In the case of Holešov, a typological approach is of limited value because of the morphological simplicity of the finds. Problems emerging from these circumstances can nevertheless be mitigated by focusing on the quantitative and spatial aspects of the studied material. The identification of general structures in the data was therefore the primary concern, and required a careful selection of tool categories that were capable of performing the tasks needed. This section explains the basis of a methodological concept that leads to the visualisation of patterns reflecting the formal relationships among archaeological finds. The advantage of this approach lies in the fact that a large amount of information can be concentrated and visually inspected in a single plot or a series of several plots.
The first step was the creation of a descriptive database that would contain information on the two-dimensional position of each grave and the list of attributes associated with the grave pits (Wheatley and Gillings 2002, chapter 2). Although a pair of coordinates referring to the central point of each grave would be sufficient to record the spatial location of every burial as-semblage, basic outlines of graves were also digitised from the site catalogue (Ondráček and Šebela 1985) to ease orientation in the site plan. A hierarchical system of tables containing the description of grave pits, burials and objects could thus be linked either to polygons or points representing individual graves in the course of the subsequent analysis. The final version of the database was realised in Microsoft Access 2000. The digital site plan was originally created in the GIS package Idrisi 32 and some of the analyses were conducted in the same software. Later the whole Geographical Information System was migrated from individual files into a personal geodatabase (still in the MS Access format) whose structure follows the specifications of ESRI. ArcGIS 10 (Shekar and Xiong 2008, 25-31) is currently the primary GIS software used by the author for data maintenance and analysis, given its long-term accessibility and support at the institution where the author is based. Statistical procedures were carried out in the program bundle Statistica, versions 5 and 6. Alternative software, including Open Source, could be used to implement the same approach.
The methods applied in this study can be divided into three major groups. The first includes several techniques belonging into the domain of spatial analysis (in a geographical sense). Simple distribution maps were of limited value but in some respects helpful for revealing a few hints worth following with the aid of more advanced data processing, such as spatial filters and trend surface analysis. Soon it became clear that if one should get beyond the inherent limits of distribution schemes it is necessary to take advantage of quantitative comparisons. What makes a really significant difference between various parts of the cemetery is the local density or intensity of selected properties; in other words the relative differences between larger areas, spatial trends and gradients (Šmejda 2003a; 2004).
|length||width||depth||lockring||potsherd||vessel||bone bead||cattle rib||faience bead||stone blade||stone arrowhead||stone flake|
The formal variability of the mortuary evidence was investigated by the second group of methods of statistical character, which define a multi-dimensional analytical space ordered by the formal attributes of the burials. For this purpose, the correlation analysis of frequently represented traits was conducted (Table 2) and its results further investigated by means of Principal Component Analysis followed by the extraction and rotation of factors (Neustupný 1993; Shennan 1997; Meyers et al. 2013, chapter 12A), in order to get mutually uncorrelated (orthogonal) axes of data variability (Tables 3 and 4). It was believed that the multivariate methods would be able to reveal a latent structure of mortuary behaviour and, when combined with the results of sexing and ageing of excavated skeletons, to shed some light on social aspects coded in the burial rite (Šmejda 2003b).
|Factor||Eigenvalue||% Variance||Cumulative eigenvalue||Cumulative % variance|
Results of these analytical techniques can be examined further by methods that have proved to be productive in the course of structural analyses and are currently getting a lot of attention in archaeology, as well as in other social sciences (Jackson 2008; Knappett 2011). They are based on structural relations within a given set of variables on the one hand and of studied entities on the other. Here the methodological apparatus of the graph theory can be applied, which studies the relationships described as networks (Neustupný 1973; Hage and Harary 1983; Gross and Yellen 1999; Sosna et al. 2013). This article works with an approach that is different in nature but can be used as a useful alternative to (or in combination with) graph theory in the course of structural analysis. The technique proposed in this article builds new unconventional models of formal space that can be made in the standard environment of Geographical Information Systems, but rather than describing landscape it is used as a tool to conceptualise research ideas and reveal non-geographical relationships in the data (Šmejda 2008).
|Variable||Factor 1||Factor 2||Factor 3||Factor 4||Factor 5||Factor 6|