The use of multi-criteria GIS to analyse the long flint blades of Sardinia, Italy

Valerio Pinna & Barbara Melosu

Figure 1
Figure 1. Long flint blades. 1: from the funerary site of Pranu Mutteddu, Goni (CA); 2–5: from the settlement site of Cuccuru Is Arrius, Cabras (OR).
Click to enlarge.
Introduction

Among applications of GIS analysis to archaeological research questions, case studies concerned with defined phenomena in discrete areas, in particular islands (Gaffney & Stančič 1991; Bevan & Conolly 2004) form an ideal subject. Located in the middle of the Mediterranean, Sardinia is thus a perfect place for testing such applications; the object of our analysis is the distribution in this region of long blades produced in a highly specialised environment at the end of the Neolithic (4000–3300 cal BC) (Figure 1). These long blades, obtained mainly through pressure flaking, are made on very high quality flint from deposits in the area of Perfugas (Sassari), where a specialist workshop producing such artefacts has been identified (Costa & Pelegrin 2004).

In our case study, GIS was used to provide an analytical tool additional to the techno-morphological study of the entire corpus of 258 artefacts. This was achieved by incorporating the information obtained from the examination of the artefacts themselves into a geodatabase. The data derived from the artefacts were linked with the data pertaining to the geo-environment of Sardinia and interrogated in terms of spatial relationships.

The construction of an environmental model in which the archaeological data can be cast required some methodological choices: the current state of palaeo-ecological and palaeo-environmental research in Sardinia makes it impossible to suggest a reconstruction of its paleo-landscape in the Neolithic. Furthermore, even the lithological data, normally considered a stable environmental element, are in our case highly diverse, stemming in part from quaternary deposits of alluvial or colluvial origin.

We have therefore used the geomorphological information currently available (Van Hove 2004; Pecere 2006), restricting the modelling of the landscape to the only sufficiently robust data, i.e. that pertaining to the relief. A reference Digital Elevation Model (DEM) was thus constructed using current contour data at a resolution of 10m (Figure 2). Such a DEM, together with the slope map (Pecere 2006) derived from it, forms the basis of all successive interrogations of the data.

Figure 2
Figure 2. Distribution of sites with long flint blades.
Click to enlarge.

There are limitations to analyses that can be carried out in such a virtual environment. Only some algorithms available in the GIS suite of analytical tools are useable: here specifically those that are based on the techno-morphological characteristics of the lithic tools or those based on the relief data.

Spatial analysis

We first undertook a Cost Surface Analysis (CSA) which, by taking into account the geomorphological characteristics of the landscape and the presence of natural or artificial obstacles, is capable of constructing thematic maps based on the amount of time an individual takes to travel over a territory and the amount of energy he expends moving around it. By subdividing the geographical space on the basis of a 'distance/cost in resources' relationship, it is possible to obtain a representation of the cost in movement within a territory which is more realistic than one based on a simple distance as the crow flies (Pecere 2006). Since the parameters used concern mainly the morphology of the territory, we based our assumptions on the fact that an increment in the contours and the slope of the terrain has to have a corresponding increase in cost of movement.

In our case study we used CSA to create a map of the island (Figure 3) which shows the regions characterised by the same degree of 'energy consumption' demanded of an individual starting from a single starting point: this point is positioned at Perfugas, the preferred area of procurement of the flint. As can be seen from this map, although it shows the presence of sites (53 in all) with long blades in all regions, the greatest concentration of sites (35 sites containing 173 blades) falls within the first three regions defined on the basis of energy consumption. If we examine the distribution of artefacts starting from the territory of Perfugas we note major concentrations in regions which possess a consumption index of 2 (25 sites with 135 blades) and 3 (9 sites with 38 blades), compared to nearer regions with a consumption index of 1 (11 sites with 34 blades); it seems therefore that contexts located some distance from the production area are well represented compared to nearer ones.

Figure 3
Figure 3. Map generated by Cost Surface Analysis.
Click to enlarge.

Next, we undertook a Kernel Density Estimation (KDE) (Baxter et al. 1997) which allows us to identify and illustrate the density of a given phenomenon. This is made possible by techniques of interpolation based on the position of points and on their reciprocal distances: compared to classical statistical approaches, it is necessary here to divide the data into zones.

In our case study we took account in our KDE of various techno-morphological parameters, each considered separately, in order to create a series of density maps referring to each characteristic under consideration: distribution of sites with long blades based on the quantity of blades recovered (Figure 4); distribution of artefacts showing intentional retouching (Figure 5); distribution of finds with cortex (Figure 6); distribution of technical pieces (cores, initially struck blades, crested blades, sub-crested blades) (Figure 7). If we examine the resulting maps it becomes evident that the retouched artefacts are spread all over the island, but especially on settlement sites; this must be related to the function of these blades (for example transform working on vegetal or animal raw materials). The other maps reveal that there is a general dearth of blades with cortex still present and a concentration of technical elements which corresponds to the primary area of production of the artefacts.


Figure 4
Figure 4. Distribution of sites according to the number of blades recovered in them.
Click to enlarge.
Figure 5
Figure 5. Distribution of retouched blades.
Click to enlarge.

Figure 6
Figure 6. Distribution of finds with cortex.
Click to enlarge.
Figure 7
Figure 7. Distribution of technical pieces.
Click to enlarge.

Conclusions

What we have described highlights the efficacy of GIS as a tool integral to the analysis of archaeological artefacts. Using statistical enquiry, each analytical computation is performed from the position in space of a given artefact and its significance in terms of understanding it in regional context. The spatial analysis of the long blades of Sardinia has made it possible to identify concentrations of some of their significant attributes and thus to document the existence of regions which, more than others, regularly feature high among the characteristics analysed.

We have sought to shine a first light on the issues connected with the circulation of resources and people within the later Neolithic Ozieri culture of Sardinia. The capillary distribution of the latter's cultural traits all over the island suggests that a complex, island-wide network was in existence, involving the circulation of ideas, primary resources and finished products. The analysis of the long blades further seems to indicate that there existed a circuit within this potential network, recognised along the plains to the west of the island (Nurra and Campidano plains), through which these specialised products were channelled towards the consumption centres, to be used either as tools on settlement sites or as grave goods on the funerary sites.

Acknowledgements

Our research was financed by the Regione Autonoma della Sardegna - Bando regionale "Borse di studio per giovani ricercatori", POR Fse 2007-2013, L. R. 7 agosto, no 7.

References

  • BAXTER, M. J., C.C. BEARDAH & R.V.S. WRIGHT. 1997. Some archaeological applications of Kernel Density Estimates. Journal of Archaeological Science 24: 347–54.
  • BEVAN, A. & J. CONOLLY. 2004. GIS, archaeological survey, and landscape archaeology on the Island of Kythera, Greece. Journal of Field Archaeology 29: 123–38.
  • COSTA, L. & J. PELEGRIN. 2004. Une production de grandes lames par pression à la fin du Néolithique dans le nord de la Sardaigne (Contraguda, Perfugas). Bulletin de la Société Préhistorique Française 101: 867–73.
  • GAFFNEY, V. & Z. STANčIč. 1991. GIS approaches to regional analysis: a case study of the island of Hvar. Ljubljana: Znanstveni Institut Filozofske Fakultete.
  • PECERE, B. 2006. Viewshed e Cost Surface Analyses per uno studio dei sistemi insediativi antichi: il caso della Daunia tra X e VI sec. a.C. Archeologia e Calcolatori 17: 177–213.
  • VAN HOVE, D. 2004. Un modèle SIG pour l'analyse socioéconomique des paysages néolithiques: application à la Calabre méridionale. Les petits cahiers d'Anatole 16. Available at: http://citeres.univ-tours.fr/doc/lat/pecada/F2_16.pdf (accessed 29 April 2012).

Authors

*Author for correspondence

  • Valerio Pinna*
    Dipartimento di Storia, Beni Culturali e Territorio, Università degli Studi di Cagliari, Piazza Arsenale 1, Cagliari CA 09127, Italy (Email: pinna.valerio@yahoo.it)
  • Barbara Melosu
    LAMPEA – UMR 6636, Université de Provence, 5, rue du Château de l'Horloge, Aix-en-Provence 13094, France (Email: barmelosu@yahoo.it)

Translated from the Italian by Madeleine Hummler, Antiquity.