Somogyi Múzeumok Közleményei 18. (Kaposvár, 2008)

FEKETE CSANÁD: Predictive archaeological modelling in Somogy county

the soil is an important standard of value, which could be vital for the settled, farmer communities. The preference of the slightly acid and neutral soils shows that most of the settlements were developed on surfaces or next to them that are siutable for agriculture. As for the physical attributes of the soil, 35 sites of the sample area can be found on the soil with strong water driving and high water content ability, 9 sites are on the soil with high water content and good water driving ability, while 7 sites are on the soil with very strong water driving and small water content ability.

17 The soil with good water content and strong accomodation ability (e.g. meadow chernozem) has good ability to supply nutrients.

18 Permanent vegetation cannot settle on the soil with very strong accomodation and slightly water content ability (e.g. shifting sands). Their ability to supply nutrients is poor, they fast swallow water and they tend to be dry out.

19 It is not surprising that there are hardly any sites in this kind of soil environment in contrast with the soil with good nutrient and water supply. Afterall the next conclusion can be drawn: most of the sites are on the soil type with slightly acid reaction, strong water driving and average water content ability. A small but significant group of the sites are located on neutral reaction soil with strong water driving ability. The sites of the later soil type group north and south of the inner village of Somogyjád. The same type of soil on the western part of the sample area seems to be less preferred, whatsmore the sites can be found - at least as the accuracy of the soil map allows - on slightly acid surface that is got wedged in the neutral reaction patches of soil. There are hardly any sites on the strongly acid soil and on the high water content soil with poor water driving ability, as well as on the sediment of the flood area neither. The advantages and weaknesses of the model The benefits of the model are considerable. The predicted area of the sites are almost perfectly modelled in some cases. Both their extension and their direction suit the polygons of the sites. These areas allways rise from their environment as islands or peninsulas, so they must have had favorable aptitudes to live on. In some cases those ridges of hills can be identified, where relatively many sites are in narrow, encompassable bands. These bands nicely indicate the regularity of the sites' alignment. The sites are stitched on the watercourses of the valleys with NS direction that peculiar to the Somogy-hilly country. There are some sites, whose territory hardly touch the predicted inhabitable zone, however the model still provides sufficient data to clear them up. There are some weaknesses of the model, as well. The first one is that only one physical parameter would be needed for the predicting of the sites: the distance from A couple of sites are located on forest soil and in the inner village. STEFANOVITS 1975, 212. STEFANOVITS 1975, 182. the watercourses. This factor is the most important one that determine the positioning of the sites.

20 Another weakness is that the model predicts such areas - in not significant ratio - that are not justified by field data. The model makes some sites possible in a long band, in the Újfalu-dűlő of the eastern part of Somogyjád, on the eastern bank of the Pogány-völgyi-víz. This area, however, is lack of sites entirely. The same observation can be made on the northwestern bank of the Bodonkúti-árok, where there is no human settlements despite the apparently favourable conditions. Checking the model on the control area I chose such an area that had territories, which were rich in sites and also those that were without them. The final result is qualified by not only the most precise modelling of the sites but the predicting of the „siteless" areas, as well. The control area consists of two condensations of sites and a no man's land between them, which makes it suitable for checking the model built on the physical paramètres of the sample area. I made the same steps as in the case of the sample area with the exception that the geographycal position of the sites (71 ) was not included because their predicting was my aim. (Table 2.) I had to face some problems that did not turn up on the sample area, when the bufferzone of the watercourses were assigned. The Bontakeszi-árok, the Fajszi-patak and their sidebranches are densely shot with different drains. These canals are apparantly used for watering those agricultural parcels that were away from natural waters. It is very important to make difference between the natural and the artificial beds because the ditches, drained in the last few decades, did not have the same role as the watercourses used to have. One part of the drains seems to be easily identifiable because they are straight or rectangular. Another part of the drains cannot be squarely distinguished from the natural waters. In this case the military surveys and the relief model were used. With the help of the Second Military Survey I determined those parts of the beds that were made artificially. After a thorough examination some drain-like beds coul be also distinguished on the military survey. Those places, where water cannot be flown under natural circumstances, can be easily identified with the help of the terrain model. These drains were not included into the assignment of the water network of the control area. (Figure 7.) Result of the model on the control area If we consider only statistical viewpoints, then the priority of the southern-eastern aspect of the sample area cannot be confirmed on the control area. While 65% of the sites (with their whole or larger territory) of the sample area are located on southern-eastern aspect surface, till It is worth to mention that some errors could occur during the digitalizing of the hydrographie data. There could be tiny sidebranches or rills where the maps do not indicate them.