Current Anthropology Volume 39, Number 1, February 1998

 1998 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/98/3801-0003$2.50

In many ways the buildings of Great Zimbabwe, its sub-

sidiary site at Khami, and their hundreds of satellites
Cognitive stand close to frontiers—of space, time, and discipline.
Although Great Zimbabwe is located near the heart of
Archaeology and the modern state named after it, there are examples of
the distinctive stone architecture pioneered there in
Mozambique, South Africa, and Botswana. Similarly,
Imaginary History at although the site is well within the large zone occupied
by the speakers of the Shona language and for decades
Great Zimbabwe 1 no competent scholar has doubted that it was their
work, many of the satellites mentioned above are in ar-
eas where the local language is now Ndebele, Sena,
Tsonga, or Venda. Temporally, though Great Zimbab-
by David Beach we’s floruit just predated the arrival of the Portuguese
on the Sofala coast and thus the beginning of the re-
corded history of the interior, there was a period when
the frontiers of archaeology and history overlapped. Al-
though the overlap is frustratingly unhelpful, it cannot
The archaeology of Great Zimbabwe in southern Africa is well be denied that Great Zimbabwe and its associated sites
developed, but supporting evidence from other disciplines is offer an outstanding opportunity for cross-disciplinary
scanty and often suspect. Consequently, the interpretation of
studies.
many features of the city that flourished ca. 1200–1500 is diffi-
cult. Thomas Huffman’s well-known reconstruction of the site’s Thomas Huffman, an archaeologist, has used pub-
cognitive archaeology relies upon misunderstood documents, du- lished Portuguese documents, Shona ‘‘oral traditions,’’
bious oral traditions, and inappropriate comparisons to arrive at and Venda anthropology, among other sources, to ex-
a picture of a city that was essentially static in its use of space plain much that is mysterious about Great Zimbabwe.
for at least two centuries. An alternative interpretation is offered
that draws upon Shona historical and anthropological sources This is not in itself a new approach to the problem. Ever
and, explicitly, imagination to show how political processes since Karl Mauch asked local Shona leading questions
within a Shona dynasty could have affected the building of the about the site in 1872 (Mauch 1871–72:190), profes-
central part of the city. This model does not pretend to explain sional and amateur researchers, whether willing or not
every feature of the site, nor does it claim to be more than one
to see the Shona as the creators of Great Zimbabwe,
of several possible models of its kind. Nevertheless, it corre-
sponds far more closely to the workings of Shona society, as it is have consulted Portuguese documents and what they
known to us from history, than Huffman’s. thought were oral traditions or speculated about con-
nections between features of the ruins and modern
david beach is a professor in the Department of History at the Shona material culture. Indeed, some of the speculators
University of Zimbabwe (Box MP 167, Harare, Zimbabwe). Born have been Shona themselves: it was a Head Native Mes-
in 1943, he was educated at the University of Cape Town (B.A., senger in 1929 who suggested that the conical tower re-
1965) and the University of London/University College of Rhode-
sia (Ph.D., 1971). His research interests cover the history of Zim- sembled a grain bin, and the fact that he was joking did
babwe and central Mozambique ca. 1500–1900. Among his publi- not stop later archaeologists from taking the idea seri-
cations are The Shona and Zimbabwe 900–1850 (London: ously (Bullock 1949:51). Nevertheless, up to the end of
Heinemann, 1980), The Shona and Their Neighbours (Oxford: the 1970s most archaeologists devoted most of their re-
Blackwell, 1994), and A Zimbabwean Past (Gweru: Mambo, search on Great Zimbabwe to conventional archaeol-
1994). The present paper was accepted 12 x 96.
ogy: excavation, dating, classification of imported goods
and local ceramics, the establishment of sequences of
wall construction, deduction of economic systems
through environmental analyses, and a host of other
pursuits.
Huffman’s work on Great Zimbabwe in the 1980s did
not abandon conventional archaeology entirely—he did
much to refine the radiocarbon dating of the site, for ex-
ample (Huffman and Vogel 1991)—but it did come to
rely heavily on non-archaeological sources such as doc-
uments and oral traditions. To a certain extent this was
unavoidable, as one more aspect of the ‘‘frontier’’ nature
of Great Zimbabwe was that, after 1976, Huffman was
based in South Africa. This meant that it would not
have been easy for him to undertake conventional ar-
chaeological research in an independent Zimbabwe in
1. Thanks are owed to David Collett, Rod McIntosh, Innocent Piki-
any case. However, his increasing reliance upon histori-
rayi, Gilbert Pwiti, Robert Soper, and Jan Vansina for their com- cal sources unfamiliar to all but a few specialists re-
ments on and criticisms of earlier drafts of this paper. quires a very brief comment upon them. Although the
47
48 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

documents and oral traditions of the Zimbabwean and tants remained in control of the land where they had
central Mozambican region are jointly superior to those been prior to these conquests in groups large enough to
of any other part of the Indian Ocean hinterland south have retained a corpus of oral tradition.
of Ethiopia for the 1500–1900 period, when it comes to This latter point requires elaboration, as oral tradi-
stone buildings a surprising fact emerges: there was and tions are frequently invoked below. In 1983 I criticized
is a distinct shortage of non-archaeological evidence Huffman’s first major article on cognitive archaeology
that the stone zimbabwe (capital sites) were the work at Great Zimbabwe as part of a more general caveat on
of the Shona. the value of oral tradition for archaeologists in Zim-
Of the thousands of Portuguese documents covering babwe (Beach 1983). Its main thrust is summed up in
the period 1506 to 1890, very few refer to stone zim- the question ‘‘What makes me think that this particular
babwe. António Fernandes’s verbal report of 1512 re- informant would know anything about this particular
ferred to a stone capital being built in the Mutapa state site?’’ In 1985 Jan Vansina’s Oral Tradition as History
without specifying whether its drystone walls were in provided a worldwide theoretical framework on oral tra-
the Great Zimbabwe style or in one of several rougher dition that rendered nearly all of its predecessors obso-
techniques still used in the north up to the 1890s (Ve- lete. If it is possible to sum up its main thrust in a few
loso 1512:182; Dickinson 1971:46–47; Leijer 1897). words, those words must include the chapter heading
Muslim traders told Diogo de Alcáçova in 1505–6 that ‘‘The Message Is a Social Product’’ (Vansina 1985:94). In
the Mutapa was living in a stone and clay capital, and 1991 I made limited use of Vansina’s framework in a
others told Vicente Pegado in 1530–38 that Great Zim- preliminary classification of Shona oral traditions
babwe was occupied by the Shona. In 1831 João Julião (Beach 1994b:244–74). Briefly, I suggested that they fall
da Silva was informed that the Changamire was occu- into six broad categories—environmental, economic,
pying Danangombe in the south-west (Alcáçova 1506: religious, family, political, and artistic. Each contains
394; Barros 1945 [1552]:393–94: Silva 1831). Diogo varying elements of myth and reality, and each has a
Simões Madeira reported that Shona considered the different time depth, ranging from general myths of the
Bedza ruins in the Zambezi Valley ‘‘a supreme piece of Bantu-speaking people to events dating from just before
work’’ (Bocarro 1899 [ca. 1631]:266–67; Beach 1987: the childhood of informants. Given a community set-
133). However, only the first of these accounts defi- tled in one place and with a reason for recording and
nitely stated that the Shona actually built, as opposed transmitting data, these time depths reflect the varying
simply to occupying, stone structures. Otherwise, in needs of society.
those parts of the northern and eastern Zimbabwean Thus, not far from Great Zimbabwe, economic oral
plateau where Portuguese observers penetrated, the traditions recalled trade routes to the coastal settle-
construction of zimbabwe in stone in the style of Zvon- ments that dated back to the mid-19th century but not
gombe, Tsindi, Tere, and Chipadze had evidently ceased earlier than the 1830s. This was presumably because
before they arrived. The only Portuguese expedition memories of earlier trade served no useful purpose. On
into the south-west, where stone building went on until the other hand, political oral traditions about migra-
at least the 17th century, did not produce any first-hand tions, conquests, and political struggles over the control
or detailed report. Khami and Danangombe were not of land dated back to the first half of the 18th century.
mentioned (Beach 1980:200–201). This was because such matters were and still are impor-
Oral traditions about the construction by the Shona tant to the people. Data from prior to ca. 1700 were very
of some late stone zimbabwe did survive into the early rare and fragmentary and usually had to be linked to
20th century. The Nambiya consistently claimed to Portuguese documents as pegs to stretch the fabric of
have built the structures of the far northwest, while the tradition. But all this presupposes a continuing commu-
Hera claimed to have built the Gombe zimbabwe but nity with a social purpose in preserving traditional data.
not those nearby (Beach 1980:75, 259). By the 1890s and The failure of oral tradition to supply evidence on the
1900s the Changamire Rozvi had rather confused and origin and meaning of Great Zimbabwe in a convincing
contradictory traditions about the south-western build- fashion was hardly surprising, as it was occupied only
ings that they had occupied up to the 1830s. They up to the first half of the 16th century. Even if the in-
claimed in some traditions to have built them (Beach habitants of the area from then until the conquest by
1980:234; Elliott 1894:110–11; Jackson 1906), but in the 18th-century immigrants had retained traditions on
other traditions they stated that the buildings were al- these subjects, there was no special reason the tradi-
ready there when they arrived from the north-eastern tions of the later groups should have incorporated and
plateau in the late 17th century (Thomas 1906, Lanning retained this information.
1906, Stuart 1906, Posselt 1904), and one version at Indeed, whereas the incoming groups such as the
least was ambiguous (Elliott 1906). However, in the cru- Duma, Nini, and southern Hera recalled the names of
cial area of Great Zimbabwe there were no clear oral their predecessors around 1700, it very often proved im-
traditions concerning the site. This was partly because possible for modern researchers to locate representa-
in the 18th century the entire southern plateau region tives of these early inhabitants (Mtetwa 1976:18–22).
was conquered by Shona groups from the north and By 1872 neither the Duma nor their Manwa predeces-
west, and virtually none of the earlier Shona inhabi- sors at the Great Zimbabwe site (also immigrants) had
 b e a c h Cognitive Archaeology and Imaginary History 49

any clear idea who built Great Zimbabwe, and migrant was vital, because it led to the discussion of the site as
labourers returning from Kimberley were beginning to an almost complete socioeconomic unit inhabited by an
spread the idea that, if Europeans built large houses at entire community (Huffman 1977:9–14). Subsequent
Kimberley, then perhaps Great Zimbabwe was of Euro- work has confirmed the importance of this approach
pean origin (Helm 1891). Moreover, oral traditions were (Sinclair 1987).
subject to change and to contamination by other It remains to be seen whether Huffman’s move from
sources. The Rozvi, for example, ‘‘rewrote’’ their oral spatial and site analysis in the 1970s to cognitive ar-
traditions in this century, often borrowing from the chaeology in the 1980s represents a third major develop-
work of archaeologists. Mutapa oral traditions about ment in research on Great Zimbabwe. According to
the formation of that state in the north also changed Huffman (1986:84), ‘‘Cognitive archaeology is the study
considerably over three centuries and incorporated de- of prehistoric ideology, that is to say the ideals, values
tails about some of the northern zimbabwe only to rein- and beliefs that constitute a society’s world-view. Cog-
force local political messages. In this century, as more nitive archaeologists use the principles of sociocultural
came to be generally known about Great Zimbabwe, anthropology to investigate such diverse things as ma-
Shona-speakers began to add it to their traditions from terial symbols, the use of space, political power and reli-
the 1940s, and those who lived near the site were just gion.’’ Further, ‘‘for analytical purposes the physical
as capable of speculation about the meaning of its archi- signposts and underlying principles of this spatial code
tecture as European visitors and archaeologists (Beach can be said to constitute a society’s ‘expressive space’ ’’
1994b:188–274; 1983:8–11). In short, even when they (Huffman 1984b:593). Concentrating mainly on the
exist, oral traditions about Great Zimbabwe and most central core of the site, where the majority of the stone
other ruins of its type have to be regarded with caution walls survive, Huffman combined personal observation
unless it can be shown how they could have survived with the published data, paying particular attention to
over a span of some six or seven centuries. the location and decoration of the walls and to the arte-
Turning to other non-archaeological aspects, there facts discovered within or near them. He then devel-
are some paradoxes that affect the discussion: Great oped a classification of them, frequently using paired
Zimbabwe was Shona, but it was not typical of Shona oppositions in structuralist style such as up : down or
settlements in historical times, and it was not even typ- left : right, and sought explanations for the patterns that
ical of its own society as is natural in a capital. (Indeed, emerged in this classification. Some of these came from
in many ways Great Zimbabwe can be regarded as an ‘‘oral traditions’’ collected at or near the site, some from
experiment rather than as a direct extension of earlier Portuguese documents, and others from elements of the
practices.) Whereas its pottery and house construction published anthropology of the Shona and also the Venda
fell well within the normal range of Shona material cul- of South Africa.
ture, other aspects were not part of historical Shona Whereas I consider such studies of vital importance,
practice. Clay or soapstone figurines, clay or soapstone I shall argue here that Huffman’s use of evidence is fre-
birds, conical towers, and open-air stepped platforms quently faulty, either because of the misreading of doc-
were not to be found in the 19th century, nor do the uments or because the oral traditions used cannot be
Portuguese documents report them. We have to recog- shown to date back to the period of Great Zimbabwe or
nise that it was not just the economic capacity of the because of the use of inappropriate elements of Venda
Shona to build such a settlement that changed after the culture. I shall go on to examine an aspect of precolonial
16th century but some of the thinking of the people as Shona society that Huffman ignores—political process.
well. Although it is extremely difficult to deduce from the ar-
chaeology of preliterate societies, it should not be ex-
cluded from our thinking about Great Zimbabwe on
those grounds. Almost the first thing that we know
The Cognitive Archaeology about Shona society from historical sources is an exam-
of Great Zimbabwe ple: the civil wars in the Mutapa state from 1490 to
1506 mentioned in the Alcáçova document cited above
As has been pointed out before, Great Zimbabwe re- would have left little if any archaeological traces but
mains a primarily archaeological problem (Garlake were clearly of great importance. I shall suggest that it
1973:76). Two crucial developments have set the scene is possible to use historical models of political process
for later work. One was, of course, the work of David to interpret Great Zimbabwe’s archaeology in a manner
Randall-MacIver in 1905. This established the date and different from that of Huffman. While these cannot be
African origin of the stone structures, and the subse- proven, they deserve to be borne in mind.
quent work of Gertrude Caton-Thompson, Roger Sum- Huffman’s cognitive archaeological work on Great
mers, Keith Robinson and Anthony Whitty, and others Zimbabwe depends upon two main articles on Great
has essentially been refining MacIver’s work. The sec- Zimbabwe itself, with subsidiary work relating to ear-
ond major advance was made by Huffman in the early lier sites such as Mapungubwe and to contemporary and
1970s. His work on the ordinary dwellings of the town later sites such as Khami. There is, naturally, a ten-
of Great Zimbabwe outside the prestige stone walling dency for the subsidiary articles to rely upon the argu-
50 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

ments of the two main ones as proven, which is hardly Huffman then proceeds to the valley ruin area: at the
surprising in view of the relative lack of detailed com- base of the hill was an open space that was probably a
ment upon them in the last decade. I therefore begin meeting place or dare. Bocarro is then cited to show
with a reexamination of these articles, commenting on that the Mutapa had such a dare. Shona rulers had
the validity of the evidence supplied (Huffman 1981, many wives. Oral traditions say that the wives at Great
1984a, b, 1986, 1987). Zimbabwe lived below the hill. The oldest walling at
the foot of the hill is in the Renders/No. 1/Mauch/
Plateau ruin area. The Renders ruin, nearest to the dare,
Snakes and Birds: Expressive Space contained a royal treasure, so the chief wife lived there.
at Great Zimbabwe The Plateau ruins probably contained grain bins, so the
central area had elite men on the hill and elite women
Huffman (1981) begins with the logical point that Great at its foot (1981:134–36).
Zimbabwe consists of a central area of stone buildings Huffman misreads Bocarro—or, rather, Bocarro’s
within two perimeter walls and an outer area (fig. 1). He source Madeira—and should have consulted the Portu-
cites four post-1870 oral traditions to the effect that the guese text.3 This shows that the enclosure described
ruler of Great Zimbabwe lived on the hill or ‘‘Acropo- had only three houses occupied by the ruler and his
lis’’ and that this was the part of the site known as zim- queen and his servants, and only these were allowed
babwe itself (fig. 2). From 1506, Shona capitals and within. Thus it was not a dare, though obviously the
burial sites were called zimbabwe. Barbosa in 1518 Mutapa had one elsewhere. Oral traditions about the lo-
wrote that the Mutapa lived in ‘‘a very high house,’’ cation of wives at Great Zimbabwe are no more reliable
thus on a hill. Important Shona rulers lived on hills, and than any others about the site, but the chief problem
rulers are associated with high ground. A fancy ‘‘royal’’ with linking the valley ruins with the wives of the ruler
spearhead-like part of the Mutapa regalia was found in on the hill is that Madeira/Bocarro states only that they
the Western Enclosure on the hill, and so were three all lived close enough together for visits for sexual pur-
Central African–type royal gongs. There is no tradition poses to be possible, which might or might not fit the
of a female ruler at Great Zimbabwe, so the hill was a distance of 650 m up and down a steep hill between the
centre of male political power (pp. 131–33). Western Enclosure and the valley ruins. Indeed, if re-
The foregoing relies upon oral traditions and docu- cent opinion—cited by Huffman—is of any value, then
ments of varying and often doubtful relevance or inter- the chief wife should be ‘‘next’’ to the ruler and his
pretation. As shown above, it remains to be proven that other wives’ houses farther away. Neither the treasure
the oral traditions are a genuine transmission of evi- nor the grain bins prove much, because they are the
dence from before 1550 and not guesswork on the part only surviving examples from a city that lasted for cen-
of informants. This fundamental point relates to the lo- turies. In any case, as will be shown later, the assump-
cation of the ruler on the hill, its specific identification tion that male and female power can be separated spa-
as the zimbabwe itself, the general association of rulers tially over such a large area is a dangerous one to make
with high ground, and the absence of any oral tradition to the exclusion of others.
about a female ruler at Great Zimbabwe. It is true that, Huffman’s next argument concerns orientation. All
in the recent past, Shona rulers usually took hills as be- Shona allocate the right side to men and the left side
ing the best defensive sites, but this was not necessarily
so before 1550. Traditions of the Changamire Rozvi
(Theal 1898:85–86 and Dames 1918:12). It appears to be an addition
state of the 1680s–1830s period, cited by Huffman, do by someone after Barbosa’s death. Barbosa’s source has to be Antó-
refer to hills, but these relate either to their capture of nio Fernandes, as no other Portuguese is said to have reached the
the stone platforms of the Khami culture in ca. 1683– Mutapa by 1517. The ‘‘very high house’’ addition could have come
96 or to a myth that explains the Mfecane invasions of from João de Barros, who says that some Mutapa houses included
poles as high as a ship’s mast (Barros 1945 [1552]:396).
the 1830s (Beach 1980:202, 234; 1994b:249). It is hardly 3. ‘‘The lodgings in which Manamotapa lives are very large and of
surprising that there is no tradition about a female ruler many houses, surrounded in a circle by a great enclosure of wood.
at Great Zimbabwe, as female Shona rulers were very They have within three lodgings, one for his person, another for
rare and confined to the north and east of the region, far the queen, another for his servants who serve him from the gates
from the site. Southern Shona informants would not be to inside; and in this way he has three gates in a great yard, one
through which the queen is served, through which no man may
likely to have heard of them. The Barbosa reference enter, except women; another for his kitchen, through which no
could equally mean a tall house rather than one on a one other than his cooks may go, who are two young principal lords
hill, but in any case the reference is almost certainly an of his kingdom, and his relatives, in whom he has most trust, and
invention by a later copyist.2 the boys who serve in the kitchen, who also are noble, of fifteen
to twenty years; [the food is then described]. The third gate serves
2. There are five versions of this part of the work of Duarte Barbosa, for the lodgings of the king, through which only the noble boys
who returned to Europe in 1517–18 and completed it before he who serve from the gates to inside may enter, who are all of fifteen
went on the Magellan expedition of 1519, during which he died. to twenty years’’ (Bocarro 1899 [ca. 1631]:267). This is quite clearly
Four of these do not contain the phrase ‘‘hua casa muyto alta not a general meeting place for the whole capital, while it is ex-
aonde ho Rei sempre esta’’ (Barbosa 1966 [1518]:360 [Portuguese]; plicit that the ruler and his chief wife live in the same enclosure,
374–76 [French]; Stanley 1886:7 [English from Spanish]; Ramusio in different houses, and that cooking was done outside. The first
1563: 288v. [Italian]). The only version that does is the text pub- sentence might refer to an outer perimeter stockade, with an inner
lished in Portuguese in 1813 and in English in 1898 and 1918 stockade for the household of the ruler.
 b e a c h Cognitive Archaeology and Imaginary History 51

F ig. 1. Great Zimbabwe: Central area.

to women, inside houses as one looks in, and designate ambages to the little, plain conical tower to the left of
separate left-hand and right-hand entrances for women the big one as one looks east. Dentelle decoration on
and men as one looks towards a rain-making enclosure. the narira or conical tower is matched by dentelle deco-
A ruler, as he faces west out of his house, however, ration on the hill or narira. The large tower represents
should have his wives on his left, so there are two con- the ruler. The right-hand ascent of the hill has a com-
tradictory orientations working at different levels. In plex gate at the bottom and dentelle decoration part
the Great Enclosure, then, the right-hand entrance way up, leading to monoliths. The left-hand ascent has
(from outside, on the west) has large and small ambages a much simpler gate decorated with a herringbone pat-
to right and left and leads directly to monoliths stuck tern leading to a slotted entrance. So there is a left :fe-
in the ground, then right to the tall, dentelle-decorated male :: right :male orientation throughout, backed by
conical tower. The left-hand entrance (from outside, on monoliths and opposed types of decoration on the walls
the north) leads left to the parallel passage past slotted (1981:136–38). The east and sunrise mean life and the
52 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

F ig. 2. Huffman’s cognitive interpretation, ca. 1250–ca. 1500.

west and sunset death. The living sleep facing east, and north side warn us that not everything at Great Zim-
the dead are buried facing west. The east of a village or babwe was necessarily due to ideology. Moreover, a
house is sacred (the chikuva pot-stack in a house is to ‘‘grand origin myth’’ recorded by Frobenius and cited by
the east) in modern Shona practice. The entire orienta- Huffman to reinforce his argument is not typical of the
tion of the site has entrances at the west, with no spe- Shona and may be a borrowing from Frobenius’s earlier
cial eastern boundary. The main enclosure on the hill work on the Kuba that was suggested to paid Shona in-
is at the west, facing west (pp. 138–40). formants (Beach 1983:11;1994b:258).
That doors may have been oriented to the west be- According to Huffman, specific parts of the hilltop
cause the prevailing winds are from the east—the orien- buildings can be assigned functions, royal and religious.
tation only later acquiring ritual sanction—and that the A house in the Western Enclosure had a ‘‘snake’’ deco-
only part of the hilltop other than the west that could ration, and snakes in Shona symbolism relate to the
have been chosen for a major enclosure was the exposed spirit world of the royal ancestors (1981:140). This con-
 b e a c h Cognitive Archaeology and Imaginary History 53

firms that the ruler lived here. The Eastern Enclosure readers for the argument to be made in his next article.
had no houses, an underground passage, and six soap- There is some inconsistency: at the Great Enclosure,
stone birds. Below it was a small cave. Birds are linked snakelike decoration in stonework means men and fer-
in tradition to Mwari, ancestral spirits, and to the tility, but earlier, in the Western Enclosure, snakelike
Rozvi. The birds may symbolise rulers, seven (counting decoration in clay had meant royal ancestors. In ex-
the one from the Western Enclosure but not that from plaining such features as an enclosure in the valley area,
the Phillips Ruin because, in Huffman’s view, that is Huffman is still accepting ‘‘oral tradition’’ over some
already established as ‘‘female territory’’) for the whole 600 years as reliable. But much of the argument rests
state over two centuries or so. Spirit mediums were on the assumption that his original division of the site
probably living in the smaller enclosures between the into male and female areas was right in the first place.
Western and Eastern Enclosures (1981:140–42). If it was not, little of it would be applicable.
The idea that the birds were mnemonic devices or Huffman goes on to make the valid point that, if there
commemorated rulers is an old one (Summers 1963:73), is a meaningful pattern at Great Zimbabwe, it should
but the fact remains that the Shona have not carved recur at Khami, and he goes on to show to his own satis-
anything to record rulers since the Portuguese came. If faction that it does (1981:145–48). At this point, how-
they had done so or made carvings of their totem ani- ever, I should stress that I am not trying to dispute that
mals, this theory would have been more tenable. These there was a pattern or patterns at Great Zimbabwe.
birds remain too atypical to support speculation of any What I am querying is the meaning attributed to it or
kind. The association of the Eastern Enclosure with them. Consequently, we now move on to his second
spirit mediums rests on a demonstrably unreliable oral seminal article on the subject.
tradition that has been uncritically repeated far too of-
ten (Beach 1994b:237; 1983:11). It could be argued from
the earliest evidence on a Shona spirit medium’s perfor- ‘‘Where You Are the Girls Gather to Play’’:
mance in the 1580s that the mediums of Great Zim- The Great Enclosure at Great Zimbabwe
babwe probably lived in the town as more or less nor-
mal persons for much of the time, accompanying the Huffman’s second article (1984a) begins by reiterating
ruler to a royal burial hill many kilometres away for rit- points made in ‘‘Snakes and Birds,’’ showing that while
ual possession (Santos 1899 [1609]:15–17). All that we the hill was ‘‘male’’ and the cluster of buildings be-
know for certain is that the Eastern Enclosure was no tween the dare and the Great Enclosure ‘‘female’’ the
one’s living area and that it was not considered neces- Great Enclosure itself contained both male and female
sary to remove the birds when the ruling dynasty left features and was accessible from both parts of the elite
the hill. As for the Western Enclosure, one decoration area. The monoliths inside the western ‘‘male’’ en-
on one house out of many such over two centuries of trance to the Great Enclosure were like those called the
occupation is very little evidence for any kind of link, ‘‘horns of the ruler’’ today on the hill. From the north-
even if it very probably was a ruler’s area at times. ern or ‘‘female’’ entrance the route along the parallel
Huffman then takes his analysis of symbolic features passage passed a stone platform that still carried cattle
into the valley; the Great Enclosure also has a sacred bones in the 1900s and grooved ambages that repre-
eastern end and chevron decoration of the outside wall sented ‘‘female’’ status. The daga structure in front of
that resembles a snake’s movement and thus, in Shona the conical tower supported many figurines. The Bemba
culture, men and fertility (1981:142–43). Apart from the and the Pedi use such figurines in female initiation
question of entrances, ambages, passages, and conical schools. Thus the Great Enclosure was a girls’ initiation
towers discussed above, there were male and female school (1984a:252–55). Huffman claims that ‘‘the most
figurines in the parallel passage and on a daga struc- appropriate model for our purpose is provided by the
ture in front of the conical-tower enclosure. The Venda Domba School (Blacking 1969). The Venda are an
stepped platform to the right of the main entrance to appropriate source, because they are descendants of the
the conical-tower area possibly represented a chikuva Zimbabwe culture: the Venda language is a recognized
(pot-stack). The striped stones represent male fertility amalgamation of Shona and Sotho, and several clans are
(pp. 143–44). In the smaller buildings between the known to have lived north of the Limpopo river in Zim-
Great Enclosure and the hill, slots, small conical babwe’’ (p. 255).
towers, and stepped platforms confirm Huffman’s idea Apart from the fact that modern opinion on such fea-
that this was a female area, and therefore a monolith in tures as monoliths is not proof and that the cattle bones
a slot here means procreation. The only bird not on the might relate to 18th-century sacrifices recalled in oral
hill represents the great wife’s male ancestors, and the tradition rather than to pre-1550 practices (Mauch
crocodile on the bird’s column safeguards the wife 1871–72:215–16),4 the sudden shift of the terms of ref-
against witchcraft during pregnancy (pp. 144–45). erence from the Shona to peoples like the Bemba and
If all of this seems a little complicated, it is only fair the Pedi is a startling one. However, if Huffman wishes
to Huffman to remember that the site is complicated.
Indeed, in dealing with the Great Enclosure, he has not 4. This source does not specify sacrifice except on the hill and
fully developed his arguments; especially where figu- ‘‘some distance away,’’ but it does not preclude the possibility of
rines and fertility are concerned, he is preparing the sacrifices just inside the northern entrance as well.
54 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

to bring female initiation into the argument, then he is the Great Enclosure was an initiation school for both
more or less forced to do this, because, quite simply, sexes, with a slightly greater emphasis on the female
there is no evidence for female group initiation among side, for the cattle bones on a cairn at the northern ‘‘fe-
the Shona in 500 years of documentation and enquiry. male’’ entrances were from sacrifices to the spirits of
The claim that the Venda rather than the Shona are the the rulers’ sisters, and the stepped platform to the right
most appropriate model is extremely misleading to any- of the conical-tower area was a symbolic chikuva, used
one unfamiliar with the ongoing discussion of Venda for symbolic kitchen-craft lessons by the school (pp.
origins.5 An Indo-European equivalent of Huffman’s 257–59).
argument would be to seek to explain an aspect of a Given that the Venda do categorise people thus in
late-1st-millennium French ruin by reference to an En- their ceremonies, the connection between them and
glish ‘‘custom’’ by claiming that English is a recognized Shona hakata is not quite as clear-cut as is implied.
amalgamation of French and Norse and that several En- Some writers offer a different categorisation, such as
glish families once lived in France. Nevertheless, Huff- living men and women as against male and female spir-
man is determined to involve the Venda in the argu- its or specifically sexual attributes of men and women
ment, because he sees a link between Venda social as against more general manhood and womanhood (Ed-
status, Shona artefacts of the recent past, and Great wards 1929:40–41; Tracey 1934:23–25; 1963:106; Si-
Zimbabwe’s walls. card 1959:26–29; Bourdillon 1976:178). However, even
Venda female initiation lessons ‘‘often’’ revolve assuming that Sicard and Huffman were right in linking
around four status categories which underlie Shona and hakata designs to Great Zimbabwe wall decorations,
Venda social organisation: old and senior man, old and the latter’s interpretation runs into trouble because of
senior woman, young and junior man, young and junior the absence of herringbone decoration in the Great En-
woman. Shona hakata or divining dice or tablets are closure. Huffman’s response is special pleading about
also categorised in this way, bearing specific designs: now-vanished features, and his assertion of the identity
old man, crocodile; old woman, double design; young of the spirits to whom sacrifices were made falls into
man, tripartite; young woman, intertwined. Along with the same category. Moreover, in arguing for a complete
Sicard (1959:27), Huffman links the designs on the ha- correspondence between hakata categories and wall
kata to the four kinds of stonework decoration at Great decorations, Huffman claims that the striped stones
Zimbabwe: old man, dentelle; old woman, herringbone; next to the conical tower, which in ‘‘Snakes and Birds’’
young man, chevron; young woman, stripes (1984a: (1981:142–43) meant male fertility, now represent the
256–57). Although there is no herringbone decoration ‘‘young woman’’ category of fertility. Changing one’s
in the Great Enclosure, it may once have decorated the mind is an essential part of research, but in this case it
small ‘‘female’’ conical tower, because the female as- looks as if parts of the evidence were being forced into
cent to the hill buildings was so decorated (p. 257). Thus place.
Huffman goes on to apply the argument to later build-
ings, Khami, Majiri, Tere, and Danangombe, though
5. This discussion has been based on oral traditions, linguistics, a
this involves replacing both dentelle and chevron pat-
very few documents, and, recently, archaeology. There are three
main theories, but they all relate to the fact that today Venda is terns by checks—making one pattern represent two op-
seen as a distinct language and not as a dialect of any of the three posed concepts that required separate patterns at Great
other languages that are neighbours to it, Sotho, Tsonga, or Shona. Zimbabwe (1984a:259–61). Following this, he argues
The first and oldest theory suggests that there was a Venda-speak- that the Shona once practised initiation over the entire
ing population living north of the Zambezi that came south rela-
tively recently, picking up Shona elements as it passed through the
south and south-west of the Zimbabwean plateau (and
Zimbabwean region and then Sotho elements when it arrived in the north-east, if Tere is included) but gave it up at an
the Soutpansberg (Wentzel 1983:169–72). A second suggests that undated event called ‘‘the collapse of the Shona states’’
Venda is a language of equal antiquity to that of its neighbours and (p. 264). This does not fit the known historical sources.6
that Sotho traces reflect Sotho proximity and Shona elements at
least two Shona immigrations (Beach 1981:211–17, 260–63). A
third shows that Mapungubwe (Leopard’s Kopje) and Moloko pot- 6. This ‘‘collapse’’ in fact only occurred in the south-west in the
tery predominated north and south of the Soutpansberg respec- 1830s, where the Changamire state had been based at Danan-
tively, being replaced after about 1450 by Tavatshena and then Let- gombe. Rozvi memories and traditions of this state, recorded in the
aba pottery. As Mapungubwe and Moloko pottery are generally 1890s and 1900s before Rozvi traditions began to change, did not
equated with early Shona- and Sotho-speakers respectively and as mention initiation. In the north-east, there was no dramatic change
one of the five modern populations making Letaba pottery today is in such regions as Maungwe, Manyika, Barwe, or modern Mutoko,
Venda, therefore the appearance of Tavatshena and Letaba pottery where there were wars with the Portuguese, or in such areas as
reflects the emergence of a completely new language, Venda, out Buhera or Makonde-Guruve that were not affected by these, and it
of a merger of Shona and Sotho (Loubser 1989:54–61). Regardless is difficult to see why initiation should have been abandoned if it
of which version is preferred, the Venda link with Great Zimbabwe had ever been practised or why it should not have been mentioned
proposed by Huffman has a fundamental problem in that it has to in Portuguese documents or in oral tradition. Essentially, the initi-
prove that the group initiation practised by the Venda today comes ation-school theory rests on the assumption that human figurines
from the Shona, who do not do such things, rather than from the can only mean group initiation. This does not mean, of course, that
neighbouring Sotho and Tsonga (or from an earlier proto-Venda the Shona had or have no concept of instruction of young people
population), who do. A test of Huffman’s theory might be found in aspects of adult life but rather that they did not and do not have
over the linguistic and political frontier in Mozambique, where an organised initiation system involving separate initiation camps
there was also settlement from Great Zimbabwe: if the Chopi had and ceremonies on the lines of those of the Xhosa, Venda, and
initiation identical to that of the Venda, Huffman’s case would be Tsonga. The nearest Shona equivalent is described in Gelfand
strengthened. (1960:36–37).
 b e a c h Cognitive Archaeology and Imaginary History 55

The Origins of the Great Zimbabwe steps where angels fear to tread by suggesting another
Cultural Pattern model using oral traditional material and documentary
evidence. As I will show, there are great dangers in do-
It has taken considerable space to summarise and com- ing so, and what I am suggesting is intended purely as
ment upon only two of Huffman’s articles on Great a working hypothesis to stimulate debate.
Zimbabwe, and even then it has been at the expense of My main objection to Huffman’s interpretation of
passing over his arguments for the extension of his Great Zimbabwe is not contained in my criticisms of
Great Zimbabwe cultural pattern to other sites. In fact, his use of specific pieces of evidence or even of his initi-
these articles are part of a prodigious number of publica- ation-school hypothesis. Where I think Huffman’s idea
tions covering a much broader field. As far as those on is weakest is in its treatment of Great Zimbabwe as a
the origins of Great Zimbabwe are concerned, they are single community, almost as a village writ large, and as
based on the published conventional archaeology. Like a static one in which the entire pattern of settlement
all other archaeologists from the late 1950s to the pres- remained more or less the same over at least two centu-
ent, Huffman sees the population of Great Zimbabwe ries. (Indeed, it assumes that the pattern was conceived
as having its origins in an ordinary mixed farming com- as a whole from the beginning but that only parts of it
munity that lived at the site for a considerable time be- were executed in ‘‘P’’-style walling for some 25–50
fore any stone buildings or urban conditions emerged. years and that the rest of it remained in its designers’
Where he differs from his predecessors is in his linkage minds until they could complete it in ‘‘P/Q’’ and ‘‘Q’’
of the site of Mapungubwe, on the South African side styles when these had evolved.) 7 There is a tendency to
of the Limpopo, to Great Zimbabwe. Up to the late think of the elite living area purely in terms of a ruler,
1970s it had been thought that Mapungubwe was a late his chief and subsidiary wives, and their religious and
derivative of Great Zimbabwe (Beach 1980:36–41). Re- social activities. There is also a tendency to look at the
dating showed that the reverse was the case: Mapun- elite area largely in terms of the walled enclosures—
gubwe flourished ca. 1075–1250, it was there that which is odd, in view of the fact that it was Huffman
striking new developments in state formation, class for- who first pointed out the importance of the unwalled
mation, and the spatial organisation of settlements be- areas—and to attempt to explain very nearly all of their
gan, and it was from there that cultural diffusion rather features and the non-functional artefacts found within
than migration spread these new ideas to Great Zim- them.
babwe, where the existing population adopted them and My model starts from the premise that Great Zim-
rapidly eclipsed the older centre. Given that Mapun- babwe was a major Shona political centre and that—al-
gubwe’s spatial organisation was very similar to that of though it may well have contained elements of older
its successor, Huffman is thus able to interpret it in the Shona political thinking or even of experimental as-
light of the earlier articles discussed above (Huffman pects that did not survive into historical times—it is
1986:315–19; Huffman and Vogel 1991:61–70). reasonable to interpret it in terms of political process.
Obviously, Huffman’s interpretation of Mapun- Naturally, it is very difficult, if not impossible, to de-
gubwe’s cognitive archaeology is still farther removed duce any kind of political process from archaeological
from the historical and anthropological evidence than evidence alone, so we have to rely upon historical evi-
that of Great Zimbabwe, but in addition there are some dence. We now know a great deal more about the work-
unexplained coincidences. Mapungubwe and Great ing of the Shona polities than we did even 15 years ago.
Zimbabwe had nearly identically oriented hills (north- Unfortunately, the new evidence relates almost entirely
west–south-east and west-east), similar ascents up to the much poorer societies of the 18th and 19th centu-
clefts in their sides, and, according to Huffman, identi- ries, for it is now clear that oral traditional evidence on
cal choices of the western end of the hilltops for royal the internal workings of societies predating 1700, such
living areas. Yet the Gumanye ancestors of the Great as the Mutapa state, is highly unreliable (Beach 1994b:
Zimbabwe builders were already living at the western 28–187, 211–43). Consequently, the model uses evi-
end of their hill before the Mapungubwe innovators set dence on politics largely from after 1700 to reconstruct
to work on theirs. Huffman’s stress on Mapungubwe as the political life of a city that had ended by 1550, and
a centre of change in Shona society appears to be un- this quite obviously has its dangers. Yet perhaps the
challengeable on existing evidence, but the coinci- dangers are not quite as great as they seem. If the Shona
dences warn us that not everything can be neatly ex- no longer thought it necessary to carve soapstone birds
plained. or to build conical towers after Great Zimbabwe ended,
it does not follow that there was a complete change in
their mode of thought, especially in matters as basic as
kinship and politics. Moreover, there was in fact no sud-
An Alternative Cognitive Archaeology and den and catastrophic end to the states of the Shona, ex-
Imaginary History cept in the south-west. Rather, there was a process of

Although certain aspects of Huffman’s use of evidence

7. All previous writers on Great Zimbabwe assumed a gradual evo-
are unsatisfactory, cognitive archaeology has definite lution of wall styles as building expanded over decades or even cen-
potential for helping us to understand the Great Zim- turies. Chipunza (1994) has forcefully shown the older view to be
babwe culture. I propose to follow in Huffman’s foot- more probably correct.
56 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

gradual impoverishment and reduction in political scale Ruler I lives at the western end of the hill (fig. 3), as
from about 1600 to about 1850, and the dividing line did many but not all of his male ancestors, who had oc-
between the larger Shona territories and the states is cupied the Gumanye site in the late 11th and 12th cen-
rarely a clear one. An examination of Shona thought turies. Considering elaborate houses and a wealthy life-
from historical sources shows many elements of conti- style insufficient to express his grandeur, he decides in
nuity from 1500 to 1900 (Beach 1994a:112–63). about 1250 to build stone walls around his home, just
Finally, it must be stressed that although the actual as the rulers of Mapungubwe had done in the past.
political succession suggested in the following model is These walls, in ‘‘P’’ style, are experimental at first and
imaginary, the kinds of political process depicted cer- frequently collapse, but soon the specialist stonema-
tainly are not. They are based on detailed studies of the sons develop their technique, and by 1300 not only
histories of nearly 100 Shona dynasties over two preco- are their ‘‘P’’ walls becoming highly durable but they
lonial centuries, with oral traditions being checked are moving towards more elaborate styles, ‘‘P/Q’’ and
against documents wherever possible. These studies ‘‘Q.’’ These walls are first built around what becomes
give us, by example, an exceptionally clear picture of the Western Enclosure, possibly replacing an earlier
the Shona attitude to politics (Beach 1994b:26–187, wooden stockade.
211–20). Essentially the Shona of the 13th century are rather
We start by assuming that about 1100 (according to like their descendants of more recent times and, for that
Huffman’s latest radiocarbon chronology [Huffman and matter, humanity elsewhere in that they are pragmatic
Vogel 1991:61–70]) a ruling dynasty of the Gumanye about theory and practice. Thus, for example, they
culture begins to become so powerful and wealthy that share with some other Bantu-speaking peoples the idea
it is able to start building solid daga houses and to imi- that the ruler ought to be secluded from the gaze of the
tate the pottery of the rulers of Mapungubwe, at that populace, but as in practice he wants and needs to take
time the great state or protostate in the region. Thus far, part in hunting, agriculture, warfare, and other activi-
the Gumanye people are relying mainly on a cycle of ties they concede the point that he will not spend all
the accumulation of cattle, women, children, young his life hidden from view. Only at certain times will he
warriors (rather than full-time paid soldiers), and thus, be hidden or speak through intermediaries, such as at
as their raiding grounds expand, more cattle. This pro- the initial reception of an ambassador from another
cess begins to cut across the trading routes from the power. On other occasions, he will be in full public
goldfields of the south-western Zimbabwean plateau to view, which is one reason he needs bodyguards (Beach
the coast, and Great Zimbabwe begins to extort a tran- 1980:101–12). Similarly, there is an idea that the ruler
sit tax from the traders, who had previously paid it to should be higher than the populace, but this does not
the rulers of Mapungubwe. Where Great Zimbabwe necessarily mean that rulers must always live on hills.
achieves direct control of gold-mining areas, its rulers They may do so, but if they want to live on flat ground
get 50% of the first part of gold extraction, and their per- then on ceremonial occasions a raised seat or even the
mission has to be asked for mining to take place at all. seating of the populace on the ground while the ruler
In addition, wherever Great Zimbabwean forces control uses a stool will suffice.8 Other differences between the-
areas with elephant herds, their rulers claim 50% of the ory and practice relate to fire and ritual suicide: It is said
ivory taken. The ruler of the state has no monopoly of that the ruler sends out fire to his subjects every year
trade, but he is the greatest trader. Moreover, coastal to test their obedience, but no one ever sees this happen
traders wishing to buy gold or ivory from the villages (Barbosa 1966 [1518]: 362). The theory that an ageing or
under Great Zimbabwe must pay a percentage of their disfigured ruler must commit suicide is just a theory;
trading goods to the state ruler. With this influx of this does not happen unless he gets a disease like lep-
wealth interlocked with the further increase of cattle rosy (Beach 1980:97). Brother-sister marriage is offi-
herds by various means, the growth of Great Zimbab- cially sanctioned for rulers, but even though sister-
we’s power is fairly rapid, and by 1270 Mapungubwe has wives have children it cannot be proven that the ruler
been completely eclipsed by the rise of its northern is the father (Beach 1980:96–97; Mudenge 1988:104–8).
neighbour (Beach 1980:42–43, 100, 110–11; Huffman There are many other examples of this kind of distinc-
1986:322). tion between theory and practice (Beach 1994a:153–58).
A particularly powerful ruler has emerged during the Ruler I’s walls around the Western Enclosure contain
last stages of the struggle against Mapungubwe. He has his own house, one for his chief wife, and one for his
benefited from the work of his father, uncles, and earlier immediate servants, just as in the Mutapa state in 1600.
ancestors, and he gains the title when he is relatively His sister-wife may well live in the Cleft Rock Enclo-
young. There may or may not have been civil war, but sure in the northern part of the Hill Complex, along
there is no fixed succession system, and this ruler has with her chosen male companion (Bocarro 1899 [ca.
faced rivalry from his elder brothers, from his uncles,
and even from descendants of his grandfather, great- 8. On a Shona capital in low-lying ground used as an alternative to
grandfather, and still earlier ancestors. Some of these a long-held mountain site, see Kuss (1882:377–79) and Storry
(1974:10–13). The royal house was in the highest part of the enclo-
lived at Great Zimbabwe, and others held large subject sure, which is not much different from the rest, and had a dais and
territories many kilometres away. We may call this suc- monolith, but the entire site is overlooked by nearby hills. On eti-
cessful ruler Ruler I. quette, see Holleman (1958:88–89).
 b e a c h Cognitive Archaeology and Imaginary History 57

F i g. 3. Imaginary historical scenario, mid- to late 13th century.

1631]:267). The rest of Ruler I’s wives live just outside However, by the time the first walls are built, Ruler
the Western Enclosure wall, down the hill a little, I is growing old, and many of his sons are fathers of large
where he can have easy access to them. The royal families. There is little extra room for them on the hill,
kitchen is also there. His many young sons and daugh- so their homesteads are spread up and down the valley
ters live in houses on the terraces down the hill, includ- inside the Outer Perimeter wall. Ruler I has no objec-
ing those that lie on the south side of the hill inside the tion to the copying of his stone walls by them or, in-
Inner Perimeter wall. As kugarira (bride-service) mar- deed, by outlying elite residents or even provincial
riage is also practised besides roora (bridewealth) mar- subrulers with city residences. Consequently, in the fu-
riage, many of these daughters have their husbands liv- ture Great Enclosure area, the No. 1, Renders, Mauch,
ing with them, and it is partly from them that the royal and East Ruin sites similar walls are built to surround
entourage of bodyguards, jesters, musicians, and others at least some of the houses of senior sons, in-laws, and
is drawn (Beach 1994a:50–53). favoured torwa (unrelated) members of the elite. More
58 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

than 20 outlying members of the dynasty and their asso- relatively small complex at this stage (fig. 4). However,
ciates build on sites around the outer edge of the grow- in the long period during which his father’s corpse is be-
ing town, including the Mutuzu, Chenga, South East, ing mummified, eventually to be buried in a sacred
and NeManwa sites. In each case, the family group has cemetery hill far away while his father’s spirit is
houses both within and outside its stone enclosure. ‘‘brought home’’ and ‘‘settled,’’ Ruler II does not move
Towards the end of the 13th century, Ruler I dies. Al- into the Western Enclosure with his wives. This is
though much the same political forces are operating in partly because his father’s house will need purification
the public and private arguments about the succession or demolition—which is one reason there are so many
as in his youth, it happens that one of his elder sons settlement levels in the Western Enclosure—and partly
has enough support to become Ruler II. The new ruler because not all of his father’s wives have gone to live
already lives in the future Great Enclosure area, still a near his grave, some continuing to live in the Western

F i g. 4. Imaginary historical scenario, late 13th century.

 b e a c h Cognitive Archaeology and Imaginary History 59

Enclosure area (Bullock 1950:176–77; Mudenge 1988: (and at least some of his sons who are in opposition to
81). them, hoping to secure the title for themselves) are
Because he is an elder son, Ruler II does not have long beaten by one of the provincial house-heads, who be-
to reign before he, too, dies (Beach 1979:48–49). This comes Ruler III. By this time, Great Zimbabwe is so im-
time, though, his brothers are keen that adelphic suc- portant that possession of it is essential for success as
cession should apply and that thus they should inherit a ruler. Ruler III comes to Great Zimbabwe and reoccu-
the title in turn; other political factors come into play. pies the Western Enclosure at the turn of the century,
Ruler I had had to overcome the opposition of his broth- evicting the previous occupants and building his own
ers, uncles, and others, who are by this time mostly liv- houses (fig. 5). By this time, the specialist stonemasons
ing in the provinces far away. On Ruler II’s death, civil have evolved the ‘‘P/Q’’ and ‘‘Q’’ styles, and consider-
war breaks out, and in the end the brothers of Ruler II able rebuilding and new building goes on in both the

F i g. 5. Imaginary historical scenario, 14th to early 16th century.

60 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

valley and the hill areas. On Ruler III’s death (perhaps were so dominated by ‘‘custom’’ that there was no room
not long after, as he is himself old, having had a long whatever for individual initiative. The birds were beau-
and successful career in the provinces before the war), tiful, but they may not have been as crucial to the
there are at least two houses operating in Great Zim- thinking of the Great Zimbabwe people as they are to
babwe politics. Ruler IV could have come from any of that of archaeologists—which may explain why they
these, but in any case it is he who undertakes the con- were left at the site. The question of ‘‘innovation’’ as
struction of the Great Enclosure in either its final or an against ‘‘custom’’ equally applies to the clay birds from
intermediary form as a living area and inner court a zimbabwe in the far south-west (Frobenius 1931:pl.
centre. 50). But these are speculations and have no more solid
Obviously, there are many variations that might have basis in historical fact than many of those criticised
applied instead of the version outlined above, but they above.
would have been variations on a theme, the remarkably
fluid process of Shona politics (Beach 1994b:28–187).
For the next 250 years, things go on in much the same
way. There is rebuilding of several structures to suit the Towards Future Debates
needs of the day as time goes on. After about 1450, how-
ever, Great Zimbabwe is facing the rivalry of the Khami As research and publication on Great Zimbabwe is now
and Mutapa states, and it is in decline. By the early 16th almost continuous, a ‘‘conclusion’’ here would be inap-
century, much of the city is no longer occupied, and the propriate. Perhaps a few observations would be more
last centre of power is in the small Western Valley En- useful. Returning to the idea of Great Zimbabwe as be-
closure, though even there elaborate housing is con- ing on several frontiers, the first problem is that of the
structed (Collett, Vines, and Hughes 1992:139–61). One frontier between archaeology and history and, by impli-
can speculate about the actual end of the state. One pos- cation, cross-disciplinary work in general. Historians of
sibility is ‘‘implosive,’’ with the city getting smaller and precolonial Africa are usually insistent that archaeolo-
smaller until, finally and figuratively, the last inhabi- gists should be aware of history, but they also insist that
tant expires next to the grave of the second-last inhabi- the reading of documents and the collection of oral tra-
tant; another is ‘‘explosive,’’ with repeated emigrations ditions are highly technical skills that require speciali-
of factions and, ultimately, the emigration of the main sed training. My criticisms of Huffman’s handling of
dynasty. Actually, both scenarios are possible, as 15th- historical evidence should be seen in that light.
century Constantinople was demonstrating at roughly Another problem is that of the frontier of time: both
that time. It was once thought that a ‘‘skeleton staff’’ Huffman and I are dealing with the question of how far
of Rozvi remained in the city, but this has been more the thought of the Shona changed over time, between
or less disproven (Mtetwa 1976:90–107). the floruit of Great Zimbabwe and now. Both of us ar-
This model may be plausible, but of course it is partly gue for a general continuity—he for orientations and I
based on the imagination. To a certain extent, it pro- for political process—but both of us see changes, in his
vides an alternative explanation for Huffman’s cogni- case the abandonment of initiation and in mine the
tive archaeology, but it does so in an entirely different abandonment of semi-experimental sculpture and ar-
way. It cannot hope to explain the decorations, conical chitecture. It is curious, however, how different our
towers, or many other features in the way that he does. continuities are. Huffman’s Great Zimbabwe Shona
However, very tentative alternatives can be suggested seem to think in clear-cut categories and to build ac-
for a few features. The objection to interpreting the cording to them; mine admit that categories ought to
stepped platforms or raised platforms in the Western exist but often act in a quite different fashion and do
and Great Enclosures as symbolic chikuva is that, as far not necessarily build anything to represent them.
as is known, such stacks occurred only inside houses, Clearly, the differences might represent the writers and
because they were not just religious focal points but not the Shona at all. Obviously, although historians and
also had a practical use for the storage of food, drink, anthropologists frequently state that ‘‘the X people’’ did
and other items (Brown 1899:199–200). It might be that or do something, this is a generalisation and is as dan-
they were used as raised seating areas for the ruler on gerous as most of these. The Shona writers themselves
those occasions when outsiders to the household did have not arrived at any common opinion on whether
enter the royal apartments (Barbosa 1966 [1518]: 360). If their ancestors were similar in thought to themselves—
we abandon for a moment the pervasive idea that the and it might not be significant if they had, because na-
Eastern Enclosure on the hill was a religious centre, we tions can also be self-deluding. Thus, Mudenge (1988:
have to ask what else might have been kept there. A 8–20) is prepared, with due scholarly caution, to accept
possibility is the royal drum, which had to be kept the basic hierarchy of modern Shona social units as hav-
somewhere when it was not being carried out to battle ing applied ca. 1500, but Chipunza (1994:58) is strongly
by 12 men as a symbol of power (Gomes 1959 [1648]: opposed to seeing the modern Shona as ‘‘living fossils’’
191). We simply do not know what the soapstone birds of the old. This contradiction between two Shona writ-
meant, but in fact they would not have strained the ca- ers is a good example of the contradictions that I see
pacity of a single carver working over a few years. We from the very moment that the Shona-speaking people
have no reason to believe that the Shona of the past became exposed to the light of history (Beach 1994a:
 b e a c h Cognitive Archaeology and Imaginary History 61

142–63), and, naturally, it foreshadows further contra- tion indices. Perhaps it is wrong to take academia seri-
dictions. ously.

james denbow
Department of Anthropology, University of Texas at
Comments Austin, Austin, Tex. 78712, U.S.A. 2 ix 97

Beach, a historian, provides a critique of archaeological

interpretations of Great Zimbabwe put forward be-
m. f. c. b ou r di l l o n tween 1977 and 1987 by Thomas Huffman. He convinc-
Department of Sociology, University of Zimbabwe, ingly shows that Huffman’s ideas have remained re-
P.O. Box MP 167, Mount Pleasant, Harare, markably constant over this long period. His critique of
Zimbabwe ([email protected]). 8 viii 97 specific aspects of Huffman’s argument would have
been improved, however, if it had included Huffman’s
Although Beach’s article does not take Huffman’s re- latest and most complete work, Snakes and Crocodiles
cent book, Snakes and Crocodiles (1996) into account, (1996). He argues, for instance, that ‘‘the chief wife
it has not been superseded by it. Huffman has not sub- should be ‘next’ to the ruler’’ rather than ‘‘650 m’’ away
stantially changed either his argument or his use of as Huffman proposed in 1981. Huffman (1996:133),
data. I have found it puzzling that academics could be however, while continuing to associate the hilltop ruins
convinced by his argument, but some have been, and with the residence of the king and his ‘‘ritual sister’’
therefore Beach’s critique is necessary and welcome. (rather than the chief wife), now argues that ‘‘the King’s
I am unconvinced by a number of Huffman’s assump- youngest wives [also] lived here, rather than spirit medi-
tions: that the thought in any society can be subsumed ums as I once thought’’ (Huffman 1996:133). The chi-
under a single symbolic system, that a few proven simi- kuva in the Great Enclosure that Beach criticizes are
larities between peoples (Venda and Shona) allow the also transformed in the later work into ‘‘speaker’s plat-
assumption of such a system, that similarities between forms’’ more consonant with Huffman’s contention
past and present Shona societies allow the assumption that the Great Enclosure functioned as a national initia-
of identical worldviews, that interpretations by con- tion school.
temporary Shona imply knowledge of life 500 years ago. The importance of Beach’s paper, however, resides on
From these assumptions Huffman argues that a central a deeper level. As he points out, Huffman treats the
institution in Great Zimbabwe is initiation, of which building of Great Zimbabwe as if it were a politically
there is no trace in any Shona society, Shona oral tradi- static community that followed a rigid and unvarying
tion, or written record. This argument of total disconti- cosmological master plan. The result is that the daily
nuity from assumptions of continuity seems to me alto- processes of accommodation and resistance through
gether too implausible. which political life was constructed are treated as if
Beach’s detailed criticism of the methodology and ar- they were simply epiphenomenal constants overdeter-
gument is convincing. He sets a rigorous standard in his mined by structural rules of spatial organization. Huff-
own work in the use of both oral and written evidence. man’s assumption of political stasis at Great Zimbabwe
If we do not always meet these standards, it is as well is not remedied but amplified in Snakes and Croco-
to be aware of where we fall short. Beach’s alternative diles, where ‘‘virtually every ruling dynasty was already
explanation of the growth of Great Zimbabwe is more integrated into a single upper class’’ (1996:190–91).
plausible, although he rightly points to its tentative and Beach’s argument is that if there are common political
unproven nature. processes to be sought, they are as likely to have in-
I am less confident than Beach is of the value of cogni- volved competition and civil war as class solidarity and
tive archaeology in situations where we have no data on cooperation. Even in his latest work Huffman seems to
how people think. In anthropology, speculation about overlook the possibility that the king’s greatest rivals
how people think went out of fashion when anthropolo- may also have been his closest relatives. On this point,
gists began to do serious fieldwork and meet people in at least, Beach’s interpretation has the benefit of being
the varieties of practical life. In real life, we often fail closer to historically documented processes of political
to understand people in spite of access to their everyday succession in southern Africa.
conversation. The subjects of archaeology are unable to His criticism of Huffman’s interpretation of the soap-
contradict our attempts to systematise their thought. If stone birds in the Eastern Enclosure as ancestral figures,
anthropologists are sometimes guilty of imposing their associating this area with ‘‘spirit mediums,’’ is well ar-
systematic constructions on the people they write gued in terms of a lack of direct linkages between Shona
about, this is much easier for archaeologists, who never oral traditions and religious practices at Great Zim-
meet those people. It would be pretentious to present babwe 500 years earlier. His disagreement with Huff-
our speculations as knowledge. man on the historical origins of initiation schools in
But perhaps I am wrong in my assumption that aca- Venda is similarly effective and seems to me to resonate
demic life should be about knowledge rather than cita- better with other linguistic and cultural comparisons of
62 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

Venda and Sotho than Huffman’s model. His historical model of actors derived from his knowledge of histori-
argument seems more tenuous, however, when it is di- cal Shona society.
rected against Huffman’s use of ethnographic analogy. Structuralist interpretations of recent settlements
Addressing Huffman’s use of male and female divisions have at least sometimes not been borne out by the phys-
symbolized by decorations on Shona divining dice to in- ical evidence. A case in point is the symbolic interpreta-
terpret decorations on stone walls, for instance, he ar- tions and actual settlement patterns near Vilanculos in
gues that there are no ‘‘female’’ herringbone designs on southern Mozambique, where the east-west, left-right
the ‘‘female’’ side of the Great Enclosure. He extends gender and senior-junior patterns outlined by A. Y. Ca-
this attack by pointing out that ‘‘some writers offer a sal and J. F. Fialho do not seem to exist as realities in
different categorisation, such as living men and women mapped family compounds (Casal 1996:74–77, 88–89,
as against male and female spirits or specifically sexual 102–4). In a similar vein, Garlake (1992:23, 25) has
attributes of men and women as against more general flatly denied the existence of deep structural patterns in
manhood and womanhood.’’ To me, however, this present-day Shona settlements and cast doubt on their
seems to strengthen Huffman’s more general argument existence in earlier times. If present-day symbolic struc-
that the divining dice reflect wider structural relation- tures are not clearly reflected in the physical evidence,
ships in Shona thought: ancestors and living, senior and why insist on discovering them in abandoned building
junior, men and women, young and old. Whether these complexes which may have been modified by several
divisions are meant to be indexed by spatial differences actors?
in stone wall decoration at Great Zimbabwe and other Huffman’s role as a researcher in Zimbabwe was es-
ruins remains to be proved, preferably through more ex- sential; he opened new fields. Some recent researchers,
cavations and analysis of archaeological materials however, have abstained from commenting on his at-
rather than through the formulation of more ‘‘sce- tempt to discover gender symbols, initiation, etc., in the
narios.’’ archaeological traces. Pikirayi (1993) manages to cite 13
I do not want to be an apologist for Huffman’s inter- publications by Huffman between 1971 and 1989 while
pretations. To rework a phrase of Shanks and Tilley avoiding the four cognitive ones mentioned above.
(1992), all too often there are no archaeological data to Cognitive approaches are not new—for example, in
form even small ‘‘resistances’’ to his interpretations, the appraisal of ‘‘primitive art’’—and very often project
which are often circular, overgeneralized, and contra- elements of the analyst’s worldview onto the past. They
dictory. As Beach argues, the modern Shona and other are often difficult to falsify and therefore remain in the
southern African peoples are not ‘‘living fossils,’’ and realm of what Popper called ‘‘conjecture’’ and others
one needs to be wary of hasty linkages that assume an have called ‘‘unconfirmed hypotheses.’’ All one can do
unvarying uniformity between modern beliefs and past in these cases is to prove misunderstanding or manipu-
practice. Beach’s point that Great Zimbabwe was not a lative selection of data. This Beach has done, but, as he
single, static community that remained unchanged for recognizes, his own proposal of looking for a sequence
over two centuries is an important one, but his alterna- of rulers is the statement of a problem rather than an
tive ‘‘processes’’ will remain on the level of ‘‘scenarios’’ answer and will be difficult to confirm.
until we have not only new interpretive models but The more open approach of systems analysis—ana-
new archaeological data. lysing building sequences, the acquisition of resources,
waste disposal, the distribution of artifacts, pottery
analysis, the rebuilding and repair of houses, roofs, and
gerhard liesegang walls, etc.—drawing on a wider range of data seems
Departamento de História, Faculdade de Letras, likely to be more fruitful and would allow the identifi-
Universidade Eduardo Mondlane, Maputo, cation of patterns that are easier to confirm. Huffman
Mozambique ([email protected]). 25 viii 97 himself has amply contributed to this field. Eventually
confirmed observations might show a pattern and sup-
Beach’s paper has two main parts on which I would like port interpretations.
to comment: a description and critique of the cognitive Unfortunately, because of the nature of the earliest
(or structuralist [cf. Garlake 1992:23, 25]) interpretation ‘‘excavations’’ and ‘‘cleaning’’ operations in 1892,
of building features at Great Zimbabwe made by Huff- 1902–3, and 1915 in the valley, the Great Enclosure,
man (1981, 1984a, b; also in Huffman 1987) and an at- and the Western Enclosure on the hill, it may be diffi-
tempt to suggest a possible periodization of the main cult to demonstrate a change in the use of certain build-
building complexes at Zimbabwe. The latter introduces ing complexes at Great Zimbabwe, although plans pub-
a model of generational dynastic activities, thus draw- lished by Garlake, Huffman (1987:32,35), and others
ing attention to the ‘‘expressive space’’ not only of a point to that possibility. Stone and daga house struc-
‘‘society’’ but of powerful rulers and individuals and the tures make past social arrangements highly visible in
groups supporting them, as well to the practical prob- the archaeological record, but to document them fully
lems of organizing households. Beach thus remains requires almost full excavation, and in some buildings
partly within the realm of cognitive archaeology but central to the argument changes in the layout of houses
moves from one vague actor (society [see Huffman were not recorded before they were destroyed. A more
1984b, 1986, as quoted by Beach]) to a more complex systematic comparison of house types allowing us to
 b e a c h Cognitive Archaeology and Imaginary History 63

distinguish chief females’ and junior wives’ houses tion of Great Zimbabwe is in many ways quite different
might be a useful step. from anything we know. This new synthesis, from my
Beach’s approach, stressing the length of occupation point of view, is where the material evidence from the
and changes in utilisation of space, may also provide a archaeological record can have a beneficial feedback ef-
stimulus for the reinterpretation of part of the ‘‘lower- fect on how we understand the past and the analogs on
class’’ living space. Unlike northern Swahili towns, the which our interpretations are based. In short, Huffman
Mozambican south-coast and Shona settlements in- does not accept oral tradition over some 600 years at
cluding Zimbabwe lacked special sanitary facilities. face value, and even if he disregarded Shona oral histo-
Bush patches, places between rocks, etc., served that ries altogether it would not be critical to his interpreta-
purpose. Therefore one would think that people would tion because he has comparative ethnographies and the
have preferred to have some abandoned compounds and archaeological record as main sources of evidence.
other covers close to their homes. Possibly the high Beach’s second objection is that there is no evidence
population estimates such as 18,000 for the 14th–15th for female group initiation, and here archaeological and
century offered in some publications (see Huffman ethnographic work in neighboring areas is significant.
1987:24) would have to be reduced to a quarter or half Beach states that Huffman’s use of the neighboring
to allow for this. Another open question is the presence Venda-speaking people as an analog for Shona practices
of Islamic elements. They were on the plateau in the is far-fetched. Ironically, his own reconstruction of
16th and 17th centuries; did they have built-up places early Venda history supports Huffman’s historical con-
of prayer at Great Zimbabwe or elsewere? nection between the Venda and the Shona. Beach’s
One final note: The latest and most complete but still model is furthermore supported by the archaeological
unfinished edition of Duarte Barbosa (Barbosa 1996) has and oral historical sequence I have constructed for the
Barbosa dying in India of old age and invalidates the Soutpansberg. He claims that there are three theories of
identification made with Magellan’s companion. If Venda origins, but in fact there are only two—the old
names do not always prove an identity, do decorative school claiming Venda origins in Central Africa (Went-
patterns always imply a specific content and meaning? zel 1983) and the more recent one (Beach 1980, Loubser
1989) demonstrating local Venda origins with strong in-
fluences from Zimbabwe. Beach and I have separately
j oh a n n e s h. n. l ou bs e r shown, in terms of oral histories and ceramic evidence,
New South Associates, Inc., East Ponce De Leon that there were at least two Shona immigrations into
Ave., Stone Mountain, Ga. 30083, U.S.A. the Soutpansberg region, the first by approximately a.d.
([email protected]). 27 ix 97 1450 and the second by a.d. 1700. Moreover, apart from
the early Sotho contribution to Venda, the Venda and
My commentary on Beach’s critique of Huffman’s the Shona have common roots in the Mapungubwe
cognitive model is informed by three years of ethno- culture. In other words, the shared cultural roots of the
graphic and archaeological fieldwork among the Venda- Venda and the Shona were strengthened during repeated
speaking people of Soutpansberg (1994–97). Beach later contacts. Persistent culture contact and shared
charges Huffman with three purported transgressions: values can be seen also in the ethnographic record, for
uncritical use of oral traditions, a mistaken Venda ‘‘ini- example, in the Nwali rain-making cult, which is based
tiation school’’ analogy, and an oversimplified presenta- in Zimbabwe.
tion of Great Zimbabwe as a static village. Viewing the It is quite conceivable that formerly shared customs
debate from a comparative anthropological perspective, survived in Venda-speaking communities in some of
I critically address Beach’s three main concerns. the more remote and rugged pockets of the Soutpans-
Beach goes to great lengths to warn the reader against berg. Such customs very likely included male circum-
the various pitfalls of oral history. Clearly, his method- cision and female initiation. Fairly recent 19th-century
ological concerns are justified, but this obviously does historical evidence indicates that the Venda adopted
not mean we should abandon oral traditions altogether. and discarded circumcision ceremonies within a fairly
This is underlined by the fact that he has recourse to short time span. The Venda chief Makhado, for exam-
oral traditions in his alternative interpretation in terms ple, adopted circumcision from his North Sotho neigh-
of political process. It is also not the case that Huffman bors virtually overnight, and there is archaeological evi-
relies solely on oral tradition in reconstructing the pre- dence in the form of stone cairns in isolated locations
historic use of space. In fact, he synthesizes oral tradi- that this practice existed before among the Venda. Par-
tion, various ethnographies, historical documents, and ticular types of initiation seem to come and go, but ma-
the archaeological record in an attempt to understand terial reminders in appropriate locations help us better
the patterning he observes in various stone-walled ruins understand indicators from the archaeological record.
in Zimbabwe and neighboring regions. In using these For example, the conical towers adjacent to the meeting
different sources as starting points he employs a variant areas of some Venda royal stone-walled settlements are
of the ‘‘direct historical’’ approach. This does not mean, testimony to female initiation. Independent corrobora-
however, that the past is necessarily a reflection of the tion includes associated figurines of the type found at
present or any of the original analogs. Interestingly, as numerous Venda settlements and at Great Zimbabwe.
Beach is at pains to point out, the cognitive reconstruc- Beach is justified in showing that Great Zimbabwe
64 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

was a dynamic settlement, a fact that has been sup- babwe with the Shona, and the way in which the evi-
ported by archaeological research (e.g., Collett, Vines, dence is presented makes it difficult to convince the
and Hughes 1992). This does not however, mean that reader that the Shona were no more than mere occu-
these developments did not hinge around some fixed pants of the buildings. Thirdly, Shona oral traditions
points in the cultural landscape. Among various Bantu- with a time span of six or more centuries are treated
speaking people in South Africa, for instance, the huts as suspect, but a hypercritical approach to this kind of
are built around the central cattle byre and change posi- evidence presents the danger of obscuring historical
tion through time. What is important is that the byre facts which may be grafted upon later traditions and
always remains central in relation to the huts. Obvi- survive intact for more than a millennium, as is argued
ously the layout changed during the occupation of by Schmidt (1996) in the case of the Haya of north-west-
Great Zimbabwe, but Beach does not convincingly ern Tanzania.
show how this changed the cultural landscape of the Beach’s subsequent critique of Huffman’s approaches
place. He is a historian, while Huffman is an anthropol- to the use of space and the meaning of Great Zimbabwe
ogist. Their ostensibly divergent interpretations are not supposedly rests on these controversial statements. It is
necessarily mutually exclusive and may actually be timely that this critique appears a year after the publica-
complementary. Obviously, a more rigorous consider- tion of Snakes and Crocodiles, Huffman’s answer to his
ation of the different building sequences at Great Zim- critics, and a review issue of South African Archaeolog-
babwe is needed to adjudicate between competing ical Bulletin to which a number of scholars including
hypotheses. The best hypothesis is the one that not Beach have contributed in reaction to the book. For
only accounts for the archaeological evidence in a con- those unable to acquire Huffman’s two major articles
sistent and economical way but also accounts for the on the subject, ‘‘Snakes and Birds: Expressive Space at
most diverse data, including wall decorations and coni- Great Zimbabwe’’ (1981) and ‘‘ ‘Where You Are the
cal towers. Beach cannot account for the whole range Girls Gather to Play’: The Great Enclosure at Great
of evidence because as a historical literalist he fails to Zimbabwe’’ (1984), Beach provides an apt summary and
recognize that preindustrial communities such as the comment.
one that inhabited Great Zimbabwe do not make rigid In the third and final part of the article Beach pro-
distinctions between sacred and profane. Accordingly, poses that since Great Zimbabwe was a major political
he fails to see that choices on where to construct ap- centre, one could use oral traditions and documentary
proaches to royal enclosures, for example, are never evidence to interpret its development in terms of politi-
made on the basis of the prevailing wind direction cal process. This constitutes an alternative model to
alone. Similarly, he does not seem to recognize the pow- that of Huffman. It supposedly fits the known archaeo-
erful constraints that sacred tradition has over idio- logical developments at the site, but it fails to explain
syncratic behavior. Great Zimbabwe may have been the meaning of public architecture at Great Zimbabwe.
an architectural experiment, but it took place within Was this architecture essentially political in outlook?
the constraints of a relatively conservative Shona Did it reflect extravagant demands of the elite upon the
worldview. local populace which the latter did not appreciate? Fur-
thermore, the changing concepts of space over centuries
are not discussed. While Beach shows that 13th-century
i nn o c e n t p i ki r a y i Shona, like Shona living centuries after them, were
History Department, University of Zimbabwe, P.O. pragmatic about theory and practice, he does not dem-
Box MP 167, Mount Pleasant, Harare, Zimbabwe. onstrate, for lack of evidence, how cognitive concepts
21 viii 97 changed over time. The model is based on post-1700
Shona politics, but it is not clear which Shona groups
Beach’s contribution serves as a useful critique of Huff- are being referred to and what links they might have
man’s theories on the meaning and use of space at Great had with the Zimbabwe culture. Finally, it is unclear
Zimbabwe. The paper opens with some controversial whether the political succession referred to in this
points relating to the distribution of the stone-built model is similar to that of the 17th or 18th century or
Zimbabwe-culture sites and the usefulness of Portu- later, how an elite developed outside the ruler’s settle-
guese written sources and oral traditions in interpreting ment on the hill, and how provincial rulers were related
these sites. However, the vagueness with which these to Great Zimbabwe. Shona political process here is an
points are presented spoils what could eventually have amorphous concept that requires formal definition and
been a coherent and convincing argument. First, Beach clarification. As it stands it is unclear how political suc-
points out that, by virtue of their distribution, the cession, wealth accumulation, observance of ritual,
stone-built zimbabwe were not all the work of Shona- etc., articulated in the functioning of Great Zimbabwe.
speakers. While it is true that the Venda built in stone, The other problem with regard to Beach’s model is its
the Ndebele and perhaps the Sena and the Tsonga are inability to identify structures such as the ruler’s
late arrivals on the Zimbabwean plateau, long inhabited house/residence and that of his immediate followers
by Shona-speakers. Secondly, Portuguese written apart from imagining that they were there. Aspects
sources are said to be unclear in linking the stone zim- such as secluded leadership, high status, ritual suicide,
 b e a c h Cognitive Archaeology and Imaginary History 65

ritual incest, etc., which are connected with some ritu- could then be set alongside those built around archaeo-
als at court, are not easily translated into units of space logical, oral historiographic, and documentary evi-
recoverable historically and archaeologically. dence. Points of correlation and divergence in spatial,
In the end the main weakness of the model is that it temporal, and cultural terms could be discerned, and a
is equally based on thin historical grounds and is thus social history could then supplement the imaginative
no different from Huffman’s ‘‘500-year-old ethno- acts of cognitive archaeology. Scholars could reassess
graphic record.’’ Clearly, post-18th-century Shona poli- (without destroying) the boundaries of space, time, and
tics were not the same as those of the 14th century. discipline at which the mazimbabwe stand, and they
Eighteenth- and 19th-century social formations were could more firmly root Great Zimbabwe itself in the
clearly more fragmented and smaller-scale (and difficult historical development of the Shona-speaking world.
to define archaeologically) than 13th- or 14th-century The methodologies of historical linguistics and com-
formations (such as Gumanye and Leopard’s Kopje), parative ethnography (Vansina 1990, Ehret 1998,
some of which led to the emergence of states. Beach’s Schoenbrun 1998) bridge the spatial and temporal par-
model implies that political processes are cyclical, and ticularity of archaeological excavation, oral historio-
change is perceived in these terms: ‘‘For the next 250 graphic evidence, and documentary (historical) records.
years, things go on in much the same way. . . .’’ The Historical linguistics and comparative ethnography,
Golden Age of the site, attained sometime during the which enter Beach’s discussion only obliquely (and,
14th century, should represent major cognitive changes even then, not as methodological keys to his main argu-
resulting in its demise and the eventual rise of the Mu- ment), require that we construct a linear historical
tapa and Torwa states. It is some of the cognitive ele- framework based on genetic language classifications.
ments of these later political formations that we must The validity of such a framework may be established by
study to understand Great Zimbabwe. The interpreta- reference to unique sound shifts and the invention of
tion of the site and the culture in general requires that new words and meanings. Inside of this classification
we employ, among other things, concepts of cultural scholars can look for the distributions of related, inter-
landscape evolution that help us understand particular locking sets of words and meanings for the cognitive
characteristics of the Shona past (Pikirayi 1998). These content of practices, ideas, and relationships which
concepts may be incorporated into the practice of ar- they think are important (such as branches of produc-
chaeology, and they should be readable in the archaeo- tion) or Shona oral historians think are important (rela-
logical record. As things stand, pre-1500 events remain tions between the living and the dead, what makes a
largely imaginary and apparently unconnected with the ruler succeed where others fail, etc.) or turn up in the
present. archaeological record (carved items, raised platforms,
and the like).
Comparative ethnographic evidence gives life to the
d av i d s c ho e n b r u n practices, ideas, and relationships for which words serve
Department of History, University of Georgia, as signs. The two methods together allow us to track
Athens, Ga. 30602-1602, U.S.A. 9 xii 1997 down an enormous variety of meanings and uses for
these signs and practices, a variety which will allow us
Beach is correct that Great Zimbabwe is an outstanding to see what was inherited and what was innovated in
site for testing interdisciplinary study. I commend him the life of Great Zimbabwe and in the lives of those who
for his careful criticism of the oral and documentary lived there and spoke a Shona language or dialect. This
materials which other scholars have mistakenly applied methodology can produce evidence which will help
in writing about Great Zimbabwe’s past, and I com- confirm or reject models like Beach’s or even Huff-
mend Huffman for producing the papers which drew man’s. If we keep the focus on Great Zimbabwe itself
from Beach the considered, knowing, and stimulating but let our gaze widen to include the entire Zimbabwe
piece published here. But even more remains to be done. Plateau, the riverine lowlands to the north, the coastal
If scholars added loanword studies, the reconstruc- lowlands to the east, the Transvaal to the south, and the
tion of bundles of interrelated cultural vocabulary, and gradually drier fringes of the Kalahari to the west, then
the proposing of etymologies to the extant classifica- we will have to consider an enormous range of linguis-
tions of Shona dialects (e.g., Ehret and Kinsman 1981), tic and ethnographic data to get a sense of the cognitive
Southeastern Bantu languages and dialects (e.g., Ownby realities of people who lived in those lands between a.d.
1985, Ehret et al. 1972), and Makuan, Nyasan, and Khoi- 1000 and 1500.
san languages (e.g., Ehret 1986), they could contextual- This method serves the study of some topics better
ize studies of Great Zimbabwe in two new ways. First, than others. Kinship systems, for example, are ex-
these additions would locate the site in the regional tremely flexible in practice, prone to having been re-
nexus of historical change expressed by linguistic diver- shaped in the course of the development of customary
sification, innovation, retention, and cross-cultural law codes in the late 19th and the early 20th century
contact. Second, they would provide a historical frame- and more usefully viewed as ideological statements
work for analysing comparatively the ethnography of than as templates for a biological reality. The same may
the Zimbabwe Plateau and beyond. That framework be said of marriage practices. Together with terms and
66 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

practices associated with political institutions, the sys- ments, but the evidence is so interwoven with the argu-
tematic study of units of social organization is better ment that it is impossible for the reader to develop any
located within the history of Shona political values. alternative interpretation. The scholarly appraisal that
Lastly, I think it important to go farther than even such an ambitious reconstruction merits and invites
Beach has gone in trying to imagine Great Zimbabwe as must thus rely on the examination of the apparent va-
a living city. He has helped us to see the growth of the lidity of each individual piece of data in the light of
archaeological remains over time. What we next need one’s own knowledge and experience. A significant
is a model of Great Zimbabwe’s life which includes score of adverse judgements can only cast doubt on the
roles for people who spoke languages other than Shona general picture without presenting anything more con-
and represented cultural traditions different from the structive. The conclusion from such a treatment is that
Shona-speaking one. In short, we need to consider the impressive facade of the ‘‘worldview’’ of Zimbabwe
whether Great Zimbabwe was the multiethnic place its society is an insecure structure whose individual ele-
involvement in trading and raiding suggests it was. ments and foundation are of doubtful solidity. Are suf-
The two comparative methods can also help here by ficient of these elements weak enough to destabilise the
revealing loanwords in Shona tongues from neighboring whole structure? One is left with the impression that
or extinct speech communities. They can also reveal there must be significance in some of the aspects he
Shona loans in, for example, Southeastern Bantu, Nya- treats, such as the symbolism of architectural decora-
san, Makuan, or Khoisan. Should such indeed be found, tion, but that his interpretation is not necessarily the
their meanings and their clustering in specific semantic right one.
fields will tell us something about the contexts for their Beach as a historian has done us a service in opening
transfer from speakers of one language group to those of up the discussion to a wider forum and showing us that
another. Recognizing unique sound changes in them Huffman’s use of documents and oral traditions is in
will help us know when, in the sequence of splits re- line with his treatment of archaeology. Blacking’s
vealed by language classifications, their transfer took (1985) critique of the domba article suggests something
place. similar for at least some of his anthropology, and it
In short, I believe that a combined program of com- would be good to have a more general assessment from
parative ethnographic and linguistic research will add an authoritative anthropologist. Meanwhile the publi-
further nuance, texture, and regional perspective to the cation of Huffman’s full treatment (1996) may bring
already considerable body of scholarship on the history even the archaeological critics out of the woodwork. It
of Great Zimbabwe and of the cultural world out of is a pity that it was not considered by Beach in the pres-
which it grew. ent article.
Beach’s own hypothetical political fairy tale has a
much narrower aim and scope but is based on a more
r ob e r t s o p e r perceptive knowledge of the working of Shona politics.
History Department, University of Zimbabwe, P.O. It is of course subject to the same limitations as he
Box MP 167, Mount Pleasant, Harare, Zimbabwe. freely admits, notably the long time lag for which he
25 viii 97 relies on a continuity of political process parallel to
Huffman’s worldview. In introducing human person-
Huffman in his writings on Great Zimbabwe and re- alities into the equation, he provides a salutary realis-
lated sites has brought an impressive breadth of re- tic and dynamic perspective to the disembodied
search to bear on extending the interpretation of the worldview, but his failure to anchor his story to any
culture beyond the limits of traditional archaeology and structural details of the different enclosures leaves it
has built up a remarkably comprehensive picture of cer- floating at some remove from the physical evidence.
tain aspects of the society. He has, however, been fortu-
nate in escaping for so long any systematic criticism of
his application of cognitive archaeology and expressive a nn b. s t a h l
space. Archaeologists in Zimbabwe who should have Department of Anthropology, State University of
tackled this have largely confined themselves to dubi- New York at Binghamton, Binghamton, N.Y. 13902-
ous mutterings among themselves and have fought shy 6000, U.S.A. 5 ix 97
of any detailed critical analysis; they have perhaps been
intimidated by the intricate weaving of the admittedly Great Zimbabwe has long captured the imagination of
equivocal archaeological evidence with anthropology Europeans, who have struggled to ‘‘make sense’’ of
and history which they feel unqualified to assess, while these spectacular stone ruins. Early literature focused
with the virtual rebirth of archaeology after indepen- on the identity of Great Zimbabwe’s makers—immi-
dence none of us had broad enough experience to chal- grant Phoenicians or indigenous Africans (Kuklick
lenge the scope of Huffman’s observations. Garlake of 1991)? Later debates focused on the economy that sus-
the old guard did indeed express his misgivings, but tained the site’s occupants—wealth generated from
only in a small guidebook. control of cattle, the Indian Ocean trade, or some com-
Huffman is adept at the selection, manipulation, and bination? At the focus of Beach’s discussion is the more
assertive presentation of evidence to support his argu- recent literature that draws on ethnographic and histor-
 b e a c h Cognitive Archaeology and Imaginary History 67

ical sources to penetrate the function and meaning of tory. Huffman does not cite them in his most recent
Great Zimbabwe and related sites. Beach’s critique con- book (Huffman 1996), which suggests that he considers
centrates on methodological issues—issues of source the issue of microchronology irrelevant. Yet time and
criticism—and how cognitive categories relate to prac- historical process are crucial to the interrogation of the
tice. Less clearly articulated is the role of theory in structuralist model that Huffman proposes, and here
shaping interpretations of Great Zimbabwe. At stake I turn to issues of source criticism to augment those
here, as Beach notes, is the issue of continuity and touched on by Beach.
change. I will comment briefly on the issue of theory, Huffman’s recent book (1996:6–8) includes a short
add some thoughts on methodology, and touch on the section on methodology in which he addresses the prob-
role of imagination in archaeology. lem of diverse sources (documents, oral tradition, and
While Huffman acknowledges certain changes in archaeology) touched on by Beach. Huffman describes
Shona society over the past 500 years (Huffman 1996: his ‘‘direct historic approach’’ as follows: ‘‘a model is
195–212), he subscribes to a structuralist model of cul- derived from the relevant ethnography of a prehistoric
ture. In his view ‘‘a relatively small set of organisational group’s living descendants, and then the model is ap-
principles operates on several levels within one society, plied back in time to the older archaeological situa-
and these can generate a wide range of features’’ includ- tions. To interpret Great Zimbabwe, then, we must first
ing spatial organization, which is thus an expression of understand more recent settlements. This procedure
‘‘a society’s world-view’’ (Huffman 1996:6). He is thus minimises and in some cases eliminates the chronolog-
predisposed to seek out similarities that suggest under- ical gap between the ethnographic model and the ar-
lying structural continuities over long periods of time. chaeological data’’ (1996:6, emphasis added). Although
This is linked to his methodological premise that Huffman disagrees with me on this (1996:6), the direct
‘‘since social and settlement organisations are different historical approach is a form of analogical reasoning,
aspects of the same thing and products of the same view one often used to stress similarities between past and
of the world, a continuity in settlement pattern is evi- present (Stahl 1993:242–43). Historical connections are
dence for continuity in social organisation and world- perceived as important in establishing the relevance of
view. Methodologically, therefore, where there is iden- an ethnographic model to the archaeological site (as
tity or at least no crucial difference between settlement Huffman suggests, to minimize or eliminate the chro-
patterns in the ethnographic present and archaeological nological gap) so that the model can be used to interpret
past, we can infer identity in world-view and social or- the site. This approach draws attention away from time
ganisation’’ (Huffman 1996:8). An important conse- and historical process—to what happened in the in-
quence of these theoretical and methodological as- tervening period (see also Hall 1984). Were the cognitive
sumptions is that time and history are relatively models of African peoples recorded by 20th-century eth-
inconsequential. Microchronology remains unimpor- nographers unmodified by historical processes over the
tant so long as the broad features of the settlement pat- past several hundred years? In order to answer the ques-
tern correlate with the cognitive model. So, as Beach tion, we need to pay far more attention to evidence from
and others (Collett, Vines, and Hughes 1992, Hall 1995) intervening centuries, including archaeological sites of
have noted, Huffman collapses centuries into a single the post-Khami period (see Hall 1984, Stahl 1994). We
temporal plane, treating Great Zimbabwe and other also need to interrogate the assumptions and contexts
sites as synchronic entities. As Hall (1995) has ob- in which ethnographic models were produced, sub-
served, broader issues are at stake here; structuralist jecting these models to source criticism (Vansina 1987,
models posit an enduring, cognitive model that is resis- 1989, 1995). Ethnographers have long stressed ‘‘tradi-
tant to change. ‘‘Is this too close for comfort to the ear- tional’’ aspects of the cultures they studied, only re-
lier assumption that Africans were incapable of change, cently paying attention to how societies are shaped by
that there could be no African history . . . ?’’ (Hall 1995: their articulation with a broader historical context
42). While Beach is clearly concerned with history and (Wolf 1982). Thus, the issue of source criticism extends
the balance of change and continuity, he too treats ‘‘the beyond the veracity of particular documents or oral tra-
Shona’’ as an enduring entity associated with a rela- ditions to a critical examination of the full range of
tively stable set of practices but is less clear than Huff- sources that undergird our interpretations of the past
man about the theoretical stance that informs his (Stahl 1993).
study. Finally, imagination has always played a role in ar-
Beach acknowledges the issue of chronology at Great chaeological and historical interpretation, a role that is
Zimbabwe, but he does not explicitly integrate Collett, increasingly acknowledged and indeed embraced (e.g.,
Vines, and Hughes’s study of architectural microchro- Spector 1993, Tringham 1991). Yet I agree with those
nology into his imaginative reconstruction. They build who believe that we are not simply free to make it up
a case for complex changes in site layout that suggest and that we must continually engage in reality checks,
‘‘changing use and significance of space’’ that ‘‘cannot assessing our imaginative scenarios against the evi-
be dismissed as events in the development of a precon- dence at hand. Those reality checks must take account
ceived plan’’ (1992:154). Beach seems sympathetic, but not just of the data that match our expectations but of
his reconstruction would have benefited from an ex- the unruly data that do not. Huffman is inclined to ig-
plicit consideration of Collett et al.’s architectural his- nore anomalies or dismiss them as superficial, idiosyn-
68 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

cratic alterations to a site’s original plan (e.g., Huffman It is important to remain clear on what this paper
1986:301). But those anomalies open up a series of alter- does and does not try to do. Clarity is not easily
native imaginative possibilities, perhaps illuminating achieved, for Great Zimbabwe is complex, and so are
struggles and tensions masked by an overarching the papers in question. Perhaps I did not make it clear
‘‘worldview.’’ Beach is sensitive to these; however, he enough—pace Loubser—that I am not simply criticis-
might have strengthened his imaginative case by link- ing Huffman’s use of oral traditions, one of the main
ing it more closely to Collett, Vines, and Hughes’s primary sources of the historian in Africa. I am criticis-
(1992) architectural history. ing many, though not all, of his uses of these and the
In closing, while scholars may disagree with the de- other great historical primary source, documents, and
tails of Huffman’s interpretation, his prolific writing aspects of the logic by which he fits these into the
moved us beyond the issues of chronology and economy tightly interlocked argument noted here by Soper.
that had dominated earlier studies, spurring us to think However, there is a more general aim than this. His-
about the site in new ways. At the same time, his con- torical primary sources require historical training,
cern with symbolism and elite structures has diverted which is why I belaboured the point that the edition of
attention away from the relationship between elites and Barbosa used by Huffman was not reliable. This is rein-
commoners to the possible tensions engendered in the forced by Liesegang: new research can revise old as-
rise of elites and to the possibility that worldview was sumptions, though the basic point about comparison of
a site of struggle. One hopes that the debate over Great texts stands. The same need for careful criticism of
Zimbabwe will not become polarized into one that fo- sources applies to oral traditions, which in 1981 were
cuses on either political process or worldview but rather being assumed to refer to Great Zimbabwe’s floruit,
generate a productive dialogue that is attentive to both. whatever Huffman writes now (Beach 1983:10). Histori-
cal training is also needed in handling negative evidence
from primary sources. Indeed, it is essential to link it to
experience, because if no historian has read or heard ev-
Reply ery source at least those who have more experience can
comment on negative evidence with more confidence
than others. This affects the question about the possible
david beach Shona origin of Venda initiation—it being agreed be-
Harare, Zimbabwe. 7 x 97 tween Loubser and me that there are many Shona links
with the Venda—and the awkward question of why
An essential first step in this reply to the comments is there is no evidence for Shona initiation and/or circum-
to establish the sequence and dating of my work and cision in historical sources.
that of Huffman (1996). I made my first comprehensive One way of testing negative evidence is to propose yet
survey of his 1980s works on Great Zimbabwe in 1991, another scenario, starting from the assumption that
as a necessary step towards a chapter on precolonial Huffman was right and that the Shona did initiate and
Zimbabwean thought (Beach 1994a:142–63). I sent the circumcise in the past. One has to explain how a Portu-
consequent paper (a first draft of this one) to Huffman. guese community that lived alongside of and intermar-
Twenty years of contact between us thus ended. At the ried with the Shona for 400 years and that was highly
same time, I learned of his forthcoming book, which sensitive to the presence of Islam never noticed the fact.
was published while this paper was already with cur- I concede that, of the hundreds of writers of thousands
rent anthropology. of documents that I and others have read in the original,
I do not believe that Snakes and Crocodiles invali- most were poorly educated and mainly interested in po-
dates this work, not least because it does not reply di- litical and economic matters, but not all can be dis-
rectly to the points made in 1991 and here. Moreover, missed thus. There were some perceptive observers
although it changes Huffman’s arrangement of his argu- among them, and their silence is significant. It is also
ment relative to the evidence—and not for the better— the duty of scholars to point out inconvenient bits of
his way of handling that evidence remains much the evidence that run counter to their argument. Checking
same, as Denbow and Bourdillon comment. Snakes and the sources again, I find exactly two Portuguese sources
Crocodiles may be the New Testament, but Old Testa- stating that the ruling dynasty of the largely Sena-
ment studies remain relevant. As it is, students can speaking territory of Barwe circumcised (and initiated?).
follow the revelations of both here and in the South The latter of these sources, both quite recent, thought
African Archaeological Bulletin, many of the commen- that this Makombe dynasty (of the wild pig totem, inci-
tators being the same. It is unfortunate that Huffman dentally) was Islamic in origin (Bravo 1937:186; Fernan-
has not responded here. Loubser’s comments are thus des Júnior 1944:22). There is also one brief statement
especially welcome, not because I regard him as an ava- that the eastern Shona dynasty of Muponda circum-
tar of Huffman—far from it—but because he is the only cised, and this was not confirmed by any other source,
commentator with direct experience of the Venda as- though one of these also claimed Muslim links with
pect of the argument and because he raises questions of Muponda (Hulley 1904; Abraham 1963:2). All of this is
evidence that must be answered before the discussion a far cry from pan-Shona initiation, and if Huffman is
moves farther afield. right then the next big question is why it stopped. Major
 b e a c h Cognitive Archaeology and Imaginary History 69

wars and invasions just won’t do. I have been showing In some ways, archaeologists do routinely build such ar-
for years that these had a limited effect, certainly no guments into their hypotheses on Great Zimbabwe.
more severe than those of Venda where initiation did Nobody seems to imagine that cloth imports only began
survive. But let us apply the assumption that the Shona at the time when the Portuguese arrived to note them,
initiated to the south-west, beyond the range of Portu- but, as Huffman (1971) has shown, physical survivals of
guese documents. In this scenario, initiation would cloth are rare and usually post-1500. For that matter,
have continued at or near the local zimbabwe until in- since 1980 I have been suggesting a military, raiding ele-
terrupted by the Ngoni invasions of the 1830s. Suppose ment as a bridge between cattle herding and control of
that, whereas the Mwari/Nwali cult continued to trade at Great Zimbabwe, long before Shona armies and
flourish in one form or another, for some reason initia- raiders appeared in the documents. This has not pro-
tion ended with the coming of Ndebele rule in 1838. voked a very strong counterattack from archaeologists,
The last initiates would have been only in early middle who can correctly point out that there is no evidence for
age by the time the missionaries arrived at Inyathi in this. Nor has my argument about Great Zimbabwean
the 1850s. Is it not odd that neither these nor the colo- sanitation problems been sniffed at (Beach 1980:42–
nial government that arrived in 1894 recorded the mem- 47).
ory of such a distinctive practice that had ended so re- However, what concerns almost all of the commenta-
cently? tors is my central argument: that the Machiavellian and
Other points about what my paper does and does not often violent Shona dynastic politics of the 18th and
attempt can be answered rather more briefly. That 19th centuries can be projected back to the time of
Great Zimbabwe developed over time, with rebuilding Great Zimbabwe, along with the rather imprecise and
and reorientation of structures, is common cause flexible pattern of thought outlined. This involves prob-
among all archaeologists except Huffman since the lems of space and time that confront Huffman and me.
1950s. My scenario simply restates this in a different How far can historical data from the north and east be
way. Nor does it try to explain everything. The problem extended to the south and south-west of the Zimbab-
with political scenarios is that they can be quite realis- wean plateau? Can historical data from after ca. 1700 be
tic for the first few stages but over time, with ever- projected back, and how far? In general, the answers are
branching possibilities, become too complex. The West- not a clear-cut yes or no, and we have to examine spe-
ern and Central Valley Enclosures stressed by Stahl do cific aspects.
show how a ruling-class political and social group re- I have long argued that Shona societies underwent
built and realigned its living space through at least five considerable changes between the 16th and the 18th
phases—if this was not a continuous process—very late century. By the latter century they were becoming
in Great Zimbabwe’s sequence. However, I cannot re- much poorer, as the decline in zimbabwe building dem-
late this to the putative houses of the dynasty in the onstrates, and more subject to intercommunity raiding,
late 13th and early 14th centuries. I do think that my as the increasing militarisation of the later Mutapa and
general scenario of successive generations building, liv- Changamire states shows (Beach 1980:148–51, 238–39,
ing in, abandoning, and reoccupying the Western and 244–47, 276). However, when it comes to internal dy-
Great Enclosures fits the picture of the Valley Enclo- nastic politics—and allowing for the fact that the
sures in Collett, Vines, and Hughes (1992). These cer- largely documentary sources on this before ca. 1700 are
tainly show how readily an inventive and flexible- less detailed than the largely oral sources for the 18th
minded group changed its ideas about the way in which and 19th centuries—the available evidence does seem
living spaces should be arranged, and I should have to show continuity. Here, pace Pikirayi and Stahl, many
made more of this. of the data on politics and thought are referenced above
Soper and Pikirayi are both concerned that my ap- (Beach 1994b:26–187, 211–20; 1994a:112–63), but this
proach cannot solve problems of identity in the occupa- can be reinforced by a vivid example.
tion and structural detail of specific enclosures. Histori- Going back to before the Portuguese established con-
ans rarely expect to be able to explain everything in trol in the north, we have the struggle between the sons
their documentary and oral sources. I would be de- of Mutapa Gatsi Rusere on his death in 1623 (Beach
lighted if, as Denbow suggests, further excavations and 1980:123; Mudenge 1988:245) and the interhouse war-
analysis could show that the rather simple categories of fare between him and two others ca. 1597–1609 (Beach
age and gender used variously by the Shona for divining 1980:125–27; Mudenge 1988:225–34). But the earliest
tablets a few centimetres long could be found in physi- case known to us, the civil war of ca. 1490–1506, is also
cal remains bridging the gap between them and stone- one of the best of the examples of the difference be-
work decorations often metres long—delighted and also tween ideology and practice (Alcáçova 1506:392–94). In
surprised. If every physical feature of zimbabwe could ca. 1490 the mutapa ‘‘Mocomba’’ faced an overmighty
be as satisfactorily explained as Huffman tries to do, it subordinate ‘‘Changamir’’ who seemed likely to usurp
would be delightful—but it should not become an over- his throne even though he was apparently not a member
riding aim. of the dynasty. The mutapa offered him the choice of
The other side of the coin to physical features that ritual suicide by poison or the elimination of his entire
we cannot explain is arguments that have no physical house. ‘‘Changamir’’ offered an enormous bribe to be
support at all. Liesegang and Pikirayi comment on this. excused this and three times refused to commit suicide.
70 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

Finally, he marched on the zimbabwe and killed the I am not sure whether the foregoing is quite what
mutapa and 21 of his 22 sons by the threat of force and Schoenbrun has in mind when he writes of ‘‘trying to
by treachery. He then ruled for four years before the sur- imagine Great Zimbabwe as a living city,’’ but I can
viving son of ‘‘Mocomba’’ rallied support from another only applaud his suggestion of further ethnographic and
state and killed him in person after a battle of more than linguistic research over a very wide area, whatever it
three days outside the zimbabwe. The sons of ‘‘Mo- does for some of the debated features of Great Zim-
comba’’ and ‘‘Changamir’’ then continued the war until babwe. Obviously, collaborative work is needed, not
at least 1506. least in establishing vocabularies that have been par-
All of this does seem to be remarkably similar to the tially recorded over 500 years. Two of the special prob-
complex struggles of more than 200 years later, yet it lems of Shona social and language studies are that
occurred at almost exactly the time that the rulers at (1) the Shona-speaking region is very large and thus few
Great Zimbabwe were imaginatively rebuilding the specialists attempt detailed comparisons with others
Western Valley Enclosure to incorporate a gold-plated and (2) the conventional wisdom is that Shona society
post and a small stone-lined tunnel between two bowl- and language are remarkably similar overall, except for
like features in the wall of a huge hut, all of which defy the Kalanga dialect cluster. The result has been a ten-
cognitive archaeological explanation so far (Collett, dency to overgeneralise on the ‘‘Shona’’ over time and
Vines, and Hughes 1992:155–57). It could of course be space and to ignore the subtler differences. Stahl cor-
argued that violent politics were confined to the north rectly notes that I have no overall theoretical stance as
and east and that things were different at Great Zim- opposed to Huffman’s. This empiricism is, I think, a by-
babwe. Yet impressive artistic and architectural product of the fact that the evidence on the Shona and
achievement are not necessarily divorced from political their neighbours over 500 years has continually warned
process: Renaissance Italy was demonstrating this—and me that what may be true for some of them is not nec-
the difference between theory and practice—at just that essarily true for all—even when their similarities en-
time. I agree that this case from the late 15th century courage generalisation.
does not prove how political process worked two centu-
ries earlier still, but I suggest that my point that it ought
to be considered as one of the variables at Great Zim-
babwe stands.
Denbow, Pikirayi, and Stahl raise the issue of con-
References Cited
flicts within and between classes at Great Zimbabwe,
a b r a h a m, d. p. 1963. ‘‘Tasks in the field of early history.’’
and Liesegang explicitly shifts our focus from the Conference on the History of Central African Peoples, Lusaka,
much-debated élite stone-built centre of the city to the pp. 1–6.
large unwalled housing areas beyond that Huffman dis- a l c á ç o v a, d. d e. 1962 (1506). ‘‘Carta de Diogo de Alcáçova
covered in the 1970s and that have revolutionised our para El-Rei, Cochin 20 novembro 1506,’’ in Documentos sobre
picture of the site. All of this fits neatly—if that word os Portugueses em Moçambique e na Africa central 1497–
1840. Edited by A. da Silva Rego and T. W. Baxter, vol. 1, pp.
is appropriate to Great Zimbabwe—with my idea of po- 388–99. Lisbon: Centro de Estudos Históricos Ultramarinos
litical process. I do not imagine that the political rival- and National Archives of Rhodesia.
ries of the owners of the different structures within the b a r b o s a, d. 1966 (1518). ‘‘Descrição da situação, costumes e
perimeter walls were divorced from the packed housing produtos de alguns lugares de Africa, ca. 1518,’’ in Docu-
mentos sobre os Portugueses em Moçambique e na Africa cen-
beyond. On the contrary, it would have been from the tral 1497–1840, vol. 5. Edited by A. da Silva Rego and T. W.
poorer housing areas that the élite house heads would Baxter, pp. 354–71, 372–81. Lisbon: Centro de Estudos His-
have drawn much of their support. I do not know if it tóricos Ultramarinos and National Archives of Rhodesia.
will be possible to link separate stone structures to the ———. 1996. O livro de Duarte Barbosa, edição crı́tica e ano-
different areas of high-density housing shown by recent tada. Vol. 1. Introdução, texto crı́tico e apéndice por M. Au-
gusta da Veiga e Sousa. Lisboa: Estudos de Históría de Carto-
drilling and survey (Sinclair et al. 1993:170–74; Beach grafia Antiga, Memórias 26. [gl]
1994a:84). This is obviously one line of research that b a r r o s, j. d e. 1945 (1552). Livro décimo da primeira decada
leads away from the polarisation feared by Stahl. In- da Asia de João de Barros. Lisbon: Agência Geral das Colónias.
deed, Liesegang raises another issue when he queries b e a c h, d. n. 1979. The Shona generation. Central African
Journal of Medicine 25:45–51.
the compatibility of a total population of ca. 18,000 ———. 1980. The Shona and Zimbabwe 900–1850. London:
with sanitation: inter- and intraclass clashes do not al- Heinemann.
ways start at the top, and if people in the high-density ———. 1983. Oral history and archaeology in Zimbabwe. Zim-
areas started to use abandoned housing areas as latrines babwean Prehistory 19:8–11.
———. 1987. Documents and African society on the Zimbab-
the result might well have been strong objections from
wean Plateau before 1890. Paideuma 33:129–45.
the descendants of the owners of these areas. Quarrels ———. 1994a. The Shona and their neighbours. Oxford: Black-
begun thus might have reached inside the perimeter well.
walls to involve the élite: this was exactly the sort of ———. 1994b. A Zimbabwean past: Shona dynastic histories
thing that occurred in the Afro-Portuguese settlements and oral traditions. Gweru: Mambo.
b l a c k i n g, j. 1985. Great Zimbabwe and domba. Man 20:
of Sena, Tete, and Zumbo in historical times (Newitt 542–45. [rs]
1973:193–94). Might urban archaeology on the Zambezi b o c a r r o, a. 1899 (ca. 1631). ‘‘Decada composta por António
tell us something about Great Zimbabwe? Bocarro, chronista de Sua Magestade, do Estado da India, dos
 b e a c h Cognitive Archaeology and Imaginary History 71

feitos dos Portugueses no Oriente,’’ in Records of south- h a r r i s, d. r. 1978. ‘‘Adaptation to a tropical rain-forest envi-
eastern Africa, vol. 3. Edited by G. M. Theal, pp. 254–342. ronment: Aboriginal subsistence in northeastern Queensland,’’
Cape Town: Government of the Cape Colony. in Human behaviour and adaptation. Edited by N. Blurton
b o u r d i l l o n, m. f. c. 1976. The Shona peoples. Gweru: Jones and V. Reynolds, pp 113–34. London: Taylor and Francis.
Mambo. [jhl, rda]
b o y d e n, s. 1980. ‘‘The need for an holistic approach to human h e l m, s. p. 1891. Report, 4 April 1891. Argief van die Neder-
health and well-being,’’ in Changing disease patterns and hu- duits Gerevormede Kerk, Cape Town.
man behaviour. Edited by N. F. Stanley and R. A. Joske, pp. h o l l e m a n, j. f. 1958. African interlude. Cape Town:
622–44. London: Academic Press. [jhl, rda] Nasionale.
b r a v o, n. s. 1937. Báruè, esboço polı́tico, histórico, admin- h o w e l l, n. 1973. ‘‘The feasibility of demographic studies in
istrativo e ethnographico da circumscripção civil do Báruè. ‘anthropological’ populations,’’ in Methods and theories of an-
Anuário da Escola Superior Colonial 18:171–94. thropological genetics. Edited by M. H. Crawford and P. L.
b r o w n, w. h. 1899. On the South African frontier. London: Workman, pp. 249–62. Albuquerque: University of New Mex-
Sampson Low. ico Press. [jhl, rda]
b u l l o c k, c. 1949. Bushman paintings, Zimbabwe and romanti- h o f f m a n, t. n. 1971. Cloth from the Iron Age in Rhodesia.
cists. NADA 26:50–53. Arnoldia 5(14):1–19.
———. 1950. The Mashona and the Matabele. Cape Town: ———. 1977. Zimbabwe: Southern Africa’s first town. Rhodesian
Juta. Prehistory 7:9–14.
c a s a l, a d o l f o y a ñ e z. 1996. Antropológia e desenvolvi- ———. 1981. Snakes and birds: Expressive space at Great Zim-
mento: As aldeias comunais de Moçambique. Lisbon: Minis- babwe. African Studies 40:131–50.
tério Ciencia e Tecnologia. [gl] ———. 1984a. ‘‘ ‘Where you are the girls gather to play’: The
c h i p u n z a, k. t. 1994. A diachronic analysis of the architec- Great Enclosure at Great Zimbabwe,’’ in Frontiers: Southern
ture of the hill complex at Great Zimbabwe. Uppsala: Societas African archaeology today. Edited by M. Hall et al., pp. 252–
Archaeologica Upsaliensis. 65. British Archaeological Reports International Series 207.
c o l l e t t, d., a. v i n e s, a n d g. h u g h e s. 1992. Dating and ———. 1984b. Expressive space in the Zimbabwe culture. Man
chronologies of the Valley Enclosures: Implications for the in- 19:593–612.
terpretation of Great Zimbabwe. African Archaeological Re- ———. 1986. Iron Age settlement patterns and the origins of
view 10:139–61. class distinction in Southern Africa. Advances in World Ar-
d a m e s, m. l. Editor. 1918. The book of Duarte Barbosa. Lon- chaeology 5:291–338.
don: Hakluyt. ———. 1987. Symbols in stone. Johannesburg: Witwatersrand
d i c k i n s o n, r. 1971. António Fernandes: A reassessment. Rho- University Press.
desiana 25:45–52. ———. 1996. Snakes and crocodiles: Power and symbolism in
e d w a r d s, w. 1929. From birth to death. NADA 7:16–42. ancient Zimbabwe. Johannesburg: Witwatersrand University
e h r e t, c h r i s t o p h e r. 1986. Proposals on Khoisan reconstruc- Press. [mfcb, jd, rs, abs]
tion. Sprache und Geschichte in Afrika 7:105–30. [ds] h u f f m a n, t. n., a n d j. c. v o g e l. 1991. The chronology of
———. 1998. The classical age of eastern and southern Africa: Great Zimbabwe. South African Archaeological Bulletin 46:
A history, 1000 B.C. to A.D. 300. Charlottesville: University 61–70.
Press of Virginia. In press. [ds] h u l l e y, t. b. 1904. Native Commissioner, Umtali, to Chief
e h r e t, c h r i s t o p h e r, m. b i n k, t. g i n i n d z a, e. Native Commissioner, Salisbury, 19 January 1904. National Ar-
g o t s c h a l l, b. h a l l, m. h l a t s c h w a y o, d. j o h n - chives, Zimbabwe, 3/33/8.
s o n, a n d r. p o u w e l s. 1972. Outlining Southern African j a c k s o n, h. m. g. 1906. Native Commissioner, Matobo, to
history: A reconsideration. Ufahamu 3:9–27. [ds] Chief Native Commissioner, Bulawayo, 28 May 1906. National
e h r e t, c h r i s t o p h e r, a n d m a r g a r e t k i n s m a n. 1981. Archives, Zimbabwe, A 3/18/28.
The dialect classification of Shona and its implications for Iron k e l l y, r. c. 1968. Demographic pressure and descent group
Age history in southern Africa. International Journal of Afri- structure in the New Guinea highlands. Oceania 39:36–63.
can Historical Studies 14:401–43. [ds] [jhl, rda]
e l l i o t t, f. g. 1906. Native Commissioner, Selukwe, to Chief k u k l i c k, h e n r i k a. 1991. ‘‘Contested monuments: The poli-
Native Commissioner, Bulawayo, 19 May 1906. National Ar- tics of archaeology in southern Africa,’’ in Colonial situations:
chives, Zimbabwe, A 3/18/28. Essays on the contextualization of ethnographic knowledge.
e l l i o t t, w. a. 1894. ‘‘The Ma-Shuna,’’ in Among the Mata- Edited by George W. Stocking Jr., pp. 135–69. Madison: Univer-
bele. Edited by D. Carnegie, pp. 109–20. London: Religious sity of Wisconsin Press. [abs]
Tract Society. k u s s, h. 1882. Notes sur la géographie de quelques régions
f e r n a n d e s j ú n i o r, j. 1944. Divagações históricas sobre o voisines du Zambèze. Bulletin de la Société de Géographie de
distrito de Tete. Arquivo Histórico de Moçambique, Secção Es- Paris 7:365–83.
pecial, 218b, Cota SE 2 III, pp. 1–107. l a n n i n g, r. 1906. Native Commissioner, Inyati, to Chief Na-
f r o b e n i u s, l. 1931. Erythräa. Berlin: Atlantis. tive Commissioner, Bulawayo, 21 April 1906. National Ar-
g a r l a k e, p. s. 1973. Great Zimbabwe. London: Thames and chives, Zimbabwe, A 3/18/28.
Hudson. l e i j e r, m. 1897. Assistant Native Commissioner, Mazoe, to
———. 1992 (1982). Revised edition. Great Zimbabwe described Chief Native Commissioner, Salisbury, 28 June 1897. National
and explained. Harare: Zimbabwe Publishing House. [gl] Archives, Zimbabwe, LO 5/4/5.
g e l f a n d, m. 1960. Mahungbwe. NADA 37:36–37. l i z o t, j. 1977. Population, resources, and warfare among the
g o m e s, a. 1959 (1648). Viagem que fez o Padre António Gomes Yanomami. Man 12:497–517. [jhl, rda]
da Companhia de Jesus ao Império de Manamotapa. Studia 3: l o u b s e r, j. h. n. 1989. Archaeology and early Venda history.
155–242. South African Archaeological Society Goodwin Series 6:54–61.
g o o d m a n, a. h. 1993. On the interpretation of health from m c k e o w n, t. 1988. The origins of human disease. Oxford:
skeletal remains. current anthropology 34:281–88. Blackwell. [jhl, rda]
[jhl, rda] m a u c h, k. 1871–72. ‘‘Journals,’’ in The journals of Carl Mauch
h a l l, m a r t i n. 1984. The myth of the Zulu homestead: Ar- 1869 to 1872. Edited by E. E. Burke. Salisbury: National Ar-
chaeology and ethnography. Africa 54(1):65–79. [abs] chives of Rhodesia.
———. 1995. ‘‘Great Zimbabwe and the Lost City: The cultural m t e t w a, r. m. g. 1976. The ‘‘political’’ and economic history
colonization of the South African past,’’ in Theory in archaeol- of the Duma people of south-eastern Rhodesia from the early
ogy: A world perspective. Edited by P. J. Ucko, pp. 28–45. Lon- eighteenth century to 1945. D. Phil. diss., University of Rhode-
don: Routledge. [abs] sia, Salisbury, Rhodesia.
72 c ur r e n t a n t hr o p o l o g y Volume 39, Number 1, February 1998

m u d e n g e, s. i. g. 1988. A political history of Munhumutapa s t a h l, a n n b. 1993. Concepts of time and approaches to ana-
c. 1400–1902. Harare: Zimbabwe Publishing House. logical reasoning in historical perspective. American Antiquity
n e w i t t, m. d. d. 1973. Portuguese settlement on the Zam- 58:235–60. [abs]
besi. London: Longman. ———. 1994. Change and continuity in the Banda area, Ghana:
o w n b y, c a r o l a n p. 1985. Early Nguni history: The linguis- The direct historical approach. Journal of Field Archaeology
tic evidence and its correlation with archaeology and oral tradi- 21:181–203. [abs]
tion. Ph.D. diss., University of California, Los Angeles, Calif. s t a n l e y, w. e. j. Editor. 1886. A description of the coasts of
[ds] East Africa and Malabar. London: Hakluyt Society.
p i k i r a y i, i n n o c e n t. 1993. The archaeological identity of s t o r r y, g. 1974. Chitungwiza revisited. Rhodesian Prehistory
the Mutapa state: Towards an historical archaeology of north- 7:10–13.
ern Zimbabwe. Uppsala: Studies in African Archaeology 6. [gl] s t u a r t, c. t. 1906. Native Commissioner, Gwelo, to Chief
———. 1998. The misconceived landscape: Written sources, car- Native Commissioner, Bulawayo, 16 February 1906. National
tographic specimens, and archaeology in northern Zimbabwe. Archives, Zimbabwe, A 3/18/28.
Uppsala: Acta Archaeologica Upsaliensis. In press. [ip] s u m m e r s, r. 1963. Zimbabwe, a Rhodesian mystery. Johannes-
p o s s e l t, j. w. 1904. Native Commissioner Charter to Chief burg: Nelson.
Native Commissioner, Salisbury, ca. 1 January 1904. National t h e a l, g. m. Editor. 1898. Records of south-eastern Africa.
Archives, Zimbabwe, N 3/33/8. Vol. 1. Cape Town: Government of Cape Colony.
r a m u s i o, g. b. Editor. 1563. Delle navigatione et viaggi. t h o m a s, t. m. 1906. Acting Native Commissioner, Insiza, to
Venice: Giunti. Chief Native Commissioner, Bulawayo, 16 May 1906. National
s a n t o s, j. d o s. 1899 (1609). ‘‘Ethiopia oriental,’’ in Records Archives, Zimbabwe, A 3/18/28.
of south-eastern Africa, vol. 7. Edited by G. M. Theal, pp. t r a c e y, h. 1934. The bones. NADA 12:23–25.
1–182. Cape Town: Government of the Cape Colony. ———.. 1963. The hakata of Southern Rhodesia. NADA 40:
s c h m i d t, p e t e r r. 1996. ‘‘Rhythmed time and its archaeolog- 105–7.
ical implications,’’ in Aspects of African archaeology: Papers t r i n g h a m, r u t h. 1991. ‘‘Households with faces: The chal-
from the 10th Pan-African Association for Prehistory and Re- lenge of gender in prehistoric architectural remains,’’ in Engen-
lated Studies. Edited by G. Pwiti and R. Soper, pp. 655–69. dering archaeology: Women and prehistory. Edited by Joan M.
Harare: University of Zimbabwe Publications. [ip] Gero and Margaret W. Conkey, pp. 93–131. Oxford: Basil Black-
s c h o e n b r u n, d a v i d l. 1998. A green place, a good place: well. [abs]
Agrarian change and social identity in the Great Lakes region v a n s i n a, j. 1985. Oral tradition as history. Madison: Univer-
to the 15th century. Portsmouth, N.H.: Heinemann. In press. sity of Wisconsin Press.
[ds] ———. 1987. The ethnographic account as a genre in central
s h a n k s, m i c h a e l, a n d c h r i s t o p h e r t i l l e y. 1992. 2d Africa. Paideuma 33:433–44. [abs]
edition. Reconstructing archaeology: Theory and practice. Lon- ———. 1989. Deep-down time: Political tradition in central
don: Routledge. [jd] Africa. History in Africa 16:341–62. [abs]
s i c a r d, h. v o n. 1959. The hakata names. NADA 36:26. ———. 1990. Paths in the rainforest: Toward a history of politi-
s i l v a, j. j. d a. 1831. Acressentamento a memória sobre a doa- cal tradition in Equatorial Africa. Madison: University of Wis-
ção do território Bandire . . . 5 agosto 1831. Arquivo Histórico consin Press. [ds]
Ultramarino, Lisbon. Moçambique Maço 21. ———. 1995. Historians, are archeologists your siblings? History
s i n c l a i r, p. j. j. 1987. Space, time, and social formation: in Africa 22:369–408. [abs]
A territorial approach to the archaeology and anthropology of v e l o s o, g. 1964 (1512). ‘‘Apontamentos de Gaspar Veloso, es-
Zimbabwe and Mozambique c. 0–1700 A.D. Uppsala: Societas crivão da feitoria de Moçambique, enviados a El-Rei, c. 1512,’’
Archaeologia Upsaliensis. in Documentos sobre os Portugueses em Moçambique e na Af-
s i n c l a i r, p. j. j., i. p i k i r a y i, g. c. p w i t i, a n d r. rica central 1497–1840, vol. 3. Edited by A. da Silva Rego and
s o p e r. 1993. ‘‘Urban trajectories on the Zimbabwean pla- T. W. Baxter, pp. 180–89. Lisbon: Centro de Estudos Históricos
teau,’’ in The archaeology of Africa: Food, metals, and towns. Ultramarinos and National Archives of Rhodesia.
Edited by T. Shaw et al. London: Routledge. w e n t z e l, p. j. 1983. The relationship between Venda and
s p e c t o r, j a n e t d. 1993. What this awl means: Feminist ar- Western Shona. Pretoria: University of South Africa.
cheology at a Wahpeton Dakota village. St. Paul: Minnesota w o l f, e r i c. 1982. Europe and the people without history.
Historical Society Press. [abs] Berkeley: University of California Press. [abs]