uv - UFS BLOEMFONTEIN BIBLIOTEEK - LIBRARY University Free State 34300004936732 '"'''' 1111' ,,'"Un1111i'v""' e"r"' s"i"'te""i' t "'" V"'"ry1s111't"a"' a,,'"t 1111' "" "" UNIVERSITY OF THE FReE STATE U~IVERSITEIT VAN DIE ((\i)) UFS VRYSTAAT VUNIVESITHI YA FREISTATA ~UV AN INVESTIGATION OF THE TRACE ELEMENT COMPOSITIONS OF GOLD FROM ZIMBABWE AND SOUTH AFRICA: IMPLICATIONS FOR TRACING THE SOURCE OF ARCHEOLOGICAL GOLD ARTEFACTS BY Robert Netshitungulwana SUPERVISOR: Prof. Marian Tredoux CO-SUPERVISOR: Dr Leon Jacobson Dissertation submitted in fuIfilIment of the requirement of the degree of Masters of Science at the University of the Free State University of the Free State Department of Geology March 2011 UniveVrsli)t'esittav~an die ,..,.:.. \$lOf:;lrH"~ :l~;N '1 8 .JUN 2013 DECLARA TION The content of this work is my own and has not previously been submitted for a degree at this or any other university. The work of other people is acknowledged by references. Robert Netshitungulwana Department of Geology University of the Free State March 2011 ii DEDICATION This thesis is dedicated to the following people: my wife, Tendani Shumani Netshitungulwana; my children, Ndiene Mukundi and Mulanga Vhuthuhawe; my parents Marubini Mercy and Tshamaano Philemon; my brothers, Aaron Mpfariseni, Azwinndini Sariel. Ntakuseni, Mmbavhalelo, Mmbulungeni Elijah, Makwarela Moses and Rofhiwa; my sisters, Rechael, Rendani, Ma todti, Lydia, Lufuno and all members of our extended family I omitted. I assume this would be our document, which will be read by generations to come, linking our grandchildren to us. lil ACKNOWLEDGEMENTS I would like to thank the following people: Professor M. Tredoux (UFS) and Dr L. Jacobson (McGregor Museum) for the opportunity they given to me of studying at the University of the Free State; Dr D. Miller for giving out a broad understanding of the artefacts; Professor H. Frimmel for sharing his broad knowledge of Witwatersrand gold; Dr A. Spath for help with the LA-ICP-MS analytical technique; Professor D. Reid for sharing his knowledge of the geochemistry of native gold; Mrs M. Waldron of the Electron Microscope Unit at UCT for her assistance with the SEM analytical technique; Miss C. Richards for proofreading and editing the grammar of this thesis and the National Research Foundation and Inkaba ye Africa for the full two years of financial assistance towards my studies. I am indebted to the following people for the provision of gold samples: A. Pather, J. Kleynhans, P. Smit, A. Johnson, G. Williams, P. van Zyl, M. O'Brien, V. Chamberlain, and F. von Berkel, all of Anglogold Ashanti; Dr P. Bender at the Transvaal Museum and L. Swan at the Museum of Human Sciences for the loan of gold ore and archaeological gold samples respectively; Dr H. Klinger of Iziko Museums in Cape Town for his help with the ore samples. I owe a favour to the following staff members at CGS, for their moral support and encouragement: Dr M. Cloete, Mr J.H. Elsenbroek, Mr S. Strauss, Mr M.L. Bensid, Mr A.E. Mulovhedzi, Mr S. Hlatshwayo and Mr M. Mmaya. My thanks are also due to T.S. Netshitungulwana (wife), Mukundi (daughter), Mulanga (son), Dr Z. Bagai, Mr M.J. Murovhi, Mr L. Moroka, Mr V.N. Chabangu and Mr M.B. Mudau for their moral support and encouragement all the way. I am grateful to my parents and the rest of the family for believing in me, through their prayers, perseverance and encouragement in my studies. Finally, I am most grateful to Rev. M.P. Kruger and his family and Mr Nndwakhulu Tshishonga and his family in Cape Town, for their wonderful financial support, prayers and encouragement regarding the duration of my studies. IV ABSTRACT The early black farmers who settled in southern Africa were involved in trading and metal technology. The history of mining for metals like iron, copper, tin and gold in southern Africa spans at least the past 2000 years. The main aim of the research was to test the viability of using gold chemistry to compare the composition of gold ores in South Africa and Zimbabwe with those of the archaeological gold artefacts from Thulamela, Mapungubwe and Great Zimbabwe. Samples from the Archaean greenstone belts in South Africa and Zimbabwe, as well as samples from ores associated with the Witwatersrand Supergroup, were used in the study. Trace element signatures were determined by laser ablation inductively coupled plasma mass speetrometry (LA-ICP-MS), a technique whereby low concentrations (down to low ppb levels) can be detected. In addition, Ag concentrations (wt %) were determined using a scanning electron microprobe, so that Ag could be used as an internal standard during the LA-ICP-MS runs to give semi-quantitative data. The most commonly occurring isotopes in gold, namely, 56Fe, 59Co, 60Ni 63CU, 66zn, 75As, 1880S, 105pd, 195pt, 202Hg, 107.109Ag, and 204, 206,207,208Pb and 209Bi, were used to construct the signatures, using their intensities in the mass spectra in counts per second (cps). Isotopic ratios were used to compare the gold ores with each other. The results show some variations in the signatures of gold from the greenstone belts and the Witwatersrand Basin. The 107Ag and 202Hg concentrations in gold from the Witwatersrand Basin are high compared to the greenstone belts. These differences have implications for the various models of gold deposition in these environments, pointing to different geochemical histories. Multivariate correspondence analysis plots for the major gold deposits show the wide group of the Barberton samples with little or no distinctive characteristics, compared to the Zimbabwean gold samples. The Witwatersrand gold plotted differently to the Barberton Greenstone Belt but closely related to the Zimbabwean greenstone belts. The ratio plot of 56Fe/107 Ag versus 202HglI07 Ag shows that archaelogical gold artefacts differ completely from the natural gold, indicating that the gold could not merely have been cold-worked, as has been suggested. This suggests that gold from anyone archaeological site could not be related to any particular or even regional source. This could be associated with the possibility of mixing of gold from multiple sources, recycling, contamination in melting and trade in items. v LIST OF TABLES TABLE I-I SUMMARY OFTHETRACE ELEMENTS SIGNATURES FOR THETHULAMELA AND MAPUNGUBWE ARCHAELOGICAL SAMPLES (AFTER DESAI, 200 I) 5 TAB LE 2-1 MACROSCOPIC DESCRIPTIONS OF THE VISIBLE GOLD SAMPLES COLLECTED IN SOUTH AFRICA. THE SAMPLE NUMBERS ARE THOSE ASSIGNED BY THE DONORS. WHERE APPLICABLE, SAMPLES ARE REFERRED TO FIGURES. CONTINUED ON PAGES 20, 21, AND 22 19 TABLE 2-2 MACROSCOPIC DESCRIPTIONS OF THE VISIBLE GOLD SAMPLES COLLECTED IN ZIMBABWE. THE SAMPLE NUMBERS ARE THOSE ASSIGNED BY THE DONORS. WHERE APPLICABLE, SAMPLES ARE REFERRED TO FIGURES. CONTINUED ON PAGE 26 25 TABLE 2-3 MACROSCOPIC DESCRIPTIONS OFTHE ARCHAEOLOGICAL ARTEFACTS COLLECTED IN SOUTHERN AFRICA. THE Z NUMBER REFERS TO FIGURE 2-20 AND WAS USED DURING THE ANALYSIS. TERMINOLOGY USED FOR OBJECTS SUCH AS RINGS: OD = OUTER DIAMETER, ID= INNER DIAMETER. CONTINUED ON PAGE 30 29 TABLE 3-1 THE OPERATING CONDITIONS OFTHE SEM FOR THE ANALYSIS OFGOLD AND SILVER CONCENTRATIONS IN GOLD 44 TABLE 3-2 A LIST OF ELEMENTS (ISOTOPES) COMMONLY OCCURRING IN GOLD, WHICH WERE SELECTED FOR ANALYSIS ON THE BASIS OF NATURAL ABUNDANCE, AS WELL AS THE ONES WITH LEAST INTERFERENCE 47 TABLE 3-3 THE LOWER LIMIT OF DETECTION (LLD (3 TIMES MEAN BACKGROUND VALUES)) FOR THE ISOTOPES SELECTED FOR ANALYSIS BY LA-ICP-MS. VALUES ARE IN COUNTS PER SECOND (CPS) AND 1-3 ARE REPEAT MEASUREMENTS OF A BLANK SAMPLE 48 TABLE 3-4 OPERATING CONDITIONS OF LA-ICP-MS IN THIS STUDY 49 TABLE 3-5 VARIATION OF OPERATIONAL CONDITIONS TESTED ON SAMPLES A AND B FROM MAPUNGUBWE 50 TABLE 3-6 COMPARATIVE ISOTOPE RESPONSE (CPS) TO DIFFERENT LA-fCP-MS TECHNIQUE OPERATIONAL CONDITIONS TESTED IN TWO GOLD SAMPLES (A AND B) FROM MAPUNGUBWE 50 TABLE 3-7 ISOBARIC INTERFERENCES AT MASS 58 AND 204, WITH THE ELEMENTS INVOLVED AND THEIR RELATIVE NATURAL ABUNDANCES 51 TABLE 3-8 GOLD AND SILVER CONCENTRATIONS, MEASURED BY SEM, IN ASETOF INTERNAL GOLD STANDARDS 53 VI TABLE 4-1 THE DATA PRESENTEDIN COUNTS PER SECOND BY LA-ICP-MS FOR SELECTED ISOTOPESAND IN WT % FOR SILVER BY SEM OFTHE WITWATERSRAND, BARBERTON AND ZIMBABWE GOLD SAMPLES. LOWER LIMIT OF DETECTION ALSO PRESENTED 57 TABLE 4-2 THE DATA PRESENTEDIN COUNTS PERSECOND BY LA-ICP-MS FOR SELECTED ISOTOPESAND IN WT % FOR SILVER BY SEM OFTHEGOLD MINES FROM PIETERSBURG, MURCHISON GREENSTONE BELTS, SABIE-PILGRIM'S REST AND KNYSNA GOLD SAMPLES. LOWER LIMIT OF DETECTION ALSO PRESENTED 58 TABLE 4-3 THE DATA PRESENTEDIN COUNTS PER SECOND BY LA-ICP-MS FOR SELECTED ISOTOPESAND IN WT % FOR SILVER BY SEM OF THE ZIMBABWEAN AND MAPUNGUBWE ARTEFACTS SAMPLES. LOWER LIMIT OF DETECTION ALSO PRESENTED 59 TABLE 4-4 SUMMAR Y OFTHE TRACE ELEMENT SIGNATURES OF GOLD SAMPLES FROM WITWATERSRAND BASIN AND THE GREENSTONE BELTS 72 TABLE 4-5 SUMMARY OF THE TRACE ELEMENTS SIGNATURES OF GOLD ARTEFACTS SAMPLES FROM ZIMBABWE AND MAPUNGUBWE 81 vii LIST OF FIGURES FIGURE I-I A DIAGRAM SHOWING VARIOUS DIFFERENT GEOSCIENTIFIC RESEARCH AREAS THAT COULD BENEFIT FROM A GOLD TRACE ELEMENT DATABASE FOR REFERENCEOR CONSULTATION PURPOSES. ........................................................................................................................................................ 3 FIGURE 1-2 A MAP OF IMPORTANT ARCHAEOLOGICAL SITES IN SOUTHERN AFRICA, WITH IMPORTANT WATERWAYS REGARDING TRADE SHOWN (SWAN, 1994; MILLER ET AL., 2000) 5 FIGURE 2-1 SIMPLIFIED GEOLOGICAL MAP OF SOUTH AFRICA, LESOTHO AND SWAZILAND SHOWING THE LOCALITY OF GOLD IN THE BARBERTON, MURCHISON, GIYANI AND PIETERSBURG GREENSTONE BELTS AND THE TRANSVAAL (SABlE-PILGRIM'S REST GOLDFIELD), WITWATERSRAND AND CAPE SUPERGROUPS (BUITENDAG, 2007) 12 FIGURE 2-2 GENERALIZED GEOLOGICAL MAP OF THE BARBERTON GREENSTONE BELT, SOUTH AFRICA, SHOWING THE LOCALITIES OF SOME MAIN GOLD MINES (AFTER DE RONDE AND DE WIT, 1994) .... 14 FIGURE 2-3 SIMPLIFIED GEOLOGICAL MAP WITH THE MAJOR GOLDFIELDS AND A GENERALIZED STRATIGRAPHIC COLUMN OFTHE WITWATERSRAND SUPERGROUP (AFTER FRIMMEL AND MINTER, 2002; FRIMMEL, 2005) 16 FIGURE 2-4 A SIMPLIFIED GEOLOGICAL MAP SHOWING DIFFERENT GOLD ORE DEPOSIT LOCALITIES AND THE ADJACENT GREENSTONE BELTS IN ZIMBABWE (BUITENDAG, 2007) 24 FIGURE 2-5 GOLD ORE FROM CITY DEEP MINE (G84) LOCATED IN THE WITWATERSRAND SUPERGROUP (SCALE DIVISIONS = 2.5 MM) 31 FIGURE 2-6 GOLD ORE SAMPLE FROM NEW CONSORT MINE (MGS 10591) LOCATED IN THE BARBERTON GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 31 FIGURE 2-7 GOLD ORE SAMPLE FROM THE AGNES MINE (MGS 101) LOCATED IN THE BARBERTON GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 32 FIGURE 2-8 GOLD ORE SAMPLE FROM BIRTHDAY MINE (MGS) LOCATED IN THE GIYANI GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 32 FIGURE 2-9 TwO GOLD ORE SAMPLES FROM MARABASTADT (MGS 112 (A, B)) LOCATED IN THE PIETERSBURG GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 33 FIGURE 2-10 GOLD ORE SAMPLE FROM GRAVELOTTE MINE (G80) LOCATED IN THE MURCHISON GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 34 viii FIGURE 2-11 THREE GOLD ORE SAMPLES FROM SABlE PILGRIM'S REST GOLDFIELD (G85): A. LYDENBURG B. GOEDEVERWACHT AT LYDENBURG C. NEW CHUM PILGRIM'S REST GOLDFIELD (SCALE DIVISIONS = 2.5 MM) 35 FIGURE 2-12 GOLD ORE SAMPLE FROM KNYSNA MINE (G III B) LOCATED IN THE CAPE SUPERGROUP (SCALE DIVISIONS = 2.5 MM) 36 FIGURE 2-13 GOLD ORE SAMPLE FROM GAIKA REEF (G24) LOCATED IN THE MIDLAND GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 36 FIGURE 2-14 GOLD ORE SAMPLE FROM THE MBERENGWA (BELINGWE) GREENSTONE BELT (G 117) (SCALE DIVISIONS = 2.5 MM) 37 FIGURE 2-15 GOLD ORE SAMPLE FROM DON SELUKWE MINE LOCATED IN THE MIDLAND GREENSTONE BELT: A IS G 123 AND B IS G 47 (SCALE DIVISIONS = 2.5 MM) 38 FIGURE 2-16 GOLD ORE SAMPLE FROM YANKEE DOODLE MINE (G 124) LOCATED IN THE MIDLAND GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 39 FIGURE 2-17 GOLD ORE SAMPLE FROM LOWER GWELO MINE (G79) LOCATED IN THE MIDLANDS GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 39 FIGURE 2- 18 GOLD ORE SAMPLE FROM VICTORIA REEF (G45) LOCATED IN THE MASVINGO GREENSTONE BELT (SCALE DIVISIONS = 2.5 MM) 40 FIGURE 2-19 GOLD ORE SAMPLE FROM ZAMBEZI (G72) (SCALE DIVISIONS = 2.5 MM) 40 FIGURE 2-20 THE ARCHAEOLOGICAL GOLD ARTEFACTS FROM GREAT ZIMBABWE 41 FIGURE 3-1 A SCHEMATIC REPRESENTATION AND A PICTURE OF THE SCANNING ELECTRON MICROSCOPE AT THE UNIVERSITY OF CAPE TOWN (UCT) (HTTP://SBIO.UCT.AC.ZAlWEBEMU/SEM_SCHOOLl) . 43 FIGURE 3-2 A SCHEMATIC DIAGRAM SHOWING THE GENERAL COMPONENTS OFTHE ICP-MS ANALYTICAL TECHNIQUE AND LASER FOR SOUD SAMPLING (AFTER Russo ET AL., 2002) 45 FIGURE 3-3 SEM GOLD VERSUS WEIGHED GOLD VALUES (WT %) OF AN INTERNAL GOLD REFERENCE STANDARD 54 FIGURE 3-4 PLOT OF THE SILVER CONCENTRATION (WT %) MEASURED FROM THE SEM VERSUS THE TOTAL COUNTS FOR 107AG FROM THE LA-lCP-MS 54 FIGURE 4-1 Box AND WHISKER PLOTS PRESENTING MEAN, MEAN PLUS STANDARD ERROR, MEAN PLUS I STANDARD DEVIATION, OUTLIERS AND EXTREME VALUES OF ALL THE DATA FOR THE GOLD IX SAMPLES FROM THE WITWATERSRAND BASIN, BARBERTON AND ZIMBABWEAN GREENSTONE BELTS. CONTINUED ON PAGES63 AND 64 63 FIGURE 4-2 Box AND WHISKER PLOTS PRESENTING MEAN, MEAN PLUS STANDARD ERROR, MEAN PLUS 1 STANDARD DEVIATION, OUTLIERS AND EXTREME VALUES OF ALL THE DATA FORTHE GOLD SAMPLES FROM PIETERSBURG AND MURCHISON GREENSTONE BELTS, SABIE-PILGRIM'S REST GOLDFIELD AND KNYSNA MINES. CONTINUED ON PAGE 67 67 FIGURE 4-3 Box AND WHISKER PLOTS PRESENTING MEAN, MEAN PLUS STANDARD ERROR, MEAN PLUS 1 STANDARD DEVIATION, OUTLIERS AND EXTREME VALUES OF ALL THE DATA FORTHE ZIMBABWEAN AND MAPUNGUBWE ARTEFACTS. CONTINUED ON PAGE 70 70 FIGURE 4-4 A TERNARY PLOT OF63CU, 66ZN AND 202HGOF GOLD SAMPLES FROM WITWATERSRAND BASIN (W'S AND REDCROSS), BARBERTON (B'S AND LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (Z'S AND BLUE BOXES) GREENSTONE BELTS. NORMALISED DATA IN TABLE 4-1 73 FIGURE 4-5 A TERNARY PLOT OF 63CU, 56FEAND 202HGOF GOLD SAMPLES FROM WITWATERSRAND BASIN (W'S AND RED CROSS), BARBERTON (B'S AND LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (Z'S AND BLUE BOXES) GREENSTONE BELTS. NORMALISED DATA IN TABLE4-1 74 FiGURE 4-6 A TERNARY PLOT OF58NI, 66ZN AND 202HGOF GOLD SAMPLES FROM WITWATERSRAND BASIN (W'S AND RED CROSS), BARBERTON (B'S AND LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (Z'S AND BLUE BOXES) GREENSTONE BELTS. NORMALISED DATA IN TABLE4-1 75 FiGURE 4-7 MULTIVARlATE CORRESPONDENCEANALYSIS AT 95 % CONFIDENCE INTERVAL FORTHE GOLD SAMPLES FROM WITWATERSRAND BASIN (W'S AND RED CROSS), BARBERTON (B'S AND LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (Z'S AND BLUE BOXES) GREENSTONE BELTS. AXIS 1 = 0.5 EIGENVALUE AND 59 % OF TOTAL. AXls2 = 0.14 EIGENVALUE AND 17 % OF TOTAL. 76 FiGURE 4-8 WARD'S METHOD (WARD, 1963) FOR CLUSTER ANALYSIS FORTHE GOLD SAMPLES FROM THE WITWATERSRAND BASIN (W'S), BARBERTON (B'S) AND ZIMBABWEAN (Z'S) GREENSTONE BELTS. ...................................................................................................................................................... 76 FIGURE 4-9 A RATIO PLOT FOR 56FEllo7AG AND 202HG/107AG FORTHE GOLD SAMPLES FROM THE WITWATERSRAND BASIN, BARBERTON AND ZIMBABWEAN GREENSTONE BELTS 77 FiGURE 4-10 BIVARIATE PLOT OF206PB VS 207PBFORTHE WITWATERSRAND, BARBERTON AND ZIMBABWEAN GOLD SAMPLES 77 FiGURE 4-11 MULTIVARlATE CORRESPONDENCEANALYSIS AT 95 % CONFIDENCE INTERVAL FORTHE GOLD SAMPLES FROM WITWATERSRAND BASIN (RED CROSS), BARBERTON (LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (BLUE BOXES) GREENSTONE BELTS AND MARABASTADT SAMPLES x (PINK SOLID BOXES) FROM PIETERSBURG GREENSTONE BELT. AXIS 1 = 0.5 EIGENVALUE AND 58 % OF TOTAL. AXls2 = O. 14 EIGENVALUE AND 17 % OF TOTAL.. 79 FIGURE 4- 12 MULTIVARlATE CORRESPONDENCEANALYSIS AT 95 % CONFIDENCE INTERVAL FORTHE GOLD SAMPLES FROM WITWATERSRAND BASIN (RED CROSS), BARBERTON (LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (BLUE BOXES) GREENSTONE BELTS AND GRAVELLOTE (G 1) SAMPLE (PINK SOLID BOX) FROM MURCHISON GREENSTONE BELT. AXIS 1 = 0.5 EIGENVALUE AND 55 % OF TOTAL. AXIS2 = 0.13 EIGENVALUE AND 16 % OF TOTAL. 79 FIGURE 4-13 MULTIVARlATE CORRESPONDENCEANALYSIS AT 95 % CONFIDENCE INTERVAL FORTHE GOLD SAMPLES FROM WITWATERSRAND BASIN (RED CROSS), BARBERTON (LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (BLUE BOXES) GREENSTONE BELTS AND SABIE-PILGRIM'S REST GOLDFIELDS (SL, SN AND SQ) SAMPLES (PINK SOLID BOXES). AXIS 1= 0.4 EIGENVALUE AND 48 % OF TOTAL. AXls2 = 0.2 EIGENVALUE AND 23 % OF TOTAL. 80 FIGURE 4-14 MULTIVARlATE CORRESPONDENCEANALYSIS AT 95 % CONFIDENCE INTERVAL FORTHE GOLD SAMPLES FROM WITWATERSRAND BASIN (RED CROSS), BARBERTON (LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (BLUE BOXES) GREENSTONE BELTS AND KNYSNA (K 1 AND K2) SAMPLES (PINK SOLID BOXES). AXIS 1= 0.14 EIGENVALUE AND 59 % OF TOTAL. AXIS2=0.I3 EIGENVALUE AND 17 % OFTOTAL. 80 FIGURE 4-15 A TERNARY PLOT OF63CU, 66ZN AND 202HGOF GOLD SAMPLES FROM ZIMBABWE (Z'S AND RED CROSS) AND MAPUNGUBWE (M'S AND SOLID PURPLE SQUARES) IN SOUTH AFRICA. NORMALISED DATA IN TABLE4-3 82 FIGURE 4- 16 A TERNAR Y PLOT OF 63CU, 66ZN AND 202HGOF GOLD SAMPLES FROM WITWATERSRAND BASIN (RED CROSS), BARBERTON (LIGHT GREEN DIAMONDS) AND ZIMBABWEAN (BLUE BOXES) GREENSTONE BELTS WITH MAPUNGUBWE (LIGHT BLUE TRIANGLE) AND ZIMBABWE (BROWN CIRCLES) ARTEFACTS. NORMALISED DATA IN TABLE 4-3 83 FIGURE 4- 17 MULTIVARlATE CORRESPONDENCEANALYSIS FOR THE GOLD ARTEFACTS FROM ZIMBABWE (Z'S) AND MAPUNGUBWE (M'S) 84 FIGURE 4-18 WARD'S METHOD FORCLUSTER ANALYSIS FORTHE GOLD ARTEFACTS FROM ZIMBABWE (Z'S) AND MAPUNGUBWE (M'S) 85 FIGURE 4-19 A RATIO PLOT FOR56FE(I05)/J07 AG AND 202HG(I05)/J07 AG FORTHE GOLD SAMPLES FROM WITW ATERSRAND BASIN, BARBERTON AND ZIMBABWEAN GREENSTONE BELTS, AS WELL AS ZIMBABWEAN AND MAPUNGUBWE ARTEFACTS 85 XI TABLE OF CONTENTS DECLARA TION II DEDICATION 111 ACKNOWLEDGEMENTS IV ABSTRACT V 1 INTRODUCTION 1 I. I RESEARCHBACKGROUND 2 1.2 PREVIOUSSTUDIES 4 1.3 PROJECTOBJECTIVES 6 1.4 GENERAL GEOCHEMISTRYOFGOLD 7 1.5 COMPOSITIONOFNATIVE GOLD 8 1.5.1 Hydrothermal Gold 9 1.5.2 Placer Gold 9 2 SAMPLE DESCRIPTIONS 11 2. I SOUTH AFRICAN GOLD ORES II 2.1.1 Barberton Greenstone Belt 13 2.1.2 Witwatersrand Supergroup 15 2.2 ZIMBABWE GOLD ORES 23 2.3 ARCHAEOLOGICAL ARTEFACTS 27 2.4 SAMPLE SELECTION 27 3 SAMPLE PREPARATION AND ANALYTICAL METHODS 42 3.1 SCANNING ELECTRONMICROSCOPE(SEM) 42 XJI 3.2 LASER ABLATION INDUCTIVELY COUPLED MASS SPECTROMETRY (LA-ICP-MS) 44 3.2.1 LA-1CP-MS procedure usedfor this work 48 3.2.2 Effects of Mass Interference 51 3.3 DATA REDUCTION AND QUALITY 52 4 RESULTS AND INTERPRETATION 56 4.1 THE MAJOR ORE PROVINCES(WITWATERSRAND, BARBERTON AND ZIMBABWE) 60 4.2 MINOR GOLD ORE DEPOSITS 65 4.3 ARCHAEOLOGICAL ARTEFACTS 69 4.4 GENERAL OBSERVATIONS AND MULTIVARlATE STATISTICS 72 4.4.1 Major Gold Ore Province 72 4.4.2 Minor Gold Ores 78 4.4.3 Archaeological Artefacts 81 5 DISCUSSION 87 5.1 CHEMICAL SIGNATUREOFTHE MAJOR GOLD DEPOSITS 87 5.2 CHEMICAL SIGNATURE OFTHE MINOR GOLD DEPOSITS 92 5.3 CHEM ICAL SIGNATURE OFTHE ARCHAEOLOGICAL ARTEFACTS 94 6 CONCLUSIONS AND RECOMMENDATIONS 100 7 REFERENCES 104 xiii 1 INTRODUCTION The early black farmers who settled in southern Africa were involved in trading and metal technology (Summers, 1969; Phimister, 1976; Oddy, 1983, 1984). The archaeological evidence shows that trading and metal technology were practised before colonizers from Europe arrived in southern Africa. Great Zimbabwe was acknowledged to have housed about 18,000 inhibitants and flourished between about AD 1290 and AD 1450 (Miller, 200 I;Huffman, 2007, 2009). It was an empire that included the present day Zimbabwe, western parts of Botswana, northern parts of South Africa and parts of Mozambique, where most of the trading activities occurred (Jacobson et al., 2002). The empire was thought to have ceased because of the over- utilization of natural resources, and internal conflicts. The chiefs migrated to the north and south in search of new sources of ivory and metals for their trading activities (Jacobson et al., 2002 and the references therein). The locations of early gold mines that were prospected, the mining techniques used, the recovery of gold from the ore, and trade were intensively discussed by Miller et al. (2000) and references therein. Maritime trade increased from small beginnings late in the first millennium AD, was well established by the loth century AD and reached a peak in the is" century AD. The internal conflicts (internecine strife) paved a way for the modern European colonizers. The Voortrekkers in the 18th century AD used old trade routes which were established from Delgoabaai and Inhaunbaue by early settlers (de Vaal, 1984; 1985). 1.1 Research Background In 1936 Letcher had already asked very important questions (Letcher, 1936). Who were the first people to discover gold in Africa? How and where did they find gold and what uses did they have for the precious metal? What became of all this gold? Where and by whom was it absorbed? Indeed many scientists have tried to answer some of the questions he asked. To address these questions it is important to look back on the introduction of metals in southern Africa. The history of mining for metals in southern Africa spans at least the past 2000 years (Oddy, 1984; Miller, 1995). That there was exploitation of metals like iron, copper, tin and gold, is supported by archaeological evidence. The earliest evidence for iron production in southern Africa was slags from the 2nd to 6th century AD sites in southern Mocambique. Also, evidence was found at Broederstroom (South Africa) dating to between the 4th and ih centuries AD, and at Divuyu in Botswana from the mid 6th century AD (Miller, 1995). The early copper mining in southern Africa spans AD 770 and 1750 for mines at Loolekop, Sealene and Kgopolwe in the Phalaborwa district. The Messina district also has evidence of early copper mining, but it has not been dated yet. The Phalaborwa and Messina district areas became major copper producers in the zo" century. Olifantspoort and the Dwarsberg in the western Transvaal and near Rooiberg were also early copper mining areas (Miller, 1995). The ancient tin sources were restricted to the mines of Rooiberg in the central Transvaal. Tin mining has been dated back to the 15th to 17th centuries AD (Miller, 1995). Tin was alloyed with copper for bronze (Cowey, 1994). 2 Humans have known about gold since pre-historie times. The earliest gold objects of all ancient civilizations were fashioned directly from native gold (Boyle, 1979), without any metallurgical treatment. In Africa, the tombs of the Pharaohs contained various gold artefacts that date back to 1350 BC (Boyle, 1979). In southern Africa, gold first appears during the Late Iron Age, after about AD 1000. Gold artefacts were found in elite burials sites such as Mapungubwe and Thulamela and also in the political centers such as Great Zimbabwe. Gold trace element fingerprint Gold mining Research Police company fields forensics Figure 1-1 A diagram showing various different geoscientific research areas that could benefit from a gold trace element database for reference or consultation purposes. Characterization of the trace and ultra-trace element signatures in native gold is generally known as gold fingerprinting and was developed about 25 years ago (Watling et al., 1994). The technology used for gold trace element analysis by laser ablation inductively coupled mass speetrometry (LA-ICP-MS), which greatly improved the accuracy of gold fingerprinting, was started in the 1990s. The earliest 3 fingerprinting of native gold by this method was done in Australia by Watling et al. (1994), and was aimed at sourcing stolen gold for return to mining companies. The gold fingerprint database has been admitted as evidence in courts of law in the case of bullion theft (Watling et al., 1994). Figure I-I shows various geoscientific research areas that could benefit from a gold trace element database for reference purposes. 1.2 Previous Studies An archaeological study done by Oddy (1984) shows that the history of mining and quarrying for metal ores in southern Africa dates as far back as 2 000 years. However, the history of indigenous metal technology remains largely unknown. This is due to some extent to uncertainty about the source of gold used to produce various artefacts, such as beads, bangles, statues and sheets, which have been recovered from the graves at archaeological sites such as Mapungubwe, Thulamela and Great Zimbabwe (Figure 1-2). Current research followed previous studies done on the archaeological gold artefacts at Mapungubwe and Thulamela (Oddy, 1984; Grigorova et al., 1998; Desai, 2001). These studies were based on the presence or absence of elements only and there was no actual data which could have been used in subsequent comparison. Table I-I from the previous studies indicates two different chemical signatures in the Mapungubwe gold artefacts. The first group was marked by the presence of strontium, mercury, rare earth elements, platinum group elements and barium. The second group has platinum group elements only and no other contaminants (Desai, 2001; Grigarova et al., 1998; Miller et al., 2000,2001). 4 Table 1-1 Summary of the trace elements signatures for the Thulamela and Mapungubwe archaelogical samples (after Desai, 2001). Elements Thulamela Mapungubwe A B Strotium ,/ Mercury ,/ ,/ Rare Earth ,/ ,/ Elements Platinum ,/ ,/ ,/ Group Elements Bismuth ,/ Barium ,/ SOUTHERN AFRICA INDIAN OCEAN 800km i N Figure 1-2 A map of important archaeological sites in southern Africa, with important waterways regarding trade shown (Swan, 1994; Miller et al., 2000). 5 Gold artefacts from Thulamela have chemical signatures marked by the presence of strontium, mercury, rare earth elements, platinum group elements and barium, similar to the Mapungubwe artefacts. It was concluded by Miller et al. (200 I) that the gold used to make certain artefacts from Mapungubwe come from one gold source, whereas gold used to make the other artefacts come from a different source. The gold source common to the metal goldsmiths of the Thulamela gold artefacts suggests that the two societies (Thulamela and Mapungubwe) had access to the same source of gold. Gold artefacts from Bosutswe showed no link to gold originating from either Mapungubwe or Thulamela (Grigorova et al., 1998). Ginwala et al. (1986) have used proton induced techniques for the determination of some trace impurities in gold objects but analytical determinations were hindered by the gold background. 1.3 Project Objectives The main aim of the research reported in this document was to test the viability of using gold chemistry to compare the composition of native gold ores in South Africa and Zimbabwe with those of the archaeological gold artefacts from Thulamela, Mapungubwe and Great Zimbabwe. Any similarities observed between the composition of the gold ores and artefacts, may help in the establishment of possible gold trade routes of archaeological importance. Gold ore samples from various localities in southern Africa, representing the most significant gold ore districts, namely, the Barberton, Murchison, Giyani and Pietersburg Greenstone Belts, the Witwatersrand Basin and the Zimbabwean craton, were procured for analysis. In order to achieve the main aim of the research, as stated above, the following tasks 6 were undertaken: (a) the development of a consistent analytical protocol for gold samples using a laser ablation inductively coupled plasma mass spectrometer (LA-Iep-MS); (b) the comparison of gold ore composition with those of the archaeological gold artefacts recovered from the ancient sites of Mapungubwe, Thulamela and Great Zimbabwe. An attempt was made to determine: • whether distinct regional sources of gold could be identified; • whether the gold used in these artefacts was soureed from multiple gold ore deposits as a consequence of complex trading routes and systems at that time; • whether gold trade between sites could be traced by identification of characteristic gold compositions; and • what metallurgical techniques, if any, were employed in the processing of gold; for example, whether systematic alloying with silver or copper was practised. 1.4 General Geochemistry of Gold This section summarizes the previous work on the general geochemistry of native gold. The trace elements present in native gold are related in some way to the processes associated with the original mineralization event. In order to understand the differences in composition of natural gold samples, it is important to review the geochemical behaviour of this metal during mineralization. Gold is a soft, yellow metal and has high electrical and thermal conductivity, exceeded only by copper and silver (Boyle, 1979; Gasparrin, 1993). In its pure state 7 gold is the most malleable and ductile of all the transition metals. Gold is a member of the group Il elements in the periodic table and when compared to the other elements in the periodic table, it shows similarities with those in group 10 (nickel, palladium and platinum) and of course the other elements in group Il (copper and silver). Gold can also form amalgams with mercury and the other elements in group 12 (zinc and cadmium) (Boyle, 1979; Gasparrin, 1993; Seward, 1984). Gold may be found in one of the following three oxidations states: the native (0), aurous (+1) and the auric (+3) states. Boyle (1979) and Seward (1984) discuss the general characteristics of gold cations, commenting that Au (I) forms a number of organometallic compounds, whereas Au (III) forms both inorganic and organometallic complexes. They comment that the solubility of gold in geochemical environments results from its general properties and that the only indisputable processes occur in gossans, where gold is either dissolved by mercury (thus forming amalgams), or reacts with Ch released by NaCl reacting with Mn02, to form chlorine complexes, which transport gold and other metals. 1.5 Composition of Native Gold Native gold contains the following major elements: 80 to 99 % gold, 1 to 15 % silver and up to 5 % copper. Minor elements present include up to I % mercury and iron, with nickel, cobalt, zinc, palladium, platinum and lead all typically at parts per million (ppm) level (Chisholm, 1979; Sie et al., 1996; Frimmel and Gartz, 1997; Allan and Woodcock, 2001 and references therein).' In nature, native gold can be found alloyed or contaminated (especially under oxidizing conditions) with other major, trace and ultra-trace elements, which provide 8 useful information about the origin and the ore formation processes of gold (Erasmus et al., 1987; Minter et al., 1993; Watling et al., 1999; Chapman et al., 2002; Guerra, 2004; Guerra et al., 2005; Raymond et al., 2005). The contaminating elements can be classified into two main categories, namely, (I) the more volatile chalcophile elements, such as zinc, cadmium, lead, bismuth, mercury, silver, and copper; and (2) the less volatile siderophile elements, such as iridium, osmium, palladium, platinum, rhenium, rhodium and ruthenium (Boyle, 1979; Roeder, 1984; Erasmus et al., 1987; Reisberg et al., 2004). During ore formation (hydrothermal or placer) the inter- element signature should reflect the unique characteristics of the processes involved in the resultant ore deposits, as will be discussed in the following sections. 1.5.1 Hydrothermal Gold In hydrothermal gold deposits, major and trace elements that can be identified in native gold may owe their presence to different types of ligands such as co', Cl, Br and HS· that were responsible for transporting gold and other metals in solution (e.g. Seward, 1984), as different types of metal complexes (Boyle, 1984; Fyfe and Kerrich, 1984). According to Fyfe and Kerrich (1984), most elements that accompany gold during fluid transportation are also concentrated during deposition/precipitation, along with gold (Groves and Foster, 1991; Groves et al., 1998). The elements that are deposited with gold include zinc, cadmium, lead, bismuth, mercury, silver, copper and platinum group elements (Reisberg et al., 2004; Nakagawa et al., 2005). 1.5.2 Placer Gold In placer gold, the mechanical weathering which causes the relocation of dispersed grains in primary gold ore deposits, e.g. the physical transport in fluvial systems, 9 often leads to an enhancement of gold purity by leaching out silver (Mosier et al., 1989; Nakagawa et al., 2005). Previous work on the general characteristics of placer gold has shown that gold grains frequently have a rim of silver depletion caused by the probable dissolution of silver in an oxidizing environment or during transportation (Chisholm, 1979). Any other base metal present in placer gold will also be removed selectively by leaching and the rate of leaching will be in the following descending order of leachability: irorc-nickelocopperc-silver (Chisholm, 1979). The platinum- group elements are not leached from native gold (Antweiler and Campbell, 1977; Siebert et al., 2005; Falconer et al., 2005). The concentration of platinum-group elements is therefore expected to be high in native gold compared to silver and the base metals in relation to the original concentration. Hypothetically, the different gold deposits selected will exhibit different chemical signatures because the deposits are from different environments of formation (hydrothermal for greenstones, modern placer for the Witwatersrand gold). The previous study (Chisholm, 1979; Antweiler and Campbell, 1977) shows that alluvial gold will be enriched in the platinum group elements and depleted in the base metal; the inverse is true for hydrothermal gold. 10 2 SAMPLE DESCRIPTIONS 2.1 South African Gold Ores South African gold ore samples in this study were obtained from the Transvaal Museum in Pretoria, Iziko Museums in Cape Town and the natural resource company Anglogold Ashanti in Johannesburg. The sample descriptions are presented in Table 2-1. The sample identity numbers that were used are those assigned by the donors and are presented as follows: MGSs are those from the Transvaal Museum; Gs are those from the Iziko Museum and 2Bs are those from Anglogold Ashanti. The samples originated from the following gold mines in South Africa: New Consort, Sheba, Alpine and Agnes mines from the Barberton Greenstone Belt; Gravellotte Mine from the Murchison Greenstone Belt; the Marabastadt Goldfield in the Pietersburg Greenstone Belt; Birthday Mine in the Giyani Greenstone Belt; City Deep Mine, Ventersdorp Contact Reef (VCR) (South Deep Mine), Carbon Leader Reef (Tautona Mine), Inner Basin Reef (West Rand Group), Contact Reef (Nolingwa Mine), B Reef (Welkom Goldfield) and the Leader Reef quartzite in the Witwatersrand Basin; the Sabie-Pilgrim's Rest Goldfield in the Transvaal Supergroup and "Knysna" gold ore from the Cape Supergroup (Figure 2-1). The geological settings for the Barberton and Witwatersrand gold ore provinces from which most samples originated are discussed in the following sections and summarized in Table 2-1. II ... GOLD LOCALITIES • TOWNS ~$[ s 1:70(m) GEOLOGICAL LEGEND (SlmpHi.d) VenR~anjNarre ~ Maine6tuy KaalfMnl. GatrlOOS, cargo caves and Kanu Gtoups E3 Ge~ SuperwDUPN;Jnwqla and NItt5t.IetIrnorpl"tc PrOVIfICa 0- ._.,,",,50"""-. -=QO\C)Ian~cI,*-antsl'loet( Mop crc.tf_4 by I Wl. '""efldog-c.-.:.-a 'Of' &cox ......-CJ"""""c:o.rpe, ....... tca._XU2_--P TtanIVaBI ~gr.o.u.p.. 'g 1.8 mm (Figure 2-5). The gold ore sample consists of milky quartz. The grains are G84 (b) City Deep Mine Same as above. <2 mm. The gold has mineralized within the quartz grains with pyrite and galena. The reddish colour on the surface of the ore is due to the presence of hematite. The ore consists of a poorly sorted quartz pebble 28602 Inner 8asin Reef The gold deposit is a modified placer. conglomerate. The pebble size ranges from I to 3 cm in diameter. The gold is finely «2 mm grain size) disseminated within the host rock. The gold deposit is located in the Witwatersrand The ore consists of well-sorted sandstone. There is a carbon 8asin, within the rocks of the Central Rand layer with cleavage perpendicular to the bedding plane. The Group of the Welkom Gold fields. The gold 28603 Carbon Leader Reef deposit is a modified placer (Robb and Meyer, gold mineralization is associated with the cleavage direction. The gold is fine grained (>2.5 mm) and 1995; Robb and Robb, 1998; Frimmel and disseminated within the host rock. Minter, 2002; Frimmel, 2005). The gold ore consists of well-rounded and angular grains of quartz. The quartz grains are> I cm and poorly sorted. The 28604 Leader Reef Same as in Carbon Leader Reef. mineral chlorine is present in the ore. Gold grains are> 2 mm. A conglomerate ore consists of poorly sorted well-rounded pebbles. The ore contains of a distinctive layer of carbon 28605 C Reef Same as in Carbon Leader Reef. approximately 2 mm thick. The gold has mineralized within this layer in between the quartz grains. The pebble grain size ranges from 0.5 to I cm. The gold grains are <2 mm. The gold is hosted in fine-grained and well sorted sandstone. The sandstone contains of a distinctive layer of carbon 28607 8 Reef Same as in Carbon Leader Reef. where gold has mineralized. The gold grains are <2 mm. The gold is associated with other sulphide minerals like 19 r---------------------------------------------------------~-----------~---------------------~------------ ..------------------------~------ ..--------------------------------------------~ Sample no. Gold deposits (Locality) Geological setting and the environment of Sample description formation pyrite. The gold is associated with the mineralized quartz vein 2B608 South Deep Mine Same as in Carbon Leader Reef. confined by pyritic ore. The gold grains are <2 mm and disseminated within the quartz vein. The ore consists of milky white quartz veins and The information of this Reef was not provided by 2B609 Birimiam Reef occasionally clear quartz crystals. The gold grains are <2 the donor. mm. Mineralization occurred within inter-locked grains of quartz parallel to the bedding. The gold deposit is located in the Barberton Greenstone Belt. The gold deposit is located between the contact zone of the Onverwacht The gold ore is dark greenish in colour due to the ferro- magnesium rock forming minerals. The gold grains are >2 MGSI0591 New Consort Mine Group and Fig Tree Group rocks. The mm. The other sulphide minerals (pyrite, arsenopyrite) are mineralization is hydrothermal and associated with present (Figure 2-6).komatiite volcanics (Cochran, 1982; Viljoen, 1984; Anhaeusser and Maske, 1986). The gold deposit is located in the Barberton Greenstone Belt, within the metasedimentary rocks of the Moodies Group within the Main MGS (?) Sheba Mine The gold ore consists of quartz. The gold grains are <2 mm Reef Complex and the Zwartkoppie ore shoot and disseminated within the host rocks. (Anhaeusser, 1974; Wagner and Wiegand, 1986; Schurrnann et al., 2000). MGS 101(a): The gold ore consists of smoky quartz, with fractures in two directions. The surface is reddish in colour, due to the presence of hematite. Occasionally yellowish in colour because of goethite present. Gold grains are >3 mm The gold deposit is located in the Barberton and occur as massive throughout the host rock (Figure 2-7). Greenstone Belt, within the metasedimentary MGS 101 (a,b) Agnes Mine rocks of the Moodies Group. The mineralization is hydrothermal (Houstoun, 1987). MGS 101(b): The gold ore consists of smoky quartz, with fractures in two directions. The surface is reddish in colour, due to the presence of hematite. Occasionally yellowish in colour because of the goethite present. Gold occurs as massive throughout the host rock, with grains >3 mm. The gold deposit is located in the Giyani The gold ore consists of smoky quartz. The reddish brown Greenstone Belt, within the rocks of BIF, colour is due to the presence of hematite. The gold is MGS (?) Birthday Mine quartzite, tremolite-actinolite schist, amphibolites disseminated within the host rock. Mineralization occurred and minor dolomites. Mineralization occurred in within the quartz vein (Figure 2-8). 20 Sample no. Gold deposits (Locality) Geological setting and the environment of I Sample description formation quartz and is associated with sulphide replacement (de Wit et al., 1992; Brandl and de Wit, 1997; Gan and van Reenen, 1997; Ward and Wilson, 1998). MGS 102: The gold ore is reddish brown on the surface due to the presence of iron rich minerals. The gold has mineralized within the highly weathered zone. The gold grains are «1.5 to 2 mm) disseminated throughout the host rock. MGS 112a: The gold ore is reddish and consists of milky The gold deposit is located in the Pietersburg quartz. The gold ore was highly weathered on the surface. MGS102,112 Greenstone Belt, within felsic, mafic and Marabastadt Goldfield Gold grains are <1.5 to 2 mm and disseminated throughout (a.b) ultramafic and volcano-sedimentary rocks of the the host rock (Figure 2-9). Uitkyk Formation. MGS 112b: The gold ore consists of clear rounded grains of quartz. The grains are cemented together by calcrete and iron cement. There is a presence of hematite and goethite giving rise to yellowish and reddish spots. The gold is visible and finely «1.5 to 2 mm) disseminated throughout the host rock (Figure 2-9). The gold deposit is hosted at the rocks of the Gravellotte Mine, Murchison The host mineral is antimony. The gold is visible and G80 Murchison Greenstone Belt. Mineralization hasGreenstone Belt massive with grains <2 mm (Figure 2- I0). occurred within the antimony mine. Lydenburg (A): Gold nuggets, five pieces. The gold grains are >2mm. The gold is associated with the heavy minerals The gold deposit occurs within the rocks of the like galena and sphalerite with some hematite (Figure 2-11). Proterozoic Transvaal Supergroup in the Malmani Subgroup of the Chuniespoort Group. Goedeverwacht (B): The gold ore is reddish to yellowish in G85 (three Sabie Pilgrims Rest Goldfield The gold is hosted in sheet-like gold-quartz- colour due to the presence of iron rich minerals (hematite samples) carbonatite-sulphide veins (Tyler and Tyler, and goethite). Gold grains are >2 mm (Figure 2-11). 1996; Harley and Charlesworth, 1996; Killick and Scheepers, 2005). New Chum Pilgrims Rest (C): The ore is reddish brown due to the presence of hematite. The gold is finely disseminated with grains <2 mm (Figure 2-1 I). G30 De Kaap Goldfield, Transvaal The information of this gold deposit was not The gold ore consists of clear quartz. The surface of the 21 Sample no. Gold deposits (Locality) Geological setting and the environment of Sample description formation found in literature. quartz is occasionally reddish due to the presence of hematite. Gold is finely disseminated with grains <2 mm. The gold mineralization has occurred along the available! fractures in quartz. MGSl11a: The ore consists of milky quartz. The gold ore is light greenish in colour with a prominent cleavage in one Situated about 17 km north west of Knysna. The direction. There is a presence of galena mineral. Gold was gold is hosted within the rocks Table Mountain visible and localized throughout the host rock with grains >2 MGS III (a,b) Group of the Cape Supergroup. MineralizationKnysna Mine mm. has occurred in quartz at the Millwood Gully, a tributary of Hominy River discovered by Thomas 8ain in 1886. MGS I11b: The ore consists of milky quartz. The gold is localized throughout the host rock with grains >2 mm (Figure 2-12). 22 2.2 Zimbabwe Gold Ores The Zimbabwean gold ore samples were obtained from the Museum of Human Sciences (formerly Queen Victoria Museum) in Zimbabwe and Iziko Museums in Cape Town. Gold ore samples were originally from the following gold mines (Figure 2-4): Reef of Gaika, Don Selukwe, Yankee Doodle and Lower Gwelo mines from the Midland Greenstone Belt; Mberengwa (Belingwe) Mine from the Belingwe Greenstone Belt; Hope Reef from the Harare Greenstone Belt; Gadzema from the Buchwa Greenstone Belt; Victoria Reef from the Masvingo Greenstone Belt; and Zambezi gold ore. Geographical coordinates information of the gold ores was not provided by the donor. Bartholomew (1990) has the list of mines with geographic coordinates, and this information was useful. Samples received with names similar to gold mines were located on a geological map (Figure 2-4). The general descriptions of the Zimbabwean gold ore samples are presented in Table 2-2. Gold ore deposits in Zimbabwe are associated with Archaean granitoid rocks (Figure 2-4) or greenstone xenoliths within the granitoids (Campbell and Pitfield, 1994; Kalbskopf and Nutt, 2003). They are classified into stratabound and non-stratabound deposits and they are distributed within the rocks of the Sebakwian Group (3.5 Ga), Bulawayan Group (2.9 Ga to 2.7 Ga) and Shamvaian Group (2.7 Ga) (Foster and Wilson, 1984; Dirks and van der Merwe, 1997; Dirks et al., 1999). The stratabound deposits comprise BIF that are intercalated with volcanics and sedimentary lithologies; whereas approximately 60 % of the non-stratabound gold deposits were derived from the mafic rocks, 17 % from ultramafic rocks, 13 % from the granitoid rocks and relatively small quantities from the felsic volcanics and sedimentary lithologies (Foster and Wilson, 1984). 23 The Zimbabwean gold mineralization is thought to have occurred in three phases. The hydrothermal phase associated with the mineralization event occurred at 2.65 ± 0.06 Ga, which was followed by another metamorphic event that occurred between 2.52 and 2.56 Ga; a post eratonic event associated with the intrusion of the Great Dyke of Zimbabwe is dated about 2.41 ± 0.07 Ga (Campbell and Pitfield, 1994). The minerals associated with gold from the granitic complexes are arsenopyrite, chalcopyrite, galena, molybdenite, pyrite, pyrrhotite, scheelite, sphalerite and stibnite (Mann, 1984). The mineralization style of the Zimbabwean greenstone belts is similar to the greenstones in the Kaapvaal Craton in South Africa and Yilgarn Block in Australia (Groves et al., 1998; Houstoun, 1987; Martin, 1993). LEGEND Waler Bodies o Gold Min•• Chrono.lratlgraphy NFault. NTl1ru.ts :' -, 'Inferrred faull. .. Inferred thrusts _ Graenslone Bells Chronoslraligr.phy Cenozoic Mesozoic _ Pal.eozole Prolorozole Archaea" Figure 2-4 A simplified geological map showing different gold ore deposit localities and the adjacent greenstone belts in Zimbabwe (Buitendag, 2007). 24 Table 2-2 Macroscopic descriptions of the visible gold samples collected in Zimbabwe. The sample numbers are those assigned by the donors. Where applicable, samples are referred to figures. Continued on page 26. Sample no. Gold deposits (Locality) Geological setting and the environment of Sample description formation Gold is mineralized at the Kwekwe Ultramafic Striation pattern was identified in the host rock. The gold has Complex of the Midland Greenstone Belt. Gold mineralized with pyrite and galena in talc-dolomite schist. The gold is hosted in talc-carbonatite schist, magnesite is visible and massive, and seems to follow the striation pattern on G24 Gaika Reef, Gwelo district rock, tonalitic gneiss (diabase, keratophyre the contact zone. The gold has a grain size of >2 mm (Figure 2-13). dykes) on greenstone margin (Bartholomew, 1990; Campbell and Pietfield, 1994). The gold is hydrothermal. The host rock is light greenish in colour and consists of quartz. The G34 (a, b) Hope Reef, Millwood gold Greenstone (hydrothermal) gold ore has carbonate minerals as it fizzes with HCl. Gold wasvisible, localized and fine grained «2 mm) and disseminated throughout the host roek. Gold is hosted in tonalitic gneiss, amphibolite, sepentine, and/or mylonite schist in gneiss of Gl17 Belingwe Mine the Mtshingwe Group In the Mberengwa The ore is greenish due to the presence of the malachite mineral.(Belingwe) Greenstone Belt (Martin, 1993; The gold is visible and consists of grains < I mm (Figure 2-14). Campbell and Pietfield, 1994). The gold is hydrothermal. The host rocks are mafic greenstone, schist, The ore consists of quartz. The quartz has vessel-like structures and serpentine, and granite in the major Archaen mineralization has taken place within these structures. Gold is Gl23 Don Mine, Selukwe shear zone at the Kwekwe Ultramafic Complex visible and finely disseminated within the host roek. The gold of the Midland Greenstone Belt. The gold is grains are <2 mm. The gold is bronze yellow in colour with a hydrothermal. metallic lustre (Figure 2-15). The host rocks are mafic greenstone, schist, The gold ore consists of milky quartz. Mineralization has occurred G47 Don Selukwe Mine serpentine, and granite in the major Archaen along the fractured zone within the quartz. Gold is associated withshear zone at the Midland Greenstone Belt. The some sulphide minerals (pyrite). Gold is visible and finely gold is hydrothermal. disseminated within the fractures of the host rock (Figure 2-15). The gold is hosted on felsites, mafic greenstone, schist, serpentine and granite on the greenstone The gold ore was greenish especially at the mineralization zone Gl24 Yankee Doodle Mine margin. The gold deposit is also hosted within where malachite is finely disseminated, along with pyrite orthe rocks of the Kwekwe Ultramafic Complex chalcopyrite. Gold was visible, likely la be finely disseminated of the Midland Greenstone Belt. The gold is (Figure 2-16). hydrothermal. 25 G92 The host rocks are amphibolite and ultramafic The gold ore consists of copper mineral (malachite), pyrite, and schist in the major Archaen shear zone of the tiny disseminated chromite with other sulphide minerals. Gold was Gadzema Mine Buchwa Greenstone Belt. The gold is finely «2 mm) disseminated, visible and localized throughout the hydrothermal. host rock. The gold ore sample has a reddish and occasionally yellowish The gold deposit is associated with the surface due to the presence of hematite and goethite respectively. 079 Lower Gwelo Mine hydrothermal process at the Midland The malachite is disseminated throughout the host rock. The gold Greenstone Belt. has mineralised within the quartz zone. Gold was visible and massive likely disseminated within the host rock (Figure 2-17) The host rocks are amphibolite and ultramafic The gold ore was greenish due to the presence of copper minerals schist in the major Archaen shear zone of the (malachite and azurite). The gold is associated with some of the G43 Gadzema Mine Buchwa Greenstone Belt. The gold is sulphide minerals like pyrite and galena. The gold is visible and hydrothermal. finely «2 mm) disseminated throughout the host rock. The gold ore sample consists of milky quartz. At the mineralized zone the quartz becomes smoky. The minerals presence in the ore The gold deposit is hosted at the rocks of the G45 Victoria Reef, Balabala includes hematite, galena and probably other sulphide minerals. Masvingo Greenstone Belt. The gold is visible, and massive, rather partly disseminated on the mineralized zone (Figure 2-18). The source and geological setting was not found The gold ore is reddish due to the presence of iron rich minerals. in literature. It can be presumed to have come The gold was visible and locally massive within the milky quartz. G72 Zambezi (?) from the northern part of Zimbabwe at the The gold in some places was disseminated. The gold grains range alluvial deposit associated with the Zambezi from <2 mm to 4 mm (Figure 2-19). river. ----- 26 2.3 Archaeological Artefacts The archaeological artefacts were soureed from the Zimbabwean eraton and were obtained from the Museum of Human Sciences (formerly Queen Victoria Museum) in Zimbabwe. The artefacts include the bangle fragments, the gold beads, the pellets, the foil fragments, the wire fragments and a tack fragment. The provenance details with the geographical coordinate system of the artefacts were received in the form of grid references. The descriptions of gold artefact samples are presented in Table 2-3 and Figure 2-20. The samples were given on the local grid in meters and were converted into proper latitude and longitude using ArcGIS software. The Mapungubwe gold artefacts samples were described in detail by Desai, (2001). In the summary of Desai's (2001) descriptions, the following was evident: (1) Mapungubwe artefacts were marked as M 1231 (A-F). (2) The artefacts types were mainly round gold beads starting from A to F. (3) The artefacts diameter was as follows: A = 3.5 mm; B = 3.2 mm; C = 2.1 mm; D = 1.9 mm; E = 2.1 mm; F = 3.4 mm. (4) The artefacts sizes were as follows: A = 2.2 mm; B = 1.5 mm; C = I mm; D = I mm; E = 1.3 mm; F = 2 mm. (5) The artefacts masses were as follows: A = 0.191 g; B = 0.099 g; C = 0.041 g; D = 0.037 g; E = 0.045 g; F = 0.20 I g. 2.4 Sample Selection A total of eighteen gold ore samples from South Africa were investigated and eleven from Zimbabwe. Not all the samples described here were analyzed because in some cases it was not possible to remove gold grains from the ore. The problem was probably due to the type of ore and the method used to remove the gold grains (see next chapter). The samples were not supposed to be destroyed or crushed, as 27 instructed by the donors. The samples which were excluded for these reasons are: Birimiam Reef, Birthday Mine, Gravellote Mine and De Kaap Goldfield in South Africa, as well as Hope Reef, Yankee Doodle Mine and Zambezi from Zimbabwe. The twenty-three artefacts from Zimbabwe archaeological artefact samples were all analyzed. 28 Table 2-3 Macroscopic descriptions of the archaeological artefacts collected in southern Africa. The Z number refers to Figure 2-20 and was used during the analysis. Terminology used for objects such as rings: OD = outer diameter, ID = inner diameter. Continued on page 30. Sample no. Colour Lustre Descriptions of the artefacts Diameter Thickness Length (mm) Mass (g) (mm) (mm) 4824 (Zl) Bronze yellow Earthly to A thin wire ring of gold, flattened on one 3.2 88 :::I 86 < 84 W== 82 I/) 80 78 76 75 80 85 90 95 100 Weighed Au value (wt %) Figure 3-3 SEM gold versus Weighed gold values (wt %) of an internal gold reference standard. 500000 ---- Ag wt. " :Ag107: " = 0.9898 400000 !. ~ 300000 ~:tI. Sr .. :zo •,!ju {•' .. 10 cp . ~ lO~ ~ -s - -- ... -- - D............0..uundE I...,." •0I.ftn, D.,*_O-wr_an.BEIWINftdD. ~ :' ~ I ',"_ ~l! --~ -ciJ 9 ~-- - Figure 4-1 Box and whisker plots presenting mean, mean plus standard error, mean plus 1 standard deviation, outliers and extreme values of all the data for the gold samples from the Witwatersrand Basin, Barberton and Zimbabwean greenstone belts. Continued on pages 63 and 64. 63 64 '..-~--~--~-~--~--, a...~.....0..MHI'td!IE I UUftaS) • OYatm tt. • [l- ~ . Il o'e" • --- --- Table 4-1. Continued. 4.2 Minor Gold Ore Deposits The data of the minor gold deposits, including Marabastadt, Gravellote and Knysna were compared with the three major ore provinces discussed in the previous section. Samples from Marabastadt have silver concentrations comparable with those of Barberton and the Zimbabwean greenstone belts (Figure 4-2). The 202Hg concentration is lower relative to the Zimbabwean samples, but comparable to those from Barberton samples, and even lower than samples from the Witwatersrand Basin (Figure 4-2). The 63Cu concentration is comparable to both the Barberton and the Zimbabwean greenstones and lower than those of the Witwatersrand Basin (Figure 4- 2). Figure 4-2 shows that a sample from Gravellote Mine has 107Ag concentration comparable to those of the Barberton and Zimbabwean greenstone belts, and lower relative to those from the Witwatersrand Basin. The 202Hg concentration is higher than samples from the Barberton and Zimbabwean Greenstone Belts, but comparable to those of the Witwatersrand Basin. The 60Ni concentration was not detected in the 65 Gravellote Mine sample. The 63Cu concentration is comparable to samples from the Barberton and Zimbabwean greenstone belts, and higher than most of those from the Witwatersrand Basin. Figure 4-2 shows that samples from Lydenburg, Goedeverwacht and New Chum- Pilgrim's Rest mines from the Sabie-Pilgrim's Rest Goldfield have 107Ag concentration higher than those of the Barberton and the Zimbabwean greenstones. The 202Hg concentration is higher than those from Barberton and Zimbabwean greenstones and comparable with those samples from Witwatersrand Basin. The two samples from Knysna Mine have silver concentration similar to those of the Witwatersrand Basin (Table 4-2, Figure 4-2). The 202Hg concentration is low compared with samples from the Witwatersrand Basin, but comparable to those from the Barberton and the Zimbabwean greenstone belts (Figure 4-2). The 60Ni concentration is lower than most of the Barberton and Zimbabwean greenstones, but comparable with those from the Witwatersrand Basin (Figure 4-2). The 63Cu concentration is lower than samples of the Barberton and Zimbabwean greenstones and comparable to those of the Witwatersrand Basin (Figure 4-2). 66 ~r---~------~------~------~-----. ~r---~----------------~----------------, - i• 'Uil :8 ,.. 'oa> = - .,.. L .,.-..,.-~f..-.-..----- •• -_~- •• --.-_--_,....,... .. -.-=._.-:- ~--------.J 3Or.-..-._-..O-.-...-_-l-Ii-.-l -..-..---------------------------, ~r_-------------------o.-W---t:-_-D--~I-..-..-.,·-O-------~Cl... ndo. 0Ubn IaHI,..., ,..., '50 lO'" '00 li II = 500 ·50 ·'00 c__",'-c"_":"_u~r-.-----~-•--..-.-..-..-.-..~..c..._..:._...c.._:_--~---,-..-..---.J _,_. ·'0111 L--..-.~..-. ,-.-.-..--~------._..:.._..~..c.._:_c_..:._--..~..-,.-.-.,-.-,,-"--*-' K~ .. ,..r---~------~--------~--------~------, .. .. elO rIO 'a = = .s L-~.......... ............... --',....., ..... ~ ~L--..-..-.n.-II-..----• .-..-..-..-..-..-..--,-..-.,-..-.._-.---~-------~ ~0r-------------------.-D..-.-..-.0.-...-..-.,-.--I----------_,o..en ~r-.-.-.-..-.-.-----------------------------,.... wnndO • Cl .... 0....... 1WIt-aD. 0uIIIlft SB 'a o -0 58... ... 'Q a.' _. o.. c___,_ ~ ~ ~ __' .,.. L-... ~------.-.~_-----_---,-, .~...- ..-:.:..-.:----_~--------' ,.. .......... ..,..... .... .....,.... ... at ~ Figure 4-2 Box and whisker plots presenting mean, mean plus standard error, mean plus 1 standard deviation, outliers and extreme values of all the data for the gold samples from Pietersburg and Murchison Greenstone Belts, Sabie-Pilgrim's Rest Goldfield and Knysna mines. Continued on page 67. 67 u rD...-...-..M..-,.;.:O..-...--sE--=r,--...-......,--.,--o.-.-------------, ..,--------------------- __------------------,a .... 0YtMtSE I YtMJSO • o..n •• .- 1.1 r e Cl lO 2. Ir o ! ! " IS • .2 •• .••. ••~----------------------------------------~ ..~----------------------------------------~ '000,----------------------------------------, 2..0.0,----------------------.-----------------,.... D MI..-0 a....antSEI MNn1SO • a....n ..o.o. •20 .00 lIlOO f !f .. 2000 R o '000 .o. 20 -.000 ~~----------~---------- .... PMludlurg lIIurchkoft ~w. ......-,-------------~Rut Knysna L_ .-..-M-'-~-U~-----.-.-~--.~---~----~-.-.-----~~~~~~----------~ 000,--------------------------------------, •2000r----------------------------------------, ... ..... 1000 '00 .... 200 2000 -2000 ..... ,---- .... .DM.t.a.n.O.. -.-.----- __-------------------------, .00,---------------------------------------,MtauseIWtaouSOoOullliHI .... lO .c- ..o o.. ii 2000 .. -0- 20 -2000 ....., L,--::-- ---::---:--:--::-- -' 'it..nbw, s.IMt Pia,tWt. ,.. .. - Figure 4.2. Continued. 68 4.3 Archaeological Artefacts The data of the artefacts recovered from the Zimbabwe and Mapungubwe sites were compared to the major ore provinces presented in Section 4.1. The twenty-three artefacts from the Zimbabwean plateau have silver concentration comparable to those of the Witwatersrand (Table 4-3, Figure 4-3). The 63Cu concentration is low compared with those samples from the greenstone belts and comparable with most of the Witwatersrand samples (Figure 4-3). Figure 4-3 shows elevated 66Zn concentration compared with the three major ore provinces. The 1880s and 195Ptisotopes have higher concentrations compared with those from the greenstones and some samples have concentrations comparable to some samples from the Witwatersrand Basin (Figure 4- 3). The artefacts from the Zimbabwean plateau have 1880s concentration more elevated than those of the Witwatersrand ore samples. The Zimbabwean artefacts have lower 202Hg concentration compared with those of the three major provinces (Figure 4-3). Table 4-3 and Figure 4-3 show that the silver concentration of artefacts analyzed from Mapungubwe is lower than that of the Witwatersrand Basin and comparable to that of the greens tones. The 63Cu concentration is comparable to those from the Witwatersrand Basin, but slightly low. 66Zn and 1880Scps were not detected. Like the Zimbabwean artefacts discussed above, Mapungubwe artefacts have 195Pt concentration elevated above those of the entire natural samples. The 202Hg concentration of Mapungubwe artefacts is also lower than that of the Zimbabwean artefacts and the natural samples. 69 ~r-------------------~--~~~--~, 12 : ~ _auSE I IIIuMSD • a..n 700 .,..... ... ",.22 • 0"IL ... li 0" :I " "" lO '00 -- ~r---------------------------------, *r--------------------------- e."""-OWNM.SEIlllta\tSO-_-----,eo ... 50 .. ... 20 lO '00 .10 '- --= -,---_-,-- ~ __ _.J ·'00 '---------=---~_----,,----,M-a-p-u-,..-. _-_--------' ZINb..... M.-glfbwtl OOOr------------------.------------------, ~r_-------------------..-.-.-------------,... 0"'0 MuMSE I"-SO • 0ueIIft .. a ..... ..O.... -sEr ..umSO.OUIIitn ~ . .c,... 7990 .. Ci) u.. 5990 • :8 3990 • .. - 1990 , -10 IlL • • .. • • 0 1000 2000 3000 4000 5000 202Hg(1 05)1107 Ag Figure 4-9 A ratio plot for 56Fe/107 Ag and 202HglI0A7g for the gold samples from the Witwatersrand Basin, Barberton and Zimbabwean greens tone belts. The ratio plot in Figure 4-9 shows that there are low 56Fe/107 Ag ratios for the gold samples from the Witwatersrand, compared to those of the greenstone belts. The greenstone belts have low 202Hg/I07 Ag ratios compared to the Witwatersrand. • Witwatersrand • Barberton Zimbabwe I 10000 •• 1000 • -• .c 11. ~ 100 I- .•. ~.If- .. "• 10 • 10 100 1000 10000 2"Pb Figure 4-10 Bivariate plot of 206Pbvs 207Pbfor the Witwatersrand, Barberton and Zimbabwean gold samples. 77 The lead isotopes were highly variable for the Barberton, Zimbabwean greenstone belts and the Witwatersrand samples (Figure 4-10). It was impossible to distinguish between the Witwatersrand and greenstone samples because of the variability. 4.4.2 Minor Gold Ores The chemistry of the minor gold samples was compared to the one of the major gold samples. The samples were not used potentially to source the archaeological artefacts because of the small number of samples involved. Figure 4-11 shows that the two samples from Marabastadt in the Pietersburg Greenstone Belt have comparable signatures to the other greenstone belt samples. In Figure 4-12, a sample from Gravellote of the Murchison Greenstone Belt indicates a comparable signature to the greenstone belts as well. From the three samples presented in the Sabie-Pilgrim's Rest Goldfield shown in Figure 4-13, two samples have a similar signature to the greenstone belt samples. The other sample shows a completely different signature but near the greenstone belts field. Figure 4-14 shows the two samples from Knysna in the Cape Supergroup also are comparable to the greenstone belt samples. 78 U t.l .. •.. + + J • +'i' + + •.. 'A U .1.5 •• .i.s G. es z.s ""Ot Figure 4-11 Multivariate correspondence analysis at 95 % confidence interval for the gold samples from Witwatersrand Basin (red cross), Barberton (light green diamonds) and Zimbabwean (blue boxes) greenstone belts and Marabastadt samples (pink solid boxes) from Pietersburg greens tone belt. Axisl = 0.5 eigenvalue and 58 % of total. Axis2 = 0.14 eigenvalue and 17 % of total. u .u. GA J • G.A. ·u .U •• .t5 .. .. .. ""Ot Figure 4-12 Multivariate correspondence analysis at 95 % confidence interval for the gold samples from Witwatersrand Basin (red cross), Barberton (light green diamonds) and Zimbabwean (blue boxes) greenstone belts and Gravellote (gl) sample (pink solid box) from Murchison greenstone belt. Axisl = 0.5 eigenvalue and 55 % of total. Axis2 = 0.13 eigenvalue and 16 % of total. 79 U 12 ... 0.' ! 0 + + + + + + 0.' U ·12 .U ~~ .,~ .,0..~' , ,~ ,~ Figure 4-13 Multivariate correspondence analysis at 95 % confidence interval for the gold samples from Witwatersrand Basin (red cross), Barberton (light green diamonds) and Zimbabwean (blue boxes) greenstone belts and Sabie-Pilgrim's Rest Goldfields (sI, sn and sq) samples (pink solid boxes). AxisI = 0.4 eigenvalue and 48 % of total. Axis2 = 0.2 eigenvalue and 23 % of total. '.6 12 0.8 0.' J 0 + + "I- +' + + u U -12 .U ~~ .,~ .,.01.0' ' r.s Figure 4-14 Multivariate correspondence analysis at 95 % confidence interval for the gold samples from Witwatersrand Basin (red cross), Barberton (light green diamonds) and Zimbabwean (blue boxes) greenstone beIts and Knysna (kl and k2) samples (pink solid boxes). AxisI = 0.14 eigenvalue and S9 % oftotal. Axis2=0.13 eigenvalue and 17 % oftotal. 80 4.4.3 Archaeological Artefacts Table 4-5 below summarizes the major dissimilarity of the archaeological artefacts from Zimbabwe and Mapungubwe in South Africa. The distinctions between the artefacts were based on either high or low concentration in cps of the following isotopes: 66Zn, 75As, 107Ag, 202Hg and the Pb isotopes. Table 4-5 Summary of the trace elements signatures of gold artefacts samples from Zimbabwe and Mapungubwe. Major distinctions Artefacts High concentration Low concentration (cps) (cps) ')As 66Zn 2U4Pb LuóPb Cl.) 107 ~ Ag ..0 LU/Pb C1:l ..0 E LULHg LUlSPb N 75As 66Zn 204Pb Cl.) ~ LvoPb ..0 107Ag 0 0Jj LU/Pb=0 0.. 208Pb 202Hg ~ The combination of the 63Cu, 66Zn and 202Hg cps in Figure 4-15 indicates the three different groups of the artefact samples. The first group is of the Mapungubwe artefacts, which is clustered into a single group discerned by high 63Cu and low 66Zn concentrations (Figure 4-15). The second group is formed by the Zimbabwean artefacts samples, which can be divided into two groups. The first group is on the intermediate region of the ternary plot. The group consists of a wide range of low 202Hg to high 202Hg concentrations samples. The second group is formed by the two 81 samples, which are characterized by high 66Zn, low 202Hg and low 63Cu concentrations. 63Cu 50 _________________________________________________ 50 .,.. 50 66Zn 202Hg Figure 4-15 A ternary plot of 63CU, 66Zn and 202Hgof gold samples from Zimbabwe (z's and red cross) and Mapungubwe (rn's and solid purple squares) in South Africa. Normalised data in Table 4-3. When the artefact samples are plotted against the major gold deposits of Witwatersrand, Barberton and Zimbabwean samples (see Figure 4-16), the following can be seen: (1) Mapungubwe samples are comparable to the majority of the Barberton samples. (2) The two samples with high 66Zn but low in 202Hg and 63Cu concentrations from Zimbabwean artefacts, are plotting with the majority of the Zimbabwean ore samples. (3) There is a wide group of the Zimbabwean artefacts 82 samples plotting next to the majority of the Witwatersrand samples. (4) The intermediate Zimbabwean artefacts plot with one sample from Witwatersrand Basin and one from Barberton Greenstone Belt. 6JCu o 50 50 o \0 + \°EPr \ DO ° ° o o DO OD o> 66Zn 202Hg Figure 4-16 A ternary plot of 6JCu, 66Zn and 202Hg of gold samples from Witwatersrand Basin (red cross), Barberton (light green diamonds) and Zimbabwean (blue boxes) greenstone belts with Mapungubwe (light blue triangle) and Zimbabwe (brown circles) artefacts. Normalised data in Table 4-3. The multivariate correspondence analysis presented in Figure 4-17 also shows that the Zimbabwean artefacts differ from the Mapungubwe artefacts. The Zimbabwean artefacts are divided into two groups as shown in Figure 4-17, but with enough samples for each group here. Figure 4-17 shows that the small group, marked yellow in the correspondence analysis plot, can be identified as having 50 % of 202Hg, about 83 70 % of 66Zn and about 20 % of 63CU concentrations (Figure 4-16). The large group can be identified as having less than 50 % of the metals (Figure 4-16). 0.6 I5 za II J,,6 zf 0.4 zb 0.2 '~" -0.2 -04 -0.6 ·0.8 Ifn -1.8 -1.5 .12 -O.S -0.6 -0.3 0.3 0.6 Axis 1 Figure 4-17 Multivariate correspondence analysis for the gold artefacts from Zimbabwe (z's) and Mapungubwe (m's). The Ward's Method (Ward, 1963) of clustering in Figure 4-18 also indicates the three major groups of the artefacts. The Mapungubwe artefacts form a group while the Zimbabwean artefacts are also divided into two groups. Only two samples from Zimbabwe artefacts are closely related to the Mapungubwe artefacts. The platinum group isotopes were mainly extremely low in concentration, except the 1880S and 195Pt for the Zimbabwean and Mapungubwe artefacts (Figure 4-3). The details of this subsection will be discussed in the following chapter of interpretation and discussion of the results. 84 2 4 6 8 ! 10 12 14 16 18 20 0 3 6 9 12 15 18 21 24 27 Figure 4-18 Ward's method for cluster analysis for the gold artefacts from Zimbabwe (z's) and Mapungubwe (m's). • Wltwat.r.rand _ Barberton Zimbabwe)( Art.,.ct, 21mbabw. J: "'.pungubw •• rt~ 100000 10000 -. .. --.: - --- --.. '0 1000 ;Ol X • r- • • +• c lOO • T ~u.. lO O~--------_---------- ~ L o 10 100 1000 '0000"'Hg(l 0')110' Ag Figure 4-19 A ratio plot for s6Fe(10s)/lo7 Ag and 202Hg(lOs)/107Ag for the gold samples from Witwatersrand Basin, Barberton and Zimbabwean greenstone belts, as well as Zimbabwean and Mapungubwe artefacts. 85 The ratio plot in Figure 4-19 shows that the chemistry of the archaeological gold artefacts differs completely from the natural gold analysed. This is based on the low Hg/ Ag ratios of the artefacts compared to the natural gold from Witwatersrand, Barberton and Zimbabwe samples. 86 5 DISCUSSION 5.1 Chemical Signature of the Major Gold Deposits One of the prerequisites in trying to trace the geological source of artefacts is to determine whether the different gold deposits exhibit different chemical signatures. Chisholm (1979) suggested that, on the basis of the major elements in gold only, it is difficult to distinguish between deposits of the same geological association. A clear chemical difference between the Witwatersrand Basin and greenstone belt gold was observed in this study, based on the analysis of major and trace isotopes in gold. The geochemical difference observed may represent different histories of the gold mineralization process. To recap from the previous chapter, isotopes which discriminate between the Witwatersrand and greenstone gold, include 56Fe, 58Ni, 63Cu, 66Zn, 107Ag and 202Hg (Table 4-4; Figures 4-4 to 4-9). The gold sample set is clearly not large enough yet for detailed work and only the most general comparisons and statements can be made. There are distinct chemical characteristics associated with individual goldfields (e.g. Hayward et al., 2005), but the current data will not be able to resolve them. Large numbers of samples are needed in order to distinguish between the individual gold mines with statistical confidence. For example, eight samples from the Witwatersrand Basin can only give an indication of the geochemical characteristics of gold in the Witwatersrand Basin as a whole. Also, only eight and ten samples were available from the Zimbabwean and Barberton greenstone belts, respectively. The geochemical difference between the Witwatersrand and greenstone gold is presented in Figures 4-4 and 4-5. These major distinctions are based on high 87 concentrations of 56Fe, 58Ni, 63Cu isotopes in native gold from the greenstone belts compared to the Witwatersrand Basin, and the higher concentrations of 107Ag and 202Hg isotopes in native gold from the Witwatersrand Basin than the greenstone belts summarized in Table 4-4. The multivariate correspondence analysis diagram also shows a clear distinction between the Witwatersrand Basin and greenstone belts (Figure 4-7). The distinction could be based on the mode of occurrences of gold. The gold mineralization associated with the granite-greenstone terranes located in the Kaapvaal and the Zimbabwean cratons occurs in dilational veins, within brittle to ductile shear zones that are controlled by hydrothermal fluids. The most likely reason why the Barberton and Zimbabwean gold have similar geochemical signatures is because of similar geological histories (Figure 4-7). The gold in the Witwatersrand Basin is generally hosted in the Archaean meta- sedimentary rocks in the Central Rand Group within conglomerate units (Frimmel and Minter, 2002; Hayward et al., 2005). These differences in mode of occurrence could explain their distinct chemical signatures, but further investigation is beyond the scope of this study because of the greater number of samples needed. Chisholm (1979) noted that most of the base metal gets leached from detrital gold in the descending order of leachability: ironc-nickeb-copperxsilver. It is expected therefore that the detrital gold composition could have low concentrations of some of these base metals. The results of the current study indicate fairly high concentration of 56Fe, 58Ni and 63Cu isotopes in most of the greenstone belt samples compared to those from the Witwatersrand Basin. The results of silver do not support the expected leaching order (Table 4-4, Figure 4-4 to 4-6), if the Witwatersrand gold was derived from the greenstone belts. 88 The silver concentration allows discrimination between the samples from greenstone belts and the Witwatersrand Basin. The silver concentration obtained from the greenstone belts in this study is within the range of 1.0 I to 4.6 wt % (Table 4-1). The silver concentration range obtained from the Witwatersrand Basin in this study is 4.2 to 9.3 wt %, within that previously reported in the literature, of 2 to 20 wt % (von Gehlen, 1983; Erasmus et al., 1987; Frimmel and Gartz, 1997; Hayward et al., 2005). The relatively high silver concentrations in the Witwatersrand Basin have implications for the various mineralization models proposed in the literature (e.g. von Gehlen, 1983; Phillips, 1987; Phillips and Meyer, 1989; Frimmel and Minter, 2002). The Witwatersrand Basin silver concentrations do not show the expected characteristics of either alluvial or hydrothermal native gold as reported in the literature. For example, a low silver concentration is expected in placer gold, probably due to the leaching of silver during the transportation processes (Chisholm, 1979; Nakagawa et al., 2005). If alluvial gold from the Witwatersrand Basin originated in the greenstone belts, simplistically it would be expected to have a lower silver concentration than gold from the greenstone belts. This study however shows this not to be the case. The majority of the samples analyzed from the Witwatersrand Basin have higher concentrations of silver than gold from the greenstones. The elevated concentrations of silver in the Witwatersrand Basin may therefore offer support for the modified placer model (Frimmel and Minter, 2002; Frimmel, 2005; Hayward et al., 2005), implying that after deposition the alluvial gold was remobilized and redistributed by hydrothermal fluids. Hence the silver enrichment in 89 the Witwatersrand gold may have been due to post-depositional hydrothermal activity. The source and the chemistry of the mobilizing fluids have been studied, but the reported findings do not support the metamorphic origin of a gold transporting fluid or of the gold itself (Frimmel et al., 1999). What needs to be resolved is whether this metamorphosing fluid was enriched in silver when entering the Witwatersrand Basin, or whether the host rocks were enriched in silver, so that when the gold and other metals were carried in solution, the silver that was originally in the host rocks was also liberated and recrystallized with gold. In order to answer these questions, a comprehensive study is needed of the chemistry of the fluids that were responsible for remobilizing gold, and the chemistry of the host rocks. This is beyond the scope of this study. The Archaean granite-greenstone terranes that hosted most of the gold deposits are older than the Witwatersrand Basin. Most of the gold deposits hosted in the greenstone belts are structurally controlled and have been subjected to post- depositional tectonic and metamorphic events, which may have impacted on the chemistry of the gold. For instance, in the Zimbabwean Craton, Campbell and Pitfield (1994) have suggested that metamorphic events after the first phase of gold mineralization could have caused silver to leach, resulting in a low concentration in the native gold from the greenstone belts. The distinguishing factor between the greenstone belts and Witwatersrand gold chemistry could be based on the subsequent tectonic and metamorphic events in both regions. The placer model for the Witwatersrand Basin could be supported if tectonic events occurred in the greenstone belts to modify their residual gold signatures after 90 alluvial gold probably originating in the greenstone belt, was deposited. Subsequent leaching of silver in the greenstone belt deposits could have occurred, resulting in the incompatible silver concentrations with the Witwatersrand Basin. Although the gold from the Witwatersrand Basin is associated with sulphides (e.g. Hayward et al., 2005), the results in the current study show evidence of 63Cu depletion from the native gold compared to the Barberton Greenstone Belt (Figure 4- 4), which could suggest an ultimate placer origin of the gold. In the hydrothermal processes copper and nickel tend to have an affinity with the sulphides, and are commonly found in minerals like chalcopyrite, arsenopyrite, etc. Copper and nickel are therefore expected to be in higher concentrations in most of the hydrothermal gold than in placer gold (Guerra, 2004). The depletion of copper and nickel noticed in samples from the Witwatersrand province does not necessarily reflect the signature of the primary source of gold, which could suggest the loss of these metals during local oxidizing conditions and could support the suggestion made by Dimroth and Kimberly (1976) and again Dimroth and Litchtblau (1978). But this explanation is incompatible with evidence that the Archaean atmospheric conditions were anoxic (Frimmel, 2005 and references therein). Mercury is a volatile element, and can be expected to leach from placer gold. As with silver, the results of the current study indicate that the 202Hg concentration is higher in the Witwatersrand Basin than in the greenstone belt, which is unexpected and does not support a simple placer origin of the Witwatersrand gold (Figure 5-4). The source of mercury in the Witwatersrand Basin has been suggested to be enrichment from the surrounding sediments as a result of mobilizing fluids, which also supports the modified placer gold theory (Oberthi.ir and Saager, 1986). 91 The distinction observed between the Witwatersrand and greenstone belts gold in the present study can be summarized. The Witwatersrand gold has elevated silver concentration and 202Hg concentration compared to the other base metal isotopes. The greenstone gold has elevated counts of base metal isotopes except for silver and 202Hg. The statistics indicate a closer association of gold chemistry between the Barberton and the Zimbabwean greenstone belts. 5.2 Chemical Signature of the Minor Gold Deposits As explained above, the native gold samples analyzed in the current study do not show clear distinctions between individual gold deposits, but chemical distinctions are evident between the Witwatersrand and greenstone belts. In comparison with the results obtained from the major gold provinces, gold samples from the Marabastadt Mine have silver concentrations and 202Hg concentrations comparable to those of the Barberton and the Zimbabwean greenstones (Figure 4-11). The gold in Marabastadt Mine was deposited within quartz veins. Additionally, the Marabastadt deposit is associated with sulphide replacement by carbonate alteration. This is a typical mineralization, which is common in the hydrothermal or the greenstone belt gold (de Wit et al., 1992; Ward and Wilson, 1998). This is evident in Figure 4-11 where multivariate correspondence analysis was used to show that the gold from Marabastadt is hydrothermal. Three gold samples that come from the Sabie-Pilgrim's Rest Goldfield have signatures resembling both the Witwatersrand Basin and the greenstone belt (Figure 4-13). The gold nuggets from Lydenburg, which were described as being associated with surface enrichment (Ward and Wilson, 1998), have lower silver and mercury 92 cps, compared to the New Chum Pilgrim's Rest gold hosted in fine-grained gossans. The Geodeverwacht Mine is associated with oxidized gold-quartz-carbonate sulphide veins and has high silver concentrations, but low 202Hg concentration due to the oxidizing conditions in the gossan. Both the Geodeverwacht and New Chum Pilgrim's Rest gold deposits have silver concentrations comparable to those of the Witwatersrand Basin. The gold samples from Knysna are hosted in the Table Mountain sandstones of the Cape Supergroup. The two samples are related to the greenstone belt signatures because of the elevated counts of 56Fe, 58CO,60Ni, 63Cu and 66Zn (Figure 4-14). The multivariate statistics plot also shows the gold from Knysna Mine closely associated with the greenstone belts (Figure 4-14). Gold mineralization here probably occurred as a result of hydrothermal processes. It has been reported that the gold is associated with quartz veins, which contain sulphide minerals like pyrite, galena and marcasite (Hammerbeck, 1976). The 202Hgconcentration is comparable to that of the Barberton and the Zimbabwean greenstones. Before discussing the archaeological artefacts in the next section, let us recap for the natural gold. There is a geochemical distinction evident in the signatures that separate the Witwatersrand gold group from the greenstone belt group. This distinction was observed in the statistics of the elemental analysis and in comparisons of the elemental signatures of 56Fe, 60Ni, 63Cu, 66Zn, 107Ag and 202Hgin native gold samples. The chemical history of anyone deposit may be complex, with various elements leached or enriched by post depositional processes. The post depositional processes may alter the original signatures of any particular deposit or occurrence of gold, 93 making it impossible to establish its ultimate geological origin on the basis of the available results. 5.3 Chemical Signature of the Archaeological Artefacts The analysis of archaeological gold artefacts was compared with the analyses reported by Anglo American Research Laboratories (AARL) and Desai (2001). The AARL research and the current research were done in order to understand the following: (a) whether gold used in these artefacts was soureed from a single gold deposit; (b) whether they were derived from more than one source as a consequence of complex trading routes and exchange systems at that time; (c) whether gold from particular sites was predominantly alluvial or reef mined, and (d) whether systematic alloying with silver or copper was practised. The AARL fingerprint results were summarized by Desai (2001). There were no data tables available to be compared with the current study, in which only samples from Mapungubwe and Zimbabwe were analyzed. The table presented in Desai (2002) was based only on the presence and absence of the elements highlighted in section 1.2. The archaeologically older Mapungubwe samples than the Thulamela samples (Miller et al., 2000) from the Limpopo Basin clustered into two groups, with one showing the presence of the rare earth elements, whereas the other group showed the presence of mercury. The common characteristics of the Mapungubwe groups were the absence of strontium and barium, and the presence of the platinum-group elements. The single Bosutswe gold artefact from Botswana showed only the presence of bismuth and barium and had a completely distinct signature (Miller et al., 2001). There were two groups of gold artefacts from the Zimbabwean plateau. One had similar characteristics 94 to the Mapungubwe samples, characterized by the presence of strontium, mercury, rare earth elements, platinum-group elements and barium. The other group from the Zimbabwean plateau was characterized by the presence of copper, silver, mercury, and lead. The archaeologically younger Thulamela samples from Kruger National Park showed the presence of all the elements outlined above, with a strong correlation to one of the Mapungubwe signatures, perhaps suggesting a common source (Desai, 2001; Miller and Desai, 2004). The AARL study scanned the entire periodic table and signatures were compared on the basis of a large number of elements (Grigorova et al., 1998; Miller et al., 2001). In the current study the comparisons are based on only a few isotopes, including 56Fe, 107Ag, 202Hg, 66Zn, 63Cu and 58Ni considered most likely to define distinct signatures. The rare earth elements were found below the lower limit of detection. The archaeological gold artefacts analyzed in this study came only from Mapungubwe and Great Zimbabwe, dating from the early Ilth to early 14th centuries AD, and the mid- 14th to mid-16th centuries AD respectively (Miller, 2002). In the current study, there are clear distinctions in the archaeological gold artefact signatures from Mapungubwe and Zimbabwe, based on 107Ag, 75As 66Zn, 202Hg and the Pb isotopes in cps (Table 4-5). Strontium, barium and the rare earth elements were not detected. These are the rock forming elements and are unlikely to bond with gold as they have no siderophile tendency (Levinson, 1974). The presence (or absence) of strontium, barium and the rare earth elements in gold artefacts therefore can give information about silicates (which may be secondary) associated with the gold, but not about the gold itself, and should not be used for classification. The Zimbabwean artefacts have higher 107Ag and 66Zn concentrations compared to the low 95 concentrations of these isotopes in Mapungubwe artefacts. There are also overlaps of the Zimbabwean and Mapungubwe artefacts in terms of platinum and mercury. The Zimbabwean artefacts are chemically distinguishable from the Mapungubwe artefacts (Figures 4-15; 4-17 and 4-18). In the current study, the Zimbabwean artefacts are clustered into two distinct groups, similar to the groups suggested by Desai (200 I). One of the Zimbabwe groups, represented by few samples, was distinguished based on the low concentrations of silver, zinc, osmium and mercury, and is similar to the Mapungubwe group. This implies that the earlier Mapungubwe material may have been soureed from the Zimbabwe plateau. The later Great Zimbabwe material probably included material from a diversity of sources, so it is not surprising that there is a wider diversity of signatures in the Great Zimbabwe material (Grigorova et al., 1998; Miller et al., 2001). Compared to the natural gold samples analyzed in this study, the Mapungubwe gold artefacts in this study were found to be chemically most similar to the gold from the Barberton and the Zimbabwean greenstone belts. It is possible that the gold at Mapungubwe and Thulamela was soureed from the nearest surrounding greens tone belts, which include the southern Zimbabwean greenstones, the Barberton, Giyani, Murchison and Pietersburg greenstone belts in South Africa. It would be premature to trace the source of gold at Mapungubwe based on the few samples analyzed in this study, but it adds to our understanding of the type of source that could be expected. The suggestion was made by Desai (2001) that the Mapungubwe gold artefacts might be reef mined and cannot be soureed to the alluvial gold ore deposits in the Transvaal vicinity because of their relatively high silver concentration. To resolve this question, alluvial gold samples need to be analyzed and compared to the Mapungubwe samples. 96 Some of the Zimbabwean artefacts are chemically similar to the gold from the Witwatersrand Basin (Figure 4-16), based on the 202Hg, 63Cu and 66Zn concentrations. It is extremely unlikely that the Witwatersrand Basin was the source of this gold. There is no archaeological evidence of mining from the Witwatersrand reefs, which is not surprising given that the gold was in finely disseminated form and not visible in quartz veins, as on the Zimbabwe plateau (Miller et al., 2001). There are also numerous archaeological gold mines recorded on the Zimbabwe plateau (Swan, 1994), so it is more likely that gold was traded to the south rather the north. This leaves the chemical similarity of the Zimbabwe archaeological material to the Witwatersrand gold as an unresolved archaeological puzzle. Statistically, there is no chemical similarity between the archaeological gold and the natural gold samples (Figure 4-17). The clear distinction between these two different sets of gold samples may have occurred as a result of the processes involved during the extraction and processing technology of the archaeological gold, or may be due to archaeological sources not represented in the modern samples. The southern African archaeological gold fabrication technology has been well documented by Miller and Desai (2004). The artefacts they studied had silver concentrations ranging from 2 % to 12 %, which did not correlate with the artefact type. The metals were cold worked and showed no evidence of soldering or welding. Miller and Desai (2004) ascribed the silver concentrations to the natural source(s), but as observed in the current study the silver concentrations in the Zimbabwean artefacts are comparable to those of the Witwatersrand Basin. If the Zimbabwe artefacts are truly derived from local material then silver may have been added. On the other hand, there is no recorded silver source in the southern 97 African archaeological record (Miller, 2002). Because of its relative volatility, silver is more likely to be lost than concentrated when melting gold in an open crucible. At present the high levels of silver in some of the Zimbabwean artefacts cannot be explained, and this remains a provocative issue for future investigation by archaeometall urgists. In Table 4-3, the high platinum-group elements in all the archaeological artefacts also cannot be explained easily, although they could point to an alluvial source in proximity to the Bushveld complex. During melting of gold, the heating process does not affect elements like platinum and osmium with high melting points, whereas volatile elements like mercury and copper may have evaporated. The low 202Hg concentration observed in most of the gold artefacts from Zimbabwe and Mapungubwe may be the result of melting under oxidizing conditions in an open crucible (Figure 4-19). Miller and Desai (2004) suggested that the metals were cold worked and the artefacts show no evidence of soldering or welding. The low mercury concentrations found in this study do not support Miller and Desai (2004) model, as the gold artefacts was heated (as least sometimes) during the process of manufacture, and has lost mercury as a result. The elevated 66Zn concentration (see Figures 4-15 and 4-16) in the Zimbabwean artefacts is also a provocative puzzle. There is also no record of indigenous zinc production in pre-colonial southern Africa. A nodule of zinc-containing alloy found at Mapungubwe was probably imported from India, via East Africa (Miller, 2002). The 66Zn in the Zimbabwean artefacts could, however, be due to melting gold in crucibles previously used to melt imported brass. 98 In summary, there are unresolved problems explaining the elevated platinum-group elements, silver and zinc levels in the archaeological gold samples. Until these have been addressed by analysis of natural alluvial samples for comparison, the question of possible silver alloying in the archaeological material has to be left open. There are also inherent difficulties in tracing gold from any particular archaeological site to source. These include the possibility of mixing gold from multiple sources, recycling, contamination in melting and trade of items (Grigorova et al., 1998; Miller et al., 200 1), as well as the need to have sampled all possible sources for comparison. For example, in the case of Great Zimbabwe it is likely that mixing occurred, because of the diverse potential sources of gold accessible to this powerful trading centre (Miller, 2002). By comparison at Mapungubwe and Thulamela, which were smaller centres, it is likely that gold originated from fewer or even only one source. This seems to be borne out by the greater diversity of elemental signatures from Great Zimbabwe. Taken together, these problems make it impossible to source the gold artefacts directly to their actual gold ore deposits on the basis of the information at hand. The current study also does not enable us to state clearly if gold from any particular site was soureed in any particular region, because the chemical signatures of the archaeological gold appear to be quite distinct from those of the natural samples currently available for comparison. This is a provocative outcome of the comparative study, and an obvious target for further research. Large numbers of samples are still needed to be used successfully in making statements based on the geochemical signatures with statistical confidence. 99 6 CONCLUSIONS AND RECOMMENDATIONS The use of the LA-ICP-MS technique to trace the geological source of archaeological gold has not been successful. There were analytical technique limitations involved; the gold sample size varies, thus making it difficult for the selection of a consistent protocol; some samples were too small and were not analyzed. For a successful comparison of the data measured by the LA-ICP-MS with the data on various elements already reported in the literature, there is a need to set up in-house standards for various elements for analysis. The gold compositional data comparison has been found useful for discriminating between gold samples from the greenstone belts (hydrothermal origin) and the Witwatersrand Basin (modified placer origin) (Table 4-4; Figures 4-4 to 4-9). The major distinctions are elevated concentrations of 56Fe, 60Ni, 63CUand 66Zn isotopes of the greenstone belts, compared to the Witwatersrand Basin, with the converse true for the 107Ag and 202Hg in the Witwatersrand Basin when compared to the greenstone belts. The comparison of geochemical data has highlighted some important questions for future research in both geological and archaeological scientific fields of study. The concentrations of silver and other elements in gold from the Witwatersrand Basin and the greenstone belts have raised the following questions. Does the relatively high silver concentration in the Witwatersrand Basin reflect the original signature of the source? 100 Was the silver concentration from the host rocks and was it enriched as a result of hydrothermal fluids responsible for remobilizing gold within the Witwatersrand Basin? Could the relatively low silver concentration in the greenstone belts be a result of post depositional tectonic and metamorphic events? The platinum-group isotopes in native gold samples from the major gold provinces were found at low concentrations. The low concentrations in some platinum-group isotopes in the Witwatersrand Basin probably occurred as a result of leaching from the gold and could have been concentrated around the vicinity. In order to understand this, there is a need for further investigation through the study of more samples. The question relating to the gold mineralization within the Witwatersrand Basin and the surrounding greenstone belts cannot be answered by the data presented in this study. There is a need for more samples to be drawn from the Witwatersrand Basin and the greenstones. The 63Cu and 60Ni concentrations in the Witwatersrand Basin are low when compared to the concentrations in the greenstone belts. This could be the result of local oxidizing conditions in the Witwatersrand Basin, which must have taken place long after deposition because the Archaean atmosphere generally is thought of as anoxic (Frimmel, 2005). But the oxidizing rationale could be support for some suggestions of less oxygenated atmospheric conditions in the Archaean (palmer et al., 1987). The issue of the Archaean atmospheric conditions within the Witwatersrand Basin cannot be addressed by data presented in this thesis. 101 The minor gold deposits cannot be soureed directly to the major gold deposits but their chemical characteristics can be associated with either the greenstone belt or Witwatersrand gold type. The gold samples from Marabastadt Mine, which was located in the Pietersburg Greenstone Belt, were successfully related to the greenstone belt gold type. The two gold samples from the Sabie-Pilgrim's Rest Goldfield were related to the Witwatersrand Basin type. However, the conclusion that they could be traced to the Witwatersrand Basin as their source would be premature. The chemistry of the gold nugget sample from the Sabie Pilgrim's Rest Goldfield was found to be similar to that of the greenstone belts, thus making it different from the other two gold samples. Obviously, more samples are needed to characterise the range of variation in this material. The chemistry of gold from Knysna was found to be similar to the greenstone gold type. The gold from Knysna is hosted in the Table Mountain sandstones, but cannot simply be placer gold. Hydrothermal processes could have been responsible for the deposition or reworking of gold in Knysna. The compositional data was not found to be very useful in sourcing archaeological gold artefacts to their ultimate source(s). Alluvial gold samples need to be analysed, for comparison with archaeological material. In this study, there are chemical distinctions observed between the Zimbabwean and the Mapungubwe artefacts. The Zimbabwean gold artefacts have higher 107Ag, 66Zn, 1880S and 195Pt concentrations than the Mapungubwe gold artefacts. Based on the current isotopic data, the gold chemistry of all the archaeological artefacts was statistically different from the natural gold samples. The difference in artefacts when compared to the natural gold signatures could be as a result of the following: (l) mixing gold from multiple sources; (2) recycling of the materials used; (3) contamination in melting or 102 manufacturing, which affected the composition (e.g. Hg boiled off); (3) trade of ore or items from more distant sources not represented in the present study. The main requirement for future research is to broaden and deepen the populations represented in the present study. This means the inclusion of many more samples from modern hard rock sources, as well as alluvial gold, and samples from historical collections representing surface material no longer available to modern prospectors. This should include investigations of the homogeneity of the individual artefacts and the natural gold. It may be done by multiple spot analyses on one archaelogical object (e.g. bead or wire ring) in order to assess variability. It might also include replicating the melting cycle and forging of the artefacts. The application can open new insights into African history. An adequate number of samples could form the basis for future studies of natural gold geochemical history, the sourcing of unprovenanced natural samples to their original sources, the identification of sources of stolen gold and the sourcing of archaeological material if this has not suffered mixing or alloying. 103 7 REFERENCES Allan, G.c. and Woodcock, J.T. (2001). A review of the flotation of native gold and electrum. Minerals and Engineering, 14,9, pp 931-962. Anhaeusser, C.R. (1974). The geology of the Sheba hills area of the Barberton Mountain Land, South Africa, with particular reference to the Eureka syncline. Economic Geology Research Unit Information Circular, University of Witwatersrand, 94, pp 1-47. Anhaeusser, c.R. and Maske, S. (1986). Mineral deposits of southern Africa. The Geological Society of South Africa, I, pp 113- I 15. Antweiler, J.C. and Campbell, W.L. (1977). Application of gold compositional analyses to mineral exploration in the United States. Journal of Geochemical Exploration, 8, pp 17-29. Bartholomew, D.S. (1990). Gold deposits of Zimbabwe. Mineral Resources Series, Zimbabwe Geological Survey, 94, pp 1-213. Boyle, R.W. (1979). The geochemistry of gold and its deposits (together with a chapter on geochemical prospecting for the element). Geological Survey of Canada, 280, pp 1- 584. Boyle, R.W. (1984). Gold deposits; their geology, geochemistry and origin. In: Foster R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, I, pp 183-189. Brandl, G. and de Wit, MJ. (1997). The Kaapvaal Craton. In: de Wit, MJ. and 104 Ashwall, L.O. (eds). Greenstone belts. Oxford University Press, Oxford, pp 581-607. Buitendag, W.L. (2007). Gold locality map. Mineral Resources Unit, Council for Geoscience, Pretoria. Campbell, S.D.G. and Pitfield P.EJ. (1994). Structural controls of gold mineralization in the Zimbabwe craton- Exploration guidelines. Zimbabwe Geological Survey, Bulletin 101, pp 1-270. Chapman, R., Leake, B. and Styles, M. (2002). Microchemical characterization of alluvial gold grains as an exploration tool. Gold bulletin, 35, 2, pp. 53-65. Chisholm, J.M. (1979). Composition of native gold. In: Glover, J.E. and Groves 0.1. Gold mineralization. The Geology Department and the Extension Service, University of Western Australia, Publication 3, pp 65-75. Cochran, A. (1982). Fluid inclusion populations in quartz-rich gold ores from the Barberton Greenstone Belt, Eastern Transvaal, South Africa. Unpublished M.Sc thesis: University of Arizona, pp 1-199. Cowey, A. (1994). Mining and metallurgy In South Africa - a pictorial history. Mintek, South Africa, pp 1-120. Desai, N. (2001). Technological, social and economic aspects of gold production and the use of Iron Age people of southern Africa. Unpublished MSc. thesis: University of Cape Town. De Ronde, C.EJ. and de Wit, MJ. (1994). Tectonic history of the Barberton 105 Greenstone Belt, South Africa: 490 million years of Archean crustal evolution. Tectonics, 13,4, pp 983-1005. De Ronde, C.EJ., Kamo, S., Davis, D.W., de Wit, MJ. and Spoon er, E.T.C (1991). Field, geochemical, and U-Pb isotopic constraints from hypabyssal felsic intrusions within the Barberton Greenstone Belt, South Africa: implications for tectonics and the timings of gold mineralization. Precambrian Research, 49, pp 261-280. De Ronde, C.E.J., Spooner, E.T.C., de Wit, MJ. and Bray, CJ. (1992). Shear zone related Au-quartz deposits in the Barberton Greenstone Belt, South Africa: field and petrographic characteristics, fluid properties and light stable isotope geochemistry. Economic Geology, 87, pp 366-402. De Vaal, J.B. (1984). Ou handelsvoetpaaie en wapaaie in oos- en noord- Transvaal. Contree, 16, pp 5-17. De Vaal, J .B. (1985). Handel langs die vroegste roetes. Contree, 17, pp 4-14. De Wit, MJ. (1998). On Archaean granites, greenstones, cratons and tectonics: does evidence demand verdict? Precambrian Research, 91, pp 181-226. De Wit, MJ., Roering, C, Hart, RJ., Armstromg, R.A., de Ronde, CE.J., Green, R.W., Tredoux, M., Peberdy, E. and Hart, R.A. (1992). Formation of an Archaean continent. Nature, 357, pp 553-562. Dimroth, E. and Kimberly, M.M. (1976). Precambrian atmospheric oxygen: evidence in the sedimentary distributions of carbon, sulfur, uranium and iron. Canadian Journal of Earth Sciences, 13, pp 1161-1185. 106 Dimroth, E. and Lichtblau, A.P. (1978). Oxygen in the Archaean ocean; comparison of ferric oxide crusts on Archaean and Cainozoic pillow basalts. Neues Jahrbuch fiir Mineralagie Abhndlungen, 133, pp 1-22. Dirks, P.H.G.M and van der Merwe, J.A. (1997). Tectonic stacking in an Archaean greenstone sequence and its control on gold mineralization at Dalny Mine, Zimbabwe. Journal of African Earth Sciences, 24,4, pp 603-619. Dirks, P.H.G.M., Kroner, A., Jelsma, H.A., Sithole, T.A and Vinyu, M.L. (1999). Structural relations and Pb-Pb zircon ages for the Makuti gneisses: evidence for the crustal-scale Pan-African shear zone in the Zambezi Belt, northwest Zimbabwe. Journal of African Earth Sciences, 28, 2, pp 427- 442. Du Plessis, SJ.P., Engelbrecht, L.NJ. and Visser, DJ.L. (1984). Geological map of the Republic of South Africa, Transkei, Bophuthatswana, Venda and Ciskei and the Africa kingdoms of Lesotho and Swaziland. Republic of South Africa. Gravity edition. Erasmus, C.S., SeIlschop, J.P.F. and Waterson, J.I.W. (1987). New evidence on the composition of mineral grains of native gold. Nuclear Geophysics, 1, I, pp 1- 23. Falconer, D.M., Craw, D., Youngson, J.H. and Faure, K. (2005). Gold and sulphide minerals in the Tertiary quartz pebble conglomerate gold placers, Southland, New Zealand. Ore Geology Reviews, 20, pp 1-21. Foster, R.P. and Wilson, J.F. (1984). Geological setting of Archaean gold deposits in Zimbabwe. In: Foster R.P. (ed). Gold' 82: The geology, geochemistry and 107 genesis of gold deposits. Geological Society of Zimbabwe, 1, pp 521-568. Frimmel, H.E. (2005). Archaean atmospheric evolution: evidence from the Witwatersrand gold fields, South Africa. Earth-Science Reviews, 70, 1-2, pp 1-46. Frimmel, H.E. and Gartz, V.H. (1997). Witwatersrand gold particle chemistry matches model of metamorphosed, hydrothermally altered placer deposits. Mineralium Deposita, 32, pp 523-530. Frimmel, H.E. and Minter W.E.L. (2002). Recent developments concerning the geological history and genesis of the Witwatersrand gold deposit, South Africa. Society of Economic Geologists, Special Publication 9, pp 17-45. Frimmel, H.E., Hallbauer, D.K. and Gartz, V.H. (1999). Gold mobilizing fluids in the Witwatersrand Basin: composition and possible sources. Mineralogy and Petrology, 66, pp 55-81. Fyfe, W.S. and Kerrich, R. (1984). Gold: Natural concentration processes. In: Foster, R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, 1, pp 99-127. Gan, S. and van Reenen D.D. (1997). Geology of gold deposits in the Southern Marginal Zone of the Limpopo Belt and the adjacent Sutherland Greenstone Belt, South Africa: Klein Letaba. South African Journal of Geology, 100, I, pp 73-83. Gasparrin, C. (1993). Gold and other precious metals: from ore to market. Springer- Verlag, pp 1-397. 108 Gihwala, D., Jacobson, L., Peisach, M. & Pineda, C.A. (1986). Proton-induced techniques for the determination of some trace impurities in gold objects. South African Journal of Science, 82, pp 258-260. Gray, A.L. (1985). Solid sample introduction by laser ablation for inductively coupled plasma source-mass spectrometry. Analyst, 110, no. 5, pp 551-556. Grigorova, B., Smith, W., Stiilpner, K., Tumilty, J.A and Miller, D. (1998). Fingerprinting of gold artefacts from Mapungubwe, Bosutswe and Thulamela. Gold Bulletin, 31, 3, pp 99-102. Groves, D.I. and Foster R.P. (1991). Archaen lode deposits. In: Foster, R.P. (ed). Gold metallogeny and exploration. B1ackie and Son Ltd, pp 63-103. Groves, D.I., Goldfarts R.J., Gebre-Marian M., Hagemann S.G. and Robert F. (1998). Orogenic gold deposits: A proposed classification in the context of their crustal distribution and relationship to other gold deposits types. Ore Geology Reviews, 13, pp 7-27. Guerra, M.F. (2004). The circulation of South America precious metals in Brazil at the end of the 17th century. Journal of Archaeological Science, 31, pp 1225- 1236. Guerra, M.F., Calligaro, T., Radtke, M., Reiche, I. and Riesemeier, H. (2005). Fingerprinting ancient gold by measuring Pt with spatially resolved high energy Sy-XRF. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 240, 1-2, pp 505-511. Gunther, D. and Hattendorf, B. (2005). Solid sample analysis using laser ablation 109 inductively coupled plasma mass spectrometry. Trends in Analytical Chemistry, 24, 3, pp 255-265. Gunther, D., Jackson, S.E. and Longerich, H.P. (1999). Laser ablation and arc/spark solid sample introduction into inductively coupled plasma mass spectrometers. Spectrochimia Acta Part B, 54, pp 381-409. Hall, M.E., Brimmer, S.P., Li, F.H. and Yablonsky, L. (1998). ICP-MS and ICP-OES studies of gold from a late Sarmatian burial. Journal of Archaeological Science, 25, pp 545-552. Hammerbeck, E.C.1. (1976). Gold outside the Witwatersrand triad. In: Coetzee, C.B. (ed). Mineral resources of the Republic of South Africa. Department of Mines, Geological Survey, Fifth Edition, Handbook 7, pp 75-92. Harley, M. and Charlesworth, E.G. (1996). The role of fluid pressure in the formation of bedding-parallel, thrust-hosted gold deposits, Sabie-Pilgrim's Rest goldfield, eastern Transvaal. Precambrian Research, 79, pp 125-140. Hayward, c.i., Reimold, W.U., Gibson, R.L. and Robb, LJ. (2005). Gold mineralization within the Witwatersrand Basin, South Africa: evidence for a modified placer origin, and the role of the Vredefort impact event. Geological Society, London, Special Publications, 248, pp 31-58. Houstoun, S.M. (1987). Competency contrasts and chemical controls as guides to gold mineralization: an example from the Barberton Mountain Land, South Africa. In: Ho, S.E. and Groves, D.1. (eds). Recent advances in understanding Precambrian gold deposits. The Geology Department and University IlO Extension: The University of Western Australia, Il, pp 147-160. http://sbio.uct.ac.za/Webemu/SEM_school/. The Electron Microscope Unit, University of Cape Town. Huffman, T.N. (2007). Handbook to the Iron Age. The archaeology of pre-colonial farming societies in southern Africa. Scottsville: University of KwaZulu Natal Press. Huffman, T.N. (2009). Mapungubwe and Great Zimbabwe: The origin and spread of social complexity in southern Africa. Journal of Anthropological Archaeology 28: 37-54. Jacobson, L., Fish, W.S. and van der Westhuizen, W.A. (2002). XRF analysis of pottery from Mutokolwe, a Khami settlement from the Soutpansberg mountains, South Africa. In: Glascock, M. (ed.) Geochemical evidence for long distance exchange. Westport, Connecticut: Bergin & Garvey, pp 215- 228. Jackson, S.E., Longerich, H.P., Dunning, G.R. and Fryer, B.J. (1992). The application of laser-ablation microprobe-inductively coupled plasma-mass-spectrometry (LAM-ICP-MS) to in situ trace-element determinations in minerals. Canadian Mineralogy, 30, pp 1049-1064. Jarvis, K.E., Williams, J.G., Parry, S.J. and Bertalan, E. (1995). Quantitative determination of platinum-group elements and gold using NiS fire assay with laser ablation-inductively coupled plasma-mass speetrometry (LA-ICP-MS). Chemical Geology, 124, pp 37-46. III Kalbskopf, S. and Nutt T. (2003). Lithological contrasts and constrains on gold mineralization in granitoids in the Zimbabwe craton: Structural controls and implications for exploration. Economic Geology Research Institute, Information Circular, 370, pp 1-24. Killick, A.M. and Scheepers, R. (2005). Controls to hydrothermal gold mineralization in the Witwatersberg Goldfield; situated in the floor to the south of the Bushveld Complex, South Africa. Journal of African Earth Sciences, 41, pp 238-250. Letcher, O. (1936). The Gold Mines of Southern Africa: The History, Technology and Statistics of the Gold Industry. Proclamination of the Witwatersrand and the Jubilee of the city of Johannesburg. 12- 43. Levinson, A.A. (1974). Introduction to Exploration Geochemistry. Applied Publishing Ltd. Wilmette, Illinois, Department of Goelogy and Calgary, USA, pp 1-924. Longerich, J. (2001). Chemometrics. In: Sylvester, PJ., (ed), Laser-ablation-ICPMS in the earth sciences: principles and applications. Mineralogical Association of Canada, Short Course Series, 29, 21-28. Longerich, H. and Diegor, W. (2001) - Introduction to mass spectrometry. In: Sylvester PJ. (ed). Laser Ablation-ICP-Mass Speetrometry in the Earth Sciences: Principles and Applications. Mineralogical Association of Canada, Short Course Series, 29, 1-19. Longerich, H. P., Jackson, S. E., Fryer, BJ. and Strong, D.F. (1993). The laser ablation microprobe-inductively coupled plasma-mass spectrometer. Geoscience Canada, 20, pp 21-27. 112 Longerich, H.P., Jackson, S.E. and Gunther, D. (1996). Laser ablation inductively coupled mass spectrometric transient signal data acquisition and analyte concentration calculation. Journal of Analytical Atomic Spectrometry, II, pp 899-104. Mann, A.G. (1984). Gold mines in Archaean granitic rocks in Zimbabwe. In: Foster, R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, I, pp 553-568. Martin, H. (1993). The mechanisms of petrogenesis of the Archaean continental crust: comparison with modern processes. Lithos, 30, pp 373-388. Minter, W.E.L., Goedhart, M., Knight, J. and Frimmel, H.E. (1993). Morphology of Witwatersrand gold grains from the Basal reef: evidence for their detrital origin. Economic Geology, 88, pp 237-248. Miller, D. (1995). 2000 years of indigenous mining and metallurgy in southern Africa - a review. South African Journal of Geology, 98, pp 232-238. Miller, D. (2001). Metal assemblages from the Greefswald areas, South Africa: K2, Mapungubwe Hill and Mapungubwe Southern Terrace. South African Archaeological Bulletin, 56, pp 83-103. Miller, D. (2002). Smelter and Smith: Iron age metal age fabrication technology in southern Africa. Journal of Archaeological Science, 29, pp 1083-1131. Miller, D. and Desai, N. (2004). The fabrication technology of southern African archaeological gold. Annals of the South African Museum, Ill, pp 79-102. 113 Miller, D., Desai, N. and Lee-Thorp, J. (2000). Indigenous gold mining in southern Africa: a review. South African Archaeological Society, Goodwin Series 8, pp 91-99. Miller, D., Desai, N., Grigorova, D. and Smith, W. (2001). Trace elements study of gold from southern African archaeological sites. South African Journal of Science, 97, pp 297-300. Mosier, E.L., Cathrall, l.B., Antweier, J.e. and Tripp, R.B. (1989). Geochemistry of placer gold, Koyukuk-Chandalar mining district, Alaska. Journal of Geochemical Exploration, 31, pp 97 -115. Nakagawa, M., Santosh, M., Nambiar, C.G. and Matsubara, e. (2005). Morphology and chemistry of placer gold from Attappadi valley, southern India. Gondwana Research, 8, 2, pp 213-222. Obethiir, T. and Saager, R. (1986). Silver and mercury in gold particles from the Proterozoic Witwatersrand placer deposits of South Africa: metallogenie and geochemical implications. Economic Geology, 81, pp 20-31. Oddy, A. (1983). On the trail of Iron Age gold. Transvaal Museum Bulletin, 19, pp 24-26. Oddy, A. (1984). Gold in the southern African Iron Age: A technological investigation of the Mapungubwe and other finds. Gold Bulletin, 17, 2, pp 70- 78. Ohata, M., Yasuda, H., Namai, Y. and Furuta, N. (2002). Laser ablation inductively coupled plasma mass speetrometry (LA-ICP-MS): comparison of different 114 internal standardization methods using laser-induced plasma (UP) emission and LA-ICP-MS signals. Analytical Sciences, 18, pp 1105-1110. Outridge, P.M., Doherty, W. and Gregoire, O.C. (1998). Determination of trace elemental signatures in placer gold by laser ablation inductively coupled plasma mass speetrometry as a potential aid for gold exploration. Journal of Geochemical Exploration, 60, pp 229-240. Palmer, J.A., Phillips, G.N. and McCathy, T.S. (1987). The nature of the Precambrian atmosphere and its relevance to Archaean gold mineralization. In: Ho, S.E. and Groves, 0.1. (eds). Recent advances in understanding Precambrian gold deposits. The Geology Department and University Extension: The University of Western Australia, Il,pp 321-326. Perkins, W.T. and Pearce, NJ.G. (1995). Mineral microanalysis by laser probe inductively coupled plasma mass spectrometry. In: Potts, PJ., Bowles, J.F.W., Reed, SJ.B. and Cave, M.R. (eds). Microprobe techniques in the earth sciences. Mineralogical Society Series 6, pp 291-323. Phillips, G.N. (1987). Metamorphism of the Witwatersrand gold fields: conditions during peak metamorphism. Journal of Metamorphic Geology,S, pp 307-322. Phillips, G.N. and Meyer, R.E. (1989). The Witwatersrand gold fields: Part II. An origin for Witwatersrand gold during metamorphism and associated alteration. In: Keays, R.R., Ramsay, W.R.H. and Groves, 0.1. (eds). The geology of gold deposits: the perspective in 1988. Economic Geology Monograph 6, pp 598- 608. lIS Phimister, LR. (1976). Pre-Colonial gold mining In southern Zambezi: a reassessment. African Social Research, 21, pp 1-30. Raymond, J., Williams-Jones, A.E. and Clark, J.R. (2005). Mineralization associated with scale and altered rock and pipe fragments from the Berlin geothermal field, El Salvador; implications for metal transport in natural systems. Journal of Volcanology and Geothermal Research, 145, pp 81-96. Reed, SJ.B. (1995). Electron microprobe microanalysis. In: Pous, PJ., Bowles, J.F.W., Reed, SJ.B. and Cave, M.R. (eds). Microprobe techniques in the earth sciences. Mineralogical Society Series 6, pp 291-323. Reisberg, L., Olivier, A., Lorand, J.P. and Ohnenstetter, M. (2004). Highly siderophile element behaviour in high temperature processes. Chemical Geology, 208, 1-4, pp 1-4. Robb, LJ. and Robb, V.M. (1998). Gold in the Witwatersrand Basin. In: Wilson, M.G.C. and Anhaeusser, c.R. (eds). The Mineral Resources of South Africa. Council for Geoscience Handbook 16. Council for Geoscience, Pretoria, pp 294-349. Robb, J.L. and Meyer F.M. (1995). The Witwatersrand Basin, South Africa: Geological framework and mineralization processes. Ore Geology Reviews, lO, pp 67-94. Roeder, E. (1984). Fluid-inclusion evidence bearing on environments of deposition. In: Foster, R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, 1, pp 129-164. 116 Russo, R.E., Mao, X., Liu, H., Gonzalez, J., Mao, S.S. (2002). Laser ablation in analytical chemistry-a review. Lawrence Berkeley National Laboratory, Beckerly, USA. Talanta, 57, pp 425-451. Schurrnann, L.W., Ward J.H.W., Horstmann U.E, Jordan LJ. and Eaton B. (2000). Carbonate dykes associated with Archaean lode-Au mineralization, Barberton Greenstone Belt, South Africa. Journal of African Earth Sciences, 30, 2, pp 249-266. Seward, T.M. (1984). The transport and deposition of gold in hydrothermal systems. In: Foster, R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, 1, pp 183-189. Sie, S.H., Murao, S. and Suter, G.F. (1996). Trace element distribution in native gold. Nuclear Instruments and Methods in Physics Research B, 109/110, pp 633- 638. Siebert, C., Kramers, J.D., Meisel, T.H., Morel, P.H. and Nagier, T.H. F. (2005). PGE, Re-Os, and Mo isotope systematics in the Archaean and early Proterozoic sedimentary systems as proxies for redox conditions of the early earth. Geochimica et CosmochimicaActa, 69, 7, pp 1787-1801. SACS (South African Committee for Stratigraphy). (1980). Part I: Lithostratigraphy of the Republic of South Africa, South West Africa/Namibia and the Republics of Bophuthatswana, Transkei and Venda. Stratigraphy of South Africa 8. Geological Survey of South Africa, Pretoria, pp 1-690. Summers, R. (1969). Ancient mining in Rhodesia and adjacent areas. Salisbury: 117 National Museums of Rhodesia Memoir, 3, pp 1-2. Swan, L. (1994). Early gold mining on the Zimbabwean plateau. Studies in African Archaeology 9. Uppsala: Societas Archaeologica Uppsaliensis, pp 1-181. Thomas, R. (2001). A beginer's guide to ICP-MS. www.spectroscopyonline.com. Spectroscopy, 16, 4-27, pp 38-48. Tyler, R. and Tyler, N. (1996). Stratigraphic and structural controls on gold Mineralization in the Pilgrim's Rest goldfield, eastern Transvaal, South Africa. Precambrian Research, 79, pp 141-169. Viljoen, MJ. (1984). Archaean gold mineralization and komatiites in southern Africa. In: Foster, R.P. (ed). Gold' 82: The geology, geochemistry and genesis of gold deposits. Geological Society of Zimbabwe, 1, pp 595-628. Viljoen, MJ. and Viljoen, R.P. (1969). An introduction to geology of the Barberton Granite-Greenstone terrain: Special Publication. Geological Society of South Africa, 2, pp 7-48. Vlachou-Mogire, C., Stern, B. and McOonnell, J.G. (2007). The application of LA- ICP-MS in the examination of the thin plating layers found in late Roman coins. Nuclear Instruments and Methods in Physics Research B, 265, pp 558- 568. Voges, F.O. (1986). The New Consort Mine, Barberton Greenstone Belt. In: Anhaeusser, c.R. and Maske, S. (eds). Mineral deposits of southern Africa, Geological Society of South Africa, 1, pp 155-162. 118 Von Gehlen, K. (1983). The morphology and silver content of gold from the upper Witwatersrand and Vensterdorp systems in the Klerksdorp goldfield, South Africa, Johannesburg. Minetalium. Deposita, 18, pp 529-534. Wagner, J.H.F. and Wiegand, J. (1986). The Sheba gold Mine, Barberton Greenstone Belt. In: Anhaeusser, C.R. and Maske, S. (eds). Mineral deposits of southern Africa, Geological Society of South Africa, 1, pp 155-162. Ward, J.H. (1963). Hierarchical grouping to optimize an objective function. Journal of American Statistical Association, 58, pp 236-244. Ward, J.H.W. and Wilson, M.G.C. (1998). Gold outside the Witwatersrand Basin. In: Wilson, M.G.C. and Anhaeusser, CR. (eds.). The mineral resources of South Africa, 16, Council for Geoscience, Pretoria, pp 350-386. Watling, R.J., Herbert, H.K, Delev, D. and Abel, l.B. (1994). Gold fingerprinting by Laser Ablation Coupled plasma mass sprectrometrotry. Spectrochimica, 49B (2), pp 205-219. Watling, R.1., Taylor, JJ., Shell, C.A., Chapman, R.1, Warner, R.B., Cahill, M. and Leake, R.C. (1999). The application of laser ablation inductively coupled plasma mass speetrometry (LA-ICP-MS) for establishing the provenance of gold ores and artefacts. Archaeopress, 6, pp 54-61. Williams, PJ. (1997). A metamorphosed, stratabound-epigenetic origin for a gruneritic Archean gold deposit, Barberton, South Africa. Ore Geology Reviews, 12, pp 135-15l. 119 Yoshihara, A. and Hamano, Y. (2004). Paleomagnetic constraints on the Archean geomagnetic field intensity obtained from komatiites of the Barberton and Belingwe greenstone belts, South Africa and Zimbabwe. Precambrian Research, 131, (1-2), pp I I 1-142. uv - UFS BLOEMFONTEIN BIBLIOTEEK. L!BRARY 120 Table 4-3 The data presented in counts per second by LA-ICP-MS for selected isotopes and in wt % for silver by SEM of the Zimbabwean and Mapungubwe artefacts samples. Lower limit of detection also presented. Locality 56Fe 59CO 60Ni 63CU 66Zn 75As 105Pd 107Ag 1880S 195Pt 202Hg 204Pb 206Pb 207Pb 208Pb 209Bi Ag wt. % Ore Province Lower limit of detection 131 3 7 13 3 13 3 107 3 3 3 29 3 10 17 220 Z1 236 11 234 370 537 4 2 449624 2 10 16 102 207 151 701 8 9 Z2 228 11 3 155 70 4 2 447811 1 11 16 34 98 74 501 8 9 Z3 182 9 2 225 108 2 2 375032 NO 11 17 200 358 400 700 8 7 Z4 1204 14 12 496 251 7 2 505827 1 7 11 45 116 111 331 8 10 Z5 355 9 6 287 137 10 2 469567 183 22 24 47 122 117 348 8 9 Z6 355 10 3 114 54 4 2 355607 38 12 18 28 63 60 171 8 7 Z7 233 11 2 164 80 8 3 288785 54 12 19 76 207 199 602 9 6 Z8 168 9 1 150 69 3 2 119917 1 11 20 31 72 69 197 7 2 Z9 180 9 2 11 6 3 2 27713 1 18 28 42 107 103 303 6 1 Z10 187 9 4 348 153 6 1 553483 63 11 19 42 107 102 304 5 11 Z11 376 15 5 187 84 7 1 547267 244 9 14 31 72 69 197 5 11 Z12 251 9 4 97 51 5 1 255374 1 12 17 35 86 83 241 4 5 Z14 515 15 9 440 196 6 1 463092 8 6 9 44 113 110 322 5 9 Z15 462 9 9 624 294 8 NO 318311 6 6 9 76 209 200 610 NO 6 Z16 317 10 5 313 155 7 NO 527842 170 8 10 59 158 155 455 NO 10 Z17 196 11 3 307 129 9 1 469049 1 9 11 92 257 248 754 4 9 Z18 1764 8 32 50 24 3 NO 337736 12 8 14 23 61 58 166 NO 7 Z19 1138 9 21 313 151 5 NO 554778 7 4 7 65 176 172 510 NO 11 Z20 173 9 4 706 336 7 1 576793 4 4 7 51 134 129 385 4 11 Z21 132 10 1 753 353 33 2 516446 2 5 7 51 134 132 384 6 10 Z22 2721 19 66 7261 3941 11 NO 1845116 11 6 7 53 140 137 403 NO 36 Z23 240 31 6 729 352 12 1 324009 4 16 29 27 62 59 169 4 6 ~ o Z24 280 7 5 203 102 12 1 363895 156 8 16 53 140 136 403 4 7 ~ Q) Valid number of samples 23 23 23 23 23 23 18 23 22 23 23 23 23 23 23 18 23 «t::: Mean 517 11 19 622 332 8 2 464916 44 10 15 57 139 134 398 6 9 c (1) Median 251 10 5 307 137 7 2 449624 7 9 16 47 122 117 384 6 9 Q) s: Minimum 132 7 1 11 6 2 1 27713 0 4 7 27 61 58 166 4 1.c (1) Maximum 2721 31 234 7261 3941 33 3 1845116 244 22 29 200 358 400 754 9 36 .c E Standard deviation (CJ) 627 5 49 1463 797 6 1 331791 73 4 6 37 72 77 181 2 6 i N Coefficient of variation (%) 121 44 257 235 240 82 30 71 166 45 42 65 51 57 45 30 71 Ma 120 5 2 172 1 9 1 62667 NO 13 3 132 377 352 1114 5 1 Mb 532 6 4 206 1 13 1 67260 NO 15 4 236 687 684 2043 5 1 Mf 121 8 2 164 1 10 2 59763 NO 12 2 72 197 191 574 7 1 Mh 148 7 1 598 1 15 1 80147 NO 7 4 101 284 273 835 10 2 (/l Mj 133 9 2 706 1 20 NO 81320 NO 3 3 98 275 275 808 10 2 ti ~ Mi 154 3 1 533 1 25 1 82220 NO 7 2 209 607 607 1802 9 2 Q) Valid number of samples 6 6 6 6 t::: 6 6 5 6 0 6 6 6 6 6 6 6 6« Mean 201 6 2 397 1 15 1 72229 10 3 142 405 397 1196 8 1 Qs) Median 140 6 2 370 1 . 14 1 73703 10 3 117 331 314 974 8 1.c ::l Minimum 120 3 1 164 1 9 1 59763 3 2 72 197 191 574 5 1 Ol c Maximum 532 9 4 706 1 25 2 82220 15 4 236 687 684 2043 10 2 ::l 0. (1) Standard deviation (CJ) 162 2 1 243 0 6 1 10165 4 1 66 198 200 593 2 0 ~ Coefficient of variation (%) 81 34 47 61 22 42 58 14 46 35 47 49 50 50 30 14 59 Table 4-2 The data presented in counts per second by LA-ICP-MS for selected isotopes and in wt % for silver by SEM of the gold mines from Pietersburg, Murchison Greenstone Belts, Sabie-Pilgrim's Rest and Knysna gold samples. Lower limit of detection also presented. Locality 56Fe 59CO 6°Ni 63CU 66Zn 75As 105Pd 107Ag 1880S 195Pt 202Hg 204Pb 206Pb 207Pb 208Pb 209Bi Ag wt. % Ore Province - Lower limit of detection 131 3 7 13 3 13 3 107 3 3 3 29 3 10 17 220Qi Marabastadt 419 25 5 1554 5 75 3 114992 2 2 169 153 286 112 165 67 2 CD Q) Marabastadt 450 NO 7 1328 3 30 1 29703 2 2 148 5 569 114 173 18 1 c 0 Valid number of samples 2 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 ii) c Mean 434 25 6 1441 4 53 2 72347 2 2 159 79 427 113 169 43 1 Q) Q...). Median 434 25 6 1441 4 53 2 72347 2 2 159 79 427 113 169 43 1 . (!) Minimum 419 25 5 1328 3 30 1 29703 2 2 148 5 286 112 165 18 1 O....l .::c::J Maximum 450 25 7 1554 5 75 3 114992 2 2 169 153 569 114 173 67 2 ~ Q) Standard deviation (cr) 22 2 160 1 31 1 60308 0 0 15 105 200 1 6 35 1 Q) 0::: Coefficient of variation (%)- 5 27 11 27 60 51 83 19 11 9 133 47 1 3 81 81Qi CD Gravellotte Mine 689 NO 92 2366 23 60 2 98920 1 1 5523 12 NO 2274 2058 49 2 Q) c Valid number of samples 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 ii) Mean 689 92 2366 23 60 2 98920 1 1 5523 12 2274 2058 49 2 c Q) !!! Median 689 92 2366 23 60 2 98920 1 1 5523 12 2274 2058 49 2o Minimum c 689 92 2366 23 60 2 98920 1 1 5523 12 2274 2058 49 2 e0n Maximum:.c 689 92 2366 23 60 2 98920 1 1 5523 12 2274 2058 49 2 ~ Standard deviation (o) ::::J :2: Coefficient of variation (%) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Lydenburg 883 415 7 296 5 18 2 10003 2 3 32 2 35 28 95 40 0 - Geodeverwacht 970 29 24 837 10 20 3 546959 2 2 316 3 180 156 211 16 11en Q) New Pilgrim/s Rest 936 6 253 446 40 20 3 375914 2 1 6752 14 14 5471 4606 11 7 eI: en Valid number of samples 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 -E .:::: Mean 930 150 95 526 18 19 3 310959 2 2 2367 6 76 1885 1637 22 6 .2> Median 936 29 24 446 10 20 3 375914 2 2 316 3 35 156 211 16 7 0::: Minimum 883 6 7 296 5 18 2 10003 2 1 32 2 14 28 95 11 0 Q) :0 Maximum 970 415 253 837 40 20 3 546959 2 3 6752 14 180 5471 4606 40 11 Cl! Cl) Standard deviation to) 44 230 137 279 19 1 1 274308 0 1 3800 7 90 3106 2572 15 5 Coefficient of variation (%) 5 153 145 53 103 7 27 88 7 43 161 111 118 165 157 68 88 Knysna Mine 2704 55 16 1441 18 7 2 214384 2 2 189 12 NO 96 150 15 4 Knysna Mine 2846 25 24 1188 18 9 3 97596 2 2 2813 51 NO 8951 7591 26 2 a. Valid number of samples 2 2 2 2 2 2 2 2 2 2 2 2 0 2 2 2 2 ::::J 0.... Mean 2775 40 20 1315 18 8 2 155990 2 2 1501 32 4524 3871 20 3 e> Median 2775 40 20 1315 18 8 2 155990 2 2 1501 32 4524 3871 20 3 Q) a. Minimum 2704 25 16 1188 18 7 2 97596 2 2 189 12 96 150 15 2 ::::J Cl) Maximum 2846 55 24 1441 18 9 3 214384 2 2 2813 51 8951 7591 26 4 Q) a. Standard deviation (cr) 100 21 5 179 0 2 1 82581 0 0 1855 27 6261 5262 8 2 oCl! Coefficient of variation (%) 4 53 27 14 0 22 23 53 13 3 124 86 138 136 38 55 58 Table 4-1 The data presented in counts per second by LA-ICP-MS for selected isotopes and in wt % for silver by SEM of the Witwatersrand, Barberton and Zimbabwe gold samples. Lower limit of detection also presented. Locality 56Fe 59CO 60Ni 63CU 66Zn 75A5 10sPd 107Ag 18805 19SPt 202Hg 204Pb 206Pb 207Pb 208Pb 209Si Agwt. % i Ore Province Lower limit of detection 131 3 7 13 3 13 3 107 3 3 3 29 3 10 17 220 City Oeep (a) 809 5 5 1383 4 8 6 475357 4 4 1158 74 38 39 129 21 10 City Oeep (b) 765 5 7 1321 3 7 8 435628 4 2 7063 74 37 39 1039 20 9 Inner Basin Reef 379 3 3 514 2 NO NO 218823 4 NO 8683 NO 434 811 901 53 4 Carbon Leader 547 8 11 589 5 8 NO 433904 3 4 7314 NO 46 37 860 27 8 Leader Reef 760 3 6 59 10 8 NO 348435 3 5 9816 63 124 78 230 10 7 C'Reef 591 4 10 31 3 13 NO 320344 3 NO 2422 NO 105 52 161 20 6 B'Reef 988 5 4 704 8 13 9 280506 3 NO 3389 97 168 188 395 71 6 c South Oeep 522 3 6 803 4 11 1 259230 3 2 8937 36 52 48 964 39 5 ,ii III Valid number of samplesco 8 8 8 8 8 7 4 8 8 5 8 5 8 8 8 8 8 "'0 Mean 670 5 7 676 5 10 6 346528 3 3 6097 69 126 161 585 32 7 C I..I.I. Median 675 5 6 647 4 8 7 334389 3 4 7188 74 79 50 628 24 6 C..f).. Minimum 379 3 3 31 2 7 1 218823 3 2 1158 36 37 37 129 10 4 Ol (ij Maximum 988 8 11 1383 10 13 9 475357 4 5 9816 97 434 811 1039 71 10 ~ Standard deviation [a) 195 2 3 502 3 2 3 93463 0 1 3300 22 133 267 392 20 2 ~ Coefficient of variation (% 29 38 40 74 57 24 54 27 9 40 54 32 106 166 67 63 28 Agnes Mine (a) 218 3 11 1345 3 6 3 89002 2 2 213 235 173 134 8708 24 2 Agnes Mine (b) 2694 3 20 205 6 6 2 174928 2 2 217 NO 103 93 99441 14 4 Sheba Mine (a) 4482 3 35 1813 2 5 2 367846 1 NO 347 NO 20 21 237 10 4 Sheba Mine (b) 8538 9 21 3814 6 76 2 102501 NO 2 264 NO 170 196 522 24 2 Sheba Mine ( c) 4748 4 33 5098 7 90 3 55642 1 2 32 9 7054 5881 14964 19519 1 Sheba Mine (d) 3608 5 11 5999 9 8 7 55408 2 3 95 330 158 13169 39634 52108 1 Sheba Mine (e) 617 2 35 4308 3 5 2 167024..... 2 NO 326 533 40 30 135 40 3 (ij New Consort Mine (a) 10000 9 4 430 9 87 3 232706 2 2 219 491 124 1315 3055 2755 5 co New Consort Mine (b) 4401 3 3 3763 6 26 5 110235 2 NO 85 153 5074 4702 10478 12372 2 cOl New Consort Mine ( c) 6273 3 29 1867 3 64 5 0 50955 1 NO 257 956 922 798 1534 1988 1 icii Valid number of samples 10 10 10 10 10 10 10 10 9 6 10 7 10 10 10 10 10 Ol ~ Mean 4558 4 20 2865 5 38 3 140624 2 2 205 387 1384 2634 17871 8885 2 CJ Median 4441 3 21 2815 6 17 3 106368 2 2 218 330 164 497 5882 1014 2 c Minimum 218 2 4 0 205 2 5 2 50955 1 2 32 9 20 21 135 10 1 t: Ol Maximum 10000 9 35 5999 9 90 7 367846 2 3 347 956 7054 13169 99441 52108 5.c (ij Standard deviation to) 3113 3 12 2000 3 37 2 99749 0 1 104 311 2524 4258 31069 16571 1 co Coefficient of variation (%J 68 58 61 70 48 99 52 71 23 28 51 80 182 162 174 187 54 Oon Mine (a) 1805 3 3 438 7 6 NO 208778 NO NO 794 11 195 183 566 500 4 Oon Mine (b) 5582 13 28 732 21 22 NO 186200 NO NO 751 22 185 175 1131 1744 4 Gadzema 4853 4 22 1517 32 7 ..... 3 118980 3 2 415 40 178 170 1521 132 2Cf) (ij Belingwe 3651 25 35 2204 11 7 2 218320 2 NO 1811 49 202 477 919 1661 4 co Lower Gwelo 6254 3 48 1074 15 6 4 100 2 2 281 NO 286 2870 6961 9940 1 cOl 0 Reef of Gaika 3686 6 53 2953 9 10 2 49707 2 2 254 46 128 110 5711 9 1 icii Victoria reef 2586 11 10 2435 13 18 2 91415 2 2 208 89 120 129 4940 48 2 Ol ~ Valid number of samples 7 7 7 7 7 7 5 7 5 4 7 6 7 7 7 7 7 CJ Mean 4059 9 29 1622 15 11 2 124785 2 2 644 43 185 587 3107 2004 3 c III Median 3686 6 28 1517 13 7 2 118980 2 2 415 43 185 175 1521 500 2 Ol ~ Minimum 1805 3 3 438 7 6 2 100 1 2 208 11 120 110 566 9 1.c III Maximum 6254 25 53 2953 32 22 4 218320 3 2 1811 89 286 2870 6961 9940 4.c E Standard deviation (o) 1598 8 19 938 9 6 1 83572 0 0 566 27 55 1014 2666 3576 1 N Coefficient of variation (%) 39 86 65 58 56 57 35 67 24 10 88 63 30 173 86 178 53 57