern lineages of this group but lasted long enough to
leave traces which still distinguish these jingundu, called Kirima, from
their southern relatives, the Songo proper. The Kirima subgroup
lives west of the upper Lui and north of the Luhando river as far as
the Kwije, roughly within the boundaries of the farthest extension
of Libolo rule. The southern and larger group called Songo proper
live along the Luhando river and east to the Kwango.48 The distinc­
tion between Songo and Kirima had much greater significance in
the seventeenth century than the few refinements in dialect noticeable
today. The earliest documentary references to the area consistently
differentiated between its two parts as the ‘greater’ and ‘lesser
Ganguellas’.49 The name came from the western Mbundu who used
the word ngangela to designate all the people living to the east of
Ndongo and, although the term indicated no specific geographical
region as the Mbundu employed it, the Portuguese generally applied
it only to distinguish the two Songo ‘Ganguellas’. Lesser Ganguella
lay to the north and corresponded to the area now inhabited by the
Kirima; Greater Ganguella was Songo proper. Since the hango
kings had ruled the northern Kirima during the sixteenth century
but had not conquered the southern Songo, it was probably the
heritage of Libolo overrule which accounted for the distinction made
not long after during the seventeenth century between these two
groups of Songo lineages.

The Mbondo after the Decline o f Libolo

As the ngola a kiluanje expanded to the south-east during the
sixteenth century, it isolated the Mbondo from Libolo, leaving the
ndala province free for the first time to develop according to local
social and political conditions. The subsequent history of this state
46Balthasar Rebello de Aragao, ‘Rellasao’; Brasio (1952-71), vi. 332-43;
also in Luciano Cordeiro (1881), iii. 15. Cumulative errors of scribes and
editors caused the word to appear as ‘chicova’ in this source.
47Assis Jr. (1951), p. 143.
48Testimony of Sousa Calunga, 21 Aug. 1969.
49For example, Cavazzi (1965), i. 214.
TH E M B O N D O A F T E R T H E D E C L IN E OF LIBOLO
108 NEW ID E A S F R O M T H E SO U T H
illustrates how the Mbundu lineages of the Lambo region, roughly
the highland promontory thrusting north-east into the Baixa de
Cassanje, took advantage of relaxed hango authority to convert
Libolo mavunga into hereditary lineage titles closely akin to their
own perpetual named positions. Since no single lineage managed to
claim the central title of the ndala Icisua for itelf, the Mbondo descent
groups ultimately settled on a form of rotating succession in which
several groups of jingundu shared serially in the exercise of power.
For reasons indicated in connection with my discussion of the
Libolo origins of the ndala title, the genealogies dealing with the
formation of the Mbondo kingdom contain a great many ambigui­
ties, but a few places in the traditions seem to offer the hope of sound
historical interpretation. Most of the variation in the titles stems
from the fact that they were once mavunga and therefore had no
‘father’ position capable of fixing them securely in relation to other
titles in the perpetual genealogies. The first terms of the names,
which derive from their original stature as mavunga, vary less than
the surnames and in fact turn out to hinge on four identifiable posi­
tions which mark four phases in the early history of the Mbondo:
Kajinga, Kikasa, Kima, and ICingwangwa.
Kajinga, it is clear, represents the local descent groups and
stands in implied opposition to the other titles in the genealogy,
nearly all of which are political titles, mostly mavunga from Libolo.
The Mbondo recognized several subdivisions within the large group
of jingundu allegedly descended from Kajinga: Ndala a Kajinga,
Nyange a Kajinga, Mupolo wa Kajinga, Mbamba a Kajinga, Zombo
dya Kajinga, ICabari ka Kajinga, and doubtless others as well.50
The location of the lands of these lineage groupings, which probably
represent the earliest remembered subdivisions in the area, are
unknown except for those of Kabari ka Kajinga and Zombo dya
Kajinga who respectively occupied domains west and east of the
middle Lui. The others seem to have lived in the highlands some­
where in the vicinity of Lambo. Nor can the historical connections
between these groups be traced, since the Mbondo have no coherent
and articulated segmentary genealogy like that of the Songo; the
attribution to these groups of direct descent from Kajinga undoubt­
edly hides a good deal of genealogical telescoping.
The second phase of Mbondo political development came with
what the genealogies describe as a ‘marriage’ between Kajinga and
50 Compare the lineages named as descendants in the narrative tradition of
Kajinga, p. 99 above.
 T H E M B O N D O A F T E R T H E D E C L IN E O F LIBOLO 109
Kilcasa.51 Kilcasa, whose title seems to have been associated with
lineage titles surnamed mwinga or ICyango a Mbashi a Kitata kya
Mukombi, may have been an early Pende lunga-Icing before the
expansion of Libolo into the Lambo area. The ‘marriage’ between
Kilcasa and ICajinga represents an early period when holders of the
kilcasa title held some influence over lineages in the area and does
not affect the significance of the later ‘marriage’ between ICajinga and
Kima a Pata.
Kama, in fact, appears next in most Mbondo genealogies and
heralds the advent of Libolo overrule. He bears several different
surnames, an expectable variability given the probable origin of the
title as the kima a pata in Libolo. The vunga title of the ndala does
not appear at all in the Mbondo genealogies, but the kima a pata
stands as a record of Libolo control in its place. The kima could
have acquired this position in the genealogies if it declined to the
status of a hereditary lineage position when the Mbondo escaped
from Libolo domination and the lineages which possessed it provided
the incumbents for the ndala kisua position, apparently through
election by a number of Mbondo descent groups who jointly chose
incumbents for the central title.52 The evidence for this assertion
comes from the near-unanimous identification of the kima as a ‘son’
either of Kajinga lea Mbulu or, depending on the variant, of the
lcikasa; Mbundu historians evidently vacillate between the two older
figures as sires of the Kima vunga for want of better knowledge of its
true origins. Since the ndala (Icisua) never became the possession of a
single lineage, it cannot appear in the genealogies of permanent
names except occasionally as an interpolation. The incumbents in the
kima a pata, apparently in the hands of an unidentified lineage,
enjoyed strategic advantages through their occupation of a place
called Kabatulcila where a mountainous ridge provides the only
access to the Baixa de Cassanje from the highlands south of Lambo.
They ruled until the 1620s when a ‘son’ position, the Icingwangwa kya
kima, replaced its ‘sire’ as the dominant Mbondo title in conjunction
with the rise of the new Kasanje kingdom then forming in the Baixa.53
Conversion of the ndala kisua from a /zizngo-appointed vunga
to a title under the control of a few central Mbondo lineages allowed
the old Mbundu tendency towards political fragmentation to
51Testimony o f Kingwangwa kya Mbashi.
52Testimony of Kisua kya Njinje, Kambo ka Kikasa, Sousa Calunga.
53Testimonies o f Mbondo group; Alexandre Yaz, 30 July 1969; Kimbwete;
Kingwangwa kya Mbashi; Sousa Calunga, 22 July, 29 and 30 Sept. 1969;
Mahashi; Kabari ka Kajinga; Apolo de Matos, 8 July 1969.
110 N EW ID EA S F R O M T H E SO U T H
reappear in Mbondo. Some of the eastern Mbondo lineages, known
as Zombo dya Kajinga, broke away from the authority of the ndala
Icisua during the period in which the Mbondo kingdom was flourish­
ing independent of Libolo influence. They built a distinct and fairly
extensive network of lineage titles in the Yongo region between the
Lui and the Kwango. The genealogies show this development as a
‘marriage’ once again, this time between the lineages of Zombo dya
Kajinga and the ancient Pende lunga-king in the area, ICayongo lea
Kupapa. Holders of lineage titles derived from Zombo dya Kajinga
evidently overwhelmed the lunga-kings since the title of ICayongo ka
Kupapa went into decline and the titles descended from Zombo dya
Kajinga ramified through Ndungu ya Zombo in the next descending
generation to five or six positions on the succeeding level. ICilamba
kya Ndungu became the most important of these in the latter part
of the sixteenth century, so that the entire Baixa de Cassanje was
known as the kina (pit) kya kilamba by 1600. The lineages of Zombo
dya Kajinga never lost touch completely with the original Mbondo
kingdom in the highlands, as indicated in the genealogies by a
number of ‘marriages’ between the old Libolo mavunga (such as
Kikululu kya Hango and even the ldma a pata) and the Mbondo
lineages beyond the Lui.
The development of the subsidiary Mbondo state of Kilamba kya
Ndungu showed most of the classic features of state-formation
among the Mbundu. It made use of the usual Mbundu network of
perpetual titles rather than developing the potential of the mavunga
on the Libolo model. It also absorbed older titles present in the
region, in this case the Pende lunga-kings, by giving the ancient posi­
tions new surnames indicative of their incorporation in the new state.
In this way, for example, the /w/iga-priest Mahashi na Pakasa, one of
the old titles near the Luhanda salt pans, became Mahashi a Mah-
ongi, from Mahongi a Ndungu ya Zombo dya Kajinga, one of the
‘sister’-titles to Kilamba a Ndungu. If the incorporation of Mahashi
indicated a movement towards the salt pans, the kingdom of Kilamba
a Ndungu also conformed to the established pattern of states tending
to concentrate near valuable economic resources.
The history of the original Mbondo state of the ndala Icisua
exhibits rather different characteristics from the other early Mbundu
kingdoms considered here. It was spawned by the implantation of an
alien vunga title emanating from south of the Kwanza, unlike the
essentially local origins of the ngola and lunga states. It survived the
decline of its original sponsor, the hango-Icings of Libolo, only to
 T H E MBO N D O A F T E R T H E D E C L IN E OF LIBOLO 111

M ap vii. Expansion of the Mbondo Kingdom of the Ndala Kisua and

Subsidiary States (c. Sixteenth Century?)

become a local title claimed by the Pende lineages of the Lambo

region. There it thrived at a time when other Libolo title-holders were
in retreat before the advancing ngola a kiluanje or, as in the Lenge
region, becoming subordinate parts of the ngola kingdom. The
location of the ndala kisua on inaccessible mountain peaks along the
escarpment which rimmed the Baixa de Cassanje may have afforded
these kings a defensive position invulnerable even to the armies of the
ngola a kiluanje. In so far as the Mbondo state expanded, it moved
away from the large western Mbundu kingdom and perhaps only
coincidentally towards the salt pans which had figured so prominently
in the earlier political and economic history of the Baixa de Cassanje.
It suffered its first reverses when the ngola a kiluanje spread south of
the ndala kisua towards the upper Lui and moved north of the
Mbondo capital into the Kambo river valley. The period of rule by
the lineages of Jinga a Ngola and Mbande a Ngola in the kingdom
of the ngola a kiluanje (1570s-1620s?) reduced the Mbondo state to
its core lineages above the escarpment near Lambo.54
51 Current work on this area by Karl Hofer may expand or modify my
reconstruction o f early Mbondo history.
 CH APTER V

T he Problem of State Form ation among

the Segmentary Lineages to the East

T he a b se n c e of large and stable kingdoms forms the most striking

characteristic of the political history of the areas east of the Mbundu.
There, in contrast to the western regions, where large and enduring
states dotted the political history of the Mbundu, as well as that of
the Ovimbundu to the south and the Kongo to the north, many
proto-kingdoms had emerged by 1600 but none had grown to great
size and most had disintegrated within a short while after their
founding.1 Such a history of small-scale and ephemeral political
structures explains why modem Songo and Cokwe lineages hold
numerous ancient titles but accord none the degree of centralization
or geographical extent comparable to the mani Kongo in Kongo,
the ngola a kiluanje among the Mbundu, or the kulembe in the
Benguela highlands. The general failure of title-holders to unite these
lineages into viable states forms the main theme of this chapter. The
principle line of the narrative follows the two exceptions which prove
the rule, the Lunda state of the mwata yamvo in Katanga and the
band of people ruled by holders of a title called the kinguri who
began their history in Lunda but ended by assaulting the very
existence of Mbundu lineages near the sources of the Kwango.
Elsewhere, there was little net increase in political scale in the period
before 1600.

Early Political History o f the Songo

Of all the Mbundu ethno-linguistic subgroups, the Songo emerge
as unique in having had no single strong kingdom.2 By 1550 or so,
1The major apparent exception to this generalization, the so-called Lunda
empire, probably did not emerge as a strongly centralized state until the end
o f the seventeenth century; even then, effective centralization extended to a
relatively small area, by no means the entire region usually shown on maps as
the ‘Lunda empire’. Without anticipating the conclusions o f this chapter, my
analysis supports the dates for the emergence of the mature state of the
mwata yamvo suggested in Vellut (1972), pp. 65-7.
2The history of the Songo is the least known o f any of the northern Mbundu.
Beyond a few scattered notes left by nineteenth century travellers, the only
 E A R L Y P O L IT IC A L H I S T O R Y OF T H E SONGO 113
the Ndongo had had their ngola a Iciluanje, the Pende had seen
several lunga-states, the Libolo a dynasty of hango-kmgs, the Lenge
a shadowy keta-state, and the Mbondo their ndala kisua. Part of the
reason for the anomalous political history of the Songo derives from
the fact that their descent groups, alone among those of the Mbundu,
belonged to the expanding system of segmentary lineages which, by
the sixteenth century, reached from the Kwanza, where the Songo
formed its western edge, at least as far east as the northern Lunda
in Katanga. Although we can obviously only guess at the extent
to which sixteenth-century Songo still approximated the ideal type of
segmentary social structure, in which genealogical ties generated by
lineage fission coincided with the spatial distribution of the descent
groups, the inconspicuousness of political structures in the area sug­
gests that little had yet disturbed the smooth operation of a decentral­
ized system which could still efficiently allocate scarce resources
(mainly land) and mobilize sufficient numbers of men for defence.
A comprehensive segmentary genealogy articulated the relation­
ships among the jingundu of the Songo and also between the Songo
and the Cokwe/Lwena and Lunda. At its summit stood the name
Mbumba a Mbulu, a symbolic figure known widely in western Central
Africa but of uncertain historical significance. The three major
subdivisions in the genealogy appear in its first descending generation,
where Ternbo a Mbumba appears as the ancestress of the Songo and
Colcwe lineages, Tumba a Mbumba as founder of an early set of
Cokwe political titles, and Ngamba a Mbumba as the progenitor of
the earliest remembered titles in the Lunda heartland near the Kalanyi
river in Katanga.3
Names in the Songo part of the articulated genealogy refer almost
exclusively to relationships between descent groups, but even the
3 All Imbangala testimonies dealing with the aetiological period of Mbundu
history concur. Mesquitela Lima (1971), p. 43, has an Ngombe/Ngombo at
the head of the Lunda genealogy. Tembo and Tumba appear in various nine­
teenth-century Cokwe traditions, sometimes ‘falling from the sky’ as the
archetypal human beings in the Cokwe world; in general, the Cokwe appear
to claim Tembo as a lineage ancestress, while the Songo attempt to distinguish
between themselves and the Cokwe by using Tembo and Tumba. See Hermann
Baumann (1935), pp. 139-40; Fonseca Cardoso (1919), pp. 14-16; F. Grevisse
(1946-7), pp. 77-8 (‘the group [Cokwe] is formed of clans recognizing the
woman Tembo as their ancestor . . . ’). Crine-Mavar (1973), p. 72, has Tumba,
Tembo, and Samba from Lunda sources.

source is Afonso Alexandre de Magalhães (1948), p. 38. Hopefully, the recent

investigations o f Karl Höfer will fill in some of the gaps in our knowledge o f
this important area.
114 T H E PR O B L E M O F S T A T E -F O R M A T IO N
incomplete versions of the genealogies which I recorded from Songo-
descended Imbangala hinted at the presence of a set of extra-lineage
names said to be males who ‘married’ Tembo a Mbumba, the
female founding ancestress of the lineages. According to the char­
acteristic idiom of the Mbundu perpetual genealogies, Tembo a
Mbumba’s ‘husbands’ stand for the politial authorities which
emerged from time to time among various segments of the Songo
jingundu. Although each of the names (a partial list includes such
figures as Kalulu lea Wambwa, Mukoso, Mwili, and Kunga dya
Palanga4) represent some form of political rule, the limited data on
Songo history make it impossible to identify most of these incipient
kings, much less determine their exact locations or place them in the
correct chronological sequence.
It seems clear, however, that most of these titles originated outside
the region occupied by the segmentary lineages and were altered as
they were incorporated as permanent named positions connected to
lineages, much as the Mbondo converted the Libolo mavunga.
Modern Songo attribute the sources of these positions to the south­
west where the independently documented dynasties of the kulembe
and the hango-Icings provide likely backgrounds for political authori­
ties of this type. Of the recorded names, only that of Kunga dya
Palanga may be provisionally identified: his title may date back to
the sixteenth century or before when one of the major figures in
Mbailundo bore the title of ‘lcungo’.s A few of these titles still
existed in the late nineteenth century when Songo reported to
travellers that a king called Kunga dya Palanga held some form
of authority over the people who lived north of the Luhando and
east of the Kwanza rivers.6 In general, however, few of the early
Songo kings managed to overcome lineage resistance to centralized
political control. The large number of titles remembered as ‘hus­
bands’ of Tembo a Mbumba and discrepancies in their positions in
the genealogy suggest that they did not exercise real power for very
long and did not extend their influence very far.

Growth o f Centralized Institutions among the Lunda

To find a parallel imposition of alien political titles on the seg­
mentary lineages of Mbumba a Mbulu, it is necessary to wander as
4Testimonies o f Alexandre Vaz, 30 and 31 July 1969; Sousa Calunga,
29 and 30 Sept. 1969; Domingos Vaz.
5A. V. Rodrigues (1968), p. 183. Palanga is also a title found most commonly
in the Libolo/Mbailundo area.
6Lux (1880), p. 96.
 G R O W T H O F C E N T R A L IZ E D IN S T IT U T IO N S 115
far from the Mbundu as the northern Lunda of Katanga where
political developments began which later affected the Songo and
eventually touched all of the Mbundu. Early Lunda history should
be analysed against a background of strongly segmentary matri-
lineages and visualized as a gradual movement through several
stages of political development characterized by progressively more
centralized structures and leading to the emergence of the mwata
yamvo in the late seventeenth century. Along the way, some Lunda
lineages opposed these changes and eventually moved away from the
centre of the Lunda state. In the process, they diffused a number of
Lunda political titles and authority symbols which became the
central positions and emblems of most Mbundu kingdoms after the
middle of the seventeenth century.
Since the analysis of Lunda history which leads to these conclu­
sions departs from the conventional interpretation of some fairly
well-known traditions, I want to preface my substantive remarks
with a review of previously available information on the earliest
periods of Lunda political history. The published oral traditions on
the most ancient phases of Lunda state-formation come from a
wide variety of sources—Lunda, Colcwe, and Pende for the most
part—and thus afford the opportunity to make comparisons between
them which might indicate the ways in which the milieu in which
each tradition has been transmitted has affected its content
and structure.7 They have, however, generally been interpreted
without adequate attention to the social and political institutions of
the people who told them. Relatively recent (i.e. since 1650) political
developments in Lunda have, for example, clearly affected the ways
in which the nineteenth- and twentieth-century Lunda recalled very
early events which, it would seem, occurred in a setting different
in many ways from political and social conditions now.
7 Still the best record o f the northern Lunda traditions is the account of
Henriquc Augusto Dias de Carvalho (1890a), the earliest recorded and includ­
ing the greatest detail on events o f the remote period. Victor W. Turner (1955)
translated portions of Carvalho’s test into English. It is closely paralleled by
a Cokwe text (perhaps contaminated with editorial interpolations from Dias
de Carvalho) reproduced in Lima (1971), pp. 42-51(7). The uncanny similarities
between Lima’s text and that o f Dias de Carvalho may be explained, however,
by the fact that Dias Carvalho’s information came largely from Lunda chiefs
whose ancestors had long lived among Cokwe, whose descendants later retold
the same version o f this tradition to Lima. See also the accounts of Leon
Duysters (1958), M. van den Byvang (1937), and Daniel Biebuyck (1957) for
later and less complete accounts. Crine-Mavar (1973) gives a recent and rela­
tively independent version from Lunda sources close to the modern royal
court.
116 T H E P R O B L E M O F S T A T E -F O R M A T IO N
Other traditions found outside the central Lunda region, such as
those of the Imbangala (whose kings claim descent from predeces­
sors who left Lunda before the changes which later altered the
content of the Lunda narratives), may preserve an archaic perspect­
ive no longer present in the Lunda homeland. Similarly, modern
Cokwe social structures resemble ancient Lunda institutions more
closely than do present conditions nearer the Kalanyi; Cokwe
traditions may therefore be more informative on the earlier phases
of Lunda history than the accounts of the Lunda themselves. I
intend to reinterpret early Lunda history by incorporating into the
analysis new and probably older traditions found among the Imban­
gala, by offering some suggestions as to the social system in which
early Lunda political development took place, and by applying to
the well-known traditions the critical techniques developed to handle
the Mbundu narratives and genealogies discussed in preceding
chapters. This approach seems justified since the Lunda histories
clearly fit the structure of genealogies and narrative episodes found
in the Mbundu traditions. Further, since the Lunda are known to
have positional succession and perpetual kinship, it seems justifiable
to handle their traditions as if they possessed characteristics similar
to those of the Mbundu.8 In particular, I assume that all names in
the genealogies refer to perpetual titles and that the narrative
portions of the traditions are separable, each from the others,
and describe historical events in metaphorical rather than literal
terms.
The traditions give every indication that the ancient Lunda had
social institutions very much like the present segmentary matri-
lineages of the Songo and Cokwe.9 Although the social institutions
of the northern Lunda appear to have changed much more in recent
centuries than those of the Mbundu,10 the practices of positional
succession and perpetual kinship come from a very early period in
8Vellut (1972), pp. 65-6, has noticed this aspect o f the traditions but has
not worked out the implications for the period before the mwata yamvo (c.
1700).
9A surprising scarcity o f reliable ethnographic data hinders reconstruction
o f Lunda social history at so great a time depth. The most important published
materials on the Lunda appear in Fernand Crine (1963), Crine-Mavar (1963,
1968, 1973), and Biebuyck (1957). Information on related groups and the
limited accessible data in travellers’ accounts allow the drawing o f some fairly
firm conclusions.
10 Crine-Mavar has stated several times that the modern Lunda are bilateral
with no strong lineages (1963, pp. 158, 165-6, and 1973, p. 69). But it remains
unclear whether all Lunda, or only the holders o f political titles, are today
‘bilateral’, and in which contexts bilateral descent obtains.
 G R O W T H O F C E N T R A L IZ E D IN S T IT U T IO N S 117
their history.11 Portions of the Mbundu perpetual genealogies dating
from the fourteenth or fifteenth centuries, for example, extend to the
Lunda, and, since these genealogies include only titles of the perm­
anent type, positional succession and perpetual kinship must have
existed in Lunda at that time. The widespread distribution of these
practices, in a broad band reaching across Central Africa, but centred
on the Lunda area, also suggests that these institutions must have
been a part of Lunda life long ago.12
Several sorts of evidence suggest that the ancient Lunda had
lineages of a segmentary type similar to those of the modern Songo.
The Lunda titles appear in the section of the perpetual genealogies
headed by Mbumba a Mbulu. Since this genealogy contains only
names referring to matrilineages, and/or the perpetual political
titles associated with these descent groups, the Mbundu habit of
locating the Lunda there indicates that they once had matrilineages
organized, at least in part, on the basis of segmentary opposition.
Recent research among the Sala Mpasu, northern neighbours of the
Lunda in Katanga, suggests that they also once had matrilineages
even though they have now abandoned them for other sorts of
social institutions.13 Some Cokwe traditions purport to describe the
conditions of life among the ancient Lunda in terms which explicitly
mention the essential features of a segmentary lineage system: a
number of relatively independent ‘chiefs’ (i.e. lineage headmen) who
occupied distinctive small territories and who banded together
under a single ‘chief’ in case of need.14 Ancient Lunda political
history therefore occurred against a background of lineages and
lineage segments occupying a relatively small area on the Kalanyi
river, each living on its own small domain (called an mpat). The
11This interpretation disagrees specifically with that o f Duysters (1958),
p. 82, who states that the characters in the Lunda traditions were persons and
not perpetual titles. Since any Inibangala would state the same opinion if asked
directly about the distinction between names and titles in Kasanje, Duysters
may have reached his erroneous conclusions through uncritical reliance on his
informants’ remarks.
12Positional succession and perpetual kinship today are found from the
Mbundu in the west through the Cokwe and northern Lunda to the Luapula
peoples and the Bemba in the east; Cunnison (1956). Crine (1963), pp. 158,
162-3, assumes that perpetual kinship and positional succession are very
ancient. See also Biebuyclc (1957), p. 794, who confirms that ties o f this type
link the most ancient surviving Lunda positions (the tubimgu chiefs).
131 am grateful to Professor William Pruitt of Kalamazoo College for gen­
erously sharing with me some of the results o f his unpublished research on this
point.
14 Lima (1971), p. 43. Crine-Mavar’s recent work seems to concur (1973),
pp. 66-74.
118 T H E PR O B L E M O F S T A T E -F O R M A T IO N
segmentary aspect of the lineage structure would have restricted the
formation of strongly centralized political institutions, much as
similar institutions hindered the rise of ldngs among the Songo. The
Lunda ‘kingdom’ at that time, far from resembling later Central
African states or its own subsequent imperial phases, consisted of
little more than a collection of lineage villages.
A set of perpetual named positions called tubungu (singular,
kabungu) provided the underlying unity for the Lunda lineages on
the Kalanyi. In addition, at the earliest remembered time another
set of political titles seems to have been present.15 Neither the exact
relationships between these positions nor their origin is clear, but
comparison of the genealogies referring to this period (approximately
the fifteenth century or so, but not datable16) reveals the existence of
a single senior title which the Lunda ranked above several related but
subordinate positions. This set of Lunda political titles pivoted on a
figure variously called Kunda a Ngamba, Konde a Matita, or some­
times Yala Mwalcu; in any case, the position is depicted as the sire
of such other positions as the kinguri, the Icinyama, and the lueji.
Uncertainty over the correct name for the senior position probably
indicates that the title had fallen into decline by the time the tradi­
tions took their present form so that its exact identity no longer
mattered to those who preserved the traditions. As the narrative to
follow shows, such seems to have been the case. Most Imbangala
historians, who use the name of the senior position only as the
patronym (or matronym) of the kinguri title held later by kings in
Kasanje, agree that it should be Kunda.17 Other Imbangala have
Kunda as ‘Konde’ and make the position a female, the ‘mother’ of
Kinguri.18 Those who make the kinguri a descendant of a female
‘Konde’ give his paternal origin as Yala Mwaku, or Mutombo
Mukulu.19 Such references to the latter title, which was an important
political position among the Luba Kaniok or Kalundwe, who lived
15As in the case o f the Mbundu, the earliest remembered period o f Lunda
history picks up the process o f state-formation in midstream. We simply have
no idea o f the phases which may have gone before.
16Miller (1972a).
17So do the Songo; Magalhães (1948), p. 35. Max Buchner (1883), pp. 57-8,
also has the surname Kunda for Kinguri. Schiitt’s ‘Kinguri kya Bangela’
(1881, p. 60), or ‘Bangala Kinguri’ (p. 70), refers to a different title, the kinguri
kya bangela, a position that appeared in the eighteenth century and had no
known connection with the Lunda kinguri position.
18Testimonies of Mwanya a Shiba, 14 June 1969; Sousa Calunga, 16 June
1969; Apolo de Matos, 18 June 1969.
19Testimonies of Apolo de Matos, 18 June and 8 July 1969; also Lima
(1971), p. 43.
 G R O W T H O F C E N T R A L IZ E D IN S T IT U T IO N S 119
north-east of Lunda, show the senior Lunda position’s connection
with the very early states known to have existed among the Luba.
Lunda sources, doing little to clarify the situation, variously por­
tray Konde as the wife of Yala Mwalcu, as a child (or grandchild)
of Yala Mwalcu, and as his sister. The official Lunda court historian
in the 1920s gave Mwalcu as the senior tubimgu chief who had a son
Yala Mwaku who in turn sired Sakalende (male) and Konde (male).
Konde then fathered ICinguri, Kinyama, and Lueji.20 Nineteenth-
century Lunda informants agreed with the Imbangala in identifying
Yala Mwaku as the father and Konde as the mother of ICinguri,
Lueji, and ICinyama.21 The Pende portrayed Yala Mwaku and
Konde Matita (female) as brother and sister and added an unidenti­
fied ‘ICavula’ as the husband of Konde Matita and father of Kinguri.22
In general, the Lunda versions giving Konde as a male seem to
reflect the modern Lunda preference for patrilineal descent among
political titles. The Pende tradition, on the other hand, ignores the
Lunda tendency to trace political legitimacy through the male line
and evidently attempts to depict Kinguri and Lueji as nephews and
hence as legitimate heirs of Yala Mwaku according to their own
matrilineal rules of descent. Placing Konde Matita as the female
link between Yala Mwaku and Kinguri accomplishes this purpose.23
If a conclusion may be drawn from such disparate data, the form yala
mwaku probably represents the Luba (male) political title and the
name ICunda (or Konde) the Lunda lineage in possession of the
position. A ‘marriage’ between them, of the sort which appears in
several of the traditions, would represent Lunda acceptance of a very
early (and otherwise unidentified) form of Luba political authority.
Other details of the traditions reinforce the impression that the
yala mwaku originated in a set of Luba political institutions which
overspread the Lunda and Cokwe some time before the period in
which the standard Lunda traditions come into clear focus. Beyond

20Duysters (1958), pp. 76, 79, 81.

21 Dias de Carvalho (1890a), p. 60.
22 Haveaux (1954), p. 21. Chinyanta Nankula (1961), p. 1, agreed with Pende
informants on the place o f Matita in the genealogy. Cf. Biebuyclc (1957), pp.
801-3, who included the name o f a Matita in his highly schematic genealogy.
23 Another genealogy (Biebuyck (1957), pp. 801-3) shows an extreme example
o f the way in which the same names shift positions relative to one another
to conform to the intent o f the person who recites it; in this case four consecu­
tive pairings o f brothers and sisters in ‘marriages’ involving Yala Mwaku,
Konde, Matita, etc., are used to dismiss this part o f Lunda history as a mythical
period before time and lineages became part o f the world. Lima’s Cokwe
tradition exhibits the same feature (1971, p. 43).
120 T H E PR O B L E M O F S T A T E -F O R M A T IO N
Yala Mwaku’s vague association with Mutombo Mukulu, other
genealogies show a marriage between Walunda wa Nyama, a female
name representing a Cokwe/Lwena linege grouping descended from
Tumba a Mbumba, and a male, Manganda a Kambamba a Musopo
wa Nyama. The male (presumably political) title was also of Luba
origin. A ‘child’ of this marriage, Kata ka Walunda or Kata ka
Manganda, in turn married ICunda to produce Kinguri and Lueji.
This marriage thus performs the same function of tracing the main
Lunda titles to a Luba (political)-Lunda (lineage) union as the
marriage of Yala Mwaku in other genealogies and confirms the
Luba antecedents of the Lunda political titles. This interpretation
would reconcile all known versions of the genealogy, both Imbangala
and Lunda, by hypothesizing a Luba origin for the earliest
remembered Lunda senior title.24
Although the traditions show that the yala mwaku enjoyed a
status formally senior to the positions held by other Lunda lineages,
they reveal little about the historical relationships between these
titles. We may surmise that the holders of the yala mwaku, in any
case, exercised little more than ritual seniority over the other Lunda
lineages since it was remembered later as a ‘first among equals’.
Real political power depended on the lineages’ manpower and
economic resources, and there is no reason to assume that these
corresponded at all times to the formal kinship hierarchy of the
genealogy.25 Since the genealogy could not alter to reflect changes in
the historical political balance, it sometimes must have lost its useful­
ness as a means of structuring relationships between the lineages.
In such instances, it became desirable either to adopt a new political
system more indicative of political realities or to abolish the senior
title which held all other positions in false subordination to it.
Because the traditions have preserved nothing about the historical
realities behind the formal genealogies, it is possible only to speculate
on what circumstances might have provoked such a restructuring of
the relationships among the Lunda political titles. None the less, a
24 Colcwe traditions also contain hints o f an early spread o f Luba titles
among the matrilineages descended from Mbumba a Mbulu; some nineteenth-
century Colcwe attributed the origins o f these names directly to ICasongo
Nyembo, the most powerful Luba state o f that period; Grevisse (1946-7),
p. 58. Haveaux (1954), pp. 28-9, notes that the Pende o f the Kasai agree that
Luba titles played a role in the early history o f the Cokwe.
25Power and seniority rarely coincided in such later kingdoms as ICasanje
or the Bemba where similarly stable systems o f titles prevented alteration of
the genealogies to reflect changing historical circumstances. See the fully
documented argument to this effect in Andrew Roberts (1974).
 Mbumba a Mbulu

G R O W T H O F C E N T R A L IZ E D IN S T IT U T IO N S
121

F ig . v Genealogy Illustrating the Alleged Origins of Ancient Lunda

Political Titles.
122 T H E PR O B L E M OF S T A T E -F O R M A T IO N
realignment did occur. The names of many titles which participated
in this revolution have probably disappeared owing to the effects
of still later changes in Lunda, but those of three positions subordinate
to the yala mwaku have survived: the kinguri,25 the lueji, and the
kinvama.27 Their later significance as central positions among the
Cokwe, in Lunda, and in Kasanje has caused them to occur in nearly
all variants of the list of early Lunda title-holders. Some versions of
the genealogy include several other positions classed as siblings of the
kinguri and subordinates of the yala mwaku. An occasional Imbang-
ala history cites vaguely a Maweje who, it is said, left Lunda even
before the kinguri. The Imbangala know nothing of Maweje’s sub­
sequent history except that he allegedly opened the trail which
kinguri and other Lunda titles later followed towards the west.28 The
Imbangala also remember a ‘brother’ of Kinguri named Munjumbo
who also left Lunda for the west.29 Others in addition to the standard
trio of Kinguri, Lueji (or Naweje, as the Imbangala call her), and
Kinyama include a Kayungo or Kalungo, Kasongo, and Iyala.30
The remainder of the list varies because historians no longer needed
to mention all these positions to account for the states familiar to
them in the late nineteenth century and after, when all known tradi­
tions were recorded. All modern lists of these positions probably
include only a fraction of the original titles.20*893

20 The meaning o f this title is not clear.

27 The word nyama, meaning meat or wild game in Kimbundu and in most
Bantu languages, occurs with some frequency as a title on the maternal or
lineage side o f the early Lunda (and Luba?) genealogies, (e.g. Kinyama,
Walunda wa Nyama, etc.). Rodrigues Neves (1854), p. 96, gave ‘Nyama’ as the
name of the ‘country’ where Kinguri had lived in Lunda, mistaking a lineage
designation for a ‘country’ since the Mbundu traditions do not distinguish
clearly between lineages and lands which often have the same name.
28 Testimonies o f Sousa Calunga, 16 June, 9 July 1969. The informant may
have confused the name of a ‘high god’ found east o f the Lunda (Mawese or
Maweje) with one o f the tubungu titles. For Maweje, see Hermann Baumann
(1936), p. 39. The name heads Lunda genealogies recorded among the central
and eastern Lunda kingdoms; Chinyanta Nanlcula (1961), p. 1; also Biebuyclc
(1957), pp. 801-3. In this case, it may represent a garbled recollection o f early
Luba influences that spread southwest from Lunda towards the Cokwe.
29 This title has a very complex history, but its origins clearly lie in Lunda;
see testimonies of Sousa Calunga, 29 Sept., 1 Oct. 1969; Domingos Vaz. The
Cokwe also recall a title o f this name that moved west from Lunda with the
kinguri', Lima (1971), p. 46.
30 Testimony of Apolo de Matos, 18 June 1969. Iyala corresponds to the
Yala mentioned by Lunda in the late nineteenth century; Dias de Carvalho
(1890a), p. 60. The two longest recorded lists o f siblings (both Lunda) include
Ndonje who turns up as a Lunda title in later Imbangala traditions; Biebuyclc
(1957), pp. 801-3; Crine-Mavar (1973), p. 71, agrees and adds others.
 G R O W T H O F C E N T R A L IZ E D IN S T IT U T IO N S 123
A set of Lunda narrative episodes reveals a shift from the non-
hierarchical political system headed by the yala mwaku to a more
centralized state headed by incumbents in the lueji position. Accord­
ing to the traditions describing this rupture in the perpetual gen­
ealogy,31 Kinguri and ICinyama, the sons of Yala Mwaku, the Lunda
king, returned home one evening after an afternoon spent drinking
palm wine. They found their father busy weaving a sleeping mat in
the courtyard of his compound. A pot of water sat beside him in
which he was soaking raffia fibres to make them pliable and suitable
for weaving. Kinguri and Kinyama, in part because of their drunken­
ness, mistook the cloudy water for palm wine and demanded that
the old man give them some to drink. When Yala Mwaku denied that
the pot contained palm wine, they became angry and beat him very
severely. Lueji, their sister, arrived just in time to attend to her
father’s last wishes before he died. Yala Mwaku willed his title to
Lueji as a reward for her faithfulness and punished Kinguri and
Kinyama for their disobedience by ignoring their claims to his
position.
Although the story as recorded in the nineteenth century continued
from tins point directly into other events, this much of the tradition
originally formed a distinct narrative episode dealing with a single
series of historical events. Interpreting the characters in the episode
as perpetual titles would mean that incumbent kinguri and kinyama,
or their lineages, became involved in hostilities against the yala
mwaku, or his lineage, and tried at some point to usurp the authority
of the senior position. The palm wine, over which Kinguri, Kinyama,
and Yala Mwaku fought, stands as a metaphor for the real, but
unstated, historical issues which prompted the junior title-holders to
eliminate their senior position. The tradition characteristically
reveals nothing about the nature of the historical dispute except to
indicate, through the symbolism of the palm wine, which the Lunda
associate with males and with political power,32 that it concerned
political authority. The names of Kinguri’s companion(s) in this
enterprise, which vary in different versions of the narrative, stand for

31 Dias de Carvalho’s account is the only published version o f the tradition

that includes more than a very sketchy presentation o f the narrative episodes;
the paraphrase given here comes from (1890a), pp. 59-75. Dias de Carvalho
wrote his version of the story in a style that obscured the original structure of
the Lunda traditions in terms o f genealogies and narrative episodes. The
following analysis attempts to reintroduce these divisions and to draw out the
historical significance o f the original structure.
32 Personal communication from Jan Vansina.
124 T H E PR O B L E M O F S T A T E -F O R M A T IO N
lineages which allied themselves to the kinguri in his wars against the
yala mwaku. The most common variants of the tradition identify the
kinguri’s main allies as the lineage which controlled the kinyama;
other versions claim that a title known as the yala also supported the
kinguri’s assault.33 The variability simply confirms that the traditions
have probably preserved an incomplete list of the groups actually
involved.
Yala Mwaku’s death moved the old Lunda political structures
from a period of relative stability into a transitional phase of inter­
necine warfare. It may be assumed that the kinguri and his allies
eventually eliminated the yala mwaku position from the Lunda
perpetual genealogy after fighting of undetermined intensity and
duration (although the scale of the conflict could not have been
large by later standards). The ‘death’ of the yala mwaku title in the
language of the narrative episodes does not refer to the assassination
of an individual incumbent, although this may have occurred one or
more times. The ‘death’ of a position in a genealogy consisting of
permanent named titles signifies that its enemies destroyed its
authority emblems and ritually eliminated it from the political
structure to which it had belonged. Imbangala traditions mention
several later cases in which opponents used such methods to abolish
titles among the Mbundu, and traditions of some of the eastern
Central African groups with positional succession and perpetual
kinship show the same thing.34
The ‘death’ of the yala mwaku freed the other Lunda lineages to
establish a new political balance unfettered by the hierarchy of
perpetual kinship which formerly had bound them in collective sub­
ordination. The abolition of a less senior title would have caused
relatively minor shifts in the status of a few Lunda lineages. Elimina­
tion of the yala mwaku meant something quite different, however,
since the other titles depended on it for legitimacy and supernatural
protection. A period of instability, hidden by the placid style of the
narrative episode, must have followed as each Lunda lineage com­
peted with the others for supremacy in the changed political environ­
ment.35 If the yala mwaku had had any direct connection with the
Luba, as some traditions indicate it did, its overthrow may also have
33Testimony o f Apolo de Matos.
34M. G. Marwick (1963), p. 389.
35Imbangala traditions obscure all interregna and periods of political
instability. The evidence that such a phase occurred here comes from the logical
discontinuity in the line of historical development (a challenge initiated by
holders o f one title, the kinguri, and a victory by those o f another, the lueji).
 G R O W T H OF C E N T R A L IZ E D IN S T IT U T IO N S 125
represented a rebellion by the Lunda lineages against the presence
of Luba political titles.
From the traditions, we must assume that lineages led by the
kinguri struggled for pre-eminence during the period of instability
and that the lineage of the lueji led the opposition to them. Since, as
the narrative stated, Lueji became heiress to Yala Mwaku’s royal
power, the lueji position must finally have built up an alliance strong
enough to defeat the kinguri and kinyama and thus to claim the
senior status formerly accorded the yala mwaku. Kinguri and Kin-
yama’s mistaken identification of cloudy water as palm wine in the
narrative episode metaphorically underlines the irony of their defeat;
they had pursued a mirage, eventually losing the battle to the lueji
despite their original challenge to the yala mwaku. The traditions do
not, of course, indicate how long these struggles continued. Since the
figures in the narrative represent permanent political positions, many
years could have elapsed between the assault on the yala mwaku
by one kinguri incumbent and later defeats administered to successors
in the same title.
The kinguri’s assault on the yala mwaku had initiated funda­
mental changes in Lunda political structures which culminated in a
new and more centralized state dominated by the lueji position. A sep­
arate narrative episode briefly outlines the formation and structure
of this successor state. Lueji was still a child when Yala Mwaku died,
in the words of the tradition, and could rule only with the counsel of
the older and wiser tubungu lineage chiefs. They granted her a form
of ritual supremacy and allowed her technical ownership of the
Lunda lands, but tubungu guardians took control of the bracelet
which embodied supreme Lunda political authority, the lukano.
The tubungu guardians moved their residences from their own
teritories to the lueji’s home domain at that time, thus creating the
first pan-Lunda capital town. The tubungu excluded the lineages of
the kinguri and the kinyama (or the yala) from the new confederation,
presumably because of their unsuccessful attempt to wrest control
of the lukano from the lueji.36 The kingdom probably retained this
form for some decades, at least, since the traditions stated that
various tubungu chiefs took turns filling the offices at the court
during this period.37 Thus, although the lueji had defeated the

36 Dias de Carvalho (1890a), p. 64.

37 Haveaux (1954), p. 4; also Dias de Carvalho (1890a), p. 64. Compare this
vague reference to the interim rule o f Lunda makota in the Imbangala kilombo
(c, 1560-1610); the period o f interim rule, which received equally summary
126 T H E PR O B L E M OF S T A T E -F O R M A T IO N
kinguri and the kinyama, the victorious title-holders had won at the
cost of their own independence. The tubungu lineage chiefs gained a
voice in the affairs of the new kingdom as the price of their assistance
during the wars. The threat of the kinguri had forced the Lunda to
develop a new political structure with a higher degree of centraliza­
tion than its predecessor(s) but one in which holders of the central
political title had to submit to a council of advisors drawn from the
Lunda lineages.
Several details mentioned in the first narrative episodes, which
stand out as possibly historical because they contrast with conditions
when the tradition was recorded, point to the conclusion that
Yala Mwalcu ‘died’ much earlier than the late-sixteenth-century
date generally assumed for the beginnings of Lunda political
history.38 Hints at the level of technology, for example, repeatedly
indicate that these developments took place before the Lunda
acquired iron of a quality high enough to manufacture broad-
bladed weapons. Lunda culture emphasized fishing rather than
hunting at the time, and Lunda hunters captured game only by
means of wooden traps rather than with arrows or spears tipped with
iron.39 The title of the yala mwaku was said to have meant ‘thrower
of rocks’,40 a praise name perhaps suggestive of a period before
the introduction of iron weaponry. Kinguri and his brother slew their
father with a wooden club rather than with the pokwe iron knife
which later became the characteristic Lunda weapon.41
The next episode in the Lunda narrative picks up the theme of
renewed Luba influence on the lueji’s Lunda state. The tradition
relates that Lueji went one day to the river ICalanyi where she found
a group of hunters camped in the forest. They had travelled there
under the leadership of a Luba noble, later known to the Lunda and
Cokwe as Cibinda Ilunga. Noticing that Cibinda Ilunga and his
companions lacked salt for the meat from animals they had slain,
Lueji oifered to provide whatever salt they might need. After a
38 For the debate on this date, see Jan Vansina (1963a), David Birmingham
(1965), Jan Vansina (1966b), and my summary in (1972a), pp. 549-51. Vellut
(1972), pp. 65-9, contains new information.
33Dias de Carvalho (1890a), pp. 60, 61; also Lima (1971), p. 44.
401 have not been able to verify Dias de Carvalho’s translation o f the name.
The significance given to it by the Lunda is more important than a literal
translation o f the words in any case.
41 D ias de Carvalho (1890a), p. 67.
treatment in the traditions, lasted at least fifty years; see above, p. 14, n. 46.
The Cokwe traditon quoted by Lima (1971), p. 44, explicitly notes that ‘some
time passed’ during this period.
 G R O W T H O F C E N T R A L IZ E D IN ST IT U T IO N S 127
lengthy conversation on the banks of the river, she invited Cibinda
Ilunga to remain in her realm. They eventually fell in love and
decided to marry, each with the permission of the councillors of
their respective kingdoms. After the marriage, Lueji gave her
liilcano, the emblem of Lunda royal authority, to Cibinda Ilunga
so that he might rule Lunda in her place.
The ‘marriage’ of Cibinda Ilunga and Lueji began another new
stage in the development of centralized political institutions in
Lunda. It united the political authority of the Luba (represented as
male) with the lineages in possession of the lueji title (represented as
female) according to a pattern already familiar from the Mbundu
traditions. Transfer of the lukano explicitly confirmed that the Lunda
adopted some new form of Luba political authority. The meat
(male) and salt (female) which Lueji and Cibinda Ilunga exchanged
reiterate the complementarity between the hunters and the Lunda
and emphasize the naturalness which nineteenth-century Lunda
attributed to the partnership of Luba-ized rulers and Lunda lineages.42
New Luba ideas and institutions spread among the Lunda at the
same time. More advanced iron-working techniques may have
arrived in company with the previously unknown kibinda hunting
society.43 Cibinda Ilunga clearly stands for the kibinda society of
professional hunters which the Lunda had lacked up to that time.44
According to the Colcwe, Cibinda Ilunga’s principal authority
emblem was a magical dagger, the leapokolo; he also came with
special bows, yitumbo45 (singular, kitumbo), and the cimbwiya
hatchet which still survives as an important symbol of political
power among the Colcwe.46 The fact that most of these innovations,
which were also symbols of the kibinda hunting association, involved
42 On the symbolism of hunters as founding kings, see Boston (1964) and
Lucas (1971).
43Edouard N ’Dua (1971, p. 39), interview (23 January 1971) with Muhunga
Ambroise explicitly confirms that Cibinda Ilunga was also a smithy.
44The southern Lunda (Ndembu o f north-western Zambia) claim that the
kibinda society (there called the wubinda bow-hunter’s cult) came from the
mwata yamvo heirs of Cibinda Ilunga in Lunda; Turner (1967), p. 280. The
experience of the Ndembu, who subsequently adopted another (wunyanga)
hunting cult, suggests that there may have been several of these societies at
different periods of the central African past, and so the introduction of this
version of the cult to the Lunda need not preclude the possibility that a similar
institution was present among the Mbundu from as early or earlier times.
Cf. pp. 50-1 above.
45 Yitumbo are a category of charms or medicines made from vegetable
substances found in the wilds frequented by kibinda master hinters; Lima
(1971), pp. 79, 303. Cf. Boston (1964), p. 124.
4GLima (1971), p. 45.
128 T H E PR O B L E M O F S T A T E -F O R M A T IO N
sophisticated iron-working techniques suggests, like the jingola of
the Samba among the Mbundu, that a technological change occurred.
Improvements in iron-working need not imply, however, that a
massive Luba military conquest took place, since the Lunda had
good reasons for voluntarily adopting Luba ideas and institutions.
Strained relations between the Lunda lineages explain why the
lueji’s people took up these new principles of political organization.
The lueji had ruled at the head of a loose confederation which had
originated as a defensive alliance against the Icinguri and the kinyama.
These two had evidently remained not far from the fringes of the
Lunda state where they constituted a continuing threat to the line­
ages which had backed the lueji.*7 Unable to destroy the Icinguri by
themselves, they must have sought new and more powerful charms,
organizational techniques, and weapons to drive their enemy away.
The figure of Cibinda Uunga symbolizes the arrival of all of these.
The danger of the Icinguri in the end forced the lueji’s Lunda con­
federacy to offer her lulcano to the holder of the main Luba title in
order to consolidate under new leadership the unity which protected
them against renewed civil warfare of the type which had occurred
during the preceding period of instability. Lunda acceptance of the
innovations represented by Cibinda Ilunga, with or without direct
military intervention by Luba armies, thus marks a major shift in
Lunda attitudes; they began to abandon the fierce independence
which had formerly typified relations among the lineages and to
move towards the more centralized social and political institutions of
the later Lunda empire.

The Spread o f Lunda Political Titles to the West

The consequences of this fundamental change in Lunda social and
political life affected most of the rest of western Central Africa.
Acceptance of Luba centralized political institutions provoked an
emigration of Lunda lineages and a diffusion of Lunda titles which,
in its earliest phase, extended west to the Cokwe and south to the
Lwena and later spread to the Mbundu and Ovimbundu. Related
movements still later reached the Ndembu along the headwaters of
the Zambezi and the people living east as far as the Lualaba.48
With the title-holders who moved west, principally the Icinguri, went
the Icibinda hunting society and several new war charms which Cibinda
47 For an explicit Cokwe statement to this effect, Lima (1971), p. 46.
48For a summary, Yansina (1966a), pp. 84-97. The later phases o f this
expansion are not discussed here as they had no immediate effect on the
Mbundu.
 T H E SPR E A D O F L U N D A PO L IT IC A L TITLES 129
Ilunga had brought to Lunda from the Luba. Given the propensity
of the historical traditions to personalize abstract historical processes,
it is probable that the movement should be viewed in part as a
diffusion of titles and only in part as the migration of individuals
which has up to now been assumed.
The historical reasons, as distinct from the non-historical plot of
the metaphorical narratives, which provoked the first expansion of
Lunda political titles seem connected with the importation of new
authority symbols from the Luba. In effect, the new charms and titles
replaced the formerly dominant set of positions which had included
the lueji, the kinguri, and the kinyama. As the old titles lost their pre­
eminence, some remained in the hands of Lunda descent groups but
acquired new names in the same way as the arrival of new techniques
of political organization had altered the titles of the old Pende
lunga-kings and some of the /zazzgo-positions among the Mbondo.
Some of the other Lunda titles, having lost their value to the nuclear
Lunda lineages, were sold or otherwise disposed of to the segmentary
descent groups living in the west where no similarly prestigious type
of political authority yet existed. In only one case—that of the kinguri
—is there clear evidence of a movement of people, and this instance
produced major political revolutions which affected all the lineages
living along a path leading from Lunda through the Colewe to the
Songo.
The fate of the lueji title in Lunda illustrates the process of name­
changing and subordination among the older Lunda titles which
remained among the lineages on the Kalanyi. The Lunda state
changed from the comparatively loose federation of iubungu chiefs
led by the lueji to a much more centralized state under the command
of a new Luba title, the mwata yamvo. The Lunda genealogies show
the mwata yamvo position as a descendant of a ‘marriage’ between a
female named Luhasa Kamonga and Cibinda Ilunga.49 According

15Luhasa Kamonga is usually described as one of Lueji’s ‘ladies in waiting’.

I cannot interpret in Lunda terms the meaning of this obvious mistranslation.
Descriptions o f the mwata yamvo as a ‘son’ o f Lueji herself come only from
non-Lunda sources and are probably not reliable. The distinction between
Lunda traditions and those from non-Lunda, mainly Cokwe, better explains
the differing descriptions o f the origin o f the mwata yamvo title than does the
largely circumstantial distinction between nineteenth- and twentieth-century
traditions pointed out by Vellut (1972), p. 66. It happened that most nineteenth-
century traditions recorded by Europeans came from Colewe sources. Abund­
ant comparative data establish the suspect nature of father-son genealogical
ties, especially when alien sources attempt to trace internal political changes in
other polities; David P. Henige (1971; 1974, pp. 71-94).
130 T H E PR O B L E M O F S T A T E -F O R M A T IO N
to the rules of the perpetual genealogies, this made the mwata yamvo
originally a Luba position subordinate to one in the possession of a
lineage known as Luhasa Kamonga. For unknown historical reasons,
this title became the most powerful title among the central Lunda
descent groups, and the tubungu were reduced to their present status
as advisers to the central royal title. The position of the lueji retained
a status distinct from the tubungu but only as a secondary title of a
new Luba position, the swana mulunda; its new name came with its
incorporation into the state of the mwata yamvo and reflected its
reduced responsibilities in comparison with its earlier primacy. The
yala mwaJcu position was also resurrected, again with a new name,
the shakala. The length of time which elapsed while these changes
took place cannot be estimated, although two centuries or more
might not fall far short of the true mark.50
The circumstances which led the holders of the kinguri title to
depart from Lunda may be seen in Imbangala narrative episodes
which suggest that the lineages behind the title found themselves
overwhelmed by Luba magic and remained only until they had
borrowed at least some of Cibinda Ilunga’s supernatural powers.51
Naweje, as the Imbangala call Lueji, had taken control of the Lunda
kingdom but ruled only as a regent in place of Kinguri who was still
a minor and unprepared to assume the royal powers which right­
fully belonged to him.52 One day, as Kinguri and Naweje were
walking along the river Lukongolo53 in Lunda, Kinguri momentarily
left his sister alone while he went into the woods in search of some
men who were making maluvo somewhere nearby.54 Not long after
he left her, a hunter called Lulcokesha appeared and spoke to

so The established time lapse comes from the calculations presented in

Miller (1972a) and Vellut (1972), p. 69. The point was also made by Dias de
Carvalho’s Lunda informants (1890a), pp. 76-7. Some nineteenth-century
Lunda informants told Dias de Carvalho specifically that the Luba had arrived
long after the first fights between the lueji and the kinguri. Dias de Carvalho
gave this information in a letter (1886, p. 135) that seems to contain a better
approximation to the original oral form o f the tradition than the reworked
version in (1890a).
51The following account is based on the testimonies o f Sousa Calunga,
21 July, 2 October 1969.
52The self-serving intention o f this Imbangala emphasis needs no special
comment.
53River not identified, but possibly an oblique reference to the Luba back­
ground o f Lunda history, since the names consists o f the common hi- prefix
for river names and the title o f the founder o f the first Luba empire (see
Vansina (1966a), p. 71).
54Fermented palm wine, tombe in Lunda.
 T H E SPR E A D O F L U N D A P O L IT IC A L TITLES 131
Naweje, offering her as a gift the tail of an elephant he had killed.55
Naweje accepted it and in return gave the hunter some food, which
he gladly accepted since he and his men had spent the previous night
in the bush with nothing to eat.
Kinguri, off in the wilderness looking for the men making mahtvo,
suddenly felt his heart begin to pound. Recognizing this as a sign
that something had gone amiss at home, he returned at once and
found Naweje in her house eating with Lulcolcesha. The requirements
of Naweje’s position prohibited men from entering her compound,
and her makota (guardians) usually kept her isolated from all males
other than Kinguri. Since Lulcokesha’s presence in the house
violated this law, Kinguri became suspicious of the stranger
and inquired about his identity. Naweje explained what had
happened and showed Kinguri the elephant tail. Kinguri immediately
recognized another transgression of Lunda custom, since only
properly installed political chiefs could possess elephant tails
which they received as tribute from their subjects. Naweje, as
regent, had no right to accept gifts which should have gone to
Kinguri.
Kinguri threatened Lulcolcesha and ordered him to leave at once.
When the hunter refused to go, Kinguri attacked him with a magical
knife (mwela) which he had inherited from his father.56 The hunter’s
head spouted fire as Kinguri appoached, and so he turned and fled
in fear. He later returned in another attempt to slay the usurper, but
this time the hunter’s mouth turned into the fangs and jaws of a
dangerous jungle cat.57 Kinguri then realized that his enemy possessed
supernatural forces much more powerful than his own. He at first
resolved to steal the charms which made Lulcolcesha so strong but
found that he could not and agreed to leave Lunda if the hunter
would first teach him his magical secrets.
Lulcolcesha agreed to the bargain and explained his charms to
Kinguri. These included something called nzungu made from the
mbamba tree58 which grows in the gallery forests along the rivers.
The nzungu not only enabled Lulcolcesha to perform the feats which

55 The bristles of an elephant’s tail were potent charms.

56 This is an anachronistic detail of a sort characteristic o f the Imbangala
traditions; the Lunda did not have the mwela at that time.
57 Either a lion or a leopard; informant was not clear on this point, probably
because the detail does not alter the meaning o f the metaphor.
58 An unidentified specific charm. The mbamba tree (imperata cylindrica
Var Thumbergii) is valued for its usefulness in preparing the yitumbo medicines
of the kibinda hunter; Lima (1971), p. 303.
132 T H E PRO BL E M O F S T A T E -F O R M A T IO N
he had used to frighten Kinguri but also parted the waters of rivers
and divined the presence of snakes and then killed them. Lukolcesha
also gave Kinguri a magical bow which allowed its owner to slay
even the most dangerous animals of the forest. Armed with these
weapons, Kinguri departed from Lunda and began his trip to the
west.
This Imbangala tradition may be reconciled with the Lunda and
Colcwe narratives of Kinguri’s departure from Lunda if we recognize
that Lukolcesha plays the role named Cibinda Ilunga in the other
histories. Both are ‘parents’ of the mwata yamvo and represent the
Luba institutions adopted by the lueji. The Imbangala use the name
Lukolcesha instead of Cibinda Ilunga (of whom they have never
heard) since their narrative episodes come from a period some time
after the original Luba titles had begun to lose their significance
relative to the mwata yamvo. Just as the old lueji position had become
the swana mulunda as Lunda political structures evolved, the Cibinda
Ilunga had lost its original name and had become the lukonkesha.
The lukonkesha came to represent the ‘mother’ of the mwata yamvo
title, just as the swana mulunda stood as symbolic ‘mother’ of the
Lunda people.55*59 The two ‘mothers’, swana mulunda and lukonkesha,
replaced Lueji and Cibinda Ilunga as metaphoric embodiments of
the fundamental paired principles of the later Lunda state, respect­
ively the Lunda lineages and Luba political authorities.60 The lukon­
kesha, although originally male, became feminine to contrast with
the masculine position of the mwata yamvo.61
Kinguri’s fondness for palm wine recurs in this version of the
tradition, indicating the essentially political significance of the
episode. Lulcokesha first appeared when Kinguri left Naweje to
search for maluvo in the woods. Since the Lunda and Imbangala
both associate maluvo with males and hence with political authority,
this part of the story apparently refers to the period when the kinguri
abandoned the federation of tubungu chiefs headed by the lueji,
represented here by Naweje, and sought independent political charms

55 Biebuyclc (1957), pp. 791, 802. The spelling lukonkesha is Lunda (and
used here to refer to the Lunda title) while Lukolcesha approximates to
Imbangala pronunciation (and is used when the Imbangala metaphor is
intended).
60 This hypothesis explains Schiitt’s confusion o f Lulcokesha with Lueji
(1881), pp. 82-3; also van den Byvang (1937), p. 43. The confusion reappears
in Vellut (1972), p. 66.
61 The Imbangala have a similarly male position (the ndala kandumbu)
that they regard as ‘mother’ o f their major political title, the kinguri in Kasanje.
 T H E SPR E A D O F L U N D A P O L IT IC A L TITLES 133
elsewhere. This detail from the kinguri’s side of the story confirms
the Lunda descriptions of events at the court of the hieji during the
period of hostility between the kinguri and the central Lunda federa­
tion. Naweje’s relationship with Lulcokesha resulted directly from
Kinguri’s search for outside aid (his addiction to palm wine)
and confirms suggestions in the Lunda narratives that the lueji
allied with the Luba in direct response to some threat posed by the
kinguri.
Lukokesha’s present of the elephant tail stands for the lueji's
adoption of Luba political institutions in opposition to the kinguri
within the framework of the existing Lunda state system, since the
Lunda believe that the hairs of the elephant’s tail possess potent
magical forces. Naweje’s acceptance of the elephant tail and her
reciprocal gift of food to Lulcokesha reiterate the union of Luba
political power and the Lunda lineages. The Imbangala traditions
use the image of a gift of food in place of the corresponding image
(salt and meat) in the Lunda traditions since this was a common
metaphor in many other contexts in Kasanje traditions.
The symbolic clash between Lulcokesha and Kinguri describes the
supernatural weapons with which both sides fought during the
historical political manoeuvring and armed clashes which must have
punctuated the conflict before the kinguri finally gave up and left
Lunda. The story clearly implies that the kinguri departed only after
the superiority of Luba institutions had become too obvious to
ignore. Lukokesha’s magic corresponds to suggestions in Lunda
and Cokwe traditions that Cibinda Ilunga introduced new weapons,
charms, and organizational techniques superior to the rudimentary
equipment of the segmentary Lunda lineages. Kinguri’s unsuccessful
attacks on Lulcokesha dramatize the inadequacy of the Lunda chiefs’
powers in comparison to those of the Luba. But the kinguri, according
to the Imbangala, did not leave before learning some of Lukokesha’s
secrets. Since it would have taken some time for innovations adopted
by the central Lunda to spread to their enemies on the fringes of the
kingdom, Kinguri’s knowledge of Lukokesha’s charms confirms
other indications that some time passed between the introduction
of Luba techniques and the departure of the kinguri.
The secret charms which Kinguri learned from Lulcokesha came
from the inventory of supernatural skills which the Imbangala
attribute to the kibinda master hunter. Kinguri’s possession of these
charms explains to them how he could have left Lunda to wander
through the wilderness which lay to the west. Professional hunters,
134 T H E PR O B L E M O F S T A T E -F O R M A T IO N
they believed, needed special magical powers which helped them
pursue large beasts through unfamiliar territory. Unskilled hunters,
who did not enjoy the protection of these special charms, would not
undertake such dangerous ventures. In particular, their fear that
snakes embodied potentially hostile supernatural beings62 meant that
hunters, who might spend days and weeks walking through thick
undergrowth, sought magical protection against them. The Luko-
kesha’s nzungu performed this function for the kinguri, whose Lunda
title included no charms of similar potency.63 The nzungu made it
possible for the kinguri to undertake his trip through the snake-
infested wilderness.
Rivers were also believed to present serious difficulties for anyone
who travelled in unfamiliar regions, and the nzungu's ability to dry
their waters had obvious advantages for the kinguri during his
journey.64 Other magical techniques belonging to the Icibinda
professional hunter guarded against dangers which the kinguri
might encounter in forests where spiritual beings hidden in the bodies
of wild animals lurked in wait of unwary travellers. The Imbangala
made a clear distinction between the natural creatures normally
encountered there and the supernatural beings in the shape of beasts
which sorcerers sent to harm their victims. Normal animals yielded
to straightforward pursuit with fire and nets and could be killed with
spear, club, or arrows. Supernatural beasts, however, succumbed
only to the elaborate precautions and special charms of the profes­
sional hunters. Since both kinds of animal looked exactly alike, the
kibinda could rarely determine beforehand which sort he might
confront, and so they always went into the woods prepared to deal
with either type. Travellers also had to be wary of supernatural
animals, since spirits tended to follow human pathways through the
wilderness. Travellers carried various charms for protection while
yibinda generally used special bows capable of slaying both natural
and supernatural enemies. Lukokesha presented such a magical bow
62 One example is the kindalandala snake o f the Mbundu mentioned in
connection with the early history of the Mbondo state, p. 96 above. The
Imbangala, for example, clear the grass from a wide area around their houses
to keep snakes (and spirits) away from their dwellings.
63 The Imbangala all agree that Kinguri was not primarily a ‘hunter’ (mean­
ing kibinda) like Lukokesha.
64 Compare the story o f Kajinga’s crossing o f the Kwanza and Muta a
Kalombo’s ability to flood streams to block Kajinga’s pursuers. Although the
cliche o f the parting of the waters is found nearly everywhere in the world, the
breadth of its distribution does not detract from its specific function in Mbundu
symbolism.
 T H E SPR E A D O F L U N D A P O L IT IC A L TITLES 135
to Kinguri, one which protected its owner both from animals and
from hostile human beings.55 Although the Imbangala acknowledge
that modern people have lost the technique of fashioning these bows,
they maintain their faith in the possiblity of once again discovering
the secret.
Other authority symbols, all of Lunda origin, also spread west
from Lunda with the kinguri. The later presence of four tubungu
insignia in Kasanje leaves little doubt that these reached the Mbundu
through the movement of the kinguri from Lunda. They included the
tuzekele (singular, Icazekele) bracelets, small metal rings denoting
lineage authority in Lunda but political authority in Kasanje,66 the
lubembe double clapperless bell,67 the mondo talking drum, and the
ngoma ya mukamba drum.68 Some other symbols found in both
Lunda and Kasanje, such as reservation of the leopard sldn for
political chiefs, occur too widely in Central Africa to permit deter­
mination of exact origins when the choice lies between such closely
related groups as the Mbundu, Luba, and Lunda.
Although only the kinguri left Lunda with the full complement of
Luba magical techniques, a number of other titled positions drifted
west at the same time. Some attention must be paid to their origins
since divisions dating from before their departure later influenced
splits within the original group of titles and led to the creation of
several Cokwe and Mbundu states. Most of the Lunda title-holders
who accompanied the kinguri belonged to lineages other than that of
their leader. The name of the kinguri’s lineage, according to official
Imbangala genealogies, was Njimba na Kakundo, a name which
also referred to the lands where the kinguri’s people had lived (their
65This bow reappears (in the hands of Kinguri’s successor, ICulashingo) in
later Imbangala traditions which come from independent non-Lunda sources;
see below, p. 192.
66Distinct from the hikano bracelet made o f human flesh which belonged to
the mwata yamvo alone.
67The lubembe is definitely Luba in origin but also associated with the
Lunda tubungu', personal communication from Jan Vansina. Its appearance in
Kasanje lends further support to the argument that some time had elapsed
after the arrival o f the Luba before the kinguri departed.
68Duysters (1957), p. 81, gave the list of insignia o f the tubungu chiefs. For
their occurrence among the Imbangala, see testimony o f Alexandre Vaz and
Ngonga a Mbande on the leopard skin (ciba ca kulwama); various testimonies
on the tuzekele, especially Mwanya a Shiba, 14 June 1969; the lubembe no
longer occurs in Kasanje, but Cavazzi (1965), i. 162, 201, mentioned it for the
seventeenth century (calling it ‘longa’); for the mondo, testimony of Apolo
de Matos, 5 Oct. 1969, and Dias de Carvalho (1890a), p. 501; testimonies of
Ngonga a Mbande, 26 June 1969; Sousa Calunga, 11 Sept. 1969, and Mwanya
a Shiba, 14 June 1969, for the ngoma ya mukamba.
 136 T H E PR O B L E M O F S T A T E -F O R M A T IO N
mpat).69 The positions which accompanied the kinguri (and later
became the makota70 who acted as guardians of the position in
Kasanje) came from two lineages apparently related to each other
but not connected to Njimba na Kalcundo. Positions called Kinzunzu
kya Malemba and two ‘nephew’ titles, Mbongo wa Imbe and Kal-
anda ka Imbe, both ‘sons’ of a sister called Imbe ya Malemba, came
from the lineage(s?) of Kandama ka Kikongwa and Kanduma ka
Kikongwa,71 The ancestral home of this group lay near the river
Lukongolo, said to be a stream somewhere in Lunda.72 The other
lineage, Kandama ka Hite, contributed the positions of Mwa
Cangombe, Kangengo, Ndonga, Kibondo kya Wulu, and Kamb-
wizo.73
Imbangala historians who prefer to emphasize the unity of the
makota with the kinguri rather than the lines which divided them
declare that all the Lunda titles came from the single ‘family’ of
Lucaze na Mwazaza,74 a female figure in the segmentary genealogies
who symbolizes a large set of Lunda matrilineal descent groups
which existed at some time in the past.73 As the maternal ancestor
of the lineages which controlled the Lundu tubungu, the Lucaze na
Mwazaza group included all of the individual lineages mentioned
by other historians as those of the kinguri and his companions. Since
69Testimonies o f Sousa Calunga, 29 Sept., 1 Oct. 1969; compare with the
Lunda traditions which specify only that the kinguri left with members o f his
own ‘family’; Dias de Carvalho (1890a), p. 76. The kinguri's lineage name
apparently no longer matters to the Lunda; this would not be surprising if
they left as early as the traditions imply and if the matrilineages later became
less important than they were at that time.
70The corresponding Lunda term is karula (pi. turula); Dias de Carvalho
(1890a), p. 70.
71Alternatively known by the title o f the major position in the lineage
Kinzunzu kya Malemba a Kawanga.
72One variant gives ‘Mukongolo’ as the name o f the lineage of all the Lunda
who came with the kinguri; testimony o f Sousa Calunga, 22 July 1969. This
appears to confuse a toponym taken from the name o f the river with the title
o f the lineage. This is a common practice. Neither name has been identified.
If, however, they consist o f the root -icongolo preceded by prefixes lu- (given
to most rivers in Lunda) or mu- (a common Bantu locative prefix), the name
could refer once again to the Luba origins of these titles; cf. p. 130, n. 53 above.
73Testimony of Sousa Calunga, 9 July, 29 Sept., and 1 Oct. 1969. This list
fails to account for one additional kota, Kahete, who later turned up in Kasanje
and apparently belonged to Kandama ka Hite.
74Testimony o f Alexandre Vaz, Domingos Vaz.
75 Cf. Biebuyck (1957), p. 815, who gives Mwazaza Mutombo as one o f the
three main dispersed Lunda groups. Mwazaza may also occur as ‘Mwasanza’,
a southern Lunda or Cokwe title. The Lucaze people live south o f the Lungwe-
bungu river in south-eastern Angola. Both Lucaze and Mwazaza point to the
Lwena/Colcwe connections for this group o f lineages. See Map I.
 T H E S PR E A D O F L U N D A PO L IT IC A L TITLE 137
the Imbangala often referred to the individual components of a group
by its broad collective name, the two versions of the genealogy
show no real inconsistency, rather only slightly different emphases.
Modern traditions probably preserve the names of only a modest
proportion of all the titles and lineages which left Lunda at the time
of the lcinguri. Further difficulties in identifying the precise composi­
tion of the party which left Lunda arise from the operation of the
so-called ‘lightning rod’ effect76 which has led later traditional
historians to enlarge the original group far beyond its historical
dimensions by adding titles of entirely different origin. Some recent
narratives claim, for example, that most of the senior nineteenth-
century political chiefs between the Kasai and the Kwango all left
Lunda with the lcinguri. The Cokwe of south-western Katanga
include Katende, Saluseke, Kandala, Kanyika lea Tembo, Cisenge,
Ndumba, Mbumba, Kapenda, ICasanje, and Kaita in their list.77 A
few of these names belong to early Cokwe titles but most refer to
latecomers who have recently become important in the region where
the tradition was told (Katende and Saluseke). The conventional set
of the major Cokwe Icings (Ndumba, Mbumba, Kanyika, and
Kandala) appears along with the title of a nineteenth-century new­
comer (Cisenge) and some of the other Lunda titles in the Kwango
region (Kapenda ka Mulemba of the Shinje and ICasanje of the
Imbangala). A published nineteenth-century Imbangala list of
Kinguri’s companions shows the same tendency; it adds the names
of several later Songo chiefs, various subordinate positions in
ICasanje, and some (but not all) of the true Lunda makota.7S
Cokwe States Based on the lcinguri
The people who held the lcinguri position moved westward from
Lunda very slowly, apparently settling repeatedly as they sought to
avoid the growing area of lineages under the influence of Luba
political institutions. The kinyama title, which seems to have left
at about the same time, moved in a different direction and eventually
came to rest among the Lwena of the upper Zambezi. The historical
developments behind the movement of the lcinguri, hidden in the
traditions behind the image of a ‘journey’,79 were continuations of
70The tendency o f founding kings to receive credit for the deeds o f their
successors; Vansina (1965).
77Van den Byvang (1937), pp. 426-7, n. 1(h), 432n. and 435. Cf, Lima (1971),
p. 46.
78Rodrigues Neves (1854), pp. 97-101, has the earliest recorded Imbangala
version o f the list.
79Cf. ICajinga’s ‘journey’ from Luanda to Mbondo, pp. 98-103 above.
138 T H E PR O B L E M O F S T A T E -F O R M A T IO N
whatever forces had initially expelled the title from the lueji’s federa­
tion of Lunda lineages. The descent groups in possession of the king­
uri pulled back from contact with the newly centralized Lunda of the
Kalanyi, entering areas where lineages organized according to the
segmentary institutions of the Cokwe and Songo gladly adopted the
kinguri and its associated magical powers. It is impossible to say
whether the kinguri title remained in the hands of the biological
descendants of the Lunda who left the Kalanyi or whether enterpris­
ing local groups took the title for themselves as it reached new
regions. If, as seems probable, the expanding segmentary lineages
of Mbumba a Mbulu were reaching the limits of available empty
land at that time, the attendant endemic conflict over scarce territories
might have led the descent groups to welcome the holders of a
prestigious title as arbiters in their continual conflicts.80
The kinguri title could not remain in a single location, however,
because even more effective political techniques based on the Luba
titles adopted by the nuclear Lunda were moving outward from the
Kalanyi close behind the kinguri. As the advancing wave of Luba
institutions reached each area where the kinguri became established,
the title drifted farther to the west and south, became fixed briefly
as an ephemeral Icing among a new group of segmentary lineages,
and then moved again as the next surge of Luba political innovations
caught up with it. Ultimately these repeated confrontations, perhaps
hastened in some places by local reluctance to accept any political
authority, led to the creation of a string of kinguri-states along a line
stretching from the Kalanyi across Cokwe territory towards the head­
waters of the Kwango and the borders of the Mbundu.
Despite the general reticence of the traditions on this level of
analysis, a number of details in the narrative episodes conform to this
interpretation. The Lunda later remembered that the kinguri's band
had moved very slowly and that it took many years for them to pass
beyond the Cokwe. The magical techniques which they had borrowed
from the Luba provided the key to their success in travelling through
unfamiliar territory. They hunted in part with traditional Lunda
snares and traps but also used the bows and arrows which Kinguri
80 This hypothesis sets a standard explanation (Vansina, 1966a, pp. 85-6)
against the background o f Cokwe/Lwena social structures. Some evidence
suggests that the lineage ‘frontier’ may have closed around this time. The
revolutions taking place in Lunda could have resulted from contact with the
Luba states in the north-east, and the Kongo to the south-west had certainly
encountered opposition from the kulembe and/or Libolo states by this time
(c. fifteenth century?).
 COKW E STATES BASED ON T H E K I N G U R 1 139
had obtained from Cibinda Ilunga/Lukokesha. A charmed dagger
called mukwale, in particular, helped them fight their way through
anyone who opposed their arrival.81 The power of the Icinguri’s
new weapons and strong Luba magic made a lasting impression on
the people who lived west of the Kasai where late-nineteenth-century
Cokwe still remembered the kinguri and his terrifying reputation.82
Such traditions as these would seem to give the viewpoint of essen­
tially stateless people witnessing for the first time the arrival of effect­
ive political authority.
Modem traditions unanimously retain this aura through their
depictions of the kinguri as a fearsome title which demanded the
lives of its own people with almost casual abandon. Kinguri, they say,
forced two slaves to kneel by his side whenever he wished to rise or
sit; he then rested his weight on daggers which plunged into their
backs.83 The image effectively portrays the awe with which the
kinsmen of the lineages regarded the kinguri. It does not suggest that
they viewed individual incumbents in the kinguri as cruel or demented
but that they feared the spiritual forces of the title and expected them
to demand drastic behaviour from their human guardians.84
The continuing opposition between the kinguri and the trailing
advance of Luba institutions is apparent in a Lunda tradition which
notes that Kinguri had left the Kalanyi in search of an army strong
enough to return and win back the power which he had lost in
81 Dias de Carvalho (1890a), p. 77. Lima (1971), p. 48, also cites Cokwe
memories o f wars which accompanied the arrival o f the kinguri in some
areas (although three o f the four adversaries mentioned, Pende, Holo, and
Shinje, refer to the history o f the kinguri in Kasanje rather than among the
Cokwe).
82 Dias de Carvalho (1890a), pp. 76, 87; also (1898), p. 28.
83 The image o f a ruler raising and lowering himself on daggers plunged into
slaves’ backs is a widespread cliché in Angola. Testimony o f Mwanya a Shiba,
15 June 1969; also various testimonies of Sousa Calunga; Also Alberto Augusto
Pires (1952), p. 1; Rodrigues Neves (1854), p. 97; Schiitt (1881), p. 79—all for
examples attributing the practice to Kinguri. Testimony o f Luciano, however,
attributed the custom to Nzinga. A. A. de Magalhães (1948), p. 33, recorded
from Songo sources that the mwata yamvo did it in Lunda. João Vieira Cameiro
(1859-61), pp. 172-3, noted that Ngola a Mbande, a seventeenth-century
Ndongo Icing, was said also to have done this. A . Bastian (1874-5), i. 313,
attributed the custom to the Duke of Sundi in Kongo. The cliché is common to
most o f the Mbundu and Kongo areas, at least.
84 Although Imbangala historians attribute Lunda origins to later and differ­
ent Kasanje rituals involving human sacrifice, these in fact came from other
sources. The idea o f Lunda origins probably came from their knowledge that
the mwata yamvo killed many people to propitiate the spirits behind their own
symbols. The Kasanje rituals, in any case, had no connection to the practices
described by the cliché.
140 T H E PR O B L E M O F S T A T E -F O R M A T IO N
Lunda.85 The Cokwe provide a detailed itinerary for Kinguri’s
‘journey’ which shows that he stopped for a ‘long time’ in at least
three places east of the Kwango.86 The title had crossed the Kasai
near the mouth of the Lonyi, a small left-bank tributary of the larger
river. It then moved up the Lonyi to higher ground and worked its
way west and south across the Luhembe, Kashimo, Lwana, Cihumbo.
Sombo, and Lwashimo rivers before making its first prolonged stop
among the lineages of an area known as Itengo (in the vicinity of the
modern city of Henrique de Carvalho). It then shifted to the region
known later as Mona Kimbundu, not far south-west of Itengo, where
it also remained for a long time. It finally continued on to the head­
waters of the Kwango and Kukumbi rivers and paused again before
moving on towards the Songo. Each of these stops represented the
establishment of a small state based on the kinguri title which the
Lunda viewed as a renewed threat to themselves.
A narrative episode which recurs in the traditions of both the
Lunda and the Imbangala seems to refer to the hostility which
marked relations between the kinguri and the Lunda at the time
the emigrants stopped near the sources of the Cilcapa river.87 Some
time after Kinguri had left the Kalanyi, Lueji sent messengers to
make contact with her brother. Kinguri received the emissaries in a
camp near a river called the Nangwiji (since renamed the Cikapa).
The Lunda begged Kinguri to return home but he refused to heed
the entreaties and rejected all further association with his former
kinsmen. Before continuing on his way, however, Kinguri renamed
the river ‘Cilcapa’ to commemorate his definitive separation. The
Imbangala today translate the word mutswalikapa as ‘we are sep­
arated’88 and allege that this event formally established the bound­
aries between the lands ruled by the Lunda and those controlled by
the kinguri. Since the border between the more recent kinguri-state
of Kasanje and the Lunda never lay along the Cikapa, this episode
must relate to a time before the kinguri settled in the Baixa de Cas-
sanje.89 This narrative episode therefore describes the formation of
85Dias de Carvalho (1890a), p. 76.
86Lima (1971), p. 48.
87See Pires (1952), p. 1, for the Imbangala; also testimony o f Domingos Vaz.
For the Lunda, Dias de Carvalho (1890a), pp. 86-90.
88Testimony o f Apolo de Matos, 4 Oet. 1969.
89This attempted reconcEiation could not have taken place after the kinguri
had reached Kasanje, since the tradition specified that it was the lueji who
sent out the call to the kinguri. Had it occurred later, the mwata yamvo rather
than the lueji would have initiated diplomatic relations between Lunda and
Kasanje.
 T H E D E C L IN E O F D E S C E N T G RO U PS 141
an earlier kingdom near the Cikapa (probably Itengo) where the
holders of the kinguri-title made a stand against reincorporation
into the expanding Lunda state.
The Lunda empire’s expansion towards the west corresponds to
the wave of Luba political institutions which drove the kinguri
farther and farther from the ICalanyi. Title-holders spread westward
from the centre of the empire to form small states which would act
as barriers between the Lunda capital and the kinguri. Several of
these rulers established themselves among the Cokwe in the move­
ment which would have brought the Luba figure of Cibinda Ilunga
to the Cokwe as a culture-hero.90 One of these rulers drove the kinguri
away from the second state on the upper Cikapa, and his title, the
mono kimbundu, has remained the name of the region the kinguri
vacated.91 The spread of title-holders from Lunda continued through
the seventeenth century as new kings settled among the Shinje
lineages living east of the middle Kwango; still others became the
western Lunda provincial governors who protected the mwata
yamvo from the new threat which later Lunda saw in the increasingly
powerful kinguri-st&te of ICasanje.92

The Decline o f Descent Groups among the kinguri’s Lunda

The years of intermittent movement and constant pressure from
advancing Luba produced fundamental changes in the social and
political structures associated with the kinguri title. The relatively
few persons who had originally left Lunda had been organized in
terms of a few closely related segmentary lineages (under Lucaze
na Mwazaza). But the rigours of repeatedly settling among new
lineages had forced the holders of the kinguri title to develop ways
of quickly integrating large numbers of people with no genealogical
connection to themselves. While the cohesiveness of the Lunda
lineages had permitted the small group to co-ordinate their activities
for purposes of defence, their chances of survival when faced with
the mobilizing techniques of the Luba depended on their ability
to attract large numbers of new and alien followers. Since social
structures based on segmentary descent groups lacked means of
integrating quantities of unrelated strangers, the need to build up a
90See Marie-Louise Bastin (1966).
91The hypothesis of a-series o f Cokwe/Lwena kingdoms ruled by the
kinguri both explains interpretive problems posed by otherwise anomalous
claims that the Cikapa once divided Kasanje from Lunda and fits with the
broader political question o f the development of the Lunda empires.
92Dias de Carvalho (1890a), p. 91.
 142
T H E PRO BL E M O F S T A T E -F O R M A T IO N

M a p vm . D ispersal o f Lunda P olitical Institutions (before c. 1600)

 T H E D E C L IN E O F D E SC E N T G ROUPS 143
large following forced the Icinguri to suppress cumbersome lineage
divisions within their band of followers.
An Imbangala narrative episode relates explicitly how the kinguri
incorporated groups of local kinsmen wherever it paused on its trip
through Cokwe lands. According to the traditional historians, by
the time Kinguri reached the area just west of the Cikapa river (now
known as Mona Kimbundu), his extreme cruelty had become worri­
some even to the Lunda followers who had left the Kalanyi with him.
They saw that his practice of killing followers every time he rose or
seated himself had greatly reduced their numbers and feared that it
might destroy them all if it did not cease.93 They felt themselves
powerless, however, to do anything except continue to furnish the
victims Kinguri demanded as long as the spirits behind the title
persisted in their need for such sacrifices.
The Lunda recognized an opportunity to replenish their supply of
potential victims when they arrived among the Cokwe who lived
near the sources of the Cikapa. The senior title-holder in this region,
Musumbi wa Mbali, had numerous subjects and was willing to send
a number of them to Kinguri where, the Lunda hoped, the Cokwe
would replace their own people as human sacrifices to their ravenous
central title. Musumbi wa Mbali suspected the real intentions of the
Lunda in requesting these people, however, and agreed to send them
only under the protection of one of his own subordinate title-holders,
Kasanje ka Kibuna (also called ICulashingo).94 Kasanje ka Kibuna
93Perhaps a reference to the decline in numbers o f full members of the
Lunda lineages relative to the number o f outsiders in the band?
94Perhaps an appointed emissary o f the vunga type, judging from the
kasanje title? The holders of this title later took control o f the kinguri position
and founded the state o f Kasanje during the 1620s. The etymology of the name
comes from shingo, Kimbundu for the back o f the neck; see the events sur­
rounding Kulashingo’s later accession o f power (p. 188, n. 32 below ); testimony
of Sousa Calunga, 23 Aug. 1969. Although this tradition establishes the origins
of ICulashingo in the region of Mona Kimbundu, published Imbangala tradi­
tions have wrongly claimed that Kulashingo came from Kisama, near the
mouth of the Kwanza river; Oliveira Ferreira Diniz (1918); testimony o f Apolo
de Matos, 18 June 1969; Pires (1952), p. 1.
The best-known but transparently erroneous etymology for this title has
led to major misconceptions about the early history o f Kasanje. Salles Ferreira
(1854-8) described Kulashingo as a ‘vassal’ o f the ‘Mwata Yamvo’. Francisco
Travassos Valdez (1861), ii. 155-6, repeated this version almost word for word,
but invented the added detail that Kulashingo had been a Lunda noble.
Vansina (1963a, p. 363) cited Valdez. Dias de Carvalho, who gained much of
his information and most o f his errors about Kasanje history from Salles
Ferreira, amplified the point by specifying that Kulashingo had been expelled
from Lunda and that he was none other than Kinguri himself (1898, pp. 15,
55). None o f this finds any confirmation in modem traditions.
144 T H E PR O B L E M OF S T A T E -F O R M A T IO N
would protect the Colcwe from any dangers presented by the Lunda
or by Kinguri. To help him, several related groups came as well,
led by his mother and her three brothers, Kibuna lcya Musumbi,
Pande ya Musumbi, and Mbumba a Musumbi.95 It was in this
manner that a Cokwe component, which later became very import­
ant in the history of the kinguri’s heirs, joined the band. Other Cokwe
groups, whose less illustrious, subsequent history has erased memory
of their incorporation, must have joined as well.
According to the structure of the Mbundu oral histories, a social
change of such fundamental importance should have been recorded
somewhere in the narrative episodes. Such a narrative appears to
have survived among the Cokwe, roughly in the area of the upper
Kasai, where the effects of abolishing the kinguri’s lineages seem to
have left the greatest impression.96
As the tradition tells the story, one of the women in ICinguri’s
group had been about to give birth when they had reached the bank
of a river said to be the Kasai. This annoyed Kanyika lea Tembo,
a companion of Kinguri who had prohibited all sexual contact,
ostensibly in order to avoid the delays caused by births and the
presence of small infants. This birth caused unusual difficulties
because the party was suffering from hunger at that time and needed
to keep moving in search of food. Even Kinguri’s magical powers
seemed to have failed, since his special bow had not provided
enough game to feed them. Although the group postponed their
crossing of the river until the woman had given birth, the mother
and baby both died. All accepted their deaths as a certain sign of the
disfavour of the ancestral spirits.
Kanyika ka Tembo appears to have been a lineage headman,
since the tradition portrays him as the uncle of the dead woman and
since her husband went to him to report the woman’s death.97
Kanyika ka Tembo berated the husband when he heard the news
and beat him severely for having violated his rule against sexual
contact with women. He concluded by ordering the man to bury his
wife and baby. This order violated all normal procedures in a society
based on lineages where the responsibility for burials belonged to
the dead persons’ kinsmen under the supervision of his lineage head-
95Compiled from testimonies o f Sousa Calunga, 16 June, 23 Aug., 29
Sept.; Sousa Calunga, ICambo ka ICikasa; Alexandre Vaz, 30, 31 July 1969;
Apolos de Matos, 18 June 1969.
96Van den Byvang (1937), pp. 433-4.
97The surname Tembo confirms the hypothesis; Tembo appears only as a
lineage name on the matrilineage side o f the Songo and Cokwe genealogies.
 T H E D E C L IN E OF D E S C E N T G RO U PS 145
man. The husband, who belonged to a different lineage, should have
had only secondary duties to perform in connection with the burial
of his wife and child.
The point of the tradition is that Kanyika ka Tembo had doubly
threatened his kinsman’s safety. He had ignored his duties to the
spiritual guardians of the lineage by his prohibition of sexual contact.
Normally, his responsibilities as lemba would have included the
distribution of the pemba (white clay) intended to safeguard the
procreativity of his kinswomen so that they might always bear
children and thus guarantee the survival of the group. He had com­
mitted a second transgression by shirking his responsibilities
in connection with the death of his niece. Failure to perform the
proper rituals could cause her spirit to return to plague her living
relatives.
The husband finally went to bury his wife at the river bank. There
he encountered a large flock of birds flying from the top of a tree.98
He managed to kill one of the birds, a bit of luck which constituted
a sign of supernatural favour, since he could not have felled a bird
on the wing without first achieving harmony with the spirits. When
he gutted the bird, he found millet and sorghum seeds in its gullet.
The seeds of domesticated plants amounted to certain evidence
of nearby human habitation and held out the possibility that the
hungry band of people might obtain food from these strangers.
This discovery revealed the husband as one favoured by the
spirits.
The leaders of the group, Kinguri and Kanyika ka Tembo among
them, assessed his experience as a sign that they should at once con­
tinue to the other side of the river. Since the stream was small
enough at that point to cross at a single leap, they began to cross on
foot. But each time one of them attempted to jump over the river
its waters rose abruptly and engulfed them, causing many people
to fall into the torrent and drown. Kinguri’s nzirngu, as well
as his magical bow, had apparently lost its power, since one of
its most important properties was to facilitate the crossing of
rivers. Their difficulties in obtaining food and crossing the river,
by implication, stemmed from Kanyika ka Tembo’s violation of
the ancient lineage customs and rituals which they had known in
Lunda.
In spite of the erratic behaviour of the river, they eventually
98 Had the translator of this tradition also recorded the species of bird and
tree, the information would have facilitated the interpretation o f the tradition.
146 T H E PR O B L E M O F S T A T E -F O R M A T IO N
crossed on a bridge of muyombo wood built by one of the other men
in the group, Ndonje." The Colcwe, like the Mbundu, regard
muyombo trees as the resting place of lineage spirits.100 The specifica­
tion of the type of wood used in the bridge emphasizes once again
that Kinguri’s difficulties resulted from the disfavour of the lineage
ancestors. The muyombo-wood bridge provided a means for the
Icinguri’s people to cross not only the river but also the metaphorical
gulf which separated them from harmony with the supernatural
world. With good relations re-established, however, they crossed
the river and quickly found a village where the people had abundant
stores of millet and sorghum. The villages received them hospitably
and gave them all they needed to eat. Ndonje and the Icinguri
then left with part of the group to continue westward and left
ICanyika ka Tembo and others to settle near the sources of the
Kasai.
This episode almost certainly reveals that the Icinguri’s band had
lost its basic organization in terms of matrilineal descent groups
through the assimilation of many unrelated foreigners. Kanyilca ka
Tembo had neglected the most basic rules of the lineages: his pro­
hibition against conceiving and bearing children eliminated the
procreative functions of their women and amounted to a crucial
step in the abolition of lineages in a matrilineal society. At the
moment that a woman bore an infant, the nucleus of a proto-
matrilineage took shape in the relationship between the mother and
the child. The Icinguri’s law against child-bearing thus constituted
the most direct means of eliminating lineages as the organizational
backbone of his following. Henceforth, all children would enter the
band through adoption or enslavement and would owe allegiance
only to the Icinguri. Motherhood and therefore kinship would cease
to exist.101
The tradition comes from descendants of Kanyilca ka Tembo,
people who had left the Icinguri rather than lose the security of their
90 ‘Ngondji’ in the published text, which certainly represents the well-known
title of Ndonje, given the numerous irregularities in the author’s orthography
of other personal names (cf. p. 156, n. 17 below).
100 Merran McCulloch (1951), pp. 75-6.
101The marked hostility to women shown by the icinguri has sometimes been
assumed to have been a result of military necessity. The argument runs that
women encumbered the military campaigns o f the band and therefore had to be
eliminated. The presence o f females did not, in fact, hinder the fighting activi­
ties o f the males. On the contrary, most African armies could fight only with
the logistical support o f their wives and children, who operated as a supply
train for the men who actually took part in battles.
 T H E D E C L IN E O F D E S C E N T G RO U PS 147
descent groups. It therefore embodied the pro-lineage biases of those
who would not tolerate life without kinship. Their hostility to the
kinguri’s new way of life appears most clearly in the salvation
achieved through reunion with the offended lineage spirits of the
muyombo-wood bridge. People taught to think in terms of unilineal
descent find it difficult to conceive of human society without lineages.
To them, any other social system appears chaotic and dangerous. The
episode implies that the hunger and other difficulties experienced by
the kinguri’s group resulted from their neglect of the principle of
lineage organization. In their view, reorganization must have afflicted
even the nzungu and the magic bow obtained from the Luba, since
these had failed to save the people from starvation.
The husband who had violated the prohibition against sexual
contact, on the other hand, had reaffirmed one of the basic principles
of the descent groups and he consequently enjoyed the blessing of
the spirit world. The omen sent to him in the form of seeds pointed
not just to a nearby source of food but also to a return to a settled
agricultural life based on lineages. The seeds symbolized both females
and the agricultural economy which the kinguri had rejected in favour
of wandering and pillaging. The Lunda closely associate women and
seed on a mundane level, since their wives customarily plant and
harvest most crops. Metaphorically, women and seeds become
equivalent since both give birth to a new life. The omen of the seeds
therefore indicated to Kanyilca ka Tembo (and to the tellers of the
tradition) that his future lay in a normal ‘civilized’ existence rather
than in the chaotic wanderings of the kinguri.
Other indications that the Lunda abandoned their lineages appear
in the consistent opposition of women to the kinguri in the traditions,
where females repeatedly threaten his emblems of royal authority.
The underlying theme that women contaminated Lunda-derived
political authority in ICasanje recurs in various traditions to point
metaphorically to the elimination of lineages. Only males who had
undergone the Lunda circumcision rites could wear the lukano.
Under certain circumstances the bracelet would not tolerate even
the presence of women nearby.102 The lukano’s incompatibility with
women is sometimes given as the reason why the lueji, a female title,
never gained full control of the bracelet while she ruled in Lunda.
One variant of the tradition makes the point explicitly, stating that
she had to entrust the lukano to her tubungu guardians during her
menstrual periods in order to avoid spoiling it and thus bringing
102 Testimonies of Domingos Vaz; Sousa Calunga, 21 Aug. 1969.
148 T H E PR O B L E M O F S T A T E -F O R M A T IO N
misfortune to her people. On one such occasion, Lueji gave the
lukano to Cibinda Ilunga, her husband, instead of to her advisers.
This angered Kinguri and led to the dispute which resulted in his
leaving Lunda.103 This narrative episode draws on the omnipresent
theme of women’s offensiveness to the kinguri title in order to explain
its departure from Lunda. It explains the origin of prohibitions
which forbade any women to touch the kinguri’s emblems or to
hold that position. Since the Lunda had lost their sacred authority
symbol to the Luba while it was in the possession of a female,
the kinguri would take great care about the presence of women in
the future.
Another tradition, perhaps derived from the same historical inci­
dent, occurs with a different plot line among the Imbangala. They use
the same image of a young woman who died during the crossing of a
river to point once again to the decline of lineages in the kinguri’s
group. According to the Imbangala, when Kinguri reached the
Kwango river, at that time called the Moa, its great width presented
a serious barrier to his progress. Kinguri himself leapt across the
river easily by virtue of his magical powers, but his companions
found the leap impossible. Kinguri had a daughter named Kwango
who wished to cross the river just as her father had done. The makota
warned against such a foolhardy attempt, saying that since she was
merely a woman she could not manage the feat. She tried, failed, and
drowned in the river, which from that time onward became known
by her name.104
Despite the differences in locale and in the superficial plot, this
story is structurally identical to that of ICanyika ka Tembo and the
muyombo-wood bridge over the Kasai. Both hinge on the relationship
between women and the difficulty of crossing rivers. Both contain
the theme of women as liabilities to Kinguri, and both use rivers as
metaphors for otherwise unexplained obstacles which faced the
migrating Lunda.105
The different locations ascribed to the event do not affect
the equivalence of the episodes, since each informant placed it on
the largest river near his own king’s lands. The Imbangala chose the
Kwango, which flows near Kasanje, while the other version of the
story came from peoples living near the Kasai. The identical structure

103 Schiitt (1881), pp. 82-3.

104 Testimony o f Domingos Vaz.
105 Cf., once again, the significance of the river in the stoiy of Kajinga’s
trip to M bondo; pp. 97-102 above.
 T H E D E C L IN E O F D E S C E N T G RO U PS 149
in both cases points out the incompatibility of women and lineages
with the survival of the kingurVs people.
Suppression of lineages allowed any outsider to join the kingurV s
people with status equal to that of all earlier members. This change
overcame the limitations on recruitment imposed by the original
kinship structure of the group and permitted unlimited assimilation
of new members. It also meant that lineage loyalties no longer diluted
the total obedience which the kinguri demanded of his people. With
lineages, the kinguri had been one title-holder in a complex network
of lineage headmen, malcota, and other lineage-controlled positions;
abolition of the lineages necessarily weakened the other titles and
centralized all authority in the kinguri. The new centralized organiza­
tion brought obvious advantages in terms of group unity and fighting
capacity which permitted the kingurVs band to survive as it passed
through unfamiliar territories controlled by enemies.
The kinguri’s radical solution evidently represented an intolerable
sacrifice to the title-holders whose positions lost influence under
increased centralization. Various chiefs abandoned the kinguri
all along their route, where they stopped, adopted a more settled
mode of life, and established themselves as new rulers over the local
lineages. The prestige of their Lunda emblems of authority made this
relatively easy to accomplish among groups without this form of
kingship. This fission process accounts for the settlement of several
chiefs later identified as Colcwe, such as Mwata Kandala and
Ndumba a Tembo.106
Through the continual incorporation of aliens, the kingurVs band
had evolved by the time it reached the upper Kwango from a small
group of related lineages into a larger and more cohesive band
lacking the particularistic and potentially divisive presence of seg­
mentary descent groups. This change both contributed to and
resulted from the continual movement of the title through a series of
Colcwe kinguri-states. This type of social and political organization
solved the problem of incorporating new recruits into the band, but
it simultaneously introduced new difficulties for its leaders as they
fought to retain the loyalty of those who resented the loss of lineages.
Disaffected headmen, such as Kanyika ka Tembo, had begun to
abandon the main group to settle again even before they reached the
Kwango. Still, the main band had found a solution to the problem of
state-formation among the segmentary lineages to the east of the
Mbundu which would have dramatic consequences in the west in
106 ‘Mwandumba’ in van den Byvang’s text.
150 T H E PR O B L E M O F S T A T E -F O R M A T IO N
the next century. Further, they had the political positions—Kula-
shingo who would become the kinguri in Kasanje, mwa Ndonje,
Munjumbo, Kabulcu ka Ndonga, and doubtless others—that by
1650 had become royal titles in all the major Mbundu states of
Angola.
 C H A P T E R VI

T he Imbangala K ilom bo-

A Radical Solution

T hree m a j o r lines of early Mbundu political development con­

verged in the upper Songo region during the middle part of the six­
teenth century to produce an entirely novel resolution of the tensions
which historically had set the particularistic Mbundu lineages against
the centralizing tendencies of Icings. Facing the ngola a kiluanje and
Libolo near the upper Kwanza, the kinguri’s lineageless band con­
fronted large and centralized kindoms for the first time since it had
fled the Luba in Lunda. Also present in the same region was an
Ovimbundu male initiation society, called the kilombo, one of the
numerous social structures described in Chapter II as cross-cutting
institutions. The Lunda title-holders who had come with the kinguri,
under pressure from the large states to the west, fused with the
kilombo some time in the sixteenth century to form powerful bands of
mobile warriors, known as Imbangala, who overspread the Mbundu
region after 1610 and ultimately settled down to found a new set of
Mbundu states which included all the major powers of the region
after 1650—Kalandula, Kabulcu ka Ndonga, Matamba, Holo,
Kasanje, Mwa Ndonje, and others.

Songo States Based on the Lunda Titles

The dissensions which had caused some followers of the kinguri to
settle among the Colcwe east of the Kwango finally split the band
apart after it had crossed the river into territory occupied by the
Songo. When the holders of positions in the once unified set of Lunda
titles went their separate ways, spreading as far as the Benguela
plateau, a few remained as ephemeral Songo positions, much like the
other titles which had preceded them in this area. The result was a
reorganization of the segmentary Songo lineages into yet another set
of short-lived states not unlike those the kinguri had headed among
the Colcwe or the coalitions of Songo descent groups established
earlier by title-holders from the south-west.
Holders of the kinguri title established a fourth small kingdom
152 T H E IM B A N G A L A K I L O M B O
among the segmentary lineages of Mbumba a Mbulu, this one just
south-east of the borders of the powerful Libolo state which, at the
height of its power at that time, extended beyond the Kwanza to the
upper Lui river. Despite some ambiguity in the traditions, this state
probably lay near the Luhando and Jombo rivers in upper Songo.
The oldest recorded Imbangala traditions identified the area of this
kinguri kingdom as near the sources of the ‘Pulo’ and ‘Lukombo’
rivers in lands occupied by Cokwe during the 1840s and 1850s.1 A
stream called Lukumbi, probably the ‘Lukombo’, flows into the
upper Jombo at approximately 10° 50' S.2 A mid-nineteenth-century
Ovimbundu tradition also concurred in identifying the region as near
the Luhando river.3 Modern Songo sources, perhaps because of their
more intimate knowledge of the region, refer to the area where the
kinguri stopped by a praise name not noted on available maps,
‘Mutonde a Kalamba Kizembe’.4 The exact boundaries of the
kinguri’s Mbundu kingdom thus remain uncertain, but all indications
place its centre in the lands of the upper Songo drained by the
Luhando and Jombo rivers.
Over a considerable number of years,5 the kinguri applied the tried
and true Mbundu technique of awarding subordinate named titles to
extend their influence over the southeastern Songo lineages just
beyond the Kirima groups controlled by the hango-kings of Libolo.
The descent groups inhabiting that region claimed a common
ancestress called Manyungo wa Mbelenge. Her descent from Tembo
a Mbumba through female links in the perpetual genealogies marked
her as the symbolic progenitor of a group of Songo lineages. A
‘marriage’ of the usual type between Manyungo wa Mbelenge and
Hango shows that Libolo kings had conquered these Songo lineages
before the arrival of the kinguri.c A second ‘marriage’, this time with
1 Rodrigues Neves (1854), pp. 98-9.
2 The ‘Pulo’ appears on no map known to me; Imbangala historians do not
today locate this state by the names o f rivers but do recall another designation,
probably a praise name, Kanzulu ka Mbwa. Testimony o f Domingos Vaz.
Cf. Rodrigues Neves (1854), p. 97, who gave ‘Kahunze’ as the praise name. The
modern Imbangala concur in locating the state vaguely west of the Kwango.
3 Magyar (1859), p. 286. Childs (1949), p. 173, argued that the Magyar
tradition did not come from Bihe, evidently basing his opinion on his inability
to obtain confirmation of the story from mid-twentieth-century informants
there. Internal evidence presented below, however, definitely identifies the
tradition as Bihe in origin.
4 A. A. de Magalhães (1948), p. 35.
5 Rodrigues Neves (1854), pp. 98-9.
0 Genealogies recited in testimonies of Sousa Calunga, 16 June, 9, 21 July
1969; Manuel Vaz; Domingos Vaz; Alexandre Vaz, 31 July 1969; Apolo de
 SO N G O STATES BASED ON T H E L U N D A T ITLES 153
the kinguri, represents the settlement of the Lunda title among the
lineages of what was then eastern Libolo.*7 Songo genealogies,
expectably the most detailed on the expansion of this state, show
‘marriages’ between the kinguri and at least two other female figures
representing lineages of the region. One of these, Kahanda, left three
descendants, Kakende, Mushinda, and ICunga, as evidence of the
former Songo kingdom of the kinguri,8 Kakende and Mushinda have
no identifiable modem incumbents, probably as a result of the name
changes which generally hinder reconstruction of Songo title histor­
ies, but ICunga had apparently been one of the earlier Libolo chiefs
settled among the Songo of the lower Luhando where he ruled
lineages which traced their matrilineal ancestry to Kavunje lea
Tembo, a title connected with the Libolo kingdom.9 The association
of these lineages with Libolo, also supported by ICunga’s location
near the Libolo outpost on the island of Mbola na Kasashe in the
Kwanza river, shows that the kinguri replaced overrule by the hango-
leings when they settled in this region. They incorporated the lineages
of the area by absorbing the older title of ICunga according to the
name-changing methods employed in the expansion of both the ngola
a kiluanje and Mbondo states.
The confrontation between the advancing kinguri and the kingdom
of Libolo seems to have provoked a period of intense civil wars
among the Lunda in which several of the secondary Lunda title-
holders abandoned the kinguri in favour of establishing their own
settled states. The Lunda emigrants had met no kingdom strong
enough to resist the supernatural powers of their leader between the
Lulua and the Kwango. There they had moved through an area of
politically unorganized lineages and could penetrate farther each

7Testimonies of Domingos Vaz; Sousa Calunga, 1 Oct. 1969; Apolo de

Matos, 18 June 1969; cf. Rodrigues Neves (1854), p. 97, who mentioned a wife
o f Kinguri but confused her name/title with that of Lueji.
8A. A. de Magalhães (1948), pp. 33-5. The Imbangala genealogies noted
in n. 6 above also show a ‘Kunga’.
9See map in Schiitt (1881). Also Lux (1880), p. 96. Testimonies o f Sousa
Calunga, 29, 30 Sept. 1969; Alexandre Vaz, 31 July 1969.

Matos, 5 Oct. 1969. The calculation o f the sequence rests on the fact that
titles derived from Manyungo wa Mbelenge, some as far removed as the third
descending generation, had already become prominent by the 1620s, notably
the Kaza (lea Hango). Kalunga lea ICilombo leya Wabo wa Hango (three gen­
erations removed from Manyungo wa Mbelenge) became powerful at about
the same time. The imputed chronology does not depend on biological life
spans but on the assumption that some time was required for a new and
dependent title to gain power in its own right.
154 T H E IM B A N G A L A K 1 L 0 M B 0
time they felt the threat of Luba title-holders behind them. But
Libolo for the first time offered serious opposition to their westward
progress. The combined pressures from pursuing Luba chiefs in the
east and resistance from Libolo in the west squeezed disaffected
elements of the kingurVs band outward towards the north and south
where no comparably powerful states yet existed, and then, as we
shall see, provoked mutiny by the remnants who remained.
Although the historical difficulties encountered by the icinguri in
Songo probably stemmed from inability to defeat the Libolo rulers
beyond the Kwanza, the traditions typically blamed his failure on
inadequate magical authority symbols. ICinguri, the traditions relate,
attempted to restore the potency of his emblems, already in serious
doubt because of the hunger and frustration his followers had
experienced in crossing the Kasai (or the Kwango?), by sending back
to Lunda for additional magic which might overcome the unexpectedly
strong opposition from Libolo. Kinguri requested one of his com­
panions, Ndonje,10to go to Lunda and return with one of the emblems
of the Lunda tubungu chiefs, a drum called the ngoma ya mukamba.
This drum was a great war charm and later became the most powerful
war drum of the Lunda mwata yamvo.11 The kinguri apparently
delegated this mission to the ndonje because his responsibilities
included care of the kinguri'1s authority emblems, an honour given
him in return for his help in building the bridge of muyombo wood
across the Kasai.
This manoeuvre may have indicated the kingurVs growing dis­
illusionment with the Luba symbols of power he had obtained from
Cibinda Ilunga, hence his attempt to restore the ancient Lunda
authority emblems of his position. The coincidence of the kingurVs
request for tubungu lineage emblems and the ndonje's association
with the muyombo wood, also a symbol of lineage solidarity in
contrast to the essentially non-lineage Luba magic, probably indicates
that the kinguri had settled, temporarily at least, as a normal lineage
state among the Songo. His award of permanent titles there supports
this hypothesis and contrasts with the lack of similar positions east
of the Kwango where lineages had disappeared.
The departure of the holder of the ndonje title ended not in the
intended revitalization of the kingurVs powers but instead led to the
creation of a new state among the lineages living along the middle
10 Ndonje means teacher or master in Kimbundu; Assis Jr. (n.d.), p. 32.
11 Mukamba in Kimbundu refers both to the dry season o f the year and to
the manioc plant. Ngoma is a widespread Bantu word for drum.
 SO N G O STATES BASED ON T H E L U N D A TITLES 155
Kwango. The ndonje neglected to search for the ngoma ya mukamba
in Lunda and abandoned the Icinguri to settle among the Minungu
people who seem to have formed a western subgroup of the
Colcwe lineages. There his position became the mwa ndonje and
remained until modern times as the title of an important Minungu
king.
Another Lunda title-holder, the mnnjumbo, left to move south
while the Icinguri remained settled in Songo.12 The full title of this
position, Munjumbo wa Ngamba or Munjumbo wa Konde, indicated
its origin as a senior Lundu tubungu title, either a brother or an uncle
to the Icinguri.13 The position had left Lunda with the Icinguri but was
separated from it in Songo, and its holders independently founded at
least two other major states, first one among the Songo and later
another in the Ovimbundu highlands. The first state resulted from the
munjumbo'’s temporary settlement as a permanent named position
among the Songo lineages living north of the Luhando along the
upper Lui;14 there its holders granted Lunda titles to the descent
groups living around them.
This Songo state then expanded towards the south-west along the
borders of Libolo and developed a secondary centre in the Benguela
plateau area then dominated by the kulembe. Traditions both from
the core of the original munjumbo kingdom in Songo and from
Ovimbundu brought under the control of this title somewhat later
confirm the general direction of this movement. The Songo on the
Mwiji river, an affluent of the upper Luhando, recalled that Mun­
jumbo came to their area from the north, that is, from the lower
Luhando.15 Northern Songo traditions state that Munjumbo went
from there to the mountains called ‘Nzambi na Ngornbe’.16Although
this name alone does not identify these mountains, circumstantial
evidence suggests that it may refer to the Ovimbundu highlands.
‘Nzambi na Ngombe’ translates as the mountains of ‘the great spirit
and of cattle’. The Ovimbundu are not only the nearest mountain
dwellers known to the Songo but are also generally associated by the
Mbundu with cattle.

12 Modern pronunciations o f the name vary from Munjumbo to Minjumbo

and Muzumbo in conformity with sound shifts characterizing these dialects.
13 Testimonies of Domingos Vaz; Ngandu a Kungu. See p 121, n 29 above.
14 Testimonies o f Sousa Calunga, 29 Sept., 1 Oct. 1969; also Domingos Vaz.
Schiitt (1881), p 111; Capello and Ivens (1882), i. 191.
15 Capello and Ivens (1882), i. 158.
16 A. A. de Magalhães (1948), p. 33. Magyar (1859), p 243, noted that the
Bihe dynasty, the best-known heirs of the munjumbo, was known as ‘kangombe’.
156 T H E IM B A N G A L A K I L O M B O
The establishment of the munjumbo states provided another
instance of Mbundu borrowing mavunga titles from beyond the
Kwanza. The traditions recall that Munjumbo acquired fearful
supernatural weapons, which may be identified as mavunga, in
order to build these kingdoms. As a Cokwe narrative episode
described these events, Munjumbo conquered the Songo of the
middle Kwanza with the aid of a magical knife or hatchet (the Cokwe
were apparently unclear as to the exact nature of this obviously alien
weapon) called mwela. The mwela had the ability to fly out of its
sheath and subdue all who resisted the wishes of its owner, crying
with a human voice while it flew about the wilderness in pursuit of its
master’s enemies. In the end, however, the magical knife turned
against Munjumbo and killed him. Munjumbo’s descendants in his
own lineage inherited the knife and ever afterwards kept it for
themselves.17
Linguistic evidence identifies the mwela as a symbol borrowed from
a state in an Umbundu-speaking area. Later ldngs of Kasanje, whose
background included strong Ovimbundu influences, used a sacrificial
knife very much like the munjumbo’s which they also called mwela.18
The Umbundu generic term for knife is mwela, while ICimbundu uses
several other dissimilar words, poko, mbele, or mukwale.19 An
alternative symbol of the same supernatural forces in Kasanje was
hatchet called kimbuya. The modern Imbangala recognize that the
shape and construction of the kimbuya differed from the common
Mbundu hatchet and suspect that it might have come from non-
Mbundu sources. The word kimbuya does not exist in ICimbundu but
in Umbundu refers to the machete or catana.20 Both words for this
weapon clearly owe their origins to the Ovimbundu region. Imban-
gala historians support the linguistic evidence by citing vaguely
known connections between the munjumbo and ‘Bailundo’.21
The very vagueness of these connections supports the hypothesis
that the munjumbo state expanded through appropriating the vunga-
titles of its south-western neighbours. The mwela or mukwale was
17Van den Byvang (1937), p. 435. The published form o f this tradition
distorted this name as well as that o f the ndonje, in this case giving ‘Mung-
andja’.
18 For the mwela in Kasanje, testimony o f Sousa Calunga, 2 Oct. 1969;
Apolo de Matos, 18 June 1969.
19Pereira do Nascimento (1903), p. 49. Alves (1951), i. 764. The Lunda
word was mukwale-, see Mattos (1963), pp. 308-9.
20A long heavy knife used both for cutting and for chopping. Testimony of
Sousa Calunga, 11 Sept.; Alves (1951), i. 706.
21Testimony o f Sousa Calunga, 1 Oct. 1969.
 SO N G O STATES BASED ON TH E L U N D A TITLES 157
everywhere used as an emblem of appointive authority and carried
no permanent heritable mandate. The traditions therefore failed
to capture and preserve a clear record of the munjumbo's ties to
the south-west, just as modern Mbondo traditions are unclear on the
links between the ndala and the hango. The historical fate of the
south-western w»?ga-holders in the munjumbo state provides a further
parallel with the history of Mbondo, since, like the Libolo ndala
among the Mbondo, the munjumbo’s Ovimbundu province survived
as an independent kingdom long after the parent Songo state had
disappeared. The south-western munjumbo state covered the northern
slopes of the Benguela plateau, where nineteenth-century Bihe kings
preserved the name Munjumbo wa Tembo as the founder of their
dynasty.22 The conclusion of the Colcwe narrative episode, in which
the munjumbo died when his own magical knife returned to slay him,
metaphorically described the defeat of the Songo munjumbo by
wwe/a-wielding holders of the south-western vunga.23 The passage of
the knife to descendants in the lineage of the munjumbo, rather than
to another political title, meant that the Songo lineages formally
reclaimed the position, thus freeing the ancestors of future Bihe kings
to build their state without restrictions from the original sponsor of
their own title. The original munjumbo declined to the status of a
local Songo position and changed its name to reflect the loss of its
ties to the other Lunda positions.24
The somewhat confused written version of the seventeenth-century
Mbundu traditions on the arrival of the Lunda in Songo substantially
confirms the evidence on the munjumbo obtainable from later
22 A. A. de Magalhães (1948), p. 33, located descendants o f the Songo
munjumbo position in the area o f Andulo, centre o f the old Bihe kingdom For
the Bihe data, see Capello and Ivens (1882), i. 191. The Bihe learned the sur­
name Tembo from the Songo after the Lunda form o f the name had moved
south, leaving only the Songo form Munjumbo a Tembo in the north Magyar
(1859), p. 266, linked the Bihe kings with the kinguri but did not mention the
munjumbo by name; there is none the less an implied connection.
23 The Imbangala genealogies add that after the munjumbo settled among
the Songo and expanded south to the Kulembe region, holders o f this title
eventually moved back towards the north, reasserting their authority over
some of the Songo lineages on the basis o f new Ovimbundu political symbols
(the mwe/d).
24 Ngandu a Kungu confirms that the modern title is basically a Songo
position. Schiitt had Munjumbo wa ICafushi as a Songo founding king who
came with Kinguri. A. A. de Magalhães (1948), pp. 33ff., gave Muzumbo as
the son o f ‘Gambo’ (Ngamba?) and Kafuti (the equivalent o f Kafushi in the
Songo dialect) lea Mvula. Capello and Ivens had Munjumbo wa Tembo as the
founder of the Songo and companion o f the founding kings o f the Cokwe
and Imbangala.
158 T H E IM B A N G A L A I C I L O M B O
genealogies and narratives. According to this tradition, which must
be handled with great circumspection,25 a band of warriors, united
under a leader misnamed ‘Zimbo’ (but probably a distorted derivat­
ive of the title of the munjumbo26) had reached an unspecified part of
the interior. Each of the subchiefs who had accompanied ‘Zimbo’
then went his own way and settled among the various inhabitants of
the region. Chiefs called ‘Ndumba’ and ‘ICandonga’ settled in
unidentified quarters, and the old Icing ‘Zimbo’ eventually died,
leaving his former vassals in their respective lands where they still
ruled as independent Icings in the mid-seventeenth century.27
All the details of this tradition connect these events with the arrival
of the kinguri among the Songo and the subsequent dispersal of the
original band of migrants. Ndonje, the tradition specified, had come
as one of ‘Zimbo’s’ subchiefs but had settled in ‘Greater Ganguella’,
the seventeenth-century name for upper Songo. The name of
Ndonje’s wife, ‘Musasa’, provides another clue to the identity of this
group, since ‘Musasa’ probably designated the matrilineages of
(Lucaze na) Mwazaza,28 descent groups still comiected by the modern
genealogies with the Lunda titles of which the ndonje was one. The
other two titles named in the 1650s, Ndumba (a Tembo dya Mbumba
a Mbulu) and (Kabulcu) lea Ndonga (‘Kandonga’) all appear in other
sources as companions of the kinguri and the munjumbo. Seventeenth-
century opinion located ‘Zimbo’s’ (i.e. munjumbo’s) original kingdom
somewhere in the unknown highlands south of the Kwanza.
A Bihe tradition recorded in the 1840s gave a quite literal descrip­
tion of these events which once again confirms this interpretation of
the munjumbo’s separation from the kinguri and his subsequent
expansion towards the south. As the Bihe told the story, Kinguri’s
band reached the river Luhando after many years of fighting their
way through hostile peoples and strange lands in the east. They
practiced customs (jkesila laws’) involving extensive cannibalism and
killed many humans as sacrifices to the spirits represented by their
chiefs. These spirits required so many victims, in fact, that the band
could not settle long in a single spot without driving away or killing
most of the neighbouring peoples. This repeatedly forced them to
25 Joseph C. Miller (forthcoming (a)).
26 Mbundu informants may have contributed to the error by pronouncing
the name ‘munjimbo’, as a result o f the tendency of eastern Kimbundu-
speakers to exchange the high front and back vowels (i and u) of Songo and
Libolo names; cf. p. 155, n. 12 above.
27 Cavazzi (1965), i. 176-7.
28 See genealogies reproduced in Figure V.
 SO N G O STATES BASED ON T H E L U N D A TITLES 159
move on in search of new populations to devastate in order to
satisfy their leader’s cruel demands.
When the band finally settled on the Luhando, they soon exhausted
the capacity of the Songo to provide sacifices and turned upon each
other to satisfy their need for human flesh. Some leaders foresaw that
internal conflict would weaken them until they would fall prey to the
vengeful enemies who surrounded them. These leaders decided to
abandon their cannibalistic customs and nomadism in favour of a
settled life based on agriculture. They formed a secret society in order
to promote those ends. The leader of the original band, the kinguri,
remained devoted to his warlike ways, however, and recognizing that
his opponents threatened to undermine his authority, opposed the
formation of their new society. A series of bloody battles followed
between the Icinguri and the members of the secret society, but no
clear victor emerged. In the end, the proponents of a settled life
deserted their warlike rivals and migrated south-west across the
Kwanza. They gradually dispersed and settled down to rule the
populations living there and adopted the local sedentary style of
living.29
The plot line of this narrative—wars between members of a migrat­
ing band over the issue of their leader’s cruel domination, movement
south-west by a part of the band, and gradual fragmentation which
resulted in the settlement of kings in Bihe and other parts of the
Ovimbundu highlands—clearly describes the dispersal of Lunda and
related titles in upper Songo. The form in which the story was pub­
lished reveals its origins among opponents of the kinguri, doubtless
the Bihe descendants of the munjumbo. The main narrative theme
clearly leads to the foundation of the Bihe kingdom even though the
written version of the tradition did not mention it specifically.30
Emphasis on the kinguri's cannibalism and cruelty might be expected
from Bihe descendants of the munjumbo who justified their abandon­
ment of the kinguri by picturing him as an inhuman monster.31
29 Magyar (1859), pp. 266ff.
30 Colcwe or Lwena sources, either directly to the European who rendered
the tradition and had travelled east o f the Kwango, or indirectly through Bihe
traders in the east, may have contributed details to the published version; for
example, Sha Kambunje, a nineteenth-century Lunda governor o f the middle
Kasai, was accorded status equal to that of Kinguri.
31 The association of the cannibalistic rituals mentioned in the tradition
with ‘icesila laws’ probably indicates the writer’s familiarity with seventeenth-
century written sources on the so-called ‘Jaga’ which generally identified
cannibalism with ‘Icesila'. Ocisila is the Umbundu form of the Kimbundu
word kijila used by Cavazzi and presumbably by the later Portuguese to refer
160 T H E IM B A N G A L A K I L O M B O
The published tradition’s term for the secret society founded to
oppose the kinguri, ‘empacasseiros’, both identified its leader as the
munjumbo and suggested one way the munjumbo’s kingdom may have
been structured. The term ‘empacasseiros’ may be equated with the
kibinda version of the professional hunting society brought from
Lunda with the lcinguri. The word empacasseiros was a Portuguese
term for elite African auxiliary troops attached to European armies
in Angola since the early seventeenth century. The Portuguese had
formed the word from ICinbundu mpalcasa, the red buffalo (bos
caffir), plus the common Portuguese suffix -eiro for the person associ­
ated with an object,32 since these mercenaries derived their elite
status from their skill in the pursuit of large game like the mpalcasa.
The concept of skilled hunters specializing in dangerous game was,
however, purely African. The men who fought as empacasseiros for
the Portuguese must also have ranked as yibinda among their kinsmen
in the villages. Since Imbangala historians emphasize that the
munjumbo held the status of kibinda while the kinguri did not,33*
identification of the lcinguri’s opponents as empacasseiros marks them
as yibinda led by the munjumbo.M A title-holder eager to build new
kingdoms would have been looking for an appropriate cross-cutting
institution on which he could base his supra-lineage political organiz­
ation. Since the kibinda society cut across existing lineage divisions,
it provided a structure that the munjumbo evidently converted to
political purposes, in conjunction with the mavunga appointive titles,
for building his new states.35
32e.g. cozinha (kitchen) H- -eiro = cozinheiro (cook).
33Testimonies o f Sousa Calunga, 29 Sept., 2 Oct. 1969. Although this narra­
tive episode may appear to describe the formation o f the empacasseiors
or the kibinda, it originally dealt with the origins of the Bihe lungs. Magyar’s
synthesis o f various traditions, African and European, produced a misleading
impression. The Portuguese, not the lcinguri's enemies, created the empacas-
seiro in its proper sense o f mercenary. The kibinda society probably diffused
with the Lunda political titles from the east.
34The opposition between the kinguri and the munjumbo also operates at
the metaphorical level. The savagery o f the kinguri made him seem like a
ferocious supernatural beast, while the munjumbo as kibinda had special
qualifications for slaying animals o f that type.
35The Bihe tradition (as well as the Songo versions o f these events) implies
that the kinguri left Songo and eventually settled in Kasanje without further
change in the composition o f his group. This dates the tradition to the period

to the ‘laws’ among the Africans in seventeenth-century Angola that they

considered cruel and unreasonable; Alves (1951), ii. 1280. Nineteenth-century
oral traditions current among the largely illiterate Portuguese population in
Angola may have contained these stories even if the writer had not read the
seventeenth-century sources.
 O R IG IN S O F TH E K 1 L O M B O 161

The Songo states of the kinguri, the mwa ndonje, and the munjumbo
illustrate themes already familiar from the earlier history of the
Mbundu lineages. Alien Lunda titles entered the region without firm
attachment to any specific groups of lineages. The Songo descent
groups then appropriated the Lunda positions and made them
centres of at least three small kingdoms by procedures which re­
sembled the Mbondo conversion of similarly lineageless mavunga
from Libolo. At least one of the Lunda positions expanded into a
major state, that of Munjumbo (a ICafushi/Tembo/Kalunga, etc.),
through awarding vunga positions based on the Ovimbundu mwela
magical knife. Others, such as the kinguri, expanded briefly by
creating subordinate perpetual titles of the native Songo type. The
dispersal and settlement of the Lunda titles as lineage positions,
provoked in this case by confrontation with Libolo, merely continued
the fission process which had begun while the kinguri's band moved
through Colcwe lands far to the east. The propensity of the Lunda
titles to revert to a settled existence suggested that the lineageless band
did not offer a congenial environment in which they could provide
effective rule. Their experience among the Songo thus presaged many
of the changes which would affect related positions north of the
Kwanza during the seventeenth century.

Origins o f the Kilombo

The remaining Lunda title-holders sought a solution to the prob­
lems of disunity which they had experienced under the kinguri by
adopting an Ovimbundu warrior society known as the kilombo. The
kilombo provided two things which the kinguri's original band had
lacked: a firm structure capable of uniting large numbers of strangers
who had evidently never replaced their lost lineages with comparably
viable social or political institutions, and military discipline capable
of defeating the large kingdoms which blocked their movement north
beyond the Luhando and west of the Kwanza. The mature kilombo,
which ultimately proved capable of defeating the most successful
Mbundu states up to its appearance north of the Kwanza, consisted of
a blend of the Lunda perpetual titles, mavunga positions originated

before the kinguri left Songo, since events immediately afterwards resulted in
abolition of the position and a long series o f wars that fragmented the band.
The Bihe tradition is therefore older than most other versions and has not
undergone the same distortions as Imbangala and other Mbundu versions o f
these events. This makes it preferred evidence for determining what transpired
in Songo.
162 T H E IM B A N G A L A K I L O M B O
among the Ovimbundu, and a warrior cult developed somewhere in
the lands of the kulembe.
In view of the evident rapidity with which the kilombo meta­
morphosed from decade to decade, and the shortage of data bearing
directly on it before the mid-seventeenth century, an attempt to
develop too detailed a description of its internal structure at this time
would probably be unwise.36 The available information does show,
however, that the kilombo first matured as an adjunct to the kulembe
Icings south of the Kwanza and that it represented an evolved form
of one of the non-lineage structures of the common type which I have
called cross-cutting institutions.37 The original and primary meaning
of the word connoted an association of males, open to anyone without
regard to lineage membership, in which members of the society
underwent dramatic initiation rituals that simultaneously removed'
them from the protective pale of their natal descent groups and
welded the initiates together as co-warriors in a regiment of supermen
made invulnerable to the weapons of their enemies.
The formidable warrior bands of the kilombo probably emerged
from a combination of Ovimbundu and Cokwe/Lwena institutions
forged when the kinguri’s band, which had become heavily influenced
by Cokwe customs during its passage through Itengo and Mona
Kimbundu, met Umbundu-speaking peoples west of the Kwanza.
The kingurVs followers had brought a powerful but relatively
unstructured cult which, linguistic evidence suggests, they had picked
up somewhere among the Cokwe. The Ovimbundu contributed
structure in the form of an early version of the circumcision camps
now found throughout central and western Angola among the
Mbundu and Cokwe as well as some Ovimbundu.
Seventeenth-century Mbundu oral traditions explicitly recorded
the creation of the kilombo from distinct Cokwe and Ovimbundu
institutions in the story o f ‘Temba Andumba’, a legendary queen who

3S Full discussion o f the later kilombo appears in Chapter VIII.

37 Seventeenth-century Europeans in Angola, heretofore the only sources
on the Imbangala kilombo, have misled all subsequent historians through their
failure to appreciate the significance o f the institution in the eyes o f the
Mbundu. As a result, the military aspects o f the kilombo (for obvious reasons
the side which mattered most to European soldiers and administrators) have
been emphasized to the exclusion o f its social and political implications for
Mbundu kinsmen. The ensuing discussion of the kilombo makes no further
reference to the usual but imprecise definitions of the word as (a) a warcamp
belonging to the so-called ‘Jaga’ (Miller, 1973a), (b) a type o f fugitive slave
settlement found in both Angola and Brazil (cf. Kent, 1965, p. 162), and (c)
any temporary camp built by nineteenth-century trading caravans in Angola.
 O R IG IN S O F T H E K I L O M B O 163
was said to have founded a large kingdom on the basis of a cult,
called an ‘execrable sect’ by offended missionaries, which contemp­
orary Europeans associated with the people whom they erroneously
knew as ‘Jaga’. The written version of this tradition, though distorted
in various ways,38 accurately specified that the ‘laws of the “Jagas” ’,
as the kilombo rites were then known, had originated somewhere in
the east and appeared with ‘Zimbo’ (the munjumbo), Ndonje, and the
others somewhere along the upper Kwanza.39 The founder of the
sect, ‘Temba Andumba’, was a daughter of Ndonje. She became a
brave warrior queen and conquered many lands. She at last grew
intoxicated with her military successes, according to the interpreta­
tion of the missionary who wrote down the tradition, and introduced
laws and rituals (yijila, singular ldjila) intended to preserve her status
as the most feared and respected ruler in Angola. ‘Temba Andumba’
proclaimed her new laws in the most terrifying ceremonies the
Mbundu could remember. First, she sent for her own infant daughter,
seized the child, and threw her into a large mortar normally used for
grinding grain. ‘Temba Andumba’ then picked up a large pounding
stick and mercilessly reduced the baby to a shapeless mass of flesh and
blood. She added certain roots, herbs, and powders to the human
remains and boiled the entire mixture to obtain an unguent which she
called maji a samba. Smearing the maji a samba on her own body and
on the bodies of her closest associates, she called her people to a
renewed campaign of terror and destruction. After devastating all the
lands within her reach, she commanded her followers to take their
own children, cut them to bits, and to eat the remains as a sign of
their devotion to the yijila laws of her kingdom.
‘Temba Andumba’, the story continued, later fell in love with a
certain Kulembe whose social position made him somewhat inferior
to the warrior queen but who equalled her in both bravery and
cruelty. Kulembe ambitiously desired to claim ‘Temba Andumba’s’
prestige for himself and so resolved to kill her and to take her king­
dom. He concealed his evil intentions for many years while ‘Temba
Andumba’ expanded her kingdom, finally gained her confidence
through flattery and feigned affection, and eventually married
her. Not long after the wedding he executed the plan he had long
hidden, inviting his wife to a ceremonial dinner customary among

38 Cf. the distortions in the story o f ‘Zimbo’, to which this narrative was
joined; Miller (forthcoming (a)).
39 Cavazzi (1965), i. 177-9. Although Cavazzi did not use the term in this
context, his ‘execrable sect’ and ‘savage laws’ clearly referred to it.
164 T H E IM B A N G A L A K I L O M B O
their people and then murdering her by putting poison into her
drink.
ICulembe successfully hid his complicity in ‘Temba Andumba’s’
death from the people of the kingdom and induced them to accept
him as her rightful successor and ruler over the adherents of the
yijila. To consolidate his authority, he sacrificed an untold number of
people in memory of the dead ‘Temba Andumba’ and performed
other deeds hypocritically intended to indicate his great piety and
grief. He then began a dramatic military campaign in conjunction
with several other brave generals (named as ‘Calanda, Caete, Cassa,
Cabuco, Caoimba, and many others’), and together they soon made
themselves masters of an area even larger than that conquered by the
famous ‘Temba Andumba’.
The significant aspects of these narrative episodes are the state­
ments that certain titles from the east joined with the lculembe to form
a new and very powerful military organization (the lcilombo). The
analysis first focuses on the position of the lcilombo chief’s first wife,
known properly as the tembanza but personified in this tradition as
‘Temba Andumba’. Titles cognate to that of the tembanza appear
throughout the Colcwe regions to the south-east, sometimes in
association with rituals identical to those called maji a samba in the
Mbundu narrative, and suggest that this part of the institutions which
formed the mature lcilombo came originally from that direction.
The evidence that the word tembanza, meaning the title of the
lcilombo leader’s principal wife, came from somewhere in south­
eastern Angola comes from an incomplete but suggestive distribution
of similar positions with the same name. The husbands of certain
Cokwe chiefs’ sisters received the title sambaza, an evident cognate of
tembanza,40 Modern Imbangala have never heard the term tembanza
in connection with the chief wife of later Icings of ICasanje, who also
were lcilombo chiefs, but they none the less feel that the term probably
came from the Colcwe.41 The senior niece of the Icing of a group of
people living now in northern Botswana (the Mbukushu42) took the
'l0 McCulloch (1951), p. 48. The prefixes sa- and ta- (te-) are equivalent in
Bantu languages; both mean ‘father’ o f the following word or name. The
absence o f nasalization in the recorded form o f the Cokwe word could result
either from an error in transcription or from a sound shift. The Cokwe sambaza
was the father o f the ldng’s heir (his sister’s son), which implies that the heir
himself was the mbanza or mbaza.
41Testimony o f Sousa Calunga, 11 Sept. 1969.
42The Mbukushu live on the Okavango river at approximately 18° S.
Guthrie does not mention Simbulcushu (the language of the Mbukushu)
specifically but Murdock groups them with the Ha and Tonga whose languages
 O R IG IN S O F T H E K I L O M B O 165
title of mambanje, also a cognate of tembanza.43 The mambanje
performed a crucial function in maintaining the well-being of the
Mbukushu kingdom since she cohabited for ritual purposes with her
uncle, the king, and shared with him the secrets of rain-making. The
preparation of these rain-maldng medicines required the murder of
some of the infant children born of the incestuous unions of Icing and
mambanje. The Mbukushu electors chose the successors to their
kings from the surviving children of this group.44 The striking resem­
blance of the Mbukushu rain-maldng ceremonies to the ritual child­
killing necessary in the preparation of the maji a samba reinforces the
identification of the tembanza of the kilombo with the mambanje title
of the Mbukushu.
The migrating lineageless band of the ktnguri probably adopted the
rituals of the maji a samba, as well as the central feminine title of the
tembanza, as they abandoned their lineage rituals among the Colcwe.
The lack of lineages deprived the Lunda of an effective means of
instilling a strong sense of membership in their band other than
through the terror of rule by the kinguri.45 The rituals of the maji a
samba, on the other hand, conferred a magical invulnerability on the
otherwise supernaturally exposed members of the band, thus sub­
stituting for the absent descent groups. The attendant rituals ex­
plained the significance of the structural change which had occurred
in two ways. In a figurative sense, the ruler’s preparation of the
maji a samba through ritual murder of her/his ‘child’ was a common
symbol for the enormity of a ruler’s power over his people. ‘Children’
in the narrative represent the subjects of political chiefs, in contrast
to their relatives who are always depicted as ‘nephews and nieces’.
The ceremony of ‘child’-killing symbolized the ruler’s absolute power
over his subjects just as the image of the kinguri murdering slaves
every time he rose or sat showed the awe in which his people held
him. In a more literal sense, however, child-killing, when practiced by

43The prefix ma- (or na- in some languages) connotes ‘mother’ of the
following name. The Mbukushu mambanje was in fact the mother o f the king’s
heir, who would be called mbanza there as well as among the Cokwe (the actual
reflex being mbanje).
44Dr. Thomas Tlou of the University of Botswana, Swaziland, and Lesotho
kindly provided information on the mambanje and the Mbukushu from his
unpublished research.
45 Compare the theory of the uses o f terror suggested by E. V. Walter (1969).
fall in Guthrie’s Zone M. See Murdock (1959), p. 365, and Guthrie (1967-72),
iii. 15. On the Mbukushu, see Thomas J. Larson (1971) and publications cited
therein.
166 T H E IM B A N G A L A K I L O M B O
an entire population, became a means of abolishing lineages, since
the murder (or denial of the social significance of a physical birth) of
children had the same structural effect on descent groups as the
kingurV s prohibition of their birth; both eliminated the ties of kinship.
The story of ‘Temba Andumba’s’ maß a samba refers to the develop­
ments noted earlier on the basis of independent evidence on the
history of the kingurVs band among the Cokwe.
The use of the maß a samba may have begun as a ritual practised
only by the leaders of the band, as it was later among the Mbukushu,
but its structural significance changed completely when it was
extended from the king (symbolized in the narrative by ‘Temba
Andumba’) to all the subjects in the kingdom. It then ceased to serve
as a means of setting the king apart from his people and instead
marked the effective end of lineages. This change, which was noted
distinctly in the seventeenth-century Mbundu narrative, illustrates
the way in which an extension of an old notion could produce a
significant political innovation.
Similar practices, evidently derived from the maß a samba of
‘Temba Andumba’, appeared much later at the courts of kings known
to have descended from titles brought across the Kwango with the
kinguri. Nineteenth-century Bihe kings, heirs of the munjumbo, had
their magical specialists take the foetus from the womb of a pregnant
woman to make an ointment which, they believed, conferred invuln­
erability.45 The seventeenth-century Imbangala north of the Kwanza,
also descended from the kingurVs original band, had exactly the same
practice.4647 The association between Lunda titles and the maß a samba
provides final confirmation that the origins of the ‘execrable sect’ lay
east of the Kwango in so far as they were associated with the kilombo
chief’s first wife.
The seventeenth-century narrative’s detailed description of the
courtship and marriage of Kulembe and ‘Temba Andumba’ describes
the union of the idea of the maß a samba with an entirely different
Ovimbundu institution apparently connected with the kulembe state
on the Benguela plateau. As in the general model of the Mbundu
perpetual genealogies, the ‘marriage’ of a female (usually signifying a
group of lineages, but here a lineageless band) to a male stands for the
subjugation of people to a political title. In this case, the ‘marriage’

46 Magyar (1859), p. 316. Chiefs’ wives elsewhere south of the Kwanza per­
formed similar rituals involving human sacrifices, but their titles distinguished
them from the tembanza. See, for example, Childs (1949), p. 20.
47 Cavazzi (1965), i. 126.
 F O R M A T IO N O F T H E IM B A N G A L A 167
represents the union of the adherents of the maji a samba cult with the
kulumbe’s warrior initiation society, the kilombo.
Linguistic evidence reveals something about the nature of the
ancient kilombo even without clear information on the social and
political institutions of the Benguela plateau at the time of the
kulembe state. Several modern Umbundu terms, all related to what
appears to be an archaic Umbundu root referring to circumcision or
to blood, occur in connection with modern Ovimbundu circumcision
rites. The most direct evidence comes from the Umbundu-speaking
M undombe people near Benguela who still called their circumcision
camp a kilombo in the nineteenth century.48 Ocilombo in standard
modern Umbundu refers to the flow of blood from a newly-circum­
cised penis; a related term, tilombo, designates a medicine prepared
from the blood and prepuces of initiates in the circumcision camp for
use in certain (unspecified) rites.49 The root -lombo which forms the
basis for all these words identifies the term kilombo as uniquely
Ovimbundu since it contrasts with the Mbundu and Cokwe word for
circumcision ceremonies, which is mukanda.50It is also distinct from a
similar-appearing root, -lumba, which means ‘wall’ in all of the major
western Angolan languages, Umbundu, Kimbundu, and Kikongo.51
Despite the possibility that the walls (lumbu) surrounding the kilombo
might have given the site its name because of their military import­
ance in keeping out enemies or because of their symbolic significance
in isolating the sanctified inner kilombo from the profane outer world,
linguistic analysis shows that lumbo and -lombo were two different
words. The word kilombo therefore indicated the origin of the
kulembe'& warrior society as an Ovimbundu circumcision camp
rather than referring to the ‘walled’ aspect of the place.

Formation o f the Imbangala

The critical development in the formation of the mature kilombo
took place when the kinguri's remaining followers, the secondary
Lunda titles known as malcota, rejected the leadership of the oppress­
ive central Lunda position and adopted the kulembe's warrior18

18 Magyar (1859), p. 23.

15 Alves (1951), i. 547.
50 Antonio da Silva Maia (n.d. (a)), p. 141; Assis Jr. (n.d.), p. 268; Cordeiro
da Matta (1893), p. 87; J. Van Wing and C. Penders (1928), p. 136.
51 Seventeenth-century Kikongo made a clear distinction between the high
vowel 0 0 found in the word for wall in all three languages and the lower vowel
(o) that occurred only in kilombo. In the Kikongo o f that time, the word for
wall was lumbu ; Van Wing and Penders (1928), pp. 136-7.
168 TH E IM B A N G A L A K I L O M B O
initiation society as the basis of a new political organization, adding
a number of wwga-positions of demonstrably Umbundu origin. The
union of the theoretically centralized kilombo with the multitude of
Lunda perpetual positions enabled the initiation society to fragment
and to spread rapidly throughout the regions south of the Kwanza as
many separate warrior bands, now called Imbangala,sz formed under
holders of new subordinate Lunda-type titles. The inherent capacity
of the perpetual positions to spawn new named titles provided a
legitimate pedigree for any warleader who could gather enough
followers to break free of existing political and lineage authorities in
the Benguela highlands. An ambitious man could adopt the organiza­
tion of the kilombo, claim charms and a title descended from another
kilombo-chief, and make his name as an Imbangala king. Even the
native Ovimbundu positions, such as that of the Iculembe, whose
reliance on mavunga showed that they had not originally exploited
the ability to nominate ‘son’ positions, adopted the Lunda technique
and awarded subordinate positions which appear in mid-seventeenth-
century documents. The fragmentation which attended the introduc­
tion of Lunda titles south of the Kwanza represented the obverse of
the centralization which the Ovimbundu mavunga had produced
among the Mbundu.
The unstable history of the kinguri's band before they reached the
Kwanza explains why the makota sought new forms of political
organization which would free them of their dependence, in terms of
Lunda political structure, on the kinguri. Defections by some title-
holders as far east as the Cokwe, and the recent departures of the
mwa ndonje and the munjumbo confirmed the seriousness of the
dissensions which had shattered the band. These were mentioned
explicitly in the Bihe tradition which described the formation of the
empacasseiro society and had been alluded to in metaphoric terms in
the story of how ICulembe clandestinely opposed the ‘execrable sect’
of ‘Temba Andumba’. The kinguri’s followers no longer accorded
their leader the complete loyalty demanded by the supernatural forces
behind his title, and—since the performance of Mbundu kings
depended on the obedience of their subjects—his charms had ceased
to perform as expected. The resulting insecurity drove the kinguri to
seek new methods of controlling his people; this explains not only his
request for the ngoma ya mukamba but also his award of new titles,52
52 The term Imbangala comes from an Umbundu root, -vangala, meaning
to be brave and/or to wander widely through the countryside; Joseph C.
Miller (forthcoming (d)).
 F O R M A T IO N O F T H E I M B A N G A L A 169
such as the kunga and others, to Songo lineages. All these measures
represented attempts to find new sources of support among the
lineages of Manyungo wa Mbelenge and Kavunje lea Tembo.
If the Lunda makota found their association with the kulembe
advantageous because it helped them to free themselves from the
burdensome kinguri, the kulembe may have found an alliance with
the numerous and martial Lunda an attractive means of resisting
either Libolo expansion or the southward advance of the ngola a
kiluanje. In either case, wars followed in which the kulembe and the
makota on one side fought against the kinguri and his remaining
Songo allies on the other. Seventeenth-century Mbundu traditions
referred to these wars in their recollection of the great conquests
which followed the kulembe’s adoption of the maji a samba in con­
junction with the ‘other brave generals’ who were, with a single
partial exception, all Lunda titled positions. ‘Calanda’ was Kalanda
ka Imbe (‘Caoimba’), ‘Caete’ was ICahete, and ‘Cabuco’ was Kabulcu
ka Ndonga, a title subordinate to the ndonga which had come from
Lunda with the kinguri-, only ‘Cassa’ (the kaza) had different ori­
gins.53
The united forces of the lcilombo then drove the kinguri north and
west, as the Bihe tradition later recalled, away from the centres of the
kulembe’s strength and toward Ndongo. The seventeenth-century
traditions confirmed that the last kinguri had died in Ndongo,54 but
not fighting against the Portuguese or the ngola a kiluanje as most
historians have assumed. He, in fact ‘died’ at the hands of the makota
who had accompanied him from Lunda, and his ‘death’ meant the
abolition of his title rather than the demise of an individual incum­
bent. These developments culminated on the island of Mbola na
Kasashe in the upper Kwanza where the Libolo kings had once
stationed one of their v«/7ga-chiefs as guardian of the eastern frontiers
of their kingdom. By this time,55 however, the ngola a kiluanje had
subsequently conquered the area and made it part of ‘Ndongo’
according to the seventeenth-century usage of the term.
The Lunda makota and the kulembe owed their victory over the
previously unconquerable powers of the kinguri to the explosive
combination of a mobile band without lineages and the assimilative
and structuring potential of the lcilombo initiation society. Lunda
53 Cavazzi (1965), i. 189. The kaza was originally a Libolo title related to the
hango but turned up shortly afterwards as a close associate of the Lunda
titles; see chapter VII.
54Cavazzi (1965), i. 190.
55c. 1550s-1560s?; Miller (1972a), pp. 560-3.
170 T H E IM B A N G A L A K 1 L O M B O
abolition, of lineages had given them the ability to incorporate large
numbers of people but this solution had simultaneously eliminated
the structural definition provided by descent groups. It thus failed to
integrate its members into an effectively unified group, as evidenced
by the constant tendencies to fission present through the history of
the band. The hostility of Colcwe and Mbundu lineages to the line­
ageless institutions of the maji a samba cult also weakened its
effectiveness as a means of forming a large but still cohesive group.
The ldlombo, on the other hand, possessed well-defined initiation
procedures which, together with the centralizing capabilities of the
Ovimbundu mavunga, compensated for the organizational weaknesses
of the Lunda band. These qualities, plus the strong magic associated
with the Lunda titles of the makota, created the large, unified, and
disciplined bands of Imbangala warriors who overran the Mbundu
later in the seventeenth century.
The Iciilembe, ironically, seems to have become one of the first
major victims of the combined Lunda/Ovimbundu ldlombo. The
seventeenth-century traditions made it clear that the resulting
expansion took place primarily under the leadership of the Lunda.
The kulembe’s once-unified kingdom disintegrated into many small
warring chiefdoms led by ldlombo chiefs, some of which later emerged
as Ovimbundu kingdoms of the eighteenth and nineteenth centuries.
The chronic instability of the Ovimbundu region not long after these
events, approximately the middle of the seventeenth century, suggests
that a power vacuum had followed the decline of some formerly
centralized state. Warlike chiefs, reported in many documents to
have been fighting one another near the headwaters of the Kuvo and
Longa rivers during the 1640s, for example, would thus have been
former subordinates of the Iculembe, freed from their overlord to work
out a new political balance which later coalesced as the Ovimbundu
kingdoms of that region. The title of the Iculembe survived these
changes, but only as one undistinguished position among many more
or less equal rulers in the area its holders had once dominated.55
Because the ldlombo structure included no perpetual named posi­
tions of the Lunda and Mbundu type, modem Mbundu traditions
totally neglect the wars and upheavals visible in documents. They
concentrate instead.on the formal aspects of struggles which altered
the relationships between the kinguri title and its makota. These
traditions tell the story in metaphorical terms reminiscent of early
56 For example, see Cadornega (1940-2), iii. 249; his spelling o f the name
was ‘Lulenbe’.
 F O R M A T IO N O F T H E I M G A N G A L A 171
Lunda history at the time of the ycila mwalcii, in which the makota
‘killed’ the position of the kinguri through application of special
magical techniques. Picking up the theme of Kinguri’s legendary
ferocity, the narratives attribute the mutiny of the makota to their
reaction against their king’s bloodthirsty demands. As Kinguri’s
incessant oppression bore more and more heavily on his own people,
the traditions recall, the makota recognized that he posed a threat to
them personally and began to devise ways to avoid the growing
danger. They finally resolved to take a desperate step: they would
attempt to slay Kinguri and take over leadership of the band for
themselves. They concentrated on searching for a means to disarm
the supernatural forces which protected their king from ordinary
sorts of danger. A simple murder of the incumbent would do no good,
as this would not affect the spirits to whom all kinguri owed their
power, and so only magical methods could abolish the problem, and
the position, entirely. In particular, they felt they must not allow
Kinguri to discover the plot or the aroused and angered forces behind
the title would certainly avenge themselves before the makota could
execute their plan.57
The makota finally chose a method which operated primarily on
the level of the symbolism of their kings’ title. The kinguri and its
attendant spirits were seen as carnivorous animals of the forest. The
kinguri’s thirst for human blood, for example, reminded them of a
lion’s roar in the night.58 Their awe of their leader resembled the fear
of men who confront wild beasts in the wilderness.59 Since the Lunda
had always hunted with pits and traps, the makota constructed a
symbolic trap of this kind, a circular enclosure of heavy stakes
situated on the island of Mbola na ICasashe where they had camped
at that time.60 The enclosure had only a single entrance (unlike all
57 Testimony of Sousa Calunga, 21 July 1969.
58 Various testimonies o f Sousa Calunga; also Mwanya a Shiba, 15 June
1969.
5'J Dias de Carvalho (1890a), pp. 60-1.
60 Full name: Mbola na ICasashe lea Masongo a Ndembi; testimony o f Sousa
Calunga, 2 Oct. 1969. Rodrigues Neves (1854), p. 99, gave the area where the
kinguri died as ‘Sunge a Mboluma’. In the nineteenth century, Sunje a Mboluma
referred to the general region on the east side o f the Kwanza near Mbola na
Kasashe; testimony o f Sousa Calunga, 1 Oct. 1969; also map o f Capello and
Ivens (1882). For geographical details on the island, Eugenio Torre do Valle
and José Velleso de Castro (1913), esp. pp. 35-41, 98-9. It was one o f the
very few toponyms known to the’earliest Europeans in an otherwise unfamiliar
region; ‘Carta do Império do Monamotapa’ located in the A .H .U ., probably a
copy of the map of João Teixeira Albernás II, 1665 (Avelino Teixeira da Mota
(1964), pp. 32-4). For the nineteenth century, Vincente José Duarte (1859-61).
172 T H E IM B A N G A L A K I L O M B O
later royal compounds of the kinguri kings in ICasanje which always
featured a second entrance as a potential means of escape for a
beleaguered ruler). On the pretext that lions roaming in the vicinity
endangered them all, the makota feigned great concern for Kinguri’s
safety and persuaded him to enter the enclosure where, they argued,
the palisade of thick stakes would protect him from danger.
The makota evidently managed to conceal the potential for treach­
ery inherent in the situation, for Kinguri failed to see the ironic
reference to himself in the purported dangers from lions.51 Trusting
his advisers and kinsmen, Kinguri did not realize that the palisade of
heavy stakes was intended to imprison the one they claimed to pro­
tect, rather than to exclude danger. Kinguri entered the enclosure and
waited while the makota outside looked for a chance to seal the single
entrance and leave their king inside to starve to death. Because
Kinguri’s supernatural powers invariably warned him of dangers
before they arose, the makota had to wait until he fell asleep and then
quickly barred the only exit from the prison.52 They lingered in the
vicinity until Kinguri died from hunger, and then left.
This description of Kinguri’s death tells relatively little about the
historical wars between Songo adherents of the kinguri and the
Imbangala warriors of the kilombo. But it does reveal a great deal
about the significance of these events for later Imbangala history.
Several aspects of the narrative express Imbangala beliefs about
proper relationships between subjects and their rulers. Supernatural
vision, for example, has remained an important ingredient of a chief’s
powers down to the present time. The tradition takes this into account
by having the kinguri fall asleep (in the metaphoric sense of negation
of his magical powers of perception) in order to explain how so
powerful and omniscient a ruler could fall victim to the makota’s
transparent plot. Starvation was an ideologically suitable means of
killing a permanent title because it did not involve bloodshed: only
human beings, according to the Mbundu theories of kingship, spill
blood when they die and the spirits of a title, the real targets of this
attack, do not do so.
Starvation symbolized not just deprivation of food for an incum­
bent but also abandonment of a title by its followers, since the
imagery of food and feeding stood as a metaphor for one of the most
basic Mbundu beliefs about their relationship to their chiefs. The612
61 Nguri means lion in some dialects o f Umbundu,
62 Testimony o f Mwanya a Shiba, 15 June 1959, for the best statement o f
the motivations behind the actions o f the makota.
 F O R M A T IO N O F T H E IM B A N G A L A 173
archetypal ‘marriages’ between male political principles and female
lineages in Mbundu genealogies establish the responsibilities of
rulers and ruled through reference to the analogous duties of a wife to
her husband. Wives produced and prepared food for their husbands
to eat, and subjects had the same obligation to their rulers. They fed
their chiefs in a literal sense by supplying the populations of their
capitals with foodstuffs, contributing labour to work the chief’s
fields, and so on. They also fed their chiefs symbolically, since
political titles (as opposed to incumbents) derived their real susten­
ance from the loyalty of their people. The tradition of Kinguri’s
‘death’ contains two levels of meaning: denial of food may have
caused the death of the title-holder from starvation, to be sure, but,
more important, abandonment by the people also abolished the title
itself.
The most common version of this episode (paraphrased above)
makes the essential point of abandonment through the image of a
straightforward denial of food to the kinguri, but other variants refer
more subtly to the same fact either through reiterating the theme of
starvation in other terms, or by incorporating the conceptual oppos­
ite, surfeit. According to some, the conspirators built a splendid new
palace for the kinguri and led him into it with great ceremony. The
palace, like the palisade in the first variant of the tradition, had only a
single entrance. When the kinguri had gone inside, the makota
blocked the door and smothered their king by pouring a great
quantity of manioc flour through a hole in the roof.63 Ironically in
this case, not denial of food and loyalty but fulfillment of the kinguri’s
excessive demands brought surfeit and death by asphyxiation.
Some Imbangala histories use a different but equivalent image
based on a cliche which usually occurs elsewhere in the corpus of
Mbundu narrative traditions.64 They explain, just as in the more
common version, how the makota imprisoned the kinguri, but they
complete the narrative by arguing that the makota continued to
provide food for the kinguri but gave him only rotten seeds unfit to
eat.65 Given the appearance of loyalty without its substance, the
kinguri soon expired.
63 Schütt (1881), pp. 79-80, 100.
04 See the departure o f Kulashingo from Portuguese Angola, below, chapter
VII.
65 Testimony of Mwanya a Shiba, 14 June 1969. Imbangala historians feel
free to transpose such cliches under the Mbundu rules o f historical composition
because the alternative images act as equivalent metaphors for the same
historical events.
174 T H E IM B A N G A L A K I L O M B O
All variants of this episode agree in emphasizing that the makota
had to employ supernatural and deceptive means to kill the kinguri.
The Imbangala sometimes use another image to describe the kinguri's
death because it emphasizes the need for deception. The makota,
according to this variant, dug a deep pit which they carefully dis­
guised with a covering of leaves and grass to give it the appearance of
solid ground. They finished their preparation of the trap by adding
the ceremonial mat which the kinguri occupied on formal occasions.
They then invited their king to receive their homage while seated on
the mat placed over the pit. The kinguri sat down and fell into the
hole where the makota buried him on the spot.66 The emphasis on
deception, which recurs in almost every variant, also suggests that the
makota assassinated the kinguri by means of certain kilombo rituals
known as kiluvia. In the kiluvia, Imbangala honoured and deceived
their prisoners of war up to the moment of their death. Like these
other captives, the kinguri never suspected his fate until it was too
late. The apparent reference to the kiluvia provides confirmation in
the oral narratives that the makota had embraced the ceremonies of
the kilombo as they rebelled against the kinguri.
Finally, the makota also broke the kinguri's alliance with the Songo
lineages of Manyungo wa Mbelenge. As one narrative episode
explains the event, the makota shut Manyungo wa Mbelenge into the
prison with the kinguri. She died before the kinguri and he, anguished
by starvation, ate part of her body before he himself died of hunger.
The tradition specifies that the kinguri ate only the upper half of the
corpse.68This episode echoes the imagery of the Bihe traditions which
told how the kinguri's cruel demands threatened to consume all his
followers.68 In terms of politics in the later kingdom of Kasanje, the
narrative showed that the kinguri had left no related titles among the
Songo and thus legitimized the authority of later kings who took the
kinguri title without the usual qualifications of kinship to its ancient
holders. Although the narrative indirectly confirms Songo claims that
titles derived from the kinguri once existed, modem Imbangala
66Oliveira Ferreira Diniz (1918), p. 93. The same cliche occurs very widely
(e.g. Balandier (1968), pp. 38, 271, n. 14, for the Kongo) and often crops up in
another context in the Mbundu traditions. This variant echoes the beastly
qualities of the kinguri, since the pit could also have served to capture large
and dangerous wild game.
67Testimonies of Kiluanje kya Ngonga; Domingos Vaz. The exact sym­
bolism o f the last detail remains obscure except to underscore the savage and
bestial nature o f the title. It may refer to the impropriety of the male kinguri
coming into contact with the genitals o f the female Manyungo wa Mbelenge.
68 Magyar (1859), pp. 266ff.
 F O R M A T IO N O F T H E IM B A N G A L A 175
historians always emphasize that, from their point of view, none of
these had any claim to succeed the Icinguri.69
The makota led their band of Imbangala south-westward some­
where on the southern side of the Kwanza river after their defeat and
abolition of the Icinguri. They no doubt found it expedient to leave the
area of their crime for fear of the angered spirit of their former ruler
and probably found no further welcome among the Songo lineages
who had allied with the Icinguri. Wars against the ngola a lciluanje
during the 1560s70 may have determined the direction in which they
moved, since their course led them directly away from the powerful
Mbundu king in the north. In their wake, the makota left a new set of
Lunda political titles in Songo centered around the munjumbo, the
nclonje, and the kunga. They also left Libolo much smaller and weaker
than they had found it, reduced from a large kingdom to a small state
occupying only the westernmost province of its former empire. They
had forced the state of ICulembe to disintegrate and had claimed
leadership of the kilombo for themselves, leaving the bulk of the
kulembe’s former lands to the munjumbo. The arrival of the Icinguri
and the formation of the Imbangala under the leadership of the
makota had caused a major revolution in the political structure of the
peoples who lived on the upper Kwanza.
The Imbangala under the leadership of the Lunda makota seem to
have travelled towards the coast south of the site of the later Portu­
guese town of Benguela. From there, they moved northward along
the coast through the 1580s and 1590s, arriving in the vicinity of the
Kuvo river in or shortly before 1601.71 There they came into close
contact with Europeans for the first time and began yet another phase
of the history of the kilombo.

1,9Pires (1952), p. 2; testimony o f Apolo de Matos, 18 June 1969.

70 Miller (1972a), pp. 560-3.
71Ibid., pp. 563-5; this group o f Imbangala probably did not drift as far
south as suggested there.
 C H A P T E R V II

T he Imbangala and th e Portuguese

T he I mb a n g a la led by the former m a k o ta of the k in g u ri made their

way north along the coast towards the Kwanza during the same years
that another band of aliens, representatives of the Portuguese crown
in Europe, were approaching the Mbundu from the sea. The simul­
taneous arrival of these two groups of outsiders at the Kwanza set the
stage for an Imbangala-Portuguese alliance in the first half of the
seventeenth century which revolutionized Mbundu political institu­
tions and geography. Together they reduced the n gola a kilu a n je from
monarchs of a robust and expanding kingdom in 1600 to nearly
powerless puppet rulers after 1630 and substituted a completely new
set of European and African states founded on the export of slaves
from Africa to the Americas. A small Portuguese state of Angola1
replaced Kongo title-holders on the coastal plain north of the
Kwanza2 and the n gola a k ilu a n je in the old central provinces of
Ndongo and Lenge, while Lunda title-holders at the heads of
Imbangala k ilo m b o bands imposed themselves near Angola where the
hango of Libolo, the Pende m alu n ga- Icings, and various subordinates
of the n gola a kilu an je had ruled before.
The written documentation available for this phase of Mbundu
political development permits more detailed reconstruction of
Imbangala and Portuguese state-formation than of their predeces­
sors’.3 Certain structural characteristics of the Imbangala k ilo m b o
that predisposed holders of the Lunda m a k o ta titles to join with the
Portuguese in the pursuit of slaves seem implicit in the evidence. But
these were most compelling for the Imbangala who settled as aliens
among the Mbundu north of the Kwanza; the different social envir­
onment south of the river allowed the k ilo m b o leaders who stayed

11 refer to the Portuguese-conquered territories in these terms in an attempt

to maintain an Mbundu perspective on events. As a result, Portuguese legal
distinctions between donataria, reiuo, conquista, and other forms o f European
rule have very little significance.
2 For the details of this part of the Portuguese conquest, see Miller (1972b).
3 This approach omits the relatively well-known histories of the expansion
o f Portuguese military and political control in Angola and the defeat o f the
ngola a kiluanje ; for these events, readers are referred to Birmingham (1966).
 F IR S T C O N T A C T S —E S T A B L IS H IN G PA T T E R N 177
among the non-Mbundu there to remain more independent in their
dealings with the Portuguese. The political institutions of the northern
Imbangala evolved in the direction of Mbundu cultural norms as the
kilombo turned into a series of settled states between 1610 and
c. 1650.45
The primary structural feature of the kilombo which influenced the
course of Imbangala contacts with the Portuguese was the character­
istic instability of relations among the Lunda title-holders, first noted
long before the kingurVs band had crossed the Kwango. It affected
their relations with the Europeans in two ways. First, rebellious
holders of subordinate titles had repeatedly sought outside sources of
support as they had broken away from the central position in the
band, as the mimjumbo had adopted vunga titles based on the mwela
magical knife and as the makota had originally embraced the
kilombo.s After the Imbangala made contact with the Europeans, the
same search for external sources of legitimacy and material aid often
brought dissatisfied seventeenth-century title-holders into alliance
with Portuguese governors, as ambitious Lunda sought to maintain
or expand, their authority over their own people. Second, the tendency
of subordinate Lunda title-holders to abandon the parent kilombo
meant that Portuguese-Imbangala contacts would result in the
establishment of multiple Imbangala states rather than a single
centralized kingdom. Portuguese governors, then facing heavy odds
as they sought to consolidate a very tenuous control between the
Bengo and Kwanza rivers, eagerly exploited the fissiparous tenden­
cies of the kilombo and ultimately helped to create the ring of
Imbangala client kings who surrounded the area of Portuguese
control by 1650.

First Contacts—Establishing the Pattern

The crew of a Portuguese trading vessel, which encountered the
Imbangala under Kalanda lea Imbe6 camped on the south bank of the
4 The parallel analysis for the Imbangala south of the Kwanza, while tech­
nically falling outside the scope o f this study, awaits the collection o f much-
needed new data.
5 Cf. the role of outside Luba authority symbols in the early history o f the
Lunda in Katanga.
6 A member o f this crew, Andrew Battell, later told the story o f his experi­
ences in Angola to the British humanist Samuel Purchas. Purchas published
Battell’s account, interpolating information gleaned from other sources about
the coast o f Africa. Although it is difficult to distinguish at times between
Battell’s observations and Purchas’s additions, there is no doubt as to the
identify o f Battell’s hosts with the people of the Lunda makota. They called
178 T H E IM B A N G A L A AND THE PORTUGUESE
Kuvo in 1601, developed a commercial partnership based on slaving
which became the prototype for all later co-operation between the
Imbangala and the Europeans. These sailors had gone to the area of
the Kuvo as participants in an extension of Portuguese slave-trading
activities from the Kongo regions in the north, where they had been
active since the early sixteenth century,7 to the Luanda bay area just
north of the Kwanza and to the southern coasts. By the turn of the
seventeenth century, the European presence near the Kwanza was
divided into two overlapping but distinct spheres of activity: govern­
ment-appointed officials claimed tenuous control over a military base
7See Vansina (1966a), pp. 45-64, or Birmingham (1966), pp. 21-41.
themselves ‘Imbangola’ (sic) (Ravenstein (1901), p. 84); the name o f the ruler
of the band was ‘Calando’ or ‘Calandola’ (pp. 31, 33, 85-6), obviously Kal-
anda lea Imbe, the third Lunda kota to rule without completing the required
initiation ceremonies. (Battell elsewhere, p. 28, called him Imbe ya Kalandula,
reversing the first and second elements o f the name.) Kalanda lea Imbe claimed
to have succeeded a great chief named ‘Elembe’ (p. 85), almost certainly the
Kulembe o f the oral traditions. The Imbangala made extensive and prodigious
use o f palm wine in their rituals, cutting down the trees (elaeis Guineensis, or
ndende in Kimbundu, according to Leite de Magalhães (1924), p. 62) to obtain
the fruit, which they ate, and the wine, which they drank. Their requirements
were so enormous that they devastated the palm groves wherever they passed,
pouring the wine over the graves o f their ancestors and using it in attempts to
contact the dead through intoxication, trances, and spirit possession. The
importance o f the palm wine probably derived from the Lunda tubungu chief’s
close association with the drink, already noted in connection with the history
o f the kinguri in Lunda. Their destructive methods o f obtaining palm wine
distinguished them from local peoples, who tapped the standing trees rather
than felling them as did the Imbangala.
Purchas’s identification o f the Imbangala with the ‘Jaga’ and even with the
Mane o f Sierra Leone (pp. 19-20) was erroneous. This passage was probably
added on the basis o f Battell’s statement that the Imbangala had come from
the ‘Serra de Leão’, or Mountains o f the Lion. The true identity o f Battell’s
‘Serra de Leão’ was not Sierra Leone in upper Guinea; the name probably
came from a reference to the kinguri, whose name (nguli or nguri) meant ‘lion’
in the Wambo dialect o f Umbundu (Alves (1959), ii. 959; testimony o f Apolo
de Matos) spoken in the most mountainous regions o f the Ovimbundu high­
lands. (The Cokwe, Kimbundu, and Kilcongo word for lion is koshi or hoje;
the Lunda use ntambo ; Dias de Carvalho (1890b), p. 347; also Chatelain
(1888-9), p. 7, for Kimbundu.) This hypothesis fits other data indicating that
the Imbangala at that time used mainly an Umbundu vocabulary. Purchas
emphasized elsewhere, contradicting his assertions about the Imbangala origin
in Sierra Leone, that only the Portuguese called them ‘Jaga’ and that no
European could have known their origins (pp. 83-4). Other details o f Battell’s
description point to an origin somewhere in the interior and a relatively
recent arrival on this part of the coast, since the Imbangala king had never seen
white men before. The low prices for which the Imbangala sold their captives
also betrayed their unfamiliarity with the coastal region near Luanda since the
slave trade had become established many years earlier there and local residents
undoubtedly knew the prevailing prices.
 F IR S T C O N T A C T S —E S T A B L IS H IN G PA T T E R N 179
on Luanda bay and over a few scattered posts along the banks of the
lower Kwanza river, while numbers of private Portuguese merchants
traded for slaves in widely scattered interior locations north of the
Kwanza and along the coasts on both sides of the river. Actual
territorial control by Portuguese government forces consisted of
little more than the fortified enclaves near Luanda bay and at
Muxima and Massangano on the banks of the Kwanza.8
A seemingly insatiable demand for African labour on the sugar
plantations in São Tomé and, more recently, north-eastern Brazil,
sustained the slave trade in both Kongo and Angola.9 Given the high
demand for slaves, and the efforts of royal slave contractors to tax
this trade at Luanda, it is not surprising that these, and doubtless
other, Portuguese sailors frequented the unregulated coasts near the
Kuvo, nor that the Imbangala camp of Kalanda ka Imbe attracted
their attention. From the ocean, they could see the large numbers of
people in the group, which they estimated at several thousand.10 The
Portuguese went ashore to determine the identity of the group,
probably with the hope of establishing a profitable trade in slaves
with them. A week’s negotiations convinced the Europeans that these
Imbangala would make suitable commercial partners. The Imbangala
seem to have accepted the Europeans’ overtures in exchange for
Portuguese assistance in crossing the Kuvo in order to attack the
people who lived on the north bank of the river.
Both partners undoubtedly embarked on this joint enterprise in
the expectation that the Imbangala would capture slaves, whom the
Portuguese would then buy and ship to the Americas. This was the
basic arrangement which formed the keystone of all later Portuguese-
Imbangala alliances. The warriors of the kilombo would supply
captives in exchange for European trade goods. The Imbangala
evidently executed a successful attack north of the Kuvo in this case,
captured slaves and found the ensuing trade to their advantage, since
they continued to raid and trade near the coast for five months.
The Portuguese derived enough profit from this arrangement to
send a party of fifty men into the interior in search of more slaves and
the Imbangala after their African partners had finally left the coast.
There a local ruler forced the European expedition to leave one of
8 Miller (1971).
5 Curtin (1969), pp. 110-6.
10Battell in Ravenstein (1901), p. 85. Although Battell specified 12,000
Imbangala, the accuracy of the figure is clearly open to question in the light
of its correspondence to the twelve ‘captains’ of the band, each ascribed a
symbolic but inexact 1,000 men.
180 T H E IM B A N G A L A AND THE PORTUGUESE
their men as a hostage in his village as a guarantee of their good
conduct while they sought the Imbangala in his lands. The Portuguese
selected as the hostage the only foreigner among them, an English
sailor named Andrew Battell, and evidently gave him up for lost since
they never returned to ransom him. Battell realized that he could not
expect rescue from his former companions and escaped frr-n
captivity to rejoin the Imbangala whom he had known near the coast.
BattelPs choice of the Imbangala as allies in preference to the local
people, and the hostility which the local chiefs showed to the Portu­
guese, clearly revealed that the Europeans and the Imbangala had
established a basis of co-operation which the inhabitants of the
region regarded as against their own interests.
The wanderings of the Imbangala during the sixteen months which
Battell spent with them in 1601-2 show how completely they domi­
nated the local people. The Imbangala wandered through the lands
between the Kuvo river and the southern bank of the Kwanza,
attacking the most powerful rulers in the region. Of the places
mentioned by Battell, the town of ‘Shillambansa’ had by far the most
significance.11 This was the capital of an important chief said to be an
‘uncle’ of the ngola a Iciluanje. Shila Mbanza’s status as ‘uncle’ meant
that he was a kota of the Ndongo kingdom, one of the guardians of
the royal symbols of authority. Further evidence of his importance
comes from the Portuguese plan, developed shortly afterwards, to
build a fortified post near his town.12 The Imbangala attack on the
shila mbanza showed that they did not hesitate to fight even the most
powerful title-holders including close allies of the ngola a Iciluanje
himself.
Such successful harassment of the local residents produced
numerous captives and quickly attracted the attention of a generation
of Europeans actively in search of slave labour for the burgeoning
plantations of the New World. At first the Imbangala showed little
respect for the elforts of Portuguese royal officials to regulate the
conduct of the slave trade in the interior in favour of the royal
contractors and sold slaves to anyone willing to pay for their prison­
ers. They thus began their participation in the Angolan slave trade as
parts of the illegal commerce run by renegade Europeans eager to
evade government monopolies and taxes. In at least one instance,
11Ravenstein attempted to identify all places mentioned by Battell, but he
often forced his analysis beyond the limits o f the data (pp. 22-7); ‘Shillam­
bansa’ is a praise name, not the name o f the title itself, since shila mbanza in
Umbundu means to ‘praise the noble’.
12Brito, ‘Rellagao breve’, AA, iii, nos. 25-7 (1937), pp. 260-1.
 F IR S T C O N T A C T S—E S T A B L IS H IN G P A T T E R N 181
they joined a group of mutineers from the Portuguese army in pursuit
of captives and booty in the Kisama region just south of the Kwanza.
The local chiefs of that region requested protection from the Portu­
guese governor of the time, João Furtado de Mendonça (1594-1602).
Furtado de Mendonça sent an expedition in search of the Imbangala
which ultimately forced them to withdraw to a defensive site where
they fortified themselves and resisted all Portuguese efforts to dislodge
them.13 The first recorded encounter between Portuguese officials and
the Imbangala thus came as the result of problems they caused for
the legal slave trade under the care of Portuguese governors at
Luanda. This experience gave royal officials first-hand acquaintance
with Imbangala fighting skills and confirmed earlier impressions that
the Imbangala might become valuable suppliers of slaves to anyone
who could win their friendship. Although no records of reactions in
Luanda survive to document the case, such experiences must have
convinced the Angola governors that the success of the ‘official’ slave
trade depended on the co-operation, or at least on the neutrality, of
the Lunda makota and their followers.
The Portuguese encountered other difficulties during this period in
Angola which must have led governors to seek ways of controlling
the Imbangala. Beyond the problems which afflicted all Europeans
who attempted to establish land bases in the deadly disease environ­
ments of tropical Africa,14 as well as the opposition of hostile African
Icings, government representatives in Angola had to contend with a
variety of interlopers who were buying slaves along the coasts both
north and south of Luanda. These ‘smugglers’, as they became known
in official circles, evaded the Portuguese royal taxes on slave trading
just as the Brazilian demand for labour rose to new heights after
1600. At the same time, the obstinate presence of an unfriendly
Kongo state called Kasanze immediately inland from Luanda forced
Portuguese governors to divert their military campaigns to the distant
interior in their search for more captives to ship to America.15 Their
simultaneous need for men and arms near the coast and in the hinter­
land stretched Portuguese forces dangerously thin on both fronts.
13Battell in Ravenstein (1901), pp. 27-8. Cadornega (1940-2), i. 52, cited a
campaign under Governor Furtado de Mendonça that fit Battell’s description
of a battle seen from the other side. The chiefs o f Libolo and Kisama evidently
called the Imbangala ‘Jingas’ or ‘Guindas’ as that time; the former name is
unexplained but had no connection with the later queen Nzinga, contrary to
Dias de Carvalho’s suggestion (1898, p. 30).
14Philip D . Curtin (1968a) summarizes theory and documents the resultant
death rates for non-immune European populations at other times and places.
15 Miller (1972b).
182 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
Bento Banho Cardoso, governor of Angola from 1611 to 1615,
explicitly indicated the strain on his resources when he noted
that supplies and men recently sent from Lisbon had not begun
to meet his requirements. He complained particularly of the lack
of soldiers and horses needed to conduct military forays in the
interior.16
Trapped between military ambitions in the interior and the need
for coastal defence, Cardoso would inevitably have sought the aid of
African auxiliary troops to compensate for the lack of men and
equipment from Europe. The talents and political position of the
Imbangala matched his requirements in almost every respect. They,
like the Portuguese, had come to the Kwanza basin as invaders who
lived by stealing from the local Mbundu farmers. The Imbangala had
a strong incentive to capture slaves for use in connection with rituals
involving human sacrifice and to replace warriors who fell in their
continual wars, and they had developed specialized military tactics
adapted to those purposes. Portuguese interest in these same skills
arose from their desire to procure slaves for export to Brazil. It was a
natural alliance, and one which the lessons learned from earlier
Portuguese-Imbangala contacts must have made obvious to all by
Cardoso’s time, some ten years or more after the first encounters.
Cardoso saw further that use of the Imbangala as mercenaries in
wars in the interior would free the slender Portuguese garrison for
coastal operations against Kasanze and the illegal slave traders.
A formal alliance between the Imbangala and the Portuguese
became effective around 1612.17 It almost immediately solved
Cardoso’s problems of insufficient means to attack the African states
of the interior, as strident objections from the Kongo king Alvaro II
made clear. He complained that the Imbangala (or ‘Jaga’ as all
sources referred to them) were ‘eating’ many of his subjects, thus
making them the first recorded victims of the new Afro-European
combination.18 More successes followed soon after; by 1615, many
16 Consulta do Concelho Ultramarino sobre as coisas que faltam no governo
de Angola para sua governação, c. 1617; A .H .U ., Angola, cx. 1.
17 Miller (1972a), pp. 567-8.
18 Letter from el-Rei do Congo Alvaro II to the Pope, 27 Feb. 1613; Arch.
Vat. Calfalonieri (Rome), t. 34, fols. 301ff.; mentioned in L. Jadin and J.
Cuvelier (1954), pp. 329-335. The ‘Jaga’ were also referred to at pp. 338, 344,
351, and 423, in complaints that continued with some regularity through the
decade. The original reference may concern the 1612 campaign against the
Dembos; see Delgado (1948-55), ii. 34. Cardoso’s first confirmation o f his use
o f the Imbangala (always under the name ‘Jaga’) came in 1615; see Auto de
Banho Cardoso, 17 Aug. 1615; A .H .U ., Angola, cx. 1, doc. 46.
 F IR S T C O N T A C T S —E S T A B L IS H IN G P A T T E R N 183
M bundu rulers south of the Bengo, who had previously resisted
Portuguese authority, were surrendering as a result of campaigns
Cardoso had waged with the aid of Imbangala mercenaries.19 The
new Mbundu vassals included the most powerful title-holders
occupying both banks of the Kwanza: Kafushi, owner of the salt
pans of Ndemba in Kisama, the kasanje of ICakulu lea Hango who
lived near Mushima, Kambambe who guarded access to the fabled
‘Mountains of Silver’ just aN /e the fall line of the Kwanza, and the
ngola a kiluanje himself.20 The Mbundu feared the Imbangala far
more than they feared the European armies which before that time
had not dared venture far beyond the Kwanza river and a very few
fortified positions along its banks.21 From the Mbundu point of view,
the Imbangala had helped the Portuguese to establish a new state in
the western portions of the ngola a kiluanje kingdom. From the
Portuguese point of view, Imbangala participation as mercenaries in
their ‘conquest of Angola’ had providentially converted a desperate
situation into a period of successful slave raiding and territorial
expansion.
The strong pressures which motivated the Portuguese to ally them­
selves with the Imbangala had their counterparts on the Iihbangala
side in the amalgam of disparate political institutions which made up
the kilombo of the Lunda malcota. Near-total centralization of
authority within the Imbangala band made alliance with outsiders an
attractive prospect to subordinated Imbangala holders of permanent
titles. The limited information available on Imbangala political
structure in the first decade of the seventeenth century22 fits the
hypothesis that a single king of the kulembe type held the only
permanent and autonomous position of power within the band,
while all other chiefs held appointive titles of the vunga type. The
formal structure of the kilombo divided the members of each Imban­
gala band into approximately twelve distinct sections, each under the
leadership of its own ‘captain’. These regiments lived and fought
more or less separately from one another; twelve separate entrances
into the joint war camp, one for each regiment, symbolized these

10Treslado dfl aviso que mandou fazer o snór. g.dor bento banho Cardozo,
21 Aug. 1615; A .H .U ., Angola, cx. 1.
20Anonymous relação, 21 Oct. 161(5?); A .H .U ., Angola, cx. 1, doe. 172.
21Letter from Andre Velho da Fonseca to el-Rei, 28 Feb. 1612; published
in Brásio (1952-71), vi. 64-70; also in AA, sér. I, iii, nos. 19-21 (1937),
71-90.
22 Good data exist only for the 1640s; these receive detailed analysis in the
following chapter.
184 T H E IM B A N G A L A AND THE PORTUGUESE
distinctions even though all clustered in the same kilombo for pur­
poses of defence.23
The ‘captains’ probably held vunga titles appointed by the single
Imbangala king. They, and the regiments they commanded, had
replaced the lineages, with which the original group of Lunda had
begun, as the basic institutions of Imbangala social structure. Since
this structure had no place for the numerous Lunda and Cokwe
perpetual titles associated with the obsolete descent groups, such as
the lcota positions or the kulashingo, the several title-holders in the
main Imbangala band had fought continually for control of the king-
ship which ICalanda ka Imbe held in 1601. As a result, the position
had passed frequently from one Lunda lcota to another during the
preceding fifty or so years. According to the traditions, the kota
Kangengo at first claimed leadership of the kilombo but ruled for only
three ‘days’ before he died, allegedly a victim of the curse which
Kinguri had uttered just before his death.24 Mbongo wa Imbe then
succeeded Kangengo as ruler of the kilombo but he could not with­
stand Kinguri’s enraged spirit and lasted only two days before he
died. Kalanda ka Imbe was the third and last of the Lunda makota to
brave Kinguri’s curse; he lived only a single day before he met the
same fate as his predecessors.25
The increasing tempo of death which haunted the makota indicates
that their control over the kilombo had dissolved into anarchy as the
title-holders fought each other for control of the single position of
power. The traditions emphasize the supernatural cause of their

23 Battell in Ravenstein (1901), pp. 28-9; this is implicit in Battell’s descrip­

tion o f the twelve captains'who had come from the earliest origins o f the band,
each with his own section o f the camp.
24 The three ‘days’ mentioned in the tradition refer to stages in the initiation
ceremonies leading to Kangengo’s full installation as ruler. These ceremonies
took place in four phases, expressed metaphorically as ‘days’, and so the
traditions may not be interpreted as indicating how long Kangengo actually
commanded the band as de facto ruler The three ‘days’ mean only that holders
o f this title lost control before any o f them managed to complete the entire
set o f prescribed rituals.
25 Testimonies of Sousa Calunga, 16 June 1969; Domingos Vaz; Apolo de
Matos, 18 June 1969. Although the Vaz tradition has reversed the sequence
usually given, of these three rulers the certainty o f other informants that Kan­
gengo ruled first has led me to accept their version of the events. Except for
Rodrigues Neves (1854), p.99, written versions o f these traditions do not mention
the period of rule by Lunda makota Rodrigues Neves said that a ‘Kasanje ka
Imbe’ assumed power directly from the kinguri. Although his informant prob­
ably meant to refer to Kulashingo, whose original title had been Kasanje ka
Musumbi, he erroneously introduced the surname Imbe in a veiled acknow­
ledgement of the missing names of Mbongo wa Imbe and Kalanda ka Imbe.
 F IR S T C O N T A C T S—E ST A B L IS H IN G P A T T E R N 185
deaths, allegedly from the curse of the kinguri, by notiag that all
three would-be rulers died at night. The mahmda thus say nothing
direct about the wars which must have occurred during this period,
but the shift from Kangengo to Mbongo wa Imbe and Kalanda ka
Imbe indicates that the band had separated into two major groups
divided along the lines of the lineages to which the Lunda titles had
once belonged. Kangengo belonged to the old lineage of ICandama ka
Hite, which lost control to Mbongo and Kalanda ka Imbe from
ICandama ka ICilcongwa and ICanduma ka Kikongwa.26 The holders
of these titles still tended to unite according to the lines of the
suppressed Idn groups they had known in Lunda.
The holders of non-Lunda titles, who had joined the kinguri’s band
as it passed through Colcwe and Libolo, had never controlled the
most important kilombo position and undoubtedly desired to recover
the prestige which they had sacrificed to the unity achieved through
exclusive use of mavunga. Some of these excluded holders of Colcwe
and Libolo positions must have recognized the Portuguese as
potentially valuable allies if they should attempt to end Lunda
domination of the Imbangala. The Portuguese could not have been
unaware of the opportunities offered by the presence of title-holders
victimized by centralization; they repeatedly exploited similar situa­
tions in later African states as they extended their influence in Angola
during the nineteenth century. Later, European support for dissatis­
fied holders of subordinate titles led to the deposition and replace­
ment of strong and independent paramount authorities by weak
puppets in more than one African kingdom.27
Although no written records confirm the Portuguese part in the
formation of the alliance with the Imbangala, the circumstances of
other better-known cases suggest a likely sequence of events. Some
time after Battell left the Imbangala, then still under the leadership of
the Lunda kota Kalanda ka Imbe, a holder of the kulashingo title led
the Cokwe/Lwena component of the group in rebellion against con­
tinued rule by the Lunda title-holders. Governor Cardoso supported
the kulashingo’s drive to control the entire band in return for the
kulashingo’s agreement to enlist the Imbangala in the service of the
governor’s military designs. The Imbangala crossed the Kwanza,
with the aid of the Portuguese, and entered the fight against the
26Testimony o f Sousa Calunga, 9 July 1969.
27e.g. ICasanje in the 1850s; I have very briefly reviewed my unpublished
evidence for this interpretation in (1973b). Portuguese merchants and military
men played an identical role in bringing an end to the Mbondo ldngdom of the
ndala kisua in the 1880s and 1890s.
186 T H E IM B A N G A L A AND THE PORTUGUESE
vassals of the ngola a kiluanje and the king of Kongo. Carsoso thus
acquired not only powerful support for his wars against the Mbundu
but also gained an ally who owed his position to the Portuguese and
who, therefore, would hopefully prove obedient to Portuguese com­
mands. The Imbangala under the kulashingo had become potentially
docile allies in the official Portuguese slave trade instead of powerful
and troublesome threats to official hegemony.
Imbangala oral traditions support the hypothesis that the kula-
shingo joined the Portuguese at least in part to overthrow the author­
ity of the Lunda makota. Since the present narratives represent the
official ideology of the kulashingo group which later founded the
Kasanje kingdom under a restored kinguri title, they bear a strong
bias in Kulashingo’s favour. They thus minimize the documented
period of rule by the Lunda and emphasize, perhaps too insistently,
the directness of the line of legitimacy from Kinguri to Kulashingo.
As the traditions relate these events, Kulashingo received the sacred
symbols of power either directly from Kinguri or from the makota
within a few days of their Icing’s death in the palisade on Mbola na
Kasashe. This version of events effectively erases the period of about
fifty years (c. 1560 to c. 1610) when the makota ruled in the name of
the kulembe. Allowing for this alteration (entirely in keeping with the
traditions’ established tendency to eliminate events which had few
effects on later political structures) the traditions accurately preserve
the major political developments of this period and show that the
kulashingo rose to power on the wings of European support.
The relevant thread of the narrative begins during the ICinguri’s
journey from Mona Kimbundu, where Kulashingo had joined the
band, to Mbola na Kasashe. The makota had long seen that Kinguri’s
insatiable demands for human sacrifices threatened the survival of
their people, and so they took a number of foreigners, including
Kulashingo, from Mona Kimbundu to offer Kinguri in place of their
own followers. Kulashingo, however, managed to escape the fate plan­
ned for him by catering obsequiously to the demands of the makota,
especially Mwa Cangombe, Ndonga, and Kangengo, the leaders
of the Kandama lea Hite section of the band, who selected the
victims who died each day under Kinguri’s knives. He had earned the
rank of kibinda, or master hunter, before he left Mona Kimbundu and
consequently spent all his time hunting in the woods where he
remained out of Kinguri’s sight and beyond the reach of the makota.
Kulashingo always gave meat to the makota when he returned from a
successful hunt in order to ensure continued favour. He also con­
 F IR S T C O N T A C T S—E S T A B L IS H IN G PA T T E R N 187
ducted a secretive love affair with Imbe ya Malemba, the mother of
M bongo and ICalanda lea Imbe, to gain the confidence of the Kan-
dama and Kanduma lea Kilcongwa group of Lunda lineages. ICula-
sliingo’s gifts of meat metaphorically portray both his personal
loyalty to the makota and the special obligations owed by yibinda to
political chiefs. His marriage to Imbe ya Malemba represented an
alliance of the Cokwe/Lwena with at least one segment of the Lunda
lineages.
When the makota resolved to kill ICinguri at Mbola na ICasashe,
Kulashingo feigned support for them but, when he returned home
from hunting one evening to discover ICinguri trapped inside the
palisade, he secretly heeded the pleas of Imbe ya Malemba, who did
not want Kinguri to die of starvation. Each night he would deposit
portions of meat from his hunts outside a small hole in the palisade
where ICinguri could reach through and take it.28 ICinguri at first
assumed that this food represented tribute from the makota and did
not immediately realize that he had been abandoned. Since the
M bundu believe that human beings live only five days without food,
the makota waited unconcernedly for their ruler to die at the end of
the prescribed period.
When they found Kinguri still alive on the sixth day, they became
suspicious and initiated a search for the person who must have
smuggled food to the imprisoned king. The makota appointed
Kambwizu to stand guard outside the palisade where he could dis­
cover the culprit. He ascended a tree overlooking the prison and
watched until he saw Kulashingo leave the food that night.
Kambwizu reported this news to the other makota and they soon
discovered Imbe ya Malemba’s complicity. They threatened the
Cokwe/Lwena and the Lunda of Imbe ya Malemba, with Kangengo
in particular berating them for their effort to save Kinguri. When
Kinguri overheard this argument going on just outside the walls of
his prison, he realized for the first time that his makota had impris­
oned him and uttered the terrible curse which later killed Kangengo,
M bongo wa Imbe, and Kalanda lea Imbe.
Kinguri died, in spite of the efforts of Kulashingo and Imbe ya
Malemba, embittered by his makota’s betrayal and grateful to
Kulashingo for his aid. These circumstances justified ICulashingo’s
right to take ICinguri’s position as king of the Imbangala. Some
variants of this narrative episode state that Kinguri named Kulashingo
28 N o variant of this tradition accounts very logically for Kulashingo’s
sudden sympathy for the kinguri.
188 T H E IM B A N G A L A AND THE PORTUGUESE
as his successor before he died, even reaching through the hole in the
palisade to place the bracelet symbolizing royal authority (the lenge
or lukano) on ICulashingo’s arm.29 Others claim that the lukano
miraculously flew off Kinguri’s arm at the moment he succumbed and
encircled the arm of ICulashingo; various makota tried to remove the
lukano, but none succeeded.30 Another variant describes how the
symbols of power instantly struck down several claimants who
attempted to put them on in spite of the Kinguri’s curse. Yet, when
Kulashingo picked up the bracelet and other regalia, he survived and
was at once acclaimed as Kinguri’s legitimate heir.31 The single
historical point which links all these variants lies in their emphasis on
the supernatural dimension of Kulashingo’s accession and the
legitimacy of his assumption of Kinguri’s power.
Other versions, only superficially contradictory to those which
depict Kulashingo as winning over the reluctant makota, underscore
the legitimacy of his authority by making him the makota’s unwilling
nominee. These traditions originated in a later period when Kasanje
political procedures required the makota to elect Kulashingo’s
successors to the position of the Icinguri. According to these variants,
after ICangengo, Mbongo wa Imbe, and Kalanda lea Imbe had tried
to wear the lukano but had failed, the remaining makota recognized
their inability to control the powers of the Icinguri and implored Kula­
shingo to take the position to save them all from death. Kulashingo
demurred by saying, ‘I am not from Lunda. The misanga (beads)
deserve the neck, not the feet.’ This proverb (sabu) represented
Kulashingo as the feet, lowest and most humble part of the body. The
makota were the neck of the Imbangala band, the part most closely
associated with the head, or the Icinguri. The sabu implied that the
honour of leadership (the misanga) should devolve upon the makota
rather than on Kulashingo. The makota, according to this variant,
insisted that Kulashingo acceed to their pleas and he finally and
reluctantly agreed, swearing, ‘If I do evil, then let me also die.’ At
this oath, the nzumbi (spirit) left Kinguri’s corpse and entered the
body of Kulashingo.32 Kulashingo in this case appears as the legiti­

29 Testimonies o f Manuel Vaz; Domingos Vaz; Mwanya a Shiba, 14 June

1969; Apolo de Matos, 18 June 1969.
30 Testimony o f Sousa Calunga, 21 July 1969.
31 Testimony of Sousa Calunga, 16 June 1969.
32 Testimony of Sousa Calunga, 23 August 1969. This version primarily
accounts for the origin of the name Kulashingo. Sbingo means neck in Kim-
bundu, and kula is to eat; the name commemorates Kulashingo’s defeat o f the
makota; cf. p. 143, n. 94 above.
 F IR S T C O N T A C T S—E ST A B L IS H IN G P A T T E R N 189
mate heir of Kinguri through selection by the makota rather than by
overtly supernatural intervention. The episode, of course, reflects and
justifies the electoral procedures in use much later in Kasanje.
Whatever the supernatural sign which legitimized Kulashingo’s
right to succeed Kinguri, most versions of this episode contend that
the makota opposed him because they hoped to abolish the position
entirely. They recited the genealogies of both Kulashingo and
Kinguri to demonstrate that, while the two titles might descend from
the same remote matrilineal group (Lucaze na Mwazaza), Kinguri
owed his authority symbols to purely Lunda antecedents (Ngamba a
Mbumba);3334Kulashingo, whose title came from different (but not
specified) sources, therefore had no right to take Kinguri’s lukano.
The makota’s refusal to honour Kinguri’s last request sealed the fate
which they had prepared for themselves by assassinating him.
According to one variant of the tradition, Kulashingo finally yielded
the symbols of royal power to the Lunda, and Kangengo, Mbongo
wa Imbe, and Kalanda lea Imbe then succeeded one another, each
dying from the effects of Kinguri’s curse. The makota then realized
that they could not resist Kinguri’s designated heir and in desperation
switched their support to Kulashingo.
According to the traditions, the Imbangala first made contact with
Europeans while this succession struggle still divided the band. The
Portuguese living at Luanda had heard rumours of Kinguri’s great
powers and wished to meet so famous a king in person. The governor
of Angola sent a messenger, called Gaspar ICanzenza, to summon
Kinguri to Luanda. ICanzenza found the Imbangala camped on the
island of Mbola na ICasashe shortly after the makota had imprisoned
Kinguri. Because the governor had addressed the message to Kinguri
personally, the makota felt unqualified to reply in his place and there­
fore explained, deceitfully, that Kinguri had fallen ill and could not
receive visitors at that moment. Since Kinguri had, in fact, died just
before Gaspar Kanzenza arrived, the makota could not leave for
Luanda until someone assumed his position and gained the right to
respond to the Portuguese invitation. Only after Kulashingo finally
assumed the title did they set out for Luanda where they met the
governor and joined the Portuguese in fighting against the ngola a
kiluanje.3*

33 Testimony o f Domingos Vaz; also testimonies of Alexandre Vaz, 31 July

1969; and Alexandre vaz. Domingos Vaz.
34 Testimonies o f Sousa Calunga, 16 June, 21 July 1969; Sousa Calunga,
Kambo ka Kilcasa; Manuel Vaz.
190 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
Although the figure of Gaspar Kanzenza is probably a fictional
prototype for nineteenth-century Portuguese messengers to Kasanje,
the presence of this role in the episode confirms the timing of the
Imbangala alliance with the Portuguese, after the makota had failed
to replace the kinguri, and shows that ICulashingo (as Kinguri’s
alleged legitimate heir) rather than the makota responded to the
Portuguese summons. The otherwise unexplained reluctance of the
makota to answer the governor’s request suggests that the Portuguese
would deal only with a holder of the kinguri position, and not with the
kilombo titles controlled by the makota. The makota’s refusal to
acknowledge that the kinguri had died probably represents attempts
by the historical holders of the Lunda positions to resist Portuguese
intervention against their control over the kilombo. They might, for
example, have spread the false rumour that a kinguri still lived, as the
traditions say they did, since this would have prevented the holder of
the kulashingo from resurrecting the kinguri title as the basis for his
Portuguese-sponsored rule over the Imbangala.35
The differing roles of the two major groups of makota in these
traditions reiterate the divisions noted elsewhere among the Lunda
35 The published versions of nineteenth-century Imbangala traditions have
neglected this event at considerable cost in terms of understanding Imbangala
history. Their omission provides an illuminating example o f the way in which
historians shifted the emphases of the malunda to reflect changing political
conditions, since circumstances in Kasanje at the time Rodrigues Neves
(1854, pp. 99-100) recorded these traditions explain the enigma very nicely.
Both Rodrigues Neves and the commander o f the 1850 Portuguese military
expedition to Kasanje (Salles Ferreira (1854-8)) made a point o f recording
Imbangala history in their memoirs o f the campaign but failed to mention the
disqualification o f the Lunda makota from the Kasanje ldngship. The Portu­
guese commanders had a strong incentive to ignore the Imbangala prohibition,
since they hoped to impose illegitimately a member o f the makota lineages as
a puppet Kasanje king responsive to Portuguese pressure. But at the same time
they had to avoid offending the sensibilities o f governors in Luanda who
occasionally tried to respect the oral charters o f the African states with which
they maintained diplomatic relations. Rodrigues Neves and Salles Ferreira,
faithful to the strategy o f support for dissatisfied subordinate chiefs against
recalcitrant kings that their countrymen had developed in the early seventeenth
century, championed the cause o f the makota lineages against the ruling heirs
o f Kulashingo. They evidently expected to find grateful allies among lineages
that normally had no chance to take power in Kasanje but which possessed
ancient titles descended from the ‘personal companions’ o f the founder of
the kingdom. If Rodrigues Neves and Salles Ferreira could conceal Imbangala
prohibitions against rule by the ‘Lunda’, their unsuspecting superiors in Luanda
might accept a king chosen from the ranks o f the makota. They could thus
install, with the approval of the Angola governor, the docile Kasanje ruler
whom they desired. They in fact succeeded in choosing a kota (ICalunga ka
Kisanga), but the Imbangala never accepted him and he died within a few
months under mysterious circumstances.
 F IR S T C O N T A C T S —E ST A B L IS H IN G P A T T E R N 191
titles at this period. Not all of them supported the assassination of the
kinguri, since the leaders of the mutiny all came from the lineage of
Kandama ka Hite. Kangengo, Mwa Cangombe, and Ndonga were
the makota who attempted to substitute the people of Kulashingo for
their own followers as sacrifices for Kinguri. Kambwizu, the koto, who
apprehended Kulashingo as he left food for the imprisoned kinguri,
also came from this group. The makota of Kandama ka Kikongwa
and Kanduma ka Kikongwa, however, had no connection with the
murder. Imbe ya Malemba’s encouragement of Kulashingo’s aid to
the kinguri symbolized their opposition to the scheme. Kulashingo
had taken care to ally himself to both Lunda factions before the
murder, but, if the traditions are read correctly, at the critical moment
he threw his support to the party of Kandama ka Kikongwa and
Kanduma ka Kikongwa with whom he and his people had a marriage
alliance through Imbe ya Malemba.
Although Kangengo and the ringleaders of the plot managed to
seize initial control after abolition of the kinguri, the political balance
soon shifted to Mbongó wa Imbe and Kalanda ka Imbe of the other
lineage. Kulashingo and his Colcwe/Lwena probably assisted them in
replacing the makota from Kandama ka Hite. Kulashingo apparently
assumed a subordinate vwwgc-position as the kasanje of Mbongo wa
Imbe or Kalanda ka Imbe at that time.36 Since this position made him
a close associate of the holders of power, yet still excluded him from
access to real authority, Kulashingo was an obvious candidate for
Portuguese support to replace the makota with more tractable
leadership. The duties of the kasanje included conduct of relations
with outsiders, either through diplomacy or war, and thus provided
the occasion for his initial contacts with the Europeans.
Since Kulashingo lacked legitimacy in terms of both prevailing
sets of political institutions, the Lunda system of perpetual titles and
the rules of the kilombo, and since the titles offered by the Portuguese
(‘Jaga’, Kyambole, etc.) had little meaning in terms of Imbangala
politics, he resurrected the title of kinguri in order to give his de facto
rule legitimacy in the eyes of his people. Although the makota had
formally abolished the title, Kulashingo could quite easily have
reclaimed it for himself by retrieving the authority symbols aban­
doned in Mbola na Kasashe, or even by fabricating credible imita­
tions of the originals. Seventeenth-century Mbundu, who would have
36 Kulashingo was also known as Kasanje ka Musumbi, Kasanje ka Kazanza,
or Kasanje ka Kibuna. Rodrigues Neves (1854), p. 99, gave the exact title,
Kasanje ka Imbe (ya Malemba).
 192 T H E IM B A N G A L A AND THE PORTUGUESE
lived through these events, explicitly recalled that Kulashingo
abandoned the kilombo, describing him as hostile to certain customs
of the makota, especially those involving the consumption of human
flesh.37 This partial description of the kulashingo's break with the
makota implied that he had adopted some other basis for his auth­
ority.
Armed with the revived symbols of the kingurVs powers and aided
by support from the Portuguese, Kulashingo emerged as leader of the
Imbangala over the opposition of the makota and other officials loyal
to the kilombo. The Portuguese awarded Kulashingo the title of
‘Jaga’, an honour of their own invention, and the name ‘Cassanje’ in
acknowledgement of the Icasanje title he had borne at the time of his
first contact with them. The resulting combination, ‘Jaga Cassanje’,
became the official Portuguese designation for all later Icings of
Kulashingo’s Imbangala band, and their followers thems fives
became known as ‘Cassanjes’.
Kulashingo soon justified his assumption of the kingurVs mantle by
winning great military victories which amply demonstrated his firm
control over the symbols of authority. Portuguese documentation
and Imbangala narratives tell the same story in somewhat different
terms. According to the traditions, Kulashingo presented himself
before a Portuguese governor in Luanda and offered to fight with him
against the ngola a kiluanje. The ngola a kiluanje, it was said, had
built an enchanted fortress which the Portuguese could not penetrate
or even locate, and the governor needed Kulashingo’s magical
weapons in order to defeat him.38 This story reflects the Imbangala
belief that guns and spears alone did not decide battles but merely
' confirmed a conclusion predetermined by charms possessed by the two
sides. It did not surprise the Imbangala, therefore, that Europeans
should have found themselves unable to discover the ngola a
kiluanje’s carefully enchanted refuge. According to these beliefs, the
Europeans’ disdain for the necessary charms would have made them
dependent on the supernatural powers of the kinguri to overcome the
spells of the ngola a kiluanje.
Kulashingo put the magical powers of the kinguri to good use. His
primary weapon was a bow (mufula) called the kimbundu. Kulashingo
had claimed the ability to vanquish any foe with his special charm,
and, since the Portuguese had never before seen such a weapon and
doubted its effectiveness, he devised a test to prove its capabilities. He
37Cavazzi (1965), i. 190.
38Testimony o f Sousa Calunga, 29 Sept. 1969.
 F I R S T C O N T A C T S—E S T A B L IS H IN G P A T T E R N 193
requested nine oxen and fired only a single arrow at the first ox. The
arrow killed the ox and then continued on a miraculously curving
path until it had killed all nine oxen and felled two thick baobab trees
as well. It finally plunged into the sea and disappeared. The Portu­
guese governor, suitably impressed with this demonstration, agreed
to accept Kulashingo’s help. The ngola a kiluanje, who understood
such things even better than the Europeans, became so frightened that
he fled at once to Pungo a Ndongo and later moved north to the
Wamba river where his descendants still live today.35*39
The modern versions of the tradition allege that these events took
place in the square in front of the governor’s palace in Luanda. Since
no palace existed in Luanda in 1612, such references are anachron­
isms incorporated into the traditions at a much later date. Even the
location of the story in Luanda, while consistent with the Mbundu
tendency to place all very early events there,40 is also false. The capital
of the ngola a kiluanje, said by the tradition to have been in Luanda,
in fact lay well in the interior, probably somewhere between the
Lukala and the Kwanza rivers.41 The historical kulashingo almost
certainly never saw the coast north of the Kwanza but remained well
up in the highlands which begin sixty to eighty miles from the coast.
Nineteenth-century traditions noted that he crossed the Kwanza not
far from Kainbambe,42 a plausible location not far from the last
recorded Portuguese contact with the Imbangala a decade earlier.
This was also a logical place to bring a large number of people across
the Kwanza, since the river west of the fall line at Kambambe
becomes too wide and deep to afford easy passage.43
35 Testimonies o f Sousa Calunga, 16 June 1969; Domingos Vaz; Mwanya a
Shiba, 14,15 June 1969. Dias de Carvalho (1898), p. 31, said that the Imbangala
received firearms at that time, but this tradition shows no evidence that the
Imbangala considered European weapons important. Guns were in very short
supply in Angola at that time and, in further refutation o f Dias de Carvalho’s
assumption, the governors used the Imbangala in part to avoid excessive reli­
ance on supplies of firearms and ammunition from Europe.
40 Cf. the opening episodes o f the story o f Kajinga, pp. 98-103 above.
41 Rodrigues, ‘História da residência dos padres’ (Rodrigues (1936)) gave
the distance from Luanda to the capital of the ngola a kiluanje as sixty leagues,
or about 180 miles; he probably overstated it somewhat. Sr. Fernando Batalha
believes that he may have found the general area of this capital along the
middle Lukala; personal communication.
42 Carvalho (1890a), pp. 77-8.
43 Most other details given in nineteenth century traditions reported by
Rodrigues Neves (1854), pp. 100-1) have little relevance to the seventeenth
century, as they derive in general from nineteenth century conditions. An
unhistorical nineteenth-century equation o f ‘Jinga’, or the queen Nzinga,
with the heirs of the ngola a kiluanje then living on the Wamba river explains
194 T H E IM B A N G A L A A N D T H E P O R T U G U E S E

Kulashingo’s Imbangala in Angola

Kulashingo’s Imbangala camped for a few years in the territory
just outside the small area under Portuguese control in the 1610s in a
prototype of all future Portuguese-Imbangala settlement patterns
north of the Kwanza. The main title-holder of each Imbangala band
settled as an Mbundu Icing replacing, in most cases north of the river,
some form of overrule by the ngola a kiluanje with a new state based
on conscription of local males into the kilombo society. Induction of
these men into the kilombo simultaneously deprived the most produc­
tive part of the local population of their former membership in
Mbundu descent groups and subjected it to the direct authority of the
Imbangala king and his vwnga-appointees. The centralized political
institutions of the Imbangala created a small but highly centralized
state in which the king could mobilize relatively large numbers of
men for fighting purposes on very short notice. In a military sense,
the Imbangala state of the kulashingo, and all the related Imbangala
kingdoms among the Mbundu, constituted mercenary camps estab­
lished on the fringes of Portuguese Angola, dominated by skilled
warriors who captured local farmers for sale as slaves in normal
times and who readily joined Portuguese expeditions to fight in
times of officially-declared wars. Unfortunately, from the perspective
of the governors in Luanda, the Imbangala also joined forces with
other traders not so closely linked to the policies of the crown’s
representatives at the capital.
Despite somewhat contradictory traditions,44 circumstantial evi­
dence points to the middle Lukala as the site of the kulashingo''s
permanent Imbangala camp since an area there, called Lukamba,
44 The traditions do not agree on the exact location where the Imbangala
lived while settled under Portuguese auspices in Angola but do concur in the
essential point that they lived somewhere near the lands then subject to Portu­
guese control. One nineteenth-century tradition (Rodrigues Neves (1854), pp.
100-1, repeated by Dias de Carvalho (1890a, p. 79, and 1898, pp. 63-4))
stated that the governor gave them lands on the middle Lukala between two
nineteenth-century towns, Ambaca and Golungo Alto. Another tradition,
dating from the early part o f the twentieth century, specified only that the place
was somewhere not far from Luanda, probably meaning anywhere west of
Kasanje (Oliveira Ferreira Diniz (1918), p. 100). A tradition from the same
area as that o f Oliveira Ferreira Diniz collected in 1969 (testimony o f Mwanya
a Shiba, 15 June 1969) made the same point by noting only that the Imbangala
had settled somewhere near where the Kwanza meets the sea.

the claim that Kulashingo fought ‘Jinga’ near Luanda and defeated her before
the Portuguese dared to come to the mainland from their based on the islands
just off the coast.
 K U L A S H I N G O ’S IM B A N G A L A IN A N G O L A 195
became an early and important centre of the slave trade.45 A
slave emporium would necessarily have developed near the main
Imbangala settlement as a consequence of their military/economic
partnership with the Portuguese. One nineteenth-century tradition
underlined the importance of commercial factors by recalling that
Kulashingo formally obligated himself to trade with the Portuguese
at that time.46
Strategic considerations would also have led the Portuguese to
settle the kulashingo's Imbangala near Ambaca. The governors of the
decade between 1610 and 1620 concentrated their military efforts on
penetrating the heartland of the ngola a kiluanje state. They first
strengthened their position on the lower Lulcala by building a new
presidio at a site called Hango, which would serve as a firm base for
launching future operations farther up the river. Lukamba (or
Ambaca) would have been the logical position for a mercenary force
stationed as a buffer between Ndongo and the weak Portuguese
forces farther down the river. The Portuguese then advanced the
fortress at Hango to a new site near Ambaca in 1617 as they mounted
their drive towards the capital of the ngola a kiluanje. The second
location of this presidio in Ambaca also points to nearby Lukamba as
the Imbangala base since the Portuguese would have wanted a
fortified post as near as possible to their African allies for purposes
of control and co-ordination.47
One consequence of the presence of ICulashingo’s Imbangala near
Angola was that they produced more slaves than official government
channels could absorb. Consequently, the second illegal trading
system thrived alongside the legal one centred on the customs house
in Luanda. Some governors encouraged these illegal ventures and

45 Some traditions (Sousa Calunga, 22 Aug., 2 Oct. 1969) call the lands
where the kulashingo settled ‘Kikanga’ (cf. Carvalho (1898), p. 66). Kikanga
means ‘market-place’; Assis Jr. (n.d.), p. 48. The oldest known tradition on this
point (1750s), noted by Leitão in Sousa Dias (1938), pp. 16-17), stated that
the ancestors of the ICasanje kings had once lived in ‘Ambaca’.
40 Schiitt (1881), p. 80. This tradition used the image o f an ivory tribute,
characteristic o f the time at which it was collected, to indicate this.
47 Nineteenth-century maps o f the Ambaca region showed a (perhaps only
coincidental) juxtapostion o f toponyms that duplicated place names also found
in ICasanje. A hill named ICasala lay south o f a region called Kasanje in
Cazengo between Ambaca and Golungo Alto. The hill had the same geo­
graphical relationship to the region as the hill called ICasala in the Baixa de
Cassanje had to the capitals of the kings of ICasanje. It is possible, though not
proved, that the coincidence o f toponyms indicated the exact spot where the
imbangala had camped. See map o f ‘Loanda and Ambaca and the Course of
ihe River Kwanza’, in Capello and Ivens (1882).
196 T H E IM B A N G A L A A N D T H E P O R T U G U E S E

M ap ix . The Imbangala and the Portuguese (c. 1600-1650)

doubtless profited from them. Manuel Cerveira Pereira (governor

from 1615 to 1617), for example, encouraged the Imbangala to over­
run the local Mbundu without restraint.48 They caused a great deal of
destruction, and local opponents of Cerveira Pereira’s policies soon
informed Lisbon that his unrestrained use of the Imbangala was
harming royal interests in the slave trade more than it was helping.
Authorities in Europe replaced Cerveira Pereira as a result of his
indiscriminate use of the Imbangala. The next governor, Luis Mendes
de Vasconcelos (1617-21), confirmed upon his arrival in 1617 that
Cerveira Pereira had left Angola in deplorable condition from the
points of view of both the Mbundu and the crown.
Vasconcelos’s initial judgement of the status of the Portuguese
conquista stemmed from revenues lost by the royal treasury. Un­
scrupulous traders, he explained, diverted slaves from Luanda to
other ports where they could evade royal customs inspectors. Rather
48 Miller (1972a), p. 569, for the irony implicit in this fact.
 K U L A S H I N G O ’S IM B A N G A L A IN ANG O LA 197
credulously, Vasconcelos added (apparently on the basis of reports
from traders engaged in the illegal trade) that many other slaves never
reached the coast at all since the Imbangala ate them on the spot to
satisfy their notorious craving for human flesh. These traders took
advantage of the exaggerated reputation of the Imbangala as canni­
bals to disguise the disappearance of slaves actually smuggled through
other ports.49 With somewhat greater basis in fact, Vasconcelos
reported that the allied Portuguese and Imbangala armies had utterly
destroyed many sobas, or hereditary Mbundu title-holders. Some
sobas had lost so many of their people that they could no longer
provide the direct tribute in slaves demanded by the crown. Vascon­
celos pleaded that he could do very little to improve the situation,
however, since he found himself opposed unanimously by local
traders who argued that the colony would fail without the assistance
of the Imbangala warriors.50 Their argument, at that period in
Angola’s history, did not greatly exaggerate the facts.
The Spanish Icing Phillip II, then extending Spanish control into the
overseas possessions of Portugal during the period of the Dual
Monarchy in Europe (1580-1640), sent an emissary, Rebello de
Aragão, at about that time to assess the position of the Portuguese
forces in Angola. As a Spaniard, Rebello de Aragão brought a fresh
non-Portuguese perspective to the controversy over the role of the
Imbangala, and his judgements may therefore be accepted with some
confidence. His 1618 report reflected Vasconcelos’s initial hostility
to the Imbangala and the traders who depended on them for their
supply of illegal slave exports. He explained that the Imbangala had
served the Portuguese governors well during the early years of the
partnership. They had proved such effective and terrifying warriors,
in fact, that the local sobas had become peaceful and docile purely
from fear. But the Imbangala had become haughty and had begun to
divert slaves which should have gone to the royal customs house in
Luanda. The Spanish inspector thus indicated their involvement in
the unofficial and illegal slave trade.
Rebello de Aragão emphasized that responsibility for the sorry
state of affairs in 1618 did not lie entirely on the side of the Africans,
since the Portuguese engaged in the illegal trade had encouraged
their cruelties and the kidnapping of local people.51 Whenever the

« Miller (1973a), pp. 134-5.

50Luiz Mendes de Vasconcelos to el-Rei, 28 Aug. 1617; A .H .U ., Angola,
cx. 1, doe. 129; published in Brásio (1952-71), vi. 283-5.
51Rebello de Aragão, ‘Relação’; Cordeiro (1881), iii. 12-17.
198 T H E IM B A N G A L A AND THE PORTUGUESE
supply of captives slowed, according to a contemporary account, the
captains of the presidios sent raiders to harass local chiefs even
without the aid of the Imbangala.52 Portuguese and Imbangala
interests in the captives taken in such raids complemented each other
perfectly: the Imbangala preferred to keep the youngest boys as
recruits for the kilombo while the Europeans would buy the females
and adult males the Imbangala did not need. The Imbangala could
take part in an unofficial raid, give some prisoners to the captain who
had permitted the foray, sell part of the remainder to private traders,
and keep the rest for their own purposes. Slaves captured and distri­
buted in this manner rarely, if ever, produced the customs duties
levied on others taken and sent towards the coast through legal
channels.
The Imbangala thus became the ambivalent keystone supporting
both arms of the double trading system which matured during the
early seventeenth century. One stream of slaves came from legal
trading and tribute authorized by the government and taxed accord­
ingly. The effectiveness of this system depended on the formal alliance
between the kulashingo and the Portuguese governors. The other
trade, in which the Imbangala played an increasingly important role
during the 1610s, depended on raiding and extortion by officials and
traders who smuggled these slaves from Angola under the cover of
assertions that the Imbangala had eaten them. By 1617, therefore, the
Imbangala had rejected their role as mercenaries employed exclus­
ively by the governor of Angola and returned to their original part as
agents of private traders.
Vasconcelos apparently attempted to bring the Imbangala back
under government control but failed, instead succeeding only in dis­
persing the main Imbangala kilombo and sending the kulashingo in
flight far into the interior. He had apparently reached Angola with a
sincere desire to collect the export duties imposed by the king and his
contractors and to eliminate the illegal trading system. Full imple­
mentation of these policies would have placed Portuguese-Imbangala
relations on a precarious footing. Vasconcelos succumbed almost at
once, however, to the reality which had turned previous governors
into enthusiastic supporters of the Imbangala and participants in the
unofficial slave trade: low salary, lack of support from superiors in
Europe, inadequate control over his own subordinates scattered
in inaccessible posts throughout the interior, and a European
52 Anonymous relação (Vasconcelos?), 31 Oct. 161(8?); A .H .U ., Angola,
cx. 1, doc. 172.
 K U L A S H I N G O ’S IM B A N G A L A IN A N G O L A 199
population unanimously opposed to any change in the status
quo.
Vasconcelos initially attacked the problem of the illegal slave trade
head on by raising an army with the expressed intent of driving the
Imbangala out of Angola. This was the campaign in which the
Portuguese moved the old fortress at Hango to its new site near
Ambaca where Portuguese cannon could cover the Imbangala
camp.53 When he reached their encampment, however, he abandoned
the expressed purpose of the expedition and joined the Imbangala in
a sweeping raid which carried far east into Ndongo and north across
the Lukala into Kongo territory. Together they attacked all the most
powerful rulers in these regions, including Keta kya ‘Labalanga’,54
the ngola a kiluanje, and some of the ndembu chiefs. This war caused
such devastation that trade came to a standstill, roads were closed to
travel, and the widespread destruction of crops brought starvation
everywhere. The Imbangala, used ruthlessly as in this case, could
overwhelm all Mbundu resistance and guarantee that the Portuguese
would not lose a major battle.
The Imbangala, who, as Vascencelos’s critics later pointed out, had
received very lenient treatment from governors up to that point, then
began to disagree with their Portuguese sponsors. Vasconcelos may
have found that he could not defeat the Imbangala and then embarked
on the 1617-18 campaign to demonstrate that their real interests lay
with the government rather than with the illegal traders. The spectac­
ular success of this raid, however, could not hide the fact that he also
intended to exert closer control over their activities than his pre­
decessors. The Imbangala probably interpreted the new presidio at
Ambaca correctly as a Portuguese attempt to bring them under
tighter reign; a fort and garrison so near their main camp might
eventually force them to abandon their participation in the illegal
slave trade. As a result of some such consideration, never identified in

53 Those who have sought ‘Manuels’ with whom the Imbangala could have
dealt at this time should not miss the captain o f the new Ambaca presidio, one
Manuel Castanho; cf. Miller (1972a), pp. 550, 569-70.
54 Keta kya ‘Labalanga’ was probably a subordinate chief of the ngola a
kiluanje who owned his title to the hango kings o f Libolo that had earlier ruled
this region. Modern traditions give a Keta kya Wabo wa Hango (‘Labalanga’?)
position related to the hango kings; testimony o f Sousa Calunga, 11 Sept.
1969. This chief definitely was not Kahete as suggested by Delgado in Cad-
omega (1940-2), i. 89, n. 1. Vasconcelos did not fight against the Imbangala in
that war as some authors have assumed; the confusion on this point has arisen
due to the similarity o f the names o f the mani Kasanze, a Kongo subchief near
Luanda, and the Imbangala (‘Cassanges’).
200 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
the extant documents, Iculashingo’s Imbangala left Angola in about
1618. Vasconcelos’s opponents reported in 1619 that the ‘greatest
captain among them [the Imbangala], valorous and powerful’ had
left the Portuguese and departed for the interior. Although this report
did not refer to the Imbangala ‘captain’ by name, the circumstances
make it almost certain that he meant the Iculashingo.55 The kulashingo
had left with so many local Mbundu (‘slaves and Christian vassals of
the Portuguese king’), as well as trade goods, that the Luanda mer­
chants suffered substantial financial losses.56
Nineteenth-century Mbundu traditions metaphorically attributed
the departure of the Imbangala to failure of the crops they received
from the Portuguese and planted when they settled in Lukamba.
Kulashingo, they reported, found that the seed had been roasted or
had become rotten and would not grow.57 The Imbangala may in fact
have received seeds from the Portuguese, probably as an inducement
to settle permanently as farmers, or the image of crop failure may
refer to the historical famine conditions mentioned in the documents,
but the primary historical content of the episode in Imbangala terms
deals with their refusal to abandon the nomadic life of wandering
raiders. The planting of seeds represented a temporary conversion to
a sedentary farming life, and crop failure constituted an omen warn­
ing them against adopting the new style of living. Fear of the spiritual
consequences of permanent settlement is thus the remembered cause
which drove the Imbangala from Lukamba. The published version of
the tradition closest to the original oral testimony draws an explicit

55 Although the published versions o f the Imbangala traditions disagree as

to the identity o f the king who led the Imbangala to Angola and then on to
Kasanje, my informants left no doubt that he held the title o f the kulashingo.
The confusion in earlier accounts comes from the Mbundu habit o f referring
to a single individual by several different names and titles. Kulashingo was
known indifferently by his proper name (Kasanje lea Musumbi, or Kasanje ka
Kibuna or ICazanza, etc.), by a praise name or throne name (i.e. Kulashingo),
tor by his title (the kinguri). Nineteenth-century informants may even have used
his (later) Portuguese title (Jaga Cassanje), Only Curt von Frangois (1888),
p. 274, accurately identified Kulashingo as the leader who took the Imbangala
from Angola to the Lui river.
56 Copia dos excessos que se cometem no governo de Angolla que o bispo
deu a V. Mg.d0 pedindo remedeo defies de presente e de futuro, 7 Sept. 1619;
A .H .U ., Angola, cx. 1, doc. 175; published in Brasio (1952-71). vi. 366-74.
57 D ias de Carvalho (1890a), p. 79. The image o f roasted or rotten seeds
operates as a cliche that may appear in such other episodes as the story o f the
malcota’s murder of the kinguri (pp. 173-175 above). One name for the area
where the Imbangala stayed in Angola comes from this episode: Lulcamba,
the name, came from a verb kukamba meaning to lack, be scarce, not suffice
(Assis Jr. (n.d.), p. 189), since the land would not bear crops.
 M A T U R E P O R T U G U E S E -IM B A N G A L A A L L IA N C E 201
contrast between a settled, farming life and the Imbangala decision
to ‘return to hunting’.58
Another nineteenth-century tradition used a different image but
still depicted a magical sign against settlement and in favour of sub­
sequent return to the wandering life of hunters. The Imbangala had
settled in Lukamba, according to this variant, and had planted their
seeds when an elephant entered their fields and ruined the crops. Since
elephants almost never appeared in that region, the Imbangala inter­
preted this misfortune as an omen. ICulashingo sent his yibinda
(specialists in hunting supernatural beasts) in pursuit of the animal
and they followed it to the future site of the kingdom of Kasanje.59
The positive omen of the magical elephant attracted the Imbangala to
hunting in this variant, just as the negative portent of ruined seeds
discouraged agricultural activities in other versions.50

The Mature Portuguese-Imbangala Alliance

The enormous success of Yasconcelos’s 1617-18 military campaign
and the hardships caused by the defection of the kulashingo brought
the pattern of European-Imbangala relations north of the Kwanza to
maturity. The Portuguese had definitively occupied the territory on
the north bank of the Kwanza as far east as their outposts at Ambaca
and later, after final defeat of the hari a Iciluanje in 1671, at Pungo
Andongo. Around these lands lay a ring of new Mbundu states
founded in the middle of the seventeenth century by Lunda title-
holders heading bands of Imbangala. The Portuguese employed these
Imbangala as mercenaries in wars and used them as their primary
suppliers of slaves in peacetime. As a result, they usually treated the
Imbangala kings as generously as possible.61 This arrangement
predominated between the Lulcala, the Kwango, and the Kwanza
from the 1620s until after 1850.
Much of this history centres on holders of Lunda makota titles who
58Dias de Carvalho (1885-6), p. 136. Later Imbangala historians may have
blamed crop failure on Portuguese treachery because o f the hostility with
which late nineteenth-century Imbangala regarded the government in Luanda.
59Rodrigues Neves (1854), p. 102.
60 Modern Imbangala traditions, in acknowledgement o f colonial political
realities, emphasized that Kulashingo left Luanda with the governor’s blessing
and that he settled in the interior as an emissary o f the Portuguese government.
61 See the relação o f Garcia Mendes Castelo Branco, 16 Jan. 1620; Ajuda,
51-VI13-25, fols. 93-95v ; published in Brásio (1952-71), vi. 446-52; also in
Cordeiro (1881), ii. Castelo Branco argued that the Portuguese king should
assist the merchants of Luanda with three casks o f wine each year which
would be given to the Imbangala to maintain their loyalty and the profitable
slave trade that sprang from it.
202 T H E IM B A N G A L A AND THE PORTUGUESE
remained in Portuguese Angola after the kulashingo had departed for
the interior. All of them clearly depended on Portuguese support for
their positions of dominance over the local Mbundu lineages. The
kulashingo’s alliance with the Portuguese between 1612 and 1619 had
exacerbated the deep rivalries which divided the title-holders in his
group of Imbangala. Few of the Lunda makota would have willingly
accepted co-operation with the Europeans on a basis which supported
the kulashingo at their expense. As claimants to independent sources
of legitimacy through the kilombo, and unwilling to recongize the
kinguri title which they had abolished once before, they dispersed
about the time the kulashingo left for the interior to found as many as
thirty separate bands near Angola. Although a few of these makota
initially reverted to open hostility against Portuguese official forces,
they achieved very little success with this policy. Ndonga, one of the
kinguri’s original makota, lingered briefly after 1618 in the devastated
parts of Ndongo but suffered defeat by a 1620-1 expedition which
followed up on the great campaigns of 1617-19; in this instance, the
Portuguese captured the chief and completely destroyed his army.52
This defeat effectively ended Ndonga’s contact with the Europeans
and caused some remnants of the band to flee with the title to the
Baixa de Cassanje where they rejoined the kulashingo’s Imbangala in
the nascent state of Kasanje.63
Kabulcu lea Ndonga, a title-holder subordinate to the ndonga,
assumed control of other parts of the ndonga’s band and reversed the
policies of his predecessor, joining with the governor of Angola in a
1621-2 campaign against the mani Kasanze near Luanda.64 ICabuku
ka Ndonga continued north after winning a victory against Kasanze
and invaded the southern Kongo provinces of Mbamba and Mpemba
where he defeated the Mbamba army and lolled both the duke of
Mbamba and the marquis of Mpemba. ICabuku ka Ndonga and the
Portuguese returned to Kongo on the side of the mani Kongo in
1623, defeating two ndembu chiefs who had revolted against their
Kongo overlords.65 The rise of the junior position to replace the
62Cadornega (1940-2), i. 90-4.
63 All Imbangala testimonies.
64 Miller (1972b), pp. 51-3.
65The extent o f Imbangala involvement in this campaign and the nature o f
their conduct may have been exaggerated by the Kongo kings. M ost descrip­
tions o f these wars come from São Salvàdor, where the Kongo traditionally
used the Imbangala as whipping boys when complaining to Europe against the
alleged injustices o f the governors o f Angola. In this case, the Kongo Icings
succeeded in obtaining the intervention o f the Pope against Vasconcelos,
resulting in his recall that same year; Jadin and Cuvelier (1954), pp. 456, 458,
 M A T U R E P O R T U G U E S E -IM B A N G A L A A L L IA N C E 203
ndonga and his support of the Portuguese paralleled the pattern
established when the kulashingo had eclipsed the makota and joined
Bento Banho Cardoso in 1612, but this partnership proved somewhat
more enduring than that of the kidashingo.
Kabulcu lea Ndonga settled down to build a semi-independent
kingdom on the southern border of the ndembu region where he
stayed on the Portuguese northern frontier as a buffer against the
hostile ndembu rulers across the Nzenza river. His strategic position
between the Portuguese and their enemies recalled the location of the
kulashingo'’s Imbangala settled near Ambaca as advance guards
against the ngola a kiluanje.
A Dutch invasion of Angola during the 1640s, part of the Thirty
Years War in Europe, brought one of the few notable—but temporary
—defections of a kabuku ka ndonga in a 200-year-long alliance. The
holder of the title wavered briefly in his loyalty to the Portuguese at
the outset of the Dutch invasion,66 but he soon returned to the side of
his old allies and in 1643 launched an expedition from his base in
Wumba north of the Lulcala to harass ndembu chiefs who had helped
the Dutch during their occupation of Luanda.67
A succession crisis in the kabuku ka ndonga’s kingdom later
allowed the Portuguese to repeat their tactic of replacing legitimate
Imbangala kings with puppet successors who depended on European
support to maintain their positions. The kingdom of Matamba,
centred on the Wamba river, had by the 1640s become one of the
most powerful eastern Mbundu states under the rule of the justly
famous queen Nzinga. She had attempted to re-establish the ngola a
kiluanje title there after the Portuguese had put puppets in the place
60 During the tenure o f Governor Pedro Cesar de Meneses, 1639-43; on the
Dutch conquest o f the coastal portions o f Angola, see Charles R. Boxer (1952).
67 Cadornega (1940-2), i. 278-9, 286-7. Consulta do Conselho Ultramarino,
17 Aug. 1644; A .H .U ., Códice 13, fols. 108-108v (Brásio (1952-71), ix. 153),
contains a commendation to the ‘Jaga’ Kabulcu for his assistance. Another
consulta o f the Conselho Ultramarino, 23 July 1644 (A .H .U ., Angola, cx. 3,
published in Brásio (1952-61), ix. 28-38) shows that Governor Antonio Abreu
de Miranda (1643-4) befriended an unnamed ‘Jaga’, very likely Kabuku ka
N dongo, in the area o f Ambaca and tried to get him to harass the Dutch. The
expedition was described in a letter from the Goverador to el-Rei, 9 Mar.
1643; A .H .U ., Angola, cx. 2.

etc. Rei do Congo to Mons. Vives, 26 Feb. 1622; Biblioteca Vaticana, Cod.
Vat. Lat. 12516, fols. 81—81v ; Brás Correia to D om João Baptista Viues,
10 Dec. 1623; ibid., fols 95—96v ; Conego da Sé do Congo to P° Manuel Rod­
rigues, S.J. (1624); Biblioteca e Arquivo Districtal—Evora, MS. CXVI/2-15,
no. 7; all published in Brásio (1952-71), vii. 3-4, 166-70, 291-7.
204 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
of the original Ndongo kings during the 1620s.68 As part of her policy
of harassing Portuguese official interests, she co-operated with the
Dutch while they held Luanda from 1641 to 1648. Kabulcu lea Ndonga
was still fighting on the Portuguese side against the Dutch in 1645 or
1646 when Nzinga’s army captured the ruling Imbangala Icing some­
where east of Ambaca. Nzinga, who also claimed allegiance to the
laws of the Icilombo, spared his life out of respect for his close
relationship to the ndonga position whose occupants she regarded as
allies. This entitled the imprisoned kabulcu lea ndonga to the rights of
blood-brotherhood which united all adherents of the kilombo. She
would not, however, release him to the Portuguese.69
Kabulcu ka Ndonga’s people, deprived of their legitimate leader,
chose as their new ruler his brother-in-law who had held the appoint­
ive w/wga-position of funji a musungo,70 one of the war leaders of the
band. Since his predecessor was known to be alive, he could not
claim full rights to the position and ruled as a regent through the
support of his wife, a daughter of the original Ndonga named
Kwanza, whom the Imbangala regarded as the rightful guardian of
the position.71 The Portuguese probably encouraged the choice of the
funji a musungo, since it gave them the opportunity to install a title-
holder only remotely qualified for his position and thus dependent on
them for support.
The new kabulcu ka ndonga, like the kulashingo, acknowledged his
debt to the Portuguese by enlisting his people in another military
campaign. Despite his lack of royal qualifications, he proved himself
an effective military commander in this attack, probably in part
because the assault on Nzinga gave his people hope of rescuing their
lost chief, then still alive but captive in Nzinga’s lands. The rescue
attempt failed and the old king eventually died in Matamba. The new
kabulcu ka ndonga then remained very close to his European sponsors,
fighting in battles against the Dutch and their African auxiliaries
during the late 1640s,72 against the ndembu chiefs in 1648,73 and finally
in 1648-9 against Panji a Ndona, the successor of the mani Kasanze
68The move o f the ngola a kiluatije to the Wamba was that mentioned in all
Imbangala traditions as taking place immediately after the arrival o f Kula­
shingo in Angola. On Nzinga, Joseph C. Miller (1975).
00 Cadomega (1940-2), i. 349-54.
70The title translates roughly as ‘sustenance o f the army’; funji is manioc
porridge and musungo is an appointive war leader o f the type used by the
Imbangala.
71 Cadornega (1940-2), i. 240.
72Ibid., i. 463-4, 490-1.
73Ibid., ii. 66-7.
 M A T U R E P O R T U G U E S E - I M B A N G A L A A L L IA N C E 205
near Luanda.71*74 The Portuguese honoured his fidelity by awarding
him the title of ‘Jaga’ and by condescendingly referring to him by the
possessive form, ‘our Jaga’.
The death in 1652 or 1653 of the kabuku ka ndonga who had fought
so loyally for the Portuguese ended for a short while the period of
close co-operation with the Europeans. His successor, who probably
had a better claim to legitimacy in the eyes of his own people, soon
abandoned the Portuguese to join Nzinga under the banner of the
kilombo. The Portuguese retaliated with a military expedition in 1655,
captured the chief, his wife (still with the name Kwanza), and all of
the major officials of the kilombo.75 Although later chroniclers
claimed that the governor forgave this kabuku ka ndonga for his way­
ward behaviour, contemporary documents show that the governor
sent him and his followers to Brazil as slaves, in keeping with the
custom of many seventeenth-century officials. The kabuku ka ndonga
who received a pardon was a later incumbent, evidently another
Portuguese puppet in the position.76
What had by that time become a standard Portuguese strategy of
installing pliable incumbents in sensitive positions again produced the
desired results. The new kabuku ka ndonga fought loyally against
several Ndongo chiefs led by Ngoleme a Keta during the rule of
governor João Fernandes Vieira (1658-61).77 The dependence of
later holders of the kabuku ka ndonga title on the Portuguese gradu­
ally increased and they simultaneously abandoned the laws of the
kilombo entirely. One incumbent finally admitted a pair of Carmelite
missionaries to his kingdom and accepted Christian baptism in the
1670s.78
By the 1680s, the kabuku ka ndonga had become a model of the
kind of affiance which Portuguese governors from the time of the
kulashingo onward tried to establish with neighbouring Imbangala
rulers. They favoured incumbents with an authentic Imbangala title
who owed their position to the Europeans rather than to the support
71 Ibid., ii. 25-6. See also Miller (1972b), pp. 53-5.
75 Consulta do Conselho Ultramarino, 13 September 1677 (A .H .U ., Angola,
cx. 2) refers back to these events but gives no date.
7" Compare Cadornega (1940-2), ii. 75-9, 498, with Consulta do Conselho
Ultramarino, 13 July 1655; A .H .U ., Livro 1 de Consultas Mixtas, foi.
187 (Códice 13). The defeated Kasanze o f 1622 had suffered the same
fate.
77 Cadornega (1940-2), ii. 164-5. Cadornega also had him fighting on the
Portuguese side against the count of Sonyo in Kongo during the early 1670s;
see ibid., ii. 278-80.
78 Ibid., ii. 426.
 206 T H E IM B A N G A L A AND THE PORTUGUESE
of their own people.79 Such kings enlisted their subjects as mercenaries
in Portuguese armies whenever officials in Luanda requested their
aid. The value of their titles necessarily declined under such circum­
stances so that the kabuku ka ndonga, for example, completely
abandoned their Imbangala position in the course of the eighteenth
century. In an echo of the familiar Mbundu pattern of changes in
titles to reflect new sources of legitimacy, the title then acquired a new
patronym, becoming Kabuku ka Mbwila, known thereafter to the
Portuguese as the ndembu Kabuku.80 The change indicated that the
kabuku had transferred their allegiance to the most powerful local
system of political titles, the neighbouring ndembu positions of
southern Kongo.81
The history of the Lunda kota position, Kalanda ka Imbe, or
Kalandula as he became known, paralleled that of the kabuku ka
ndonga in providing another example of how much Imbangala kings
north of the Kwanza depended on the Portuguese for the security of
their positions. The occupants of the kalanda ka imbe had abandoned
the kulashingo when he left for the interior and, like the kabuku ka
ndonga, had settled near Portuguese Angola. According to tradition,
an incumbent named Kashita (not otherwise identified) swore
allegiance as a Portuguese vassal, the ‘Jaga’ Kalandula, during the
conquest of Lulcamba.82 The kalanda ka imbe could have provoked
the kulashingo’s departure for the interior by convincing the Portu­
guese that he, as a legitimate leader of the kilombo and successor to
the kalanda ka imbe in charge of the band as late as 1601, might lead
the Imbangala in the service of Portugal more effectively than the
kulashingo. A Portuguese switch to co-operation with the kalanda ka
imbe might explain the documentary reference to Vasconcelos’s
‘abuse’ of the Imbangala, that is, abuse of the followers of the kula­
shingo. It would also explain the kulashingo’s decision to take the
kinguri title and seek his fortune in the far interior, leaving control
over the kilombo to the kalanda ka imbe and others who had gained
the favour of the Portuguese from which the kulashingo had tried to
exclude them.

79 Ibid., passim.
80 ‘Noticias do paiz de Quissama . . . ’ (1844), p. 124.
81 A .H .A ., Códice 240, passim. The Mbwila was the most powerful of the
numerous ndembu chiefs north o f Ambaca.
82 José Maria Mergú (Capitão-chefe o f Ambaca) to Gov. Geral José Rod­
rigues Coelho do Amaral, 1 Dec. 1856; Boletim Oficial de Angola, no. 585
(13 December 1856). This independent tradition agrees with other sources in
placing the Imbangala near Lulcamba in 1617-18.
 M A T U R E P O R T U G U E S E -IM B A N G A L A A L L IA N C E 207
The kalanda lea imbe, in any case, settled as the ‘Jaga’ Kalandula on
the frontiers of Ambaca where he guarded against the ndembu chiefs
to the north-west and against Nzinga’s Mataraba to the east. His
position there made him another in the line of Imbangala client
states which ringed Portuguese territory to the north. During the
1640s, Kalandula faithfully fought the Dutch in the company of the
Portuguese and Kabuku lea Ndonga.83 Much of this warfare centred
on control of a major trade route which ran from Nzinga’s kingdom
of Matamba through the ndembu region to the Dutch traders active
on the coast north of Luanda. Most of the slaves who left Angola
during the Dutch occupation came down this trail, and the Portu­
guese, Dutch, and their respective African allies fought continuously
to control it.
The kalandula abandoned the Portuguese only once, when he went
over to Nzinga with the Icabuku ka ndonga in 1653.84 Since the
kalandula occupied an important position along the Portuguese nor­
thern defences, the governor in Luanda immediately tried to win him
back.85 The Portuguese could not defeat him as they had crushed the
Icabuku ka ndonga, nor did an opportunity arise to manipulate the
succession to his title, but they eventually employed diplomatic
methods to regain his services. The terms of a 1656 treaty, in which
Nzinga formally renounced her thirty-year enmity against the Portu­
guese, required that she return the kalandula to Portuguese vassal-
age.86 The later kalandula, often in association with the Icabuku ka
ndonga, repeatedly sent their army to fight in Portuguese wars.
The location of the two former Imbangala chiefs’ lands on the
north side of the Lulcala above Ambaca87 kept them dependent on
Portuguese support since it left them in an exposed position near the
powerful ndembu and Kongo chiefs to the north. The kings of Kongo
83 Gov. Sousa Chichorro to el-Rei, 8 Dec. 1656; A .H .U ., Angola, cx. 4.
84 Cadornega (1940-2), ii. 75-9; Cavazzi (1965), ii. 33-4, said that Nzinga
defeated Kalandula . He gave the date, wrongly, as 1657.
85 Luis de Sousa Chichorro to el-Rei; 29 July 1656; A .H .U ., Angola, cx. 4.
8CCapitulos e Pazes que fas a capitão manoel frois peixoto Como embaixador
nesta corte da Rainha Donna Anna de Souza por mandado do senhor govern­
ador e capitão geral destes Reinos Luis Martins de souza chichorro Retificados
pelo capitão Joseph Carrasco; A .H .U ., Angola, cx. 4; published in AA,
sér I, ii, nos. 7-8 (1936), 9-14; Cavazzi (1965), ii. 332-3 (doc. 46). See also
letters from Luis de Sousa Chichorro, 14 Oct. and 8 Dec. 1656; A .H .U .,
Angola, cx. 4.
87 Cadornega (1940-2), iii. 244-5. The exact location o f the kalandula's
lands probably changed from time to time. The territory was known as Kitukila
in 1656 and lay on the boundary o f the lands o f Nzinga to the north and of
Ndongo to the south.
208 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
killed at least one kalandula in 1658 as part of a general harassment of
chiefs loyal to the Portuguese.88 Other kalandula fought against the
ndembu Nambo a Ngongo in the 1660s,89 accompanied the Portu­
guese expedition to Sonyo (in Kongo) under João Soares de Almeida
in 1670,90 and again against the ndembu Mbwila in 1693.91 The
Portuguese granted the title of ‘Ngola a Mbole’ or ‘Kyambole of the
Portuguese king’ to the kalandula and provided them with arms and
supplies iii return for their participation in many military expeditions
throughout the eighteenth and even into the nineteenth century.92
Nzinga, ruler of Matamba and pretender to the title of ngola a
kiluanje after the defeat of the Ndongo kingdom of the same name,
adopted the rites of the kilombo in the 1620s and considered herself
Imbangala. But her kingdom developed in ways quite atypical of the
other Imbangala states elsewhere in northern Angola since she was
able to maintain a much more consistent opposition to Portuguese
activities in Angola, a record equalled only by such southern Imban­
gala kings as the kidembe and Kakonda.93 Matamba turns out to be
the exception which illuminates the fundamental reason why most
Imbangala Icings maintained such close relations with the Portuguese.
Nzinga, alone among the Imbangala rulers of the north, claimed
political authority (i.e. certain local titles in Matamba) derived from
the indigenous Mbundu system of titles. Her local sources of legiti­
macy, although never secure, allowed her to command her own
people with marginally more security than the alien Imbangala
bands and thus to behave more independently than the holders of
exotic Lunda titles who never gained the confidence of the lineages
they claimed to rule.
The economics of the slave trade also contributed to Nzinga’s
independence of Portuguese control before 1656. The trade route
which developed during the 1630s from Matamba through the
88 Gastão Sousa Dias (1942), pp. 39, 103, 105,
89 Cadornega (1940-2), ii. 191—6; Kalandula was not mentioned by name, but
there is little doubt that he was the ‘Jaga’ referred to.
90Feo Cardoso de Castello Branco e Torres (1825), pp. 202-3; also Levy
Maria Jordão, visconde de Paiva Manso (1877), p. 254.
91 Gonçalo da Costa de Meneses to el-Rei, 25 Apr. 1693; A .H .U ., Angola,
ex. 11. Again, the kalandula was not mentioned by name but the identification
is quite certain.
92Numerous references in A .H .A .; e.g. Gov. D . Antonio de Lancashe to
D . Francisco Agostinho Rebelo, Jaga Callandulla, 14 Aug. 1775; A.H.A.,
Códice A -14-4, fols. 89—90v ; also Termo de juramento e vassallagem a Sua
Magestade que presta o Jaga Calandulla, D . Manoel Affonso, como abaixa se
declara, 20 Oct. 1870 (Boletim Oficial de Angola, no. 50 (10 Dec. 1870), 706).
93 See section following.
 M A T U R E P O R T U G U E S E - I M B A N G A L A A L L IA N C E 209
nckmbu towards the Dutch on the coast ensured her access to Europ­
ean trade goods regardless of her relations with governors in Luanda.
She competed for slaves with the Iculashingo’s neighbouring kingdom
in Kasanje, and their rivalry had reached the stage of armed conflict
just before the Dutch invaded Luanda in 1641.94 The Dutch occupa­
tion of Angola cleared the way for large-scale exports of slaves from
Matamba and fuelled Nzinga’s rise to the stature of the most power­
ful ruler in the interior during the 1640s. She established a virtual
monopoly over the slave trade from the interior during that decade.95
Her ascendency lasted, however, only until the Portuguese expelled
her European allies, the Dutch, in 1648 and reopened trading with
Kasanje during the 1650s. The shift in the slave trade to Kasanje
prompted her to reach a reconciliation with the Portuguese in 1656.
The kingdom of Kasanje under the Iculashingo, like Matamba under
Nzinga, claimed a connection with the Imbangala Icilombo but found
its ultimate sources of legitimacy in the kinguri title brought from
Lunda. In the case of Kasanje, the successors of the Iculashingo
preserved a degree of independence from Portuguese influence during
the 1630s and 1640s through alliances with the mwa ndonje and some
of the Songo titles of Lunda origin. The slave trade became important
to Kasanje only after 1648 when renewed Portuguese hegemony near
the coast restored the second major slave trading network in Angola,
i.e. the olflcial one. Kasanje controlled the inland terminus of tins
system, which ran through Portuguese possessions to the south of the
parallel illegal trade route from Matamba to the ndembu and their
allies. The great distance which separated Kasanje from the seat of
Portuguese power in Luanda enabled the kinguri to pursue relatively
independent policies even after the new trading system linked them
once again to the Portuguese. In addition, the eastern Mbundu of
Kasanje seem to have had greater respect for the Lunda titles than
their western fellows, as the proliferation of related titles in Songo
had demonstrated. Kasanje also incorporated a variety of native
M bundu positions drawn from the Libolo area.96 All of these con­
ditions made Kasanje, like Nzinga’s Matamba, unrepresentative of

Cadornega (1940-2), i. 205, 207; also Cavazzi (1965), ii. 79.

1,5Report o f Pieter Mortamer, 1643; Archief van de Eerste West Indische
Compagnie, 46; S. P. Honore-Naber (1933). The Dutch claimed that they
bought 12,000 to 13,000 slaves per year during the early 1640s, nearly all of
them from Nzinga. I am indebted to Dr. Phyllis Martin for pointing this
reference out to me.
M These and other points related to the foundation of the Kasanje kingdom
will appear in future studies based on my 1969-70 research in Angola.
210 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
the more common case of Imbangala bands closer to the coast who
depended heavily on Portuguese assistance for their survival.

The Imbangala South o f the Kwanza

The close, almost symbiotic, relationships between the Imbangala
and the Portuguese north of the Kwanza river contrast strikingly
with the barely broken record of hostility which the warriors of the
kilombo showed to Europeans south of the river. An explanation for
the difference must be sought in terms of the Mbundu and Ovimbun-
du social backgrounds against which the respective groups operated.
In so far as the southern kilombo bands have been revealed in the
written records, the Imbangala bands on both sides of the river had
similar political structures composed primarily of institutions derived
from the kulembe and the vz/nga-positions, both of which had origin­
ated among the Ovimbundu. If, as the histories of several other types
of title and symbol among the Mbundu have suggested, the conse­
quences of a political innovation depend on its compatibility with the
social institutions in which it takes root, the Ovimbundu-derived
kilombo should have found a more cordial reception among the
peoples living south of the Kwanza than among the alien lineages of
the Mbundu. Whereas the hostility of the Mbundu lineages had
driven most northern Imbangala into the arms of the Portuguese, the
congeniality of the kilombo among the Ovimbundu allowed the
southern Imbangala to remain fiercely independent of European
interference.
Imbangala opposition to Europeans south of the Kwanza did not
derive from a want of Portuguese attempts to duplicate their arrange­
ments with the Iculashingo among the kilombo bands who had re­
mained among the Ovimbundu. Manuel Cerveria Pereira, the former
governor in Luanda who had made such excessive use of the Imban­
gala from 1615 to 1617, attempted to extend his alliance with them to
the south side of the Kwanza when he left office in 1617. He mounted
an expedition which sailed south towards the Kunene in search of
Imbangala bands. His venture foundered almost immediately on a
decided lack of Imbangala interest, a problem which became a
perennial hindrance to nearly all later Portuguese attempts at
accommodation south of the Kwanza. Cerveira Pereira intended to
use the Imbangala to promote Portuguese penetration of the lands
behind a second seaport at Benguela Velha (near the mouth of the
Kuvo), just as the Iculashingo had opened the way to the Mbundu
interior. He had good reason to expect to 'find Imbangala there,
 T H E M IB A N G A L A SO U T H OF T H E K W A N Z A 211
since they had been reported in that vicinity at least as recently as
1601.
Cerveira Pereira failed to encounter Imbangala with whom he
could initiate a profitable slave trade near Benguela. On account
of the absence of African allies, and because of the extreme aridity of
the land, he continued southward until he discovered a group of
Imbangala camped near a river which he called the ‘Murombo’.97 He
convinced these Imbangala that they should raid the local inhabitants
for captives which they would exchange for Portuguese trade goods,
much as the kulashingo’s Imbangala had traded with him near
Luanda. This agreement did not work out as well as the Portuguese
had hoped, since these Imbangala co-operated only to the point of
selling their captives during the day, but then stole them back as soon
as night fell. The Imbangala king (whose name was not recorded)
also sheltered other slaves who had escaped from the Portuguese and
eventually collected about thirty refugees in his camp. When Cerveira
Pereira accused the Imbangala chief of duplicity, the kilombo leader
denied any knowledge of the escaped slaves and threatened to attack
the Portuguese if they did not treat him with more respect. His
independent attitude so provoked Cerveira Pereira that the Portu­
guese commander attacked the Imbangala encampment to recover
his losses. The European forces twice failed to storm the hill where
the Imbangala had built a fortified refuge before some eighty men and
two horsemen managed to penetrate the defences. The Portuguese
won the ensuing battle, captured the chief, beheaded him, and took
back most of the trade goods and slaves which they had lost. They
also enslaved as many Imbangala warriors as they could capture.
Cerveira Pereira shortly afterward attempted another slave-trading
alliance with a different Imbangala band but experienced no greater
success. In this case a chief, ‘Ka Ngombe’,98 at first accepted the
partnership but broke away as soon as he realized that Cerveira

97 Probably not the Balombo; see Map X.

98 N ot the same ICangombe given by Capello and Ivens as the fourth chief
of Bihe (1882, i. 158); Gladwyn M. Childs (1970), p. 245, placed his reign at c.
1780-1805. The form of the name suggests a missing initial element, as in
(Kalunga) ka Ngombe, since ngombe means ox and occurs as a surname with
some frequency south o f the Kwanza. Alternatively, ‘Kangombe’ may have
been the mwa Cangombe, one of the Lunda makota o f the ICandama ka Hite
lineage who had apparently lost control o f the Imbangala to Mbongo wa Imbe
and Kalanda ka Imbe somewhere in this part o f Angola. The defeat and
possible expulsion o f the title from the band (not recorded in the traditions)
could have caused the name to remain there after the other makota continued
on their way north.
212 T H E IM B A N G A L A AND THE PORTUGUESE
Pereira intended to make himself the dominant partner. Cerveira
Pereira, perhaps wiser from his experience with the Imbangala near
the ‘Murombo’, tried to limit ‘Ka Ngombe’s’ potential for trouble­
making by stationing him in an out-of-the-way spot and limiting his
forces to only fifty men. ‘Ka Ngombe’ recognized the danger, how­
ever, and fled to the hills under the excuse that he had gone to search
for more booty. He stole some of the Portuguese cattle as he departed
and took them to his camp on an inaccessible hilltop. The Portuguese
managed to storm and sack the camp but then beat a hasty retreat to
their main base near the coast under continual harassment from the
regrouped Imbangala." ‘Ka NgombeY Imbangala, like their counter­
parts near the ‘Murombo’, could not have recovered from an initial
defeat to mount sustained resistance without the support of local
residents. Cerveira Pereira’s expectations of a profitable trade had
not been fulfilled.
The only southern Imbangala bands who co-operated with the
Portuguese on the same basis as those north of the Kwanza were a few
who lived near Benguela Nova during the 1620s, notably the kilombo
of ‘Angury’100 and Kapingena. It was their willingness to provide
slaves to the Portuguese that distinguished them (like most Imban­
gala) from non-Imbangala chiefs; the Portuguese noted explicitly
that they traded with the Ovimbundu primarily for foodstuffs and
other supplies but bought only slaves from the Imbangala.101 Neither
‘Angury’ nor Kapingena appeared by name in documents again,
after the 1620s, although they may have been among the unidentified
Imbangala auxiliaries who aided the Portuguese during the wars of
the 1650s near Benguela.102 Another Imbangala king in the same
region, Kashana, offered his support when a Portuguese relief
expedition stopped in Benguela in 1645 on its way from Brazil to aid
the beseiged Portuguese who had fled to Massangano after the Dutch
capture of Luanda.103
55Representação o f Manuel Cerveira Pereira, 2 M y 1618; A .H .U ., Angola,
cx. 1, doc. 141; published in Brásio (1952-71), vi. 315-19; also in Cordeiro
(1881), iii.
100Nguri? There is no probable relation to the kinguri except that both
names may have come from the Wambo word for lion.
101Relação of Fernão de Sousa, 22 Apr. 1626; A .H .U ., Angola, cx. 2;
published in Brásio (1952-71), vii. 436-8. This distinction lasted until at least
the end of the eighteenth century; see J. Pinheiro de Lacerda (1845), p. 488.
102 Cadornega (1940-2), ii. 43. Cadornega elsewhere (ii. 250) identified the
‘Jagas’ o f Benguela as Kabeto, Kalunga lea ICingwanza, ICasindi, and Ngulu.
103Relação da viagem de Sotomaior em socorro de Angola; B.N .M ., MS.
8187, fols. 37—60v; published in Artur Veigas (1923), 18-23; also A A, sér II,
i, nos. 3-6 (1943-4), 145-53; Brásio (1952-71); ix. 374.
 T H E IM B A N G A L A SO U T H OF T H E K W A N ZA 213
The eagerness of Kashana and the other kilombo chiefs in this area
to establish friendly relations with the Portuguese offers a glaring
exception to the hostility of other Imbangala south of the Kwanza
and indicates that some powerful enemy may have driven those who
lived near Benguela to seek alliance with the Europeans. The most
likely source of their anxiety was another Imbangala king, Kakonda,
who lived beyond the mountains east of Benguela. Kakonda con­
sistently fought any European penetration of his lands throughout
the seventeenth century and thus fits the normal pattern of Portuguese-
lmbangala relations south of the Kwanza. The Portuguese tried
repeatedly to penetrate his stronghold but failed except for the
construction of a short-lived post there in 1684. Kakonda drove them
out the following year, however, and the wars resumed and con­
tinued until well into the eighteenth century.104
To the north of Benguela, a number of Imbangala bands identifi­
able as related to the kulembe constituted an impenetrable obstacle
to Portuguese advances beyond the coastline. The Portuguese fought
them intermittently at least from the 1620s, when a report described
the region as full of fierce and warlike Imbangala who frustrated the
hopes of governors who had tasted the fruits of their alliance with the
Imbangala north of the Kwanza and wanted to duplicate their
successes in the south. These southern Imbangala refused either to
trade with the Portuguese or to pay tribute. They lived in fortified
caves where they stored water and food to withstand seiges.105 The
kulembe had established a strong kingdom on the upper Longa river
after the munjumbo had driven him out of the mountains to the south
and had become pre-eminent among them. His new domain extended
towards the coast where the holder of a subordinate vunga-title, the
sungo dya kulembe, ruled the Sumbi people who lived near the mouth
of the Kuvo.106 The central kulembe lands, and those of two sub­
ordinates, Nambo a Mbungo and Lunga dya Kafofo, lay beyond the
southern borders of Halco and Libolo.
These Imbangala kings regularly made life miserable for any local

104 See, for example, Governador de Angola D om João de Lancastre, 3

Apr. 1688 (copy of 20 Apr. 1690); A .H .U ., Angola, cx. 11; also Cadomega
(1940-2), ii. 176-9.
105 Relação da Costa de Angola e Congo, pelo ex-Governador Fernão de
Sousa, 21 Feb. 1632; Ajuda, 51-VIII-3, fols. 11—18v ; published in Brásio
(1952-71), viii. 129.
106 Cadomega (1940-2), iii. 249. The sungo was probably a ‘war leader’ (the
musimgo)-, cf. the title of the ftm ji a musungo among the Imbangala of Kabuku
lea Ndonga, pp. 204-6 above.
214 T H E IM B A N G A L A AND THE PORTUGUESE

sMnssangano
Combnmbo
l ib o l o
RIM BA HAKO
✓Ngonga a . Lunga d y a '
Kahanga ? K a fo fo
(1 0 5 0 s)
Benguola'
Velha C y / M u n i d y a
(1617) / Ngombe
/ ^
Quicombo

Movement ol the K a p in g e n a ?
Portuguese expedition /ie o n « ,\
of 1045 (IbZU S )
- Attacks by
Sungo dyn Ktilembc K akonda
50 100 150 km
1
L_U------------- _______ |

M ap x. Imbangala South of the Kwanza (Seventeenth Century)

chiefs who submitted to European blandishments, and most Portu­

guese records from this area chronicle little more than their conflicts
with the kulembe and his subordinates. The kulembe’s Imbangala
raided chiefs loyal to the Portuguese in the early 1640s and fought an
army commanded by Diogo Gomes de Moraes in 1643.107 These
battles showed that the Portuguese armies could defeat the southern
Imbangala on occasion, but their aftermath emphasized that they
could not exterminate them; in classic guerilla style, the kulembe
resumed his raiding of the loyal chiefs as soon as Moraes’s army had
left. The next year, another Portuguese expedition went out from
Massangano to fight the kulembe once again.108 His persistent

107 Cadornega (1470-2), i. 312-13. See also the ‘Catalogo dos governadores
do Reino de Angola’, published in Feo Cardoso de Castello Branco e Torres
(1825), p. 175; and Mattos (1963), p. 275.
108 Cadornega (1940-2), i. 344-6, 354.
 T H E IM B A N G A L A SO U T H O F T H E K W A N Z A 215
resistance to the Portuguese had by that time given him a reputation
as the most powerful Imbangala chief south of the Kwanza. One
governor referred to him, with some exaggeration, as ‘that great Jaga
Lulembe [sic] who has conquered from here to Mozambique’,109 and
Mbundu traditions from that time depicted him in the same glowing
terms.110
Nambo a Mbungo and Lunga dya ICafofo, the kulembe’s provincial
governors, continued fighting against Portuguese penetration in
later years, turning up on the side of the rebellious Kisama chief
Kafushi lea Mbari in the 1650s.111 Another Imbangala chief in the
Rimba province of southern Libolo, Ngonga a ICahanga, opposed
the Portuguese at about the same time, but he apparently suffered an
unrecorded defeat at some time since he eventually declared himself
a vassal of the Portuguese crown.112 Other hostile Imbangala,
unidentified, but very likely connected with the kulembe, roamed the
province o f ‘Gemge’113 which lay east of the Sumbi. The extensive and
consistent Imbangala opposition to European activities in this region
contrasted dramatically with the relative docility of the Imbangala
kings among the Mbundu.
The best-known example of the kulembe’s opposition to European
penetration of the area south of the Kwanza occurred when the
western branch of his Imbangala defeated the 1645 Portuguese
expedition sent from Brazil to relieve the Portuguese forces trapped
by the Dutch in Massangano. The commanders of the expedition,
Teixeira de Mendonça and Lopes Sequeira, landed their forces near
the mouth of the Kilcombo river. From there, they intended to march
overland towards Massangano across territory controlled by the
kulembe. Their choice of landing site put them at an initial disadvant­
age, since it forced them to make the same difficult crossing over the
Kuvo river which had stopped the kalanda lea imbe’s Imbangala
nearly a half-century earlier. Nzamba, the ruler of friendly Sumbi
living south of the Kuvo, had made an offer of aid which probably
influenced the Portuguese strategic planning. The known hostility of
the kulembe’s subjects farther north and the availability of African
allies, still a major concern to the Portuguese everywhere in Angola
near the Kikombo may have made this the best of the available alter­
natives after all.
105Relação da viagem de Sotomaior; Brâsio (1952-71), ix. 374.
110 Cavazzi (1965), i. 188-90.
111 Cadomega (1940-2), ii. 90-1, 103.
112Cavazzi (1965), i. 207.
113Njenje? N ot identified.
216 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
Teixeira de Mendonça expected to obtain additional help in this
area from an Imbangala chief who had fled from his original lands
north of the Kuvo to the southern bank of the river after a dispute
with other Imbangala. This chief, Muni dya Ngombe, had good
reasons for seeking an alliance with the Portuguese, since he hoped to
secure their aid in a planned return across the Kuvo to take revenge
on his enemies. Muni dya Ngombe had therefore promised to help
the Portuguese cross the river and subdue the lands between there and
Massangano, which he accurately pictured as filled with potentially
hostile Imbangala. His scheme fit the pattern of friendly Portuguese-
Imbangala relations which Cerveira Pereira and others on the
Portuguese side had pioneered, and which Teixeira de Mendonça had
already witnessed in Benguela. The Portuguese plan to relieve
Massangano depended, as had so many others of their designs, on
Imbangala co-operation. The co-operation of these mercenaries
loomed important enough in the minds of these commanders to out­
weigh the other disadvantages of their chosen strategy.
Although this joint Portuguese-Imbangala project began auspici­
ously, the enemy Imbangala once again demonstrated their ability to
regroup in the friendly territory south of the Kwanza and in the end
overwhelmed the united forces of the relief expedition. The men of
Muni dya Ngombe and Nzamba constructed a bridge across the
Kuvo as agreed and when they completed it the combined Portu­
guese, Sumbi, and Imbangala force began the trip northward towards
Massangano. They conducted an initial victorious campaign against
Imbangala chiefs who opposed, as predicted, their advance to the
north bank of the river. Then they marched in the direction of
Massangano until they encountered a second group of Imbangala,
veterans of the kulembés battles against Gomes de Moraes in the
previous year. These Imbangala, drawing on the experience gained in
earlier confrontations with the Portuguese, joined those defeated near
the Kuvo, massed their joined forces in a stone fortress facing a river,
and administered a crushing defeat to the Portuguese. Muni dya
Ngombe’s Imbangala deserted at the first sign of the impending loss
and left the Portuguese to fight the battle alone; 103 out of 106
Portuguese died.114
Only later in the 1670s did the Imbangala of the kulembe recognize
the advantages to be derived from participation in the European
slave trade. They began to seek contact with the Portuguese for
purposes of limited commerce during that decade. These arrange-
114 Relação da viagem de Sotomaior; Brásio (1952-71), ix. 374.
 T H E I M B A N G A L A SO U T H OF TH E KW A N ZA 217
ments began when an Imbangala chief called Kasambe attacked the
Sumbi ruler of the Kilcombo region.115 A port had developed near the
mouth of the river where Portuguese ships took on ivory and slaves
obtained from the peoples who lived upriver. The Sumbi near the
coast had established themselves as middlemen in this trade between
the kulembés Imbangala, who up to that time had remained in the
interior, and the Portuguese who stayed near the coast. Kasambe’s
attack indicated that the kulembés people had ended their aversion to
direct commercial contacts with the Portuguese and now desired to
capitalize on the opportunities offered by the trade in slaves. They
therefore wished to bypass the obstacle represented by the middlemen
Sumbi of the Kilcombo. The Sumbi chief resisted the Imbangala
attempt to eliminate his position in the trade and also defeated a
Portuguese expedition commanded by Gaspar de Almeida, thus
preserving for the time being the established commercial pattern and
his own profits.116
Other Imbangala who lived near Libolo, the only Mbundu regions
south of the Kwanza, at first fought the Portuguese with a determina­
tion equal to that of the others on that side of the river. But they
seemed unable to maintain their opposition whenever they entered
the regions inhabited by Mbundu. Descending from their homes in
the Ovimbundu region known as Tunda,117 they destroyed crops and
raided Hako and Libolo chiefs who had declared their loyalty to the
Portuguese at least since the 1620s, perhaps out of fear of their war­
like southern neighbours. The Tunda Imbangala apparently gained
control of the area shortly after 1620 and seriously damaged Portu­
guese interests by disrupting the slave trade which ran north through
Cambambe to the Portuguese posts just beyond the Kwanza. The
bishop Mascarenhas, interim governor of Angola in 1623-4, sent an
expedition under Lopo Soares Laço to bring the Libolo border with
115His name was spelled ‘Cacabe’ or ‘Caçabe’; Kasambe is the most likely
reconstruction. At some earlier date, the Portuguese had come into conflict
with Kasambe and had defeated him. Cadornega (1940-2), i. 365, mentioned a
German active in Angola who was known as ‘Casabe’ because he had fought
and beaten a ‘Jaga’ of that name somewhere south o f Libolo.
116Cadornega (1940-2), ii. 291-4. Although the Imbangala o f the Iculembe
influenced the customs and institutions of several o f the later Ovimbundu
kingdoms, they established permanent dynasties in only one or two instances,
probably Ciyalca and Wambo; Childs (1940, p. 188; 1964, p. 374).
117For the location o f Tunda, see Relação da Costa de Angola e Congo;
Brásio (1952-71), viii. 121; and Bispo D . Simão Mascarenhas to el-Rei, 3 Feb.
1624; Brásio (1952-71), vii. 199-203. I have followed Brásio’s interpretation
o f the extremely difficult script of this document (cf. Delgado’s reading in
Cadornega (1940-2), i. 113, n. 1).
218 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
Tunda back under Portuguese authority. Lopo Soares Laço fought
against the Imbangala chief, Nzenza a Ngombe, defeated him, and
captured others, among them one called ‘Bango-bango’.118
This victory temporarily brought the Imbangala and Portuguese
together in common opposition to the local chiefs, on the model of
the Imbangala north of the river at about the same time. The local
Hako and Libolo, like the Mbundu on the other side of the Kwanza,
but unlike the Ovimbundu to the south, consistently opposed which­
ever outside invader, Portuguese or Imbangala, posed the greater
immediate threat to their autonomy. They had initially allied them­
selves to the Europeans out of fear of Nzenza a Ngombe, but when
Nzenza a Ngombe’s defeat tipped the balance of power locally in
favour of the Portuguese, the local chiefs immediately switched sides
and began to follow the leadership of Kafushi lea Mbari in resistance
to the Portuguese. Fernão de Sousa, the next governor in Angola,
completed the cycle of political realignments by employing his former
enemy, Nzenza a Ngombe, to bring Kafushi ka Mbari and his
rebellious allies back to the side of the Portuguese.119 This strategy
enabled de Sousa to control the region without great expense at a
time when Dutch threats to Luanda demanded the concentration
of all Portuguese forces near the coast. From the Imbangala point of
view, alliance with the Portuguese gave them security in the midst of
Mbundu lineages hostile to the kilombo and protected them from the
growing power of the descent groups who had rallied behind Kafushi
ka Mbari.
The later history of ‘Bango-bango’, who had been captured in
1624, underlined the contrast between Imbangala hostility on the
south side of the Kwanza and co-operation on the north bank of the
river. ‘Bango-bango’ settled north of the Kwanza in Uamba and gave
up the Imbangala way of life to become a loyal kilamba, or captain of
the African auxiliaries who fought in Portuguese armies.120 He, like
118Bispo D . Simão Mascarenhas to el-Rei, 3 Feb. 1624; Brásio (1952-71),
vii, 199-203. The document does not make it clear whether Laço fought with
or against Nzenza a Ngombe. Silva Correa (1937), i. 238, and Feo Cardoso
de Castello Branco e Torres (1825), p. 164, specify that he opposed Nzenza
a Ngombe and defeated him. Delgado in Cadornega (1940-2), i. 113, n. 1,
added an Imbangala chief ‘Cazanga’ among those defeated; this seems to be an
error, as the document reads ‘Caça . . . ’ and probably refers to ICaza, then in
disrepute among the Portuguese (p. 219 below).
119 Fernão de Sousa to el-Rei, 22 Aug. 1625; A .H .U ., Angola, cx. 2, and
Ajuda, 51-VIII-30, fols, 321-32F ; Brásio (1952-71), vii. 359-68.
120Technically, any African authority who received lands within Portuguese
territory by grant from the governor rather than by hereditary right (the term
contrasted with that o f soba, which designated indigenous title-holders); most
 T H E IM B A N G A L A SO U T H OF T H E KW ANZA 219
the kabuku ka ndonga, and the kalandula, gave loyal service to the
governors of Angola for many years, notably against the Dutch in
the 1640s. He feigned desertion to the Dutch in 1641 and caused them
considerable damage before escaping back to the Portuguese. The
king of Portugal later repaid his services with a mercê of the Order
of Christ; the former Imbangala ‘Bangobango’ accepted Christian
baptism as João Bango on that occasion.121 His movement to the
north side of the Kwanza had deprived him of the local support on
which the southern Imbangala depended; this converted him to a
supporter of Portuguese advances against the Mbundu.
The movements of the kaza, a title originally subordinate to the
hango of Libolo, illustrate the security which Imbangala chiefs found
south of the Kwanza. The kaza ka hango, a perpetual title granted to
1V1bundu lineages near the former centre of Libolo, belonged to people
living somewhere opposite the Portuguese post at Cambambe.122
The holders of this title had adopted the kilombo and become
quite powerful in their own right before 1620, when the Portu­
guese respected them as some of the more important Imbangala
chiefs of that region. A kaza ka hango joined the Portuguese during
the period of their military successes in the company of the Icula-
shingo, crossing the Kwanza to fight near Massangano on the side of
a Portuguese army under captain Luis Gomes Machado. The
Imbangala king then refused to co-operate further, or ‘rebelled’ in
Portuguese terms, and settled his people in Ndongo. Another
Portuguese military expedition drove the kaza ka hango out of
Ndongo in 1621, showing that even the most powerful Imbangala
bands had difficulty resisting the Portuguese outside their home
territory.123

121 Cadornega (1940-2), i. 237-8, 247-8, 463, 514. A subordinate position,

Malange a Bangobango, also existed; A .H .A ., Códice D-20-1, fol. 133v ;
published in Dr. Carlos Dias de Coimbra (1953) and in Brâsio (1952-71), x.
59-62. Bangobango himself had aided the Portuguese cause in Angola for some
thirty years (i.e. since about 1620) and his ‘family’ tradition of service was said
to date from the time o f Paulo Dias de Novaes (1575?), long before the other
Imbangala had reached the coast. The other data available on Bangobango
show that such effusive praise exaggerated the true facts.
122Relação da Costa de Angola e Congo; Brásio (1952-71), viii. 121.
123 Cadornega (1940-2), i. 88-94; also PQ Mateus Cardoso, 16 Mar. 1621;
B.N.L., Cx. 29, doc. 26; published in Brásio (1952-71), vi. 566-9. cardosa helps
establish the date by mentioning that the governor had been absent in the
interior pursuing a long series o f wars.

yilamba fought with the Portuguese in exchange for their lands. The arrange­
ment amounted to a Portuguese version o f the vunga.
220 T H E IM B A N G A L A AND THE PORTUGUESE
The kaza fled back to the south side of the Kwanza and used the
firm basis of support he commanded there to take his revenge
throughout the 1620s. He first aided the ngola a kiluanje Mbande a
Ngola when he opposed Portuguese penetration of Ndongo and then
sided with the queen Nzinga in political manoeuvering after his
death. As part of his policy of alliance with Ndongo kings, the kaza
had agreed to protect Mbande a Ngola’s ‘son’ (a title-holder?) and
heir to the njola a kiluanje title from capture by the Europeans. After
Nzinga replaced Mbande a Ngola, however, the kaza handed this
‘son’ over to Nzinga, thus permitting her an opportunity to have him
assassinated in a bit of treachery highlighted in most Portuguese
accounts of Nzinga’s rise to power. From the kaza’s point of view,
his decision to turn over his hostage to Nzinga maintained a consist­
ent policy of support for whichever incumbent ngola a kiluanje
seemed likely to use the position to oppose the Portuguese. The kaza
remained in Ndongo only as long as Nzinga seemed capable of
holding off the Europeans, however, for a Portuguese victory in 1626
once again sent him in flight to the other side of the Kwanza. He at
first sent word from there that he wished to make his peace with the
Europeans, once and for all,124125but then broke all contact with the
Portuguese shortly after 1630. Imbangala traditions show that he
moved to the Baixa de Cassanje where he played an important part in
founding the state of Kasanje.
Finally, one aspect of Nzinga’s behaviour during the 1620s con­
firms the tendency of Imbangala Icings to flee south of the Kwanza in
time of distress. She had supplemented her position as ngola a
kiluanje with a symbolic marriage to the kaza125 which gave her the
position of tembanza (first wife) of the kilombo chief. This crucial
Imbangala office, heir to the functions attributed to the ‘Temba
Andumba’ of the traditions, entailed preparation of the maji a samba
and enabled Nzinga to assert leadership over the remnants of the
kulashingo’s Imbangala after they dispersed in about 1619. Her claim
to this position also explains the strong influence which she seems to
have exerted over the Icalandula and the Icabuku lea ndonga from time
to time during the 1640s and 1650s.
Her alliance with the Imbangala also served a strategic purpose by
providing her with a safe refuge near the southern Imbangala on the

124 Cadornega (1940-2), i. 142; also Relação do Governador de Angola, c.

1627-8; Ajuda, 51-VHI-30, fols. 247-60v ; published in Brásio (1952-71), vii.
526-7.
125 Cavazzi (1965), i. 259, and ii. 70-2.
 T H E M IB A N G A L A SO U T H O F T H E K W A N ZA 221

Kwanza river islands of Kindonga whenever Portuguese pressure

drove her out of her redoubts north of the river.126 She retreated to
these areas in and beyond the Kwanza several times during the 1620s
and also fled at least once (in 1629) towards the Imbangala of kula-
shingo, then already settled in the Baixa de Cassanje.127 Nzinga moved
north to the ancient kingdom of Matamba only after her strategy of
taking refuge among various Imbangala groups had failed to protect
her position in Ndongo.

Conclusions
The history of the Imbangala contacts with the Portuguese pro­
vides by far the best-documented example of state-formation among
the Mbundu. Imbangala mercenary armies formed the backbone of
the Portuguese expeditions which put Ndongo on the defensive and
then replaced Mbande a Ngola with puppet ngola a kiluanje in
Pungo Andongo, leaving Nzinga free to make a claim to possession
of the title from her new base in Matamba. Some of the makota who
left ICulashingo created new states as a defence line protecting the
Portuguese against hostile kingdoms to the north and east. A major
Imbangala state arose in Kasanje which, with Mátamba, became the
major supplier of slaves for the trade which supported the Portuguese
state in Angola until the middle of the nineteenth century. All the
kingdoms which emerged from the ashes of the Angolan wars,
Angola itself, Kalandula, Kabulcu, the later kings of Jinga, Kasanje,
Holo, and mwa Ndonje (as well as several Ovimbundu ldngdoms
south of the Kwanza) owed their origins to the rulers of the Icilombo.
These became the dominant states of eighteenth-century Angola,
completely replacing the earlier kingdoms of Ndongo, Libolo, and
Kulembe.
Both Portuguese and Imbangala represented similar challenges of
major proportions when viewed from the perspective of the Mbundu
kinsmen, whose ancestors had preserved the autonomy of their
descent groups against such diverse threats as mavunga title-holders
126Fernão de Sousa to Francisco de Castro, 8 Apr. 1628; Ajuda, 51-VIH-31,
fol. 171v ; published in Brásio (1952-71), vii. 549-50; also Relação do Govern­
ador de Angola; Brásio (1952-71), vii. 526-7.
127Cadornega (1940-2), i. 165-6. Cadornega’s description o f the battle
between Nzinga and the Portuguese leaves no doubt that it occurred in the
region o f the Lambo highland. Nzinga retreated down the trade route that led
to Yongo, where the kulashingo had settled, descended the escarpment of
Katanya near the capital o f the ndula kisua, crossed the Baixa de Kafushi, and
ascended the rocky spur o f Kabatukila on her way back to the islands o f
Kindonga.
222 T H E IM B A N G A L A A N D T H E P O R T U G U E S E
from Libolo and the centralizing kingdom of the ngola a kiluanje.
Neither of the new invaders had lineages of the type which the
Mbundu regarded as fundamental to human society. Both came as
aliens from far outside Mbundu territories. Neither derived their
living from agriculture, as did the Mbundu, and both stole or traded
for the produce of local farmers.
The Portuguese and Imbangala perceived their common interests,
at least in terms of economic and military expediency, if not in
full appreciation of their more subtle similarities in the eyes of
the Mbundu. They joined forces to establish a new set of slave­
trading states which in the most fundamental sense differed relatively
little in their impact on the Mbundu, whether they were run by
Europeans, as was Angola, or by Africans, as were all the rest.
The precarious positions of both parties north of the Kwanza
kept them in a firm alliance against the subversive desires of the
Mbundu lineages.
The Imbangala south of the river provided the test case which
validates this hypothesis, since the kilombo there thrived in the more
suitable social institutions of the Ovimbundu. Although we cannot
yet analyse in detail how Ovimbundu social structures tolerated the
kilombo in a way that the Mbundu descent groups did not, most
Imbangala there found local supporters who enabled them to
maintain a stand-offish, if not hostile, attitude to the Portuguese until
the end of the seventeenth century and later. These facts suggest that
the point would reward further investigation.
Finally, the role of the Imbangala in the Portuguese ‘conquest’ of
Angola helps to explain some of the major features of European
history in this part of Africa during the seventeenth century. The
strengths and drawbacks of the Imbangala political structures and
the relationship of the Imbangala to the Mbundu and Ovimbundu
populations on either side of the Kwanza illuminate much of the
history of the Angolan wars, both against African Icings and against
European rivals. The Imbangala gave Portuguese armies their first
consistent successes against the Mbundu in the decade after 1610. The
warriors of the kilombo made possible the increase in the slave trade
which converted the colony of Angola from a backwater of the
Kongo into an area of major Portuguese economic and political
interest. Opportunities to fight at the side of Imbangala armies
influenced such specific strategic decisions as the location of the fort
at Ambaca and the landing of the 1845 mission sent to relieve
Massangano. The Imbangala not only enabled Portuguese officials to
 C O N C L U SIO N S 223
initiate large-scale wars designed to capture people for the legal
slave trade but also encouraged the development of smuggling
which attracted the Dutch and others to the coast north of
Luanda. Imbangala mercenary states protected the frontiers
which finally emerged from the conflicts of the early decades of
that century.
 C H A P T E R V III

Institutionalizing Political Innovation

The later history of the Lunda title-holders who headed the Icilombo
war camps brings into clear focus a theme which underlies the entire
course of Mbundu political history: the contrast between the evident
inventiveness of aspiring political authorities, who created numerous
new titles and emblems of power in their search for political hege­
mony, and the infrequency of successful innovations which spread
widely enough and lasted long enough to merit the designation of a
‘state’. The Imbangala bands, for example, had stopped and settled
briefly in several locations—even from imperfect documents and oral
traditions, we know of several places among the Songo, others near
the Atlantic coast as Kalanda lea Imbe’s Icilombo worked its way
north, and again under the kulashingo in conjunction with the
Portuguese north of the Kwanza. Yet none of these pauses lasted
long enough to fit our largely intuitive definition of a ‘kingdom’. On
the other hand, several holders of related Lunda positions—the
kabuku lea ndonga, the Icalandula, the Kasanje kinguri, and others—
established permanent political structures on the fringes of Portu­
guese Angola which have become accepted as important ‘states’ in
the period after 1650. The same contrast between frequent innovation
and uncommon permanency recurs in the earlier history of the
kinguri title, when at least three incipient kingdoms existed briefly
among the Cokwe east of the Kwango.
The northern Imbangala kings managed to impose themselves as
permanent rulers of the Mbundu lineages in part through their
celebrated alliance with the Portuguese, as the differences between
their history and that of their counterparts south of the Kwanza
demonstrate. But even as northern Imbangala rulers fought for
Portuguese governors and traded with European merchants, they
also struggled to survive by abandoning the most distinctive aspects of
the Icilombo and adopting local Mbundu ideas and practices. In the
end, the Imbangala Icings survived as rulers of the Mbundu only by
trading the alien political institutions, which had originally enabled
them to effect the revolution in Mbundu politics described in the
preceding chapter, for local ideas, institutions, and symbols. The
 K I N A N D N O N -K IN 225
history of the Imbangala Icilombo, therefore, paralleled (in so far as
the available evidence allows comparison) the history of the ngola,
which had originally been integrated into the Mbundu system of
descent groups before it flowered briefly as an instrument of central­
ization in the ngola a kiluanje ldngdom and then yielded again to the
parochial machinations of lineages within the state. The mavunga,
which first reached the Mbundu as agents of the distant Libolo Icings,
had undergone the same transformation as descent groups converted
them to little more than local lineage titles by the beginning of the
seventeenth century.
The histories of all major Mbundu states in the period before 1650
emphasize that the critical step in the process of state-formation was
not the invention of techniques capable of expanding political scale
or increasing centralization; it lay instead in the transition from an
ephemeral cross-cutting structure to a more permanent political state
able to impose some sort of subservience on the Mbundu lineages. In
short, the institutionalization of political innovation constituted the
key problem for would-be kings who faced a social environment
dominated by strong kin groups. Since relatively scanty information
on the earlier Mbundu states precludes detailed examination of this
problem in connection with the ngola and the vunga, this point has
been left for examination in terms of the Imbangala Icilombo, for
which ample documentation exists.
The major Imbangala political institutions clearly changed rapidly
during the first fifty years of their history among the Mbundu, drop­
ping or modifying most of the alien elements in their heterogeneous
structure and retaining only the Lunda titles which had come from a
background most similar to that of the Mbundu social structures of
which they became a part. Many aspects of the original Icilombo were
thus lost. The contrast between the fate of the Icilombo among the
Mbundu and its history in its native area south of the Kwanza, where
it underwent far fewer modifications, once again proved that state-
formation was to be understood at least as much in terms of the
environment in which it occurred as in the nature of the political
institutions themselves.

Kin and Non-lcin

One of the key innovations which gave the Imbangala Icilombo its
overwhelming military superiority in relation to the Mbundu was the
absence of particularizing and divisive lineages of a type which had
historically retarded the expansion of political and social scale among
226 I N S T I T U T I O N A L I Z I N G P O L IT IC A L I N N O V A T IO N
the Mbundu. Evidence from a variety of sources supports the hypo­
thesis that the Imbangala bands reached Angola without descent
groups,1 Nineteenth- and twentieth-century traditions collected from
descendants of the Imbangala indicate that lineages disappeared
when the kinguri abolished kinship among his followers before they
had reached the Kwango river.2 Seventeenth-century Europeans
repeatedly emphasized such superficially bizarre Imbangala laws as
those prohibiting the birth of children. These laws, clearly among the
most important yijtla laws of the kilombo, effectively denied the
procreative function of women and, because the mother-child
relationship formed the crucial kinship link in the matrilineal social
systems of central Africa, eliminated Idnship within the kilombo.
It is not necessary to accept the often exaggerated European stories
of sexual abstinence, infanticide, and the abandonment of children in
the bush to understand how these yijila produced the intended effect
of suppressing descent as an element of social structure among the
Imbangala. Births continued to occur in practice, and many children
survived the formal proscriptions against their presence, but the
yijila declared all such infants illegitimate in Imbangala terms and
absolutely denied them social status within the kilombo. Some infant­
icide may have taken place as a result of the low prestige of these
children but, contrary to the claims of horrified and often gullible
seventeenth-century missionaries, by no means all Imbangala infants
died from exposure. Illegally pregnant women could circumvent the
letter of the yijila, which specified that no children could be bom
inside the kilombo, by temporarily leaving the confines of the walled
encampment to give birth, thus placing their infants outside the
formal community of the warrior initiation society. Children born
within the sacred enclosure, on the other hand, caused great alarm
among the Imbangala since they violated the ritual purity of the
place, and in such cases the Imbangala killed both mother and child
in order to destroy the incipient matrilineage created by the birth.3
1 All indirect, but conclusive when considered together. The nature o f the
available sources readily explains why direct confirmation could not be expected
from them. Seventeenth-century Europeans possessed only the limited ethno­
graphic insights o f their time and could not have described the unique char­
acteristics o f Imbangala social and political organization in modern technical
terminology. N or could the Mbundu genealogies preserve a record o f this
aspect o f Mbundu history since they trace only lineage and political titles o f the
sort that the Imbangala abolished.
2 See Chapter V.
3 Cavazzi (1965), i. 181. Cavazzi, like most Europeans, mistakenly interpreted
the yijila literally to mean that they permitted no births at all.
 K IN A N D N O N -K IN 227
The Imbangala also refused to acknowledge the significance of
birth as a mode of structuring society by denying the fertility of their
females. One way in which they expressed this idea was to attempt to
banish the fact of menstruation, the most obvious symbol of female
fertility. They claimed, for example, that the appearance of a men­
struating woman during preparations for important enterprises could
doom their prospects for success.4 Since many central African
peoples excluded menstruating women from participation in specific­
ally male activities, this practice alone did not distinguish the
Imbangala from other societies with lineages. But the Imbangala
developed the general belief to extremes generally not found among
their neighbours; according to the seventeenth-century traditions,
‘Temba Andumba’s’ warriors would give up the battle and surrender
to their enemies if one of their women began to menstruate during a
military campaign. Even the potential kinship represented by a
menstruating woman evidently contaminated the lineageless environ­
ment of the Icilombo.
Most Imbangala ceremonies carried out this ideology by pro­
hibiting the attendance of any female whatsoever. The Imbangala
allegedly observed this proscription so literally that they refused to eat
female flesh in rites involving cannibalism. Later in the seventeenth
century, even after the Imbangala had adopted many Mbundu
descent group ceremonies unrelated to the original Icilombo, they still
did not use the flesh of women for sacrifices to the lineage ancestors.
They had also reintroduced female fertility ceremonies, but never
performed them within the walls of the Icilombo.5 Among the Ovim-
bundu, where the practices of the Icilombo survived somewhat longer
than they did among the Mbundu, the Bihe Icilombo excluded all
women from its rituals as late as the 1840s.fi Imbangala ‘adulterers’,
that is, warriors apprehended in intimate contact with any woman,
had to pay much higher fines than such offenders among the neigh­
bouring peoples with lineages; a seventeenth-century Kimbangala
suffered death, and Europeans who violated Imbangala females had
to pay goods amounting to the value of a slave, a fine equivalent to
the life taken from one of their own men.7 These rules against illicit
sexual intercourse enforced the prohibition against procreation on
males as well as on women and had the effect of discouraging the
development of lineages within an Imbangala band.
Proof that the Imbangala consciously intended their anti-female
•' Ibid., i. 183. 5Ibid.
0 Magyar (1859), pp. 275, 312-13. 7Cadornega (1940-2), iii. 269.
228 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
prohibitions only to abolish lineages and had no rule against the
presence of women generally lies in the fact they excluded only their
procreative role and depended on women for certain other purposes.
The Imbangala welcomed women as long as they remained outside of
the kilombo and did not affect the lineageless social structure of the
band. They had no recorded objections to the ritual and economic
functions of their women, for example, since Imbangala armies
always included a great many female camp-followers, wives, and
slaves.8 In their ritual cycle, Imbangala warriors depended on the
tembanza, wife of the leader of the kilombo, for preparation of the
maji a samba which conferred invulnerability on them. Women
performed important duties in ceremonies which Imbangala title-
holders and war leaders conducted before every important battle.9 On
these occasions, they had ritual intercourse with their wives, and their
ability to consummate the sexual act demonstrated their firm com­
mand of the supernatural powers attributed to their office. Ritual
intercourse thus acted as an omen which foretold the outcome of the
impending battle; successful copulation guaranteed that the Imban­
gala would win.
The Imbangala king’s chief wife participated in a complementary
ritual held after the battle which symbolically reaffirmed the people’s
loyalty to their chief. She accepted the heart or brain of a slain enemy
from the leader of the war camp and ate it in a public ceremony. If
she became nauseated or could not eat it, the Imbangala assumed
that she had been sexually unfaithful during her mate’s absence and
put her to death.10
Ritual intercourse of this type had a primarily symbolic function

8 Cavazzi (1965), i. 219.

9 Ibid., i. 183, 185. The distinction between the customs o f the kilombo,
which excluded women (except for the tembanza) and came from the Ovim-
bundu, and the Lunda origins o f most northern Imbangala kings’ titles explains
why females had significant ritual roles in association with these rulers. The
rituals attached to the Lunda positions symbolized the submissive ‘female’ role
o f the people in opposition to the ‘male’ dominance of the chief and matched
the symbolic role o f female figures evident in Mbundu musendo dating at least
from the sixteenth century.
The correspondence o f the Imbangala tembanza to the Mbulcushu mambanje
(pp. 164-5 above) suggests that the Imbangala probably used the children
o f these unions in the preparation o f the maji a samba. This story o f ‘Temba-
Ndumba’, who lulled her own child to make the first maji a samba, makes this
point explicitly.
10 Cavazzi (1965), i. 185. This ceremony recalls the ‘ritual meal’ o f Kulembe
and ‘Temba Andumba’ described in the seventeenth-century narrative on the
origins o f the kilombo.
 K IN AND N O N -K IN 229
and did not give rise to lineages. It could thus exist in the Icilombo
despite the ban against kinship in Imbangala society. Despite the
superficially exotic appearance of this ritual, its structure conformed
closely to oath-talcing ceremonies found everywhere in central Africa
and the resemblances allow some inferences about the significance of
the ceremony. In the analogous Mbundu ‘poison oath’, a kimbanda
diviner who specialized in administering the oath gave a drink con­
taining a poisonous substance to a person accused of a crime.
The Mbundu believed that spiritual forces under the control
of the kimbanda revealed the guilt or innocence of the individual
through the drink, killing the guilty and sparing the innocent.
The Imbangala merely substituted human flesh for the beverage
used by others. Since the health of the Imbangala chief’s wife
symbolized the spiritual health and unity of his people, sexual
unfaithfulness revealed faulty loyalty of his people in the political
sphere.
The Imbangala extended their efforts to eliminate lineage struc­
tures by abolishing circumcision, a genital mutilation which, among
other central African groups, functioned as the male equivalent of
rites intended to guarantee female procreativity. The Mbundu
regarded circumcision as the main prerequisite to the attainment of
adult male status in their society.11 Only adult males, that is, circum­
cised men, could marry and thus provide legitimate heirs for their
wives’ lineages. Circumcision guaranteed the fertility of males, just
as corresponding lineage ceremonies protected the fecundity of
females. Babies born of an uncircumcised male, the Mbundu there­
fore assumed, could have issued only from witchcraft or from sorcery,
and since such children could bring great harm to their lineages, few
Mbundu women willingly engaged in sexual intercourse with
uncircumcised men. In Mbundu terms the Imbangala abolition of
circumcision ceremonies had the same effect as the prevention of their
women from bearing children: it eliminated the possibility of
legitimate offspring and, hence, of lineages.
Although the amateur ethnographers of the early seventeenth
century noted neither the presence nor the absence of circumcision,
it almost certainly existed among non-Imbangala at that time.
Observers first mentioned it as generally characteristic of the peoples
11 Although these reconstructed seventeenth-century beliefs are inferences
projected back in time from interviews conducted in 1969, the close connection
o f these customs with the lineage genealogies known to survive from the six­
teenth century (and earlier) justify the attribution o f these ideas back at least
to the period under consideration.
230 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
of northern Angola in about 1690.12 The intimate relationship of
circumcision to the basic Mbundu lineage structure, to their marriage
customs, and to beliefs about fertility, however, suggests that they
practiced it long before the Imbangala arrived shortly after 1600.
The nineteenth-century distribution of western Angola groups
without circumcision conformed to the pattern of otherwise un­
related descendants of the seventeenth-century Imbangala and
suggested that they had been responsible for elimination of the
custom when they overspread the region. Except for the modern
inhabitants of Kasanje, the most direct Mbundu heirs of the seven­
teenth-century Imbangala,13 the modern distribution of peoples
without circumcision centres near the probable home of the kulembe
kings who had originated the kilombo. Nineteenth-century travellers
generally did not mention circumcision camps among the Ovim-
bundu, who still have very few except in the southern and south­
eastern regions least affected by the sixteenth- and seventeenth-
century Imbangala.14 The people of Wambu, who have preserved
several aspects of earlier Imbangala social and political structures,
had no circumcision at all as late as the mid-nineteenth century.15
Peoples on both banks of the Ngango river (‘Kibala and Libolo’)
still did not practice circumcision in the 1920s.16
The Imbangala oral narratives and genealogies add circumstantial
evidence in support of the hypothesis that the Imbangala had
abolished lineages and circumcision in tlieir bands. The chapter on
the early history of the kinguri showed that descent groups probably
disappeared before the Lunda and Colcwe in his band had reached
the upper Kwango. This hypothesis would explain why the Lunda
12Cadomega (1940-2), ii. 260, specified that Mbundu boys underwent
circumcision at the end o f the seventeenth century at the ages o f about six to
nine years. The presence o f these rites was confirmed for the late eighteenth
century in Mattos (1963), p. 337, and ‘Noticias do paiz do Quissama . . . ’
(1844), p. 212. Lopes de Lima (1846), iii. 200-1, appears to have fabricated
his statement that the ‘Jaga’ (meaning the Imbangala) were the only circum­
cised peoples in Angola. The allegation came in the context o f his argument,
equally fallacious, that the ‘Jaga’ had North African, specifically Muslim,
antecedents and had introduced the custom o f circumcision (closely con­
nected with Islam in the minds o f strongly Roman Catholic nineteenth-century
Portuguese) to the Kongo and Mbundu. Cameiro (1859-61), pp. 175-6,
emphatically corrected Lopes de Lima on this and a number o f other points.
13Various testimonies o f Sousa Calunga; only the Lunda Icings o f Kasanje,
i.e. the incumbents in the explicitly non-Imbangala kinguri title, underwent
circumcision, and then only at the time o f their installation.
14 Merran McCulloch (1952), p. 44.
15 Magyar (1859), pp. 159, 162.
16Leite de Magalhaes (1924), p. 68.
 K IN A N D N O N -K IN 231
makota adopted the kilombo; its proscription of circumcision con­
firmed the elimination of kinship from their lineageless band. The
abolition of circumcision in turn forced the makota to ‘murder’ the
Icinguri title in Mbola na ICasashe since the Lunda rituals of that
position required that its occupant be circumcised, just as all later
incumbents in ICasanje were also circumcised, uniquely among their
Imbangala fellows. The yijila of the kilombo had come into direct
conflict with rule by a Lunda tubungu position, and the triumph of
the kilombo required elimination of the kinguri title.
Libolo, the seventeenth-century Mbundu name for the kingdom
ruled by the hango and connected indirectly with the kulembe,
further associates the abolition of circumcision with the regions
where the kilombo began. The word libolo means foreskin in Kim-
bundu17 and indicates that the northern Mbundu derived their name
for this kingdom from its people’s most notable physical character­
istic in their eyes; it was the state of ‘those with foreskins’. Since the
word occurred in the earliest known documents of the sixteenth
century, the linguistic evidence suggests that the kingdom of Libolo,
probably influenced very early in its history by the kulembe, had
abolished circumcision before 1600.18
The unsettled Imbangala way of life and their need for constant
military preparedness may have contributed to the original decision
to eliminate circumcision. Imbangala bands moved constantly about
tire countryside as they exhausted the food supplies in the regions
where they passed. They could therefore rarely pause long enough in
one place to conduct the lengthy ceremonies connected with the
circumcision camp. Since they maintained the kilombo in a state of
permanent readiness for war, they could not exempt young men, the
able-bodied warriors of the band, from active military service for the
time needed to complete the months-long seclusion in the bush.19
Imbangala attitudes towards women, which denied their procreative
function even while depending on them for other ceremonial and
17 Kilima or ICirima (the name for the Songo lineages once incorporated in
the Libolo kingdom) means an uncircumcised person in Umbundu; Alves
(1951), i. 525. The word apparently does not exist in ICimbundu.
18 The hypothesis that the Mbundu o f Libolo abolished their lineages under
the influence of the kilombo explains why these groups at one time lost their
place in the Mbundu aetiological genealogy. (Miller, 1972 unpub., chapter II,
appendix E, pp. 228-39.) Without circumcision, and therefore lacking lineages,
the Libolo descent groups found no place in a genealogy consisting solely o f
names symbolizing ldn groups.
19 Similar conditions seem to have prompted the Zulu kings to eliminate
circumcision in times of extreme duress; Glyn Charles Hewson (1970), p. 64.
232 I N S T I T U T I O N A L I Z I N G P O L IT IC A L I N N O V A T IO N
practical duties, taken together with their elimination of male cir­
cumcision, amount to near-conclusive proof that the Imbangala
bands which descended on the Mbundu had abandoned the single
most important element of Mbundu society, their descent groups.
The Imbangala requirements of mobility and military preparedness
provided functional reasons which may have impelled them to take
these drastic steps. The absence of descent groups implied, however,
the need for some equivalent structure which could regulate and co­
ordinate the activities of the large numbers of people found in a
kilombo located in regions filled with hostile members of Mbundu
lineages.

The Kilombo as a Military Machine

The Imbangala kilombo, in its true sense of an initiation society or
warrior fraternity to which all Imbangala males belonged, replaced
descent groups with a strongly centralized political structure and
intensive military training. The initiation ceremonies which climaxed
the introduction of new members into the Imbangala bands made
their camp much more than a walled place used for defence; it
became a sacred spot associated with military training and weapons,
accessible only to initiated male members of the society. The yijila
served as rules and regulations designed to develop capable and
skilled warriors in place of the songs and chants taught in normal
circumcision camps. The maji a samba, which the first ‘Temba
Andumba’ had created by killing her own daughter and grinding her
remains into a mass of blood and flesh, conferred unity and courage
on Imbangala soldiers through the belief that it made the members of
the kilombo invulnerable to harm.
The conscription of new members into the kilombo through initia­
tion ceremonies not connected to kinship gave the Imbangala an
unlimited capacity to assimilate large numbers of new warriors. Their
ability to expand the size of a band without restriction distinguished
them from the lineages of the Mbundu, which could grow through
reproduction only at a much more limited rate. In order to ensure
that new members of the kilombo would be young enough to respond
to intensive training and could serve efficiently as soldiers, as well as
for ideological reasons connected with their denial of kinship, the
Imbangala accepted only uncircumcised young males as initiates.
This requirement produced the often cited Imbangala tendency to
sell or kill all captives other than boys and very young men who had
 T H E K I L O M B O AS A M IL IT A R Y M A C H IN E 233
not yet undergone circumcision in their own lineages and who thus
qualified for induction into the kilombo.20
The initiation ceremonies which converted these uncircumcised
youths into Imbangala warriors took place inside the kilombo walls
in front of reliquaries containing bones preserved from the pre­
decessors of the living Imbangala king.21 The king and his consort,
the tembanza, guarded these sacred relics while they supervised the
rituals of initiation. The Imbangala troops, all properly initiated
warriors, assembled before their leader and began by demonstrating
their martial skills in a mock battle. The ‘mothers’22 of the young men
then brought the candidates from a place outside the kilombo where
they had remained hidden up to that point. The Imbangala soldiers
confronted the initiates at the gates of the enclosure with drawn
bows, as if to defend the kilombo from attack. But then, instead of
assaulting the boys, they touched them lightly on the chest with
arrows to signify a ritual slaying, or capture as prisoners of war. On
the following night, the soldiers finally annointed the youths’ bodies
with maji a samba, the ointment which the tembanza made from
human fat, and brought them inside the kilombo.2324
The leaders of the kilombo concluded the initiation by performing
rites of an unspecified nature to confirm the validity of the initiation
ceremony by recourse to the supernatural forces of the kilombo 24 The
kilombo initiation clearly fit the classic model of rites de passage
throughout the world which mark an individual’s attainment of a
new status within his society. In this case, the youths first suffered a
ritual death at the hands of the Imbangala warriors, then spent a
20 The unusually young age (5-9 years) at which the Mbundu were circum­
cising youths during the 1680s may have been a response to Imbangala kid­
nappings; initiation at so early an age would have made their youths ineligible
to become Imbangala and might have decreased their attractiveness before
they were old enough to serve as warriors.
21 Battell in Ravenstein (1901), p. 32; Cavazzi (1965), i. 182; Cadornega
(1940-2), iii. 223.
22 This reference to ‘mothers’ may have been symbolic and, in any case,
need not have contradicted other evidence that the Imbangala did not recognize
the social relationship o f motherhood. On the other hand, it may have reflected
an adaptation to the Mbundu lineages already re-emerging among some
Imbangala by the 1650s.
23 This procedure paralleled the course o f the circumcision camp in Mbundu
culture, where boys spent several months concealed in the bush before formally
re-entering Mbundu society under the guidance of their mothers in a two-day
ceremony. The role of the Imbangala ‘mothers’ also followed that of the
initiates’ mothers in the Mbundu rituals.
24 Cavazzi (1965), i. 181-2, specified that these rituals involved sexual
intercourse between the ‘mothers’ of the initiates and unidentified partners.
234 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
night in an intermediate status unrecognized by the kilombo, and
finally entered the enclosure and the society as newly adult Imbangala
warriors.2526
The crucial phase of the kilombo initiation ceremonies, which was
believed to confer invulnerability on the new Imbangala warriors,
consisted of rubbing maji a samba on their bodies. Seventeenth-century
traditions explicitly noted that the maji a samba had made the legendary
‘Temba Andumba’ and all her followers invulnerable to harm.25 The
story of the maji a samba, although apocryphal in its details like most
Mbundu narratives, accurately connected the Imbangala custom of
infanticide and its sociological consequence, the elimination of
lineages, to their military strength. By associating the Imbangala
warriors’ alleged invulnerability with the murder of their own chil­
dren, the traditions metaphorically attributed their military effective­
ness to the tightly integrated lineageless social structure of the
kilombo.
The metaphorical interpretation of kilombo rules makes Imbangala
customs appear less grotesque than most writers have portrayed
them. The Imbangala did not literally ‘kill’ their children27 but ‘slew’
them only in the sense of abolishing their lineage affiliation during the
initiation rituals. This step separated them permanently (i.e. ‘lolled’
them) from their relatives and rendered them eligible for conscription
into the kilombo without the encumbrance of kinship or kinsmen. In
a similarly metaphorical vein, the Imbangala forbade women to
enter the kilombo only in the sense of excluding them from the
initiation rites reserved for young male warriors. There were, in fact,
25 Battell, in Ravenstein (1901), pp. 32-3, noted that the Imbangala marked
children with a collar as a stigma but removed it when youths lolled their first
man; this made them ngonso, or initiated warriors. These details probably
refer to aspects of the same initiation ceremonies not mentioned in Cavazzi’s
later description.
26 Cavazzi (1965), i. 177-8. Battell noted, in Ravenstein (1901), p. 33, that
ICalanda lea Imbe’s attendants rubbed him with human fat. Although he did
not specify the term for this ointment, it was almost certainly the maji a samba
explained by Cavazzi. Joaquim John Monteiro (1875), ii. 155-7, found that
mid-nineteenth-century Ovimbundu living in the interior behind N ovo Red­
ondo still anointed the bodies of their chiefs with human fat.
27 Purchas, in Ravenstein (1901), pp. 84-5, seems to have interpolated this
remark into Battell’s account, arguing that the Imbangala literally killed their
children in order that the infants would not encumber the military operations
of the warriors. The rationale is obviously implausible. Cavazzi (1965), i. 227,
cited individual cases of abandonment of children, but this did not constitute
proof that the Imbangala killed all their children since abandonment was a
normal means o f ridding a community o f deformed or ‘bewitched’ infants in
many parts o f Africa and Europe.
 T H E K I L O M B O AS A M IL IT A R Y M A C H IN E 235
no Imbangala women at all in a technical sense, since no female could
participate in the initiation ceremonies. This left all females accom­
panying the band with the status of non-Imbangala outsiders.28
The custom of ‘killing’ women who gave birth inside the kilombo
probably meant in practice that most women lived in houses located
outside the walls of the sacred ritual centre. They bore and raised
infants there, but these children, even though biologically the sons
and daughters of Imbangala males, had no social connection with the
kilombo society until formally initiated. Open secular towns or camps
built around a central sacred place accessible to only a few qualified
initiates occurred commonly in Africa. Mbundu circumcision camps
and hunters’ retreats, although not located in the towns, conformed
to the same sociological and physical model. The innermost quarters
of chiefs’ compounds provided another example of the phenomenon.
All constituted secluded retreats where qualified initiates retired to
conduct secret rituals.
The kilombo initiation endowed Imbangala kings with a monopoly
of power and authority never attained by their Mbundu counter­
parts. The multicentric social structures of the Mbundu included
many different types of independent specialized ritual leaders and
political title-holders, each wielding limited authority over a defined
sphere of activity. Some Mbundu officials had lineage titles respons­
ible for an ngola, a hinga, or pemba, while others held non-lineage
positions and ruled through other authority emblems. No single
chief’s authority comprehended all aspects of Mbundu life. But,
since the monolithic Imbangala social structure contained no
lineages, no independent officials competed with the kings for an
individual’s loyalties. All authority became centralized under a single
title-holder, and this concentration of power distinguished them from
their Mbundu predecessors.29
The Imbangala kings’ dual competence in both the temporal and
supernatural spheres further distinguished them from Mbundu kings
like the ngola a Icihianje or the w/iga-holders from Libolo. Mbundu
political officials’ relations with their subjects extended only to
temporal matters and left relations with the supernatural to yimbanda
diviners and the lineage officials. Imbangala kings, on the other hand,
28The modern Imbangala o f ICasanje still claim that their women are all
Songo, Mbondo, Pende, and so on, and that the only true Imbangala are
males.
29Anonymous relação (c. 1619) (Cordeiro (1881), v. 10) implied that the
unusually comprehensive authority o f the Imbangala chiefs distinguished them
from other African political authorities known to the author o f the document.
236 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
assumed the supernatural functions which non-political officials
performed for the Mbundu. Among the Imbangala, an absence of
mediumistic symbols comparable to the wooden figurines which
maintained artificial contact between Mbundu chiefs and their
limited supernatural resources clearly distinguished their kings from
those of the Mbundu.30 European observers, who only partially
comprehended the differences between the Imbangala kings and
Mbundu holders of political titles (the sobas, as they were known),
used precisely this aspect of their position to distinguish them, noting
that the Imbangala chiefs resembled ‘priests’ more than ‘political’
officials in the usual sense. On the African side, a nineteenth-century
Bihe tradition drew the same contrast quite explicitly by recalling
that the Imbangala kings resembled ‘priests’ more than ‘chiefs’.31
Nearly all seventeenth-century sources implicitly recognized a strong
distinction between soba and ‘Jaga’ as they referred to the Imbangala
kings.
Lower-ranking Imbangala officials had no independent authority
which might detract from the power concentrated in the position of
the single titled king. Unlike the Mbundu, who had characteristically
organized states through the use of semi-autonomous subordinate
perpetual titles controlled by lineages, the Imbangala employed only
political officials of the mavunga type.32 They probably brought these
positions, such as the ilunda, a mani lombo, the ikota, a tandala, and an
ngola a mbole, from Libolo.33 The vunga titles made all other officials
dependent on the Imbangala kings since each ruler could appoint and
dismiss incumbents at his own discretion without regard for the
claims of other lineages or corporate groups. The mavunga repre­
sented another fortunate complementarity between the kingurVs
lineageless Lunda band and the kilombo of the Iculembe. These
titles thus contributed to the absolute discipline which the Imbangala
kings imposed on their followers.
The tightly-knit social and political structure of the kilombo gave
seventeenth-century Imbangala armies a crucial tactical advantage
over their Mbundu enemies. Translated into military terms, the

30 Battell in Ravenstein (1901), p. 86.

31 Magyar (1859), pp. 315-16. The seventeenth-century equivalent of this
attitude influenced Cavazzi’s description o f the Imbangala and led him to
refer to them continually as a religious sect and to emphasize the religious
aspects o f their customs.
32 See chapter IV for the Libolo mavunga.
33 A full description o f these officials and their functions appears in Miller
(forthcoming (d)).
 T H E K I L O M B O AS A M IL IT A R Y M A C H IN E 237
unified structure of the Imbangala band created a relatively well-
integrated field organization capable of executing disciplined and
co-ordinated manoeuvres during major battles. Most Mbundu
armies, by contrast, consisted of a loose amalgam of small groups
drawn from the lineages subject to whatever Icing had raised the
military force. Each unit of such an army operated under the immedi­
ate control of its own hereditary leaders, tactical and ritual, who
often had no particular military skills. These, in turn, had no experi­
ence in carrying out orders received from the king.34 Most Mbundu
armies, as a result, attained only a fragile over-all unity and achieved
minimal co-ordination which left them vulnerable to any enemy who
could execute well-planned attacks.
All Imbangala troops received substantial military training during
their initiation into the lcilombo and could respond to directives from
the single Imbangala Icing with a degree of co-ordination unattainable
under the conditions present in Mbundu armies. The Imbangala king
and his vi/nga-holders, the most prominent of whom acted as war
leaders (the ngola a mbole), divided Imbangala warriors into squad­
rons, each under the command of an appointed leader, the musungoP
These misungo held their positions by virtue of demonstrated tactical
ability rather than by hereditary right and thus made effective field
commanders. Their close association with the king enabled them to
follow his directives quickly and accurately.
Perhaps the most important strategic advantage of the kilombo was
its capacity to co-opt unlimited numbers of alien males without
diminishing the cohesiveness of the Imbangala army. Any male, no
matter what his ethnic origins or how he came to the Imbangala,
became a fully qualified Imbangala warrior by demonstrating his
personal ability to fight and by completing the kilombo initiation
rites. Freed from the need to observe complex and restrictive kinship
rules in recruiting new members, the Imbangala quickly built up
armies large enough to overwhelm any opponent. Defeat and even
capture rarely destroyed the Icilombo’s ability to resist since its
leaders could regenerate their forces by co-opting as many new
warriors as they required.36
The contribution of training and discipline to Imbangala military
superiority became most apparent on the battlefield. Long and
34 Mattos (1963), p. 247, has an excellent description o f such an army, c.
1800.
35 Modem Mbangala retains musimgo, a personalized form of the same root,
meaning an appointed war leader; testimony o f Apolo de Matos, 6 Oct. 1969.
35Cf. Cerveira Pereira’s experiences near Benguela, chapter VII.
238 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
continual wars had taught the Imbangala how best to deploy the
misungo and their squadrons, but the model Imbangala battle plan
differed little from those of the Mbundu. One unit, a scouting party
called the pumbo,37 ranged ahead of the main army to seek initial
contact with the enemy. It tried to locate him in time to allow the
main body of Imbangala forces to ready itself for battle before the
enemy discovered them. The main Imbangala army moved more
slowly and with considerably less mobility than the pumbo since it
included the women and children of the soldiers, as well as a great
deal of non-military baggage. The main Imbangala force differed
little from Mbundu armies in this respect. When the pumbo reported
contact with the enemy, commanders hid the women, children, and
baggage in a nearby woods. They then arrayed the warriors in a
convenient elevated position and waited for their enemies to attack.38
The basically defensive Imbangala manoeuvres did not differ signifi­
cantly from tactics employed by non-Imbangala armies when they
adopted a defensive stance.39
When the battle finally began, the two opposing hordes clashed in
apparent confusion by European standards. Each Imbangala warrior,
like his enemies, wore identifying insignia which distinguished friend
from foe during the hand-to-hand fighting. The Imbangala in general
used the same weapons as the Mbundu: bows and arrows, with or
without poison, lances, stabbing spears, hatchets, knives, and so on.40
Their armaments varied at different times and places, as one Imban­
gala group or another occasionally developed special techniques
based on a particular weapon. Those who fought against the Portu­
guese in ICisama during the 1640s seem to have made especially
effective use of a war hatchet.41 Another group of Imbangala south of

37 Miller (forthcoming (d)).

38 For example, Cordeiro (1881), v. 11, 13.
35 Few records describe autonomous Imbangala offensive formations other
than as parts o f a Portuguese army, and so no comparison can be made with
the tactics of Mbundu or Kongo armies on the offensive (three assault waves).
The records leave the impression that the Imbangala rarely initiated an attack
but relied on defensive strategies.
40 Cavazzi (1965), i. 154, listed all these weapons for the Imbangala and
‘Mushikongo’ alike, clearly implying that they were not distinguishable in this
respect. Battell in Ravenstein (1901), p. 29, listed their weapons as ‘bows,
arrows, and darts’.
41 Relação de Viagem de Socorro de Angola de Teixeira de Mendonça e
Lopes Sequeira, B.N.M ., MS, 8187, fols. 61-65v ; published in Brásio (1952-
71), ix. 332-45; Veigas (1923), pp. 13-18; AA, sér. II, i. nos. 3-4 (1943-4),
135-44. Compare the hatchet (‘casengula’) mentioned by Battell in Raven­
stein (1901), pp. 32-3.
 T H E K I L O M B O AS A M IL IT A R Y M A C H IN E 239
Kisama introduced a longbow so powerful that an archer had to
plant one end on the ground and brace it against his feet in order to
draw the bowstring. These Imbangala also employed an arrowhead
uniquely shaped in the form of a halberd.42
The actual battle between an Imbangala army and its opponents
rarely lasted very long,43 since the superior discipline and execution
of the kilombo quickly proved decisive. The fighting ended as soon as
the first small unit on either side gave way and the mates of the
defeated squadron abandoned their resistance and fled in disorder. It
was at this point that Imbangala training gave their warriors a crucial
advantage over their enemies, since their belief in their own invulner­
ability and faith in the maji a samba instilled a confidence that they
could not lose. They tended to fight better than their opponents as a
result, and the Mbundu, accepting superior Imbangala performance
as proof of their invincibility, had little desire to resist what they
regarded as inevitable defeat.
Although the end of the battle decided the direct clash between
massed armies, it marked only the beginning of the damage inflicted
on the losers. The defeated side lost relatively few soldiers during the
hand-to-hand-fighting, but they sustained serious losses as the victori­
ous army pursued the fleeing remnants for days, stealing supplies,
kidnapping the women and children who accompanied the troops,
killing, and taking as many prisoners as possible for sale as
slaves.44
The brevity of the actual battle showed that both the Imbangala
and their enemies both accepted it as an omen. They regarded the
physical combat as only a small part of a much larger war strategy
based on supernatural preparedness. During the days and weeks
before a battle, the Mbundu conducted rituals which, they believed,
could predetermine which army would prevail, arming themselves
with the best magical charms available, waiting for omens to indicate
the most propitious moment to attack, and cementing their good
relations with spiritual forces which could turn the actual battle in
their favour. The war leaders who performed these ceremonies thus
had already executed their most important duties long before the
actual fighting began. The emphasis on magic accounted in part for
the lack of attention to more mundane matters such as battlefield
discipline. Both sides entered the fray in hopes that their magical
12Cadornega (1940-2), ii. 104.
13Cavazzi (1965), i. 219.
14 See descriptions o f battles, ibid. i. 154-5, 185, 193, 217-19.
240 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
preparations had given them invulnerability or had disarmed their
opponents’ charms.
At the same time, they feared that they might have overlooked a
potentially fatal detail and accepted the possibility that their enemies
had discovered charms even more powerful than their own. These
doubts explain why the Mbundu fled at the first sign of weakness on
either side; they accepted it as a sign that their preparations had
failed and that supernatural forces had doomed that army to defeat.
The defeated unit’s comrades, assuming that further resistance was
futile, therefore fled from what they viewed as inevitable defeat. This
attitude resembled their approach to divination ceremonies, since
they regarded the actual battle as a decisive omen which revealed
once and for all which army had justice on its side. Victory
turned less on force of arms than on supernatural revelation of
truth.
In the area of morale, the Imbangala had an overwhelming
advantage which made the crucial difference in a situation where the
failure of a single small unit could prove decisive. The kilombo
rituals kept Imbangala warriors in a state of perpetual supernatural
readiness through their strict observance of the yijila laws. This
explains why they laid such extreme emphasis on constant and exact
adherence to the yijila', they kept the Imbangala morally and super-
naturally fit as well as in a state of disciplined physical prowess.
Individual charms carried by each warrior gave additional super­
natural assistance. Feathers, horns, bones, animal claws, bird beaks,
and body paint, they believed, conferred the strength and invulner­
ability needed to win wars.45
The Mbundu, usually occupied with hunting, agriculture, and
lineage affairs, could not keep themselves similarly ready for war.
They instead had to perform lengthy and elaborate ceremonies on
the occasion of each engagement to bring themselves to the proper
degree of preparedness. These required many days of drumming and
dancing which gave their opponents advance notice of an impending
attack. The Imbangala, on the other hand, could fight Mbundu
armies without warning and beat their opponents before they felt
prepared to defend themselves. The cumulative effect, of all these
factors made the Imbangala certain victors in these literal wars of

45 Ibid. i. 219; if this practice served to identify individual Imbangala

soldiers in the manner of European armies’ emblems and insignia, as Cavazzi
supposed, it also was closely connected to the customs o f yibinda professional
hunters who used such charms for protection against supernatural animals.
 T H E K I L O M B O AS A M IL IT A R Y M A C H IN E 241
nerves. Mbundu and Kongo fears of the Imbangala confirmed the
importance of their strategic advantages since the mere rumour of
their presence was enough to send large armies fleeing in complete
disarray.46
The Imbangala warcamp itself had less direct military significance
than most contemporary observers claimed. It was true that the
kilombo often took the form of a fortified town surrounded by
wooden palisades,47 but the use of fortified towns probably did not
distinguish the Imbangala from their neighbours. Africans every­
where built brush piles and stake palisades around their homesteads
and villages as basic means of defence in unsettled times.48
These palisades only protected the Imbangala from enemy assaults
or provided an offensive base under certain rather limited conditions.
The kilombo provided an effective means of making war only against
opponents too powerful to defeat by outright attack but weak enough
not to beseige the kilombo. In such cases, the main Imbangala army
remained inside the kilombo while small squadrons (the pumbol)
engaged the enemies in frequent skirmishes. They gradually weakened
the opposing army to the point where the main Imbangala army
could pierce his defences. When the proper time came, they executed
this assault by sending out a diversionary force to attack the town
just as if they intended to provoke another small-scale engagement.
The main body of Imbangala troops remained hidden in the kilombo
until their enemies came out into the open and then rushed out to win
a quick and decisive victory.49
Since the Imbangala did not use the kilombo in this fashion against
large massed armies of Portuguese or other opponents, the walled
palisade was effective only against relatively weak chiefs whose
limited resources might permit them to harass the outskirts of an
unfortified village but would not allow them to storm a walled refuge.
The function of the kilombo as a military redoubt, while real enough
in some cases, does not alone explain the widespread successes which
the Imbangala achieved against large Mbundu, Kongo, and Portu­
guese armies.
16 Relação do Bispo do Congo (?) to el-Rei, 7 Sept. 1619; Brásio (1952-71),
vi, 375-84; also Fernão de Sousa to el-Rei, 28 Sept. 1624; Brásio (1952-71),
vii. 255.
17 Battell in Ravenstein (1901), p. 20; Cavazzi (1965), i. 191, has a rather
schematic drawing o f a kilombo ; the drawing has been reproduced in Kent
(1965).
48 The construction o f fortified villages in unsettled regions was often noted
by later travellers; it was obviously a very ancient response to danger.
45 Battell in Ravenstein (1901), pp. 31-2.
242 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N

Men and Non-Men

The Imbangala north of the Kwanza achieved their overwhelming
victories at least in part as a result of their ability to terrify their
opponents into submission, sometimes without having to face any
serious opposition at all. Their fundamental technique involved the
manipulation of symbols and rituals which identified them in the eyes
of the Mbundu as supernatural beings, or at least as superhumans.50
The Imbangala pretensions to non-humanity added to their apparent
invincibility by placing them theoretically beyond the range of the
material weapons at the disposal of their enemies. The means they
used become intelligible through comparing the characteristics which
the Mbundu used to classify beings as humans or non-humans51 with
known Imbangala customs.
The Mbundu defined humanity according to the attributes they
themselves possessed: sedentary farmers grouped in lineages which
collectively controlled specified parcels of land. The three comple­
mentary elements of land, lineages, and agriculture made civilized
human life possible. Membership in a lineage endowed a living being
with the qualities of humanity through the blood which he shared
with his kinsmen, the individuals united to him through the females
of his ngundu. Complex supernatural bonds united each kin group to
its lands and integrated the people and territory into a single collect­
ivity. The relationship between land and people meant, for example,
that the spirits of deceased lineage members remained part of the
living kin group and rested easily only when their bodies lay in the
lineage’s own lands.
The category of non-humans included all other beings: animals, the
spiritual forces of a closely related but invisible supernatural world,
and most other people unrelated to the Mbundu.52 The Mbundu
basically regarded non-humans with fear because they, as human
beings, had no physical means of controlling the actions of non­
humans. Weapons provided effective protection against normal
animals and human enemies, but, the Mbundu believed, supernatural

50 M. Crawford Young (1970), pp. 987-90, provides an excellent discussion

o f the effectiveness o f magic as a tactic in the 1964 Congo rebellions. He
stresses that ‘magical’ techniques can work only when the enemy believes in
them.
51 This discussion, like that dealing with Mbundu beliefs about circumcision,
derives from Mbundu attitudes expressed in 1969. The same logic justifies its
extension to their seventeenth-century ancestors.
52 Cf. Middleton’s discussion o f similar attitudes (1965).
 MEN A N D N O N -M E N 243
beings invulnerable to arrows and knives often disguised themselves
in the bodies of men or beasts. Since no visible marks distinguished
these supernatural beings, they presented special difficulties for the
Mbundu who consequently took special precautions to protect them­
selves from unidentifiable but potentially dangerous strangers or
supernatural beasts. These precautions took the form of rituals
performed by such magical specialists as the yibinda master hunters
or by lineage officials.
No human being could survive outside the protective shield which
these rituals established around the Mbundu lineages. Kinsmen had
to act, for example, with one eye constantly cocked towards the
spirits of their deceased ancestors. Only charms and magic, and the
mediation of diviners skilled in communication with non-human
beings, enabled them to influence the capricious characters of the
supernatural world. The duties of the lemba dya ngtindu included
rituals in which he represented the concerns of the living to the
ancestors of the kin group. Each lineage had a kibinda master hunter
who dealt with non-human beings which appeared in the form of
large beasts. The diviners or yimbanda communicated with other
types of spiritual beings.
Mbundu kings, by contrast, were relatively vulnerable to super­
natural dangers since they dealt only with human spheres of activity
and ruled primarily through straightforward coercion. Still, political
chiefs ruled their primarily temporal domains in part through control­
ling supernatural forces associated with their titles. They often
employed yimbanda to fabricate special wooden figures (yiteka,
singular kiteka) which gave a limited ability to ward off the dangers
which they saw in the non-human part of the cosmos. Few, however,
possessed the skills of the kibinda which Lunda and Cokwe positions
like the mmjumbo and the kulashingo possessed.
The attention which the Mbundu evidently devoted to protection
from the supernatural in all spheres of their lives suggests how
susceptible they must have been to Imbangala tactics which deliber­
ately portrayed the warriors of the kilombo as the opposites of human
beings in as many ways as possible. The Imbangala invaders may have
borne a superficial resemblance to normal humans in the eyes of the
Mbundu, but Mbundu cosmology made ample allowance for the
deceptiveness of physical appearances and, as a result, the Mbundu
readily accepted the Imbangala as supernatural beings. Because they
had no specialists or techniques to deal with non-humans like the
Imbangala, they dreaded the Imbangala even more than they feared
244 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
other such beings, such as large forest animals or ordinary spirits,
which their established magical specialists, lineage officials, and
yibinda could combat. The Imbangala image of non-humanity thus
explains some of the fear with which the Mbundu (and some Euro­
peans53) regarded the invaders. In conjunction with tactical super­
iority and Mbundu beliefs about the function of magic and charms in
warfare, this accounts for the effectiveness of the Imbangala against
Mbundu kings and their armies.
The Imbangala built their rituals and symbolism of non-humanity
on ideas already present in the cultures of western Central Africa, in
conformity with a pattern which has appeared at numerous other
times and places. Most Central African political authorities made
limited use of such imagery to set themselves apart from their sub­
jects by commiting deeds strictly forbidden to normal humans.
Ritual incest, murder, and other acts symbolically established a king
as something more than human and thus qualified him to rule normal
people.54 They also separated the king from his kinsmen. These
beliefs normally did no more than establish the ruler as a neutral
arbiter believed to exercise impartial judgement in disputes involving
his former lineage or clan, but in the case of the Imbangala they
reinforced the absence of kin groups in the Imbangala kilombo.
African history records a number of other groups who used the
symbolism of non-humanity in similar ways to overawe their opposi­
tion and to forge tight bonds of unity in the face of hostility (e.g.
Mau Mau, Majimaji, the Simba of the Congo in 1964, and so on, to
name only recent and well-known examples).
Cannibalism, a favourite means of describing non-human out­
siders in cultures all over the world, provided an underlying theme in
many of the yijila rituals and customs of the kilombo which symbolic­
ally negated the humanity of the Imbangala. Their use of cannibalism
drew on analogies which the Mbundu saw between cannibals and
wild beasts or witches, all non-human beings which lived on human
flesh. Lions and other large animals obviously attacked and ate
humans from time to time. Witches, the Mbundu believed, nourished
themselves on the flesh of their victims; they materialized on dark
nights and danced slowly around the bodies of their human prey
53 Mbundu attitudes towards the Imbangala penetrated even to Europeans
who knew little o f the specific symbolism used by the Imbangala to convey their
non-human status to the Mbundu. European sources almost without exception
compared the Imbangala to the Devil or to ‘savages’, the seventeenth-century
European antitheses o f civilized Christian humanity.
54 Cf. the Mbukushu king mentioned in chapter VI above.
 MEN A N D N O N -M E N 245
before devouring them slowly, piece by piece.55 Cannibalistic rituals
thus effectively portrayed the inhumanity of the Imbangala in terms
which no Mbundu could fail to perceive.
Contrary to the rumours of Imbangala bloodthirstiness which
circulated in seventeenth-century Angola, the warriors of the Icilombo
actually consumed human flesh in only a limited number of rituals,
most of which fit familiar models of ceremonies found elsewhere in
Africa but without the distinguishing element of cannibalism. The
addition of cannibalism converted these ceremonies, performed else­
where as expressions of humanity, into manifestations of the non­
human status of the Imbangala participants. One such ritual, which
took place in advance of all battles, tested the courage of Imbangala
warriors and the effectiveness of their supernatural charms just before
they would be needed. In this ceremony, the Imbangala king, as the
foremost warrior in the Icilombo, killed a youth within the confines of
the Icilombo; the kings’ councillors then killed two adults inside the
kilombo and two more outside the walls. They mixed the flesh of the
five56 human victims with meat from five cattle, five goats, and five
dogs killed inside the kilombo and the flesh of five more of each
species sacrificed outside. The assembled warriors, the king, and his
advisers then ate the mixture, and each subordinate chief repeated
the entire sequence of killings before leaving the kilombo to fight.5758
Successful completion of this series of sacrifices demonstrated the
readiness of the chiefs and warriors to defeat their enemies.
A related ceremony, or another part of the same ritual cycle,
apparently detected potential cowards before they could endanger
their companions through faint-heartedness on the battlefield. Before
all important military engagements, a kimbanda specialist who had
charge of preparing the war charms, built five bonfires in and around
the kilombo.5&The kimbanda ran a long cord from the main bonfire,
which burned in the large open space in front of the king’s dwelling,
into the surrounding cleared area. The Imbangala warriors then
danced around the fire, taking care not to step on the rope which lay
on the ground beneath their feet. If any soldier touched the cord, all

55 Testimony o f Apolo de Matos, 7 Oct. 1969; compare with Redinha

(1958), p. 89.
56 The number five is quite rare in Central African symbolic systems but
reappears in modem Ovimbundu ceremonies accompanying chiefs’ burials;
Verly (1955), p. 683.
57 Battell in Ravenstein (1901), p. 34.
58 The use o f the number five may link this ceremony (reported about 1650)
with that reported about 1600; both were held immediately before wars.
246 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
the others present immediately killed him and ate his body on the
spot.59 The structure of this ceremony resembled divination rituals
found elsewhere in central Africa; in this case, the rope represented a
medium leading to the supernatural forces symbolized by the fire and
theoretically singled out disloyal and faint-hearted warriors before
they could betray their comrades on the field of battle.
From the Imbangala point of view, this and other rituals purged
their band of any member who might violate the strict yijila laws of
the Icilombo. Several reported customs document the seriousness
with which the Imbangala took these laws. Later in the seventeenth
century, for example, Nzinga’s relatively Mbundu-ized Imbangala
still sacrificed and ate humans at annual harvest ceremonies adapted
from Mbundu rituals, but they consumed only the bodies of persons
who had offended the spirits of the yijila.60 According to Imbangala
belief, the maji a samba conferred invulnerability on the members of
the kilombo only so long as they complied strictly with the yijila.
Strict observance of these rules was the psychological key to the
behaviour of Imbangala warriors who neither wavered on the battle­
field nor, as they believed, died from wounds or disease. The Imban­
gala therefore assumed that cowards, fallen warriors, and even
invalids and elderly people whose imperfect physical condition
revealed symptoms of mortality and humanity had broken the yijila
and forfeited the non-human status conferred by initiation into the
kilombo. The Imbangala consequently ate their own fallen soldiers
and shunned the elderly and the sick in order to purify their war camp.
Cannibalism also resolved several ideological difficulties which the
homeless wandering of the Imbangala created for people whose
background lay in sedentary agricultural communities. A widespread
Central African belief held that the spirits of unburied kinsmen
wandered the earth causing illness and death among the living.61
Moreover, Mbundu standards of proper burial demanded much more
than simply placing a body in the ground; the spirit did not rest
easily unless it lay in the lands of its own lineage. There the dead
person’s kinsmen could assume responsibility for both the body and
the spirit which lingered near it. Since possession of land occupied by
kinsmen gave the only assurance of freedom from troubles caused by
these spirits, the Imbangala exposed themselves to the malice of
59 Cavazzi (1965), i. 200-1.
60 Ibid. i. 89, 123.
61 Ibid. i. 212 confirmed this belief for seventeenth-century Mbundu. The
same idea appeared in the Cokwe narrative o f Kinguri’s crossing o f the Kwango
(van den Byvang (1937), pp. 433-4).
 MEN A N D N O N -M E N 247
their own dead when they abolished their lineages and abandoned their
homes. They could not bury the corpses of their comrades or their
enemies and they therefore ate their bodies in the hope of destroying
their potentially troublesome spirits.62
This hypothesis explains several Imbangala habits noted but mis­
understood by contemporary Europeans. Imbangala warriors care­
fully retrieved the bodies of slain adversaries from the field of battle
and brought them to the kilombo where they até the corpses in a
communal feast. Some attempted to wound their enemies in special
ways to identify which soldier had done the killing. European
observers noted that these markings enabled each warrior to claim
the body later at the kilombo. They presented specified portions of the
corpses, such as the heart and head, to their chiefs in recognition of
their status as the ‘head’ or ‘heart’ of the kilombo.63 The custom of
maiming enemies in ways which identified the assailant did not derive
from gluttonous desires for human flesh (as critics of the Imbangala
charged), but came instead from each warrior’s need to tame the
spirits of all the opponents he had killed by claiming and eating their
bodies. Without lands of their own, the Imbangala could protect
themselves against retribution from the spirits of their unburied
enemies only in this fashion.
A newTmbangala king and his councillors always sacrificed and
ate a human being at his installation as ruler of the kilombo.M Human
sacrifices during a king’s initiation (without the consumption of the
victim’s body) did not distinguish the Imbangala from most other
Central African peoples. Such ceremonies occurred widely to
demonstrate the ruler’s powers of life and death over his people and
to set him above other title-holders in his realm. Kings usually took
the life of their human sacrifice with a special sacred weapon which
symbolized the supernatural forces entrusted to the ruler’s hands.65
Human sacrifices thus tested the chief’s control over the spirits behind
62 The failure to bury fallen comrades carried through the non-kinship
ideology o f the Imbangala; cf. Horton (1967), p. 68.
63 Cavazzi (1965), i. 183, 217. Monteiro (1875), ii. 155-7, found the same
practice east o f N ovo Redondo in the nineteenth century.
64 N o seventeenth-century observer recorded the installation o f an Imbangala
king, but evidence survives in the form o f rituals conducted by the Imbangala
of Kasanje in the nineteenth century. My field notes generally confirm the
descriptions given by Rodrigues Neves (1854) and Salles Ferreira (1854-8).
One may also compare the ceremonies described for Bihe by Magyar (1859),
pp. 270-7, for the mid-nineteenth century.
65 In the case o f the Imbangala, perhaps the ‘casengula’ mentioned by
Battell in Revanstein (1901), pp. 32-3. Cf. the munjumbo's mwela (p. 156
above).
248 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
his magical weapon. His victim’s death demonstrated these spirits’
capacity both to protect and to punish the king’s subjects. The
Imbangala king and his advisers merely extended this common
ceremony to include consumption of a part of the body.
Cannibalism strengthened the discipline of the Icilombo, since each
warrior derived strong encouragement to respect the laws of the band
and to resist the temptation to flee from their enemies from his
knowledge that violation of the yijila meant instant death and the
disgrace of being eaten. Even if a coward escaped detection in the
preliminary dance around the bonfire, he might still be killed and
eaten afterwards if he revealed his loss of courage during the battle.
On the theoretical plane, cannibalism purified the Icilombo by erasing
all contaminating physical remains of unworthy comrades. Imbangala
and non-Imbangala alike believed that spirits lingered near the
bodies they had once inhabited. These spirits could return to plague
the living unless the physical remains associated with the spirit were
destroyed. Imbangala rituals therefore specified that they had to
destroy the entire bodies of such offenders by eating them.66
Imbangala and Europeans alike exaggerated the extent of the
cannibalism practiced in such ceremonies. Some observers claimed
that the Imbangala ate only people and refused to touch the flesh of
cattle or goats.67 Others recorded that the Imbangala ravenously
consumed the bodies of all their enemies to relieve their hunger and
fed ‘chiefly’ on human flesh. Even though the list of Imbangala rituals
involving the consumption of human flesh probably remains in­
complete, the evidence does not support such extreme claims. The
Imbangala may have eaten parts of relatively large numbers of human
bodies in certain rituals, especially those conducted after large
battles, but they habitually lingered in fertile areas (such as ‘Cali-
camba’ in Kisama) precisely because they found an abundance of
cattle and grain there. They did not raise cattle themselves but they
regularly stole from the herds of their neighbours, presumably for
meat to eat since no record survives of Imbangala mounted on oxen.68
66 Monteiro (1857), ii. 155-7, attributed this practice to unidentified groups
living east o f Novo Redondo in the nineteenth century (perhaps the Sele, who
still possessed a reputation for cannibalism seventy years later; Harnbly
(1934), p. 120); they lived not far from the places where the Icilombo originated.
67Purchas in Ravenstein (1901), p. 84. Cavazzi (1965), ii. 188, varied the
idea by arguing that the Imbangala ate people solely for nourishment.
0BBattell in Ravenstein (1901), pp. 25, 30. see parecer o f Francisco Leitão,
4 Dec. 1643; A.G.S., secretarias provinçales, maço 2639; published in Brâsio
(1952-71), ix, 86-7; also Leite de Faria (1952), p. 240. Leitão stressed that the
‘Jaga’ (or Imbangala) ate meat other than human flesh, as if to rectify rumours
 MEN A N D N O N -M E N 249
The Imbangala in fact ate only the bodies of enemies they killed and
preferred to capture as many women, men, and children as possible
alive to sell as slaves to Europeans or to augment their own numbers.
Much of the exaggeration surrounding the extent of Imbangala
cannibalism came from European misunderstandings of the idiom in
which the Mbundu described the Imbangala. Africans commonly
attributed cannibalistic practices to all their enemies, including
Europeans, and this habit certainly contributed to the formation of
fantastic beliefs about Imbangala cannibalism among those who did
not know the situation first-hand. Further sources of confusion lay
in the semantic field of the Kimbundu word which included the
European meaning of ‘to eat’. Unlike European words for ‘eating’,
the corresponding Bantu term had much broader senses which could
apply to capture or appropriation of another’s possessions for one­
self. Enemies thus ‘ate’ their captives by killing them, enslaving them,
or actually consuming parts of their bodies. Chiefs ‘ate’ the tribute
which they received from their people whether it consisted of food­
stuffs or of palm cloths and ivory.
From the Mbundu point of view, the Imbangala ‘ate’ captives they
stole from the lineages in the sense that they incorporated individuals
into their bands through a ritual death and transformation into non­
human beings. This meant that captured Mbundu abandoned their
lineage affiliation forever and became lost permanently to their kin
groups. In addition, some Africans undoubtedly played on the
credulity of the Europeans when fighting against them. On at least
one occasion, they brought out ‘pots’ (probably mortars used for
grinding flour) and told the naive Portuguese that they intended to
put all prisoners taken alive into the pots and cook them in order to
eat them.69 Only rare perceptive observers saw how restricted their
cannibalism was and pointed out that nearly all peoples in Angola
practiced similar customs, though to a more limited degree. One such
writer noted the use of cannibalism in Benguela early in the seven­
teenth century and emphasized that all the groups living there used it
in ceremonies for communication with their ‘ancestors’.70
69 Brito, ‘Rellaqao breve’, in Albuquerque Felner (1931), pp. 17-18.
70 Cordeiro (1881), v. 18; the author o f the anonymous document stated the
point as if to correct an erroneous but generally current impression to the
contrary.

to the contrary then current in Europe. The argument that the Imbangala
turned to the consumption o f human flesh because their bands grew beyond
[he capacity o f the land to support them will not hold; Battell in Ravenstein
(1901), p. 30.
250 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
The Imbangala also set themselves apart from normal human
beings with distinctive physical markings which supplemented
cannibalism as indications of their supernatural status. Nearly all
peoples of Angola used certain types of bodily mutilation, especially
scarification and chipping or filing of the teeth to signify their
humanness and to produce physical distinctions mirroring the
cultural and social differences between them.71 The Mbundu matri-
lineal kin groups, for example, achieved a sense of community and
identity through such techniques. Lineage members, however, never
pulled teeth or amputated parts of their own bodies, since they re­
served such markings for non-humans (either superhuman, as chiefs,
or subhuman, as criminals and slaves). These beliefs allowed the
Imbangala symbolically to emphasize their non-humanity by
extracting at least the upper two incisor teeth and perhaps the lower
incisors as well.72
The Mbundu saw dental extraction in particular as significantly
different from their own practices of filing or chipping teeth. They
pulled teeth only from the jaws of deceased occupants of permanent
titles and associated these teeth with access to strong supernatural
forces. A dead chief’s successors commonly preserved his teeth as a
relic which helped them to communicate with his spirit; the exact
tooth to be extracted varied from office to office, but in the case of the
kinguri of Kasanje later Imbangala always removed an upper
canine.73 During the burial rituals of a kibinda master hunter, the
surviving yibinda took an upper molar from their dead colleague’s
jaw and kept it in a shrine along with other possessions of the dead
hunter.74 Chiefs and yibinda both underwent special initiations which
placed them outside the lineage structure of the society and gave them
a semi-human status not unlike that claimed by the Imbangala.
Tooth extraction in Mbundu thought symbolically ran directly
counter to dental mutilation and effectively reinforced the image of
the Imbangala as non-human beings. It cut across lineage lines, while
71 Antonio de Almeida (1937).
72 Battell in Ravenstein (1901), p. 34, noted that the Imbangala extracted
four teeth. Cadornega (1940-2), iii. 225, specified the upper incisors and
added that the lower incisors were removed only on certain (unspecified)
occasions. Cavazzi (1965), ii. 240, mentioned only the upper incisor teeth; he
also pointed out (i. 171) that Nzinga’s Mbundu-ized Imbangala no longer
extracted but merely filed the upper incisors by the middle o f the seventeenth
century.
73 Testimony o f Sousa Calunga; also R. Verly (1955), pp. 689, 690, for a
record o f the identical custom in Libolo.
74 Testimony o f Sousa Calunga.
 D IS A P P E A R A N C E O F T H E K I L O M B O 251
mutilation reinforced them, and applied only to the dead in contrast
to the symbolic functions of mutilation which was practised on the
living.
Many recorded rituals of the kilombo society thus contributed to
the Imbangala image of non-humanity by reversing the humanizing
functions of normal Mbundu circumcision rites, converting Imban­
gala initiates from persons into non-human beings. The maji a samba,
prepared from the bodies of children, symbolically covered the
Imbangala warriors with death. Cannibalism equated the Imbangala
with witches and supernatural beasts. Proscriptions against births
in the kilombo denied the most basic biological functions of
human beings. The absence of circumcision further underlined the
differences between Imbangala and the circumcised Mbundu, and
dental extraction completed their connection with the supernatural
world. Even if the effects of these practices on the Mbundu
cannot be documented directly, they must have contributed to
the terror which the Imbangala caused in the ranks of Mbundu
armies.

Disappearance o f the Kilombo

Ironically, the same quality of alienness which enabled the Imban­
gala to sweep over most of the Mbundu during the early years of the
seventeenth century also made it impossible for the Imbangala to
build a permanent state on the basis of the kilombo rituals and laws.
Thus, although the kilombo proved revolutionary in its ability to
destroy existing Mbundu political structures, it had already proved to
be an ephemeral method of political and social organization by 1650
and played only a small part in forming the permanent Imbangala
states among the Mbundu.
The Imbangala kings, on the other hand, made a far more lasting
impression, since many of them survived into the twentieth century as
rulers of durable Mbundu states. Strict observance of the yijila
declined rapidly and the Imbangala adopted customs practically
indistinguishable from those of the people among whom they settled.
Even the modern Imbangala of ICasanje, the most direct Mbundu
heirs of Kulashingo and the Lunda makota, today retain almost none
of the characteristics which distinguished their seventeenth-century
forebears.
The assimilation of the imbangala kilombo into the Mbundu
lineages may be traced by comparing the unique features of each to
show decreasing differences through the seventeenth century.
252 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
Increasing similarities between the two would have resulted from
Imbangala borrowings from the dominant Mbundu culture and
would reveal rapid change in the relatively short span of forty to
seventy years. Although it is theoretically difficult to distinguish
certain widespread Central African customs which the Imbangala
may have brought to Angola and only coincidentally shared with the
Mbundu from distinctive Mbundu habits which they borrowed
directly from the local population after their arrival north of the
Kwanza, some tentative conclusions still emerge from the available
data. Several beliefs held by the Imbangala of the later seventeenth
century, for example, have such intimate connections with uniquely
Mbundu institutions that the Imbangala could only have obtained
them after they settled north of the river.
The evidence in support of this point includes a number of words
attributed to the Imbangala at different times after 1600. These show
a progressively higher proportion of Kimbundu terms later in the
century. The Imbangala probably had no distinctive language of
their own when they first reached Angola, since the band at that time
consisted of a mixture of different peoples co-opted all along the path
of the advance of the kinguri to the west: perhaps a few remaining
Lunda and Cokwe/Lwena, more Songo, many Ovimbundu, and
undoubtedly others as well. The earliest recorded ‘Imbangala’ word-
list came from the period before they had crossed the Kwanza river.75
This limited and inaccurate vocabulary consisted mainly of words in
use among the Portuguese sailors all along the western coast of
Africa. It included no uniquely Kimbundu words but had one
clearly Umbundu term {imbangala). Wordlists from the middle of the
seventeenth century, some fifty years later, show a mixture of
Kikongo-Portuguese terminology, Kimbundu words, and Umbundu.
These reflected a stage in the evolution of the Imbangala vocabulary
(and culture) after it had begun to evolve under the influence of
contact with Europeans who spoke Kilcongo and Portuguese and
with African speakers of Kimbundu. By the end of the century,
however, several terms noted earlier in other languages had shifted
to their Kimbundu equivalents (mani lumbu to mwene Inmbn, ilunda
to Icahu, and so on).76 The basic vocabulary of the modern Imbangala
represents the final phase of the modification of the dialect in the
direction of Kimbundu and exhibits no significant differences from
other eastern Kimbundu dialects except for a reported vague resem-
75 Battell in Ravenstein (1901).
76 Detailed analysis in Miller (forthcoming (d)).
 D IS A P P E A R A N C E O F T H E K I L O M B O 253
blance to the Libolo dialect, which presumably shows some similari­
ties to neighbouring Umbundu dialects.77
In contrast to the deliberately inverted customs of the original
kilombo, many Imbangala beliefs reported by the middle of the
seventeenth century differed little from those found among the local
Mbundu lineages. Although some of these have also been reported in
Central Africa generally, their emphasis on female fertility and child­
bearing, for example, shows that the Imbangala had already aban­
doned their early laws prohibiting the birth of children to members of
the kilombo. The Mbundu-ized Imbangala of the later period inter­
preted the birth of twins as an omen of misfortune and performed
rituals after such a birth to purify the mother and her kinsmen.78 If a
child’s upper teeth appeared before its lower ones, the Imbangala
took this as a sign that the child was a witch.79 The Imbangala of
Nzinga celebrated the first menstruation of young girls as evidence
that they were ready to bear children.8081Since a woman’s kinsmen
placed this kind of importance on her fertility only in a unilineal
society where the kin group’s survival depended on her ability to bear
children, these beliefs indicate a return to matrilineages and con­
comitant abandonment of the kilombo.61
Mid-seventeenth-century Imbangala veneration of deceased relat­
ives also indicated that they had adopted lineages of the Mbundu
type. Ceremonies identical to those described for the Imbangala
during the 1650s occur today in Mbundu descent groups and may be
assumed to have come from local roots before the arrival of the
Imbangala in Angola.82 Mbundu attitudes towards ancestors derived
from the idea of positional succession, which placed strong emphasis
on communication between the living incumbent in a name or title
and the spirits of his predecessors.83 Although one observer noted the
77 Comparative 200-word vocabularies o f Mbangala, Mbondo, Songo,
Jinga, Ambaldsta, and related Kimbundu dialects; for the Libolo-Mbangala
connection, Magalhães (1922), p. xxiii.
78 Cavazzi (1965), i. 182.
78 Ibid. i. 121, 181.
811Ibid. i. 183-4.
81 The lineages tends to assume greater responsibility for female fertility in
matrilineal societies where the descent group’s survival depends on its own
women; in many patrilineal systems, since the group propagates itself through
its males, it is less concerned with their wives (who are members o f other kin
groups); Robin Fox (1967), pp. 119-20.
8- Linguistic evidence supports this point; Miller (forthcoming (d)) under
mukwa a Icushingilisa. The connection o f these rituals with the pemha and other
very old lineage symbols confirms the argument.
B:* Although some present-day Luba have similar practices (Vansina (1966b),
254 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
practice of ‘ancestor’ veneration among the Imbangala before they
made contact with the Mbundu, he specified that the ceremonies
occurred in connection with political officials (probably Lunda
titles), not lineage ancestors of the Mbundu type.84 His description of
the related rituals made no mention of spirit possession, and he
probably mistakenly applied the name of ‘ancestor worship’ to
ceremonies which maintained contact between the incumbents in
Lunda positions and the previous occupants of their titles.
The distinctively Mbundu form of spirit possession operated
exclusively within the context of lineage ceremonies and dealt only
with the spirits of deceased kinsmen (jinzumbi, singular nzumbi).
Since the jinzumbi harmed only their own relatives, an offending spirit
always belonged to the same lineage as the person afflicted. Responsi­
bility for the ceremonies designed to relieve his suffering therefore
fell to lineage officials who conducted the rites before the assembled
members of the kin group. Seventeenth-century Imbangala spirit
possession rituals possessed all these features; the only Imbangala
priests who used spirit mediums were the nganga a nzumbi, specialists
in dealing with jinzumbi ancestor spirits described as ‘deceased kins­
men’ who bothered living Imbangala through dreams or by causing
sickness or death.85 The nganga a nzumbi first excavated the grave of
the dead relative accused of causing trouble. He always found the
grave easily accessible since it lay in lineage lands occupied by the
living members of the ngundu. He then examined the body to see if it
had begun to decompose. A decomposing corpse showed that the
nzumbi had left the body to wander contentedly elsewhere in the
supernatural world. But if the nganga a nzumbi found that the corpse
had not decomposed as much as expected, he announced that he had
detected the spirit responsible for sickness and death in the kin group,
blaming the dead relative for having been an undiscovered witch
while alive. The priest in this case performed rites to ensure that the
spirit should not again visit the living.86
If trouble persisted even after the nganga a nzumbi had taken these
steps, he arranged a spirit possession ceremony in which a medium
attempted to communicate with the nzumbi to discover the cause of
84 Battell in Ravenstein (1901), pp. 34-5.
85 Cavazzi (1965), i. 203-4.
88 Testimonies o f Sousa Calunga; Apolo de Matos, 6, 7 Oct. 1969.

p. 423), the century and a half and the institutional changes that separated
the Imbangala settled in Angola from the Luba make a historical connection
between the two a less likely explanation than an origin in local customs.
 D IS A P P E A R A N C E O F T H E K I L O M B O 255
his displeasure. The entire living membership of the afflicted person’s
lineage witnessed the rituals which attended this effort. Under the
guidance of an nganga a nzumbi and the other officials of the lineage,
the assemblage consumed intoxicants and performed dances and
chants calculated to bring on the trancelike state necessary to contact
the dead. Eventually, some member of the lineage (today, often a
woman specializing in such matters) felt himself possessed by the
spirit his kinsmen wished to reach. This individual was called a
mukwa a Iciishingilisa.87 The mukwa a Icushingilisa transmitted
questions from the living to the nzumbi and in turn interpreted the
answers of the supernatural to his relatives.88 The close association of
the mukwa a Icushingilisa with Mbundu lineage rituals indicated that
the Imbangala had adopted Mbundu jingundu by the middle of the
seventeenth century.
The spirits of the dead evidently communicated with living Imban­
gala in other typically Mbundu ways as well. Imbangala grave-
keepers could feel the will of the spirit whose grave they guarded
through their own emotions ;89 a period of moroseness, for example,
might send a grave-keeper in search of a diviner to discover the cause
of the spirit’s displeasure. The Mbundu also believed that jinzumbi
ancestor spirits appeared to the living in the form of animals pos­
sessed by spirits. An Imbangala nganga a nzumbi, summoned to
relieve sufferings caused by such a spirit, often placed a snare near the
house of the afflicted patient and baited it with special substances
intended to attract the brothersome nzumbi. The nganga a nzumbi
then presented any small animal which fell into the trap as proof that
the spirit had returned to harass his victim but had fallen into the
snare. He then destroyed the animal with special techniques believed
to put an end to the spirit concealed in the animal’s body.90
The other types of spirit mentioned by seventeenth-century
sources as unique to the Imbangala in fact represented adaptations of
Mbundu beliefs which had survived from much earlier periods. Most
of these had some connection with earlier political systems which had
become obsolescent by 1650 as a result of the changes brought by the
Samba and the ngola during the preceding 100 years. The Imbangala
living north and east of the upper Kwanza (the modem Songo and
87 The word given by Cavazzi, shingila, to denote the spirit medium is
today mukwa a kushingilisa (from the verb kushingila, indicating the act
of being possessed by an nzumbi); testimony o f Alexandre Vaz and Ngonga a
Mbande.
“» Cavazzi (1965), i. 128-9.
Ibid. i. 203-4. 90 Ibid.
256 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
probably the Mbondo and Pende), for example, believed that super­
natural beings resided in the springs of rivers. In the seventeenth
century, they explained that water flowing from these springs
represented the tears of once great spirits who had become sorrowful
because their human followers had abandoned them. The tears of the
female spirits formed the lakes of the region and those of the males
had become rivers.91
Widely held Central African beliefs about water spirits make exact
identification of these beings uncertain, but they present obvious
similarities to the malunga of the early Pende who had lived in the
same region where these spirits occurred. Modern Mbundu have
numerous equivalent water spirits; some, for example, claim to have
seen white supernatural beings emerge from rivers in Kasanje.
Other river spirits cause floods which destroy the crops and posses­
sions of th'ose who offend them.92 The names of some of the spirits
mentioned in the seventeenth century survive in the present as the
names of the rivers and lakes where the spirits allegedly once lived.93
A second type of local spirit attracted the loyalties of Imbangala in
areas formerly influenced by the state of Libolo. The observer who
reported these beliefs called these spirits ‘idols’, implying that carved
figurines of the type called mukishi in Kimbundu represented these
supernatural beings in the real world. These ‘idols’ occurred in male-
female pairs and, although they had no apparent connection with
ancestors, some communicated with the living through the medium of
a mukwa a kushingilisa. Each claimed adherents only in a specified
geographical region. The people in Kisama obeyed the priests who
spoke for idols called Navieza and Kasuma; Ibundo was the domin­
ant spirit of this type in Libolo itself. Kasuto and Nlcishi held the
loyalty of people in the ‘Ganguellas’, roughly the Songo area east of
Pungo Andongo.
Available data give only a vague idea of the nature of these spirits,
91 Ibid. i. 215. ‘Spirits’ tears’ seem to be a common cliche explaining the
disappearance of no longer dominant belief systems, some probably associated
with earlier political systems.
92 Testimony o f Sousa Calunga.
93 Notably Sashia, which is a river and lake, and the Kwango (river); the
‘Bala’ may be any one o f several rivers called Mbale. The Unga, Mwala, and
Lamba remain unidentified, probably owing to subsequent changes in the
names o f the bodies of water where they were believed to live. The Mbundu
told Cavazzi that these spirits had once resided elsewhere but had fled to take
refuge in the waters at the time o f the arrival o f the Imbangala. This story
probably meant that these spirits had once stood behind political authorities,
such as the lunga-kings, who declined in importance when the Samba authority
emblems entered northern Angola.
 D IS A P P E A R A N C E O F T H E K 1 L O M B O 257
but several factors suggest that these ‘idols’ may have represented
distant memories of former Libolo political officials. Mbundu
miilcishi figurines occurred mainly in association with political chiefs,
and the localization of these ‘idols’ in specific areas formerly con­
trolled by the hango recalls the positions of Libolo territorial govern­
ors. The occurrence of the idols in male-female pairs is reminiscent
of the figures in the political genealogies united by symbolic marriages
between political authorities and lineages. Each pair of idols,
according to this hypothesis, represents a survival from earlier forms
of political authority, probably Libolo overrule in these cases. Later
conquests by the ngola a kiluanje and the Imbangala had reduced
them to the status of priestly wielders of supernatural authority, who
remained faithful to the spirits of their old positions but no longer
commanded much temporal authority in relation to newer sets of
titles and spirits.
The re-emergence of lineages among the Imbangala by the middle
of the seventeenth century, Imbangala veneration of local spirits and
their concomitant incorporation of older, non-Imbangala, political
titles and symbols, their emphasis on female fertility and procreation
and—most obviously—the fact that they had ceased to wander
through the countryside and had settled as farmers among the
Mbundu all show the extent to which the original Imbangala
kilombo had yielded to the indigenous institutions and ideology of the
M bundu.
The kilombo had represented an effective response to the extremely
unsettled conditions among the Mbundu at the end of the sixteenth
century and the beginning of the seventeenth century, but it provided
an inefficient means of establishing permanent control over restive
Mbundu kinsmen. Disruptions accompanying the decline of the
kulembe state system and the rise of the ngola a kiluanje, as well as
the opening of the Atlantic slave trade and the development of the
Portuguese territory of Angola, created a place for fast-moving
armies which could rapidly assimilate people cast adrift by the
instability of the times. But the imbangala ideology of non-human-
ness and the rigorous conditions of life in the kilombo differed too
dramatically from the backgrounds of the people they incorporated.
Life as Imbangala placed too great a strain on individuals to endure
for very long without alteration in the direction of more familiar
institutions. The wandering style of Imbangala life contradicted the
basic belief systems and material culture of recruits who had grown
up in villages of sedantary farmers. As these recruits grew into
258 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
positions of responsibility in the kilombo, they gradually replaced
the yijila with normal Mbundu lineage structures and abandoned the
artificial denials of human biology and psychology on which the
kilombo depended.
The same ability to assimilate aliens, which had accounted for
Imbangala military strength, ironically contributed to the disappear­
ance of the kilombo in the longer run. Prohibitions against the birth
of Imbangala infants meant that bands tended to sustain themselves
by assimilating boys from surrounding non-Imbangala villages, and
the continual infusion of Mbundu novices quickly diluted whatever
alien cultural and linguistic characteristics the Imbangala brought to
an area. The Imbangala also obtained most of their women from
non-Imbangala living near them, so that most of the youths brought
into the kilombo, whether born of Imbangala women or not, had
been raised in homes where the culture and language were basically
Mbundu. This continual replenishment of trained Imbangala by raw
Mbundu recruits had produced its predictable effect by 1650 when,
slightly more than a single generation after the first Imbangala had
crossed the Kwanza river, many Mbundu words and customs had
already replaced the Cokwe/Lwena, Lunda, and Ovimbundu institu­
tions brought from beyond the river.
The demonstrated capacity of the Mbundu jingundu to absorb
alien rulers without losing their own characteristic identities also
contributed to the decline of the kilombo. The basic incompatibility
between the appointive titles used by the Imbangala and the perma­
nent positions of the Mbundu placed the invaders at a disadvantage
among the Mbundu; their precarious position there contrasted with
their security south of the river and explains why they readily
accepted Portuguese support in the former case but regularly opposed
European penetration in the latter.
Finally, fortuitous effects of Portuguese administrative policies
indirectly strengthened Mbundu lineages at the expense of the
kilombo. The Luanda authorities adopted the Mbundu sobas, usually
either lineage headmen or holders of lineage-controlled political
titles, as the official ‘native authorities’ in Angola. Portuguese
officials preserved the positions of the sobas (though not, of course,
the tenure of individual incumbents) by making them responsible for
tribute and taxes payable to the European administration. Itinerant
merchants supported those outside the sphere of direct Portuguese
administration by accepting them as their main trading partners.
Since these sobas represented lineages, their increasingly secure
 D IS A P P E A R A N C E O F T H E I C I L O M B O 259
status tended to preserve the Mbundu jingundu at the expense of the
Imbangala kilombo.
Nearly every factor which worked against survival of the kilombo
had a counterpart which encouraged preservation of the titles of the
Imbangala kings. Just as the kilombo disappeared in part because it
embodied ideas and institutions alien to the Mbundu, the kings’ titles
rested on Lunda and Cokwe concepts of positional succession already
familiar to the Mbundu. In particular, the Mbundu social and politi­
cal system, like that of the Lunda and Cokwe, rested on the theory
that the living occupant of a title ruled through his control of
physical relics preserved from his predecessors.
The similarity between the Imbangala and the Mbundu titles
appeared most clearly in their common association of bones with
spirits. An Mbundu kimbanda (specialist in magic) who wished to
dispose of a troublesome spirit, though not particularly that of a
chief, did so by locating the bones of the body, digging them up, and
burning them.94 Rituals connected with the Imbangala kings made
similar use of bones. They preserved the bones of dead chiefs in boxes
called misete (singular, musete) which they treated with veneration in
public ceremonies.95 The bones enabled them to communicate with
the spirits of these positions in a manner which the Mbundu recog­
nized without difficulty. The Imbangala in particular used the bones
to request the aid of deceased kings to bring military victories.
Because their unsettled life style kept them from claiming lands of
their own, they carried the bones in boxes rather than burying them
as the Mbundu did.96
Much of the symbolism surrounding Imbangala kings was common
to many Central African states beyond the Mbundu and Lunda
systems of titles. One of the common Central African symbols
involved the use of ‘peacock’ feathers in the head-dresses of Imban­
gala kings.97 ICalanda ka Imbe wore ‘peacock’ feathers when first
MTestimony o f Apolo de Matos, 6 Oct. 1969.
1,5 Cavazzi (1965), i. 185, 208; Cadornega (1940-2), iii. 228.
■is Preservation o f chiefs’ bones distinguished modern Ovimbundu from
Mbundu; Verly (1955), p. 690. But R. de Sousa Martins (1973) reported very
similar practices from the ndembu region of southern Kongo. Battell’s descrip­
tion (in Ravenstein (1901), pp. 34-5) of Imbangala ‘graves’ and ‘relatives’
sacrificing to ancestral spirits is inconsistent with nearly all other data on
Imbangala burial customs.
IJ7 The recurrent identification of the feathers in question as those o f a ‘pea­
cock’ does not specify which bird they might have come from, as the true
peacock does not occur in Angola. The koshi, ngwadi, or ngumbi identified by
Plancquaert (1971) in the lower Kwango region is properly a francolin;
personal communication from Mr. Kenneth P. Enright.
260 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
encountered by Europeans south of the Kwanza.98*‘Peacock’ feathers
have also been widely reported in association with chiefs elsewhere in
Angola. An old woman tended fifty ‘peacocks’ called the ‘njila
mukiso’ near the grave of the shila a mbanza, a non-Imbangala title-
holder in Kisama." In the sixteenth century, the mani Kongo and the
ngolaa kiluanje both used ‘peacock’ feathers as symbols of royalty.100
Twentieth-century Yaka associate the same symbol with their Lunda
chiefs, although it has no connection with chiefs of other origins.101
A nineteenth-century Lunda chief, Kibwiko, who lived near the
Kasai river, had a feather head-dress made from the plumes of the
‘peacock’.102
The use of white feathers to symbolize peaceful intentions on
formal occasions and ‘red’ ones to indicate war provided further
common ground between seventeenth-century Imbangala kings, as
well as the later kinguri of Kasanje,103 and other Central African
kings. It occurs today at least among the Lunda, Colcwe, Minungo,
Yaka, Tyo, Ovimbundu, and Kuba, as well as among the Mbundu.104
The ‘red’ feathers in these cases appeared ‘purple’ rather than ‘red’ in
the Mbundu system of colour perception and came from the ndua
bird, a type of plantain eater which had strong supernatural
significance.105 They believed that the ndua lived in the forests
far from human habitation where its loud, hoarse, unbirdlike
cry warned travellers of dangers, both natural and supernatural.
It could frighten robbers, animals, and spirits waiting to molest
unwary passers-by. The Mbundu interpreted its presence near civiliza­
tion as a bad omen, and believed that misfortune always followed,
98 Purchas in Ravenstein (1901), pp. 1, 86.
55 Batteil, ibid., pp. 26-7, 86. The name was obviously njila a mukishi
(‘the birds o f the mulcishi charm’), but his attribution o f the practice to the
Imbangala is doubtful owing to the description o f suspiciously similar circum­
stances at the court o f the ICisama shila a mbanza. The trait was not elsewhere
ascribed to seventeenth-century Imbangala but is very common among modern
Mbundu.
100 Lopes in Bal (1965), p. 62; also Jean Barbot (1732), p. 520.
101 Plancquaert (1971).
102 Joaquim Rodriques Graça (1854-8), p. 125.
103 Cavazzi (1965), i. 219.
104 Redinha (1963), pp. 48-9; Capello and Ivens (1882), for the mwata yamvo
in the 1870s; Hambly (1934), p. 135; testimonies o f Alexandre Vaz, 31 July
1969; Apolo de Matos, 5 Oct. 1969; personal communication from Jan
Vansina.
105 Monteiro (1875), i. 74-9, correctly identified the species as Corythaix
Paulina, adding that it was found in wooded areas all over Angola. Mr. Enright
adds that the name applies in Lunda to two species o f arboreal Louries or
turacos ( Corythaix Schaloui and Musophaga Cossae).
 D ISA PPEARANCE OF THE K I L O M B O 261
especially if the bird perched on the roof of a house and called
out.105
The Mbundu associated the ndua with blood. Its wing feathers are
blood-red in colour, and contain pigmentation which dissolves in
water containing ammonia; it then runs like blood from a wound.
The blood which trickles from the nose of a corpse was also called
ndua. The same word designated various diseases believed to result
from ‘excess blood’.106107 The eighteenth-century Ambakista traders
applied the term to poison oath ceremonies conducted in Kasanje.108
The similarity between the Imbangala kings’ connection of the ndua
bird with blood and with the supernatural world and its significance
in Mbundu cosmology showed a common level of symbolism
which facilitated the settlement of Imbangala rulers north of the
Kwanza by making them more acceptable to the descent groups
they ruled.
Other customs associated with the Imbangala title-holders which
occurred widely in Central Africa included court etiquette requiring
persons assembled before the king to applaud and show great
pleasure whenever their ruler sneezed or showed any other involun­
tary reflex.109 The modern Imbangala explain that a sneeze indicates
contact between the living chief and the supernatural powers behind
his position. Since the most important functions of an incumbent in a
perpetual title concerned his communication with the spirit world, a
sneeze indicated that he was actively performing his duties. This news
deserved the applause of the people.
Imbangala kings preserved only two recorded distinctive traits
possibly indicative of their Lunda origins. The incumbent Icalanda Jca
imbe had extensive tattooing all over his body in 1601.110 Scarification
associated him with the east, since it was relatively rare among the
Mbundu and Ovimbundu but highly developed in eastern Angola
among the Colcwe and Lunda.111 The Imbangala also painted the

106 Monteiro, ibid. The later incumbents in the kulashingo position in Kas­
anje kept the carved figure o f an ndua on the peak o f their compound roof.
107 Assis Jr. (n.d.), p. 32.
108 These oath ceremonies, for which Kasanje was renowned throughout
western Angola, probably evolved from the ibundu or kanu ‘drink’ noted in the
seventeenth century (Cadornega (1940-2), ii. 402-3; Cavazzi (1965), i. 190);
also Dias de Carvalho (1890a), p. 69. The name probably stemmed from a
belief that the gall bladder o f the ndua was extremely poisonous; Alfredo de
Albuquerque Felner (1940), ii. 14.
109 Cavazzi (1965), i. 193.
110 Battell in Ravenstein (1901), p. 33.
111 Almeida (1937), p. 79.
262 IN S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
kalanda ka imbe’s body with a white powder;112 this was probably the
pemba, a white clay associated with chieftainship, as opposed to
lineages, primarily among the Lunda, Luba, ICuba and in the forest
kingdoms of Zaire. Among the Mbundu, however, pemba belonged
only to lineage officials. This marked the origin of its use by the
Imbangala kings as somewhere to the north-east and probably in
Lunda.
Two other customs noted in the documents connected Imbangala
kings with the Ovimbundu antecedents of the kilombo. Seventeenth-
century Imbangala allowed only their noble title-holders to lie on
cattle skins.113 In Angola, only the kingdoms south of the Kwanza
associated cattle with nobility; there mourners wrapped the body of
their dead chief in the skin of an ox before transporting his corpse to
its burial place.114 The Imbangala sometimes hung their human
sacrifices upside down during the seventeenth century, probably in
order to facilitate the drainage of blood from the body. They col­
lected this blood in vessels and drank it in certain unspecified
ceremonies.115 This custom has been reported elsewhere only in
ICalconda, one of the southern Ovimbundu kingdoms descended from
Imbangala founders.116
The consonance of ideology and ritual associated with the Imban­
gala kings’ titles and the existing Mbundu political systems helped
the rulers to survive despite the failure of the kilombo to replace the
lineages. The simultaneous success of the kings and failure of the
kilombo thus reflected the different origins of the two major com­
ponents of the Imbangala bands, Lunda titles and an Ovimbundu
warrior initiation society. The kulashingo had therefore made a
portentous decision when he discarded the kilombo rules in favour
of the Lunda title of the kinguri in order to join the Portuguese.
The later growth of the ICasanje state, based on the Lunda position,
may have been a partial result of his choice. Elements drawn from
two such disparate systems could remain together only as long as
adversity forced them to unite. The relative tranquillity which
followed each time the Imbangala settled inevitably allowed them
to drift apart, as it did soon after the makota and the kulashingo
settled among the Mbundu.
112 Battell in Ravenstein (1901), pp. 33-4.
113 Cavazzi (1965), i. 190.
114 Testimony of Sousa Calunga, 10 Sept. 1969, for Libolo. Magyar (1859),
p. 316, for Bihe. Alberto Ferreira Marques (1949), p. 14, for Kakonda.
115 Cavazzi (1965), i. 187.
116 Ferreira Marques (1949), p. 14.
 D ISA PPEARANCE OF THE K I L O M B O 263
The Imbangala title-holders found themselves unable to establish
rapport with their Mbundu subjects by means of the kilombo and
turned for support to the Portuguese who, for reasons of their own,
welcomed Imbangala settlements on the periphery of their own
territories. Acceptance by the Europeans contributed to their
survival since their status as intruders ironically gave them certain
rights under Portuguese law not conceded to the Mbundu sobas.111
Governors of Angola, for example, rarely demanded labour or
supplies from Imbangala kings without a formal treaty specifying the
terms and conditions of co-operation. They in turn made it clear that
they regarded the kings as dependent on Portuguese approbation for
their positions. The slave trade, carefully controlled in the mutual
interest of these rulers and their European commercial partners,
provided economic underpinnings for the role defined by formal
diplomacy.
Seen from the perspective of Mbundu history, finally, the Imban­
gala invasion continued the pattern of repeated but unsuccessful
attempts by outsiders to impose centralized authority on the Mbundu
lineages. Just as the jingandu had survived previous challenges by the
vi/nga-kings, by Libolo, and by the ngola a kiluanje, they resisted
incorporation info the kilombo. The lineages had either taken control
of earlier overlords’ titles for themselves, as they had done with the
Libolo mavunga and most positions derived from the ngola a kiluanje
or adopted new forms of political organization to counteract the
centralizing tendencies of their current rulers, as they had done with
the original form of the ngola. In the case of the Imbangala, the
inherent weaknesses of the kilombo made it especially vulnerable to
Mbundu opposition.
The ability of the jingundu to dilute the power of their kings para­
doxically left them open to further invasion by limiting their capacity
to offer unified resistance. The lineages had gradually taken control

117 The stereotyped contrast between friendly, peaceful sobas and the warlike
intrusive ‘Jaga’ ran consistently through Portuguese sources from the six­
teenth to the nineteenth century. Examples: Fernão Martins (1591); B.N.L.,
Coleção Pombalina, códice 644, fol. 334; published in Brásio (1952-71), iii.
433-4, and Fernão Guerreiro (1930-42), ii. 191. Also Cadornega (1940-2),
passim, for the seventeenth century, Thomas Bowdich (1824), p. 25, from the
eighteenth century, and Duarte (1859-61), p. 134, for the nineteenth. It was
this vague sense of the word ‘Jaga’ to denote any warlike invader that led to
confusion of the Imbangala with other unrelated marauders in Mbundu and
even Kongo areas; for the latter, see Oliver de Bouveignes and J. Cuvelier
(1951), pp. 63, 71-2, where at least two distinct unidentified enemies of the
Kongo mani Mazinga were called ‘Jagas’.
264 I N S T I T U T I O N A L I Z I N G P O L IT IC A L IN N O V A T IO N
of the titles subordinate to the ngola a Iciluanje even as the central
kings spread their authority more and more widely during the six­
teenth century. The increasing autonomy of the jingundu rendered the
kingdom incapable of mounting solid resistance to either the Portu­
guese or the Imbangala. When the two invaders combined their
forces in pursuit of slaves, they doomed the older Mbundu kings to
defeat.
Although the Mbundu lineages had emerged victorious over the
kilombo, they could not similarly banish the Imbangala kings. Many
headed the Mbundu-ized Imbangala kingdoms which prospered as
the primary suppliers of slaves sent to European traders throughout
the eighteenth and part of the nineteenth century. Holo, Jinga,
Kasanje, Mbondo, Kalandula, Kabulcu lea Ndonga, as well as various
southern kingdoms such as Bihe, Kakonda, and Wambo, all thrived
under holders of Imbangala titles settled on the fringes of Portuguese-
controlled territory. These states, formed by the combination of
Imbangala kings and Mbundu kinsmen, dominated the political and
economic history of Angola for the next two centuries, until the
ending of the slave trade during the 1850s once again shifted the
balance of power in favour of the lineages and sent even the Imban­
gala kings into a decline from which they have never recovered.
 C H A P T E R IX

Conclusions

B e c a u s e t h e several examples of state-formation found in the early

history of the Mbundu all occurred in a relatively unvarying social
and ideological background, they may be usefully compared to elicit
some tentative generalizations about ways in which kingdoms may
emerge in an environment featuring strong descent groups and, in
particular, where people think in terms of perpetual kinship and
positional succession. Although the constant social and ideological
context of these cases simplifies the historian’s task in one way, by
reducing the number of variables he must consider, at the same time it
complicates his job by limiting the applicability of his conclusions to
cases where demonstrably similar circumstances obtain. The patterns
which emerge from the history of the Pende /wnga-kings, the kulembe,
Libolo and its vunga appointive chiefs, the ngola a kiluanje, Lunda
perpetual titles, the Imbangala Icilombo, and the contacts between
them should therefore be applied elsewhere only to the extent that the
relevant aspects of Mbundu culture are also present. My present
feeling—though not yet fully documented—is that roughly similar
conditions existed throughout most of the matrilineal area of the
southern African savannas.
This disclaimer requires explicit emphasis at the outset since,
whatever the broader applicability of some of the points drawn from
the experience of the Mbundu, the historian should always phrase his
conclusions with an eye to the precise conditions under which they
hold. If a wider comparative dimension emerges from these cases, it
probably lies in the opposition which the Mbundu saw between their
kings, or other forms of political structure, and their kinsmen, the
members of the strongly corporate Mbundu unilineal descent
groups. The value of examining several cases from a single society
lies in the capacity of this methodology to illuminate the variety of
ways in which states emerged from a background of unilineal descent
groups and the strength these unilineal descent groups showed in
their responses to the growth of states in their midst.
266 C O N C L U SIO N S

Defining a ‘State’
The fundamental question of what constituted a ‘state’ in the
context of Mbundu society has been deferred to the conclusion in
order to develop a definition based on Mbundu theory and practice
rather than on categories derived from Western experience. It is clear
that the Mbundu had no conceptual category equivalent to that of
‘state’ in European terms but instead made a dichotomous distinction
between their descent groups (and the related titles, etc.) and a
residual category which included all other forms of what I have
termed ‘cross-cutting institutions’. Accepting the Mbundu categoriza­
tion of their social institutions as a starting-point, a two-stage
definition of ‘state’ follows. We may initially term all the hunting
societies, circumcision camps, curing cults, and other organizations
which related people to one another beyond the bonds of kinship as
the Mbundu recognized it as ‘political’ in the sense that they existed
outside the realm of purely ‘social’ relations as defined by the descent
group structure.1 The Mbundu typically expressed these relationships
in terms of networks of permanent named positions; usually thought
of as ‘descendants’ of single ‘founders’, which tied the lineages (and
the people in them) to one another.
But the category of ‘political’, thus phrased, is a negative one,
defined as not involving the descent groups, and is broader than
anything which could usefully be termed a ‘state’ since it includes a
great many small-scale and ephemeral phenomena. Some degree of
duration seems necessary to distinguish the ‘states’ which appear in
Mbundu political history, as well as size superior to the other institu­
tions present in the society. Both the characteristics of size and
duration, which separate Mbundu ‘kingdoms’ or ‘states’ from other
non-kin-based institutions, are relative and ultimately subject to the
historian’s judgement. No clear line divides those ‘states’ which were
large enough and sufficiently long-lasting to merit the designation
from other structures relegated to the conceptual limbo of transitory
and minute cross-cutting institutions. Precisely this difficulty marred
the discussion of the incipient states which appeared from time to
time in the history of the Songo, for example. ‘States’ thus occupy one
corner of a two-dimensional plane defined in terms of size and
duration and are not qualitatively distinguishable from other non­
lineage, i.e. ‘political’, institutions.
1 Although I arrived at this definition o f ‘political’ on empirical grounds, my
subsequent reading in political anthropology has shown me that it is a common
theoretical concept; e.g. Lucy Mair (1962), p. 24.
 D E F I N I N G A STATE 267
The vagueness of the lines distinguishing ‘states’ from other
‘political’ structures, which might be called by a variety of terms
(‘statelilce institutions’, pre-states, or even proto-states occur as
possibilities but show the fruitlessness of a quest for a single useful
one), can be seen in the development of the Icilombo from its origins
as one cross-cutting institution, a circumcision camp, in a context of
many, into a dominant instrument of state-building in the hands of
the Imbangala kings. Having attained the status of ‘state’, the
Icilombo then metamorphosed into no more than a single relatively
minor aspect of the court rituals of Imbangala kings who built
enduring states on the basis of other political techniques. Similarly,
some of the political mavunga originally associated with the state of
Libolo lost their function as elements of the large kingdom and
became the focus of local religious veneration among the Songo, still
‘political’ in the sense that they attracted unrelated disciples from
many lineages but hardly describable as ‘states’ in an environment
where the ngola a kiluanje and the Imbangala had become the
dominant forms of state organization.
Not all characteristics of the Mbundu states defined in this way
coincided with the categories which have customarily been used to
define ‘states’ in Africa.2 Mbundu states, for example, had no
necessary territorial organization of the sort most writers have sought
to discern in African kingdoms. Some Mbundu states, especially
those based on the lunga and—to some extent—the Libolo mavtmga,
defined the authority spheres of political title-holders in terms of
geography rather than in terms of people. But such indisputable
kingdoms as the ngola a kiluanje and the Icilombo at its mobile height
defined themselves in terms of people rather than land. The notion of
‘marriages’ between title-holders sanctioned by the king and descent
groups, expressed most clearly in the idiom of the genealogical
traditions, made this point quite explicit. In fact, Mbundu cosmology
reserved control of land and certain forms of authority over the
people who lived on defined parcels of it to the lineages rather than
to non-descent group political authorities. Each lemba dya ngundu, if
he headed a landowning lineage, determined the rights of both
kin and non-kin to till the soil, hunt in its forests and woodlands,
fish in its streams, and harvest the fruits of its trees. The reliance of
the early theorists on a ‘territorial’ definition of state authority
probably represented their first approximation to what, at least in
the case of the Mbundu, closer examination has suggested may
2 Based on Fortes and Evans-Pritchard (1942), p. 5, and Vansina (1962a)
268 C O N C L U S IO N S
have been a definition of authority over persons other than one’s
kinsmen.
Centralization, another frequently cited defining characteristic of
African states, varied greatly among Mbundu kingdoms, ranging
from the barely incipient concentration of authority found among the
Pende lunga-kings through intermediate degrees of central control in
the ngola a kiluanje and Libolo states to near-total centralization in
the Imbangala kilombo. Despite the evident trend towards greater
centralization through time, all these structures were equally ‘states’
according to Mbundu standards. Nor was there a positive correlation
between the degree of centralization in a kingdom and its durability,
as the history of the kilombo demonstrated, since the Mbundu line­
ages clearly preferred less centralized political institutions to states
with all power concentrated in a single theoretically omnipotent Icing.
The degree of centralization also varied from time to time in the
history of each state depending on the talent and resources available
to its king relative to the peripheral title-holders in the provinces.
Thus Libolo, theoretically among the more centralized of the early
Mbundu states, owing to its development of the appointive mavunga
titles, at times failed to retain control over outlying title-holders like
the ndala kisua of Mbondo.
Monopoly over the legitimate use of force has ranked as another of
the distinguishing aspects of states. Yet among the Mbundu this
criterion failed to differentiate states from other sorts of institutions,
both social and political. The Mbundu expanded the conventional
Western sense of the term ‘force’ to include the magical coercive
techniques which backed the authority of nearly all their political and
social titles in addition to the simply physical means of coercion
usually denoted by the term. But neither the magical forms of coer­
cion nor the use of armed men were the monopoly of kings. The men
at the heads of Mbundu states had constantly to strive for and assert
their de facto superiority in charms and arms, since other sorts of
title legitimately gave access to independent forms of coercion and the
holders of these titles exercised them whenever they could. From this
perspective, one important theme in the history of Mbundu states
was constant competition between holders of different sorts of title
in a pluralistic universe of forces. The assumption that a ‘state’ had
exclusive access to coercive techniques would leave the historian at a
loss to explain the rise of subordinate positions to ascendancy, the
avidity with which the Mbundu searched continually for new forms
of authority, and the multiple powers claimed by the most successful
 D E F I N I N G A STATE 269
Mbundu chiefs. Moreover, relatively few Mbundu states regarded
themselves as dependent on the use of violence or even as specialists
in i ts exercise, since only the Imbangala kilombo maintained a standing
army and was prepared to conduct warfare on anything approaching
a permanent basis. It may be accurate to speak of individual title-
holders seeking to consolidate their positions through abrogating the
use of force to themselves, but the concept has little meaning at the
larger level of the state.
Jn some ways the most striking aspects of the Mbundu states before
1650 were the plurality and diversity of the institutions found within
them. In addition to the set of titles which served to define the king­
dom, all these states also included lineage titles, positions derived
from earlier forms of political authority which had spread in the
area, ad hoc officials such as war leaders and emissaries appointed in
response to specific circumstances, as well as prophets, diviners, and
other (nominally) religious specialists whose claims on the super­
natural constantly threatened to detract from the equally mysterious
powers of more explicitly political officials.
The kingdom of the ngola a Iciluanje has left the best evidence in
support of this point. The ngola a Iciluanje state, narrowly defined,
consisted of a network of titles based on jingola authority emblems,
but the kingdom actually worked in terms of the corporate lineages
which controlled the jingola. These lineages had predated the expan­
sion of the ngola a Iciluanje and continued to exist through the state’s
history. At times, judging from the evidence available, lineage
officials used their control over the state-titles based on the ngola to
play an important role in the politics of the kingdom. Simultaneously,
non-lineage positions based on such earlier forms of authority as the
malunga and the Libolo mavunga received important commissions
from the central kings and/or provided centres of localism opposed to
the tendencies towards increased centralization represented by the
central kings. War leaders, appointed by the ngola a Iciluanje for the
conduct of a single campaign, must have represented another potent
element in the balance of power in the state, even though the record
of their influence has not survived in the available evidence. We may
assume, from comparison with other times and places, that un­
recorded prophets also arose from time to time and had to be taken
account of by those who presumed to control the state.
The emphasis on plurality and diversity of authority leads to
several related general points. For the historian, who is less interested
in typologies of state-structures than in examining the processes by
270 C O N C L U SIO N S
which states come into being and change through time, the tension
created by the simultaneous presence of a variety of competing forms
of authority has significant heuristic value. A major theme of
Mbundu political history, for example, concerned the constant
tension between the particularism of the descent groups and expan­
sions in political scale promoted by the kingdoms. In such areas as
Songo, particularism seems to have won out most often, while
farther west among the Libolo, Ndongo, Lenge, and others, king­
doms occasionally managed to suppress lineage loyalties in favour of
a wider political perspective. The growth of the usual sort of Mbundu
kingdom from a single local position to a widespread network of
related titles may also be interpreted as a different type of confronta­
tion between authority emblems in a pluralistic universe. The rise of
the ngola a Icihtanje must have appeared locally as a contest between
the holders of the ngola and those with mavunga, malunga, and what­
ever other emblems may have commanded the loyalties of Mbundu
on the eve of its expansion.
The process of forming Mbundu states necessarily produced this
diversity of authority since expanding kings tended to incorporate
rather than eliminate earlier titles. Later states therefore grew progres­
sively more complex as each one introduced one more (temporarily
dominant) element to the universe of title-holders in any given
area. Later Icings incorporated and reused, as it were, older titles
because of the loyalty that these positions commanded from local
peoples, who presumably accepted new authority forms more readily
when they came cloaked in familiar garb. This prototypical version
of what the British later enshrined as ‘indirect rule’ may have been
necessary to incorporate successfully new areas into an expanding
state. Non-literate peoples, like the Mbundu, seem to dread abrupt
and total changes in their thought and society,3 and this sort of
state mitigated the appearance of change in the theoretically static
world of positional succession and perpetual kinship.

How and Why Mbundu States Were Formed

The history of the Mbundu provides a number of explanations for
the formation of states in a so-called ‘stateless’ environment which
presume no outside stimulus or inspiration from Hamites or anyone
else. In fact, the degree to which local social and ideological con­
ditions affected the structures of almost all Mbundu states precluded
the possibility that simple contact with alien ideas or institutions
3 Horton (1967).
 HO W AND WHY M B U N D U STATES F O R M E D 271
would have significant results for Mbundu politics. Nearly all
Mbundu states can be accounted for in terms of older, perhaps
‘political’ but still non-state institutions found in the society, as the
Ici/ombo grew from Ovimbundu circumcision ceremonies and the
ngola a kiluanje developed out of an earlier form of the ngola. The
assumption which lies beneath all the explanations I have offered is
that people tended to create institutions in response to felt needs,
specifically that the desirability of contact between unrelated mem­
bers of different lineages forced the Mbundu to find ways of structur­
ing relationships between non-lcinsmen. These relations, by definition,
were political, and the variety of state institutions already noted
emerged in response to this need.
The assumption of a functional need4 to relate people without
blood ties explains why political institutions emerge, but it does not
account for why some but not others emerge as dominant. That is, it
does not explain why ‘states’ in the narrower sense of the term
appeared. Mbundu history shows that, whatever the structural
tendencies prompting people to create political structures, non-
structural historical circumstances determined which of the myriad
cross-cutting institutions grew to sufficient size and lasted long
enough to be termed ‘states’. No doubt, more than one factor
promoted the growth of any single kingdom, but for purposes of
summarizing the experience of the Mbundu it may be useful to
classify the sorts of historical circumstances which were important,
illustrating each type with examples drawn from Mbundu history.
(1) control over a scarce but valuable resource—the relevant
Mbundu examples include the authority which some of the Pende
lunga-kings exercised over the salt pans of the Baixa de Cassanje, as
well as the iron-ore deposits in the Nzongeji river valley controlled
by the ngola a kiluanje. The ngola a kiluanje Icings seem also to have
moved toward control of the Baixa de Cassanje salt deposits as they
expanded their kingdom and to have claimed other salt sources
located south of the Kwanza in Kisama. The economic value of these
natural resources stimulated just the sort of social circumstances in
which political institutions might appear and led to the eventual
emergence of states. A rare but necessary resource such as iron or
salt presumably attracted unrelated persons from a large area in
4 My use o f these ‘functionalist’ insights does not imply acceptance o f the
early functionalist tendency to emphasize the static and harmoniously inte­
grated aspects o f a society. Cf. the uses made o f tension by Max Gluckman or
the dialectical opposition of elements in African societies in Coquery-Vidro-
vil.ch (1969).
272 C O N C L U SIO N S
search of the desired commodity. As these unrelated individuals
congregated about the salt pans or iron diggings, the etiquette
provided by purely kinship links would not have sufficed to regulate
their contacts with one another which, we may assume, would not
always have been friendly.
As disorders increased among strangers flocking to the site of salt
or iron supplies, a number of solutions might theoretically have been
possible, including the establishment of a market-place authority
like that associated with the ‘port of trade’ defined by Karl Polanyi.5
In such a case, a local authority might originally have regulated
contacts between strangers present at the resource without claiming
any authority over them after they left to return to their own descent
groups. The crucial shift from limited and local ‘port of trade’
market regulation of this sort to a kingdom would have occurred
when the market authorities had accrued wealth and prestige enough
to allow talented or ambitious local governors to convert their
economic capital into political authority, forming a state claiming
jurisdiction over adjacent and even distant lineages as well as over the
natural resource itself.6 These incipient kings could then have
awarded titles derived from their own position to formalize cle facto
links between outlying lineages and themselves, in effect licensing
their trading partners to share in the political wealth. If force was
involved in this process, as it undoubtedly was, the ability of Mbundu
kings to convert their economic resources (i.e. wealth derived from
possession of the resource) into human capital by purchasing slave
retainers provided aspiring rulers with a means of acquiring the man­
power they would need to convince dissidents of the legitimacy of
their authority.7
(2) military or strategic position—at least one Mbundu example,
that of the Mbondo, suggests that occupation of a strong defensive
position may have led to the creation of a Idngdom centred on a site
such as that of the Katanya escarpment above the Baixa de Cassanje.
The circumstances promoting the adoption of a state form of
political organization in this case would have been the same as those
found at the type of economic resource described in category (1)—the
s Karl Polanyi, C. M. Arensberg, H. W. Pearson (1957), and Karl Polanyi
(1963). The relevant aspect of the concept that I emphasize is the independent
market-policing authority in its political function, rather than the economic
(price-administering) functions to which Polanyi devoted most attention.
6 A conversion analogous to the distinction drawn between subsistence-
oriented and market trade in David Birmingham (1971).
7 On slaves and ldngs, Miller (forthcoming (b)), and Terray (1974).
 HOW AND WHY M B U N D U STATES F O R M E D 273
presence of a variety of strangers, in this case refugees from lineages
gathered together for self-defence, who needed to find a basis for
co-operation where kinship ties did not extend.
Once states existed, whatever the causes which first brought
strangers together, a variety of new factors encouraged the formation
of new states at defensible sites, especially if they lay in the interstices
between large neighbouring kingdoms. In point of historical fact, the
Mbondo ndala kisua emerged under the latter circumstance, since its
location atop the mountains of Katanya placed it on the frontier
where Libolo influence confronted the Pende toga-kings of the lower
Baixa de Cassanje.
(3) institutional innovations capable of attracting manpower—the
case of the Imbangala kilombo shows the potential for state-formation
in the ability to recruit large numbers of men and to co-ordinate their
activities. The amassing of manpower played an especially critical
role in the particularistic and divisive atmosphere created by the
strong Mbundu descent groups. The kilombo, which was able to
suppress nearly all potentially fragmenting claims from lineage or
other types of authority, represented the fullest development of this
technique in known Mbundu history. But the same element of
institutional innovation was present in nearly all examples of
Mbundu state-formation, since the institution of slavery equally
represented a means of divorcing people from their old (in this case,
lineage) ties and attaching them to new structures (usually the Icing,
in the instance of Mbundu kingdoms). Slaves were therefore closely
associated with Icings throughout Mbundu history.
(4) ideological innovation—the power of an idea should not be
underestimated in the history of state-formation among the Mbundu,
since an essential component of every title-holder’s political authority
was his ability to convince people that he not only had a right to rule
but also possessed supernatural methods of implementing the powers
he claimed. In this sort of environment, the rise and fall of kingdoms
depended in a very real way on the elaboration of ideas to which
people responded with loyalty. The history of the kilombo, where the
Imbangala ideology of non-humanness at first terrified their oppon­
ents into submission but then failed to retain their loyalties because of
its extreme break with prevailing Mbundu cosmology, provided
the most dramatic example of both success and failure in this
regard.
It was the Mbundu kings’ political use of ideology which obscured
the conventional Western distinction between ‘force’ and ‘magic’,
274 C O N C L U SIO N S
since magical components bulked large in a Icing’s panoply of forces.
The idiom of the political traditions, which described wars exclusively
in terms of charms and magic, showed this clearly, as did the more
tangible concern of Mbundu war-leaders for lengthy magical prepara­
tion when they confronted the always ready Imbangala warriors with
their shield of invulnerability provided by the maji a samba. Correlat­
ive to this point, and also to the importance of organization and
mobilization as political techniques, it may be added that superior
physical weaponry seems not to have played a significant role in state
formation among the Mbundu. Ideological superiority, on the other
hand, may have been an important means of attracting manpower
and thus closely related to the factors discussed under (3).
(5) outside allies—the importance of an outside ally in the forma­
tion of Mbundu states appears in several cases. From the point of
view of an Mbundu lineage, a local political title-holder’s authority
over his kinsmen depended on the support of outsider kings who had
granted the position. Conversely, the power of distant Libolo kings
provided a local wzga-holder’s only guarantee of superiority over the
lineages. The history of the Mbondo again provides an example,
since the lineages of Lambo claimed the ndala kisaa and other Libolo
vi/nga-positions when the expansion of the ngola a kUuanje cut them
off from their sponsors south of the Kwanza. In a somewhat different
sense, alliances between local kings and outsiders may have been
important in the early history of the Lunda, where reactions to and
utilization of the neighbouring Luba provided a continuing theme in
the evolution of the lueji’s ldngdom, the growth of the mwata yamvo,
and the expulsion of the kinguri. The best-documented example of
state-formation by alliance with an outside power occurred when the
Imbangala and Portuguese made contact in a situation where neither
could have succeeded without the aid of the other. This principle
worked on numerous occasions in the history of Portuguese-
Imbangala relations: kulashingo’s use of the Portuguese to gain
ascendancy over the holders of the Lunda makota positions, the
Portuguese alliance with the kulashingo in beginning their territorial
conquests between 1610 and 1620, and later the dependence of such
Imbangala kings as the kabuku ka ndonga and the kalandula on the
Portuguese as the kilombo withered away.
(6) commercial monopolies—the ability to control the movement
of commodities (analytically distinguishable from control of the
economic resource itself) lay at the basis of the slave-trading states
built by later Imbangala rulers in Matamba and Kasanje. The
 HOW AND WHY M B U N D U STATES F O R M E D 275
Portuguese state of Angola thrived on control of another stage in the
same commerce in slaves. Although these cases superficially res­
embled states based on control of a natural resource in that both
were ‘economic’ in nature, they were historically different since
simple control of a resource was capable of generating a state in an
environment devoid of large-scale political institutions. The trading
state, on the other hand, appeared only later when other states
already existed capable of buying or selling sufficient quantities of a
commodity to support this type of parasite kingdom. A trading state
like Angola usually lay between two older states (Kasanje, Matamba,
etc., on one side, and Portugal and Brazil on the other), one in con­
trol of production and the other in charge of distribution or consump­
tion. It was thus historically derivative in ways that a king in control
of a valuable resource was not.
(7) agricultural surplus—no sedentary state could appear without
agricultural techniques capable of producing a surplus to support the
agriculturally non-productive specialists in magic, war, and arbitra­
tion who ran the state machinery. The absence of an agricultural
surplus does not seem to have limited the opportunity to develop
states in the recoverable history of the Mbundu, yet neither does the
simple abundance of food in itself seem to have prompted the forma­
tion of a state. Production in excess of local populations’ need for
food may in any case be treated as a special subcase of the category
(1) of possession of a valuable natural resource, since the historical
relationship between agricultural plenty and the formation of states
depended on creating Icings to regulate access by strangers to the
surplus. That is, extra food would not have been produced, whatever
the potential of the combination of land, climate, and cultigens,
unless farmers had the opportunity to trade it to outsiders who either
had less suitable lands or had specialized in other economic activities
such as trade, crafts, and so on. This category is worth mentioning
separately only because the Imbangala represented the opposite
limiting case. They theoretically grew no food of their own and lived
entirely by preying on the production of others. The history of the
Imbangala showed only that states may exist without any agricultural
base whatsoever if they move continually, seeking new areas to
plunder as they devastate the regions where they have passed.
(8) technological superiority—technological superiority, construed
in the narrow but conventional sense of possession of superior
weapons, may have led to the creation of Mbundu states when the
ngolci arrived if the Samba in fact introduced the Mbundu to
276 C O N C L U SIO N S
techniques of forging iron into broad-bladed weapons. The evidence
on this point, however, is less than clear, and technological superiority
generally seems to have been less important than the ideological and
organizational techniques mentioned under categories (3) and (4). In
particular, Portuguese firearms seem to have made little difference in
the ability of the Europeans to form a state before they combined
these weapons with conventional African weapons in the hands of
their Imbangala allies.
(9) individual genius—the nature of the data, both oral and
written, makes it impossible to assess the role of ‘great men and
women’ in the history of the Mbundu. But all of the eight categories
mentioned above implicitly allow for the role of individual human
genius in the formation of Mbundu states. Clearly, it took an excep­
tional person to convert an early ‘port of trade’ into a full-fledged
kingdom, just as only human inventiveness could devise a successful
new symbol of authority, convert a circumcision camp into a kilombo,
or bring two potentially hostile mercenary camps into a mutually
beneficial alliance. The checkered record of the Portuguese-Imban-
gala alliance, as it passed through the hands of governors and
Imbangala Icings of varying talents, shows how great a difference the
identity and genius of the partners made. The influence of ‘great men
and women’ is implicit in all Mbundu political history, but the
sources highlight it only in exceptional cases.8
The historical process of state-formation may be categorized in a
variety of other ways, of which I shall mention two. First, three
logically exhaustive categories constructed in terms of a new state’s
relation to its predecessors comprehend all the cases of state-forma­
tion known among the Mbundu.9 (a) Primary states may arise
through the conversion of a purely local political institution, as
happened in the case of the kilombo and the ngola and probably in the
rise of the early lurtga-kings in control of salt pans in the Baixa de
Cassanje. These states are distinguished by the extent to which they
were based on local ideas and practices, and they may be uncom­
monly stable and enduring for that reason, (b) Some secondary
states, such as the ndala kisita of the Mbondo, originated as break­
aways from older states. These states had to adapt their original
8 The major exception is, o f course, queen Nzinga o f Matamba; Miller
(1975).
9 This is basically the distinction between ‘primary’ and various types o f
‘secondary’ states favoured by cultural evolutionists; e.g. Morton H. Fried
(1960), and Lewis (1966) for an extension o f this sort o f typology to African
examples.
 HOW A N D W HY M B U N D U STATES F O R M E D 277
alien institutions to local circumstances if they were to survive. The
history of the Imbangala kilombo north of the Kwanza provided
abundant evidence both of the process of disintegration, as the
holders of the makota titles fled the kulashingo after 1619, and of their
subsequent tendency to embrace Mbundu ideas and institutions, (c)
Other states appeared independently of outside control but in
imitation of institutions present nearby. This process, known to
anthropologists as stimulus diffusion, accounted (possibly) for part
of the history of the successive kinguri- states east of the Kwango and
probably for some of the Imbangala bands which appeared farther
west during the late sixteenth and early seventeenth centuries. Given
the deficiencies in the historical record, it may be difficult to disting­
uish stimulus diffusion from the extension of an existing state struc­
ture through the award of political titles by one Mbundu lineage to
another. Title-holders may have fabricated false links with neighbour­
ing titles, indistinguishable in the genealogies from real ones, when
historically they merely copied the idea without the formal award of
a title.
The second alternative method of describing Mbundu state-
formation develops conceptual themes which seem to have recurred
in several cases. I choose to deal with three of these: the concept of
the ‘outsider’, the importance of manpower, and the function of the
king as arbiter. The history of the Mbundu leaves little doubt that
states resulted in most cases from the presence of people defined
according to local standards as ‘outsiders’.10 For the Mbundu, only
kinsmen qualified as ‘insiders’, while all non-relatives had the status
of ‘outsiders’ with regard to the underlying lineage structure of the
society and the fundamentally kin-oriented cosmological system.
M bundu political structures, by definition, were organizations which
structured relationships between ‘outsiders’, or non-lcin. Mbundu
kings were themselves outsiders, removed from their descent groups
and kinsmen through initiation ceremonies which placed them in a
non-lineage limbo where they acted as theoretically neutral arbiters
in disputes which divided competing groups of kinsmen.
Kings mediated relations between ‘outsiders’, whether they acted
as market authority over traders congregated at a salt pan, as judges
between unrelated lineages in their kingdom, or as war leaders against
warriors from an alien neighbouring people. Kings could justify their
monopoly of trade in commercial states since theirs was the responsi­
bility for regulating contact between the subjects of the kingdom (the
10 Here my debt is to Lucas (1971) and Igor Kopytoff (1971).
278 C O N C L U SIO N S
insiders) and traders who came from the outside. This definition of a
king’s power accounts for the limitations on their powers since their
authority extended only to relationships between people who re­
garded one another as ‘outsiders’. They could not interfere in the
internal functioning of the constituent institutions of their states,
notably the descent groups in the case of the Mbundu. Hence the
plural nature of authority in Mbundu kingdoms. Political history
among the Mbundu was to a large extent the history of ‘outsiders’,
the kings, in their attempts to extend their authority over the relations
between strangers. And the Mbundu traditions’ unanimous attribu­
tion of state-founding to such ‘outsider’ hunters and conquerors as
Ngola Inene and Cibinda Ilunga provides metaphorical confirmation
that the Mbundu themselves saw things in this way.
The second major theme, that of manpower, derives its importance
from the fact that Mbundu society was non-literate and hence non-
technological. Europeans later built states and expanded their range
of political hegemony by applying administrative and military tech­
nology simply not attainable under the conditions of the oral
Mbundu society. Thus, Mbundu kings could not create institutions of
ideological penetration in the absence of pervasive communication
technology, nor instruments of physical coercion without the ability
to train, equip, and provision large armed forces, nor firm adminis­
trative control without bureaucratic structures based on written
records and instructions. Mbundu Icings could erect states only on the
basis of sheer manpower, their ability to co-ordinate the raw labour
of relatively large numbers of people in pursuit of ends they defined.
The critical importance of absolute members of people in state-
formation gave slaves their important position at the heart of many
Mbundu kingdoms. The unusual successes of the Imbangala de­
pended on recruitment and organizational techniques which gave
them literally incredible (in the context of Mbundu ideas) advantages.
The numbers of people surrounding a powerful king at his populous
capital were the visible manifestation of this principle and were the
only sources of strength which distinguished him from the impover­
ished and lonely priest or holder of an obsolescent position, whose
ancient and grandiose title theoretically ranked him above the suc­
cessful king except for the fact that he had no followers. The portrait
of the proud but ragged African ‘king’ recurs often in the accounts of
later travellers who met these individuals, and such ancient Mbundu
‘kings’ as the /zwga-holding Butatu persisted in this form long after
abandonment by their followers had caused them to lose effective
 C O N C L U SIO N S A BO U T M B U N D U H IST O R Y 279
power. The legend of the Icinguri’s death provided an explicit Mbundu
acknowledgement of the importance of manpower.
Third, the emphasis on the pluralistic structure of Mbundu states
identifies the essential role of the king as arbitrator between the
competing groups in the kingdom. As the chief judicial official, he
weighed the conflicting claims of lineages to find peaceable settle­
ments and enforced his decisions through whatever sanctions he
could muster. The important implication of the king’s arbitration
function is that it emphasizes the initiative of the local population in
creating a kingdom and accepting the king, in deliberately introduc­
ing stratification of a kind which did not exist where there were no
states. It consequently removes the need to develop ‘conquest’
hypotheses to explain why people would assume a position subordin­
ate to a monarch. The neutrality imparted by the king’s status as
‘outsider’ supported his reputation as an arbiter capable of rendering
impartial judgement.

Conclusions about Mbundu History

In addition to conclusions about the general process of state-
formation, as represented by the Mbundu, some fresh insights into
the course of Mbundu history emerge from the story of these king­
doms. In so far as the Mbundu may be said to have existed as a group
separate from their neighbours, it seems useful to view the formation
of Mbundu states in terms of the interaction of three adjacent ethno­
cultural complexes, Kongo, Ovimbundu, and the Lunda/Cokwe/
Lwena area of strong segmentary lineages, each with certain dis­
tinguishing characteristics above and beyond the elements of langu­
age, economy, and thought which they shared. The Iuiiga was a
nearly indigenous form of Mbundu political organization, whose
origins may be traced back to ideas either so old or so basic to
Mbundu culture that external sources of influence cannot be identi­
fied. Most later Mbundu political history, however, consisted of
successive waves of influence emanating from the neighbouring areas:
the ngola came from Kongo, the mavunga and later the kilombo from
the Ovimbundu', and political titles connected with the Imbangala
came to the Mbundu from the Lunda and Colcwe.
ft should be emphasized that this model is a relative one, ultimately
derived from the arbitrary focus on the Mbundu as the subjects of the
study. Presumably a parallel model of a central area with some
indigenous state forms and a number of influences from neighbouring
regions would also have emerged had I chosen the Kongo, the
280 C O N C L U SIO N S
Ovimbundu, or the Cokwe/Lwena as protagonists. Some indication
of how an alternative focus would have yielded a parallel analysis
may be seen in the brief review of Luba influences on the Lunda in
Chapter V. It may also be pointed out that the study as it stands
presents only one side of what was in fact a two-way process, since
Mbundu political structures presumably were influencing their neigh­
bours as much as their neighbours influenced the Mbundu. The only
glimpse of the complementary process was seen in the spread of titles
based on the ngola far south of the Kwanza into Ovimbundu areas.
From what we have seen of state-formation among the Mbundu,
it would seem futile to identify, as many studies claim to have done,
the ultimate ‘origins’ of states anywhere in Africa. The ‘earliest’
Mbundu states are no longer visible in the surviving evidence, and
processes of state-formation which characterized the Mbundu
suggest that political (if not ‘state’) institutions have existed, almost
by definition, since the remote coalescence of the economic and social
bases of Mbundu society as it existed by the sixteenth century.
‘States’, too, were therefore very old, and the keys to the most ancient
phases of Mbundu political development await archaeological
investigation rather than the conventional search for documents and
oral traditions. The available evidence paints only a selective picture
of the conversion of some (not all) very old political institutions to
larger-scale kingdoms but does not reveal the ‘origins’ of states for
the Mbundu. In a certain sense, there was very little ‘new’ in Mbundu
political history, since most of the story consisted of adaptations of
existing ideas and institutions to new purposes.
Neither simple migration nor diffusion hypotheses make much
sense of state-formation in the case of the Mbundu. The ngola, the
kinguri, and the vunga show how institutions spread without a
corresponding movement of people, particularly not the large-scale,
rapid mass migrations which lay behind such theories as the ‘Hamitic
hypothesis’ and its derivatives. In addition, the relative insignificance
of weaponry makes so-called ‘conquest’ theories seem unlikely in the
light of the Mbundu experience; arms may account for victories on
the battlefield, but they are insufficient to explain the much more
complex process of forming a state. The tendency of the Mbundu
traditions to personalize abstract ideas accounts for the appearance
of ‘migrating conquerors’ in the oral histories, and the role of the king
as an ‘outsider’ explains why he was said to have come from far away.
In actual fact, the appropriation by local ambitious and clever men of
someone else’s good idea seems a far more likely explanation of most
 C O N C L U S IO N S ABO U T M B U N D U H IS T O R Y 281
early Mbundu states. It was thus the idea or the institution which
travelled in most cases, while the basic population of the Mbundu
region has remained relatively stable for a very long time. Otherwise,
it would be difficult to account for the obvious antiquity of most of
the lineages in the region.
Diffusion hypotheses, while closer to historical fact in some ways,
must be applied very carefully, since the experience of the Mbundu
shows that the simple availability of an idea diffused from the outside
did not guarantee its implementation or long-term success. The
Mbundu transformed both the mavunga and the kilombo, for
example, after they adopted them as methods of political organiza­
tion. In the case of the kilombo, the transformation was so dramatic
that it lost its distinctive characteristics and became assimilated to the
prevailing local type of political title. Thus diffusion did not explain
state-formation but merely provided the opportunity for local
innovators to change an outside idea into a form which they could use
to create new states. State-formation occurred in the adaptation and
modification processes rather than in the spread of the idea. The key
lay in the institutionalization rather than in the innovation, since new
ideas and institutions appeared constantly, both from within and from
without, but few left a permanent mark on Mbundu political history.
Mbundu state-formation never took place as an ‘event’ in the sense
of introducing fully-developed state structures into an area and
simply implanting them in a form which remained largely unaltered
until picked up much later by written sources and modern ethno­
graphers. This false impression of state-formation is, of course, a
logical corollary of the old and discredited assumption that little
changed before the coming of Europeans disturbed the static equili­
brium of African state forms and caused them to evolve for the first
time. The notion of states as composed of multiple historical layers is
useful because it forces us to recognize that Mbundu state structures
resulted from lengthy historical development, as successive waves
of political innovation altered the organizational components of
Mbundu society. Each stage differed from its predecessor in the
addition of one or several new elements and in the changed roles of
the older titles and symbols. Only a hypothesis which takes specific
account of time and sequence can explain the often bewildering com­
plexity of mature Mbundu and other African states. The often noted
multiplicity of officials, internal contradictions, and reduplication of
functions resulted from the nature of a multi-layered state in which
much was added but relatively little was dropped.
282 C O N C L U SIO N S
The history of the Mbundu adds another nail to the coffin which
already encases simplistic evolutionary hypotheses which posit such
unidirectional movements as development from simple to complex
forms, from small-to larger-scale institutions, from primitive com-
munalism to feudal states, or whatever else. Mbundu political history
; moved in no single direction but consisted of an irregular alternation
between the triumph of institutions based on the loyalties of kinship
and those articulating the demands of kings. Kings repeatedly rose-to
claim the services of Mbundu kinsmen, who persistently saw them­
selves first as members of their lineages and only secondarily, and
usually under duress, as obedient to,outside authority. Many of the
kings fell foul of the lineages in the long run. Ancient and powerful
lunga-kings’ titles, for example, survive today as minor descent group
positions in remote corners of the area their holders had once ruled.
The ngola a kilmnje kingdom foundered in part on the particularistic
rivalries of lineages within its domain. The kilombo, most notably,
disappeared because it implied too extreme a restructuring of the
underlying Mbundu kinship structure. The lineages were constantly
in the background of Mbundu political history, and the long record
of state-formation and disintegration which has formed the subject of
this study-has as yet done little to woo Mbundu kinsmen away from
their descent groups.
It is not really ironic to conclude a study ostensibly concerned with
states with the remark that the real history of Mbundu political
institutions lies in the non-political milieu in.which they existed. The
differences in the history of the kilombo among the strongly segment­
ary lineages of the eastern Mbundu, among the western Mbundu
under the sway of the ngola a Iciluanje, and among the non-Mbundu
south of the Kwanza demonstrate more clearly than any other case
the importance of the societal background in understanding state-
formation. As the antiquity of states among the Mbundu, and the
elusiveness of the distinction between non-state, statelike, and state
political structures demonstrate, the potential for states was inherent
in the milieu, and the conditions which stimulated and guided the
emergence of mature states likewise were found in the social and
economic surroundings. If any single lesson emerges from the history
of the Mbundu, it is that a fascination with the customary but false
opposition between ‘states’ and ‘stateless societies’ has lured histori­
ans away from the local bases of most states in a futile search for
Hamitic and other equally chimerical exotic bringers of civilization.
 Bibliography

I. LIST OF ORAL TESTAMONIES CITED 1

B ango B w ila M w a N d on je
Im bangala Group M w anya a M uhim ba
K abari ka Kajinga M w a n y a a S h ib a — 1 4 ,1 5 June 1969
K asanje ka K anga ndala kandumbu — see Sousa
K asanje ka N zaje C alunga
K ijinda ka N o k en a N gan d u a K ungu
K iluanje kya N gon ga N g o n g a a M bande— 26 June 1969
(see also K asanje ka K anga)
Kilundula
N g u n za a K asanje
K im bw ete
N zaje
K ingw anga kya M bashi
S ok ola
ICisua kya N jinje, K am bo ka
Kikasa, Sousa Calunga Sousa C alunga— 16 June; 9, 21,
22, 24, 26, 28 July; 21, 22, 23
K itubiko A ugust; 10, 11, 29, 30 Septem ­
K ulashingo ber; 1 , 2 , 10 O ctober 1969
L uciano, W encislau Sousa Calunga, K am bo ka K ikasa
M ahashi V az, A lexandre— 30, 31 July 1969
M atam ba, Luis V az, A lexandre, and D o m in g o s
V az
M atos, A p o lo de— 1 8 ,1 9 , 20 June;
7, 8 July; 4, 5, 6, 7 O ctober 1969 V az, A lexandre, N g o n g a a
M bande
M bande a N g o n g o
Vaz, D o m in g o s
M bondo G roup
V az, D o m in g o s, A lexandre V az,
M bum ba a K asam bi N g o n g a a M bande
M ushiko a K ingw angw a V az, M anuel

II. LIST OF PUBLISHED WORKS CITED

A breu e o m in g o s de. See Brito.

B r it o , D
A b s h ir e , D M ., and M ic h a e l A . S a m u e l s , eds. 1969. Portuguese
a v id
Africa: A Handbook. N ew Y ork.

1 The designations refer to an index o f tapes and notes in the possession of

the author. In general, the name o f the major informant(s) identifies the
interview; however, in cases where more than one interview was held with
the same individual, dates indicate the second and all following sessions.
284 B IB L IO G R A P H Y
A iry Shaw , E. K. 1947. ‘The Vegetation of Angola’. Journal of Ecology,
XXXV, 23-48.
A jayi, J. F. A de and M ichael C rowder, eds. 1971-3. History of West
Africa. 2 vols. London.
A kinjogbin , I. A. 1971. ‘The Expansion of Oyo and the Rise of Dahomey,
1000-1800’. In Ajayi and Crowder, eds., History of West Africa, I,
.... 309-43.____ ,
Alagoa, E. J) 1970. ‘Long-distance Trade and States in the Niger
Delta’. JAH, XI.3, 319-30.
---------------- . 1971a. ‘The Development of Institutions in the States of
the Eastern Niger Delta’. JAH, XII.2, 269-78.
---------------- . 1971b. ‘The Eastern Delta City States and their Neighbors,
1600-1800’. In Ajayi and Crowder, eds., History of West Africa, I,
269-303.
A lbuquerque F elner, A lfredo de. 1933. Angola; Apontamentos
sôbre a ocupação e início do estabelecimento dos Portugueses no Congo,
Angola e Benguela, estraídos de documentos históricos. Coimbra.
---------------- . 1940. Angola; Apontamentos sôbre a colonização dos
planaltos e litoral do sul de Angola. 3 vols. Lisboa.
A lmeida, A ntonio de. 1937. Sôbre mutilições étnicas dos aborígenes de
Angola. Lisboa.
A lves, P e A lbino. 1951. Dicionário etimológico Bundo-Português.
2 vols. Lisboa.
A ngola . Serviços Meteorológicos. 1955. O clima de Angola. Luanda.
A rmstrong, R. G. 1960. ‘The Development of Kingdoms in Negro
Africa’. JHSN, II. 1, 27-39.
Assis J unior, A ntonio de. n.d. Dicionário Kimbundu-Portugues.
Luanda.
A tkins, G uy. 1954. ‘An Outline of Hungu Grammar’. Garcia de Orta,
II.2, 145-64.
---------------- . 1955. ‘A Demographic Survey of the Kimbundu-Kongo
Language Border in Angola’. BSGL, LXXIII. 7-9, 325-47.
B al, W illy, ed. 1965. Description du Royaume de Congo et des Contrées
Environnantes (par Filippo Pigafetta et Duarte Lopes-1591). Louvain
et Paris.
Balandier, G eorges. 1968. Daily Life in the Kingdom of the Kongo.
Trans. Helen Weaver. New York.
---------------- . 1970. Political Anthropology. Trans. A. M. Sheridan
Smith. New York.
Barbot, J ean. 1732. A Description of the Coasts of North and South-
Guinea, and of Ethiopia inferior, vulgarly Angola . . . London.
Baseheart, H arry W. 1972. ‘Traditional History and Political Change
among the Matengo of Tanzania’. Africa, XLII.2, 87-97.
B astian, A. 1874—75. Die Deutsche Expedition an der Loango-kiiste.
2 vols. Iena.
 B IB L IO G R A P H Y 285
B attell, A ndrew . See Ravenstein, ed.
Baumann, H ermann. 1935. Lunda: Bei Bauern und Jägern in Inner-
Angola. Berlin.
---------------- . 1936. Schöpfung und Urzeit des M enschen in M ythes des
Afrikanischen Völker. Berlin.
B aumann, H ermann, and D. W estermann. 1962. L es Peuples et
civilisations de TAfrique. Trans. L. Homburger. Paris. (Originally
Völkerkunde von A frika, Essen, 1940)
B eattie, J ohn . 1964. O ther Cultures. London.
Beidelman, T. O. 1970, ‘Myth, Legend, and Oral History: A Kaguru
Traditional Text’. Anthropos, LXV.l, 74-97.
Biebuyck, D aniel. 1957. ‘Fondements de l’organisation politique des
Lunda du Mwaanta Yaav en territoire de Kapanga.’ Zaire, XI. 8,
787-813.
Birmingham, D avid. B. 1965. ‘The Date and Significance of the Imban­
gala Invasion of Angola’. JA H , VI. 2, 143-52.
---------------- . 1966. Trade and Conflict in Angola. Oxford.
---------------- . 1970. ‘Angola and its Hinterland’. In Gray and Birming­
ham, eds., Pre-Colonial African Trade, pp. 163-73.
Bohannan. P aul . 1953. ‘Concepts of Time Among the Tiv of Nigeria’.
S W J A , IX. 3, 251-62.
B ohannon, P aul, and L aura. 1953. The Tiv o f Central Nigeria. London.
B ontinck , F r , ed. 1970. Diaire Congolais (1690-1701) . Louvain et Paris.
Boston, J. S. 1964. ‘The Hunter in Igala Traditions of Origin’. Africa,
XXXIV.2, 116-25.
---------------- . 1969. ‘Oral Tradition and the History of the Igala’. JA H ,
X.l, 29-44.
Bouveignes, O liver, de and J. C uvelier. 1951. J érőmé de Montesarchio.
Namur.
Bovill, E. W. 1968. Golden Trade o f the M oors. 2nd rev. ed. London.
(Originally Caravans o f the O ld Sahara, London, 1933.)
B owdich , T homas. 1824. A n Account o f the Discoveries o f the Portuguese
in the Interior o f Angola and M ozambique. London.
B oxer, C harles R. 1952. Salvador de S á and the Struggle fo r B razil and
Angola, 1602-1686. London.
B randão, P aes. 1904. ‘Diário da marcha do chefe do Conselho do
Libollo, Tenente Paes Brandão, à região da Quibala’. P A , XI. 22-5,
76-9, 137-40, 223-37, 288-91, 349-55, 406-12, 481-5.
B rásio, A ntônio . 1952-71. M onum enta M issionária Africana-Âfrica
Ocidental (Sér. I). 11 vols. Lisboa.
B rito, D omingos de A breu e. 1931. Um inquérito à vida administrativa de
Angola e do Brazil. Ed. Alfredo de Albuquerque Feiner. Coimbra.
B uchner, M ax. 1883. ‘Das Reich des Mwata Yamvo und seine Nachbar­
länder’. Deutsche Geographische Blätter, VI, 56-67.
C adornega, A ntonio de O liveira de. 1940-2. História Geral das
286 B IB L IO G R A P H Y
Guerras Angolanas, 1680. Eds. José Matias Delgado and M. Alves da
Cunha. Lisboa.
C altanisetta, F ra L uca da. See Bontinck .
C annecattim, Bernardo M aria de . 1804. Diccionario da lingua bunda
ou angolense, explicada na lingua portugeza e latina. Lisboa.
------------------ . 1854. Collecção de observações grammaticaes sobre a
lingua Bunda ou Angolense e Diccionario da lingua Conguenza. 2nd ed.
Lisboa.
C a pello ,H., and R. I v e n s . 1882. From Benguella to the Territory o f
Yacca. Trans. A. Elwes. 2 vols. London.
C ardoso, F onseca. 1919. E m terras do M oxico: Apontam entos de
etlmographia angolense. Porto.
C arneiro, J oão Vieira . 1859-61. ‘Observações feitas em 1848 relativas a
diversos objectos que lhe pareceram não exactos no 3o volume dos
Ensaios sobre a Estatística das Possessões Portuguezas da Africa
Occidental pelo Conselheiro José Joaquim Lopes de Lima’. A C U , II,
174-9.
C arneiro, R obert L. 1968. ‘Slash-and-burn Cultivation among the
Kuikuru and its Implications of Cultural Developments in the Amazon
Basin’. In Yehudi A. Cohen, ed., M an in Adaptation, The Cultural
Present, Chicago, pp. 131-45. (Reprinted from Antropologica (Cara­
cas), Supp. No. 2 (1961).)
Carvalho, H enrique A ugusto D ias de. See D ias de C arvalho.
‘Catalogo dos G overnadores do R eino de A ngola’. See M onteiro
de M orais(?).
Cavazzi de M ontecuccolo, P e J oão A ntónio. 1965. Descrição histórica
dos Três reinos do Congo, M atam ba e Angola. Trans, and ed. Pe
Graciano Maria de Luguzzano. 2 vols. Lisboa.
C hatelain, H éli. 1888-89. Grammatica elementar do Kimbundu ou
lingua de Angola. Genebra.
---------------- . 1894. ‘Folk-Tales of Angola’. M em oirs o f the American
F olk Lore Society, I, 1-315.
C h il d s , G l a d w y n M. 1949. Umbundu Kinship and Character. London.
---------------- . 1964. ‘The Kingdom of Wambu (Huambo), A Tentative
Chronology’. JAH, V. 3, 367-79.
---------------- . 1970. ‘The Chronology of the Ovimbundu Kingdoms.’
JAH, XI. 2, 241-8.
Chinyanta N ankula, M wata K azembe XIV. 1961. Historical Tradi­
tions o f the Eastern Lunda. Trans. Ian Cunnison. Lusaka. (Rhodes-
Livingstone Cummunication no. 23.)
C ohen, D avid W. 1972. The Historical Tradition o f the Busoga: Mitlcama
and Kintu. Oxford.
C oimbra, D r . C arlos D ias de. 1953. Livro de Patentes do Tempo de
Salvador Correia de Sá e Benevides. Luanda.
 B IB L IO G R A P H Y 287
CoQUERY-ViDRoviTCH, C atherine . 1969. ‘Recherches sur un mode de
production africain’. L aP ensée, no. 144, 61-78. Trans. Susan Sherwin,
in Martin A. Klein and G. Wesley Johnson, eds., Perspectives on the
African Past, Boston, 1972, pp. 33-51.
C ordeiro, L uciano, ed. 1881. Viagens, explorações e conquistas dos
Portugueses: Colleção de documentos. 6 vols. Lisboa.
C o r d e ir o da M a t t a . J. D. 1893. Ensaio de diccionario kimbundu-
portuguez. Lisboa.
C orrea, A ntonio da Silva. See Silva C orrea.
C r in e , F e r n a n d .1963. ‘Aspects politico-sociaux du système de tenure
des terres des Luunda septentrionaux’. In Daniel Biebuyck, ed.,
African Agrarian System s (International African Seminar), London,
pp. 157-72.
C r in e - M a v a r , B r u n o . 1963. ‘Un Aspect du symbolisme Luunda:
l’association funéraire des Acudyaang’. M iscellanea Ethnographica
{M R A C , no. 46), Tervuren, pp. 79-108.
— ------------ . 1968. ‘A propos de la sterilité artistique des Lunda’.
Études congolaises, XI. 2, 59-67.
---------------- . 1973. ‘Histoire traditionnelle du Shaba’. Cultures au
Zaire et en Afrique, I, 5-108.
C unha M attos, R. J. da. See M attos.
C unnison, I an . 1956. ‘Perpetual Kinship: A Political Institution of the
Luapula Peoples’. R L J , no. 20, 28-48.
C urtin , P hilip D. 1968a. ‘Epidemiology and the Slave Trade’. Political
Science Quarterly, LXXXIII. 2, 190-216.
---------------—. 1968b. ‘Field Techniques for Collecting and Processing
Oral Data’. JAH, IX. 3, 367-86.
------------- :—. 1969. The Atlantic Slave Trade: A Census. Madison.
D avidson, Basil. 1959. L ost Cities o f Africa. Boston.
---------------- . 1961. The African Slave Trade. Boston.
D e c k e r , H. C. de. 1939. ‘Die Jagazüge und das Königtum im mittleren
Bantugebiet’. Zeitschrift fü r Ethnologie, LXXI. 4-6, 229-93.
D e l a fo sse , M a u r ic e . 1912. Haut-Sénégal-Niger. 3 vols. Paris.
•— -------------. 1931. Negroes o f Africa. Washington.
D elgado, R alph . 1948—55. História de Angola. 4 vols. Lobito.
‘D escoberta de A ngola e Congo : R elações de A ngola , tiradas do
cartório do C ollegio dos P adres da C ompanhia’. 1883. B SG L, IY.
6, 300-4; 7, 338-52; 8, 364-86.
D ia s de C a r v a l h o , H e n r iq u e A u g u s t o . 1885-6. ‘Expedição ao Muata
Yanvo (Cartas à Sociedade de Geographia)’. B SG L , V. 8, 476-92; VI.
3, 133-62.
---------------- . 1890a. Ethnographia e história tradicional dos povos da
Lunda. Lisboa.
---------------- . 1890b. M ethodo pratico para faliar a lingua da Lunda
contendo narrações históricas dos diversos povos. Lisboa.
288 B IB L IO G R A P H Y
---------------- . 1898. O Jagado de Cassange na P rovinda de Angola,
M emória. Lisboa.
D iniz , J osé de O liveira F erreira. See O liveira F erreira D in iz .
D rake, S t . C lair. 1959. ‘The Responsibility of Men of Culture for
Destroying the ‘Hamitic Myth’.’ Presence Africaine, special issue.
Paris.
D u a r t e , V ic e n t e J o sé . 1859-61. ‘Noticias de alguns dos districtos de que
se compõe esta província (Angola)-Districto do Duque de Bragança’.
A C U , II, 123-8, 129-35, 141-4, 145.
D uysters, L é o n . 1958. ‘Histoire des Aluunda’. Problèmes de VAfrique
Centrale, XII. 40, 75-98.
E d w a r d s , A d r ia n C. 1962. The Ovimbundu under Two Sovereignties.
London.
E hret, Christopher. 1971. Southern N ilotic H istory. Evanston.
E stermann, C arlos. 1956. Etnografia do sudoeste de Angola. 2 vols.
Lisboa.
E vans-P ritchard , E. E. 1939. ‘Nuer Time-Reckoning’. Africa, XII. 2,
189-216.
-----------. 1940. The Political System o f the A nuak o f the Anglo-Egyptian
Sudan. London.
---------- . 1949. The Sanussi o f Cyrenaica. Oxford.
‘E xplorações dos P ortuguezes no interior d ’A frica meridional’.
1843. A M C , III. 5, 162-90; 6, 223-40; 7, 278-97; 9, 423-39; 10, 493-
506; 11, 538-52.
‘E xplorações dos P ortuguezes no sertão d ’A frica meridional’.
1844-45. A M C , IV. 7, 286-300; 8, 303-14; 9, 334-43; 10, 377-81; 11,
397-408; V. 1, 19-26; 2, 63-77; 3, 108-120; 4, 149-64; 5,199-208; 7,
264-78; 9, 321-37; 10, 364-72; 11, 428-37; 12, 468-81.
F a g a n , B r ia n M. 1969. ‘Early Trade and Raw Materials in South
Central Africa’. JA H , X. 1, 1-14.
F age, J ohn D. 1961. A n Introduction to the H istory o f W est Africa.
London.
---------------- . 1965. ‘History. In Robert Lystad, ed., The African World,
New York, pp. 40-56.
■
---------------- . 1974. States and Subjects in Sub-Saharan African H istory.
Johannesburg.
F allers, L loyd A., ed. 1964. The King’s M en; Leadership and Status in
Buganda on the Eve o f Independence. New York.
---------------- . 1965. Bantu Bureaucracy : A Century o f Political Evolution
among the Basoga o f Uganda. 2nd ed. Chicago. (Original edition,
Cambridge, 1956).
F aria, P e F rancisco L eite de. 1952. ‘A Situação de Angola e Congo
apreciada em Madrid em 1643’. P A , IX. 52, 235-48.
F eierman, Steven. 1974. The Shambaa Kingdom, A H istory. Madison.
F elner, A lfredo de A lbuquerque. See A lbuquerque F elner.
 B IB L IO G R A P H Y 289
F eo C ardoso de C asteixo B ranco e T orres, J oão C arlos. 1825.
Memórias contendo a biographia do vice almirnate Luis da Motta Feo e
Torres, A história dos governadores e capitaens generaes de Angola
desde 1575 até 1825, e A Descripcão geographica e política dos reinos de
Angola e de Benguella. Pariz.
F erreira, F rancisco de Salles. See Salles F erreira.
F erreira D iniz , J osé de O liveira. See O liveira F erreira D iniz .
F erreira M arques, A lberto. 1949. ‘Contribuição para o estudo da
etnografia dos povos da Lunda’. MA, 26-27, 77-85; 28, 13-20.
F ic a l h o , C o n d e de. 1947. Plantas Úteis da África portugueza. 2nd ed.
Lisboa.
F in l e y , M. I. 1965. ‘Myth, Memory, and History’. History and Theory,
IV. 3, 281-302.
F o r d e , D a r y l l . 1967. ‘Anthropology and the Development of African
Studies.’ Africa, XXXVIII. 4, 389-406.
F o r d e , D a r y l l and P. M. K a b e r r y , eds. 1967. West African Kingdoms
in the Nineteenth Century. London.
F o r t e s , M e y e r . 1969. Kinship and the Social Order: The Legacy of L. H.
Morgan. Cambridge.
F ortes, M eyer, and E. E. E vans-P ritchard , eds. 1940. African
Political Systems. London.
Fox, R o b in . 1967. Kinship and Marriage. Baltimore.
F r a n ç o is , C u r t v o n . 1888. ‘Geschichtliches bei den Bangala, Lunda und
Kioko’. Globus, LIII, 273-6.
F r ie d , M o r t o n H. 1960. ‘On the Evolution of Social Stratification and
the State’. In S. Diamond, ed., Culture in History, New York, pp.
713-31.
G a l e y , J ea ,n - C l a u d e . 1973. ‘L’Etat et le lignage’. L ’Homme, XIII. 1-2,
71-82.
G o o d y , J a c k . 1968. ‘The Myth of a State’. Journal of Modern African
Studies, VI. 4, 461-74.
---------------- . 1971. ‘The Impact of Islamic Writing on the Oral
Cultures of West Africa’. CEA, XI. 3 (no. 43), 455-66.
G o o d y , J a c k and I a n W a t t . 1963. ‘The Consequences of Literacy’.
CSSH, V. 3, 304-45.
G o s s e w e il e r , John. 1939. Carta fitageografica de Angola. Lisboa.
G o u v e ia , D o m in g o s H. G. 1956. Reconhecimento da Baixa de Cassanje.
Lisboa.
G raça, J oaquim R odriques. 1854-58. ‘Viagem feita de Loanda com
destino às cabeceiras do Rio Sena, ou aonde for mais conveniente pelo
interior do continente, de que as tribus são senhores, principada em 24
de Abril de 1845’. ACU, I, 101-14, 117-29, 133-46. (Republished as
‘Expedição ao Muatayanvua’. BSGL, IX. 8-9 (1890), 3654 68;
excerpted in Boletim Oficial de Angola (1855) and in AA, sér. II, II.
9-10 (1945), 225-39.)
290 B IB L IO G R A P H Y
G ray, R ichard , and D avid Birmingham, eds. 1970. Pre-Colonial
African Trade. London.
G reenberg, J. 1949. ‘The Negro Kingdoms of the Sudan’. Transactions
o f the New York Academ y o f Sciences, ser. II, XI. 4, 126-35.
---------------- , 1972. ‘Linguistic Evidence Regarding Bantu Origins’.
JA H , XIII. 2, 189-216.
G r ev isse , F. 1946-7. ‘Les Traditions historiques des Basanga et de leurs
voisins’. Bulletin du centre d ’étude des problèmes sociaux indigènes,
II, 50-84.
G uebels, L éon . See de Bouveignes.
1930-42. Relação anual das cousas que fizeram os
G u e r r e ir o , F e r n ã o .
padres da Companhia de Jesus nas partes da índia Oriental das cartas
dos mesmos padres que de la vierão. 2nd ed. 3 vols. Coimbra. (First ed.,
Lisboa, 1605.)
G uthrie, M alcolm. 1967. The Classification o f the Bantu Languages.
London.
---------------- . 1967-72. Comparative Bantu: A n Introduction to the
Comparative Linguistics and Prehistory o f the Bantu Languages. 4 vols.
Famborough.
H a m b l y , W. D. 1934. The Ovimbundu o f Angola. Chicago. (Field
Museum of Natural History Publication No. 329: Anthropological
Series, XXI. 2.)
H auenstein, A lfred. 1960. ‘Le Serpent dans les croyances de certains
tribus de Test et du Sud de l’Angola’. Estudos Etnográficos, I, 217-34.
---------------- . 1964. ‘Les Voyages en caravane des Tjiaka.’ Anthropos,
LIX, 926-32.
---------------- . 1967a. ‘Considerations sur le vase sacré ombia yohasa
de la residence royale de Civonga de la tribu des Hanya.’ Anthropos,
L X II, 907-36.
---------------- . 1967b. L es Hanya. Wiesbaden.
H a v e a u x , G. L. 1954. L a Tradition historique des Bapende orientaux.
(IR C B , XXXVII. 1.) Bruxelles.
H eintze, Beatrix. 1970. ‘Beitrage zur Geschichte und Kultur der
Kisama (Angola)’. Paidetima, XVI, 159-86.
H enige, D avid P. 1971. ‘Oral Tradition and Chronology’. JA H , XII. 3,
371-90.
---------------- . 1974. The Chronology o f Oral Tradition. Oxford.
H e u s c h , L u c de. 1972. L e R oi ivre ou Vorigine de Vétat: m ythes e t rites
bantous. Paris.
H omburger, L ilias. 1925. ‘Le Group sud-ouest des langues bantoues’.
M ission Rohan-Chabot (Angola et Rhodesie) 1912-1914, III. 1. Paris.
H o n o r é - N a b e r , S. P. 1’. 1933. ‘Nota van Pieter Mortamer over het
gewest Angola, 1643’. Bijdragen en Mededeelingen van het historisch
Genootschap gevestigd te Utrecht, LIV, 1-42.
H o r t o n , R o b in . 1967. ‘African Traditional Thought and Western
Science’. Africa,'■XXXNW. 1, 50-71; 2, 155-87.
 B IB L IO G R A P H Y 291
---------------- . 1969. ‘From Village to City-State’. In M. Douglas and
P. Kaberry, eds., M an in Africa. Garden City, N. Y., pp. 38-60.
---------------- . 1971. ‘Stateless Societies in the History of West Africa’.
In Ajayi and Crowder, eds., H istory o f West Africa, I, 78-119.
H u f f m a n , T. N. 1972. ‘The Rise and Fall of Zimbabwe’. JA H , XIII.3,
353-67.
H ymes, D. H. 1960. ‘Lexicostatistics so far’. CA, I. 1, 3-44.
I nternational I nstitute of A frican L anguage and Cultures. 1930.
A Practical Orthography o f African Languages. London.
J adin, L. and J. C uvelier. 1954. L'ancien Congo d ’apres les archives
Romaines (1518-1640). (1RCB , XXXVI. 2.) Bruxelles.
J ones, G. I. 1963. ‘European and African Tradition on the Rio Real’.
JA H , IV. 3, 391-402.
J ordão, L evy M aria. See P aiva M anso.
K aberry, P hyllis. 1957. ‘Primitive States’. British Journal o f Sociology,
VII. 3, 224-34.
‘K arte zur Ü bersicht der R eise J oaquim R odrigues G raca’s’. 1856.
Petermanns Geographische M ittheilungen, II, Tafel 17.
K atoke, I srael K. 1971. ‘The History of Pre-colonial States of North­
western Tanzania, An Introduction’. Journal o f World H istory, XIII.
3-4, 512-14.
K e n t , R. K. 1965. ‘Palmares: An African State in Brazil’. JA H , VI. 2,
161-76.
K iwanuka , M. S. M. 1969. ‘The Evolution of Chieftainship in Buganda,
ca. 1400-1600’. University o f E ast Africa Social Sciences Council
Conference (1968)1969), H istory Papers, vol. III. Kampala.
K opytoff, I gor . 1971. ‘Ancestors as Elders in Africa’. Africa, XLI. 2,
129-42.
K ottak, C onrad P. 1972. ‘Ecological Variables in the Origin and
Evolution of African States: The Bugunda Example’. C S S H , XIV. 3,
351-80.
L adeiro M onteiro, R amiro. See M onteiro.
L angworthy, H arry W. 1971. ‘Conflict among Rulers in the History of
Undi’s Chewa Kingdom.’ Transafrican Journal o f H istory, I. 1. 1-23.
L arson, T homas J. 1971. ‘The Hambukushu Migrations to Ngamiland’.
African Social Research, XI, 27-49.
L eitão, M anoel Correia. See Sousa D ias, G astão.
L eite de F aria, pe F rancisco. See F aria.
L eite de M agalhães, A ntônio. 1924. D istrito do Cuanza-Sul: Geographia
Histórica, Físcia, Política e Economica do D istrito; Relatório do
Governador (do Cuanza-Sul) . Lisboa.
L évi-Strauss, C laude. 1966. The Savage M ind. Chicago.
L ewis, H erbert. 1966. ‘The Origins of African Kingdoms’. CEA, VI. 3
(no. 23), 402-7.
292 B IB L IO G R A P H Y
L ima, M esquitela, 1971. Fonctions sociologiques des figurines de culte
H am ba dans la société et dans la culture Tshokwé ( A ngola). Luanda.
L l o y d , P eter C. 1954. ‘The Traditional Political System of the Yoruba’.
S W J A , X. 4, 366-84.
---------------- , 1965. ‘Political Structure of African Kingdoms: An
Exploratory Model’. In Michael Banton, ed., Political System s and
Distribution o f Power. (ASA Monographs No. 2.) London.
---------------- . 1968. ‘The Political Development of West African
Kingdoms’. JA H , IX. 2, 319-29.
L opes, D uarte. See B al.
L opes de L ima, J osé J oaquim. 1846. Ensaios sôbre a statistica das
possessões portuguezes na Africa Occidental e Oriental, vol. III:
Angola e Benguela. Lisboa.
Lux, A nton E. 1880. Von Loanda nach Kimbundu. Ergebnisse der
Forschungsreise in äquatorialen W est-Africa (1875-1876) . Wien.
M c C ulloch, M erran . 1951. The Southern Lunda and R elated Peoples.
(Ethnographic Survey of Africa: West Central Africa, Pt. 1.) London.
---------------- . 1952. The Ovimbundu o f Angola. (Ethnogra phic Survey of
Africa: West Central Africa, Pt. 2.) London.
M c C a l l , D a n ie l F . 1964. Africa in Time Perspective. Boston.
M c G affey, W yatt, 1970. Custom and Government in the Lower Congo.
Berkeley.
M agalhães, A fonso A lexandre de. 1948. ‘Origem dos Basongos’.
M A , 8, 33-8.
M agalhães, A ntonio M iranda . 1922. M anual de línguas indígenas de
Angola. Loanda.
---------------- . 1934. ‘Os Ambundos de Angola e o percurso provável da
sua migração para aquela hoje nossa colónia.’ Trabalhos do I o Con­
gresso Nacional de Antropologia Colonial, Porto - Setem bro de 1934.
2 vols. Porto, II, 537-57.
M agyar, L ászló. 1859. Reisen in Süd-A frika in den Jahren 1849 bis
1857. Vol. I. Pest und Leipzig.
M aia, A ntonio da Silva, n.d.a. Dicionário Elementar P ortugm s-
Omumbuim-M ussele. Cucujäes.
----------- :---- . n.d.b. Dicionário Rudimentar Português-Kimbundo.
Cucujäes.
M a ir , L u c y . 1962. Primitive Government. Baltimore.
M anuel, F rank E., and F ritzie P. 1972. ‘Sketch for a Natural History
of Paradise’. Daedalus, XI. 1, 83-128.
M artins, R. de Sousa. 1973. ‘Etno-história da Quibaxe-Dembos-
Angola: occupações antigas’. Trabalho (Luanda), no. 43, 93-178.
M arwick , M. G. 1963. ‘History and Tradition in East Central Africa
through the Eyes of the Northern Rhodesian Cewa’. JA H , IV. 3,
375-90.
 B IB L IO G R A P H Y 293
M a t t e n k l o d t (ed. H. B a u m a n n ). 1944. ‘Die Kisama.’ Koloniale Völker­
kunde (Wiener Beiträge zur Kulturgeschichte und Linguistik), VI,
71-108.
M a t t o s , R. J. da C u n h a . 1963. Compêndio Historico das Possessões
Portuguezas na Africa. Rio de Janeiro.
M e l o , F r a n c is c o de. See F ic a l h o .
M id d l e t o n , J o h n . 1965. The Lugbara of Uganda. New York.
M il l e r , J o s e p h C. 1970. ‘Cokwe Trade and Conquest’. In Gray and
Birmingham, eds., Pre-Colonial African Trade, pp. 175-201.
---------------- . 1972a. ‘The Imbangala and the Chronology of Early
Central African History’. JAH, XIII. 4, 549-74.
---------------- . 1972b. ‘A Note on Kasanze and the Portuguese’, Canad­
ian Journal of African Studies, VI. 1, 43-56.
---------------- . 1973a. ‘Requiem for the ‘Jaga’.’ CEA, XIII. 1 (no. 49),
121-49.
---------------- . 1973b. ‘Slaves, Slavers, and Social Change in Nineteenth
Century Kasanje’. In Franz-Wilhelm Heimer, ed., Social Change in
Angola, Munich, pp. 9-29.
---------------- . 1975. ‘Nzinga of Matamba in a New Perspective’. JAH,
XVI. 2, 201-16.
---------------- . Forthcoming (a). ‘Cavazzi’s History of the ‘Jaga’.
History in Africa.
---------------- . Forthcoming (b). ‘Imbangala Lineage Slavery’. In Igor
Kopytofif and Suzanne Miers, eds., Slavery, Wardship and Serfdom in
Africa (provisional title).
---------------- . Forthcoming (c). ‘Kasanje Kinglists—Theory and Facts:
Miscalculating African History?’ In David P. Henige, ed., Essays on
Oral Tradition and Chronology (provisional title).
---------------- . Forthcoming (d). ‘Words, History and the Imbangala’.
M ir a n d a M a g a l h ã e s , A n t o n io . See M a g a l h ã e s .
M o n t e ir o , J o a c h im J o h n . 1875. Angola and the River Congo. 2 vols.
London.
M o n t e ir o , R a m ir o L a d e ir o . 1973. A família nos musseques de Luanda.
Luanda.
M o n t e ir o de M o r a is , J o ã o .(? ) ‘Catalogo dos governadores do Reino de
Angola, com uma nota previa noticia do principio da sua conquista e
do que nelle obrarão os governadores dignos de memória.’ In Feo
Cardoso de Castello Branco e Torres, Memórias-, also Academia Real
das Sciencias (Lisboa), Colecção de Noticias para a História e Geo­
grafia das Nações Ultramarinas, III. 1 (1826) and reprinted in A A, sér
I, I. 1 (1937).
M o t a , A v e l in o T e ix e ir a da. 1964. A Cartografia antiga da Africa
Central e a travessia entre Angola e Moçambique, 1500-1860. Lourenço
Marques.
294 B IB L IO G R A P H Y
M outa, F., and H. O’D onnell. 1933. Notícia Explicativa do Esboço
geológico de Angola. Lisboa.
M urdock . G eorge P. 1959. A frica: Its Peoples and their Culture H istory,
New York.
N eves, A ntonio R odrigues. See R odrigues N eves.
‘N oticias do paiz de Q uissama e do E xercito que foi a castigar os
gentios daquella província . . . ’ 1844. AMC, sér. VI. 4, 119-27.
O’F ahey, R ex S. 1970. States and State Formation in the Eastern Sudan.
(African Studies Seminar Paper No. 9.) Khartoum.
O got, B. A. 1964. ‘Kingship and Statelessness among the Nilotes’. In
Vansina, Mauny, and Thomas, eds., H istorian in Tropical Africa, pp.
284-304.
O liveira F erreira D iniz , J osé de. 1918. Populações indígenas de
Angola. Coimbra.
O liver, R oland. 1968. ‘Western Historiography and its Relevance to
Africa.’ In Ranger, ed., Emerging Them es pp. 53-60.
O liver, R oland, and J. D. F age. 1962. A Short H istory o f Africa.
Harmondsworth.
P aes B randão. See B randão.
P aiva M anso, L evy M aria J ordão, Visconde de. 1877. H istoria do
Congo (D ocum entos), 1492-1772. Lisboa.
P ereira do N ascimento, J. 1903. Diccionario Portuguez-Kimbundu.
Huilla.
P igafetta, F elippo . See B al.
P inheiro de L acerda, J. 1845. ‘Noticia da Cidade de Benguella, e dos
costumes dos gentios habitantes daquella Sertão’. A C U , V, 486-8.
P lancquaert, M. 1971. L es Yalca: Essai d'histoire. (M R A C , no. 71.)
Bruxelles.
P ocock, D. F. 1964. ‘The Anthropology of Time Reckoning’. Contribu­
tions to Indian Sociology, VII, 18-29. Reprinted in John Middleton, ed.,
M yth and Cosmos, Garden City, N.Y., 1967, pp. 303-14.
P olanyi, K arl. 1963. ‘Ports of Trade in Early Societies’. The Journal o f
Economic H istory, XXIII, 30-45.
P olanyi, K arl, C. M. A renberg, and H. W. Pearson, eds. 1957. Trade
and M arket in the Early Empires. Glencoe, 111.
R anger, T erence, O., ed. 1968. Emerging Themes o f African H istory.
Dar-es-Salaam.
E. G., ed. 1901. The Strange Adventures o f Andrew Battell o f
R a v e n s t e in ,
Leigh in Angola and Adjoining Regions. London.
R ebello de A ragão, Balthasar. See C ordeiro, vol. III.
R edinha, José. 1958. Etnossociologia do Nordeste de Angola. Lisboa.
---------------- . 1961. ‘Distribuição étnica de Angola’. M A , 107-12, 3-22.
Reprinted as Distribuição étnica de Angola: introdução, registo étnico,
mapa. Luanda.
 B IB L IO G R A P H Y 295
---------------- , 1962. ‘Quem são os Ambundos?’ M A , 183-5, 37-48.
—------------- . 1963. ‘Insignias e simbologias do mando dos chefes
nativos de Angola’. MA, 22-3, 37-46.
---------------- . 1968. ‘Subsídio para a história e cultura da mandioca
entre ospovos do nordeste de Angola’. B IIC Â , V. 1, 95-108.
R ib e ir o da C r u z , J o sé . 1940. N otas de Etnografia Angolana. Lisboa.
R ic h a r d s , A u d r e y . 1940. ‘The Political System of the Bemba Tribe—
North-Eastern Rhodesia.’ In Fortes and Evans-Pritchard, eds.,
African Political System s, pp. 83-120.
R o b e r t s , A n d r e w . 1974. A H istory o f the Bemba. London.
R o d r ig u e s , A. V. 1968. ‘Construções bantas de Pedra em Angola’.
B IIC A , V. 2, 169-89, photos, maps.
R o d r ig u e s , P e F r a n c is c o . 1936. ‘História Inédita de Angola (Manu­
scrito do Século XVI)’. Arquivo Histórico de Portugal, II. Reprinted in
A A , sér. II, XVII. 67-70 (1960), 189-216, and Brasio, M onumenta, IV,
546-81.
R o d r ig u e s N ev es , A n t o n io . 1854. M em ória da Expedição a Cassange.
Lisboa.
R o t b e r g , R o b e r t I. 1965. A Political H istory o f Tropical Africa. New
York.
R o w e , J o h n . 1970. ‘Major Themes in African History’. In John N .
Paden and Edward W. Soja, eds., The African Experience. (3 vols.),
Evanston, I, 154-76.
S a h l in s , M a r s h a l l D. 1961. ‘The Segmentary Lineage: An Organiza­
tion of Predatory Expansion.’ American Anthropologist, LXIII. 2,
322-45.
S a lles F e r r e ir a , F r a n c is c o de. ‘Memória’. 1854-8. A C U , I, 26-8.
Republished in Boletim Oficial de Angola, No. 696 (20 April 1853), as
‘Memória sobre o sertão de Cassange’.
S a l v a d o r in i , V. 1969. ‘Biblioteche e archivi d’Angola’. Bolletino della
Associazione degli A fricanisti Italiani, II. 3-4, 16-29.
S a n d e r s , E d it h . 1969. ‘The Hamitic Hypothesis: Its Origin and Func­
tions in Time Perspective’. JA H , X. 4, 521-32.
S c h e u b , H a r o l d . 1975. Xhosa Ntsom i. Oxford.
S c h ü t t , O t t o . 1881. Reisen in Siidwestlichen Beclcen des Congo. Berlin.
S e l ig m a n , C h a r l e s G. 1957. Races o f Africa. 3rd ed. London. Originally
published 1930.
S il v a C o r r e a , E l ia s A l e x a n d r e da. 1937. H istória de Angola. 2 vols.
Lisboa.
S il v a M a ia , A n t o n io d a . S ee M a ia .
1970. ‘Some Considerations Relating to the Forma­
S m it h , A b d u l l a h i .
tion of States inHàusaland’. J H S N , V. 3, 329-46.
---------------- . 1971. ‘The Early States of the Central Sudan’. In Ajayi
and Crowder, eds., H istory o f W est Africa, I, 158-201.
S m it h , M. G. 1960. Government in Zazzau. London.
296 B IB L IO G R A P H Y
Sousa D ias G astão. 1934. Relações de Angola: Primórdios da occupação
portuguesa pertencentes ao cartório do Colégio dos Padres da Com­
panhia, de Luanda, e transcritas do códice existente na Biblioteca
Nacional de Paris. Coimbra.
---------------- . 1938. ‘Uma viagem à Cassange nos meados do século
x v r ir . B SG L, LVI. 1-2, 3-30.
---------------- . 1942. A Batalha de Ambuila. Lisboa.
Southall, A. W. 1953. Alur Society. Cambridge.
Southwold , M artin . 1961. Bureaucracy and Chiefship in Buganda; The
Development o f Appointive Office in the H istory o f Buganda. (East
African Studies, no. 14.) Kampala.
Statham, J ohn C harles B aron. 1922. Through A ngola-A Coming
Colony. London.
Stevenson, R obert F. 1968. Population and Political System s in Tropical
Africa. New York.
Suret-C anale, J ean. 1964. ‘Les Sociétés traditionnelles en Afrique
noire et le concept du mode de production asiatique’. L a Pensée, no.
117, 19-42.
T ardits, C laude. 1973. ‘Parenté et pouvoir politique chez les Bamoum
(Cameroun)’. L ’H om me, XIII. 1-2, 37-49.
T eixeira da M ota, A velino. See M ota.
T erray, E mmanuel. 1974. ‘Long-Distance Exchange and the Formation
of the State: The Case of the Abron Kingdom of Gyaman’. Economy
and Society, III. 3, 315-45.
T homas, E. V. and D. Sapir. 1967. ‘Le Diola et les temps: Recherches
anthropologiques sur la notion de durée en Basse-Casamance’.
Bulletin de TInstitut Fondemental de VAfrique Noire, sér. B. XXIX,
321-424.
T orre do Valle, E ugenio and J osé Velleso de castro. 1913. 0 Cuanza
desde Quibinda a Cumanga. Lisboa.
T orres, J oão C arlos F eo C ardoso de C astello B ranco e. See F eo
C ardoso.
T ravassos Valdez, F rancisco. 1861. S ix Years o f a Traveller’s L ife in
Western Africa. 2 vols. London.
T urner, Victor W. 1955. ‘A Lunda Love-story and its Consequences’.
R L J , no. 19, 1-26.
---------------- . 1967. The Forest o f Sym bols; Aspects o f Ndem bu Ritual.
Ithaca, N. Y.
---------------- . 1968. The Drums o f Affliction. Oxford.
Valdez, F rancisco T ravassos. See T ravassos V aldez.
van den B yvang, M. 1937. ‘Notice historique sur les Balunda’. Congo,
I. 4, 426-38; 5, 548-62; II. 2, 193-208.
van G eel, J oris. See van W ing .
V ansina, J an . 1962a. ‘A Comparison of African Kingdoms’. Africa,
XXXII. 4, 324-35.
 B IB L IO G R A P H Y 297
---------------- . 1962b. L ’Evolution du royaume rwanda des origines a 1900
(ARSOM, XXVI. 2.) Bruxelles.
---------------- . 1962c. ‘Long-distance Trade Routes in Central Africa’.
JAM , III. 2, 375-90.
---------------- . 1963a. ‘The Foundation of the Kingdom of Kasanje’.
JA H , IV. 3, 355-74.
---------------- . 1963. Geschiedenis van de Kuba. (MRAC, no. 44.)
Tervuren.
---------------- . 1964a. L e Royaum e kuba. (MRAC, no. 49.) Tervuren.
---------------- . 1964b. ‘Process Models in African History’. In Vansina,
Mauny, and Thomas, eds., H istorian in Tropical Africa, pp. 375-89.
----------------. 1965. Oral Tradition. Trans. H. M. Wright. Chicago.
---------------- . 1966a. Kingdoms o f the Savanna. Madison.
---------------- . 1966b. ‘More on the Invasions of Kongo and Angola by
the Jaga and the Lunda’. JA H , VII. 3, 421-9.
---------------- . 1968. ‘The Use of Ethnographic Data as Sources for
History’. In Ranger ed., Emerging Themes, pp. 97-124.
---------------- . 1969. ‘The Kingdom of the Great Makoko.’ In Daniel F.
McCall, Norman R. Bennett, and Jeffery Butler, eds., Western African
H istory (Boston University Papers on Africa, IV.), New York, pp.
20-44.
---------------- . 1970. ‘Cultures through Time’. In Raoul Naroll and
Ronald Cohen, eds., A H andbook o f M ethod in Cultural Anthropology.
Garden City, N. Y., pp. 165-79.
---------------- . 1971a. ‘Les Mouvements religieux kuba (Kasai) a
Pepoque coloniale’. E H A , II, 155-87.
---------------- . 1971b. ‘Once upon a Time: Oral Traditions as History in
Africa’. Daedalus, C. 2, 442-68.
---------------- . 1973. The Tio Kingdom o f the M iddle Congo, 1880-1892.
London.
---------------- . 1974. ‘Comment: Traditions of Genesis’. J A H , XIV. 2,
317-22.
V a n s in a , J a n , R . M a u n y , and L. V . T h o m a s , eds. 1964. The H istorian in
Tropical Africa. (International African Seminar.) London.
v a n W i n g , J. and C. P e n d e r s . 1928. L a Plus Ancien Dictionnaire Bantu.
Louvain.
V e ig a s , A r t u r . 1923. ‘Duas tentativas da reconquista de Angola em
1645’. Revista de Historia, XII. 45-48, 75-45, 5-23.
V e l l u t , J e a n -L u c . 1972. ‘Notes sur le Lunda et la frontiere Luso-
Africaine (1700-1900)’. E H A , III, 61-166.
V e r h u l p e n , E. 1936. Baluba et Balubaises du Katanga. Antwerp.
V e r l y , R. 1955. ‘Le ‘Roi divin’ chez les Ovimbundu et Kimbundu de
1’Angola’. Zaire, IX. 7, 675-703.
W a l t e r , E. V . 1969. Terror and R esistance : A Study o f Political Violence.
New York.
298 B IB L IO G R A P H Y
W eeckx, G. 1937. ‘La Peuplade des Ambundu (District du Kwango)’.
Congo, I. 4, 353-73; II. 1, 13-35; 2, 150-66.
W h e e l e r , D o u g l a s and R e n £ P e l is s ie r . 1971. Angola. London.
W h it t l e s l e y , D. S. 1924. ‘Geographic Provinces of Angola: An Outline
based on Recent Sources’. Geographical Review, XIV, 113-26.
Y o u n g , M. C r a w f o r d . 1970. ‘Rebellion and the Congo’. In Robert I.
Rotberg and Ali A. Mazrui, eds., Protest and Power in Black Africa,
Oxford, pp. 968-1011.

I I I . L IS T O F U N P U B L IS H E D W O R K S C IT E D

B a s t in , M a r ie - L o u is e .1966. ‘Tschibinda Ilunga: Heros Civilisateur’.

Diss. Université Libre de Bruxelles.
B ir m in g h a m , D a v id . 1971. ‘The Origins and Consequences of Market-
Oriented Trade in Africa’. Paper read to the East Africa Social
Research Conference, Dar-es-Salaam.
C o h e n , D a v id W. 1971. ‘Emergence and Crisis: The States of Busoga in
the Eighteenth and Nineteenth Century’. Makerere University,
Kampala.
H e w s o n , G l y n C h a r l e s . 1970. ‘Shaka’s Kingship and the Rise of the
Zulu State, 1795-1828’. MA Thesis, University of Wisconsin.
L u c a s , S t e p h e n A n d r e w . 1971. ‘The Outsider and the Origin of the
State in Katanga’. Paper presented to the Annual Meeting of the
African Studies Association, Denver.
M il l e r , J o s e p h C. 1971. ‘The Dual Slave Trade in Angola’. Paper
presented to the Annual Meeting of the African Studies Association,
Denver.
---------------- . 1972. ‘Kings and Kinsmen: The Imbangala Impact on
the Mbundu of Angola’. Diss. University of Wisconsin.
M issã o d e I n q u é r it o s A g r íc o l a s d e A n g o l a . (Reports on farming
techniques.) Luanda.
N’D ua, E d o u a r d . 1971. ‘L’lnstallation des Tutschokwe dans l’empire
Lunda, 1850-1903’. Mérnöke, Grade de Licencié en Philosophie et
Lettres: Groupe Histoire, Université Lovanium de Kinhasa.
P ir e s , A l b e r t o A u g u s t o . 1952. ‘História dos Sobados do Posto de
Cinco de Outubro’. Xá-Muteba.
R e q u ie r . 1930. ‘Rapport d’enquéte sur la chefferie de Kasongo-Lunda’.
Manuscript in Fonds ethnographiques, M.R.A.C.
R o n s m a n s (?) n.d. ‘Remarques relatives à l’histoire des Bayaka (Notes
sur l’étude de Dequenne et les remarques de Roelandts)’. Manuscript
in Fonds Ethnographiques, M.R.A.C.
S a l a z a r , S ig u r d v o n W il l e r . n.d. (c. 1965?) ‘Bondos e Bângalas:
Subsídios etnográficos sobre as duas tribos, recolhidos na área da
Circunscrição do Bondo e Bángala, do Distrito de Malanje—Angola’.
Dissertação para o acto de licenciatura pelo I.S.C.S.P.U. 2 vols.
 B IB L IO G R A P H Y 299
V az, M anuel, 1970. ‘História dos Bângalas’. Manuscript in possession
of author.
W n.d. ‘Notes sur les populations du district du Kwango’.
a u t e r s (? ).
Manuscript in Fonds Ethnographiques, M.R.A.C.
X a v ie r M a r t i n s , F e r n a n d o B a r r o s . 1963. ‘Origem dos Povos Lunda:
Lenda e História’. Manuscript taken from ‘Relatório da inspecção
Circumscrição Administrativa do Cassai-Sul, em 1963 pelo Inspector
Adm.or Xavier Martins’. Lunda. (In possession of Vitor António dos
Santos, then secretário do Posto do Quela, Distrito de Malanje.)
IV . L IS T O F A R C H IV E S C O N S U L T E D
Lisbon, Portugal Arquivo Histórico Ultramarino (A.H.U.)
Arquivo Nacional da Torre do Tombo (A.N.T.T.)
Biblioteca da Ajuda (Ajuda)
Biblioteca da Sociedade de Geografia de Lisboa
(B.S.G.L.)
London, England, British Museum (B.M.).
Luanda, Angola. Arquivo Histórico da Câmara Municipal de
Lunda (A.H.C.M.L.)
Arquivo Histórico de Angola (A.H.A.)
Arquivo da Câmara Episcopal (A.C.E.)
Arquivo do Quartel Geral (A.Q.G.)
Tervuren, Belgium. Section Ethnographique, MuséeRoyal de FAfrique
Centrale (M.R.A.C.)
 Glossary
(African Language Terms Used in the Text)*

baka a musendo plural of m uku a musendo

jingola plural of ngola
jingundu plural of ngundu
jinzum bi plural of nzumbi
jisabu plural of sabu
kabungu Lunda title, the most ancient known, of the type
of the kinguri, the lueji, and others
kibinda professional hunter, an Mbundu initiated status
entailing possession of supernatural powers to
deal with large wild animals.
ko ta Mbundu title, the elders of a lineage entrusted
with guarding lineage positions; in the context of
a kingdom, court officials attending the king,
often the electors of the occupants of the royal
position
lcumbu Imbangala praise name
kijila in general, a prohibition of any kind; with
specific reference to the Imbangala kilombo, one
of the laws which all members had to respect
scrupulously
kilómba title of Pende kings who ruled the Baixa de
Cassanje before the arrival of the Imbangala
kilombo the warrior initiation society of the Imbangala;
also the sacred enclosure into which only members
might enter; also the Imbangala war camp
kimbanda an Imbangala magical specialist; there were
various types of kimbanda depending on the type
of spirit which they treated and the means they
employed
lemba an Mbundu lineage official, ‘uncle’ of the matri-
lineage (or the ngola)
lukano the Lunda bracelet, symbol of royal authority
lunda Mbundu tale, narrative in form, of which some
deal with the names in the formal genealogies
* Terms occurring only once or in the footnotes have not been included in
this glossary.
 G L O SSA R Y O F A F R I C A N L A N G U A G E T E R M S 301
lunga ancient Mbundu authority symbol, associated
with the Pende in particular; believed to come
from the sea and have close connections with
bodies of water.
m aji a samba the magical ointment believed to confer invulner­
ability on Imbangala warriors
m akota plural of Icota
malemba plural of lemba
malunda plural of lunda
malunga plural of lunga
maluvo fermented palm wine, any intoxicant
mavunga plural of vunga
milemba plural of mulemba
misete plural of musete
misungo plural of musungo
mpat Lunda term for the territories of the ancient
Lunda lineages
m ukishi a figurine representing supernatural forces attrib­
uted to some types of Mbundu political titles
m uku a musendo professional Imbangala historian
nndcwa a kushingilisa Mbundu lineage spirit medium (usually female)
mukwale knife in Kimbundu; also a special knife attributed
to some kings
mulemba tree closely associated with Mbundu lineages and
lineage headmen
musendo Imbangala formal genealogy
inusete Mbundu reliquary
musungo appointed warleader
muyombo tree believed to be the residence of lineage spirits
ndua a plantain eater of complex symbolic significance
in many parts of West-Central Africa
nganga magical specialist, of various types
ngola a piece of iron held by most Mbundu lineages as
an important authority emblem; associated with
the Samba people
ngundu Mbundu matrilineage
nzumbi Mbundu lineage ancestor spirit; by extension, the
spirits of deceased occupants of permanent
positions
nzungu a magical weapon believed to belong to Kinguri
pem ba the white clay powder distributed by the Mbundu
lineage headmen to their nieces (female members
of their lineages)
302 G I,O SSA RY OF A F R IC A N L A N G U A G E TERM S
pokw e a fighting knife, especially associated with the
nineteenth-century Lunda
sabu Mbundu proverb
soba Portuguese-appointed chief (plural, sobas)
takula red powder which lineage headmen distribute to
the males of their lineages
tubungu plural of kabungu
vunga an appointive title introduced from Libolo among
the Mbundu, distinct from the hereditary perpet­
ual positions
yibinda plural of kibinda
yijila plural of kijila
yilamba plural of kilam ba
yimbanda plural of kimbanda
 Index

Adam and Eve (as aetiological Benguela current, 32

figures), 55, 58 Benguela plateau. See Ovimbundu
Aetiological genealogy. See Genealo­ highlands
gies, aetiological Benguela Velha (Portuguese town),
Afonso I (Kongo king), 66 210
Agricultural technology, 35-6 Bihe (state), 157, 166, 227
Akwaluanda (people), 40 Botswana, 164
Almeida, Gaspar de, 217 Brazil, 179, 182, 212, 215
Almeida, João Soares de, 208 Buganda (state), 8
Alur (people), 29n Bunyoro (state), 6
Alvaro II (Kongo king), 182 Bushbuck, 69
Ambaca (Portuguese post), 40, 194n, Butatu a ICuhongo kwa Wutu wa
195, 199, 201, 207 Nyama (political title), 70-1, 72,
Ambaca (region), 93 73, 87, 96
Ambakista (language, people), 40,41
Ambundu (people), 40, 42 Cain and Abel (as aetiological
Andulo (modern town), 36, 157n figures), 55
Angola (state), 3, 178—83. See also Cambambe (Portuguese post), 217,
Portuguese 219
‘Angury’ (political title), 212 Cannibalism, 158-9, 182, 227, 244-9
Anuak (people), 29n Cardoso, Bento Banho, 182, 185, 203
Atlantic Ocean, 39 Carmelites, 205
Authority, symbols of, as evidence, Carmona (modern town), 34
57-8; Mbundu notions of, 52-3, ‘Cassanje’, 192. See also Kasanje
86-7, 235, 268-70; Imbangala Cattle, symbolism of, 105
notions of, 235-7 Cerveira Pereira, Manuel, 196, 210-
Aragão, Rebello de, 197 12, 216
Chameleon, avoidance of, 68-9
Baixa de Cassanje (region), 34, 38,41, Chronology, early ngola states, 78,
42, 70, 72, 73, 76, 80, 84, 86, 87, 92, 79; early Kongo, 79; early Lunda,
93, 96, 99, 104, 108, 109, 110, 111, 125-6, 133; Songo, 152-3n; kil-
140, 202, 220; as salt source, see ombo, 175, 186; in oral traditions,
Kihongwa river, Lutoa river see Oral Tradition—Imbangala,
Baixa de Kafushi (region), 72, 99, chronology in
221n Cibinda Ilunga (political title), 126-7,
Balanga (palanga) (political title), 128, 129-30, 138-9, 148. See also
103-4, 114n Lukonkesha
Balombo river, 21 On Cihumbo river, 140
Bandundu (Zaire province), 42 Cikapa river, 140, 141, 142
‘Bango-Bango’ (political title), 218-19 Cimbwiya (authority symbol), 127
Batteli, Andrew, 177-8n, 180, 185 Circumcision camps, as cross-cutting
Bemba (people), 45 institutions, 51; and origin of
Bengo/Nzenza river. See Nzenza Icilombo, 162, 167
(Bengo) Circumcision, of kinguri, 147; Imban­
Benguela (Portuguese town), 81, 167, gala abolition of, 229-33
175, 212, 216 Cisenge (political title), 137
304 IN D E X
Cliches, in narrative episodes, 23; Fang (people), 29n
examples criticized, 101 (rising Fission. See Descent groups
river), 139, 145 (daggars), 173, Fula dya mbande (political title), 17
200-1 (rotten seeds), 256n (spirits’ Functionalism, 4, 8-9
tears) Ftmji a musimgo (political title), 204
Climate, 32 Furtado de Mendonça, João, 181
Congo (Zaire) river, 31, 32
‘Conquest theory’ of state formation, Genealogies (musendo), method of
2, 4-5, 8, 280-2 recording, viii; historical meaning,
Cosmology, Mbundu, 45-6, 56, 242- 16-21, 46; aetiological, 19-22, 39,
4, 250-1, 253-64 passim, 267 41-2, 65, 74-5, 78-9; segmentary,
Cokwe (language), xii, 38 113-14, 117
Colcwe (people), and Mbundu, 34, 36, —political, and «gofa-titles in, 68;
37,38-9; and perpetual kinship and examples criticized, 75, 78, 90,
positional succession, 45; and 91-2; and v;/;;,ga-litles in, 94—5,
ICalunga, 61; and aetiological 97-8, 108
genealogy, 76, 113; early political Ghana, ancient (state), 6
history, 112; social structure, 116; Golungo Alto (Portuguese town),
and Lunda titles, 128-9; later 194n
chiefs, 137; and kinguri, 137-41, Gomes de Moraes, Diogo, 213, 216
162; and Minungu, 155; men­ ‘Ganguellas’ (region), 158
tioned, 29n, 52, 151 Gaspar Kanzenza, 188-9
Cross-cutting institutions, 43, 50-2, ‘Gemge’ (region), 215
88, 266-7 Hako (people), 41, 42
Curing cults, 50 Hako (region), 81, 93, 102, 213, 217-
18
Dande river, 32, 42 ‘Hamitic hypothesis’, 1, 4-10, 280-2
Descent group headmen {malemba Hango (political title), in Libolo,
dya ngimdu), in genealogies, 47; 90-4; and Songo, 114; and kaza,
duties and authority, 46-8, 52-3, 219; mentioned, 110,113,129, 152,
243; and lunga, 60-2; and ngola, 153, 157, 176
68, 69; mentioned, 59, 63 Hango (Portuguese post), 195, 199
Descent groups (jingundu), in African Hanya (people), 81
states, 1; in Imbangala genealogies, Hari (people), 61
17-18, 20, 21, 47-8; in narrative Hari a kiluanje (political title), 84, 85,
episodes, 24-5; described, 43-9, 86, 201
49-52 passim; antiquity of, 47, 59; Headmen. See Descent group head­
loyalty commanded by, 48, 50, 53; men
fission, 53; among Lunda, 48, 116— Henrique de Carvalho (Portuguese
18; among Songo, 113—14; among town), 140
Cokwe, 146-50, 154, 165-6; and History, Imbangala perceptions of,
kilombo, 225-32 13-14, 56
Dias de Novaes, Paulo, 83 Holo (people), 42, 139n
Diviners (yimbanda), 50, 52-3, 229, Holo (state), 151
235, 243, 259 Hungu (people), 38, 39, 41, 42, 55, 60
Documentary sources, 26-7 Hunting societies. See Kibinda
Dutch, 203, 204, 207, 209, 212, 215,
219 lkota (political title), 236
Ilamba (region), 80, 92, 219
Ilunda (political title), 236
Elders, descent groups. See Makota Imbangala (people), o f Kasanje,
(descent group elders) xi-xii, 20, 42, 104, 137, 230, 231;
Empacasseiros, 160, 168 rituals, 166; stymology of name,
Ethnographic data. See Method­ 168n; kilombo ideology, 225-64;
ology, ethnographic data mentioned, xii, 52, 55
 IN D E X 305
Imbe ya Malemba (descent group), ICalanyi river, 113, 117,126,138, 139,
136, 187, 191 140, 141
Infanticide, 226, 234 ‘ICalonga’ (state), 61
Informants. See Methodology Kalulu ka Wambwa (political title),
Interlacustrine states, 1, 6 114
Interviews. See Methodology Kalunga (aetiological figure?), 59,
Iongo (modern town), 72 60-1
Iron technology, in sixteenth century, Kalunga ka Kilombo (political title),
36; and Samba, 67, 74-5, 76; and 153n
political power, 80-1 Kalunga ka Ngola (political title), 79.
Iron, sources of, and states, 86, 87, See also Kalunga ka kabila
271-2 Kamana ka mbande (political title), 17
Ishi (village lands), 42 Kambala ka mbande (political title),
Itengo (region), 140-1, 162 17
Iyala (yalci) (political title), 122, 124, Kambambe (region), 92, 193
125 Kambambe (political title), 183
ICambo river, 17, 34, 39, 42, 66, 70,
Kabari ka Nzungani (descent group), 84, 92, 97, 111
17, 18 ICambo ka Mana (region), 66
Kabari ka ICajinga (descent group), ICambwizo (political title), 136, 187,
108 191
ICabatukila (region), 102, 109, 221n ICandala (political title), 137
Kabele ka Hango (political title), 97 ICandama ka Hite (descent group),
Kabila ka ngola (political title), 17 136, 185, 186, 191
Kabuku ka Mbwila (political title), ICandama ka ICikongwa and ICand-
206 umaka ICikongwa (descent groups),
Kabuku ka Ndonga (political title), 136, 185, 191
150, 158, 164, 169, 219, 220 Kangengo (political title), 136, 184-5,
Kabuku ka Ndonga (state), 151, 186, 187, 188, 189, 191
202-6, 207 ‘ICa Ngombe’ (political title), 211-12
Kabungu (political title), 118, 125-6, ICangongo ka Pango (political title),
129-30, 133, 154 73
Kafushi ka Mbari (political title), 81, Kangunzu ka ngola (political title), 17,
86, 183, 215, 218 19, 79
ICahanda (descent group), 153 ICanje (political title), 71
ICahete (political title), 136n, 164,169 ICanyika ka Tembo (aetiological
Kahunze (praise name), 152n figure), 137, 144-7
Kaita (political title), 137 ICapenda ka Mulemba (political title),
ICajinga ka Mbulu (aetiological fig­ 137
ure), 55, 60, 97, 98-103, 108, 109 Kapokolo (authority symbol), 127
ICakende (political title), 153 Karl (people), 42
Kakonda (political title), 208, 213 Kasai river, 137, 140, 144, 146, 148,
ICakulu ka Hango (political title), 93, 154
183 ICasala (region), 73n, 195n
Kalcimga ka kabila (political title), ICasambe (political title), 217
17. See also Kalunga ka Ngola Kasanje (political title), 103-4, 137,
Kakwako (region), 36 143n, 183, 191-2
Kalala Ilunga (Luba state-founder), Kasanje (state), oral histories, ix; and
60 present study, xi; and vtmga, 95;
Kalanda ka Imbe (political title), and formation, 109, 200, 202; and
Portuguese, 177-81; and kilombo, Lunda authority symbols, 135;
184-5, 187-9, 191; mentioned, 60, during 1640s, 209-10; and Icaza,
136, 164, 169. See also ICalandula 220; circumcision in, 231; men­
ICalandula (state), 151, 206-8, 219, tioned, 2 0 ,139n, 141,148, 151,156,
220 164, 172, 174, 188-90
306 IN D E X
Kasanje ka Kibuna (political title). ICikato kya Kajinga (descent group),
See Kulashingo 99
Kasanje ka mbambo (diviner), 104 Kikombo river, 215-17
Kasanje ka ngongo (political title), Kikululu kya Hango (political title),
104-5 110
Kasanje ka Yela (political title), 104 ICilcungo kya Njinje (political title?),
Kasanze (state), 181, 182, 202, 205n 72
Kasenda ka Swaswa (political title), Kilamba (political title), 219
84 ICilamba kya Ndungu (political title),
Kashana (political title), 212-13 79
Kashimo river, 140 Kilombo, misunderstood, 162n; and
ICasongo (political title), 122 Lunda makota, 167-75, 230-1; and
Kasongo Nyembo (state), 120n • Portuguese, 176-223 passim ; struc­
Kata ka Walunda (ka Manganda) ture, 183; among Ovimbundu, 222,
(political title), 120 227; among Mbundu, 224-64
Katanga (region), 37, 112, 113, 115, Kilombo kya Kasenda (descent
117, 137 group), 84
Katanya (region), 96 Kiluanje kya Hango (political title),
ICatende (political title), 137 92
Katumbi ka N gola a Nzenza (political Kiluanje kya Samba (aetiological
title), 98-9, 100-1 figure), 55
Kavunje ka Mbulu (aetiological Kiluanje kya samba (political title),
figure), 55, 153, 169 65, 85
Kayongo ka ICupapa (political title) Kiluanje kya Samba (state), 75, 76, 79
(or Kayongo ka Butatu), 72, 110 Kiluvia, 174
Kayungo (Kalungo) (political title), Kima a Papa (political title), 99,
122 101-2, 108, 109, 110 . . .
Kaza (ka hango) (political title), 93, Kimbanda. See Diviners
153n, 169,’219-20 Kimbundu (language), viii, xii, 38,
Kazanga (region), 99 39-40, 252
Kazekele (authority symbol), 135 Kimbuya (authority symbol), 156
Kembo (region), 73 Kindonga (islands), 221
Keta (state), 80, 86, 113 Kingongo kya mbande (political title),
Keta kya Labalanga (political title), 17
199 Kinguri (political title), in Lunda,
Kibala (state), 81 118-28; leaves Lunda, 130-5,
Kibala (region), 230 147-8; Lunda descent group of,
Kibinda (hunter), as cross-cutting 135-6; among Cokwe, 137-41,
institution, 51-2, 160; authority 143-9; in Kasanje, 140, 141, 147,
of, 52-3, 240n, 243-4, 250; among 150, 171, 174, 209, 231, 250; and
Lunda, Cokwe, 127, 128-9; King- circumcision, 147; among Songo,
uri and, 133-5, 138-9; among 151-5, 158, 159, 161, 168-9, 172;
Ovimbundu, 160; Kulashingo as, suppression, 169-74, 187-91; and
186 kulashingo, 186-93; mentioned, 112
Kibondo kya Wulu (political title), Kinguri kya Bangela (political title),
136 118n
Kibuna kya Musumbi (political title), Kingwangwa (kya kima) (political
144 title), 108, 109
Kihongwa river, 37 Kinyama (political title), 118-26, 128,
Kijila , 158, 163-6, 226-48 passim 138
Kikango kya Hango (political title), Kinzunzu kya Malemba (political
97 title), 136
Kikasa (kya Kilcululu) kya Hango Kirima (people), 106
(political title), 97, 108, 109 Kisama (region), 34, 36, 39, 66, 77,
Kikasa kya Ndala (political title), 97 81, 87, 94, 143n, 181, 183, 215, 256
 IN D E X 307
Kisua kya Kajinga (descent group), Kwanza river, xiii, 32, 34, 36, 37, 39,
99 41, 66, 75, 77, 80, 83, 89, 90, 91, 92,
Kitelca, 243 93, 96, 98, 100, 102, 103, 104, 110,
Kiteke kya Bengela (political title), 113, 114, 151, 152, 156, 159, 161,
81, 86, 93 163, 175,176,178-83, 193, 201, 206
Kitumbo (authority symbol), 127 Kwije river, 34, 41, 93, 95, 106
Kitushi (political title), 103, 105 Kwilu river (Kwango tributary), 66
ICoko na Mumbi (political title), 73 Kwilu river (Kasai tributary), 66n
Konde (a Matita) (political title),
118-19 Laço, Lopo Soares, 217-18
Kongo (language), 38 Lambo (region), 97, 102, 108, 109,
Kongo (people), 32-4, 37, 38-9, 42 111, 221n
Kongo (state), and Luanda region, Lemba dya ngola (political title), 63,
36, 39; early growth, 41, 79; and 68
Imbangala, 182-3, 185-6, 199, 202; Lenge (authority symbol), 188
mentioned, 8 ,1 1 2 ,138n, 181, 207-8 Lenge (people), 41, 42, 55, 60, 76, 80,
Kongo dya Hango (political title), 97 86, 93, 94, 110, 113
Kongo dya Mbulu (aetiological fig­ Libolo (people), 38, 41, 42, 55, 113,
ure), 17-20 passim, 55, 60 256
Kori river, 99, 102 Libolo (region), 37, 100, 213, 217-18
Kuba (people), 59, 60 230
Kuba (state), 8 Libolo (state), history, 90-106; and
Kubango-Okavango river, 32 ngola a kiluanje, 93; and Mbondo,
Kukumbi river, 140 95-106; and Songo, 106, 152; and
Kulashingo (political title), origins, kinguri, 151-4, 161; decline, 175;
143-4; as kinguri, 150, 186-93, 202; etymology o f name, 231; men­
and Portuguese, 185-98, 200-1, tioned, 89, 104, 105, 108, 110,
202, 206, 210-11, 219; in ICasanje, 138n, 169, 175, 219
209, 220-1; mentioned, 184, 203 Lineages. See Descent groups
Kulembe (political title), early Longa river, 34, 38, 81, 170, 213
history, 89—90; and Libolo, 91-2; Lonyi river, 140
and Songo, 114; and Lunda, 155, Lopes Sequeira, 215-16
169-71; and ‘Temba Andumba’, Lualaba river, 129
163-4, 166, 168; and Imbangala, Luanda (island, bay), 32, 36, 59,
210-17 passim ; and mimjumbo 213; 179
mentioned, 96, 104, 105, 112, 138n, Luanda plateau, 34, 35, 70, 75
208, 230-1 Luanda (Portuguese town), 39, 181,
Kulturkreise (diffusionists), 4, 8, 29 189, 192-3, 202, 203, 204, 205, 209,
Kunda (a Ngamba) (political title), 212
118-20 Luanda, modern, 40, 42
Kunene river, 32, 210 Luba (people), 60
Kunga (dya Palanga?) (political title), Luba (state), 61, 118-20, 127-8,
114, 153, 175 132-3, 138
Kurinje lcwa Samba (aetiological Lubembe (authority symbol), 135
figure), 55 Lucala (modern town), 70
Kuvo river, 34, 36n, 170, 175, 178, Lucaze (people), 136n
179, 180, 210, 213, 215-17 Lucaze na Mwazaza (descent group),
Kyango kya Hango (political title), 97 136-7, 141, 158, 189. See also
Kwale river, 34, 41 ‘Musasa’
Kwango (person), 148 Lueji (political title), early history,
Kwango river, 32, 34, 36, 41, 43, 61, 118-19; later history, 123-6, 147;
62, 66, 67, 72, 73, 76, 106, 110, 113, and mwata yamvo 129-30; and
137, 140, 141, 148, 149, 151, 153, Cibinda Ilunga, 130-1, 133; and
154, 166, 177, 201 kinguri, 140-1; mentioned, 122
Kwanza (person), 204-5 Luhanda river, 37, 70, 72, 73, 96
308 IN D E X
Luhando river, 34, 39, 41, 92, 102, Mahongi) (political title), 47, 72,
106, 114, 152, 153, 155, 158, 159, 110
161 Mahongi a Ndungu ya Zombo dya
Luhasa ICamonga (descent group), ICajinga (descent group), 110
129-30 Maji a samba, 163-6, 220, 228, 232-4,
Luhembe river, 140 239 246
Luhinga river, 41 M akota (descent group elders), 18,
Lui river, 34, 37, 42, 59, 61, 70, 72, 73, 47, 48, 52
75, 83, 84, 106, 108, 110, 111, 152, Makota (of Icinguri), in Lunda, 135-7;
155 and kilombo, 164, 167-75; and
Lukala river, 32, 201; upper, 41, 42, Portuguese, 176-210; states based
65, 75, 79, 207; middle/lower, 40, on, 201-8, 251-64; mentioned,
41, 65, 75, 76, 78, 80, 82, 92, 193, 148, 149
194 Makota (of lueji), 131
Lukamba (region), 92, 194-5, 200-1 Malanje (modern town), 17, 40
Lukano (authority symbol), 125, 127, Mali, ancient (state), 6
128, 147-8, 188-9 Malimda. See Narrative episodes
Lukokesha (political title). See Cib- Mamba (snake), 95-6
inda Ilunga Mambanje (political title), 165
Lukonkesha (political title), 130-1, Manganda a ICambamba a Musopo
133-5, 138-9 wa Nyama (political title), 120
Lukongolo river, 130, 136 Mani kongo (political title), 112
Lukumbi river, 152 Mani lombo (political title), 236
Lukunga (political title), 93 Manyungo wa Mbelenge (descent
Lulua river, 153 group), 152, 169, 174
Lumbo (mwene lumbo), 105 Marimba (modern town), 17
Lunda (language), xii Marimba (people), 81
Lunda (people), 34, 45, 52, 113, 114- Marriage alliances, 49
18 Mascarenhas (bishop), 217
Lunda (state), expansion, 111, 135-7, Massangano (Portuguese town), 179,
141, 154; and Mbundu, 115, 122, 212, 214, 215-16, 219
135; and Luba, 127-8, 132-3; Matamba (state), 42, 66, 71, 74, 75,
mentioned, 20, 112 79, 151, 203-5, 207, 208-9, 221
Lunda (narrative episode). See Nar­ Matamba (region), 79, 82, 83
rative episodes Matamba a Mbulu (aetiological
Lunga (authority symbol), origins, figire), 55, 60, 75
55, 58; historical meaning of, 56, Maweje (political title), 122
58; as lineage emblem, 59-63, Mbailundo (state), 98, 102, 114, 156
87-8, 235; states based on, 67, 68 Mbamba, 131
70-3, 76, 86, 87, 96, 175; and Mbamba (province), 202
ngola, 69; and water spirits, 256; Mbamba a Kajinga (descent group),
mentioned, 73, 89, 94, 103, 110, 108
113, 129, 267 Mbande a kingongo (political title), 17
Lunga dya Kafofo (political title), Mbande a ngola (political title), and
213-15 ngola a kiluanje, 84, 111; and
Lungu dya Hango (political title), 93 kaza, 220; mentioned, 17, 19, 82
Lushimbi river, 79 Mbangala (language), xii
Lutoa river, 37, 73 Mbango (mountain), 71n, 72
Lwana river, 140 Mbekesa a Lukunga (descent group),
Lwashimo river, 140 80. See also Lukunga
Lwena (people), 129 137. See also Mbola na kasashe (political title), 93,
Cokwe 104, 169
Mbola na Kasashe (region), 169, 171,
Machado, Luis Gomes, 219 186-91, 231
Mahashi na Pakasa (Mahashi a Mbombo ya Ndumbu, 17, 18
 IN D E X 309
Mbondo (people), 41, 55, 60, 113, Mundombe (people), 34, 167
114, 129, 157, 256 Muni dya Ngombe (political title), 216
Mbondo (state), vimga in, 95, 97; and Munjitmbo (political title), origins,
Libolo, 95, 98-106; expansion, 122; states based on, 155—61, 168,
108-9; decline, 185n; mentioned, 175; and Iculembe, 213; mentioned,
20, 90 150, 163
Mbongo wa Imbe (political title), Mupolo wa Kajinga (descent group),
136, 184-5, 187-9, 191 99, 108
Mbukushu (people), 164-5, 166 ‘Murombo’ river, 210-11
Mbulu wa ICurinje (aetiological ‘Musasa’ (descent group), 158. See
figure), 55 also Lucaze na Mwazaza
Mbumba (political title), 137 Muselce (region), 92
Mbumba a Kavenge (political title), Musenclo. See Genealogies
81 Mushima (region), 183
Mbumba a Mbulu (aetiological fig­ Mushinda (political title), 153
ure), 113, 116, 138, 152 Musumbi wa Mbali (political title),
Mbumba a Mbundo (political title), 143
81 Mitsungo (military title), 237-8
Mbumba a Musumbi (political title), M usw i (political title), 66, 67, 75
144 Muta a Kalombo (legendary figure),
Mbumba a Nyasi (legendary figure), 99, 101
105 Mutombo Mulculu (political title),
Mbundu subgroups, sixteenth cent­ 118, 120
ury, 32, 37-9; modern, 39-42 Mutonde a Kalamba ICizembe (praise
Mbwila (political title), 208 name), 152
Metaphor, 23 Muxima (Portuguese post), 179
Methodology, tape recordings, vii- Muyombo, 146, 147, 148, 154
viii, ix; informants and interviews, Muzumbo (political title), 155n
viii- ix, xiii; ethnographic data, Mwa Cangombe (political title), 136,
ix - x, 26. See also Genealogies 186, 191, 211n
Migrations, 28-9 (Mwa) Ndonje (political title), 146,
Minungo (language), 38 150, 154-5, 158, 163, 168, 209
Minungo (people), 155 (Mwa) Ndonje (state), 151,154-5,161,
Moa river, 72, 148 175
Mona a yanga, 52 Mwata Kandala (political title), 149
Mona Kimbundu (region), 140, 141, Mwata yamvo (political title), 112,
142, 162, 186 115, 129-30, 132, 139n, 141, 154
Monclo (authority symbol), 135 Mwela (authority symbol), 131, 156—
Mozambique, 215 7, 177
Mpat, 117 Mwiji mwa ngola (political title), 17,
Mpemba (province), 202 19
Mucari (modern town), 17 Mwili (political title), 114
Muhi wa kabila (political title), 17,18. Mwilu (political title?), 66
See also Muhi wa Ngola
Muhi wa N gola (political title), 79. Nambo a Mbungo (political title),
See also Muhi wa kabila 213-15
Mulcari river, 79 Nambo a Ngongo (political title), 208
Mukishi, 256-7 Nangwiji river. See Cikapa
Mukoso (political title), 114 Narrative episodes (malunda), record­
Mulcwa a kushingilisa, 255, 256 ing methods, viii; historical mean­
Mukwale (authority symbol), 139 ing of, 21-6; mentioned, 26, 27
Mulemba (authority symbol), 48, 53, —interpretations, seventeenth cent­
59, 60, 61, 63, 64 ury, 75-6, 157-8, 162-7; Mbondo,
Mumbanda a Mbulu (aetiological 98, 103; Lunda, 123-8 passim,
figure), 55, 60, 79 139-41, 147-8; Imbangala, 130-7,
310 IN D E X
Narrative— coni. Ngola a mbande (political title), 17,
143-7, 148, 154, 170-5, 186-93, 139n
230-1; Colcwe, 144-8, 156-7; Ngola a mbole (political title), 236,
Bihe, 158-60 237
Naweje. See Lueji Ngola a kabila (political title), 17
Ndala a Kajinga (descent group), 108 N gola Inene (aetiological figure), 55,
Ndala a Kikasa (political title), 97 56, 74, 75, 78, 79
Ndala kandumbu (political title), ix Ngoleme a keta (political title), 80, 86,
Ndala kisaa, origins, 95, 96-8; ety­ 205
mology, 105-6; in Mbondo, 109- Ngoma ya mukamba (authority sym­
10; decline, 185n; mentioned, 20, bol), 135, 154, 155, 168
113, 157 Ngonga a ICahanga (political title),
Ndambi a ngola (political title), 17- 215
19 passim, 79, 82-3 Ngonga a mbande (political title), 17
Ndemba (region), 77, 183 Ngungu ya Mbuku wa Hango
Ndembu (language), 38 (political title), 92
Ndembu (people, o f Angola), 39 Njimba na Kakundo (descent group),
Ndembu (people, o f Zambia), 45,129 135-6
Ndembu (political title), 199, 202-3, Njinje a mbande (political title), 17
204, 206, 207, 209 Nkanu (people), 66
Ndonga (political title), 136, 186, 191, Nyange a Kajinga (descent group),
202, 204 99, 1Q8
Ndongo (people), 41, 42, 55, 60, 76, Nzamba (political title), 215-16
113, 169 Nzenza (Bengo) river, 32, 36, 39, 41,
Ndongo (region), 219-21 183, 203
Ndonje. See (Mwa ) ndonje Nzenza a Ngombe (political title),
Ndua (bird), 260-1 100-1 , 218
Ndumba a Tembo (political title), Nzenza ya Mahashi (descent group),
137, 149, 158 47, 72
Ndungu ya Zombo (descent group), Nzinga (Ana de Sousa), 42, 93, 97n,
110 139n, 193-4n, 203-5, 207, 208-9,
Negage (modern town), 17, 34 220-1
Ngama a Ngola (political title), 81, 93 Nzongeji river, 36, 41, 76, 77, 86, 87
Ngamba a Mbumba (aetiological Nzungi a N gola (political title), 79.
figure), 113, 189 See also Nzungi ya Kabila
Nganga a nzumbi, 254-5 Nzungi ya kabila (political title), 17.
Ngangela (people), 34, 106 See also Nzungi a Ngola
Ngango river, 81, 90, 92, 93, 230 Nzungu, 131, 134, 145, 147
Ngola-Musuri. See Musuri Nzumbi, 13
Ngola (authority symbol), historical
meaning, 58; and descent groups,
63-70, 87-8, 235; origins, 65-7; Oral traditions, and structuralism,
states based on, 73-86; mentioned, 9, 23-4. See also Narrative epi­
89, 127 sodes, Genealogies
Ngola a kiluanje (political title), 17- —Imbangala, method o f collection,
20 passim, 66, 92, 94, 112, 113, 169 xi; historical meaning of, 11-26;
N gola a Kiluanje (state), rise, 75-9; chronology in, 14-15, 18-19, 25-6;
expansion, 79-82, 106, 111; inter­ and social and political institu­
nal politics, 82-6; and Libolo, 90, tions, 15-16; and Lunda history,
92, 93; and Kajinga ka Mbulu, 98, 116
99—100; and Icinguri, 151; and — Mbondo, criticism, 101
Lunda makota, 175, 180; and —Lunda, criticism, 115-16, 120-2
Imbangala/Portuguese, 176, 182, —Cokwe, and Lunda history, 115-16,
185-6, 192-3, 195, 199; and 117
Nzinga, 220; mentioned, 94 Orthography, xii, xiii
 IN D E X 311
Ovimbundu (people), and Mbundu, Samba (people, of modern Zaire), 68
xiii, 34, 38-9; modern, 96; and Samba a Ngola (aetiological figure),
Lunda titles, 129; and kilombo, 55, 65, 74
210-21, 222, 227; circumcision Sambaza (political title), 164
among, 230 São Tomé, 179
Ovimbundu highlands, leniembe in, Segmentary lineages, 49
89-90, 155, 168; mimjumbo in, 155; Seles (people), 34
origins of kilombo in, 161-7; Shakala (political title), 130
mentioned, 34, 36, 90, 96, 103, 112, ‘Shillambansa’ (shila mbanza) (polit­
210-21 passim ical title), 180
Ovimbundu states, early, 112; later, Shinje (language), 38
90, 96 Shinje (people), 137, 139n
Slaves, 272, 273, 278
Paka (people), 42, 61 Slave trade, 178-83, 195-200, 207,
Palm wine, 123, 125, 132-5, 177-8n 208-9, 211-12
Pande ya Musumbi (political title), Soba (political title), 197, 218n, 236,
144 258, 263
Panji a Ndona (political title), 204 Sombo river, 140
‘Peacock’, 259-60 Songhai (state), 6
Pemba, 48, 53, 60, 63, 145, 235, 262 Songo (people), descent groups, 49;
Pende (people), limga-states among, and Libolo, 106,152; early history,
70-3; and ngola a kiluanje, 80; 112-14; and kinguri, 161, 172,
mentioned, 41, 42, 52, 55, 60, 79, 174-5; mentioned 41, 42, 52, 55,
83, 84, 90, 92, 95, 96, 110-11, 113, 92, 108, 116, 118, 209, 256
129, 139n, 176, 256 Sonyo (province), 205n, 208
Pende (modern people, of Zaire), 42 Soso (language), 38
Perpetual kinship and positional Soso (people), 66
succession, and ngola, 68; and Sousa, Fernão de, 218
political titles, 81-2; among Lunda, Spirit possession, 50, 253-5
116-17; mentioned, 45-7, 87 Spirits, ancestor, 253-5; other, 255-7
Pokwe, 126 State-formation, xii, 1—11, 29
Political titles, explained, 17-19 ‘Stateless’ societies, 1-3
‘Port of trade’, 272 States, definitions, 3-4
Portuguese, and ngola a kiluanje, 10, Structuralism, 9
11, 76, 82-3, 85, 86, 93; in Angola, Sudanic empires, 1
41; in narrative episodes, 98 ‘Sudanic state’ hypothesis, 7-8, 10
Praise names, 16, 126 Suku (people), 41, 66, 67
Proverbs, 16, 188 Sumbi (people), 215—17
Plan bo, 238-41 Sungo dya kidembe (political title),
Pungo a Ndongo (region), 193, 201 213
Swana mulunda (political title), 130,
Quicombo (bay), 104 132
Swaswa dya Swali (descent group),
Residence patterns, 44-5 72, 83-4, 87
Rimba (region), 215 Swela (people), 102
Rwanda (state), 8
Takula, 48
Sala a Hango (political title), 92n Tandala (political title), 236
Sala Mpasti (people), 117 Teixeira de Silva (modern town),
Salt, and trade, 36; and political 36
power, 70, 72, 73, 76, 77, 81, 84, 87, Teixeira de Mendonça, 215-16
96, 271-2 ‘Temba Andumba’, 162-7, 168, 220,
Saluseke (political title), 137 227, 232, 234. See also Tembanza
Samba (people?), 65-70, 74, 75, 76, Tembanza (political title), 164-5, 220,
79, 80, 82, 83, 127 228, 233-4
312 IN D E X
Tembo a Mbumba (aetiological Walunda wa Nyama (descent group]
figure), 113-14. See also ‘Temba 120
Andumba’ Wamba river, 41, 77, 98, 99-100, 19
Tumba river, 79 Wambu (state), 104, 230
Tumba a Mbumba (aetiological Witchcraft eradication, 50
figure), 113, 120 Women, role in Icilombo, 227-9,234-;
Tumba dya N gola (aetiological fig­
ure), 75, 76, 79 Xa-Muteba (modern town), 73
Tunda (region), 217
Tyo (people), 66, 67 Yala mwaku (political title), 118-25
126, 130, 171
Yaka (people), 61
Umbundu (language), 38, 39
Yivo ya ICajinga (descent group), 9!
Ushi wa Nzumbi (political title),
Yongo (region), 72, 73, 110, 221n
73
Zaire river. See Congo
Vasconcelos, Luis Mendes de, 196- Zambezi river, 32, 128, 138
201 Zazzau (state), 8
Vegetation patterns, 34-5 ‘Zimbo’ (political title), 158, 163
Vieira, João Fernandes, 205 Zombo dya ICajinga (descent group)
Village structure, 43-4 108, 110
Vimga (political title), history, 89- Zulu (people), 1
110 passim; and munjumbo, 156-7, Zunda dya Ngola (aetiological figure)
160-1, 177; in kilombo, 168, 170, 75, 79
183-4, 236-7; mentioned, 114, 168, Zundu dya Mbulu (aetiological fig
204, 267 ure), 55, 60, 76