Explaining post-depositional
processes that affect Bronze Age
survey distribution patterns in the
Raganello basin, Calabria
A pilot study of the Contrada Damale
W. Jelmer Wubs
Student number: 1781359 12-8-2015
Master Thesis: Final Version
1
Table of Contents
1. Introduction ... 3
1.1 The pilot study area ... 6
1.2 Problem analysis ... 10
1.3 Research questions ... 11
1.4 Structure thesis ... 12
2. Literature study... 13
2.1 Survey: site and off-site ... 13
2.2 Post-depositional processes ... 17
2.3 Ploughing and the plough zone ... 19
2.4 Agricultural terraces ... 23
2.5 Comparisons: How others deal with the problem ... 25
3. Spatial analysis ... 29
3.1 The effects of agricultural processes on the archaeological record ... 30
3.2 The data ... 35
3.2.1 The GIS data ... 35
3.2.2 The survey data ... 37
3.2.3 Pre-processing data: analysis of the units ... 40
3.2.4 Pre-processing data: placing the sketch data into GIS ... 44
3.3 The soil deflation and inflation model ... 47
3.3.1 Modelling ... 48
3.3.2 Results ... 52
3.4 The proximity to terrace wall model ... 55
3.4.1 Modelling ... 55
2
3.5 Application of other data ... 64
4. Discussion ... 67
4.1 Discussion of the spatial analysis results ... 67
4.2 Conclusions ... 69
5. Summary ... 74
Literature ... 76
Appendix I: Overview survey units of the pilot study area ... 83
3
1. Introduction
Survey archaeology has become a worthwhile discipline to study the material remains of the past. Initially the lesser cousin of the excavation, the more intensive and systematic landscape-oriented approaches of the 1980s onwards (Attema et al. 2010, 15) have propelled the discipline into a modern and respectable scientific pursuit in its own right. With its maturity however came a range of methodological problems (Sullivan et al. 2007, 322), such as survey bias, field conditions and (post-)depositional processes. Central issue to these problems is: how do we explain the way the distribution patterns of our survey finds were formed?
Explanations for survey distribution patterns have been based on assumptions on soil processes, settlement patterns and depositional factors. One of these assumptions is the idea that archaeological material gets scattered or distorted through post-depositional processes (De Haas 2012, 62), most commonly natural erosion and/or anthropogenic processes. While the effects of post-depositional processes on the archaeological record are not without a basis (Taylor 2000, 23-24; Cavanagh & Mee 2007, 11-12), they often go untested for most archaeological survey projects (De Haas 2012, 62). Since survey archaeology is landscape-oriented and its strengths lies in its large quantities, survey data is typically collected and stored for a particular area (a ‘unit’: see below) as a whole (Banning 2002, 76). This way of data collecting is usually considered enough for answering most of a projects research questions. However, in trying to get a more in-depth understanding of the effects post-depositional processes have on the archaeological record, a more detailed account is needed on the exact position of each find. This information more often than not does not exist due to a lack of time, money, interest and resources (De Haas 2012, 77-79).
4 concerns survey data specifically, a ‘site’ should be understood as a concentration of finds in the distribution pattern.
The ‘unit’ is a term that refers to a partitioned area in space that provides the framework for an archaeological field survey. Typically these are geometrical shapes of the same size, most commonly that of rectangular quadrats (a grid). A unit can be non-geometric (Banning 2002, 81) when the terrain is not suitable for geometric shapes. This often happens when one has to deal with small, irregularly-shaped fields. Since survey projects collect and store the data for the unit as a whole, units are the spatial resolution one has to operate in (Banning 2002, 76).
The issue of post-depositional processes is of great importance for the field surveys carried out by the GIA (Groninger Institute of Archaeology) in the Raganello basin in Northern Calabria, Italy (fig. 1.1). The two main projects are known as the Raganello Archaeological Project (RAP: 2000-2010) and the Rural Life Project (RLP: 2010-2015). They are under the supervision of dr. Martijn van Leusen. The two projects have produced a large amount of data on
5 protohistoric (mainly Bronze Age) pottery distributions. Personally, I have been involved with the RLP only, but in quite a capacity. I have participated in five fieldwork campaigns ranging from two weeks to two months. Work within these campaigns include field (re-)surveys, geophysical surveys of various kinds (primarily with the magnetic and magnetic susceptibility instruments) and small-scale excavations of a number of protohistoric sites. Outside of these campaigns (and not counting the work for this thesis), I have worked on processing and interpreting some of the geophysical data and on studying the properties of soil samples that were collected in a laboratory. This has made me familiar with the region and aware of the methodological problems surrounding the data.
The fact that the two GIA projects are aimed at the protohistory makes the role of post-depositional processes more important than the average Mediterranean survey. Most archaeologists that operate in the region are interested in the later periods with a higher site complexity. Even when small and simple sites are found, there is a preference to focus on the minority of special sites with a richer assemblage (e.g. temples, cities, cemeteries) (NWO 2013). The theories and methodologies in Mediterranean surveys are developed with these complex sites in mind. Protohistoric sites in the Raganello basin on the other hand consists primarily of small surface pottery scatters spread out over the landscape (Attema et al. 2010, 91). The implications of the distortional effect of post-depositional processes can therefore be greater for protohistoric sites then their later successors. The smaller sizes of these sites means this needs to be explored at a higher spatial resolution, such as within a survey unit. This does not indicate that the problem should be ignored for the other periods. Post-depositional processes are after all non-discriminatory, and while they operate for a shorter time the later the period is, the archaeological record is still affected by them.
6 researched by both the RAP and RLP. The following section discusses the pilot study area in more depth.
1.1 The pilot study area
The pilot study area, the Contrada Damale, is an area located on the foothills of the Serra del Gufo, a mountain overlooking the coastal plain of Sibari (figs. 1.21 & 1.3) (De Neef et al. 2012, 15). It has been categorized as forming a part of the marine terraces of the Raganello study area, a succession of step-wise platforms and associated alluvial fan delta-deposits. These are the result of a combination of tectonic uplift and sea level fluctuations during the Quaternary (Feiken 2014, 25-26). The ground consists of old fluvial sediments and debris from the nearby limestone.
1
The (hillshaded) DEM (or layers that are derived from it) shown in all the figures of the Contrada Damale in this thesis is based on LiDAR data supplied by KUL/VITO (University of Leuven/Vlaamse Instelling voor Technologisch
Onderzoek) in 2008 to the GIA. The DEM itself is taken from the GIS database of the RAP & RLP with permission
from dr. Martijn van Leusen.
7 The clayey soil contains a lot of stones and is primarily used for the production of olives and grain. A defining feature are the agricultural terraces that were built on the slopes of the Contrada Damale (De Neef et al. 2012, 15). They are still in use to this day and have an important role in the demarcation of field boundaries and the formation of the current slope. With the advent of mechanized ploughing the local elevation differences between field boundaries in the Contrada Damale have increased (De Neef et al. 2012, 15-16). The agricultural terraces are a main focus of this thesis.
The GIA has conducted extensive field surveys of the Contrada Damale and identified a number of sites (fig. 1.2). As the fields are small and irregular, most of the survey units conform to their shape. There are just a few units that are of a rectangular quadrat. The surveys are a continuation of research interests by the GIA in the Sibaritide and its hinterland since 1991. The initial focus was on excavations of the ancient settlement of Timpone della Motta, a hill on the northern bank of the Raganello river and inhabited since the Bronze Age. Over time researchers became interested in the surrounding area of the site, knowing that the catchment area and hinterland of the settlement could be just as important for understanding the past. The RAP had extended the research area, covering the entirety of the Raganello basin from the Sibari coastal plain to the valley that runs inland towards the Pollino Mountains (fig. 1.1) (Attema et al. 2010, 81-82). The survey results of the RAP showed a large amount of evidence for protohistoric
8 habitation of the hills and mountains, signaling a long-term period of use during the Bronze Age. This includes cave sites and open-air sites up to an altitude of 1600 meters (Attema et al. 2010, 103). The protohistoric activity in the hills and mountains became the area of investigation for the RLP.
The most common identifier for protohistoric activity in the Contrada Damale (and the Raganello basin as a whole) is a material known as impasto. This is a general term used for protohistoric handmade pottery. The impasto from the study area is characterized by its red and black coloring and impure composition (fig. 1.4). Most of the impasto pottery found during surveys is broken and severely worn, sometimes not larger in size than a thumbnail. The majority of the material is dated to the Middle through Final Bronze Age (1700 – 1000 BC) (Attema et al. 2010, 91-94; see table 1.1 for the full chronology). This thesis uses the term Bronze Age to cover the entire period for simplicity’s sake, but do note that the Early Bronze Age (2300 – 1700 BC) is most likely not covered with the impasto material.
The Bronze Age people living in the Contrada Damale occupied small rural structures,
9 sometimes clustered in groups of two or three. No evidence for a larger nucleated settlement has been found. They most likely lived off what they produced themselves through husbandry. It is unclear how the interregional connections were established between the foothills and the coastal area, in part due to a lack of information on habitation of the coastal plain (if there was any). This is because the possible pre- and protohistoric material in the Sibaritide is covered with a thick alluvial layer and can therefore not be found on the surface. There is evidence to suggest that the region became part of a larger exchange system during the Recent and Final Bronze Age, which saw more pressure on the local agricultural system. This could explain the increase of material from this period in the Contrada Damale (Attema et al. 2010, 89, 91-94). The foothills seem to have been depopulated during the Late Iron Age and into the Archaic period, as hardly any material from those periods was found (Feiken 2014, 38). It is not until the Hellenistic period that the foothills are occupied again (Attema et al. 2010, 103).
The fact that Bronze Age material can be found on the surface means that at least some of the archaeological material is within reach of the plough. This is supported by evidence collected on the Contrada Damale. Corings and profile pits show that several protohistoric layers can be found at varying depths within the first meter of the soil. The parent material (C horizon) is at most at a depth of 1.5 m, and usually much less than a meter. There are also occurrences of the bedrock appearing close the surface due to erosion, sometimes even at a depth of +/- 0.25 m (Sevink & Den Haan 2012). Geophysical research with the magnetic gradiometer has also
Name main period Name period Name sub period Date
Protohistory Bronze Age Early Bronze Age 2300 – 1700 BC Middle Bronze Age 1700 – 1350 BC Recent Bronze Age 1350 – 1200 BC Final Bronze Age 1200 – 1000 BC
Iron Age Early Iron Age 1000 – 850 BC
Late Iron Age 850 – 750 BC
Colonial Archaic period 750 – 480 BC
Classical period 480 – 325 BC
Hellenistic Early Hellenistic 325 – 200 BC
Late Hellenistic 200 – 30 BC
10 revealed a number of features associated with the protohistoric period (De Neef et al. 2012, 17-19). Since the instrument is generally only capable of picking up signals of at most one meter in depth, it means the features are located within the first meter of the soil. This does not mean that there are no protohistoric features outside of the gradiometer’s range, but based on the depth of the parent material, this seems unlikely. Even if we are to assume that some features were dug into the parent material, the evidence shows that a major part of the protohistoric material is within the first meter of the soil. This means that the archaeological record is at great risk by post-depositional processes, in particular ploughing.
1.2 Problem analysis
Post-depositional processes have an important role in the formation of the amount of material that can potentially be found on the surface. Regarding the analysis of survey data, the problem exists in the extent of the displacement through these processes. At what point become the patterns in the distribution too unrecognizable to be able to identify it as a site, or at least as something that can be meaningful attributed to any activity in the past? And even when patterns can be recognized, how do we know that there is a direct correlation with an archaeological source beneath the surface? These issues are especially important when considering a sloped agricultural landscape, such as the Contrada Damale, where the effects of post-depositional processes are stronger due to the slope and anthropogenic influences. The downslope movement of artefacts is a real possibility (Cavanagh & Mee 2007, 12) and with it a range of interpretative problems.
11 What is needed to test the two models is the exact position of individual finds from a survey, as that is the only way to determine the actual displacement of artefacts. The usual survey method of collecting and storing data, that for the unit as a whole, is not enough: it is the movement of artefacts within a unit that is of interest. The survey data from the RAP and RLP is for the most part collected in the same way as most surveys: as a quantity in a unit. What makes this particular data set somewhat different is that during the collection process attention was paid to the exact positions of impasto finds (impasto being the most common type of Bronze Age pottery in the region). Because of the usual constraints of time, money and resources, the positions of these finds could only be recorded through marking them on a sketch plan of the unit or grid. This type of ‘sketch data’ is not ideal, but it is data nonetheless and the closest one could feasibly use in order to determine the exact position of pottery. It follows then that part of this thesis is concerned with finding out whether this ‘sketch data’ is a reliable source for a spatial analysis.
1.3 Research questions
The research questions set out by this thesis are the following: Main question:
What post-depositional processes can explain the distribution of Bronze Age pottery in GIA survey data from the Raganello basin?
Sub questions:
- What models exist to explain the post-depositional processes that affect the distribution of pottery, in particular those of a similar period and region as the protohistory in the Raganello basin?
- Which processes affect the pottery distribution in the Contrada Damale and how?
- What is the reliability of the available survey data in relation to determining the exact positions of individual finds?
12 1.4 Structure thesis
13
2. Literature study
The following is the literature study of this thesis. First some concepts about survey archaeology are discussed, especially the terms ‘site’ and ‘off-site’. Then the post-depositional processes are described in a more general context, before focusing on two of these processes that are important for the study area, namely ploughing and agricultural terraces. In here the potential effects of these processes on the archaeological record are explored. An overview of how other researchers have tried to deal with the problem makes up the final section of this chapter.
2.1 Survey: site and off-site
Archaeological field survey at its core can be defined as “studying the distribution of surviving features, and recording and possibly collecting artefacts from the surface” (Renfrew & Bahn 2008, 95). Though the practice of field survey has been around since the nineteenth century, for the longest time it was seen as a method to locate sites to dig, not as an important research tool on its own (King 1978, 4). The first major changes to this line of thinking happened in the 1960s and 1970s (Cherry 2004, 23), but even then the views on archaeological survey were rather simplistic. There was a generally held assumption that a direct correlation existed between the spatial distributions of survey finds in the field and the spatial distributions created by the behaviour of people from the past (Ammerman 2004, 177). Archaeologists have been properly challenging this assumption from the 1980s onward, though this also led to scepticism by some about the value of the data (Banning 2002, 10). Others on the other hand argue in favour of the discipline, not by focusing on its weaknesses but on its strengths (Cherry 1984, 117). An increasingly strong emphasis is placed on the role of archaeological survey and its importance to regional studies (Cherry 2004, 23-24).
14 is especially important for protohistoric societies such as the people from the Bronze Age in the Raganello basin, where rural life was effectively the only life.
The nature of rural sites is very different from urban sites. Add to this the factor ‘time’ and the changing complexities of societies (Renfrew & Bahn 2008, 178-181), the debate about what the term ‘site’ even refers to opens up. This can range from an entire settlement or graveyard to a simple storage pit or even a find spot. When dealing with survey archaeology, the finds are more dispersed and a site commonly refers to a minimum number of individual finds on the surface (typically pottery or flint) within a small area. What this means though can differ from person to person. In his study of the Biferno valley, Barker (1995, 138) sharply differentiates between find spots (1-5 finds), possible sites (6-20) and sites (20+). Bevan and Conolly (2004, 129) on the other hand, for their Kythera survey, give a more vague description, using ‘site’ as “a convenient and shorthand term to refer to clusters of artefacts that on the basis of their composition and contextual association, are assumed to represent the visible material remains of either short or long-term, and often multi-phased, places of human settlement.” Prieto (2011, 75) broadens the term by defining a site as a locus of deliberate human activity, including both artefacts and ecofacts as possible indicators. Ultimately, the definition of the word ‘site’ is a matter of interpretation (Alcock et al. 1994, 138) and the purpose one has for it.
15 The term ‘off-site’ implies a certain disconnect from the site ‘proper’ and thus a disconnect from deliberate human activity. This makes it seem like nothing of real importance can be gained from studying this material, but is this true? Several models have been brought forth to explain off-site distributions by Bintliff and Snodgrass (1988), which have become rather popular in the years following (De Haas 2012, 60-61). A funny explanation (‘model one’) is that of the “pot that fell of the donkey’s back”, in other words material that was left behind because it got broken or lost en route. While this has undoubtedly happened, it is too incidental of an occurrence to be the major cause of all off-site distributions. The second model relates off-site distributions to areas used less intensively then permanent occupation sites. This interpretation is suited for a pattern with some degree of local concentrations, indicative of minor activities. It has been tested and found convincing for off-site material from North American surveys (Bintliff & Snodgrass 1988, 507-508), but it has been rather easily dismissed for Mediterrenean surveys. Bintliff and Snodgrass (1988, 508) discuss the models in relation to their “carpet-like” distribution, but do not detect local concentrations and are therefore not considering it. More curiously, Alcock et al. (1994) do not discuss the model beyond mentioning it, even though they themselves bought up that local concentrations exist. Yet this model might warrant further examination as a possible explanation for off-site material. It has been brought forth that there are questions surrounding minor non-permanent activities in protohistory, for example that of pastoralism (Van Leusen & Attema 2002, 20). More extensive study of off-site material in relation to this second model could therefore provide some answers.
16 The fourth model is known as the ‘manuring hypothesis’. It proposes that farmers collected manure and, together with household rubbish, regularly spread it across their cultivated fields as fertiliser. The ceramic remnants of this practice is what we find in the fields today. Historical sources attest to the existence of this (Bintliff & Snodgrass 1988, 508). Parallels of this interpretation can be found in Roman and medieval north-western Europe (Alcock et al. 1994, 143). The manuring hypothesis has become quite popular as an explanation for off-site distributions (De Haas 2012, 62), but at the same time received plenty of criticism. Alcock et al. (1994, 166) argued against attempting to explain off-site material in terms of a single mechanism such as the manuring hypothesis. One should explore multiple options on a case-by-case basis. Even if we were to assume that the cause is depositional, there are a number of other cultural processes besides manuring that could account for the distributions. These processes include, but are not limited to, cultivation practises, pastoralism and industrial activities (Barker 1995, 46). There are also doubts about the availability of manure in more isolated ecological settings or about regions where the availability of water, not nutrients, is the limiting factor on crop production (Cherry et al. 1991, 50).
It should be noted that Bintliff & Snodgrass (1988, 506) base their hypothesis primarily on materials from the Classical and Late Roman periods in Boeotia, Central Greece, which are markedly different from Bronze Age Italy. In fact it is often presumed that manuring is an indication of intensive agriculture, associated with urban centres (Given 2004, 14). So even if manuring might be a satisfactory explanation for later periods, it might not be for protohistoric rural societies. On the other hand, if other types of processes are involved in the creation of Bronze Age off-site material, then it can be reasonably deduced that they may also apply for other periods.
17 concentration of survey finds that are indicative of some correlation with an original source beneath the surface.
2.2 Post-depositional processes
Site formation processes, both on- and off-site, are divided into depositional and post-depositional. The depositional processes (‘disposal’) can be considered in rate, duration and manner. Abandonment or burial for instance is different than a long-term period of manuring (Given 2004, 19). Understanding the way material was left behind is important in trying to understand the archaeological record, but it is just as important to be aware that material is never found in exactly the same way as it was deposited. Even when a site is still in situ, post-depositional processes will have filled it up, altered its composition or destroyed and moved parts of it. Transport, transformation and destruction are the main components when talking about the post-depositional processes that affect the archaeological record. Concerning the spatial patterns of archaeological surveys, the main factor is the transport of material. The other two components do play a role in the pattern creation process, but mostly in terms of the absence of material. (Banning 2002, 72).
An important subsection of the post-depositional processes are those that are geomorphological. These are typically divided into sedimentation and erosion. They are complementary and can occur at different scales (Feiken 2014, 27). Depositional regimes of sedimentation can cover artefacts and sites (Lewarch & O’Brien 1981, 301) and can contain artefacts from elsewhere as well (Given 2004, 18). The two main forms of sedimentation in hilly and mountainous landscapes are colluviation and alluviation. Colluvial deposits are the result of erosion due to heavy rainfall, forming a loose heterogeneous mixture at the base of slopes. Alluviation is the landscape forming process by rivers and creates well-sorted and homogeneous material (Feiken 2014, 30).
18 ones related to gravitational forces of the material itself. These processes occur without the direct aid of water, ice or air. Examples include landslides and debris flow. A main difference between water erosion and mass movement is that the first is often a more gradual process, whereas the later typically transports larger amounts of material in a short time (Feiken 2014, 28-30).
Human interference into the environment has a strong influence on erosion, typically increasing its effects. The intensification of land use in particular has gained a great deal of attention. Geo-archaeological research that focusses on erosional effects in the past deals with how this relates to land use practices, rural economies and population sizes (e.g. Van Andel et al. 1990; James et al. 1994; Wilkinson 1999). This is flipped for geo-archaeological research into modern land use, as it tries to explain how it affects erosion and/or spatial distributions of archaeological material (e.g. Boismier 1997; Schörner 2012). The increase in availability of GIS has allowed archaeologists to use erosion models, but it was found that the ones developed for non-archaeological purposes were lacking (Conolly & Lake 2006, 203-204). Many make use of the RUSLE (Revised Universal Soil Loss Equation) (Feiken et al. 2011), such as its application for the Zakynthos Archaeological Project (Gouma et al. 2011, 2715) or its adaptation for the Troina Erosion Model (Fitzjohn & Ayala 2011, 9). Only recently was a model, called CALEROS (CALabria EROSion model) developed that moves away from RUSLE and it is the first to simulate long-term landscape developments (Feiken 2014, 125). While these models often do what they are supposed to do for their specific area, they are either not applicable to other regions or (in the case of CALEROS) this has not yet been tested. They also give insufficient information on the effects of erosion at a higher spatial resolution.
19 processes. Certain authors (e.g. Lewarch & O’Brien 1981, 308-311; Bintliff & Snodgrass 1988, 508; Banning 2002, 73) are dismissive of their effects, but they typically deal with large and complex sites/landscapes and are not really bothered with the variation within a unit. For small protohistoric sites however, such as occur during the Bronze Age of the Raganello basin, the influence of localized human activities may be greater than some suspect. Ploughing and terracing are the two main aspects that are theorized to have a strong impact on the formation of the survey data in the pilot study area and they are discussed in more detail below.
2.3 Ploughing and the plough zone
Cultivation has an impact on the distribution, size and condition of artefacts (Cavanagh & Mee 2007, 12), primarily through the process of ploughing. The practice of ploughing has been around since ancient times, but the effect of pre-modern equipment on the soil is relatively unknown due to a lack of information on the subject (Boismier 1997, 10). It has been recognized that the disturbance of archaeological sites increased since the advent of mechanized ploughing (Lewarch & O’Brien 1981, 312-313). This is because the modern plough reaches a greater depth and has therefore a greater potential to disturb and destroy archaeological layers (Taylor 2000, 17). With the existence of large-scale ploughing, the notion of the ‘plough zone’ started to emerge.
The plough zone is the entire layer of the soil that has been affected by ploughing. It is through ploughing that the archaeological record enters the upper layers of the soil and subsequently makes its way to the surface (Taylor 2000, 16). This is one of the reasons archaeologists prefer to survey on a ploughed field. At the same time, repeated processes of ploughing rebury artefacts that had been brought up to the surface, meaning that the distributions are never the same after each ploughing event. Artefact distributions on the surface form only a fraction of the entire population in the plough zone (Ammerman 1985, 34), with the majority of estimates ranging about 5-6%. The material in the plough zone in turn forms only a small section of the entire archaeological record (Schörner 2012, 34). This often leads archaeologists to wonder: how do we extract meaning from plough zone material (see: Francovich & Patterson (eds.) 2000)?
20 extent of the displacement. At what point does ploughing disrupt the distribution patterns too much? There is a tendency for archaeologist to minimize the effect, often assuming that after a while equilibrium will occur (Odell & Cowan 1987, 460) even though there is no actual evidence to support this (Yorston et al. 1990). Experimentations and simulations have been carried out in an attempt to deal with the problem, but they do not all come to the same conclusions. This is in part due to the variance in both design and duration (Boismier 1997, 2).
An often-quoted experimental study by Odell and Cowan (1987) made use of exactly a 1000 numbered flaked stones, weighted to the nearest 0.1 g each. The stones were all measured and painted blue for maximum visibility. They were then planted in a level field about 10-15 cm below the surface in a grid at 50 cm intervals, thus knowing the exact ‘starting’ position of each stone. What followed were fourteen ploughing events over a period of two years. A survey was carried out after each ploughing event and all finds were recorded, but left in place for the next round. They also recorded the type of plough used and the ploughing patterns employed. What they found was that the average cumulative horizontal displacement of the material was just over 2 m. The displacement was slightly stronger in favour of the east-west ploughing directions, but not by much. This study can be seen as evidence that ploughing does not disrupt the distribution patterns too much, but it can hardly be applied universally. Odell and Cowan make use of stones, not pottery, which has different sizes, weights and densities. More importantly, the experiment was carried out on a level field, not a terrain with elevation differences. The potential impact the elevation factor can have is thereby neglected. As a last point, while fourteen ploughing events is relatively speaking a large number compared to most studies, a two-year period is still rather short. Considering that modern ploughing has been around for decades, its long-term effects are still unaccounted for. Even Odell and Cowan (1987, 468) themselves admit that an equilibrium of sorts had not yet been reached.
21 events there were exactly in the six year period, which is the weak point of this study. Ammerman estimates that there were between three and six events before the first observation. The results are rather different from Odell and Cowan (1987). At the first observation alone, tiles had already moved up to 4 m from their original positions, with an average of 1.36 m. This average had increased to 2.19 m by the third observation. At the end of the fourth observation, most tiles were no longer within 2 m of their starting positions, though few had moved as much as 5 m. The tiles had generally moved along the east-west ploughing direction, which runs at a right angle to the slope. At that point it was found that downslope movement did not affect the displacement too much; up to 1 m at most. This view changed when a different ploughing direction was taken before the fifth observation. The farmer had now ploughed north-south downslope. The majority of the tiles had moved more than 5 m downslope, some even more than 15 m. By the sixth visit (which was brief), only seven tiles were still visible, but all of them had moved at least 15 m downwards. It became clear that even a single episode of downslope ploughing could alter the distribution patterns greatly. One can only imagine what the effects would be if the downslope ploughing was followed by heavy rainfall, which would have caused a large event of soil erosion. What makes Ammerman’s study difficult to compare with Odell and Cowan is the depth the material was placed in. Even though the results make it clear that displacement through ploughing can be great, the results may be different if Ammerman had placed the material a few cm deeper.
22 position: In a 200 year simulation of a real artefact scatter, the centre of the distribution had moved away from its subsurface features and a second cluster had emerged to the north of the main one. These results show that ploughing has a continuous dynamic influence on artefact distributions. The notion that an equilibrium will occur appears to be false.
Boismier (1997) has modelled the effects of ploughing as well, creating a computer simulation model called TILLSIM (Tillage Simulation). It incorporates information from both archaeology and agricultural engineering. TILLSIM consists of three components, namely pattern characteristics (artefact size, spatial position, etc.), field properties (field size and slope) and process attributes (displacement and surface occurrence). The simulation was run on artificial point patterns. There were quite a large number of results, but broadly speaking it was found that surface pattern characteristics are primarily the result of slope gradient and the number of simulated ploughing events. Just as Yorston et al. (1990) showed, artefact dispersion by ploughing is a continuous process. Slope processes created a systematic downslope bias into the patterns. This effect only increased with an increasing slope gradient. The size of the artefact did not seem to matter, with different size classes being displaced similar distances. The number of artefact clusters decreased over time, meaning that the identification of archaeological patterns will continuously become more difficult. Boismier postulates (p. 223) that closed artefact populations contained completely within the plough zone will have lost their integrity of cluster composition within a period of 25 to 50 years of ploughing.
As shown by these studies, it appears that survey material increasingly disperses over time by ploughing. This effect is strongly enhanced by the presence of slopes, where material moves downwards relatively fast. This does not mean that survey material is inherently flawed. As long as there is still an archaeological source beneath the soil that gets periodically touched by the plough, new material will continue to enter the plough zone. Even if the entire archaeological record is within the plough zone, concentrations can still be detected. What this does reveal though, is that loose dispersed material that is typically categorized as ‘off-site’, may come from some type of archaeological site after all. In areas with extensive ploughing, the off-site patterns may not be the cause of manuring, as suggested by Bintliff & Snodgrass (1988), but the direct result of human-caused post-depositional processes.
23 comm., 2014). Widespread use of more advanced methods of agriculture began in southern Italy in the 1980s (Prieto 2011, 72). Presumably this was the same for the Contrada Damale. This is why full dispersion of the archaeological material has likely not happened yet. The question remains to what extent ploughing does affect the distribution patterns in the study area. This is explored further in chapter 3.
2.4 Agricultural terraces
A defining feature of the Contrada Damale are the terraced fields. Terraces are an important economical investment in agricultural landscapes around the world, particularly in tropical to semi-arid environments. They are constructs to change hillslopes into stepped areas of relatively flat ground (Bevan & Conolly 2011, 1303). This has the additional effect of improving cultivation by reducing soil erosion, the collection of water in their hydrological infrastructure and the preservation of vegetation (Koster et al. 2011, 1). Terraces can vary widely in shape, complexity and degree of labour involved (Bevan & Conolly 2011, 1303). The terraces in the Raganello area are described by Koster et al. (2011, 2) as followed, which they base on a classification in Frederick et al. (2000):
- Braided terraces, which have ends that slope together to form a continuous zigzag slope, mostly used for cereal cultivation
- Pocket terraces, i.e. semi-circular benches used for the cultivation of trees
- Parallel terraces, which follow the contours of slopes and are used for a variety of products
As is a common problem with agricultural terraces, it is not really clear how the Raganello terraces were constructed or even how old they are. It is stated that the first terraces in southern Italy were constructed during the Iron Age with the arrival of the Greek colonists (Koster et al. 2011, 1). This could be the same for the Raganello basin, but there is no direct evidence for it. There is much debate to what extent the current terraces still resemble the original terraces. Additionally, there is a lack of knowledge on the composition of the terraces and what archaeology is preserved within (Koster et al. 2011, 1-2).
24 it seems that terraces were in a constant cycle of decline, repair/construction and maintenance. This process is generally associated with the increase and decrease of populations: the larger a population becomes, the more arable fields are needed, which results in more soil erosion, thus the bigger the incentive to maintain terraces. Once a population declines again, the terraces decline with it (Wilkinson 1999, 183). However, this is not necessarily true. From data on Antikythera (admittedly from a more recent period, namely late-18th to mid-20th century AD), it has been shown that a population increase did not result in an intensification of terracing. On the contrary; more terraces were built during a population decline. It appears that political stability and favourable economic conditions were the deciding factors in terracing. Which makes sense, as constructing terraces would require some organizational effort.
Since terraces are notoriously difficult to date, there is no direct archaeological evidence for a relationship between population growth and terrace construction (Bevan & Conolly 2011, 1304-1305). What is known is that without repair and maintenance, a terrace will eventually lose its structural integrity and collapse (Van Andel et al. 1990, 383.) This means that, while population growth may not have been the incentive, something caused the people of the past to maintain them. The fact that they still exist today attests to this.
But how does the construction and maintenance of terraces affect the archaeological record, particularly that which existed before the terraces were built? Assuming that the first terraces indeed appeared in southern Italy during the Iron Age, their very construction could have been disastrous for the Bronze Age layers beneath. In building the terrace, soil must be removed and deposited elsewhere. The archaeological material within that soil is moved with it. Even if we are to assume that the removed soil is re-used in the construction of the terraces, or deposited close-by, the integrity of its archaeological source remains lost. At the same time, agricultural terraces can protect the archaeology within or beneath, due to their relative low susceptibility to erosion (Koster et al. 2011, 2-3). Artefacts move around in the soil, particularly the plough zone, due to erosion and ploughing. Upon reaching the terrace wall, artefacts are stopped in their tracks. By restricting the movement of artefacts, terraces can thus actually serve to conserve sites rather than destroy them (Cavanagh & Mee 2007, 12). In seems then that if archaeological layers survive the initial construction of terraces, they will continue to exist.
25 remain in the same field. Terraces create, as it were, a micro-environment. The potentially disastrous effects of ploughing set out in the previous section can therefore be limited. Artefacts can still disperse quite strongly within the same field, but rarely moves beyond it. This is assuming of course that the terraces remained in decent condition and no disastrous erosion events have occurred. Another point of interest is that typically not every field in a terraced landscape is necessarily demarcated by terraces. Soil and artefact displacement through erosion and ploughing can still have free reign in these type of fields (Bevan & Conolly 2011, 1312).
With agricultural terraces functioning as a restricting factor in the movement of soil and artefacts, it is possible that archaeological material ends up cluttering behind the terrace walls. This can create a new reservoir of material in the plough zone that may not have been there before the post-depositional processes came into effect. Such a reservoir can be the cause of a concentration in the survey distribution patterns near a terrace wall. These concentrations could subsequentlybe identified as sites, even though the material originally came from elsewhere. Whether this is something that actually happens remains a matter of debate and will be explored by the spatial analysis of this thesis.
2.5 Comparisons: How others deal with the problem
The problem of the effects ploughing and terraces have on archaeological distributional patterns exists for many archaeological surveys in the Mediterranean and beyond. In order to give more context, this final section of the chapter goes into how other people have tried to deal with the problem.
26 An example of a Mediterranean survey project that does not explicitly discuss the issue in its publication is that of Northern Keos (Cherry, Davis & Mantzourani 1991). That this is the case for this particular project is interesting, as it is otherwise considered very thorough in trying to understand how its data set came to be. Its site collection strategies and important matters like visibility are well documented. Some of the contributors themselves (particularly Cherry) have discussed the problem in other works. That the effects of post-depositional processes are not mentioned in the main work for Northern Keos may be indicative of the lack of importance that is generally placed on them.
Barker (1995) and his team did an extensive survey of the Biferno valley, an area in Molise, central-southern Italy. The survey encompasses all periods, from prehistory to modern times, and was set out to document the long-term settlement history of a typical Mediterranean landscape (p. xvi). The project’s primary concern is that of the development of human settlement, not that of agricultural practices. Nevertheless, Barker acknowledges the impact ploughing had on his survey record. In order to get a grasp of the scale of the problem in the Biferno, they resurveyed a four km2 area in the upper valley which they had surveyed a few years previously. Focusing on the Samnite and Roman periods (and deliberately excluding off-site data), Barker notes many discrepancies between the two sets of data. But as he was more interested in the regional scale, his overall principal feature of dense settlement near a specific landmark still upheld. In his view then the effects of ploughing are minimal, as surveys could still provide critical data for the understanding of long-term settlement history (p. 49-51). Something he seems to ignore though is that the effect can be different for prehistoric and protohistoric material and there is no reflection on the issue of generalization between different periods.
27 indiscriminate scatter by 2001 (p. 74). The problem was initially bypassed by choosing in advance not to survey those areas where displacement was too great (p. 77). Over the years however, they became aware that the problem could not be ignored and a resurvey project was set up, which reexamined sites surveyed in the twenty years previous (p. 88). While this data could have answered some interesting questions regarding the effects of ploughing, their concern lay mainly with the recovery rate of sites (p. 91). It was found that in the twenty year period, roughly a quarter of the sites had been lost. There remains a sense of optimism though, as it is pointed out that most of these sites would not even have been discovered without ploughing in the first place (Thompson 2004, 81; 83). There seems to be no further exploration of how ploughing has affected the spatial distribution of a site itself, instead choosing to focus on the area as a whole.
Moving on to the island of Cyprus, two subsequent survey projects have taken place there: the Sydney Cyprus Survey Project (SCSP) (Given & Knapp 2003) and the Troodos Archeological and Environmental Survey Project (TAESP) (Given et al. 2013). The people who were involved with the SCSP were already quite aware of the problems ploughing can cause, as attested by an article by Given (2004, 18), but they do not seem particularly concerned with it. This is a glaring lack of interest, as they were very detailed in recording other elements that affect distribution patterns, such as visibility and recovery rates (Given & Knapp 2003, 49-56). This was rectified for the TAESP, in which they did their own plough zone experiments (Gibson in: Given et al. 2013, 39-41). They looked toward general trends of lateral and vertical movement of tiles, which only gave them results on the recovery rates. The actual movement of individual tiles in response to ploughing remained unexplored. Gibson notes how further research into this is necessary (p. 41), but for the TAESP as a whole there does not seem to be any further critical analysis on how this affects their own data.
29
3. Spatial analysis
The main goal of the spatial analysis is to find out if there is a correlation between the distribution of protohistoric pottery and post-depositional processes, particularly those induced by humans. This chapter outlines the methodology, the data, the processing steps and the results. All GIS processes discussed have been carried out in ArcMap, developed by Esri.
It was shown in the previous chapter that people have primarily tried to use experiments and simulations to try to explain the effects post-depositional processes have on the survey data. These are good initiatives and provide us with valuable information. However, rarely is actual survey data tested against real human-induced changes to the landscape. The spatial analysis in this chapter attempts to do this through the development of two models that represent the effects of ploughing and terracing. The models are known as ‘soil deflation/inflation’ and ‘proximity to terrace wall’. The reasoning for using these specific models are discussed in their respective sections.
To actually test the models, the exact positions of individual finds is needed. This type of data is often lacking due to the way survey data is collected and stored: as a quantity of finds within a unit. In a way this is also true for the RAP and RLP, which stores its data in the same fashion. It is only because sketches were made of the approximate positions of the finds that a test of this nature can be done.
30 3.1 The effects of agricultural processes on the archaeological record
With the information from the literature study, several hypothetical scenarios can be devised for the pilot study area on how finds concentration patterns from archaeological surveys can relate to agricultural post-depositional processes. A schematic overview is provided in figures 3.1 through 3.4. All scenarios assume that sites are located in sloped agricultural fields and uses Bronze Age material as the example.
The first group of scenarios (the ‘A’ scenarios: fig. 3.1) is about the relation between the relative depth of the plough zone and the distribution of the archaeological record. A site that is completely within the plough zone (scenario A1) gets completely disturbed by the process of ploughing. In time full dispersion is likely to occur, but it will not happen immidiately. It is therefore still possible to have a concentration of finds at the original position of the site. The center of the concentration might be off when compared to the position of the site due to repeated processes of ploughing in the same direction. A site that is partially within the plough zone (scenario A2) gets only partially disturbed by the process of ploughing. The material that is ploughed up will eventually disperse, but new material will continue to enter the plough zone. This is the most likely scenario that results in a concentration of finds at the original position of the site. A site that is completely beneath the plough zone (scenario A3) will not get disturbed by the process of ploughing. It is not likely that material will enter the plough zone. It should be noted that material from beneath the plough zone may be brought up through other processes (such as construction or digging), but this is incidental. It is therefore unlikely a concentration of finds will be found at this location. Something to be aware of is that the depth of the plough zone is not stable and may change over time through erosional processes, continuous ploughing or the introduction of a new type of plough. The scenario concerning a particular site should for that reason not be considered fixed.
31 of the terrace. The B1 scenario then demonstrates what happens if a site is deep enough to remain untouched by the plough on the upper side of the terrace, but not deep enough for the lower side. In that case a concentration may be found at the lower side, but not the upper side. When the plough zone reaches the site at both sides of the terrace (scenario B2), concentrations can exist on both sides as well. This is possible when the terrace wall is not particularly high. When the site no longer exits at the lower end of the terrace (scenario B3), a concentration is only possible at the higher end. The three scenarios imply that a site gets partially destroyed by the construction of the terrace, but there is always the possibility that the site is left untouched when the terrace is built on top of it. This can either be when the site is located too deep for it to be disturbed by the process or the entire terrace is constructed in its entirety over the slope. The second method of terrace construction would require the use of material from elsewhere to fill up the raised section. Scenario A3 can in those instances be applied, with the added bonus of protection by the terrace. The other end of the spectrum is also a possibility, in that the construction of a terrace completely destroys the site and the material gets redeposited elsewhere.
Figure 2.1: Schematic representation of the relative depth of the plough zone and how this relates to the distribution of Bronze Age (BA) material. Not to scale.
Scenario A1: BA site completely within the plough zone. Full dispersion a possibility, though a concentration can still be present on the surface at the original position of the site.
Scenario A2: BA site partially within the plough zone. Full dispersion not likely, as new BA material continues to enter the plough zone. Likely to be a concentration present on the surface at the original position of the site.
32 Another issue to be aware of is that ploughing rarely happens up till the edge of the terrace wall. The immediate area near the terrace wall may then remain unaffected. When the terrace is constructed slightly uphill from the site, material that was removed for the construction of the terrace can be deposited on top of the site. If this does not happen, then the site is at a greater risk for getting exposed to the plough zone.
Something interesting can happen when a terrace is constructed some distance downslope from the site (the ‘C’ scenarios: fig. 3.3). When a site is not dispersed (scenario C1) the concentration will most likely be at the original position of the site. When a site gets partially dispersed however (scenario C2), there is the possibility the material will be transported downslope. It will eventually come upon the terrace wall and be stopped in its tracks. When this happens enough over time, a significant portion of the site will end up near the terrace wall. Concentrations can then be found at both the original position of the site as well near the terrace.
Figure 3.2: Schematic representation of the construction of a terrace on a Bronze Age (BA) site and how this relates to the distribution of Bronze Age material. Not to scale. Variations of the A scenarios can be applied.
Scenario B1: Terrace cuts partially through the site, removed material gets displaced at unknown location or is re-used in raising terrace. Plough zone does not go through site at upper side of terrace, does at the lowered side of terrace. BA material likely not to enter the plough zone in upper side of terrace, it does at the lower side of terrace. Concentration possible at lower side of terrace.
Scenario B2: Same situation as B1, except that plough zone also cuts through site at upper side of terrace. BA material likely to enter the plough zone on both sides of terrace. Concentrations possible at both sides of terrace.
33 This may give the appearance of two separate sites, when in actuality it all comes from the same source. If the original site is completely disturbed (scenario C3), the concentration may even only occur at the terrace. Archaeologists may then get the wrongful impression that the original site is located there.
Matters get even more complicated when an additional site is added to the situation (the ‘D’ scenarios: fig. 3.4), with site A located some distance uphill from the terrace wall and site B located at the terrace wall. When dispersion is at its minimum (scenario D1) the material remains concentrated at its original position. However, when site A gets partially dispersed (scenario D2), some of its material may intermingle with the material from site B at the terrace wall. Two concentrations could still be present, but the one at the terrace wall will be more profound. Full
Figure 3.3: Schematic representation of a Bronze Age (BA) site some distance away from the terrace and how this relates to the distribution of Bronze Age material. Not to scale. Variations of the A scenarios can be applied.
Scenario C1: Site not dispersed. Concentration likely at the original position of the site.
Scenario C2: Site partially dispersed. Terrace wall restricts movement of dispersed material further downslope. Concentrations possible at both the original position of the site and at the upper side of the terrace.
34 dispersion of site A (scenario D3) will only add more material to site B. Scenarios D2 and D3 are not only problematic in defining the position and quantity of sites, but also cause issues in the dating and classification. The intermingling of sites does not even have to occur near terraces, but can already happen when they are within close proximity. Several conditions need to be met before the D scenarios will occur (e.g. proximity, ploughing duration, depth and direction, other soil processes), but that does not mean they should be ignored.
What all these scenarios show then is that in an area of terraced agricultural fields a concentration of finds does not necessarily correlate to one single site or to its original position. It should be stressed that the scenarios should not be favored one over the other or that all these
Figure 3.4: Schematic representation of two Bronze Age (BA) sites, one some distance away from the terrace, one beneath the terrace, and how this relates to the distribution of Bronze Age material. Not to scale. Variations of the A and B scenarios can be applied.
Scenario D1: Sites A and B not dispersed. Concentrations likely at both original positions of the sites. Material corresponding to its source.
Scenario D2: Site A partially dispersed, site B not dispersed. Terrace wall restricts movement of dispersed material further downslope. Concentrations possible at both original positions of the sites. Material at Site A corresponds to it source and at site B a mixture of both site A and B materials.
35 things necessarily happen. It is to demonstrate that the potential is there for it to happen. It is precisely for this potential that the problem of agricultural post-depositional processes needs to be explored and better understood.
3.2 The data
The data used for this thesis can be divided in two main components: GIS data and survey data. This section outlines the data and assess it for its reliability. In addition it is shown how the survey data, which partially only exists in paper form, is transformed into a useable GIS layer. Some of the decisions regarding the data set were made on the basis of an initial GIS analysis. In order to not unnecessarily break up the discussion of the data, the results of these decisions are shown. The GIS data proper is discussed first, as a part of it was needed for the transformation of the survey data. The use of the data is through permission of dr. Martijn van Leusen and drs. Wieke de Neef.
3.2.1 The GIS data
Certain aspects of the GIS data was created by myself (see below), other data sets (or layers) were already existent. The base data that is used for this thesis is discussed here. The coordinate system used is ‘WGS 1984 UTM Zone 33N’.
A high-resolution digital elevation model (DEM) is essential. This is a term used for models of the Earth’s surface, though technically it can refer to any measurable variable that varies continuously over space (Wheatley & Gillings 2002, 107). A DEM of the surface topography can be used as the base data for a wide variety of models (Conolly & Lake 2006, 101). The DEM for the research set out in this thesis is raster-based and derived from LiDAR data of the Raganello basin supplied by Leuven University (KUL) and VITO (Vlaamse Instelling
voor Technologisch Onderzoek) to the GIA in 2008. LiDAR stands for Light Distance And
36 From the DEM a hillshade can be derived (fig. 3.6). This is a shaded relief of a surface raster over which a theoretical light source has been placed. The azimuth and altitude settings of the light source for the hillshade used by this thesis are set at 315 degrees and 45degrees respectively. The result is a good visual representation of the topography of the landscape. This makes the hillshade a useful visual tool to identify features and locations.
Another important element is the survey units data set, which consists of polygons that cover the areas that have been surveyed. The majority of the units are shaped after the lay-out of the fields. The survey unit numbers can be associated with the data in the RAP & RLP database. Since all survey data comes from within these units, they must be used to mark off the area that can actually be used for the study. All RAP and RLP units within the Contrada Damale were selected and made into a new polygon layer. This left a total of 328 as the base number of units (figs. 3.5 & 3.6).
37
3.2.2 The survey data
During the RAP surveys finds were usually recorded by unit. This is a common method within survey archaeology and usually serves its purpose (that of identifying sites) well enough. However, this thesis requires more information than that, as it focusses on the positions of the pottery finds within the units and its relation to post-depositional processes.
Ideally, the exact positions of all individual finds would have been recorded with the aid of a total station or a similar device. This is extremely time-consuming and it is understandable that it did not happen during the RAP surveys. Yet some recording of the positions of finds occurred through marking the find spots on the sketch plan of the unit. This was done to get a better sense of the visibility problems. Through charting the pottery, a decision could be made on studying a unit further through a total survey. This was only done for impasto (with a few exceptions for special finds), the pottery most commonly associated with the Bronze Age of Calabria.
38 The surveys of the RLP were for the most part resurveys of the RAP units, though it also covers a few fields that were previously not surveyed by the RAP. The project’s aim with these resurveys is to get a better understanding of survey biases. The RLP has produced sketch plans of units, both new and resurveys ones, and both have been used for establishing the exact positions of individual finds. The RLP has also produced exact recording of finds with a total station. This happened on areas that were total surveyed with a high finds concentration in order to understand the exact spatial relations of such a ‘site’. The entirety of the unit is not covered with these recordings. Only a small number of sites were recorded in this matter. The sheer quantity of these recordings concentrated in a small area would unfairly skew the results in their favor. The total survey data is for that reason not used for the creation of the individual finds data set.
Sketches are not the most reliable form of data. They are often inaccurate, simplistic representations of reality, used for reference only (fig. 3.7). This does not make them automatically useless. The idea behind using the sketch data is that with a large enough number of units and individual finds, trends in the data appear. At the same time one should be blind to all its potential faults and errors. The sketch data is analyzed and assessed on its reliability. Units that are deemed unreliable or simply not useful (due to the lack of a sketch) will be taken out of the data set (sections 3.2.3 & 3.2.4).
It is important to be aware that the data is also affected by survey biases2. A whole range of factors influences the number of finds that are done in the field, such as visibility, survey conditions and experience. The importance of survey bias has been recognized early on by the GIA projects and steps have been taken to record the factors involved. Through this it became clear that there is a range of locally significant visibility factors that can be assessed within a framework of both geomorphological and land use/land cover variation. Examples include factors as vegetation, stoniness and the condition of the soil (Attema et al. 2010, 172). Weather conditions are equally important, as the contrast between sunlight and shade changes through the day (Feiken 2014, 36-37) and rain alters the condition of the soil. Then there are the surveyors themselves, whose level of experience (Attema et al. 2010, 172) and physical and mental ‘readiness’ can differ. One person may be inclined to find and identify more material then
39 another, even though the visibility factors are the same. This ‘surveyor factor’ is unfortunately much more difficult to quantify. Then there is the coverage of each field, which can differ per unit. Excluding the total surveys, most units in the pilot study area had a coverage of 20%, though there are instances where the coverage is 40-50%.
To take away from the survey bias and coverage is that the number of individual finds in each unit is always less than the potential number of finds that were present on the surface. This is one of the reasons finds are usually recorded per unit, so that the number of finds can be corrected for coverage (there is no universal method yet for the correction of survey bias). When using the exact positions of actual individual finds from a standard survey, correcting for coverage is not an option. This means that theoretically, units with a higher coverage or better survey conditions are better represented on an individual finds data set. One way to combat this bias is to use a large quantity of units in which the potential skewing effects of the individual properties of a unit are balanced out.
An overview of all the survey units in the Contrada Damale can be found in appendix I. This includes information on whether impasto was found, whether a sketch was made and whether the impasto was marked on the sketch. Other types of information has been included as well, such as the condition (wear) of the impasto, the type and coverage of the survey involved and who filled out the survey form. This information is needed for the pre-processing of the data (sections 3.2.3 & 3.2.4).
40
3.2.3 Pre-processing data: analysis of the units
With both the RAP and RLP, a total number of 327 units were surveyed in the pilot study area. Technically there are 328 units within the GIS units layer, but unit 4165 exists twice. Both refer to the same field, except that one of them covers the entire field, while the other only a part of it. It became clear that the sketch of unit 4165 refers to the smaller of the two. The one that covers the entire field has thus been taken out of the data set. The remainder of that field is still covered by other units, so no actual data was lost.
Due to resurveying of the same fields in later years and then being assigned a new unit number, some of the units overlap. This could favor the fields that have been surveyed more often, but there is not much to be done about this: resurveys did not cover the older units exactly the same and did not necessarily produce better results. It is often the case where most of the finds were marked for the resurvey and not the original survey. Deciding which units to keep or remove would then become too arbitrary. The issue occurs over the entire Contrada Damale and in that sense does not favor one area over the other. The overlapping of some survey units is therefore considered an acceptable bias.
No impasto was found in 72 of the units. This is based on the finds forms associated with the units. As it is just as important to know where no impasto was found as where it was found, these units have been used for the analysis in the GIS.
For 12 units there were no sketches made, though impasto was found according to the forms. Similarly, there were 16 units where impasto was found, but no finds were marked on the sketch plan. It is most likely that the persons filling out these forms simply forgot to do so. Both type of units are not useful, as it is impossible to know where on the unit the impasto was found. They have thus been removed from the data set.
Part of a unit had not been surveyed in three instances. Impasto had been found in the rest of the unit and this has also been properly marked on the sketches. It was decided to use these units for the data set, treating the areas that were not surveyed as simply containing no impasto.
