King Croesus’ Gold Excavations at Sardis and the History of Gold Refining Andrew Ramage and Paul Craddock with contributions by M.R. Cowell, A.B, Gegkinli, D.R. Hook, M.S. Humphrey K. Hyne, N.D. Meeks, A.P. Middleton and H. Ozbal Published for The Trustees of The British Museum by = BRITISH MUSEUM Jill PRESS In association with ARCHAEOLOGICAL EXPLORATION OF SARDIS Harvard University Art MuseumsTo our wives, Nancy and Brenda gon nese axons Croesus’ gold : excatE Sart ai AA =37881* © 2000 The Trustees of the British Museum and the President and Fellows of Harvard College First published in 2000 by British Museum Press A division of The British Museum Company Ltd 46 Bloomsbury Street, London WC1B 3QQ {in association with the Archaeological Exploration of Sardis (as Monograph 11) A catalogue record for this book is available from the British Library ISBN 0-7141-0888-X Designed and set in Apollo by Andrew Shoolbred Printed in Great Britain at Cambridge University PressContents Contributors 7 Acknowledgements 8 Prologue A. Ramage and P-T. Craddock 10 Excavation of the gold refinery at Sardis 10 The process of gold refining 10 Summary of the historical evidence 11 Golden Sardis A. Ramage 4 Sardis in context 14 Lydian metallurgy and the beginnings of coinage 17 Sources for the Lydian gold 19 Value and quantity of gold 20 Storage and use 21 Digging at Sardis 23 Historical Survey of Gold Refining: 1 Surface Treatments and Refining Worldwide, and in Europe Prior to AD 1500. PT. Craddock 27 Surface enhancement of gold artefacts: the precursor to gold parting 27 ‘The inception of gold parting 31 Literary evidence for cementation parting 32 Historical Survey of Gold Refining: 2 Post-medieval Europe PT. Craddock 54 Renaissance Europe: Probierbiichlein, Biringuccio, ‘Agricola and Ercker 54 Later history and descriptions 65 Gold refining by sulphur, antimony and mineral acids 67 ‘The Excavations and Finds A. Ramage n ‘The goldworks 72 The altar 72 Details of the goldworking area 81 ‘The two dumps 86 The finds 88 Chronology 94 Technical summary 96 Scanning Electron Microscopy of the Refractory Remains and the Gold N.D. Meeks 99 Introduction 99 Excavated Materials and Methods of Examination 99 Excavated material 99 ‘Methods used to examine the materials 101 Examination of Refractory Sherds and Furnace Bricks 102 Description of sherds 102 Discussion 126 Examination of Gold Foils and Dust 145 Gold fragments from the Archaeological Museum, at Manisa 145 The purified gold grains 151 Conclusions 151 Scientific Examination of Some Ceramic Materials and Samples of Litharge A.P. Middleton, D.R. Hook and ‘M.S. Humphrey 157 Introduction. 157 ‘Materials and methods 157 Observations 158 Interpretation 166 Summary 1686 Contents 7 10 Scientific Examination of the Lydian Precious ‘Metal Coinages M.R. Cowell and K. Hyne 169 Introduction 169 Previous analyses 169 Flectrum coins 172 Gold and silver coins 173 Platinum group element inclusions 173 Conclusions 173 Replication Experiments and the Chemistry of Gold Refining P-T. Craddock 175 Previous experiments and replications 175 Active agents in the salt cementation process 180 Chemistry of the salt cementation process 181 Physical chemistry of the parting process 182 Examination of the Sardis Gold and the Replication Experiments A.E. Geckinli, H. Ozbal, P-T. Craddock and N.D. Meeks 184 Examination of the Gold from Sardis 184 Gold samples selected for examination 184 Methods 184 Results and discussion 185 Scanning electron microscope study 185 Replication Experiments and Examination of the Experimentally Treated Metal 187 Form and examination of the test pieces 187 Mechanism of the removal of silver 187 Conclusions 188 Reconstruction of the Salt Cementation Process at the Sardis Refinery PT. Craddock 200 Introduction 200 Source of the metal treated at the refinery 200 Parting vessels 202 Cementation furnaces 203 Cementation mixture 203 Evidence for graduation or quartation 205 Operating parameters of the process 205 Efficiency of the separation and gold recovery 207 Recovery of silver from the debris of the process 208 Summary 209 Epilogue: The Significance of the Sardis Refinery in the Classical World A, Ramage and PT. Craddock 212 APPENDICES 1 Inventory and Descriptions of the Gold Samples Compiled from Sardis records by A. Ramage 215 2. Inventory and Descriptions of Finds of Equipment and Supplies A. Ramage 221 Bellows nozzles or tuyeres 223 Bread tray with vitrification 226 Bread tray without vitrification 227 Coarseware with vitrification 228 Coarseware without vitrification 229 Plainware 229 Brick 229 Brick: pieces of glazed surface 230 Vitrified materials 230 Crucible rim 231 Metal 231 Litharge cakes 231 Lead dribble 232 3 Karly History of the Amalgamation Process PLT. Craddock 233 Gold extraction 233 Silver extraction 235 4 The Platinum Group Element Inclusions PLT. Craddock 238 Early references to PGE inclusions and their removal from gold 240 Implications of the survival of PGE inclusions for the interpretation of the Sardis process 242 5 Assaying in Antiquity P.T. Craddock 245 Specific gravity 245 Fire assay 246 Touchstone 247 Summary 249 6 Concordance between Sardis Excavation Codes/ Descriptions and British Museum Codes 251 Technical Glossary 253 Bibliography 257 Index 269Contributors M.R. Cowell Department of Scientific Research, British Museum, Great Russell Street, London WC1B 3DG PT. Craddock Department of Scientific Research, British Museum AB. Geckinli Faculty of Chemical and Metallurgical Engineering, Istanbul Technical University, 80626 Maslak, Istanbul DR. Hook Department of Scientific Research, British Museum MS. Humphrey Department of Scientific Research, British Museum K. Hyne Formerly Department of Scientific Research, British Museum N.D. Meeks Department of Scientific Research, British Museum AP Middleton Department of Scientific Research, British Museum H, Ozbal Bosphorus University, 80815 Bebek, Istanbul ‘A. Ramage Department of the History of Art, Cornell University, Ithaca, New York 14853Acknowledgements We are indebted to many people for their encouragement and unstinting help, across a broad spectrum of academic disciplines, both in the humanities and the sciences. Archaeological ‘The late Professor G.M.A. Hanfmann and Professor Craw- ford H. Greenewalt, Jr, his successor as Field Director, deserve particular thanks for recognising the importance of the discovery and supporting the field excavations and the subsequent analytical investigations in the laboratory. We had much encouragement also in the early interpreta- tion of the finds from the late Professor Cyril S. Smith of the Massachusetts Institute of Technology. He believed in the main conclusions but played the helpful sceptic by pointing out anomalies and discrepancies. We also stand in the debt of Sidney M. Goldstein, a conservator at Sardis in 1968 and a fellow graduate stu- dent, He helped in the beginning, discussing the interpre- tation of the little burnt rings that turned out to be hearths. He shared Andrew Ramage’s excitement in the discovery and they both realised its importance even ‘though they were not able to define the precise activity at first. In his capacity as a conservator, he organised the finds and took on the duty of compiling lists of the differ- ent categories and individual pieces of gold and ceramic materials collected in 1968 and 1969. These lists form the basis for the identification of items mentioned in the day- to-day field records. Richard E. Stone, the Senior Conservator in 1968, must be recognised as the person most responsible for guiding Ramage and Goldstein towards the specific interpretation of gold refining for what was clearly some sort of metal- ‘working process. He was at that time a graduate student in the Conservation Program at the Institute for Fine Arts, New York University. ‘We should also recognise the contribution of our Turk- ish colleagues in the field, in particular the team of work- ‘men responsible for the actual digging. One of them in particular, the late Huseyin Bal, can be commended for a sharp eye and a delicate trowel. The Archaeological ‘Museum in Manisa has been the recipient of their finds, and we are also grateful to the late Kemal Ziya Polatkan, director at the time of the discovery, and to successive directors (Kubilay Nayir and, since 1986, Hasan Dedeoglu), who have been most helpful in permitting Goldstein and Ramage to study the pieces of gold and other refinery materials stored in their care, and for allowing the gold to travel to Istanbul for scientific study. Celalettin Sentiirk and Teoman Yalginkaya facilitated the physical transfer of gold specimens to the University of Istanbul in 1993. Elizabeth Gombosi took many of the original field pho- tographs and Michael Hamilton printed a large proportion. Elizabeth Wable (in consultation with Sidney Goldstein) made many of the drawings reconstructing the process, and Catherine Alexander drew many of the tuyeres and refractory items. Sardis recorder for 1997, Ellen Roblee, and assistant recorder for 1998, John Vonder Bruegge, whether at Sardis or back at the office, helped in the retroactive inventory project of the technical material. Laura M. Gadbery, Associate Director and Head of the Sardis office in Cambridge, assisted in the preparation of the new object cards. To all of them, many thanks. ‘The original support and financial contributions that made the Archaeological Exploration of Sardis possible have by now been rehearsed several times, as has the gen- erosity of the Republic of Turkey, through the Department of Antiquities and Museums, in affording us the privilege of working at Sardis. That is the background to all our endeavours. For the discovery and study of the gold refin- ery a few years may be picked out (1967-70, 1975, 1987 and 1993) when the support offered through the continua- tion of the excavations was crucial. These years cover the actual discovery and excavation of the refinery (1967-70), a short study season by Sidney M. Goldstein (1975) and two study seasons by Paul T. Craddock (1987 and 1993). ‘Andrew Ramage was active at Sardis for all seasons except 1975 but did not engage in gold research other than that entailed in his original discovery until 1987. Major sup- port in those years came from The Old Dominion Founda- tion, the Ford Foundation, the Loeb Classical Library Foundation of Harvard University and the National Endowment for the Humanities,’ a federal agency whose | NEH grants 167-0-56, H68.0-61, H69-0-23 supported genera field work in 1967, 1968 and 1969 respectively. RO-111-70-3966 supported fieldwork publications, as did RO-21414-87.views do not necessarily concur with those set out here. Individuals whose contributions have been generous and consistent include Mr and Mrs David Greenewalt, the late Dr Edwin H. Land and Mrs Land, Mr Thomas B. Lemann, Mrs Guy Smallwood and Mrs Gustavus F. Swift. Scientific ‘Much of the scientific work was carried out at the Depart- ‘ment of Scientific Research of the British Museum, and the various contributors to this volume wish to thank their colleagues within the Department for help and discussion over a wide range of topics. In particular, Dr LC. Freestone is thanked for his help in interpreting the structure of the refractories. We also should thank our colleagues in other departments, notably the Department of Coins and Medals, for assistance in selecting the Lydian coins for scientific examination from the collections of the British Museum, and allowing them to be analysed. We are also grateful to Professor A.A. Gordus of the Department of Chemistry, University of Michigan, for allowing us to publish here the analyses of gold foils, etc. from the excavations and of Lydian coins from the Ashmolean Museum, Oxford, per- formed many years ago. We wish to thank Dr Jack Ogden for help and advice over all matters auriferous, freely given from his ency- clopaedic knowledge of the subject. We are grateful to Dr Justine Bayley, of the Ancient Monuments Laboratory of English Heritage, for imparting her knowledge of the exca- vated remains of parting, not least for teaching Paul Crad- dock to recognise the distinctive features of the debris of salt cementation. We must also thank Professor E.T. Hall, of the Research Laboratory for Archaeology and the His- tory of Art, Oxford, both for his encouragement whilst a Trustee of the British Museum, and for his scientific advice, based on his own important experiments on gold Illustration acknowledgements “Acknovlegement is due tothe following for permission to reproduce photographs and other ilustrations: “Archaeological Exploration of Sardis for Figs 1.1 t0 1.3, 1.5, 1.6, 1.9 t01.11, 4.1 to 4.53, 4.55 £04.57, A21 t0 A210 British Library Board for ig. 2.6 Deutsches Archéologisches Institut Athens for Fig 4.54 Department of Metallurgy and Materials Engineering, Istanbul ‘Technical University for Figs 9.1 to 9.23 Hirmer Verlag Miinchen for Fig. 1.7 ‘Trustees of the British Museum for Figs 1.4 and 1.8, 5.1 105.64, 6.1 10 6.10, 7.1, 9.24 t0 9.34, 10.1 0 10.5, AL.1 te AL10 Acknowledgements 9 refining, which he has allowed us to quote. We also wish to thank Dr Alessandra Giumlia Mair for discussion and translation of the ancient texts in Chapters 2 and 3. We are grateful to the laboratories of Johnson Matthey Ltd for making available to us J.H.F. Notton’s unpublished internal, report on his gold-refining experiments. ‘The work on the gold foils from the Archaeological Museum in Manisa was conducted at the laboratories of the Department of Metallurgy and Materials Engineering of the Istanbul Technical University and partially sup- ported by the Archaeometry Research Center of the Bosphorus University and the Archaeological Exploration of Sardis. Our thanks go to Dr V. Giirkan, who undertook the macrophotography of the gold foils, and to H. Sezer for his assistance with the SEM examination, The macropho- tography of the material examined in the British Museum, was undertaken by Anthony Milton of the British Museum Photographic Service, Editorial Our readers, Dr Oscar White Muscarella and Professor Mike Wayman, and our editors, John Day, Katherine Kiefer and Nina Shandloff, on both sides of the Atlantic, have persuasively but firmly helped us to smooth out the organ- isation of the book with its mixed historical and scientific components. Andrew Ramage's wife, Nancy Hirschland Ramage, was in on the discovery from the very beginning and made drawings of the cupels and of the furnaces, Many times recently she has read over his text and offered countless suggestions for improvement as well as encour- aged him to persevere when things looked bleak. Andrew Ramage Paul Craddock August 1999Prologue A. Ramage and P°T. Craddock Excavation of the gold refinery at Sardis ‘The discovery of the gold refinery is one of the most important finds made in the course of the Harvard-Cornell excavations, The installations are at the moment unique in the classical world and of fundamental importance for the interpretation of several short or inconclusive passages from ancient authors around the world. The details of the sequential discoveries and day-to-day strategies are sum- marised below to produce a synthetic picture of the work- shop and its procedures. "At the time of the excavation, the expedition conserva- tor, who had extensive technical expertise, helped us to make the connection between multiple small hearths with associated lead oxide and the possibility of a gold-refining process. Subsequently, we saw the reddened masses of mud brick as furnaces for another, and even more vital stage in the parting process. This was on the basis of written accounts from the sixteenth century that apparently had retained a tremendous amount of traditional information and practice, ‘The place of Lydia in the centre of western Anatolia and the expansive policy of its kings meant that its capital, Sardis, became the focus of diplomacy and hostilities at the end of the seventh century and into the sixth and fifth centuries ac. Enormous wealth was generated during that period from conquest and from the exploitation of the alluvial gold in the River Pactolus and other rivers nearby. It is, therefore, of great significance to have found one of the installations where the natural gold-silver alloy was processed to provide pure gold and pure silver as is implied by the tradition in the ancient world that the Lydians were the first people to employ gold and silver for coins, The workshop was found in a domestic area, and the fittings and equipment give the impression of domestic items pressed into service for special requirements. Thus, the furnaces give the appearance of a battery of domestic ovens and the vessels for the parting process are derived from domestic jugs, ‘The dating of the refinery fits very closely with that of the reign of King Croesus (561-547 nc), who became a symbol for enormous wealth into our own time. The body of the text here contains an explanation of the chronology, as the circumstances of the discovery and the actual artefacts or workshop residue are described. Detailed scientific examination of some of the finds, undertaken in the British Museum's Department of Sci- entific Research, elucidated the parameters of the refining process. When combined with the evidence found in historical texts describing various methods of refining gold, the archaeological evidence and the results of the scientific examination create a detailed picture of the gold- refining process used at Sardis. The processes of gold refining Gold occurs naturally as minute particles of metal in the primary quartz veins in which it formed. This is known as primary gold. Where these deposits have eroded, the gold can be released, reconcentrated by the action of water and deposited in beds of sands or gravels. This is secondary or placer gold. The gold, being of very different density to the other materials, is often concentrated within very specific areas of these sands or gravels. Sometimes these areas are quite small but still of very great value, containing large quantities of gold in near-surface deposits of easily worked sand or gravel in former or existing stream beds. It isbelieved that the Pactolus contained such deposits, and although nothing is known about the actual mining methods, their discovery and exploitation took place over a relatively short period, creating enormous wealth. How- ever, the deposits were soon exhausted, and the riches of Croesus became but a memory to succeeding generations. Almost all gold occurs naturally containing some silver. This can vary enormously but typically lies between about 5% and 40% by weight. The only other metal found in natural gold above trace levels is copper, which, however, is only rarely found in quantities greater than 1% or 2%. With the advent of coinage, where both the weight and purity of the metal were guaranteed, it was clearly imperative to be able to refine the gold. The base metals, notably copper, could be removed from the gold by cupel- lation. The impure metal was melted together with lead and the resulting mixture subjected to a continuous blast of air at high temperature, typically in the region of 1100 °C. This completely oxidised the lead and any other base metals present, to form a molten mass of lead oxide ~ litharge — with the other metal oxides absorbed within it, but leaving the two noble metals, gold and silver, unaffected and separate. Different methods were necessary to separate the gold and silver, an operation known in English as parting. Before the discovery of mineral acids, the only method of parting gold and silver was by cementation with acidic salts. In a cementation process, the principal reactants are in the solid state, and the usual salt used in antiquity seems to have been sodium chloride (common salt), but others, notably alum, ferric sulphate and potassium nitrate (nitre, also known as saltpetre), were also used alone or in combination. These were joined in the Middle Ages by sulphur parting, using either elemental sulphur or, more usually, an iron or antimony sulphide, The active agents were often supported on an inert medium or carrier, typi- cally of clay or brick dust, which was especially useful if the reagents melted during the reaction. The mixture of active agents and the carrier was known as the cement, The finely divided, impure gold was placed, carefully surrounded by the cement, inside a ceramic container. This was known as the parting vessel, and was usually specified as being of earthenware, the relatively open structure of which would have been thermally more resilient and more porous than finewares. The parting vessel was then heated for prolonged periods at tempera- tures below the melting point of the impure gold, but the other reagents could be either solid or molten. Note that, although cementation processes are nominally solid state, at the operating temperature the active agents have a high Prologue 11 vapour pressure, even though still solid, and in reality it was their vapours which attacked the solid metal. Thus, in the salt cementation process, which is believed to have been employed at Sardis, vapours of chloride ions and chlorine at elevated temperature attacked the surfaces of the impure gold, penetrating deeply into the interior of the metal along the grain boundaries. The process was per- formed for some hours, or even days according to some early descriptions, until the gold was purified. The silver ‘was converted to silver chloride, which is also volatile, and was absorbed by the inert carrier, if present, and in the walls of the parting vessel and furnace, The silver could be recovered from these various materials by smelting them with lead to absorb the silver salts, followed by cupellation to release the now-pure silver. Acid parting began in the Middle Ages, initially using nitric acid, which was replaced in the post-medieval period by sulphuric acid. Here, the solid impure gold was attacked by the hot aqueous solutions of the acid. Once again, the active reagents attacked preferentially down the grain boundaries, In order to facilitate the parting process, it was normal practice to add considerable quantities of silver to the impure gold, a procedure known as graduation or quartation. The reason for this seemingly quixotic practice of adding more of the metal they were trying to remove was to ensure that there was abundant silver within the body of the gold which the acid would eat away, creating considerable porosity throughout the gold, and thereby ensuring that all of the metal was exposed to the acid. In fact, this procedure originated earlier with the salt cementation process, where it was used as a special extra refinement, with the extra metal, usually copper, added after the first cementation. However, the greater porosity created in the gold was not as necessary for cementation parting, which operated at much higher temperatures than the acid parting process, with the active reagents being present as vapours rather than as liquids, Summary of the historical evidence A number of important points emerge from the broad survey contained in the historical Chapters 2 and 3 and in Appendices 3 to 5, together ranging over some very disparate subjects and sources. Surface enrichment ‘The deliberate surface enrichment of gold artefacts seems to have been a widespread practice almost since the incep- tion of goldworking. It also seems that corrosive salts12. Prologue — such as common salt, alum, iron sulphates and nitre ~ were used, which were capable of removing some of the silver from the gold, as well as the base metals. These salts ‘were subsequently to be the active agents in the cementa- tion refining processes, but at high temperature rather than in an aqueous medium. Everywhere, except possibly in India, the introduction of a gold coinage seems to have provided the stimulus for gold refining. Without the requirement of gold of guaran- teed purity for the specific needs of coinage, there was no incentive to refine gold. The surface of finished gold artefacts could be enhanced quite satisfactorily without the loss of weight attendant on full refining. Surface enrichment could be achieved much more easily with mineral acids, but cementation methods continued well into the twentieth century, long after cementation had been abandoned for gold refining. Cementation ‘The basics of cementation refining are very similar around the world. To a degree, this was inevitable, dictated by the physical chemistry of the process. A variety of active salts was used in antiquity, notably common salt and sulphates in the form of alum or iron sulphates. The chlorides and sulphates were used both alone and in combination; experiments performed in the eighteenth century suggest that a combination of salts could operate at lower tempera- tures. Nitrates in the form of impure saltpetre may have been used in ancient India, although apparently not in the West until the Middle Ages. Some recipes suggested the use of copper, lead or zinc sulphates, but these are unusual and uncertain. ‘There is little evidence for the use of either elemental sulphur or metal sulphides, such as marcasite or pyrites, for the refining of gold in classical antiquity, although they may have been components of recipes for surface treat- ments in late antiquity. The use of sulphur and sulphides, notably antimony sulphide (stibnite) only became preva- lent in the medieval world. Similarly, strong mineral acids were only employed from the Middle Ages in Europe and the Middle Fast, and even then were rarely used alone for primary gold refining until well into the post-medieval period. Some early recipes for gold-refining cements specify the addition of copper and even lead salts. These are repeated and expanded by many of the later accounts, There is a suggestion in some of the ancient descrip- tions that two-stage processes were used. This is certainly true of the first reliable and detailed accounts, such as that of Theophilus in the twelfth century and those of authors, such as Ercker, during the Renaissance. In the second, more rigorous refining stage of the cementation, known as graduation, copper was often added. Almost all of the medieval and later recipes state that reagents have to be mixed with an inert support medium, powdered brick, sometimes with additional clay, being very popular, but no such materials are found in the earlier descriptions, notably that of Agatharchides, c. 116 nc. The gold to be refined needed to have a large surface area exposed to the reagents. The granules of freshly won gold could be used as they were, or the gold could be ham- mered into thin foils, or be granulated by pouring the molten metal into water. The Probierbiichlein, compiled in the early sixteenth century, describes the addition of lead to the gold to facilitate the granulation, The common descriptions through the ages of the part- ing vessel concur that it should be of earthenware rather than of fineware ceramic. Some descriptions actually specify a cooking pot, and Agricola recommends a vase- shaped pot, not dissimilar to the Lydian coarseware cooking pots at the Pactolus North refinery which seem to have been selected for use as parting vessels. Similarly, several accounts describe the use of earthenware potsherds for various tasks. Almost all the descriptions specify a lid should be fitted and sealed in place with clay, although in the Japanese and some of the later Indian processes, the cementation mixtures were not contained in a pot at all. The function of the lid was to prevent loss of the volatile silver salts, and both Indian and Japanese commentators note the loss of silver. ‘The various descriptions of the parting furnaces, which become ever more detailed from late antiquity through the Middle Ages, have strong similarities both amongst themselves and with the furnaces excavated at Sardis, suggesting a long continuity. After parting, the gold was washed and melted, and the silver recovered from the spent cement. This was almost always done by cupellation ~ although note the reference in the Probierbiichlein to the amalgamation process for the recovery of silver from the parting cements. Cupellation was the standard process for the recovery and refining of silver throughout the Old World, and as such was not usually described in any detail in the early accounts of gold refining. The Indian saltpetre process is an exception, where it does seem that significant quantities of gold also had to be recovered from the cement along with the silver, and that a special process was necessary. ‘There are several excellent accounts of the salt cemen- tation processes in the major nineteenth-century metallur- gical texts. Percy's Gold and Silver: Part One, published in 1880, is pre-eminent, although by then the process hadceased to be used except, as Percy correctly stated, possi- bly in Japan. Percy’s long description includes an impor- tant and detailed section on the process in classical antiquity (pp. 397-402), which was the first major discus- sion on the subject. He concluded that a method of separating gold from silver was known and practised in the time of Strabo: but centuries before that period the ancients must have been able to effect such a separation; for other- wise it is impossible to account for the remarkable purity of the gold of many of their coins. Percy describes the chemistry of the processes in terms of fixed and immutable elements and of Dalton’s atomic theory. His descriptions are thus very different from the theoretical explanations of the processes that had gone before. They are, in fact, the first modern descriptions, and as such form an appropriate starting point for the scientific investigation of the remains excavated at the Sardis refinery. Amalgamation ‘The treatment of metals with mercury has a long history, stretching back into the first millennium sc, but there is no evidence for the extraction, in antiquity, of either gold or silver from their ores by amalgamation, There are descrip- tions from Roman times of the recovery of gold from gilded artefacts, but the earliest unequivocal descriptions of the treatment of gold ores with mercury to extract the gold are medieval. Silver in metallic form was already being recov- ered from its finely ground ores and slags in the later first millennium ap. The recovery of silver minerals from cupel- lation debris by amalgamation was a late medieval devel- opment, and the successful treatment of silver ores only began in the sixteenth century in the Americas, It is possible that before mercury was used to extract gold, lead may have been used in the same manner. There is certain archaeological evidence from Roman times for the use of lead to extract the gold from the auriferous slags created by the smelting of pyritic gold. It is possible that the minute particles of gold from primary deposits could have been extracted from the finely ground ore with lead, although the literary evidence for this is tenuous and as yet there is no archaeological evidence. Platinum group element inclusions Platinum group element (PGE) inclusions, which are so Prologue 13 apparent in some of the Lydian gold coins, were described by some ancient classical authors either as adamas or adamans. Pliny gives the most detailed descriptions, although confusingly describing gemstones at the same time. Plato mentions adamans and implies that it could be refined from the gold. The Mappae Clavicula, compiled in the post-Roman period, contains some elaborate recipes using lead, which describe the removal, recovery and even use of the inclusions. Assaying The ability to determine the composition of the gold was clearly essential to any meaningful refining process. Simple heating could reveal the presence of other metals in the gold if they were present in quantity, and the potential methods for quantitative determination were specific gravity, fire assay and the touchstone. ‘The principles of the specific gravity method seem to have been understood in antiquity, but were rarely, if ever, used. The method ascribed to Archimedes was correct in theory, but would not have been practicable. It was not until the development of proper precision balances during the Renaissance that the method could have been of practi cal application. Fire assay is a broad term covering both cupellation and parting methods. Cupellation alone could be used to estimate the quantities of base metals and as such seems to have been the standard method from remote antiquity in the Middle East, where the sources describe the weight loss of suspect gold on fire refining to constant weight. However, cupellation alone would be of no use to estimate the silver content, for which the fire assay would have had to include a cementation stage. This would involve considerable time, skilled effort and specialised equipment, and in practice does not seem to have been a common method. From classical antiquity at least, until the recent past, the usual method of assay was by touchstone. Prior to the introduction of inineral acids in the Middle Ages, touch- stones could only be really accurate with binary alloys of gold-silver or gold-copper and could not easily quantify alloys of gold with copper and silver in a combination that was usually intended to preserve the colour of the gold. ‘Thus, in practice, the ancient and medieval worlds had no routine effective method of determining the composi- tion of gold adulterated with a combination of metals. This probably explains the proliferation of artificial and alchemic golds.CHAPTER 1 Golden Sardis A. Ramage ‘The wealth of the Lydian kings and their capital at Sardis was legendary from their own time until the present day. The gold itself is no legend, however, and small quantities can still be recovered from the sands and gravels of the River Pactolus and other nearby streams. Most of the stories about the wealth of Sardis centred around Croesus, who was the last king of an independent Lydia. He served as the model among the Greeks for an extraordinarily rich man who was not able to achieve happiness or success in spite of his wealth. He was, however, assumed to be responsible for changing the Lydians’ use of coins made of an alloy of gold and silver to coins that were pure gold or pure silver. ‘The Harvard-Cornell excavations at Sardis have brought to light a series of installations that would have been capable of separating the major components of the mixed metal to enable the Lydians to produce the silver and gold coins for which they were so renowned, Sardis in context Gold and Sardis ~ the words are inextricably connected in the ancient record. Poets and historians frequently refer us to topographic details of Sardis and Lydia (Figs 1.1 and 1.2) and recount personal stories of its legendary kings. They cast the riches of Sardis, especially those of Croesus, who ruled the Lydian empire from 561 to 547 Bc, in an allegori- cal and moralising mode. The name of Sardis carried with it the weight of many ancient comments and the lessons to be learned from the progression beyond wealth to luxury and pride. Very little is known about other Lydian cities from archaeological excavation, although the names of ‘many places in Lydia are recorded by Herodotus and espe- cially Xenophon and Strabo, and several towns that issued coins during the Roman era seem to have retained their Lydian names. There are clumps of burial mounds within the region traditionally known as Lydia, but at present no definitive associations can be made with habitation sites of the Lydian period, which are very hard to discern in the modern topography.’ ‘The lesson regarding wealth was exemplified by the Greek historian Herodotus’ account of the meeting between Solon and Croesus, where the latter was discon- certed to find that riches were not to be equated with happiness, Croesus had asked Solon, a famous Athenian philosopher, who was the happiest mortal, and had expected him to say, ‘Croesus’. But no such luck, Solon had answered with the name of Tellus, an obscure dead Athen- ian, Solon maintained that no one could be called happy before his death because there might be some untoward incident to turn things upside down. Such a reversal did in fact happen to Croesus, who lost his son in a hunting accident and forfeited his empire to King Cyrus the Great of Persia as a result of his own over-confidence. Even though ancient critics and modern historians have shown that this specific encounter with Solon would have been impossible, because Solon lived a generation before Croesus, it nevertheless gives an excellent flavour of the Greeks’ attitude towards those who lived beyond their borders and did not speak Greek. The epithet barbaroi, literally denoting non-Greek-speaking groups, became more or less synonymous with Lydians and especially Persians. The Lydian kings and later the whole people, to the Greeks, were effete and given to luxury. This condition ‘was, supposedly, brought about by self-indulgence. Schol- ars have speculated that this was a rhetorical manoeuvre to show the Greeks as morally superior; it runs all through Greek history writing. Other Greek writers had the same attitude towards Lydian wealth and specifically their gold, at least as far back as the time of King Gyges, who had ‘usurped the kingly power in the early seventh century 3c,Golden Sardis 15 ~} ay Fig, 11a (above) Easter Mediterranean lands. Fig. L.1b (left) Sardis and central Lydia,16 Golden Sardis Fig. 1.2 Mount Tmolus and the region of Sardis. for the contemporary Greek poet, Archilochus, pointedly spurns both wealth and power:* I do not care for the wealth of Gyges rich in gold. Envy has never taken hold of me. I am not vexed at the divine order nor do I long for a tyrant’s power. These things are far from my eyes. ‘The poet Pindar brings out his own ambivalence towards {gold in one of his victory odes (Olympians 1.1), by declaring “Water is best..! but going on to describe the shining attrac- tion of gold. In context, this may have to be interpreted metaphorically, but the opposition of the two substances, water and gold, is in the same spirit as that of contrasting the simple Greeks and the luxurious Lydians. ‘The expansion of Lydian power at the expense of their neighbours to east and west seems to have been mainly the work of King Alyattes, the father of Croesus, in the late seventh and early sixth centuries. His predecessors, Gyges and Ardys, had been engaged in fighting off the nomadic Kimmerian tribes, who moved in from the Balkans and south Russia and ravaged western and central Anatolia for much of the seventh century.® The physical circumstances of Sardis are not referred to in Herodotus’ romantic account of Gyges’ coming to power and at present we cannot characterise a particular level in the excavations asbelonging exactly to the period of Gyges.* It was left for Croesus to use and display his riches at Sardis and, by his over-confidence, to become the unwitting cause of the great conflict between the Persians and the Greeks, which was the central subject of Herodotus’ enquiry in his History. Of the actual luxury of the Lydian court at Sardis itself, we have only a few concrete examples in jewellery (Fig. 1.3), fine metalwork and ivories. The rest must be imagined: the fine fabrics, furniture, scents and culinary delights alluded to but hardly ever described by the ancient writers of any era. In fact, the gold and its effects seem to have monopolised the attention of ancient authors, Excavated items, with their usual bias against organic materials, have to form the largest source of information about ordinary Lydians because the lives and habits of the kings and aristocrats, if that is how they should be charac- terised, are the only topics worth mentioning. Lydian metallurgy and the beginnings of coinage OF the flashiest source of this luxury that fascinated the Greeks, the gold bullion and coin of Croesus, we have recently had a tantalizing glimpse. Not the treasury, not the royal jeweller’s atelier, or even the mint has come to light. The Harvard-Cornell Sardis Expedition has, how- ever, found a yet more fascinating place: a workshop where the grains and dust of native gold were processed to produce the raw material for those fabulous treasures. In many ways, this offers more of a challenge than admiring the superb finish or awesome weight of the worked prod- ucts, because one can now wonder at the imagination required to bring the unlikely-looking raw material to a usable form, and at the skill required to produce gold of an almost perfect fineness. This fineness is all the more remarkable when one considers that the original raw material, alluvial gold, contained substantial amounts of silver and sometimes copper.’ The goldworking facilities had a finite life, even though the installations we found may have been reused several times. The product seems to have been made in small batches, to judge by the size of the furnaces. But small as the operation must have been, it supplies us with precious information about ancient gold-refining processes that has been found nowhere else. It has been assumed that what was retrieved from the Pactolus was a natural alloy of gold and silver that contains up to 40% silver. But the analyses presented here, as well as earlier studies, indicate that the composition was vari- able, some of the alloy having a much higher gold content. Golden Sardis 17 Fig. 1.3 Gold earring in the form of a lamb: length 10 mm, height 10 mm. It was found near the gold refinery at Sardis. We should allow for variation in the natural conditions that cause concentrations to build up in the gravels. Bolin,’ who claimed that the Lydian kings were debasing their currency, made an unwarranted assumption: that all the gold alloy found in the Pactolus was of the same composi- tion. If this were so, then his argument that the Lydians were purposely debasing the coinage might stand. But evidence would suggest that there was variation in the raw material from the Pactolus (and presumably the other streams), and that silver had to be added to create an artificial alloy, electrum, of constant purity. Thus, Bolin’s argument becomes unnecessary. Itis generally agreed that the Lydians invented coinage (as we understand it) for the Mediterranean world, but it is a matter of vigorous debate exactly when this happened. At some point in the seventh century sc — the range is from the early years to the very end of the century ~ they decided that creating and marking a series of small lumps of electrum at a consistent weight would be useful. The purpose for it or the advantage to be gained is perhaps less easily discovered than the date of its inception. A serious drawback to this system is that one cannot readily assess18 Golden Sardis the amount of gold within a lump, even though the colour isa rough guide. An even more important innovation was to institute a bimetallic series of coins, presumably using gold and silver separated from alluvial gold in the first instance, augmented by pure silver from other sources (Fig. 1.4, p. 129). Knowledge of this momentous change is laconically preserved for us by Herodotus, who reports simply that ‘they [the Lydians] were the first people that we know to employ minted coins of gold and silver...’ The recent discoveries offer evidence that may shed light, at least indirectly, on this question, By now the ultimate use of these lumps of metal as, coins has been declared, even if that was not the original purpose of their creation. The use of agreed weights of precious metals as payment or as expressions of wealth goes back in the Near Bast at least 2000 years before Croesus. The use of coins, however, and the implication of even a partial monetary economy, is much more recent. ‘The Lydians seem to have started this progression to a monetary economy with the electrum issues, and in essence perfected it with the new coins of gold and of silver. A difficulty for numismatists lies in finding an appro- priate time for initiating the electrum series, and King Aly- attes, the father of Croesus, is a frequent candidate for this honour." His hostile dealings with the Medes, whose terri- tory was directly to the east of Lydia, and the need to pay his own army, might have suggested coinage as a solution, Paying mercenaries has been suggested frequently as a reason for creating coinage. The logic is that even at a very low wage the lump sums owed by the month or at the end of a campaign would become considerable, and even the small electrum pieces represented a notable store of value in the seventh- and sixth-century economy. The next ques- tion is where to place the introduction of the bimetallic system referred to by Herodotus. ‘The first coins were made from natural alluvial gold to which silver was added to create an electrum containing approximately 55% gold, 45% silver and 1% or 2% copper.’ As Herodotus says, the Lydians created a coinage of pure gold and pure silver. This change is usually said (but without any explicit proof) to have taken place in the reign of Croesus, that is to say between 561 and 547 ac ‘There are strong contrary opinions, however, that put the initiation of Lydian gold and silver coins after the fall of Sardis and into the Persian era." ‘The same types of Lydian coinage were continued under the Persians at Sardis, but the actual change from electrum to separate coins of gold and of silver has been attributed by scholars to Croesus and an independent Lydia." The gold-working installations and the chronolog- ical indicators now discovered at Sardis make it clear that the parting of gold and silver, and thus the capability for issuing a coinage using the separate metals, was actually practised at the appropriate time for Croesus to have been the instigator. It is much more awkward to have the Persians, who had no experience with currency, suddenly take advantage of the Lydian innovation to issue a series of coins with Lydian devices before inventing their own stan- dard type. None of the Persian standard types is in fact like the early Lydian or the disputed type. Thus we can confirm the hypothesis that the Lydians under Croesus initiated the bimetallic system of coinage, and even pro- pose that the Sardis refining installations made its intro- duction possible. Herodotus’ description of the offerings of Croesus at Delphi makes it clear that the parting of gold and silver from natural gold must have been familiar to the Lydians and to the Greeks, as the use of the words lewkos (white) and apephthos (literally, boiled down), for natural and refined gold, indicates." The evidence of the processes used in the refining establishment and its probable date make the attribution of the change in the composition of the coinage to Croesus more likely than ever. Even in the early seventh century ac the Lydians must have had access to gold, whether tribute or locally pro- duced, both because of the specific reference of Archilochus, above, and because of more prosaic refer- ences to the gold mixing bowls of Gyges that could be viewed in the treasury of Cypselus at the sanctuary of Apollo at Delphi." Generally speaking, we suppose the gold was sepa- rated from the sands and gravels in which it lay by the traditional method of washing. It was in the form of ‘dust’ Fig. 1.5 Native gold from the River Pactolus.or quite small particles (Fig. 1.5). Nuggets are the excep- tion, which is why they have commanded so much atten- tion over the years. Howard Crosby Butler, who excavated at Sardis between 1910 and 1914, even mentions that his team occasionally found bits of gold when they were digging pits in the bed of the Pactolus." They dug over 1000 tombs, most of which had already been looted. Nevertheless, in the few that were undisturbed, they found many pieces of gold jewellery.” Sources for the Lydian gold ‘The source of the gold at Sardis is near the Palaeozoic massif of Mount Tmolus, ancient metamorphic rocks and intrusive veins of quartz, located to the south of Sardis, in a more recent conglomerate and its associated alluvium." In earlier times, the action of the River Pactolus could have concentrated the sparsely disseminated gold from the allu- vium (derived from the conglomerate) by actively rework- ing large quantities of the gravels. The present stream is sluggish, but tectonic movements such as the major Golden Sardis 19 Fig. 1.6 Temple of Artemis at Sardis, loking east tovrards the ‘Acropolis, Excavation house at lft earthquake of ap 17, which partially buried the Temple of Artemis (Fig. 1.6), situated by the side of the stream, may have also altered the old channel, thus permanently chang- ing the gradient.” ‘The ancient authors generally refer to the gold-bearing sands of the Pactolus or the Hermus as the principal source of the gold. The tale of Midas’ release from his ‘golden touch’ by washing at the source of the Pactolus tends to corroborate this idea."* Curiously enough, these sources do not make much mention of gold in connection with the Phrygians, in their heyday during the eighth century 1c, and finds of precious metals at the city mound of Gordion and in the tombs there have been meagre." There are, how- ever, ancient accounts of actual mining in the area of ‘Mount Sipylus and we have a personal report of ancient workings in the foothills of Tmolus, immediately to the south of Sardis.” For the time being, we have not been able20 Golden Sardis to identify the actual spots referred to. Certainly all the modern research to ascertain whether any gold remains in the area has concentrated on the conglomerate and the alluvium derived from it. These surveys have shown that the conglomerate forming the Necropolis hill on the west bank of the Pactolus at Sardis still contains gold, even if at ‘very low concentrations. We can assume that the same is true for the matching hill used as the acropolis of ancient Sardis, and in that case the Lydians and Croesus were quite literally sitting on a gold mine. There is no direct evidence of the organisation or whereabouts of other production centres. One cannot tell, for example, whether the conversion of locally mined gold dust was the main activity or whether refining ‘scrap’ coins was important too. Both forms of metal were cer~ tainly processed at the Pactolus North refinery. It is known ‘that all the streams issuing into the Hermus valley from the conglomerate lying up against Mount Tmolus continue to carry gold (see Fig. 1.2)."" We may infer that the quantity from each of these sources in ancient times matched that of the Pactolus, for which we have such vivid descriptions. There are scattered accounts of the actual mining of gold in some areas of Lydian-controlled territory, notably in the Troad near the town of Astyra on the Hellespont:* [Gold mines] are now scant, being used up, like those on Mt. Tmolus in the neighbourhood of the Pactolus River. Strabo XIII.1.23, translated by H.L. Jones Another passage lists the sources of various rulers’ wealth: The wealth of Tantalus and the Pelopidae arose from the mines round Phrygia and Sipylus; that of Cadmus from those round Thrace and Mt. Pangaeus; that of Priam from the gold mines at Astyra near Abydus (of which still to-day there are small remains; here the amount of earth thrown out is considerable, and the excavations are signs of mining in olden times); and that of Midas from those round Mt. Bermius; and that of Gyges and Alyattes and Croesus from those in Lydia and from the region between Atarneus and Pergamum, where is still a small deserted town, whose lands have been exhausted of ore, Strabo XIV5.28, translated by H.L. Jones Some of this text is corrupt, but it is interesting to note that the final area mentioned is not far from Adramyttion, where Croesus was ‘governor’ as a young man. Another snippet seems to confirm the Lydians’ hold over the Helle- spont: Abydus [near Astyra] was founded by Milesians, being founded by permission of Gyges, king of the Lydians; for this district and the whole of the Troad ‘were under his sway .. Strabo XJII.1.22, translated by H.L. Jones We may assume also that the Greek cities conquered by Croesus paid tribute in precious metal (presumably silver), but we cannot tell whether it would have come as bullion or coin, Value and quantity of gold The statements of Herodotus, if taken at face value, list more than 5 tonnes of gold given to the Oracle of Apollo at Delphi in the form of bricks or statues of a declared weight. Some have doubted that Herodotus was right or that Croesus actually did part with that much gold, on the grounds that the various measurements are irreconcilable with prevailing weight standards and the natural weight of the materials. We should not, however, regard the Lydian kings or Croesus as confidence tricksters, as some have suggested.” On the contrary, we maintain that Croe- sus was initiating a higher standard that would facilitate ‘commercial exchange with the neighbouring Ionians. ‘After the elaborate list of offerings at Delphi (some of which may have been discovered in the French excavations of 1939), Herodotus gives his account of the sack of Sardis and Cyrus’ treatment of Croesus. There are several differ- ent accounts of their relations, both upon the first capture of Croesus and subsequently. The story of Croesus on the pyre is attested very early in the fifth century ne by the painting signed by Myson on a red-figured amphora in the Louvre (Fig. 1.7). It beautifully matches Herodotus’ story, where Croesus is going to be burned to death by Cyrus because he was the enemy king. However, when Cyrus changed his mind, but could not put the fire out, Croesus prayed to Apollo, who sent a miraculous rainstorm that did, Bacchylides has the most coherent story, and Pindar mentions it too ~ all before Herodotus’ own account.” Herodotus also tells of gifts that Croesus made to various oracular sanctuaries of Apollo, and laconically observes that the dedications to Milesian Branchidai (i. the oracle of Apollo at Didyma) are equal in weight and similar to those at Delphi.” We should also remember that, in addition, Croesus contributed to a particularly expen- sive part of the Temple of Artemis at Ephesus ~ the column, drums carved with figures, of which many broken fragments are now in the British Museum (Fig. 1.8). Lydian riches were not confined to royal wealth, because Nicolas of Damascus recounts that when Croesus was young he tried to borrow a substantial sum of money from a private citizen by the name of Sadyattes.” Given hisFig, 1.7 Attic red-figured amphora by Myson, showing Croesus on ‘the funeral pyre, ¢, 500 ac: height 58.5 em, name, which is the same as that of Croesus’ grandfather, he was probably within the aristocratic circle, but his access to precious metal, even if it were not coined, adds to the impression that it was not closely held by the royal purse. Croesus had an unnamed citizen put to death for conspir- ing to elevate Croesus’ half-brother to the throne and confiscated the citizen's considerable private resources for his own dedications at Didyma.” The Pactolus North refinery is likely to have been capable of processing several hundred kilograms of gold each year, which would have been sufficient for tens of thousands of coins. Even so, it is likely that other facilities lay elsewhere, probably at the edge of town and in other valleys, since the fumes of lead oxide generated during the cupellation process, used to recover the silver, were extremely hazardous. Storage and use We have a quite explicit list of the tribute brought to Cyrus the Great by the various peoples of the Persian empire.” Almost all the assessments are in talents of silver, Golden Sardis 21 except for various contributions in kind and 360 talents of gold dust from India valued at 4680 Euboeic talents of silver. Herodotus tells a marvellous story about an Athenian aristocrat, Alemaeon, who made a fool of himself by putting on special clothes to get all the gold that he could carry away from Croesus’ storerooms. It goes like this: Alemaeon, son of Megacles, gave all the assistance in his power to the Lydians who came from Croesus at Sardis to consult the oracle at Delphi; and Croesus, when the Lydians told him of the good service he had rendered, invited him to Sardis and offered him, as a reward, as much gold as he could carry on his person at one time. Alcmaeon thought of a fine way of taking advantage of this unusual offer: he put on a large tunic, very loose and baggy in front, and a pair of the widest top-boots that he could find, and thus clad, entered the treasury to which the king's servants conducted him. Here he attacked a heap of gold dust; he crammed into his boots, all up his legs, as much as they would hold, filled the baggy front of his tunic full, sprinkled the dust all over his hair, stuffed some more into his, mouth, and then staggered out, scarcely able to drag ‘one foot after another and looking, with his bulging cheeks and swollen figure, like anything rather than a man. When Croesus saw him he burst out laughing, and gave him all the gold he was carrying, and as much again in addition Herodotus VI.123, translated by A. de Sélincourt This story indicates that much of Croesus’ store of gold was in the form of dust, which is not so easily used but it is more easily divided and apportioned than large ingots, especially when a considerable amount is to be allotted to the coinage. It is possible, also, for the parting to be done before the dust has been melted into lumps. The introduc- tion of reliable gold and silver coins allowed for easier provision of small but measured amounts of precious metal ina form that did not require a second weighing or melt- ing. This would have been much more economical but it is hard to prove, given that most of the metallic evidence from Sardis consists of thin pieces of gold foil of varying purity. They must have been melted once already before they could be hammered into foil. Herodotus says that Gyges donated six golden mixing bowls, weighing a total of 30 talents, to the oracle at Delphi, as well as an indeterminate but larger weight of silver and other gold items.” These items are described as “the Gygads’, but elsewhere the Gygads are taken to mean the coins of Gyges, just as Croeseids are supposed to be the coins of Croesus or Daries those of Darius.” This argument22 Golden Sardis, Fig. 1.8 The face ofa woman from one of the lower column drums of the Archaic Temple of Artemis at Ephesus, c. 530 acis difficult to sustain if coins had not been invented by this time, as most scholars claim. It is not necessary that all this ‘metal came from local sources; it could well have been trib- ute from the subjects of the Lydians’ widening empire. Gold, however, was less accessible than silver to contem- porary Greeks. Given the prominence of the Hermus and the Pactolus in the ancient texts, we may assume that some of the gold was local and that there was a big find in the time of Croesus or a more concerted effort to exploit the ore deposits. Certainly Sardis itself rather than Lydia as a whole is coupled with Bactria as the source for the gold used in decorating the palace of Darius at Susa.” ‘There is still a possibility that a much bigger area of goldworking lies in the unexcavated area to the north. ‘There are, however, two important questions here: Where ‘was the river in Lydian times? Would the Lydians have had the goldworks right in the middle of town? It is hard to imagine Herodotus not mentioning it in the section of his narrative where he described the burning of the city during the revolt of the Greek cities of Ionia against the Persians; there he explained that the agora or marketplace was divided by the river.” Digging at Sardis ‘The Harvard-Cornell team is not the first group to excavate at Sardis. In 1910, a team directed by Professor Howard Crosby Butler of Princeton University, but financed by the ‘American Society for the Excavation of Sardis, began work in the area of the Temple of Artemis. The clearing of a huge Hellenistic temple was not Butler's primary aim but he started at a spot known to have ancient remains, because of its two standing columns. It was also near the river, which ‘was said to have flowed through the agora or marketplace of the Lydian city and was, therefore, preferable to begin- ning with any of the several large Roman buildings, whose remains were still visible above ground.” The physical situation of the city at the crossroads of a traditional east-west route to the interior and down to the Aegean Sea and of a north-south route, as well as its mythological rep- utation, made it a prime target for archaeological investiga- tion. Unfortunately, the First World War interrupted the plans and an attempt to reopen the excavations in 1922 was foiled by the hostilities of the Turkish War of Indepen- dence. Even though Butler's team cleared the temple and investigated a great number of Lydian graves, the residen- tial areas of the Lydian city continued to be elusive. It was left to the Harvard—Cornell team to look once more for the Lydian city. There is, in fact, a tenuous but real connection between the Harvard-Cornell expedition and Butler's Golden Sardis. 23 team, because in 1914 Professor George Chase of Harvard joined them to study the pottery found in the previous four seasons. In 1938, citing the pressure of administrative duties, he asked Professor George M.A. Hanfmann to col- laborate with him. In 1948, Hanfmann visited Sardis and determined that most of the pottery and ‘small finds’ had been lost or destroyed and that there was not sufficient ‘material for a substantial publication as had been planned. He does say that this trip caused him to decide that ‘a new excavation at Sardis was needed’.” The new excavations, the Archaeological Exploration of Sardis, were set up by Professor G.M.A, Hanfmann (Fogg Art Museum, Harvard University) and Professor AH. Detweiler (College of Architecture, Cornell Univer- sity). The sponsoring institution was the American Schools of Oriental Research, where Detweiler was president. ‘The prime objective of any research at Sardis is the Lydian city of Croesus and his predecessors ... which remains completely unknown.”* Hanfmann emphasised the Lydian era because, even though the city was prosperous and influential under the Persians, the Hellenistic kings and the Romans, it never again matched the ‘Golden Sardis’ of the Lydian kings. When the current excavations were finally started, in 1958, the professed aim was to find and explore the remains of the Lydian city, which are, after all, the unique aspect of Sardis in its 1300 year history as an urban centre. ‘This had been Butler's aim also but first the clearing of the temple and then the war sidetracked his plans. ‘The original aims of the expedition have been achieved ina spectacular way and several important finds from the Roman era should be included in the list of discoveries, Detailed knowledge of just a few areas of the Lydian city has only shown how much more there is to be learned about its growth and status from the very beginning of the Lydian empire. The general urban development is poorly Known because, unfortunately, the monumentality and extent of the later buildings at Sardis itself meant that fre- quently the Lydian remains were obscured, or even altered or destroyed, by Roman complexes. These have required their own careful investigation and in several cases could not be removed just to facilitate exposing the Lydian levels. An example of this situation can be observed on the overall plan of the excavation sector Pactolus North, where the presence of a substantial late Roman townhouse pre- vented excavation of domestic units of the Lydian period (Fig. 1.9). We can assume the continuity of the Lydian levels beneath the Roman building because of the rich remains of houses, and of some kind of public building, under the excavated Roman street, just to the south. We cannot easily reconstruct these Lydian domestic remains24 Golden Sardis because the encroachment of the river at the west side has removed much of the evidence. The goldworks we have explored lies at the north-west corner of the excavated area known as Pactolus North (PN) (Figs 1.10 and 1.11).” The main industrial spaces represent only a small proportion of the area exposed at PN: roughly 900 out of a total of 2500 square metres, not including the purely Roman or Byzan- tine remains just referred to. Most of the excavating was done in the summers of 1968 and 1969, although small tests and a limited amount of analytical work were contin- ued thereafter. What we have uncovered must have been quite a small operation. It is unlikely that this one area represents the complete gold-refining operations of Croe- sus or any other Lydian king. Some of the precious metal produced at the PN refin- ery still exists in the form of coins and jewellery dug up ‘over the years by farmers and grave robbers; much of it is now dispersed in many museums around the world. In addition, untold amounts must still remain underground in Lydia itself, as the recent publication of Lydian treasure from looted tombs has shown." Finding the goldworks of Croesus was an unexpected and welcome bonus but much too specific as a research strategy. The field director might as well have announced that the expedition was looking for any one of the famous areas of the site as a specific purpose. In a sense, the approach from the very beginning could be said to have included looking for any of those desirable but elusive prizes. Fig. 1.9 Overall plan of Pactolus North. The Lydian remains are delineated with bold ‘stone-wall’ lines, the Roman and later remains in outline PACTOLUS NORTHGolden Sardis 25 Fig. 1.10 (eft) Pactolus North, looking south-east, The goldworks are in the foreground, the Acropolis in the left background. Fig. 1.11 (below) Pactolus North, looking north-west. The River Pactolus flows in the background, infront of the poplars and ather greenery.26 Golden Sardis Notes 1A. Ramage and N.tl, Ramage (1971, pp. 143-60). 2. Archilochus 15 (Pedley 1972, no. 40). 3, Herodotus I.15-16; Strabo 1.3.21, XIL4.8, XIV.1.40 (Pedley 1972, nnos 52, 49, 50, 51, with additional references), 4 Herodotus Sf. King Kandaules was so proud of his wife's beauty that he persuaded his bodyguard, Gyges, to observe her ‘undressing, She saw him and made it clear privately that he must cither kill the king and marry her or she would have him executed. He took the hint and founded the Mermnad dynasty. ‘Topkaya (1984, p. 116), Bolin (1958, pp, 22-37). Herodotus 1.94 ‘Weidauer (1975) discusses this at length in Chapter 2. Cowell etal (1998); Bolin (1958), See also pp. 169-73, 10 See p.97, note 15. 11 For example, Kraay (1976, p. 30}; Seltman (1955, p. 62). Main sources of evidence: Herodotus 1.94; Pollux, Onomasticon 1X.83 (Redley 1972, no. 80}, 12 Herodotus 94, 13 Herodotus 113-14 (Pedley 1972, no. 1), 14 Butler (1922, p. 16, foomote 1} 15 Published by €, Densmore Curtis (1925), and now in the Archaeo- logical Museum, Istanbul, 16 For the general regional geology, ee Brun et a. (1971, pp. 225-55), 17 Topkaya (1984) 18 Ovid, Metamorphoses X1.142. 19 Young (1981) and Kobler (1995), But we should be wary of absolute judgments in the light of the hoard of Lydian electrum coins found in 1963 and published by Bellinger (1968), and the fact that a final report on the buildings and finds from the city mound has not ‘yet appeared, 120 For Mount Sipylus: Pausanias VIL.24.7. For Mount Tmmolus: Letter from Clarence A. Wendell (US Minerals Attaché in Ankara) to G.M.A. Hanfmann, 11 August 1968: ...the other [photograph Bn 2 B 25 26 7 29 30 31 32 3 ey 35 36 37 38 borrowed ftom De Birgi isa photograph of an ancient mine opening on Mount Tmalus. The ancients followed a gold-bearing, ‘quartz-arsenopyrite vein in mica schist rock country, This ‘confirms my conclision that I had scen old mine dumps in this area when I took my stroll with Professor Detweiler in 1967! birgi (1949), ‘Attempts to reopen the mines were made in the late nineteenth ‘century but were unsuccessful Bolin (1958, p. 23) ‘Amandry (1939 and 1977). Bacchylides IL; Pindar, Pythians 1,184. Herodotus’ account isin Book 1.86, Herodotus 1:2. [Nicolas of Damascus, Jacoby (1923), 90 F 65 (Pedlley 1972, no. 64). Herodotus 1.92 Herodotus 11.90-97. Herodotus 1.14. Thirty talents weigh 1135 kg. Pollux, Onomasticon I.87: “Highly prized is the Gygaean gold, as are the staters of Croesus: (Pedley 1972, no. 80) Dsf (Darius, Susa F) 35-55 (Pedley 1972, no. 303, with additional bibliography and explanations), Herodotus V.101f: [The fonians] were prevented from sacking the place ... The Pactolus is the river which brings the gold dust down from Tmolus, It flows through the market at Sardis, and then joins the Hermus, which, in its turn, flows into the sea’ (Translated by A.de Sélincourt) Butler (1922, p- 3. Hangmann (1972, p. 10) -Hangimann (1972, p. 11) quoting from his grant proposal to the Bollingen Foundation ‘This slightly odd name is meant to distinguish it from Pactolus Clif, a smaller sector about 200 m to the south of Pactolus North, agen and Gztark (1996)CHAPTER 2 Historical Survey of Gold Refining 1 SURFACE TREATMENTS AND REFINING WORLDWIDE, AND IN EUROPE PRIOR TO AD 1500 PT. Craddock There are no detailed accounts of gold-refining processes extant until several centuries after gold production had ceased at Sardis. Even the one surviving ancient account, made in the second century xc, which does contain some practical detail,"* is very possibly conflated (see pp. 34-5) Otherwise, the descriptions from classical antiquity are sparse in the extreme, such that in some instances it is not even certain that gold refining is being described.’ Most of the literature considered here relating to the ancient processes originates from Burope or the Middle East, but there are other important early sources, notably from India, that are sometimes overlooked by classical scholars.* However, through the Middle Ages and the Renais- sance there are descriptions of the gold-refining processes in ever greater detail.” From a careful study of these and the remains excavated from the refinery at Pactolus North (PN) in Sardis, it seems that the procedures changed but little over the centuries and across continents. Thus the written descriptions of the methods, together with the rationale for carrying them out given by the more recent technical authors, are invaluable for our understanding of the ancient processes, This chapter concentrates on the historical evidence for the refining of gold by cementation prior to the Renais- sance, Other related chemical processes in the separation, assaying and refining of gold are dealt with separately in Appendices 3 to 5. The parting processes left the purified gold as metal, but the silver which it contained as an impurity was left in the form of silver salts from which metallic silver had to be recovered. At the PN refinery, this was carried out by the process of cupellation, which was the standard method used throughout antiquity to recover silver both from base metal and its ores, The production of silver was not the prime concern of the PN refinery, and because of this and because there is an enormous literature on cupellation methods, including several detailed surveys," the subject will not be considered further in this chapter. By bringing together the literary evidence from around the world, some features which might individually be ignored or judged as erroneous, gain significance when they are found to appear in a number of independent accounts. These include such varied subjects as the intro- duction of lead and other base metals or metal salts at early stages in the processes, and an awareness of the platinum element group inclusions in gold (Appendix 4, p. 238). These subjects, taken in conjunction with the finds from the PN refinery, suggest new interpretations that have transformed our knowledge of the early methods of col- lecting and refining gold, Surface enhancement of gold artefacts: the precursor to gold parting ‘The chemistry of the processes employed to enhance the surface of base gold are closely related to those used in the gold cementation processes, and thus it is appropriate to discuss them as a preamble, especially as the techniques considerably predate the first true refining processes. Although complete separation of silver from gold was probably not practised in antiquity prior to the Lydian period, there is some evidence for at least the partial removal of silver from gold alloys right back almost to the inception of the use of gold. Thus Shalev’ believes that the cast and hammered gold rings found at the Nahal Qanah Cave, dated to the fourth millennium nc, were deliberately treated to enhance the gold content at the surface. Specific gravity studies suggested that the metal of the rings usually contained between 65% and 75% by weight of28 Historical Survey of Gold Refining (1) gold but at the surface this had typically risen to between 80% and 95%, the remainder being of silver. This might be thought of as nothing more than the usual surface enhancement that occurs during prolonged burial, but the detailed study of the surface topography by scanning electron microscopy suggested to Shalev that the light hammering (in fact it seems to have been more of a plan- ishing operation than true hammering) had occurred after the surface enhancement. He postulated that the rings: could have been passed through a surface oxidizing process, perhaps in an open fire on salty sand with NaCl. After the surface oxidation, the objects could be cleaned in natural organic acid to obtain the desired golden yellow colour, and their surfaces hammered in order to achieve their final shape and to regain their bright appearance, Further evidence for the deliberate adulteration and surface treatment of gold has now been discovered, dating from the third millennium xc. The three famous gold chis- cls excavated from the royal cemetery at Ur’ have recently been examined after it was noticed that the golden surface was peeling away in some areas, revealing much more coppery metal beneath.’ Analysis of the larger chisel, Registration No. WAA 121349, showed the body metal at a depth of 1 mm typically to contain about 45% gold, 10% silver and 45% copper. By contrast, the surface examined had a gold content of 83%, with 9% silver and 8% copper. Clearly, both the copper and the silver had been preferen- tially removed from the surface layers, The surface was compact and heavily burnished, showing that the deple- tion had taken place before the final polishing of the gold and not whilst the metal was buried in the ground. Thus the chisels are a good and very carly example of the process known as depletion gilding, ‘These are very similar to the better known processes of Central and South America, including depletion gilding of the tumbaga alloys." In some of these treatments, copper alone was removed from the surface using organic acids derived from various fruit juices, but in other processes copper and silver were removed, which required much more active chemicals. According to descriptions made by the Spanish priest Fray Bernardino de Sahagiin in sixteenth-century Mexico," this was done by a combina- tion of the application of alum and of pastes comprising a yellow earth and common salt, heat, and the burnishing of the base gold: ‘And when it was cast, whatsoever kind of necklace then it was burnished with a pebble. And when it had been burnished, it was additionally treated with alum; the alum with which the gold was washed [and] was rubbed was ground. And a second time the piece entered the fire; it was heated over it. And when it came forth, once more, for the second time it was at once washed, rubbed, with what was called ‘gold medicine’. It was just like yellow earth mixed with a little salt; with this the gold was perfected; with this it became very yellow. And later it was polished; it was made like flint, to finish it off, so that at last it glistened, it shone, it sent forth rays. In this treatment, it is likely that the alum was potassium alum, and the yellow earth was a hydrated ferric sulphate, Which are all very corrosive. Before the distillation of min- eral acids in the Middle Ages, corrosive sulphates or nitrates were widely used in the Old World as well as in the New. In the classical technical and alchemic literature, impure ferric sulphates were often referred to as misy (see p. 36). Bergsoe" carried out replication experiments based on the methods described by Sahagin, using alloys of gold and copper. He obtained a satisfactory surface with alloys containing 50% and even 25% gold, but the enriched surface peeled off an alloy containing only 13% gold. Techtmann' also carried out some experimental treat- ments to replicate the early American treatments, using an artificial alloy containing 60% gold, 30% silver and 10% copper, an alloy not far removed from the Sardis alluvial gold apart from the rather high copper content. The metal was hammered out and in the first experiments treated with a cement of brick dust and common salt in a sealed container, heated to about 350 °C, After only 10 minutes, a surface layer of almost pure gold had formed, and if the process was continued for much longer there was a tendency for the surface to peel off. Similarly, the gold-refining experiments of Notton” using common salt alone as the reactant demonstrated how easy it was to sub- stantially remove the silver at quite moderate tempera~ tures, Lechtmann thought that the cement in her first experiments had been too strong. Indeed, it was more reminiscent of the traditional European cementation salt parting processes rather than the Mexican process where the cement was just smeared on the surface of the metal, and with no mention either of a container or of heat in the second stage. One suspects that Lechtmann did not appre- ciate beforehand just how easily the silver could be removed. Accordingly, in the next experiment, an alloy containing 40% gold, 48% silver and 12% copper was coated with pastes of either ferric sulphate and common, salt or ferric sulphate, common salt and ferric oxide, and left at room temperature for two days. After this, the paste ‘was washed off, revealing a surface blackened by copper and silver oxides. These were removed by washing in hotbrine, exposing a reddish-brown, rather porous surface which could be consolidated by burnishing, giving a smooth compact yellow surface. Lechtmann also suggested that heating to about 270 °C achieved the same effect, but it is not clear why this should be so, indeed the heat would be more likely to cause diffusion of silver from the interior to the surface, effectively reversing the whole treatment. This method was much less rigorous and did not totally remove the silver except in the very surface, and would seem to have replicated the surfaces on the Ur chisels much more closely than the first experiments had done, There are several descriptions of surface enhancement processes in some of the late Roman technical manuscripts such as the well-known Leyden papyrus, which was found at Thebes in Egypt: Recipe 14 For treating gold, or for thoroughly purify- ing it and making it brilliant. Misy 4 parts; alum 4 parts; salt 4 parts. Grind with water and having coated the gold with it place it in an earthenware vessel put in a furnace and luted with clay and heat until these sub- stances have become molten, then withdraw it and scour carefully. ‘These methods of surface enhancement were often used on coins. Thus the thirteenth-century Islamic treatise of ‘Mansiir ibn-Ba‘ra describes how the gold flans were chem- ically treated before being struck:” After the gold flans are made round, they are placed in a pot and heated to redness. Pulverized Indian salt in a little fresh water is added to it. Then a fire is kindled under it with such intensity that the salt dissolves (melts?) as lead does. It [the molten salt] is decanted into a mould; then the dinars are removed to be washed with cold water, then fine sand. They are dried ina pot above a gentle fire and then stamped. ‘The Indian salt referred to is likely to have been saltpetre, principally potassium nitrate (KNO,), together with a little sodium chloride and ammonium salts." It would melt at quite low temperatures and remove some of the silver from the surface by the formation of silver nitrate and oxide." It is very likely that the ‘Indian salt’ is to be identified with the Indus earth of the Arthasastra (described on p. 34), which ‘was also used in gold treatments at a much earlier date. Some of the traditional Japanese methods for surface enhancement of gold coins were observed by Gowland during his work at the Imperial mint in the later nineteenth century, and his descriptions are valuable for giving an insight into the ancient processes: Owing to the large proportion of silver which the coinage alloys contained, the coins were still nearly Historical Survey of Gold Refining (1) 29 White in colour, and it was necessary to give them a surface of gold. This was effected not by any ordinary gilding process, but by dissolving out the silver from the upper layers of the alloys (Fig. 2.1). The coins were first painted with a mixture composed of iron and copper sulphates, potassium nitrate, calcined sodium chloride, and resin made into a paste with water. They were then carefully heated to redness on a grating fitted over a charcoal fire (to convert the surface silver to silver chloride). After this they were immersed in a strong solution of common salt (to dissolve the silver chloride), washed with water, and dried. Their surfaces now consisted of a layer of pure gold. This process was followed by the old workers in gold, and with trifling modifications is still practised at the present day. During his travels in West Africa in the early nine- teenth century, the explorer Mungo Park observed gold dust being melted under the ashes of corn cobs, which were alkaline. This was done to give a better surface appearance to the metal when it had solidified.” Following the introduction of mineral acids in the Middle Ages, surface treatments with mineral salts were only gradually replaced by acid treatments to remove copper and silver from the surface, and jewellers contin- ued to use the older methods until comparatively recently, even in Europe. Thus, for example, Howard,” in the late eighteenth century, described gold ‘cleaning’ as follows: Some boil the pale (i.e. silver-rich) gold in a solution of these salts (verdigris, i.e. copper acetate, sal-ammoniac with vitriol, ie. iron sulphate, or alum) made in vinegar; others moisten the salts with vinegar into the consistence of a paste, which is spread upon the gold, and the metal laid on burning coals till the mixture is bumt off. The sprinkling of the salts in powder upon the moistened gold, practised by some, does not answer so well. In the early nineteenth century, Gill” stated that: It is a very curious circumstance, that the best work- men in this branch of jewellery have at this day no other menstruum for giving the last high finish in colour to their beautiful articles, than the compound salts of alum, nitre and common salt ‘These tended to be traditional recipes of family firms and as such were closely guarded secrets that were often lost on the demise of the firms. Bury* has reported one of the very few that was published. The gold artefacts were boiled in an aqueous solution containing equal amounts of alum and common salt and twice the amount of saltpetre,30 Historical Survey of Gold Refining (1) Fig. 2.1 Colouring gold coins in the Japanese Imperial ‘Mint, mid nine- teenth century. (From W. Gowland, ‘Trans. Proc. Jap. Soc. 13 (1915), facing p. 32) ‘The resulting surface depended on the duration of the treatment, but afterwards the surfaces were treated by scratch-brushing or burnishing to give the required finish. Much the same effect could be achieved more easily with mineral acids but it was claimed that these had a ‘dimin- ished lustre.” The technical author Alfred Hiorns was much con- cerned that many traditional processes of metal finishing treatments were in danger of being lost through not being published. As head of the Metallurgy Department of the Birmingham Municipal Technical School around the turn of the nineteenth century, he was well placed to record the treatments. In his Mixed Metals he describes the treatment of gold alloys both with acidic solutions of salts, such as potassium nitrate and common salt, and with anhydrous salts used in a molten state." The mixtures of salts speci- fied include two parts potassium nitrate to one part each of common salt and alum, and another recipe containing equal parts of potassium nitrate, sal-ammoniac and borax. For the former mixture Hiorns states that: ‘These substances are ground to a fine powder, well mixed and placed in a previously heated blacklead (graphite) ‘colour’-pot .... It is well to get the pot nearly red-hot before placing the ‘colour’ in it. The mixture must then be constantly stirred with an iron rod. It will first boil up as a greenish liquid, then solidify, and afterwards boil up a second time and become thoroughly fused, having 2 brownish-yellow. colour. At this stage the work, which has been previ- ously annealed and dipped in dilute aquafortis (nitric acid), is dipped in the ‘colour’, being suspended on a silver or platinum wire, the latter being preferred, and kept in motion for about a minute and a half, then immersed in boiling water containing a little aquafor- tis. The immersion and swilling are again repeated, when the articles possess a beautiful colour. They are then washed in hot water containing a little potash, and finally dried in warm boxwood saw-dust. Surface enhancement processes continued to be described through the twentieth century. Thus Fishlock’s standard book on metal finishing” describes cements of alum, potas- sium nitrate, zinc sulphate and sodium chloride. These were mixed to a slurry in which the objects to be treated were dipped or it was applied as a paste. The coated objects were then heated over a charcoal or coke fire and plunged into water. The use of zinc sulphate is interesting, and recalls its inclusion in some parting mixtures listed by ‘Agricola and Ecker, and possibly in some of the Islamic sources." Alternatively, the objects could be immersed in the molten reagents until the desired degree of surface enrichment had taken place. Another treatment was with a boiling aqueous solution of potassium nitrate, alum and common salt, in which the objects to be treated were immersed, for periods of between 10 and 20 minutes.‘These recent treatments would have been familiar to crafts- men treating base gold millennia before. The inception of gold parting The evidence from remote antiquity for the deliberate surface treatment of gold artefacts which removed both silver and copper, coupled with the comparative ease with which silver could have been totally removed from gold using either common salt or a salt/corrosive iron sulphate mixture, suggests that gold refining was well within the technical capabilities of the ancients long before the Sardis refinery. Thus the possibility must be considered that true gold refining has a much longer history. First, it must be understood that there was no a priori reason for the ancients to refine gold or to consider what ‘came out of the ground as impure. Our idea of pure gold as a single and precisely defined element is based on the rela- tively modern scientific concepts of the nature of elements, and in particular on the Law of Constant Composition, by which each element has precise, invariant and unchange- able properties. To us, this seems no more than stating the obvious, but the ancients did not have such a concept of an ultimate, pure elemental material.” Thus metals such as gold, coming from various sources, could have widely differing properties but still be gold. Given the widely held belief that metals ‘grew’ in the ground, it would seem only logical to expect the properties of the metal to depend on their environment. For example, a light-coloured natural alloy from a given locality would be regarded as the gold of that place, rather than necessarily a natural alloy of rather high silver content. Refining would be a long and expensive process to improve the colour but reduce the weight of the gold, whereas surface treatment would be much easier and enhance the colour whilst preserving the weight. The discovery of the application of depletion gilding to the Ur chisels shows that, even in the third millennium nc, the technology existed to remove at least some of the silver from the surface of gold-silver alloys. Clearly, the whole object of the treatment was to make the metal look golden at the surface whilst preserving as much of the ‘weight as possible. ‘There seems to have been genuine confusion in antiquity between the status of ‘real’ native gold alloys and adulterated golds or electrums, even though they could be physically and chemically identical. This confu- sion persisted right through into the post-medieval period and beyond, as exemplified by Ercker’s comments on the improvement or graduation of gold (see p. 64). As Historical Survey of Gold Refining (1) 31 discussed in Appendix 5, p. 247, the principal method of assay was by the touchstone, which was essentially a sur- face method. The method of specific gravity measurements, which seems much more obvious and infallible to us, was apparently little employed, long after Archimedes’ famous, if apocryphal, discovery of the principle of the method, until balances of sufficient accuracy were developed during the Renaissance in Europe. Thus although the pro- cesses of gold refining were likely to have been known from remote antiquity, little or no natural gold alloys were treated until the introduction of coinage created the need for pure metal or at least metal of consistent quality. In short, this was a technology awaiting a use. Direct evidence of gold refining Perhaps the most direct evidence for the refining of gold in antiquity is the purity of the surviving metal. In general, ancient goldwork contains appreciable silver compatible with unrefined natural gold, but Nicolini,” for example, has claimed that the high purity of some ancient gold arte- facts indicated they had been made of refined metal. How- ever, these high-purity pieces are not part of a consistent series but rather isolated examples amongst otherwise impure metal. There is the additional complication that the analyses themselves may not be representative of the true composition of the metal.” From such analyses of gold antiquities as have been performed, there is no evidence prior to the introduction of coinage for the use of a true cementation process that would have totally removed silver and any other metals from throughout the gold. Without coinage, there was, of course, little incentive to refine and thereby reduce the weight of metal. In an interesting paper, Ogden” has shown that there are marked differences in the composi- tion of gold artefacts made by those who produced a gold coinage and those who did not. It would seem that before coinage was adopted the unrefined gold was employed with little or no treatment, but where coinage was pro- duced then, although the gold used in artefacts might still be base, it was usually deliberately debased by the addi- tion of regulated quantities of silver and/or copper to refined gold. Particularly striking is the comparison between the composition of the gold artefacts of the Greeks, who had gold and silver coins, and that of the gold artefacts of their contemporaries the Etruscans, who did not. Similarly, the pre-Colombian civilisations, expert in the surface treatment of quite base gold alloys (see p. 28), certainly had the technical knowledge to purify gold, yet this was never done. Without the concept of coinage, there was no motive.32. Historical Survey of Gold Refining (1) Indirect evidence of gold refining ‘The material excavated at Sardis is the earliest surviving physical evidence for the parting of gold and silver. Other, more indirect archaeological and literary evidence has been put forward to suggest that gold parting was prac- tised even earlier. These claims will now be discussed. ‘There are many ancient texts from both Mesopotamia and Egypt referring to the refining of impure gold."* Unfor- tunately, they are accounting records of quantities rather than technologies, thus usually they state no more than the weight of the metal before refining and the weight loss after refining. Without exception, no details are given of either the metal with which the gold was debased or of the refining process, beyond that it was often performed by fire in the furnace. Sometimes the gold was subjected to ‘more than one refining operation, as in this example of the first millennium »¢ from the archives of Erech:” Five minas of gold, five weight measures of revenue of the king were put into the fire, At the first firing two thirds mina, five shekels of gold disappeared. It was reduced to four minas fifteen shekels. In the second firing, half a mina, two shekels of gold were lost to yield three and two thirds minas, three shekels of gold. It might be thought that as gold was naturally alloyed with silver and that generally ancient goldwork only contains small amounts of copper and other metals, it is most likely that these refining operations refer to some form of parting process by which the silver was removed. However, gold alloyed with copper was quite widely used in both Mesopotamia and Egypt at various times, and copper or lead would be the obvious metals with which to debase gold fraudulently. There is also some evidence, discussed in Appendix 3, p. 235, that lead may have been used in antiquity to extract very finely divided gold from the ore, and the resulting gold could still have contained some lead. ‘Thus, in all probability, the early records of refining opera- tions refer to cupellation, which would remove base metals such as copper and lead but not affect the silver. It is noticeable that there is no reference to the recovery of the metal removed from the gold, rather suggesting that it was not silver. Even in Roman times, Pliny could still note that gold was refined by roasting with lead, that is a cupellation process,” although elsewhere he describes the true parting processes (see pp. 35-6). The general consensus of the lit- erary sources would seem to be that gold-silver parting was not practised before the mid first millennium sc — that is, the time of the Sardis refinery." ‘A different approach has been suggested by the requirements of gold used for the production of gold leaf. It has been stated that in order to produce true gold leaf” gold of a high purity was required, and thus the occur- rence of very thin gold leaf could be taken as an indication that gold refining was being practised. Thus Oddy states that: ‘The invention of gold leaf was impossible before the perfection of methods for the purification of gold as only pure gold, or gold rich alloys, free from certain impurities can be beaten out to produce the thinnest, leaf. The introduction of gold refining is generally put at around 2000 nc in Mesopotamia, but it is quite possible that a better estimate could be made from a study of the dates of surviving objects which are covered in {traces of} gold leaf Gold was beaten into thin sheets or foils from the very beginning of its use, and by the second millennium sc in Egypt true gold leaf was being produced." However, although ductile metal substantially free of lead or copper would have been a prerequisite and could have been produced by cupellation, it is quite possible to produce acceptable gold leaf from gold containing small amounts of copper and appreciable quantities of silver, and some of the gold leaf from antiquity is decidedly impure.® Thus the introduction of gold leaf cannot be taken as an indication that gold parting was being practised Literary evidence for cementation parting Unequivocal literary references to, or descriptions of, the parting processes ail post-date the Sardis refinery, but do show that the technology was in use around much of the Old World well before the end of the first millennium ne, usually where a gold coinage was also in production, the apparent exception being India. None of the descriptions is very detailed, and some- times they are difficult to interpret or fully understand, due variously to their brevity, style or translation. In the main, they seem to describe the treatment with common salt and/or corrosive sulphates, often in the presence of added base metal. Modern commentators, who have tended to be working on individual texts in isolation rather than taking all the texts together, have suggested that the use of mixed reagents and especially the presence of additional base metals is likely to be an error. However, instructions listing these ingredients recur again and again, including the much later reliable and detailed accounts from the European Renaissance (given on pp. 58-9 and 63-4), Taken with the evidence for the presence of both lead and copper, apparently at early stages in the process atthe Sardis refinery, these early references now require care- ful reappraisal. Herodotus (1.14 and 50}* describes in some detail the donations of precious metal made by both Gyges and Croe- sus to various temples and sanctuaries: Gyges offered a lot of silver, but also much gold. (114.1) Croesus offered half-ingots, six palms long and three palms wide and one palm thick. (1.50.1) Their number was 117 and four of these were of purified gold, each of the weight of two and a half talents while the other half-ingots of white gold had a weight of two talents. (1.50.2) It is maybe significant that whereas the offering of Gyges are just described as being of gold, Herodotus specifically states that some of the donations made by Croesus were either of white gold (which could be a reference to the alloy containing about 45% silver found in the electrum coins) or of refined gold. Taken together, these perhaps suggest that the refining process began with Croesus. There are several early descriptions relating to the purification of gold from India, although all are rather confused. A lost work, Indika by Ktesias, written about 400 nc, apparently contained descriptions of both gold refining and the production of crucible steel, but these seem to have got conflated in the surviving abridgement made by Photios, a Byzantine bishop, some 1200 years later.” The section on gold reads: Concerning the Spring which is filled every year with alluvial gold, from which 100 earthen pitchers are drawn up annually. The pitchers must be earthen since the gold when withdrawn is in a solid state, and it is necessary to break them to extract it. The spring is of a square shape with a perimeter of 16 cubits [i.e. 24 ft] and a fathom [ie. 6 ft] in depth. Each pitcher weighs a talent [i.e. 57 Ib]. Commentators have long considered the so-called spring should be regarded as a furnace, but suggested that it should be iron that was in the pots, as ‘storage (of gold) in pitchers would not affect its physical condition’ However, if the spring was indeed a furnace, then treatment of alluvial gold in the solid state in sealed earthenware pots sounds very like the later descriptions of the cementation parting of gold. The furnace dimensions are rather large, but the square shape is similar both to the excavated PN furnaces and later descriptions given below and in Chapter 3. Several descriptions of the treatment of gold, includ- ing refining processes, are to be found in the Arthasastra in the sections outlining the duties of a mine manager.” There Historical Survey of Gold Refining (1) 33 are several references to the purification of gold, including the following at 2.13.5-8: OF the best [varieties], the pale-yellow and the white are impure, He should cause that because of which it is impure to be removed by means of lead four times that quantity. If it becomes brittle by the admixture of lead, he should cause it to be smelted with dried lumps of cow-dung. If it is brittle because of [its own] rough- ness he should cause it to be infused in sesamum-oil and cow-dung. This should be a reference to the cupellation of gold to remove base metals such as copper, but there is a problem in that gold normally contains silver and it does specifi- cally state that the impure gold to be so treated was pale or white, which strongly suggests that silver was indeed the impurity, and of course cupellation would not remove it, However, the whole description is similar to the much later passage in the Mappae Clavicula, where it is implied that it was the removal of the silvery-white platinum group ele- ment inclusions that is described." The quantities would be appropriate to form the gold-lead intermetallic com- pound, Pb,Au, and of course cupellation would remove the PGE inclusions. The very next section of the Arthasastra at 2.13.9 contains a direct reference to the refining of mined gold: [Gold] produced from the mines, becoming brittle by the admixture of lead, he should turn into leaves by heating and cause them to be pounded on wooden anvils, or should cause it to be infused in the pulp of the bulbous roots of the kadalt and the vajra plants. ‘Once again, mine gold should contain silver not lead, unless the lead was being deliberately added to the finely ground gold ores to gather up the precious metals in a process akin to the later liquation process, only operating on the powdered ore instead of on molten copper, as dis- cussed in Appendix 3, p. 235. Hammering out the impure gold into leaves could be significant in that it forms one of the stages in parting processes, but this does not seem to be a reference to a cementation process. The most convincing reference to true cementation is at 2.13.47 in a section on gold working rather than gold production, and would be the earliest description of the cementation process if the Arthasastra, or at least this passage of it, is indeed of the fourth century sc: Ornamental gold of the best kind, possessed of excel lent colour, passed through an equal amount of lead, turned into leaves by heating [and] made bright with Indus-earth, becomes the base of blue, yellow, white, green and parrot-feather colours.34 Historical Survey of Gold Refining (1) Indus earth is normally taken to be the saline soils from the great river valleys of India that were latterly collected at specific times of the year and used as a reagent to patinate the zine-rich bidri wares," or as a feedstock for the production of gunpowder.” The earth is typically rich in common salt, nitre and ammonium salts, and would form an ideal cement for the parting process (see p. 29 for the use of ‘Indian salt’ in medieval Islamic gold treatments) ‘The further reference to the prior treatment with lead, for whatever purpose, is interesting. Thus the sequence would seem to be cupellation, hammering into foils, followed by parting with saltpetre. This sequence is strikingly similar to the stages one believes took place at Sardis to refine serap gold: cupellation, hammering into foils, followed by cementation, but using common salt instead of saltpetre. Returning to the classical world, reference is made to gold refining by Plato in the Politicus. However, it is only given as a metaphor, and from this the impression is con- veyed that the process was difficult.” No technical detail is given, although copper, silver and even the PGE inclusions were apparently to be removed. ‘The best known and certainly the most detailed description of the cementation process as practised in antiquity is that quoted by Diodorus Siculus” in the first century nc, froma more detailed but now lost work, On the Erythraean Sea, by the grammarian and geographer Agath- archides of Cnidus, which was written in the second cen- tury nc, and apparently based on first-hand observation.” ‘The whole quotation is on gold production in Egypt. The first part is a quite detailed description of the mining of the gold using firesetting to weaken the host rock. Fire- setting was only used against hard homogeneous rock and thus it was the mining and processing of gold from a primary quartz deposit that was described here rather than the working of secondary placer deposits such as those from the River Pactolus. This could be of significance in the choice of processing technique adopted, particularly those to separate the gold from the gangue, as the gold in the primary deposits would tend to be much more finely dispersed than in the secondary placers. Note also that gold from a primary source would have no PGE inclusions (see Appendix 4, p. 238). The description continues with the beneficiation and purification processes: In the last steps the skilled workmen receive the stone which has been ground to a powder and take it off for its complete and final working; for they rub the marble (ce. the mine gangue) which has been worked down upon a broad board* which is slightly inclined, pour- ing water over it all the while; whereupon the earthy matter in it, melted (washed) away by the action of the water, runs down the inclined board, while that which contains the gold remains on the wood because of its weight. And repeating this a number of times, they first of all rub it gently with their hands, and then lightly pressing it with sponges of loose texture they remove in this way whatever is porous and earthy, until there remains only the pure gold-dust. ‘Then at last other skilled workmen take away what has been recovered and put it by fixed measure and ‘weight into earthen jars, mixing with it a lump of lead proportionate to the mass, lumps of salt and a little tin, and adding thereto barley bran; thereupon they put on it a close-fitting lid, and smearing it over carefully with mud they bake it in a kiln for five successive days and as many nights; and at the end of this period, when they have let the jars cool off, of the other matter they find no remains in the jars, but the gold they recover in pure form, there being but little waste. ‘This is clearly a description of the salt cementation process,” and equally clearly it is related to the process practised at the PN refinery some 400 years earlier. It is noticeable that the otherwise complete description omits any mention of turning the mined gold into foils prior to parting, strongly suggesting that the gold dust was treated in that condition. There is in fact only one ancient western source that states that the gold dust should be made into foils for the parting process.”* The remains from Sardis, as shown in subsequent chapters, also suggest that the freshly mined gold was treated directly without first being converted to foils. ‘The quite unambiguous additions of controlled quan- tities of lead and also of some tin are noteworthy, especially in view of similar statements in some other early descrip~ tions and the presence of lead on the surfaces of the part ing-vessel sherds from the PN refinery, and of tin on some of the gold globules and a foil (see pp. 112 and 186} The function of the lead and tin in the cement has always proved problematic, and hence the accuracy of the whole quotation has been questioned. Convincing modern parallels are hard to find. Percy records a process practised up to the nineteenth century at Oker, near Goslar in the Harz mountains (Germany), for the recovery of trace amounts of gold from silver using litharge and sulphur.” However, the function of the lead in that process was specifically to remove the overwhelming quantities of silver as economically as possible and the process could not work at all effectively on metal containing more than a few per cent of gold. Some scholars, notably Healy, suggested that the two processes of parting and of cupella- tion, to recover silver from the spent cement, had been.conflated, but unless the account is totally mixed up, it is quite clear that weighed quantities of lead and a little tin were part of the contents sealed in the parting vessel Halleux" suggested a double process of cementation and cupellation proceeding simultaneously. This is ingenious but chemically impossible, as the whole point of the cementation is to turn metallic silver into a salt and cupellation recovers metallic silver from its salts. Burstein suggests that the lead was added to recover precious metal from the residuum of the ground ore after washing, but before cementation. Whilst this interpretation is techni- cally quite feasible it does require a gross conflation in the section dealing with the recovery of the impure gold. Some of the practical manuals published in Germany in the early sixteenth century® describe the addition of lead to the molten impure gold in order to facilitate its granulation when poured into cold water prior to parting, In Agath- archides’ description, however, the washed gold seems to have been poured straight into the cementation vessels, Several later sources do describe lead (zinc?) or copper salts as components of the parting cement.” Notton,“ who carried out gold-parting replication experiments based on Agatharchides’ description, found that the addition of lead metal actually impaired the removal of silver from the alloy of base gold with 55% silver and 7.5% copper that he used. He suggested that if lead really was added then its function could have been to remove any silicious gangue material from the gold. How- ever, overall, Notton seems to have doubted whether lead and tin were really added to the charge. The addition of barley bran is not mentioned in other early descriptions.” It has been suggested that the organic material was necessary as a source of hydrogen ions to create hydrogen chloride as the active reagent in the process." However, the process works perfectly well with common salt alone, as Notton demonstrated, and the addi- tion of any organic material, by making the atmosphere inside the vessel more reducing, would actually inhibit the reaction. It is possible that the bran or straw was the external fuel rather than an internal reactant, as both Pliny and Strabo in their descriptions (given below) of gold refining state that a fire of chaff should be used rather than charcoal. The use of straw as a fuel in the gold refining is also mentioned in the Babylonian Talmud, which states that a clay container with openings was used to let in the flame of the barley straw fire.” However, straw or chaff does not seem to be a very convenient fuel for a process lasting for five days and nights, and once again it would require a gross mistake in Agatharchides' text as it now stands. Historical Survey of Gold Refining (1) 35 ‘Thus it does appear that the ingredients as listed in the surviving accounts are impracticable. As this was a real industrial process, it would appear that there is something wrong with these descriptions. The most likely fault is conflation and it is possible that, as Burstein suggests, lead ‘was used to recover further gold from the crushed ore after washing, followed by cupellation and cementation, Alter- natively, the refining process itself was twofold: first the gold was cupelled with the lead and tin and barley bran, followed by parting with common salt. Traditional cupella- tion or fire assay often uses lead oxide with flour as a reducing agent." This explanation does at least account for all the ingredients, but it should not have been necessary to have cupelled the raw gold prior to parting, and the order in which the ingredients are listed in the existing accounts gives no support for a putative twofold process with the metal and bran first, followed by the common, salt. It is important to note that there is no reference to brick dust, clay or any other inert carrier, and at the termi- nation of the process the jars contained pure gold and very little else, Noting this absence, Notton carried out one experiment with no brick dust and found that the process worked perfectly well but the silver was almost entirely lost from the sealed parting vessel, and onto the furnace walls, reminiscent of the walls of the Sardis furnaces. The specification of a closely fitting and luted lid is important, as there is no direct evidence that the Sardis vessels were lidded. All subsequent detailed descriptions specify that the parting vessel should be sealed, and on the literary evidence alone it is likely that the Sardis vessels were also lidded. Pliny refers to the purification of gold several times in his Natural History." The first reference is at 33.60: ‘The first proof of quality in gold is however its being affected by fire with extreme difficulty; besides that, it is remarkable that though invincible to live coal(s) made of the hardest wood it is very quickly made red hot by a fire of chaff, and that for the purposes of purifying it itis roasted with lead. ‘This is in a passage on assaying, and the process is one of cupellation that by itself would not remove any silver in the gold. The second description is at 33.69 in sections on gold mining, where it states that: ‘The extracted material is pounded, washed, burnt and ground. The powder from the mortar is called scudes; the silver, which comes out of the furnace, is called sweat (sudor). The dirt, which is thrown.36 Historical Survey of Gold Refining (1) out of the furnace, is called scoria in the case of all metals. The one in the case of gold is pounded and melted again. The bowls are made of tasconium, this is a white earth similar to clay and no other (earth) can stand the bellows and the fire and the red-hot mate No details of reagents are given but the treatment of gold, possibly from a primary context, which produced a silver scum might suggest a parting process. However, no active salts are mentioned, and it is more likely that Pliny had in mind the smelting and recovery of gold from the ore rather than a parting process. The reference to the reworking of the slags (probably with lead although this is not stated) to recover more of the gold is extremely interesting and recalls the Roman gold-smelting process excavated at Trés Minas in northern Portugal.” There are more detailed descriptions of gold parting in sections dealing with the medicinal uses of gold (33.84), where it states that: It (gold) is also roasted with a double weight of salt and three times weight of misy and again with two portions of salt and one of the stone which is called schiston. So it draws poison (virus) from the things which are burnt with it in an earthen vessel (vas), itself being pure and uncorrupted. The ash left is preserved in an earthen pot (olla), mixed with water and smeared on the face (cures) eruptions, but it is better to wash it away with lomentum. Pliny was describing here a double process using mixed reagents, in this instance first common salt and iron sulphate, followed by common salt with tale (an inert hydrated mixed silica/magnesia, SiO,/MgO, mineral). Experiments by d’Elhuyar in the late eighteenth century,” using a variety of reagents either singly or in combination, showed that cementation could be made to work at a much lower temperature using a mixture of common salt and sulphates, either in the form of gypsum (calcium sulphate) or barytes (barium sulphate), than when the same reagents were used singly. ‘The translation given in the Loeb edition has a number of inaccuracies. Misy is translated there without comment as copper pyrites, a mixed iron/copper sulphide,” whereas in fact it is correctly translated as a mixture of corrosive sulphates, with ferric sulphate predominating. Schiston is just translated as ‘splittable’; in fact from other contexts it is likely to be tale, although Healy” suggests alum. The translation also suggests that the vessel in which the reac tion took place was specifically an earthenware cooking ot, that is an olla, whereas the word used is vas, which is the word for an ordinary pot; only the residue was to be kept in an olla, Halleux" believed that Pliny was working from an original Greek text and that virus should be read as the equivalent of the Greek ios: that is, rust rather than poison. Lomentum is translated by Rackham as ‘lotion’ but in fact has the specific meaning of bean meal.”” In the third description, which occurs at 34.121, the preparation and uses of misy are described in detail.” The short reference states that: A mixture of misy is employed in the magical purifica- tion of gold. At 35.183, in sections dealing with the preparation and properties of alum, a mixed potassium aluminium sul- phate, Pliny notes that black alum (most likely to be a mixed iron potassium sulphate) was used in the dyeing of dark cloth and that it could be used to purify gold. Note that in the last two references parting is referred to only in the specific context of the uses of the misy or alum and thus the presence of other salts (such as common salt) in the process is not excluded. Strabo, in his Geography,” describes gold refining by the Galatae, who inhabited the Cevennes and Pyrenees, as follows: ‘The product of smelting the gold and refining it with a kind of ‘styptic’ earth is ‘electrum’. They further smelt this mixture of gold and silver: the silver is burned away and the gold remains. This type of alloy (that is the electrum) is easily fused and hard like stone. It is for this reason that the gold is preferably melted by a fire fuelled by chaff since the flame is gentle and suitable for an alloy which yields and fuses easily; a charcoal fire, on the other hand, consumes a lot since it over-melts the gold and volatilizes it because of its intensity. Strabo's Greek term otvatnp@ons jn, styptic earth, is likely to have been a corrosive sulphate, such as alum or ferric sulphate, the Greek equivalent of the Latin misy.” Note that the styptic earth is apparently used only in the first part of the process to produce refined electrum, although in fact it only makes sense as part of the parting cement. The explanation why chaff was to be preferred is also somewhat misunderstood; a low temperature was nec- essary in order not to melt the finely divided electrum, thereby reducing the surface area exposed to the cement. As already noted (p. 35), chaff seems an unlikely fuel for a ptocess of long duration. Recipes for the refining of gold featured in some of theAlexandrian alchemic literature of late antiquity have been collected and discussed by Halleux.” The following unusually clear and concise recipe is taken from the Pseudo-Democritus, which has its origins in the first centuries ap, but only survives in a later Syriac version, from which a translation was made for Berthelot:”” Du sel et de I'alun, chacun une once, du vitriol noir de Chypre, 2 grammes; broie ensemble et mets dans un creuset. Plonges-y des lames d'or. Dispose par couches de la chalcite [] entre les lames. Mets sur des charbons ardents et souffle, Quand le produit sera fondu, tu verras quel produit tu obtiendras. Once again, this is a mixed chloride/sulphate process, this time using both alum and ferruginous sulphates. Note the reference to the gold being in the form of sheets (leaves?). This is apparently the first such specification in western gold-refining literature ‘The Mappae Clavicula® has a typically rather confused account of salt cementation in recipes 2 and 3, described as methods of making or ‘extending’ gold Again making gold Melt one ounce of silver, half an ounce of copper and one ounce of gold. Again, take sand and press it onto a level place. Cool until itis dry, and mix again some salt and roast it in the furnace for a day and a night. After- wards take it out and wash it until the salt runs off; and again dry it and knead it in vinegar and set it aside for a little until it absorbs it and dries out. Then again put into the furnace a piece that has not been washed and do this once and again; knead it in vinegar every time you put it into the furnace. Now you ought to put it into the furnace four or five times until it becomes almost as if it is all cooked away; and when you take it out use a silver withdrawing tool, which in Greek is called elquison. Take a weight equal to the former amount, mix it all together and grind it. Then melt separately the two kinds of material that you have concocted [that is the residual gold and the salt-containing cement which now contains silver] and sprinkle them gradually [onto a lead bath?] until it is used up. Then cool it and you will find that hard lead has been made. Melt this together with cepsonium, that is, kneaded ashes. As is shown according to the Key, psomion is ash kneaded with water, which you lay underneath in the furnace to the thickness of a finger. The next chapter, 3, continues: Again Take only a little for experiment when you do it once until you learn it thoroughly. Take one ounce of Historical Survey of Gold Refining (1) 37 reddish Cyprian copper in solution [? posios], one ‘ounce of quite good silver, and melt it with chaff, until ‘when hammered out it does not make a noise, and then melt it together with one ounce of gold and the same amount of natron. Then turn face to face two little bowls, that is two hollow earthenware pots and put inside them the widened [that is, hammered out] melt that has been prepared, and mix in antisma. What had been a little bead of copper is now turned into a little bead of four ounces of silver. In the bead we find more than an equal amount of gold. [Take] one part of Pontic sinopia (a ferruginous clay containing hematite), two parts common salt, grind them all together, lay the sheets on the bottom and sprinkle [the sinopia and salt mixture] over them and coat them with pot clay so that, they cannot breathe, Put fire under them until you feel it is all right. Take them out and you will have the very best gold. At first reading, these two descriptions would seem to be conflations of the ‘extension’ of gold by the addition of silver and copper in the ratio of two parts to one respec- tively, and of the refining of gold by parting followed by cupellation. However, these descriptions could be the earliest reference to graduation, followed by refining. That is, the practice of adding more of the metal to be removed in order to allow an open porous structure to develop throughout the metal grains during the refining process, ensuring that all the silver or base metals in the gold were exposed to the parting agent.* Smith and Hawthorne believed recipe 3 to be just a surface treatment to give the impression that the metal was pure gold throughout. However, salt cementation was usually associated with complete parting, especially when the metal was in the form of foils. It would anyway have been useless to produce surface treated foils as the surface enhancement would have been lost when they were melted up for use, Treatments that merely enhance the surface are only of use on the finished artefact. ‘Thus there are three possibilities: © A conflation of gold extending and refining. Note: Chapters 2 and 3 are entitled respectively ‘Again making gold’ and ‘Again’, following “Making the most gold’, which is the title of Chapter 1. © Surface treatment, but the products are respectively molten or foils. © Graduation, Of these, the third seems most reasonable and requires least alteration of the text as it has come down to us,