Material Culture and dialectics of Identity and Power:
Material Culture and dialectics of Identity and Power: Towards a Historical Archaeology of the Rozvi in South-Western Zimbabwe By Machiridza Lesley H. A dissertation submitted in fulfilment of the requirements for the degree MA (Archaeology) In the Department of ARCHAEOLOGY AND ANTHROPOLOGY at the UNIVERSITY OF PRETORIA FACULTY OF HUMANITIES SUPERVISOR: Professor Innocent Pikirayi APRIL 2012 © University of Pretoria To God, my family and the Rozvi people ii ABSTRACT The desire to attach identities (e.g. ethnic, gender, race, class, nationality etc.) to material culture has always featured at the core of archaeological inquiry. Archaeologists share the view that material culture is an active cultural agent that can reflect complex ideas that operated in the minds of prehistoric agents when carefully examined. These ideas were often shaped by dynamic social interactions and they sometimes manifested through stylistic patterns or material culture variation at archaeological sites. In Zimbabwe, various archaeological identities have been defined but Rozvi identities remain the most problematic. This study, therefore, revisits the Rozvi subject in the light of contemporary ideas on ethnicity, agency and material culture. Rozvi identities are probed from material culture at Khami and Danamombe sites, which are also linked with the Torwa historically, thus historical archaeology largely informs this investigation. Through documentary and fieldwork research results, I found that Rozvi identity construction processes were extremely fluid and sophisticated. Diverse elements of culture (both tangible and intangible) were situationally invoked to mark Rozvi ethnic boundaries. Whilst ceramics at Khami were diverse and complex, Danamombe pottery became more simple, less diverse or homogenous. Polychrome band and panel ware however still occurred at Danamombe, but in very restricted numbers. Perhaps the production and distribution of polychrome wares was controlled by Rozvi elites as part of their ideology and power structures. On the contrary, beads, dry-stone walls, and status symbols became more diversified at Danamombe than at Khami. However, dhaka structures show no difference between the two research sites, where mundane stylistic differences manifesting at Danamombe, the former Rozvi capital, are perceived as demonstrative of ethnic objectification. KEYWORDS: material culture, style, identity, ethnicity, agency, social interaction, historical archaeology, Rozvi iii DECLARATION I declare that this dissertation is my own unaided work. It is submitted for the degree of MPhil in the University of Pretoria, Pretoria. It has not been submitted before for any other degree or examination in any other university. Lesley Hatipone Machiridza day of 2012 iv TABLE OF CONTENTS List of Figures List of Plates List of Tables …………………………………………………………………….. …………………………………………………………………….. …………………………………………………………………...... viii x xi CHAPTER ONE …………………………………………………………………….. Background and Conceptual Approaches ………………………………………… 1.1 Introduction ………………………………………………………………….. 1.2 Problem formulation ………………………………………………………… 1.3 Research aims ………………………………………………………………... 1.4 Definition of terms ………………………………………………………… 1.4.1 Nyai/VaNyai ……………………………………………………. 1.4.2 Torwa …………………………………………………………... 1.4.3 Guruuswa ………………………………………………………. 1.5 An Overview of Rozvi Socio-Political Complexity ………………………… 1.6 The Zimbabwe culture: an overview from Great Zimbabwe to Rozvi state…… 1.7 Organisation of this work ……………………………………………………. 1.8 Conclusion …………………………………………………………………... 1 1 1 5 6 8 8 9 9 10 17 20 22 CHAPTER TWO ……………………………………………………………………. Physiography of south-central and western Zimbabwe …………………………………………………………………………………………... 2.1 Introduction ………………………………………………………………….. 2.2 Topography and Drainage …………………………………………………... 2.3 Geology and Soils …………………………………………………………... 2.4 Climate ………………………………………………………………………. 2.5 Vegetation …………………………………………………………………... 2.6 Current land use patterns ……………………………………………………. 2.7 South-central and western Zimbabwe cultural landscape: opportunities and constraints …………………………………………………………………………. 2.8 Overview …………………………………………………………………….. 23 CHAPTER THREE ………………………………………………………………… Analytical frameworks ……………………………………………………………… 3.1 Introduction ………………………………………………………………….. 3.2 Defining Historical Archaeology ……………………………………………. 3.3 Contextualising notions of Identities, Ethnicity, Power, Agency and Material Culture……………………………………………………………………………… 3.4 Varying perspectives on ethnicity in archaeology……………………………... 3.5 Defining agency in archaeology……………………………………………….. 3.6 Towards a definition of Rozvi identities …………………………………….. 3.7 Conclusion …………………………………………………………………... 44 44 44 45 …………………………………………………………………... 68 CHAPTER FOUR v 23 23 25 28 30 34 35 37 42 51 53 59 61 66 Research Methodology ……………………………………………………………… 4.1 Introduction ………………………………………………………………….. 4.2 Written sources ……………………………………………………………… 4.3 Oral traditions ………………………………………………………………... 4.4 Ethno-historical sources …………………………………………………….. 4.5 Archaeological approaches ………………………………………………... 4.5.1 Ceramic analysis ………………………………………………... 188.8.131.52 Vessel form …………………………………………………….. 184.108.40.206 Lipform ………………………………………………………… 220.127.116.11 Fabric …………………………………………………………... 18.104.22.168 Exterior colour ………………………………………………….. 22.214.171.124 Decoration technique …………………………………………... 126.96.36.199 Decoration placement …………………………………………... 188.8.131.52 Decoration motif ………………………………………………... 184.108.40.206 Sherd thickness ………………………………………………… 4.5.2 Analysis of Beads ………………………………………………………... 220.127.116.11 Classification …………………………………………………... 18.104.22.168 Material ………………………………………………………… 22.214.171.124 Techniques of manufacture …………………………………….. 126.96.36.199 Shape ……………………………………………………………. 188.8.131.52 Structure ………………………………………………………... 184.108.40.206 Colour …………………………………………………………... 220.127.116.11 Diaphaneity …………………………………………………….. 18.104.22.168 Decoration ……………………………………………………… 22.214.171.124 Size ……………………………………………………………… 126.96.36.199 Bead analysis ………………………………………………….. 4.4.3 Architectural analysis of dry-stone walls ………………………………... 188.8.131.52 Wall number ……………………………………………………. 184.108.40.206 Material ………………………………………………………… 220.127.116.11 Form ……………………………………………………………. 18.104.22.168 Techniques ……………………………………………………… 22.214.171.124 Entrance ………………………………………………………… 126.96.36.199 Decoration ……………………………………………………… 188.8.131.52 Classification and analysis ……………………………………… 4.5.4 Dhaka structures …………………………………………………………... 4.5.5 Miscellaneous finds ………………………………………………………... 4.6 Conclusion …………………………………………………………………... 68 68 69 70 72 74 74 75 76 77 77 78 78 78 78 78 79 79 80 80 81 81 81 82 82 82 82 82 83 83 83 84 84 84 84 84 85 CHAPTER FIVE ……………………………………………………………………. Documentary and ethnographic data on the Rozvi and Torwa ………………….. 5.1 Introduction ………………………………………………………………….. 5.2 Data from Portuguese documents …………………………………………... 5.3 Oral traditions ………………………………………………………………... 5.4 Traditions of Rozvi origins and power ………………………………………... 5.5 Rozvi traditions of the Mwari religion…………………………………………. 5.6 Rozvi traditions of chieftainship and land rights………………………………. 86 86 86 86 91 92 92 95 vi 5.7 Rozvi traditions centred on totemic identity…………………………………… 5.8 Traditions of the Rozvi dialect as a symbol of identity………………………... 5.9 Traditions of Rozvi military power…………………………………………….. 5.10 Data from creative writing……………………………………………………. 5.11 Rozvi ethno-historic data……………………………………………………... 5.12 Overview of Rozvi identity markers………………………………………….. 98 100 100 101 106 112 CHAPTER SIX ……………………………………………………………………… Archaeological Data Presentation: Material culture as Rozvi witnesses ………... 6.1 Introduction ………………………………………………………………….. 6.2 Ceramic analysis and results ………………………………………………… 6.3 Bead analysis and results ……………………………………………………. 6.4 Dry-stone wall analysis and results ………………………………………….. 6.5 Dhaka structure description and analysis ……………………………………. 6.6 Description of miscellaneous finds ………………………………………….. 6.7 Conclusion ………………………………………………………………… 114 114 114 115 141 147 154 157 157 CHAPTER SEVEN ………………………………………………………………… Discussion and Conclusions: Rozvi identities in the past and present …………... 7.1 Introduction ………………………………………………………………….. 7.2 Contextualising Torwa and Rozvi ethnic identities ………………………… 7.3 An agency oriented archaeological characterization of Rozvi ethnicity ……. 7.4 Challenges of defining identities and power archaeologically ……………… 7.5 Directions for future research ………………………………………………... 159 159 159 160 167 173 174 REFERENCES …………………………………………………………………….. APPENDICES ……………………………………………………………………… Appendix A: Data capture sheet for Torwa/Rozvi ceramics ……………………. Appendix B: Data capture sheet for Torwa/Rozvi beads ………………………... Appendix C: Architectural data capture sheet for Khami and Danamombe ……. Appendix D: Interview questions for the Rozvi ………………………………… Appendix E: Detailed descriptions of analysed vessels from Khami and Danamombe ……………………………………………………………………….. 176 a a b c d vii e List of Figures Figure Fig. 1.1 Map showing the distribution of Khami-phase or Torwa sites that later defined the core Rozvi state during the late 17th century (After Pikirayi 2001)……………………………………………………………… 2 Fig. 1.2 Map indicating a much broader distribution of Rozvi settlements both north and west of Mafungautsi plateau during the early 18th century (After Pikirayi 2001)………………………………………………………. 3 Fig. 1.3 Map highlighting Butua/Guruuswa modified from Petrus Plancius, 1594 (in Randles 1958)………………………………………………………… 10 Fig. 2.1 Location of the research Area …………………………………… 24 Fig. 2.2 Location of Khami, Danamombe and other Khami phase sites …….............................................................................................................. 25 Fig. 2.3 A satellite topographical map of Zimbabwe adopted from (www.wikipedia.org/wiki/Geography_of_Zimbabwe) …………………… 27 Fig. 2.4 The Agro-ecological regions of Zimbabwe, dotted areas in the map are regarded as unsuitable (Adopted from Vincent and Thomas 1961) 33 Fig. 4.1 Vessel lipform profiles (After Phillipson 1976)………………….. 77 Fig. 4.2 Bead shape profiles (Modified from Spector 1976)……………… 81 Fig. 6i Graph showing the frequency of vessel shape categories for Khami and Danamombe ………………………………………………………....... 118 Fig. 6a Pots with tall necks and pots with concave necks ………………… 119 Fig. 6b Pots with short vertical necks and out-turned rims ……………….. 119 Fig. 6c Pots with in-sloping and out-turned rims …………………………. 119 Fig. 6d Pots with short concave necks ……………………………………. 119 Fig. 6e Constricted pots with in-sloping rims …………………………….. 120 Fig. 6f Constricted pots with no necks …………………………………… 120 viii Fig. 6g Constricted pots with no necks …………………………………… 120 Fig. 6h Constricted pots with no necks …………………………………... 120 Fig. 6i Constricted bowl ………………………………………………….. 121 Fig. 6j Slightly constricted bowls and an open shallow bowl ……………. 121 Fig. 6k Deep hemispherical and deep straight sided bowls ……………… 121 Fig. 6ii Graph showing lipform frequencies from Khami and Danamombe 122 Fig. 6iii Graph showing vessel fabric frequencies from Khami and Danamombe ………………………………………………………………. 123 Fig. 6iv Graph showing the distribution of exterior vessel colour …………………………………………………………………………….. 124 Fig. 6v Bar graph showing sherd thickness for Khami and Danamombe 125 Fig. 6vi Graph showing ceramic classes at Khami and Danamombe …… 139 Fig. 6vii Graph showing bead colour frequencies from Khami and Danamombe ……………………………………………………………… 142 Fig 6viii Graph showing frequencies for bead classes at Khami and Danamombe................................................................................................... 146 Fig. 6ix Graph showing wall height frequencies at the research sites ……. 149 Fig. 6x Graph showing the frequencies of wall types at Khami and Danamombe ……………………………………………………………….. 152 Fig. 6o An artistic impression by Posselt of a prehistoric dhaka hut with a cross-wall and internal feature, done following excavations at Great Zimbabwe (After Ndoro 2001)…………………………………………….. 155 Fig. 6p Plan of Khami ruins Hut Cb3 with both a cross-wall and outside radial walls (After Robinson 1959)………………………………………... 156 Fig. 6q Dhaka hut from Khami with outside radial walls but no internal cross-wall (After Robinson 1959)…………………………………………. 156 ix List of Plates Plate Page Plate 5.1 Some of the senior Rozvi members and an assistant researcher who participated in the interviews conducted in Bikita……………………………………………………… 107 Plate 5.2 Images showing Duma custodians, and a Rozvi representative to Mabweadziva, demonstrating rain-making rituals at Tsime Rapfupajena the sacred well ………………... 109 Plate 5.3 “Ndiro yedanda” the wooden plate that is now used by the Rozvi when they perform their rain-making traditions ……………………………………………………… 110 Plate 5.4 Chironde mountain, one of the several mountains that the Rozvi attach their historical origins on. There are claims that several Rozvi burials are scattered on this mountain, but these have not yet been ascertained………………………………………… 112 Plate 5.5 The Rozvi “dare” court at Shupai Makotore‟s homestead with a single R-style dry-stone wall. This is where all important Rozvi matters are dealt with and it constitutes the starting point for rain-making customs associated with Mabweadziva ……………….. 112 Plate 6.1 Reconstructed globular pot, Khami site museum……………………………….. 139 Plate 6.2 Constricted polychrome pot, Danamombe site museum ……………………….. 139 Plate 6.3 Polychrome vessel at Danamombe site museum ……………………………….. 140 Plate 6.4 Pecked stones Khami site museum ……………………………………………... 153 Plate 6.5 Remains of a massive dhaka structure at the summit of Danamombe‟s Main platform …………………………………………………………………………………… 155 x List of Tables Table Page Table 2.1 Summary of temperature variations within the broad research area for the year 2010 (Source: Zimbabwe Meteorological Department) …………………. Table 2.2 A summary of Agro-ecological zones (After Vincent and Thomas 1961) 31 32 Table 6.1 Khami Hill Complex vessel shape categories ………………………………….. 117 Table 6.2 Danamombe Kitchen midden trench vessel shape categories …………………. 118 Table 6.3 Summary of lipforms from Khami and Danamombe ………………………….. 122 Table 6.4 Summary of vessel fabric from Khami and Danamombe ……………………… 123 Table 6.5 Summary of vessel exterior colour from Khami and Danamombe ……………. 124 Table 6.6 Summary of sherd thickness from Khami and Danamombe …………………... 125 Table 6.7 Summary of decoration technique and technique combination at Khami ……... 127 Table 6.8 Summary of decoration technique and technique combination at Danamombe 127 Table 6.9 Summary of vessel types Khami Hill Complex ………………………………... 138 Table 6.9.1 Summary of vessel types Danamombe Kitchen midden trench ……………... 138 Table 6.9.2 Summary of glass bead colours Khami Hill Complex ………………………. 141 Table 6.9.3 Summary of glass bead colours Danamombe Kitchen midden trench ………. 142 Table 6.9.4 Summary of bead classes Khami Hill Complex ……………………………... 145 Table 6.9.5 Summary of bead classes Danamombe Kitchen midden trench ……………... 145 Table 6.9.6 Summary of wall height Khami and Danamombe …………………………… 148 Table 6.9.7 Summary of dry-stone wall decoration at Khami and Danamombe …………. 151 Table 6.9.8 Summary of dry-stone wall types at Khami and Danamombe ………………. xi 152 ACKNOWLEDGEMENTS There were many driving forces behind the conceptualization and successful completion of this dissertation. Firstly, I would like to acknowledge God for providing the life, a perfect health and the willpower to always forge ahead. My initial interest on the Rozvi was inspired by my late grandmother who was a Rozvi, belonging to the Basvi house in Wedza district. She passionately narrated several mythical accounts about the Rozvi, which triggered my earliest interest on the history of the group. I later joined the University of Zimbabwe where my interest on the subject blossomed due to the unequivocal academic support I got from Dr Gerald Mazarire, Dr Robert Soper, Dr Seke Katsamudanga, Dr Paul Mupira and Professor Gibert Pwiti. These scholars gave me the initial support to pursue the subject seriously from an archaeological perspective. When I initially started this investigation in 2004, I doubted myself but the numerous discussions I held with these individuals spurred me on remarkably. After completing the initial stage, I felt there was nothing more to research on the Rozvi but Professor Innocent Pikirayi greatly renewed my strength and passion for the Rozvi archaeology. I am especially grateful to Professor Pikirayi because he first facilitated my registration for this Masters degree with the University of Pretoria in 2009. His support came at a time when there was no hope because of the serious financial problems I was facing at the time. I formally started this particular research in March 2010 and Professor Pikirayi did not tire in sourcing more funds for my studies. I therefore benefitted financially through a university bursary which he organised to cover my tuition and other basic research needs. I would also like to thank him for the numerous research materials that he made available whenever I failed to access it from Zimbabwean libraries. He also dedicated his precious time towards discussing and making critical comments aimed at improving the quality of this work. Thus this research took shape due to the critical advice received from my supervisor Professor Pikirayi. Although he may not agree with everything written, I sincerely hope he appreciates my debt to his advice. I however take full responsibility for all errors, weak points and poor arguments presented in this study. I want to thank Dr Munyaradzi Manyanga whose concern for my academic progress and welfare has been incessant. I really cherish the provocative questions and comments he made throughout xii the course of this research. I would also like to acknowledge the hospitality he provided whenever I visited South Africa. Indeed his office and home provided a tranquil setting for discussions, thinking and writing during my stay in South Africa. At the Midlands State University, I benefitted from the unwavering support of my departmental Chairperson Rtd. Lt. Col. Niya Mtombeni. Unfortunately, time and space does not permit me to list all the sacrifices he made to see the completion of this study. I also am grateful to the Research Board Committee at the Midlands State University for releasing grants that enabled me to conduct several field trips. For assistance in conducting research among the Rozvi in Bikita, I thank Chief Marozva and Chief Mukanganwi. I especially want to thank several Rozvi members who gladly participated in the interviews and provided very insightful information for this study. At the National Museums and Monuments of Zimbabwe (NMMZ), I would like to thank the executive director, Dr Godfrey Mahachi for granting me the permission to access all museum collections, and to research at Khami and Danamombe. Happinos Marufu, Godhi Bvocho, Henry Chiwaura, Clapperton Gutu, Charity Ndlovu, Lonke Nyoni, and Blessed Magadzike did a sterling job when I conducted research in museums. Several Midlands State University students volunteered to assist in the collection of data and this enabled the smooth running of this project. I am indebted to all these individuals for their unique contributions to this study. Special thanks also go to K.C. Muringaniza and Russell Kapumha for producing some of the maps that were used in this study. I also want to extend my gratitude to the staff at National Archives of Zimbabwe (NAZ) for assisting me to access all documents and files relevant to my research. I had the honour to discuss and share ideas related to this research with other scholars like Dr William Zivenge, Dr Nhamo A. Mhiripiri, Mr Godfrey T. Ncube, and Ancila Nhamo. My family has always been a pillar of support throughout the course of my studies, I appreciate the emotional, financial and spiritual backing that they have always provided. Finally, I have to make mention of a special person who made very special contributions during the entire duration of this research. That person happens to be Rozvi, and I really appreciate the fact that she understood and respected my studies by tolerating my prolonged absence from home. The person who made these sacrifices is none other than my beloved wife. At times it is impossible to xiii remember everyone who does well to us, so I would like to end by acknowledging who-so-ever might have been let down by my poor memory. xiv Chapter One Background and Conceptual Approaches “Examination… aspects of social and cultural identity- especially when bolstered by textual evidence for the ancient meanings assigned to these categories- should prove to be a productive area for future study” (Stein 1998:14). 1.1 Introduction This study explores the history and archaeological identity of the Rozvi people of south-western Zimbabwe, who have remained a subject of speculation from the early 18th century until the present. In particular, the unquestioned association of Khami-phase sites such as Khami and Danamombe with both the Torwa and Rozvi people has led to more academic confusion and an over-simplistic understanding of Rozvi material indices. In fact, the Rozvi are a very complex and dynamic socio-political grouping that first mobilized in north-eastern Zimbabwe before spreading south-west wards to invade and assimilate the already weakened Torwa community during the 1690s. Around the 18-19th centuries there were more Rozvi dispersals as some of the members migrated from their core south-western territories into the north, west, east and southern parts of the Zimbabwean plateau (see Figs 1.1 and 1.2). Those who moved southwards become Singo among the Venda in South Africa, in the west and north-west they became Kalanga, Nambya and Shangwe respectively. Those in the south-east became Karanga, in the east Ndau, and in the broader eastern and central parts they became part of the Zezuru and Manyika communities. It therefore becomes difficult to ignore the Torwa and the rest of the postRozvi houses now scattered across the Zimbabwean plateau and beyond, when tracing Rozvi ethnicity because all their customs and traditions still constitute a sophisticated Rozvi system. Nonetheless, only the Gumunyu houses in Bikita were investigated in this study because it was not feasible to explore all the Rozvi descendants cited above. 1 Figure 1.1. Map showing the distribution of Khami-phase or Torwa sites that later defined the core Rozvi state during the late 17th century (After Pikirayi 2001). 2 Figure 1.2. Map indicating a much broader distribution of Rozvi settlements both north and west of Mafungautsi plateau during the early 18th century (After Pikirayi 2001). Initially, when this research was conceptualised at Honours level (Machiridza 2005), the assumption was since the Rozvi are well remembered in oral traditions, recorded in documentary sources and strongly linked to well known archaeological sites, then defining their archaeological identities would be fairly easy. My objectives therefore, were basically to critique previous views concerning the Rozvi, identifying associated material culture through the use of historical documents and analysing this material culture from archaeological sites attributed to the Rozvi. As it turned out, defining the Rozvi archaeologically was no easy task because the most powerful Rozvi dynasty simply took over many Khami type sites that were formerly inhabited by the Torwa. Hence, it was difficult to clearly separate Torwa and Rozvi material culture on the basis of my initial narrow theoretical and methodological approach to ethnic 3 identities and their relationship to material objects. Nonetheless, the Rozvi are generally defined as a broad political grouping that rose to power during the 1670s in north-eastern Zimbabwe, from people of the moyo (heart) totem under the leadership of Changamire Dombo (Beach n.d., 1980, 1994b). Although the Rozvi are well remembered as the last dynasty that ruled from the well known Khami type sites, their wide distribution across the Zimbabwean plateau may not have been synonymous with their political power (Beach 1994b). It was only the major Changamire Rozvi group of the 1690s-1830s that migrated to settle in south-central and western parts of Zimbabwe that had significant influence, otherwise the rest were simply fragmented small groups (Beach 1994b). Thus the south-western Rozvi became the hub of political power in the Rozvi confederacy, which was a decentralised state system controlling certain territories governed by Rozvi chiefs from around 1695 up to the 1830s (Beach 1980; Mazarire 2003, 2009). These Rozvi alliances survived for quite some time due to manipulation of certain elements of tradition until the Nguni groups arrived in the 1830s. According to Pikirayi (1997a) archaeologists have assumed that the Rozvi are too recent a phenomenon to deserve any special archaeological attention and hence the obvious archaeological bias towards much earlier periods of the Zimbabwean past. This archaeological bias has mainly been influenced by the belief in the primacy of the written text for periods regarded as recent or historical. Thus only historians have been grappling with Torwa and Rozvi issues since the 1970s, such that their contributions now form the bulk of our reference material on these subjects (see Beach n.d., 1980, 1983a, 1994a, 1994b, 1995; Bhila 1972, 1982a, Mazarire 2003, 2009; Mudenge 1974a, 1974b, 1988; Ncube 2004). Perhaps the lack of meaningful archaeological research has forced historians to take a step further, and speculate about the Rozvi and associated archaeological signatures (e.g. Beach 1980, 1995; Mudenge 1974a:23). Despite the significant strides made by historians, the Torwa and Rozvi periods remain poorly understood and that has had a negative bearing on how the Rozvi are defined archaeologically. Historians have been describing these socio-political entities in terms of dynasties and this is not always materially manifested. At least from their contributions, it is clear the Rozvi started as a small group that became diverse, widespread and very dynamic before they disappeared into 4 oblivion in the late 19th century. In view of this fluid nature of these people, this study views the Rozvi not merely as a dynastic or historical name but rather a “dynamic ethnic identity” that marked political status, which was crucial in shaping power relations. This view and conception of Rozvi holds great potential in the analysis and synthesis of the subject, whose references are currently fragmented, conflicting and scattered. This study therefore argues that developments noted above are strong indicators that there is a wide research gap in as far as the Rozvi archaeology is concerned. Thus the whole investigation starts from the problematic poser; “Who were the Rozvi?” This question is prioritised because only when it is satisfactorily answered, can we be in a position to provide a meaningful definition archaeologically. This dissertation focuses on how the Rozvi used material culture to structure and express their identity. 1.2 Problem formulation The Rozvi are widely documented in historical sources but their archaeological identity remains a mystery. Wide ranging views have been raised about their identity and their socio-political nature (e.g., Beach n.d., 1980, 1994a, 1994b, 1995; Bourdillion 1976; Caton-Thompson 1931; Hall 1909; Mudenge 1974a, 1974b, 1988; Posselt 1935; Randall-MacIver 1906; Robinson 1959, 1966; Schofield 1948). Historians and archaeologists do not quite agree on the origins, dates, and distribution of settlements associated with the Rozvi. In particular the relationship between Torwa and Changamire Rozvi states remains poorly defined (Pikirayi 2001). The proper origins, as well as group identity for the cluster of sites in the south-central and western regions of the country near the present town of Gweru, regarded as the core settlements of both the Torwa and Rozvi rulers, has not been ascertained. The status of Rozvi at Khami remains a contested issue as some archaeologists continue to associate Rozvi with Khami, for instance Robinson (1959, 1966) and van Waarden (1998), while historians like Beach (1980: 190-2) argue that the Rozvi never lived at Khami. Several sites in the south-central and western region of Zimbabwe are strongly linked to Rozvi ruling dynasties (Beach 1980). However a majority of these have never been excavated to verify these historical claims. Pikirayi (1993, 1997a) and Pwiti (1996a) point to the lack of serious archaeological attention to sites relating to the Khami phase, a development that continues to 5 undermine our comprehension of archaeological sites attributed to the Zimbabwe culture 1. Robinson (1966) noted that further knowledge on the Rozvi would be impossible until far more fieldwork and excavation is done within a wider area. This research largely focuses on fieldwork and the analysis of museum collections from the two archaeological sites. Fieldwork and analysis of evidence will be strongly guided by current theory and method in the study of agency and ethnicity in prehistory. The major concerns of this study relate to defining relationships between people namely the Rozvi and material culture at the sites they inhabited. This is obviously a difficult task because identifying ethnicity from material remains of past societies is not so easy despite having multiple archaeological approaches at our disposal. To begin, ethnicity is a form of identification based on subjective elements of history and culture (Kobylinski 1989; Emerson 1997; Isbell 2000; Preucel 2000). In most cases various elements of culture are used to signify the identity of an ethnic group but some of these elements are not „real‟ or „objective‟. Even those elements of culture that are visible and accessible through archaeology, do not form permanent foundations of identification because ethnicity is a dynamic process. Hence tracing ethnicity becomes a problem because archaeology as a discipline is premised on interpretations that are objectively derived from material remains of long gone societies. Therefore defining a plastic identity such as that of the Rozvi through material remains demands great caution because it entails considerable subjectivity. The problems associated with distinguishing the Torwa and Rozvi identities via material remains at their settlements possibly led many scholars to take comfort in lumping their archaeology as generally similar. This study argues that identities are actively communicated through material culture as noted by Wells (1998), hence Rozvi identities should be investigated through patterned differences in material culture. 1.3 Research aims This study aims to understand the construction processes of Rozvi identities, their use of power and symbolism. Such an appreciation of the Rozvi creates a suitable platform to trace their 1 The phrase Zimbabwe culture tries to define the monumental architecture, settlement organisation, ceramic traditions and myriad material culture found at sites that have a historical or cultural connection with Great Zimbabwe monument found in south-central Zimbabwe (Pikirayi 1993, 2001). In this regard even sites that differ slightly in architecture like Khami phase ones are considered part of this culture (Beach 1980, 1983a; Huffman 2000, 2007; Pikirayi 2001; Pwiti 1996a). 6 archaeological signatures from sites in south-central and western Zimbabwe. Therefore material remains such as pottery, house structures, stone walls, and other artefacts from Danamombe and Khami are examined to denote possible Rozvi ethnic markers. It is clear that when the Rozvi invaded the south-central and western areas of Zimbabwe during the 1690s they assimilated the Torwa population (Beach 1980:195). However, there are no firm statements about what happened in the archaeological record when the Rozvi incorporated the Torwa. In fact, the lack of a clear definition of the Torwa identity is the major challenge when trying to separate them from Rozvi. Thus it is crucial to also interrogate the Torwa by using Portuguese documents and ethnographical records because a clear conception of their society and identity will in a way unlock Rozvi archaeological identity. Currently the Torwa and Rozvi archaeology is lumped together even though these state systems existed at different times and differed in their sociopolitical and possibly economic characters. The Torwa state was generally a centralised and stable political entity while the Rozvi state was less centralised, unstable and ideologically sophisticated (Beach n.d., Mazarire 2003, 2009). The ideological differences between the Torwa and Rozvi certainly have serious implications on how we may interpret their material record. It is likely that Rozvi material symbols varied extensively as their political influence was more widespread and dynamic than that of the Torwa. The first objective of this study is to describe the construction of Rozvi identities, and their relationship to power and power relations. This objective was formulated against the background that previous researchers discussed these socio-political identities without paying attention to their real meanings and associated cultural markers that transformed simple names into group identities. The Torwa are basically remembered as the major inhabitants of Khami phase sites from an archaeological perspective, but we know very little about them historically. While the Rozvi history has been fairly addressed their archaeology remains speculative, hence this study hopes to review and explain the dynamics associated with these socio-political entities. The second objective of this investigation is to identify and interpret the material indices that are associated with the Rozvi in south-central and western Zimbabwe. This objective was inspired by the view that material culture is an active agent in the communication of socio-political meanings and messages (Hodder 1982; Ndoro 1991; Wells 1998). A careful and contextual analysis of various forms of material culture from the research sites aided with documentary and 7 ethnographic sources should enable the identification of objectified or signified Rozvi material indices. This follows the premise that ethnicity is always expressed through cultural symbols that are consciously invoked by group members to mark their boundaries or difference from “others”. Finally this study seeks to generate a detailed synthesis of historical and archaeological literature on the identities of the Rozvi. This objective stems from the fact that there is contrasting and inadequate information concerning the Torwa and Rozvi identities especially from an archaeological perspective. Historical and archaeological views on the Torwa and Rozvi are currently fragmented and at times conflicting, but if they are harmonised a refreshing account of these identities may be established. Until now there has never been any serious attempt to synthesise this information, hence this study hopes to close the gaps and stimulate research interest in that respect. 1.4 Definition of terms This study has encountered the following terms: Nyai, Torwa, and Guruuswa. These are briefly explained to illustrate how their meanings vary widely in available literature. 1.4.1 Nyai/VaNyai: This term originally applied to “Shona” highveld inhabitants, primarily as a local name describing people by their location on the Zimbabwean plateau (Mazarire 2009). Later, the term acquired a political dimension referring to soldiers of the Mutapa state (Mudenge 1988). Since then it was adopted to denote young men who established a dependent relationship with a „principal baron', in order to work for wives through a “Karanga2” custom called kutema ugariri. Their duties ranged from domestic activities to political tasks like guards, errand runners, spies, councillors, messengers or warriors (Beach 1983a; Lancaster 1974; Mazarire 2009). These Nyai could accumulate wealth and power through time and eventually expand their baron's influence or even carve their own territories. Hence it eventually became difficult to distinguish Nyai followers from their political leaders and political state systems like Rozvi. According to Mazarire (2009) these power structures were volatile and were the cause of political instability sparking from succession disputes leading to the rise and fall of several states 2 Karanga is the historical name of almost all the people we now call Shona. This term is preferred in this study because before the nineteenth century, ancestors of present day Shona people were described as Karanga by the Portuguese. 8 like Mutapa, Torwa, Rozvi, among others. All in all Nyai/VaNyai were active agents that were influential in shaping power relations in pre-colonial state systems. 1.4.2 Torwa: The origins of the name Torwa and what it implied remain a major challenge for both historians and archaeologists today. Mazarire (2009) uses the term loosely in references to inhabitants of the southwest who spoke an ancient Kalanga dialect. This approach does not clearly depict the real meaning of the name Torwa, which was first documented around the 1490s following a revolt from the Mutapa (Randles 1979). Holleman (1952) noted that (vatorwa) meaning foreigners constituted the second class in an organised “Shona” society regardless of their rank, position or length of residence in a tribe. Vatorwa did not mean social inferiority or subordination to machinda (the chief‟s relatives), but the term simply distinguished the chief's lineage from people who came from elsewhere (Hollman 1952:15). Therefore the Torwa rebel who founded the Guruuswa/Butua dynasty could have been nicknamed mutorwa when he settled with Changamire somewhere in the Mutapa state during the 1490s. When that Torwa dynasty in Guruuswa expanded into a state, leaders who assumed the throne also assumed the dynastic title Torwa, their subjects and material culture was also automatically coined Torwa by outsiders to that community (see Beach 1980; Pikirayi 2001; Randles 1979). One may thus tentatively suggest that the term Torwa became an ethnic identity owing to this historical background in its usage. 1.4.3 Guruuswa: This term appears quite often in historical sources dealing with the history of the Mutapa and Rozvi states (see Beach 1980; Mudenge 1974a, 1988; Zachrisson 1978). The territory referred to is difficult to define with certainty because it varied widely among those who tried to locate it. According to Beach (1976:3) people between the lower Ruya and Mazowe rivers, locate it to the west; those in Mount Darwin and Guruve areas place it to the south; while those in Mafungautsi plateau and the Mupfure-Munyati confluence point towards the eastern direction. Mudenge (1974a) sometimes calls it Mbire/Guruuswa when dealing with areas that were briefly governed by the rebellious Changamire of the 1490s. According to Beach (1980:197) Guruuswa was just a Karanga environmental term for high ground with tall grass in references to land running from the Zambezi Escarpment southwards. This region is also historically known as Butua or Butwa by the Portuguese while other scholars prefer to loosely define it as Guruuswa meaning high ground with tall grass. Randles (1979) suggested that Butwa 9 or Vatwa meant land inhabited by strangers (vatorwa) to the Karanga people. Thus it meant Khoisan country or the Torwa area because these people were somehow different from the Karanga in terms of their biological make-up and cultural attributes. The exact geographical boundaries of Guruuswa varied through time hence Portuguese cartographic maps generally located Butua/Guruuswa as territories on the south-western regions of the Mutapa state that spread from below the Zambezi River, going as far as the Shashi-Limpopo basin and adjacent areas (see the marked area on Fig. 1.3). Figure 1.3. Map highlighting Butua/Guruuswa modified from Petrus Plancius, 1594 (in Randles 1958). 1.5 An Overview of Rozvi Socio-Political Complexity When examining Rozvi identities, we cannot avoid delving into aspects of social complexity because we need to first appreciate the nature of their socio-political organisation in historical period contexts on the Zimbabwean plateau. Worse still, the dynamic nature of Rozvi makes it difficult for one to simply describe their complexity as either a chiefdom or state system. Therefore before determining whether they were a chiefdom or state, it is crucial to thoroughly review these two concepts first. Anthropologists, historians and archaeologists have for long 10 argued over the nature, origins and development of complexity (e.g., Cohen 1978; Eriksen 1991; Isbell 2000; McIntosh 1999; Preucel 2000; Stein 1998; Yaeger and Canuto 2006). According to McIntosh (1999) a significant volume of literature produced between the 1940s-1960s was greatly influenced by Fortes and Evans-Pritchard's (1940) classification system of prehistoric societies from lineage segments, chiefdoms and finally state systems with a centralised administration. The sequence towards complexity started with simple (autonomous villages) sometimes referred to as “ethnic units”, “tribal” and “tribe”, “primitive” or “uncivilised” societies. These were basically conceived in terms of exclusive, territorially bounded, selfconscious clusters sharing common language, history and culture (Isbell 2000; McIntosh 1999; Spear 2003). Such kind of a classification emphasised evolution as the natural history of all humanity, thereby simplifying the development of societies as merely universal across the globe (Isbell 2000). Unfortunately, this evolutionary model of socio-political complexity would create serious problems if applied in the examination of Rozvi complexity because their formation processes were very sophisticated. Terminology has always been a common challenge especially when scholars attempt to adopt a classification system developed elsewhere. During the 1970s there was growing concern over use of terms like “civilised” or “uncivilised”, “primitive”, “tribal” and “tribe” (Isbell 2000; McIntosh 1999). All these terms are currently viewed as unsuitable, because they carry pejorative connotations for African societies. Although the concept of “tribe” (autonomous village communities) remains widely used in anthropological and archaeological literature, its meaning remains elusive. In most instances the autonomous villages are treated as the second stage in the linear development of societies from bands, to autonomous villages, chiefdoms and finally state systems (Isbell 2000; Stein 1998). Isbell (2000:248), further notes that the concept of tribe in archaeology has been taken to mean villages or homogenous, natural communities imagined as a self-sufficient farming villages. Southall (quoted in McIntosh 1999:2), notes that in Africa the distinction between tribes and chiefdoms has been problematic because many tribes behaved like chiefdoms. Emberling (1997) notes that the term tribe is ambiguous because it continues to be defined using both cultural characteristics (where drawing boundaries using social markers is problematic), and political characteristics (as groups existing within states or other peripheries). This concept of tribe therefore remains an issue of serious academic concern. 11 Despite these debates surrounding the notion of a tribe, Rozvi socio-political complexity was far developed to be defined as a tribal formation. Following the trajectory of political evolution, these tribes were replaced by “chiefdoms”, which are an ambiguous platitude meaning everything and nothing simultaneously. Carneiro (1981) defines them as pre-state societies, while Service (1962) generally refers to “chiefdoms” as a stage of cultural evolution. McIntosh (1999) argues that they constituted the second phase of development from simple villages to compound (chiefdoms). Oberg (quoted in McIntosh 1999) defined chiefdoms as polities united under permanent control of a chief. While Stein (1998) explores the concept to some reasonable depth, as having multiple meanings due to its prolonged usage and broad meaning that constitutes “simple” and “complex” chiefdoms. Stein notes that there are “simple” and “complex” chiefdoms, which develop as a result of economic inequalities generated and maintained through strategies employed by elites or chiefs. Therefore, it is a concept often used to mark the origins of social complexity for hunter-gather-fisher societies and food producing societies because of the ability of certain individuals to create and sustain economic inequalities through the monopoly of surpluses (Stein 1998:8). He further notes that the individuals hoping to become chiefs can use prestige goods, control of land, pastoral capital or even labour mobilised through kinship ties or ritual as economic means to gain power. This conceptualisation of a chiefdom somehow befits Rozvi complexity as it relates to the role of agency in socio-political transformation. However most models describing complexity at the chiefdom level have a bias towards behavioural elements, which portray chiefdoms as homogeneous, static and confined to specific territories. Scholars such as Kristiansen (quoted in Stein 1998:8) frustrated by these cultural evolutionary frameworks, argue that the chiefdom concept is too vague, and it obscures critical transformations in the development of complexity and sounds more like a variant of tribal societies. Pikirayi (1993) says that the term is problematic since its meaning ranges from societies ruled by chiefs to political hierarchies involving a redistributionary economic system. Southern African archaeologists have not been spared either from the debate of defining boundaries between chiefdoms and states. Pikirayi (1997, 2001) rightly argues that Zambezian chiefdoms and states exercised functional interdependence and such a scenario makes it difficult 12 to separate these social organisations chronologically. In southern Africa, we cannot easily distinguish chiefdoms and state systems because resources that ensured the survival of a state were drawn from the wider territory controlled by chiefs who also needed to draw resources from the state to survive. This interdependence sustained a fluctuating and dynamic situation that facilitated the development of chiefdoms into state systems, or caused the collapse of state systems due to competition from chiefdoms that took charge of their own resources (Pikirayi 2001). Pwiti (1996b) acknowledged that it is difficult to differentiate chiefdoms from states but made great efforts to explore Musengezi chiefdoms and the Mutapa state in northern Zimbabwe. Generally chiefdoms in southern Africa are considered as settlement organisations that mark the beginning of socio-political complexity starting from around 800-1000 AD (Pikirayi 1997, 2001; Pwiti 1996a; van Waarden 1998). Chiefdoms are thus treated as settlements that show transformations from Early Farming Communities (EFC), which were dominated by ideologies of equality that facilitated the survival of non-stratified and self-sufficient villages. New economic opportunities brought about by foreign trade through the Indian Ocean coast, and increase in cattle herds led to the re-negotiation of ideologies by certain individuals who wanted to mask their economic advantages by legitimising socio-political inequality (Pwiti 1996b). Although the Rozvi are traced to a certain rebellious chief from the Mutapa state (Changamire Dombo dated between the 1670s-1696), their socio-political organisation was far too complex to be merely equated to these early chiefdoms because their political power was not tied to economic incentives. Pwiti (1996b) further characterised Musengezi chiefdoms in northern Zimbabwe as led by a few individuals with access to strategic resources that they used to build economic and political power. Using Musengezi sites in northern Zimbabwe, he suggested that the settlement pattern for chiefdoms should have large sites covering at least two hectares and smaller ones of less than half a hectare. In southern Africa Later Iron Age sites like Mapela, Bambadyanalo (K2), Leokwe hill, Taukome, Toutswe, Ntabazingwe and Monk‟s Kop among others have been classified as chiefdoms because they show evidence of long distance trade, cattle wealth, class distinctions between the elites and commoners, and ideologies legitimised by rituals and religion (Calabrese 13 2000; Denbow et al 2008; Hall 1987; Huffman 2005; Pwiti 1996a, 1996b). While the Rozvi operated in certain ways like chiefdoms cited above, they differed in that they were more heterogeneous, scattered territorially and sophisticated ideologically. Rozvi elites never entirely depended on monopolising economic resources to maintain their grip on power yet they still managed to control their dispersed subjects. In fact, they systematically manipulated the Mwari religion, mythological versions, land-rights and investiture ceremonies in order to gain power over subjects. This alone confirms that they were more advanced than other ordinary chiefdoms. In this regard, their socio-political structures were too advanced to be simply defined as chiefdoms. Therefore the “chiefdom” concept has had mixed reactions possibly because it lacks a clear definition thus meaning different things to different people at the same time. Elsewhere in Africa the concept of chiefdoms has been further complicated by colonialism which invented or constructed new meanings in order to divide and rule their colonial subjects (McIntosh 1999; Pikirayi 1993; Spear 2003). Another possibility for the multiple meanings of “chiefdom” is it was crafted for Hawaii and other Polynesian states and simply applied to a totally different socio-political context leading to many complications. However the lack of a better term to define complex societies, which do not qualify to be called states and the shift from the redistribution-based model to a conflict-based model of power relations, has kept the concept favourable in archaeological applications (Stein 1998:10). Already historians in Zimbabwe are using the concept of “dynasty” in place of “chiefdom” in attempts to refer to defined legitimate lines of rulers with varying powers in specific territories (Pikirayi 1993). Although “dynasties” were common in discussions of the Zimbabwean precolonial state systems like Mutapa and Rozvi, the term does not adequately replace the “chiefdom” concept in archaeological terms. Nonetheless for purposes of this study the “chiefdom” concept is equated to a dynasty, a term that better reflects the nature of polities from the 15th to 19th centuries. Mazarire (2009:2) argues that much of complexity after the decline of Great Zimbabwe during the 15th century was a result of fragmentation into several dynamic, adventurous and sophisticated dynasties. Several dynasties that broke from Great Zimbabwe were pivotal in the development of complexity throughout the Zimbabwean plateau from the 15th 14 to the 19th centuries (Mazarire 2009). Hence, when we discuss “post-Great Zimbabwe” states, we are basically talking about the Mutapa dynasties, Torwa dynasties, Rozvi dynasties, and many other small dynasties that are too many to mention. These dynasties operated in the same manner as chiefdoms, which developed as a result of economic inequalities and unstable power relations that were created and manipulated by elites, chiefs or dynastic leaders such as Changamire Dombo of the Rozvi. The development and survival of states on the other hand has been explained in terms of the intensification of long distance trade, cattle production, population growth, ideological factors among other aspects (see Garlake 1973, 1982; Hall 1987, Huffman 1981, 2005, 2008a, 2008b; Pwiti 1996a; Sinclair 1987; van Waarden 1998). Others have characterised state systems as having five-level hierarchy of leadership constituting; a family head, a headmen, petty chiefs, senior chiefs, and then a king such as the case with Mapungubwe and Great Zimbabwe (van Waarden 1998). Pwiti (1996a:19) adds that monumental architecture is a major indicator of dominance by those in power, and this signals complexity at the level of a state as well. These universal models to the development of complexity in southern Africa cannot adequately address the Rozvi because they constantly reorganised themselves in order to remain relevant to varying historical contexts. Hence throughout their history as a political formation, their socio-political qualities fluctuated between both chiefdoms and state systems. Theoretical debates are also centred on the origins and nature of “states” or “communities” which mark the third phase of cultural evolution. Sinclair et al. (1993) defines a “state” as a political entity demonstrating sophisticated cultural achievements in terms of settlement structure, religion, military and diplomatic organisation, external trade, agriculture, natural resource exploitation and management, and most importantly the ability to extract tribute. American anthropologists and archaeologists are opting for the term “community” instead of “state” and “community” is further subdivided into real and natural (primordial or local) communities and imagined (fleeting, politicised or fluid) communities (Isbell 2000; Yeager 2000). The “community” or “natural community” was conceived in terms of shared residence, life experiences, goals and sentiments constituting functional and predictable human behaviour (Isbell 2000). These notions of social complexity are also premised on passive “holistic” 15 evolutionary, functionalist, adaptationist, and mechanistic models. Perhaps the Torwa sociopolitical complexity was synonymous with a “natural community”, while the Rozvi was an “imagined community”. However it has been demonstrated that social complexity can also develop in the absence of significant economic control, as was the case with the Rozvi initial rise to power. McIntosh (1999) notes that since the 1970s another critical anthropological and archaeological theoretical vein was operational, emphasising that people are not uniform automatons merely reacting to changes in the external world but rather active participants in the formation of social realities. This theoretical tide has risen significantly to challenge the behaviourist mode in explaining social complexity (Isbell 2000; McIntosh 1999; Stein 1998). Proponents of this theory argue for complexity in terms of “agency” or “imagined communities” which are volatile, dynamic, integrated by gratifying experiences, desires and intentions, and most importantly never isolated but interacting and adjusting to prevailing social contexts (Isbell 2000; Preucel 2000; Yaeger and Canuto 2000). As a result of such interactions, identities or group „collective consciousnesses‟ developed on the basis of common history, sentiments, knowledge, goals, and interests (Isbell 2000; Stein 1998). Powerful members defined meanings of all social relations in order to construct identities of commonality or distinctiveness from „others‟ (Cohen 1978; Eriksen 1991; Emerson 1997). This is the model that closely relates to Rozvi origins, use of power and their fluctuating nature. These elites basically exploited two kinds of strategies namely „network-based‟ and or „corporate-based‟ in order to legitimize and sustain their power. These strategies led to either a „hierarchy‟ (unequal and stable) or „heterarchy‟ (decentralized and fluctuating/unranked with related elements to one another and having potential for ranking in a number of ways) kinds of social organisation (Stein 1998). Thus the notion of “agency” which places emphasis on political leaders as agents and strategisers in the development of state systems or complexity is central to this research (McIntosh 1999; Preucel 2000; Yaeger and Canuto 2000). This recent and refreshing theoretical approach allows archaeologists to interrogate complex issues such as identities, social relations, representation of power, as well as the material record for contradictions intended to influence others (Isbell 2000; Yaeger and Canuto 2000). This study 16 therefore conceptualises the Rozvi as a state system or “imagined community”. This agency approach to Rozvi complexity is suitable because it is flexible enough to accommodate all the other relevant notions to socio-political complexity. Therefore in terms of socio-political complexity, the Rozvi were not necessarily a chiefdom but a very dynamic and decentralised state system. Their state had no clear political boundaries because membership was most often imagined; meaning members simply had to perceive themselves as different from “others”. 1.6 The Zimbabwe culture: an overview from Great Zimbabwe to Rozvi state The Torwa and Rozvi are generally remembered as the major occupants of Khami phase sites, which mark the second phase of the Zimbabwe culture. Hence it is necessary to briefly outline the sequence of the Zimbabwe culture so as to contextualise the Torwa and Rozvi in these cultural processes. Several identities also centre on Zimbabwe culture sites, and the Rozvi identity features quite prominently at such ruins and that triggers concerns requiring clarification. The phrase Zimbabwe culture tries to define the monumental architecture, settlement organisation, ceramic traditions and myriad material culture found at sites that have a historical or cultural connection with Great Zimbabwe monument found in south-central Zimbabwe (Pikirayi 1993, 2001). In this regard even sites that differ slightly in architecture like Khami phase ones are considered part of this culture (Beach 1980, 1983a; Pikirayi 2001; Pwiti 1996a). So far the phrase appears to have limited significance to local historians as it remains confined to archaeological studies. Perhaps its unpopularity among historians, could be that it tends to generalise or simplify an otherwise sophisticated and prolonged process of cultural transformation and dynamism. This of course follows the argument by Beach (1980) that archaeologists study Zimbabwean prehistory as though it was populated by pots and bones instead of people. As noted already the Zimbabwe culture has been presented in a linear trajectory whereby the fall of a major state subsequently led to another, thereby creating a simple and straight forward linear sequence of states. According to Huffman (2008a, 2008b) the Shashe-Limpopo basin experienced a dry climatic period around AD 1200-1250 and this impacted negatively on agricultural production for Mapungubwe, the first state system in southern Africa. However, the demise of Mapungubwe can not be sorely explained on climatic changes as proven by recent 17 research. Smith (2005) notes that political factors, competition for trade and the growth of Great Zimbabwe should be given much attention if we are to fully understand the decline of Mapungubwe. Again Denbow et al. (2008) registers similar sentiments regarding the demise of Mapungubwe, hence such trends have led to a marked revision of the previous views about Mapungubwe and Great Zimbabwe. Therefore regarding the fall of Mapungubwe, (Huffman 2008a:2043) suggests that natural disasters such as the dry spell at the end of the 13th century, that impacted negatively on agricultural production could have undermined the king‟s authority. Many believed that the ancestors chose sacred leaders and if they were not happy with a king's reign they communicated through natural misfortunes (Huffman 2008a). This could have triggered political instability at Mapungubwe because a king associated with misfortune had to be replaced by another sacred leader acceptable to the ancestors and God. Huffman (2005:55) notes that after the fall of Mapungubwe around AD 1300 some people went southwards while others scattered northwest. Therefore probably the dynasty from Mapungubwe introduced class structures and new social organisation at Great Zimbabwe (Huffman 2008b). However, this is not to say that the shift in population from Mapungubwe led to the rise of Great Zimbabwe. In fact, Great Zimbabwe was on the fringe of Mapungubwe, the people inhabiting the site were different; they took full control of gold and ivory trade leading to their development as a regional centre of power ahead of Mapungubwe (Huffman 2005, 2008b). However social organisation once typical at Mapungubwe was transferred to Great Zimbabwe probably by an early dynasty from Mapungubwe, which introduced several other ideological and technological innovations from Mapungubwe. Huffman (2005) also highlights that ceramic evidence at Great Zimbabwe does not quite confirm the assumption that there was a population shift from Mapungubwe to Great Zimbabwe. If such inconsistencies in material culture are to be given serious or critical archaeological attention, our understanding of southern African prehistory could otherwise be enhanced. Around AD 1430-50 Great Zimbabwe was also abandoned due to the rise of two peer states, the Mutapa in the north and Torwa in the southwest (Pikirayi 1993, 2001). The Mutapa state is well understood from both historical and archaeological dimensions (see Abraham 1959; Beach 1980, 1983a, 1994a, 1995; Bhila 1982a; Mudenge 1988; Pikirayi 1993; Pwiti 1996a; Randles 1979). 18 Pikirayi (1993) argues that during the 15th century, Harare and Musengezi traditions gradually changed in to Zimbabwe tradition settlements typical of the Mutapa state. The transformation was probably brought about by local developments with only minor influence from Great Zimbabwe. There was no massive population shift, but some Mutapa dynasties simply took full control of trade and agricultural production in northern Zimbabwe (Pikirayi 1993). According to Beach (1983a) the Mutapa state was already declining by the 1650s due to civil wars and Portuguese interference such that by 1720 its territory was only confined to the Zambezi lowlands of Zumbo and Teve. Pikirayi (1993) also offers an alternative archaeological explanation for the collapse of the Mutapa state by highlighting archaeological evidence that signalled the impact of conflict on the Mutapa state during the 19th century. Round about the same dates that Mutapa state developed in AD 1450, the Khami/Butua area was undergoing rapid urbanism. van Waarden (1998) argues Khami does not show any evidence of large scale migration from Great Zimbabwe, meaning it developed as a competitor to Great Zimbabwe leading to its eventual abandonment. She further argues that urbanism/complexity at Khami was so sudden indicating that the Torwa state had already ceased dealing with Great Zimbabwe by taking control over its own cattle and natural resources. Although the Torwa state was arguably the second largest state system in southern Africa owing to the wide distribution of Khami type sites, no rigorous investigation has yet been undertaken on it both by historians and archaeologists. Historians have largely ignored it on the basis of limited Portuguese sources and unreliable oral traditions. The least they have attempted to do is to state that it rose around 1450 and declined in 1644 due to a civil war and that 1683 marked the last date for a Torwa ruler (Beach 1980, 1983a, 1995; Bhila 1972; Mudenge 1974b; 1988). Pwiti (1996a) therefore pointed out that there is a wide gap to fill in as far as the Torwa state is concerned. There is even more confusion surrounding developments in the south-west after the demise of the Torwa and rise of Rozvi during the 1680s. Although it is agreed that Khami type sites had extended in the eastern direction to the current Insiza district to establish sites dated to the 17th century, little more than speculative knowledge about the Torwa exists so far. All that has been confirmed to some reasonable level of certainty is that Danamombe became the capital of the Rozvi state from the 1690s until the state collapsed around the 1830s. The rest of Khami phase 19 sites are merely linked to Rozvi from a descriptive angle, and no serious scholary work has yet been done regarding those site clusters. Be that as it may, this Rozvi period is the major focus of this study, which in particular seeks to explore developments in the south-west from an historical archaeology perspective. This section has also given an outline of the Zimbabwe culture, which shows that the Rozvi rose to power when many prominent sites had already been constructed and abandoned. Therefore their association with some of the monuments like Great Zimbabwe as shall be presented later on in this study, suggests that their identity was carefully and consciously constructed. 1.7 Organisation of this work This chapter has presented the Rozvi in the context of previous research attempts in defining them and their archaeology. The discussion develops further by outlining the broad research goals and the various theoretical underpinnings that will be applied for this dissertation. This chapter further problematises Rozvi socio-political complexity because they have been variously defined as a tribe, chiefdom, dynasty, state, and empire. A review of these notions of sociopolitical complexity was undertaken as part of efforts to define, describe, and contextualise Rozvi political influence within southern Africa, and Zimbabwe in particular. Chapter 2 explores the environmental settings of the south-central and west by paying particular attention to the geology, vegetation and rainfall patterns in order to understand prehistoric, historic and present settlement patterns. The prime concern is to understand how the environmental and cultural setting created opportunities and constrained the Rozvi political leadership and their subjects. Chapter 3 presents the theoretical frameworks in historical archaeology in view of the Rozvi and related researches done elsewhere. It defines the flexibility and potential of multi-disciplinary approaches in examining prehistoric societies. In this case I attempt to define how such approaches relate to studying notions of identity, ethnicity, power and the use of material culture in reinforcing ideologies and symbolism. 20 Chapter 4 discusses an agency oriented methodology, by systematically outlining the research procedures that were followed to define Rozvi archaeological identities. In this case, myriad archival sources, and contemporary folklore in the form of Shona praise poetry and novels based on historical narratives are integrated as part of oral traditions in order to deduce Rozvi identities. This chapter explores the appropriate archaeological approach for denoting Rozvi ethnicity as well as its associated challenges. Chapter 5 presents and analyses the data extracted from Portuguese ethno-historic sources, oral traditions (gathered from documentary sources and post-Rozvi houses in Bikita district of southeastern Zimbabwe), data from past ethnographic records, as well as information gathered from historical sources that at least examined both the Torwa and Rozvi pasts. Chapter 6 deals specifically with the analysis of archaeological evidence from Khami and Danamombe. In particular patterns observed from a comparative analysis of ceramics, beads, architectural features of dry stone-walls, mud house remains, and other special finds from both research sites are used to make inferences about Rozvi material indices. Inferences rather than assertive remarks are made on the meaning of material culture because the approach hereby adopted is purely subjective. It is not possible to attain an objective interpretation of the past using material culture because differences in any aspect of culture can be signified to distinguish one ethnic group from the other. Emberling (1997) also argues that these cultural features may vary throughout economic and political systems without having significant associations with any specific social group. Chapter 7 draws the dissertation together by synthesising archaeological data with documentary and ethnographic analogies. The chapter also explores the challenges of defining ethnicity and power through material culture. The value of ethno-historical investigations on post-Rozvi groups is evaluated for purposes of guiding and stimulating future research. This study concludes by harmonising current contradictions contiguous to Torwa and Rozvi historiographies, and it also highlights possible avenues for future research. 21 1.8 Conclusion This chapter has provided a general context in which the investigation of Rozvi archaeological identity is to be undertaken. In particular, this study problematises research issues that have been taken for granted for so long yet they are pivotal in our full appreciation of the Zimbabwean precolonial past. Questions such as “Who were the Rozvi or Torwa?” may sound very easy to address yet they demand careful thinking to be satisfactorily answered. For instance, the dynamic nature of the Rozvi demands a full appreciation of the role played by individuals in legitimising power and elite privileges. Precisely, Rozvi socio-political complexity may only be defined through the agency theory because it is flexible enough to explain their dynamic nature. It is only through the application of such dynamic models that we may attempt to tease Rozvi identities through material culture. Since the Rozvi were conscious beings who acted rationally in varying socio-political contexts, their symbols of identity were not randomly generated/produced but situationally expressed. Thus the prime concern of this investigation is to denote such Rozvi archaeological signatures from sites that they inhabited after conquering the Torwa. Therefore Rozvi material symbols shall be inferred from the archaeological sites of Khami and Danamombe, which are historically associated with the Torwa and Rozvi rulers respectively. 22 Chapter Two Physiography of south-central and western Zimbabwe “By 2000, historical archaeologists generally no longer viewed landscapes as static backdrops of human action but rather as places that are created and imbued with diverse meanings, disparate ideologies, and variant perspectives” (Orser 2010:115). 2.1 Introduction This chapter describes the environment of the research area (see Fig. 2.1), which is south-central and western Zimbabwe, primarily defined by the Gweru-Shangani and Gwai river systems flowing towards the Zambezi in the northwest, and the Thuli-Runde rivers, flowing towards the Limpopo and Save rivers in the south and south-east respectively. These river systems originate from Zimbabwe‟s central watershed, which is a highland or plateau region reaching up to 1500 m in places. The Matobo Hills immediately south of the modern city of Bulawayo are unique physiographic features that mark the north-western extremes of the upper Runde-Thuli catchment area. The Thuli River continues to define the south-western boundary as it flows in a south-eastern direction towards the Limpopo basin. The Runde River, whose source is the city of Gweru, marks the south-eastern limits of the research area. T South latitude. In this area, you find former capitals of the Torwa and Rozvi such as Khami and Danamombe (Dhlodhlo) (see Fig. 2.2). Though these two sites are of primary interest to this study, the broader south-central and western area is preferable because the associated communities had complex symbiotic relationships with the wider cultural and natural landscapes. So far the geographical location of the Rozvi state is disputed as some scholars consider it as an empire covering the entire Zimbabwean plateau (Beach n.d.). Beach (n.d, 1994a, 1994b) has however undertaken systematic research on the Rozvi and his research findings led him to radically revise this over-blown picture. He thus pointed out that the Rozvi rose in the northeastern plateau somewhere near the Murehwa area, the territory formerly occupied by the Mutapa and Manyika states. Although significant Rozvi dispersals took place between the 17th and 18th centuries, the Rozvi metropolis was largely centred in the south-central and west from around the 1690s until the 1860s when the state finally succumbed to Nguni incursions. Owing 23 to unique physical and cultural attributes peculiar to the region, south-central and western Zimbabwe has witnessed prolonged and continuous occupation spanning from the Middle Stone Age to the present (Walker 1995:13). Ndoro (2001) argues that cultural landscapes should be viewed as arenas of political discourse, where perceptions of nature, the environment and sense of a place are always changing due to cultural and political influences. This chapter therefore describes the physical environment of the south-central and western regions of Zimbabwe in order to explore how humans were constrained by it in their socio-economic and political interactions. 27°1'00" 27 27°31'30" 28°2'00" 28 28°32'30" 29°3'00" 29 29°33'30" 30°4'00" 30 30°34'30" -18 -18 18°26'30" 18°26'30" N 18°57'00" -19 ng ani 18°57'00" -19 S ha G w er 19°27'30" 19°27'30" u Gweru G w 19°58'00" -20 19°58'00" -20 a i hu i li 20°59'00" -21 60 27°31'30" 0 20°59'00" -21 Matobo Hill Ranges 27 27°1'00" 20°28'30" nd Lu # T 20°28'30" Bulawayo 28 28°2'00" 60 28°32'30" 29 29°3'00" 120 180 29°33'30" 30 30°4'00" 240 Key River City Figure 2.1. Location of the research Area. 24 Great Dyke 300 Km 30°34'30" Figure 2.2. Location of Khami, Danamombe and other Khami phase sites. 2.2 Topography and Drainage The topography of Zimbabwe is divided into three broad categories, which include the highveld extending from the south-west to north-east of Zimbabwe (1,200 m and above), the middleveld (600-1,200 m) and finally lowveld, which is generally below 600 m above sea level (Kay 1970; Ngara et al. 1998). The highveld and middleveld categories dominate the Zimbabwean plateau but these gradually drop into the Makgadikgadi basin in eastern Botswana. It is however quite 25 difficult to distinguish the highveld and middleveld regions in the research area because the topography is almost similar in many ways (see Fig. 2.3). However in view of the above classification by Kay (1970) and Ngara et al. (1998), the region aligned along the central watershed with heights ranging above 1,200 m defines the highveld area. This highveld region gradually progresses into the middleveld as it extends towards the western extremes of the country, and on its adjacent sides as we progress either north-west wards towards the Kalahari of Hwange or southwards towards areas like Kezi and Zvishavane. Northern side of the watershed is generally marked by gently flat rolling plains with scattered elongated ridges, a number of hills, granite outcrops and small kopjes (Watson 1960). Some of these landforms provided natural foundations for a number of Khami phase monuments that are commonly found in the area today. Much further north in adjacent parts of the research area, certain sections of these plains gradually fall into the depressed region of the Zambezi Valley. Due to the generally flat gradient, the Gwai River splits into the Umguza, Bembesi, Koce and Khami Rivers that flow in a north-western orientation (Amm 1939). The north-western gradient of the plains influences the general direction of the Gweru and Shangani Rivers, which eventually join with the Gwai River. These gentle slopes with soft Kalahari sands have also created almost parallel courses for most river systems flowing towards the Zambezi Valley and hindered their maturity as most of them cease flowing at the end of the rainy season. 26 Figure 2.3. A satellite topographical map of Zimbabwe adopted from (www.en.wikipedia.org/wiki/Geography_of_Zimbabwe). In contrast to the north, the topography on southern parts of the watershed is characterised by steep slopes that give low-lying, undulating country dotted by prominent ridges, hills, inselbergs, whalebacks, dwalas (dome shaped rock outcrops), castle kopjes and caves. These landforms resulted from long term weathering and erosion processes that gradually exposed the underlying granites, which retained the forms of their molten origin (Garlake 1987). Due to the general resistance to weathering processes, the granite outcrops remained prominent when softer overlying deposits were removed through weathering and erosion processes. Presently these huge granite blocks have cracked and split due to weathering processes to form several rugged hills, while other hills less affected by weathering have maintained their original smooth curved surfaces. These landforms now feature prominently on the landscape thereby disrupting the rolling plains, which characterise the research area especially from approximately 80 km east of the city of Gweru and beyond, going south-westwards. This area coincides with the highest density of archaeological sites in the country (approximately 0.01 sites per square kilometre), and they number up to 20 with the large ones being Danamombe, Manyanga, Zinjanja, and Naletale. These ruins were either built on or near granite landforms possibly for cultural, 27 architectural or religious reasons. The hills gradually become numerous further south-westwards from the Insiza district. Around a 100 km distance going south-westwards from Insiza, the granite outcrops increasingly dominate the entire landscape through a series of repeated hill and valley portions that are commonly known as the Matobo Hills (Danamombe Management Plan 2004; Kay 1970). South of the Matobo Hills along the Umzingwani River, banded ironstones and harder layers of sedimentary rocks also give rise to prominent hills and ridges on the western margins of granite masses (Amm 1939). In terms of drainage, most rivers originate from depressions located between various landforms to flow towards the Save and Limpopo basins in south-eastern courses. The Runde river drains into the Save basin, while rivers like Mwenezi, Umzingwani and its principle tributary the Insiza drains the central parts of the research area. Finally, the Thuli River defines the western extreme of the research area. The very steep gradient south of the watershed has enabled rivers to easily cut through the Kalahari sands to reach the hard basement schists and granites below, thereby signalling greater stages of river maturity. Moyo et al. (1998) notes that generally the west and south-western parts of Zimbabwe have the least amount of runoff with an annual average ranging between 17-19 mm. This implies that most rivers in this region flow only during the rainy season and possibly immediately after meaning they will be dry during the greater part of the year. 2.3 Geology and Soils Zimbabwe's geological formation is extensively characterised by the Precambrian rock formation dated between 570 million and 3,8 billion years ago. Schuter (2006) describes Zimbabwe‟s geological history and highlights that the craton constitutes vast areas of granitic gneisses, and younger granitods. After the occurrence of several metamorphic processes, this basement complex was intruded by the Karoo Super-group made up of sediments and volcanic lavas particularly found in the Zambezi and Save-Limpopo basins. This Karoo era was largely characterised by terrestrial detrital sediments (feldspathic sandstones, grits, shales, coal beds, among others). Later on, the Karoo sediments were overlain by basaltic flows and rhyolites that produced gabbros, granophyres, granites and syenites. Basically, intrusive granite and gneisses cover approximately 53% of Zimbabwe while the greenstone belt rich in gold, silver, copper, iron and many other precious minerals constitutes 7% of the rocks in the country. It therefore 28 follows that settlement patterns in the research area were influenced by these geological variations as most sites have been found near important mineral deposits. Although it is difficult to distinguish the high and middleveld regions in terms of underlying geology, the watershed region is generally dominated by dolerites while the middleveld regions flanking it on both sides constitute Precambrian granitoid rocks overlain by Carboniferous to Triassic Karoo sedimentary rocks with extensive basaltic intrusions. The highveld region is largely associated with the Great Dyke subsystem comprising ultramafic rocks (serpentine and pyroxenite) and mafic rocks. Hence, dolerite, greenstone, banded ironstone, serpentine, granite and schist rocks are quite common along the highveld and adjacent sections. Possibly this widespread distribution of granite and related rocks facilitated the construction of several Zimbabwe culture monuments found in the research area. The belts of schist in this region are also vital in terms of mining and farming activities (Moyo et al. 1998; Nobanda et al. 1998). Soils derived from granite are usually deep, clayey and acidic when rainfall is above 800 mm, but if it is less than that, then the soil becomes sandy. Since the research area receives about 750400 mm of rainfall annually, much of the soil from granite rocks tends to be sandy and less productive for cropping (Marongwe et al. 1998). Apart from granite derived soils, the middleveld sections of the research area are also characterised by small patches of soil formed on aeolian deposits, which are sandy. Soils on the northern margins of the middleveld are the Kalahari sandy type, hence they are generally well drained and they have a poor moisture retention capacity. In fact, prehistoric communities living around Mafungautsi plateau made good harvests of tobacco from these sandy soils (Ncube 2004). Deep red clays derived from mafic rocks are largely found along the watershed as they derive from the harder and more resistant rocks like dolerite, and usually do not show nutritional problems (Nobanda et al. 1998). Dry stone-walled structures constructed during the historical period like Naletale and Danamombe, as well as several commercial farms were established around these soils, most probably to enhance chances of exploiting them for cultivating crops, building mud huts, and as raw materials for ceramics manufacture. These clay soils are often localised in weathered seeps and at the bases of catenas. In terms of soil groups, south-central and western Zimbabwe is dominated by Ferisiallitic (Ferralic Cambisol/Ferralic Arenosol), Siallitic (Cambisols, Luvisols and Gleysols), Lithosol (Eutric Leptosol), Regosol (Ferri-Luvic Arenosol) and Sodic (Gleyic 29 Solonetz and Sodic Soloonchak) soils (Nyamapfene 1991). Ferisiallitic soils are widely distributed along the Great Dyke as such they have high agricultural potential because rainfall is relatively high. Currently maize and other crops are commonly grown in these regions (Nyamapfene 1991). Siallitic soils also have a very high agricultural potential but they are affected by the aridity of the environments in which they occur. In the research area they are distributed around Gweru, Bulawayo, Hwange, Insiza and Matopos (see Nyamapfene 1991:58). In situations where water is available for irrigation some very high levels of crop production can be attained (Nyamapfene 1991:59). The remaining four soil groups cited above are generally poor in terms of fertility; hence they support cattle ranching and wildlife activities (Matowanyika et al. 1998:144-5). South-central and western Zimbabwe has been widely known for livestock production and wildlife exploitation since prehistoric times as evidenced by large quantities of cattle and wildlife bones recovered from archaeological sites like Khami and Danamombe. 2.4 Climate Ngara et al. (1998) notes that the climatic conditions for Zimbabwe are influenced by both topography and distance from the equator. As such Zimbabwe lies within the semi-arid to arid region in which rainfall is both sporadic and spasmodic. Basically Zimbabwe experiences dry and wet seasons, which have been further sub-divided by Ngara et al. (1998) into, a rainy season mid-November to mid-March, a post rainy season mid-March to mid-May, a cool dry season mid-May to August and finally a hot dry season from September to mid-November. Rainfall and temperature variations in the research area are largely influenced by prevailing physical attributes like relief and altitude. High areas such as the Eastern Highlands generally have the coolest temperatures and the highest precipitation, while low to deep areas like the Zambezi Valley are very hot and dry. Since the research area is characterised by a rather subdued highveld region that is flanked by middleveld regions on both sides, temperature zones usually range in the warm to hot categories. Maximum temperatures in the country are usually recorded in the hottest summer months starting from the end of October to November; however areas falling along or near the watershed have relatively lower temperatures than the adjacent middle and lowveld. On the contrary, minimum temperatures in the country are recorded in winter particularly between May and July; again the watershed experiences relatively lower temperatures as shown on Table 2.1 below: 30 Table 2.1 Summery of temperature variations within the broad research area for the year 2010 (Source: Zimbabwe Meteorological Department) Station January February March April May June July August September October November December Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Byo. 28.9 17 28.4 17.5 27.1 15.5 25.9 14.8 24.8 11.8 21.6 7.3 21.7 8.2 24 8.5 30 12.5 32.7 16 30.1 17.3 28 16.8 Gweru 28.7 15.2 28.3 15.5 26.4 17.3 26.3 13.1 23.5 8.9 20.6 4.6 21 5.4 23.3 3.9 29 7.5 31.4 12 29.5 14.6 27 15.2 Kwek. 30.6 18 30.3 18 29.1 16.4 28.6 15.8 27.1 11.2 23.6 7.1 23.8 8.6 26.7 8 31 12.2 34.3 16 31.6 18.2 29 17.8 Zvish 33.4 19.7 32.4 20 29.9 17.9 28.9 16.6 32.9 12.8 24.1 8.1 23.4 8.8 25.2 8.7 31 12.8 33.4 17 32.9 18.8 31 18.7 W Nic. 34 18.6 32.6 19 31 17.2 28.6 15.8 28 11.4 24.4 4.9 23.8 6.8 25.6 7.5 42 12.5 34.4 17 32.1 18.6 32 18.2 Plmtr. 29.4 18.5 28.5 18.1 28 16.4 26.7 15 25.5 12.2 22.1 8.1 22.3 9 24.9 9.2 30 14.7 33.5 18 30.4 17.3 28 16.9 Lupn. 32.4 24.9 31.7 18.8 31 17 30.7 15.9 29.2 10.3 25.8 4.8 26.3 6.7 29.6 5.1 34 11.2 51.5 16 34.7 18.2 32 18.5 Key: Max= Maximum monthly average temperature Byo= Bulawayo Kwek= Kwekwe Min= Minimum monthly average temperature Zvish= Zvishavane W Nic= West Nichleson 31 Plmtr= Plumtree Lupn= Lupane Vincent and Thomas (1961) defined a number of agro-ecological zones primarily on the basis of mean annual precipitation, temperature and relief as a framework for land-use and planning (Fig. 2.4). It should however be noted that this colonial land classification system is rather static because it does not have archaeological, historical or temporal validity since climate change is a crucial factor (Manyanga 2006). The agro-ecological zoning is so general in nature and it does not take into consideration some variations that occur within regions thereby undermining precision. For instance the main problem related to the agro-ecological zoning, especially for semi-arid areas is that of resolution as their zoning ignored variability that occurs along topographic catenas (variations of soil structure/texture along a transect especially from a river bank to a hilltop). This agro-ecological classification gives the impression that certain regions were uninhabitable, harsh and unproductive for crops and domestic animals yet prehistoric settlements thrived in such regions. Region V in southern Zimbabwe is considered unsuitable for agro-pastoral activities in the classification yet that territory was formally part of Mapungubwe the first civilisation in southern Africa. Table 2.2. A summary of Agro-ecological zones (After Vincent and Thomas 1961). 32 Figure 2.4. The Agro-ecological regions of Zimbabwe, dotted areas in the map are regarded as unsuitable (Adopted from Vincent and Thomas 1961). The research area therefore receives annual rainfall of between 750-400 mm (Marongwe et al. 1998; Pikirayi 2001, Vincent and Thomas 1961). In this regard, the highveld region defining the watershed‟s highest regions mainly confined on the eastern side of the research area is generally associated with 750-500 mm annual rainfall. The rest of the south-western highveld and adjacent parts falling into the middleveld regions receive erratic rain ranging from 500 mm and below per annum. Using the summarised agro-ecological regional classification for Zimbabwe by Vincent and Thomas (1961), the research area would fall into regions III and IV respectively, while region II forms only two small isolated zones as depicted in (Fig. 2.4). The rainfall amounts and high temperatures typical of these regions are thus considered as only suitable for livestock production and partial dry-land cropping (Marongwe et al. 1998). 33 Manyanga (2006:46) has however successfully demonstrated that agro-ecological zones can be misleading in certain ways if we are to consider agricultural activities that were, and are still being practised in such so-called harsh zones. In fact, the agro-ecological classifications stem from modern perceptions that have been uncritically accepted and applied in interpreting past socio-economic activities. There is no doubt that subsistence agriculture often thrives very well in these so-called dry and unsuitable zones especially if drought resistant crops like sorghum and millet are planted, hence we should not be rigid about farming activities that can be carried out in these areas. Prehistoric communities also had an open access system that enabled them to exploit various resources provided by the different landscapes and today commercial agriculture based on the production of various crops under irrigation is thriving well in these so-called “harsh” regions (Manyanga 2006; Pwiti 1996a). Thus when we seek to interpret how past communities survived in these regions, there is need to adopt a much open approach. 2.5 Vegetation Zimbabwe has five phyto-regions supporting two main vegetation types namely; woodlands and grassland (Nobanda et al. 1998). The highveld with rainfall amounts ranging from 700-1500 mm and nutrient poor soils is characterised by Miombo woodland dominated by species like Musasa (Brachystegia spiciformis), Mupfuti (Brachystegia boehmii), Munhondo (Julbernardia globiflora), Mukwa (Pterocarpus anglolensis) and Muchacha/Muhacha/Muchakata (Parinari curatellifolia). According to Poilecot (2007) Miombo woodlands are also characterised by tall Andropogoneae grasses such as Hyparrhenia filipendula, H. Cymbraria, Hyperthelia dissolute, Andropogon gayanus, Schizachyrium and many other species. As rainfall decreases in the southcentral and western parts of the country, Miombo is replaced by Acacia woodlands largely dominating the south-western extremes of the research area. These woodlands are common in the dry regions that receive rainfall of 750 mm and below. Usually these trees grow on red clay soils derived from metasediments and metavolcanics (Poilecot 2007). Typical species are Muunga (A. Karoo), A. nilotica, Uhlahla (A. Gerrardi) and iPucula (A. Rehmanniana), Katopa (A. Nigrescens), A. tortilis and the associated shrub species range between 2-6 m high (Nobanda et al. 1998; Poilecot 2007; Wild 1952). Because these woodlands are scattered, most often there is a relatively good grass cover. 34 The other dominant woodland in the research area is Mopane (Colophospermum Mopane) largely found on low-lying parts with rich clay soils from granite rocks. Grass cover in these woodlands is very poor as it comprises annuals like bristle grass/stick grass (Aristida adscensionis), A. scabrivalvis, A. rhiniochlora, Erogrostis rogersii, Enneapogon cenchroides and Heteropogon melanocarpus (see Danamombe Management Plan 2004; Poilecot 2007; Timberlake & Mapaure 1999; Wild 1952). Terminalia-Combretum woodlands are also common in the research area but they are found in a mixture of trees and shrubs on sandy soils that usually receive less than 600 mm of rainfall annually. Again grass cover is poor but it comprises of Eragrostis viscova, E. Rogersii, Aristida congesta, A. Scabrivalvis, A. Stipitata and Pogonarthria squarrosa (Chapano 2002; Poilecot 2007). These tree species and grasses were exploited as housing materials, furniture material, wood for crafts, firewood, traditional medicines and some of the species even provided good browse for wildlife and domestic stock found in the region. The vegetation in the research area is also a reflection of the underlying geology as volcanic rocks like dolerite found along the Great Dyke constitute toxic minerals that inhibit the growth of certain tree species (Marongwe et al. 1998). On the other hand, adjacent areas of the Great Dyke largely dominated by granite are characterised by rolling grass-lands and sparse tree growth. Thus the only areas with dense vegetation cover are those with capping of Kalahari ferricrete and sand (Tyndale-Biscoe 1949). Historically the highveld area running all the way from the Zambezi and spreading to the south-west was commonly known by the term Guruuswa, meaning it was dominated by tall grass. These grasslands presented the best pastures for diverse faunal species and thatching material for houses, while some of the plants yielded wild fruits that often augmented food supplies for different communities that inhabited the region. Some of the animals like elephants were often hunted for meat, skin and ivory, but ivory eventually became a crucial resource for external trade between the 7th-10th century AD as is explained in the preceding sections. 2.6 Current land use patterns In the Gwai-Shangani basin, clear divisions could be drawn in terms of land use before the Fast Track Land Reform Programme (FTLRP) initiated in 2000 (Cobo et al. 2009; Kwashiri n.d.). The areas aligned along the central watershed including areas adjacent to the cities of Gweru and Bulawayo were initially set for white commercial farming, while certain sections of lands were 35 adopted for mines, and national parks. Most of the white commercial farmers specialised in ranches for cattle (to produce beef and dairy) and wildlife (for meat, trophies and tourism) at places like Somabhula and Bembezaan (Kreuter and Workman 1996). Some commercial farmers practiced irrigated horticulture but at a small scale because the regions were generally perceived as less suitable for such economic activities (Love et al 2005; Nyamapfene 1991). In the northern extremes of the research area at places like Gwai and Shangani the colonial government also acquired land for commercial forestry production specialising in Baikiea plurijua hardwoods, Mkusi (Teak) and Gum trees (Eucalyptus species) (Kwashirai n.d.; Nyamapfene 1991). Commercial honey production was sometimes practised at some of these forest estates (Nyamapfene 1991:48). Apart from these activities, several mines were established in areas like Silobela, Shangani, Zvishavane and Hwange to exploit various minerals like gold, copper and coal. Commercial farms found along the central watershed of the country benefited from Fersiallitic soils, which are very productive for crop production and they are often associated with gold deposits (Nyamapfene 1991:74). These soils are widely distributed on the central zone of the research area but land use possibilities are greatly undermined by rainfall unreliability and limited alternative water sources. The commercial farms extended southwards into the upper Runde-Tuli catchment area, which falls into Matabeleland South province. Commercial agricultural activities north of Kezi as well as in adjacent districts like Insiza, Matobo and Somabhula were generally confined to livestock rearing and the cultivation of drought resistant crops (Love et al. 2005). Less arable lands or sandy soils, which were characterised by erratic rainfall and limited natural resources were designated as African native reserves by the colonial government. This was done first in 1898 through the Land Ordinance Act and then later through the Land Apportionment Act of 1931. In the research area, these native reserves are to be found scattered on the peripheries of commercial farms, national parks as well as forest and game reserves. Agriculture is the main economic activity undertaken by these smallholder farmers (peasants) in communal areas. Most of the peasants in the upper Runde-Tuli catchment area are restricted to livestock production in the form of few cattle, sheep and donkeys but most of them specialise in small stock in the form of goats (Love et al. 2005). This scenario is almost similar to the Gwai36 Shangani communal areas in terms of livestock production but only differs in that cattle are the most dominant stock instead of goats because of the abundance of Mopane trees and sweet grasses especially around Gwai and Shangani (Kwashirai n.d.; Nyamapfene 1991). In terms of crop production, the peasants specialise in the growing of drought resistant crops like sorghum, millet, rapoko, cow peas, cotton and maize short season varieties (Love et al. 2005; Nyamapfene 1991; Kreuter and Workman 1996). Some communal members specialise in panning of alluvial gold from rivers while others make crafts of wood targeting tourists who sometimes visit the Rhodes Matobo and Hwange national parks as well as other interested buyers. After the FTLRP, some of the communal farmers were resettled in the former white commercial farms, where they are experimenting in a number of economic activities (Cobo et al. 2009). These “new farmers” are trying a variety of crops both at subsistence and commercial scales. Research has also established that the resettled farmers own more cattle than their counter-parts in the communal areas (Cobo et al. 2009). In areas where mineral deposits like gold and chrome were enclosed in former white commercial farms, resettled communal farmers are actively mining these in order to augment their income. Therefore the FTLRP brought about significant agro-ecological and socio-economic changes to the research area and the country at large. By and large, these changing land use patterns create a strong basis for comparing and interpreting modern economic activities with those of historical communities such as the Rozvi under study. 2.7 South-central and western Zimbabwe cultural landscape: opportunities and constraints South-central and western Zimbabwe is generally a dry region such that one would not expect it to have attracted significant human settlements. On the contrary, it has been intensively inhabited since Stone Age times (Garlake 1987; Walker 1995). This continuous cultural sequence especially from the Middle Stone Age dating from around 200,000 years ago to the historical period may best be explained in terms of survival strategies adopted by the various community leaders in managing that environment. The model for survival since prehistory was to extract resources from a much broader or wider area in order to avoid over-exploiting locally available resources. Strategic resources like wild floral and faunal species, mineral resources and domestic livestock were central elements in establishing networks and sustaining local economies that spanned through the entire culture history of south-central and western Zimbabwe. These socioeconomic and political networks were very complicated and diverse for the wide spectrum of 37 cultural phases represented. Therefore a careful analysis of any cultural group, state system or centre of power in that region reveals that contacts were very widespread and this strategy created the impetus for the subsequent rise of powerful and influential cultures like Khami, commonly associated with Torwa and Rozvi states. South-central and western Zimbabwe has one of the largest concentrations of hunter-gatherer sites in the country, the sub-region and the world at large (Matobo Hills Management Plan 2004). The Matobo hills provide an outstanding example of the cultural impact that Khoisan foraging groups had on the landscape. Stone tools and rock art paintings are commonly found in the granite shelters which were ideal for processing different floral and faunal species acquired from the landscape (Walker 1995). Since food was seasonally available, the foragers possibly migrated around the Matobo hills, eastern Botswana, and even as far as the middle Limpopo valley in search for food. Rock art from all these areas depict the same spiritual and natural themes, that were symbolically expressed through animals, humans, trees, plants, and various abstract designs, meaning these hunter-gatherers were in close contact with each other. Around AD 650 agro-pastoralists arrived in the south-central and western region and exerted more pressure on land and other natural resources. Foragers adapted to that development by establishing exchange ties with their new counter-parts until they finally got absorbed (Manyanga et al. 2009; Mitchell 2002; Phillipson 2005). According to Manyanga et al. (2009) the transition from hunting and gathering was by far complex than previously thought, but just the spiritual fusion between the Khoisan and Bantu farmers is of critical importance to this study. The Khoisan were the initial land owners, hence their ancestral spirits had power to either bless or curse that land (Manyanga et al. 2009). A complex system of oaths, taboos, and curses formerly governing the way Khoisan communities inhabited and exploited their environment. These San traditions were gradually and selectively adopted by the incoming Bantu and passed down through many generations until today. Intangible values are known to have great power in shaping the way community members relate to each other, their environment, and the spirit world (Githitho 2005; Ndoro 2001). Ndoro (2001) points out that the African culture does not separate humanity and nature, hence granite boulders, hill outcrops, and rock shelters have always been associated with rainmaking and belief systems associated with the monotheistic deity, Mwari. The whole concept of “sacred leadership” had a 38 lot to do with the topography of the landscape. Around the 13th century AD, the emerging elite at Mapungubwe and other Zimbabwe culture sites developed an ideology that legitimised class distinctions commonly termed “sacred leadership”. This ideology of “sacred leadership” was premised on the centralisation of political and ritual power by the leaders who closely associated themselves with, the land and Mwari (God). According to Arens and Karp (quoted in McIntosh 1999:17) African communities believe that the source of power resides in interactions between the natural, social and spiritual realms. Therefore religious beliefs fostered by the landscape, had a very strong bearing on how prehistoric people exploited their environment and settled within it. Early farmers in south-central and western Zimbabwe are defined by the ceramic tradition or culture “Zhizo” dated between 700 and 1300 AD. It was found throughout the area formerly dominated by hunter-gatherers. In Zimbabwe Zhizo people occupied the Matobo hills, Tabazikamambo and Ntabazingwe, while maintaining close trading ties with Schroda, a major trading site in the Limpopo (Pikirayi 2001; van Waarden 1998; Whitelaw 1997). Zhizo communities in Zimbabwe also interacted with Toutswe and Bosutswe in eastern Botswana which eventually grew into powerful states due to strategic management of their cattle herds and regional networks of trade (Calabrese 2000; Denbow et al. 2008, Reid et al. 1998). The environment provided zebra, wildebeests, sable, giraffe, eland, elephant, leopard and kudu, animals that were hunted for meat and products like skins, bones and ivory used for trade. The pastures were exploited for livestock rearing and the Mopane leaves provided good browse for cattle during dry seasons and droughts (Denbow et al. 2008). Denbow et al. (2008) pointed out that Bosutswe yielded ceramics from Great Zimbabwe and Khami ruins. Thus local crops could have been supplemented through local trading networks with areas experiencing better harvests. However complexity in eastern Botswana largely resulted from control of cattle wealth and possibly metallurgical products rather than crop production and external trade. Around the 10th century AD ivory demand for the Indian Ocean/East Coast trade was declining and that affected Schroda, which was a major ivory trading site in the Shashe-Limpopo basin. Subsequently the demand for gold was on the increase; hence Leopard's Kopje cultures developed as the major suppliers of gold. Thus as from the 10th century AD most of Africa's gold was exported through Sofala (Huffman 2005). van Warden (1998) argues that much of this gold probably came from mines or alluvial deposits in south-western Zimbabwe. Many sites in 39 Matabeleland associated with Woolandale pottery are mostly found near gold, iron, tin and copper mines, and that denotes the sources of gold (Robinson 1966; Summers 1961). Huffman (2005) also notes that various resources like copper, gold, ivory and salt which were crucial in the development of Mapungubwe came from Botswana and southern Zimbabwe. Therefore between the 11 and the 13th centuries AD, south-central and western Zimbabwe played a pivotal role in the development of major states. Extensive connections between the interior and the Shashe-Limpopo intensified from 1000 to 1220 AD leading to the rise of chiefdoms commonly called Leopard's Kopje cultures. This culture again becomes prevalent in south-central and western Zimbabwe and the neighbouring areas at sites like K2/Bambandyanalo, Ntabazingwe, Sekukwe Kopje, Nyangabgwe Hill, and Hill Site in Botswana. The hills provided an opportunity for leaders to built elite settlements, sometimes with stone walls to emphasise their status to commoners. This cultural development significantly spread and building techniques were greatly elaborated between the 15th and 17th centuries. Many stone walled enclosures in south-central and western Zimbabwe are still carelessly classified as Leopard's Kopje, Great Zimbabwe tradition, Khami phase or Rozvi period sites. However the general assumption is that these sites could have developed as satellite settlements for either Mapungubwe or Great Zimbabwe in order to extract trade resources from the interior. When Great Zimbabwe declined around 1450 AD, some dynasties are believed to have moved northwards to found the Mutapa state whilst others took a westward shift to establish the Torwa state based at Khami (Pikirayi 1993, 2001; Pwiti 1996a). van Warden (1998) and Huffman (2008a) however argue that there is no evidence for a large scale migration from Great Zimbabwe to Khami and that urbanism there was so rapid meaning the local Woolandale chiefdoms were already advanced. Summers (1961) also argued that during the 15 th century, a new type of stone-walling, ceramic type, copper, bronze and gold appeared at both Great Zimbabwe and Khami. Whatever the case, people are known to move into territories they have some form of knowledge or contact with. On this basis it is likely that the old trading contacts, and tribute payment relations established between Woolandale chiefdoms and Great Zimbabwe triggered the westward shift. This brought about closer interactions between the two cultures leading to the rise of Khami around 1450 AD. 40 Basing on the extensive distribution of Khami phase sites and the economic strategies adopted by previous cultures in relation to the geography of the area, one may tentatively suggest that Khami phase sites represented a series of decentralised chieftainships. The assumption is that some of the most influential Leopard's Kopje sites in that region simply transformed to another level of complexity by taking advantage of external trade, local/regional trading, cattle wealth, and the tributary system. Previously resources from the south-central and west had been exploited to support other state systems that were far and hardly returned enough benefits accrued from external contacts to the major suppliers of local trade goods. The origins of Khami and the Torwa state however remains a subject of scholarly debate. Historians argue that Khami or the Torwa state was founded by a rebel from the Mutapa state hence the name Torwa (meaning a stranger). According to Beach (1994a), around 1490 a certain Changamire rebelled from the Mutapa and was captured and assassinated in 1494. However his son fled to the south-west and founded the Torwa state and continued fighting until about 1547. Going by historical facts, south-central and western Zimbabwe had contacts with the Mutapa state as well, although such contacts were portrayed as generally hostile. Very little is known about trade in Butua which could have been carried out by Swahili traders operating in the north and north-eastern parts of Zimbabwe (Summers 1961). All we can safely say is that at different historical periods, Khami and Danamombe were in some form of contact with sites in northern Zimbabwe like Ruswingo rwa Kasekete, Dambarare, Ruhanje (Luanze), and Hwangwa (Luangwa) (Mudenge 1977; Pwiti 1996a; Robinson 1966). Khami has not received adequate research coverage, and until today the only substantial excavations were done in the 1950s by Keith Robinson. Related major Khami phase sites in that region have either been haphazardly excavated or never at all meaning we cannot make any firm statements regarding the archaeology of the area as yet (Robinson 1966). The only necessary comment for now is that, local and external trade as well as cattle wealth must have had a major economic and political impact in the region. These factors probably influenced the evolution of Leopard's kopje cultures into Khami phase settlements associated with magnificent well decorated retaining-wall monumental structures that remain scattered in south-western Zimbabwe. It remains unclear, how much the Torwa had gone in terms of monumental construction and material culture production when the Rozvi arrived from the north-east around 1695 to conquer 41 and assimilate that state (Beach 1995:27). After Changamire Dombolakonachingwango (Dombo, for short) or Chikurawadyembewu rose to power and waged four major battles in the north-east between the 1680s-90s he chose to migrate to south-western Zimbabwe (Beach 1980). Several reasons for that decision have been suggested, but possibly the Rozvi migration to south-central and western Zimbabwe during the late 17th century was inspired by the need to draw power from certain parts of that landscape which are still commonly associated with the seat of Mwari (God) and ancestral spirits. Butua/Butwa was also far from Portuguese traders who were a constant threat to local states. The area also provided rich pastures for Dombo's cattle which were his springboard to power, again a lot of gold was extracted from that region, and lastly the Torwa state was already weakened by civil wars since 1644 such that it was easy for Changamire Dombo to take over the entire state. Beach (1994a) actually states that the Rozvi state was literally built on foundations laid by the Torwa state. The Rozvi developed into a very powerful state between the 1690s and 1830s by maximising on what nature had to offer and creating several ideologies aimed at manipulating people's perceptions of the entire cultural landscape in order to reinforce their identities. South-central and western Zimbabwe was gradually altered and redefined to fit into new meanings that reinforced values for the Rozvi community. When Nguni groups arrived during the 1830s they defeated the Rozvi and established their dominance in south-central and western Zimbabwe which lasted until the 1890s when European colonialism took over the economic and political system of the region. Several commercial farms were carved out for wildlife conservation, dairy and cattle ranching with the entrenchment of European rule. Until today south-central and western Zimbabwe remains a suitable environment for hunting, livestock production and several mining activities, all of which traces back to prehistoric times. 2.8 Overview The research area is basically characterised by a central highveld or watershed region that drops in altitude as it extends towards eastern Botswana. This highveld is bordered by the middleveld on both the northern and southern sides. The geological formations in the area generally determine the relief, soils, drainage and economic activities for the different localities. Variations in altitude resulting from underlying rock formations also have a direct bearing on the climatic and ecological conditions in the respective areas. These natural characteristics either directly or 42 indirectly shaped the economic, spiritual/social, and political organisation of people residing in these environments. Nonetheless, people are never confined or constrained by their environment, but rather they adapt to it by strategising in multiple ways. Thus despite the generally harsh climate, poor soils and low drainage characterising the south-central and western region of Zimbabwe, it remained central to all major prehistoric state developments in the sub-region. The natural landscape offered various opportunities in terms of mountains, hills, flora and faunal species that were inter-twined with the social and spiritual realms leading to a unique cultural landscape. This prolonged interaction between people and nature laid the foundations for the powerful historical Rozvi state, which capitalised on several natural and cultural domains such as pastures, mineral wealth, Torwa stonewalled structures, the Mwari religion in the Matobo hills, among other things to legitimise their actions and power. 43 Chapter Three Analytical Frameworks “... theoretical breath is what gives the contemporary practice of historical archaeology its interpretive power, historical insight, and relevance” (Orser 2010:115). 3.1 Introduction This chapter is structured in a manner that demonstrates the strength of historical archaeology in studying identities and meanings associated with material culture. In this case the argument that identities are fluid social constructions that are better understood by being responsive to analytical concepts developed in disciplines like history, political science, social anthropology, geography, and linguistics comes to the fore. The climax of this chapter then addresses the dynamics surrounding ethnicity, agency, power, material culture and the Rozvi. Most importantly this study argues that the Rozvi are best understood as an ethnic group because their identity was consciously constructed, imagined, and sustained through the strategies employed by Rozvi elites. These elites were rational agents that capitalised on ethnic consciousness as a means to gain power over “others”, and that ethnic awareness was premised on the use of ethnic markers drawn from various elements of culture including material culture. It therefore emerges from such an exploration that the relationship between ethnicity and material culture is quite sophisticated. Throughout the history of archaeology, material culture has always been attributed to cultural, tribal, ethnic, national and racial identities. However the relationship between identities and the material record has varied markedly from the initial simplistic views to more advanced and sophisticated approaches. The initial phase was dominated by the culture-historical framework largely concerned with chronology and definition of cultures and ethnic groups on the basis of object typologies and their geographical distribution. Similar traits in material culture from specific sites and regions gave rise to culture provinces, which were equated to ethnic groups and tribes. This led to the methodological principle that cultural continuity was synonymous with ethnic continuity, hence with the aid of „direct ethno-archaeological methods‟, modern ethnic groups could be traced to their primordial origins (Jones 1997; Veit 1989). This meant that if one could establish material culture that constituted typologies as well as identifying an ethnic group that once occupied such sites or territories, then an ethnic group would have been defined archaeologically. Of course, 44 such approaches are now widely condemned as the last three decades have witnessed remarkable theoretical and methodological re-adjustments in the conceptualisation of identities, and ethnicity (Jones 1997). This chapter therefore zeroes in on such theoretical concepts and employs them in the study of the Rozvi archaeological identity. 3.2 Defining Historical Archaeology The conceptualisation of historical archaeology has varied widely in terms of its research themes, theoretical approaches, and appropriate methodologies. Due to varying theoretical and methodological approaches, scholars across the globe generally employ methods conditioned by their regional and local experiences. All factors cited above, together with the exponential growth that still characterises the discipline today have made it impossible to universally define it (see Pikirayi and Pwiti 1999; Orser 2010). Since it is ever discovering new ways of opening up its subject matter, it has also become impossible to distinguish it from related disciplines. According to Emberling (1997:301) defining a term is to limit it because definitions by nature are rigid and absolute than membership of aspects encompassed by them. Apparently historical archaeology was unsystematically carried out from the 1920s until the 1960s, when major developments in social history and ethnic awareness took place in North America. Thus scholars only started grappling in systematically defining it as the study of European material culture done with the aid of documents seeking to interpret the spread of this culture into non-European areas from the 15th century until the industrial period (e.g. Leone 1977; Schuyler 1970, 1978; South 1977). Posnasky and Decorse (1986) also defined historical archaeology, as archaeology undertaken in periods or for areas in which the principal source of information is documentary evidence. These definitions were problematic in several ways because they implied historical archaeology was only about studying European material culture, using text and not oral traditions, thereby limiting research to describing “who made what, when, where, and how it was used.” All this implied only material culture of literate (European) communities was relevant for study in historical archaeology. Secondly, these perceptions totally excluded the use of oral traditions, which though problematic presents untapped potentials because they address other issues often ignored or forgotten by written documents (Paynter 2000). Finally, the primacy of textual sources in the definition of historical archaeology confused many as to whether the field‟s prime 45 concern was either history or anthropology when in reality it was a mixture of various disciplines. Schmidt (1977, 1983) argues at length about the value of oral traditions and the need to integrate them in defining historical archaeology especially for African research issues. Considering that most parts of Africa were unexplored by documentary sources until around the 18-19th centuries, and that many African people still maintain viable indigenous belief systems, oral traditions should now form an integral part of the definition of historical archaeology. In the Zimbabwean context, only the northern and eastern parts were covered by Portuguese documents during the 16th century, otherwise written sources only covered the south and south-western parts from the 19th century (Beach 1987). Therefore much of what we know about the history of the south and south-western parts of the country comes from reconstructions of oral traditions, thereby indicating their integral role for historical archaeology in Zimbabwe. However this is not to say that one should carelessly apply oral traditions in the practice of historical archaeology. Vansina (1965, 1996) suggests that a critical comparative method, which exploits ethnographic, linguistic and archaeological research, should be employed as a check for oral information. Even up to the 1980s certain scholars were still attempting to define historical archaeology by emphasising on fussy aspects that seemed critical in distinguishing it from history and anthropology. Once again this approach was not fruitful because of the discipline‟s close similarities with history and anthropology, culture and action, ethnography and ethno-history, as well as its dual coverage of the past and present (Paynter 2000). It would be extremely difficult to define historical archaeology on the basis of academic research boundaries because it is a field of study that is always seeking to open up its subject matter by being responsive to intellectual tides in broader disciplines of anthropology, history, and contemporary academic ideology or “political science in the case of Zimbabwe” (Paynter 2000). Phillipson (1985) raised the concern that the informal use of the term historical archaeology has continued at the expense of clarity possibly because of this dynamic nature of the discipline. This fluid nature of historical archaeology has also led other scholars to question whether it is processual or post-processual in nature (Paynter 2000). Such questions clearly demonstrate the confusion that has been created by vast literature dealing with different forms of historical archaeology. Therefore, any definition that tries to distinguish historical archaeology from history, anthropology, sociology, linguistics and any other social science or humanity subject is bound to face myriad problems. This broad nature of historical archaeology led Pikirayi and Pwiti (1999:75) to highlight that it is a study of 46 remains of past societies with the aid of historical evidence, embracing the interests of diverse scholars from archaeology, anthropology, history, geography, folklore, etc. While Pikirayi (1993) defined historical archaeology as a study of sites that can be interpreted with the aid of historical sources (oral and documentary), historically datable artefacts and anthropological information, his definition does not apply to this research because it excludes the notion of ethnicity. In this case, historical archaeology is thus defined as a multi-disciplinary approach incorporating anthropological, historical and archaeological concepts so as to examine the relationship between material culture and concepts of discourse, resistance and identity construction. The approach hereby adopted seeks to determine how material culture fits within systems of meaning and action, by ignoring the meaningless boundary between prehistory and historical archaeology and covering the past and present in a bi-directional perspective in order to present a simple and coherent account of the past (Connah 2007; Paynter 2000; Orser 2008). Orser (2010) observed that due to the rapid growth of historical archaeology many practioners have been forced to concentrate on four key research themes namely; analytical scale, capitalism, vectors of inequality, as well as heritage and memory. He also makes it clear in his discussion of these critical research areas that they are not mutual but closely intertwined, and that these are not the only themes being addressed currently by historical archaeologists. Hall and Silliman (2006) concur that historical archaeology has grown significantly to branch into the following key areas; European colonial expansion, capitalism, the interplay between word and text, as well as text and artefact. In chapter 8, Silliman explains how archaeologists are using historical archaeology to study labour and its relationship to identities like class, race, ethnicity and gender (Silliman 2006:150). He argues that people rationally adopt any of the above listed identities in their various labour struggles, hence when interpreting material culture associated with such identities, it is crucial to appreciate the social context because identities are complex constructions. This research therefore follows the “vectors of inequality” theme, which basically attempts to discern social identities of past peoples using their physical remains. According to Orser (2010), vectors of inequality simply mean that past communities had multifaceted structures of inequality commonly termed identities. Currently archaeologists understand identities as social categories produced through negotiation of power relations in distinct socio-historical settings 47 (Voss and Allen 2008; Voss 2008). These identities are viewed as situationally mutable and contested categories that are open to interpretation, hence the ways in which they are expressed depend on spatio-temporal environments (Orser 2010). This new conceptualisation of identities in historical archaeology marks a paradigm shift from previous analyses that correlated identities in textual sources with excavated artefact associations as though identities were fixed positions that people occupied. Another major flop in this approach to identities was the failure to realise that identities are more about ideas that people have concerning themselves and often imbue selected objects/things from their culture with particular meanings. Therefore in seeking to define identities using material culture, we are actually trying to denote symbols and their meanings for past communities. Edmund (quoted it Schmidt 1983:72) summed it up by stating: „Ideas are more important than things; creative imagination is deeply entangled with the formulation of verbal concepts; archaeologists need to appreciate that material objects revealed by their excavations are not „things in themselves,‟ nor are they just artefacts- things made by men-, they are representations of ideas.‟ Schmidt (1983:73) reinforces this idea by noting that beliefs, symbolism, cosmology and ritual life are accessible to archaeologists especially in Africa, hence archaeologists should work towards discerning non-material meaning attached to material patterning through space. Unfortunately there has been a strong reaction against interpretations that follow a symbolic system of analysis (Schmidt 1983). This brings us to the issue of a methodological framework in historical archaeology that seeks to examine both the intangible and tangible social aspects of past communities. Interestingly, historical archaeology has also been widely defined on the basis of methodologies and this approach has not been very successful either (Paynter 2000; Orser 2008). The major challenge of defining historical archaeology as such has a lot to do with the nature of the discipline, which borrows theories from other disciplines. This flexible nature of the discipline has created a “basic methodology” that varies widely from one scholar to the other depending on the subject under investigation (Orser 2010:116). According to Paynter (2000:16) while historical archaeologists agree that “objects talk about their meaning if placed in their proper contexts,” there are great disagreements regarding the relevant contexts, perspectives and perceptions for meanings and whether interpretations are an end in themselves or part of a larger enterprise. As a result of these conflicting perspectives, approaches towards meaning alone, include structural analysis, 48 contextual analysis, and dialogical analysis, Foucauldian approaches, analysis of ideology, studies of ritual, analysis of „double consciousness,‟ analysis drawn from humanistic anthropology and Hermeneutic readings, among others (Paynter 2000:16). For purposes of clarity, only the relevant approaches to this study are briefly explored below. Structural analysis relate to the analysis of structures of thought or „worldview‟. These in turn influence the invention of codes and rules that find expression in material culture as symbolic concepts. These structural codes and symbols condition human action because they carry certain meanings to members of the society, but these meanings change according to context and association of their visibility. Structural analysis therefore tries to deduce the various meanings imbued in material culture. On the other hand, contextual analysis refers to a methodological procedure that examines the environmental setting of material culture. This analysis is premised on the assumption that „culture explains culture‟, meaning the objects are contained within a culture so their meaning is fixed to that culture and can only be revealed through appreciating such contexts (Gamble 2001). This approach to interpreting meaning depends on a high density of data, the richer the data then the better because meanings of objects are always embedded in context (Gamble 2001). The dialogical analysis is basically an interpretive method that closely analyses spoken and written utterances or actions for their inherent communicative significance. It is an approach that focuses on discourse through oral and written evidence but it goes beyond that by seeking to understand the objectives behind the way people spoke and acted. Foucauldian approaches are analytical approaches inspired by the works of a French social theorist Michel Foucault. His central argument was that power and knowledge are interrelated (Mulligan 2003). Thus every human relationship is a struggle and negotiation for power through the medium of discourse, which operates by rules of exclusion. He defined discourse as systems of thought composed of ideas, attitudes, and courses of action, beliefs and practices. He further argued that discourse legitimises power through the construction of „current truths‟ that control power relations. Therefore knowledge is the creator of power and creation of power (Foucault 1980). As such, objects control discourse as nodes in complex webs of knowledge, power and other ideas (Foucault 1972). Foucauldian approaches are closely related to ideological analysis. Ideology has been defined as a body of ideas that reflects social needs and aspirations of an individual, group, class or culture. Ideology has also been defined as a set of doctrines, beliefs or 49 ideas that form a comprehensive vision for a political, economic or social system so as to offer change or conformity to a set of ideals (Mulligan 2003). Ideological analysis therefore tries to infer meanings from objects, structures or symbols and styles associated with particular social agents and processes that produced them. However, currently there are so many approaches to ideological analysis in archaeology. Studies of ritual have been popular for symbolic interpretations of stonewalled sites and rock art in southern Africa (see Huffman 1981, 1996; Lewis-Williams 1980, 1981). Analysis drawn from humanistic anthropology examine humans in the context of knowledge of symbols in their various contexts, identity and its dynamism; while Hermeneutic readings refer to the circular character of interpreting the meaning of text by referring to its cultural, historical and literary contexts. Therefore most of these approaches to “meaning” mentioned above are closely related although they stress on different subject areas or themes. These approaches aimed at understanding meaning in the past are relevant for this study because the Rozvi identity was never fixed but varied extensively throughout their political dominance. Since Rozvi ethnic identity was situational in nature, sorely adopting a structural analysis that often relies on direct application of the „ethnographic present model‟ would lead to problems in data interpretation because identities are not static, but complex, dynamic and mixed constructions. Cohen (1978) has defined the ethnographic present as an analytical framework that assumes what is observed at a point in time can be turned into a timeless „system‟ whose parts work closely to sustain the whole. Huffman (1996) also defined the ethnographic present as a model that examines living descendants of prehistoric groups, oral and documentary evidence to study societies that existed some 500 years ago on the Zimbabwean plateau and adjacent areas. This was the approach he used to interpret symbolism, use and meaning of space within Zimbabwe tradition walled monuments. The main problem with the ethnographic present model is that it tends to major on circumstantial similarities between the past and present, while paying piecemeal attention to elements of difference that could be key to a clear appreciation of the past (Beach 1998). Beach (1998) further argues that the model is often abused by scholars who are not too familiar with the use of primary historical sources to the extent that even irrelevant evidence is sometimes manipulated or carelessly accepted in interpreting the past. The other weakness of the model relates to the 50 way it implies that chronological and historical aspects have little bearing on the general social structure of societies. Mazarire (2003) also raises the concern that chieftaincies in Africa have largely been derived from their ethnographic present as if history is constructed from the present. Thus ethno-archaeology presents the most appropriate alternative approach for analogical reasoning in this study, because it is more sensitive to context and historical processes (Emberling 1997:312; Hodder 1982). Ethno-archaeology is more dynamic because it accommodates techniques from history, anthropology, ethnology and other related disciplines. Historical archaeology has opened up more research avenues as evidenced by the multiple themes and approaches that now characterises the discipline. Several successful historical archaeological projects have so far been undertaken in different parts of the world. Numerous studies have also been done in Africa, and they are summarised in a number of works (see Connah 1975; Hall and Markell 1993; Kinahan 2000; Lane and Reid 1998; Pikirayi 1993; Posnansky 1969), among many others. Hence this study is only another dimension to its practice in the southern region of Africa. 3.3 Contextualising notions of Identities, Ethnicity, Power, Agency and Material Culture New and exciting questions concerning the notion of identities are increasingly being raised in historical and social anthropological studies today (see Bowser 2000; Gosselain 2000; Hodder 1982; Wells 1998). This follows the realisation that the study of identities holds great potential in opening new avenues when interpreting social organisation and past political realities. Identities have been defined as multifaceted and dynamic socially constructed attributes that vary through space and time for purposes of distinguishing an individual/group from the larger social context (Lucas 2004; Wells 1998). An identity is not a fixed or static attribute inherent in an individual/group, but a dynamic, flexible and contingent entity that is always in a state of change. Identities are situational and dynamic by nature, meaning they are invoked strategically in varying contexts because they define parameters of interaction during individual/group relations (Stone 2003). Since identities serve the purpose of defining and providing a connection/means of interaction for an individual/group, an infinite list of identities such as (race, class, gender, ethnicity, national, regional and family/lineage, political faction, names, nick-names, totems, status and professional identities) has been gradually developed by several anthropologists and 51 archaeologists. In terms of identity formation, Yaeger (2000:126) highlights that local community identity often begins through daily practices that represent nodes of similarities to members. Through such „practices of affiliation‟ a sense of community often unconscious can develop and become more concrete to members ultimately leading to natural communities. These are homogenous, stable and closed social organisations usually characterised by personal daily interactions. Alternatively, when there are no face-to-face daily interactions, but shared perceptions of common origins and belief systems, then imagined communities usually come into being. Castells (2004) highlights there are three forms and origins of identity construction namely; legitimising identity, resistance identity and project identity. Legitimising identity refers to identities constructed by ethnic elites who will be seeking to justify and sustain their power to a wider society through the use of “common social experience” discourses and ideologies. Mhiripiri (2008) elaborates that legitimising identity is initiated by powerful members seeking to extend and rationalise their advantages in social relations. There is also resistance identity, which relates to elite constructed consciousness aimed at mobilising the so-called “victims” to rise against some oppressive system for their own freedom. Resistance identity is thus generated by agents who feel oppressed, devalued, stigmatised or stereotyped by the logic of domination. Sometimes resistant identities may develop into legitimising identities, when concerned actors manage to become the dominant forces in society (Castells 2004; Mhiripiri 2008). Project identity is constructed through the manipulation of any material culture with potential to reinforce a new identity for those actors seeking to alter the social structure. All these notions to identity formation suit Rozvi origins well in the sense that they have progressed from resistance identity, to legitimising identity, and finally to project identity. A detailed coverage of these identities and their relevance to the Rozvi will be explained in more detail in chapter seven. Wells (1998:243) argues that identity is the material outcome of a series of choices made by an individual regarding the character of material culture made and used daily. This assumption explains why archaeologists have been grappling to attach identities to material objects since the inception of the discipline. In line with this trend, cultural and ethnic identities have always been the focus for study, but scholars increasingly became sceptical of methods used in defining such identities. Despite these concerns, historical archaeologists have devised numerous, diverse, and sophisticated analyses aimed at understanding identities as they are produced through negotiation 52 of power relations in distinct social contexts, rather than as fixed entities (Allen and Voss 2008; Emberling 1997; Stone 2003; Orser 2010; Wells 1998). Although archaeologists generally agree that there is a positive correlation between artefact distribution and social boundaries, interpreting these distributions has always been a contentious issue (Gosselain 2000). However archaeologists are now paying more attention to modern theories of ethnicity and identity, which acknowledge the interactive, heterogeneous and dynamic nature of processes through which social differences are formed. By examining identities as processes of identification and differentiation rather than as fixed positions that are occupied permanently by individuals, we are now in a better position to infer on the materialisation of such identities. In this regard, the Rozvi are hereby examined as an ethnic group but it is important to first define, describe and explain their identity construction processes before interpreting their material culture. More so, owing to the complexity of the notion of ethnicity, subsequent sections will briefly explore this concept for purposes of contextualising it in this study. 3.4 Varying perspectives on ethnicity in archaeology Throughout the history of anthropology and archaeology, terms like “ethnic and ethnicity” have always centred at the heart of research enquiry (Jones 1997). Despite the lack of an explicit definition, these terms continued to attract more attention and interest across human sciences. Since few people defined what they meant by ethnic or ethnicity, these concepts increasingly became ambiguous as they were uncritically applied to a wide range of socio-cultural groups in the past. Emberling (1997:301) mentions that the term ethnic was derived from the Greek word “ethnikos” meaning a number of people or animals living and acting together (implying cultural similarity). Isbell (2000:244) argues that in common anthropological usage, „ethnic‟ implies a group was once co-residential. As a result, until the 1970s ethnic groups were perceived as synonymous with tribes, racial and cultural groups constituting real/natural communities. The natural communities were conceptualised as relatively static, conservative, homogenous, and closed social units maintained by residential proximity and common daily experiences (Canuto and Yaeger 2000). Under this primordial approach, ethnicity was considered as largely unconscious of structure but isolation, adaptation, shared residence, experience, knowledge, sentiments and goals brought about group solidarity. This view of ethnicity was problematic in 53 that it presented ethnic units as natural, static, homogenous and permanent entities that are basically unified by emotional attachment. Iron Age studies in southern Africa by scholars like Huffman (2005, 2007) perceived ethnicity (especially for Zhizo communities making Leokwe ceramics) as constituting autonomous kin-based groups with a shared culture. This approach to ethnicity and ethnic groups seriously confined research because it became difficult to distinguish ethnicity from other social identities like kinship and researchers became obsessed in defining „general similarities‟ at the expense „differences,‟ which otherwise held more value in investigating past realities. Calabrese (2005) however favours an interaction approach to ethnicity, which has great potential for archaeologists interested in interpreting the dynamics between material culture and ethnic identities. These aspects are covered to some reasonable depth in the foregoing discussion. Francis (1947) and Emberling (1997) comprehensively described ethnogenetic and maintenance processes, which are commonly achieved through constant negotiation and reification of differences, but once that stops ethnicity ceases to exist as an organising structure. Ethnicity can originate in hunter-gatherer societies, complex state systems and in capitalist or modern societies. Since this study is focusing on a pre-colonial group that eventually developed into a complex state system, ethnicity related to socio-political complexity is prioritised. According to Emberling (1997) when a state conquers or assimilates formally independent groups, the victor or victims may initiate ethnicity for various reasons. He also mentions that migration (forced or voluntary) into new areas may also influence a sense of distinctiveness from one another or enhance a sense of unity due to movement and isolation. It is also possible that ethnicity can emerge in complex socio-political state systems as a form of resistance when leaders mobilise members against perceived oppressive ideologies or forces (see Preucel 2000). It should be borne in mind that in any of these ethnogenetic processes, about five conditions should exist if ethnicity is to develop as a rallying point for group membership (Emberling 1997). Firstly, ethnic members should see themselves as having a common ancestry or a common descent (cultural, genealogical, genetic or kinship). Secondly the kinship idiom should include other broader social categories like extended family, lineage or clan. They should speak at least a mutually intelligible language and there should be some hierarchical relationship to other nearby groups or states. Last but not least, there should be some competent and 54 knowledgeable individuals who act to stimulate perceptions or feelings of common interests and goals for group membership (see Bentley 1987; Cohen 1978; Emberling 1997; Francis 1947; Isbell 2000; Preucel 2000; Norton 1993; Stone 2003). Essentially, ethnic consciousness is most often manufactured and legitimised through discourses of common ancestry, mythology, ritual practise, buildings, objects, and landscapes by ethnic elites. The discourse of a common ancestry is vital for ethnic consciousness because it suggests collective memory of former unity, territorial origin and political control (Emberling 1997; Francis 1947). In the absence of such a common history, agents actively construct fictitious accounts through myths that comment on certain historical events that prevailed in metaphorical terms. Anthropologists now commonly agree that mythical traditions about the remote past are strong indicators of ethnic affinities. According to Finley and Vansina (quoted in Loubser 1988:302) mythical accounts should be understood as stylised representations of past societies streamlined by selective memory into certain historical themes that are valuable for the group. The interaction school of thought propounds the second perspective to ethnicity, which argues that ethnicity stems/emerges from contexts of competition among individuals or groups. In this case ethnicity is invoked as a tool by these competing parties and this is why some scholars term it the „instrumentalist approach‟ because it treats ethnicity as a situational tool used to attain social, political and economic goals by particular individuals or groups (Emberling 1997). From the 1970s theoretical approaches were changing to perceive ethnicity as one of the most important structuring principles that has no existence apart from inter-ethnic relations (Emberling 1997; Stone 2003). Current social theory holds that identities like ethnicity are actively constructed by individuals who are fully conscious of various cultural practices that can be exploited to either include or exclude other community members. In the broadest sense, they define ethnicity as a socio-political consciousness involving general process of identification and differentiation commonly used to gain power over political, social, and economic issues by individuals or groups (Cohen 1978; Emberling 1997; Eriksen 1991; Stone 2003). Jones (1997:108) argues that ethnic groups are rarely a reflection of the sum total of similarities and differences in cultural trait lists, but they are self-conscious and self-defining groups based on real and imagined cultural difference. The underlying idea in all these views about ethnicity is that it is a constructed identity that is strategically expressed in varying contexts through 55 carefully selected elements of culture that either include or exclude members in competition for resources. Thus, ethnicity cannot exist outside of competition, power relations and defined cultural systems. This constructivist, situational, contingent and dynamic nature of ethnicity informs us that it is an identity that requires in-depth explanation. The explanations should largely centre on clarifying how ethnic groups originate, survive through time, and eventually sink into oblivion. According to Stone (2003:35) ethnic groups cannot exist in the absence of competition for power, and this means ethnicity is an identity that is called into operation in opposition to one or more groups. Usually those who invoke or construct ethnic identities are referred to as „ethnic leaders‟, „ethnic elites‟, „catalysts‟ or most commonly „agents‟ (see Emberling 1997; Francis 1947; Stein 1998; Stone 2003). Norton (1993) demonstrated that these agents behind the construction of ethnicity should be sensitive to social experience or cultural context in order to draw relevant discourses of common interests. Failure to appreciate common social experiences by agents often results in failure to gain social support by these leaders. These agents are therefore responsible for selectively using cultural elements to create and sustain an ideology of shared sentiments and exclusion when competing for scarce resources. The selected elements of culture used to mark boundaries or differences between groups are variously termed, „allegiance to some external object‟, „elements of tradition‟, „language games‟, „socio-cultural diacritics‟, or „cultural symbol or practice‟ as objects for contemplation (Bentley 1987; Cohen 1978). All the terms refer to elements of culture, like religion, language, political ideology, and distinctive territory, common scale of values centred on some government, class or race and totemism as signifiers of ethnic identity (Bentley 1987; Cohen 1978; Eriksen 1991; Francis 1947; Norton 1993). Ethnic markers are diverse in nature because membership into an ethnic group is not a given but a negotiated position by those within and outside the group. Ethnicity comes from both ascription to by outsiders and self-identification by group members, meaning there has to be agreement from outside and inside (Emberling 1997:304). Ethnicity therefore becomes a dynamic and complicated process of identification in the sense that multiple ethnic markers may be used for identification by the same group depending on the nature of advantages being sought. For these reasons ethnicity is not an absolute or objective identity but a fluid, fleeting and situational entity. 56 A central element of this research is derived from the concept of „imagined communities‟. This notion emphasizes that ethnicity is constructed through competing discourses about power and its representation in the social, economic, ideological, military and political relationships. Power has been defined as an abstract, volatile, and fluid phenomenon exercised either collectively or individually by people over each other or their environment to pursue and attain goals (Mann 1986; Stein 1998; Robertshaw 1999). In this case both physical/coercive and creative/ideological power shall be analyzed in order to understand how the Rozvi influenced power relations. Stein (1998) also notes that research trends are shifting towards the conflict-based model that focuses on agency and power relations in describing the working of complex societies. This model seeks to identify social groups and define how they were organized so as to determine their power relations. It would therefore be vital to understand how the Rozvi constructed and reconstructed their identities throughout their political dominance. Owing to this nature of ethnicity, Stone (2003) notes that participation in an ethnic group provides support and organisation or rather a political power block that can be deployed to assert political rights. Emberling (1997:310) says that ethnic emphasis gives larger political support because in ethnicity an individual member has potential to represent the whole group thereby dramatically increasing its political influence. Thus if a number of individuals in positions of influence choose to emphasise that particular ethnic identity in structuring political and economic processes, then the group‟s influence spreads. In this study, this neatly fits Rozvi expansionist policy, whereby they identified different members in positions of influence like chiefs, religious leaders and the Nyai (Mwari religion messengers), whom they assimilated and exploited in governing „others‟. Hence ethnic consciousness is largely a product of powerful agents who define meanings of social relations in order to legitimise and sustain their power. Stein (1998) envisages that these agents exploited two kinds of strategies namely „network-based‟ and or „corporate based‟ in order to guard against their power. If the corporate strategy was deployed then a hierarchy (unequal and stable) social organisation developed. On the other hand, if a network-based strategy was deployed then a heterarchy (decentralised, fluctuating and unranked) social organisation with related elements to one another and having potential for ranking in a number of ways developed. In this case both strategies cited above were employed by Rozvi agents during political dominance. 57 These perspectives on power relations in complex societies have great potential in the analysis of how Rozvi elites were organized and constantly adapted to change. All these complexities demonstrate that actors in the past were highly reflexive and conscious of social structure, power asymmetries, negotiation of power and identity by themselves and others in pursuit of power and resources. In view of these renewed interests into notions of ethnicity and power, the past two decades have been characterised by increasing dissatisfaction with deterministic and mechanical approaches to identity analysis. Thus the interaction school of thought has been criticised for ignoring the importance of habitus and culture as constraining forces when individuals negotiate, manipulate and construct meanings for identification purposes. Interactionists argue that ethnic groups are best understood by examining subjective trait lists of „difference‟ rather than objective traits enclosed by boundaries (Barth 1969; Cohen 1978; Emberling 1997). Barth (1969:10) observed that ethnicity is defined on the margins, rather than the centre of a homogenous group because ethnic groups define themselves vis-avis other groups. He further elaborated that it is the ethnic boundary that defines the group, not the cultural stuff it encloses (Barth 1969). It therefore follows that, the interaction school of thought emphasises that individuals have freewill to select and objectify any element of culture to mark difference from others without any social or cultural constraints. This is problematic because, even though individuals can exercise their choices freely, sometimes their actions are unconsciously performed due to cultural influence. On the other end, the primordial school of thought emphasises habitus as the central issue in the study of ethnicity. Those who subscribe to the habitus concept argue that when examining ethnicity we should not focus on subjective symbols that mark difference, but on how members identify each other objectively (Bentley 1987; Bourdieu 1977; Jones 1997; Shennan 1989). The habitus concept is based on assumptions that there are certain sets of underlying rules with lasting predispositions that condition human behaviour, perceptions of the world, and interaction starting at family level to the larger society, and that solidarity is reinforced through daily activities (Stone 2003). Although valid in highlighting how culture conditions human action, this approach exaggerates the role of unconscious predispositions as forces for ethnic affiliation. Worse-still the habitus concept largely promotes redundant normative and essentialist paradigms in explaining social organisation and culture change. However the habitus concept 58 should not be totally disregarded but its strengths should be merged with those of the interaction school of thought in order to develop an agency oriented approach. 3.5 Defining agency theory in archaeology Thus Antony Giddens was inspired in such a manner to propound the agency theory, which tries to balance notions of interaction and habitus. Agency theory is premised on the assumption that a dialectical relationship exists between agency and structure. In that regard, agency refers to human action inspired by rational ideas, while structure encompasses institutions and traditions that constrain human action. These constraints could be gender, race, status, material culture among others, but individuals respond differently to these hindrances in their efforts to attain goals (Dornan 2002; Joyce and Lopiparo 2005; Stone 2003). In general, agency theory emphasises that different individuals or groups are highly conscious of different social categories that condition their behaviour and aspirations, but do not just give in to these impediments. Just like any other ground breaking theoretical analysis, the agency approach has been quickly adopted and applied in sociology, anthropology, linguistics, and archaeology among other disciplines (see Dornan 2002; Joyce and Lopiparo 2005). The ultimate result of such a trend is that the theoretical premise becomes too broad to the extent of losing clarity. It is crucial to pay special attention to how agency is understood purely from an archaeological perspective in order to integrate it in the analysis of Rozvi identities. Dornan (2002) reviewed agency in archaeology, which is summarised here for purposes of this study. The analysis of agency has focused on three key issues namely; resistance and rationality, intentionality and consequence, and finally the unit of analysis, and these are also outlined below. The notion of resistance and rationality has been approached from two different dimensions. One school of thought equates agency to resistance because Bourdieu (1977) and Giddens (1984, 1991) centralised resistance to class inequalities. Subscribers to this premise agree that agency is any action that does not conform to normative traditions, hence mundane action can be imbued with a sense of resistance to those who want to maintain tradition (Dornan 2002:318). Agency is viewed as resistance to oppression and domination coming from customs, traditions, transcendental will, or any other obstacles. Rationality on the other hand views agency as related to interests, and motives of past actors in dealing with ideology, power, identity, tradition and many other constraints (Dornan 2002). Although these two approaches differ in their views of 59 agency the general view is that agency is about intentionality and constraints that need to be constantly manipulated or negotiated. The issue of intentionality versus consequence tries to explain social or cultural changes that are associated with agency. The argument raised for intentionality states that agents are intentional beings who are motivated to bring about change. Hence in archaeology, when we examine changes in material culture we should understand that as structure that was consciously manipulated by agents. Often the question on the reproduction of oppressive structure through actions intended by agents to resist it has been raised. Hence, this question can only be addressed by adopting the habitus concept. Regarding consequence, tensions centre on the fact that change in (social/material culture) is not intentional but accidental, that is, while agents are fighting against structure, certain unintended changes may happen. This is also a valid view because it counters determinism in the explanation or interpretation of historic changes. The unit of analysis problematises the definition of an individual in agency, because individuals are difficult to access archaeologically. However theoretical advancements in the analysis of identity and ethnicity construction in archaeology are opening new windows because they shift focus from individual members to the examination of indigenous socio-political systems, sources of power, and competition for scarce resources (including power) that individuals participate in (Dornan 2002; Orser 2010; Stone 2003). Ethnic identities are either invented or imagined through the efforts of certain agents who realise that there is power in numbers. As such these individuals exploit ethnicity as a role that calls for common views, interests, feelings, goals and behaviour that oppose the so-called “others”. In order to resonate with cultural experience, symbols are drawn from culture but new meanings are usually added. Material objects, landscapes and associated physical features, cultural features (stone-walled monuments, mounds and prominent houses etc.) are commonly employed as symbols of identity. This is because some of these symbols can be used to mark difference, they can also be distributed strategically and given meaning in the process of usage. Most importantly objects or landscape features are commonly used to signify identities because they live longer than humans, hence they can carry the same message or different messages for multiple generations of people (Zivenge 2010 pers.comm.). 60 3.6 Towards a definition of Rozvi identities Determining who the „Rozvi‟ were requires a careful examination of all the spatial and temporal contexts in which they were mentioned. Although many of our sources about the Rozvi are entirely external in the sense that views about the group came from outsiders, a comparative analysis of these sources is useful. At least now we know that before the 1680s there was no mention of the name Rozvi because such an identity did not exist on the Zimbabwean plateau. Mazarire (2009) states that although a certain Changamire (personal name unknown) broke away from the Mutapa state in the 1480s-90s, he was not the founder of the Rozvi people. If ever the Rozvi were in existence during the 15th century, the Portuguese would have certainly mentioned them as they did concerning the Torwa. According to Beach (n.d., 1994a) it is difficult to prove that this early Changamire was related to Changamire Dombo and the Rozvi, because his descendants were last heard of in 1547 when they were defeated by the Mutapa. By assuming that there was a connection between this early Changamire and Changamire Dombo, we are led to treat the Rozvi as having originated from Butua when they launched attacks on the Mutapa and Portuguese in the late 17th century (see Mudenge 1974b, 1988). Mazarire (2009) also notes that the early Changamire of the 1490s was a Torwa ruler who broke away from the Mutapa. This means that the identity of the early Changamire remains highly contested especially when trying to relate it to the Torwa or Rozvi respectively. In fact, any attempts to trace the Rozvi to this early Changamire should be viewed as failure to appreciate that Rozvi identity was a constructed consciousness, and not a simple kinship or genealogical concept that could be easily traced through blood ties. Changamire Dombo who is commonly credited as the Rozvi founder, is variously traced to the central and north-eastern parts of the Zimbabwean plateau namely; Mbire or Njanja, Murewa or the general lands between Nyadire and Mazowe rivers, or those territories between Mutapa and Manyika but close to Manyika (Beach n.d., 1980, 1994a; Mudenge 1988). He started his political career as an ordinary herdsman of Kamharapasu Mukombwe who was a leader of the Mutapa state (Mudenge 1988). During the 1650s the Portuguese were expanding their prazos (farms) into the Mutapa state, thereby forcing the locals to pay kuruva (a form of tax meant to force Africans to work freely for Portuguese farmers). In addition to that, Africans were also forced to work in the mines just like slave labourers, hence these developments facilitated the revival of „nyai 61 clientelism‟ (Mazarire 2009). Nyai clientelism was part of a social and military system that characterised the Mutapa state, which revolved around provincial barons who were eying the royal Mutapa title (Beach 1983a; Mazarire 2009). These provincial barons attracted young men seeking cattle to use as bride price. In most cases these young men were loyal and served as military clients who furthered political interests of their leaders (Beach 1983a:265). As such the nyai evolved into major warrior classes that varied in size and fighting skills, thereby creating volatile power structures that were scattered throughout the plateau. Under this system leaders with plenty of cattle wealth such as Changamire Dombo attracted more youngsters and became more powerful in the process. According to Beach (1983a) Changamire Dombo‟s group emerged as the most successful nyai faction. As a result, by the early 1680s Changamire Dombo had rebelled from the Mutapa state to start his own political campaigns. Thus in 1684 Changamire fought the Portuguese at Maungwe and emerged victorious. He continued his raids on the Mutapa state; particularly at Dambarare, Mukaranga, Ruhanje, Hwangwa and Manyika between 1684-1695, causing total destruction on many Portuguese forts (Mudenge 1977, 1988). Through these raids, Changamire managed to eject the Portuguese from Manyika before he finally moved south-west to take over the weakened Torwa state in 1696 (Beach n.d., 1980, 1983a, 1994a, 1994b, 1995; Mazarire 2009). The nyai may be better understood through the notion of agency, which though complex, yields valid insights on social formations. Thus the most appropriate agency approach could be one which argues that the past was not only shaped by collective action but by independent individuals pursuing purposeful, rational action and sometimes responding to unconscious dispositions. As such the nyai would fit into the practical rationality approach to agency. This approach allows us to interpret notions of power inequalities and resistance. Basically it argues that leaders do not impose change, but social change is the unintended consequence of struggle and resistance against the accumulation of social power by competitors (Dornan 2002). In this regard nyai clientelism was more about the ability of certain individuals to manipulate other members in order to bring about collective action that did not necessarily bring about collective interests but individual aspirations to reality. 62 Rozvi identity therefore emerged from a context of intense competition for economic and political power involving local African chiefs and the Portuguese who were interfering with African affairs. Zachrisson (1978) observes that the kingdom of Guruuswa was under the Torwa dynasties until 1683 when Changamire took over Torwa, Manyika and larger sections of the Mutapa state. Changamire Dombo also drove away the Portuguese and it was during that time that his group of fighters earned the nickname Rozvi, meaning destroyers or despoilers (Zachrisson 1978:14). Mudenge (1974b:26) points out that Portuguese reports of these early wars did not use the name Rozvi to describe Changamire and his followers, but simply called them „terrible Karanga warriors‟. It has also been mentioned that the name Rozvi could have been derived from “kurozva” meaning destroyers (Bullock 1927; Posselt 1935). This is an acceptable view, but meanings of names and their associated transformations in usage in any culture are better appreciated through the concept of historical epistemology (Zivenge 2010 pers. comm). Many names for either people or things associated with a particular culture often carry meanings relating to particular historical events. On the contrary, any names borrowed from other cultures lack cultural/historical meaning (Zivenge 2010 pers. comm.). Pfukwa (n.d.:1) also postulates that a name is basically an onomastics label or part of speech but with time it acquires attributes that make it more than just a linguistic attribute. He adds that the concept of descriptive backing (which constitutes propositions concerning the identity of the name bearer) takes the name from the linguistics domain into the social sciences (Pfukwa n.d.). In the social sciences, a name determines social relations and social relations determine naming patterns and processes (Allen 1983; McDowell 1981). Josephs (quoted in Pfukwa n.d.) argues that a name expresses an aggregate of identities, which can be current or previous; hence naming can be viewed as an act of self-perception, self-praise or disarming the hostile „others‟. As such a name can be an expression of identity whose meaning is in constant motion from the time that it is adopted and given meaning. Therefore naming processes and patterns are closely related to ethnic identification and power because they are often used by different agents to take social, cultural or political space (see Allen 1983; Pfukwa n.d.; Squire 2006). Through time the Rozvi identity acquired so much power as a result of the military history registered by Changamire Dombo, mythological versions constructed to glorify the group, religious alliances forged by various Rozvi leaders, as well as acquired social and political responsibilities like allocating land to newcomers and installations of local chiefs. 63 As a result, this constructed ethnic identity became so influential in shaping power relations during and even after Rozvi dominance. Zachrisson (1978:16) highlights that certain chiefs like Tumbare, and Chihunduro had to change their totems to moyo of the Rozvi in order to reinforce their political power. Bhila (1982a:93) also adds that some Manyika rulers invoked the Rozvi identity in succession disputes and for self-protection against Nguni invaders. The Portuguese also invoked the Rozvi name in their trading relations because they appreciated the authority and power of such an identity (Bhila 1982b). Hence the Rozvi ethnic identity was a reckoned political and economic weapon because it remained relevant to varying socio-political contexts. This fluidity fits quite well into the notion of ethnicity especially considering that it was an identity that eventually became a source of both political and economic power. Several other groups not originally part of Changamire Dombo's followers were co-opted into the Rozvi system thereby creating a dynamic, heterogeneous, and powerful identity for both the ruling elite and their followers. The plastic, fluid, fleeting and dynamic nature of the Rozvi befits the instrumentalist approach to ethnicity. Zachrisson (1978) observed that the Rozvi were not a homogenous group because they had different totems and they had no genealogical ties with the Changamire dynasty though sharing the ruling of the dynasty. There were also some Rozvi groups of the moyo totem who were related to the ruling dynasty but never took part in ruling and most often they lived like ordinary Shona people commonly found on the borders of the Rozvi core area (Zachrisson 1978). According to Beach (n.d.:7-8), Rozvi was a term used by the Portuguese to identify soldiers and supporters of Changamire rulers of the 18th century whose state was called Rozvi. The Shona used it in the same way but extended it to several groups though it originally applied to Changamire dynasty of the moyo totem and various branches of that family scattered across the plateau (Beach n.d.). Beach (1994b) refers to these as minor Rozvi groups because they shared the same totem and territorial origins with Changamire Dombo, but never actively participated in Rozvi state political matters. During the late 18th century a number of Rozvi dynasties broke away from the south-west thereby creating new relationships and forging political alignments, all that further spread the name far and wider (see Beach n.d., 1980; Ncube 2004). It therefore becomes very difficult to give a universal definition for the Rozvi because that identity meant different things through time and space. 64 Currently it is widely accepted that the expression of ethnicity often utilises or sometimes revolves around, material culture manifestations (Harrison 1999). Naturally it follows that material culture signifying Rozvi ethnicity could also have been characterised by spatio-temporal variations as well. Hall (quoted in Meskell 2002:286) further argues that identities are constructed within discourse and are produced in specific historical and institutional sites within specific discursive formations and practices. Therefore this situational ethnicity thrives on ideologies that take advantage of particular elements of cultural traditions or customs as symbols to mark boundaries for both members and non-members. The process of signifying these ethnic cultural symbols is what is often termed 'objectification' of culture (Norton 1993). These various boundaries are the reason why ethnic groups are considered dynamic, because boundaries are constantly negotiated in order to either include or exclude others for specific interests. Since the Rozvi were a mobile community that was always joining and at times assimilating new members, then it is likely that their conceptual boundaries were ever fluctuating. At times the Rozvi used material culture to communicate these social boundaries. Oral traditions highlight that sites like Khami, Danamombe, Matendere, Gorongwe and Great Zimbabwe were the work of the Rozvi (Bullock 1927). Ncube (2004) also recorded several traditions that mentioned that post-Rozvi groups that migrated to the north-west during the 19th continued building several Khami-type structures. Obviously, these traditions hardly considered the chronology of these sites, because traditions simply sought to transmit Rozvi ideologies. Beach (1980:196) argues that all sites in the Insiza district were built by the Torwa and the Rozvi only made minor additions. He further argues that during Rozvi political dominance, Khami was under a Kalanga Ndumba ruler and not the Rozvi. Traditions referring to the Mutinhima Rozvi house that lived in the Mulungwane hills are often mistaken as referring to Khami (Beach 1980:192). Therefore the profuse use of dry-stone structures and mountain imagery as Rozvi identities in traditions possibly signalled Rozvi ideological power. According to Huffman (1996:24) oral traditions emphasise that ruins had to have grandeur of mountains because mountains symbolized political and spiritual authority. Sinamai (2003) observed that meanings of zimbabwe sites always changed because they were used in making power visible to subjects. Traditions identifying the Rozvi with all prominent dry-stone structures were not necessarily wrong but they were simply commenting on the political power of Rozvi elites metaphorically. All pre-historic studies perceive major ruined monuments in southern Africa as former dwellings 65 of the rulers who were wealthy and powerful. Rozvi elites could have perpetrated these views in oral accounts by simply accepting undue credit they were given for the construction of all ancient dry-stone structures. In view of the foregoing discussion, it can be stated that Rozvi ethnic identity was characterised by myriad mythical traditions for purposes of reinforcing various attributes of the group‟s unique qualities. Apart from common ancestry and myth, ritual practices are often manipulated or objectified from culture by ethnic elites in order to integrate people of diverse backgrounds (Preucel 2000; Stone 2003). These are particularly crucial for ethnic identification because they emphasise objects, meals, dressing and even behavioural practices as symbols for group identification in the public arena. These ritual practices have great emotional strength and they can be carried out whenever there is need to include or exclude „others‟ in the public domain. Discourses related to material culture are also critical for ethnic consciousness. These are addressed in chapter five of this study. Be that as it may, Rozvi discourses were manufactured in order to define and use that power to gain political, economic and social advantages over others. 3.7 Conclusion This chapter did not propose any new theory or method in the study of identities in archaeology, but only emphasised the analytical concepts that may be appropriate for defining Rozvi identities. Therefore archaeologists should critically adopt, elaborate, and apply these analytical concepts from social and cultural disciplines. Fortunately, historical archaeology has evolved in that respect, and it has been adopted to examine the archaeological identity of the Rozvi. Their identity encompassed ideologies, mythology, taboos, heroes, religion, totems, material culture (etc.), thus notions of ethnicity, power, agency and symbolism automatically become central to their study. Material culture is particularly crucial in defining identities because it actively communicates social meanings. Subsequently, archaeologists no longer perceive stylistic attributes in material culture as merely functional or passive, but as active symbols that carry social messages. Although typological classifications are becoming unpopular in identity studies today, they are however integrated into this study in order to denote artefact varieties. Formally, varieties were discarded as ambiguous classes needing no attention but that has since changed. They are now regarded as vital in tracing identities because salient and mundane objects often signify agency. According to Pikirayi (2007:298) agency as a methodology should be the subject 66 for present and future research because of its potential in interpreting material culture on the basis of historical and cultural processes. 67 Chapter Four Research Methodology “... past complex social interactions that we are interpreting cannot simply be explained through measurements, percentages and ratios. Real meaning will be derived from sustained contact research into local communities and their connections to the broader region, thus a diachronic, multi-scalar approach” (Cohen n.d.:17). 4.1 Introduction It is crucial to adopt an agency approach when studying ethnic groups in archaeology because historical processes generally determine ethnic generation and expression. Thus an adequate appreciation of Rozvi social practices and relations is very critical because ethnicity is first and foremost a social and political identity. Therefore a qualitative research methodology starting from a robust survey of archival sources (ethno-historic sources by the Portuguese and related historical documents) on Rozvi related dynasties and their geographical locations marked the initial stage of this research. This was aimed at gathering information to distinguish the Rozvi by marking their political boundaries or spatial geographical limits. This stage was also necessary in investigating the possible nonmaterial characteristics that could have been vital to Rozvi ethnic generation. This initial phase helped in determining the social/cultural and political discourses that marked Rozvi ethnogenesis, as well as the maintenance and disappearance of their ethnic identity in relation to material culture indices. This discourse analysis was also critical in defining the Torwa who preceded the Rozvi at both sites being investigated in this study. Basically, a discourse centred approach strengthened the analysis of sociological contexts that influenced variations in meanings ascribed to material culture. Ethnicity is a self-conscious identification that is continuously produced in efforts to remain relevant to different social and historical contexts (Yinger 1985:162). This scenario obviously yields different cases from an archaeological perspective because material culture often used in the communication of the same ethnic identity may vary both quantitatively and qualitatively (Jones 1997). Following this premise, subsequent phases of the research were characterised by both quantitative and qualitative approaches involving fieldwork, and analysis of museum collections. Fieldwork took two forms, namely, an ethno-historical study on post-Rozvi houses in Bikita and architectural documentation of dry-stone walling at Danamombe and Khami sites. 68 Interviews (ethnographic inquiries) were specifically conducted to obtain additional information on how the Rozvi perceived themselves in relation to material culture and “other” people. Architectural documentation sought to establish the inherent stylistic aspects for comparative purposes and tracing Rozvi identities. Museum collections like ceramics, beads and other finds from the two sites were also examined so as to trace Rozvi ethnic expression. Other secondary sources like dissertations, journals and published books, with relevant data on material culture from Khami and Danamombe were exploited in order to enrich the interpretation of Rozvi material indices. 4.2 Written sources Documentary sources were consulted in order to identify „discursive formations‟ (sets of statements that govern strategic choices, object appearance and their distribution, see Mulligan 2003:429) by comparatively analysing statements and determining their regularities about the Rozvi. This archival analysis was also aimed at establishing Torwa and Rozvi origins, their geographical locations and their „objects of discourse‟ (material culture signified as part of ideological processes), which are those things that can be identified, observed, classified and analysed archaeologically. In this case several visits were made to the National Archives of Zimbabwe (NAZ) in Harare, libraries at institutions like the Midlands State University (MSU), University of Pretoria (UP), University of Zimbabwe (UZ), and the Zimbabwe Museum of Human Sciences (hereafter ZHMS) in Harare. However Portuguese primary sources that mentioned the Torwa and Rozvi were extremely rare at NAZ, hence some extracts found in this study were drawn from secondary sources like Axelson (1960), Mudenge (1974b) and Randles (1979) who used archival repositories outside Zimbabwe. This detailed documentary review was useful in establishing most of the various elements of culture that were objectified in the construction of Rozvi identities. Subsequent archaeological research was also partly influenced by findings from this initial approach. Nonetheless, many scholars find it difficult to reconcile information from documentary sources with that from the archaeological record something that Beach (1980) and Pikirayi (1993) have done reasonably well. Beaudry (1988:1) states that many researchers assume archives are lacking in prehistory, thus research designs appear as though scholars want to discover whether archaeological sites properly reflect documents or vice versa. Other scholars also argue that documents should be 69 viewed as problematic sources of information in and of themselves requiring careful study and interpretation (see Little 1992; Paynter 2000; Preucel 2000; Orser 2009). Mazarire (2010 pers. comm.) also highlighted that many archives were not meant for African history, hence when we study them we are not studying the history but the politics of the construction of local history. He went on to add that history has suffered a lot from imposed theories and methods to the extent that we read more about what historians thought about people, rather than what the people thought about themselves. However due care was exercised in the use of documents as useful guides in defining identities in archaeology. In view of the above concerns, a „middle-range theory‟, which suggests a comparison of results from documentary study and the material record, was adopted. Middle-range theory was collaborated with „appropriate epistemology‟ (the use of multiple lines of evidence in the verification of data from an insider‟s perspective). Integrating these two concepts facilitated the development of an analysis that treated historical sources and archaeological evidence as independent but complementary disciplines in the study of identities. 4.3 Oral traditions Historians interested in the pre-colonial past of south-central and western Zimbabwean communities such as the Torwa and Rozvi have also turned to oral traditions to overcome the lack of written sources. These traditions were reviewed in order to understand how the Rozvi identified themselves in relation to their origins, religion, settlement and material culture, political roles, kinship ties, and military power. Oral traditions are generally defined as unwritten forms of information covering categories like narratives, legends, anecdotes, proverbs and historical lays (Vansina 1965). Later, Vansina (1985) redefined oral traditions as documents of the present, which also carry messages from the past. In this case he considered „oral materials‟ or „oral data‟ referring to proverbs, poetry/songs, lengthy historicized text, or epics as part of oral traditions (Henige 1988:232). Henige (1988) shares the same sentiments as he defined oral traditions as a genre constituting „oral materials‟. This concept of „oral materials‟ allows us to exploit sources like poems, novels, songs and any other forms of traditional literature for purposes of soliciting historical information. Such an approach has been very popular among historians Vansina (1965, 1985, 1996) being the most outstanding, hence it was adopted in this study in order to analyse several views about the Rozvi from Shona praise poetry and historical novels such as Nhume yamambo and Misodzi dikita neropa both written by Nobert Mutasa. 70 A lot of oral traditions were also collected by colonial administrators, travellers, and some anthropologists who sought information for their own use and reached conclusions that could have been biased (Curtin 1969). These old traditions however remain valid because they were collected before societies were seriously affected by modernization and political change (Curtin 1969). Fortunately for the Rozvi, a reasonable amount of oral traditions which communicated indigenous views were recorded in the Native Affairs Department Annual from as early as the 1920s until the 1980s (see Beach and Posselt 1986). These annuals largely dealt with African affairs from a contemporary perspective, making them invaluable sources of information. As such, for purposes of this study, oral accounts were collected in two main phases, with the initial one focusing on archival files at NAZ and other related works by scholars like Abraham (1959), Beach (1976, 1983b), Bhila (1982a), Bullock (1927), Fortune (1956), Hodza and Fortune (1979), and Robinson (1959). The second phase relates to traditions, collected as part of an ethnohistorical investigation that was undertaken among post-Rozvi houses now living in Bikita. The ethno-historical study was conducted in Bikita because this is where we find the direct historical descendants of the Rozvi who once resided at one of the archaeological sites under investigation. According to Spear (2006:309) oral traditions constitute a methodological tool for reconstructing pre-colonial history. Often oral sources such as testimonies, life histories, and oral traditions have a broad relevance to all African history (Spear 2006). This means that if we are to trace the history of pre-colonial groups that did not have the skill of writing, oral traditions may provide useful insights into their past. Philips (quoted in Spear 2006:306) notes: “We must… use African sources in reconstructing the African past… to understand the ideas of the African time and place we study, to realize how Africans of the past conceptualized the world around them… and to try to figure out how they would have thought about the changes that were happening around them.” A number of scholars acknowledge their value in tapping local knowledge and epistemology as certain traditions were preserved and passed on for political, social and economic functions. Beach (1976) observed that oral traditions constitute informal learning processes often devoted to the histories of ruling dynasties (lines of descent and land rights) rather than the general masses. These traditions can either be myths commenting on current events in metaphorical terms or accurate historical facts (Beach 1976; Loubser 1988). In most cases oral traditions often constitute the bulk of written records but many scholars still view them with considerable 71 scepticism (Curtin 1969). Several concerns like loss of human memory, local language barriers, individual agendas and biases, potential feedback from written sources, chronology or actual date problems, willingness and performance of the interviewee(s), historical context, professional background of oral traditions collectors among other reasons have been cited as major challenges (Beach 1976; Curtin 1969; Spear 2006). Beach (1983b:30) points out that: “…, African oral tradition does not consist of a pure stream of information flowing unhindered from the past to the present, waiting to be tapped by academics.” Vansina (1965) also argues that most oral traditions contain kernels of historical truth, but if a tradition has internal contradictions, or goes against facts established in other sources of evidence then it should be regarded as unreliable. Philips (quoted in Spear 2006:306) continues: “It is not sufficient simply to accept a particular scholar‟s conclusions, but, as with all sources, we must learn to treat them critically by becoming literate in a methodology and epistemology of disciplines far removed from our own.” An ethno-historical study was thus undertaken among post-Rozvi houses in Bikita in order to minimise biases associated with oral traditions and to strengthen analogical reasoning. 4.4 Ethno-historical research Ethno-archaeology is closely related to ethnicity, which is the primary focus of this study. Agorsah (1990:192) defined ethno-archaeology as a tool used to explain and understand past human behaviour on the basis of models derived from observed cultural processes in traditional and modern societies. However, simple collection and description of ethnographic data by archaeologists is not ethno-archaeology, and in certain instances “direct analogies” derived from such approaches are limiting (Agorsah 1990; Gould 1980; Gould and Watson 1982; Hodder 1982). Hodder (1982) therefore proposed the use of “relational analogies”, which can be equated to ethno-historical studies such as the one conducted by Gould (1966). Wylie (1973:708-9) defined ethno-history as a methodological framework that combines anthropology with history through historical text, oral traditions, and ethnography in order to broaden interpretive horizons for past social processes. Henceforth an ethno-historical study was undertaken at post-Rozvi communities namely; the Gumunyu houses in order to aid the interpretation of material evidence from Danamombe. 72 This data was collected through oral field techniques that included using pre-structured questions (see appendix D), employing a research assistant, identifying and selecting the interviewee(s) according to age and knowledge. Archival sources (historical, ethnographic, linguistic and anthropological) were examined in order to select post-Rozvi houses with direct links to archaeological sites in south-western Zimbabwe. It emerged from the documentary survey that the Rozvi of Bikita had been fairly documented over the years as the direct descendents of core Rozvi settlements in south-central and western Zimbabwe, hence they provided the best research potential for this study. Only elders of seventy years and above (both male and female), currently living in the rural areas were targeted because many still rely on knowledge passed on to them by their elders. The researcher and his assistant also approached local chiefs in Bikita, namely Marozva and Mukanganwi for permission to conduct research in their communities. The other reason for approaching the chiefs was to appraise them about this research so that they could advise and direct us to the relevant Rozvi elders. All names of senior Rozvi descendants as well as their homes were accurately provided to us and that worked well to our benefit. We spent about two days introducing ourselves, explaining the purpose of our visit to the Rozvi members and making appointments to formally meet our targeted Rozvi informants at their homes. Data gathering started on the third day, as we visited different homes and interviewed our informants separately. An Olympus audio recorder and digital camera were used to capture the data, so the data was accurately recorded and all senior Rozvi elders in Bikita numbering up to eleven individuals participated in the study. Arrangements were also made for us to visit their sacred places and to take photographs of their customs and associated material objects. The study was completed on the fifth day and all the files and photographs were downloaded on a computer before final notes on the research were taken down. Ethno-historical information was particularly useful in defining Rozvi conceptions of power, dynastic lineages, and defining the social significance of material and non-material cultural features. Oral accounts also enhanced information on the cultural symbols and possible material markers for Rozvi identities. These accounts were treated with extreme caution as they were analysed against each other and then related to documentary evidence. The only setback encountered was that some Rozvi cultural material could not be photographed, for instance, the regalia for a Rozvi representative to Mabweadziva (major sacred rain shrine in the Matobo Hills) was said to be in Gokwe, while 73 other objects associated with Rozvi customs were either broken or hidden at the homes of other Rozvi descendents. 4.5 Archaeological approaches This study targeted Danamombe because it was a Rozvi capital, Khami was never re-inhabited after the 1644 civil war hence it largely represents Torwa material culture. Therefore a detailed comparative analysis of material culture from both Khami and Danamombe was systematically conducted in an attempt to establish exclusive Rozvi material symbols. The underlying methodological principle for this study being that differences in material culture patterning between Khami and Danamombe possibly signified Rozvi ethnicity. Emberling (1997) highlights that any aspect of material culture can be used to mark ethnic boundaries. Following this premise, archaeological evidence in the form of ceramics, beads, dry stone-walls, mud huts/dhaka structures, and other miscellaneous objects from both sites was systematically analysed and described. 4.5.1 Ceramic analysis This study had to rely on museum collections previously excavated at the sites of Khami and Danamombe. Keith Robinson‟s intensive excavations at Khami during the 1950s yielded a lot of ceramics that were selected for purposes of this study. A total of 504 ceramic sherds mainly diagnostic ones from the Hill complex were intuitively selected from the dozens of boxes at the ZMHS in Harare. The major limitation with the ceramic sample from Khami was that stratigraphic context could not be established from the existing museum files. As such ceramics from Khami were largely analysed for their distribution spatially without referring to stratigraphic context. Ceramics from Danamombe were sampled from the 2005 National Museums and Monuments of Zimbabwe (NMMZ) excavations funded under the theme; Field School for Archaeology and Related Human Sciences. Most of the Field School collections are at the Zimbabwe Military Museum in Gweru and their stratigraphic context is fairly well documented. The “Kitchen midden” trench yielded the majority of material evidence out of all trenches excavated during that fieldwork, hence it was automatically chosen for study and 3 299 sherds were analysed. 74 The methodology of ceramic analysis for this study is premised on the notion that laws/operative-value systems govern both individual and small social group behaviour (Gifford 1960; Huffman 2007). As such, potters act in response to these value-sanctioned systems and not their abilities or imaginations for fear of economic and psychological stress (Gifford 1960). These constraining values can be recognised in objects like ceramic types and varieties if only the proper classification is adopted (Gifford 1960). Huffman (2007:104) echoes similar sentiments as he argues that artists are accountable to each other and the community as they make pots. As they try to please the community they repeat codes, designs and colours in various ways that reflect style, which is easily recognised by outsiders as an identity. Identities derived from such stylistic forms are not connected through blood ties but interactions in social spheres like marriage patterns (Huffman 2007). In order to define ceramic styles at the research sites, a multi-dimensional approach as suggested by Huffman (1980, 1989, and 2007) was employed. This system is premised on the selection of ceramic dimensions that are relevant for a particular research purpose. Huffman (2007) further argues that when analysing ceramic style it is crucial to relate dimensions like profile, design layout/organisation and decoration motif to get an idea about ceramic types. However due to the high rate of sherd fragmentation it was difficult to determine full motif layout. In order to address this challenge vessel form, technique and motif combinations were used to get an insight of the types in both assemblages. Rozvi ethnicity may be inferred from any of the ceramic attributes that were considered in the analysis as long as the attribute could be used to mark difference from the so-called “others”. Highly visible dimensions of ceramics such as vessel form and decoration are of major importance in this case; hence the rare elaborately decorated ceramic types should be crucial in identifying the Rozvi. Since the multi-dimensional approach depends on the recognition and use of all attributes of a ceramic collection, a data capture sheet was designed with dimensions of vessel form, lip form, fabric, exterior colour, decoration technique, decoration placement and motif (see appendix A). 184.108.40.206 Vessel form: For comparative purposes, the primary unit of analysis was the vessel, thus potsherds both diagnostic (rim, neck, shoulder or a combination of these) and undiagnostic (plain body or base fragments) were initially sorted into profile classes. These profile classes were 75 further combined into vessel shape categories as defined by Pikirayi (1993), Pwiti (1996a), Shepard (1956), and Sinclair (1987). Seven vessel shape categories were given as follows: Category A: R; Bo/Ba; Bo; Ba (any vessel shape) Category B: R/N; N Category C: Sh/Bo/Ba; Sh/Bo; Sh Category D: R/N/Sh/Bo/Ba; R/N/Sh/Bo; R/N/Sh;N/Sh/Bo/Ba;N/Sh/Bo; N/Sh (independent restricted vessels) Category E: R/N/Bo/Ba; R/N/Bo; N/Bo/Ba; N/Bo (dependent restricted vessels) Category F: R/Sh/Bo/Ba; R/Sh/Bo; R/Sh (restricted vessels with simple contour-constricted bowls) Category G: R/Bo/Ba; R/Bo (unrestricted vessels with simple contour-open bowls) Key: R =Rim, N =Neck, Sh =Shoulder, Bo =Body and Ba =Base 220.127.116.11 Lipform: This was recorded as the extreme top edge that defines the vessel mouth. Recording these followed Phillipson (1976) who observed that there are rounded, tapered, squared, externally thickened, internally thickened, bilaterally thickened, bevelled, and fluted lipforms (see Fig. 4.1) below. 76 Figure 4.1. Vessel lipform profiles (After Phillipson 1976). 18.104.22.168 Fabric: This was defined as the clay to temper ratio in the ceramic paste. The material used as temper and its grain size generally influences variability in ceramic texture (Shepard 1956). Fabric was recorded subjectively as course, medium or fine after a careful visual assessment of a potsherd in cross-section. 22.214.171.124 Exterior colour: Vessel colour is influenced by several factors such as composition of the clay, firing (the atmosphere, temperature, duration), absorption of stains during use, deposition of soot when placed on fire, deposition of substances when discarded or even 77 accidental reheating in the event of fire outbreaks (Shepard 1956:103). Due to these various factors only exterior vessel colour was recorded because it was the most important in view of my research aims. In instances where colour varied on the same potsherd, the dominant colour was recorded. The Munsell Soil Colour Charts could not be accessed during the analysis hence visual judgement was applied in defining colours. Colours used in the study for plain and decorated vessels were black, grey, greyish-brown, brown, and reddish-brown, soot was recorded as a surface coating because the primary or secondary colour was underneath. Undiagnostic fragments were simply classified according to colour after which they were simply counted. 126.96.36.199 Decoration technique: This refers to the method of applying design(s) /motif(s) on different parts of a vessel. A wide range of objects, paints and substances were used in the processes of decorating pots, hence decoration techniques were defined on the basis of the tool or method of applying the motif. 188.8.131.52 Decoration placement: The actual location of decorations on the vessel was recorded as decoration placement. The analytical categories used in recoding the data included inside the lip, on the rim, inside lip to the rim, on the neck, shoulder, body or base. 184.108.40.206 Decoration motif: According to Huffman (2007) this is the entire design organisation on a vessel that is often categorised according to the dominant decoration technique and its various combinations. These motifs were recorded through illustrations on the data capture sheet and comments were made for each motif. 220.127.116.11 Sherd thickness: This is the widest part of a sherd that was measured in cross section. Measurements were taken on the maximum points of diagnostic sherds only and this was done using a digital callipers. This was done specifically to check if there were any significant differences between the two assemblages. 4.5.2 Analysis of Beads In the past, beads were strategic resources often possessed by a few individuals who strategically used them in efforts to influence social and power relations (Huffman 2007; Pwiti 1996a, 1996b; Wood 2005). It is against this background that beads from both Khami and Danamombe were analysed to denote Rozvi ethnicity. Once again Keith Robinson‟s 1950s museum collections 78 from Khami ruins were targeted for study. The ZMHS in Harare has several such boxes, hence a combination of stratified and random sampling procedures were employed to select beads for analysis. Initially, all boxes with beads from Khami Hill complex were selected and opened. This was followed by matching catalogue numbers for beads in museum boxes with those in Keith Robinson‟s 1959 publication. Finally only 1 462 beads that matched Robinson‟s database of finds from Khami ruins were selected for study. Beads from Danamombe were intuitively sampled from the 2005 NMMZ “Field school” collections at the Military Museum in Gweru. The Kitchen midden trench yielded 866 beads which were fairly adequate for purposes of this study. Glass, metal and shell beads were all analysed in terms of context, material, manufacturing technique, shape, structure, colour, diaphaneity, decoration, and metrical attributes relating to size (see appendix B). 18.104.22.168 Classification: Beads were initially classified according to material, which included glass, metal and shell. The procedures for describing and classifying beads as laid out by Bvocho (2001), Kidd and Kidd (1970), Kinahan (2000), Mupira (1991), Tapela (1998), and Thondhlana (2005) were reviewed and selectively employed in defining bead class varieties. The method of manufacture was the primary criteria of classifying glass beads while; appearance (shape, structure, colour, diaphaneity, decoration and size) was secondary. All shell beads had the same method of manufacture and shape; hence they were treated as constituting a single class with minor variations based on size. These shell beads were then analysed quantitatively so as to infer on their historical and social variability. Metal beads were primarily classified according to material and techniques of manufacture, (shape and size) were used to define variations. 22.214.171.124 Material: Visual inspection methods were used to identify the materials used to make beads. Glass and shell beads are fairly easy to identify, however problems arose when I dealt with metal beads without composition equipment. Thondhlana (2005) proposed a valid system for describing metal beads visually and this criterion was employed for this study. Thus copperbased metal beads with a yellowish colour were identified as brass, metal beads with a reddish colour were recorded as bronze, copper shows a brownish or greenish colour if corroded, and beads made of tin are whitish in colour. Shell beads were manufactured from materials such as achatina and ostrich eggshell but these were classified as shell for purposes of simplicity. 79 126.96.36.199 Techniques of manufacture: Glass beads were described following definitions outlined by Kinahan (2000) who identified three distinct methods: drawing, winding and moulding. Hand-drawn tubes of molten glass produced drawn beads, which were then cut into various lengths while some of the sharp edges were rounded by reheating. These are identified by elongated bubbles and longitudinal fibres in the glass running parallel to the perforation. A magnifying glass was used under bright conditions to check for these characteristics but these were difficult to define with certainty. Wound beads were made by winding molten glass around a thin wire. Identifying characteristics included bubbles and fibres arranged in spiral fashion around the perforation. Then mould-pressed beads were made by squeezing molten glass between two halves of the mould. These are regular in terms of shape, colour and shiny appearances. Some have a distinct ridge or seam that is formed when glass is squeezed in the mould. Techniques for metal beads manufacture were defined following Thondlana (2005). Bending referred to (folding done haphazardly to produce an angled form), folding (curving blanks of metal into smooth rounded profiles), casting (using moulds to shape molten metal), and chopping (using sharp object to cut tubes into small pieces). Shell beads were made by grinding the edges of shell until they were circular and smooth after which the centre was perforated by a sharp object from both sides. Therefore these shell beads were simply recorded as ground. 188.8.131.52 Shape: This dimension relates to forms derived from the longitudinal and transverse sections of beads (see Fig. 4.2). The following basic shapes were recorded: cylinders (profile is parallel to perforation thereby giving a tubular form), barrels (rounded/convex profile ends tend to be flattened), oblates (width greatest at the centre tapering equally towards each end), spheroids (length equal to width, which gives a rounded profile), disks (thin to flat in length with small diameters), and bicones (two symmetrical cones with a common base). These shape definitions were derived from Kinahan (2000), Mupira (1991), Spector (1976), Tapela (1998), and Thondhlana (2005). 80 Figure 4.2 Bead shape profiles (Modified from Spector 1976). 184.108.40.206 Structure: This was only valid for glass beads as it refers to layers that can be noticed visually. Thus beads were described as either single/simple layered or compound/multi-layered. 220.127.116.11 Colour: The descriptions for metal beads have already been dealt with under classification. Colours for glass beads were initially described following Munsell colour order in Kinahan (2000) Appendix I. Some of the names were too long and confusing, hence for purposes of simplicity these codes were dropped in favour of generic colours like blue, red, yellow, green, white, black, pink, and white which has lost its lustre. Beads with two or more colours were simply recorded as mixed in colour. 18.104.22.168 Diaphaneity: Relates to visual effects of natural light on glass beads. Under such conditions beads can be opaque, translucent or transparent. 81 22.214.171.124 Decoration: This was only common for glass beads where designs of various colour sheds were added to the beads as part of enhancing surface finish. 126.96.36.199 Size: Bead diameter is crucial in determining size. Although measurements of bead length were recorded, these were not critical in defining size variability. Metrical dimensions for size ranges were as follows: small, 2-4mm; medium, 4-6mm; large, 6-10mm; and very large, 10mm and above. 188.8.131.52 Bead analysis: The data was classified and tabulated using Ms Excel in order to denote statistical patterns. 4.5.3 Architectural analysis of dry-stone walls Groups of people use material culture to express their identities and southern Africa is no exception to this. Huffman (2007) and Pikirayi (2007) argue that houses, mural art and a wide variety of objects often constitute an integrated design field with symbolic meaning to the group and these repeated codes are recognised by outsiders as part of group identity. It therefore follows that ethnicity is sometimes communicated through architectural style (Emerson 1997). Following this premise dry-stone walls at both Khami and Danamombe were subjected to analysis. Khami has several platforms, which could not all be examined due to time and resource constraints. As such purposive sampling strategies were employed to select only the Hill complex, Cross ruin and Passage platform for detailed study. It was assumed that these platforms constituted the core settlements of the elites in the past; hence they provided the best opportunities for research. The same sampling procedures undertaken at Khami were repeated at Danamombe in order to select the Main platform, Vanguard/Northern platform, and the Outer/Lower enclosure from the rest of the dry-stone structures at the site. A digital camera, tape measure, and data capture sheets (see appendix C) were used in the recording process. The data capture sheet was designed with the following variables: wall number, construction material, architectural techniques, entrance, and decoration. In total 193 walls were recorded at Khami, while 82 walls were recorded at Danamombe. 184.108.40.206 Wall number: Site plans with wall numbers for both sites could not be found hence a numbering system was devised during fieldwork. Recording and numbering only focused on the exterior side of walls because decoration was commonly done on the outside, and these were the 82 only accessible sections especially for retaining walls. Numbering started from walls either at the entrance or close to it, and the first wall was recorded as 1. If the same wall showed slight changes along its length it was recorded as 1b, but if this wall was joined or abutted by another wall then it was labelled number 2. Thus numbering of walls increased cumulatively, and each section of the wall that carried a different number (e.g. 1, 1a, 1b, 1c, 2, 2a, 3) was counted as a separate entity. This numbering system was employed for all platforms and enclosures that were analysed at both sites. 220.127.116.11 Material: This referred to the type of rocks used in constructing the walls. The most commonly used rock type was granite, associated rock varieties like laminar schists, ironstone, basalt and sandstone. These non-granite rocks were commonly used to decorate the walls but at times they simply constituted the building material for certain walls. These non-granite rocks were therefore recorded as (other) because they were so variable in terms of their distribution. 18.104.22.168 Form: This variable addressed the structural appearance of the walls as some were either retaining (Rt.) or free-standing (Fs.); wall heights were also measured as part of this dimension. Retaining walls were constructed adjacent to landforms such as hills. Quite often the hill was ultimately submerged by a series of artificial platforms that sometimes accommodated mud huts on top. On the other hand free-standing walls are those walls that form enclosures because they are built as erect independent structures. Wall height was taken from the highest point of each wall; the zero reading of the tape measure was placed where the wall emerges from the ground. In order to deal with problems of secondary infill caused by erosion processes, a thorough visual inspection of the soil texture, grass cover as well as the general gradient was done before taking any readings. Thus readings were taken from the ground upwards and measurements were rounded to the nearest 0.01m. The following wall height ranges were then used to analyse the data: very short= 0-0.50m, short= 0.50-1m, medium= 1-1.50m, high= 1.50-2m, and very high= +2m. The major limitation encountered when measuring wall height was that some sections had been restored while others had collapsed badly. These alterations negatively influenced the quality of data collected. 22.214.171.124 Techniques: These described various construction skills such as producing dressed blocks (d/bl.), neat coursing (n/co.) and battered back (bb) walls. Dressed blocks were those rocks shaped into regular forms for purposes of constructing neat wall faces. Neat coursing 83 defined the skill of careful selection and laying-out regular rock blocks to form a series of horizontal wall courses. The skill of reinforcing walls by slanting the top sections backwards, such that when viewed in cross-section the wall angle would be more than 90˚ was described as battered back. 126.96.36.199 Entrance: This was the access point of either an enclosure or platform usually defined by two separate walls with either rounded or squared profiles. A single platform or enclosure could have a series of such entrances. However most of these entrances were difficult to define especially for those sections seriously affected by collapses. 188.8.131.52 Decoration: The additional designs that formed different patterns on the exterior walls were also recorded. Those captured on the recording sheet included: linear friezes of dentelle (Lf/d), chevron, cord, check, herringbone (Hr), isolated courses of non-granite stone forming bands of a different colour, and small rectangular panels with a herringbone infill (Srp/h.inf.). Each wall was examined for each of these design patterns. 184.108.40.206 Classification and analysis: The primary dimensions used in classifying walls were form, construction technique, and decoration motif. The other attributes on the architectural data sheet were subordinate to the above two variables. Earlier contributions towards the classification of dry-stone walls by Garlake (1970), Summers et al. (1961), and Whitty (1961) were reviewed and integrated for purposes of this study. 4.5.4 Dhaka structures House remains made of mud/dhaka at Khami and Danamombe were just described. Broader comparisons of dhaka structures from the two research sites were done through reviewing literature such as Caton-Thompson (1931), Chakanyuka (2001), Dhliwayo (1989), Hughes (1997), and Robinson (1959) who also wrote about dhaka structures in south-western Zimbabwe. 4.5.5 Miscellaneous finds These constituted the various archaeological finds such as ornaments, imported porcelain and other rare finds like gold beads that are said to have been recovered and looted from Danamombe by early treasure hunters. These unique finds were simply described and assessed for their 84 economic, social and political significance because rare precious goods are usually manipulated by the elite and ethnic groups to signify power and status. 4.6 Conclusion This chapter has set out various parameters that were used to define the archaeological identity of the Rozvi. In this regard a multi-faceted methodological approach focusing on extracting documentary information, oral traditions and ethnographic data was carefully crafted. This followed the notion that ethnicity is both a social and political construction that exists within discourse and ideologies that are constantly produced and re-produced in specific historical and institutional contexts. Therefore when examining ethnicity we need to first pay attention to these ideational, social and subjective dimensions before attempting to tease such identities through material culture. More so the analysis of multiple sources of archaeological evidence such as ceramics, beads, dry-stone walls, dhaka structures and other miscellaneous finds from the two archaeological research sites was specifically done to deal with the fleeting nature of ethnicity. This agency oriented methodology creates opportunities to explore the way individuals manipulate material culture and how they are constrained by it in efforts to attain goals in varying social contexts. The next chapter presents data extracted from documentary sources, ethnographic records, creative writing and post-Rozvi descendents. 85 Chapter Five Documentary and ethnographic data on the Rozvi and Torwa “Naturally, it is very difficult if not impossible, to deduce any kind of political process from archaeological evidence alone, so we have to rely upon historical evidence” (Beach 1998:55). 5.1 Introduction The objective of this chapter is to establish a foundation upon which social and political identities of the Rozvi can be traced through material culture. Although the major focus is the presentation of data about the Rozvi, views about the Torwa are included albeit only in passing since very little is known about them before and after they were assimilated by the Rozvi during the 1690s. We are fortunate that a fairly sufficient historical record has been generated by the Portuguese, indigenous communities through their oral accounts, historians, ethnographers, and creative writers. In addition, ethno-historical data was gathered from the Gumunyu (post-Rozvi houses or direct living descendents of the historical Rozvi) located in south-eastern Bikita district of Zimbabwe. A review of these various sources of data is aimed at addressing Rozvi origins, and their possible symbols of identification. The data presented in this chapter shows that the Rozvi originated in north-eastern Zimbabwe and their rise to power was closely associated with the activities of a great military leader. Rozvi identities were defined around the Mwari religion, their military prowess, political power, totemism, mythological versions, material culture and natural features like mountains and trees, dialect, performance of traditional roles like rainmaking customs, appointment of Chiefs, and awarding of land rights to newcomers. 5.2 Data from Portuguese documents The earliest literature on pre-colonial communities on the Zimbabwean plateau may be traced back to Portuguese writings that gradually accumulated from the beginning of the 15th century following their arrival on the East African coast. Around 1530-31, they had settled at Sena and Tete replacing the Swahili as the dominant trading partners with African communities. It was through these myriad interactions between the Portuguese traders and Africans that documents with links to the Torwa and Rozvi were generated. While Portuguese documents have greatly aided research, Wills (1985:19) argues that Portuguese sources are often misleading because of inherent biases and the writers‟ misunderstanding of the local situation. Pikirayi (1993:108) adds 86 that some of these documents are vague, repetitive and confusing. Beach (1980:197) further argues that these sources were not generated to be informative about the geography or history of the local people because the interests of the writers were purely wealth oriented. Nevertheless a legacy of Portuguese notes, official directives, narratives, a ҫ ova, who wrote a letter to the King of Portugal on the 20th of November 1506 stating: Mocomba, father of Quesarymgo, had a favourite who was a great lord in his kingdom and who ruled the whole kingdom exiling, beheading and acting in all things as king, and was called Changamire and was the king‟s chief justice and the word for justice is amyr or as we [the Portuguese] say it „governor‟. And this amyr owned many towns and villages in the kingdom given him by the king. And within his lands the amyr became great through the power he held in the kingdom and in this way acquired many people. As he was eager to remove the threat of this over-mighty governor, the king sent him a cup of poison with orders to drink it. Changamire refused, assassinated Mocomba and ruled for four years as usurper. He was in turn killed by Quesarymgo, the only survivor of Mocomba‟s twenty-two sons whom Changamire had massacred. But Quesarymgo was not able to command the allegiance of Changamire‟s vassals, who grouped under the leadership of one of Changamire‟s relations, Toloa; and so there was war between the two factions (Randles 1979:6). The Changamire governor cited in the account above was possibly related to the „Toloa‟ (Torwa) because his followers eventually submitted to Torwa authority after he was assassinated. This first Changamire is dated between 1480 and 1494 Portugal (da Silva Rigo and Baxter 1540-1560 (Vol. VII): 173), Changamire was described as a lord and king who ruled over an ivory territory and always won in war but somehow got defeated. This Changamire was not the Rozvi founder as there is no mention of that name or identity before the mid-18th century. Mudenge (1974b:27) comments: In short there is no evidence from Portuguese sources that the Rozvi as a nation bearing the name „Rozvi‟ existed before the advent of Dombo. Previous writers have at times unfortunately used eighteenth-century Portuguese documents clearly referring to Dombo (1684-95) as if they refer to the Changamire of c.1480-c.1492. It is partly this 87 error which has given the impression that there are Portuguese documents which show the existence of the designation Rozvi prior to Dombo‟s advent. At present, however, we simply do not know of the existence of such sources. ҫ ova is that the Torwa existed somewhere on the Zimbabwean plateau, hence total silence about the Rozvi proves that they were not yet a political entity. Although the Portuguese did not directly interact with Butua, they certainly had a reasonable appreciation of the geography and its people. In that respect, Randles (1979:7) cites Gaspar Veloso‟s letter of 1512 addressed to the King of Portugal stating that: ... between the land of the Monomotapa and Sofala all kings bore allegiance to the Monomotapa, but further inland (still on the plateau, of course) there was another king who had risen in revolt and with whom he was at war: the king of Butua. The latter was as powerful as the Monomotapa and his country held great reserves of gold. Thus Gaspar Veloso implies that while the Mutapa state dominated much of the plateau, the Torwa controlled a reasonable share of that territory further inland probably in the south-west. Father Francisco de Monclaro (da Silva Rigo and Baxter 1497-1840 (Vol. VIII): 283-529) also noted that the kingdom of Butua was extremely rich in gold, silver, ivory, and cattle. This meant that rather than being a mere vassal of the Mutapa, it was an aut They have other mines in a region named Toroa, also called the Kingdom of Butua, ruled by a prince [chief] named Burrom, a vassal of Benomotapa (Randles 1979:8). This source implies Torwa was synonymous with Butua, but an anonymous document of 1683 points out: ...the kings of Butua bear the title of Toroa (Randles 1979:8). This document informs us that „Toroa‟ meaning Torwa was a title for kings who governed in the lands of Butua. However there is a big difference between 1552 and 1683, which implies that political circumstances might have changed resulting in what was simply a name for a region becoming a dynastic title. Hence origins of the title „Torwa‟ has, and will remain a subject of speculation among scholars as no one has yet come across a Portuguese source that explores its background to some reasonable depth. Scholars such as Randles postulate that: 88 The word „Toroa‟ refers to someone who is incorporated into a group without having any family ties therein; he may be a „slave‟ or perhaps a „vassal‟ – in this case, a vassal of the Monomotapa (Randles 1979:8). ҫ ova‟s account and current knowledge on the Torwa subject, it is difficult to go by Randles‟s assumption cited above, especially considering that the Torwa were already in existence when the revolt by Changamire took place in the 1490s. We are told that after Changamire was killed by Quesarymgo, his followers submitted under the leadership of his Torwa relatives. Therefore, Torwa as a dynastic title or group identity was already in use when the first Changamire rebelled from the Mutapa. One may only suggest that the term Torwa acquired new meanings after rebellions and migrations occurred in the Mutapa state during the 15th-16th centuries. Possibly such historical developments resulted in the dynastic title being applied to both leaders and commoners in the lands of Butua, especially by outsiders to that kingdom. As time passed, Portuguese activities and interference in African affairs became more rampant, especially during the 17th century. The Portuguese capitalised on the death of a missionary in the Mutapa state especially between 1600 and 1633, hence by the 1640s they had more influence in African internal affairs. Beach (1995:27) notes that some locals in the Mutapa state had to flee to the mountains in order to avoid being forced to mine gold by the Portuguese. Following a succession dispute in the Torwa state, one of the defeated brothers immediately asked Sismundo Dias Bayao, who was linked with the feira of Manyika to bring his army to the south-west. The Portuguese responded to the call and the civil war of 1644 ensued thereby scattering the Torwa communities from their capital, Khami. Thus Beach (1994a:96) rightly argues that the 1644 civil war at Khami should not be entirely attributed to the Portuguese. Accordingly, Portuguese interference coupled with rampant succession disputes worsened the local situation such that by the 1670s the political atmosphere was extremely volatile on the Zimbabwean plateau. It is around that time that a certain Changamire mobilised a strong army and rose against the Mutapa, Portuguese traders, and other locals whom he considered as threats to his political power. Dionizio de Mello e Castro and Antonio Pinto de Miranda explained: While in charge of King Mucombue‟s herds, Changamire was instructed to drive them into the land of Orobze [Mello e Castro; Orobu in Pinto de Miranda]; there he rose in rebellion against the king having gathered together a great army; in consequence he is today the most powerful, the most dreaded and the most respected of all the chiefs 89 who refuse to obey the emperors of Monomotapa, whom he has on several occasions defeated and routed in pitched battle. He is the absolute ruler over Abutua... (Randles 1979:16). Axelson (1960:179) identifies this second Changamire to lands adjoining Butua where he built a very powerful army that attacked Mutapa, Butua, and the Portuguese at Maungo/Maungwe and other forts between the 1680s and 90s. Through these victories, Changamire gained so much respect from the Portuguese, Mutapa leaders and several other chiefs who automatically chose to adhere to his command and power (Axelson 1960). In 1685 Caetano de Mello de Castro reported Changamire Dombo had many kings and chiefs under his command that caused problems for the Portuguese (Mudenge 1988:287). In 1758 Ignacio Caetano Xavier reported that Changamire‟s lands were populated by the Borobzes (Randles 1979:16). Interestingly, the name Rozvi only started appearing frequently in Portuguese sources around the mid-18th century suggesting it was never in use before then. This proves that it was only constructed after some major sociopolitical events that had occurred around the late 17th-18th centuries and since then it became synonymous with power. Randles (1979:16-17) cites an anonymous manuscript, which he dates to the end of the 18th century stating that: ... Urobze lies a long way from Manica, it would take a month to reach it, it is said to be nearer the Cape of Correntes. The Africans call it Goromucuro; it lies to the west of Manica. The Kingdom abounds in the rolling veld plains [campinas vistozas planices]. There are huge herds... There are many birds called „emas‟ [ostriches]. There are also very few trees, if at all, and firewood is replaced by cow-dung dried in the sun... Although the manuscript cited above mixes the name of the followers of Changamire with their later core-territories, the description given in the account befits Butua, which was under the Torwa dynasties before they were assimilated by Changamire and his followers around 1696. Several Portuguese documents covering the 18th century corroborated the image of the Rozvi as fierce, powerful and warlike thus proving they were a force to reckon with at the time. Axelson (1960:182) affirms the power of Changamire‟s army by citing their 1693 and 1695 attacks on Portuguese forts of Dambarare, Manica, and Macequece (Masekesa) as extremely vicious (see Bhila 1982a). It therefore follows that the name Rozvi was constructed following Changamire Dombo‟s military campaigns and this is confirmed by the appearance of the name in Portuguese sources that date from the mid-18th century. 90 5.3 Oral traditions Oral traditions vary widely from myths to relatively factual accounts devoted to ruling dynasties, hence some of these traditions are examined in order to trace Rozvi ethnicity. Beach (1983b:16) highlighted Aaron Marwodzi‟s account recorded sometime before 1924, which envisaged that: ... Rozvi, Korekore and Sotho... starting at Lake Tanganyika and crossing the Zambezi. They all had the same zhou totem, and as they moved into Zimbabwe different groups were allowed by the Rozvi to split off and settle: Korekore, Nhowe, Zvimba... The Rozvi eventually reached the Sotho country (the Sotho having presumably overtaken them) and drove them out, and reached the land of Chihunduro soko, took his “Zimbabwe” country and settled there, separating themselves from the Kalanga, the Lilima, and the Humbe... Marwodzi‟s account conflicts with other traditions because it identified an external southern origin for the Rozvi, cites a unique totem for Rozvi, mentions Chirisamhuru as the first Mambo (Chief), located the core Rozvi area to central Zimbabwe rather than the south-west, and his account portrays Rozvi as a homogenous and static ethnic group just like the Sotho, which is problematic. Beach (1976) cites another oral account documented by Abraham: ...NeMbire‟s daughter Negupo is said to have married Mutota Churuchamutapa, “not to be confused with the later Mutota Nyatsimba, who founded the Mutapa dynasty.” Their daughter Senwa or Nehanda had in due course a soncalled ChikuraWadyembeu-by an unknown father. ChikuraWadyembeu was installed as the first “Rozvi” ruler at Zimbabwe. Mutota Nyatsimba who founded the Mutapa state in the north succeeded Chikura at Zimbabwe early in the fifteenth century, was probably his son (Beach 1976:2). This tradition basically stated that the earliest „nuclear Shona‟ group dates back to the 14th century and it was led by Nembire who came from Lake Tanganyika to settle north of the Zambezian plateau but controlled lands as far as the Limpopo (Abraham 1959; Beach 1976). In this case Nyatsimba Mutota is portrayed as a son of ChikuraWadyembeu indicating some confusion because these individuals lived at different times in history. Beach (1976) adds that the association of Mutota, the Rozvi and Zimbabwe ruins indicates that this tradition dates back to the 17th century and not 14-15th centuries. Factual errors in the above traditions could have been deliberately manufactured by Rozvi elites in attempts to legitimise their power. 91 5.4 Traditions of Rozvi origins and power According to Beach (1976:3) post-Rozvi groups found south of Rusape share traditions that ChikuraWadyembeu otherwise known as Changamire or Dombolakonachingwango (often shortened to Dombo) was the founder of the Rozvi. These traditions mention that this leader rose from somewhere around Murewa area before moving southwest to conquer Chiwundura and Tumbare (pre-Rozvi occupants of Khami-type sites). Beach (n.d.:10) also notes that there are other traditions, which trace this first Changamire to a location on the trade route to Sofala. Tumbare, Chihunduro, Nechasike or Nechadzike and other former rulers of the Torwa state were integrated in the Rozvi ruler-lists (Beach 1976:3). This tradition is plausible because it resonates with other historical sources like some Portuguese documents already explored above. Hodza and Fortune (1979:217) also collected traditions from several Rozvi chiefs in Zimbabwe and noted the following: ... When the Rozvi reached this country they found Munhumutapa ruling it. They attacked him and drove him away. This was the first fight in which they had ever engaged. After Munhumutapa‟s flight, they subdued all the chiefs who had been subject to him. They are said first to have gone north, to Pfura, then east, to Nyanga, then south to Manyika and Bikita, and finally west to Mberengwa (Belingwe). Tradition explains that the name Rozvi was given to them because they disturbed the lives of everyone, man and beast alike. All this is said to have happened under the rule of the first chief whom they remember, called Changamire. ... The warlike nature of the Rozvi is vividly depicted as marking their initial rise to power. Most importantly traditions highlight or emphasise the connection between particular historic events and the name Rozvi. This data agrees with Portuguese sources discussed above and it shows that the term Rozvi as an identity was historically constructed. 5.5 Rozvi traditions of the Mwari religion Oral traditions also consistently identified Rozvi origins and power with the worship of Mwari, which they did through people, trees, objects, among other things. After reviewing several oral traditions about the Rozvi, Hodza and Fortune (1979:217) pointed out: The Rozvi, like many other groups, say that they came from a land called Guruuswa and from a place in that land called Manyanga... Their traditions tell us that they were accompanied on their travels by a voice which they called Tovela which led them on their way, keeping them safe from dangerous places. The voice could speak from any object it chose and we are told that it spoke through grass or trees or through a little child, unable as yet to speak for 92 itself. The voice of a presence whom some say was that of the first Rozvi ever created, the founding father of the clan. Others say he was the first person ever to be created and that after his death his spirit accompanied the different groups from Guruuswa to the countries they chose. Tovela not only protected his people on their travels but gave them food as well. When they clapped in homage and supplication to him at a tree, portions of porridge, pots of milk, and combs of honey would come out of the ground. He also gave them medicines to become invincible but not inaudible to their foes and to be long-lived. Tovela came to be called by other names by different groups such as Mwari... The quotation above highlights how religious discourse was used as both an economic and political tool by the Rozvi. The interesting aspects that emerge from the extract are that the Rozvi were Mwari‟s favourite children, whom he provided with everything in mysterious ways and protected from all enemies using supernatural power. Hence the Rozvi identity was synonymous with Mwari, power, wealth and authority over “others”. This association of Rozvi origins, power and governance with Mwari was very common in oral accounts. Fortune (1956:69) also made the following important observation: One of the remarkable features of the Rozi story lies in the personal nature of their relationship with Mwari. How insistent he was on their acknowledging that they lived, ruled and conquered only by his favour and help! Even the magic horns lost their power when this was withdrawn. Any hubris or trust in their own unaided strength led to the immediate withdrawal of his favour and consequent disaster... Even Posselt recorded a fragment of a Rozvi praise poem that was recited to him in honour of Mwari. The poem was revised and translated by Fortune (1956:83) and it reads as follows: Mwari nkulu Mbibanyi weshongwe. Mudzana parubge. Huruva ikatunga Vezha novuTebele. Nkukutu unga rutombo, Dziva levula latekayila. Ikavi mibvumbi. Great Mwari. Piler up of the towering rock-mountain. One who dances on rock. Dust rises. Vendaland and Ndebeleland. Hard one, you are like a rock. Pool of water that has moved from side to side. And has become mist. The historical origins of the concept of Mwari (the sky god) remains a subject of debate among scholars (see Beach n.d.; Daneel 1970; Ranger 1974). Daneel (1970:16) however noted that Mwari was primarily concerned with fertility of women and the land, but his interests in tribal politics only started when the Rozvi rose to prominence. Tales collected by Bullock (1927:40) state that Rozvi emissaries claimed they knew Mwari; they were sent by him and were his children. There were also claims that the Rozvi monarchy was called into being by Mwari 93 (Daneel 1970:22). Daneel (1970:40) further noted that those who claimed to be children of Mwari often carried symbolic objects such as a gourd cup and wore the insignia of black beads. Bhila (1982a) also recorded traditions in Manyika that identified the Rozvi through the Mwari religion and portrayed that religion as superior to all others. The Rozvi worshipped Mwari by killing a beast under a tree and they would leave it there the whole night beside some beer. The following day people ate the meat and drank the beer, while asking the spirits to intercede for them to Mwari. In the prayer they asked for what they wanted, clapped their hands and sat down after which they averred that mudzimu (the ancestral spirits) had replied them (Bhila 1982a:109). These prayers took place at Matonjeni or in special huts below the mountains in Mutasa until the 19th century (Bhila 1982a). Mwari worshipping was also prominent in the Matobo hills but this was a much later development (Beach n.d). In the 18th century the Singo dynasty introduced Mwari to the Venda who had always worshipped in caves. Later some of these Venda communities migrated to the Matobo hills during the late 19th century and introduced a modified Mwari religion (Beach n.d.:32). Other traditions go on to mention that when the Rozvi had grown so much in power, they became proud and refused to make an offering of cattle to Mwari and this led to their demise (Beach n.d.; Gann 1965). Traditions recall the strong link between the Rozvi and the Mwari religion because Rozvi chiefs exploited and elaborated the cults for political purposes. Kuper (1954:31) highlighted that rainmaking enhanced the political influence of chiefs; as such the ritual position of Rozvi Mambos (Chiefs) therefore became more advanced than any other. Bourdillion (1976) also observed that chiefs‟ political power often arises from the religious power of their ancestors because even those members not related to the chief also speak of spirit guardians of the chiefdom as their ancestors who had the duty of providing rain and caring for the crops. Daneel (1970:23) noted that the Mwari religion played a centralizing role during the Rozvi confederacy (around the 17th-19th centuries) such that the religion became closely identified with the Rozvi. Therefore the Mwari religion was manipulated as an identity by the Rozvi elites because they wanted to elevate their status to the highest spiritual realms in order to reinforce their political superiority and power. This assumption stems from the fact that several traditions linking the Rozvi with Great Zimbabwe and other prominent ruins have been recorded by different scholars. In 1872, Carl Mauch collected traditions at Great Zimbabwe that also linked 94 the Rozvi with the site and Mwari. Basically, these traditions stressed mythical elements highlighting Rozvi power and pride (Beach n.d.:2). Bullock (1927) also recorded traditions from Nemakonde area emphasising the Rozvi had a strong system of governance, which enabled them to control other local groups and to build ancient structures scattered across the country. Bullock was told tales about Rozvi attempts to build scaffolding to reach and catch the moon for making a „ndoro‟ conus shell for their king. He was also told about Rozvi attempts to move Mutirikwi, Urungwe, Nyandoro and other mountains so as to build them again at Zimbabwe. These mythical traditions were strategically crafted by Rozvi elites to signify a unique past associated with great ambitions, power, wealth and supernatural goals. Fortune (1956:82) highlights these Rozvi old tales: ….while living at the court they tried to pile up poles saying: “We will take the moon to make a platter for the Mambo.” They were killing oxen for making thongs to bind the ladder-steps. They spent a year doing these things but afterwards the logs unexpectedly rotted and fell down, and all the people in the sky died…. …. Now it was that after they were wasted in that way, they said: “We go and take the big rock called Makono so that it may be sat on by Mambo.” And they dug it putting logs underneath to prise up that huge rock and so it fell and crushed them; and they all died. They left the place saying: “The rock worsted us; let us go to Ndarikure.” Having gone to dig again so that they might take it to the court so that it might go and be sat on by Mambo, it thereupon fell again and crushed them and all died. 5.6 Rozvi traditions of chieftainship and land-rights Oral traditions also strongly identified the Rozvi through their political roles of installing chiefs and giving land to new-comers. Bourdillion (1976:27) highlighted that the Rozvi dominance over surrounding dynasties was vividly visible in the installation ceremonies of modern Shona people who sought the ritual approval of a member of the Rozvi clan. Bullock (1927:286) stated that Mavhudzi, a hereditary priest of the Rozvi, had the responsibility of appointing chiefs but he had to be accompanied by a Rozvi of the moyo/moyondizvo (heart) totem. This genuine/ legitimate Rozvi would then pronounce: “You … are Chief. Look after your people. I give them to you.” After consulting „mutibvu‟ (custodian) of „chiremba‟ (the red cap) symbolizing a crown, the new chief would then be given regalia of the dead chief, iron rod or sceptre, assegai and the bow of his forefathers (Bullock 1927). Hodza and Fortune (1979:7) furnished us with a detailed Rozvi 95 installation poem, which they also translated but it has been shortened for purposes of this study as follows: Ndini MuRozvi ndimire pano. Uri pamberi pavanhu, ndinoti, Wave’she nhasi. I am the Rozvi who stand here. As you stand before the people, I call you Chief today. Ndakuturika chiremba chechiremekedzo pamberi pavanhu I place on you before the people the symbol of honour. Takuturura chidawo chako nhasi. I take off from you your own praise name today. Ndinopedzisa ndichiti, Heyi tsiva yevhu, ipfumbate maoko ako ose Finally I say, Here is a handful of soil, hold it in both hands. Izvi zvinoratidza nokuonesa kuti uchatonga nokudya pasi. That shows you will rule the land and eat its fruit. The important points that emerge from the above poem are that, Rozvi-appointed chiefs were given „chiremba‟ a symbol of power in the form of a white shell, „ndoro‟. The „ndoro‟ or symbol of power was often worn on the chest or forehead, while the „chiremba‟ could have been a feathered head-dress (Hodza and Fortune 1979). Rozvi appointed chiefs also changed their „zvidawo‟ praise names to Rozvi related ones for greater dignity, worth and political identification. Again, customs such as the holding of a handful of soil as a symbol of the right to possess land, and the holding of a handful of flour in a pool as a test for the spirits approval of a potential chief distinguished the Rozvi from other chieftainships. Bhila (1982a:18-19) gives another detailed account of Rozvi investiture rituals in Manyika. He noted that a Rozvi envoy took a piece of rope, which acted as a mystical link between the Rozvi, the new king and his people, and after that people would feast. If there was a pool the chief was immersed in water but if not, he would be simply sprinkled in an act of purification. After that a fuko (black and white cloth) was used to cover him, the black colour symbolizing his vices while white represented virtues. The chief‟s royal authority was symbolized by emblems of power namely; a parimwe/crown (a flat circular piece of iron with a hole in the centre to accommodate a string that would tie it on the left arm or bound it on the forehead), magutakudya (glittering conus button), maparimwi (a battle axe made of black polished wood inlaid with brass), and a conus shell symbolizing the chief‟s charismatic qualities. Finally the chief was given emblems of authority which included a bow and sheaths of arrows that probably belonged to his predecessor. He was also given a Rozvi girl, and then a set of laws were passed in relation to payment of tribute (Bhila 1982a). Again in Manyika, a new Rozvi chief was supposed to eat food cooked with pebbles from the stomach of a male crocodile that he had personally caught so that he 96 would symbolically become a crocodile (Bhila 1982a; Bullock 1927; Huffman 1996). This meant the Rozvi-appointed chief would be dangerous, ferocious, brave, and expected to live longer and experience growth throughout life, as well as communicate with the spirit world at the bottom of the pool. The practice of installing chiefs was also common among the Nambya who are also of Rozvi descent because they broke away from the central Rozvi core-area in the south-west. Ncube (2004:12) highlighted traditions stating that the installation of a chief among the Nambya entailed offering the candidate a portion of soil to symbolize his ownership of the land. This new ruler was also given a string of beads from the previous Sawanga (Chief) abbreviated to Wange as a badge of office (Ncube 2004). Beach (1982:6) noted that in 1929 when thoughts of reviving Mambo were muted, a ceremonial stick (tsvimbo) was sent from Goronga to Mavhudzi for installation purposes. Thus, the installation ceremony procedures varied through space and the symbolic objects for chieftainship were diverse among the Rozvi groups. Apart from carving an identity on the basis of installing chiefs, the Rozvi were known for granting land rights to incoming groups. Bourdillion (1976:85) stated that the association of a chief and land symbolized in installation ceremonies meant there was a strong bond between the spirits, the land and the chief. Zachrisson (1978) also noted that giving land rights was one of the Rozvi‟s sources of power, when they lost that grip, their identity also declined. Bhila (1982a) further mentions that the Rozvi distributed land to their feudatories, settled boundary disputes, and gave tributary rulers totems such as shumba (lion), shava (eland), nzou (elephant), and bonga (wild cat) in an effort to integrate and consolidate heterogeneous clans living in newly acquired lands. However it was not necessarily a Rozvi sub-chief who was installed to found a new Rozvi sub-dynasty. The vanquished simply had to respect Rozvi customary law, pay tribute and report regularly on state matters to the Rozvi emperor (Bhila 1982b:1). Hence the Rozvi did not always conduct these installation ceremonies and they did not necessarily appoint all the chiefs who were under their jurisdiction. Kinship ties among the Karanga have always been defined through clan names, mitupo, and subclan names, zvidao, often related to geographical areas suggesting common origins. Bourdillion (1976:38) adds that „mitupo‟ and „zvidao‟ are often „praise names‟, but they have religious and symbolic connotations, which make them more than just names. At times these names can be 97 changed simply to conceal the identity of the group or to adjust its relations with other groups (Bourdillion 1976). Loubser (1988:303) elaborates that „mitupo‟ do not necessarily constitute corporate bodies, genealogically related people, or territorially defined units because members of the same „mutupo‟ can either be royals or commoners. Oral traditions often mention status differences, and the status of each „mutupo‟ depends on political circumstances such that the history of status grades can be tentatively reconstructed through „mitupo‟ (Ralushai 1977:59). Thus oral traditions often identified pre-colonial groups such as the Rozvi through territorial regions, ruling houses, totems and sub-totems. These kinship attributes are well remembered in oral accounts and they are most effectively used in group identification as well as in distinguishing the „major Rozvi‟ from the „minor Rozvi‟ groups. Most oral traditions also point out that the titles „Changamire‟ and „Mambo‟ were associated with Rozvi leaders who were largely of the moyo (heart) totem (Beach n.d., 1976, 1983a, 1994a; Mazarire 2009; Zachrisson 1978). Hence the moyo totem was commonly used to define kinship ties among the Rozvi. 5.7 Rozvi traditions centred on totemic identity Beach (1994a) mentions that traditions collected during the late 19th century from a southwestern perspective indicate that the Rozvi rose from an east to north-east direction that point towards Murewa. That area was a moyo nuclear zone (meaning a territorial region with a concentration of people bearing the moyo totem) since it was enclosed by non-moyo dynasties before Changamire broke away from the Mutapa to establish his power-base in the south-west around the 1690s. This migration triggered a series of other movements that further complicated the moyo totem identity as non-moyo groups were also assimilated by the prominent Changamire dynasties, while the weak moyo groups that either remained behind or migrated to other areas where they eventually got absorbed by non-moyo dynasties. Although the moyo totem was generally accepted as a common denominator for Rozvi identification, its value in that respect was only situational as other non-moyo houses took over senior offices in the Rozvi administration (Zachrisson 1978). Thus there are far too many moyo totem houses that are mentioned in oral traditions, but not all of them controlled power relations among the Shona. The south-western Rozvi dynasties remained dominant, while the other moyo groups that remained in the north-east as well as those that migrated to the central plateau and south-eastern parts got submerged by different dynasties during the 18th century. Oral traditions state that when 98 the Rozvi arrived in the south-west, Chibundule, Chihunduro, Chiwundura or Chihunduru, Tumbare, Nichasike or Nichakadzike among others were the main ruling dynasties that were defeated and incorporated into the Rozvi leadership (Beach 1994a, 1976; Mazarire 2009; van Warden 1998; Zachrisson 1978). The Rozvi moyo totem became a fashion for some chiefs who strived to be identified with the Rozvi Mambos (Chiefs), as the case with Chihunduro-shoko, Tumbare-bhebhe, Mavhudzi-shava, Nerwande-shoko and Bepe-dziva/mbedzi, (see Beach 1994a; Mazarire 2009; Zachrisson 1978). These changed their “zvidawo (praise-names) to Mavhudzishava/moyondizvo, Tumbare-moyo/bhebhe, and Tumbare-moyo/mitombo, Chihunduro- moyo/shoko, Mangwende-moyo/muzukuru, and Svosve-soko/moyondizvo vhudzijena (Beach 1994a; Hodza and Fortune 1979; Zachrisson 1978). Thus Rozvi praise names and totems were never fixed or static; they always changed whenever the balance of power shifted as the case with Chihunduro-shoko, Tumbare-bhebhe, Mavhudzishava, Nerwande-shoko and Bepe-dziva/mbedzi. Some totems and praise names were invented or constructed as individuals sought to adjust to shifting power relations. Hodza and Fortune (1979:221) cited some of the Rozvi „zvidawos‟ that emerged when the Rozvi were integrating other chiefdoms and losing their grip on power as, Moyo Sinyoro (senhor), Moyo Chirandu, Moyo Wakapiwa, Moyo Mutaurwa, Moyo Muturikwa, Moyo Shavamombe, Moyo Mupfuyiwa, Moyo Mushoriwa, Moyo Muvhimwa, Moyo Muzivikani, Moyo Varasa Mugonderwa, Moyo Chipfuyamiti and Moyo Ndebele. These shifting totemic identities reflect the fluctuating nature of Rozvi identities in varying socio-historical contexts and that clearly resonates with the notion of ethnicity. During the first third of the 19th century the south-western Rozvi disintegrated into several dynasties. Resultantly, Chihunduru, Tumbare, Mavhudzi and Nerwande retreated north-west to become shoko (monkey) among the Nambya of Hwange and shava (eland) among the Shangwe of Mafungabusi Plateau (Mazarire 2009). The Singo who settled among the Venda are also believed to have broken away from the south-west between 1750-1800 as some of their early chiefs had names like Lozi (Rozvi), Mambo, Dyambeu and Thoho-ya-Ndou (Loubser 1988). Oral traditions trace the Singo from Matangoni (Matonjeni?) mountain (assumed as near Danamombe, the Rozvi capital), while others point towards Mberengwa mountain and some traditions emphasise the big lakes of Central Africa (Loubser 1988). Although they were Rozvi, 99 their totem was not moyo but singo, which also indicates that the moyo totem was no longer the sole attribute for defining kinship among the Rozvi. Even the rest of Rozvi groups from the south-west were simply identified through house names, and not totems, showing that Rozvi kinship ties were no longer entirely defined through totemic attributes. Rozvi houses in southeastern Bikita district of Zimbabwe were simply termed Jiri and Gumunyu, while those that set for Makoni in north-eastern Zimbabwe became Chiduku and Tandi, in upper Save there was the mention of house names like Tumbare, Nyashanu, Svosve, Chingombe, Mbava/Ruzane, Gwangwava-Musarurwa and Mutekedza among others as Rozvi (Beach 1976, 1983b; Mazarire 2009). 5.8 Traditions of the Rozvi dialect as a symbol of identity Oral traditions also cite dialect as one of the major characteristics that distinguished the Rozvi from their subjects. Linguistically, Rozvi was treated as part of Nyai and Kalanga dialects by Dock (1931). Kuper (1954) observed that the Rozvi dialect was still spoken by the „old people‟ near Bikita, while in other areas the Rozvi spoke local dialects like Zezuru, Karanga among others. Even scholars like Dock (1931), Posselt (1935) and Fortune (1956) noted that there was a dialect called Rozvi, which was scattered among various Shona groups like Jiri, Mbava, Chiduku, Mbire and Hungwe among others. Dock (1931) however noted that it was difficult to obtain a pure Rozvi speech during the time that he wrote his work, but stated the Rozvi dialect could be traced to a northern origin, and that it was closely related to Kalanga, Lilima and Nambya dialects. Makuvise from Makuvise village located among the Jiri Rozvi of Bikita district narrated Rozvi history in the Rozvi dialect as he recalled: Mushure maizozo kwakavuya Dziti laiyi Mcecenyani ndo kubva tabaya nalo ndo kulitata. Mushure malo kwakavuya limwe Dziti laiyi Zhishiri ndo kubayana nalo tikalitata… mailolo Dziti kwakavuya limwe laiyi… After those same events there came a Nguni called Mcecenyani and forthwith we fought him and chased him off. There came again another Nguni called Zishiri (Huge Bird?) and we fought with him and drove him off…there came another called…(Fortune 1956:71-2). 5.9 Traditions of Rozvi military power Finally oral traditions emphasise Rozvi identity on the basis of military strength. Changamire Dombo is well remembered as the first leader of the Rozvi and he is said to have been a leader 100 who was able, vain and cruel (Bhila 1982b). Documentary evidence and oral traditions confirm that Changamire had the most powerful army on the Zimbabwean plateau during the late 17th and early 18th centuries, which enabled him to be victorious in all major battles. Other traditions mention that the name Rozvi was derived from “kurozva” meaning “to destroy to nothing” (Bullock 1927; Posselt 1935). Since the name Rozvi only started appearing in Portuguese sources after Changamire Dombo had waged several battles, it is highly likely that his army made up of several clans earned itself the nickname „vaRozvi‟, meaning the destroyers. Fortune (1956:83-4) recorded traditions stressing some qualities of the Rozvi army: …It is the drum of Tumbale. Another name for it was Dulitsire. This is the drum of Tumbale which, when it has sounded, it appears that the army has come out to go and kill...This drum used to be played by Tumbale himself. He was a servant of the Mambo. He was a person who had come out with the Rozi venerating Bepe...This drum was such, that when it was beaten, there used first (to sound) the gun of Mambo which was called Cidzimbenhema. It used to be such that when it was sounded the whole land heard, and all knew that there was a force at the court... ... And the Rozi sees that the force has come, and it is announced: “Tumbale and Basi go and see the force which has come.” And they go to see and stand... “It has come to fight.” Go now Basi and Tumbale instruct your children as you go and fight.” Today you let the bees and they go and fight today with the army that has come.” Then they go then it is then to go and let out the bees from the magic horns... (When enemies had come to attack the Rozvi Tumbale and Basi were instructed to find out and report their findings. After the report they were then asked to instruct others to fight the enemy by letting out bees from magic horns) own emphasis. We learn from these traditions that Tumbare was a senior Rozvi army official, and that guns were owned by chiefs, who often used them to communicate military messages. The association of magical powers or trickery with Changamire and other post-Rozvi groups is very common in traditions; this could have been done to portray the army as sanctified by the ancestral spirits and Mwari. Thus, oral traditions highlighted the power of the Rozvi army by emphasising supernatural dimensions to its strength in order to portray it as invincible to any natural enemy. 5.10 Data from creative writing This research also considered information from creative writers as crucial in the examination of the Rozvi. Of course, many particularly historians would doubt the value of these sources in any attempt to understand the past because of concerns surrounding imagination, mythology and distorted historical facts. However, Shona praise-poetry constitutes an important component of African history because it celebrates clan origins, ancestors, memorable events and many other 101 socio-economic qualities of past generations. In this case nhetembo dzerudzi literally “praises of the clan” are important because they are uttered to ancestral spirits midzimu, the whole lineage or clan, and to the clansman who would have performed a good act (Fortune and Hodza 1974). The imagery by which thanks are conveyed varies from totems (animals or parts of living creatures) such as the zebra (tembo), eland (mhofu/shava), baboon (mukanya), monkey (soko), leg (gumbo) or heart (moyo), (Fortune and Hodza 1974). Usually praises follow the qualities of the animals such as its beauty, swiftness, and strength, but for the Rozvi the moyo (heart) is central to any living creature, hence it possibly signifies the central political role of the Rozvi in the past. Among the Shona, totems and clan praise poetry constitute a genre of oral traditions that convey cultural messages, values and histories of clan members (Pongweni 1996). Even though clan praise poetry is classified under creative or imaginative works, its role among the Shona has a very strong historical impression since it often addresses geographical origins, ancestral legends and political roles of particular clans (Kahari 1990). Therefore totemism and praise poetry constitute complex traditions that promote sentiments of attachment among group members, ritualised collective consciousness and respect for symbolic objects of group identity (Pongweni 1996). Totemism and praise poetry also performs social functions among the Shona such as facilitating exogamy in marriages, acting as social lubricants emphasising respect, serving to distinguish „them‟ from „us‟ (ethnicity), which enhances respect, love, cooperation and support; they also act as oaths whenever there is need to demonstrate sincerity or truthfulness of a report or promise, and finally they act as taboos with magical sanctions that promote solidarity through sharing of food and showing commendable behaviour towards fellow kin-members (Pongweni 1996). Thus both clan praise poetry (nhetembo dzerudzi) and sexual oriented praise poems (zvirevereve zvomugudza) metaphorically depict attributes like a common descent and substance as socially defined and not biologically or genealogically imposed. In this study, I confine myself to examining totems and praise poetry associated with the Rozvi so as to make infererences on how they imagined themselves and how they were perceived by others. Kahari (1990:103) observed that the moyo totem was associated with praise names like Moyondizvo/Bvumavaranda, Nematombo, Sinyoro, Wakapiwa, Muzukuru, Chirandu, Sithole, Wadyegora, Mavhudzi, Rwanza, Masukume, Dhewa, Tumbare and Zuruvi, thus indicating the complexity of the moyo clan. Historically and even in contemporary circles, the 102 names cited above are often widely used in identifying the Rozvi constituency. Hodza and Fortune (1979:223) adds that: Clan praises of the Rozvi recall their once extensive power and wealth. In them the function of the praises as propaganda and self-advertisement comes clearly to the fore ... The material of the praises is threefold, containing references to (i) the Rozvi wealth in cattle, (ii) their political power, and (iii) ancestral sites and names. Among the latter the names and praises of kings figure prominently. Rozvi political power is often vividly expressed through praise poetry, for instance Fortune and Hodza (1974:69) cited and translated the following Rozvi praise-poem: Bvumavaranda, „ You who receive the homage of your subjects, Vakadzi vachiza vomene; women coming of their own accord ; Jengetanyika; Keeper and master of the land ; Mwene wavanhu; Lord of the people ; VaChuru-chamipfunde-manji; Mound surrounded by abundant sorghum ; VaNyoka haisvosvwi. Those who are known as “ A snake is not to be provoked. Kuisvosva inoruma munhu. If you provoke it, it will bite someone”.‟ This example summarises most of the important political qualities of the Rozvi as great leaders with many subjects. This praise-poem is acknowledging that the Rozvi were the major custodians of the land. If they were land custodians, then their ancestral spirits were the guardians of the land thus reinforcing Rozvi control of other groups. It also emphasises that the Rozvi were the rulers of the people, hence their authority was unquestionable. The imagery of the mound of sorghum depicts the Rozvi as care-givers of their many subjects and there are several versions of oral traditions that highlight the same qualities. Finally, the snake metaphor carries the skilful and dangerous qualities of the Rozvi, which largely befits the Rozvi army. Another poem highlights the Rozvi dialect, power, wealth and great ancestral heroes. Shoko (2001:19-20) recorded and translated the poem shown below: Bhepe Nzinda Tumbare Wejila Jena VaChulu ChaNhave VaChisavase VaChigala nakabhanga VaChichengeta munongolo Chilailo chakanye ishomwe VaChilima mapfunde manji Aisava Tumbare Resident of the Royal capital He who drapes in white apparel Of the mountain Possessor of red soils He who carries his knife around Keeper of an impala Whose home supper is marula fruit Who dresses his wives in black Grower of plenty of sorghum But for Tumbare The Rozvi offspring would have perished in fire (war) 103 Bwana bwomuRozvi Bwaisapera kutsva mumoto Aiwa kani Tumbare Wejila Jena Maita zvenyu Bhepe Yea Tumbare He who dresses in white Thank you The poem above highlights a number of interesting aspects about the Rozvi; in particular it was recited in a Rozvi dialect. Tumbare, a former Rozvi military official is being praised on the basis of his dressing (white cloths also cited heavily in Moyo/Sinyoro praise poetry), weapons (knives), and history in warfare. His wives are said to have dressed in black clothes and this possibly signifies the value of cloth among the Rozvi in the past. Cloth as dressing material was rare when compared to animal skins, perhaps this scarcity made cloth a symbol of identity among the Rozvi back then. Unfortunately it was difficult to determine the exact date when this poem became part of Rozvi traditions. Assuming that it dates to the times when the Rozvi were the prominent ruling class in pre-colonial times, it informs us about dressing styles or fashion preferences for the Rozvi. Fashion styles relate to ethnic identities, hence these stylistic aspects could be very crucial elements of Rozvi identities. The poem also emphasises special qualities associated with the Rozvi, and these include royal residences, land ownership and their ability to produce surplus food for their subjects. A related Rozvi praise poem comes from Mutasa (1991:5) and it goes as follows: “Moyondizvo, Bvumavaranda! “Moyondizvo, Bvumavaranda! Vakadzi vachiza vomene; Women coming of their own accord; Muti unokopa chirimo! The tree which sprouts in the dry season! Kukopa zhezha unokonze ndove; In the rains it causes a swamp; Mambo asingatandavari; King who does not rest; Kutandavara mvura inonaya mubvumbi, If he rests, the rain would turn to mist, Nyamusimira mombe! One who pledges cattle! Kusimira mbudzi nemhuka urombo, Pledging goats and animals is poverty, Vakapera nenda! Those who perished on towers to the moon! Vagadzi voushe hwavaranda venyu isu!” Those who ordain rulers among their subjects!” The above Rozvi praise poem highlights several attributes of the group‟s identities. Firstly, it highlights the praise names, “Moyondizvo/Bvumavaranda”. It also emphasises the character of Rozvi Mambos as dangerous if provoked. Again the Rozvi are portrayed as very rich in cattle. The stanza “Vakapera nenda!” has been interpreted in multiple ways, some argue it means those who were ravaged by lice, while others suggest it means those who were killed by spears. This study favours the one, which suggests „Those who perished on their way to the moon‟, when 104 they built scaffolding using timber in attempts to get the moon as a present for their king or Mambo. This was a common myth that created strong sentiments of solidarity among the Rozvi. Nobert Mutasa, a local novelist also carried out extensive historical research on the Rozvi and his efforts culminated in the publication of two novels entitled Nhume Yamambo (1990) and Misodzi Dikita Neropa (1991). In both novels the voice of the narrator claims to be the voice of history that creates a quasi-historical story of the Rozvi. Chiwome (1996) noted that some fiction writers adopt an approach which, idealises the past in a manner that re-affirms the African identity. These writers usually construct their stories in a traditional context by using myths, symbols and characters as typical images of old Africa (Chiwome 1996). Although these novels are based on fictitious stories, there are certain instances where realistic events, characters or places are mentioned. At times these realistic aspects are sometimes exaggerated such that, readers not familiar with such details may be mislead into accepting such stories as true accounts of the past. This is the case with Nobert Mutasa‟s two novels cited above. His stories have very strong parallels with historical facts to the extent that it is difficult to distinguish fact from fiction. Actually, Chiwome (1996:76) writes: After independence, Mutasa published Nhume Yamambo (1990) which gives a realistic picture of the Shona past with its myths and legends. The story is based on extensive research about the culture of various inhabitants of the country in the hey day of Rozvi rulership. He adopts an Africanist perspective which sheds light on Shona politics and cosmology... At this stage, the author displays knowledge of significant Shona cultural movements. Thus Mutasa (1990, 1991) managed to identify several prominent Rozvi houses like Mutinhima, Jiri, Gumunyu, Tumbare, Washaya, Vhudzijena, Mavhudzi, Nerwande, Ruzane, Chireya, Chirisamhuru, Dyembeu, Dhewa, Govo, Samuriwo, and Changamire Dombo among others. Mutasa (1991:11) also wrote: “Nokuda kourozvi hwakapinza hwavatungamiriri ava hwokugona kuumba nyika ino sezvairi nhasi nemiganhu yayo, rudzi rwavo rwakazonzi vaRozvi vourozvi... Asi maPutukezi akarozverwa mapurazi aakanga apamba akavati ndivo vaRozvi vokurozva.” “Because of their good quality of governance that shaped this country and its boundaries, their clan was named Rozvi of intelligence and brilliance... But the Portuguese who lost their prazos to this group, which they had forcibly acquired named them Rozvi the destroyers.” 105 Although this was simply a novel, the shifting meanings concerning the name Rozvi and its origins was well put across. Beach (n.d:15) argues that by end of the 17th century a Torwa Kalanga Ndumba sub-chief who had married Changamire Dombo‟s sister was ruling from the Khami area. The author also identified the Torwa as well as their chiefs and areas of settlement accurately, for instance chief Ndumba of Khami was cited in the story. Most importantly sources of Rozvi power as well as their political wisdom were well articulated. Mutasa (1990, 1991) clearly noted that it was difficult to separate the Rozvi from their military background, their relations with Mwari and their ties with several chiefs scattered across the plateau. It was highlighted that the Rozvi used the diplomacy of giving their subject chiefs Rozvi girls as wives, but these wives represented the interests of their Rozvi family kin. In this way the Rozvi maintained their grip on surrounding subjects. The role of Mwari in the Rozvi governance cannot be over-stated and their fall from power was well articulated as stemming from conflicts with agents of Mwari, which was tantamount to disrespecting Mwari himself. The myths of the tower to the moon, building and moving mountains as well as the construction of a massive dry stonewalled monument depict strong symbolism and power among the Rozvi (see Beach n.d, 1976; Bullock 1927; Mutasa 1990, 1991). Nobert Mutasa further identified all Khami-type sites in the south-west and mentioned symbols of Rozvi chieftainship and power such as mountains and monumental architecture. Although the stories are fiction based, the historical parallels are amazing such that these may be useful sources for analogical reasoning. 5.11 Rozvi ethno-historical data A follow-up was made on post-Rozvi houses that left the core south-central and western area around the late 19th century to settle in the south-eastern district of Bikita in Zimbabwe. These were traced through the use of historical and ethnographic sources such as Beach (1983b) and Fortune (1956) respectively. The ethno-historical study sought to establish how the living Rozvi descendents perceive themselves in relation to their non-Rozvi neighbours as well as clarifying how they define their identity through material culture. The research revealed that communities living in Bikita still maintain very strong identity sentiments that particularly distinguish the Rozvi from the Duma, who constitute the majority population in the district. Interestingly, the Duma and the Rozvi share the same moyo (heart) totem. However they differ in terms of zvidawo (praise names). All Rozvi elders interviewed belonged to the Gumunyu house that shares 106 “Dewa/Moyondizvo/Bvumavaranda” as praise names while the Duma share praise names like “Gono and Mamvura” (see Plate 5.1). One of our informants called Pikirayi Makotore further argued that the Rozvi moyo (heart) is not specific to any animal species/living creature. He elaborated that the moyo (heart) totem for the Duma differs from that of the Rozvi in that it relates to some particular species. Although this view could not be ascertained, it indicates that the Rozvi have a very strong sense of difference from their neighbours. Plate 5.1. Starting from the right is Pikirayi Makotore (Rozvi), immediately next to him is Fani Rumbabwa Chisi (senior Rozvi “muzukuru” nephew), followed by Chabwera Makotore (Rozvi), the man with a hat on the knee is Lucky Tonderai Ngundu (Rozvi), on the extreme left is Rtd. Lt. Col. N. Mtombeni (assistant researcher). The other two young man in the picture did not participate in the interviews (Photograph by the Author, August 2011). In terms of customs, the Rozvi emphasised that they once spoke a “Rozvi dialect” which distinguished them from the Karanga, but today they generally practice the same customs as their neighbours. Such customs include venerating the graves of their ancestors “kutsvaira makuva evakafa”, performing traditional dances such as the “Ngororombe” festival, asking the family ancestral spirits to protect living family members “kupira gono”, abstaining from working the fields as a community on a single chosen day per week “kueresa chisi”, performing traditional ritual functions “mutambo webira”, and participating in traditional rain-making ceremonies 107 “kuita mukwerera”. Most importantly, it emerged that the Rozvi are considered as the natural leaders in rain-making ceremonies for the local communities in Bikita. Dziwairai Ushe is the current Rozvi representative who leads rain-making ceremonies even at the Duma shrine called “Tsime raPfupajena” (Pfupajena‟s sacred well) located in Chigure area (see Plate 5.2). He is also the only Rozvi representative from Bikita who goes to Mabweadziva (the major rain-making centre located in the Matobo hills in south-western Zimbabwe). The informants stated that the regalia that has been traditionally given to each representative to Mabweadziva includes mucheka wegungwe (a gown with white on one side and black on the other), bhandi dzvuku (a reddish-like cotton belt), and tsvimbo (ceremonial wooden stick). After returning from Mabweadziva with instructions, the Rozvi representative would need other objects like ndiro yedanda (a wooden plate, Plate 5.3) for putting “zviyo” millet used to brew traditional beer, mukombe wedanda (drinking wooden container), huha (ceremonial axe), chibako chefodya (snuff calabash) and tswanda (a woven bowl) to make an offering to the ancestors. Fani Rumbabwa Chisi also stressed that among the Rozvi, a traditional reed-woven surfer “rusero” is important when performing rain-making ceremonies because it symbolises the open character of the Rozvi and their ability to separate good from evil. After the ancestral offering, mutoro (traditional beer for rain) is brewed by women who would have passed their menopause. 108 Plate 5.2. The left picture shows Dziwairai Ushe (a Rozvi representative to Mabweadziva) inside “Tsime Rapfupajena” at Chigure demonstrating rain-making customs. The other two men in the picture are the Duma custodians of the sacred well. The right picture shows the sacred well at close range as well as broken clay pots and a fragment of a gourd cup (Photograph by the Author, August 2011). Once the beer is ready, it is shared from makate (very big black clay pots) and zvirongo (smaller black clay pots), by using either gourd cups or mukombe wemuti (drinking wooden container). These Rozvi traditions are often done while sitting under a particular muhacha/muchakata (Parinari curatellifolia) tree. However, if the Rozvi representative conducts rain-making rituals for the Duma and other people, he goes to Tsime raPfupajena at Chigure, where he drinks water from the sacred pool using a gourd cup. After he has finished he then pours more water into clay pots which are then carried by very old women to other senior members who also drink at a reasonable distance from the sacred pool. Objects needed for such rituals include black clay pots, uta nemiseve (bow and arrows), zvisanho (wooden hammers), chidende (gourd cup) and huha (a ceremonial axe). After these rituals people then go back to the venue of the traditional ceremony to drink mutoro (traditionally brewed beer for the rain). It was pointed out that the Duma now rely heavily on the Rozvi for such ceremonies because when they tried it alone they failed dismally. Pikirayi Makotore further emphasised the importance of Rozvi customs over “others” by claiming that fields for the Rozvi always receive bumper-harvests as compared to those of non-Rozvi people who do not respect Rozvi ritual customs. This view could not be verified but it 109 is vital in highlighting how the Rozvi always use knowledge as a means of gaining power over others. Plate 5.3. “Ndiro yedanda” the wooden plate that is now used by the Rozvi when they perform their rain-making traditions. This is currently kept at Shupai Makotore‟s homestead (Photograph by the Author, August 2011). This Rozvi dominance among the Duma was even cited for past military battles, and traditional songs where the Rozvi always lead in singing. We also learnt that a Rozvi representative was in charge of filling his mouth with water before blowing it through the bottom of all drums to be played during a ritual function. This was done as part of informing the spirits about the ritual play they were about to perform. Fanu Rumbabwa Chisi and Pikirayi Makotore further stressed that the Rozvi were responsible for installing chiefs before the colonial government usurped such powers. All the Rozvi informants stated that this power to lead others was acquired from Rozvi ancestors and Musikavanhu/Mwari (the sky God). When appointing chiefs, the Rozvi representative needed dehwe resimba/ingwe (leopard skin), uta nemiseve (bow and arrows), tsvimbo (ceremonial stick), and masese (traditional beer) which was then poured on the chief after a poem was narrated. Fani Rumbabwa Chisi recited the Rozvi poem for installing chiefs as follows: Nhasi takugadze Ushe; Wekubata ivhu raBambo vako! Muvhu mune zvakawanda, Zvirema, mapofu, mhuka, nyoka, zvese nemakudo; Today we install you as Chief; To guard the soil of your Fathers! The soil has a lot of things, The crippled, the blind, animals, snakes, and even unruly elements; 110 Zvese zviri muvhu nhasi ndezvako! Unobata vanhu nezviri muvhu zvakasikwa naMwari; All that is in the soil today belongs to you! You rule your people and all in the soil, which was created by God; We give you this handful of soil, We pour on you this traditional beer Saka tinokubatisa iri ivhu, Tichipendera nezvinobva muvhu! which comes from the soil as we thank the soil! It also emerged that mountains were very central to Rozvi history and identities (see Plate 5.4). Our informants such as Fani Rumbabwa Chisi, Pikirayi Makotore, Lucky Tonderai Ngundu, and Dziwairai Ushe emphasised that mountains found in Bikita such as Chironde, Chiono, Chakadya, Gwindingwi, Bokokwi and Baradzanwa were central to the Rozvi because they signify the earliest Rozvi settlements, Rozvi ancestral burials and mysterious events associated with the spirit-world. They also argued that some of the mountains provided the best defence for the Rozvi during times of conflict in the past. This was quite an interesting point because it reflects how elements of nature are objectified to signify ethnic identities through various discourses ascribed to them. The Rozvi in Bikita also have their dare (traditional court) located at Shupai Makotore‟s homestead where they gather for initial rain-making ceremonies and other important family events (Plate 5.5). Interestingly, this traditional court has R-style dry-stone walls which indicate that walling could have been an important aspect of Rozvi customs from the past to the present. Unfortunately, the Rozvi only considered the dry-stone wall as part of their traditional court, they do not attach special meanings to the walls. On the whole, all the Rozvi informants could fairly highlight the material objects associated with their traditions. Dziwairai Ushe also emphasised that the Rozvi often wore a metallic bracelet on their hands called “shambochena”. We however never saw one among the Rozvi we interacted with during the course of our research. 111 Plate 5.4. Chironde mountain, one of the several mountains that the Rozvi attach their historical origins on. It is claimed that several Rozvi burials are scattered in this mountain but the researcher did not get an opportunity to survey the mountain for verification purposes (Photograph by the Author, August 2011). Plate 5.5. The Rozvi “dare” court at Shupai Makotore‟s homestead with a single R-style drystone wall. This is where all important Rozvi matters are dealt with and it constitutes the starting point for rain-making customs associated with Mabweadziva (Photograph by the Author, August 2011). 5.12 Overview of Rozvi identity markers The data shown above testifies that Rozvi was a consciously constructed identity that gradually acquired multiple meanings throughout the course of history. Initially, there was no mention of 112 that name until the mid-18th century when a series of important socio-political events associated with Changamire Dombo and his followers had occurred. A review of the earliest written sources has proven that a great military leader called Changamire suddenly rose to power through a series of military campaigns backed by young fighters. In no time, that military formation was transforming itself into a formidable ruling dynasty popularly known as Rozvi. Eventually, Rozvi became a dynamic identity as it referred to Changamire Dombo‟s close kin, the co-opted (Torwa and other groups), core territories of the political entity, chiefs appointed by the Rozvi and the list goes on. Both documentary sources and oral traditions revealed that Rozvi elites capitalised on discourses and ideologies centred on religion, mythology, installation of chiefs and land-rights to sustain their ethnic identity and dominance in power relations. At some point in history, totemism and the Rozvi dialect acted as attributes of their ethnic identification. Most importantly, various traditional customs and material culture were subjectively objectified to signal Rozvi ethnicity. Generally, Rozvi installation ceremonies included; the red cap, iron rod or sceptre, assegai, a bow and sheaths of arrows, piece of rope, black and white cloth (fuko), a string of beads, ceremonial stick, battle axe, conus button, conus shell, a flat circular piece of iron with a hole in the centre, offering a Rozvi girl, a handful of soil as well as performing several crocodile rituals. The ethno-historical study confirmed some of the Rozvi customs and it also revealed the dynamic character of the Rozvi. All the aspects cited above served to distinguish the Rozvi from their subjects. Therefore, several elements of tradition were carefully selected and objectified by the Rozvi elites in order to carve an ethnic identity. In the next chapter, I present the archaeological data. 113 Chapter Six Archaeological Data Presentation: Material culture as Rozvi witnesses “Material culture is a storehouse of signs that can be brought out actively to mark difference” (Clark 2005:440). 6.1 Introduction This chapter strives to probe material culture from Khami and Danamombe for possible differences that can be attributed to Rozvi ethnicity. Previous chapters have shown that Torwa dynasties lost both their power and major settlements to the Rozvi in 1696. As a result, controversy revolves around defining specific Torwa and Rozvi archaeological signatures at these Khami-phase sites in south-western Zimbabwe. Beach (1994a:96) and Huffman (1996:51) argue that the Torwa civil war of 1644 resulted in the abandonment of Khami (Torwa capital 1450-1644) and it was never re-inhabited by the Rozvi. According to Hegmon (1992:527) in most cases ethnicity is associated with distinct stylistic differences in material culture. Thus, a thorough comparative study of material culture from Khami and Danamombe could possibly expose changes at the latter site signalling Rozvi ethnicity. Several scholars like Clark (2005), Gosselain (2000), Jones (1997), Stone (2003), and Wells (1998), have argued that ethnicity is expressed through style in objects and structures. Therefore this study critically engages all perspectives towards style because a multi-dimensional approach is more valid for social group studies (see Hegmon 1992). Theoretically, there is isochrestic style/variation associated with technological aspects of material culture (Hegmon 1992; Sackett 1990). Isochrestic style can provide resources for ethnic identification when taken up and objectified through human agency. These objectified stylistic attributes (symbols) are referred to as emblemic style or iconological style, which actively communicates messages about ethnicity, power, power relations, individual/group identity and status (Hegmon 1992). Style is the general appearance/design of any artefact or structure that was acquired from the manufacturing process, with little or no bearing on its function (see Bowser 2000; Hegmon 1992; Jones 1997; Sackett 1990; Wiessner 1983, 1985). Style is typical and representative of groups, hence it reflects human behaviour (Huffman 2007). It is important to remember that there is great ambiguity involved when dealing with identities such as ethnicity. We cannot simply equate material culture style to an ethnic group because that relationship is extremely fluid and 114 sophisticated (Hodder 1982). However since identities are usually defined by subjective boundaries or differences in cultural entities, it is generally assumed that a cluster of different stylistic attributes or mundane artefacts signified ethnic symbols (Emberling 1997; Stone 2003; Wells 1998). Thus any element of culture whether tangible or intangible with potential to mark difference from the so-called “others” was most likely adopted and manipulated by ethnic elites to symbolise or express group identity. Following these premises, multiple material culture in the form of ceramics, beads, architectural structures and other miscellaneous finds from the two sites were examined in order to trace trifling changes or slight stylistic differences that could possibly signal Rozvi ethnicity. 6.2 Ceramic analysis and results It is crucial to take into account all previous ceramic analysis on the Khami and Danamombe assemblages before presenting ceramic data for this study. Caton-Thompson (1931) observed that the lower occupation levels at Danamombe had a lot of undecorated globular jars and bowls that were polished in black, gray and red ware. She further noted that this distribution of ceramics was almost unchanged throughout the stratigraphy as the upper levels had isolated samples of polychrome band and panel ware. Schofield (1948) also recognised that Danamombe comprised of undecorated spherical pots either with or without necks constituting two variants namely, R3 (1a) and (1b). These ceramic variants sometimes occurred in association with polychrome wares. R3 (1a) was exclusive to Danamombe and it had thick spherical pots with black, red and reddish brown burnished surfaces, while (1b) constituted thinner pottery with a red burnish, but more often it was either black or graphite burnished. Summers (1959) excavated at Danamombe and made similar observations to the above scholars, but he emphasised that black graphite wares were confined to the lower levels. Thus previous scholars generally came up with consistent descriptions of ceramic assemblages from Danamombe. According to Robinson (1959) ceramics from Khami comprised of three principal types. The first one being Form A consisting of spherical pots without necks and sometimes decorated, then Form B characterised by spherical pots with thick rolled rims and short concave necks, but rarely decorated. Finally Form C was described as having spherical pots with tall vertical, concave or convex, conical, bowl-shaped or funnel-shaped, and fluted necks. The neck or 115 shoulder regions for these pots were either highly decorated with polychrome band and panel motifs or left plain. Polychrome band and panel wares appeared throughout the stratigraphy at Khami but a higher frequently was in the upper levels (Robinson 1959). Hughes (1997) also excavated at Khami and identified Period I occupation levels dominated by globular pots with tall and short necked jars, and Period II characterised by constricted globular pots with little or no necks. Generally ceramics from Khami and Danamombe had almost similar vessel profiles, however there are two differences worthy of comment. Firstly Khami ceramics were more profusely decorated than those at Danamombe, and Robinson‟s Form C ceramics were either absent or extremely rare at Danamombe, where some form of specialisation in the manufacture of globular constricted pots with thick sherds, but without necks occurred. On this basis, it is now appropriate to present ceramic data for this particular study. A multidimensional approach was adopted in the analysis of ceramics from Khami and Danamombe. This systematic and replicable procedure for classifying pottery has been popularised by Huffman (1980, 1989, and 2007) and Soper (1971) to study ceramics for purposes of defining cultural and ethnic groups in southern Africa. This approach initially classifies attributes such as vessel form, lipform, fabric, exterior colour, decoration technique, decoration placement, decoration motif and sherd thickness into general clusters. Following Bowser (2000), Hegmon (1992), Hodder (1982), and Plog (1983), combinations of individual stylistic attributes such as vessel form/profile and decoration designs were prioritised in defining Rozvi ethnic expression. Huffman (2007) argues that correlating dimensions like vessel profile, design layout and motif is key in defining ceramic types that can be equated to people and their languages. However the relationship between such types and ethnicity is not automatic or straight forward, as ethnicity is subject to both social and historical contexts. While it could have been more appropriate to use Huffman‟s multidimensional approach, the fragmentary nature of the sherds made it impossible to accurately reconstruct design layout. As such vessel profile and motif dimensions were used as the primary means of defining ceramic types, eventually decoration layout became subordinate to these two dimensions. Technological attributes such as fabric were also incorporated in the analysis because they are part of isochrestic style (see Sackett 1990; Hegmon 1992). For comparative purposes vessel shape categories were used as highlighted in Chapter 4. 116 The number and percentage of each shape category for both sites is shown in Tables 6.1 and 6.2 below. In addition the frequencies for all shape categories are illustrated in the graph on Fig. 6.1. The data from Danamombe indicates that in terms of temporal variation, category E (dependent restricted vessels) corresponding with Form III (Hughes 1997) and Form A (Robinson 1959) generally increased in frequency through time. While that was happening category D (independent restricted vessels) corresponding with Form I and II (Hughes 1997), Form B and C (Robinson 1959) gradually decreased in frequency at the same site. In contrast, these shape categories both fall within the 20% frequency range at Khami thereby showing that some slight spatial and temporal changes in terms of vessel shape occurred at Danamombe. The reason why Khami dominates in terms of all the other shape categories may be that the sample used was by far less fragmented as it was from the Hill complex. On the other hand, Danamombe had the most fragmented sherds in category A, which constituted 88.33% of the entire sample. Category F (restricted vessels with simple contour-constricted bowls) was either absent or too restricted at the two sites. All shape profiles from Khami and Danamombe are shown on Fig. 6a-k below. In particular, Fig. 6a-c represent shape profiles exclusive to Khami, while Fig. 6d-k shows shape profiles from both research sites. The latter profiles were combined because sherds recognised from the Khami assemblage were similar to those identified at Danamombe (see appendix E for a detailed description of vessel shape profiles presented below). Table 6.1 Khami Hill Complex: vessel shape categories C D E F G TOTAL 63 82 92 85 84 13 0 419 Frequency 15.04% 19.57% 21.96% 20.28% 20% 3.10% 0.00% 100% A B Key: A-G are the broad vessel shape categories outlined in Chapter 4 117 Table 6.2 Level Surface 1 2 3 4 5 6 7 8 9 10 11 12 13 Total Frequency Danamombe Kitchen midden trench: vessel shape categories A B C D E F G TOTALFrequency 50 1 3 3 1 58 1.76% 370 370 11.22% 404 23 2 429 13% 375 1 37 6 419 12.70% 539 8 13 10 570 17.28% 431 1 2 2 2 438 13.28% 185 1 11 8 23 4 232 7.03% 192 12 2 10 24 1 241 7.31% 162 6 9 1 10 2 190 5.76% 125 5 3 8 62 2 205 6.21% 12 2 4 3 12 33 1% 41 4 2 5 15 67 2.03% 4 4 12 20 0.60% 24 3 27 0.82% 2914 35 35 46 239 30 0 3299 100% 88.33% 1.06% 1.06% 1.40% 7.24% 0.91% 0% 100% Key: A-G are the broad vessel shape categories outlined in Chapter 4 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% Khami 40.00% Danamombe 30.00% 20.00% 10.00% 0.00% A B C D E F G Fig. 6i: Graph showing the frequency of vessel shape categories for Khami and Danamombe. 118 Pots with tall necks, and pots with short in-sloping, vertical, and concave necks. Fig. 6a: Pots with tall necks and pots with concave necks. Fig. 6b: Pots with short vertical necks and out-turned rims. Fig. 6c: Pots with in-sloping and out-turned rims. Fig. 6d: Pots with short concave necks. 119 Globular pots with very restricted concave necks. Fig. 6e: Constricted pots with in-sloping rims. Fig. 6f: Constricted pots with no necks. Globular pots with no necks; plain/simple rims and rolled/out-turned rims. Fig. 6g: Constricted pots with no necks. Fig. 6h: Constricted pots with no necks. 120 Pots with tall necks, and pots with short in-sloping, vertical, and concave necks. Fig. 6i: Constricted bowl. Fig. 6j: Slightly constricted bowls and an open shallow bowl. Fig. 6k: Deep hemispherical and deep straight sided bowls. 121 The two ceramic assemblages were also analysed in order to establish the general character of lipforms. Table 6.3 and Fig. 6ii below shows that rounded and tapered lipforms were popular for both assemblages, the slight variations in frequency could be a result of sampling strategies used in this study. Khami also dominates in terms of internally thickened, bilaterally thickened, beveled and fluted lipforms. Conversely, squared and externally thickened lipforms were more common at Danamombe than at Khami thereby signaling spatial variation/changes between the two sites. As illustrated on Fig. 6ii below, internally thickened, beveled, and fluted lipforms were totally absent at Danamombe. Most globular constricted pots at Danamombe had externally thickened rims while squared lipforms were common for bowls. These vessel forms had a strong bearing on the lipform differences between the two sites. The significance of these minor stylistic changes at Danamombe should be carefully examined in terms of Rozvi ethnic signification. However it is also acknowledged that the potters‟ preferences could have influenced such stylistic variations. Table 6.3 Site Name Ro. 159 58.03% 164 Danamombe 45.18% Ke y: Khami Ro.= Rounded Summary of lipforms from Khami and Danamombe Ta. Sq. Ex. Th. In. Th. B. Th. Be v. Fl. Total 87 8 3 2 4 10 1 274 31.75% 2.92% 1.10% 0.73% 1.46% 3.65% 0.36% 100% 124 39 35 0 1 0 0 363 34.16% 10.74% 9.64% 0% 0.28% 0% 0% 100% Ta.= Tapered In. Th.= Internally thickened Sq.= Squared Ex. Th.= Externally thickened B. Th.= Bilaterally thickened Bev.= Bevelled Fl.= Fluted. Fig. 6ii: Graph showing lipform frequencies from Khami and Danamombe. 122 Vessel fabric was also analysed for both assemblages in order to define technological preferences in the manufacture of ceramics at the two sites. Results are shown on Table 6.4 and the frequencies on Fig. 6iii. Course fabric was not widely used especially at Danamombe where it constitutes a meager 1.06% of the entire assemblage. A notable difference between the two assemblages relates to the use of medium fabric, which gave a figure of 86.03% for Danamombe against 45.36% from Khami. It is also evident that fine fabric was widely used at Khami constituting 46.57% as compared to just 12.88% from Danamombe. This data therefore reveals that there were significant technological differences between the two research sites. As noted earlier on, certain technological attributes provided good resources for agents to signal ethnic identities through „emblemic style‟ (Hegmon 1992). Table 6.4 Site Name Khami Danamombe Summary of vessel fabric from Khami and Danamombe Course Medium Fine Total 40 225 231 496 8.07% 45.36% 46.57% 100% 35 2838 425 3298 1.06% 86.05% 12.89% 100% 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% Khami 40.00% Danamombe 30.00% 20.00% 10.00% 0.00% Course Medium Fine Fig. 6iii: Graph showing vessel fabric frequencies from Khami and Danamombe. 123 The two assemblages were also compared in terms of vessel exterior colour and the results are presented on Table 6.5 and Fig. 6iv below. It is apparent from the data that both assemblages were generally well fired as the black, greyish-brown and reddish-brown colours dominate the assemblages. Poorly fired vessels which are identified by a greyish colour were less frequent at both sites. One major difference between the two assemblages is that 15.14% of sherds from Danamombe were tainted with soot as compared to just 3.42% from Khami. This difference was a result of the contexts from which the samples were drawn, the assemblage from Khami was from the Hill Complex, while Danamombe ceramics came from the Kitchen midden, which most likely had a lot of broken cooking vessels. Plog (1983) argues that cooking pots are less likely to be used to express social messages because of their short lifespan. Rather, storage pots or serving bowls are frequently used to express social messages or identities because they survive longer than cooking pots and they carry different decorative designs (Plog 1983:138). Table 6.5 Site Name Summary of vessel exterior colour from Khami and Danamombe Black Grey G. Brown R. Brown Soot ? Total 158 36 108 119 15 2 438 Khami 36.07% 8.22% 24.66% 27.17% 3.42% 0.46% 100% 834 191 1212 464 482 0 3183 Danamombe 26.20% 6% 38.08% 14.58% 15.14% 0% 100% Key: G. Brown= Greyish brown R. Brown= Reddish brown ?= Eroded surface 40.00% 35.00% 30.00% 25.00% 20.00% Khami 15.00% Danamombe 10.00% 5.00% 0.00% Black Grey G. R. Brown Brown Soot ? Fig. 6iv: Graph showing the distribution of exterior vessel colour. 124 Khami and Danamombe assemblages differed in terms of sherd thickness. Small sherds were less frequent for both assemblages as Khami had 7.95% while Danamombe just had 4%. Khami had the greatest frequency of medium sherds while Danamombe had a higher frequency in terms of very thick sherds (see Table 6.6 and Fig. 6v). These spatio-temporal variations could have been a result of several factors, but it is likely that the globular constricted pots common at Danamombe largely influenced the variations in sherd thickness. A bigger pot generally has a thicker sherd to suit its function and enhance its durability, but this technological factor can also indirectly reflect group identity especially if treated as an emblemic stylistic preference (see Hegmon 1992:129). Table 6.6 Summary of sherd thickness from Khami and Danamombe Site Name Small Medium Large Total 21 194 49 264 Khami 7.95% 73.49% 18.56% 100% 14 162 174 350 Danamombe 4% 46.29% 49.71% 100% Key: Small= >6mm Medium= 6-9mm Large = 10mm + 80.00% 70.00% 60.00% 50.00% 40.00% Khami 30.00% Danamombe 20.00% 10.00% 0.00% Small Medium Large Fig. 6v: Bar graph showing sherd thickness for Khami and Danamombe. Decoration was first categorised according to dominant technique, which was then cross tabulated to establish motif combinations (see Table 6.7-8). At Khami graphite burnishing was the most frequent decoration technique constituting 51.06%, while burnishing or polishing of plain vessels was the most common technique at Danamombe constituting 95.90%. Applied designs/complex panel motifs were the second most dominant technique at Khami constituting 125 32.33%, while these constituted only 0.45% at Danamombe. Burnishing/polishing was limited to just 6.94% of the Khami assemblage. Punctates and comb-stamps are absent from the Danamombe assemblage while these were present at Khami albeit in restricted numbers. The most dominant motif combinations at Khami are applied designs, as incisions, graphite burnishing and red ochre painting were combined into various polychrome motifs. From Table 6.8, it is apparent that burnishing and polishing remains dominant but graphite burnishing, red ochre painting and engravings were combined into different motifs. Therefore although polychrome band and panel motifs were present in both assemblages, their occurrence and frequency significantly varied (see Fig. 6l-n for decoration motifs). The contexts from which these samples were chosen could have influenced these kinds of results. In that case there is room to assume that these changes could possibly indicate the influence of socio-political forces on the potters. This follows the view that decorative designs on pottery convey complex and ambiguous information about ethnic identities, power or status in social relations (Hegmon 1982:525). Therefore the few decorated or even the dominant plain sherds at Danamombe could have served as Rozvi ethnic symbols. A critical discussion of these stylistic variations in relation to Rozvi ethnicity is presented in Chapter 7. 126 Table 6.7 Summary of decoration technique and technique combination: Khami D.Tech. Inc. Pun. B/Po. Gb. Pai. C/st. Engr. App. Total Freq. Inc. 14 47 13 74 22.36% Pun. 1 5 6 1.81% B/Po. 23 23 6.95% Gb. 105 105 31.72% Pai. 9 9 2.72% C/St 3 6 1.81% Engr. 1 1 0.30% App. 107 107 32.33% Total 14 1 23 169 13 3 1 107 331 100% Freq. 4.23% 0.30% 6.94% 51.06% 3.93% 0.91% 0.30% 32.33% 100% key: D.Tech.= Decoration technique Inc.= Incision Pun.= Punctates B/Po.= Burnished polished Gb.= Graphite burnished Pai.= Red ochre paint C/st.= Comb-stamped Engr.= Engraved App.= Complex designs including several technique combinations Table 6.8 Summary of decoration technique and technique combination: Danamombe D.Tech. Inc. Pun. B/Po. Gb. Pai. C/st. Engr. App. Total Freq. Inc. 1 1 2 0.11% Pun. 0 0.00% B/Po. 1705 1705 95.84% Gb. 27 27 1.52% Pai. 13 14 27 1.52% C/St 0 0.00% Engr. 1 6 3 10 0.56% App. 8 8 0.45% Total 1 0 1706 41 20 0 3 8 1779 100% Freq. 0.06% 0.00% 95.90% 2.30% 1.12% 0.00% 0.17% 0.45% 100% key: D.Tech.= Decoration technique Inc.= Incision Pun.= Punctates B/Po.= Burnished polished Gb.= Graphite burnished Pai.= Red ochre paint C/st.= Comb-stamped Engr.= Engraved App.= Complex designs including several technique combinations 127 Fig. 6l: Incised and engraved motifs from Khami and Danamombe. 128 1. Single incised horizontal line. 2. Multiple horizontal incised lines. 3. Single horizontal and multiple oblique incisions forming a triangular design. 4a. Horizontal band of incisions with wavy lines forming loops, one with multiple oblique incisions. 4b. Similar to 4a but incised oblique lines more distinct. 5. Engraved horizontal lines with cross hatched engravings. 6. An angled rectangular design with multiple cross hatched engravings. 7. Horizontal and oblique engraved lines forming triangles with cross hatched lines. 8. Multiple alternating oblique incisions forming a triangular design. 9a. Almost similar to 8 but with cross hatched lines. 9b. Similar to 9a but cross hatched incisions on the left side. 9c. Fine line incised horizontal band with multiple oblique and horizontal incisions forming triangular panels. 10. Fine line diagonal and oblique incisions with small triangular designs. 11. Horizontal band of herringbone incisions. 12. Engraved cross hatched lines forming a triangular panel. 12a. Horizontal engraved band with engraved cross hatched lines. 12b. Similar to 12a but defined by fine line incisions. 12c. Engraved lines forming triangle designs, some triangles have multiple cross hatched lines. 129 Fig. 6m: Polychrome band and panel motifs from Khami and Danamombe. 130 13a. Oblique punctates forming a single horizontal band. 13b. Incised lines forming a single horizontal band with multiple oblique punctates filled with graphite burnish. 13c. Similar to 13b but punctates are cross hatched and filled with graphite burnish. 13d. Multiple oblique punctates forming a horizontal band separated from a lower band of red ochre paint by a fine line incision. 13e. A narrow horizontal band of graphite burnish separated by a fine incised line from a band of punctates forming a herringbone pattern. 14a. Vertical to oblique comb-stamp impressions. 14b. Horizontal band of comb-stamped impressions separated by a fine line incision from a lower band of graphite burnish. 15. Horizontal band of stab and drag impressions filled with graphite burnish. 16a. Horizontal band of red ochre separated from another band of graphite burnish by an engraved line. 17a. Alternating horizontal bands of graphite burnish with a band of red ochre in the centre. All bands are separated by fine line incisions. 17b. Similar to 17a but found at Danamombe. 17c. Similar to 16a but band of graphite is on top. 17d. Oblique band of alternating red ochre bands with a single band of graphite burnishing at the centre. All bands are separated by fine line incisions. 18a. Horizontal incised bands of graphite burnishing and red ochre. Lower band defined by multiple oblique incised lines. 18b. Multiple horizontal incised bands with oblique incised lines. The central band has red ochre infill while the lower one is graphite burnished. 18c. Alternating horizontal bands of graphite burnishing and red ochre defined by engraved lines. The lower band has incisions forming triangles or chevron patterns? 18d. Broad horizontal band with alternating bands of graphite and red ochre forming a herringbone design at the centre. The upper part is engraved with alternating bands of red ochre and graphite burnishing, while the lower part has horizontal engravings and oblique fine line incisions filled with graphite burnish. 131 19. A single horizontal and multiple oblique engraved lines that are alternating to form interlocking triangles filled with red ochre and graphite burnish. 20a. Band of graphite burnish separated by fine line incisions from a horizontal band with red ochre infill. The lower band has fine line cross hatched incisions. 20b. Similar to 20a but lower cross hatched band filled with graphite burnish. 20c. Same as 20a but cross hatched band is at the top. 21a. Red ochre band defined by horizontal fine line incisions with fine line cross hatched triangles pointing downwards. 21b. A single horizontal band joined by two lower interlocking diagonal bands with red ochre infill. Fine line incisions define the bands and a small cross hatched triangle pointing downs defines the lower left part. Both the upper and lower sections are filled with graphite burnish. 21c. Incised diagonal bands with red ochre infill joining a single horizontal band with graphite burnishing. Fine line cross hatched incisions define the top central part of diagonal bands, while multiple vertical incisions appear on both sides of the diagonal bands with graphite burnish infill. 21d. Red ochre triangular band defined by incisions and cross hatching at the centre, lower part simply graphite burnished. 21e. Incised triangular bands filled with alternating graphite burnish and red ochre infill. The central triangle has fine line cross hatched incisions. 21f. Same as 21e but central triangle has a single vertical incised line and multiple horizontal incisions with red ochre infill. 22. Alternating chevron or zigzag incised bands with alternating graphite burnish and red ochre infill. 22a. Horizontal band of red ochre defined at the top by cross hatched horizontal and vertical incisions with graphite burnish infill. 22b. Horizontal incised bands with alternating graphite burnish and red ochre infill. Upper part defined by cross hatched fine line vertical and horizontal incisions with red ochre infill. 23a. Alternating horizontal bands of graphite and red ochre, defined at the top by multiple oblique incised lines forming alternating triangles. Upper middle triangle with no incisions is graphite burnished. 23b. Similar to 23a and b but defined by both engraved and incised lines. 132 24. Almost similar to 23a but turned up-side down, upper part defined by a broad engraved band with cross hatched engravings filled with red ochre. 25a. Oblique fine line incisions defined by diagonal alternating bands of red ochre and graphite burnishing. 25b. Similar to 25a but diagonal bands and oblique incisions forming an up-right triangular design. 25c. Similar to 25b but triangular design pointing downwards. 26. Diagonal engraved bands with alternating graphite burnish and red ochre infill forming a triangular design. 133 Fig. 6n: Complex panel motifs from Khami. 134 27a, 27b. Complex wrapped fibre or bead impression motifs. 28, 29. Complex comb-stamped motifs. 30, 31a-d. Complex polychrome engraved motifs. 32a-c, 33. Complex polychrome incised motifs. 34a-b, 35a-b, 36a-c. Complex polychrome incised and engraved motifs. Diagnostic sherds were further categorised according to combinations of vessel form and decoration motif or surface treatment. Highly fragmented sherds presented challenges in establishing decoration layout/organisation for both assemblages. However as a counter measure, vessel classes already defined by Hughes (1997) and Robinson (1959) were consulted and used in efforts to define types charactering Khami and Danamombe assemblages. On this basis a total of 6 vessel types and 14 sub-types were defined from both assemblages. Initially, sub-types for all classes were established but those that were not represented in the analysed samples were ignored completely in the list below. Thus recognised types and sub-types were described as follows: Class 1. Shouldered pots with long vertical, convex and funnel shaped necks; plain. C1a. Shouldered jars with long, vertical, convex and funnel shaped necks; complex motifs on the neck and shoulder. C1b. Shouldered jars with long fluted and vertical necks; bands of graphite and/or red ochre on the rim, neck, shoulder or entire body. Class 2. Shouldered pots with short vertical, convex necks, and in-sloping rims; plain. C2a. Same profiles as C2 but with bands of incised motifs on the neck, graphite inside rim or entire body. C2b. Same profiles as C2 and C2a but with simple bands of red ochre or graphite burnish inside rim, or entire body. Class 3. Spherical constricted pots with no necks and simple rims; plain. C3a. Same profile as C3 but with band of either graphite or red ochre inside rim and/or motifs on the shoulder. C3b. Spherical constricted pots with little necks or out-turned rims; plain. 135 C3c. Same as C3b but with band of graphite or red ochre inside rim and/or complex motifs on the upper shoulder. C3d. Spherical pots with thickened rims/ rolled rims; plain. C3e. Same profile as C3d but with band of graphite or red ochre inside rim and/or complex motifs on the upper shoulder or entire body. Class 4. Constricted bowls with squared and rounded lips; plain. C4a. Variant with band of graphite or red ochre inside rim. C4b. Variant with entire body graphite burnished. Class 5. Deep open hemispherical bowls with plain to slightly out-turned rims; plain. C5a. Variant with band of graphite or red ochre inside rim. C5b. Variant entire body graphite burnished. Class 6. Open shallow bowls with rounded lip forms; plain. C6a. Variant with red ochre or graphite burnished band inside rim or entire body. For a comprehensive appreciation of the distribution and frequency of these classes at Khami and Danamombe see Table 6.9, 6.9.1, and Fig. 6vi. As can be discerned from the graph below, ceramic classes 1, 1a, 2, 2a and 5b were exclusive to the Khami assemblage. While Khami had 14 well represented ceramic classes, Danamombe only had 11. The data suggests that tall necked globular vessels, as well as shouldered pots with short vertical, convex necks, and in-sloping rims with decorations were absent or rare at Danamombe. The only necked pots at Danamombe matching the above shape profiles were plain (class 2), and they constituted a paltry 9.62% when compared to 15.25% from Khami. On the contrary, spherical constricted pots with little or no necks were prevalent at Danamombe and the frequency margins significantly differ (see Classes 3, 3b, and 3d on Fig. 6vi). Plate 6.2 and 6.3 shows some of the neck-less spherical globular polychrome pots that were recovered from Danamombe. It is also noticeable that Danamombe dominates Khami in terms of Class 3a but the margin of difference is relatively lower. Class 4 constricted hemispherical bowls with square and rounded lipforms are exclusive to Danamombe. Furthermore deep open hemispherical bowls with plain bodies and out-turned rims, as well as those with a slip of red ochre or graphite burnishing inside the rim remain dominant at Danamombe whilst they are rare at Khami. It is difficult to read too much into these bowl 136 variations because variation is due to the individual potters „hand‟. The sample for class 6 (open shallow bowls) is very low for both assemblages, hence it indicates there was no significant change. Plate 6.1 below shows what was probably a tall necked vessel from Khami ruins. 137 Table 6.9 C1 Summary of vessel types: Khami Hill Complex C1a 32 C2 16 C2a 45 C2b 44 C3 60 C3a 4 C3b 2 C3c 32 C3d 11 C3e 15 C4 26 C5 0 C5a C5b 3 0 C6 4 Total 1 295 10.85% 5.42% 15.25% 14.92% 20.34% 1.36% 0.68% 10.85% 3.73% 5.08% 8.81% 0% 1.02% 0% 1.36% 0.33% 100% Table 6.9.1 Summary of vessel types: Danamombe Kitchen midden trench Cont. C1 C1a C2 C2a C2b C3 C3a C3b C3c C3d C3e C4 C5 C5a C5bC6 Total Freq.% Surface 1 1 3 2 2 9 2.88 Level 1 0 0 Level 2 1 2 10 9 1 2 25 8.01 Level 3 3 13 6 14 2 6 44 14.1 Level 4 2 11 3 12 2 30 9.62 Level 5 2 2 4 1.28 Level 6 4 8 11 1 6 1 1 32 10.26 Level 7 4 10 10 1 1 26 8.33 Level 8 2 5 6 4 1 5 2 4 29 9.29 Level 9 7 17 21 16 4 2 67 21.47 Level 10 0 0 Level 11 3 3 5 5 16 5.13 Level 12 4 2 7 6 19 6.1 Level 13 3 4 4 11 3.53 Total 0 0 30 0 0 36 6 96 7 85 10 20 20 1 0 1 312 Freq. % 0 0 9.62 0 0 11.54 1.92 30.77 2.24 27.24 3.21 6.41 6.41 0.32 0 0.32 100% Key: Cont.= Context 138 35.00% 30.00% 25.00% 20.00% Khami 15.00% Danamombe 10.00% 5.00% C1 C1a C2 C2a C2b C3 C3a C3b C3c C3d C3e C4 C5 C5a C5b C6 0.00% Fig. 6vi: Graph showing ceramic classes at Khami and Danamombe. Plate 6.1: Reconstructed globular pot, Khami site museum (Photographed by the Author, June 2011). Plate 6.2: Constricted polychrome pot Danamombe site museum (Photographed by the Author, June 2011). 139 Plate 6.3: Polychrome vessel at Danamombe site museum (Photographed by the Author, August 2011). The major observations in terms of ceramic results are that some slight changes occurred in the Danamombe assemblage. Several pot forms found at Khami were phased-out and there was a reduction in the manufacture of short vertical and convex necked pots. On the other hand, Danamombe experienced an increase in the manufacture of large globular pots either with very short necks or non at all. The data also shows that Danamombe experienced a marked reduction in the variety of lipforms that were observed from the Khami assemblage. There is a distinct increase in externally thickened and squared lipforms in the Danamombe assemblage. Thus the majority of these pots at Danamombe had externally rolled rims, and very thick sherds made with medium fabric while their external surfaces were commonly burnished or polished. Only a limited number of Danamombe ceramics were lavishly decorated, in fact polychrome band and panel designs were extremely rare at the site. Tentatively these stylistic changes can either be attributed to the potter‟s design preferences or to the influence of ethnic elites seeking to further their socio-political goals. 140 6.3 Bead analysis and results Beads were also analysed for purposes of tracing differences or changes that can be attributed to Rozvi ethnic expression. However beads are largely a reflection of market forces during particular historic times, hence their implications in defining ethnic identities are very subjective. In this case, glass beads were first analysed according to colour in order to assess colour preferences for the two assemblages (For results see Tables 6.9.2 and 6.9.3 below). The most dominant colour is blue and it registered 47.6% at Danamombe and 31.2% at Khami (see Fig. 6vii below). This is closely followed by green which records 30% at Khami and 17.5% at Danamombe. Black beads are fairly common at Khami as shown by the frequency of 25.4% but that figure drops significantly at Danamombe where a single bead was recorded. Another interesting development is that the frequency of red beads increases at Danamombe to 16.3% from a mere 3.4% recorded at Khami. Only a single pink bead was found at Danamombe in level 11 and such was not found at Khami. The rest of the colours like white, Indian red on white and blue on white account for less that 5% at both sites thereby suggesting their limited occurrence and distribution. The analysis of bead colours does not quite reflect significant differences between the two sites, hence it is difficult to explain such a trend in terms of ethnic expression. Table 6.9.2 Blue Summary of glass bead colours: Khami Hill Complex Red Yellow Green White Black Pink I/B/W. Total 336 37 76 323 13 273 1 18 1077 31.19% 3.44% 7.06% 29.99% 1.21% 25.35% 0.09% 1.67% 100% Key: I/B/W.= Indian red on white and blue on white 141 Table 6.9.3 Summary of glass bead colours: Danamombe Kitchen midden Cont. Blue Red Yellow Green Black Pink White White? I/B/W. Total Level 1 6 6 2 2 16 Level 2 0 Level 3 3 1 1 5 Level 4 11 2 3 16 Level 5 9 8 3 20 Level 6 9 5 3 17 Level 7 7 2 3 1 13 Level 8 0 Level 9 7 2 1 4 1 15 Level 10 8 3 2 1 14 Level 11 3 3 6 8 1 1 4 26 Level 12 16 4 3 1 24 Level 13 0 Total 79 27 14 29 1 1 3 7 5 166 Freq. 47.59% 16.27% 8.43% 17.47% 0.60% 0.60% 1.81% 4.22% 3.01% 100% Key: Cont.= Context White?= White bead that has lost its patination Freq.= Frequency I/B/W.= Indian red on white and blue on white 50.00% 45.00% 40.00% 35.00% 30.00% 25.00% 20.00% 15.00% 10.00% 5.00% 0.00% Khami Danamombe Fig. 6vii: Graph showing bead colour frequencies from Khami and Danamombe. In the final analysis, the various bead attributes were combined into classes following procedures defined in Chapter 4. This procedure was applied for glass, metal and shell beads in order to identify and bring out more clearly spatial and temporal changes in the occurrence of the beads. 142 As already noted above, material culture stylistic differences are often associated with human agency and ethnic influence. Thus the two assemblages were classified and they produced a total of 7 classes with 15 sub-classes that were described as follows: Class 1. Drawn glass bead; simple small oblate (blue, green, yellow, black, white, pink). C1a. Drawn glass bead; compound small oblate (blue on white, Indian red on white, red, green, white). Class 2. Drawn glass bead; simple small barrel (blue, red, green, yellow, white). C2a. Drawn glass bead; compound small barrel (blue, yellow, red). C2b. Drawn glass bead; simple small cylinder (black, white, red, green, blue). Class 3. Wound glass bead; simple large sphere (green). C3a. Wound glass bead; compound large barrel (blue, red). Class 4. Molded glass bead; simple large hexagon (blue). Class 5. Folded copper bead; small oblate. C5a. Folded copper bead; medium oblate. C5b. Folded copper bead; large oblate. C5c. Folded copper bead; small barrel. C5d. Folded copper bead; medium barrel. C5e. Folded copper bead; large barrel. C5f. Folded copper bead; small cylinder. C5g. Folded copper bead; medium cylinder. C5h. Folded copper bead, large cylinder. C5i. Folded copper bead; small ring. Class 6. Chopped bronze bead; small barrel. Class 7. Ground shell bead; small disc. C7a. Ground shell bead; medium disc. C7b. Ground shell bead; large disc. *NB C= Class, while sub-types for each bead class are represented by letters a, b, c, etc. Bead classes were specially defined to establish the variation of bead styles at the two research sites. It is quite clear that class 1a with a frequency of 65.7% is the most dominant and it only occurs at Khami, the trend remains the same for class 1b, although the frequency is much lower 143 at 3.2%. This shows that drawn oblates were absent or extremely rare at Danamombe, which appears to have had a preference for drawn barrels constituting 17.2% of the total sample. In contrast Khami only registered 1.2% for class 2a glass beads. All the other glass bead classes were rare at both sites indicating they were either not readily available on the market, or their distribution was carefully controlled. Copper beads were less common for both sites with the highest frequencies being 6.2% for Class 5c at Khami and 5.1% for Class 5g at Danamombe. It is however important to note that the occurrence of copper bead classes was more frequent at Danamombe than at Khami. Shell beads remained fairly common at both sites especially the smaller ones in class 7a. A prominent difference between the two assemblages is that medium sized beads were restricted to Danamombe and these only appear in level 6. Large shell beads featured at both sites but they were more prevalent at Danamombe, where they almost occur in all levels. The frequency of class 7c was however decreasing through time as shown on Table 6.9.5. Basing on the results shown on Fig. 6 viii, one can arguably suggest Danamombe had more bead varieties than Khami and this stylistic development could possibly inform us about the role of material culture in ethnic expression. 144 Table 6.9.4 C1 % Freq. Summary of beads classes: Khami Hill Complex C1a C2 C2a C2b C3 C3a C4 C5 C5a C5b C5c C5d C5e C5f C5g 5h C5i C6 C7 C7a C7b Total 960 47 18 0 4 4 1 2 16 0 90 0 2 1 0 0 0 6 1 281 0 29 1462 65.66 3.21 1.23 0 0.27 0.3 0.1 0.14 1.09 0 6.2 0 0.14 0.07 0 0 0 0.41 0.07 19.2 0 1.98 100 Key: C = Class Table 6.9.5 Summary of beads classes: Danamombe Kitchen midden trench Cont. C1 C1a C2 C2a C2b C3 C3a C4 C5 C5a C5b C5c C5d C5e C5f C5g C5h C5i C6 Lv. 1 9 6 1 7 2 Lv. 2 3 Lv. 3 4 1 6 Lv. 4 16 16 2 Lv. 5 19 2 10 Lv. 6 16 1 9 Lv. 7 13 19 Lv. 8 Lv. 9 14 1 Lv. 10 14 6 Lv. 11 20 6 6 Lv. 12 24 4 Total 0 0 149 9 8 0 0 0 0 1 0 23 2 0 44 3 16 2 % Freq. 0 0 17.2 1.04 0.9 0 0 0 0 0.12 0 2.66 0.23 0 5.1 0.35 1.85 0.23 Key: C = Class Cont.= Context 145 C7 C7a C7b Total 17 15 48 81 123 18 77 56 63 90 0 108 0 12.47 21 123 378 14.2 43.65 42 3 26 82 112 149 109 74 15 83 32 139 866 100 70 60 50 40 Khami 30 Danamombe 20 10 0 C1 C2 C2b C3a C5 C5b C5d C5f C5h C6 C7a Fig 6viii: Graph showing frequencies for bead classes at Khami and Danamombe. Beads were basically chosen because they constitute a significant category of material culture that can be manipulated by agents to reinforce ethnicity. Apart from their secondary function of decorating ceramics for instance Musengezi wares (see Pikirayi 1993), beads were acquired or manufactured for ornamentation purposes. Today beads are no longer simply viewed as part of fashion preferences but they are perceived as symbols of wealth, power, status and ethnicity (see Bvocho 2005; Huffman 2007; Musesengwe 2009). Wood (2005) argued that in the past beads were controlled and strategically redistributed by the elites in efforts to create and secure alliances as well as to increase prestige. Hegmon (1992) also argued that carefully controlled goods were often imbued with ideological meanings by elites in the past. The results of bead analysis presented above show that there was a change from drawn oblates at Khami to drawn barrels and cylinders at Danamombe. In addition, Danamombe had a wider variety of folded copper beads than Khami. It also appears that medium and large shell beads were more common at Danamombe than they were at Khami. One can argue that Danamombe enjoyed a wider variety of bead types and sub-types than Khami. Thus it is possible to interpret these changes in terms of ethnicity. This assumption follows the view that ethnicity is often expressed through differences in fashion styles, hence many bead varieties could be examined as indications of the existence of such social processes. In this case bead stylistic differences are manifesting more at Danamombe than Khami, thus such a trend can be interpreted as evidence of 146 intentions to mark social differences by ethnic agents. However, other factors like trade patterns should be considered as well when interpreting such material evidence. 6.4 Dry-stone wall analysis and results Dry-stone walls were examined following procedures presented in Chapter 4. In particular, drystone walls at the centre of the two research sites were selected because they were the most stable, decorated, accessible, and they are widely associated with the elite who constituted the centre of ideological constructions. Khami is the second largest monument after Great Zimbabwe and its architectural styles are perceived as a combination of Leopard‟s Kopje and Great Zimbabwe building modes (Pikirayi 2001). Generally the ruins are scattered within a radius of 34 km of the core area that is lined-up along the western bank of the Khami river and dam (Hughes 1997). A few other ruins are located on the eastern bank, hence Khami has a total of 14 ruins that reflect very close architectural affinities. Robinson (1959) and many other scholars associate the central part with elite occupation and the peripheral areas with low status people. The Hill Complex forming the core area is the most prominent structure and it is bound by both poorly coursed and neatly coursed retaining walls with elaborate decorations, free-standing walls are quite rare. It has 3 main platforms that appear like massive terraces and these accommodate remains of huge huts on their edges. In addition it has two entrances of parallel passages that reach the hill summit from the south-eastern and north-western sections respectively. Approximately 25yards (22.86m) north-west of the Hill complex lies the Cross ruin which was also chosen for study. It is roughly circular in plan and it has neat retaining walls covering a diameter of 80ft (24.38m) and a height of 30ft (9.14m), (Robinson 1959). The third and last platform that is examined in this study is the Passage ruin which lies approximately 500m south of the Hill Complex. This is also circular in form and it rises to a height of 8-10ft (2.43-3.04m), its central part is bisected by a parallel passage. This ruin comprises of both free-standing and retaining walls, however decorations, and massive hut remains are limited in occurrence. As already highlighted in Chapter 4, for comparative purposes, the other dry-stone walls were selected from Danamombe the core Rozvi settlement. Danamombe is much smaller than Khami in terms of site size, however it was constructed on a low granite outcrop of about 1.490m and its dry-stone structures cover an estimated area of 700 hectares. The Vanguard and Main platforms define the northern sections and core area of the site respectively. The Main platform is the most 147 prominent structure that has a few remains of mud huts, a single northern entrance and passage that reaches the summit in an almost straight manner. Consequently, this passage divides the Main platform into two, namely the eastern and western sub-rectangular sections. While the eastern platform covers an area of 2.8 square meters with a height of 3m, the western platform only covers 900 square meters and its wall height ranges around 6m (Pikirayi 2001). The eastern and western faces on the Main platform have three tiered stone walls that are characterised by some impressive decorations. The Vanguard and Main platforms are also flanked on their eastern, southern and western sides by numerous small enclosures with undressed and poorly coursed walls. Only two enclosures of such type were chosen for purposes of this study. The enclosure located immediately north-east of the Main platform and south-east of the Vanguard platform, as well as the very wide enclosure (courtyard) on the western side of the Main platform were sampled for detailed analysis. Various attributes of dry-stone walls at Khami and Danamombe were recorded and analysed for purposes of determining whether there were any changes in the building styles. As shown on Table 6.9.6 and Fig. 6ix, very short walls below 50cm were exclusive to Khami though their frequency was also very low recording just 5.68% of the total sample. Apart from that there was general consistency in the height of walls at both sites indicating there was general continuity in the building traditions. Table 6.9.6 Summary of wall height: Khami and Danamombe Site name Dry-stone wall height categories Total 0-0.49m 0.50-0.99m 1-1.49m 1.50-1.99m 2.00m+ Khami 10 77 52 27 10 176 Frequency 5.68% 43.75% 29.55% 15.34% 5.68% 100% Danamombe 0 27 34 19 2 82 Frequency 0% 32.93% 41.46% 23.17% 2.44% 100% 148 50.00% 45.00% 40.00% 35.00% 30.00% 25.00% 20.00% 15.00% 10.00% 5.00% 0.00% Khami Danamombe Fig. 6ix: Graph showing wall height frequencies at the research sites. Table 6.9.7 below also shows the distribution of wall decorations at the research sites. Owing to massive collapses of some sections and limitations of sampling procedures, wall decoration could not be quantified. In order to avert the problem of over simplification, decorations identified at each site were simply recorded in terms of presence or absence and captured in the recording form by the (x) symbol on Table 6.9.7. While Danamombe had six of the seven decoration motifs, Khami only had three. These findings are consistent with observations from more extensive studies by Garlake (1970) and Robinson (1959:11). Going by such findings one can tentatively suggest that Danamombe had more decoration motifs than Khami. Perhaps this happened because Rozvi community members sought to actively communicate different messages related to ethnicity, power, power relations, status, gender among other individual and group identities. In order to appreciate the development of wall decoration motifs, we should recognise that herringbone represents the more complex type of cord, either in continuous or discontinuous horizontal lines or panels. As shall be noted in the next chapter, slender spatiotemporal changes may also be subjectively attributed to the role of agents seeking to reinforce their ethnic identities. In order to get a more focused understanding of wall types at the two sites, architectural attributes such as form, construction technique and decoration motif were combined. This classification was adapted from earlier classifications by Summers et al. (1961) and Whitty (1961). Therefore the classification model produced the following types: 149 A. PQ style: Retaining walls with dressed blocks, partial to neat coursing and battered back profiles; plain. B. Q style: Retaining walls with dressed blocks, neat coursing and battered back profiles; decorated. C. R style: Retaining walls without dressed blocks, neat coursing and decorations. D. PQ style: Free-standing walls with battered back profiles, partial to neat courses of dressed blocks; plain. E. Q style: Free-standing walls with battered back profiles, partial to neat courses of dressed blocks; decorated. F. R style: Free-standing walls with neither neat courses nor dressed blocks; could be battered back plain. The distribution and frequency of these wall types at each site is shown on Table 6.9.8 and Fig. 6x below: 150 Table 6.9.7 Decoration Summary of dry-stone wall decorations at Khami and Danamombe Khami Danamombe x x x x x x x x x Key: Ic/n-g.= Isolated courses of non-granite stone forming bands of a different colour. Srp/h.inf.= Small rectangular panels with a herringbone infill. x= Denotes decorations identified during fieldwork for this particular study. 151 Table 6.9.8 Summary of dry-stone wall types at Khami and Danamombe Type Form Construction Technique Deco. Khami Danamombe Retain. Free-st. D/Blo. N/Co. B/Bck. Total Frequency Total Frequency x x x x A x x x 114 60% 20 24.69% x x x x x x x x B x x x x 28 14.74% 8 9.88% x x x x x C x x x 19 10% 7 8.64% x x x x x x D x x x 17 8.95% 19 23.46% x x x x x x x x E x x x x 10 5.26% 9 11.11% x x x x x F x x 2 1.05% 18 22.22% x TOTAL 190 100% 81 100% Key: Retain.= Retaining wall Free-st.= Free standing wall D/Blo.= Dressed blocks N/Co.= Neat coursing B/Bck.= Battered Back Deco.= Decorated 70% 60% 50% 40% Khami 30% Danamombe 20% 10% 0% A B C D E F Fig. 6x: Graph showing the frequencies of wall types at Khami and Danamombe. 152 It is clear that plain PQ retaining walls are dominant at Khami as they constitute 60% while Danamombe just had 24.69%. Although Khami continues to dominate Danamombe in terms of retaining wall types B and C, the margins of difference are narrow thus suggesting there was general continuity for those particular walls. A remarkable difference between the two sites is that types D and E are more frequent at Danamombe than at Khami. This data testifies that freestanding walls were common at Danamombe than at Khami and these were more profusely decorated. Finally R style free-standing walls were less frequent at both sites but Danamombe somewhat recorded more of these. Basing on these observations it becomes very difficult to simply accept earlier notions that Khami and Danamombe possess very similar material culture that does not show important changes. Certainly there were negligible spatial and temporal variations between the two sites that can be accounted for by using diverse perspectives, one of which being ethnicity. Plate 6.4: Pecked stones, Khami site museum (Photograph by the Author, June 2011). The researcher also took interest in some of the pecked or scratched stones (see Plate 6.4 above) in the Khami site museum because they bear symbolic designs also found on pots and dry-stone walls. Robinson (1959:11) observed that most of these stones were recovered from the debris at the Khami hill ruin 2, hence they probably formed part of the dry-stone wall building material. He suggested that these stones were contemporaneous with wall building and that they could have had ritual or other meaning. While conducting dry-stone wall surveys at Khami, the researcher also came across these isolated stones on wall courses on the Hill Complex. Huffman (1996:31) also found isolated pecked stones with rows of dots, circle and dots, as well as lines 153 with circle and dots at Danamombe. Huffman (1996:29) suggested that these designs symbolised snakes (pythons) and crocodiles associated with pre-19th century sacred leaders for the Venda, Shona, and Rozvi. Although we may never know with absolute certainty what these symbols meant or represented to the associated communities, we know they constituted complex codes communicating social or political messages. While these symbols remained the same, meanings ascribed to them by various people could have varied from gender, ethnicity, class, status, age and myriad other socio-political identities. In that case they become relevant for purposes of this study. 6.5 Dhaka structure description and analysis. Traditional house remains were also examined because they were built time and again, and they reflect architectural styles that can be correlated to group identities (Huffman 2007). The researcher did not conduct any mathematical analysis of dhaka structures because the data was largely acquired from secondary archaeological sources. Scholars like Caton-Thompson (1931), Chakanyuka (2001), Dhliwayo (1989), Hughes (1997), Muringaniza and Ruwitah (1996), Ndoro (2001), Robinson (1959), Rudd (1984) and Wieschhoff (1941) have written about dhaka structures at Zimbabwe tradition sites and their observations were exploited for purposes of this study. The analysis revealed that the methods of construction were relatively consistent at both sites as dhaka, wooden poles, and sometimes dressed granite stones were used as building materials (Caton-Thompson 1931; Chakanyuka 2001; Dhliwayo 1989). Robinson (1959) envisaged that massive circular undecorated huts (ranging between 2.5-9m in diameter) were built for chiefs on the edges of terraced platforms (see Plate 6.5 below). Most hut foundations were built on kerbs while their walls were very thick (approximately 30-50cm). For this reason they were rarely reinforced by spaced wooden poles, which were commonly part of the associated dhaka radiating walls. Robinson (1959:23) further explained that huts often built with a framework of light sticks that were smeared with mud both on the interior and exterior sides were commonly associated with commoners. Dhliwayo (1989:41) also postulated that huts with cross-walls (Figs 6o and 6p) were rare at both Khami and Danamombe, as each site only recorded a single hut. On the whole, circular huge huts either without or with outside radiating dhaka walls abutting to stone walls (see Fig. 6q below) were reasonably frequent at both sites. 154 Consequently one may tentatively suggest that dhaka remains from the two sites did not experience any noticeable architectural changes. Perhaps changes between the two sites could have been distinguished in terms of roofing techniques if only these structures were less prone to agents of deterioration. Plate 6.5: Remains of a massive dhaka structure at the summit of Danamombe‟s Main platform (Photograph by the Author, August 2011). Figure 6o An artistic impression by Posselt of a prehistoric dhaka hut with a cross-wall and internal features, this was drawn following excavations at Great Zimbabwe (After Ndoro 2001). 155 Figure 6p Plan of Khami ruins Hut Cb3 with both a cross-wall and outside dhaka radial walls (After Robinson 1959). Figure 6q Dhaka hut from Khami with outside radial walls but no internal cross-wall (After Robinson 1959). 156 6.6 Description of miscellaneous finds. Throughout the centuries materials considered precious, beautiful and rare have contributed to the power base of leaders (Huffman 2007; Pwiti 1996b). Bvocho (2005:410) states that in Zimbabwe ornaments like brass and copper wire for armlets and leg rings were worn as symbols of wealth and importance in the community. Robinson (1959:113) also points out that it is very difficult to find prized objects in the archaeological record because they were hardly abandoned or thrown away. Accordingly, various local and exotic dear finds recovered from both Khami and Danamombe were comparatively examined in order to make inferences on Rozvi ethnicity. Basing on Robinson (1959), outstanding finds from Khami included fragments of blue and white Nankin China, 6 fragments of very course glazed ware decorated in blue and red colour. He also found a metal sheeting decorated with a scroll design, a few iron, bronze and gold bangles, a gold pellet, a copper finger ring with fragments of gold, iron and bronze spear-heads, and a crescent shaped axe-head having an inlaid copper disc in the centre (Robinson 1959). Conversely, Danamombe yielded raw gold, a Portuguese silver chalice, Dutch gin bottles, a ring, bell, medallion, copper or brass bangle, flint-lock, gun parts, two cannons, plenty of gold beads, wire tacks, chain-work, and Nankin China ware (see Muringaniza and Ruwitah 1996; Pikirayi 2001; Sibanda 1999). Scholars like Beach (1974) and Mudenge (1974a) mentioned several other precious goods that Rozvi chiefs had access to through their tight control of trade and the tributary system. These marked differences in the distribution of costly goods at the two sites signify that the Rozvi had more material symbols to exploit as sources of wealth, status, power and ethnicity. Therefore the greater variety of precious finds that were confined to the site of Danamombe can be interpreted as part of Rozvi material symbols. 6.7 Conclusion Even though results from the data presented above show patterns of continuity and change in material culture, the ethnic implications are quite complex. Be that as it may, minor changes in material culture and prestigious rare finds at Danamombe represent the potential symbols of Rozvi ethnicity. Whilst Khami shows diversity in terms of ceramic types and their decorations, Danamombe experienced a marked reduction in ceramic stylistic varieties. The analysis of beads however revealed a different picture indicating that Danamombe had a greater variety and frequency of beads than Khami. In terms of glass bead colours, black was quite common at 157 Khami but only a single one was found at Danamombe. Red beads however appear more frequently at Danamombe than at Khami. Generally varieties of copper and shell beads were fewer at Khami than at Danamombe, whether this was ethnic related or otherwise remains open to discussion. Dry-stone walls also show some minor but interesting stylistic improvements at Danamombe, however causes for such changes could be ethnic related or otherwise. Of all the data sets examined, dhaka structures reflected the most consistent picture of general continuity in style at both sites. Undoubtedly, Danamombe was much richer in terms of prestigious finds than Khami. The next and final chapter synthesises historical, ethnographic and archaeological data in order to define Rozvi ethnicity and its material correlates. 158 Chapter Seven Discussion and Conclusions: Rozvi identities in the past and present “... the richest results from fieldwork stem from the integration of oral, written, and iconographic materials and the testimony from places and objects into a single rich, multifaceted reconstruction that cannot be achieved by using any of these sources on its own” (Vansina 1996:136). 7.1 Introduction In this chapter, Rozvi ethnic identities are explored and defined in order to carefully distinguish and interpret their material culture from that of the Torwa. Only tentative remarks are made about the Torwa because not much is known about them both historically and archaeologically. It has already been highlighted in this work that ideas (ideologies and discourses) are more important than objects (see also Cohen n.d.; Paynter 2006; Schmidt 1983). Therefore it is critical to first establish the various ideas that characterised the Rozvi before attempting to define their possible implications on material culture. Results from documentary and ethnographic sources revealed that the Rozvi also imbued ethnic meanings to various objects including natural features like mountains and trees. Thus, ideas surrounding Rozvi ethnicity were not entirely fixed on archaeological material alone. They were sometimes pinned on objects that hardly feature in the archaeological record that we often so much rely on for ethnic answers. It is also crucial to note that at times archaeological data does not show distinct stylistic changes that can be confidently attributed to ethnic influence. Therefore defining ethnicity through material culture is a very complex and subjective procedure that will not yield absolute or empirical conclusions. For purposes of clarity, theoretical concepts relating to agency, structuration, and interaction are systematically applied in defining Rozvi archaeological signatures. In this way, historical, ethnographic and archaeological data are synthesised in a manner that exposes Rozvi identity construction processes, power relations, and material culture symbolism. The dissertation concludes by highlighting challenges and prospects of studying power and identities such as ethnicity in archaeology. 159 7.2 Contextualising Torwa and Rozvi ethnic identities In Chapter 1, I highlighted that the poor or unsystematic definition of the Torwa and Rozvi has seriously affected any meaningful archaeological attempts on such subjects. In fact, it is difficult or almost impossible to define the archaeology of the Torwa or Rozvi if their identities are not well conceptualised historically. Thus questions raised earlier on about the identity of the Torwa and Rozvi are revisited and clarified in this section. The Torwa are more difficult to define because, to date there is incomplete historical detail about their past. Nonetheless, for purposes of clarity a few remarks are made in connection with their past. Although Torwa history remains obscure, documentary data in Chapter 5 has shown that there are about three different perspectives regarding their name or identity. In one of the versions, it was mentioned that a certain Changamire of the 1490s rebelled from the Mutapa state to rule briefly in the southern Mbire and Guruuswa provinces. After a period of four years he was subdued and killed by the Mutapa but his son and his Torwa relative escaped and continued fighting until 1547-8 (see Beach n.d., 1980, 1983a, 1995; Mudenge 1974a; Garlake 1973). Following their defeat, nothing was herd about them until the 1644 civil war. This version implies that the Torwa identity could have developed from the name of a certain individual who rebelled from the Mutapa during the late 15th century, but it is difficult to determine exactly how this identity transformation occurred. In this regard, it is crucial to review the other perspectives concerning Torwa identities. Beach (1980), Pikirayi (2001:200) and Randles (1979) share the view that, Torwa probably meant rebels or outsiders (vatorwa) to the Mutapa state who were located in the Guruuswa/south-western periphery. Randles (1979) further suggests that the Portuguese referred to south-western Zimbabwe as Butua, and in Swahili “tua” means “disgrace”, hence “Bu-tua” had such connotations. Randles (1979) also notes that the people that lived in Butua or Vatwa were culturally closer to the Khoi and San, who are now scattered in parts of Botswana and South Africa. It should however be emphasised that the identity of the people in question was not imagined, rather their difference from the Karanga could have resulted from mere geographical isolation. Therefore it is assumed that the Karanga in the north degraded people in Butua as strangers and barbarians because of their cultural differences (Randles 1979:20). However it is widely believed that the Torwa succeeded the Leopard's Kopje culture that was associated with an ancient Kalanga dialect (see Beach 1980; Huffman 1974; Mazarire 2009). Interestingly, 160 Ncube (2004) omitted the Torwa from his list of people who inhabited the south-west, only citing the Kalanga, Nambya, Leya and Tonga. Robinson (1959) never mentioned the Torwa when he suggested that pots at Khami were made by the Humbe and Lilima women who came from beyond Plumtree. It appears Robinson was trying to ascribe the pots to people of a Sotho descent though he generally credited the Leopard‟s Kopje cultures and Rozvi people for the manufacture of ceramic wares at Khami. Generally historians believe that the Torwa disappeared due to the arrival of the Rozvi who assimilated the entire community (see Beach 1980, 1995). Perhaps the Torwa were easily forgotten in history because they were a creation of historical processes and they simply failed to negotiate their identity in changing socio-political circumstances. Many historians and archaeologists regard the Torwa as the builders of Khami-phase sites in south-western Zimbabwe (see Beach n.d., 1980, 1983a, 1983b, 1994a, 1994b, 1995; Ncube 2004; van Waarden 1998). On the other end, data from oral traditions merely highlight the names of Torwa rulers such as Chihunduru/Chiwundura, Tumbare, and Ndumba (Beach n.d.; Mazarire 2009). These views neither aid a clear appreciation of Torwa identities nor their material culture. Unfortunately, this is the perspective that has been very popular among archaeologists because it allows them to directly equate material culture at Khami-phase sites to the Torwa. This kind of approach is extremely limiting because it oversimplifies past communities by treating them as closed and fixed entities. Whatever the case, it has emerged from this study that the name Torwa had several meanings that varied through space and time. It is however important to note that without adequate historical information, we can only add more speculative views on the subject. Nonetheless, the biggest question remaining is, “How do we distinguish Rozvi material culture from that of the Torwa?” On that note, I now turn on to the problem of defining Rozvi identity construction processes, their sources of power and how their identities influenced power relations. A lot of confusion was created following the projection of numerous and conflicting views about the Rozvi over a very long period of time (e.g. Abraham 1959; Axelson 1960; Beach 1976, 1983a, 1983b, 1994a, 1994b, 1995; Bullock 1927; Hall 1909; Mudenge 1974a, 1974b, 1988; Posselt 1935; Randles 1979; Robinson 1959, 1966). The analysis of oral traditions, Portuguese documents, ethnographic and historical sources confirmed that opinion has varied widely 161 regarding the identity of people mentioned in history and tradition as Rozvi. It has taken far too long for scholars to realise that the Rozvi were a creation of dynamic historical events and processes. Data extracted from the earliest Portuguese sources indicates that Rozvi origins can be traced back to the late 17th century military campaigns that were propagated by a certain Changamire and his warriors. It was through these successful military campaigns that Changamire gained a lot of respect and power among local communities on the Zimbabwean plateau. By the mid-18th century the name Rozvi, which never existed before was in use and it referred to Changamire and all his immediate followers. Until today, oral traditions recall the name of that Changamire as Dombolakonachingwango (shortened to Dombo) of the north-east as the Rozvi founder. Changamire Dombo capitalised on historical conditions prevailing in north-eastern Zimbabwe during the late 17th century to mobilise people against Portuguese oppression and domination. As such, the notion of ethnicity comprehensively explains such Rozvi origins and rise to power. Ethno-historic data in Chapter 5 has shown that various elements of culture were situationally used by the Rozvi to distinguish themselves from their non-Rozvi neighbours. During the 19th century some senior Rozvi chiefs had access to a few guns in the form of small cannons (Mudenge 1974a:378). Oral and documentary evidence presented in Chapter 5 highlighted guns as part of Rozvi symbolic material. Guns depicted the power of Rozvi chiefs because they were exclusively owned by a few individuals who held positions of power. The relationship between ethnicity and power has already been discussed in Chapter 3, and it was argued that ethnicity is a political tool widely exploited by both the dominant and subordinate groups aiming to legitimise and further their interests. Stone (2003:35) also stressed that ethnicity is an organising structure that cannot exist in the absence of competition for power because it is always invoked in opposition to one or more groups. Hence ethnicity should therefore be conceptualised as an ideological concept that is constructed in specific historical contexts by rational agents. As noted in Chapter 3, resistance identity can be constructed through the actions of rational beings or agents (Mhiripiri 2008). Shennan (1989) argues that ethnicity comes from contexts of competition, and such contexts reinforce common interests in opposition to others. In such scenarios, agents carefully manipulate ethnicity as a resource for mass mobilisation through manufacturing discourses of resistance against oppressive forces. Mulligan (2003:431) stresses 162 that the notion of ideology is closely tied to creative agents with a political project. These two premises cited above are applicable to contexts that were prevailing in pre-colonial Zimbabwe when the Rozvi rose to power. In particular, equilibrium hierarchical power structures evolved due to Nyai formations and these created the best conditions for the generation of ethnic consciousness. Nyai clientele‟ power structures were volatile and they were the cause of political instability sparking from succession disputes among pre-colonial states (Mazarire 2009). Thus Changamire Dombo sprang to power around that time after terrorising the Portuguese, Mutapa and several other locals with his Nyai army (see Axelson 1960; Beach 1983a, 1995; Mazarire 2009; Mudenge 1988). It is likely that Changamire Dombo initially constructed an ethnic consciousness through a discourse of resistance to the Portuguese and all other local threats to his political and economic goals. Consequently, a series of successful military campaigns against these groups earned him and his followers the nickname “varozvi”, meaning destroyers. That name Rozvi was gradually accommodated, negotiated and transformed in order to control power relations (see Bhila 1982a:110). Oral traditions highlighted several accounts that link Rozvi origins, power, governance as well as their ultimate demise with the Mwari religion. In fact, these claims were ideologies of ritual practise deliberately created to promote common sentiments among the widely scattered Rozvi members. In this way, their military background was gradually mystified or shaded-off thereby allowing them to develop fully as a legitimate political entity. Documentary and field data also revealed that the Rozvi led in rain-making customs at Mabweadziva (Matobo Hills) and at Tsime raPfupajena in Bikita. Apart from these religious roles, ethnographic data also highlighted that the Rozvi were widely consulted in traditional dance performances, as well as in the appointment of new chiefs and issuing of land rights. Thus the Rozvi gradually redefined their military past by projecting a unique Rozvi religious identity through oral accounts, praise poetry and ritual functions in order to gain legitimacy. The Rozvi eventually gained a central position in society through their strategic manipulation of discourses centred on the Mwari religion (see Bourdillion 1976; Daneel 1970; Fortune 1956). The Mwari religion enabled the Rozvi to claim ownership of various landscapes, dry-stone structures and even diverse forms of material culture that sustained their socio-political interests. On this basis, it can be argued that the construction of Rozvi power closely relates to Foucauldian approaches which state that power and knowledge are interrelated (Mulligan 2003). Foucault (1980) further argues that discourse legitimises power through the 163 construction of „current truths‟ that control power relations. Therefore knowledge is the creator of power and creation of power. It has also emerged from the forgoing discussion that ethnographic data collected in Bikita relates quite well with earlier observations made about the Rozvi in documentary sources. In particular, the living Rozvi in Bikita distinguish themselves from their neighbours through their close control of the Mwari religion. Generally, religious ideologies reinforce political power as was the case for many pre-historic societies in Africa (see Huffman 1996; McIntosh 1999; Pikirayi 2001; Schoeman 2006). In southern Africa the ideology of “sacred leadership” which suggested that a king‟s power was based in part on his connection with the royal ancestors and God, greatly sustained the construction of identities around class distinctions in archaeology (Huffman 1996). Therefore religion is central to our understanding of ethnicity, and it is actually impossible to separate it from ethnicity (Yinger 1985). Preucel (2000) points out that ritual activity is good for integrating or consolidating people of diverse backgrounds. Furthermore, Yaeger (2000:126) postulates that a strong sense of horizontal comradeship is often cultivated through “practices of affiliation” that manifest through shared origins or belief systems. Rozvi elites thus centralised and consolidated their power by claiming that they were the children of Mwari (God), and they were ordained by him to rule others (see Beach n.d.; Bullock 1927; Daneel 1970). Henceforth, discourses and ideologies of the Mwari religion were quite pivotal in the construction and sustenance of Rozvi identities and power. Several traditional objects like a cotton gown with white on one side and black on the other, a reddish-like cotton belt, ceremonial wooden stick, wooden plate, drinking wooden container, ceremonial axe, snuff calabash, gourd cup, black beads among other cultural material were signified in Rozvi religious traditions. When Rozvi political power declined their religious influence remained strong in areas like Manyika. Even ethnographic data gathered from Bikita demonstrated that Rozvi knowledge of the Mwari religion still gives them an edge over their Duma counterparts. Thus through the Mwari religion, Rozvi elites managed to maintain a strong system of governance over other local chiefs. Basically the close connection between the spiritual world and political formations implied that people and land were fruitful through such dialectical relationships (Huffman 2006; Schoeman 2006). This general belief was exploited and manipulated by Rozvi elites to attain their personal political interests. The data in Chapter 5 also demonstrated that Rozvi ritual 164 ceremonies were often associated with the consumption of beer and food. Isbell (2000:256) explains that ritual and festive events provide perfect opportunities for consuming food and sharing experiences. She further elaborates that sharing of food is one of the basic human acts that enhances positive experiences and sentiments of group solidarity. This is why many ethnic identities are asserted in meals that require certain foods while prohibiting others (Isbell 2000). Thus during crocodile rituals in Manyika, a new Rozvi chief had to eat food cooked with pebbles from the stomach of a male crocodile (Huffman 1996; Bhila 1982a). Perhaps isolated pecked stones found at Danamombe symbolised complex religious connections between Rozvi kings, crocodiles and power. Therefore the Rozvi were a dynamic group that progressively evolved from a mere military formation, to a religious group that eventually developed into a powerful decentralised state system. Following oral and ethnographic data presented in Chapter 5, one can also argue that the Rozvi constructed and legitimised their ethnic identities through mythological versions. Since there was no community in history associated with the Rozvi identity prior to Changamire Dombo‟s rise to power, an imagined past had to be justified through chaining-out discourses of a common ancestry. Mythology promoted shared perceptions of common origins and belief systems to a dispersed Rozvi community. Precisely, mythical stories about Rozvi journeys led by Mwari, attempts to catch the moon for their king, and efforts to dig up mountains and move them to their royal capital were highlighted in oral traditions. These myths were ideological constructions aimed at legitimising an otherwise imagined Rozvi origins and past (see Abraham 1959; Beach 1976, 1983b; Bullock 1927; Hodza and Fortune 1979; Posselt 1935). In a way, these accounts glorified the Rozvi as very powerful and rich because they invested in massive and expensive projects. In addition, mythical accounts were crucial in portraying the Rozvi as people who were ambitious, brave and very proud of themselves. Therefore mythical versions were an important component of Rozvi identity construction processes and they greatly enhanced Rozvi control of power relations. The institution of marriage was manipulated by Rozvi elites in efforts to legitimise their ethnic identity and to negotiate power relations. Several chiefs appointed by the Rozvi were sometimes given young Rozvi girls as wives (see Beach n.d.; Bhila 1982a; Mutasa 1990, 1991). This strategy served two fundamental needs for the Rozvi system of governance; firstly it promoted 165 kinship ties by “blood”. The word blood is in quotes because ethnicity is a social category that is hardly defined by such biological attributes. In fact, the Rozvi manipulated mitupo (totems) using the moyo (heart) totem as the core referent to promote strong sentiments of kinship. This is why Loubser (1990) argues that totems should be seen as small scale ethnic units. Hence among the Rozvi, marriage ties and totems were objectified to justify a historically constructed identity. Secondly, inter-marriages between Rozvi women and distant chiefs naturally strengthened and legitimised Rozvi political influence in such areas (Bhila 1982a). A Rozvi wife was expected to act like a secret agent by constantly giving political feedback to the Rozvi metropolis (Mutasa 1991). Thus through the manipulation of traditional institutions such as marriage, the Rozvi administration gradually became rooted in moral force and not coercion. This system worked well for their decentralised state which was made up of a number of ordinary dynasties. In addition, discourses of severe punishment to rebellious subjects were cultivated in reinforcing social and political solidarity in the Rozvi confederacy. This data is crucial in the sense that we gain a much broader understanding of the Rozvi identity construction processes as well as the related ideological systems. Oral, ethnographic and documentary sources also pointed out that the Rozvi once spoke a certain dialect called “Rozvi”. However, it should be emphasised at this point that while the Rozvi dialect was an important attribute of identification in the past, it can never be inferred from the material record. The same holds true for multiple discourses about punishments that would be imposed on anyone who challenged Rozvi instructions. Bhila (1982a) pointed out that the rigour with which Changamire meted out his punishment deterred rebels among his subjects. If feudatories delayed tribute payment or kept silent about parts of royal game that were entitled to a Rozvi king, there would be serious repercussions. These discourses were exaggerating the attitude of Rozvi elites towards tribute defaulters and rebellious subjects in order to enhance Rozvi control of power relations. Thus discourses of punishment legitimised Rozvi sanctions against those who chose not to support Rozvi political and economic interests. Preucel (2000:64) also highlighted how the pan-Pueblo leaders exploited rhetoric of severe punishment to generate ethnic consciousness and legitimise sanctions against those who refused to accept ancestral ways. Rozvi praise poetry such as the one by Hodza and Fortune (1974:69) in Chapter 5 of this study is a good example of how Rozvi power was projected among communities. 166 Project identity which relates to the active manipulation of material culture by rational agents seeking to alter the social structure also befits the Rozvi. In the course of articulating various discourses and ideologies, different elements of material culture were carefully adopted and given new meanings by Rozvi ethnic agents. At least now we can characterise some of the objects that were historically objectified by the Rozvi in their various traditional customs with some certainty. Through the use of historical and ethnographic sources, I managed to highlight different material culture that was imbued with symbolic meaning for purposes of signifying Rozvi political power, investiture ceremonies for chiefs, and rain-making customs (see Beach 1983b; Bhila 1982a; Bullock 1927; Hodza and Fortune 1979; Huffman 1996; Ncube 2004). Interestingly, most of the signified features or material objects were commonly found or used in daily activities among traditional societies. Their values only changed when they were actively used in particular contexts to communicate ethnic values. While archaeologists concur that identities are played out in the world of objects, they differ remarkably in terms of approaches to determine the meaning of objects in the past. Cohen (n.d.:4) argues that aspects we see as important today might never have been meaningful to the creators and users of such objects. This scenario explains why Gamble (2001) argues that meanings of objects are embedded in context. Thus determining the meaning of an object requires a thorough examination of associated context by using a high density of data (Gamble 2001). On that note, I now shift attention towards inferring Rozvi ethnicity through material culture variation at Danamombe. 7.3 An agency oriented archaeological characterisation of Rozvi ethnicity This section tries to address the problem of distinguishing the Rozvi from the Torwa through material culture variation at Khami and Danamombe. The discussion critically examines results of stylistic variation in archaeological data already presented in Chapter 6. Archaeologists generally agree that individuals in the past were not uniform automatons but they acted rationally (through knowledge and reason) in their struggles to attain social, political and economic goals (see Dornan 2002; Stone 2003; Yaeger 2000). However these individuals did not necessarily exercise free-will or choices because some of their interests and actions were constrained by structure (institutions, traditions, and material culture). In order to overcome these various sociopolitical constraints, ethnicity was often situationally invoked or assumed by certain members in negotiating power and power relations (Emberling 1997; Stone 2003). As noted already in 167 Chapters 3, the agency approach which argues that there is a dialectical relationship between agency and material culture is valid in interpreting Rozvi archaeological indices. Agency is about the search for interests and motives of past actors in relation to ideology, power, identity, tradition and associated constraints for dynamic historic groups (Dornan 2002). Wells (1998:240) actually notes that material culture is not just a means by which people interact, but it is an active agent used to construct meanings and communicate ideas. Since ethnicity organises or structures groups on the basis of perceived boundaries or differences from the so-called others, slight stylistic changes in material culture could possibly denote ethnic symbolism. This notion follows the premise that style in material culture was an active component of power manipulation strategies by agents (Hegmon 1992:128). However stylistic variations in material culture should not be blindly equated to ethnic symbolism because changes in material culture can result from myriad processes. Ceramics are widely believed to be active agents in the expression of identities, and for this reason they were analysed to trace Rozvi ethnicity. The analysis of ceramics indicated that Khami and Danamombe assemblages were not homogenous but there were some stylistic variations at the latter site. While Khami had diverse vessel types, Danamombe experienced a marked reduction in ceramic diversity. Complex decorations observed at Khami became rare at Danamombe as the majority of vessels were simply burnished or polished. Potters at Danamombe began specialising in the manufacture of plain spherical pots with little or no necks, and constricted hemispherical bowls with square and rounded lipforms. Several reasons can be offered for these stylistic variations, but the most suitable interpretation is that Rozvi agents sought to mask intra-site social differences by promoting the manufacture of a simple ceramic style. In order to assimilate the Torwa community, material culture in the form of plain ceramics had the most potential to integrate the “new community” because profusely decorated ceramics could have otherwise emphasised internal social boundaries. The presence of other ceramic types at Danamombe which were similar to types found at Khami can be interpreted as the unintentional reproduction of structure. However ethnographic data collected in Bikita indicated that the Rozvi do not use pottery to mark social boundaries. Basically ceramics in the Bikita ethnographic context served basic functions, ordinary pots only acquired symbolic values in particular contexts such as during 168 times of Rozvi ritual or customary activities. Therefore archaeological ceramic changes noted above should not be treated as empirical evidence of Rozvi ethnicity. Theoretically, the role of private or domestic material culture such as ceramics in the expression of ethnicity has been widely debated. Stone (2003) states that portable objects like ceramics can be used to situationally express ethnic identities because they can either be hidden or brought out in varying ethnic interactions. Bowser (2000) also argued that women could signify their political alliances through the active use of ceramic style in the domestic context. Emberling (1997:311) stressed that although pottery style may indicate ethnicity, it does not always constitute significant social difference between groups. In southern Africa, studies by Huffman (2007), Pikirayi (1997b, 2007), and Ndoro (1996) have demonstrated that ceramic style carries symbolic massages relating to fertility issues, ethnicity and other group identities. Therefore from a different perspective, the minor stylistic changes noted at Danamombe could be explained through the agency theory. Changes in material culture either result from intentional or unintentional actions by agents seeking to resist power by others (Dornan 2002). Thus the dominance of spherical constricted pots with little or no necks associated with (classes 3 to 3e) at Danamombe could have resulted from Torwa resistance of Rozvi domination. It has been argued that objects with some degree of decorative elaboration termed “complex artefacts” represent “networks of significance” because they communicate information about the identity of the people who made and used them (Wells 1998:240). Perhaps polychrome classes 3a, 3c, 3e were manipulated in such a manner at inter-site level by Rozvi elites because of their rarity at Danamombe. This development could possibly account for the wide distribution of such ceramic styles on the Zimbabwean plateau (see Mupira 2007; Pwiti 1996a; Robinson 1966). Beads constituted another important category of archaeological data that was used to trace Rozvi ethnicity. The results of bead analysis indicated that Khami had a prevalence of drawn oblates, which were replaced by drawn barrels and cylinders at Danamombe. Apart from that, Danamombe also enjoyed a wider variety of copper beads when compared to Khami. More so medium and large shell beads were more common at Danamombe than at Khami. Generally, in terms of bead types and sub-types Danamombe had a wider variety than Khami. Bead differences between the Khami and Danamombe assemblages are quite apparent but their implications for Rozvi ethnic identities should be cautiously approached. Ethnographic data cited 169 by Bullock (1927) hinted that vaNyai (Rozvi messengers to Matobo Hills) wore black beads as part of their religious regalia. Gelfand (1959) also mentioned that Rozvi religious leaders used large red beads as part of their ritual material culture. This data at least informs us that the Rozvi signified black and red beads in their religious customs. Although the material used to make such beads was not clarified, the fact that their colours were emphasised is crucial to this study. Archaeologically, only a single black bead was found at Danamombe while about 273 were analysed from Khami. Robinson (1959) actually highlighted that common glass bead colours at Khami were black, Indian red, blue, green, and white. The occurrence of red beads at the two sites also shows another distinct difference because Danamombe witnessed a marked increase in red beads than Khami (see Fig. 6viii). Thus these minor stylistic variations in archaeological data seem to confirm views derived from ethno-historic sources that black and red beads were situationally objectified in Rozvi religious traditions. Pikirayi (2003:149) states that beads of several colours were brought to Africa through the Indian Ocean Zone, but Indian Reds were very common among Africans during the 16-17th centuries. Bhila (1982a) also noted that while the popularity of black, yellow, and green-blue beads varied from one region to the other, red and black beads were very popular among Africans. In particular, smaller beads were in greater demand and they were highly priced (Pikirayi 2003). It is however important to note that beads do not carry functional but symbolic significance, so they were commonly acquired and used to signal power, wealth and identities such as ethnicity (see Huffman 2007; Wood 2005). Personal ornaments such as jewellery are highly visible and they can be worn to signal ethnic identity by individuals (Wells 1998:244). However when it comes to the interpretation of bead stylistic variation at Danamombe there are several other issues to consider. Firstly the variation of glass beads was influenced by production and exchange patterns. We know that glass beads were not locally produced but they were imported from Asia and Europe, so these factors also had a bearing on the distribution of bead types at the two sites under investigation. It is also important to note that ideology, gender, status, and religious factors could have affected the distribution of locally manufactured beads in various ways at the two research sites. Therefore when we try to interpret bead stylistic variation at archaeological sites, we need to pay particular attention to their distribution in well defined archaeological contexts. Usually beads that are accidentally deposited into the archaeological record do not inform us much about past social processes. 170 Rozvi ethnicity was also inferred through monumental architectural styles because architecture often expresses ideological structures and symbolic codes. In that case it is possible to infer ideas and meanings behind the built dry-stone structures. The data presented in Chapter 6 indicates that Danamombe had less retaining walls and more free-standing walls than Khami. Additionally, dry-stone wall decoration motifs were more diverse at Danamombe than at Khami. R style free-standing walls are also more prevalent at Danamombe than at Khami, and such walls were even identified during ethnographic fieldwork at the homestead of Shupai Makotore in Bikita. As such one can argue that there were some significant architectural stylistic differences between the two research sites. The most interesting development is that both the most complex and the most inferior building styles occurred at Danamombe. Perhaps such kind of cultural variation indicates that some ideological forces inspired by Rozvi elites were at play. Manyanga et al. (2009:10) argued that stone architecture was often ideologically manipulated by the elite to express power over nature, resources and humans. Therefore profusely decorated dry-stone walls could have been constructed to signal the power of Rozvi rulers who most likely inhabited such structures. Ethnographic data collected in Bikita also indicated that the Rozvi still use a traditional court defined by R-style walls for their rituals and customs. Huffman (1996) also interpreted the western enclosure (built in R-style) at Danamombe as a Rozvi courtyard. Although R-style walls were poor in quality, they could have been objectified to symbolise Rozvi traditional courts in certain contexts. Unfortunately, those specific contextual meanings are inaccessible to archaeologists if nothing was documented during those specific historic times. Emberling (1997:325) postulated that ethnicity can be marked by any aspect of material culture although household structures are more commonly used. Stone (2003) further comments that permanent objects have a more public signal of ethnicity than portable objects. Hegmon (1992) also concurs that highly visible material culture often indicates group or ethnic identity, while material in private settings convey ritual or belief systems. Stylistic changes that occurred at the site of Danamombe should be critically examined in relation to Rozvi ethnicity. However the most significant architectural change was that walls at Danamombe became more profusely decorated. Joyce and Lopiparo (2005:372) suggested that innovations of constructions should be seen as the exercise of agency, while repetition of design should be recognised as the 171 reproduction of structure. Hegmon (1992) argues that increasing diversity in decoration often reflects increasing social identities and differences. In this work I argue that changes in architectural styles expressed elite strategies to manipulate ideologies relating to identities and power. Chief among the causes of change was the desire by agents to express their identities which only manifested when ethnic members generated or emphasised differences from others. Rozvi elites could have enforced these stylistic changes simply to emphasise their ethnic identity to members who lived outside their core settlement. Unlike ceramic styles which became less complex at Danamombe, dry-stone wall styles became more elaborate in order to communicate ethnic messages to more distant members and their non-Rozvi subjects. Unfortunately, it is impossible for us as yet to determine the specific messages that were communicated by these decorations. Wells (1998) explains that material culture carries several meanings that change through time because people always imbue new meanings to objects they make, acquire and use. Interestingly, designs on walls were also common on ceramics thereby suggesting these were signs and symbols commonly used to communicate social or ideological messages (see Deacon et al. 1999; Huffman 2007; Pikirayi 2007). Thus meanings and messages carried by these designs can only be inferred by delving deeper into notions of “semiology of signification” and “semiology of communication” (see Hooper-Greenhill 1995:57). Of course such endeavours are beyond the scope of this study, hence wall designs exclusive to Danamombe possibly denote Rozvi ethnicity. However, it is important to note that other factors like wall function and masonry techniques of the builders could have influenced such changes but these cannot be sufficiently explored in this study. Dhaka structures or remains of traditional mud huts and miscellaneous finds also constituted the other archaeological indices used to trace Rozvi ethnicity. Huffman (2007) pointed out that mud huts often carry symbolic designs that communicate messages related to group identities. While this view is valid, dhaka structures at both sites under investigation did not show any stylistic changes. The absence of archaeologically visible stylistic boundaries is difficult to interpret when dealing with ethnicity because ethnic identities are dynamic concepts that thrive on objectified differences or boundaries. Henceforth it is also difficult to infer Rozvi ethnic traits from dhaka structures that show general similarities between the two research sites. Various other special finds from Danamombe were examined in terms of their significance as elite status 172 symbols. Wells (1998) suggested that personal ornaments, weapons, pottery, metal vessels and numerous other objects play active roles in creating the user‟s identity. In the context of Khami and Danamombe, it appears rare and precious goods were tightly controlled and exclusively owned by the elites. While differences are apparent in the data presented in Chapter 6, the listed objects at Danamombe are not absolute evidence of Rozvi ethnic symbols. Emberling (1997:320) argues that elite goods should not be equated to ethnic goods, rather one can only talk of elite emulation of prestigious or foreign goods before suggesting any ethnic control. In the light of such arguments, it is hereby proposed that Rozvi elites emulated a lot of foreign and local prestigious goods presented in Chapter 6. Objects such as gold beads and bangles, conus shells, blue and white Nankin China ware, guns, iron and bronze spear-heads, and crescent-shaped axe-heads having inlaid copper disc in the centre among other prestigious goods signified Rozvi material culture. However since these objects were not easily accessed by others, they could have been easily objectified as Rozvi ethnic symbols. In fact, oral, documentary and ethnographic data reviewed in Chapter 5 has proven that we can have an infinite list of material culture that was used to denote Rozvi ethnic boundaries because each socio-historical context had a unique set of symbols. For instance, documentary and ethnographic data indicated that natural features like mountains and trees, as well as ethnographic objects like wooden plates, gourd cups, woven bowls, and reed-woven surfers among other materials were situationally projected as Rozvi symbols. Most of these symbols are either impossible or difficult to recover from the archaeological record. By and large miscellaneous goods used during Rozvi installation ceremonies for chiefs and those recovered from their royal capital Danamombe could have served as Rozvi ethnic markers in the past. Be that as it may, we should not be blind to the fact that these objects were not permanent ethnic markers but they were situationally or contextually invoked to mark ethnic differences in dynamic socio-political processes. 7.4 Challenges of defining identities and power archaeologically The greatest challenge for archaeologists studying ethnicity is that it is difficult to recover the specific ideas and perceptions that surrounded material culture (Jones 1997). Since ethnicity was continuously produced and reproduced to mark “difference”, archaeologists cannot simply distinguish ethnicity from mere spatial continuity and discontinuity in material culture. Meskell (2002:286) states that the fluidity of identities creates problems for archaeologists especially in 173 contexts lacking adequate historical evidence. Cultural identity and ethnicity are not clear-cut concepts, as such they are difficult to tease even with textual or ethnographic data (Emberling 1997; Hegmon 1992; Wells 1998). Hence, these views clearly reflect some of the factors that affected the definition of Rozvi ethnicity through material culture patterning at Danamombe. While we know that Danamombe was a Rozvi capital, we are still ignorant about the material culture that was already in place when the Rozvi usurped power from the Torwa. This is because the earliest Portuguese sources were totally silent about the archaeology of the Torwa and Rozvi. On the other end, oral traditions either distorted or over-simplified references to Rozvi settlements and their material culture. Only historical archaeology can resolve this problem by piecing together all the fragmented sources of data dealing with such a past. Although stylistic variation between Khami and Danamombe was equated to notions of ethnicity and power, the relations between these concepts is dynamic and extremely sophisticated (see Emberling 1997; Hegmon 1992; Wells 1998). In particular, ethnic categories may continue while style changes or styles may change while expressing the same ethnic identity (Jones 1997). At times the same style may even communicate multiple messages about power structures within a community. Therefore stylistic meanings were never fixed and their connection with ethnicity and power was temporal and fluctuating. Hegmon (1992) stated that in archaeology we generalise about style because it is impossible to get the specific historical meanings. Nonetheless, if stylistic distribution is larger than local scale of production it permits the recognition of information exchange about patterns of sameness and difference (Emberling 1997; Wells 1998). 7.5 Directions for future research Historical archaeology and agency approaches should remain pivotal in the study of identities because of their flexibility and interpretive capabilities. These frameworks have interpretive power, historical insight and relevance. Hence, they should guide further research at Khami phase sites in the Insiza district which are generally associated with both Torwa and Rozvi identities. In addition, systematic fieldwork should be conducted at these Khami phase sites at a much broader scale in order to better appreciate the nature of stylistic spatio-temporal variation. This study revealed that the relationship between ethnicity and material culture is extremely complex. While pottery at Danamombe became less diverse and more homogenous, dry-stone 174 walls and beads at the same site showed increasing diversity. Perhaps these stylistic variations were partly influenced by the disturbed contexts from which some of the samples were drawn. It therefore follows that future research should have a firm grip over context of archaeological assemblages because notions of identities and ethnicity are closely connected to provenance. Therefore more sites in the Rozvi core area should be systematically excavated in order to recover archaeological data with proper stratigraphic context. Since identities are complex constructions that existed in the minds of historic actors who subjectively imbued different material culture with meanings, our horizons into the past need to be broadened through the exploitation of multiple sources of evidence. However researchers should guard against imposing their own imaginations on archaeological assemblages because meanings of material culture are best known to the past social actors themselves. In future, researchers should both exploit and capitalise more on knowledge derived from ethnographic and historical sources when tracing identities through material culture. Clearly, ethnographic data gathered from documentary sources and the living Rozvi in Bikita was very insightful in terms of exposing processes of material culture manipulation and ethnic mobilisation. In that regard, if more post-Rozvi houses scattered across the Zimbabwean plateau and beyond are examined more promising results may be obtained. Once a rich ethnographic database is generated, the analysis and interpretation of archaeological data from other sites in the Rozvi core area such as Naletale, Manyanga, and Zinjanja among others can be done in a much fruitful or rewarding manner. Future research should pay closer attention to the dialectical relationship between ethnicity and material culture in order to articulate Rozvi ethnogenetic and maintenance processes. 175 REFERENCES Abraham, D. P. 1959. „The Monomotapa‟, Dynasty NADA, 36: 58-84. Agorsah, K. E. 1990. „Ethnoarchaeology: The Search for a Self-Corrective Approach in the Study of Past Human Behaviour‟, The African Archaeological Review, 8: 189-208. Allen, I. L. 1983. The Language of Ethnic Conflict. New York: Columbia University Press. Amm, F. L. 1939. „Geology of the country around Bulawayo‟, Southern Rhodesia Geological Survey Bulletin, 35: 1-307. Axelson, E. 1960. The Portuguese in South-East Africa, 1600-1700. Johannesburg: Witwatersrand University Press. Barth, F. 1969. Ethnic Groups and Boundaries. Boston: Little, Brown. Beach, D. N. (n.d.) „The Rozvi and the Changamire state: The origin of a myth: The historiography of the Rozvi‟, Henderson Seminar paper, 34. Beach, D. N. 1974. „Ndebele Raiders and Shona Power‟, The Journal of African History, 15 (4): 633-651. Beach, D. N. 1976. „The Mutapa Dynasty: A Comparison of Documentary and Traditional Evidence‟, History in Africa, 3: 1-17. Beach, D. N. 1980. The Shona and Zimbabwe 1900-1850. Gweru: Mambo Press. Beach, D. N. 1983a. „The Rozvi in Search of Their Past‟, History in Africa, 10: 13-34. Beach, D. N. 1983b. Zimbabwe plateau and its peoples. In History of Central Africa 1, Birmingham D. and P. M. Martin (eds.), 248-277. New York: Longman. Beach, D. N. and Posselt, F. W. T. 1986. “NADA” and Mafohla: Antiquarianism in Rhodesia and Zimbabwe with Special Reference to the Work of F. W. T. Posselt‟, History in Africa, 13: 1-11. Beach, D. N. 1987. Documents and African society on the Zimbabwean plateau. Paideuma 33: 129-145. Beach, D .N. 1 994a. The Shona and Their Neighbours. Oxford: Blackwell Beach, D. N. 1994b. A Zimbabwe Past: Shona Dynasty Histories and Oral Traditions. Gweru: Mambo Press. Beach, D. N. 1995. Zimbabwe before 1900. Gweru: Mambo Press. Beach, D. N. 1998. „Cognitive Archaeology and Imaginery History at Great Zimbabwe‟, Current 176 Anthropology, 39 (1): 47-72. Beaudry, M. C. 1988. Introduction. In Documentary Archaeology in the New World, Beaudry, M.C. (ed.), 1-3. Cambridge: Cambridge University Press. Bentley, C. G. 1987. „Ethnicity and Practice‟, Comparative Studies in Society and History, 29 (1): 24-55. Bhila, H. H. K. 1972. „Trade and the survival of an African polity: The external relations of Manyika from the sixteenth to the early nine-teeth century‟, Journal of Central Africa Historical Association, 3: 11-28. Bhila, H. H. K. 1982a. Trade and Politics in a Shona Kingdom. The Manyika and their African and Portuguese Neighbours, 1575-1902. London: Longman House. Bhila, H. H. K. 1982b. „The Portuguese Legacy in south-central Africa: A reflective essay on the Debates originating from use of Portuguese sources‟, Conference on Zimbabwean History: Progress and Development, Department of History University of Zimbabwe, 1: 1-10. Bourdieu, P. 1977. Outline of a theory of Practice. Cambridge: Cambridge University Press. Bourdillion, M. F. C. 1976. The Shona Peoples: An Ethnography of contemporary Shona, with special reference to their religion. Gwelo: Mambo Press. Bowser, B. J. 2000. „From pottery to politics: An Ethnoarchaeological study of political factionalism, Ethnicity, and Domestic pottery style in Ecuadorian Amazon‟, Journal of Archaeological Method and Theory, 7 (3): 219-248. Bullock, C. 1927. The Mashona (The indigenous Natives of Southern Rhodesia). Cape Town: Juta and Co. Ltd. Burrett, R. S. 1998. „Naletale: A Khame type settlement near Gweru‟, Heritage Zimbabwe 17: 138-150. Bvocho, G. 2001. A study of ornaments from Mwenezi and Malumba, south eastern Zimbabwe. BA Honours dissertation, University of Zimbabwe. Harare. Bvocho, G. 2005. „Ornaments as social and chronological icons. A case study of southeastern Zimbabwe‟, Journal of Social Archaeology, 5 (3): 409-423. Calabrese, J. A. 2000. „Interregional interaction in southern Africa: Zhizo and Leopard‟s Kopje relations in Northern South Africa, Southwestern Zimbabwe and Eastern Botswana, AD 10001200‟, The African Archaeological Review, 17 (4): 183-210. Calabrese, J. A. 2005. Ethnicity, Class, and Polity: the Emergency of Social and Political Complexity in the Shashi-Limpopo Valley of Southern Africa, AD 900-1300. PhD Thesis, 177 University of Witswatersrand. Johannesburg. Carneiro, L. R. 1981. The chiefdom: precursor of the state. In The Transition to Statehood in the New World, Jones, D. & R. R. Kautz (eds.), 37-79. Cambridge: Cambridge University Press. Castells, M. 2004. The power of identity. Volume 2. Oxford: Blackwell. Caton-Thompson, G. 1931. The Zimbabwe Culture: Ruins and Reactions. Oxford: Clarendon Press. Chapano, C. 2002. „A checklist of Zimbabwean grasses‟, Report No. 16. Harare: South African Botanical Diversity Network. Chakanyuka, C. 2001. An analysis of house remains from Zimbabwe Tradition sites in the Insiza district, Central Zimbabwe. BA Honours dissertation, University of Zimbabwe. Harare. Chikamba, R. 2004. Moyo Chirandu. In Shaurai: Nduri dzazuro nedzanhasi, Mhonde, R. P. Bvute, O. & Mareya, L. (eds.), 18. Harare: Priority Projects Publishing. Chiwome, E. M. 1996. A Social history of Shona Novel. Harare: Mazongororo Paper Converters. Clark, B. J. 2005. „Lived Ethnicity: Archaeology and Identity in Mexicano America‟, World Archaeology, 37 (3): 440-452. Cobo et al. 2009. J. G. Cobo, G. Dercon, C. Monje, P. Mahembe, T. Gotosa, J. Nyamangara, R. J. Delve & G. Cadisch, „Cropping Strategies, Soil fertility investment and land Management practices by smallholder farmers in Communal and Resettlement Areas in Zimbabwe‟, Land degradation and development, 20: 492-508. Cohen, D. R. n.d. Tracing rootlets of modern political discourse in Botswana: the role of archaeology in the creation of identities. Available at: http: traumwerk.stanford.edu:3455 (Accessed 03/07/10). Cohen, R. 1978. „Ethnicity: Problem and Focus in Anthropology‟, Annual Review of Anthropology, 7: 379-403. Connah, G. 1975. The archaeology of Benin. Oxford: Clarendon Press. Connah, G. 2007. „Historical archaeology in Africa: An Appropriate Concept?‟, African Archaeological Review, 24: 35-40. Curtin, P. D. 1969. „Oral Traditions and African History‟, Journal of Folklore Institute, 6 (2/3): 137-155. 178 Daneel, M. L. 1970. The God of the Matopo Hills: An essay on the Mwari Cult in Rhodesia. Paris: Mouton and Company. Deacon et al. 1999. D. Deacon, M. Pickering, P. Golding & G. Murdock, Researching Communications: A Practical Guide to Methods in Media and Cultural analysis. New York: Oxford University Press. Denbow et al. 2008. J. R. Denbow, J. Smith, N. M. Ndobochani, K. Atwood, K. & D. Miller, „Archaeological excavations at Bosutswe, Botswana: cultural chronology, paleo-ecology and economy‟, Journal of Archaeological Science, 35: 459–480. Dhliwayo, A. W. 1989. A study of dhaka houses from the Great Zimbabwe tradition sites. BA Honours dissertation, University of Zimbabwe. Harare. Doke, C. M. 1931. A Comparative study in Shona Phonetics. Johannesburg: The University of Witswatersrand Press. Dornan, J. L. 2002. „Agency and Archaeology: Past, Present and Future Directions‟, Journal of Archaeological Method and Theory, 9 (4): 303-329. Emberling, G. 1997. „Ethnicity in Complex Societies: Archaeological Perspectives‟, Journal of Archaeological Research, 5 (4): 295-344. Emerson, T. E. 1997. Cahokia and the Archaeology of Power. [E-books] Tuscaloosa: University of Alabama press. Available at: http://www.questia.com/library/book/cahokia-and-the-archaeology-of-power-bythomas-e-emerson.jsp (Accessed 23/10/09). Eriksen, T. H. 1991. „The Cultural Contexts of Ethnic Differences‟, Man, New Series, 26 (1): 127-144. Fortune, G. 1956. „A Rozwi: Text with translation and Notes‟, NADA, 33: 67-84. Fortune, G. and Hodza, A. C. 1974. „Shona-Praise Poetry‟, Bulletin of the School of Oriental and African Studies, University of London, 37 (1): 65-75. Foucault, M. Translated by Sheridan Smith, A.M. 1972. The Archaeology of Knowledge. London: Tavistock. Foucault, M. 1980. Two Lectures. In Power/Knowledge: Selected Interviews, Gordon, C. (ed.). New York: Pantheon. Francis, E. K. 1947. „The Nature of the Ethnic Group‟, The American Journal of sociology, 52 (5): 393-400. Gamble, C. 2001. Archaeology: The Basics. London: Routledge. 179 Gann, L. H. 1965. A History of Southern Rhodesia: Early days to 1984. London: Chatto and Windus. Garlake, P. S. 1970. „Rhodesian-Ruins- A Preliminary Assessment of their Styles and Chronology‟, The Journal of African History, 11 (4): 495-513. Garlake, P. S. 1973. Great Zimbabwe. London: Thames and Hudson. Garlake, P. S. 1982. Great Zimbabwe: described and explained. Harare: Zimbabwe Publishing House. Garlake, P. 1987. The Painted Caves: an introduction to the prehistoric art of Zimbabwe. Harare: Modus Publications. Gelfand, M. 1959. Shona Ritual: with special reference to the Chaminuka cult. Cape Town: Juta and Co. Ltd. Giddens, A. 1984. The Constitution of Society. Cambridge: Polity. Giddens, A. 1991. Modernity and Self-Identity: Self and Society in the late Modern Age. Stanford: Stanford University Press. Gifford, J. C. 1960. „The Type-Variety Method of Ceramic Classification as an Indicator of Cultural Phenomena‟, American Antiquity, 25 (3): 341-347. Githitho, A. N. 2005. The sacred Majikenda Kenya forest of coastal Kenya: Traditional conservation and management practices. In Traditional conservation practices in Africa, Joffroy, T. (ed.), 61-67. Rome: ICCROM. Gosselain, O. P. 2000. „Materializing Identities: An African Perspective‟, Journal of Archaeological Method and Theory, 7 (3): 187-217. Gould, R. A. 1980. Living Archaeology. Cambridge: Cambridge University Press. Gould, R. A. and Watson, P. J. 1982. „A dialogue on the meaning and use of analogy in ethnoarchaeological reasoning‟, Journal of Anthropological Archaeology, 1: 355-381. Gutu, C. 2007. An analysis of Pottery from the Site of Danan'ombe, South-west Central Zimbabwe. BA Honours Dissertation, University of Zimbabwe. Harare. Hall, R. N. 1909. Prehistoric Rhodesia. London: T. Fisher Unwin. Hall, M. 1987. The Changing Past: farmers, kings and traders in southern Africa 200-1860. Cape Town: David Philip. 180 Hall, M. and Markell. 1993. Introduction: Historical archaeology in the Western Cape. In Historical archaeology in Western Cape, Hall, M. & Markell, A. (eds.), 3-5. South Africa Archaeological Society Goodwin Series 7. Hall, M. and Silliman, S. W. 2006. Introduction: Archaeology of the Modern World. In Historical Archaeology, Hall, M. & Silliman, S. W. (eds.), 1-22. Malden: Blackwell Publishing. Harrison, S. 1999. „Cultural Boundaries‟, Anthropology Today, 15 (5): 10-13. Hegmon, M. 1992. „Archaeological research on style‟, Annual Review on Anthropology, 21: 517536. Henige, D. 1988. „Oral, but Oral What? The Nomenclatures of Orality and their Implications‟, Oral Tradition, 3 (1-2): 229-238. Hodder, I. 1982. Symbols in Action: Ethnoarchaeological Studies of Material Culture. New York: Cambridge University Press. Hodza, A. C. and Fortune G. 1974. „Shona-Praise Poetry‟, Bulletin of the School of Oriental and African Studies, University of London, 37 (1): 65-75. Hodza, A. C. and Fortune G. 1979. Shona Praise Poetry. Oxford: Oxford University Press. Holleman, J. F. 1952. Shona Customary Law. Cape Town: Oxford University Press. Hooper-Greenhill, E. 1995. Museums and their visitors. London: Routledge. Huffman, T. N. 1974. The Leopard‟s Kopje Tradition. Salisbury: National Museums of Southern Rhodesia (National Museums and Monuments of Southern Rhodesia Memoir 6). Huffman, T. N. 1981. „Snakes and birds: expressive space at Great Zimbabwe‟, African Studies, 40: 131-150. Huffman, T. N. 1989. „Ceramics Settlements and Later Iron Age Migrations‟, African Archaeological Review, 7: 155-182. Huffman, T. N. 1996. Snakes and Crocodiles: Power and Symbolism in Ancient Zimbabwe. Johannesburg: Witwatersrand University Press. Huffman, T. N. 2000. Mapungubwe and the origins of the Zimbabwe culture. In African Naissance: The Limpopo Valley 1000 Years Ago, Leslie, M., Maggs, T. (eds.), 14–29. (South African Archaeological Society Goodwin Series 8). Huffman, T. N. 2005. Mapungubwe: Ancient African Civilisation on the Limpopo. Johannesburg: Witwatersrand University Press. 181 Huffman, T. N. 2007. Handbook to the Iron Age: The Archaeology of Precolonial Farming Societies in Southern Africa. South Africa: University of KwaZulu-Natal. Huffman, T. N. 2008a. „Climate change during the Iron Age in the Shashe-Limpopo Basin, southern Africa‟, Journal of Archaeological Science 35: 2032-2047. Huffman, T. N. 2008b. „Mapungubwe and Great Zimbabwe: The origin and spread of social complexity in southern Africa‟, Available online 3 December. Hughes, G. 1997. „Excavation of the peripheral area of settlement at Khami, 1989‟, Zimbabwea, 5: 3-21. Isbell, W. H. 2000. What we should be studying: the „imagined community‟ and the „natural community. In The archaeology of Communities: New World perspective, Canuto, M. A. and J. Yaeger (eds.), 243 -266. New York: Routledge. Jones, S. 1997. The Archaeology of Ethnicity: Constructing identities in the Past and Present. London: Routledge. Joyce, R. A. and Lopiparo, J. 2005. „Doing Agency in Archaeology‟, Journal of Archaeological Method and Theory, 12 (4): 365-374. Kahari, G. 1990. The rise of the Shona novel: A study in development, 1890-1984. Gweru: Mambo Press. Kay, G. 1970. Rhodesia: A Human Geography. England: University of London Press. Kidd, K. E. and Kidd, M. A. 1970. A classification system for glass beads for the use of field archaeologists. Ottawa, Parks Canada: Canadian Historic Sites: Occasional Papers in Archaeology and History, 1: 45-89. Kinahan, J. 2000. Cattle For Beads: The archaeology of historical contact and trade on the Namib Coast. Studies in African Archaeology 17. Uppsala: Societas Archaeologica Uppsaliensis. Kobylinsky, Z. 1989. An ethnic change or socio-economic one? The 5th to 6th centuries AD. in the Polish lands. In Archaeological Approaches to Cultural Identity, Shennan, S. J. (ed.), 303312. London: Unwin Hyman Ltd. Kreuter, U. P. and Workman, J. P. 1996. „Cattle and Wildlife Ranching in Zimbabwe‟, Rangelands, 18 (2): 44-47. Kuper, H. 1954. The Shona. In The Shona and Ndebele of Southern Rhodesia: Ethnographic survey of Africa, Southern Africa Part IV, Forde, D. (ed.), 9-37. London: International African Institute. Kwashirai, V. C. n.d. Poverty in the Gwai Forest Reserve, Zimbabwe: 1880-1953. http://www.globalenvironment.it/kwashirai.pdf. (Accessed 16/06/11). 182 Lancaster, C. S. 1974. „Ethnic Identity, History, and “Tribe” in the Middle Zambezi Valley‟, American Ethnologist, 1: 707-730. Lane, P. and Reid, A. 1998. Historical archaeology in Botswana. In Ditswa Mmung: The Archaeology of Botswana, Lane P., Reid, D.A.M. & Segobye A. (eds.), 161-176. Gabarone: Pula Press. Leone, M. 1977. The new Mormon Temple in Washington, D.C. In Historical Archaeology and the Importance of Things, Ferguson, L. (ed.), 43-61. Washington, DC: Society for Historical Archaeology. Lewis-Williams, J. D. 1980. „Ethnography and iconography: aspects of southern San thought and art‟, Man, 15: 467-482. Lewis-Williams, J. D. 1981. Believing and seeing: symbolic meaning in southern San rock art. London: Academic Press. Little, B. 1992. Text-aided archaeology. In Text-Aided Archaeology, Little, B. (ed.), 1-6. Boca Raton, F L: CRC Press. Loubser, J. H. N. 1988. Archaeological contribution to Venda Ethnohistory. PhD Thesis, University of Witswatersrand. Johannesburg. Love, D. and Jonker, L. (eds.) 2005. An overview of the Mzingwane Catchment, Zimbabwe a contribution to the WaterNet Challenge Program Project 17. Integrated Water Resources Management for Improved Rural Livelihoods: Managing Risk, mitigating drought and improving water scarce Limpopo Basin. WaterNet Working Paper, No. 1. Lucas, G. 2004. An Archaeology of Colonial Identity: Power and Material Culture in the Dwars valley, South Africa. New York: Kluwer Academic. Machiridza, L. H. 2005. Setting the parameters for reconsidering the Rozvi archaeological identity in South-Western Zimbabwe: A Historical Archaeology Perspective. BA Honours dissertation, University of Zimbabwe. Harare. Management Plan 2004. Danamombe National Monument. National Museums and Monuments of Zimbabwe. Management Plan 2004. Matobo Hills World Heritage Site. National Museums and Monuments of Zimbabwe. Mann, M. 1986. The sources of Social Power. Cambridge: Cambridge University Press. Manyanga, M. 2006. Resilient Landscapes: Socio-environmental dynamics in the ShashiLimpopo basin, southern Zimbabwe c. AD 800 to the present. Uppsala: Department of Archaeology and Ancient History, Societas Archaeologica Uppsaliensis. 183 Manyanga et al. 2009. M. Manyanga, G. Pwiti & S. Chirikure, Conceptualizing Forager and Farmer Interactions in the Shashi-Limpopo Basin: Preliminary thoughts from surveys in southern Zimbabwe and eastern Botswana. Unpublished Seminar Paper, Department of Anthropology and Archaeology, University of Pretoria. Matowanyika et al. 1998. J. Z. Z. Matowanyika, N. Marongwe, J. M. Made, E. Dengu, M. Mukahanana, V. Vhudzijena & M. Moyo, Land Resources. In The State of Zimbabwe‟s Environment, Chenge, M. Sola, L. & Palecz, D. (eds.), 140-212. Harare: Government of the Republic of Zimbabwe, Ministry of Mines, Environment and Tourism. Mazarire, G. C. 2003. „Changing Landscape and Oral Memory in South-Central Zimbabwe: Towards a Historical Geography of Chishanga, c.1850-1990‟, Journal of Southern African Studies, 29 (3): 701-715. Mazarire, G. C. 2009. Reflections on pre-colonial Zimbabwe, c.850-1880s. In Becoming Zimbabwe: A history from pre-colonial period to 2008, Raftopoulos, B. and A.S. Mlambo (eds.), 1-38. Harare: Weaver Press. McDowell, J. H. 1981. „Toward a Semiotics of Nicknaming‟, Journal of American Folklore, 97 (37): 1-17. McIntosh, S. K. 1999. Pathways to complexity: An African perspective. In Beyond Chiefdoms: Pathways to Complexity in Africa, McIntosh, S.K. (ed.), 1-30. Cambridge: Cambridge University Press. Meskell, L. 2002. „The Intersections of Identity and Politics in Archaeology‟, Annual Review Anthropology, 31: 279-301. Mhiripiri, N. A. 2008. The Tourist Viewer, the Bushmen and the Zulu: Imaging and (re)invention of identities through contemporary visual cultural productions. PhD Thesis, University of KwaZulu-Natal. KwaZulu-Natal. Mitchell, P. 2002. The Archaeology of southern Africa. Cambridge: Cambridge University Press. Moyo et al. 1998. N. A. G. Moyo, D. Mazvimavi, S. Mtetwa, B. N. Dube, A. T. Mhlanga, & W. Mhlanga, Water Resources. In The State of Zimbabwe‟s Environment, Chenge, M. Sola, L. & Palecz, D. (eds.), 214-268. Harare: Government of the Republic of Zimbabwe, Ministry of Mines, Environment and Tourism. Mudenge, S. I. G. 1974a. „An Identification of the Rozvi and its Implications for the History of the Karanga‟, Rhodesian History. 5: 19-32. Mudenge, S. I. G. 1974b. „The role of foreign trade in the Rozvi Empire. A reappraisal‟, Journal African History, xv (3): 373-391. Mudenge, S. I. G. 1977. „Eighteenth-Century Portuguese Settlements on the Zambezi and the Dating of Rhodesian Ruins: Some reflections on the Problems of Reference dating‟, The International Journal of African Historical Studies, 10 (3): 384-393. 184 Mudenge, S. I. G. 1988. A Political History of the Munhumutapa. Harare: Zimbabwe Publishing House. Mulligan, S. P. 2003. “An Archaeology of Ideology?” A Reply to Howarth‟, Journal of Political Studies, 51: 429-435. Mupira, P. 1991. A classification of imported glass beads from some Iron Age Traditions in Zimbabwe. BA Honours dissertation, University of Zimbabwe. Harare. Mupira, P. 2007. „Farming communities on Murahwa‟s Hill, Central, Eastern Highlands Zimbabwe‟, Zimbabwea, 9: 63-82. Murimbika, M. 2006. Sacred Powers and Rituals of Transformation: An Ethnoarchaeological Study of Rainmaking Rituals and Agricultural Productivity during the Evolution of the Mapungubwe State, AD 1000 to AD 1300. PhD Thesis, University of the Witwatersrand. Johannesburg. Muringaniza, J. S. and Ruwitah, A. R. R. M. 1996. „Ethnography and archaeological interpretation of use of space at Zimbabwe Tradition sites‟, Zimbabwea, 4: 23-33. Musesengwe, B. 2009. Beads from the kitchen enclosure Danan‟ombe National Monument: An integrated approach. BA Honours dissertation, Midlands State University. Gweru. Mutasa, N. 1990. Nhume Yamambo. Gweru: Mambo Press. Mutasa, N. 1991. Misodzi Dikita Neropa. Gweru: Mambo Press. Ncube, G. T. 2004. A History of Northwestern Zimbabwe 1850-1960. Kadoma: Mond Books. Ndoro, W. 1996. Towards the Meaning and Symbolism of Archaeological Pottery Assemblages. In Aspects of African Archaeology: Papers from the 10th PanAfrican Association for Prehistory and Related Studies, Pwiti, G. & Soper, R. (eds), 773-779. Harare: University of Zimbabwe Publications. Ndoro, W. 2001. Your Monument our Shrine: The preservation of Great Zimbabwe. Studies in African Archaeology 19. Uppsala: Societas Archaeologica Uppsaliensis. Ngara et al. 1998. T. Ngara, W. Zhakata, V. Mugova, B. Makunike & S. Musingarabwi, The Atmosphere. In The State of Zimbabwe‟s Environment, Chenge, M. Sola, L. & Palecz, D. (eds.), 101-137. Harare: Government of the Republic of Zimbabwe, Ministry of Mines, Environment and Tourism. Nobanda et al. 1998. N. Nobanda, M. Moyo, A. Murwira, E. B. Wiken & R. Mkwanda, Ecoregions: A First Approximation. In The State of Zimbabwe‟s Environment, Chenge, M. Sola, L. & Palecz, D. (eds.), 406-462. Harare: Government of the Republic of Zimbabwe, Ministry of Mines, Environment and Tourism. 185 Norton, R. 1993. „Culture and Identity in the South Pacific: A Comparative analysis‟, Man, New Series, 28 (4): 741-759. Nyamapfene, K. 1991. Soils of Zimbabwe. Harare: Nehanda Publishers (Pvt.) Ltd. Orser, Jr. C. E. 2008. „Historical archaeology as a Modern-World archaeology in Argentina‟, International Journal of Historical Archaeology, 12: 181-194. Orser, Jr. C. E. 2010. „Twenty-First-Century Historical Archaeology‟, Journal of Archaeological Research, 18: 111-150. Paynter, R. 2000. „Historical and Anthropological Archaeology: Forging Alliances‟, Journal of Archaeological Research, 8 (1): 1-37. Pfukwa, C. n.d. Liberation struggles and Terrorist wars: some reflections on Chimurenga. Unpublished Seminar Paper, Midlands State University. Gweru. Phillipson, D. W. 1976. The Prehistory of Eastern Zambia. Nairobi: British Institute in East Africa. Phillipson, D. W. 1985. African archaeology. Cambridge: Cambridge University Press. Phillipson, D. W. 2005. African Archaeology. 3dr Edition. Cambridge: Cambridge University Press. Pikirayi, I. 1993. The Archaeological Identity of the Mutapa State: Towards a historical archaeology of northern Zimbabwe. Studies in African Archaeology 6. Uppsala: Societas Archaeologica Uppsaliensis. Pikirayi, I. 1997a. Research trends in historical archaeology on the Zimbabwean plateau and adjacent margins. In Caves, Monuments and Texts: Zimbabwean Archaeology Today. (Studies in African Archaeology 14), Pwiti, G. (ed.), 143-157. Uppsala: Department of Archaeology and Ancient History. Pikirayi, I. 1997b. The pottery of the historical Mutapa state: An examination of the local ceramics from some trading centres in northern Zimbabwe. Zimbabwea, 5:30-47. Pikirayi, I. and Pwiti, G. 1999. „States, traders and colonists: Historical archaeology in Zimbabwe‟, Historical Archaeology, 33 (2): 73-89. Pikirayi, I. 2001. The Zimbabwe Culture; Origins and Decline of Southern Zambezian States. New York: Altamira Press. Pikirayi, I. 2003. Shifting Patterns of Trade in the Western Indian Ocean Zone: Northern Zimbabwe, 1500-1700 AD. In Climate Change, Trade and Modes of Production in Sub-Saharan Africa. (Studies in African Past-3), Chami, F. & Pwiti, G. (eds), 129-154. Dar es Salaam: Dar es 186 Salaam University Press. Pikirayi, I. 2007. „Ceramics and group identities: Towards a social archaeology in southern African Iron Age ceramic studies‟, Journal of Social Archaeology, 7 (3): 286-301. Pongweni, A. J. C. 1996. Shona praise Poetry as role negotiation: The battles of the clans and the sexes. Gweru: Mambo Press. Posnasky, M. and Decorse, C. R. 1986. „Historical Archaeology in Sub-Saharan Africa-A Review‟, Historical Archaeology: Journal for the Society of Historical Archaeology, 20 (1): 114. Pwiti, G. 1996a. Continuity and Change. An archaeological study of farming communities in northern Zimbabwe AD 500-1700. Studies in African Archaeology 13. Uppsala: Societas Archaeologica Uppsaliensis. Pwiti, G. 1996b. „Peasants, Chiefs and Kings: A model of the Development of Cultural Complexity in Northern Zimbabwe‟, Zambezia, xxiii (i): 31-52. Plog, S. 1983. „Analysis of Style in Artifacts‟, Annual Review of Anthropology‟, 12: 125-142. Poilecot, P. 2007. Eragrostis species of Zimbabwe. Montpellier: Editions Cemagref. Posnansky, M. 1969. „Bigo Bya Mugenyi‟, Uganda Journal, 33: 125-150. Posselt, F. W. T. 1935. Fact and Fiction. Bulawayo: Books of Rhodesia. Preucel, R. W. 2000. Making Pueblo Communities: Architectural discourse at Kotyiti, New Mexico. In The archaeology of Communities; New World perspective, Canuto, M.A. and J. Yaeger (eds.), 58-77. New York: Routledge. Ralushai, V. N. M. N. 1977. Conflicting accounts of Venda history in particular reference to the role of Mutupo in social organisation. PhD Thesis, Queens University. Belfast. Randall-McIver, D. 1906. Medieval Rhodesia. London: McMillan. Randles, W. G. L. 1958. South east Africa and the Empire of the Monomotapa as shown on selected printed maps of the sixteenth century. Studia 2, 103-163. Randles, W. G. L. 1979. The Empire of Munhumutapa from the fifteenth to the nineteenth century. Gweru: Mambo Press. Reid et al. 1998. A. Reid, K. Sadr & N. Hanson-James, Herding traditions. In Ditswa Mmung: the archaeology of Botswana, Lane, P., Reid, A. & Segobye, A. (eds). Gabarone: Pula Press and The Botswana Society. 187 Robertshaw, P. 1999. Seeking and Keeping Power in Bunyoro-Kitara Uganda. In Beyond Chiefdoms: Pathways to Complexity in Africa, McIntosh, S.K. (ed), 124-135. Cambridge: Cambridge University Press. Robinson, K. R. 1959. Khami Ruins. Cambridge: Cambridge University Press. Robinson, K. R. 1961. „An Early Iron Age site from the Chibi district, southern Rhodesia‟, South African Archaeological Bulletin, 16: 147-164. Robinson, K. R. 1966. The archaeology of the Rozwi. In The Zambezian Past, Stokes, E. & R. Brown (eds.), 3-27. Manchester: Manchester University Press. Rudd, S. 1984. „Excavations at Lekkerwater ruins, Tsindi Hill, Theydon, Zimbabwe‟, The South African Archaeological Bulletin, 39 (140): 83-105. Sackett, J. R. 1990. Style and ethnicity in archaeology: the case of isochrestism. In Uses of style in Archaeology, Conkey, M. Harstorf, C. (eds.), 32-43. Cambridge: Cambridge University Press. Service, E. R. 1962. Primitive Social Organisation: an evolutionary perspective. New York: Random House. Schoeman, S. H. 2006. Clouding Power? Rain-control, Space, Landscapes and Ideology in Shashe-Limpopo State Formation. PhD Thesis, University of the Witwatersrand. Johannesburg. Schuter, T. 2006. „Geological Atlas of Africa: with notes on Stratigraphy, Tectonics, Economic Geology, Geohazards and Geosites of Each Country‟, Chapter 4, 248-251: Springer Berlin Heidenlberg. Available at: http://books.google.co.zw/books?id=w1hLhf6bGscC&printsec=frontcover&source=gbs_ge_sum mary_r&cad=0#v=onepage&q&f=false (Accessed 07/04/10). Schimdt, P. R. 1977. Historical Archaeology: a structural approach in an African culture. Contributions in inter cultural and comparative studies No. 3. Connecticut: Greenwood Press. Schimdt, P. R. 1983. „An alternative to a strictly materialist perspective: a review of historical archaeology, ethnoarchaeology and symbolic approaches in African archaeology‟, American Antiquity, 48 (1): 62-79. Schofield, J. F. 1948. Primitive Pottery, an introduction to South African ceramics, prehistoric and protohistoric. Cape Town: South African Archaeological Society. Schuyler, R. 1970. „Historical and historic sites archaeology as anthropology: basic definitions and relationships‟, Historical Archaeology, 4: 83-89. Schuyler, R. 1978. Historical Archaeology: a guide to substantive and theoretical contributions. New York: Baywood Publishing Co., Farmingdale. 188 Shennan, S. J. 1989. Archaeological approaches to cultural identity. In Archaeological Approaches to Cultural Identity, Shennan, S. J. (ed.), 1-32. London: Unwin Hyman Ltd. Shepard, A. O. 1956. Ceramics for the Archaeologist. Washington, DC: Carnegie Institution. Shoko, E. D. (ed.) 2001. Midlands State University: Inauguration and Installation Ceremony, 17 March. Sibanda, C. 1999. A guide to monuments in Central Region, (1993 Central Region Museum File copy). da Silva Rego, A. and Baxter, T. W. (eds.) 1971. Documents on the Portuguese in Mozambique and Central Africa, 1540-1560, (7 vols.). Lisbon: Centro de Estudos Historicos Ultramarinos and the National Archives of Rhodesia and Nyasaland. da Silva Rego, A. and Baxter, T. W. (eds.) 1962-1975. Documents on the Portuguese in Mozambique and Central Africa, 1497-1840, (8 vols.). Lisbon: Centro de Estudos Historicos Ultramarinos and the National Archives of Rhodesia and Nyasaland. Silliman, S. W. 2006. Struggling with Labor, Working with identities. In Historical Archaeology, Hall, M. & Silliman, S. W. (eds.), 147-166 . Malden: Blackwell Publishing. Sinamai, A. 2003. Cultural shifting Sands: Changing Meanings of Zimbabwe sites in Zimbabwe, South Africa and Botswana. Available at: www.international.icomos.org/victoriafalls2003/papers/C3-7%20-%20Sinamai.pdf (Accessed 09/05/11). Sinclair, P. J. J. 1987. Space, Time and Social Formation: a territorial approach to the archaeology and anthropology of Zimbabwe and Mozambique, c.0-1700 AD (sun 9). Uppsala: Societas Archaeologica Uppsaliensis. Sinclair et al. 1993. P. J. J. Sinclair, I. Pikirayi, G. Pwiti & R. Soper, Urban trajectories on the Zimbabwean plateau. In The Archaeology of Africa: food, metals and towns, Shaw, T., P. J. J. Sinclair, B. Andah & A. Okpoko (eds), 705-31. London: Routledge. Smith, J. 2005. Climate Change and Agropastoral Sustainability in the Shashe-Limpopo River Basin from AD 900. Doctoral thesis, University of Witwatersrand, Johannesburg. Soper, R. 1971. „A general review of the Early Iron Age in the southern half of Africa‟, Azania, 6: 5-37. South, S. 1977. Method and Theory in Historical Archaeology. New York: Academic Press. 189 Spear, T. 2003. „Neo-Traditionalism and the limits of invention in British Colonial Africa‟, Journal of African history, 44: 3-27. Spector, J. D. 1976. „The Interpretive potential of Glass Trade Beads in Historical Archaeology‟, Historical Archaeology, 10: 17-27. Square, C. J. 1996. „Returning to Paradise through Naming: The Incantation of Names. in B. Breyten Breytenbach‟s Return to Paradise‟, Nomina Africana, 10 (1&2): 79-97. Stein, G. J. 1998. „Heterogeneity, Power and Political Economy: Some Current Research Issues in Archaeology of Old World Complex Societies‟, Journal of Archaeological Research, 6 (1): 144. Stone, T. 2003. „Social Identity and Ethnic Interaction in the Western Pueblos of the American Southwest‟, Journal of Archaeological Method and Theory, 10 (1): 31-67. Summers, R. F. H. 1959. „Excavations at DhloDhlo ruins‟, Unpublished report, Archaeological survey, Museum of Human Sciences, Harare. Summers, R. F. H. 1961. „Southern Rhodesia Iron Age: First approximations to the history of the last 2000 years‟, Journal of African History, 2 (1): 1-13. Summers et al. 1961. R. F. H. Summers, K. Robinson & A. Whitty, „Zimbabwe excavations‟, Occasional Papers of the National Museums of Rhodesia, 3 (23A): 15-332. Tapela, M. C. 1998. An archaeological investigation of ostrich eggshell beads in Botswana, Botswana Journal of African Studies, 15 (1): 60-74. Timberlake, J. and I. Mapaure. 1999. Vegetation survey of Maramani/Tuli area, southwest Zimbabwe: Unpublished report, CESVI Lowveld Project. Thondhlana, T. 2005. Style, Space and Time: a critical analysis of the chronology and spatial distribution of copper and copper beads from Zimbabwe Iron Age sites. BA Honours dissertation, University of Zimbabwe. Harare. Tyndale-Biscoe, R. 1949. „The Geology of the country around Gwelo‟, Southern Rhodesia Geological Survey Bulletin, 39: 1-145. Vansina, J. 1965. Oral Tradition: A study in Historical Methodology. Chicago: Aldine. Vansina, J. 1985. Oral Tradition as History. Madison: University of Wisconsin Press. Vansina, J. 1996. Epilogue: Fieldwork in history. In Pursuit of history: Fieldwork in Africa, Adenaike, C. K. & Vansina, J. (eds.), 127-147. Portsmouth:Heinemann. van Waarden, C. 1998. The Late Iron Age. In Ditswa Mmung: The Archaeology of Botswana, 190 Lane P., Reid, D. A. M. & Segobye A. (eds.), 115-160. Gabarone: Pula Press. Veit, U. 1989. Ethnic concepts in German prehistory: A case study on the relationship between cultural identity and archaeological objectivity. In Archaeological Approaches to Cultural Identity, Shennan, S. J. (ed.), 35-56. London: Unwin Hyman Ltd. Vincent, V. and R. G. Thomas. 1961. An agricultural survey of Southern Rhodesia Part 1. Salisbury: Government Printer. Voss, B. L. 2008. The Archaeology of Ethnogenesis: Race and Sexuality in Colonial San Francisco. Berkeley: University of California Press. Voss, B. L. And Allen, R. 2008. Overseas Chinese Archaeology: Historical foundations, current reflections, and new directions, Historical Archaeology, 42 (3): 5-28. Walker, N. J. 1995. Late Pleistocene and Holocene Hunter-gatherers of the Matopos. An archaeological study of change and continuity in Zimbabwe. Uppsala: Department of Archaeology and Ancient History, Societas Archaeologica Uppsaliensis. Watson, R. L. A. 1960. „The Geology and Coal Resources of the Country around Wankie‟, Southern Rhodesia Geological Survey Bulletin, 48: 1-57. Wells, P. S. 1998. „Identity and Material Culture in the Later Prehistory of Central Europe‟, Journal of Archaeological Research, 6 (3): 239-298. Wieschhoff, H. A. 1941. The Zimbabwe-Monomotapa culture in southeast Africa. Menasha, Wisconsin: George Banta (General Series in Anthropology 8). Whitelaw, G. 1997. Iron-using agriculturalists of eastern and southern Africa. In Encyclopaedia of Precolonial Africa: Archaeology, History, Languages, Culture, and Environments, Vogel, J.O. (ed.), 444-455. Walnut Creek: Alta Mira Press. Whitty, A. 1961. „Architectural style at Zimbabwe‟, Occasional Papers of the National Museums of Southern Rhodesia, 3 (23A): 289-305. Wiessner, P. 1983. „Style and Social Information in Kalahari San Projectile Points‟, American Antiquity, 48: 253-276. Wiessner, P. 1985. „Style as Isochrestic variation? A reply to Sackett‟, American Antiquity, 50: 160-166. Wiessner, P. 1990. Is there a Unity to Style? In The Uses of Style in Archaeology, Conkey, M. W. & C. Hastorf (eds.), 105-112. Cambridge: Cambridge University Press. Wild, H. 1952. A southern Botanical Dictionary of Native and English Plant Names. Salisbury: Government Printer. 191 Wood, M. 2005. Glass Beads and Pre-European Trade in the Shashe-Limpopo Region. Masters thesis, University of Witwatersrand. Johannesburg. Wylie, K. C. 1973. „The Uses and Misuses of Ethnohistory‟, The Journal of interdisciplinary History, 3 (4): 707-720. Yaeger, J. and M. A. „Canuto. 2000. Introducing archaeology of communities. In The archaeology of Communities; New World perspective, Canuto, M. A. and J. Yaeger (eds.), 1-15. New York: Routledge. Yinger, J. M. 1985. „Ethnicity‟, Annual Review Sociology, 11: 151-180. Zachrisson, P. 1978. An African Area in Change. Belingwe 1894-1946: A study of Colonialism, Missionary activity and African response in Southern Rhodesia. Gothenburg: Department of History in Gothernburg. 192 APPENDICES APPENDIX A: Data capture sheet for Torwa/Rozvi ceramics Site name: Acc. TP/TR Layer No Vessel shape categories Lip forms Fabric 1 2 3 4 5 6 7 ro ta sq ex.T in.t b.t bv fl comb co md fn Exterior Colour fr bk g gbr b a Decoration Technique Decoration placement rbr st other in pu b/po g/b pa c-st w/fb en l r n s b bs Motifs Comments APPENDIX B: Data Capture Sheet for Torwa/ Rozvi Beads Site Name: Size Acc. TP/TR Key Lev Material M. Tech Shape Structure Colour Diaphe. Deco. >5 mm Length Categories 5-8 8-14 mm mm 2-4 mm Width Categories 4-6 6-10 mm mm Comments Acc.-Accession Number; TP/TR - Test Pit/Trench; Lev-Level; M.Tech.-Manufacturing Technique; Diaphe-Diapheneity; Deco-Decoration b APPENDIX C: Data capture sheet for dry stone-wall stylistic attributes Site name: ………………………………. Wall No. Material G Oth Dry stone-wall Architectural attributes Techniques Entrance Form Rt. Fs. Site section: ……………………………............................................................... Height d/bl. n/co. bb Rd Sq Lf/d. Chevron Decoration Hr Cord Check Ic/n-g Srp/h.inf. NB* Wall No.- wall number, G- granite, Oth- other, Rt.- retaining wall, Fs.- free-standing wall, d/bl.- dressed blocks, n/co.-neat coursing, bb- battered back, Rd- rounded, Sq- squared, Lf/d- linear friezes of dentelle, Hr- herringbone, Ic/n-g – isolated courses of non-granite stone forming bands of a different colour, Srp/h.inf.- small rectangular panels with a herringbone infill. c APPENDIX D: Rozvi Interview Questions Name of respondent: ………………………………………………………………….. Age & Sex: ………………………………………………………………….. Occupation: ………………………………………………………………….. Dynastic house: ………………………………………………………………….. Date of interview: ………………………………………………………………….. 1. Who are the Rozvi? 2. Which Rozvi houses do you know and where did your house come from? 3. Explain the difference, if any, between the Rozvi and the Karanga. 4. Explain to me any Rozvi customs which you remember. From where did the Rozvi acquire their political power? 5. Which Rozvi religious or ceremonial materials from the past do you know and what were/are their significance? 6. Which are the most important aspects of Rozvi traditional customs still practiced today? d APPENDIX E: Detailed descriptions of analysed vessels from Khami and Danamombe Pots with tall necks, and pots with short in-sloping, vertical, and concave necks. 1. Tall necked pot with a funnel and fluted neck, lipform slightly out-turned and tapered. 2. Pot with tall vertical neck, lipform out-turned and rounded. 3. Pot with tall slightly in-sloping neck, out-turned rim and rounded lipform. 4. Pot with tall in-sloping neck? Rounded lipform and out-turned rim. 5. Tall necked pot with in-sloping neck? Vertical squared lipform, rim externally thickened and bevelled. 6. Short necked pot with in-sloping neck, rounded lipform and slightly thickened rim. 7. Pot with short vertical neck, rounded lipform and out-turned rim. 8. Same as 7 but rim more out-turned and tapered. 9. Same as 7 and 8 but rim slightly out-turned and tapered. 10. Pot with short vertical neck, rounded lip and slightly externally thickened rim. 11. Same as 10 but rim slightly out-turned and tapered. 12. Same as 10 but neck much shorter. 13. Same as 11 but lip form slightly tapered 14. Same as 12 but sherd thickness is smaller. 15 Same as 14 but lip almost squared. 16. Pot with short in-sloping neck, bilaterally thickened rim and rounded lip. 17. Pot with short in-sloping neck, externally rolled rim and rounded lip. Globular pots with very restricted concave necks. 1. Globular pot with very short neck, externally thickened rim and lip slightly squared. 2. Same as 1 but rim externally rolled and lipform rounded. 3. Same as 1 & 2 but rim distinctly out-turned and lipform rounded. 4. Globular pot with a very short neck, out-turned rim and tapered lip. 5. Same as 5 but neck is more concave, while the lip is rounded. 6. Same as 5 but neck distinctly concave. 7. Same as 6 but sherd thickness slightly smaller. 8. Same as 7 but neck region more defined. 9. Globular pot with very a short vertical neck? Lipform tapered. 10. Same as 5 but with tapered lip. 11. Same as 7 but lip distinctly tapered. e Globular pots with very restricted in-sloping necks. 1. Constricted globular pot with in-sloping rim, vertical squared lipform. 2. Constricted globular pot with externally thickened rim and rounded lip. 3. Constricted pot with in-sloping bilaterally thickened rim and rounded lip. 4. Same as 1 but lipform rounded. 5. Constricted pot with simple vertical rim and tapered lip. 6. Constricted pot with plain in-sloping rim and squared lip. 7. Constricted pot with bilaterally thickened rim and rounded lip. 8. Constricted pot with simple in-sloping rim and vertical rounded lip. 9. Constricted pot with in-sloping rim and tapered lip. Globular pots with no necks; plain/simple rims and rolled/out-turned rims. 1. Globular constricted pot with no neck, externally thickened rim and rounded lip. 2. Same as 1but rim more vertical and tapered. 3. Globular constricted pot with tapered lip. 4. Same as 3 but rim more vertical and lip is rounded. 5. Same as 4 but rim is more vertical and raised. 6. Globular constricted pot with vertical rim and tapered lip. 7. Globular constricted pot with a slightly externally thickened rim and rounded lip. 8. Globular constricted pot with in-sloping rim and slightly raised rounded lip. 9. Same as 8 but lip slightly tapered. 10. Globular constricted pot with externally thickened rim and rounded lip. 11. Globular constricted pot with horizontal rim and tapered lip. 12. Globular constricted pot with in-sloping rim and tapered lip. 13. Globular constricted pot with in-sloping and externally thickened rim, lip rounded. 14. Same as 13 but rim almost vertical and lip slightly tapered. 15. Globular constricted pot with bilaterally thickened rim and almost squared lip. 16. Same as 15 but with an oval moulded boss decorated with deep broad line diagonal incisions. 17. Same as 14 but rim more vertical and lipform rounded. 18. Globular constricted pot with in-sloping externally thickened rim and rounded lip. 19. Globular constricted pot with in-sloping heavily rolled rim and rounded lip. 20. Globular constricted pot with horizontal rolled rim and rounded lip. 21. Same as 20 but rim heavily rolled. 22. Same as 21 but rim bilaterally thickened and lip slightly tapered. 23. Globular constricted pot with bilaterally thickened rim and squared lip. f Constricted, open, hemispherical and straight sided bowls. 1. Constricted bowl with externally thickened rim and rounded lip. 2. Slightly constricted hemispherical bowl with out-turned thickened rim and rounded lip. 3. Same as 2 but rim not thickened. 4. Shallow bowl with undifferentiated rim and rounded lip. 5. Deep hemispherical bowl with rounded lip. 6. Same as 6 but lipform squared. 7. Same as 6 & 7 but lip square to rounded. 8. Slightly open hemispherical bowl with rounded lip. 9. Same as 8 but rim slightly thickened pointing outwards and lipform rounded. 10. Deep straight sided bowl with square to rounded lip. g
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
166 As chapters 4 and 5 are so closely aligned, the... will not be reiterated here but chapter 4 will be...