Internet Archaeol. 26. Healey and Campbell. The Obsidian from Domuztepe

The use of obsidian at Domuztepe is interesting. It is by no means an essential raw material for tool manufacture – flint is plentiful in the surrounding area, as revealed during an archaeological survey of the plain (E. Carter pers. comm.), and in the limestone hills to the south (Garrard et al. 2004, 147). Nevertheless, obsidian is found in most contexts of the site and was used not only for tool manufacture but also for jewellery, mirrors and vessels. Although other material was also obtained from a long distance, for example sea shells, obsidian is the only one that is preserved in large quantities and was used for both mundane and more personal items. The initial geochemical analysis (of 54 pieces) undertaken by J-L Poidevin (CNRS, Magmas et Volcans, Institut de Géologie, Clermont-Ferrand) indicates that it originated from eight different sources, representing most of the obsidian source areas in use at this time (Fig. 1), exemplifying the widespread contacts of this cultural milieu. What this does not tell us is the relative importance of each of these sources in the assemblage, nor the circumstances of acquisition of the obsidian, whether by direct access or through exchange, or whether redistribution centres were involved (e.g. Renfrew 1975). Nor does it tell us who participated in the processes of procurement, which may have been embedded in other activities; possibilities include but are not limited to herders, pastoral nomads (Crawford 1978; Chataigner 1998, 308), traders, specialists or simply individuals with social ties to external groups (Copeland 1995, 5-6; Astruc et al. 2007, 335-40). We do not always know whether obsidian reached Domuztepe as finished objects, part prepared cores or raw blocks, nor how it was worked and used in the different contexts on the site. These are key issues if we are to begin to consider some of the key archaeological questions highlighted above.

The logistics of characterising such a large assemblage are huge. There are few facilities in Turkey for analysis of this quantity of objects and, even if the cost was not prohibitive, we would be unlikely to obtain permission to export such a large number for provenance analysis elsewhere. Furthermore some of the objects are in the Kahramanmaraş Museum and, quite rightly, cannot be exported or destructively analysed. We therefore needed to devise a method of investigating the obsidian in a comprehensive, non-destructive and time- and cost-effective manner in a field laboratory, and one that could be replicated in future seasons.

Our initial approach was to group the obsidian according to its visual characteristics and using a pilot study of some 1000 artefacts to compare these groups to the characteristics of the obsidians that had already been analysed and attributed to source.

Although visual characteristics are not a completely reliable indicator of origin – obsidian can vary in colour within a single source/flow and we do not yet understand the full extent of this variation (Flude pers. comm.) – it has long been clear that certain colours predominate at some sources (Balkan-Atli 1994). Indeed, Renfrew (Renfrew et al. 1966; 1968) recorded the colour of the objects which he analysed and this also correlates with his geochemical groupings. He was thus able to establish that green obsidians were peralkaline and could therefore only come from two sources (Renfrew et al. 1966; Renfrew 1977; see also Cauvin 2000, 169). Others have also found a correlation between colour and source – for example Maeda for the obsidians found in the Rouj Basin sites (Maeda 2003; pers. comm.) and Cauvin for other sites (Cauvin et al. 1986), although for a more cautious view see Cauvin (2000, 165-9). In the Mediterranean, colour has also been successfully used as a discriminator where physical characteristics and geochemical analyses go hand in hand (Tykot and Ammerman 1997; Lugliè et al. 2007). We also note similar studies in Latin America (Braswell et al. 2000).

The main characteristics on which the groupings were based at Domuztepe included colour in transmitted light (in order to standardise our attribution we used a consistent tungsten daylight bulb throughout), translucency, inclusions, flow-banding (cf Skinner 1999). This enabled us to establish a number of broad groups, the replicability of which was established and refined by a number of different observers. We are also in consultation with a colour scientist to improve our objectivity in describing and recording colour.

Figure 2: Selection of the different colours of obsidian from Domuztepe. Images are at different scales

The main colour groups are: green, browns (including translucent light brown and a darker, more treacly brown, which we have subsequently separated), opaque black, black with a red edge, mahogany or reddy-brown and various translucent greys (Fig. 2).

Thirty-six samples from each of the main colour groups were analysed geochemically by ICP-MS to confirm the utility of the groupings and the utility of this approach. Subsequently, using these results as a basis, in a blind-test, a further 18 samples were allocated by two independent observers to sources prior to geochemical analysis. The results, summarised in Table 1, confirmed that there is a significant correlation to the sources.

All the green obsidian belongs to the Bingöl/Nemrut Dağ sources (difficult to distinguish even using geochemistry) (Poidevin 1998; Chataigner 1994; 1998) and 88% of artefacts assigned to this source by geochemistry were also successfully assigned by colour alone. This is not a surprise as this particular correlation is well established (see above). Other correlations are even more encouraging. Overall, nearly 90% of two colour groups (brown and black with a red edge) were correctly assigned to the calcalkaline Bingöl sources and 78% of artefacts assigned by geochemistry could also be assigned by colour alone. Similarly, the two translucent grey colour groups were 83% accurate in specifying the Göllüdağ-east geochemical source and, overall, 64% of the artefacts in this group were also successfully categorised by colour. Overall the study suggests that we ought, even without refining the methodology further, to be able to provide some indication of provenance for c. 75% of the obsidian at Domuztepe with a level of accuracy of c. 85%. Although this is far from providing perfect provenance data, it is a level of uncertainty that can be accommodated in subsequent analysis of changing obsidian use. It tells us, for example, that the bulk of the obsidians came from sources far to the east of the site, but that Cappadocian sources provided a significant minority of the obsidian (Fig. 3).

Figure 3: Chart showing proportion of different colours of obsidian from Domuztepe and their potential source attributions

Furthermore, we have been able to demonstrate that certain morpho-technological characteristics are characteristic of some of the groups which we identified on the basis of colour. We tested technological attributes (striking platform and bulb type, directionality of flaking and dimensions) in each colour group using standard non-parametric tests (Mann-Whitney U-test and chi-squared tests). We also recorded morpho-typological characteristics (full details in Healey 2000, 132-9 and appendix 2). The main significant differences are as follows.

Figure 4: Distribution of different coloured obsidian at Domuztepe by phase

In addition, it is clear that some of the obsidians are being reduced on the site probably from pre-formed cores, but that others may have come in as finished objects.

Figure 5: Jewellery, mirror and vessels including unfinished beads and bead blanks made of obsidian. 1-2: beads (no. 1 probably broken in manufacture). 3: bead blanks. 4: incised pendant/seal. 5-6 links. 7: obverse and reverse of a mirror. 8-9: vessel fragments

Cores are present only in the dominant colour groups (greys, browns and greens). The core-to-blade ratio suggests that the green obsidian was more carefully conserved than the grey. By contrast the rarer red obsidians only occur as blades and a few flakes, but there are four vessel fragments of this obsidian colour. There is evidence that at least some of the jewellery and other ground and polished items were made at the site. Among the pendants and beads of obsidian there are some unfinished forms, such as a bead that broke while being perforated (Fig. 5, 1) and in 2005 a bead-making workshop was uncovered that included obsidian bead blanks and associated debris (Fig. 5, 3) of translucent brown and grey colour, among other materials (Campbell 2005). Similarly, the presence among the fine, thin-walled vessels of two thicker, less well-finished forms may suggest that they may have been locally manufactured (Fig. 5, 8-9).

3. The Obsidian from Domuztepe