Project for the topography of ancient Achaia, Greece: Quantitative analysis and visualisation of the results of the intensive surface survey at Kamenitsa. Helene Simoni and Kostas Papagiannopoulos Kamenitsa Alissos, 25002 Vrachneika, Greece 1. Introduction This paper describes quantitative studies of archaeological data derived from an intensive survey in Achaia, Greece. The survey took place within the framework of the Project for the Topography of Ancient Achaia, organised by the Centre of Greek and Roman Antiquity of the National Hellenic Research Foundation in collaboration with the Archaeological Service of Patras (Rizakis, 1992; Petropoulos and Rizakis, 1994). Extensive research has been conducted in the ter ritory of Patras (the modern capital city of Achaia), but this intensive survey was restricted to an area of c. 6 km , round the villages of Kamenitsa, Thereianou, Alissos and Kaminia. It was implemented by young students and graduates of archaeology (Papagiannopoulos & Zahos, 1994).
ι Figure
1. Western
s
Achaia.
;
, ;
)
. \ \*
ν
.
The s t u d y - a r e a (fig. 1) is bounded to the west and south by the Peiros river, to the rth by the gulf of Patras and to the east by the main high way leading from the coast to the hinterland, very close to the industrial zone of Patras. Its geographical position is and has always been a decisive fac tor for the development and the evolution of the cultural landscape. The area lies by the coast, at the inmost part of the
S f y s been crossed by major roads that link the northern and north-eastern part of the Peloponnese with the western and south western part. The Peiros was navigable in the past. Although, nowadays, the quantity of water has declined, especially during the summer, the local authori ties are planning the construction of a dam to provide for the irrigation needs of the fields. The relief, low and gently undulating, allows a wide variety of crops to be grown on the slopes round the villages. ulf
1 1 h a s
d w
One year before the commencement of the intensive survey the coordinators of the project Dr Rizakis and Dr Petropoulos, together with the authors, had carried out an extensive one. Papagiannopoulos (1990) had also performed additional investigations. The research provided us with significant traces of ancient settle ments, not in accordance with the poor historical evidence; the area was away
from the main ancient and Byzantine centres. Nevertheless, Mycenaean tombs, late Roman graves, multi-period settlements etc. were located. 2. The Intensive Survey at Thereianou The research group decided to set up an intensive survey that would focus on a restricted area in order to explore and clarify the spatial organisation and test the methodology. Coming from a traditional archaeological background we had very little experience of quantitative analysis of archaeological data, but the type of research and the quality and quantity of the recovered material encouraged us to use statistics. The first attempts were made by an engineer, Nikos Thanassouras, at that time a student of the Polytechnic School of Athens, in an area of 1,5 km round the village of Thereianou (Zahos et al., 1996).
Figure 2. Thereianou:
the survey area with the 15 X 50m transects,
plotted
on it.
The area was divided into blocks of 120x50m (fig. 2). Every block was divided into eight transects of 15x50m. Eight field walkers, one in each transect, walked at a distance of 15m from each other. The number of the observed fragments of pottery (the sherd count) in each transect was recorded and then corrected according
to the visibility of the ground (fig. 3 a). The counts from all transects were used for the analysis and classification of the data.
I l l
High
Figure 3. Thereianou: visibility.
intensive
survey
Medium
results,
Low
π
Not surveyed
a) Surface
pottery
corrected
for
In order to plot the data on a map of the study area they were classified into a number of classes which would be sufficient to give a clear outline of the sites, their halo and the off-site area, while avoiding too much detail which would produce fuzzy broken patterns distorting our observation. The data were classified into 4 ranges: [MIN, Xm-S], [Xm-S, Xm], [Xm, Xm+ S], [Xm+ S, MAX] Where MIN is the minimum, MAX is the maximum, Xm the mean and S is the Standard Deviation of the sherd counts from all transects.
However in this case MIN > Xm-S, so only three intervals were used: [MIN, Xm], [Xm, Xm+ S], [Xm+ S, MAX] (fig. 3 a)
High
Figure
3.
b) Surface
pottery
corrected
Medium
and smoothed
Low
Not surveyed
at every two
columns.
A more comprehensive visualisation of the results was achieved by smoothing every two and four neighbouring transects using their average value (fig. 3 b). None of the maps derived in this way could be a "true" representation of the whole pattern. Each smoothing had a different impact on the presentation of the sites and their peripheries and as a result the same processing could be insightful for one pattern and deceptive for another. The analysis of the results was possible only through the combination of different smoothings and a good knowledge of the landscape itself (Zahos et a l , 1996: 31). 3. The Intensive Survey at Kamenitsa During the following archaeological campaigns in 1991 and 1993 the survey trans ferred to the area around the nearby village of Kamenitsa. An area of approxi mately 6 k m was plotted in a grid comprising of "squares" of 150 χ 150 metres with a changing orientation (fig. 4). We applied this method with the particular geomorphology of the region in mind. The goal was to keep the field walkers at 2
15m from each other, which would not have been possible with uniform squares plotted on this terrain, which is not flat. In this way we had better information of our position in the landscape and of the possible existence of small sites. Each square was divided into 30 transects sized 15 χ 50 metres and each field walker walked along the long axis in the centre recording everything within a visual range of one metre.
Figure 4. Kamenitsa:
the survey area with the 150 X 150 grid plotted
on it.
The choice between a grid or modern field boundaries for the application of a field survey is a common topic of discussion among archaeologists from all over Europe, especially those coming from the North but working in the South.
We are definitely in favour of the grid for several reasons: • Since Greek independence, last century, there has been no cadastral map of the state. • The existing field units that the farmers of the area exploit are very small and cannot be compared with the broad plains and valleys of other European landscapes. • The boundaries of modern fields are not necessarily the same as those present in the past and there is no extensive work on thecadastres of the research area that could give us a clear picture of land use in the recent past, let alone in antiquity. The existing field boundaries are useful only for the representation of the recent or modern economic and cultural landscape. • We intended to survey the landscape as a single continuous surface divided into smaller arbitrary but regular units, with a detailed description for each. The field boundaries tend to distort this picture of the landscape. • It is easier to import the data into a GIS environment. This is much easier to do for a uniform grid with more or less even geometrical shapes than for tiny polygons of irregular shape. In this way, our basic units are of similar size and we maintain a uniform resolution of our data and control of the terrain. The volume of the archaeological material led us to use a database, rather than a conventional paper-based system, to keep track of the data. Our database was recorded in Excel 4.0 for Windows, in which all the statistical calculations were performed. Again, the same ranges were used: [MIN, Xm], [Xm, Xm+ S], [Xm+ S, MAX]. However, this time the sherd count was corrected taking into consideration not only the visibility but also the accessibility of the fields (that is how many metres each fieldwalker had walked) since very often only a part of the 50m long transect was accessible due to thick vegetation or modern construction. The 3 classes correspond respectively to a low, medium and higher density of sherds. Treating the surface as a continuous archaeological landscape the same formulas were applied for the overall calculation and evaluation of the data. By recording not only the archaeological evidence but several environmental parameters, we could keep a control on our data as well as our methodology. For example, after 2 years of field walking three times per year, late October - early November, late April, late June early July, it became obvious that the best collections were possible in the autumn, just after the ploughing of the fields and the first rainfalls, and we gradually limited the number of expeditions to a major one at this time of year. The same software was used for drawing the map. The structure of the spreadsheets resembled to the structure of the grid; each spreadsheet cell represented one transect on the ground. Although the maps that were produced did not demonstrate the geography faithfully and accurately, since in reality the transects are not rectangular but, rather, parallelograms with a varying orientation, they proved to be very enlightening and were extensively utilised during the fieldwork for assessing what had been done so far and making decisions on what to do next (fig. 5).
Figure 5. Kamenitsa: intensive tery corrected for visibility.
survey
results
of the period
1991-93.
Surface
pot
Since the database and the " m a p s " were two separate Excel files we were not able to make them work together automatically. Changes in the database were auto matically made by simply changing or complementing the formulas we used, but this was not possible for the file of the " m a p s " . This would have required regular manual corrections of more than 7.000 entries! For the delineation and the interpretation of the archaeological patterning no smoothing was used. Instead, those parts of the grid that seemed to present higher occurrences of archaeological material were surveyed more intensively at a second stage with the aim of defining the time span of the artifacts and the extent as well as the type of the site in every period. We used transects of 10 χ 10m that were walked completely. Again the same statistical analysis was performed to quantify and visualise the distribution of pottery across the landscape.
Figure 6. Kamenitsa: the 10 X 10 grid of the «site Z» superimposed over the pottery pattern of the intensive survey a) using the raw data, b) corrected for visibility and accesibility
4. The need for GIS Our financial resources were extremely limited and so was our time. All the participants worked on a voluntary basis. Additionally, GIS technology was not accessible to the study group so we did not plan the project with the needs of a GIS application in mind. However, we thought that the inclusion of the whole study area in a GIS would provide a control of all our previous analysis and would give our study added potential. Since we are using volunteers, so far only 1/3 of the grid at Kamenitsa exists in a digital format. Digital data is also available on two Classical sites: Omega (Andrinopoulos & Simoni, in press) and Zeta. The site Zeta was discovered during the intensive field work (fig. 6 a, b). The raw data yielded a long band (c.1500 m ) with high concentration of pottery and a periphery round it, that was cut by a rural road. After correction for accessibility the band decreased in size by half ( 7 5 0 m ) and it was limited to the south. At this stage it was considered useful to survey the whole area round this concentration u s i n g a 10x10m grid. Guided by the density of the pottery in the squares, research was directed to the n o r t h and Northwest, finishing where the sherd counts decreased. 2
2
The environmental parameters (time, conditions of ground and light) were favourable. The survey took place mostly on cloudy days, late in the morning and the ground was m o s t l y ploughed and without shadow. T h e r e fore, the visibility of Figure 7. Site Z: The 10 X 10 survey results. Surface potthe surface was above tery; a) raw data b) corrected for visibility and accessibility.
average. W h a t is worth noting is that the conditions were similar for neighbouring squares. In this way the difference in the abundance of material can be explained archaeologically and we can reach more reliable conclusions. Comparing the results of the original survey using 15 χ 50m transects and the new survey on a 10 χ 10 m grid it is clear that a second "site" is evident to the NE, while the already existing "site" has moved slightly to the east (Fig. 7 a, b). These differences are easily interpreted because of the different (minimum- maximumStandard Deviation- Average) values that are used for the statistical analysis. In the case of the original transects, counts from the whole study area are estimated. In the case of 10 χ 10 m grid, we use only the counts from this "site" itself. The area of the concentration varies between 400 and 1500 m according to the way in which the data is manipulated. We have seen already above the results from the survey us ing 15 χ 50m tran sects. Processing of the data derived from the 10x10m grid yields two separate smaller concentra2
tions covering 700 m in total. Correction for visibility and ac cessibility does not al ter the picture. Using only counts of tiles, the area would be 800 m (corrected) to 900 m (raw) (fig. 8 a, b ) , w h e r e a s the highest counts of the rest of the sherds oc cupy a smaller area varying between 40 0 m (raw) and 2
900m (corrected) (fig. 9 a, b). The average of all these values is 831.25m which is close to the area es t i m a t e d u s i n g the counts of tiles. H o w e v e r , the aver age area as estimated by the total collection of pottery (700 m ) is 8. Site Z: The 10 X 10 survey results. Tiles recorded; a) raw data b) corrected for visibility and accessibility
n 0 t
v e r
^ different.
However, the pattern of the sherds excluding tiles is not very reliable for general conclusions since sherds are spread all over. This may be the result of their movement in the ploughzone, facilitated by the inclination of the slope from Northeast to Southwest and the fact that sherds are usually lighter and smaller than tiles. The different apparent size of the concentrations, as evidenced by high raw and corrected sherd counts, is biased due to the different conditions u n d e r w h i c h each square was searched. It is obvious that certain high concentrations appear after the correction especially when the c o n d i t i o n s of field w a l k i n g h a v e been hard e.g.: when the visibility has been Figure 9. Site Z: The 10 X 10 survey results. Surface pottery low, the ground dry, except for tiles; a) raw data b) corrected for visibility and the sky c l o u d y at accessibility noon and part of the square unapproachable because of vegetation or rural road. On the other hand, very favourable conditions such as high visibility, at a sunny noon, on a wet freshly ploughed surface, tend to minimise the final corrected numbers and shrink the area of high concentrations. This is the reason for the two slightly different pictures that the maps with raw and corrected data display. It is a pity that another rural road to the SW of Zeta prevented the continuation of the 10x10m survey to this direction, since high concentrations of pottery were recorded there in the 15x50m transect survey. A grab collection from the area across the road included many diagnostic sherds of the same (Classical) period.
The next stage of this work will focus on the study of the landscape in different periods. The study of the settlement pattern in each period is to a certain extent based on the known published literature, but to a greater extent on the analysis of the ceramic material that has been collected in the fields during the survey (Zahos and Papagiannopoulos, in press). Several archaeologists specialising in different fields and periods are together preparing the publication of the intensive survey, which is to be expected within the next 2 years. Our work with GIS has prompted us to examine certain theoretical aspects of our research that we had not considered earlier due to the lack of a suitable tool. Operations like interpolating, filtering, reclassifying and displaying the data, that used to be time consuming, can now be performed in seconds. The danger behind this enhancement is that we may make a spectacular but deceptive combinations of options and modules. Theoretical problems may arise which slow down an otherwise fast procedure. The question is: how can we manipulate the systems in order to depict reality more closely without biasing the procedures to give us the results that we want? This is very problematic, and any answer will be so as well, since there is not one reality, but a set of subjective observations performed by a group of people. Acknowledgements The study presented above would not have been possible without the contribu tion of several friends and colleagues. We would like to thank the more than 100 colleagues who participated at the survey and collected the necessary information for this paper and especially V. Tsakirakis and Ar. Koskinas from the Landscape Archaeology Group (Athens). The Regional Development Centre of W. Achaia provided the necessary equipment. Thanks are owed to Darja Grosman and Brane Music (department of Archaeology, University of Ljubliana), Kristof Ostir (Academy of arts and science, Ljubljana), and Kostas Mahairas (Regional Development Centre of W. Achaia) for their support on technical matters. Kostas Mahairas prepared the figure 1; to Zoran Stancic for his patience during long and very constructive conversations with the authors about this paper in particular and GIS in general; and to John Peterson who read and improved the language of the text. References Andrinopoulos A. and Simoni H., (in press). "Εφαρμογή Γεωγραφικού Πληροφορικού Συστήματος στην αρχαιολογική έρευνα της Δυτικής Αχαΐας (G.I.S. application on the archaeological survey of western Achaia)," Proceedings of the International Conference on Ancient Dyme, DymaiaVouprassia, Kato Achaia, 6-8 October 1995. Papagiannopoulos Κ. B., 1990. "Επιφανειακή αρχαιολογική έρευνα στην περιοχή Αλισσου - Θερειανου Αχαΐας (Archaeological surface survey in the Alissos - Thereianou region, Achaia)," TOMOS TIMHTIKOS Κ. N. TRIANTAPHYLLOU. Patras, 539-553.
Papagiannopoulos K. and Zahos G., 1994. "Intensive survey," in Petropoulos M. and Rizakis Α., "Settlement patterns and landscape in the coastal area of Patras; preliminary report", Journal of Roman Archaeology 7, 187- 189. Petropoulos M. and Rizakis Α., 1994. "Settlement patterns and landscape in the coastal area of Patras. Preliminary report," Journal of Roman Archaeology 7,183- 207. Rizakis A.(ed), 1992. Paysages d'Achaie I: le bassin du Peiros et la plaine occidentale. ΜΕΛΕΤΗΜΑΤΑ 15, Athenes. Zahos G., Papagiannopoulos K., Simoni H. and Thanassouras N., 1996. "Πειραματική εφαρμογή εντατικής έρευνας στην περιοχή Θερειανου Αχαίας (Experimental application of intensive survey of T h e r e i a n o u area, Achaia)," Pyxida 1/1995, Athens, 24-34. Zahos G. and Papagiannopoulos K., in press: "Η εντατική επιφανειακή έρευνα στην Αχαΐα: μια αλλη προσέγγιση (The intensive surface survey in Achaia; a different approach)," Proceedings of the International Conference on Ancient Dyme, Dymaia- Vouprassia, Kato Achaia, 6-8 October 1995.
