PREVIOUS   NEXT   CONTENTS   HOME 

Summary | Introduction | Method of Analysis | Sample for analysis | Results of the functional analysis | Analysis of the functional results | Statistical analysis of use-wear data | Ethnographic data | Users | Change in edge angles | Relations between the four phases | Discussion

6.10 Relations between the four phases

PCA eigenvectors and scores were calculated for the first three principal components for each of the four phases (morphological, macroscopic, microscopic and ethnographic) after modifications to the dataset described previously. The eigenvectors can be related to the original variables making the greatest contribution to each and scores can be related to each other by correlation (Jeffers 1978). Thus it is possible to relate, using correlation, between the principal component scores in different phases and find an overall indication of the relationship between groups of tool properties. In the following the tool properties contributing most strongly to the principal component is shown in brackets. Morphology PC1 (tool size and edge thickness) was related to microscopic PC1 (polish invasiveness, distribution and development) (r= 0.431***) and to all three anthropology PCs (r>0.259*). Morphology PC2 (shape and edge angle) was related to anthropology PC2 and 3 (user and edge angle unused) (r>0.299*) while morphology PC3 (edge length) was correlated (r=0.252*) with macroscopic PC3 (ventral macrofracture type) and anthropology PC1 (time used and use) (r=0.358*). Macrofracture PC1 (all variables) was correlated (r=0.433***) with microscopic PC2 (striations, microfractures and linear features). Macrofracture PC2 and 3 (Dorsal and ventral macrofracture type respectively) were related to anthropology PC2 (r>0.303*). Microfracture PC2 and PC3 (microtopography) were related to anthropology PC3 (r>0.265*).

From this we can deduce that the thickness of the edge appears to be influential in the development of macro and microfractures. These fractures appear more related to properties of the edge than the use to which the edge was put. And 'user' correlates with the size and edge of tools, which suggests that selection of tools is a personal matter and that individual assessments of requirements vary.

The PCA scores suggested that the physical properties most important in influencing use of tools were in fact those that had been isolated in the earlier analysis, as variables which were related to other variables within the same phase of properties. This indicated strong general relationships within and between phases of tool properties and uses and users and so a cluster analysis was employed to detect any distinctive groupings of similar tools. The sixteen variables found to be best related to each other were used – user and use from the anthropology group, edge angle, edge length, edge thickness, mass and Z shape from the morphology group, dorsal and ventral macrofractures from the macroscopic group and microfractures, microfracture type, microtopography, polish distribution, polish distribution type, microlinear features, and polish invasiveness and development from the microscopic group. Four distinct clusters of tools were found and the centroids for each property in each of clusters A to D are shown in Table 17.

Cluster A contains cubic to columnar shaped, small tools with long edges and wide edge angles. These pieces were used for cutting rattan, and cane grass, but mostly bamboo. Polish distribution was almost always continuous and not invasive (polish distribution level 'edge only'). Linear features were absent and polish was quite well developed (polish development level 'B').

Cluster B consists of thick edged, large cubic tools with relatively long edges and wide edge angles. They were almost all used on wood. Polish was contained continuously and evenly to the edges and the development of the polish was low (polish development levels 'A+–B'). Linear features were absent.

Cluster C is made up of small, squarish, intermediate to planar tools with low edge angles and thin edges. Most of these tools were used on rattan, vine, or grass. They were used for a long time and their polish distribution patterns were almost always 'edge only/asymmetric'. Polish was well developed and invasive and occurred both continuously and intermittently along the used edge. Parallel and perpendicular linear features were quite common.

Cluster D consists of small, planar tools with low edge angles and thin edges. They had all been used for a short time. These tools were used for butchering, grass and vine working. Polish distribution was always continuous and the polish distribution type was either 'edge only even', or 'edge only asymmetric'. Linear features were absent and the polish was generally quite invasive (polish invasiveness '<0.5D') and relatively well developed.

The uses to which all the artefacts in each cluster had been put were examined, in order to assess the functional significance of the clusters, and to detect any differentiation between bamboo, grass, rattan and vine. Two uses occurred almost equally across three of the four groups: bone working and paring tree fern pins (clusters A, C and D) [The small number of tools used on bone (3) limits the relevance of this observation.] Wood-working occurs almost exclusively with cluster B tools. While bamboo, rattan, vine and grass occurred in the other three clusters (A, C and D), bamboo working was strongly dominant in cluster A, vine was dominant in cluster C and rattan and grass were dominant in cluster D. The butchery tools occurred only with tools in cluster D. Bamboo was worked on with long, thick edges. The polish was relatively well developed (polish development group B). This is in keeping with work on a relatively hard material.

The morphology of the used edges in clusters C and D were similar, being thin with low edge angles. Cluster C tools had all been used for long periods of time while Cluster D tools had been used for short periods of time. The polish on both groups of tools had some similarities of distribution but the polish on cluster C tools was less developed than cluster D. Linear features were only present in cluster C. As the edge angles and general morphology of the tools in both clusters was so similar, the differences in polish patterns must be related to the raw materials worked on, with grass and rattan producing generally more invasive and more developed polish than vines.

 PREVIOUS   NEXT   CONTENTS   HOME