Alba Regia. Annales Musei Stephani Regis. – Alba Regia. Az István Király Múzeum Évkönyve. 20. 1980 – Szent István Király Múzeum közleményei: C sorozat (1983)
Tanulmányok – Abhandlungen - Éry Kinga, K.: Comparative statistical studies on the physical anthropology of the Carpathian basin population between the 6–12th centuries A. D. p. 89–141.
a few important data (for example body height) could not be included in the calculations because of the different computational methods applied by various authors. A further difficulty arose from the limited power of applied statistical methods in the overall characterization of each of the measurements and discreta of the populations concerned. This is why the following study concentrates on the recognition of tendencies. Detailed evaluation will require the exploitation of new resources of data and the use of more improved methodologies. B) As in the earlier research this work was based on the calculation of distances. The following study however, uses exclusively the measurements of males, as only this gender was represented in larger series. Inclusion of female data might have blurred possible patterns emerging for the wider comparison aimed at here. At the same time, questions posed in this paper could fully be answered on a population level using the measurements of only male series as welK 1 ). The main criterion in entering series into the statistical analysis was sample size. In order to provide an acceptable basis for evaluation at least ten cases per measurement were required with the exception of six important smaller samples which were used to fill temporal or territorial gaps between the major samples. The average sample size however, was even in these cases above ten( 2 ). The composition of basic data used to characterize the populations of the Carpathian Basin thus may be summarized as follows: Avar Period 22 samples, 9th century A. D. 1 sample, Period of the Hungarian Conquest 4 samples, Period of the Árpád Dynasty 18 samples. Earlier habitation areas of the 6—12th centuries populations of the Carpathian Basin were represented by data from the area east of the Carpathians. It is a historical fact that in addition to the Avar Period, the early history of the Hungarians is also closely associated with the Eurasian steppe belt. Thus, 63 eastern samples come from various parts of the USSR. They range from the Minusinsk Basin to the Eastern Carpathians and from the Kama river region to the Pamir Mountains. Temporally they represent the Late Bronze Age (cca. 15th century B. C.) and the Period of the Mongol invasion (13th century A. D.). Unfortunately, due to the lack of sufficient data, populations which lived before the Late Bronze Age in that area could not be studied. Populations north and west of the Carpathians were not represented in such an abundance of chronological (1) This is proved by the results of the following examination. The number of samples containing female skulls were sufficient for analysis in 83 of the 120 series. The female metric characteristics of these were converted into male equivalents using the A 1 e к s e e v — D e b e с procedure (1964). On the basis of 12 cranial measurements 93 percent of the calculated distances between female and male skulls resulted in similarities onaP^ 1.0% significance level. (2) These six, relatively small samples are the following : No. 6 : Lugovo, No. 34. Lower Kama region (Cheganda and Mazunino groups), No. 35 : Upper Kama region (Mitinsk and Demenki), No. 92: Tiszaderzs, No. 98: Nitra-Lupka, No. 104: Besenov. varieties. Due to their possible connections with the 6—12th century history of the Carpathian Basin eight "western Germanic" (4—7th century A. D.) and three "western Slavic" (9th century A. D.) series were also included in the comparison. Unfortunately no series of sufficient number have been published for areas south of the Carpathians (Romania, Bulgaria, Yugoslavia). As far as the Carpathian Basin is concerned, a similar lack of complexity is characteristic of the data from earlier periods than the 6th century is also a fact. Aside from a few Late Roman Period (4—5th century A. D.) series from Eastern Transdanubia sufficient information is not available. Similarities between these Roman Period series allowed their inclusion into the same sample, while there are only sporadic or no data on "Hun —Germanic", "Sarmatian", Iron Age and Bronze Age populations. The situation is even worse in regards to the osteological data from earlier periods. This is why the continuity of the 6—12th century A. D. population in the Carpathian Basin could not be accurately traced back to the earlier times. C) The analysis was again carried out using Penrose' s distance method (1954) (see also Knussmann 1967). The following ten cranial measurements provided input for the statistical calculations: greatest cranial length (Martin 1), greatest cranial width (Martin 8), smallest frontal width (Martin 9), basion-bregma height (Martin 17), facial length (Martin 40), facial width (Martin 45), upper facial height (Martin 48), orbital width (Martin 51), orbital height (Martin 52), nasal width (Martin 54). The standardization of mean values was carried out using Thoma's (1978) mean sigma values (erg). The significance of generalized distance data (CR) between each pair of samples was tested by Rahman's (1962) method. The 1.0 percent level of significance was chosen as a criterion for similarity (y? * 99.0). Thus, generalized distances equal or smaller than 0.197 were regarded as indicative of similarity between particular series. Needless to say, this similarity criterion is only conventional. In fact, sometimes larger values may blurr actual similarities just as smaller distances do not necessarily guarantee similarity. This is why a purely quantitative evaluation of the obtained values cannot be sufficient. One must also pay attention to the structure of interrelationships established between the series by this computational procedure. Obviously, the success of the method introduced by Penrose is just as dependent on multiple hypothesis testing and critical interpretation of the results as any of the other multivariate techniques. Whenever a considerable body of data is subjected to statistical comparison, clustering is instrumental in achieving the right orientation in the mass of values measuring distances between each pair of cases. The dual sequential procedure (Creel 1968) is one of the many possibilities which may be presented in the form of a dendrogram. It thus helps with an easier understanding of the interrelationships between series. The disadvantage of using dendrograms however, is that due to their two dimensional structure they cannot always be efficiently used in demonstrating all directions and details of relationships. Sometimes even characteristics of secondary importance assume 90
