vessels: inside and outside

VESSELS: INSIDE AND OUTSIDE PROCEEDINGS OF THE CONFERENCE EMAC '07 9TH EUROPEAN MEETING ON ANCIENT CERAMICS * 24-27 OCTOBER 2007, HUNGARIAN NATIONAL MUSEUM BUDAPEST, HUNGARY

Edited by: Katalin T. Biró Veronika Szilágyi Attila Kreiter

VESSELS: INSIDE AND OUTSIDE Proceedings of the Conference EMAC '07 9th European Meeting on Ancient Ceramics 24-27 October 2007, Hungarian National Museum, Budapest, Hungary

INTERNATIONAL SCIENTIFIC COMMITTEE

S.Y. Waksman (Lyon, France) I. Dias (Lisbon, Portugal) M. Vendrell (Barcelona, Spain) E. Starnini (Genova, Italy) P. M. Day (Athens, Greece) M. Maggetti (Freiburg, Switzerland) M. Martinón-Torres (London, England) V. Kilikoglou (Athens, Greece) LOCAL ORGANISING COMMITTEE

B. Bajnóczi, IGCR HAS M. Balla, BME T. Bezeczky, Vienna K. Gherdán, ELTE H. Herold, Vienna A. Kreiter, FSCH Zs. Mersdorf, Archeosztráda Kft. F. Pintér, FSCH Gy. Szakmány, ELTE V. Szilágyi, II HAS, ELTE K. T. Biró, HNM, ACE K. T. Bruder, HNM G. Tomka, HNM M. Tóth, IGCR HAS

Printed in 2009 Budapest, Hungary Published by the Hungarian National Museum H-1088 Budapest, Múzeum krt. 14-16 Publisher and Editor-in-Chief Tibor Kovács, General Director of HNM © Copyrights with the authors © Editors K.T. Biró, V. Szilágyi, A. Kreiter © Hungarian National Museum Cover design: Ágnes Vári Printed at T-MART Press, Budapest ISBN 978-963-7061-67-7 Printed with financial support of the Italian National Archaeometry Society (AIAr)

EDITORIAL PREFACE

EDITORIAL PREFACE PROCEEDINGS OF THE CONFERENCE EMAC'07 BUDAPEST - VESSELS: INSIDE AND OUTSIDE 24-27 OCTOBER 2007 The 9th European Meeting on Ancient Ceramics (EMAC) took place in Budapest, 24-27 October 2007, organised by the Hungarian National Museum. 98 experts from 20 countries took part in this event, presenting 43 lectures and 62 posters. The present volume contains the scientific papers submitted for publication, altogether 36 communications. The communications conveyed the latest results in the field of pottery archaeometry, with special attention paid to the petrographical aspects of the study of archaeological ceramics. The Conference was organised in thematic sections. The structure of the Proceedings follows these sections. The lectures, posters as well as the papers submitted reflect the recent advancement and state of the art on the subject. The Conference had a special importance for Hungarian research. Pottery archaeometry and even more, pottery petroarchaeology has been rather understudied in Hungary until very recently. Apart from pioneering studies by a few experts like Márta Balla or Tamás Bezeczky, generations of archaeologists have spent their life investigating pottery from a traditional typo-chronological view without considering the power of interdisciplinary studies towards a better understanding of material culture. Though most archaeological descriptions include a gross characterisation of the physical qualities of the constituting matter ranging from colour, grain size and sometimes temper, the interdisciplinary investigation of the raw materials and their use in drawing archaeologically relevant conclusions is a recent development in ceramic studies in Hungary. Organising EMAC and paying due attention to these problems transformed the way of thinking about pottery. By now, we have a generation of young experts working in the field of pottery analysis using scientific methods adopted mainly from geosciences. For the occasion of the EMAC meeting we could publish a special volume of the electronic archaeometry journal Archeometriai Műhely / Archaeometry Workshop www.ace.hu/am 2007/2 with an annotated bibliography on pottery archaeometry. The results also serve as a starting point for further studies and projects, several of them with international collaboration. The editors recommend the present volume to anyone interested in archaeological pottery; archaeologists, analysts, conservators and ethnographers, as well as any interested reader who wants to know more about ceramic vessels, inside and outside. ACKNOWLEDGEMENTS The editors of this volume wish to express their gratitude to those who made the publication of this volume possible. First of all, the Local Organising Committee and the Scientific Committee helped enormously in the organisation of the meeting, and took a major role in the revision of the submitted manuscripts. Financial support for the publication of this volume was received from the Italian National Archaeometric Society (AIAr) and many thanks must go to the President of the AIAr, Professor Mauro Bacci. We want to express our special thanks to Elisabetta Starnini, mediating towards AIAr and helping the publication of this volume in many ways. Supporters of the Conference and Proceedings are listed on the List of Supporters. We are also indebted to our collaborators making the cover design (Ágnes Vári) and printing (T-Mart Press). The electronic version of the manuscripts in colour is available at the Conference Website (http://www.ace.hu/emac07/). We are also grateful for hosting (Archaeocomp Association) and design (Gábor Telcs). The electronic version of the proceedings and some conference memorabilia are included on a Conference CD, also designed by them. Editors Katalin T. Biró Veronika Szilágyi Attila Kreiter

3

EMAC'07 BUDAPEST - VESSELS: INSIDE AND OUTSIDE

4

GREENE & HARTLEY: FROM ANALOG TO DIGITAL - RADIOGRAPHY OF ARCHAEOLOGICAL POTTERY

FROM ANALOG TO DIGITAL: PROTOCOLS AND PROGRAM FOR A SYSTEMATIC DIGITAL RADIOGRAPHY OF ARCHAEOLOGICAL POTTERY 1 1 2

A. Greene – 2C. Hartley

University of Chicago, Department of Anthropology, 1126 E 59th St., Chicago, IL 60637, [email protected] University of Chicago, Department of Anthropology, 1126 E 59th St., Chicago, IL 60637, [email protected]

Abstract: In this paper we present the results of three years of research and experimentation with the digital radiographic analysis of archaeological potsherd assemblages, with particular attention to discerning and distinguishing techniques of vessel formation. In contrast to previous digital radiographic efforts which have primarily been used to evaluate museum objects or archaeological finds of particular heritage import, the authors offer a digital radiographic application for the analysis of large archaeological potsherd datasets (n > 500), the basic fragmentary data of traditional archaeology. We describe the significant improvements over older analog techniques, the types of formation mechanics discernable through radiography, and demonstrate the way digital image manipulation can identify and discriminate between different formation strategies. The particular imaging protocols for producing image sets of maximum quality are delineated and the authors outline the post-processing tools that take advantage of the metricmatrix qualities of digital imagery. Keywords: radiography of material culture, potsherds, assemblage, digitization

2005; Rye 1977). This work is contextualized within a host of other radiographic studies of archaeological fauna (Ambers 2005), human bone (Davis 2005), paper (Daniels and Lang 2005), metals (Lechtman et al. 1975), and textiles (Yoder 2008). Pottery, in particular potsherds, however, have received comparatively less attention than have ceramic objects of art or particularly enigmatic museum pieces, certain notable exceptions notwithstanding (Braun 1982; Carr 1990; Heinsch and Vandiver 2006; Vandiver 1987). In the last quarter of the twentieth century, this limited work on potsherds showed much promise and innovative thinking, including intriguing indications that the radiographic examination of pottery could distinguish (1) mechanical formation techniques of vessels (Vandiver 1987; 1988) and (2) potentially identify minerals over incredibly vast swaths of the container (Braun 1982; Carr 1990), permitting regional mineral sampling through an essentially nondestructive analytical program.

INTRODUCTION The archaeometric understanding of pottery formation techniques has benefited greatly from instrumental (Courty and Roux 1995), ethnographic (Longacre 1991), and experimental (Wallaert-Pêtre 2001) examinations of archaeological ceramics in recent decades. Formation practices interest scholars of pottery because they bridge important boundaries in the production process between the procurement of raw materials and the final achievement of a ceramic “product.” Radiography provides a particularly powerful tool in the investigation of formation techniques due to its ability to reveal the traces of mechanical actions like coil-building, slabbuilding, and wheel-throwing, which determine a vessel’s shape and texture, and are critical for understanding the organization of production practices in general (Carr 1990; Hamon, Querré, and Aubert 2005; Heinsch and Vandiver 2006; Lang and Middleton eds. 2005; Vandiver 1987, 1988). In this study, we focus on the relatively new field of digital radiography (DR) because it offers the additional benefits of: (1) incredibly rapid data acquisition speeds lasting less than 10 seconds and (2) immediate post-processing capacity in transforming radiographic images, obviating the need to scan film radiographs in order to make use of powerful digital tools (Lang & Middleton 2005; O’Connor and Maher 2001). These improvements over the previous analog technique now enable a truly systematic and assemblage-based radiography of formation techniques to take place.

These early studies often utilized analog film techniques in hospitals or made use of instruments like the Xerox Corporation’s “Xeroradiograph.” By the 1990s, clear parameters for the production of high quality analog imagery were widely used and available (Lang and Middleton 1997). Since the debut of more robust digital equipment and techniques two decades ago, however, the obsolescence of xeroradiography (Lang and Middleton 2005) has not resulted in substantial advancements in radiographic imaging techniques for pottery and the general analysis and interpretation of their radiographic imagery. In fact, experimentation with digital radiography (DR) and X-ray computed tomography (XCT or CT) has tapered off considerably (Applbaum and Applbaum 2005; Lang and Middleton 2005; Vandiver et al. 1991).

Over the last half-century, ceramics have occupied an important position in archaeometric approaches using radiographic analysis (Braun 1982; Carmichael 1990; Digby 1948; Glanzman and Fleming 1986; Middleton

5


Table 1 Optimal parameters for digital radiography determined from initial results. Parameters kiloVolts (kV) milliAmperes (mA) Integration Time (IT) Source-to-Object Distance (SOD) Source-to-Detector Distance (SDD)

Possible Settings

Optimal Settings

100, 200, 300, 400

~275

0-4 (dependant on kV)

~0.1

1, 2, 4, 8, 16 seconds

8 seconds

95, 126, 160, 164, 191, 225, 229, 256, 294 cm

95 cm

219, 284, 349 cm

219 cm

This is most certainly due to the extreme difficulty and complexity of properly configuring such instruments (Casali 2006), but it is also in line with the traditional tendency of using radiographic analysis for the study of art objects, and not for the in-depth, assemblage-based analysis so essential to archaeological research. This study constitutes, to our knowledge, the first systematic application of digital radiographic techniques in potsherd assemblage analysis since the methods became available in the 1980s.

EXPERIMENTAL The first phase of our technique development project involved the determination of settings for digital radiographic potsherd analysis that maximize the advantages of the technology. In order to gain the degree of accuracy required for consistent, sample-to-sample, assemblage-based analysis, image acquisition settings must reliably produce images of a high, measurable quality. Without the ability to produce comparable images with similar quality for each potsherd, a dataset with numerous cases would be largely useless. Lang and Middleton (2005) published a recommended set of parameters for the analog radiography of pottery, but in the course of our research these parameters quickly revealed themselves to be unsuitable for the newer digital equipment. Whereas analog film radiography and xeroradiography of pottery utilized long integration periods, requiring the artifact to sit in a relatively low kVp environment while the film soaked up radiation for minutes at a time, this kind of exposure would quickly saturate (and eventually damage) a digital X-ray detector.

Fig. 1 Comparison of over-attenuated and over saturated radiograph images. Figure 1a displays the lack of detail exhibited by a potsherd that has not received enough kVp or integration time. Figure 1b displays the white background and “panelization” lines of a detector that has been oversaturated with too much energy.

Concomitantly, the lower kVp necessary for long analog integration times results in too much attenuation of the source, resulting in overly dark imagery.

6


Table 2 Comparison of X-ray Digital Radiography and Computed Tomography. Stage Position 1 2 3 4 5 6 7 8 9

Source-to-Detector Distance (cm) 349 349 349 284 284 284 219 219 219

Source-to-Object Distance (cm) 225 256 294 160 191 229 95 126 164

Magnification Factor 1.55 1.36 1.19 1.78 1.49 1.24 2.31 1.74 1.34

The possible integration times were determined by the hardware settings of the detector—for this experiment, we selected 5 different integration times of the 8 possible: 1, 2, 4, 8, and 16 seconds. Finally, the SODs and SDDs were organized into 9 different combinations of locations for the X-ray tube and sample stage because these also have essentially infinite combinations (Table 2, Fig. 2).

In addition to these difficulties, the daily and weekly variability in the beam output of an X-ray tube introduces additional variation in potential attenuation, such that analysts working with a large dataset of potsherds acquiring data over a long period of time must have the ability to measure and calibrate the consistency of image quality as they work. So not only was it necessary to produce specific parameters of integration time, kV, and mA, we had to introduce a technique for measuring radiograph quality across a potential dataset, image-byimage.

All possible permutations for kV/mA, integration time, SOD, and SDD, were then tabulated, resulting in 189 possible combinations. Of these, we selected a random sample of 50 combinations (or a 27% sample) in order to conduct our parameter testing2. To calculate image quality, the project utilized the metric of Modulation Transfer Function, or MTF value (cf. Casali 2006: Appendix B; Fujita et al. 1992; Pham 2006), essentially the number of line pairs per millimeter (lp/mm) resolvable at different experimental settings.

Working in the X-ray Computed Tomography (XCT) Laboratory at Argonne National Laboratory (ANL) in the United States, our primary experimental setup in this endeavor consisted of a Phillips 420 kVp X-ray tube, which has small and large filaments generating a 1500µm spot-size and a 4500µm spot-size respectively. The tube is paired with a Perkin-Elmer 1640-A X-ray detector that has a 200µm resolution and measures 2048 pixels square. As a 16-bit digital detector it has the ability to discriminate 4096 shades of grey1.

Because we were interested in the ability to distinguish the sharpness of ancient potsherds, and not the typical line pair gauge used in radiographic calibration, each test image contained the same sherd, cut so that one edge was straight and the MTF could be reliably calculated using an edge-spread function (ESF) (Casali 2006; Pham 2006)3.

As mentioned above, the parameters included in our protocol study were: (1) X-ray tube kilovolts and milliamperes, (2) the integration or “exposure” time on the detector, (3) the Source-to-Object Distance (SOD), and (4) the Source-to-Detector Distance (SDD). The possible settings used for each parameter are summarized in Table 1. Because the number of possible voltage/ampere combinations (producing the kVp of the tube) is essentially infinite, we selected settings at intervals of 100 kV (e.g.: 100, 200, 300, and 400 kV) paired with different sets of integration time, SOD, and SDD. In the course of our testing, we determined the mA settings that produced viable images when paired with these kV settings—in other words, some combinations of kV and mA setting produced either over-attenuation of the X-ray beam or over-saturation of the detector, rendering the images useless for analysis (Fig. 1).

To measure MTF, the fifty test images were run through a software module written by the project in Interactive Data Language (IDL), entitled the “Sherd Image Viewer and Analysis” Program 2 (ShIVA2). The program imports imagery into its display module (Fig. 3a), samples the pixel values across the sherd edge (Fig. 3b), calculates the MTF value for each successive image via fast Fourier Transforms (cf. Fujita et al 1992; Pham 2006), and outputs the MTF value and a graph of its distribution (Fig. 3c).

7


Fig. 2 The configuration of source, object, and detector stages at Argonne National Laboratory’s X-ray Computed Tomography laboratory. Nine combinations of positions were used to develop the most appropriate instrumentation for the DR of ancient potsherds.

When plotted together, the fifty test images suggested optimal settings of kVp and stage positions, which prompted further parameter experimentation with 25 additional setting cohorts in the 200-300 kV range in setup position 7. Those final experiments indicated that the following parameters would be ideal for the DR of ancient ceramic potsherds: an integration time of 8 seconds (Fig. 4a), a kV between 250 and 275 (Fig. 4b), a mA setting of approximately 0.1 to 0.15, and the positioning of the source and object as close together as possible (Fig. 4c). These results are further summarized in Table 1. At present, this is the best combination of parameters to acquire ancient potsherd imagery, yielding a maximal MTF value around 0.6. It is important to note that extensive variability existed across the range of setting permutations, so that the highest MTF value was not necessarily the best indicator of a reliable or preferred cohort of settings. For example, while the highest MTF values were achieved at setup positions 3 and 9 (Fig. 4c), the low level of magnification available in those positions made them less preferable for potsherd visualization overall.

The second phase of the experiment involved the identification of productive digital filters that would identify the most relevant features of discrimination in the ceramic images. Important diacritica in this endeavor include: the presence and organization of joins, coils, or rings, the orientation and disposition of voids and elongate particles of temper (to help identify the use of wheel throwing or forming), and the presence of mold traces, often shown by the cavities left by textiles and basketry during the molding process (Heinsch and Vandiver 2006; Rye 1977; Vandiver 1987, 1988). Using statistical and filtering modules present in Interactive Data Language (IDL), we composed several texture and gradient filters which identify and classify the factors of maximum density difference into distinct groups and then display them color-coded on a new image (cf. Deemer and Metzger 2006).

RESULTS This study of radiographic pottery techniques produced two primary results: (1) a set of standardized parameters for the mass digital radiographic imaging of archaeological potsherds and (2) new analytical methods for the bulk analysis of vessel formation techniques through digital manipulation and analysis routines of the resultant datasets.

In addition to standardizing the image acquisition process to produce images of a consistent quality and resolution, the project also developed post-processing tools in ShIVA2 for data normalization and analysis.

8


Using ShIVA2, the analyst identifies the region of interest (ROI) on the “raw” image (the sherd and the background or scale), eliminates the background values around the sherd, and normalizes the histogram within the ROI. Figure 5 shows the results of this process through the normalization of an Early Bronze Age (ca. 3500-2600 B.C.) sherd from Azerbaijan. A properly normalized image (Fig. 5b) provides an instant visual improvement over a “raw” digital radiograph (Fig 5a) and is a prerequisite for further filtering analyses of sherd subregions and structural features. Observe the difference between the lighter analogue xeroradiograph image in Figure 5d and the normalized DR image in Figure 5b that utilizes the maximum dynamic range. Particular areas can be further contrast-enhanced by the use of a bounding box (Fig. 5c). Only after images are normalized can filtering tools provide analytical insight into the structural aspects of various vessel formation techniques. The routines written into ShIVA2 filter each dataset based on texture and gradient aspects, identifying the primary cohorts of variability and classifying them by color. These filters are able to identify features such as joins, particle orientations, clay gradients, layer boundaries, and variation in inclusions frequencies over different regions of a particular sherd. In Figure 6, a progression from photograph, to "raw" radiograph, to normalized radiograph, concludes with a filter named “entropy” (Figure 6d) that has identified linear stacking in the wall of a first millennium B.C. bowl from Tsaghkahovit, Armenia. This patterning in the vessel wall is characteristic of a coil or ring-building technique (Gelbert 2005; Sall 2005), an important revelation as most tablewares from Achaemenid Armenia are assumed to be wheel-thrown (Khatchadourian 2008). DISCUSSION While these new parameters for the digital radiography of ancient potsherds enable significant data collection to now take place, the delineation of a systematic strategy of radiographic data acquisition is also essential. To ensure that an assemblage analysis project utilizing radiographic techniques remains connected to other aspects of traditional and archaeometric pottery analysis, it is important that DR analyses are directly informed by other research methods, particularly in the constitution of the original assemblage dataset. Pre-existing knowledge regarding the objects of study resulting from traditional ceramic analysis techniques (morphological, stylistic, and fabric analysis), compositional analysis, or SEM examination form an essential background from which to draw samples and organize a dataset that is as “representative” as possible. This will allow for a focused analytical program that will offer data on formation techniques that can immediately inform the ordering of production typologies and assemblage variability in general.

Fig. 3 Potsherd visualization (a), MTF calculation (b), and MTF output (c) from the ShIVA2 program. Here we distinguish the normalization or pixel values from the acquisition of consistent quality images (MTF values). In a photographic analogy, MTF can be compared to image focus and pixel value normalization can be understood as contrast enhancement. If the images are not sufficiently focused, no amount of contrast enhancement will be able to improve their clarity and usefulness. Pixel normalization takes advantage of the full “dynamic range” of the X-ray detector, what Casali calls “…the ratio of maximum to minimum detectable signal (2006: 55-56).” During the normalization or “equalization” process, all of the pixel values within the sherd are stretched across a histogram based on the highest and lowest values.

9


Fig. 4 Plots of the resultant MTF calculations from the test image dataset.

10


Fig. 5 Demonstration of the ShIVA2 normalization technique from “raw” radiograph (a), to contrast-enhanced radiograph (b). Figure 5c demonstrates the targeted contrast enhancement feature that applies a bounding box. Figure 5d shows the same potsherd imaged with an older analog xeroradiograph technique.

resolution techniques such as microfocus radiography or CT (Casali 2006; Lang and Middleton 2005; Lang et al. 2005).

Once the dataset has been properly selected and stratified, we suggest a telescoping, multistage method for the implementation of DR techniques that exploits the powerful resolution of the method, while acknowledging the complex data management imperatives involved in assemblage-based analysis.

CONCLUSION While archaeologists and archaeometrists have been aware for several decades of the analytical potential of assemblage-based radiography, it is only now, with the full digitization of the data acquisition, normalization, analysis, and storage of these massive datasets, that the large-scale radiographic examination of potsherds is possible. Traditional ceramic analysis and radiography may well characterize the production techniques for one particular vessel, but the analytical value of those results can only be evaluated in light of that vessel’s position within a broader ceramic assemblage.

In the first stage, a large assemblage of pottery, or a sample of a larger assemblage, can be imaged using the standardized parameters for digital radiography outlined earlier. Easy manipulation of these images using Photoshop or IDL programming allows the investigator to quickly mine the data for multi-scalar evidence of production techniques, documenting patterning of voids and inclusions, gradient features in the clay matrix, and structural joins not visible externally. As a second stage of analysis, the investigator can then put together a shorter list of objects for further analysis using higher

11


Fig. 6 Demonstration of the ShIVA2 “entropy” filter revealing stacked bands in a potsherd wall. The figure shows the sample as it moves from photograph (a), to “raw” radiograph (b), to normalized radiograph (c), to filtered radiograph.

tools4. These tools will need to be augmented through high volume “batch-oriented” routines and high capacity storage systems, to effect the efficient analysis of pottery formation techniques. Future work is intended to perform the same protocol and application development for highresolution microfocus and computed-tomographic techniques. This scale of analysis provides archaeologists with the ability to gain a systematic and detailed picture of the variability in formation techniques and shed new light on the organization of past pottery industries.

If there is any essential conclusion to be drawn from the past 30 years of ceramic ethnoarchaeology, it is that pottery is situated within enormously complex and highly variable production organizations; even on the simplest social level they are true “industries” (Balfet 1984; Leeuw 1993; Mahias 1993; Wallaert-Pêtre 2001). From a statistical and anthropological viewpoint, the significance of results is inexorably linked to the construction of a representative dataset. An analytical perspective such as this demands closer working relationships with field archaeologists, as the composition of an assemblage-level dataset is most dependant on the expertise and research design of excavators.

ACKNOWLEDGEMENTS The authors would like to express their gratitude to collaborators Dr. William Ellingson, Dr. Christopher Deemer, and Richard Koehl at the Argonne National Laboratory computed tomography facility for their attentive training and support. Special thanks are also extended to Dr. Pavel Avetisyan and Dr. Ruben Badalyan at the Armenian Institute of Archaeology and

This technique development project has laid the groundwork for assemblage-based digital radiographic analysis of potsherds by providing the foundational tools for data acquisition, normalization, and analysis—made available through a modular and adaptable set of software

12


BALFET, H. (1984): Methods of Formation and the Shape of Pottery. In The Many Dimensions of Pottery, eds. S.E. van der Leeuw & A.C. Pritchard, pp. 171-202. University of Amsterdam, Leiden.

Ethnography, Dr. Chen Xingcan at the Institute of Archaeology, CASS, Dr. Li Liu at Latrobe University, MaryFran Heinsch at the University of Chicago, and Dr. Ludmilla Koryakova at the Institute of History and Archaeology, Ural State University, Ekaterinburg, Russia for providing access to the ceramic samples included in this study. Thanks are also due to Dr. Adam T. Smith, Dr. Michael Dietler, Dr. Shannon Dawdy, Dr. Nicholas Kouchoukos, and Dr. Michael Chinander at the University of Chicago, Lori Khatchadourian at the University of Michigan, and Dr. David Peterson, Idaho State University, for their helpful thoughts and ongoing support.

BRAUN, D. P. (1982): Radiographic Analysis of Temper in Ceramic Vessels: Goals and Initial Methods. Journal of Field Archaeology 9:183-192. CARMICHAEL, P. H. (1990): Nasca pottery construction. Ñawpa Pacha 24:31-48. CARR, C. (1990): Advances in Ceramic Radiography and Analysis: Potentials. Journal of Archaeological Science 17:13-34.

NOTES CASALI, F. (2006): X-ray and Neutron Digital Radiography and Computed Tomography for Cultural Heritage. In Physical Techniques in the Study of Art, Archaeology, and Cultural Heritage, eds. D. Bradley & D. Creagh, vol. 1, pp. 41-123. Elsevier, Amsterdam.

1

While the average human eye can only discriminate approximately 20-30 shades of grey in an image (Russ 2007: 92), each digital-matrix image pixel posses a value between 0 and 4095 that can be (1) contrast-enhanced across the wide dynamic range of the instrument, allowing easier human viewing, and (2) statistically manipulated to show the greatest factors of variability, filtering and classifying the most relevant data sub-groups (Casali 2006; Lang and Middleton 2005; Lang et al. 2005).

COURTY, M. A. & V. ROUX (1995): Identification of Wheel Throwing on the Basis of Ceramic Surface Features and Microfabrics. Journal of Archaeological Science 22(1):17-50. DANIELS, V. & J. LANG (2005): X-rays and paper. In Radiography of Cultural Material, eds. J. Lang & A. Middleton, pp. 96-111. Elsevier Butterworth-Neinemann, Burlington, MA.

2

We also experimented with the use of metal screens over the detector and filters in front of the X-ray source. Different metals and thicknesses affect the quality of the X-ray beam in different ways, but it was eventually determined that the best way to maximize image contrast was to operate without screen or filter (Lang and Middleton 2005).

DAVIS, R. (2005): Radiography: archaeo-human and animal remains. Part I: Clinical radiography and archaeo-human remains. In Radiography of Cultural Material, eds. J. Lang & A. Middleton, pp. 130-149. Elsevier ButterworthNeinemann, Burlington, MA.

3

The ceramic materials used in the study were drawn from the Making of Ancient Eurasia (MAE) collection, an interdisciplinary collaboration between scholars at the University of Chicago and ANL that studies ceramic and metal objects from the Neolithic, Bronze, and Iron Ages of Central China, the Russian Steppe, and the South Caucasus (Koryakova and Epimakhov 2007; Liu et al. 2002; Smith et al. in press).

DEEMER, C. & Z. METZGER (2006): Automated flaw detection of ceramic microturbine components using highspeed x-ray computed tomography. Presented at ASME Turbo Expo 2006: Power for Land, Sea and Air. Barcelona, Spain. DIGBY, A. (1948): Radiographic examination of Peruvian pottery techniques. In Actes du xxviiie Congrès International des Américanistes, Paris, 1947, pp. 605-608. Musée de l’Homme, Paris.

4

The binary version of the ShIVA2 program is available at http://mae.uchicago.edu

FUJITA, H., D.-Y. TSAI, T. ITOH, K. DOI, J. MORISHITA, K. UEDA & A. OHTSUKA (1992): A simple method for determining the Modulation Transfer Function in digital radiography. IEEE Transactions on Medical Imaging 11(1):34-39.

REFERENCES AMBERS, J. (2005): Radiography: archaeo-human and animal remains. Part II: Radiography of animal remains. In Radiography of Cultural Material, eds. J. Lang & A. Middleton, pp. 150-154. Elsevier Butterworth-Neinemann, Burlington, MA.

GELBERT, A. (2005): Reconnaissance des techniques et des méthodes de façonnage par l’analyse des macrotraces: étude ethnoarchéologique dans la vallée du Sénégal. In Section 2, Archaeometry; Symposium 2.1, Pottery Manufacturing Processes: Reconstitution and Interpretation, eds. A. L. Smith, D. Bosquet & R. Martineau, pp. 67-78. Archaeopress, Oxford.

APPLBAUM, N. & Y. H. APPLBAUM (2005): The use of medical computed tomography (CT) imaging in the study of ceramic and clay artifacts from the ancient near east. In X-rays for Archaeology, eds. M. Uda, G. Demortier & I. Nakai, pp. 231-245. Springer, Dordecht, The Netherlands.

13


GLANZMAN, W. D. & S. J. FLEMING (1986): Pottery: fabrication methods. In The Late Bronze and Early Iron Ages of Central Transjordan: The Baq’ah Valley project, 1977-1981, University Museum Monograph No. 65, ed. P. E. McGovern, pp. 164-177. University Museum, University of Pennsylvania, Philadelphia.

complexity in the Yiluo region, north China. Journal of Field Archaeology 29:75-100. LONGACRE, W. (editor) (1991): Ceramic Ethnoarchaeology. University of Arizona Press, Tuscon. MAHIAS, M. (1993): Pottery Techniques in India: Technical Variants and Social Choice. In Technological Choices: Transformations in Material Culture Since the Neolithic. P. Lemonnier (ed.), pp. 157-180. New York, Routledge.

HAMON, G., G. QUERRE & J. AUBERT (2005): Techniques de fabrication de ceramiques du Néolithique moyen I en Armorique (France). In Section 2, Archaeometry; Symposium 2.1, Pottery Manufacturing Processes: Reconstitution and Interpretation, eds. A. L. Smith, D. Bosquet & R. Martineau, pp. 127-138. Archaeopress, Oxford.

MIDDLETON, A. (2005): Ceramics. In Radiography of Cultural Material, eds. J. Lang & A. Middleton, pp. 76-95. Elsevier Butterworth-Neinemann, Burlington, MA.

HEINSCH, M. & P. VANDIVER (2006): Recent Xeroradiographic Analysis of Kura-Araxes Ceramics. In Beyond the Steppe and the Sown: Proceedings of the 2002 University of Chicago Conference on Eurasian Archaeology, eds. D. L. Peterson, L. Popova, & A. T. Smith, pp. 382-394. Brill, Leiden.

O’CONOR, S. & J. C. MAHER (2001): The digitisation of Xradiographs for dissemination, archiving and improved image interpretation. The Conservator 25:3-15. PHAM, T. (2006): Appendix B - Model-based point spread function estimation. In Spatiotonal adaptivity in superresolution of undersampled image sequences. Unpublished PhD thesis, Delft University of Technology.

KHATCHADOURIAN, L. (2008): Social Logics Under Empire: The Armenian ‘Highland Satrapy’ and Achaemenid Rule, ca. 300-600 BC. Unpublished PhD Thesis, University of Michigan.

RUSS, J. C. (2007): The image processing handbook. 5th edition. CRC Press, Boca Raton, Florida.

KORYAKOVA, L. N. & A. V. EPIMAKHOV (2007): The Urals and Western Siberia in the Bronze and Iron Ages. Cambridge University Press, Cambridge.

RYE, O. S. (1977): Pottery Manufacture Techniques: X-ray Studies. Archaeometry 19(2):205-211. SALL, M. (2005): Cultural Contacts and Technical Heritage in Senegambia. In Section 2, Archaeometry; Symposium 2.1, Pottery Manufacturing Processes: Reconstitution and Interpretation, eds. A. L. Smith, D. Bosquet & R. Martineau, pp. 57-66. Archaeopress, Oxford.

LANG, J. & A. MIDDLETON (editors) (1997): Radiography of Cultural Material. 1st edition. Elsevier ButterworthNeinemann, Burlington, MA. LANG, J. & A. MIDDLETON (editors) (2005): Radiography of Cultural Material. 2nd edition. Elsevier ButterworthNeinemann, Burlington, MA.

SMITH, A. T., R. BADALYAN, P. AVETISYAN, L. MINC, & A. GREENE (In Press): The Archaeology and Geography of Transcaucasian Societies I: Regional Survey in the Tsaghkahovit Plain, Armenia 1998-2000. Oriental Institute Press, Chicago.

LANG, J. & A. MIDDLETON (2005): Radiography: Theory. In Radiography of Cultural Material. 2nd edition., eds. J. Lang & A. Middleton, pp. 1-19. Elsevier Butterworth-Neinemann, Burlington, MA.

VANDIVER, P. (1987): Sequential Slab Construction; A Conservative Southwest Asiatic Ceramic Tradition, ca. 7000-3000 B.C. Paléorient 13(2):9-35.

LANG, J., A. MIDDLETON, J. AMBERS & T. HIGGINS (2005): Radiographic images. In Radiography of Cultural Material. 2nd edition., eds. J. Lang & A. Middleton, pp. 2049. Elsevier Butterworth-Neinemann, Burlington, MA.

---. (1988): The Implications of Variation in Ceramic Technology: The Forming of Neolithic Storage Vessels in China and the Near East. Archeomaterials 2(2):139-174.

LECHTMAN, H., L.A. PARSONS & W.J. YOUNG (1975): Seven Matched Hollow Gold Jaguars from Peru's Early Horizon. Studies in Pre-Columbian Art and Archaeology 16. Dumbarton Oaks, Trustees for Harvard University, Washington, D.C.

VANDIVER, P., W. A. ELLINGSON, T. K. ROBINSON, J. J. LOBICK & F. H. SEGUIN (1991): New Applications of XRadiographic Imaging Technologies for Archaeological Ceramics. Archeomaterials 5(2):185-207.

LEEUW, S. V. D. (1993): Giving the Potter a Choice: Conceptual Aspects of Pottery Techniques. In Technological Choices: Transformation in Material Culture Since the Neolithic. P. Lemonnier (ed.), pp. 238-288. London, Routledge.

WALLAERT-PÊTRE, H. (2001): Learning How to Make the Right Pots: Apprenticeship Strategies and Material Culture, A Case Study in Handmade Pottery From Cameroon. Journal of Anthropological Research 57(4):471-493.

LIU, L., XC. CHEN, YK. LEE, H. WRIGHT & A. ROSEN (2002): Settlement patterns and development of social

YODER, D. T. (2008): The use of “soft” X-ray radiography in determining hidden construction characteristics in fiber sandals. Journal of Archaeological Science 35(2):316-321.

14

HEIN & AL.: GREAT POTS ON FIRE: THERMAL PROPERTIES OF ARCHAEOLOGICAL COOKING WARE

GREAT POTS ON FIRE: THERMAL PROPERTIES OF ARCHAEOLOGICAL COOKING WARE 1

A. Hein – 1,2N.S. Müller – 1V. Kilikoglou

1

Institute of Materials Science, N.C.S.R. “Demokritos”, 15310 Aghia Paraskevi, Athens, Greece Department of Archaeology, University of Sheffield, Northgate House, West Street, Sheffield S1 4ET, UK

2

Abstract: In the past, experimental studies of the functionality of archaeological cooking pots were focussed mainly on thermal shock resistance. Other properties, such as the thermal conductivity, and their relation with specific ceramic fabrics were discussed rather generally without being directly investigated. In the present study a modified Lees’ disk setup was used in order to determine the thermal conductivity of a series of experimental ceramic disks. The effect of production parameters on the thermal conductivity was assessed by examining mixtures of two different base clays, one non-calcareous and one calcareous, which were tempered with different amounts of granite or phyllite and fired at three specific temperatures: 550°C, 850°C and 1050°C. The selection of the examined ceramic fabrics was based on archaeological evidence from the Aegean Bronze Age. The most important parameters emerged to be degree of vitrification and porosity of the ceramic matrix. Keywords: cooking pots, thermal conductivity, heat capacity, temper

INTRODUCTION Among the most significant advantages of ceramic materials are their remarkable thermal properties. Ceramics are heat resistant, i.e. they remain inert up to considerably high temperatures and they present relatively high thermal shock resistance and heat capacity. For these reasons they were the most common materials for various applications involving the use of heat: ranging from daily life, such as cooking, to specialised technical processes, as for example metallurgy or glass making. Depending on the application, however, different properties were required regarding the heat capacity and particularly the thermal conductivity. It can be argued that in the case of cooking pots, for example, ceramics with comparably high thermal conductivity were required, in order to provide optimum heat transfer. But also mechanical properties such as strength and toughness had to be suitable (Tite et al. 2001, Müller et al. this volume).

Fig. 1 Schematic model of a ceramic microstructure as multiphase composite: In this example the phases are the clay body, the voids (black) and two different kinds of nonplastic inclusions (white and light grey). The individual phases affect the heat transfer in the ceramics differently. when applying an external heat source (Skibo et al. 1989; Schiffer 1990).

The physicochemical properties of ceramics depend on the nature of the raw materials, on the clay paste processing and on the firing technology. Indeed, various approaches can be observed in ancient functional ceramics, towards achieving suitable material properties. As for the thermal properties of cooking pots, most of the discussion so far was focused on thermal shock resistance, with the thermal conductivity referred to only in side notes as being controllable trough wall thickness (Tite & Kilikoglou 2002; Broeksman et al. 2004). The development towards thinner walls, which enhanced the heat transfer, was explained also with changes in nutrition patterns (Braun 1983). An actual approach to quantify heat transfer in cooking pots was the definition of 'heating effectiveness' as a parameter, which describes the rate at which the temperature of a pot’s content is raised

During simulated cooking tests, water was heated in replicate miniature vessels over an open flame under sufficiently controlled conditions. Nevertheless, 'heating efficiency' is a rather arbitrary parameter being a complex product of heat conductivity, heat flux, heat capacity, permeability and shape of the vessels. Therefore, it does not provide thorough information about principles that control these properties. Another approach to assess the thermal properties of ceramics is the use of computer simulations (Hein & Kilikoglou 2007). Here, the ceramic microstructure is regarded as a multiphase composite, with the clay body, the voids and the particular inclusions being the different phases, all of them having specific thermal properties (Fig. 1). The voids for example act as heat barriers.

15


Table 1 – Clay mixtures used for the experimental ceramic disks: The base clays came from Pikermi and Kalami (both in Greece) and they were tempered with different amounts of granite and phyllite material. In the last columns the total porosities of the ceramic disks fired at different temperatures are listed.

PIK -UT PIK-G10 PIK-G40 PIK-P10 PIK-P40 KAL-UT KAL-G10 KAL-G40 KAL-P10 KAL-P40

base clay

temper

calcareous calcareous calcareous calcareous calcareous non-calcareous non-calcareous non-calcareous non-calcareous non-calcareous

non-tempered 10% granite 40% granite 10% phyllite 40% phyllite non-tempered 10% granite 40% granite 10% phyllite 40% phyllite

550 ºC 32 30 29 30 30 34 32 31 33 32

porosity in % 850 ºC 1050 ºC 32 33 31 32 31 32 31 33 33 34 33 11 31 13 30 22 34 12 32 19

particular ceramics can be developed. These models can then be tested for their simulated performance during use, in terms of heat transfer and thermal stress. The results demonstrate the importance of parameters controlled by ceramic technology and constituting the thermal performance of the final products. In order to verify results from computer simulations, the thermal properties of test ceramics with controlled microstructures were measured in laboratory experiments, which will be presented in this paper. Furthermore, the determined data will provide the basis for further, more complex simulations which consider also the shape of cooking vessels, following an approach applied in a recent study on smelting furnaces used in metallurgy (Hein & Kilikoglou 2007). It is hoped that the data provided by this work will provide a further parameter in the discussion of cooking ware ceramic technology and the choice of particular raw materials.

Fig. 2 Principle of the measurement of thermal conductivity with a Lees’ disk setup: The ceramic disk to be measured is placed between a heat source with controlled and regulated temperature and a heat conductor which transfers heat to the ambient air. The heat loss of the heat conductor in relation to its temperature is known. Therefore the thermal conductivity of the test disk can be determined from the temperature difference between heat source and heat conductor after steady state is reached.

EXPERIMENTAL In order to study the thermal conductivity of typical cooking ware pottery experimental ceramic disks were prepared from ten different clay mixtures, with noncalcareous or calcareous base clays and different amounts of granite or phyllitic temper materials, each fired at three different temperatures (Table 1) (Müller et al. this volume). For the determination of their thermal conductivity a modified Lees' disk setup was used (Fig. 2). The examined sample in the shape of a flat disk is in contact with a heat source at one side and with a heat conductor at the opposite side. The temperature of the heat source is controlled and stabilized during the measurement.

However, while some properties as e.g. the effective heat capacity of the composite can be estimated in a straightforward way as the sum of the heat capacities of the individual phases weighted by their volume fraction, the effective thermal conductivity is affected also by the spatial distribution and the shape of the individual phases. Therefore, based on observations multiphase models of ceramic structures are developed and tested with computer simulations for their effective thermal properties. If the properties of the base material components are known, computer models of

16


Fig. 3 Thermal conductivity of the disks from the non-tempered clays at specific temperatures: Presented are the measured values for the non-calcareous Kalami clay (left) and the calcareous Pikermi clay (right). The symbols correspond to the firing temperatures: ♦- 550°C, ▼ - 850°C and ● - 1050°C. thermal conductivity of the ceramics manufactured from the calcareous clay is actually slightly higher than one of the ceramics produced from the non-calcareous clay. However, when fired at 1050ºC, the thermal conductivity of the non-calcareous ceramics was approximately twice as high as the value of the calcareous ceramics. One explanation for this is the clearly lower porosity of the high-fired non-calcareous ceramics (Table 1). The effect of porosity, however, as it can be estimated, is too small to explain the entire difference. Another explanation is provided by the difference in the development of the ceramics’ microstructure during firing. In the case of the calcareous clay a fairly continuous development of the microstructure, reflected in the degree of vitrification at different firing temperatures, is observed. In contrast to this, the non-calcareous clay experiences only small changes in its microstructure at firing temperatures of 550ºC and 850ºC, but when fired at 1050ºC a high degree of vitrification with extensive areas of solid glass is observed.

Beforehand, the heat loss of the heat conductor into the environment depending on the temperature q&loss (T ) is estimated. The heat loss from the rim of the sample disk can be neglected provided that its thickness is small compared to its diameter. In this way the heat flux through the sample can be determined. When steady state is reached, the temperature difference between heat source and ceramic surface in contact with the heat conductor, dT = T0 – Tsurf , provides the thermal conductivity of the ceramic, taking into account the area A and the thickness x of the sample: (1)

k (T ) = q& loss (T ) ⋅

x A ⋅ dT

The heat loss of the brass disk was determined by heating it up to a temperature of 400ºC and afterwards leaving it for cooling down while recording its temperature. During cooling the surface, which is in contact with the sample during a normal measurement, was placed on a thermal insulator. Therefore, the measured temperature decay allows for the determination of the heat flux q&loss (T ) from the brass

Effect of tempering According to Table 1, only in the case of high-fired ceramics produced from clay mixtures with the noncalcareous base clay the total porosity was significantly different from all other ceramic disks, i.e. clearly smaller. Therefore, the absolute amount of porosity cannot be the only reason for differences of the effective thermal conductivity. It remains to be discussed up to which extent observed differences in pore shape or crack development at different firing temperatures and with different tempering materials (Müller et al. this volume), affect the effective thermal conductivity. However, the measurements indicated a difference between granitic and phyllitic temper. For ceramics fired to 550°C and 850°C, moderate granite tempering appeared to increase the thermal

disk to the environment at specific temperatures.

RESULTS AND DISCUSSION Effect of clay type and firing temperature As for the non-tempered ceramics, both clay types show increased thermal conductivity with increasing firing temperature (Fig. 3). In the case of the noncalcareous clay a significant increase in thermal conductivity was observed between 850ºC and 1050ºC. For firing temperatures at 550ºC and 850ºC, the

17


optimum heating rate also depended on the supposed use of the cooking vessel.

conductivity, while tempering with phyllite appeared rather to decrease the effective thermal conductivity. An explanation might have been the different nature of the tempering materials: granite as a plutonic rock with low porosity has a higher thermal conductivity than the base clay, while phyllite possesses an anisotropic thermal conductivity. Due to the production process of the disks, the platy phyllite particles are oriented parallel to the surfaces, perpendicular to the heat transfer direction, resulting in decrease of thermal conductivity.

Thermal stress Thermal shock resistance is a complex parameter affected by heat transfer, thermal expansion, strength and toughness. Furthermore, it depends on the vessel shape. After determining the basic materials properties (Müller et al. this volume) all these parameters can be considered and evaluated in computer models of the vessels, the simulated use of which can be examined for example with the Finite Element Method (FEM) (Hein and Kilikoglou 2007).

An exception of these observations was the case of the high-fired ceramics. Here, the effective thermal conductivity decreased with both tempers, particularly using the non-calcareous base clay. This behaviour is probably due to the development of cracks that act as heat barriers. Moreover, it must be considered that the non-calcareous base clay presents a comparably high thermal conductivity presumably in the same range as the thermal conductivity of the granite temper.

CONCLUSIONS The present work demonstrates the role of raw materials (base clay and temper), manufacturing process and firing temperatures in the thermal conductivity of the ceramic products. At the same time it became evident that the thermal performance of the ceramic material is not in a straightforward relationship with each of the above parameters, but the effect is complex and sometimes the effect of one of them obscures the effect of another. Therefore, when explaining archaeological ceramic technology and choice of materials in view of thermal performance one should consider all the parameters collectively and furthermore availability of materials but also existing traditions. The next step after the determination of the basic material parameters will be their employment in computer models in order to simulate the use of the cooking vessels and their behaviour under thermal stress.

In summary, for low to intermediate firing temperatures, moderate tempering with a material that has higher thermal conductivity than the base clay indeed seems to increase the effective thermal conductivity of the ceramics, at high firing temperatures this effect seems to be disturbed by crack development due to thermal expansion. Increased amounts of temper appear to diminish the effective thermal conductivity, probably due to the presence of manufacture induced macro-pores resulting in a microstructure in which the ceramic matrix is less connected. Heat capacity Heat capacity affects, together with thermal conductivity and density, the rate of the temperature increase during heat transfer under transient conditions, i.e. when the temperature of the material is not in equilibrium. This relation is expressed by the parameter thermal diffusivity α, which is the quotient between the thermal conductivity k and the heat capacity per unit volume ρ ⋅ cp. The slope of the heating curve and particularly the time after which steady state is reached depend on the thermal diffusivity of the ceramics. The higher the thermal diffusivity, the faster the temperature increases and equilibrium is reached. Therefore, the measurements with the modified Lees’ disk in principle also provide information about thermal diffusivity or the heat capacity, respectively. This, however, is still subject to further examination. In the present study the effect of heat capacity might be neglected due to rather small variations of the heat capacity of the particular clays and temper materials. As for the examined clay mixes, thermal conductivity and density are expected to affect the thermal diffusivity to a greater extent. After all, the

ACKNOWLEDGEMENTS This study was partly funded by the Institute for Aegean Prehistory (INSTAP).The support of N.S.M. through an IKY (Greek State Scholarship Foundation) scholarship is gratefully acknowledged, as is an ORSAS Scholarship awarded by Universities UK.

REFERENCES BRAUN, D. (1983): Pots as tools, in Archaeological hammers and theories (eds J.A. Moore and A.S. Keene), 107-134, Academic Press, New York. BROEKSMAN, T., ADRIAENS, A.. & PANTOS, E. (2004): Analytical investigations of cooking pottery from Tell Beydar (NE-Syria), Nuclear Instruments and Methods in Physics Section B, 226 (1-2) 92-97.

18


HEIN, A. & KILIKOGLOU, V. (2007): Modeling of Thermal Behavior of Ancient Metallurgical Ceramics, Journal of the American Ceramic Society 90, 3, 878 – 884.

SKIBO, J.M., SCHIFFER, M.B. & REID, K.C. (1989): Organic-tempered pottery: an experimental study, American Antiquity, 54 (1), 122-146. TITE, M. S., KILIKIOGLOU, V. & VEKINIS, G. (2001): Strength, toughness and thermal shock resistance of ancient ceramics, and their influence on technological choice. Archaeometry 43 (3):301-324.

MÜLLER, N.S., KILIKOGLOU, V., DAY, P.M., VEKINIS, G & HEIN, A.: The influence of temper on performance characteristics of cooking ware ceramics, this volume. SCHIFFER, M.B. (1990): The influence of surface treatment on heating effectiveness of ceramic vessels, Journal of Archaeological Science, 17, 373-381.

TITE, M. S. & KILIKOGLOU, V. (2002): Do we understand cooking pots and is there an ideal cooking pot?, in Modern trends in scientific studies on ancient ceramics (eds. V. Kilikoglou, A. Hein and Y. Maniatis), 1–8, BAR International Series 1011, Archaeopress, Oxford.

19


20

MARA & AL.: 3D-ACQUISITION OF ATTIC RED-FIGURED VESSELS AND MULTI-SPECTRAL READINGS OF LEKYTHOI

3D-ACQUISITION OF ATTIC RED-FIGURED VESSELS AND MULTISPECTRAL READINGS OF WHITE GROUND LEKYTHOI IN THE KUNSTHISTORISCHES MUSEUM VIENNA 1 1

H. Mara – 2E. Trinkl – 3P. Kammerer – 3E. Zolda

IWR - University of Heidelberg Im Neuenheimer Feld 368 69120 Heidelberg Germany, [email protected] 2 Austrian Academy of Sciences, Institute for Studies of Ancient Culture, [email protected] 3 Vienna University of Technology, Institute for Automation Pattern Recognition & Image Processing Group [email protected], [email protected]

Abstract: Motivated by archaeological requirements we are developing an automated system using 3D-acquisition based on structured light for documentation of ancient ceramics. Furthermore we are developing a system for art-historic analysis of medieval paintings using multi-spectral readings of colour pigments. Documentation of polychrome pottery of the Collection of Greek and Roman Antiquities of the Museum for History of Art in Vienna (KHM) required the combination of both systems for documentation, classification and (virtual) restoration tasks. Therefore we show the combined methods of our systems for digital, contact-free, radiation-free acquisition of 3D-models including multi-spectral readings of painted ceramics. Keywords: 3D-acquisition, multi-spectral readings, pottery, 3D-model, Corpus Vasorum Antiquorum

INTRODUCTION As we all know, the analysis of ceramics reveals information about the age, trading relations, advancements in technology, art, politics, religion and many other details of ancient cultures.

Fig. 1 (a) Attic red-figured chous KHM IV 1043. (b) Attic white ground lekythos KHM IV 3745 (© KHM).

21


Fig. 2 3D-acquisition of a vessel in the Kunsthistorisches Museum Vienna.

Fig. 3 UV/VIS/NIR Spectrometer (Perkin Elmer Lambda 900).

This is often a time consuming process and requires a lot of skill and manpower of experts. We are developing an automated system for documentation of pottery to help archaeologists to document their finds efficiently and accurately, one which can be used for further (computerized) research. The base of documenting pottery is a vertical intersection which is called a profile line. The intersections are usually drawn by hand. The PRIP-group provide a computer aided documentation method by using an automated system for acquisition and documentation of ceramics using a 3D-scanner based on the principle of structured light (DePiero & Trivedi 1996; Liska 1999). It was successfully tested on sherds as well as on unbroken vessels in interdisciplinary projects both in excavations and in museums (Kampel & Sablatnig 1999; Cosmas et al. 2001). In parallel the same group worked on a second system for analysis of medieval paintings in which pattern recognition techniques are used to detect and to classify brush strokes of under-drawings (Kammerer et al. 2007; Asinger et al. 2005).

The method which is used for the estimation of volumes and geometry will be of special interest for future work. METHODS 3D-Acquisition and Processing For acquisition we use a Konica-Minolta 3D-scanner based on the principle of structured light (Tosovic 2002) having a resolution of 20% CaCO3) clay from Pikermi (Attiki, Greece) and a non-calcareous clay, from Kalami (Crete, Greece).

145


RESULTS Fracture strength

Fig. 1 Cross section of replicate briquettes, containing 10% of granite (a) and 10% of phyllite temper (b) respectively. Fractions with a particle size of 15 µm Monocrystalline quartz (v%) Polycrystalline quartz (v%) Mica (v%) Feldspar (v%) Rock fragment (v%) Opaque minerals + organic material (v%) Accessories (v%)

Vörs (n=3)

Kup (n=3)

average

st.dev.

average

st.dev.

66,4

1,59

55,2

1,6

7,7

6,64

14,0

3,3

10,4 4,0

8,71 3,44

6,5 6,4

1,0 0,7

3,6

3,30

12,5

2,2

8,0

6,18

5,3

1,9

Traces

traces

On the other hand the use of organic tempering is proved by the presence of carbonized plant remains (Fig.6), the amount of which may vary significantly and was an important parameter for the classification.

RESULTS AND DISCUSSION Petrography The classification of the fragments was based on textural features - grain size, porosity - measured with the thin section analysis - and non-plastic component/matrix ratio, amount of organic tempering. The microscopic analysis proved the reliability of the macroscopic groups, which are the followings: (i) group of compact, low porosity (13%) samples with relatively coarse (silt-fine-sand) grain size and with scarce vegetal remains – fragments of the floor; (ii) group of samples with higher porosity (8-15%), silty average grain size and with a large amount of vegetal remains – fragments of the wall; (iii) samples with a layer of 0,5-1 cm large pebbles on the bottom and with low porosity (1-3%) – fragments of kilns. In most of the groups smoothed, occasionally painted surfaces can be found on some fragments. The three groups are identical at both sites. In Fig. 3 one example of each group of the Kup site is shown. The mineralogical composition of the daub samples is very homogenous on both sites and within all the groups, the results of the semi-quantitative analysis of the nonplastic components are given in Table 1. The differences between the two sites clearly originate from the different sedimentology of the areas. In each group, the grain size and the quantity of the nonplastic clasts in the matrix can be very diverse (Fig. 4, 5) even in one piece, but their distribution is serial. There are no signs of artificially added non-organic temper except for burnt, maximum 1 mm large fragments of former plasters or ceramic fragments.

Based on the observations by SEM-EDAX the very finegrained matrix consists of 10-20 μm size grains of the same types of minerals as the larger non-plastic components (Fig. 7) and various amounts of disperse carbonate. With this technique we could find another type of organic material in the samples – identified by their high phosphorous content – in the form of irregular shaped nodules of 100 μm scale, which may be of animal or vegetal origin. The white painted layer with a thickness of a few 10 μm has a chemical composition similar to apatite, which suggests bone grist as raw material (Fig. 8) the use of which was discovered at other archeological sites as well. (e.g. Çolak et al. 2001). Comparison with local soils A granulometric and mineralogical investigation and comparison of the daub and soil samples of both sites were carried out by means of binocular microscopy and XRD analysis. In those parameters that can be observed with a stereomicroscope, such as the mineralogical composition and the roundness of the clasts, the daub and soil samples originating from the same sites proved to be similar. In Figs. 9 and 10 the diffractograms of XRD analysis of four representative daub samples and four soil samples can be found. The analysis has proved that they contain approximately 70% quartz and smaller quantities of micas (the 10Å phase), feldspars and in some cases, carbonates (calcite and less dolomite).

182

KOVÁCS & AL.: DAUB: BETWEEN POTTERY AND SEDIMENT

Fig. 9 X-ray diffractograms of four representative daub samples. (1.1.4/1 and 1.7.3/2 from Kup-Egyes site; 3.1.8 and 3.1.9. from Vörs-Máriaasszonysziget site)

Fig. 10 X-ray diffractograms of four soil samples. ( VMSz. 43.; VMSz. 70., VMSz. 71. from Vörs-Máriaasszonysziget site and KE.XII/2003/1. from Kup-Egyes site)

The only differences emerged in the clay mineral content of the daub and the soil samples: there are no clay minerals in the daub, not even in the