Alba Regia. Annales Musei Stephani Regis. – Alba Regia. Az István Király Múzeum Évkönyve. 23. 1984-1985 – Szent István Király Múzeum közleményei: C sorozat (1987)

Tanulmányok – Abhandlungen - Choyke, A. M. – Bartosiewicz László: Animal exploitation and its relationship the bone deposition at Lovasberény-Mihályvár. p. 7–18.

The factor analysis itself was used to re-define these correlations in a synthetic way. Animal species of major economic importance at this site may be grouped in terms of three principal components with a latent root exceeding 1. A rotation of this theoretical structure was performed to satisfy the simplicity criterion, that is, to achieve the best fitting pattern. Since the obtained factors are related to individual variables (species) by positive factor loadings (Table 4), they are considered unipolar (HÁRMAN 1967). The first and most characterise factor represents 28.9% of the total variance involved in the studied phenomenon and is defined by leftovers from the consumption of domestic pig, caprines and cattle. Thus, it may be labeled "Animal keeping" and is indicative of the combined deposition of bones resulting from the exploitation of domestic animals. Similarly, the second factor, with a 25% explanatory value, may be termed "Hunting", referring to the consumption of red deer. The third significant factor describes 22.8% of the total variance after Varimax rotation, and is mostly influenced by the presence of bones from the two large domestic ungulates, cattle and horse. It is probably indicative of a functional trait of the features concerned and was thus termed "Butchering site". The intended meaning of this keyword was, that while the previous two factors express relations to alternative meat procurement strategies, features best associated with the third factor probably served as deposits for bones removed from large carcass sections during primary butchering, which did not actually make it to the other two types of mixed deposits. The lack of remains of large game animals associated with this factor may be indicative of occasional off-site butchering of red deer, wild pig and probably aurochs. From the viewpoint of seriation the first two factors may be of particular interest. Although features indicative of the consumption of meat from domestic animals, summarized by the "Animal keeping" factor are apparently disturbed by the presence of "wild cattle" bones among the species with largest factor loadings, one should reckon with a certain degree of misidentifications of this animal due to its considerable morphological similarity and size overlap with domestic cattle (Bökönyi 1962). In the next step of the calculations each feature was assigned to all three factors to differing degrees. First, the significance of deviations from the norm set by previous calculations was evaluated. In Table 5 Mahalanobis distances (represented by yj

1 values) from each feature to the centroid of all these cases for the original data (first column), factor scores (second column) and their difference (third column) are listed. (The x

2 values were divided by their degrees of freedom, 8, 3 and 5 respectively). This list illustrates the outstanding faunal composition of Pit 55, House 61 and House 65. While the first of these features may have been some sort of an "industrial deposit" or "sacrificial pit", the two houses are characterized by a great number of horse and sheep/goat bones respectively. The factor scores listed in Table 6 show to what degree a particular feature may be associated with each of the separate factors which form independent continua in and of themselves. Ranking along these three dimensions will be interpreted as a form of seriation in subsequent conclusions. Using such a seriation results in perspectives which have already been expressed when labeling the factors themselves. In Table 6 the highest positive factor score for each case is marked by italics, as indicative of association with one of the three factors. Cases with all negative values were not taken into consideration, as being related to the three types only by negative evidence. Such non-characteristic deposits may equally be the result of random primary deposition or secondary disturbance resulting from subsequent prehistoric activities. Aside from these features however, twenty others could be classified as related to the meat exploitation and consumption of domestic animals, while ten are characterized by evidence of hunting. Another eleven features seem to be related to the butchering of predominantly large domestitates including the deviant pit 55. Conclusions and Suggested Application Although the concentration on species frequencies, rather than skeletal elements per species in this study prevented the possibility of direct (within skeleton) correlations between the variables, the first hand chronological significance of possible groupings may be biased by a number of effects that vary in nature. Taphonomic bias. Although a number of non-diagnostic factor scores were considered as being indicative of secondary deposition, such influences can not be intuitively sorted out without consideration of the site's stratigraphy. Prehistoric re-occupation and construction activities may contribute a considerable amount of distorsion in this sense (Choyke 1984). Whether this tendency is expressed by uniformly negative factor scores to a satisfactory degree remains to be answered by further analyses, even though the factor loading matrix in Table 4 contained nothing but unipolar, positive factors (White — Thomas 1972; Thomas 1978). Behavioral bias. The exact interpretation of seriations obtained by this method may be difficult to arrive at beause of potential variability in Prehistoric behavior. The hunting/animal keeping dichotomy is only one of the hypotheses generated to help understand subsistence strategies. It is possible, however, that this distinction is rather functional than chronological (for example shows locations devoted to infrequent but major hunting feasts), just as the label „Butchering site" marking features with the occurrence of more horse bones, may be indicative of diachronic change rather than a distinguishable behavoir, such as localized dismemberment of large domesticates. Computational bias. The method chosen for analysis transformed a number of discrete data (species/bone frequencies) into three major continuous series. There was a risk that trivial correlations, such as the random but simultaneous occurrence of relatively underrepresented species may be overemphasized by this technique. It was hoped, that biases of this nature were largely avoided by the exclusion of less frequent and not necessarily meat purpose animal species from the calculations (dog, for example). While, it may be argued, many of the characteristics established by this complex and computer-bound procedure could have been recognized by visual appraisal of the basic data matrix, it was hoped that the three continuous synthetic variables obtained will facilitate subsequent quantitative analyses. 12