The scientific study of archaeological ‘stone’ includes obsidian, chert, flint, marble, turquoise, jade and so on. Each rock type can be identified and classified according to its mineralogy, structure, texture and, when present, its fossil content. This may allow the pressures, temperatures and other environmental conditions in which it was formed to be deduced. The nature of minerals and their development is discussed in the section on ceramic raw materials (see Sections 4.2 and 4.2.3), so will not be repeated here.
Flint is basically a silica-rich (siliceous) material. Silica is often also the chief component of glass, but instead of being entirely amorphous like glass, flint and chert are composed of masses of minute interlocking quartz crystals and fibres with water-filled micropores between them (i.e. it is cryptocrystalline like chalcedony and chert). Flint can also contain amorphous silica called opal. Chert is characterised by a splintery fracture and flint by a marked conchoidal fracture, the latter giving millions of man-made flint artefacts their characteristic appearance. Sometimes the quartz crystals can be replaced by other substances; flint also contains microfossils. It is probably formed as part of a post-depositional process in carbonate sediments in which silica is redistributed. Chalk is composed of fossilised skeletons of millions of minute marine organisms. Some non-carbonate minerals are also preserved in the flint. The chemical composition of the flint is therefore partly a reflection of its formational mechanism, partly the chemical composition of its host sediment and also partly the extent to which the formational mechanism has gone to completion’. Flint and chert occur in a variety of forms. In Britain chert occurs most often in tabular masses and flint in nodules, though at Grimes Graves in Norfolk tabular masses of flint were worked as well as some nodular wall-stone. Flint and chert cannot be distinguished by their crystallinity.
One of the better known sources of flint in Europe is the Neolithic site of Grimes Graves in Norfolk which is thought to have been used for a short period around 2000 BC. This mine is in a well-drained heathland environment in eastern England which is rich in chalk and in the Cretaceous was a flat sea bed. The only visible clue on the disturbed ground surface for the existence of subterranean flint mines is a series of shallow depressions. These depressions are where the vertical shafts have been backfilled with material and a certain amount of subsidence has occurred. There were between 350 and 500 shafts so the activities were on a massive scale. However, this scale of activity was not, by any means, especially unusual with at least 250 mine shafts at Easton Down and Martin’s Clump in Hampshire. The mining of flint would have been like mining many other materials that occur in horizontal seams, either rock or mineral. At Grimes Graves a series of vertical conical shafts of 10–15 m have been dug through the surface frost-shattered flint down into better quality unaltered flint.
In following the beds of flint, miners dug horizontal galleries from the vertical shaft of between 3–7 m in length. In some environments the flint could be broken away from the walls by heating it up by lighting fires next to the walls of the galleries. Once the flint was hot, cold water would have been thrown onto it, causing it to crack. Some evidence of fire has been found at the foot of the shafts at Grimes Graves which may have been used to break up large masses of extracted flint. However, the softness of the chalk and the brittleness of the flint at Grimes Graves suggests that fire would not necessarily have been used as a means of breaking up the flint in the galleries. Once broken into suitably sized chunks the flint would have been loaded into baskets and hauled up ladders resting on the sides of the shafts to the surface. The large chunks would later have been broken down into suitably sized nodules for working into implements. It is clear from archaeological investigations in flint mines that Neolithic man was only interested in a small proportion of the many layers of flint through which the shafts were sunk.
Although Grimes Graves has been researched in some detail, the level of production based on the number of filled-in mine shafts detected is dwarfed by the site of Rijkholt-St Geertruid in the Netherlands which dates to c. 3000 BC. Here some 5,000 mining shafts have been found. On the assumption that all the mines were used for 500 years and continuously (both rather big assumptions) the excavator, P.W. Bosch calculated that some 150 million Neolithic flint axes would have been produced.
In the case of the chert mine at Wadi el-Sheikh some 160 km south of Cairo in Egypt, the complexity of how the working deposits accumulated over time was considered (see Figure 6.1; Weisgerber 1987). Here both open and underground quarrying techniques were used with a primary phase in the New Kingdom, and with some evidence of older workings. Some of the tool marks found in the mines have different characteristics from those found in European mines: they appear to have been made with metal tools. It is suggested that this strengthens the interpretation that there were both stone- and metal-using phases of mining. The mining activity produced large dumps of debris of up to 3 m in height – which is likely to have been the original appearance of other ancient quarries such as Grimes Graves in England, Rijkholt-St Geertruid in the Netherlands or Spiennes in Belgium. The preservation at Wadi el-Sheikh is so good in the desert environment that ‘it is as if they have only recently been abandoned’ (Weisgerber 1987: 168). Interestingly, stone grooved hammers similar to those found in European prehistoric copper mines were found in the mining debris, but the groove was close to the end rather than in the middle, as in European examples. Chert picks were also found.
Figure 6.1 The nineteenth-century map of chert mins at Wadi el Sheikh, Egypt.
Some archaeologists rely on a visual identification of stone but the results which include their perception of colour can not only be arcane but also difficult to communicate. A chert type, for example, is obviously formed as a result of a natural process and although a degree of visual homogeneity may result, this is not guaranteed. Chert is a siliceous material deposited or formed by secondary solution and re-deposition over a restricted period of time in a restricted geographical area. One way through the minefield of subjective perception of colour is to use colour charts. But colour is only one attribute of stone and often an unreliable source indicator. A more comprehensive characterisation of stone is usually required. For example, Luedtke (1978) has found that if chert types were formed close in time and space they tended to have similar proportions of trace elements, as determined by neutron activation analysis; by using discriminant analysis she was able to distinguish between cherts with different sources in the North American mid-west.
As with other work which attempts to source rocks in order to build up exploitation and distribution patterns during different periods of prehistory and history, it is important to properly sample geological sources so as to include the full range of physical variation for the chert type. By doing this the aim is to determine the full range of compositional variation of the source before archaeological samples are investigated scientifically. Although of course it is still possible to chemically analyse archaeological artefacts made out of rock in an attempt to identify different compositional groupings, for rocks like cherts only the degree of compositional variation within and around a source may provide the evidence necessary to begin to identify the source’s chemical signature. Chemical variations can exist within formations and this is why it is so important to sample sources such as quarries adequately. Once adequately characterised, it becomes possible to show, through the analysis of artefacts, how quarries were used over time.
A recent example of this is where scientific analysis has accompanied one of the most comprehensive investigations of a stone quarry, on a large scale. This has been the Roman quarry at Mons Claudianus in the Eastern Desert of Egypt (Peacock and Maxfield 1997). Here survey and excavation of the quarries as well as adjacent buildings, settlements and fort has placed the quarrying activities in a solid regional and environmental context. Chemical and penological analysis of the grey plutonic rocks has been very successful and has produced a means of sourcing the rock and therefore distinguishing it from similar rocks in other regions.
However, a comprehensive study of chert and chalcedony use and distribution in Belize has made a contribution which illustrates both the potential and the limitations of using an appropriate trace analytical technique to study geological reference groups and associated artefacts made from it (Cackler et al. 1999). As with flint, the colour of the chert did not have a particularly significant role to play in its characterisation (Cackler et al. 1999: 394). In this case, variations in chert colour were found to be due to weathering. By investigating statistically the geological samples, the only distinction that could be identified analytically was a distinction between northern Belize chert, Crooked Tree chalcedony (which was lower in most trace elements than chert) and an unidentified material from Kichpanha. Hudler et al. (1996) were however able to distinguish between chert quarries at Colha in Belize due to micro-variation in its composition, but this would have a restricted applicabilty to the sourcing of archaeological artefacts.
By using neutron activation analysis and examining the structure of the data produced using factor analysis, Hoard et al. (1992; 1993) have shown that it is possible to distinguish between groups of siliceous materials from the White river valley in the USA. These cryptocrystalline siliceous materials (chalcedony and chert) of the central Great Plains were used to make chipped stone tools. Since specimens from the three known sources at Flattop Butte in north-east Colorado, Table Mountain in east central Wyoming and the White river badlands of South Dakota are visually indistinguishable, this was a distinct step forwards (Hoard et al. (1993) Figure 2) showing the value of such work.
Steatite (or soapstone) is a rock mainly composed of talc and minor amounts of other minerals such as chlorite. It was used for the manufacture of objects such as beads in Mesopotamia (Moorey 1994: 75–6). It was probably used for the manufacture of vessels on sites of the Halaf period through to the fifth millennium BC (ibid: 39–40). Kohl et al. (1979) used X-ray diffraction analysis and other means to investigate 375 samples of ‘soft stone’, predominantly from carved artefacts (but including specimens from potential sources) of the late fourth millennium to early second millennium BC from south-west Asia, including Mesopotamia.
In the eastern United States steatite was extensively mined in prehistory. It was used in the late Archaic period and possibly in the early Woodland period for the manufacture of lugged bowls. Sedentary farmers later used it for the manufacture of smoking pipes and decorative items. The use of neutron activation analysis to attempt to source steatite in the area has met with success. Allan et al. (1975) used the relative concentrations of rare earth elements (the lanthanide group of chemical elements with atomic numbers of between 57 and 71) to chemically characterise steatite according to the quarry that it came from. Even the analysis of steatite from the same zone but from quarries several miles apart could be distinguished. They used a technique familiar to geochemists: first the rare earth concentrations are normalised to average chondritic values, then the ratio of the steatite values to chondrites is plotted again the atomic number of rare earth elements such as lanthanum (La), samarium (Sm) and europium (Eu). The reason why normalised values are used is that elements with even atomic numbers tend to occur at higher concentrations than those with odd atomic numbers. This is due to relative nuclear stability and relates to the origin of the element. The average concentration of each element in chondritic meteorites is used for normalising the results, because the abundance of elements in these meteorites is usually taken as the average abundance in the solar system. Geological processes cause fractionation of rare earth elements and it is this fractionation of one relative to the others that can provide a means of linking their chemical characteristics to a source for the steatite. Test analyses on steatite artefacts from the Shenandoah valley has provided useful evidence for two groups exploiting widely separated sources (Allan et al. 1975: 80).
Flint was used to manufacture some of the earliest tools used by man that have survived. Because it can be worked to produce sharp blades with characteristic conchoidal fractures, it was used to make, among other artefacts, including hand axes, blades, scrapers and arrowheads. Some of the questions which archaeologists are keen to answer in relation to flint use and characterisation are:
• How was flint used in ancient societies?
• Why were particular flint sources used rather than others?
• Over what distances did the flint travel from its source?
Programmes of archaeological science research can be constructed to answer these questions. To answer how flint was used in ancient society, a wide range of archaeological features must be taken into account, including where the flint was found on the site, the kind of site the flint was found on, how the flint artefacts were made, what ‘significance’ they held in the archaeological assemblage, what the flint artefacts were used for and what other lithics were used (e.g. stone axes) on the site? Abrasion studies, cut marks on bones and forensic archaeology can all contribute to the overall interpretation in different ways.
One of the questions that can be investigated with chemical analysis is an attempt to establish what distance the flint travelled from its source. This relies on the premise that it is possible to source the flint chemically to a particular location or zone, and to build up a distribution zone from it to the archaeological sites on which it was used or worked. To some extent this also assumes that no flint mines have been destroyed by natural processes, such as by coastal erosion. It would be hoped that the geological conditions in which the flint formed would be such that the trace elemental composition would be characteristic of the geological stratum or age. As mentioned above, slight differences in the characteristics of formation, such as the chemical structure of the host sediment and the degree to which its formational process has gone to completion, should lead to specific compositional signatures for flint found at different (vertical and horizontal) locations. Indeed Cowell et al. (1980) found a consistent compositional variation between layers of flint at Grimes Graves.
Some of the earliest chemical work on flint was carried out using emission spectrometry (Sieveking et al. 1970) and neutron activation analysis (NAA), which is capable of measuring some elemental concentrations to below 1 ppm. Both techniques have been used to investigate the possibility of chemically characterising flint sources in Britain (Aspinall and Feather 1972; Aspinall et al. 1975; De Bruin 1972). This showed that it was sometimes possible to distinguish in general between broad geological areas in which flint occurs, such as the relatively high levels of thorium detected in flint from Grand Pressigny (France) which distinguished it from the flint from Dutch sites of St Gertrude and Rijkholt, from all the English flint tested and from Spiennes in Belgium. Also, the thorium concentrations found in Grimes Graves flint has a mean and standard deviation which was diagnostic. However, subsequent research using atomic absorption spectroscopy (AAS) (Craddock et al. 1983) has found that it is possible to chemically characterise flint more precisely, linking chemical compositions to specific areas within England: East Anglia, the South Downs and Wessex. One of the reasons why this work was more fruitful is that AAS can be used to detect a somewhat different suite of elements. In this case the elements detected were aluminium, iron, magnesia, potassium, sodium, calcium and lithium, at levels above 50 ppm. Apart from the obvious high silica levels the elements which provided the characterisation were thought to be present in the non-carbonate minerals (alumino-silicates, phosphates etc.), unreplaced low magnesia calcite and trapped interstitial water (Cowell 1979; Craddock et al. 1983: 138).
They chemically analysed flint derived from 11 mining areas (including 2 nearby mines in their ‘South Downs A’ group and 4 in their ‘South Downs B’ group). Craddock et al. (1983) compared these to the chemical compositions of 400 polished flint axes, of which 300 derived from the presumed 3 main areas in which the flint mines are located: the South Downs, Wessex, and East Anglia. Through the chemical analysis of 100 axes derived from excavated Neolithic sites within these areas Craddock et al. showed an apparent discrimination between the flint derived from these mines based on their relative concentrations of iron and aluminium. They were able to indicate, on the basis of the proportion of flint axes found in each region, which mine the flint was derived from. This produced some intriguing and some unexpected results. Although it might be assumed (somewhat simplistically) that the area in which the axeheads were found would have supplied the flint to make them, in two of the three cases this was not so (see Figure 6.2). In the South Downs region more than 50% of the flint axes analysed derived from the South Downs area. However, more than 50% of the axes found in the Wessex region were also made from South Downs flint and rather more than a third of the axes found in the East Anglia region were made from South Downs flint.
Figure 6.2 The relative proportion of different kinds of flint (from the South Downs, Wessex, East Anglia, other areas and unclassified sources) used for making flint axes which have been found in these same areas of flint sources
© The British Museum.
Grimes Graves (an East Anglian mine) apparently supplied less than 10% of the flint used for axes in the East Anglian area which is about the same proportion as the Wessex flint used. In all, 67% of the axes were assigned to a South Downs source. One explanation for these unexpected findings is one of dating. Grimes Graves was used for a short period in the latest Neolithic. Other flint mines were exploited much earlier: for example, some of the South Downs flint mines in the fourth millennium BC. Since considerably fewer axes were apparently used in the late Neolithic, it is perhaps not surprising that few axes made from flint mined in the late Neolithic have been found. About one quarter of the flint axes which were analysed in this study remained unsourced, indicating that further work is necessary to locate the sources, though they may now be destroyed. However, the results clearly provide an excellent step forward showing that flint which was used to make axes travelled some distance from its source. Nonetheless, as Craddock et al. point out (1983: 160) chemical analysis of flint artefacts other than hand axes might have produced the allusive analytical proof for the use and distribution of Grimes Graves flint. When compared to the results of stone axe provenance studies (see Section 6.8 below), reasons other than the working properties of the flint may need to be put forward for the use of ‘alien’ flint, such as cultural, in areas where flint is already available. Another result of this research is that it shows how important it is to use an analytical technique which can measure the appropriate range of elements at the appropriate level of sensitivity for the characterisation of the material. (This may only become clear when research has already shown that the results from using a particular technique are inappropriate or unpromising.) Clearly a comparison of the chemical compositions of flint artefacts from dated archaeological contexts with those from (where possible) seams of contemporary dated flint which have definitely been exploited by man is a priority.
Subsequently, inductively coupled plasma atomic emission spectroscopy has been used for the chemical analysis of flint with another successful outcome. Thompson et al. (1986: 247) showed that some elements, such as barium and beryllium, for example, provided a distinction between flint deriving from Cissbury and flint deriving from Black Patch, Church Hill and Harrow Hill.
Consigny and Walter (1997) provide a promising set of results for the chemical analysis of flint and chert from the Paris basin. Although not fully published, they indicate that by using particle-induced X-ray emission (PIXE) they are able to chemically distinguish siliceous materials from the same geological layer but different geographical regions (see Figure 6.3). Most work had focused on the central part of the Paris basin in the Ile-de-France using techniques such as micropalaeontology. The results showed that Upper Cretaceous (Senonian and Turonian) cherts associated with chalk and Tertiary cherts associated with lake limestone had been used. It is easy to distinguish between the two kinds, but it was found to be much more difficult to distinguish between cherts of the same type from different locations. In order to address this problem the research involved sampling geological deposits from a range of geographical locations, including from the valleys of the rivers Eure, Orge, Loing, Lunain, Yonne and Vanne. Both Upper Cretaceous and Tertiary formations were sampled. The research involved taking samples of raw chert and flint, where possible, from the same and different geological formations (i.e. both horizontal and vertical sampling). It was found that the further apart the deposits of the same geological stratum were, the greater the difference in trace elements detected in each (Consigny 1992). This was found to be true of chert and flint deposits of the same geological age: the trace element characteristics were found to gradually change as the sampling point moved away from a location along a single geological stratum.
Figure 6.3 The distribution of chert and flint in the Paris basin and surrounding area and the principal sites mentioned in the text.
Three groups of chert which probably derived from Champigny limestone were examined. They derived from the archaeological sites of Etiolles and Donnemarie-Dontilly, La Fouillotte and from another site, Villejuif. Etiolles is a late Magdalenian site in the centre of the Paris basin, on the right bank of the Seine to the south-east of Paris. There is evidence here for the manufacture of flint blades from large cores (Pigeot 1987). Donnemarie is situated 80 km to the south-east of Paris. It is an Upper Palaeo-lithic open air settlement which used long blades. Chemical analysis of the cherts found at the sites, probably associated with Champigny limestone, showed that they could be distinguished chemically. The chert from Donnemarie is characterised by the presence of germanium and uranium, that from Etiolles by the presence of germanium, uranium and zirconium, and that from another location, Villejuif (for which no locational information is given), by unusually high levels of uranium. These elements have been used in the principal components plot in Figure 6.4.
Figure 6.4 A principal components analysis of chert from Etiolles, Villejuif and Donnemarie.
Flint from two phases of use at Etiolles was also analysed which led to a more complex and only partially-resolved set of results. They used the same criteria to chemically characterise the flint as for the chert: presence or absence of elements and/or the levels of trace elements. Consigny and Walter discovered three compositional types. One group characterised by uranium, arsenic and germanium could only be broadly sourced to the Tertiary limestone (Consigny and Walter 1997: 341). Analysis of the second type, a golden brown flint, revealed three compositional types of which only one could be sourced to the valley of the river Loing which had been suggested before the analysis had been carried out on the basis of macroscopic characteristics. A source for the second compositional type of golden coloured flint could not be suggested. The third type, which is characterised by niobium, could not be matched with any geological samples examined, but it was found that flint from a Magdalan-ian site of Abri du Lagopède at Arcy-sur-Cure about 150 km to the south-east was also characterised by this element. The authors suggest that the sites are unlikely to be linked, but more sensibly they suggest that flint found at the two sites was derived from the same source of flint. However, in spite of the fact that niobium is a rare trace element in flint, until the source of flint containing niobium is located, it is difficult to be sure how close together the sources were which provided flint for the two sites.
This research by Consigny and Walter provides a glimpse of what might be possible: it provides a possible framework for research in other flint-rich environments, such as across the English Channel in England. Given that a level of flint characterisation has been achieved using neutron activation analysis, ICPAES and especially AAS there appears to be a great potential to build on this work. Clearly a thorough geological survey of flint and chert sources is an important component in the investigation of the sources that may have been used in antiquity.
To the north of these French geological formations, but of associated age, Holmes and Harbottle (1994) have used neutron activation analysis to examine Lutetian limestone from the area around Paris. They found that while petrography was able to localise it to a stratum belonging to several quarries, NAA was able to produce finer distinctions between quarries belonging to the same geological stratum. They also mention that their findings can have significant art-historical implications in sourcing the limestone used to build and embellish Nôtre-Dame Cathedral, for example.
Research into the sourcing of obsidian, a stone which it might be claimed is one of the better candidates for chemical characterisation, has produced some exceptional results. Obsidian studies can be viewed as representing a microcosm of the way in which archaeological materials research has developed and progressed, starting with the use of relatively small data sets due to the limiting and time-consuming nature of the scientific techniques used, through to the use of more highly automated techniques which are able to generate archaeologically coherent data sets (Jope 1989). Importantly if the research project is set in an appropriately broad framework, results from the chemical analysis of obsidian have not only enabled it to be sourced, but have also provided evidence for the ways in which obsidian from different sources was used during different periods of prehistory.
Obsidian was one of a range of lithics used to make tools. By taking the use of other lithics into account the study of chemically characterised obsidian use brings a range of archaeological factors into play. These include the proportion of tools from a dated site assemblage made from obsidian; the changes in obsidian source exploitation over time; the size and function of the sites on which the obsidian was found; the kind of trade/exchange networks involved in the distribution of obsidian; the reasons why a specific type of obsidian was preferentially used. Combined with an archaeological survey and excavation of obsidian-using sites, as well as a comprehensive survey of the obsidian sources, the complex archaeological information which can be derived from such studies contributes in an important way to mainstream prehistoric archaeology. Some of the early work with obsidian had already provided important inferences about trade/exchange systems (Renfrew et al. 1965; 1966; 1968).
Obsidian is a naturally-occurring glass which is formed during volcanic eruptions. It is almost entirely a glass which formed as a result of rapid chilling at the earth’s surface; some obsidians contain up to 15% crystals. The lavas which produce obsidians generally contain greater than 66% silica (they are rhyolitic in composition), and are therefore classified as acid rocks. Some have a more basic composition and are called dacite or trachyte. Chemically, obsidians can be defined according to the balance of alkalis (soda and potassium oxide), silica, aluminia and calcium oxide; the ratio of alkalis to aluminia distinguish between peralkaline’ and ‘subalkaline obsidians, for example (Macdonald et al. 1992). One of the main compositional differences between obsidian and a typical man-made soda-lime glass is the much higher level of aluminia in obsidian, which normally occurs at levels above 10% (as opposed to about 2–5% in most man-made glass) and the much lower levels of calcium oxide (generally less than 2%, as opposed to 6–8%). One result of these compositional differences (especially the differences in alkali and aluminia levels) is that obsidian melts at much higher temperatures than man-made glass. It is therefore a material which man has not, until recently, attempted to modify by heattreatment. Obsidians weather at their surfaces as a result of interaction with water in the environment (hydration) and under some (restricted) circumstances this can be used as a form of dating (Ericson 1988), although any success has mainly been restricted to its application in the US.
Obsidian has been used for the manufacture of blades and arrowheads, and also for the manufacture of bowls. The quality of obsidian is clearly important in determining what it was used for and Renfrew et al. (1965) noted that in Classical contexts it was used for the manufacture of seals, tesserae, mirrors and statues.
In some instances obsidians are apparently sufficiently homogeneous and chemically distinct for them to be characterised chemically in terms of each eruption and lava flow (Gordus et al. 1968; Michels 1982). The reason for this homogeneity is its rapid solidification from a liquid state, preventing any mixing. Bowman et al. (1973) have however discovered variability in some elements due to magmatic mixing prior to eruption, although for the characterisation of archaeological obsidians and their sources this has apparently not yet proved to be a problem. Having said that, those who collected obsidian in prehistory may have only returned to a limited number of loci on an obsidian flow (or perhaps a single locus), which would lead to limited variation in the chemical composition of the obsidian used to make artefacts. Jack (1976) used less than five analyses as a means of ‘characterising’ most of the major obsidian sources in California because it was assumed that its compositional homogeneity made a more detailed geochemical survey unnecessary. Hughes (1994) has however investigated closely the variation in the chemical composition of obsidian in the Casa Diablo region of California. Although this obsidian source had been assumed to be geo-chemically homogeneous, analysis of 200 samples from twenty different locations revealed the existence of two or possibly three potentially different sources (Lookout Mountain, Sawmill Ridge and Prospect Ridge). Hughes (ibid.: 268) points out that this more specific characterisation provides archaeologists with the opportunity to resolve temporal, spatial and production histories involving each obsidian type. It could also allow archaeologists to investigate whether more than one source of obsidian was used by a prehistoric community at any one time; currently the obsidian artefacts which can be traced to the Casa Diablo source were derived from sites which appear only to have used a single source at a given time. These kinds of studies alert us to the possibilities of geochemical variations, perhaps even within ‘single’ flows (Pollard and Heron 1996: 87). Hughes (1994) working in California, Tykot (1997a) working with Sardinian obsidian, and Yellin (1995) working with Anatolian obsidian, all emphasise the importance of chemically analysing a sufficient number of obsidian samples to monitor any geochemical variations at different volcanic sources; these three authors also underline the importance of recording precisely the locations from which samples of obsidian are removed for analysis during field surveys.
Because some geological sources of obsidian are identifiable in the field today, it can be possible to relate the composition of a specific obsidian source to the composition of obsidian artefacts made using material from that source. Obsidian was used at least as early as Mesolithic times in the Mediterranean (Hallam et al. 1976; Blackman 1984), and earlier still in other parts of the world (Michels et al. 1983; Merrick and Brown 1984; Gowlett and Henderson, forthcoming; Perlès 1987).
The first techniques used for obsidian analysis were optical emission spectroscopy and petrology, and these were later followed by techniques such as neutron activation analysis, electron probe microanalysis, X-ray fluorescence spectroscopy and, most recently, inductively coupled plasma emission spectrometry (Heyworth et al. 1988). The results of these analytical techniques have indicated that trace and minor element compositions have most discriminative value for obsidians in the Mediterranean area. In the western Mediterranean, traces of lanthanum, scandium and caesium in particular have provided a means of fingerprinting obsidian sources. In the analysis of middle Stone Age East African obsidians minor levels of calcium and iron have distinguished between sources. It follows therefore that it should be possible from the chemical analysis of obsidian to work out distribution zones in a particular period, the use of obsidian relative to other chipped stone resources, and to suggest mechanisms of distribution. Much of the work by Renfrew and co-workers and by Ammerman has attained these objectives. One result of these studies is clear: the contacts which can be inferred between the source of the obsidian and the points at which the obsidian (artefacts) have been found, based on their chemical constituents are in most cases completely new.
It was not until c. 7500 BC that obsidian artefacts were moved in any quantity, or with any regularity, beyond the immediate environs of the source areas in the Near East. Cann and Renfrew (1964) and Renfrew et al. (1966; 1968) discerned two principal source areas for obsidian in the Near East and the sites which they supplied during the seventh and sixth millennia BC. Chemical characterisation was based on trace levels of caesium, thorium and scandium. Renfrew and co-workers referred to these sources and the sites in their immediate vicinity as ‘supply zones’. The supply zone in central Anatolia was labelled the Cappadocian source, based on the volcanoes near Agicol and Çiftlik. The second supply zone, in eastern Anatolia (Armenia), is based on the volcanic sources of Nemrut Dag and Bingol. Supply zones were defined as the area within which sites have chipped stone industries consisting of greater than 80% obsidian (expressed in terms of the number of artefacts). It is worth emphasising the importance of this idea of proportion for the chipped stone industry, because it is a way of standardising data so that different site assemblages can be directly compared, irrespective of the numbers of artefacts involved. Wright (1969: 48) has suggested that, in some instances, the total weight of obsidian would provide an alternative means of defining the proportion of material involved, although even this suggestion has potential problems; both sets of information provide different levels of meaning and one way forward is to present both % weight and % number of artefacts with distance from source. However, chemical characterisation is the basis of this work. From the work of Renfrew et al. (Figure 6.5), it is apparent that the central Anatolian supply zone extends up to 250 km from the source, whereas the Armenian supply zone extends over 300 km from the source. Beyond the supply zone is the ‘contact zone’, where the fall-off in the amount of characterised obsidian is exponential and is therefore plotted on a logarithmic scale in Figure 6.5. (This means that the scale on the left [% obsidian] increases by a factor of 10 with each major subdivision).
Figure 6.5 The percentage of obsidian (logarithmic scale) in the chipped stone industry versus distance from source for early Neolithic sites in the Near East. The shaded area is the supply zone; the straight lines indicate approximately exponential fall-off of obsidian from the contact zones. Triangles represent sites in central Anatolia and the Levant supplied by the Cappadocian source; squares represent sites in the Zagros mountains area supplied by Armenian sources. T.S. = Tell Shernsharah; JA = Jarmo; BO = Bouqras; TG = Tepe Guran.
It is evident that with the sharp fall-off in % obsidian beyond 250 km or 300 km, at the edge of the supply zone, the mechanisms involved in its distribution in the contact zone must be somewhat different. Where there is a higher percentage of obsidian around the source within the supply zone it can be suggested that any trade that occurred was intensive, and possibly that the obsidian consumers travelled to the source themselves. At least it can be suggested that the people who procured the obsidian probably formed an integrated and dependent part of the supply system (Figure 6.5). In the contact zone band the trade falls off, and indeed it is possible that the amount of obsidian traded was deliberately withheld to inflate its value. The term Renfrew (1975; 1977) has applied to this form of trade or exchange is ‘down-the-line’. This form of trade involves the transmission of ‘goods’ from one settlement to another in a linear fashion and is normally applied to land trade. It is unfortunate that other components of the chipped stone industry in the same regions of the Near East have not apparently been studied in the same way, since this could partially provide us with a fascinating series of competing zones for different types and qualities of materials. It would then be possible to broach the tricky concept of relative value.
Wright (1969: 51) has suggested that this model of supply and contact zone would not work for Armenian material if obsidian from separate phases of sites are considered. Wright (ibid.: 51) also suggested that a consideration of different functions of sites in the study area might make a difference; sites which specialised in obsidian artefact production could have a significantly larger number and weight of obsidian artefacts, including debitage associated with production, and in making comparisons with non-production sites the artefact types should be included. Torrence (1986: 97–8) has also emphasised this approach combined with a consideration of relative proportions of stone type found in specific site phases. Thus, a comparison of % obsidian (however this is expressed) between contemporary sites which specialise in obsidian artefact production and between those which have other functions in the zones(s) around the source of obsidian would appear to offer a means of comparing like with like, where the % weight of artefact assemblages (including debitage) can be directly compared. Nevertheless, the distribution patterns published by Renfrew et al. reflect regular but not necessarily regulated behaviour during the seventh and sixth millennia BC and there need not necessarily have been any formal exchange or trade organisation involved. Their research blazed the trail for subsequent research in obsidian analysis and is especially significant for its contribution to modelling distribution mechanisms.
In an analytical study of Anatolian obsidian by Gratuze (1999) using a very sensitive analytical technique, ICP-MS with the sample being introduced by laser ablation, he claims to have detected nine separate Cappadocian sources. By comparing his results with those produced by other workers who have used neutron activation analysis and X-ray fluorescence spectrometry he has also been able to confirm some attributions they made to specific sources where there had been an element of doubt. Gratuze (ibid.) also claims that laser ablation ICP-MS will replace NAA and XRF as a virtually non-destructive technique (see Section 2.2). However, the level of destruction that is involved in using NAA/XRF and ICP-MS in liquid mode, as used by Tykot (1997a), is evidently acceptable.
In Italian contexts (Ammerman 1979) has followed an inclusive approach to the sourcing and distribution of stone, one which takes into account the types of site which produced the stone (and whether they specialised in working it), the different artefact types and the relative sizes that have been found, the proportion of different stone types found on each site as well as the distance from stone sources. Early work on obsidian sources in the western Mediterranean has shown that four principal obsidian sources could be chemically defined, mainly by using neutron activation analysis and based on the lanthanum to scandium and caesium to scandium ratios. Figure 6.6 shows a plot of ppm of barium versus ppm of zirconium in obsidians from a range of sources in the Mediterranean and Africa. The figure underlines how obsidian from Sardinia, Lipari and Pentellaria can be distinguished, but also the extent that they differ compositionally from Anatolian sources. Hallam et al. (1976) later showed that it was possible to distinguish compositionally between the four basic sources which are all on islands near Italy: Sardinia, Lipari, Pantelleria and Palmarola, the latter in the Pontine islands. Let us now look at these four main geochemical groupings in terms of their geographical distribution (Figure 6.7). The four zones given are what Hallam et al. (1976) have referred to as obsidian interaction zones. These are defined as ‘the area within which sites derive 30% or more of their obsidian from the same specific sources’. Of course by increasing the percentage for the definition of an interaction zone from 30% to 50%, we could produce a rather different, more contracted pattern leading to a different archaeological interpretation. Since Hallam et al. (1976) published their paper, Thorpe et al. (1984) have shown that Pantellerian obsidian reached southern France, where it was excavated from Copper Age contexts.
Figure 6.6 Parts per million barium versus parts per million zirconium detected in obsidian samples from a range of sources in the Middle East, in the Mediterranean and in Africa.
Figure 6.7 The distribution of chemically characterised obsidian in and near Italy and southern France; SA = Sardinian A obsidian; SB = Sardinian B obsidian; SC = Sardinian C obsidian; P.l = Palmarolan obsidian; Pa = Pantellarian obsidian. The different zones are obsidian interaction zones
(Cann et al. 1968; reproduced with kind permission of the authors and the Prehistoric Society).
Francaviglia (1988) noted that Neolithic and late Bronze Age peralkaline obsidian found in Sicily could also be sourced at Pantelleria. Francaviglia (ibid.) analysed 143 obsidian samples from Pantelleria, a considerable sample size, and found that there were at least five different obsidian sources on the island, though the exact location of some were uncertain. Not only does this work show how important a large sample size is, it also modifies the distribution patterns given in Figure 6.7, which specifically excludes the occurrence of Pantellerian obsidian from Sicily in the Neolithic and Copper Ages; a distribution map which included this new data would need to incorporate the percentage of obsidian from each source per site, and provide results for obsidian use in each period. An important consideration is the analysis of a statistically acceptable proportion of obsidian from a site in order to establish within-site compositional variability and to provide the basis for an estimate of the proportion of obsidian used from each source. Naturally, to be confident of establishing a reliable pattern of distribution, it is essential to analyse a sufficiently large amount of obsidian from well-dated archaeological contexts. Some early analytical projects on obsidian suffered from inadequate sampling, though in fairness this must be seen in the context of the more time-consuming and labour-intensive nature of carrying out the chemical analyses at that time.
Figure 6.7 is a conflation of different time period distributions for sub-periods within the Neolithic and Copper Ages (Hallam et al. 1976: Figure 4). While the locations of the origins of obsidian plotted in the distribution patterns are well worth working out, it is quite possible that the four sources were exploited at different intensities and for different lengths of time through the period considered, from c. 5000 BC to c. 2000 BC. Nevertheless, the work by Hallam et al. laid important foundations for later work and, as we have seen, these distributions have been modified by subsequent research. What we should aim for in each project should be the analysis of a sufficient number of well-dated obsidian artefacts which can be treated as representative of specific periods rather than of a broad sweep of time. This would lead to fine-tuning of the data and more specific archaeological interpretations. An interesting feature of Figure 6.7 is the Palmarolan zone, which lies within the Liparian interaction zone. Its existence can probably be explained in technical terms: it may exist because Liparian obsidian occurs in much larger lumps producing longer blades than obsidian from the Palmarolan zone, making it more attractive and saleable over a larger area (Hallam et al. 1976: 99). The distribution of obsidian from Lipari actually extends beyond the Palmarolan zone and perhaps this is an example of competitive trading. Renfrew (in Hallam et al. 1976) suggests that sea travel would have been avoided. However, Ammerman (1979: 102) has now produced strongly conflicting evidence.
For the southern part of Italy, in Calabria, Ammerman has made a substantial contribution to the archaeological interpretation of chemically characterised obsidian. By carrying out an extensive survey of the Neolithic sites in the area (Figure 6.8) he provided a dense distribution of sites where obsidian has been found. Obsidian was found at over 200 sites dating mainly to the Neolithic and Copper ages, though some was also found in Mesolithic contexts. Ammerman closely studied the total lithic assemblage from these sites. Using chemical analysis, he has established that Lipari was the primary source with an alternative at Palmarola on the Pontine island. The lithic material from the sites on the west coast of Italy ranged from about 100 to several hundred pieces, and many assemblages contained over 90% obsidian. On the other hand, on the east coast near the town of Crotone, the proportion of obsidian was less than 40%. When compared to the pattern we saw for the Near East, the rate at which obsidian declines away from the source is much greater. Here it falls by 50% in less than 100 km, yet in the Near East the percentage of obsidian remained at 80% within 300 km of the source. Some sites on the east coast of Calabria contained as little at 10% obsidian in their lithic assemblages, and others lying at about the same distance from Lipari contained more than 30%. This pattern is therefore quite different from the monotonic down-the-line movement of obsidian, the suggested model for obsidian exchange in the Near East.
Figure 6.8 The result of a survey of Neolithic sites in Calabria, Italy showing the number of sites on which obsidian has been found.
Possible archaeological interpretations of the pattern (Ammerman 1979; Ammerman and Andrefsky 1982) seen on the east coast of Italy are: (1) some sites take a greater part in the long distance exchange networks of obsidian; or (2) the patterns result from local differences in the availability of chert; or (3) we are seeing the results of multiple local exchange systems which cannot as yet be disentangled. If movement of obsidian had taken place across the land avoiding sea transport, as Hallam et al. (1976) suggest for this area, then we should expect that, as the supply of obsidian moved up the west coast of Calabria, the amount available would decrease, due to loss or deposition among the sites. However, Ammerman’s work has indicated that obsidian is present up the west coast of Calabria in uniform proportions. He suggests a radial distribution of obsidian by sea from Lipari, and that we do not need to invoke a down-the-line mechanism for its distribution. It also seems probable that obsidian reached the east coast of Calabria over land and, as we have noted, there are probably several local factors which might have caused its uneven distribution. The working assumption had been that all sites participated to the same extent and in the same way in the exchange networks. If this had been the case the same pattern of lithic utilisation would be observed at sites which were of equal distance from the source. Ammerman (1979) has shown this not to be the case and that the proportions of different sizes of obsidian artefact sizes and by-products are important (Ammerman and Andrefsky 1982). Greater proportions of obsidian, including more pieces in a reduced form, occurred at sites closer to the coast. The research by Ammerman and Francaviglia shows how important large-scale analytical projects are; both have extensively modified the distribution patterns from individual obsidian sources built up by earlier work in the area.
In the study of Neolithic obsidian from Gaione in the Po plain, northern Italy Ammerman et al. (1990) have suggested that obsidian which reached the site deriving from different sources was possibly treated as having a different value according to its origin. The highly translucent blades of Liparian obsidian may have been valued more as special prestige items (they show few signs of edge damage) rather than as utilitarian objects. The Sardinian obsidian circulating as cores was probably regarded as much more utilitarian.
Another (complementary) approach to obsidian studies has been taken by Tykot (1997a). By focusing on the Sardinian source of obsidian in the Monte Arci volcanic complex he has shown, by using electron microprobe and ICP-MS, that there were nine chemically distinct sources; other workers focusing on archaeological obsidian had already detected three of these (Hallam et al. 1976). Tykot (1992) had already shown that five distinct sources were to be found from the chemical analyses of archaeological artefacts. He identified nine distinct geological sources by first carrying out a comprehensive field survey of potential obsidian sources of which, five had been used for making prehistoric tools. It is interesting to note that the quantity and quality of the obsidian at each site varies (Tykot 1997a: 468–72). Clearly an investigation of any correlation between the quality of the obsidian and the function it was put to is the next step. Tykot (1997b) does, however, underline the variation in intensity of use of each Monte Arci sources and also that the obsidian was used on different sites at different times during the Neolithic in the western Mediterranean. Both Francaviglia (1986) and Tykot (1997a) show that the major Mediterranean sources – Sardinia (A, B and C), Lipari, Pantelleria, Palmarola, Melos and Giali can be distinguished on the basis of their major elemental compositions.
Other recent work has shown that other sources of obsidian were exploited at Szöllöske and Malá Toroña in the Carpathians in southeastern Slovakia. Carpathian obsidian was found to have been used in eleven out of twelve samples from late Neolithic and Bronze Age contexts at Mandalo, Macedonia, tested by Kilikoglou et al. (1996) using neutron activation analysis; the twelfth sample was found to have derived from the ‘anticipated’ source of the Greek island of Melos. One of the important archaeological inferences from this work is that there was interaction between Macedonia, central Europe and the Aegean in the late Neolithic. Two other occurrences of Carpathian obsidian, each single pieces, were found in north-eastern Italy. The first was at a Neolithic cave site, Grotta Tartaruga, but in this case one of the ‘expected’ more local sources of obsidian had been used predominantly, that of Lipari (Williams-Thorpe et al. 1984). The second was at an early Neolithic site of Sammardenchia (Randle et al. 1993). In both cases the obsidian was from a distant source even though there were nearer, local, sources, raising the question of why it was selected for use – perhaps its fracture characteristics or its colour created the demand for it.
The Greek islands also form part of a volcanic zone in which obsidian has been found. The island of Melos, in particular, has been the focus of an analytical study. It has been found that Melos formed the source for most obsidian found in Neolithic and early Bronze Age Aegean contexts (Renfrew et al.1965); Renfrew and Aspinall (1990) detected its use in Upper Palaeolithic and Mesolithic contexts at Franchthi Cave in the Argolid. Renfrew (1972: 442–3) has suggested that a direct access model is appropriate for the exploitation of Melian obsidian. In this model consumers travelled directly to the source to procure their obsidian; no movement of obsidian to intermediate locations or the involvement of middlemen needs to be invoked. Torrence (1986: 105) has suggested that this form of access may have continued into the Bronze Age and included specialist knappers seeking out suitable nodules of obsidian on Melos.
However, in considering the distribution of Greek Neolithic obsidian Perlès (1992: 119) urges that consideration of a full range of materials is necessary in order to interpret fully systems of exchange and organisation. It is clear for example, that ‘prestige goods’ and utilitarian goods are likely to be exchanged/traded using different systems – though at the same time possibly relying on the same initial contacts. For obsidian, a complex picture emerges. During the early and middle Neolithic preformed cores were imported to village sites and blades made there. While prismatic cores were produced at Melos for export during the late and final Neolithic, they were also produced at sites like Saliagos, where a probable workshop for their manufacture has been found. Furthermore, Perlès (ibid.: 128) suggests that the same core may have been used for blade production on more than one site, though this must be difficult to prove absolutely. This does, however, remind us that such core reductions may have occurred in a rather more complex sequence than may at first be assumed. In considering the distribution of stone use in the Greek Neolithic, Perlès (1992) has included andesite, emery and honey-coloured flint, and she also considered the type of site involved in stone procurement. This approach, though painstaking, is clearly one of the ways forward for this area of research, since it also involves the difficult concept of value.
Before leaving discussion of obsidian it is worth briefly pointing out the potentially different archaeological interpretations provided by use of trace element analysis and major element analysis. Figure 6.9 is a map of what we have called obsidian interaction zones in the El Chayel area of Guatemala built up from trace element analysis (Michels 1982): some rather confused overlapping zones are produced. If major elements are used for obsidian characterisation instead, the zones become discrete and geographically definable (Figure 6.10). This pattern can now be used for the interpretation of other features of the archaeological system. These results underline the importance of initially determining a wide range of chemical elements in obsidian analysis. It might be argued that the selection of major elements to provide a ‘true’ characterisation, and the rejection of a trace elemental fingerprint on archaeological grounds removes the objectivity from the process of characterisation and that one is in danger of getting involved in a circular argument (i.e. if the type of characterisation of the obsidian agrees with other archaeological features that particular characterisation is correct). However, it must be realised that successful chemical fingerprinting of obsidian using trace elements in the Mediterranean area does not necessarily mean that such a means of characterisation would also be valid for other areas of the world, given different geochemical environments. The study of obsidian compositions in the North Mexico basin (Findow and Bolognese 1982) provides a further example of useful archaeological results.
Figure 6.9 The distribution of obsidian for the El Chayal source, Guatemala based on trace element analysis. Each sub-system is labelled with a letter; each locality complex affiliation with a number.
Figure 6.10 The distribution of obsidian for the El Chayal source, Guatemala based on major element analysis. Each sub-system is labelled with a letter; each locality complex affiliation with a number as defined in Figure 6.9.
Stonehenge on Salisbury Plain in Wiltshire, southern England is one of the most famous prehistoric sites in Europe. The first written description of Stonehenge was as early as the twelfth century, in the chronicle of Henry of Huntingdon (Chippindale 1983). Today the site consists of an imposing outer circle of giant stone uprights which originally held crosspieces and which were built using silicified sedimentary quartz rocks, called sarsens (silicretes). The use of prepared stone joints, including mortice and tenon and tongue and groove, have apparently been borrowed from woodworking technology. This outer circle enclosed a horseshoe of smaller bluestones of largely basaltic origins (basaltic ‘spotted’ dolerite). Scientific investigation of the stones has experienced a long history, mainly because the site of Stonehenge has an enduring fascination in European prehistoric studies. Stonehenge consists of a series of constructional phases dating to the third and early second millennia BC. The three principal ones (Cleal et al. 1995) consist of:
(I) A typical henge with a bank and ditch; a series of 56 Aubrey holes which held timber posts (Lawson 1997) formed a circle within the bank and ditch. None of the large stones were present yet; the sediments at the bottom of the Aubrey holes contained no debris from stoneworking. This first phase probably dates to the later Neolithic around the end of the fourth and into the third millenia BC (Bayliss et al. 1997)
(II) The second phase saw the silting of the ditch; cremation burials were cut into Aubrey holes. A number of timber structures were built. Atkinson’s excavations (1979) revealed holes which had apparently been only partially dug in the north-western quarter. This phase started in the late third millennium and is therefore of early Bronze Age date (Bayliss et al. 1997).
(III) In the third phase the stone settings were constructed. It would originally have consisted of 85 stones, of which 51 dressed stones remain today. Both the outer circle of pairs and the inner horseshoe were surmounted by stone lintels. At the entrance to the embanked area two further stones were erected; one, ‘the Slaughter Stone’, survives. An oval arrangement of stones was constructed within the horseshoe arrangement; the stone holes which provide the evidence for this suggest 18 uprights were involved. The largest stones were used in the central pair of uprights in the horseshoe, the surviving one standing to 6.7 m in height and with 2.4 m below ground it weighs around 50 tonnes. The Sarsen Circle was in place by 2620–2480 cal BC; the remodelling of the blue-stones into the circle and horseshoe occurred by 2280–2030 cal BC and 2270–1930 cal BC (Bayliss et al. 1997: 56).
A later sub-phase of building involved a further configuration of bluestones, with a horseshoe being constructed within the sarsen horseshoe and a circle between the sarsen horseshoe and the sarsen circle. Later still a circle of (Y and Z) holes was dug outside the sarsen circle and it is clear that they were not used for stone uprights. The remains that can be visited today date to c. 1800 BC.
A lot has been written about the origins of the stones at Stonehenge and there are many arguments about their geological origins. There is no argument about the sources of the sarsens found there. They are widely encountered in southern England and are considered to have been the result of silicification affecting sand in the Lower Tertiary formations (Summerfield and Goudie 1980). Green (1997: 5–6) argues convincingly that sarsens of sufficient size are unlikely to have been deposited by glacial action of Salisbury Plain and that they would have derived from the Marlborough Downs more than 20 km to the north.
The source of the bluestones, a spotted dolerite, has been established with almost complete confidence to the Preseli mountains in south-west Wales. Petrological investigations started as early as 1878 (Maskelyne 1878). The means by which the stones arrived at Stonehenge, however, has provided meat for a long string of research papers and the use of an intriguing range of scientific techniques. The stones were either deposited in the area of Stonehenge on Salisbury Plain by glacial action, although this area today is virtually devoid of such stones, or they were transported from south Wales by prehistoric man. If the latter interpretation is accepted, then clearly the implications for the organisation of labour are enormous; Colin Renfrew has suggested that 30 million man-hours would have been needed to construct the site.
Recent analytical research by Williams-Thorpe and Thorpe (1992) has involved a detailed investigation of the sources of the bluestones. By chemically analysing 15 Stonehenge monoliths and 22 excavated volcanic bluestone fragments from Stonehenge, they have shown that 30 are basaltic dolerite, something that had been known for a while (Thomas 1923), but that, in addition, 5 are rhyolite, 5 are other volcanic materials and 3 are sandstone. They also investigated the possible sources of the bluestones. Eleven of the dolerites that they tested were found to have derived from at least 3 different loci in the Preseli Hills; the spotted and unspotted dolerites derived from an area within a radius of 1.75 km, or possibly 0.5 km (Green 1997: 7). On the other hand the rhyolites came from 4 other sources which were fairly close to the Preseli Hills, 1 about 8 km from the edge of the dolerite distribution, 2 in northern Pembrokeshire and a fourth more generally in Pembrokeshire. Twenty-two volcanic bluestone fragments (9 dolerites and 13 rhyolites) which had been excavated from Stonehenge were also chemically analysed. All except one of these could be sourced to the same area as most monolith samples, in the eastern Preseli Hills. The remaining one, a rhyolite sample, derived from a locus about 2 km away in north Pembrokeshire. This research clearly revealed incontrovertible evidence for the exploitation of stones from the same general area, but that within that area, stone from a number of sites was used.
One particular question therefore needs to be addressed: if the stones were deliberately quarried and transported by man why did they not stick to one source? For the stones that have been tested the majority did derive from a small number of sources which were relatively close to each other. If most of the stones had derived from a range of widely spaced locations the inference would be that the stones were probably transported by glacial action long before Stonehenge was constructed.
However, there is only equivocal evidence that the Anglian glaciation of 400,000 years ago deposited material on or near Salisbury Plain (Green 1997: 8–9). A more likely interpretation is that its route from the Irish Sea, and perhaps from as far as Scotland stopped at the Bristol Channel. Williams-Thorpe and Thorpe suggest, nevertheless, that the bluestones were glacial erratics dropped by an icesheet close to Salisbury Plain. The available evidence for the use of stone to build other British and Continental prehistoric megalithic structures indicates that stones were generally only moved over distances of 5 km or less (Thorpe and Williams-Thorpe 1991). However, Green (1997: 8) raises the point that if the ice sheet had traversed a range of durable rock types why was it that rocks from a relatively limited geographical area were deposited on Salisbury Plain. Green’s own investigation of terrace sediments of rivers draining Salisbury Plain (1973) including 50,000 pebbles, where bluestone erratics would be expected to turn up, has drawn a blank. An early reference by De Luc (1811) to volcanic rocks being strewn across Salisbury Plain has been used as evidence that they were available in the early nineteenth century, so would also have been available in prehistory. However, this reference has now been discounted (Green 1997: 8) because it refers to deposits in the Midlands.
Up to this point, the balance of argument would appear to be in favour of the transport of stones from south Wales by humans, not by glacial action. The results of yet another scientific technique, chlorine-36, can now be brought into play. The principles of this technique are that the chlorine-36 atoms start to decay as soon as the rock face is exposed and that the ratio of chlorine-36 and its stable form can provide an estimate of the period for which the rock was exposed. Early results from the analysis of a fragment of bluestone excavated by Richard Atkinson (Bowen 1995) suggest that it was first exposed towards the end of the last glaciation. This might indicate that it could not have been transported by glacial action – unless, that is, it was transported by earlier glacial action. Another possible (likely) scenario could result from the bluestones being worked at the site of Stonehenge, whether they had been transported there by man or by glacial action, producing the same result. Clearly when chlorine-36 determinations are carried out on a much larger number of samples, a more balanced result will be obtained (Green 1997a: 305).
Moving on to other studies, the examination of British Neolithic stone axes by thin-section petrology has led to the identification of several sources of stone for the axes. While the mineralogical and geological description of many stone axes has been carried out, only a small proportion of the possible natural sources have been similarly characterised. Two such sources and associated distribution zones are axes of group VII, an augite-granophyre outcrop at Graig Lwyd, Gwynedd, Wales (Keiller et al. 1941), and group XX (Shotton 1959), an epidotised ashy grit which probably has its source in the Charnwood Forest area of Leicestershire in England. Though the exact source in Charnwood Forest is unlocated, fragments apparently also occur in Iron Age and early medieval pottery (see sections 4.4.2 and 4.5.2.2) (Cummins and Moore 1973; Cummins 1979; see also Clough and Cummins 1988: maps 7 and 17). Cummins (1979) produced a cumulative axe frequency graph which is the inverse of the decay function used by Renfrew for obsidian. It is evident that the number of axes in the distribution zone starts tailing off markedly after about 300 km from the source. The reasons why the distribution of axes extends so far from the source may be related to their essential use in forest clearance in the Neolithic, or alternatively their function as prestigious items. Careful measurement and statistical and contextual analysis of the axe groupings by Hodder and Lane (1984) indicates that group VII, and group XX axes get sharper with distance from source, and that for group VII in particular the length and cutting edge of the axes changes in a regular way. Both group VII and group XX axes decrease in length with distance from source, the axes being sharpened or broken down and re-used for what may have been utilitarian functions. Also, certain larger axes found close to the sources apparently show fewer signs of use; a possible prestige non-utilitarian function can be suggested for them.
Darvill (1989) published a discussion of the circulation of stone and flint axes. Although his data agree with Hodder and Lane’s (1982) data in showing that the overall length of group VII axes decreases from source and also that the greatest density occurs around the source, he points out (as Ammerman does with obsidian) that unless stone, flint axes and other stone types are studied together as a complete assemblage, the relative contributions of specific sources to the axe population of a given area will be false and the interpretation of national distribution patterns misleading. There may however be difficulties in making direct comparisons between the distributions of different materials since, when broken, flint may be more easily re-used than a stone axe, and flint is rather more difficult to chemically characterise (see Section 6.5 above; see also Matiskainen et al. 1979). Darvill also tentatively identified three main types of distribution patterns: regional, local and waterborne, the last being a distinctively coastal/riverine variation of the regional distribution pattern. He has suggested gift exchange, seasonal migrations and pastoral communication as distribution mechanisms, though exchange between cultivators and herdsmen cannot be ruled out. The use of the term ‘trade’ as an explanatory term for axe distributions by Cummins (1974) seems inappropriate; the actual social and economic contexts in which Neolithic stone axes changed hands are difficult to prove, though the distribution patterns are nonetheless worth attempting to explain.
Work on group VI axes of the Great Langdale/Scafell Pike area of Cumbria, originally designated group VI on the basis of its petrology (Keiller et al. 1941), has advanced through a comprehensive survey (Quartermaine and Claris 1986; Bradley and Edmunds 1988; 1993), with the result that mineralogical and textural definitions of the sites where axes were made are now much better than those made in the original study. The tuff was found to have various new mineralogical characteristics (epidote and scapo-lite) which enabled Claris et al. (1989) to localise the origin of the stone used. This has helped to characterise and locate the sources of the stone in a way which had not been possible before. McVicar (1982) interpreted the gradual decrease in axe length from the source among communities in the Western Isles as indicating that access to the supply was relatively easy and reasonably direct. The distribution showed a marked fall-off towards more distant communities; as in the previous example, the axes were probably being kept longer because damaged specimens were resharpened before being finally discarded. Access to stone may have been reasonably easy close to the source, but not as the distance from the source increased.
Bradley (1984) has interpreted several other characteristics of stone axe distributions. His recognition of clusters of ‘foreign’ axes at the outer limits of axe distributions (e.g. group VII axes from north Wales) may constitute evidence for some form of ‘directional trade’ under political control and increasingly specialised treatment afforded to stone axes (e.g. conversion to mace heads) once they have left their areas of origin may turn out to be a significant factor (Bradley 1990: 303). In general, there is little evidence for a strong ‘political’ direction from axe production at quarries, and although henge monuments can be sited close to major lithic sources, with little apparent evidence for influencing the ‘axe trade’, there is some evidence for specialised deposits of artefacts in and around ceremonial centres (Bradley and Edmonds 1988: 183). As a means of examining the functionality of stone axes Bradley et al. (1992) have investigated the tensile strength of various stone sources used to make axes. For example, they have found that the Langdale and Graig Lwyd axes had a higher tensile strength than porcellanites from North Antrim, Northern Ireland (see below). Sheridan (1986: 20) has pointed out that the distribution of group IX porcellanite axes extended well into Britain, and this is in spite of their lower tensile strength. On the other hand group VI Langdale axes with greater tensile strength rarely occur in Ireland (Jope and Preston 1953). There are several ways of interpreting this result. It is possible that the difference in tensile strength was not sufficiently great to affect the use to which they were put. Alternatively if the ‘use’ was purely ritual then other characteristics such as colour and texture might have been considered to have had greater importance and determined the axe distribution.
Indeed, consideration of the tensile strength of stone axes in Great Langdale, Cumbria and other areas by Bradley et al. (1992) has provided an important means of assessing whether the use of stone with a particular strength played an important part in determining the kind of rock used or whether other explanations need to be put forward. The tensile strengths determined for Great Langdale rocks ranged between 34.2±3.2 MPa and 42.0±2.1 MPa which are higher values than most major rocks types from Britain. They showed that although on a local level in Cumbria (especially in Great Langdale) there was some evidence for the use of stone with greater tensile strength to make axes, the evidence also showed that better quality rock was not worked on a larger scale than the lower quality rock. In general, the situation they discovered was far from simple. Loft Crag, for example, which was easy to reach and which had rock with the second highest tensile strength out of eight quarry locations tested (Bradley et al.: 1992, Table 1), was only used to a limited extent. Sites where rocks with some of the lowest tensile strength had been extracted and worked covered about the same area as sites where rocks with the highest tensile strength had been extracted and worked. One suggestion is that production of axes in different locations in Great Langdale may have been controlled by particular individuals or communities (Bradley and Suthren 1990) and that access to some outcrops was restricted for some reason. It was clear from this research that utilitarian considerations were just one of a range of reasons why particular stone sources were exploited and that they certainly did not determine the form of Neolithic ‘axe trade’. The work with northern Irish porcellanite also serves to underline this finding in a powerful way.
A major source of stone used for the manufacture of Neolithic axes was porcellanite, a hard contact metamorphic rock which formed from the contrasting primary lithologies of highly weathered basalt and Jurassic clay. Knowles (1903; 1906) and Mallory (1990) have clearly recognised the evidence for the working of porcellanite into axes at Tievebulliagh near Cushendall and at Brockley on Rathlin Island, Co. Antrim, Northern Ireland. Keiller et al. (1941) allocated this to group IX. At Tievebulliagh, Mallory (1990) showed that scree materials appeared to have been used for making axes, whereas for the second source, at Brockley, fire setting was almost certainly used in the mining operations.
Meighan et al. (1993) have determined trace element levels in porcellanites from North Antrim using wavelength-dispersive X-ray fluorescence analysis and determined their mineralogical compositions using X-ray diffraction. In addition they used strontium isotope determinations to attempt to determine provenances. The trace element analyses showed that they could distinguish between porcellanites from first Portrush and second Tievebulliagh, and Brockley on Rathlin Island. Their formation by contact metamorphism of Jurassic clay and Tertiary basalt respectively led to these geochemical differences. Although formed under similar conditions, Tievebulliagh and Brockley are 26.5 km apart and on opposite sides of a major Tertiary ‘divide’, the Tow Valley Fault, so for them to have exactly the same major/trace element and isotopic signatures would be exceptional. Meighan et al. (1993: 27, Figure 2a) have demonstrated, although with rather small numbers, that a plot of the relative levels of strontium and calcium does discriminate not only (as expected) between Portrush and Tievebulliagh, but also between Brockley and the other two. Strontium isotope analysis also distinguishes between the three sources (ibid.: Figure 2b).
In comparison with the Great Langdale, Cumbria, source many more examples of porcellanite axes have been found: some 6,400 compared to 1,612 (Clough 1988: 4). Martyn Jope (1952) carried out early work on the occurrence of porcellanite, with the number of examples being increased by Sheridan (1986) and by Cooney et al. (1994). Given that porcellanites were used so extensively in the Neolithic, if the characterisation by Meighan et al. (1993) holds up for samples which derive from securely dated archaeolgical contexts, then it will be possible to examine the form of distribution patterns associated with each of the three porcellanite sources: whether they display boundedness, whether there is a chronological distinction in the use of the three sources and whether they occur on particular site types.
Over and above the work with stone axes the results of these research projects are a good example of how easy it might be to make false assumptions about the use of raw materials in the past based solely on their material properties and to ignore aspects of human behaviour which determine their use and which are unconnected to positivism. Ethnographic studies of stone axe production make this clear (Burton 1987), even though some workers ignore completely the results of research which shows that approaches based purely on features found in capitalist societies (Steinberg and Pletka 1997) are inappropriate.
Allan, R.O., Luckenbach, A.H. and Holland, C.G. (1975) ‘The application of instrumental neutron activation analysis to a study of prehistoric steatite artefacts and source materials’, Archaeometry 17, 1: 69–83.
Ammerman, A.J. (1979) ‘A study of obsidian exchange networks in Calabria, World Archaeology 11: 95–110.
Ammerman, A.J. and Andrefsky, W. (1982) ‘Reduction sequences and the exchange of obsidian in Neolithic Calabria’, in J.E. Ericson and T. Earle (eds), Contexts for Prehistoric Exchange, New York: Academic Press, pp. 149–172.
Ammerman, A.J., Cesana, A., Polglase, C. and Terrani, M. (1990) ‘Neutron activation analysis of obsidian from two Neolithic sites in Italy’, Journal of Archaeological Science 17: 209–220.
Aspinall, A. and Feather, S.W. (1972) ‘Neutron activation analysis of prehistoric flint mine products’, Archaeometry 14, 1: 41–54.
Aspinall, A., Feather, S.W. and Phillips, A.P. (1975) ‘Preliminary analyses of French flint samples’, in F.H.C. Engelen (ed.), Proceedings of the Second International Symposium on Flint, Maastricht, Netherlands, 8–11 May 1975, Nederlandse Geol. Vereniging, Maastricht (Staringia no. 3), pp. 42–46.
Atkinson, R.J.C. (1979) Stonehenge: Archaeology and Interpretation, Harmondsworth: Penguin.
Bayliss, A., Bronk Ramsey, C. and McCormac, F.G. (1997) ‘Dating Stonehenge’, in B. Cunliffe and C. Renfrew (eds) Science and Stonehenge, Proceedings of the British Academy 92 Qxford: Oxford University Press, pp. 39–60.
Blackman, M.J. (1984) ‘Provenance studies of Middle Eastern obsidian from sites in Highland Iran’, in J.B. Lambert (ed.), Archaeological Chemistry II, American Ceramic Society, Advances in Chemistry Series no. 205, Columbus, Ohio: American Ceramic Society, pp. 19–50.
Bowen, D.Q. (1995) ‘Late Cenozoic Wales and south-west England’, Proceedings of the Ussher Society 8: 208–213.
Bowman, H.R., Asaro, F. and Perlman, I. (1973) ‘Composition variations in obsidian sources and the archaeological implications’, Archaeometry 15: 123–127.
Bradley, R. (1984) The Social Foundations of Prehistoric Britain, London: Longman.
Bradley, R. (1990) ‘Perforated stone axe-heads in the British Neolithic: their distribution and significance’, Oxford Journal of Archaeology 9, 3: 299–304.
Bradley, R. and Edmonds, M. (1988) ‘Fieldwork at Great Langdale, Cumbria 1985–1987: preliminary report’, The Antiquaries Journal 68: 181–209.
Bradley, R. and Edmonds, M. (1993) Interpreting the Axe Trade. Production and Exchange in Neolithic Britain, Cambridge: Cambridge University Press.
Bradley, R., Edmonds, M., Meredith, P., Smith, J. and Edmonds, M. (1992) ‘Rock physics and the Neolithic axe trade in Britain’, Archaeometry 34, 2: 223–233.
Bradley, R.,Edmonds, M., Meredith, P., Smith, J. and Edmonds, M., Suthren, R. (1990) ‘Pétrographie analysis of hammerstones from the Neolithic quarries at Great Langdale’, Proceedings of the Prehistoric Society 56: 117–122.
Burton, J. (1987) ‘Exchange pathways at a stone axe factory in Papua New Guinea’, in The Human Uses of Flint and Chert, in G. de G. Sieveking and M.H. Newcomer (eds), Proceedings of the Fourth International Flint Symposium, Brighton Polytechnic, 10–15 April 1983, Cambridge: Cambridge University Press, pp. 183–191.
Cackler, P.R., Glascock, M.D., Neff, H., Iceland, H., Pyburn, K.A., Hudler, D., Hester, T.R. and Chiarulli, B.M. (1999) ‘Chipped stone artefacts, source areas, and provenance studies of the Northern Belize chert-bearing zone’, Journal of Archaeological Science 26, 4: 389–397.
Cann, J.R. and Renfrew, C. (1964) ‘The characterization of obsidian and its application to the Mediterranean region’, Proceedings of the Prehistoric Society 30: 111–133.
Chippindale, C.C. (1983) Stonehenge Complete, London: Thames and Hudson.
Claris, P. and Quartermaine, J. (1989) ‘The Neolithic quarries and axe factory sites of Great Langdale and Scafell Pike: a new field survey’, Proceedings of the Prehistoric Society 55: 1–26.
Cleal, M.J., Walker, K.E. and Montague, R. (1995) Stonehenge in its Landscape, London: English Heritage.
Clough, T.H. McK. (1988) ‘Introduction to the regional reports: prehistoric stone implements in the British Isles’, in T.H. McK. Clough and W.A. Cummins (eds), Stone Axe Studies, The petrology of prehistoric stone implements from the British Isles vol. II, Council for British Archaeology Research Report no. 67, London: Council for British Archaeology, pp. 1–11.
Clough, T.H. McK. and Cummins, W.A. (eds) (1988) Stone Axe Studies, The petrology of prehistoric stone implements from the British Isles vol. II, Council for British Archaeology Research Report no. 67, London: Council for British Archaeology.
Clough, T.H. McK. and Cummins, W.A. (1988a) ‘The petrological identification of stone implements from the east Midlands: third report’, in T.H. McK. Clough and W.A. Cummins (eds), Stone Axe Studies, The petrology of prehistoric stone implements from the British Isles vol. II, Council for British Archaeology Research Report no. 67, London: Council for British Archaeology, pp. 41–44.
Consigny, S. (1992) Méthode géochimique et origine des silex archéologiques, Memoire de Maîtrise de Préhistoire de l’Université de Paris I.
Consigny, S. and Walter, P. (1997) ‘Flint origin and ion beam analysis: archaeological results in the Paris basin, France’, in R. Schild and Z. Sulgostowska (eds), Man and Flint, Proceedings of the Vllth International Flint Symposium, September 1995, Institute of Archaeology and Ethnology, Warszawa:Polish Academy of Sciences, pp. 337–342.
Cooney, G., Mandal, S. and O’Carroll, F. (1994) ‘Review of the Work of the Irish Stone Axe Project 1991–1993’, unpublished report, Dublin.
Cowell, M.R. (1979) ‘The archaeological and geochemical implications of trace element distributions in some English, Dutch and Belgian flints’, in F.H.C. Engelen (ed.), Proceedings of the Third International Symposium on Flint, Maastricht, Netherlands, Nederlandse Geol. Vereniging, Maastricht (Staringia no. 6), pp. 81–84.
Cowell, M.R., Ferguson, J. and Hughes, M.J. (1980) ‘Geochemical variations in East Anglian flint with particular reference to Grimes Graves flint mines’, in E.A. Slater and J.O. Tate (eds), Proceedings of the 16th International Symposium on Archaeometry and Archaeological Prospection, Edinburgh 1976, Edinburgh: National Museums of Antiquaries of Scotland, pp. 80–89.
Craddock, P.T., Cowell, M.R., Leese, M.N. and Hughes, M.J. (1983) ‘The trace element composition of polished flint axes as an indicator of source’, Archaeometry 25, 2: 135–164.
Cummins, W.A. (1974) ‘The Neolithic stone axe trade in Britain’, Antiquity 48: 201–205.
Cummins, W.A. (1979) ‘Neolithic stone axes: distribution and trade in England and Wales’, in T.H.McK. Clough and W.A. Cummins (eds), Stone Axe Studies, Council for British Archaeology Research Report 23, London: Council for British Archaeology, pp. 5–12.
Cummins, W.A. and Moore, C.N. (1973) ‘Petrological identification of stone implements from Lincolnshire, Nottinghamshire and Rutland’, Proceedings of the Prehistoric Society 39: 219–255.
Darvill, T. (1989) ‘The circulation of neolithic stone and flint axes: a case study from Wales and the mid-west of England’, Proceedings of the Prehistoric Society 55: 27–44.
De Bruin, M., Korthoven, P.J.M., Bakels, C.C. and Groen, F.C.A. (1972) ‘The use of non-destructive activation analysis and pattern recognition in the study of flint artefacts’, Archaeometry 14, 1: 55–63.
De Luc, J.A. (1811) Geological Travels. Volume III. Travels in England, London: F.C. and J. Rivington.
Doran, J.E. and Hodson, F.R. (1975) Mathematics and Computers in Archaeology, Edinburgh: Edinburgh University Press.
Ericson, J.E. (1988) ‘Obsidian hydration rate development’, in E.V. Sayre, P. Vandiver, J. Druzik and C. Stevenson (eds), Materials Issues in Art and Archaeology, Proceedings of the Materials Research Society, vol. 123, Reno, Nevada 1988, Materials Research Society, Pittsburgh, pp. 215–224.
Findow, F.J. and Bolognese, M. (1982) ‘Regional modeling of obsidian procurement in the American Southwest’, in J.E. Ericson and T.K. Earle (eds), Contexts for Prehistoric Exchange, New York: Academic Press, Chapter 3.
Francaviglia, V. (1988) ‘Ancient obsidian sources on Pantelleria (Italy)’, Journal of Archaeological Science 15: 109–122.
Gordus, A.A., Wright, G.A. and Griffin, J.B. (1968) ‘Obsidian sources characterized by Neutron Activation Analysis’, Science 161: 382–384.
Gowlett, J. and Henderson, J. (forthcoming) ‘Scientific analysis and sourcing of obsidian from East Africa’.
Gratuze, B. (1999) ‘Obsidian characterization by laser ablation ICP-MS and its application to prehistoric trade in the Mediterranean and the Near East: sources and distribution of obsidian within the Aegean and Anatolia’, in J. Henderson, H. Neff and T. Rehren (eds), Proceedings of the International Symposium on Archaeometry, University of Illinois at Urbana-Champaign (UIUC), Urbana, Illinois, 20–4 May 1996, Journal of Archaeological Science 26,8: 869–882.
Green, C.P. (1973) ‘Pleistocene gravels and the Stonehenge problem’, Nature 243: 214–216.
Green, C.P. (1997) ‘Stonehenge: geology and prehistory’, Proceedings of the Geologists7 Association 108, 1: 1–10.
Green, C.P. (1997a) ‘The provenance of rocks used in the construction of Stonehenge’, in B. Cunliffe and C. Renfrew (eds) Science and Stonehenge, Proceedings of the British Academy 92, Oxford: Oxford University Press, pp. 257–270.
Hallam, B.R., Warren, S.E. and Renfrew, A.C. (1976) ‘Obsidian in the western Mediterranean: characterization by neutron activation analysis and optical emission spectroscopy’, Proceedings of the Prehistoric Society 42: 85–110.
Heyworth, M.P., Hunter, J.R., Warren, S.E. and Walsh, J.N. (1988) ‘The analysis of archaeological materials using inductively-coupled plasma spectrometry’, in E.A. Slater and J.O. Tate (eds), Science and Archaeology, Glasgow 1987 British Archaeological Reports series 196, Oxford: British Archaeological Reports, pp. 27–40.
Hoard, R.J., Bozell, J.R., Holen, S.R., Glascock, M.D., Neff, H. and Elam, J.M. (1993) ‘Source determination of White river group silicates from two archaeological sites in the Great Plains’, American Antiquity 58, 4: 698–710.
Hoard, R.J., Bozell, J.R, Holen, S.R., Glascock, M.D., Neff, H. and Elam, M.J. (1992) ‘Neutron activation analysis of stone from the chadron formation and a Clovis site on the Great Plains’, Journal of Archaeological Science 19: 655–665.
Hodder, I. and Lane, P. (1982) ‘A contextual examination of Neolithic stone axe distribution in Britain’, in J.E. Ericson and T.K. Earle (eds), Contexts for Prehistoric Exchange, New York: Academic Press, Chapter 10, pp. 213–235.
Holmes, L.L. and Harbottle G. (1994) ‘Compositional characterization of French limestone: a new tool for art historians’, Archaeometry 36, 1: 25–39.
Hudler, D., Iceland, H., Cackler, P.R. and Glascock, M.D. (1996) ‘A sourcing study of preceramic constricted unifaces in northern Belize’, a paper presented at the 1st Annual Meeting of the Society of American Archaeology, 10–14 April 1996, New Orleans.
Hughes, R.E. (1994) ‘Intrasource chemical variability of artefact-quality obsidians from the Casa Diablo area, California’, Journal of Archaeological Science 21: 263–271.
Jack, R.N. (1976) ‘Prehistoric obsidian in California I. Geochemical aspects’, in Taylor, R.E (ed.), Advances in Obsidian Glass Studies, Park Ridge, New Jersey: Noyes Press, pp. 183–217.
Jope, E.M. (1952) ‘Porcellanite axes from factories in north-east Ireland: Tievebulliagh and Rathlin. Part I. Archaeological Survey’, Ulster Journal of Archaeology 15: 31–55.
Jope, E.M. (1989) ‘Preface’, in J. Henderson (ed.), Scientific Analysis in Archaeology and its interpretation, Oxford University Committee on Archaeology, Monograph no. 19 and UCLA Institute of Archaeology, Archaeological Research Tools 5, Oxford: Oxford University Committee on Archaeology, pp. xi–xv.
Jope, E.M. and Preston, J. (1953) ‘An axe of stone from Great Langdale, Lake District, found in County Antrim’, Ulster Journal of Archaeology 49: 19–32.
Keiller, A., Piggott, S. and Wallis, F.S. (1941) ‘First report of the Sub-Committee of the South-Western Group of Museums and Art Galleries on the petrological identification of stone axes’, Proceedings of the Prehistoric Society 7: 50–72.
Kilikoglou, V., Bassiakos, Y., Grimanis, A.P. and Souvatzis, K. (1996) ‘Carpathian obsidian in Macedonia, Greece’, Journal of Archaeological Science 23: 343–349.
Knowles, W.J. (1903) ‘Stone axe factories near Cushendall’, Journal of the Royal Anthropological Institute 33: 360–366.
Knowles, W.J. (1906) ‘Stone axe factories near Cushendall’, Journal of the Royal Society of Antiquaries of Ireland 36: 383–394.
Kohl, P.L., Harbottle, G. and Sayre, E.V. (1979) ‘Physical and chemical analyses of soft stone vessels from Southwest Asia’, Archaeometry 21, 1: 131–159.
Lawson, A.J. (1997) ‘The Structural History of Stonehenge’, in B. Cunliffe and C. Renfrew (eds) Science and Stonehenge, Proceedings of the British Academy 92, Oxford: Oxford University Press, pp. 15–38.
Luetke (1978) ‘Chert sources and trace-element analysis’, American Antiquity 43, 3: 413–423.
Luedtke, B.E. (1979) ‘The identification of sources of chert artefacts’, American Antiquity 44, 4: 744–757.
Mallory, J.P. (1990) ‘Trial excavations at Tievebulliagh, Co. Antrim’, Ulster Journal ofArchaeology 53: 15–28.
Maskelyne, N.S. (1878) ‘Stonehenge: the petrology of its stones’, Wiltshire Archaeological and Natural History Magazine 17: 147–160.
Matiskainen, H., Vuorinen, A. and Burman, O. (1979) ‘The provenance of prehistoric flint in Finland’, in Y. Maniatis (ed.), Archaeometry, Proceedings of the 20th International Symposium, Athens, Greece, Oxford: Elsevier.
McDonald, R., Smith, R.L. and Thomas, J.E. (1992) Chemistry of the Subalkalic Silicic Obsidians, US Geological Survey Professional Paper 1523, Washington: The US Geological Survey.
McVicar, J.B. (1982) ‘The spatial analysis of axe size and the Scottish axe distribution’, Archaeological Reviews from Cambridge 1, 2: 30–45.
Meighan, I.C., Jamison, D.D., Logue, P.J.C., Mallory, J.P., Simpson, D.D.A., Rogers, G., Mandal, S. and Cooney, G. (1993) ‘Trace element and isotopic provenancing of north Antrim porcellanites: Portrush – Tievebulliagh – Brockley (Rathlin Island)’, Ulster Journal of Archaeology 56: 25–30.
Merrick, H.V. and Brown, F.H. (1984) ‘Rapid chemical characterization of obsidian artefacts by electron micro-probe analysis’, Archaeometry 26: 230–236.
Michels, J.W. (1982) ‘Bulk element composition versus trace element composition in the reconstruction of an obsidian source system’, Journal of Archaeological Science 9: 113–123.
Michels, J.W., Tsong, I.S.T. and Nelson, C.M. (1983) ‘Obsidian dating and East African Archaeology’, Science 219: 361–366.
Moorey, P.RS. (1994) Ancient Mesopotamian Materials and Industries, Oxford: Clarendon Press.
Peacock, D.P.S. and Maxfield, V.A. (1997) Survey and Excavation at Mons Claudianus, volume 1 Topography and Quarries, Le Claire: Institut Français d’archéologie orientale.
Perlès, C. (1987) Les Industries Lithiques Taillées de Franchthi (Argolide, Grèce), Tome I, Présentation générale et industries paléolithiques, Excavations at Franchthi, Greece, Fascicule 3, Bloomington and Indianapolis: Indiana University Press.
Perlés, C.(1992) ‘Systems of exchange and organisation of production in Neolithic Greece’, Journal of Mediterranean Archaeology 5: 115–164.
Pigeot, N. (1987) Magdaléniens d’Etiolles, économie de débitage et organisation sociale unité d’habitation, 25e supplément à Gallia-Préhistoire.
Pollard, A.M. and Heron, C. (1996) Archaeological Chemistry, London: Royal Society of Chemistry.
Quartermaine, J. and Claris, P. (1986) ‘The Langdale axe factories’, Current Archaeology 102: 212–213.
Randle, K., Barfield, L.H. and Bagolini, B. (1993) ‘Recent Italian obsidian analyses’, Journal of Archaeological Science 20: 503–509.
Renfrew, C. (1972) The Emergence of Civilisation, London: Methuen.
Renfrew, C. (1975) ‘Trade and action at a distance: questions of integration and communication, in J. Sabloff and C.C. Lamberg-Karlovsky (eds), Ancient Civilization and Trade, Albuquerque: University of New Mexico Press, pp. 3–59.
Renfrew, C. (1977) ‘Alternative models for exchange and spatial distribution’, in T. Earle and J.E. Ericson (eds), Exchange Systems in Prehistory, New York: Academic Press, pp. 71–90.
Renfrew, C. and Aspinall, A. (1990) ‘Aegean obsidian and Franchthi Cave’, in C. Pèrles (ed.), Les Industries Lithiques de Franchthi (Argolide, Grèce), Tome II: Les Industries du Mésolithique et du Néolithique Initial, Fasicule 5, Bloomington and Indianapolis: Indiana University Press, pp. 258–270.
Renfrew, C., Cann, J.R. and Dixon, J.E. (1965) ‘Obsidian in the Aegean’, Annals of the British School of Athens 60: 225–247.
Renfrew, C., Dixon, J.E. and Cann, J.R. (1966) ‘Obsidian and early cultural context in the Near East’, Proceedings of the Prehistoric Society 32: 30–72.
Renfrew, C., Dixon, J.E. and Cann, J.R. (1968) ‘Further analysis of Near Eastern obsidians’, Proceedings of the Prehistoric Society 30: 319–331.
Sheridan, J.A. (1986) ‘Porcellanite artifacts: a new survey’, Ulster Journal of Archaeology 49: 19–32.
Shotton, F.W. (1959) ‘New petrological groups based on axes from the West Midlands’, Proceedings of the Prehistoric Society 25: 135–143.
Sieveking, G. de G., Craddock, P.T., Hughes, M.J., Bush, P. and Ferguson, J. (1970) ‘Characterization of flint mine products’, Nature 228: 251–254.
Steinberg, J.M. and Pletka, B.J. (1997) ‘The value of flint in Thy, Denmark’, in R. Schild and Z. Sulgostowska (eds), Man and Flint, Proceedings of the Vllth International Flint Symposium, September 1995, Warszawa: Institute of Archaeology and Ethnology, Polish Academy of Sciences, pp. 337–342.
Summerfield, M.A. and Goudie, A.S. (1980) ‘The sarsens of southern England: their palaeoenvironmental interpretation with reference to other silicretes’, in D.K.C. Jones (ed.), The Shaping of Southern England, London: Academic Press.
Thomas, H.H. (1923) ‘The source of the stones of Stonehenge’, Antiquaries Journal 3: 239–260.
Thompson, M., Bush, P.R. and Ferguson, J. (1986) ‘The analysis of flint by inductively coupled plasma atomic emission spectroscopy, as a method of source determination’, in G. de G. Sieveking and M.B. Hart (eds), The Scientific Study of Flint and Chert, Proceedings of the Fourth International Flint Symposium held at Brighton Polytechnic 10–15 April 1983, Cambridge: Cambridge University Press, pp. 243–247.
Thorpe, R.S. and Williams-Thorpe, O.W. (1991) ‘The myth of long-distance megalithic transport’, Antiquity 65: 64–73.
Torrence, R. (1986) Production and Exchange of Stone Tools, Cambridge: Cambridge University Press.
Tykot, R.H. (1992) ‘The sources and distribution of Sardinian obsidian’, in R.H. Tykot and T.K. Andrews (eds), Sardinia in the Mediterranean: A Footprint in the Sea. Studies in Sardinian Archaeology Presented to Miriam S. Balmuth, Sheffield: Sheffield Academic Press, pp. 57–70.
Tykot, R.H. (1997a) ‘Characterization of the Monte Arci (Sardinia) obsidian sources’, Journal of Archaeological Science 24: 467–479.
Tykot, R.H. (1997b) ‘Mediterranean islands and multiple flows: the sources and exploitation of Sardinian obsidian’, in M.S. Shackley (ed.), Method and Theory in Archaeological Volcanic Glass Studies, Advances in Archaeological and Museum Science Series, New York: Plenum Press.
Weisgerber, G. (1987) ‘The ancient chert mines at Wadi el-Sheikh (Egypt)’, in G. de G. Sieveking and M.H. Newcomer (eds), The Human Uses of Flint and Chert, Proceedings of the Fourth International Flint Symposium held at Brighton Polytechnic, 10–15 April 1983, Cambridge: Cambridge University Press, pp. 165–172.
Williams-Thorpe O.W., Warren, S.E. and Courtin, J. (1984) ‘The distribution and sources of archaeological obsidian from southern France’ Journal of Archaeological Science 11: 135–146.
Williams-Thorpe O.W., Warren, S.E. and Thorpe, R. (1992) ‘Geochemistry, sources and transport of the Stonehenge Bluestones’, in A.M. Pollard (ed.), New Developments in Archaeological Science, Proceedings of the British Academy 77, Oxford: Oxford University Press, pp. 133–161.
Williams-Thorpe O.W., Warren, S.E. and Nandris, J.G. (1984) ‘The distribution and provenance of archaeological obsidian in central and eastern Europe’, Journal of Archaeological Science 11: 183–212.
Wright, G.A. (1969) Obsidian Analyses and Prehistoric Near Eastern Trade: 7500 to 3500 B.C., Anthropological papers of the Museum of Anthropology, University of Michigan no. 37, Ann Arbor: University of Michigan.
Yellin, J. (1995) ‘Trace element characteristics of Central Anatolian obsidian flows and their relevance to prehistory’, Israel Journal of Chemistry 35: 175–190.