The Role of Ore Geology and Ores
in the Archaeological Provenancing of Metals.
School of Earth Sciences, University of Birmingham, Edgbaston
The degree of processing needed for the manufacture of metals means that matching them to their raw material sources is difficult. Attempts to do this based on comparisons between the major and minor element geochemistry of metal and potential ores and more recently on their stable isotope geochemistry have met with limited success. A major reason for this is that ore sampling strategies have either been neglected or based upon an incomplete understanding of the mineralogical and geochemical complexities inherent in any metalliferous ore deposit.
All ore deposits are spatially zoned and each zone has its own set of mineral assemblages and its own distinct geochemical characteristics. In addition, deposits are zoned in time and many are the result of multiple and often genetically unrelated mineralisation events. Mineralogical changes at mine sites continue after mining, forming a new zone named the supragossan. Here, the removal and gain of elements leads to new mineral associations whilst the original composition of the mined ores is radically altered. In particular an understanding of the origin and behaviour of lead in the supragossan zone is an essential first step in stable lead isotope provenancing studies at mine sites.
The provenancing of arsenical copper and gold nicely illustrates these issues. Only by comprehensive sampling of a wide range of potential copper ores, plus stable isotope data, is it possible to suggest that Ross Island, Ireland, was a source for copper arsenic bronzes (type A metal) in the Chalcolithic of the British Isles. Primary gold grains from a small gold deposit at Calliachar, Scotland show a wide variation in chemistry between the main zones, demonstrating the need for extensive sampling of multiple grains. Gold in placer deposits potentially comprises three components, proximal and distal detrital grains and in situ gold and, as their relative amounts will vary throughout the deposit, bulk analyses rather than detailed analyses of individual gold grains would appear to be a better method of provenancing prehistoric gold.
Keywords. Ore deposit geology, Provenancing studies, Gold, Lead, Ross Island.
Provenance studies, in the sense of matching stone to its geological origin or, in the case of a processed artefact matching its component raw materials to their respective geological origins, potentially remains a very powerful tool in the armoury of archaeologists. The techniques for 'scientific provenancing' (as opposed to 'stylistic provenancing') for lithics and ceramics based on petrography or geochemistry are routine, well established and often successful. Indeed, recently it has been shown that a combination of 'total petrography' (the use of transmitted and reflected light techniques so that all the mineralogical and textural data are collected) alongside standard major and minor element geochemistry, can provide very precise provenancing of lithics (Ixer 1994). There is some suggestion that artefacts can be matched to individual outcrops or even within an outcrop down to a scale of tens of metres (Ixer 1994; Ixer 1996; Ixer 1997).
The provenancing of ceramics is harder than lithics because they comprise a mixture of clay and temper. Both components may be differently sourced and will have undergone different degrees of processing. However, the final ceramic remains a mechanical mixture of lithic fragments (still largely unaffected by any processing and so petrographically recognisable) within thermally altered clays.
With metal artefacts, provenancing becomes yet more difficult as the raw materials are homogenised into a chemical whole - the metal or metal alloy. For metalwork, therefore, provenancing studies have been almost exclusively geochemical; initially, comparing the major or trace element chemistry of 'ores' and metal and, most recently, by studies based on their stable lead isotope signatures. The results of these endeavours have been subject to much criticism, notably by Budd et al. (1994; 1996) who question the validity of such studies. It is because of these doubts and because the degree of processing that is required to produce a metal obscures much about the raw materials used in their manufacture, that a complete understanding of the origins of the raw materials, especially the ores, is needed. This falls into the branch of geology called mineral deposit studies that deals with the recognition, classification and the genesis of mineral deposits.
Introduction to Mineral Deposit Geology.
Exploitable metalliferous mineral deposits are rare, geochemical aberrations often with complex mineralogies; there are classes of mineral deposit that are represented by a single example and many have less than fifty, for example Carlin-type gold deposits or the Zambian Copperbelt-type copper deposits. Even commonplace mineral deposits, those with hundreds of examples worldwide, are geochemical anomalies. This is because most geological/geochemical processes work towards the gradual homogenisation of the Earth's crustal composition but, ore deposits, almost by definition, comprise aberrant concentrations of metals, having 102 to 104 times the average concentration of those metals in the crust. These abnormal concentrations rarely form as the result of a single process but normally require multiple events. This is because a sequence of different concentration mechanisms are needed to up-grade non-ore rocks through a series of proto-ores finally into an exploitable deposit. These mechanisms may be related to each other but equally the final deposit may be the result of the superimposition of more than one genetically unrelated mineralisation event. This means that, for example, a deposit may comprise copper mineralisation that was differently sourced and of a different age from any lead-zinc mineralisation that lies next to it. In addition, for most mineralisation events the ore fluids are an evolving system with mineral precipitation taking place in response to changes in physico-chemical conditions, especially drop in temperature.
The majority of ore deposits are geochemically and therefore mineralogically variable; this variation occurs in space and time and is known as spatial zoning and paragenesis, respectively. Spatial zoning can be investigated and described in the field by plotting the distribution of ore types onto a map, section or plan but, whilst some aspects of temporal zoning can be determined in the field by collecting structural data, it is mainly studied in the laboratory by interpreting the textural relationships between minerals under the title of paragenetic studies.
Zoning and paragenesis are two aspects of the same phenomenon and hence can be cogenetic (Park & MacDiarmid 1970). It is vital for archaeologists who wish to match ore deposits and their ores to metal, using trace element geochemistry or stable isotope methods, to recognise the complexities of both these aspects of mineralisation. This will ensure that only those mineral associations/ores that have archaeological significance are sampled and, equally, that none are overlooked.
CLICK HERE FOR FIGURE 1.
Spatial Zoning within an Orebody
FIGURE 1a shows a representation of the spatial zoning in a base metal sulphide vein whilst FIGURE 1b shows zoning in a massive stockwerk deposit where the primary ore/proto-ore is finely disseminated within its host rock and the exploitable ore is present within a zone of supergene enrichment. As the figures show, the mineral associations/ores fall into a fourfold framework. It should be kept in mind that, within each of these four divisions, there might be further sub-divisions.
These associations are: -
In should be noted that each of the four zones can carry very different ore types which in turn may need different mining and beneficiation techniques to exploit them. At its very simplest, friable carbonate-oxide ores need to be treated differently from massive sulphide ores. Hence with time and the continued exploitation of an orebody the ores will change from those found in gossans down through the supergene enrichment zone into the primary ores. Potentially each change in ore-type will be accompanied by a change in the composition of any metal produced from that ore and there may be cases where early and late metal from the same mine will have very different geochemical signatures and are not recognised as having the same source.
1. Host rock/wallrock assemblages - these are pre-mineralisation assemblages often comprising simple sulphides or oxides. They are not ore but some wallrocks may be used as fluxes.
2. Primary (hypogene) mineralisation assemblages. Commonly there is more than one primary mineralisation event and the ores resulting from any one event may be zoned. Normally assemblages formed at high temperatures are separated in time and space from those that formed at low temperatures. The assemblages comprise many minerals including sulphides, arsenides, sulphosalts and oxides intergrown with 'unwanted' minerals known as gangue and often in complex textural relationships.
3. Secondary (supergene) mineralisation assemblages. These are formed by the alteration of primary ores below the water table. They are especially important in copper and silver deposits, where they form very rich and hence very exploitable ores. The assemblages are simpler than the primary ores; they are dominated by 'simple' sulphides and replacement textures are characteristic.
4. Oxidation/weathering assemblages. Ores within the oxidation zones are often known as gossans (iron-hat) and are present on the surface. Brightly coloured carbonates, sulphates, arsenates, native metals etc are present within brown to black iron oxides and abundant hydrated iron oxides - the latter are collectively known as 'limonite'.
Temporal Zoning within an Orebody
Although FIGURE 1 shows the spatial relationships of the various mineral assemblages and ores within an orebody it cannot show their temporal variation, known as their paragenesis. A paragenetic diagram is the usual way of showing the evolution of a mineral assemblage with time and one is shown in FIGURE 2. Here the paragenesis for the copper and lead mineralisation at the Bronze Age Great Orme Mine, partially based on Ixer & Davis (1996), is shown. The diagram employs the usual conventions so that relative time moves from left to right; the minerals are listed in order of their first appearance and the width and thickness of the bars indicate the length of time of deposition and relative amount of mineral deposited.
CLICK HERE FOR FIGURE 2.
From the diagram it can be seen that at the Great Orme the limestone-dolostone wallrocks only carry a little pyrite and TiO2 minerals, that there are two separate mineralisation events, an earlier lead-rich and a later copper-rich one, which are followed by extensive supergene and oxidation zone alteration dominated by carbonates and limonite. Although the diagram separates the two primary mineralisation events, it cannot by itself indicate the length of time between them in absolute terms, nor can it suggest if the events were cogenetic or genetically unrelated. In the Great Orme example it is other evidence that shows the lead mineralisation is Mississippi Valley-style in origin and probably Permo-Triassic in age, whereas the copper mineralisation is an example of the copper-dolomite association and Jurassic-Cretaceous in age (Ixer & Davies 1996; Ixer & Stanley 1996; Ixer & Budd 1998). Hence, at the Great Orme Mine, an unrelated episode of copper mineralisation is superimposed upon earlier lead mineralisation so that the mine carries two independent sets of ore. The archaeological implications of this, in particular on sampling procedures and their role in the use of stable lead isotopes to provenance copper metal, are discussed at length by Ixer (2000).
Once the ore geology and its complexities have been understood it becomes possible to collect, for archaeological purposes, in situ representative samples of all the potential ores alongside the other mineral associations. This is best done together with an economic or mine geologist and as much care should be taken in selecting the mineralised material as in excavating any archaeological artefacts (Ixer 1995). Whilst this is feasible for mines that are currently being exploited or abandoned mines with good access to the mined ore, it is often not possible for historical and prehistoric mine sites where the ore has been removed.
Material collected from abandoned mines and mine sites should be treated with extra caution. It is useful to consider the mine and mine dumps as belonging to a new and discrete geochemical/mineralogical zone. This zone, christened the supragossan, is the result of post- mining geochemical and hydrogeological processes that take place both underground and on the surface, namely in the spoil, mine tailings and slags. This zone shares many aspects with, and analogies to, the post-depositional changes found in artefacts recovered from excavations.
The Supragossan Zone.
One of the main consequences of mineral exploitation (mining) is to change the local hydrological conditions by altering the chemistry and flow rates of the groundwaters. Underground, mining features like drives, shafts and stopes can provide pre-eminent pathways for groundwater flow especially if the country rocks/host rocks are impermeable (Younger 1995). The chemistry of this groundwater is radically altered as it passes through these open systems especially if mining has exposed pyrite (and other iron sulphides) to oxidation, which then acidify that groundwater. This acidification (lowering of the pH to between 2 - 4) and the passing into solution of base metal ions (iron, copper, lead and zinc) plus other elements like arsenic is the reason that many mine waters are both grossly polluted and highly reactive. These groundwaters not only remove metals but can also reprecipitate them, sometimes in new combinations as new mineral species, so that the surface geochemistry/mineralogy of the accessible underground 'ore' is not the same as that originally mined. The extent of this alteration zone depends on the porosity and permeability of the rocks but in many cases it will not be very thick.
On the surface, spoil heaps, rock waste (and slagheaps) react with rainwater and groundwaters in a similar way. Allard et al. (1989) have shown that the release of metals from mine waste and slags is promoted by lowering the pH of waters penetrating them but more especially by an increase in the redox potential when the material is exposed to air in the presence of water. Crushing increases by many orders of magnitude the surface area of mineralised rocks and if spoil heaps are left unreclaimed then oxidation, drainage and surface outflow occur in rapid succession (Younger 1995). Extreme geochemical conditions are created that lead to the formation of unusual suites of alteration products in the spoil. Similar processes occur underground as described by Jenkins & Johnson (1993). Many spoil heaps, especially those from base metal mines, have suffered rapid, intense and extreme degrees of alteration by geochemically aggressive fluids under pH and Eh conditions that are very different from those associated with the primary (and most of the supergene) mineralisation. Although many of the mineral assemblages found in spoil heaps will be similar to those found in the gossan zone of the exploited ore (if it had one) there will be very important differences that can distort archaeological interpretations. For example, the recent proposal that copper arsenates, which are commonly present in spoil material (and some gossans) but rare in primary ores, were the original ores for the production of arsenical bronzes (Pollard et al. 1991; Budd et al. 1992), has been challenged on these grounds by Ixer & Budd (1998) and by Ixer & Pattrick (in press).
It should be remembered that the spoil has been anthropogenically selected. Richer and run-of-the-mill grade ores will be missing, and the original metal ratios of the deposit will be distorted. Cwmystwyth in central Wales is a good example, for here, at this Bronze Age copper mine, the excavated spoil is galena-rich and copper-poor (Timberlake 1995), quite different from the ore that was exploited.
Bachmann (1982) has similar concerns with regard to slags stating that 'A slag may undergo more or less extensive weathering depending on composition, climate and duration of deposition - leaching changes the overall composition of slag significantly. Chemical analyses therefore must be interpreted with care'.
These problems can be overcome. On many mine sites paragenetic studies of the available material will distinguish between the mineralogy of the supragossan zone and the original ore assemblages, or at least will allow those earlier assemblages to be mineralogically reconstructed. Restoring the original geochemistry of the ores and especially the relative importance of their minor elements is far more difficult.
Lead in the supragossan zone.
The popularity of stable lead isotope ratios as a provenance tool means that an understanding of the geochemistry of lead within the post-mining, supragossan zone of a mine site is essential. Normally lead moves slowly in the environment and most groundwaters carry very little lead in solution but at mine sites it, alongside other heavy metals, is easily leached, although often it does not move very far before being reprecipitated within secondary lead-bearing minerals.
Gee et al. (1997) investigating this phenomenon compared Roman, Medieval and Post-Industrial Revolution lead smelting slags. They were able to show that the major controls on secondary lead precipitation are pH and the availability of Fe, Mn, P and CO2, with lead being adsorbed onto iron or manganese oxides ( 'limonite' and 'wad' respectively), or incorporated into members of the phosphate-bearing, pyromorphite group of minerals or the carbonate minerals cerussite and hydrocerussite.
Poorly crystalline iron and manganese oxides/hydroxides are present at nearly all mine sites (from the oxidation of iron-bearing sulphides etc) whilst sources of phosphorus (bone, wood ash and other organic material) are commonplace at prehistoric mines, as are sources of carbonate. Therefore, within the supragossan zone there is an increased potential for the precipitation of secondary lead minerals from circulating groundwaters onto the spoil, slag or metal. It is vital that the origin of this lead is known. If lead remobilization is localised then the isotopic ratios of the lead in the secondary minerals remain the same as the initial ones from the ore (Faure 1986) and the movement has few archaeological consequences. By contrast, if the original ores are lead-poor, as are many copper or tin ores, and if the local groundwaters are lead-bearing, having come into contact with a nearby lead source, then it is the isotopic signature of the groundwaters that will be measured, not that of the ores (assuming some precipitation of the lead onto the ore). Hence it is important to know the isotopic signature of the local groundwater at the site and doubly so if one or more of the ore, flux, slag or metal are foreign to the excavation site, as their true signature may be swamped by that of the local, post-mining groundwater. Lead-shot, which is quite often found in modern gold placers, certainly in the highlands of Scotland, is an extreme example of this problem. Similarly if the deposit carries lead-rich and lead-poor ore assemblages then the isotopic signature of the lead-rich ores will dominate the groundwater and subsequently the supragossan material, leading to the potentially erroneous conclusion that the lead-rich ores alone were exploited.
The need for understanding the ore geology, for archaeologically dedicated collecting of ore samples and for having a detailed knowledge of the mineralogy can be clearly demonstrated by the studies at Ross Island, Killarney, Ireland. Preliminary descriptions of the geology and ores, and discussion of the importance of this Chalcolithic copper mines site for our understanding of the origins of metallurgy in the British Isles, are given in Ixer & Pattrick (1995; in press), Ixer & Budd (1998) and O'Brien (1995; this volume). It is the earliest known copper mine in the British Isles and has a very wide range of potential ores (in terms of their metal contents and grain-size). Its main importance, however, lies in being the only known Bronze Age copper mine in the British Isles with arsenic-rich ores (tennantite and/or arsenopyrite) capable of producing arsenical copper, namely Chalcolithic type A metal, as defined by Northover 1982, (this is equivalent to Coghlan and Case (1957) type 1 metal).
CLICK HERE FOR FIGURE 3.
Although some lead isotope data have been reported by Ixer & Pattrick (1995) FIGURE 3 shows all the data for the mineralised specimens from Ross Island together with plots of Chalcolithic type A artefacts (Rohl 1995).
Plotted onto FIGURE 3 are three galena-poor, disseminated, mixed chalcopyrite-tennantite ores hosted within limestones (Western Mine ore); three massive-pyrite-chalcopyrite plus minor amounts of galena and sphalerite ores taken from two sulphide caches excavated from the Bronze Age work camp (Ore from Excavation) and three massive galena-sphalerite plus some pyrite and chalcopyrite ores taken from the adjacent site of Blue Hole, probably also exploited in the Bronze Age (Blue Hole ore). The 'Muckross Ores' are from similar massive sulphide mineralisation to Blue Hole but a couple of kilometres away; they are not believed to be potential Bronze Age ores.
The figure shows an overlap between the massive sulphide ores from Ross Island and type A metal. Hence, at first sight the ore mineralogy (the copper ores are tennantite and/or arsenopyrite-rich), the dating of the mine and adjacent work camp (Beaker pottery and the C14 dates) (O'Brien 1995) and lead isotope data all concur with the suggestion that Ross Island supplied metal for type A artefacts. However, the data suggest that it was the massive ores rather than the chalcopyrite-tennantite, carbonate-hosted ores that were smelted, since the latter plot away from the artefacts, being more uranogenic than either the massive sulphide ores or the artefacts.
It should be noted in passing that the isotopic differences between the massive sulphide ores (with 1 wt% galena or more) and the limestone-hosted copper ores (with lead in the ppm range) provide a fine example of the isotopic divergence between lead-rich and lead-poor ores, that occurs in geologically old deposits, even when the ores began with the same initial lead ratios. This phenomenon, which is described in more detail by Macfarlane (this volume), is of extreme importance in provenance studies. The high Pb206/Pb204 ratios, showing that the limestone-hosted copper ores are uranogenic, are also to be expected, as much of the mineralisation is associated with carbonaceous matter and detrital heavy mineral grains along stylolitic junctions.
The archaeological interpretation of FIGURE 3 is more difficult than the mineralogical explanation of the differences shown by the ores. If it is assumed that the differences between the lead isotopes of the ores is real, then the simplest interpretation is that only the massive ores were exploited. However, this runs counter to the archaeological evidence, for it is the copper-tennantite ores that are present within the Bronze Age mine workings and make up the majority of the 'ore' found with the excavated work camp (O'Brien 1995). It seems more likely that the in situ chalcopyrite-tennantite ores are not representative of the Bronze Age copper ores that were taken from Western Mine and that these (richer) ores had sufficient galena to give the same ratios as those seen in the type A metal. It is also worth noting that, if only the disseminated, chalcopyrite-tennantite ores had been sampled in line with the excavation evidence, there would have been no isotopic match between Ross Island ores and Chalcolithic arsenical copper metal. In order to try to constrain the problem better, a far larger suite of samples, including richer chalcopyrite-tennantite ores, has been analysed in co-operation with the British Geological Survey.
Although it is best to use slag or worked ore for provenancing rather than ores, there is no slag from secure Bronze Age contexts at Ross Island. Very rare samples of burned ore are present within the Bronze Age stratigraphy (for example specimen 1242) (Ixer & Pattrick in press). However, their void spaces are infilled with post-Bronze Age lead carbonate (cerussite) so that any lead signature from this material would be seriously compromised and so lead determinations on this material will not be attempted.
Ross Island shows that mineral deposits are zoned in space and time (the zoning continuing after mining) and comprise many different mineral assemblages some of which are/were ore. In terms of provenancing (the matching of the correct ores to metals) there is probably no such thing as a simple ore deposit nor is there a simple ore collecting strategy.
These truths will be illustrated again by using gold and the 'simplest' problem, namely the provenancing of unalloyed gold artefacts.
Introduction to the provenancing of gold
As stated in the introduction the provenancing of metals and alloys is inherently difficult. This is because the degree of processing that goes into the manufacture of an artefact is inversely related to the chances of a successful match between that object and its raw materials. This being so, the simplest metal to provenance should be unalloyed, single-sourced (obtained from a single mine site) gold.
Large numbers of quantitative and semi-quantitative major and minor element analyses for European gold artefacts have been published by Hartmann (1970; 1982). Taylor and her co-workers have had some success in using these analyses to match and classify these artefacts (Taylor 1980; Warner 1993) but there has been little progress in identifying their geological sources. Indeed recently Taylor et al. (1997) have acknowledged that 'major constituents and major trace analyses were inadequate for characterising specific ore sources' and have proposed that qualitative, multi-elemental, association patterns will prove to be better at pin-pointing ore sources. The use of increasingly more sophisticated analytical techniques will certainly provide better data on individual gold grains, but this in itself can add little to further provenance studies as long as an incomplete understanding of the complexities of gold mineralisation remains the main reason for poor or inadequate sampling.
Hard rock sources
Gold is a rare, but widely dispersed, metal occurring in many different mineral associations throughout a variety of geological environments (Boyle 1979; 1984). Although the majority of present day, economic gold deposits are primary, hard rock deposits associated with acid, intermediate or basic igneous rocks or found within a wide range of metamorphic terrains, much of the gold mined in prehistory was won from sedimentary placers in recent river gravels. These secondary deposits are formed by the mechanical concentration of gold and other minerals that survive physical and chemical weathering of a primary deposit and are then transported by water or ice to a new location and deposited.
Each major class of primary gold deposit has its own mineralogical and geochemical characteristics and exploration geologists spend much time in learning to recognise and identify these in order to find new sources of the yellow metal. The mineralogical and geochemical differences between classes of primary gold deposit not only affect their ore associations but also are manifested in a number of different ways within individual gold grains. Firstly, native gold grains can enclose other mineral phases, as small, 1-10µm diameter, inclusions; amongst these are a wide range of commonly occurring sulphides, sulphosalts, oxides and silicates, plus more unusual selenides and tellurides. Usually these minerals are part of the main mineral assemblage but some phases are only found within gold grains. However, in either case the minerals are of genetic significance and Leake et al. (1997), who list over fifty opaque phases from gold grains in Scotland, suggest that inclusion identification may be the most diagnostic tool in distinguishing between gold deposits. Secondly, the gold may itself vary, most notably in its fineness, chemical composition and homogeneity. Fineness is the numerical measure of the parts per thousand of gold in any gold-bearing alloy, so that 800 fine gold is 800 parts gold to 200 parts other metals: any naturally occurring gold-silver alloy that is 800 fine or less is called electrum. Silver and copper are the most common metal alloys but in a few gold occurrences palladium or mercury are important. Finally, in many deposits single gold grains are chemically inhomogenous with silver-rich areas separated from more gold-rich areas either in a random fashion or zonally but with a recognisable pattern (Leake et al. 1993).
These mineralogical phenomena are universal; they are part of the primary mineralisation and are largely preserved during the weathering and erosion of the primary deposit and transport of the gold grains to their placer (Leake et al. 1993; Leake & Chapman 1996). Hence, for a number of decades it has been a standard and successful exploration technique to use placer gold grain characterization to predict the parent bedrock gold mineralisation- a form of genetic provenancing.
Potentially therefore, single gold grains show sufficient mineralogical and geochemical variation that they uniquely reflect the type of gold mineralisation or even an individual gold deposit. However, gold in primary deposits is a complex mineral and as Leake & Chapman (1996) state 'there may be significant variation in the composition of gold derived from the different parts of the same mineralisation'. Few studies have attempted to examine and quantify the extent of these primary gold variations within a small deposit and ever fewer have extended the research to include variations between primary gold mineralisation and any associated placer. A gold prospect at Calliachar-Urlar Burns, Scotland was chosen to attempt such a study as the prospect is pristine and small enough that comprehensive and representative sampling from drill-core, surface outcrops and stream sediments was practicable. In addition, the richness of the gossan (100-400ppm gold has been reported) and the presence of gold grains up to 0.9 grams in weight in nearby streams (Chapman 1994) suggest that the prospect could have been mined successfully in prehistory had it been discovered.
Calliachar Burn and Urlar Burn lie approximately 1km from each other near Aberfeldy in the Highlands of Scotland. Here, fourteen, thin, quartz-galena-pyrite ± base metal sulphide ± gold veins and their overlying gossans lie upstream of a small gold placer. The mineralogy of all the mineral assemblages, including analyses of their associated gold grains, is reviewed and described in Ixer et al. (1997). Gold at Calliachar shows all the expected mineralogical features so that although many gold grains are inclusion-free, others enclose pyrite, galena and sphalerite and a few have 'unusual antimony sulphosalts'; the gold is alloyed with silver and mercury; and the primary and supergene gold grains are compositionally inhomogeneous with internal fineness variations.
CLICK HERE FOR FIGURE 4.
Native gold (strictly speaking electrum together with a few gold alloys) from the different mineral associations from Calliachar and Urlar Burns are plotted in terms of their major metals, gold, silver and mercury in FIGURE 4. Each mineral association is listed in its paragenetic order from the oldest at the top to the youngest at the bottom. The figure demonstrates a very large variation in gold compositions between these associations so confirming the views of Leake & Chapman (1996). Some of this variation is routine, namely the overlap between the core composition of some of the placer grains and in situ hypogene and supergene (gossan) gold. The increase in fineness with time from hypogene electrum grains to supergene electrum grains to high fineness gold rims surrounding placer grains is also commonplace. Less expected are the marked differences between the gold compositions found at Calliachar and those of Urlar Burn approximately one kilometre away and in particular the differences in mercury content. These variations clearly demonstrate: that even in a small (easily mineable), primary gold deposit very different ores are present in different parts of the orebody and so potentially would yield metals with quite different minor and trace element geochemistries; as a consequence all parts of the orebody need to be sampled and many analyses are required, not just one or two, to show these complexities in their entirety. Indeed, the results from Calliachar strongly indicate that the twenty samples per ore deposit that are believed to be enough for lead isotope provenance studies (Reedy & Reedy 1998; Sayre et al. 1992) would be an insufficient number for a comprehensive geochemical characterisation of all the potential metal producing associations. If the amount of geochemical variation in gold between Calliachar and Urlar Burns is a common phenomenon (and it may be), then Budd & Haggerty (1995) are correct in challenging the concept of a single regional (or even quite local) chemical, gold signature as suggested by Taylor (in Thwaite 1995).
The requirement for extensive sampling is not restricted to primary gold deposits but is equally valid for placer deposits, as they too are not homogeneous in terms of the grade and composition of their ores. Some placer gold deposits are the result of the weathering of a single, primary bedrock gold source but even here the placer will carry gold grains with significant fineness variations, that will reflect the differences in composition between the hypogene and supergene zones and the degree of complexity of the original source. Most placers are more complex as they are multi-sourced with gold coming from nearby hardrock sources (proximal grains) as well as distant ones (distal grains) plus perhaps gold grains already modified by the sedimentary cycle and entering their second or third generation placer. In addition most placers have an in situ (authigenic) gold component in the form of gold-rich rims about the detrital grains and/or forming discrete nuggets; this gold is not detrital but is formed by the precipitation of gold from groundwaters.
Proximal, distal and authigenic gold grains not only vary in their chemistry but also in their size and shape and Bowles (1988) in an accessible review, summarises the chemical and morphological changes that take place in alluvial gold grains during their passage from bedrock to placer. In broad terms proximal grains, those close to the hardrock source, are angular, have bright, scratch-free surfaces, still have minerals attached to them or show sharp scars where they have been lost and have thin or missing gold-rich rims. During transport the surface of the grains becomes heavily scratched and matte and the grains are hammered, folded and kneaded until after travelling for approximately 100 kilometres they are finally deformed into one of two shapes, rods /cylinders or flakes/plates/sheets. At the same time the core composition of the grain increases in fineness and there is an increase in the thickness of the gold-rich rim. Most gold placers therefore comprise three different sorts of gold; angular, equidimensional, low to moderate fineness, proximal grains; cylindrical or flake-like, high fineness, distal grains; and very fine gold as rims enclosing detrital grains and/or present as discrete nuggets.
The absolute and relative amounts of the three components will vary throughout the placer, since all three classes of gold differ in their hydrodynamic properties and in grain size and because the precipitation of authigenic gold depends on very local geochemical conditions. Hence, authigenic nuggets may be the initial gold taken from the surface of a placer, whereas the bulk of the gold might be detrital and mined from concentrations to be found in irregularities at its base (Bowles 1988). Therefore as the ratios of proximal: distal: authigenic gold vary throughout the placer, with each component carrying its own independent and distinctive geochemical signature, so the composition of the gold metal produced from that part of the placer will change. Indeed, since the detrital and non-detrital components have very different geneses and are unlikely to be geochemically (including isotopically) alike, artefacts made from them may not be recognised as coming from the same place but assigned to different sources.
Finally, there is the question of what constitutes gold ore in a placer. This needs to be addressed before a placer can be correctly sampled. In gold placers ore cannot be represented by individual gold grains (unless they are large) but comprises collections of them. Perhaps as in forensic work, bulk gold, namely one to ten grammes collected from a single part of the placer, needs to be geochemically analysed and compared to an artefact. Where this is not possible the number of gold grains needed to adequately represent all of the gold components must be established before extensive analyses are warranted. For some single-sourced, proximal deposits this may only require a few grains but there will be others where the complexities of gold their components may be such that the deposit cannot be included in any meaningful provenancing programme. There may even be a case where, if the artefact were made exclusively from authigenic (nugget) gold, none of which remains, trace element or stable isotope analyses of groundwater might prove a better provenancing tool than detrital gold grains.
For the minerals exploration industry, detailed mineralogical and geochemical data from single, placer gold grains are of great value as a genetic provenancing tool (indicating that the gold is associated with a porphyry copper deposit or has ophiolitic affinities) as exemplified by Leake and co-workers (1993) rather than being geographically specific (namely that the mother-lode crops out fifteen kilometres upstream), whilst the bulk chemical analyses of the same gold grains have little diagnostic worth. This contrasts with archaeometallurgy where the situation is reversed. Here artefacts are more likely to be successfully provenanced (given a specific geographical origin) if their chemistries are compared with those of bulked gold samples rather than with individual gold grains.
So it seems that although the provenancing of prehistoric gold does not have some of the problems associated with other metals, especially alloys like arsenic- or tin-bronzes, it remains difficult. It, like all other metal provenancing, depends upon recognising and sampling the correct ores. This in turn requires a full understanding of the ore geology of the deposit and this must never be thought of as simple, however tempting that may be.
Trace element and isotopic provenancing of metals can only be as good as the ore sampling that it is based upon. This in turn depends on a proper understanding of the ore deposit geology, natural geochemical processes within the supra-gossan zone and the distinction between mineral samples and exploited ore. Once all this has been achieved then geochemical studies have much to contribute, both in terms of provenancing ores/mines to artefact but also temporally within the mine itself. Once the mineral zoning has been established in three dimensions, then for mines with centuries or millennia of exploitation it may be possible to determine their mining history, rather than just provenancing metal from their ores. Combining detailed and precisely located ores from within the mine and associated spoil and slag heaps with trace element and isotopic fingerprinting and if all this can be matched to datable metal artefacts it may lead to a proper understanding of the exploitation history of a mine.
Drs R.A.D.Pattrick and C.J.Stanley are thanked for their mineralogical prowess and for the many high quality gold analyses from Calliachar-Urlar Burn. In a similar vein Dr J. Naden provided numerous analyses of gold placer grains from throughout the world and introduced me to the complexities of genetic gold provenancing. Germane criticism by Drs. J. Tellam and R. Chapman helped to modify the text, as did the incisive referees' comments. C.R.Ixer provided the computer skills for this paper.
ALLARD, B., I. ARSENIE, K. HAKANSSON and S. KARLSSON. 1989. In J.P. Vernet (ed.): Proceedings of an International Conference on Heavy Metals in the Environment: 221-224. Edinburgh: CEP Consultants.
BACHMANN H. 1982. The identification of slags from archaeological sites. Occasional Publication 6. London: Institute of Archaeology.
BOWLES, J.F.W. 1988. Mechanical and chemical modification of alluvial gold. The Australian Institution of Mining and Metallurgy Bulletin and Proceedings 293: 9-11.
BOYLE, R.W. 1979. The geochemistry of gold and its deposits. Bulletin of the Geological Survey of Canada 280.
1984. Gold deposits: their geology, geochemistry and origin, in R.P. Foster (ed.): Gold '82. Geological Society of Zimbabwe Special Publication 1: 183-189. Rotterdam: A.A. Balkema.
BUDD, P., D.GALE, A.M.POLLARD, R.G. THOMAS and P.A.WILLIAMS. 1992. The early development of metallurgy in the British Isles. Antiquity 66: 677-686.
BUDD, P. and R. HAGGERTY. 1995. Old Gold. New Scientist 147: 50.
BUDD, P., R. HAGGERTY, A.M. POLLARD, B. SCAIFE and R.G. THOMAS. 1996. Rethinking the quest for provenance. Antiquity 70: 168-174.
BUDD, P., B.SCAIFE, T.TAYLOR and R.G.THOMAS. 1994. Untangling the web: some new views on the origins of prehistoric metallurgy. Historical Metallurgy 28: 98-102.
CHAPMAN, R. 1994. The geological origins and occurrences of gold in Britain and Ireland. Department of Adult and Continuing Education Course Notes. Leeds University.
COGHLAN, H.H. and CASE, H.J. 1957. Early metallurgy of copper in Britain and Ireland. Proceedings of the Prehistoric Society: 91-123.
GEE, C., M.H.RAMSEY, J. MASKALL and I. THORNTON. 1997. Mineralogy and weathering process in historical smelting slags and their effect on the mobilisation of lead. Journal of Geochemical Exploration 58: 249 - 257.
HARTMANN, A. 1970. Prähistorische Goldfunde aus Europa - Spektralanalytische Untersuchungen und der Auswertung - Studien zu den Anfängen der Metallurgie 3. Berlin: Gebrüder Mann Verlag.
1982 Prähistorische Goldfunde aus Europa II - Spektralanalytische Untersuchungen und deren Auswertung - Studien zu den Anfängen der Metallurgie 5. Berlin: Gebrüder Mann Verlag.
IXER, R.A. 1994. Does ore petrography have a practical role in the finger-printing of rocks? in N.Ashton and A. David (ed.): Stories in Stone. Lithic Studies Occasional Paper 4: 10-23. Oxford: Lithic Studies Society.
1995. Ancient Mining: Modern Collections, Abstract, Mineralogy in the Field 1995. A one- day symposium organised by The Russell Society in association with the University Museum at Keble College Oxford.
1996. Ore petrography and archaeological provenance. Mineralogical Society Bulletin 113: 17-19.
1997. Detailed provenancing of the Stonehenge Dolerites using reflected light petrography - return to the light, in A. Sinclair, E. Slater and J.Gowlett (ed.): Archaeological Sciences 1995. Oxbow Monograph 64: 11-17.
2000. Potential and realizable ores from the Great Orme Mine. Journal of the Great Orme Exploration Society 1. 7-13.
IXER, R.A. and P.BUDD. 1998. The mineralogy of Bronze Age copper ores from the British Isles: implications for the composition of early metalwork. Oxford Journal of Archaeology, 17: 15-41
IXER, R.A. and J. DAVIES. 1996. Mineralization at the Great Orme Copper Mines, Llandudno, North Wales. UK Journal of Mines and Minerals 17: 7-14.
IXER, R.A.F. and R.A.D. PATTRICK,. 1995. Cu-Pb-Zn-Co-Ag mineralization at Ross Island - Muckross Mine, Killarney, Ireland. Abstract. Mineral Deposits Study Group meeting. Manchester.
IXER, R.A. and R.A.D.PATTRICK In press. Copper-arsenic ores and Bronze Age mining and metallurgy with special reference to the British Isles, in P.T.Craddock and J.Lang (ed.): The Prehistory of Mining and Smelting. London: British Museum Press.
IXER, R.A.F., R.A.D. PATTRICK AND C.J. STANLEY. 1997. The geology, mineralogy and genesis of gold mineralization at Calliachar-Urlar Burn, Scotland. Transactions of the Institution of Mining and Metallurgy. (Section B: Applied Earth Science) 106: B98-B107.
IXER, R.A. and C.J. STANLEY. 1996. Siegenite-bearing assemblages found at the Great Orme Mine, Llandudno, North Wales, Mineralogical Magazine 60: 978-982.
JENKINS, D.A. and D.B. JOHNSON. 1993. Abandoned metal mines: a unique mineralogical and microbiological resource. Journal of the Russell Society 5: 40 - 44.
LEAKE, R.C., D.J.BLAND and C.COOPER. 1993. Source characterization of alluvial gold from mineral inclusions and internal compositional variation. Transactions of the Institution of Mining and Metallurgy. (Section B: Applied Earth Science) 102: B65-B82.
LEAKE R.C. and R.J. CHAPMAN. 1996. Microchemical characterization of alluvial gold from Scotland. Abstracts Volume. Mineralisation in the Caledonides. Institution of Mining and Metallurgy Meeting. Edinburgh.
LEAKE R.C., R.J. CHAPMAN, D.J. BLAND, E.CONCLIFFE and M.T. STYLES. 1997. Microchemical characterization of alluvial gold from Scotland. Transactions of the Institution of Mining and Metallurgy (Section B: Applied Earth Science) 106: B85 - B98.
NORTHOVER, J.P. 1982. The exploration and long distance movement of bronze in Bronze Age and Early Iron Age Europe. Bulletin of the Institute of Archaeology 19: 45-72.
O'BRIEN, W. 1995. Ross Island and the origins of Irish-British metallurgy in J. Waddell and E. Shee Twohig (ed): Ireland in the Bronze Age: 38-48. Dublin: Stationary Office.
PARK, C.F. and R.A.MACDIARMID. 1970. Ore Deposits. 2nd Edition. San Francisco: W.H.Freeman and Co.
POLLARD, M., R.G.THOMAS, D.P.WARE and P.A. WILLIAMS. 1991. Experimental smelting of secondary copper minerals: implications for Early Bronze Age metallurgy in Britain, in E. Pernicka and G.A.Wagner (ed.): Archaeometry '90: 127-136. Basel.
REEDY, C.L. and T.S.REEDY. 1988. Lead isotope analyses for provenance studies in the Aegean region: a re-evaluation, in E.V. Sayre et al (ed.): Materials issues in art and archaeology: 65-70. Pittsburgh: Materials Research Society.
ROHL, B.M. 1995. Application of lead isotope analysis to Bronze Age metal work from England and Wales. Unpublished D. Phil thesis. Oxford University.
SAYRE, E.V., K.A.YENER, E.C.JOEL and J.L. BARNES. 1992. Statistical evaluation of the presently accumulated lead isotope data from Anatolia and surrounding regions. Archaeometry 34: 73-106
TAYLOR, J.J. 1980. Bronze Age Goldwork of the British Isles. Cambridge. Cambridge University Press
TAYLOR, J.J., R.J. WATLING, C.A. SHELL, R.A.IXER, R.J. CHAPMAN, R.B. WARNER and M.
CAHILL. 1997. From gold ores to artefacts in the British Isles: A preliminary study of a new LA-ICP-MS analytical approach, in A.Sinclair, E. Slater and J.Gowlett (ed.): Archaeological Sciences '95. Oxbow Monograph 64: 144-154.
TIMBERLAKE, S. 1995. 'Copa Hill' and 'Llancynfelin Mine'. Archaeology in Wales 35: 40-43
THWAITES, T. 1995. Gold fingerprints spot the fakes. New Scientist 147 (1985): 24.
WARNER, R.B. 1993. Irish Prehistoric goldwork: a provisional analysis. Archaeomaterials 7: 101- 113.
YOUNGER, P. 1995. Polluted groundwater in abandoned coalfields: some hydrogeological and geochemical idiosyncrasies. Fourth annual conference on groundwater pollution. 11pp. London, March 1995.
Originally Published 1999 in 'Metals In Antiquity'
(Eds. S.M.Young, M.Pollard, P.Budd and R.A.Ixer)
POB 920, Oxford, England, OX2 7YH
Tel/fax: +44 (0)1865 311914
Return to Homepage