JOURNAL
OF ANTHROPOLOGICAL
ARCHAEOLOGY
4, 73- 115 (1985)
Background to the Ghana Empire: Archaeological
Investigations on the Transition to Statehood in the
Dhar Tichitt Region (Mauritania)
AUGUSTIN HOLL
Department of Ethnology and Prehistory, University of Paris X-Nunterre,
92001 Nanterre Cedex, France
Received December 21, 1984
In this paper, an attempt is made to fit together various kinds of data related
to the process of the formation of the Ghana empire, the earliest West African
state. An overview of research on this topic outlines some of the major problems
concerning state formation in this region. This is followed by a discussion of the
key politicoeconomic features and their dynamic relations in the process of social
differentiation. The use of this theoretical framework helps to make sense of the
Dhar Tichitt archaeological record. After careful examination of the prehistoric
data, it seems that, before the advent of the Ghana empire, trends towards state
formation were already in motion in the Dhar Tichitt region o 1985 Academic
Press, Inc.
INTRODUCTION
This paper deals with problems related to the transition to statehood
in Western Africa, and specifically, the development
of the so-called
Ghana empire, considered to be the first West African state. As archaeologists, we ought to demonstrate the archaeological visibility of modes of
social behavior and link them to various kinds of societal dynamics in
time and space.
Our approach is multifaceted.
After a short overview of the state of
research on Ghana empire, some key issues are delineated. Methodological devices and a general theoretical framework of state development
are constructed to help in inference building. Finally, in the last part of
the paper, this model is applied to the Dhar Tichitt archaeological record.
I. THE GHANA EMPIRE: AN OVERVIEW
From a wide range of research including ethnohistory, history, and
archaeology, the Ghana empire is known to have been the first state
organized in Western Africa. Takrur and Kawkaw are among the first
73
0278-4165185 $3.00
Copyright
0 1985 by Academic Press, Inc.
All tights of reproduction
in any form reserved.
74
AUGUSTIN
HOLL
West African states to be mentioned
in Arab sources (Thilmans and
Ravise 1980) and the earliest allusion to the Ghana state appears in the
Arabian historical record around 800 A.D. (Bathily 1975: 4). The territorial range of the empire is not well known (McIntosh 1981), but it seems
that its southern borders were on the river Niger in the east and the river
Senegal in the west. Its northern border seems to have been situated
between the 21” and 22” northern latitude. On the whole the territory was
an irregular shape of approximately
200,000 to 300,000 km2, with the core
in the Aouker region (Fig. 1).
The state sociopolitical organization was articulated around a monarch,
called manga, who was at the same time the chief priest of the Snake
Cult. His bureaucracy
centered in the capital town Koumbi-Saley
(Bathily 1975; Mauny 1961; Devisse 1977). However, considering the ecological setting of this town, some scholars (Posnansky 1982:21, 22) think
that it was an unsuitable location for a royal, capital city and that this
town must have been only a commercial entrepot. This position seems
untenable theoretically
as well as empirically.
It cannot be based on
ecology and considerations
of carrying capacity, because there is no
clear-cut difference in population between a royal, capital town and a
1
FIG. 1. Territorial extent of the Ghana Empire at its climax in the IOth-11th centuries
A.D. 1, Site of Koumbi-Saley; 2, site of Tegdaoust; 3, region of Mema; 4, Dhar Tichitt
region.
BACKGROUND
TO THE
GHANA
EMPIRE
75
large commercial entrepot. Archaeological investigations in the surroundings of Tegdaoust (the Aoudaghost of Arab sources) have revealed the
presence of large-scale storage facilities (Robert-Chaleix
and Richir 1983),
which were used to feed the large, nearby city. Even if we still lack
adequate archaeological
information,
Koumbi-Saley
still stands as
Ghana’s capital city.
Districts or provinces were administrated by offtcials called tunku. At
the climax of the Ghana empire, between 900 A.D. and 1100 A.D., the
bulk of state wealth came from trans-Saharan trade. Gold and slaves were
exchanged for salt and luxury items from North Africa (Thomassey and
Mauny 1956:140; Robert-Chaleix
1983). The society was stratified and
composed of four main groups. From top to bottom, there was the monarch and his kin, the nobles, bureaucrats, and district offtcials; then free
men-traders,
farmers, herders, and craftsmen; and finally slaves (Sylla
1977:85-86, Bovill 1978:80-81). This social stratification has been further
confirmed by Al-Bakri’s description of ceremonies of the royal court and
erection of a tumulus at the death of the manga, and excavations of a
large graveyard located in the southwest of Koumbi-Saley
(Mauny 1961;
Devisse 1977:89-91). In this cemetery large tombs with colonnades are
composed of one central chamber in which a high ranked individual was
buried, with additional secondary chambers, where other corpses have
been found. According to Devisse (1977:89) it seems that other corpses
were buried at the same time as that of the central personnage. However,
it can also be argued that the monument was family property in which
every member was buried at his death. Whatever the case may have been,
the most significant feature is the difference between monumental tombs
and the simple tomb types. Their presence alone is an additional argument
in favor of social hierarchy in the Ghana empire.
Archaeological
research in the Ghana region has been going on for a
long time. Most work has focused around the discovery of major cities
of West African medieval states (McIntosh and McIntosh 1980). In the
1960’s, research was widened to include some spatial patterning analysis
of the excavated mounds. The excavations of Tegdaoust are the best
example of this relatively new research strategy (Vanacker 1979; Saison
1979; Polet 1980; Robert-Chaleix
1980; Devisse 1983), and indicate that
urban layout appears to have depended on social stratification.
There
were elite quarters, which yielded the bulk of luxury items in large
houses, and craftsmens’ quarters, which yielded major concentrations of
pottery manufacture and metal working in relatively small houses. The
capital Koumbi-Saley
seems to have been based on the same model, with
an additional part, the monarch’s palace and shrine. According to the
ethnohistorical
record (Sylla 1977:86; Wa Kamissoko 1975:67), KoumbiSaley was divided into three main parts: the monarch’s residence located
76
AUGUSTIN
HOLL
in the east of the city, the center settled by traders, craftmen, and foreigners, and the west occupied by slaves and livestock.
More recent archaeological data (Robert-Chaleix
1983; Haaland 1980)
illustrate some broad trends in the regional distribution of natural resources and/or division of labor. The region of Mema yielded evidence
of large-scale ironworking in an area considered to possess the best iron
ore in that part of West Africa (Cisse 1975: 11). This development may be
related to satisfying an increasing demand for iron equipment. The small
sites of Chiguettomi,
Sboueh Leadi, and Moulay Arbad (Robert-Chaleix
and Richir 1983:349) yielded evidence of large-scale storage of agricultural products, probably to feed the large population of the nearby city
of Tegdaoust .
Hypotheses
Concerning
State Development
in Ghana
Numerous hypotheses have been presented concerning Ghana state
development.
Some have already been dismissed as fanciful and some
others have a better empirical basis. But all of them emphasize one part
of the total socioeconomic system, and consider it to be the prime mover.
The first group of hypotheses concern the so-called Hamitic hypothesis
critically reconsidered by Bathily (1977:22). Drawing conclusions from
criteria such as divine kingship and the presence of iron technology and
from the misinterpretation
of the ethnohistorical
record, the proponents
of the Hamitic hypothesis considered the Ghana empire to have been
founded by a family coming from the Middle East. This crude diffusionist
point of view has been completely rejected (Goody 1980:19).
The second class of hypotheses emphasizes the role of trans-Saharan
trade in the process of Ghana empire formation. The state is considered
to have been created in order to secure peace, and bureaucratic structures
for the sake of transactions between North Africa and the Sahelian zone.
Thus, the Ghana empire is considered to be an epiphenomenon,
mercantilistic in nature (Mauny 1961; Gellner 1977; Bovill 1978; Hopkins
1980:20). These hypotheses must face some major objections. First, large
cities already existed in West Africa around 250 B.C., long before the
advent of the classic medieval trans-Saharan trade (McIntosh and McIntosh 1980, 1983; Devisse 1983). Second, in such models, social systems
are considered passive. Thus, the linkage between trade and the preGhana social system have not been explicated.
Munson’s (1971, 1980) prehistoric research in the Dhar Tichitt led him
to suggest a new hypothesis on the origin of Ghana state formation.
Having noticed strong similarities between the dwelling patterns of Dhar
Tichitt and the present brush and wattle villages of the Mande-speaking
peoples, and also striking resemblances between the prehistoric ceramics
BACKGROUNDTOTHEGHANAEMPIRE
77
and the present Soninke pottery manufacture, Munson concluded that
the present-day Soninke are descendents of early prehistoric inhabitants
of the Dhar Tichitt region. According to him, the Lybico-Berbers,
who
destroyed the Dhar Tichitt tradition, “may have expanded farther south
where they found ready at hand rich fields of gold and reservoir of slaves.
They may have organized the trans-Saharan trade to their profit. Trading
activities may have created the necessary conditions for the beginning of
the process of state formation in Ghana” (in Bathily 1977:31).
This version of Ghana state origins, although better based empirically,
oversimplifies the processes involved in the transition to statehood, and
does not take into consideration the nature of the transactions implied in
the trans-Saharan trade.
New insight into the history of relations between central empires and
nomads in the Middle East in antiquity has shown that the direct control
of the latter by the former was always temporary. Central monarchs were
always looking for a “gentleman’s
agreement” with native elites of border
areas, mostly by gift exchanges and clientship relations (Briant 1982). By
so doing, they secured peace and trade expansion; otherwise, in case of
absolute administrative
control, the cost of transactions may have become more and more prohibitive (Blanton 1976).
Goods included in trans-Saharan trade transactions were limited by
several constraints:
The variety of commodities traded accross the Sahara was limited partly by the
length of the journey-the
desert part alone taking over six weeks-which
meant
that highly perishable goods could not be taken and partly by the high transport
charges which are said to have at least doubled the cost of most goods carried and
meant that it was worth while to carry only things of small bulk and high intrinsic
value (Wickins 1981:148).
The southward flow of goods consisted of certain foodstuffs, such as
dates, wheat, dried grapes, and nuts, and above all, commodities
like
swords, horses, books, beads of glass, shell, or stone, salt, spices, perfumes, manufactured iron tools, copper, woolen clothes, turbans, aprons,
and fine enamelware. Slaves, gold, cotton clothes, hides, skins, leather
goods, ivory, pepper, and kola nuts were the main northward-bound
goods (Mauny 1961; Hopkins 1980; Trimingham
1980:10-l 1; Wickins
1981).
Gold and slaves were the main goods of the Sahelian zone. The procurement system necessary to acquire them was not a simple one. Foreign
traders or political authorities were obliged to deal with middlemen, and
the middlemen were mainly local elites. The slave trade often involved
raiding, but many scholars (Wickins 1981:164-65; Thornton 1982; Hopkins 1980) agree that African middlemen controlled the landward side of
the trade, The same argument is applicable to gold in West Africa. Gold
78
AUGUSTIN
HOLL
mining zones were all located in the southern marshes of the Ghana
empire. According to Bathily (1977: 114- 1 IS), great empires had not taken
control over the gold production per se. In fact, they only had control
over the transactions, but production always stayed out of the might of
great emperors. While explaining the nature of the gold trade transactions, Bovill (1978:79-85) expressed the same idea:
The merchants once tried
capturing one of the timid
it was three years before
because they had no other
to discover the source of the gold by treacherously
negroes. He pined to death without saying a word, and
the negroes would resume the trade, and then only
way of satisfying their craving for salt. (Bovill 1978:82).
If local social systems were at least partially involved in the transSaharan trade, they had probably witnessed some transformations;
these
changes may have been quantitative and/or qualitative. Local households
were full participants in the transition towards statehood, not mere passive structures out of the historical realm. As has been recently demonstrated by Warmer (1983596) in his sociological analysis of Precolonial
Bamenda:
Everywhere on the plateau, the household was the primary unit of production and
consumption. These households were varied in size
. But in all of them kinship
assumes the role of relations of production. It determines who may have access
to resources and means of production, organized the process of production and
the distribution of goods and services.
The households were differentially integrated in the interregional trade
network, directly or via the village and/or chiefdom levels. Their dynamic
interactions in time and space was one of the major features of the rise
and fall of paramount chiefdoms on the Bamenda plateau. These observations shed new light on the mechanism of trade in ancient societies and
its interactions with social systems. Munson’s position, being very general, failed to demonstrate these dynamics.
In foeusing on the transition to statehood in West Africa, and specif’itally the advent of state formation in Ghana, we have to deal with multiple
definitions and try to look for bonds or links between them. In Bateson’s
(1984: 19) words, “the structure which binds is a metastructure,
that is a
structure of structures.”
In the perspective taken here, this means a
sound, comprehensive definition of statehood. Therefore, the building of
analytical concepts of state formation should involve spelling out the
range of social behaviors implied and the alternative possibilities
for
human actions.
ii. DEFINITIONS,
The problem
both qualitative
THEORY, AND METHOD
of transition to statehood is multifaceted and includes
and quantitative
transformations
of human social sys-
BACKGROUND
TO THE
GHANA
EMPIRE
79
terns. The implications
are economic, political, religious, demographic,
and ecological. To grasp this complexity, we need some clearly stated
definitions, a theory which links the different components and a method
which allows us to tit factual evidence together.
I. Definitions
The phrase transition to statehood refers to the study of a process of
social change from a former social stage A to a latter social stage B. This
evolution is a process requiring a great deal of time, and the opportunities
to observe it ethnographically
or in the ethnohistorical
record are rare
and very limited in scope. Thus the analysis can only be inferential.
There is a huge amount of literature on the problem of development of
state formations (Steward 1955; Fortes and Evans-Pritchard
1940; Wright
and Johnson 1975; Johnson 1976; Wright 1977; Blanton et al. 1981; Jones
and Kautz 1981; Haas 1982; Friedman and Rowlands 1977; Cahiers
d’e’tudes ufiicaines, Special Issue 1982; Eisenstadt et al. 1983; Quilter
and Stocker 1983; Claessen 1984; etc.). In their now-classic book on
African political systems, Fortes and Evans-Pritchard
(1940) presented
two main types of political organization of African societies: the segmentary system and the centralized one. They stated that “those who consider that a state should be defined by the presence of governmental
institutions will regard the first group as stateless societies and the second
group as primitive states” (1940:5), and the transformation from the first
type to the second was briefly explained in terms of invasion or conquest
(194O:lO). It now seems that the situation is not so simple, and according
to Alexandre (1982:229) relations between stateless societies and centralized ones can no longer be based upon arguments of mutual exclusiveness.
Attempts at a universally acceptable definition of the state are a waste
of time and energy. The numerous definitions already used by various
scholars merely need to be operationalized;
in other words, we need to
give them an archaeological content and visibility. But archaeological data
are ambiguous, and their meaning can only be demonstrated by clearly
stated archaeological test implications.
Following Haas (1982:3), a state can be defined “initially
in the most
general terms as a society in which there is a centralized and specialized
institution of government;”
the stage just below state level will therefore
show trends towards centralization
and specialization. These sociopolitical organizations have been granted various names in the anthropological
and historical literature: chiefdoms, pristine or “asiatic” states, statelets,
or segmentary states.
According to Carneiro (198 1:45), a chiefdom is an autonomous political
unit comprising a number of villages or communities under the permanent
80
AUGUSTIN
HOLL
control of a paramount chief. He considers chiefdoms to be the only route
to the state; their emergence was a qualitative step, and everything that
followed, including the rise of states and empires, was in a sense merely
quantitative. Pristine or “asiatic” states emerged from the tribal system
(Friedman and Rowlands 1977:216-220); this concept is explicitly restricted to the earliest state formations. The size of the “asiatic” state
may not exceed an area of 20 to 30 km radius with a population of around
10,000. The size of this political unit may depend largely on the ability
to centralize the economy and to prevent accumulation of labor and surplus in peripheral areas.
Riley, quoted by Doolittle
(1984:13) characterizes statelets as possessing ranked rather than egalitarian societies with a ruling class, economic life being based on irrigation agriculture and heavily oriented toward trade.
In his discussion on the Alur system and political theory, Southall
(1970:246-247) stated that:
the distinction between state and segmentary organisation is theoretically valid,
and at abstract level intermediate forms demands no separate category. But in any
scheme of classification which claims empirical relevance, the criteria of legitimate
isolation are different and any empirical form which has a certain frequency, stability and structural consistency must receive due consideration.
Hence, he called this intermediate form a segmentary state, which is
a political system combining localized lineage segmentation with specialized political institutions, while trying to trace its development from
the interaction of contrasting social structures, and pointing out its implications for the theory of the state.
From this series of definitions, it is clear that there are major lines of
agreement between the authors quoted. Whatever the main emphasis of
each of them, the chiefdom, the pristine or “asiatic” state, the statelet,
and the segmentary state all appear to be supracommunities
and regional
phenomena. These phenomena have spatial correlates in the form of settlement patterns, which can be tested archaeologically
(Blanton et al.
1979).
Transformations
in other spheres of social activities are also implied.
Social stratification, labor intensification and exchange, demography, and
environmental
conditions interact to produce new patterns. Any study of
state formation must take into consideration the dynamics of each analytical unit and its connections with other variables within a systemic
framework.
2. Theory
To avoid the “prime-mover”
bias, a systemic approach is needed to
analyze the problem of the transition to statehood in Africa (Phillipson
BACKGROUNDTOTHEGHANAEMPIRE
81
1979; Butzer 1981; Bisson 1982; Eisenstadt et al. 1983). In explaining
change in human societies, one ought to deal with the “core features,”
those characteristics that are not epiphenomenal,
but basic to all societies, permitting valid comparisons (Blanton et al. 1981: 17).
In general terms, the problem dealt with in this paper can be stated as
scale, integration, and complexity, or from another point of view, structure, system, and dialectics (Blanton et al. 1981; Godelier 1977, 1980;
Kowalewski et al. 1983).
Scale refers to size of the unit being analyzed; this unit may be a spatial
one, for instance, a region, a site, a compound, or a house; it may also
be a social one, that is a household, a community, or an aggregate of
people. Thus, scale may vary according to the analytical level of investigation.
Integration refers to the dynamic relations between units- their interdependence, which may be conceptualized as a system-wide regulation
of information flow. This integration may be of several kinds-economic,
political, social. “Depending on the kind of component units, the connections are established as flow of material, energy, information
or
people. The greater the flow through interconnecting
channels, the
greater the interdependence.”
(Blanton et al. 1981:20).
Complexity refers to the extent to which there is functional differentiation among societal units, and is therefore obviously connected to integration. Theoretically,
complexity
can be partitioned
into two main
components: horizontal and vertical differentiation;
the former refers to
functional specialization among parts of equivalent rank, and the latter
to a hierarchy of functionally diverse parts of a societal system.
According to Blanton et al. (1981:231):
It may seem that by this discussion we have unnecessarily complicated the issue
of state origins. But on examination our statements concerning the concepts of
vertical and horizontal differentiation, integration, and scale are much less abstract
and much more operational than the key concepts of many of the current hypotheses of state origins.
Hence, whatever evolutionary
model one assumes, it is obvious that
the process of state formation implies increasing scales of inclusiveness
in every sphere of human experience. A discussion of some components
of societal activities: political, economic, and ideological factors, will
show the analytical potentials of the Blanton et al. scheme.
Political factors. Causal explanations of increasing political control in
human societies are commonly made on the basis of a division between
“conflict theories” and “integrative
theories” (Haas 1981, 1982).
From the point of view of “conflict theories,” the acquisition of political power by a social group is through the control of critical resources.
82
AUGUSTIN
HOLL
Thus, this group is in a position to enforce obedience and manipulate
power for its benefit.
theories” emphasize coordination of
On the other hand, “integrative
different social components to minimize the cost of decisions for the
benefit of the whole society.
This distinction is better illustrated by the current debate in anthropology over the process implied in social hierarchy, social stratification,
or, in the terms used in this paper, vertical differentiation.
Discussions
on the Indian caste system lead to emergence of two traditions of explanations: the ritualists and the power theorists.
According to Dumont (1966, 1977, 1983), one of the major proponents
of the ritualist’s position, the underlying logic of the Indian caste system
is the opposition between the pure and the impure; thus, as one moves
from the top to the bottom of the social hierarchy, purity decreases. The
positions of individuals in relation to each other are accordingly defined,
and power forced to express itself in religious terms is asserted to be
under the predominance of values, that is, religion or status.
Political theory persists to identify itself with a theory of power, that is, to consider
a minor problem as a fundamental one, which is contained by the relations between
“power” and values or ideology (Dumont 1977: 19).
The power theorists (in Borgstrom 1977:327) consider caste as directly
comparable to other forms of social stratification. It is therefore only an
instance of a more general phenomenon, based on rules of recruitment
and interaction that are not ordained by values, but defined by the ruling
strata in the society; hence status is secondary to power. Some marxist
scholars share this point of view (Kubbel in Gellner 1977).
According to Borgstrom (1977:325) these extreme positions both fail to
explain how power is related to value in the Indian caste system. After
all, both claim to say something about the same institution,
the caste
system, and if both are really necessary to understand that phenomenon,
this must mean that they are situational and should be referable to a
common basis that accounts for both.
In general terms, social stratification refers to all forms of social inequality and may include caste and rigid occupation classes as well as
age and sex stratification.
Narrowly defined, it may only deal with specific kinds of inequality in which society-wide strata are obviously recognized. Social stratification is thus a basic element of social organization
in all human and some animal societies; “interpersonal
and intergroup
relations of dominance and submission, rank or hierarchy appear wherever people live together” (Cancian 1976:227).
For our purpose, social stratification is better explained in terms of
responses to changes of scale in a society’s decision-making
process
(Johnson 1982); in situations where sufficient numbers of people accu-
BACKGROUNDTOTHEGHANAEMPIRE
83
mulate and have to share space, considering the limited capacity of the
human brain to process information, and the inability of humans to be at
more than one place at the same time, social creation of stratification
occurs.
It appears that conflict versus integrative and status versus power rationales may be present at the same time within any sociopolitical context, mainly because the origin of a specific political condition may have
little or nothing to do with the perceptions of those being ruled (Jones
and Kautz 1981:20).
Economic factors. Discussion of economic factors implied in the transition to state organization is often made at the macro level of trade,
external communication,
and environmental
complementarity.
The importance of this macro level cannot be ignored; however, analysis on this
scale tends to underemphasize the contribution of lower level economic
components such as households.
If economy is defined as a sphere of social activities dealing with the
procurement of resources and their distribution and consumption for the
maintenance and reproduction of the society as a whole, in many cases
the household appears to be the primary socioeconomic entity which is
horizontally and/or vertically differentiated.
One of the critical features of the household economy or Domestic
Mode of Production (Sahlins 1968, 1976) is the recruitment of the labor
force. From this point of view, in agricultural, self-sustaining societies,
the forces of production are essentially tools and land. They are relatively
easily accessible to all members of the social group, but the skills necessary to achieve various socioeconomic tasks “provide those who possess them with genuine authority over the laymen since the continuation
of the group depends on this knowledge” (Meillassoux
1980a:137). This
knowledge is often acquired with seniority, supporting the fundamental
senior/junior relationships. The man/woman relationship is not parallel to
it but it has its own dynamics which may be complementary
or contradictory in the household structure. Considering the overlapping nature of
generational structure in all societies, one may agree that there are always
at least three generations present in a living society: the old, the middleaged, and the young.
Societies need some mechanism which allows individuals to give to the old and
have confidence that they in turn will be the recipients of gifts when they are old.
Money, ritual knowledge, clubs, land, kinship groups, and a host of others are all
social arrangements which help to inspire this confidence and to achieve an optimal
allocation of consumption (Walsh 1983647).
In this perspective, kinship relations in a household are at the same
time relations of production, and two types of circulation of goods and
services can be inferred. The seniors receive the juniors’ prestations and
84
AUGUSTIN
HOLL
then supervise the whole of the group output. In this particular case, the
generalized reciprocity model (Sahlins 1968, 1976) is misleading and the
concept of delayed reciprocity seems more appropriate. According to
Bouju (1984:19), the prestation/redistribution
cycle is the expression of
an asymmetrical and complementary
social relation. In that relation every
transfer of goods or services is a response to the simple obligation included in the set of rights and duties implied by kinship status.
The successful household leader may therefore have a large number of
people dependent on him in order to secure labor force, large amount of
output, authority, and prestige. The qualitative link between the household structure and the community level is based upon these features. In
the anthropological
literature, surplus production is sometimes considered to be the key variable in vertical social differentiation.
In a review
of theories of social stratification in sociology and anthropology, Cancian
(1976:230- 1) persuasively argued that the idea that stratification resulted
from the creation of a material surplus is not only a simplistic view, but
also has little support since its definition in absolute terms is not workable. But in relative terms this concept may be useful as an indication of
an increased output production beyond the subsistence requirements of
a production unit. Economic differentiation
based on control of people
is not operative in terms of so-called “surplus” production but only in
terms of prestige arising from a large compound (Bohannan and Bohannan 1968:223). Correlated with the acquisition of prestige goods, a
high food-output
production
strengthens the position of a household
leader as a potential community leader.
These prestige goods may often be in exotic raw materials, shell, stone,
or other socially valued matter (gold, iron, etc.); they may also be in
livestock, cattle, horses, or camels, for instance. Whatever the case may
be, prestige goods are not only valued on a material basis, but, belonging
to the higher level of exchange hierarchy, their social value is predominant. They are insignia of wealth and success (Iteanu 1983: 178). It is thus
possible for an individual to invest wealth if he “converts up” into a
morally superior category; to convert subsistence wealth into prestige
wealth and both into women is the “economic aim of individual Tiv.
Such conversion is the ultimate type of maximization”
(Bohannan and
Bohannan 1968:234-235).
If the dynamics of the household economy are correlated to the sphere
of prestige goods exchange, it may be one possible way to increase vertical differentiation
in a community. A successful household leader with
a large following and a large amount of output production who converts
its wealth into prestige items may become, on the “big-man”
model, a
man of prestige and gain authority over his fellows at the community
level and/or beyond it.
BACKGROUND
TO THE GHANA
EMPIRE
85
Ideological factors. The place of ideological factors or schemes of social representation
in the transition to statehood cannot be underestimated. Working on African data, Eisenstadt et al. (1983: 1237) found that
the emergence of states in Africa is not only the result of structural differentiation of political roles, but is also strongly correlated to symbolic
differentiation
of the political sphere through its concrete manifestations
in different types of centers. That is, according to the fusion or separation
of secular and religious offices, or their various combinations, the forms
of emergent states are different. The same idea is presented by Friedman
and Rowlands (1977) in their notes towards an epigenetic model of the
evolution of civilization.
The very existences of technology, subsistence activities, and social
relations affect societies as they are symbolically mediated (Sahlins 1980;
Coe 1981; Freidel 1981; Keatinge 1981); elite ideologies and religious
practices are more than simple epiphenomena
of the techno-environmental-political
realm.
In the absence of naked coercive force, religious sanctions provided a path toward
the development of political centralization
. . The amount of resources both
human and natural, expended on religion in the form of temple construction and
tribute must have a profound effect on the configuration of early state economies
as well as on the forms of political organization developed to organize and direct
the exploitation of these resources (Keatinge 1981:187).
In this regard, the senior/junior relationship previously examined in the
household socioeconomic dynamics may provide one possibility of understanding the lineage system’s ideology. For example, land allocation
systems, or, more generally, the access to means of production,
is
strongly controlled by a segmentary society’s ideology, particularly in
cases of communal ownership (Bouju 1984:104).
This short review does not pretend to have reviewed the totality of
factors involved in the transition to statehood. Other factors were certainly involved in this process, among which were demography and environment. However, within the framework outlined, a theory of primary
state development in West Africa can be presented.
A theory of primary state development in West Africa. The development
of primary states involved transformations in scale, integration, and complexity of the whole society, but transformations in higher scales are not
mere summations of the lower-level ones, the whole being always more
than the sum of its component parts. From this point of view, a hierarchical model of state development includes an analysis of at least three
levels: the primary social unit, the household; the community, which is
an aggregate of a varying number of households; and the region, functionally defined as a two-dimensional
space in which a patterned behavior, a network or system, occurs (Tourney 1981:470).
86
AUGUSTIN
HOLL
At the household level, kinship relations, recruitment of labor force,
senior/junior relationship, with its corollary delayed reciprocity, output
intensification,
and prestige goods circulation are in a state of punctuated
equilibrium
and link the primary socioeconomic
entity to community,
regional, and/or extraregional networks. Population increase and generational segmentation allowed territorial expansion (Sahlins 1961, 1969;
Bohannan and Bohannan 1968; Pollet and Winter 1980; Bouju 1984).
Given enough time, one would expect a wide distribution of historically
related communities
in a given region. The most successful household
leader, having a large following and a large output production, may therefore be in a position to acquire elite goods, greater distributive capacities,
and prestige, incentives to vertical differentiation
among households.
At the community level, cooperation and competition are both present.
Household leaders with the greatest distributive capacities and prestige
have better opportunities
to reach, directly or indirectly, positions of
authority or community managers, organizing communal works for the
benefit of the whole community (Harris 1979:98). Their way of living may
not be very different from that of their dependents but
. . under the guise of “elite goods” they collectively have at their disposal a
surplus directly or indirectly produced by the surplus labour of the juniors and
they use this surplus to control the reproduction of the lineage groups and correspondingly the reproduction of the dependence of these groups on themselves
(Dupr6 and Rey 1980~195).
Broadly defined, any society can be conceptualized
as an exchange
system (Schneider 1979:192); the control of some key nodes of the exchange network may allow gains in wealth and power. Due to community
dynamics, land availability,
population increase, conflicts, and, given
enough time, daughter communities may be created in the surroundings.
In farming and agropastoral economies these new settlements may depend, at least in their pioneer stage, on the accumulation and storage of
foods and seeds of their former communities (Meillassoux 198Ob). These
relations may be of various natures, political, economic, ideological, or
all of them, and may have created regional networks.
At the regional level, communities
are horizontally
and/or vertically
differentiated.
To ensure their social reproduction,
communities
are involved in multiple exchange networks. These nodes of differential concentration of people and shelters in the continuum of population distribution over the area of a region may lead to segregation, this concept
referring to the extent to which households or groups of households are
independent. Exchange of people, prestige items, and tribute may be
channeled through the networks thus created.
BACKGROUNDTOTHEGHANAEMPIRE
87
. . . Linkages between specialized subsystems may take the form of transactions
in the context of central institutions. Chieftainships, governments and markets are
examples of such central institutions (Blanton 1976:251).
Differential concentration of people and historical factors such as “anteriority”
(e.g., historically
earlier arrival) may lead to a hierarchy of
settlements, which may have been reinforced by tribute flow from lower
to higher level centers, and prestige goods flow in the inverse direction
(Friedman and Rowlands 1977; Steponaitis 1978, 1981). These features
have relatively clear archaeological correlates which can be partly evidenced by locational analysis.
In the model presented here, an attempt is made to build a subtle but
strong inferential device which will allow us to make sense of the Dhar
Tichitt archaeological record. The model has fragments of a deterministic
approach to social systems, but, following Harris (1979:108), it is argued
that it is not by denying the existence of a deterministic
component in
social processes that one will improve that situation; and for the sake of
intelligibility,
it is necessary to delineate such a component. The empirical
test implications
of this model include an archaeological analysis of a
household’s dwelling unit through the spatial distribution of artifacts and
domestic facilities, an intrasite analysis through the distribution of household dwelling units at the scale of a site, and, finally, the consideration
of settlement patterns at the regional scale, with a discussion of patterns
of land use.
3. Method
Most researchers agree that the processes involved in the transition to
statehood are composed of many sets of relations which are stochastically
combined. Hence, a methodological
device is needed to grasp this complexity. The approach followed here falls generally into the research
strategy of cultural materialism outlined by Harris (1968:659):
The essence of cultural materialism is that it directs attention to the interaction
between behavior and environment as mediated by the human organism and its
cultural apparatus. It does so as an order of priority in conformity with the prediction that group structure and ideology are responsive to these classes of material
conditions.
Therefore, after the description of environmental
conditions during the
Dhar Tichitt neolithic occupation, an attempt is made to show how subsistence patterns and group structure were organized. While emphasizing
archaeological visibility of patterned durational behaviors, we will heavily
rely on actualistic studies, in ethnography as well as human geography,
to gain knowledge on features like land use patterns and subsequent set-
88
AUGUSTIN
tlement patterns, constraints
decision-making
structures.
imposed upon agropastoral
III. AN ARCHAEOLOGICAL
TICHITT
1. Dhar
Tichitt Region:
HOLL
Physical
CASE STUDY:
REGION
economies,
and
THE DHAR
Setting and Paleoclimatology
The Dhar Tichitt region, forming part of the sandstone cliff series of
southeastern Mauritania, is situated between 18”20’/18”27’ N. latitude and
9”05’/9”30’ W. longitude, in the present southwestern part of the Sahara
desert. The study area measures 44 by 15 km (660 km*). A line of steep
sandstone cliffs, standing on the average 60 m above mainly sandy flats,
and a series of interdunal depressions formerly filled with freshwater
lakes, form its main topographic features. The small plateau is dissected
by narrow valleys of seasonal streams or wadis. The hydrographic regime
was an endoreic one, the wadis emptying into lakes located in sandy
interdunal depressions (Fig. 2). This hydrographic network was fed by
rains and the water table.
Palaeoecological
studies (Hugot 1977; Munson 1981; Ho11 1983) have
shown that the main climatic feature of the Dhar Tichitt from 4500 to
2000 B.P. was the existence of two contrasting seasons: an important dry
season of 7 to 9 months and a short wet season with stormy rains. The
vegetation was savanna like, with a light tree component inhabited by a
fauna composed of herbivores-antelopes,
hippopotamus-as
well as
lions.
During the later Holocene, this region was settled by populations of
farmers and livestock herders, who complemented
their subsistence by
hunting, fishing, and collecting of wild plants resources. Bulrush millet
(Pennisetum
sp.) was the only cultivated plant attested in the archaeological record, while cattle and sheep/goats were the herded animals. In
radiocarbon years, this neolithic occupation lasted from 3850 + 250
(Hugot 1979:850) to 2170 + 105 B.P. (I 3819), when increasing trends
towards desertification
made the region less suitable for permanent
human settlements.
2. Archaeological
Record
The archaeological record of the Dhar Tichitt region is composed of
many kinds of cultural material data, from individual artifacts in stone,
bone, and ceramic to massive architectural features, comprising dwelling
structures in dry stone masonry. Fauna1 and botanical remains and rock
BACKGROUND
TO THE
GHANA
EMPIRE
89
90
AUGUSTIN
HOLL
drawings are also present (Amblard 1981; Beyries 1981; Delneuf 1981;
Ho11 1983; Hugot 1979; Munson 1971).
Based on a general survey in 1980, (Rapport preliminaire
n.d.), the
analysis of low-altitude aerial photos (IGN, Mission MAU 3025, 1980),
and Munson’s (1971) previous research, a total of 46 sites have been
mapped in our 660-km2 study area. The distributions
of ecofacts and
artifacts at these sites were considered in relation to the Dhar Tichitt
ecosystem. Consequently two categories of sites could be distinguished.
The first category is composed of 43 sites located on the plateau and
characterized by built structures of dry stone masonry. These are considered as main settlement types (Ho11 1983, 1985a).
The second category includes three sites (Sites 1, 45, and 46) located
in the sandy interdunal depressions and characterized by scatters of archaeological material without any built structures. They have been interpreted as dry season camps (Fig. 3).
In this study only the first category of sites is considered. These sites
are aggregates of compounds of various forms and surface area. Each
compound is materially delimited by a dry stone wall. The compound
thus appears to be the primary spatial unit in the Dhar Tichitt neolithic
system.
3. Subsistence Base and Socioeconomic
Organization
Subsistence activities of the Dhar Tichitt populations can be partitioned
into their major component parts, that is, millet cultivation, cattle and
ovicaprine herding, hunting, fishing, and wild plant collecting. Socioeconomic organization,
referring to the acquisition of resources, and their
distribution and consumption,
includes the means of production: necessary land, technology, and skills. A related topic includes the social relations of production, that is, the relative positions of individuals with
respect to each other in the productive process.
This agropastoral system is composed of two main cycles of activities
which must be wisely combined: agricultural tasks and pastoral activities.
Both have periods of high labor input which can occur at the same time
or at different times. The annual calendar may therefore present some
bottleneck periods of intensive labor input. In these periods, the labor
force is a critical factor (Hall, 1985b) and the primary socioeconomic unit
must be organized in order to cope with that situation.
The household level. Households are the level at which social groups
articulate directly with economic and ecological processes; they are a
product of a domestic strategy to meet the productive, distributive, and
BACKGROUND
TO THE
GHANA
91
EMPIRE
Y
5
a
l
92
AUGUSTIN
HOLL
reproductive needs of its members, and their nature varies considerably
from society to society. An archaeological definition of a household encounters many difftculties, mainly because coresidence is only one of its
criteria. Nonetheless, in this paper, the core of the household is assumed
to live in the same dwelling unit. In the Dhar Tichitt archaeological record, the compound, being a well-delimited
spatial unit, can be equated
with a minimal social aggregate. Due to the distributional
regularity of
compounds, it is possible to consider them to be archaeological remains
of households.
The major portion of fieldwork for this study was carried out at Site
38. It is composed of 200 compounds, extending over an area of 12 ha
(600 x 200 m). Individual compounds vary in size from 200 to 1000 m2,
the number of inner dwelling units varying from two to seven (Fig. 4).
The excavation of compound 50E allows one to gain insight into the
spatial organization of a domestic unit.
Compound SOE is an oval-shaped feature of 450 m2; it is composed of
a dry stone wall of 87.6 m perimeter, 1.5 m average height, and 1.4 m
average thickness, five individual dwelling units, one storage facility, and
one hearth (Fig. 5). The distribution
of grinding equipment and stone
anvils/hammerstones
in the dwelling units may indicate a coresidence of
I
-_c
p
Circumvallation
Stone
walls
wall
1,11,111 Livestock
mmnm Escarpment
FIG. 4. Site 38.
enclosures
BACKGROUND
TO THE
GHANA
93
EMPIRE
5m
0
0
Hearth
0
Stones
FIG. 5. Compound
=
Stone
walls
50 E.
five nuclear families in compound 50E, leading to a minimal estimate of
ten adult inhabitants (Ho11 1983).
Due to the multiplicity
of productive tasks implied in an agropastoral
economy, a nuclear family may not be a viable socioeconomic
unit.
Larger familial units, that is, extended or polygynous families, are therefore the most common minimal entities in these kinds of societies, as
“large households have a potential for great flexibility
in dealing with
diverse or scattered economic opportunities
that require simultaneous
labor” (Wilk and Rathje 1982:632).
The community level. A community
is an archaeological abstraction;
it is based on the assumption that groups lived in discrete spatial spots
in which they left clusters of material remains. In the Dhar Tichitt archaeological record, sites are aggregates of compounds, or what is socially the same, communities
are aggregates of households. The distri-
94
AUGUSTIN
HOLL
bution of compounds according to their size does not show any clustering.
However, according to our model, one would expect the number of inner
dwellings to indicate the number of household members, the largest being
the wealthiest in terms of productive capacity. Cooperation and competition may be considered as two major aspects of social interaction at the
community level.
Cooperation involves the existence of special purpose groups. In the
study area, the presence of circumvallation
walls around sites 6, 15, 38
and 42 and also large livestock enclosures in the near proximity of the
same sites are evidence pertaining to the existence of some communal
undertakings. The decision-making
process can only be inferred. In this
regard, livestock herding may have been, at least partially, a communal
affair. The erection of a circumvallation
wall, whatever its functions, is
obviously the outcome of a decision enforced for the benefit of the community as a whole. Considering household dynamics, wealth expressed
in a large following, allowing the control of labor force, the ability to have
a larger output of production (allowing redistributive capacities), and the
conversion of wealth into prestige items may allow a successful household’s leader to gain authority and thus enable him to organize and
manage communal affairs. The archaeological correlates of these features
may be large compounds and the presence of exotic items interpreted as
prestige goods. In Site 38, the largest compounds measure 900 to 1000
m2; the number of inner dwelling units in each amounts to seven; their
minimum number of occupants may have been fourteen, a number well
under twenty, which is the average number of household members computed from some West African ethnographic observations (Lericollais
1970:125; Pollet and Winter 1980:332; Bouju 1984).
Prestige goods found in the Dhar Tichitt archaeological record are exclusively beads made of exotic raw materials: carnelian, a translucent red
chalcedony, and amazonite, a pale-green slightly translucent microcline
(Mauny 1956:141; Munson 1971:324-325;
Amblard
1981:483; Ho11
1983:238). The source of carnelian is still unknown; the nearest known
source of amazonite is, according to Mauny (1956), in the Hoggar, about
1600 km east-northeast of Dhar Tichitt, or, according to Amblard (1981),
in the Mauritanides
geological formation near Tidjikja,
about 200 km
northwest of Dhar Tichitt. Whatever the case may be, the control of this
long-distance exchange network was probably one important factor in the
vertical differentiation among households at the community and regional
levels.
Regional level. Differential aggregation of populations according to the
distribution of natural resources, kinship, and/or political affiliation leads
to specific patterns of settlements at the regional level. The archaeological
visibility of these features is through locational analysis (Vita-Finzi and
BACKGROUND
TO THE
GHANA
95
EMPIRE
Higgs 1970; Vita-Finzi 1978; Johnson 1977, 1980, 1981; Steponaitis 1978,
1981). Land available in the Dhar Tichitt can be partitioned into three
broad categories, according to their potential use (Table 1, Fig. 6). This
information is gathered from maps (IGN 1965, Tichitt: feuille NE 29XV)
and checked during the 1980 and 1981 field seasons.
Arable land well suited for millet cultivation is composed of light, welldrained, sandy soils of the Piedmont and more clayey coils located on
the plateau. The first class of arable land may have allowed the cultivation
of short-maturation-cycle,
low-yielding
millet by decrue techniques
(Munson 1971), archaeologically
attested by small, dry-stone dams across some wadi valleys. The second class of arable land may have been
used for the cultivation of 120-day-maturation-cycle,
high-yielding millet.
Arable land amounts to 57.34% of the study area.
Relatively good grazing land may sometimes be intermingled with arable land. But to avoid interference and perhaps the destruction of highly
valued food, pasture areas may be located in the surroundings, on the
plateau as well as in the sandy interdunal depressions. This category of
land amounts to 15.88% of the total land available (Table 1, Fig. 6).
The third category of land is composed of rocky surface areas, and is
classified as rough grazing and poorly arable; it amounts to 26.82% of the
total. An interesting observation can be made by looking at Figure 6. All
settlements are located on this third category of land; this feature can be
interpreted as a strategy to minimize the loss of arable and good grazing
land while being in optimal locations with respect to the availability of
sandstone, used as building material.
Patterns of settlement may reflect strategy of resource acquisition, kinship, political affiliation, or all of these. The Dhar Tichitt settlements
present a three-tiered hierarchical pattern, which includes five nodes or
clusters of sites.
LAND
USE POTENTIALS:
Arable
DISTRIBUTION
land
Relatively
good
grazing
Ha
%
1
2
3
4
5
8622
8950
7050
3270
10,500
45.98
59.66
73.43
60.55
57.69
-
Total
38,392
57.34
10,fm
Cluster
TABLE
OF LAND
Ha
%
3750
1050
20.00
7.00
1350
4450
25.00
24.45
-
15.83
1
USE CATEGOIUES
AMONG
SITE CLUSTERS
Rough
grazing and
poorly arable
Ha
6378
5000
2550
780
3250
17,958
%
Ha
%
34.01
33.83
26.56
14.44
17.85
18,750
15,000
9600
5400
18,200
28.00
22.40
14.33
8.06
27.18
26.82
66,950
_:
.:.
I
0
‘5km
,
l
_,..,.
Main center
I
4
Hamlet
FIG. 6. Dhar Tichitt: Distribution
..;._..,
* Homestead
of land use categories.
,...
BACKGROUNDTOTHEGHANAEMPIRE
4. Settlement
Hierarchy
97
and Dynamics
Our sample of 43 sites can be partitioned into three classes according
to their sizes, expressed in surface area and/or number of compounds.
The largest sites referred to as rank 1 sites, number four. Their surface
area and number of compounds vary from 3.5 ha and 120 compounds for
Site 15 to 12 ha and 200 compounds for Site 38, including Site 30, 4 ha
and 140 compounds, and Site 42,4 ha and 180 compounds. This class of
sites presents some specific archaeological features: with the exception
of Site 30, the three other sites have circumvallation
walls and large
livestock enclosures in their near proximity. All of them are located in
what may be considered optimal areas, just over or near the cliff escarpment at major wadi mouths, overlooking the sandy interdunal zone (Fig.
7) They are also clustered in what may be considered the regional core
area; their relative spacing from each other, in straight lines, varies from
16 km between Sites 15 and 38 to 5 km between Sites 38 and 42. The
mean distance is 9.58 km, with a range of 11 km and a standard deviation
of 5.18 (n = 6, Table 2). In this analysis, these sites are referred to as
main centers or villages.
Middle-size sites are referred to as rank 2 sites; this class is composed
of 10 sites, each of them containing a minimum of 20 compounds and a
maximum of 50. They are distributed in two main clusters, spatially well
segregated (Fig. 8): an eastern cluster of four sites (Sites 36, 37, 39, and
40), and a western cluster of six sites (Sites 6, 9-13). These sites, which
are only aggregates of compounds without any other characteristic archaeological feature, Site 6 excepted, are considered as hamlets.
There are 29 small sites refered to as rank 3 sites. They are characterized by the aggregation of a maximum of 19 compounds, and are evenly
FIG. 7. Dhar Tichitt: Distribution
of rank 1 sites.
98
AUGUSTIN
TABLE
DISTANCES
HOLL
2
(IN KM) BETWEEN RANK
1 SITES
Site
Site 15
Site 30
Site 38
30
38
42
11.5
16
14
5.5
5.5
5
distributed all over the region. But, they show a trend toward a preferential location in the hinterland, along wadis and the western peripheral
area (Fig. 9). In this study they are considered as homesteads.
This settlement hierarchy is more meaningful if we consider it in relation to the chronology of the Dhar Tichitt archaeological record. One
must keep in mind that we are studying processes requiring a great deal
of time; the chronological scheme will therefore be built accordingly.
The chronology is based upon a sample of 25 radiocarbon dates (Table
3); an alternative chronology (Munson 1971) was not satisfactory (Ho11
1983:186- 198). It is worth noting that the radiocarbon sample of dates is
biased toward rank 1 sites, particularly Sites 38 and 30. However, comparison of dates with the various categories of sites shows some interesting trends (Table 4).
In this regard and keeping in mind the household and community segmentation in time previously discussed, it is interesting to find that rank
1 sites seem to be the earliest settlements in the region, followed by rank
2 sites and finally rank 3 sites. In this last class, one radiocarbon date
from Site 17, 3700 + 130 B.P. (GX 1890), measured on a shell sample,
seems unreliable. This trend may evidence population flow from earlier
G
G
:
FIG. 8. Dhar Tichitt: Distribution
of rank 2 sites.
BACKGROUNDTOTHEGHANAEMPIRE
FIG. 9. Dhar Tichitt: Distribution
99
of rank 3 sites.
settlements to new areas in the region. The process involved may have
been the search for new agricultural lands, game, and fuel.
’
At the regional scale, the settlement hierarchy is dominated by four
centers, each with a minimum of 120 compounds, that is, more than twice
the number of compounds of the largest rank 2 sites, Site 6. This is a
primate model (Haggett 1973:121; Johnson 1980:236) of settlement hierarchy (Fig. IO), composed of several nodes. These nodes or clusters of
sites, which may be considered as subregional social entities, will now
be analyzed in more detail.
Subregional nodes: Using settlement hierarchy data, the rank size rule
model, nearest neighbor rationale, and topographic location for minor
settlements like Sites 8, 31, and 41, five subregional nodes or site clusters
have been found in our study area (Fig. 11). Each of these site clusters
presents a primate model, but their sizes in numbers of sites and territorial
range vary (Fig. 12).
Cluster 1 is the largest and is composed of 12 sites distributed over an
area of 18,750 ha (Fig. 11, Table 1). Site 15 is its main center, Sites 9-13
are hamlets and Sites 14 and 16-19 are homesteads (Fig. 12a). Distances
from site to site (Table 5) vary from 1 to 12 km (n = 66, m = 4.59, Y =
11, SD = 2.42), and from the main center to the rest of settlements (Tables
6and7)fromlto8km(n
= 11,m = 3.72,r=
7,SD = 1.87).Population
estimates extend from a minimum of 1310 to a maximum of 2620 (Table
8), and land use potentials (Table 1) show the presence of 45.98% arable
land, 20.0% relatively good grazing, and 34.01% rocky surface, rough
grazing, and poorly arable land.
Cluster 2, centered around Site 30 as main center, is composed of 11
sites distributed over a territory of 15,000 ha (Fig. 11, Table 1). There are
no hamlets in this cluster, and the remaining settlements (Sites 20-W
100
AUGUSTIN
TABLE
AVAILABLE
3
RADIOCARBON
Site
Sample
38
46
38
17
38
45
45
38
45
45
30
4.5
38
38
38
45
30
12
38
30
Dunal zone? (not specified)
Bone
Bone
Charcoal
Shell
Bone
Bone
Charcoal
Potsherd
Charcoal
Charcoal
Charcoal
Charcoal
Bone
Charcoal
Bone
Charcoal
Charcoal
Charcoal
Bone
Charcoal
Bone
Charcoal
Charcoal
Charcoal
Charcoal
38
4
11
HOLL
DATES
14C date
3850
3830
3776
3700
3490
3465
3425
3400
3350
3205
3205
3190
3122
3100
2975
2950
2885
2780
2760
2700
2610
2600
2430
2330
2170
2
f
k
2
2
2
2
250
120
120
130
50
160
130
2
2
2
e
f
f
f
2
f
+
k
”
2
k
k
2
f
110
95
105
110
120
105
110
100
140
140
160
115
110
105
80
105
105
Lab. No.
?
GIF 2884
DAK 52
GX 1890
MC 427
GX 1421
GX 1889
TL CRIAA
13561
GX 1323
I 3564
I 3562
DAK 203
I 3565
DAK 187
I 3563
GX 1888
GX 1325
DAK 190
GX 1324
GIF 4110
GX 1326
GIF 6083
13566
13819
are homesteads (Fig. 12b). Distances from site to site (Table 5) vary from
1 to 10 km (n = 55, m = 4.60, r = 9, SD = 2.25), and from main center
to other settlements (Tables 6 & 7) from 2.5 to 9 km (n = 10, m = 5.60,
r = 6.5 SD = 1.80). Population estimates range from a minimum of 1050
to a maximum of 2100 (Table 8), and land use potentials evidence the
control of 59.66% arable land, 7.0% relatively good grazing, and 33.83%
rocky surface, rough grazing, and poorly arable land (Table 1).
Cluster 3 is composed of 10 sites, the main center being Site 38. Sites
36,37,39, and 40 are hamlets, and Sites 31-35 are homesteads (Fig. 12~).
This site cluster covers an area of 9600 ha (Fig. 11, Table 1). Distances
from site to site (Table 5) vary from 1 to 12 km (n = 45, m = 5.02, r =
11, SD = 3.02), and from the main center to the rest of settlements (Tables
6and7)from1to10km(n=9,m=4.11,r=9,SD=2.86).Population
estimates range from a minimum of 1550 to a maximum of 3100 (Table
8), and land use potentials show the control of 73.43% arable land, 26.56%
rough grazing and poorly arable land, and the relative absence of good
grazing land (Table 1).
BACKGROUND
TO THE
TABLE
RADIOCARBON
CHRONOLOGY
Rank 1
Site
GHANA
4
AND
SETTLEMENT
HIERARCHY
Rank 2
14C date
Site
38
38
38
38
30
38
38
38
30
38
3850
3776
3490
3400
3205
3122
3100
2975
2885
2760
30
38
2700 2 115
2430 2 80
Rank 3
14C date
+ 250
f 120
” 50
f
f
f
f
f
t
105
120
105
110
140
160
12
11
101
EMPIRE
Site
14C date
17
3700 f 130
3
2600 2 105
4
2330 + 105
2780 z!z 140
2170 ? 105
Cluster 4 has the smallest number of sites. It is composed of four sites
scattered over a territorial range of 3270 ha (Fig. 11, Table 1). Hamlets
are absent. Site 42 is the main center, and Sites 41, 43 and 44 are homesteads (Fig. 12d). Distances from site to site (Table 5) vary from 1 to 3.5
1
FIG.
2
3
10. Primate model of settlement hierarchy.
Rank
102
AUGUSTIN
HOLL
FIG. 1I. Dhar Tichitt: Distribution
of site clusters.
km (n = 6, m = 1.91, r = 2.5, SD = 0.78), and from main center to
other settlements (Tables 6 and 7) from 1 to 2 km (n = 3, m = 1.50, r =
1, SD = 0.40). Population estimates range from a minimum of 950 to a
maximum of 1900 (Table 8), and land use potentials evidence the presence
of 60.55% arable land, 25.0% relatively good grazing, and 14.44% rocky
surface, rough grazing, and poorly arable land.
Cluster 5, composed of six sites, is the most eccentric one. It lacks a
rank 1 site. The highest-ranked settlement is Site 6, a rank 2 site numbering 36 compounds with a circumvallation
wall and livestock enclosures; the remaining five settlements, Sites 2-5 and 7, are homesteads
(Fig. 12e). These settlements are widely distributed over an area of 18,200
ha (Fig. 11, Table 1). Distances from site to site (Table 5) vary from 2.5
to 13.5 km (n = 15, m = 6.46, r = 11, SD = 3.39), and from the main
center to the rest of settlements (Tables 6 and 7) from 2.5 to 12.5 km (n
= 5, m = 6.50, r = 10, SD = 3.72). Population estimates range from a
minimum of 375 to a maximum of 750 (Table 8), and land use potentials
are 57.69% arable land, 24.45% relatively good grazing, and 17.85% rocky
surface, rough grazing, and poorly arable land (Table 1). This cluster
appears to be less densely occupied than the others; the implications of
this pattern will be discussed below.
Intercluster analysis. In this intercluster analysis, some major features
discussed in the previous part of the paper will be highlighted. These
include the settlement hierarchy, distribution of land use categories, and
prehistoric populations. By so doing we hope to show the presence or
absence of population pressure, however relative this concept may be,
and discuss its implications.
The settlement hierarchy over all the Dhar Tichitt presents a primate
model; the four higher level sites are of relatively equal weight, dominated
BACKGROUNDTOTHEGHANA
Size
200
EMPIRE
103
Size
200
5
x
E 100
a8 150 .\_
B
5 50
E
z
Site
1
Size
200
Clustir
a
Site
?
z 150
a
5
::
E 100
22 150 /
z
1 50
E
2
15
l3
38
c
Cl”&
s3
. J-y
Cider
23
Rank
b
Sizer
200
22
1..,
: 150
a
Cl00
x
I
8 50
d
5
5
1
30
\
1
Rank
iz0 100
6
$ 50
IL
1,
Site
Rank
Site
1
42
aue:er
d3
---hank
d
C
Size+
200
?
- 150.
:
x
E 100.
?I
6
I’“1
%,*
1
Rank
Clue:er
s3
e
FIG. 12. Primate models of settlement hierarchy in individual site clusters.
by the 12-ha surface area and 200 compounds in Site 38. These settlements are distributed in such a way that each of them dominates a subregional node of sites. The degree of intercluster variability can best be
seen by examining the constituent parts of each site cluster. Two clusters,
Clusters 1 and 3 (Figs. 12a and c) present three-tiered settlement hierarchies composed of one rank 1 site, five and four rank 2 sites, respectively, and five rank 3 sites. Clusters 2 and 4 are devoid of rank 2 sites
and therefore contain one rank 1 site, and 10 and three rank 3 sites,
104
AUGUSTIN
TABLE
HOLL
5
STATISTICAL DATAON DISTANCES BETWEEN SITES IN EACH CLUSTER
No. of
Cluster
sites
1
2
3
4
5
12
11
10
4
6
No. of
measurements
Mean
66
55
45
6
15
Range
4.59
4.60
5.02
1.91
6.46
I1
9
11
2.5
11
Variance
SD
5.87
5.07
9.14
0.61
11.54
2.42
2.25
3.02
0.78
3.39
respectively (Figs. 12b and d). Cluster 5 is dominated by a rank 2 site,
with five rank 3 sites (Fig. 12e). This last cluster seems to have been in
a process of vertical differentiation
when the Dhar Tichitt system collapsed. This argument may be strengthened by a close examination of
land use potentials, population
distribution,
and potential agricultural
yields.
According to the distribution
of potential land use categories among
clusters (Table l), Cluster 5 possess the greatest amount of arable land,
followed by Clusters 2, 1, 3, and 4. For the good grazing category, Cluster
5 has the largest proportion, followed by Clusters 1, 4, 2, and 3, which
has none. And finally, for the rocky surface category, suitable for rough
grazing and poorly arable, Cluster 1 controls the greatest amount, followed by Clusters 2, 5, 3, and Cluster 4. In the region as a whole, Cluster
1 controls the largest territorial
range, followed by Clusters 5, 2, 3,
and 4.
TABLE
DISTANCESBETWEENTHEMAIN
Cluster
Sites
15-14
15-16
15-17
15-10
15-13
15-11
15-18
15-9
15-12
15-19
15-8
1
km
1
1.5
2
3
3
4
4
4.5
5
5
8
Cluster
6
CENTERAND~THER~ETTLEMENTSIN
2
Cluster
3
Cluster
EACHSITECLUSTER
4
Cluster
5
Sites
km
Sites
km
Sites
km
Sites
km
30-25
30-24
30-26
30-22
30-23
30-27
30-21
30-20
30-28
30-29
2.5
3.5
4.5
5
5
6
6.5
7
7
9
38-40
38-39
38-37
38-31
38-32
38-36
38-33
38-35
38-34
1
1.5
2
2.5
3
4
5.5
7.5
10
6-5
6-7
6-4
6-3
6-2
2.5
3
5.5
8.5
12.5
42-43
42-41
42-44
I
1.5
2
BACKGROUND
TO THE
TABLE
STATISTICAL
Cluster
1
2
3
4
5
DATA
ON DISTANCES
No. of
sites
4
6
105
EMPIRE
7
THE MAIN CENTER AND OTHER SEITLEMENTS
IN EACH SITE CLUSTER
BETWEEN
No. of
measurements
12
11
10
GHANA
Mean
Range
11
10
3.72
9
3
5
4.11
5.60
1.50
6.50
Variance
7
6.5
9
SD
3.50
3.24
8.20
0.16
13.85
1
10
1 .a7
1.80
2.86
0.40
3.72
The distribution of population estimates (Table 8) shows that Cluster 3
is the most densely occupied, followed by Clusters 1, 2, 4, and 5.
Thus Cluster 5, controlling
the largest amount of arable and good
grazing land, is at the same time the least populated. This striking discrepancy is stronger if we examine potential agricultural yields.
Our computations of average millet yield per hectare are based upon
quantitative data published by several authors and on agricultural practices of peoples inhabiting African climatic zones characterized by two
contrasting seasons. The average millet yields published vary from 250
to 870 kg/ha in Eastern Africa (Schneider 1979:72) to 600 to 700 kg/ha
among the Soninke Dyahunu in Mali (Pollet and Winter 1980:332), including 630 kg/ha among the Dogon in Mali (Bouju 1984: 137) and 300 kg/
ha of the Serer in Senegal (Lericollais
1970: 127). We considered 300 kg/
ha to be a reasonable minimum yield, agricultural techniques employed
being relatively simple. Using this minimal value, the potential millet
yield per year of the Dhar Tichitt may have approximated
11,517 metric
tons, varying from a minimum of 981 metric tons in Cluster 4 to a maximum of 3150 metric tons in Cluster 5 (Table 9). A more realistic insight
may be gained by cross-correlating these potential millet yields with theoTABLE
POPULATION
Sites
ESTIMATES
8
FOR EACH
Compounds
CLUSTER
Population estimates
Cluster
No.
%
No.
%
Min
1
12
11
10
27.90
25.58
23.25
9.30
13.95
262
210
310
190
75
25.02
20.05
29.60
18.14
7.16
1310
2620
24.25
1050
1550
2100
3100
1900
28.70
750
6.94
2
3
4
5
Total
4
6
43
1047
950
375
5400
MaX
10.800
%
19.44
17.59
INTERCLLJSTER ANALYSIS:COMPARISONS
TABLE
9
OFINTERCLUSTERTERRITORIALRANGE,AVAILABILITYOFARABLELAND,POPULATION
POTENTIAL MILLET YIELDS, ANDTHEORETICAL
CARRYING CAPACITIES
Territorial
range
Cluster
Arable
land
Potential
millet
yield
Population
estimates
ha
%
ha
%
Min
1
2
3
4
5
18,750
15,000
9600
5400
18,200
28.00
22.40
14.33
8.06
27.18
8622
8950
7050
3270
10,500
22.45
23.31
18.36
8.51
27.34
1310
1050
1550
950
375
Total
66,950
38,392
ESTIMATES,
5400
Max
2620
2100
3100
1900
750
10,800
%
24.25
19.44
28.70
17.59
6.94
kg
2,586,600
2,685,OOO
2,115,OOO
981,000
3,150,OOO
11,517,600
%
22.45
23.31
18.36
8.51
27.34
Theoretical
carrying
capacity
(wople)
7874
8173
6438
2986
9589
35,060
$
s
2
2
3
F:
107
BACKGROUNDTOTHEGHANAEMPIRE
retical carrying capacity, human food requirements, and population concentrations. Millet is the staple food of many African populations and its
nutritional constituents are varied and relatively rich. One hundred grams
of bulrush millet (Pennisetum sp.) is composed of 10.60 g water, 12.47 g
protein, 5.00 g fat, 2.80 g fiber, and 67.13 g carbohydrates (Schneider
1979:70, Table III). A person/day consumption of 0.9 kg (Bouju 1984: 137)
seems a reasonable average quantity of food intake.
The computation of these quantitative estimates shows that the theoretical carrying capacity of the Dhar Tichitt is well beyond the effective
actual population, and the gap between the two suggests interesting observations (Table 9): the smaller the gap, the stronger the pressure. In
this regard Clusters 3 and 4 may have actually witnessed a relative population pressure, and the need for new lands in that core area may have
lead to occupation of Cluster 5. This inference is independently strengthened by two radiocarbon dates from Sites 3 and 4, 2600 + 105 BP (GX
1326) and 2330 + 105 B.P. (I 3566), respectively, and the incipient vertical
differentiation
of settlements in Cluster 5.
Vertical differentiation evidenced by settlement hierarchy probably had
political correlates. The main center of each cluster may have been the
decisive step in the intracluster decision-making
process. The various
clusters being of relatively equal strength, their mutual interaction may
have included cooperation and competition.
We have no clear archaeological evidence for cooperation, but it may have concerned exchanges
of peoples. Competition
may have concerned not only the quest for and
protection of agricultural land and livestock, but also game and aquatic
resources (Harris 1984) and paths to fuel. Circumvallation
walls attested
in the main centers of four of the five clusters may be interpreted as
pertaining to intercluster warfare, thus reinforcing the positions of these
centers as decision-making
centers.
CONCLUSION:
CENTRALIZATION
TO STATEHOOD
AND TRANSITION
Primate models of settlement hierarchy may have two main lines of
explanation. First, it may be considered as indicating an increasing trend
toward sociopolitical centralization.
It may also pertain to secondary centralization centered on lower-ranked district settlements after the collapse
of a former larger state formation.
The Dhar Tichitt archaeological record falls under the first explanation,
if we consider the absence of earlier clearly established human settlements in that region and the later archaeological,
ethnohistorical
and
historical data of the so-called Ghana empire.
108
AUGUSTIN
HOLL
In this paper, we so far have discussed several features from household
demography and socioeconomic
dynamics to settlement hierarchy and
probable intercluster
warfare. An additional factor, the environment,
must now be considered.
The Dhar Tichitt region was settled during the later Holocene c.a. 4500
to 2000 BP, when the Sahara margins witnessed increasing trends toward
desertification.
Thus the region may actually have been environmentally
circumscribed during the prehistoric occupation under discussion. According to Carneiro (1981&t), the circumscription
theory runs as follows:
As population density increases and arable land comes into short supply, fighting
over land ensues. Villages vanquished in war, having nowhere to flee, are forced
to remain in place and to be subjugated by the victors.
From this point of view, the decreasing quantity of water input from
both rain and groundwater table and consequent incomplete pedogenesis,
the depletion of game, aquatic resources, and fuel, difficulties in livestock
herding, which may have led to wider-range nomadism, increasing agricultural labor input, and diminishing food output may have produced the
development of new adaptative mechanisms. In this regard, the development of a symbolic mediator of stress in the form of rainmaking and
its correlated snake cult seems a reasonable possibility. The general distribution of these features in Africa is strongly correlated with the distribution of the climatic pattern of two contrasting seasons. In the specific
case of the formation of the Ghana empire, the ethnohistorical
record
clearly shows the links between the monarch, the snake cult, rainmaking,
and problems related to water availability
(Bathily 1975). In the Wagadu
legend, a deal was made between the monarch and a huge snake living
in a sacred well behind the palace. Each year, by the end of the dry
season, the most beautiful virgin had to be sacrificed to the snake; reciprocally the snake had to give sufficient rains full of gold nuggets, necessary conditions of population wealth. Maley (1981:528) considers this
legend to be a metaphorical
representation of the typical stormy rain
formation process, based upon the similarities of form between the hurricane-like whirl which precedes rains and the snake. The region being
devoid of any perennial river, the people’s wealth is exclusively dependent upon rains.
Keeping in mind the environmental
stress in the Dhar Tichitt, a southward shift of populations in search of moister areas seems a reasonable
possibility. This process is not based upon firmly established archaeological data, and we must avoid making a direct link between Dhar Tichitt
and the Ghana Empire. In the present state of research, there is a @IOto 800-year gap between the Dhar Tichitt archaeological record and the
BACKGROUND
TO THE GHANA
EMPIRE
109
earliest archaeological manifestations of medieval Ghana. Additional archaeological investigations in this area are needed to settle the question.
However, the scanty archaeological
data recovered from Tegdaoust
preurban levels and Koumbi-Saleh
earliest levels, dated, respectively,
around 600-700 A.D. (Devisse 1983:214) and 1400 rf: 160 B.P. (LY 1610)
(Sutton 1982), pertain to an agropastoral socioeconomic
system composed of cattle and ovicaprine herding (Bouchud 1983) and agriculture.
Finally, when all the data are assembled, it is possible to argue that the
Dhar Tichitt archaeological record reflects the development of rank, indicating the beginning of the development of a primary state. More likely,
it was a simple chiefdom, or simple ranked society, with some tendency
towards emergence of complex chiefdom organization.
“Simple chiefdoms are those in which social control activities are exercised by figures
drawn from an ascribed local elite subgroup; these chiefdoms characteristically have only one level of control hierarchy above the level of the
local community”
(Wright 1984:42). The diachronic connections with
Ghana state formation is still unknown, due to a lack of systematic archaeological survey in the area between Dhar Tichitt and Ghanean towns
like Koumbi-Saleh
and Tegdaoust. Future research will probably settle
this question.
ACKNOWLEDGMENTS
I thank Anick Coudart, Serge Cleuziou, Jean-Paul Demoule, Jerome Dubouloz, Professor Jean-Louis Huot, Michael J. Ilett, Thomas E. Levy, Catherine Per&, Henry Wright,
and two anonymous reviewers who read, commented on, and improved earlier versions of
this paper. Responsibility for its contents, however, rests entirely with me.
REFERENCES
CITED
Alexandre, Pierre
1982 African political systems revisited. Cahiers d’ktudes Africaines XXII, 87-88
(3-4):299-230.
Amblard, Sylvie
1981 La civilisation nkolithique du Dhar Tichitt- Walata li cravers son matkriel lithique. Doctoral thesis, University of Paris I, Paris.
Bateson, Gregory
1984 La nature et la pen&e. Seuil, Paris.
Bathily, Abdoulaye
1975 A discussion of the traditions of Wagadu with some reference to ancient Ghana.
Bulletin de I’IFAN, sCrie B 37: l-94.
1977 Intervention. In Acres du Collogue Histoire et Tradition Orale, pp. 114-115.
Fondation SCOA pour la recherche scientitique en Afrique noire, Paris.
Beyries, Sylvie
1981 Releve’s des gravures rupestres sur le Dhar Tichitt. Memoire de DEA. University of Paris X, Paris.
AUGUSTIN
110
HOLL
Bisson, Michael S.
1982 Trade and tribute: Archaeological evidence for the origin of states in South
Central Africa. Cahiers d’Etudes Afiicaines,
XXII, 87-88 (3-4):343-361.
Blanton, Richard E.
1976 Anthropological studies of cities. Annual Review of Anthropology
51249-264.
Blanton, Richard E., Jill Appel, Laura Finsten, Stephen Kowalewski, Gary Feinman, and
Eva Fisch
1979 Regional evolution in the valley of Oaxaca, Mexico. Journal of Field Archaeology 6:367-390.
Blanton, Richard E., Stephen A. Kowalewski, Gary Feinman, and Jill Appel
A comparison
of changes in three regions.
Cambridge
1981 Ancient Mesoamerica:
University Press, Cambridge/New York.
Bohannan, Paul, and Laura Bohannan
1968 Tiv Economy.
Longmans, London.
Borgstrom, Bengt Erick
1977 On rank and heirarchy: Status in India and elsewhere. Archives EuropPennes
de Sociologic
18:325-334.
Bouchud, Jean
1983 Paleofaune de Tegdaoust. In Tegduoust III: Recherches
sur Aoudaghost.
Catnpagnes
1960-1965,
pp. 355-363. Editions Recherches sur les Civilisations,
Pd.
Bouju, Jacky
1984 Graine de I’homme.
enfunt du mi/. Societe d’Ethnographie,
Paris.
Bovill, E. W.
1978 The go/den trade of the Moors. Oxford University Press, London.
Briant , Pierre
1982 Efaf etpasteurs
au Moyen-Orient
ancien. Cambridge University Press/Editions
de la Maison des Sciences de I’Homme, London/Paris.
Butzer, Karl W.
1981 Rise and fall of Axum, Ethiopia: A geo-archaeological interpretation. American
Antiquity
46471-495.
Cancian, Frank
1976 Social stratification. Annual Review of Anfhropology
5~227-248.
Cameiro, Robert L.
to sfnrehood
in the New
1981 The chiefdom: precursor of the state. In Transition
World, edited by G. D. Jones and R. R. Kautz, pp. 37-79. Cambridge University Press, Cambridge/New York.
Cisse, Youssouf
orale (Bamako),
pp.
1975 Intervention. In Acres du collogue Histoire et Tradition
10-18. Fondation SCOA pour la recherche scientifique en Afrique noire, Paris.
Claessen, Henri J. M.
1984 The internal dynamics of early states. Current Anthropology
25:365-379.
Coe, Michael D.
1981 Religion and the rise of the Mesoamerican state. In The rransition to statehood
in the New World, edited by G. D. Jones and R. R. Kautz, pp. 157-171. Cambridge University Press, Cambridge/New York.
Delneuf, Michele
1981 La cPramique
nColirhique
du village d’dkreijit
(Mauritanie).
Mtmoire de maltrise, Universite de Paris X, Paris.
Devisse, Jean
1977 Intervention. In Actes du collogue Histoire er Tradition
orale (Bamako),
pp.
89-91, Fondation SCOA pour la recherche scientifique en Afrique noire, Paris.
BACKGROUNDTOTHEGHANAEMPIRE
1983
111
Les fouilles de 1%2 a 1%5: Stratigraphie et architecture. In Tegdaoust Ill.
Recherches sur Aoudaghost. Campagnes 1960-1965, pp. 33-244. Editions Recherches sur les civilisations, Paris.
Doolittle, William E.
1984 Settlements and the development of “statelets” in Sonora, Mexico. Journal of
Field Archaeology 11:13-24.
Dumont, Louis
1966 Homo Hierarchicus: le systeme des castes et ses implications. Gallimard, Paris.
1977 Homo Aequalis: genese et epanouissement de l’ideologie economique. Gallimard, Paris.
1983 Essais sur l’individualisme:
une perspective anthropologique
sur l’ideologie
moderne. Editions du Seuil, Paris.
Dupre, Georges, and Philippe Rey
1980 Reflections on the relevance of a theory of the history of exchange. In Relations
of production,
edited by D. Seddon, pp. 171-208. E Cass, London.
Eisenstadt, Samuel N., Michel Abitbol, and Naomi Chazan
1983 Les origines de I’etat: une nouvelle approche. Annales: Economies Societes
Civilisations 6: 1232-1255.
Fortes, Meyer, and E. E. Evans-Pritchard
1940 Introduction. In African political systems, edited by M. Fortes and E. E. EvansPritchard, pp. l-23. Oxford University Press, London/New York.
Freidel, David A.
1981 Civilization as a state of mind: The cultural evolution of the Lowland Maya. In
The transition to statehood in the New World, edited by G. D. Jones and R. R.
Kautz, pp. 188-227. Cambridge University Press, Cambridge/New York.
Friedman, Jonathan, and Michael J. Rowlands
1977 Notes towards an epigenetic model of the evolution of civilization. In The evolution of sociaf systems, edited by J. Friedman and M. J. Rowlands, pp. 201276. Duckworth, London.
Gellner, Ernest
1977 Class before state: The Soviet treatment of African feudalism. Archives europeennes de sociologic 181299-322.
Godelier, Maurice
1977 Horizons, trajets marxistes en anthropologie (Vol. I). E Masptro, Paris.
1980 The concept of “Asiatic Mode of Production” and Marxist models of social
evolution. In Relations of production, edited by D. Seddon, pp. 209-257. F.
Cass, London.
Goody, Jack
1980 Technology, tradition and the state in Africa. Cambridge University Press,
Cambridge/New York.
Haaland, Randi
1980 Man’s role in the changing habitat of Mema during the Old Kingdom of Ghana.
Norwegian Archaeological Review 13:31-46.
Haas, Jonathan
1981 Class conflict and the state in the New World. In The transition to statehood
in the New World, edited by G. D. Jones and R. R. Kautz, pp. 80-102. Cambridge University Press, Cambridge/New York.
1982 The evolution of prehistoric state. Columbia University Press, New York.
Haggett, Peter
1973 L’analyse spatiale en geographic humaine. Armand Colin, Paris.
Harris, Marvin
1968 The rise of anthropological theary. Routledge & Kegan Paul, London.
AUGUSTIN
112
HOLL
1979 Cannibales
et monarques:
essai sur l’origine
des cultures.
Flammarion, Paris.
1984 Animal capture and Yanomamo warfare: Retrospect and new evidence. Journal
of Anthropological
Research
40: 183-201.
Hall, Augustin
1983 Essai sur l’economie
neolithique
du Dhar Tichitt (Mauritanie).
Doctoral thesis,
University of Paris 1, Paris.
1985a Subsistence patterns of the Dhar Tichitt Neolithic (Mauritania). Review of African Archaeology,
in press.
1985b Unite de production et unite de consommation dans le ntolithique du Dhar
Tichitt (Mauritanie). L’Homme,
in press.
Hopkins, A. G.
1980 An economic history of West Africa. Longman, London.
Hugot, Georges
1977
Un secteur quaternaire
lacustre manuritanien:
Tichitt. Mtmoires de I’IMRS I,
Chambery.
Hugot, Henri-Jean
1979
Le ne’olithique
saharien.
Doctorate d’Etat thesis, University of Paris X, Paris.
Iteanu, Andree
1983
La ronde des &changes:
de la circulation
aux valeurs chez les Orokaiva.
Cambridge University Press, Editions de la Maison des Sciences de I’Homme, Cambridge/Paris.
Johnson, Gregory A.
1976 Early state organization in Southwest Iran: Preliminary field report. In Proceedings of the IVth Annrtal Symposium on Archaeological Research in Iran,
edited by E Bagherzadeh, pp. 190-223. Tehran.
1977 Aspects of regional analysis in archaeology. Annual Review of Anthropology
6:479-508.
1980 Rank-size convexity and system integration: A view from archaeology. Economic
Geography
56234-247.
1981
Monitoring complex system integration and boundary phenomena with settlement size data. In Archaeological
approaches
to the study of complexity,
edited
by S. E. van der Leeuw, pp. 143-188. Amsterdam.
1982 Organisational structure and scalar stress. In Theory and explanation
in archaeology,
edited by C. Renfrew, M. J. Rowlands and B. A. Seegraves, pp.
389-421. Academic Press, London/New York.
Jones, Grant D., and Robert R. Kautz
1981 Issues in the study of New World state formation. In The transition to statehood
in the New World, edited by G. D. Jones and R. R. Kautz, pp. 3-34. Cambridge
University Press, Cambridge/New York.
Keatinge, Richard W.
1981 The nature and role of religious diffusion in the early stages of state formation:
An example from Peruvian prehistory. In The transition to statehood in rhe New
World, edited by G. D. Jones and R. R. Kautz, pp. 172-187. Cambridge University Press, Cambridge/New York.
Kowalewski, Stephen A., Richard E. Blanton, Gary Feinman, and Laura Finsten
1983 Boundaries, Scale, and Internal Organization. Journal of Anthropological
Archaeology
2:32-56.
Lericollais, Andre
1970 La deterioration
39:113-128.
McIntosh, Susan K.
d’un terroir Sob, en pays Serer (Senegal). Etudes Rurales
37-
BACKGROUND
1981
TO THE GHANA
EMPIRE
113
A reconsideration of Wangara/Palolus, island of gold. Journal of African History
22:145-158.
McIntosh, Susan K., and Roderick J. McIntosh
1980 Prehistoric investigation uf Jenne, Mali. British Archaeological Reports, Oxford.
1983 Djenne-Jeno, cite saris citadelle. Lu Reclterche 148: 1272- 1275.
Maley, Jean
1981 Etudes palynologiques duns le bassin du T&ad et pal~oclamatofogie actuetle,
ORSTOM, Paris.
Mauny, Raymond
1956 Perles ouest-africaines en amazonite. Bulkfin IFAN, sCrie B 18 (I -2): 140- 147.
1961 Tableau gPograhique de I’ouesl ufricain au Moyen age d’oprks Ies sources
&rites, lo tradition et I’orche’ologie. Memoires de I’IFAN, 61, Dakar.
Meillassoux, Claude
1980a “The economy” in agricultural self-sustaining societies: A preliminary analysis.
In Relations of Production. edited by D. Seddon, pp. 127- 157. E Cass, London.
1980b The social organization of the peasantry: The economic basis of kinship. In
Relations of production, edited by D. Seddon, pp. 159- 169. F. Cass, London.
Munson, Patrick J.
1971 Tichitt tradition: A late prehistoric occupation in Southwestern Africa. Ph.D
thesis, University of Illinois, Urbana. Champaign, Illinois.
1980 Archaeology and the prehistoric origins of the Ghana Empire. Journal of Africon History 21~457-466.
1981 A late holocene (c. 4500-2300 B.P.) climate chronology for the Southwestern
Sahara. Paleoecology of Africa 13:53-60.
Phillipson, D. W.
1979 Migration, ethnic differentiation and state formation in the Iron Age of Bantu
Africa. In Space, hierarchy and society. edited by B. C. Bumham and J. Kingsbury, pp. 205-214. British Archaeological Reports, Oxford.
Polet , Jean
1980 Fouille d’un quartier de Tegdaoust (Mauritanie orientale): urbanisotion, orchifecrure, utilisation de I’espace construit. Doctoral thesis, University of Paris I,
Paris.
Pallet, Eric, and Grace Winter
1980 The social organization of agricultural labour among the Soninke (Diahunu,
Mali). In Relations of production, edited by D. Seddon, pp. 331-356. E Cass,
London.
Posnansky, Merrick
1982 Some reflections of comparitive nature on first milienium A.D. complex societies in west Africa. Paper presented at the 25th annual meeting of the African
Studies Association on “Complex societies of west Africa, their archaeology
and early history,” Washington, D.C.
Quilter, Jeffrey, and Terry Stocker
1983 Subsistence economies and the origins of Andean complex societies. American
Anthropologist 85:545-562.
Rapport Preliminaire
1980 Rapport Preliminaire de la Mission de recherche prehistorique de Tichitt 1980,
Prospection du Dhar Tichitt-Aratane. Paris, unpublished.
Robert-Chaleix, Denise
1980 Une concession “mt!diCvale” d Tegdaoust: implantation, t%olution d’une unit6
d’hubitation. Doctoral thesis, University of Paris I, Paris.
114
AUGUSTIN
1983
Cdamiques
Aoudaghost.
decouvertes
Campagnes
HOLL
B Tegdaoust. In Tegdaoust
III. Recherches
sur
pp. 245-325. Editions Recherches sur les
1960-1965,
civilisations, Paris.
Robert-Chaleix, D., and Cl. Richir
1983 Un site jusqu’a present inconnu dans le Rkiz: Moulay Arbad. In Tegdaoust III.
Recherches
sur Aoudaghost.
Campagnes
1960-1965,
pp. 335-350. Editions
Recherches sur les civilisations, Paris.
Sahlins, Marshall D.
1961 The segmentary lineage: an organization of predatory expansion. American An1968
1969
thropologist
Tribesmen.
63~322-345.
Prentice Hall, New Jersey.
Land use and the extended family in Moala, Fiji. In Environment
and cultural
behavior,
edited by A. P. Vayda, pp. 395-415. The Natural History Press, New
York.
1976
Age de Pierre, age d’abondance:
l’economie
des societes primitives.
Gallimard.
Paris.
1980 Au coeur des sociPtPs: raison utilitaire
et raison culturelle.
Gallimard, Paris.
Saison, Bernard
1979 Fouille d’un quartier
artisanal
de Tegdaoust
(Mauritanie
orientale).
Doctoral
thesis, University of Paris I, Paris.
Schneider, Harold K.
1979
Livestock
and equality
in East Africa,
the economic
basis for
social
strucfure.
Indiana University Press, Bloomington/London.
Southall, Aidan W.
1970 Alur society: A study in processes and types of domination.
Oxford University
Press, London/New York.
Steponaitis, Vincas P.
1978 Location theory and complex chiefdoms: A Mississipian example. In Mississipian Settlement
Patterns,
edited by B. D. Smith, pp. 417-453. Academic
Press, London/New York.
1981 Settlement hierarchies and political complexity in nonmarket societies: The
formative period of the valley of Mexico. American Anthropologist
83:320-363.
Steward, Julian H.
1955 Theory of culture change: The methodology
of multilinear
evolution.
University
of Illinois Press, Urbana/Chicago/London.
Sutton, J. E. G.
1982 Archaeology in West Africa: A review of recent work and a further list of
radiocarbon dates. Journal of African History 23:291-313.
Sylla, Diarra
1977 Intervention. In Actes du colloque Histoire
et Tradition
orale (Bamako),
pp.
85-86. Fondation SCOA, Paris.
Thilmans, G., and A. Ravise
1980 Protohistoire
du Sene!gal: Sintiou-Barn
et les sites dufleuve.
Tome II. Mtmoire
de I’IFAN 91, Dakar.
Thomassey, P., and R. Mauny
1956 Campagne de fouilles de 1950 a Koumbi Saleh. Bulletin IFAN, s&ie B 18(12):117-140.
Thornton, John K.
1982 The Kingdom of Kongo, ca. 1390- 1678: The development of an African social
formation. Cahiers d’e’tudes africaines XXII, 87-88, (3-4):325-342.
Tourney, C. P.
1981 Pattern and scale in archaeological regions. In Networks
of the past: Regional
BACKGROUND
TO THE
GHANA
EMPIRE
115
interaction in archaeology, edited by P. D. Francis, F. J. Kense, and P. G. Duke,
pp. 467-476. University of Calgary, Calgary.
Trimingham, John S.
1980 The influence oflslam upon Africa. Longman, London/Beirut.
Vanacker, Claudette
1979 Fouille d’un quartier de Tegdaoust. Mtmoires de I’IMRS 2, Paris.
Vita-Finzi, Claudia
1978 Archaeological sites in their setting. Thames & Hudson, London.
Vita-Finzi, Claudio, and Eric S. Higgs
1970 Prehistoric economy in the Mount Carmel area of Palestine: Site catchment
analysis. Proceedings of the prehistoric society 36: l-37.
Wa Kamissoko
1975 Interventions. In Acres du colloque ifistoire et Tradition orale (Bamako), pp.
29-38, 65-67. Fondation SCOA, Paris.
Walsh, Carl E.
1983 Saving in primitive economies. American Anthropologist 85543449.
Warnier, Jean-Pierre
1983 Sociologic du Bamenda Prkoloniai (Cameroun). Doctorate d’Etat thesis, University of Paris X, Paris.
Wickins, Peter
1981 An economic history of Africa from the earliest times to partition. Oxford University Press, Capetown.
Wilk, Richard R., and William L. Rathje
1982 Household archaeology. American Behnvioral Scientist 25:617-639.
Wright, Henry T.
1977 Recent research on the origin of state. Annual Review of Anthropology 6:379397.
1984 Prestate political formations. In On the evolution of complex societies: Essays
in honor of Harry Hoijer, 1982, edited by Timothy Earle, pp. 41-47. Undena
publications, Malibu.
Wright, Henry T., and Gregory A. Johnson
1975 Population, exchange and early state formation in Southwestern Iran. American
Anthropologist 77:267-289.
Augustin Ferdinand Charles Holl