M. Járó - L. Költő szerk.: Archaeometrical research in Hungary (Budapest, 1988)

Analysis - BARTOSIEWICZ László: Water-sieving experiment at örménykút, site 54

the 25 mm water sieving becomes inefficient below the 1 cm size limit. This, however, is in part the consequence of the method of measurement: splinters with a greatest length of 1 cm may be as slender as 1 to 15 mm, thus they are easily washed through the 25 mm mesh. Fig.l Frequency polygons showing the size distribution of fragments recovered by the three different kinds of techniques. Continuous line: hand collection; dashed line: water-sieving with a 2.5 mm mesh; dotted line: water-sieving with a 0.8 mm mesh. Line shading serves to show the overlapping size ranges of water sieving carried out on two levels of refinement. In the next step of calculations cumulated frequency data from Table 1 were studied as a function of increasing refinement expressed by the declining order of size categories. A plot of these values against each other would result in a curvilinear relationship which may be approached using an exponential function (Brody 1945). Curve fitting within the sub-samples of hand collection and water-sieving with 25 and 0.8 mm mesh respectively was facilitated by the use of a logarithmic transformation. Thus, decimal logarithms of the basic data outlined linear trends which could be evaluated in terms of a linear regression analysis. This method served to measure the degressive tendency apparent within the methods of recovery at increasing refinement. Potential intersection points between the three functions would also define size intervals beyond which one recovery technique becomes more productive than the other. Finally, these observations could be judged in terms of correlations and statistical probabilities which are indispensable in appraising the accuracy of estimations.