Programme of the third Joint Chapter Meeting

Archaeology is one of the most “digitized” social sciences. Nowadays almost all data archaeologists register during their work is in a digital form. In the field, in the lab, and during analysis the use of a computer is an absolutely must.
At the age when young people leave high school, they are experts in the use of desktops, laptops, tablets and mobile phones. But when they leave university with a degree in Archaeology are they well prepared for the reality of archaeological work? What are the skills and knowledge we should teach our students to let them confidently face the modern, digital practice of life as a professional Archaeologist?

Since six years the Universities at Mainz conduct a regular joint study program in the fields of surveying, spatial information technology and geophysical prospection with special focus on archaeological applications. The program is offered to graduate students from several archaeological disciplines as well as geoinformatics and surveying and is currently integrated into three different Master-Programs at two different universities.
The talk will discuss experiences gained during the courses emphasizing challenges caused by the special interwoven structure of the program. The multiple dependencies on the one hand demand a strict course schedule and on the other hand disallow a consecutive design of the teaching and learning content. The heterogeneous prior knowledge of the students require special measures to balance workload while obtaining new and reinforcing present competences.
In order to meet these challenges, a modular core curriculum has been developed involving an anticipated archaeological scenario that allows for teaching most aspects on the basis of a common, simple but also narrative and telling data basis not confined to a specific archaeological region or period.
Students with a wide variety of skills and competences have been taught how to analyse this dataset in different learning environments (team work, e-Learning, teacher-centred). Experiences with the different learning approaches as well as the lessons learned during the past six years will be discussed. The settings of the courses within different curricula and the lack of practical application of the newly acquired knowledge in most practical and academic settings are an issue.

Digital archaeology is an integral part of the curriculum at Saxion – University of Applied Science. The students learn to work with software that is most often used by the companies and institutes in the Netherlands. The idea behind this is to enable the students to participate easily in the work process during internships, but also after graduation. Teaching is not only focussed on learning to use the software, but it is also expected in other courses that the students are able to apply the knowledge they gained. Furthermore, the teaching of digital tools is accompanied by courses where students learn the analogue counterpart. They do for example not only learn to document using RTS or RTK-GPS, but are also able to use conventional methods such as measuring band or a prisma. An interesting challenge is to teach students digital tools, while quite a few have the same aversion to these that can be found with most archaeologists. The first part of the paper will focus on the curriculum and more important what steps are taken to ensure that it is always actual without too much changes.
A fascinating aspect of teaching in a world where everyone has a smartphone or uses various digital tools on a daily basis, is that most students have difficulties understanding the basics of computers and software. The more easy technology is available, the less it is understood. The smartphone of the students can also be used to make teaching more interactive. In the second part of this paper I will demonstrate such a tool.

Without the use of IT-based methods archaeological fieldwork is hardly imaginable today. GPS, databases, GIS, UAVs and 3D-scanners are on the rise to replace conventional methods nearly completely. With these new technologies also the demands on the archaeologists changed completely. They have to handle a wide range of technical methods to meet the requirements of a modern excavation or survey.
So it is no surprise that many archaeologists try to specialize in that field in the light of better chances in the tight labor market. But mainly they do this independently without any professional advice. It is surprising that a professional archaeological training is offered at no point in Germany that satisfies the requirements of the potential employers and the desire of students for a specialization in the field of digital documentation and analysis. Previous offers in this field are mostly confined to basic topics and usually they are not an integral part of the curriculum but taught only in specialized and more or less limited fieldschools. As another disadvantage these courses are mostly not conducted by professionals but by archaeologists with a broader user-knowledge.
To face this problem the Landesamt für Denkmalpflege Baden-Württemberg and the eScience-Center of the University of Tübingen started a common program in teaching digital methods in archaeology. This program will allow to specialize on these new methods without neglecting the fundamentals of the different archaeological disciplines from the prehistoric to the near-eastern archaeology. The program is not an own course of studies, it is an extension to the traditional archaeological programs.
In our opinion the direct practical application of the learned techniques and methods in the field is from a very high importance to internalize them. To avoid the limitation of these practical exercises to some short weeks in the summer we decided to concentrate to a nearby study area in the domain of Rottenburg at the Neckar, where we find an extremely wide range of human traces from the Stone Age up to the middle Ages. Thereby it is possible to connect theoretical and practial lessons in a very tight way.
In this paper we will present the current state of the project and the experiences we made so far. Also we will discuss the need of collaboration in this field with other heritage authorities and universities.

The best place for teaching is where the actual material under study is close at hand. Student involvement in archaeological fieldwork is therefore a fundamental part of teaching the archaeological practice, from understanding stratigraphy and recognizing artefacts to documentation processes and methodological issues. However, in the context of archaeological fieldsurvey, transmission of knowledge is often much more difficult, for various reasons. Students are split up in separate teams, roaming the landscape with often miles between them, overstretching teachers and specialists. Pottery sherds are often so weathered that they can only be studied after proper cleaning afterwards. Equally important is that the data gathering process of archaeological fieldsurvey – with often abstract sampling designs – is much less tangible and easy to explain as with archaeological excavations. All this can easily turn an archaeological fieldsurvey project into an undertaking where students do little more than walk their tracts and pick up or count artifacts, ‘encouraging’ a very passive learning stance. The OpenArchaeoSurvey project aims to overcome this problem by making use of recent technical developments and supply teams with devices that allow exchange of data and direct communication between themselves, teachers and specialists using mobile broadband connections. This involves a free open source Python plugin for a mobile QGIS application that uses a client-server setup to exchange collected data, images and text messages. PostgreSQL/PostGIS records, geotagged photographs and chat messages can be synchronized with use of a simple toolbar and assessed within the QGIS interface. As said, where this has obvious benefits for efficient data-management for any fieldwork project, the main goal is to improve the potential for participants to contribute and learn. Specialists can provide direct feedback, complex data can easily be exchanged, results directly analyzed. This provides opportunities for field teams to continuously tap into available expertise. Apart from the possibilities for mobile learning, this can be expected to improve the fieldwork and results in general.

A unique characteristic of the Neanderthal Museum is its role as museum and extra-curricular learn-center for pupils and interested laypeople on the one hand, and its role as research institution on the other hand. Due to that double-function digital media are used in different ways on various scales and with different objectives.
Since 2010 the Neanderthal Museum possesses a structured light scanner (Breuckmann) which is used both for research activities as well as for didactic purposes. In the context of research conducted in the framework of different projects archaeological remains (such as lithic and organic artefacts as well as cave art) and anthropological remains are digitized to document and analyze the respective objects but also for cultural heritage reasons.
With regard to anthropological workshops held at the museum as well as for classes at the University, we started to set up a digital archive of 3D surface scanned, representative artefacts which are to be used for teaching purposes. Recently, a new software (tetra4D) has been utilized to generate PDFs of 3D scanned artefacts to give people easy access to those data in the course of workshops. Additionally, a new software has been developed in close cooperation with the University of Cologne that should enable users to easily visualize and conduct simple analysis of 3D scanned archaeological and anthropological remains.
With its online platform NESPOS the Neanderthal Museum has a database at hand that is used as storage place for 3D objects and allows access to digital data for laypeople and experts likewise.
The presentation will describe the different ways in which digital media are used at the Neanderthal Museum and outline possibilities and restrictions in dealing with digitized archaeological contents for laypeople on the one side and academic colleagues on the other side.

In archaeological computing, methods for calculating similarities and identifying patterns have been proposed and applied for a long time and include a wide range of approaches.
Highlights of such applications sometimes involve sophisticated mathematical models and generate stunning results. After shortly introducing such a highlight application, the talk will focus on fairly simple concepts for calculating similarities and identifying patterns. The adequate application of some frequently used similarity and distance measures will be discussed first as well as different transformations of interval scale data. Such transformations are often applied routinely, but are they really necessary and do they serve the purpose? Often distance or similarity calculations are the basis of a method for identifying patterns, and inappropriate dissimilarity measures may invalidate the results of the method applied. For many archaeological data sets, simple and easily intelligible approaches provide adequate insights into the structure of the data. Detecting the pitfalls of such approaches is much easier than with highly complex black box methods. One of the aims of this talk is to encourage the audience to think about the methodology before pressing a few buttons in one of the user-friendly readily available software tools. Some of the methods applied and discussed in this talk have been in the archaeological toolbox for a long time but it seems quite outdated to use them today. This will be checked by re-analysing a data set comprising of 30 fibulae: An initial study of this data set was published in 1970, the objective at that time was to show the potentials and limits of methods like Principal Components Analysis and several cluster analysis algorithms. Another part of the talk will focus on methods involving local averaging or more technically speaking low pass filtering. This approach is simple and by no means new, but has a wide variety of applications in archaeology. In general, its purpose is to identify the global trend of the data. Due to time limitations, it is not possible to give a comprehensive overview over all methods for calculating similarities and identifying patterns relevant for any archaeological data. Still the talk covers a wide range of applications including comparison and classification of artefacts like pots or fibulae, identifying groups in big data sets with more than 500,000 objects, pattern recognition to detect archaeological features in LiDAR images, least-cost distances and Llobera’s measurement of prominence.
Though the focus of the talk is on fairly simple methods, these are not necessarily well-known. Most of the results presented were created by applying freely available software and a spreadsheet program.

This paper is part of a broader project that seeks to introduce the digital humanities community with a range of computer and quantitative methods that can be applied for a better understanding of archaeological collections. This comprises, among others, the following research subjects:
• Application of clustering techniques for unsupervised learning
• Automatic recognition of artefact features through the application of computer vision algorithms
Our main objective has been to promote generic methodologies, which can be applied to many other projects. In this paper we will focus on a new kind of typological analysis based on a quantitative procedure called Spectral Clustering. This technique uses Graph Theory to analyse the eigenstructure of an affinity matrix in order to partition data points into disjoint clusters. The original algorithms were developed a decade ago by mathematicians and machine learning professionals. To the best of our knowledge, this technique has not been applied before in archaeology despite its proven performance in partitioning a collection of artefacts into meaningful groups.
We argue that Spectral Clustering yields better results than more traditional approaches such as single and total linkage numerical taxonomy, as well as k-means. The technique is especially useful in exploring archaeological collections because it is especially suitable to analyse categorical data.
As a study case we choose a collection of stone masks found in the remains of the Sacred Precinct of Tenochtitlan, the main ceremonial Aztec centre, located in Mexico City. The schematic features of these objects set them apart from other artefacts with more naturalistic style. This has attracted the attention of many specialists and during the last three decades the style of these items have been the subject of intense debate. Through the application of Spectral Clustering we were able to segment this collection into well-defined groups. In the future, this could lead to a better typology of this collection.

The complexity of scientific data in archaeology calls for elaborated methods for the automatic extraction of knowledge from such sort of data. Data Mining offers a variety of techniques for data understanding and pattern extraction with supervised learning (clustering of data instances into groups of similar instances) and unsupervised learning (classification of new instances into predefined categories) being among the most popular ones. The vast amount of work in Data Mining is on pattern extraction algorithms, however the effectiveness of these algorithms heavily depends on the chosen feature space representation; redundant features for example might lead to a deterioration of the performance of the extracted models.

In this work, we deal with the problem of evaluating whether and how the inclusion or exclusion of a feature affects the extracted data mining patterns. Our data stems from isotope measurements of animal findings, namely oxygen, strontium and lead.

The data was collected as part of the DFG FOR 1670 (“Transalpine Mobilität und Kulturtransfer”), which focuses on cremated finds in the Central European Alps region as data source. The motivation for this study is that oxygen isotopes are not stable through cremation and therefore, measurements of the oxygen isotope in samples from material that has been subjected to cremation cannot be used for further analysis. Our question is whether the cremated finds can be used to establish the same patterns as the uncremated ones, given that oxygen after cremation is not usable. Understanding the effect of the cremation upon the extracted patterns and whether such a procedure incurs loss of information is of great importance for the project and in general. In this work, we examine the effect of oxygen omission in patterns extracted through supervised and unsupervised learning tasks, namely clustering and classification models, by establishing a mapping of the results with and without oxygen isotopes and evaluating their differences and commonalities.

Our results show that pattern stability is not influenced by the omission of oxygen and therefore both cremated and uncremated samples can be employed to enrich the training set for further analysis.

Thousands of small flint pieces cover the sandy surface at the margins of the sabkha (a dry palaeolake) next to the oasis of Tayma, Saudi Arabia, an archaeological site investigated by a Saudi-German collaborative project (by the DAI and the Saudi Commission for Tourism and Antiquities / SCTA), funded by the German Research Foundation (www.dainst.org/de/project/tayma?ft=all).
Dense waste heaps resulting from the production of carnelian beads during the Chalcolithic period (approx. 5th to 4th millennia BC) have been recorded over the years in this area. Automated computer applications are necessary to handle the enormous quantities of objects, simply because manual recognition and counting are not sufficient for investigating a significant portion of the entire material.
The spatial distribution of these clusters has been mapped during the 2012 surface survey using Quantum GIS on an Android tablet. By extrapolating the results from one sampling site, the total number of discarded flint drills was estimated approx. 3 million. This number serves as a raw correlate for calculating the active productive forces necessary to wear down these tools while perforating discoid carnelian beads.
Based on these results, analysis proceeded on a more detailed level: Samples from numerous waste heaps were sifted and scanned in high resolution: items of flint, carnelian, sandstones and anhydrite in various shapes and sizes ranging from 1 to 60 mm were recorded. By connecting different freely available software for image processing, the (raw) material was identified, quantified, and measured automatically. This method is efficient but also reductive.
The talk will focus on the variables generated by this analysis and on conceptual problems which occur during this process, such as object recognition via image segmentation, material recognition via supervised classification and similarity calculation via minimal area difference between 2D-silhouettes of objects.
Finally I want to discuss how size-frequency distribution diagrams, fractal dimensions and similarity calculations of the bead-maker’s assemblage can serve as correlates for specific human-matter interactions and approximate the total commodity output of investigated production sites, despite the taphonomic effects which have disturbed the original dump deposits during the last 6000 years.

The automated detection of archaeological sites in remotely sensed data has recently become an important research topic due to the increasing availability of aerial and satellite images as well as ALS and radar data for archaeological research. To efficiently analyse large datasets, automated methods are required to perform routine tasks, such as screening large image data sets for known types of archaeological sites. However, available tools for the automated detection of modern buildings or roads usually fail to detect archaeological sites even if they are partially preserved above the ground. This is because they are often fragmented and of greatly varying size and shape, and the contrast to their surroundings is often very low due to the low height of ruined walls and the presence of random non-archaeological structures of similar spectral properties in their immediate surroundings.
In this paper we present a new approach to detect a specific category of archaeological sites in aerial and satellite images of the Silvretta Alps on the Swiss-Austrian border, at an altitude of 1500 to 2500 masl. A key aim of our research project in this area is to trace back the seasonal use of the alpine zone for pasture to its prehistoric origins. This requires an inventory of archaeological sites associated with alpine pastoralism, such as abandoned livestock enclosures. Our goal is to develop algorithms that allow a rapid screening of remotely sensed images for candidate sites of this type in order to guide and assist archaeological fieldwork in difficult terrain.
In order to be useful, our approach requires a high true-to-false detection ratio and needs to be computationally fast and easy to use. We base our approach on geometric cues of our target objects, which can be described as linear features meeting at approximately right angles. Since the target objects are located in open grassland, we first filter out wooded and urban areas based on their high texture contrast. We then extract linear features using morphological filtering. In order to identify groups of linear features that meet the geometric definition of our target objects, we define convexity and angle constraints that are used to assign a rectangularity measure to each group. This measure effectively expresses the likelihood of the group at hand corresponding to a target object. The result is a map indicating candidate locations of target objects in our study area and their corresponding likelihood values.
At the current state of our project, our approach yields good results even for incomplete instances of livestock enclosures, e.g. where walls are partially or completely missing. Additional work is required to reduce the number of false positives. Furthermore, we plan to test our approach on data sets from other mountainous areas with similar archaeological features to further improve its robustness. The long-term goal of our research is to provide a method of rapid, efficient and reliable screening of large image data sets for a predefined category of target objects prior to fieldwork.

In the past decades micromorphology has become an indispensable tool in (geo)archaeological research. It provides information on mineral and organic components, their nature, spatial distribution, and possible diagenesis. This supports the identification of site formation processes. Although micromorphology traditionally is a qualitative approach, quantitative analyses add valuable information. The application of the Feature Analyst® for ArcGIS® is a method to quantify micromorphological results.
The Feature Analyst® provides an automated feature extraction (AFE) model, usually applied for the interpretation of satellite and aerial images to identify e.g. buildings, roads, or vegetation types. It follows an object-based approach. Thus, the classification is not only based on the values of selected pixels but also on their surroundings, shapes, and clustering of pixel groups. Often, an automatic image classification with only one image is not reliable as different components resemble each other in color. The Feature Analyst® allows to combine transmitted plain polarized illumination images and cross polarized images. This makes results more reliable and precise e.g. a differentiation between quartz and pore space is possible. Apart from this, using the Feature Analyst® makes it possible to skip image mosaicking and allows to work on the entire thin section. This generates an easier and quicker workflow.
The Feature Analyst® enables the user to extract features in a short amount of time and to subsequently refine the classification, if necessary. The automatic classification approach allows for the determination of total volumes, size distribution, and sizes of single objects, as well as further statistical analysis. It proved to be a valuable tool in assessing percentages of flint, limestone, quartz, bones, and pore space. The interpretation of large numbers of thin sections is facilitated by the batch processing option, which allows for the application of developed classification algorithms to different data sets. This increases the comparability of thin sections, especially regarding the shares of individual mineral and organic constituents extracted, as well as fabrics and groundmasses.
With the help of the Feature Analyst® the efficient analysis of thin sections becomes possible, especially in terms of a quantitative assessment and precision. Classification results can seamlessly be integrated in any workflow for visualisation purposes.

This work presents the first results of the analysis on the settlement patterns of hunter-gatherer groups during Middle Palaeolithic in the western part of the Cantabrian region using the Site Catchment Analysis tool created inside the CRC 806 “Our way to Europe”. The study area comprises the territory between the basins of Nalón and Sella rivers. Most of the well-known Mousterian archaeological sites in this area, as El Sidrón, La Viña, El Conde or La Güelga are located in caves. However, there are also open air sites that belong to this period as El Barandiallu, located in the valley of Aboño river in the province of Asturias. In this site more than 1400 remains of stone tools in local quartzites and quartzs were recovered, and different types of non-local flint. The relationships between El Barandiallu and other Mousterian open air sites in this area as Llagú, Bañugues and El Caxili will be analysed, as well as the connections with the aforementioned karstic sites. For this purpose a cost-distance raster is calculated for El Barandiallu in order to enlighten the settlement patterns of Neanderthal groups in this area.

Furthermore, it is also important to highlight that the main problem of these open-air settlements, aside of all the taphonomical postdepositional processes, is that the record is only limited to the lithics. For this reason it is essential to study the lithic raw materials in order to link these kinds of archaeological sites with the related flint outcrops. The appliance of Least Cost Path Analysis will allow us to interpret the mobility and the lithic raw material procurement strategies used by Neanderthal groups between Nalón and Sella rivers.

The techno-typological characterization of all these Mousterian sites is of particular interest, in which due to deficiency and lack of numerical datings, we must resort to this type of analysis in order to contextualize this human occupation chronostratigraphically. In this case, the existence in several of these lithic assemblages of bifacial macro-tools (handaxes and cleavers) is particularly interesting since it is a specific feature of several of the local Mousterian sites that help us to establish more correlation elements, together with the presence and distribution of the different types of non-local lithic raw materials, which exist in most of these contexts.

Thus, the main elements combined in this SCA approach and in the processing of this information with GIS are: lithic raw materials, technological and typological characterization of lithic assemblages and the association between El Barandiallu with other open-air and karstic sites between Sella and Nalón rivers.

Despite the strong maritime character of the Netherlands, now and in the past, maritime archaeology has become a threatened research specialism: the nature of archaeological studies is frequently too narrowly focused on terrestrial archaeology, the discipline has a relatively small amount of active scholars and is still perceived to be engaged in antiquarianism. Archaeological studies that do target the maritime past are particularly focused on shipwrecks that are often documented as isolated and material objects, without considering historical, political, social and geographical context. It means that there is an urgent need for new interdisciplinary approaches in modern day maritime research. To bridge these boundaries, several researchers, including Christer Westerdahl, proposed to use a new tool, called the maritime cultural landscape: “the whole network of sailing routes, with ports, havens and harbours along the coast, and its related constructions and other remains of human activity, underwater as well as terrestrial”. This concept, based on Scandinavian archaeology, had a profound impact on maritime archaeology as the maritime cultural landscape approach is suitable for spatial studies instead of only studying individual sites or major excavations.

Changing the focus from object- and shipwreck orientated maritime archaeological studies to more integrative and spatial studies on the maritime cultural landscape forms the core of my PhD-research. The study area in this research is defined as the province of Flevoland, a reclaimed part of the former Zuiderzee with a highly dynamic past: from land into lake, into sea, into polder. Flevoland is often referred to as ‘the largest ship graveyard on land in the world’, as it contains over 420 remains of wrecked ships. I will focus on the interrelation between landscape development (geomorphology), occupation (eroded settlements and former islands) and shipping (wrecks) from a spatial point-of-view between 1100 and 1400 AD. A density analysis (ArcGIS) of wrecks in Flevoland has already shown distinct patterns and relations between wreck sites, seaports and submerged settlements. Examining these patterns in a GIS will lead to a better understanding of the landscape of the former Zuiderzee.

The idea that the dynamics of human action and movement through the landscape can offer insights into human phenomena, such as settlement genesis and growth, state formation and socio-political organization, underlies a great number of archaeological geospatial analyses. Especially, in recent years the increasing popularity of Network Science in archaeology has encouraged the development of novel approaches that use dynamic network models to study human interaction and the evolution of settlement hierarchies in the prehistoric landscape (Knappett 2011, Bevan and Wilson 2013). Such models are computer simulations which as a rule are created with ad hoc programming, although on some occasions the results of GIS-based procedures, such as least-cost path analysis, can also be used as model inputs. At a very basic level the aim of dynamic geographic networks is to simulate the “flow” of goods and people within a given settlement distribution over time, so as to explore the spatial dynamics that gave rise or contributed to the collapse of settlement hierarchies and networks of interaction in the long term.
This paper builds on pilot work on spatial interaction models (Bevan and Wilson 2013) in order to investigate aspects of socio-political organisation in South-Central Crete and identify patterns of settlement hierarchy at the period just before the emergence of the first palace of Phaistos. Although previous works have looked into spatial interactions at large spatial scales, the focus here will be upon small-scale and short-term settlement patterns that incorporate also the results of archaeological surface surveys. Furthermore, unlike previous works in which results have been validated mainly against observed settlement sizes, the produced models will be discussed and assessed with reference to a wider body of archaeological data that incorporates also the results of a statistical analysis of similarities in archaeological artefacts.
References:
Bevan, A. & Wilson, A. (2013) Models of settlement hierarchy based on partial evidence, Journal of Archaeological Science, 40(5), 2415–2427.
Knappett, C., Rivers, R., Evans, T., 2011. The Theran eruption and Minoan palatial collapse: new interpretations gained from modelling the maritime network. Antiquity 85, 1008-1023.

The question ‘Which timbers originate from the same tree?’ aims for insights into historic timber economics and construction processes. Here it is addressed to wooden linings from early Neolithic wells excavated in the Rhineland region in Western Germany.
Neolithic wooden relics, especially from the early Neolithic Linear Pottery Culture (LBK, 5500 to 5000 BC), are very rare in Central Europe. In the Rhineland wooden samples from three early Neolithic well sheetings excavated in Erkelenz-Kückhoven (EK, n=86), Arnoldsweiler (AW, n=36) and Merzenich-Morschenich (MM, n=24) have been dendrochronologically analysed. The felling dates of the youngest trees are 5090 BC (EK, waney-edge), 5097±10 BC (AW, sapwood, preliminary date) and 5052±5 BC (MM, sapwood).
Nearly all boards of the linings are made of oak (Quercus spp.) and only dated tree-ring series from this species are used for the analysis. A data set from recent trees serves to check the reliability of the results. We apply two established approaches to analyse the tree-ring width sequences in regard to their common origin.
1. Visual comparison of the growth-curves and statistical analysis of their correlations (e.g. Tegel et al. 2012).
2. The ‘dendro-allocation’ by Mom et al. (2011).
Starting from Mom et al. (2011) one of us (GR) refined some aspects of the computation using the package vegan (Oksanen et al. 2013) written in the statistical computing language ‘R’ (R Core Team 2012). The results of this procedure are checked against those of the above mentioned approaches. An R-code for the analysis can be provided by GR.