UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The Middle Palaeolithic of present day Romania : a critical review Pop, Cornel Marian 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_pop_cornel.pdf [ 1.64MB ]
Metadata
JSON: 24-1.0073841.json
JSON-LD: 24-1.0073841-ld.json
RDF/XML (Pretty): 24-1.0073841-rdf.xml
RDF/JSON: 24-1.0073841-rdf.json
Turtle: 24-1.0073841-turtle.txt
N-Triples: 24-1.0073841-rdf-ntriples.txt
Original Record: 24-1.0073841-source.json
Full Text
24-1.0073841-fulltext.txt
Citation
24-1.0073841.ris

Full Text

THE MIDDLE PALAEOLITHIC OF PRESENT DAY ROMANIA: A CRITICAL REVIEW by CORNEL MARIAN POP B.A., The University of British Columbia, 2010  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Anthropology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2013  ©Cornel Marian Pop, 2013  Abstract This thesis presents a critical review of research on the Middle Palaeolithic record of present day Romania. Through an analysis of the historical development of the discipline as well as qualitative and quantitative assessments of the published data, I argue that despite a history of research that spans over a century, the Romanian Middle Palaeolithic record is poorly understood, even from a culture historical perspective. Although over 240 locations with potential Middle Palaeolithic finds have been identified to date, of which 120 have yielded artifacts confidently classified as Middle Palaeolithic, only approximately 21 of these have been systematically excavated and adequately published. These sites constitute a sample shown to evidence statistically significant patterns that are most likely a reflection of research bias rather than underlying archaeological phenomena. Nevertheless, through a careful exploratory analysis of the data promising directions for future research are identified, as well as a series of patterns which reveal that Neanderthal populations likely had limited adaptive capacities that rendered them unable to cope with the harshest climatic regimes. Based on currently available figures, Romania appears to have been sparsely populated by Neanderthals throughout, its Middle Palaeolithic record yielding less than approximately 100,000 lithic artifacts (including debris), the majority of which were recovered from but two levels at a single site: Ripiceni-Izvor.  ii  Preface This thesis is original and independent work by the author, Cornel M. Pop.  iii  Table of Contents  Abstract ..................................................................................................................................................... ii Preface ..................................................................................................................................................... iii Table of Contents ..................................................................................................................................... iv List of Tables............................................................................................................................................. vi List of Figures .......................................................................................................................................... vii Acknowledgements ................................................................................................................................ viii Dedication ................................................................................................................................................ ix 1.  Introduction ......................................................................................................................................1  2.  Data and Methods.............................................................................................................................4  3.  A brief history of Romanian Middle Palaeolithic research ...............................................................6 3.1.  4.  3.1.1.  Márton Roska (1880-1961) ...............................................................................................7  3.1.2.  Nicolae Moroşan (1902-1944) ..........................................................................................8  3.1.3.  C. S. Nicolăescu-Plopşor (1900-1968) and the early socialist period ................................9  3.2.  Research after Nicolăescu-Plopşor .........................................................................................14  3.3.  Research after socialism..........................................................................................................17  3.4.  Summary .................................................................................................................................17  The Romanian Middle Palaeolithic record as it is known today .....................................................18 4.1.  General overview ....................................................................................................................18  4.2.  Chronology ..............................................................................................................................22  4.3.  Fauna and Homo remains .......................................................................................................24  4.4.  Lithic Industries .......................................................................................................................25  4.4.1.  Typical Mousterian ..........................................................................................................25  4.4.2.  Mousterian of Acheulean Tradition (MTA) .....................................................................26  4.4.3.  Denticulate Mousterian ..................................................................................................27  4.4.4.  Eastern Charentian or “Quartzitic Mousterian”..............................................................27  4.5. 5.  Early Palaeolithic research in Romania .....................................................................................6  Summary .................................................................................................................................29  A quantitative assessment of the Romanian Middle Palaeolithic record .......................................30 5.1.  General overview ....................................................................................................................32  5.2.  Climate and the distribution of sites .......................................................................................35 iv  6.  5.3.  Raw materials and assemblage variability ..............................................................................37  5.4.  Summary .................................................................................................................................46  Conclusions .....................................................................................................................................47  Works Cited .............................................................................................................................................50 Appendix A: A partial database of the Romanian Middle Palaeolithic record........................................57 Database goals and availability: ..........................................................................................................57 Variable selection criteria: ..................................................................................................................58 Database structure and contents:.......................................................................................................59 Variable definitions: ............................................................................................................................59 Primary variables:............................................................................................................................59 Auxiliary variables: ..........................................................................................................................65  v  List of Tables Table 1: Distribution of different raw material types across levels excavated at open-air and cave sites .....................................................................................................................................................................35 Table 2: Distribution of predominant raw material types across levels deposited in different climatic contexts .......................................................................................................................................................35 Table 3: The distribution of climatic conditions across the different levels and sites for which valid data is available. .....................................................................................................................................................37  vi  List of Figures Figure 1: Composite map of Romania showing: a) the location of the 21 adequately researched sites analyzed in this thesis, b) a kernel density map of the 120 locations where lithics attributable with confidence to the Middle Palaeolithic have been found, and c) the distribution of karst formations. .....22 Figure 2: Distribution of Middle Palaeolithic levels across the 21 sites analyzed in this thesis. ................33 Figure 3a,b: Incidence of formal tools (a) and the diversity of tool types (b) across levels characterized by the use of different raw material types. ................................................................................................41 Figure 4a,b: Ratio of retouched pieces to platform remnant bearing flakes (PRBs) across levels characterized by the use of different raw material types...........................................................................43 Figure 5a,b: Relationship between the percentage of retouched pieces and the volumetric density of lithics. ..........................................................................................................................................................44 Figure 6a,b: Relationship between (a) the incidence of retouch and fine-grained raw materials and (b) between the incidence of fine-grained raw materials and the volumetric density of lithics. ....................45  vii  Acknowledgements I would like to thank my supervisor, Dr. Mike Richards, for his guidance, encouragement, and support at all stages of my research; without his help the writing of this thesis would not have been possible. I would also like to thank Dr. David Pokotylo, my committee member, for his constant willingness to help and for introducing me to the world of statistical analysis in a way that made it a thoroughly enjoyable experience. I would like to thank Dr. Marin Cârciumaru for being extremely accommodating and helpful, in particular for letting me participate in fieldwork with his team in 2011 on a short notice, for allowing me to view the collections from Bordul Mare and Peştera Cioarei, and for sharing with me raw material samples and much of his personal bibliography; I cannot emphasize my gratitude enough. I would also like to thank his student, Dr. Elena-Cristina Niţu, for being equally helpful and accommodating. Furthermore, I would like to thank Dr. Elena Terzea, Dr. Vasile Boroneanţ, Dr. Maria Bitiri, Dr. Gheorge Lazarovici, and Dr. Mircea Anghelinu for taking the time to meet with me to discuss the history of Romanian archaeology and the Romanian Palaeolithic record. Many thanks are also due to Dr. Radu Ioviţă and Dr. Adrian Doboş for allowing me to participate in fieldwork as a volunteer for the Lower Danube Survey for Palaeolithic Sites project in 2010, for the many helpful discussions we had, and for allowing me access to bibliographic materials. Finally, I would like to express my enormous gratitude to my wife, Cecilia Pop, and to my parents, Veronka and Nicolae Pop, for their continued support and patience. Without their support my research would have been impossible. This research was funded in part by a Joseph-Armand Bombardier CGS – Master’s Award scholarship (award number #766-2011-1156) from the Social Sciences and Humanities Research Council of Canada, and a University of British Columbia Faculty of Arts Graduate Entrance Scholarship.  viii  Dedication To my wife, for her unending patience and support.  ix  1. Introduction A number of recent publications have brought to light intriguing aspects of the Middle and early Upper Palaeolithic record of present day Romania. These include the discovery of some of the earliest anatomically modern human (AMH) remains in Europe at Cioclovina, Peştera Muierii, and Peştera cu Oase (Soficaru et al. 2007; Soficaru et al. 2006; Trinkaus et al. 2003), evidence for possible interbreeding between AMHs and Neanderthals (Soficaru et al. 2006), and signs of behavioural continuity across the transition to the Upper Palaeolithic (Riel-Salvatore et al. 2008). Moreover, it has been suggested that there is good evidence for the presence of transitional industries (Horvath 2009; Cârciumaru 1999), and a late survival of the Mousterian (Cârciumaru et al. 2007). The discoveries regarding the Middle Palaeolithic proper have been equally interesting: Cârciumaru et al. report on the discovery of ochre containers at Peştera Cioarei (Cârciumaru et al. 2002; see also Cârciumaru and Ţuţuianu-Cârciumaru 2009) and evidence for bitumen use and hafting at Râşnov – Gura Cheii (Cârciumaru et al. 2012), while Păunescu (1993) long ago reported numerous and varied Middle Palaeolithic habitation structures at Ripiceni-Izvor. Furthermore, Neanderthal phalanges have been unearthed at Ohaba Ponor (Gaál 1928; Gaál 1943) and Livadiţa (Terzea 1979: 114; see also Cârciumaru 1999: 65-70), and footprints have been discovered at Vârtop (Onac et al. 2005). Other, more controversial finds have been interpreted as evidence for bone tool use (Păunescu 1989; Cârciumaru 1999; Horvath 2009), possible cooking pits (Cârciumaru 1999: 77), and even a cave bear cult among Neanderthals (Lascu et al. 1996). Considered together, these discoveries show that the Romanian Palaeolithic record has much to contribute to current debates regarding the capabilities and lifeways of Middle Palaeolithic hominins and the colonization of Europe by anatomically modern humans. However, in spite of the recent publication in English and French of syntheses, reviews, and articles that have dealt at some length with the Romanian Palaeolithic (Cârciumaru et al. 2007; Cârciumaru 1989, 1999; Păunescu 1989; Horvath 2009; Mertens 1996; Riel-Salvatore et al. 2008), relatively little is known about the Middle Palaeolithic record of this interesting region, particularly outside Eastern Europe. The aforementioned publications certainly provide a wealth of information regarding the various lithic industries that have been identified to date, the geochronology of Middle Palaeolithic habitation, and the history of research (particularly Horvath 2009) but, with the exception of Riel-Salvatore et al. (2008), they provide a purely qualitative assessment of the record and generally fail to evaluate the impacts of methodological and theoretical biases (but see Horvath 2009). Furthermore, with the exception of Mertens (1996), none of these publications focus exclusively, or even primarily, on the Middle Palaeolithic. As a result, basic information such as the number and richness of Romanian Middle 1  Palaeolithic sites remains elusive. More importantly, the general lack of critical perspectives on past research renders the proper identification of issues and problems affecting current interpretations of the record, as well as of the most promising areas for future research, difficult. The aim of this thesis is to present and critically assess the information available on the Romanian Middle Palaeolithic from both a qualitative and a quantitative perspective. In adopting a critical stance I do not aim to judge the competency of Romanian researchers or to compare the Romanian case with that of other parts of the world; such judgements or comparisons would be unjustified and could, in any case, only be conducted by fully taking into account the difficult geopolitical, economic, and social circumstances of Romania in the twentieth century. Rather, my goal is to highlight issues and problems that should be considered before including Romanian data in regional or continental scale analyses, and to identify useful avenues for future research. The important discoveries that have been made to date, as well as the geographic position of Romania along the Danube corridor, postulated as an important migration route for early anatomically modern humans into Europe (e.g. Mellars 2004), demand renewed research efforts in the area; my goal is to provide an informative discussion that can hopefully facilitate such efforts. My overall analysis is organized into three main chapters. One core objective pursued here is the identification and evaluation of issues pertaining to the generation and interpretation of past results, and as such a considerable portion of this thesis (Chapter 3) is devoted to a discussion of the history of research. A brief qualitative assessment of the current state of knowledge regarding a) the distribution of known versus well-researched sites, b) the chronology of the Middle Palaeolithic record, c) the various industries identified to date, and d) the faunal record and Homo discoveries is then presented in Chapter 4. In that chapter, sites which have been systematically excavated, are well-published, and have yielded more than a handful of lithics are also identified. Chapter 5 is devoted to a second core objective, namely the presentation and quantitative analysis of data available for the relatively wellresearched sites, with a primary focus on the role played by climate and the use of different raw materials in the variability evidenced in assemblage composition and the distribution of sites. Unlike previous quantitative studies of the Romanian data (e.g. Riel-Salvatore et al. 2008; Doboş 2010), which have focused primarily on testing specific hypotheses and which have considered only a sub-set of known sites, I analyze the Romanian data as a whole through the conceptual lens of exploratory data analysis (EDA). The methods and data used in this thesis are discussed in Chapter 2. Based on the analyses outlined above, I argue that the known Middle Palaeolithic record of Romania is neither particularly rich (the sum total of recovered lithics, including shatter, is likely less than 100,000) 2  nor extensively studied, even from a culture-historical perspective. As shown in Chapter 4, many regions have been inadequately researched, and the number of systematically excavated and well-published sites (roughly 20 to no more than 40, depending on the criteria used) relative to the number of locations thought to contain reliable traces of Middle Palaeolithic habitation (approximately 120) is low. Moreover, the sample constituted by well-known sites is biased, rendering the interpretation of patterns evidenced in the data problematic. Still, it is shown that the available data allow for some important observations to be made, concerning both promising future directions and the results of past research, which suggest that Neanderthal populations had a limited capacity to adapt to harsh climatic regimes and that the territory of present day Romanian was sparsely inhabited throughout the Middle Palaeolithic.  3  2. Data and Methods This thesis is largely based on an extensive review of the Romanian literature, as well as on insights derived from three months of fieldwork conducted in 2010 and 2011, which included surveying in southeastern Romania (Dobrogea) as a volunteer with the Lower Danube Survey for Palaeolithic Sites (LoDanS) project (2010), and in the Haţeg region of western Romania as part of a team led by Professor Marin Cârciumaru (2011). Fieldwork also included visits to many of the known Middle Palaeolithic sites1, collection of raw material samples2, and discussions with several of the leading twentieth century Romanian researchers, including Dr. Marin Cârciumaru, Dr. Maria Bitiri, Dr. Elena Terzea, and Dr. Vasile Boroneanţ. The quantitative analyses to which Chapter 5 is devoted are based on data compiled mostly from Păunescu’s comprehensive synthesis and repertory of Romanian Palaeolithic finds (Păunescu 1998, 1999a, 1999b, 2000, 2001). These data have been made available in the form of an electronic database to facilitate further exploration (see Appendix A). Păunescu’s works were chosen because they provide quantitative data on the various assemblages that are compatible and therefore useful for interassemblage comparisons. Păunescu analyzed much of the lithic material recovered from the various sites according to common criteria and following the same analytical standards, providing typological and technological descriptions of the lithic components that generally included core, shatter, debitage and tool counts, as well as quantitative data on the incidence of particular raw materials and different platform and core types. Other data, such as retouch types, mass of the lithic material, and the size distribution of various pieces is also provided in some cases, but inconsistently so. These works are certainly not perfect, as Păunescu’s identification of the lithic materials is at times questionable, and his analyses were in some cases based on a relatively small sub-set of the recovered lithics (e.g. at Remetea-Somoş). Moreover, no information on the volume of excavated sediments is provided (although this is true of the vast majority of Romanian publications), and in specific cases other publications provide more details – Cârciumaru et al.’s (2000) analysis of the assemblages from Peştera Cioarei, for example, provides such details as the minimum number of cores from which the flaked  1  The visited sites included: Boineşti, Boroşteni – Peştera Cioarei, Buşag, Nandru – Peştera Curată, Nandru – Peştera Spurcată, Ohaba Ponor – Bordul Mare, various locations around Gurahonţ (Iosăşel), Brotuna – Coasta Cremenii, Râşnov – Gura Cheii, Peştera – Peştera Mare, Peştera Muierii, Peştera Hoţilor, and Cheia – Gura Cheii. The purpose of the visits was to get a sense for the placement of the sites on the landscape and for the type of raw materials available in their immediate vicinity. No sediments were examined, although the profile at Ohaba Ponor – Bordul Mare was photographed with permission from Dr. Cârciumaru. 2 Most of the raw material samples were obtained from Dr. Cârciumaru.  4  implements were derived. Overall, however, for the purposes of conducting country-wide comparisons Păunescu’s works are the only viable sources of quantitative data. Data compiled from Păunescu’s repertory were complemented in some cases with data obtained from other sources. Data on the volume of excavated sediments, for example, were estimated with apparent success by Riel-Salvatore et al. (2008) for a series of (mostly) cave sites from the Southern Carpathians, and these data were incorporated into the present analysis. Data on the climatic context in which the various levels were deposited, on the other hand, were compiled from Cârciumaru’s palynological studies (e.g. Cârciumaru 1980). Data on the lithic components from Boroşteni – Peştera Cioarei were compiled mostly from Doboş’s PhD thesis (2010), as Cârciumaru’s analysis of the lithic components are incompatible with the lithic data available for other sites, and Păunescu only provides a very summary description of these components. For this same reason, data for the site of Zăbrani were compiled from Tuffreau et al. (2007). Finally, coordinates and absolute elevations for many sites were recorded by the author in 2010 and 2011 with the aid of a Garmin Oregon 450 GPS; in other cases this information was extracted from the existing literature or, where such information could not be found in published sources, it was estimated with the aid of Google Earth. The resulting database includes information on the 21 sites identified in Chapter 4 as systematically excavated and reasonably well published. Although every attempt was made to minimize errors, the fact that multiple sources were used, that in some cases information had to be extracted from lengthy descriptions that varied considerably in detail, and that for several sites (e.g. Boineşti, Buşag, Remetea Somoş) Păunescu only provides information on a sub-set of the recovered lithics, virtually guarantees that some of the included information is inaccurate. Despite its shortcomings, however, the database likely contains the best available data for inter-site comparisons. Statistical analyses of the database were conducted from a perspective grounded in exploratory data analysis (EDA), a data-driven (as opposed to hypothesis driven) approach popularized by John Tukey (1977). EDA focuses on developing an intimate knowledge of the structure and properties of the data through rich descriptions, a skeptical attitude, and a focus on the unexpected and problematic. It is, in a sense, “detective work” (Tukey 1977: 1) that emphasizes the use of robust analytical methods and the identification of interesting patterns that can lead to the generation of new hypothesis (see Behrens and Yu 2003; Good 1983). This approach was adopted because the quality of the Romanian data is currently unknown, and EDA facilitates the detection of problems and biases; moreover, the ability to generate new hypotheses on the basis of the peculiarities of the data is considered here to be particularly important in guiding future research. 5  3. A brief history of Romanian Middle Palaeolithic research This chapter presents an introduction to, and a review of, the history of Romanian Middle Palaeolithic research. It constitutes a discussion, rather than a formal historical analysis, of various aspects of past research that are considered to be important in understanding the archaeological record as it is known today. Many vital aspects of the history of archaeological research in Romania are only superficially broached, as the scope of this thesis prevents an in-depth treatment. Much has been written, for example, on the impacts of nationalism, politics, social context, and economic systems on archaeological research (e.g. Shanks and Tilley 1989; Hodder 1991; Trigger 2006; Murray and Evans 2008). As is the case in other parts of the world, the historical trajectory of Romanian archaeology cannot be understood without considering these factors. However, I limit myself here to simply pointing out some specific instances in which these factors influenced the direction of research. For an in-depth, theoretically sophisticated analysis of the history of Romanian archaeology, the interested reader is directed to the works of Anghelinu (2003; 2004). 3.1.  Early Palaeolithic research in Romania  Riel-Salvatore et al. (2008: 400) state that the archaeological study of some Palaeolithic caves in Romania was “well underway by the mid-19th century”, but this assessment is overly optimistic and somewhat misleading. Archaeological investigations (as opposed to antiquarian collections) began mostly in the second half of the nineteenth century, and the first discovery of stone tools associated with Quaternary fauna occurred in 1876, when Koch identified lithic implements associated with Mammoth remains near the town of Rădaia (Cârciumaru 1999: 7; Păunescu 2001: 15). Research into the Palaeolithic period in the nineteenth century was sporadic and tangential at best, marked mostly by the occasional discovery of lithics by geologists, speleologists, and paleontologists, although some archaeologists did manifest a marginal interest in the topic3. It was only in the first decades of the twentieth century that a handful of scholars, mostly trained in the natural sciences, began investigating the Palaeolithic period in a systematic way. This early period, which ended with the start of the Second World War was, despite the general lack of institutional support and funding (Anghelinu 2004), one of the most fertile periods of research, in which Romanian scholars maintained close associations with  3  According to Anghelinu, Tocilescu attempted in his Dacia înainte de romani (1880) to verify the three periods proposed by Worsaae on the territory of former Dacia, adopting also Lubbock’s division of the stone age into Palaeolithic and Neolithic (Anghelinu 2004: 96). Other archaeologists of the period, such as Odobescu and Bolliac, also discussed the existence of the “age of stone” in Romania as early as 1867, but all lithic industries, including Neolithic and later, were considered by those scholars, according to Nicolăescu-Plopşor, to be Palaeolithic (Nicolăescu-Plopşor 1938: 41-42). 6  researchers in Western Europe, particularly France, and contributed to a significant degree of information flow on the international scene. 3.1.1. Márton Roska (1880-1961) One of the first and most important of these early researchers was Márton Roska, a student of Hungarian archaeologist Béla Posta at the University of Kolozsvar (Cluj). Through his travels to Western Europe Roska became acquainted with French and German archaeology4, and between 1910 and 1914 he conducted, as an employee of the Institute of Archaeology and Numismatics of Cluj, a series of excavations in the western part of present day Romania (Transylvania, then part of the AustroHungarian Empire), which covered the whole historic and prehistoric gamut. Of particular importance in this period was his 1911 excavation of the Palaeolithic site of Cioclovina, where he discovered lithic implements that he attributed to the Mousterian. This site is well-known in current literature due to the discovery, in 1940, of an early AMH cranium that has been the subject of much recent research (e.g. Harvati et al. 2007; Soficaru et al. 2007). Other sites which Roska investigated in the 1910s include Gălăteni, Peştere - Peştera Igriţa, and Şardu, but these sites have received little attention following the initial reports and are poorly known in the literature today. Roska’s activities were stalled by the First World War, in which he actively participated as an officer in the Austro-Hungarian army (Gáll 2010: 289). He returned to the Institute in 1920, but by then it was under the Romanian leadership of D. M. Teodorescu and Roska had been demoted to a simple assistant. Nevertheless, Roska resumed fieldwork in 1921, with an almost exclusive focus on Palaeolithic sites. In 1921 he returned to Cioclovina, and in 1923 he began excavating the site of Bordul Mare at Ohaba Ponor (Gáll 2010: 290), where he recovered the first Neanderthal remains (three phalanges) to be identified in the country (Gaál 1928; Gaál 1943). It was in 1924, however, that he began engaging once again in extensive fieldwork. That year the University of Cluj made available a generous sum (30,000 lei), unprecedented until then for Palaeolithic research in Romania, with the occasion of Henri Breuil’s visit, with whom Roska was to investigate a large number of sites in the region (Roska 1924; Breuil 1925). Overall, Roska discovered some 27 locations with lithic material that he attributed to the Palaeolithic, although a few of these finds were later re-interpreted by him as Mesolithic (see Păunescu 2001 for a review).  4  According to Gáll Roska visited museums in Paris and Berlin in 1908, and participated in fieldwork in Frankfurt am Main and in the International Archaeology and Anthropology Conference held in 1912 in München (Gáll 2010: 288).  7  In terms of theory and methods, Roska’s earliest works are grounded in the evolutionist perspectives of de Mortillet, whose subdivision of the Palaeolithic into Chellean, Mousterian, Solutrean, Aurignacian and Magdalenian he adopted fully (Roska 1912). By 1924 Roska’s views showed little change, as his classification of assemblages was still based on unilinear evolutionary models (Roska 1924). However, by 1933 a shift in perspective, which brought his work more in line with that of other Palaeolithic researchers of this period, is evidenced, as Roska now considered the Mousterian to be an industry rather than an epoch, one which manifested itself in different contemporaneous traditions throughout Europe (Roska 1933). In his classification of assemblages, Roska employed a variety of criteria that went beyond the simple analysis of type fossils; these included manufacturing techniques, associations with faunal remains, stratigraphy, and comparisons with other sites in the region (Roska 1912). Roska conducted relatively small and carefully planned excavations and, based on the material he collected at Bordul Mare at least (which I had an opportunity to view in 2011), it appears that he carefully labeled and recorded all lithic finds. Moreover, he collected both typical artifacts and atypical pieces; although latter researchers dismissed many of these as lacking intentional modification (e.g. Păunescu 2001), recent re-analyses of some of the materials collected by Roska suggest that his ability to identify artifacts was better than previously thought (Elena-Cristina Niţu 2011: personal communication), and that a re-analysis of the largely ignored but very rich collections he produced is warranted. In any case, due to his status as a Hungarian outsider, Roska’s work had a limited influence on subsequent research, and much of his work was dismissed, often unfairly, by later scholars. 3.1.2. Nicolae Moroşan (1902-1944) Another important Palaeolithic researcher of this early period was Nicolae Moroşan, who focused on the region of Moldova in north-eastern Romania. Although trained as a geologist under the supervision of Ion Simionescu at the University of Iaşi, Moroşan specialized in Palaeolithic research in Paris under the tutelage of Marcel Boule and Henri Breuil (Moroşan 1931: 223), benefitting from a bursary granted to him by the Romanian Academy in 1929 (Moroşan 1931: 223; Brudiu 1997: 22). During his years of study in France he prepared a regional synthesis of the Palaeolithic in Moldova, which he successfully defended as a PhD thesis at the Geology Department of the University of Bucharest in 1933 (Brudiu 1997: 22). This synthesis, published in 1938, was based on insights derived from the study of over 20 sites, which Moroşan had either discovered (e.g. the Middle Palaeolithic sites of Ripiceni-Izvor and Ripiceni-La Adăpost) or systematically excavated for the first time during the 1920s. Together with Nicolăescu-Plopşor’s synthesis of the Romanian Palaeolithic as a whole, which was published the same year, Moroşan’s paper set the standard for rigorous research methodology as well as the foundations for later multidisciplinary research. 8  Moroşan, as Peyrony and other French archaeologists of the time, showed little interest in the inner workings or properties of Palaeolithic cultural systems, and focused instead on elucidating the succession and distribution of lithic industries on the basis of strictly empirical facts and at a regional scale. Far from being a shortcoming, however, the attention devoted to lithic implements and depositional contexts resulted in some important insights that had a net positive effect on the interpretation of the Palaeolithic record. A paper coauthored by Moroşan and Renée-L. Doize, for example, provided a very interesting analysis of the mechanisms through which different natural processes, such as exposure to rapid temperature changes, can lead to breakage patterns that resemble intentional knapping, as well as of the ways in which the effects of such natural processes may be differentiated from the effects of human agency (Doize and Moroşan 1931). Moroşan’s focus on environmental factors, quality of raw materials, and the context of finds, grounded in a thorough training in the natural sciences, meant that although the identification of industries was largely based on diagnostic artifact types, the identification of these as such rested on a variety of criteria beyond simple tool morphology. These criteria are evidenced in two papers, published in 1931 and 1934, and are largely similar to those of Roska (manufacturing technique, stratigraphic context, and faunal associations), although based on a more rigorous research methodology and a more thorough understanding of these various factors (Moroşan 1931, 1934). 3.1.3. C. S. Nicolăescu-Plopşor (1900-1968) and the early socialist period Although Moroşan’s ideas had a significant impact on the discipline5, his institutional influence was limited due to his arrest by Soviet troops in 1941 and his premature death in 1944. It was therefore another scholar, C. S. Nicolăescu-Plopşor, who came to dominate Romanian Palaeolithic research until his own death in 1968. Unlike earlier researchers, Nicolăescu-Plopşor emerged within the mainstream archaeological tradition, and had considerably more institutional support. Nicolăescu-Plopşor began his career with an almost exclusive focus on the history and prehistory of his native region of Oltenia (Doboş 2005), devoting a series of publications to the topic in the 1920s and 1930s. Although his research covered a variety of issues and fields, including ethnography, folklore, and literature, it is his efforts to establish a local (pre)historic sequence that are of interest here. His attempts to uncover the origins of the local inhabitants led him to an increasing interest in the deep prehistory, including the Palaeolithic, the first vestiges of which he discovered in 1923 (Păunescu 2003: 129). As early as 1924 he published a periodization of the local prehistory that included the Palaeolithic, 5  In regards to his work on the chronology of the Pleistocene, Nicolăescu-Plopşor noted that Moroşan may rightfully be called the “founder of Rumanian Palaeolithic geochronology” (1961: 8). Păunescu also states that Moroşan can be considered the founder of modern complex Palaeolithic research in Romania (Păunescu 1998: 16).  9  the latter subdivided according to the paleontological criteria proposed by French archaeologist J. Dechelette in 1912 (Doboş 2005: 235). Gradually the Palaeolithic period became his main focus, and in 1932 he defended his PhD thesis on the topic at the University of Bucharest (Păunescu 1998: 17), where he had studied with German trained Ioan Andrieşescu (Anghelinu 2004: 148)6, the first director of the prehistoric section of the National Museum of Antiquities. In it he criticized the work of earlier, non-professional pioneers for failing to properly consider environmental factors in their studies, for simply adopting, as Roska did, chronologies developed for Western and Central Europe. The local quaternary conditions differed, according to Nicolăescu-Plopşor, from those prevalent in other parts of the world, and therefore Romanian archaeologists should strive to delineate local geological periods with the greatest precision possible by studying thoroughly their characteristic climate, flora, fauna, as well as human industries (NicolăescuPlopşor 1938: 45), before embarking on a study of Palaeolithic cultures. His thesis is in great part an attempt to do this, followed by a critical review of the available data. True to his multidisciplinary approach, which built upon but extended the work of earlier Romanian researchers such as Roska, Nicolăescu-Plopşor’s thesis shows a remarkable degree of methodological rigour, and even sophistication for the period. He reviewed Roska’s finds and concluded that the objects attributed by the latter to the lower Palaeolithic are little more than caprices of nature, explainable in terms of local natural processes. Unfortunately, despite the attention paid to geological and environmental processes, Nicolăescu-Plopşor’s approach to the study of material culture was almost entirely descriptive, in that he focused on accurately identifying and describing the variability of human industries rather than on explaining it. One significant exception to this is the explanation he provides for a quartzite based Mousterian found in the southern Carpathians: “the lack of flint forced the Mousterian inhabitants of [the Bordul Mare] cave to use quartzite and bone, which resulted in certain new [artifact] forms that Palaeolithic technological studies will have to carefully address” (1938: 94 – my translation). Given the paucity and unreliability of the available data, and Nicolăescu-Plopşor’s unwillingness to uncritically adopt western and central European definitions of geological periods, his empirical focus is hardly surprising. While Nicolăescu-Plopşor departed from French scholarship to some extent, he was nevertheless unable to escape its influence to any significant degree (Doboş 2005: 241). Although he  6  Andrieşescu had completed his post-doctoral studies in Germany, under the supervision of Gustaf Kossinna and Hubert Schmidt (Anghelinu 2004: 127).  10  framed it almost exclusively in evolutionary terms7, Nicolăescu-Plopşor adopted much of Breuil’s scheme, with whom he had studied in France for over a year (Roşu 2004: 103), and indeed, the majority of his citations regarding methodology and interpretation are French sources, although these are closely followed by German ones. Importantly, his main sources of information regarding the Romanian Palaeolithic are the works of French trained Moroşan (16 citations) and those of M. Roska (15 citations). Shortly after the publication of Nicolăescu-Plopşor’s thesis in 1938, archaeological research was stalled by the Second World War, which brought profound changes to Romania. Whereas before the war the local elites sought to situate themselves in Western intellectual circles, adopting western approaches while at the same time working towards developing a unique national identity, the 14 years (1944-1958) of Soviet military occupation and the instauration of a Soviet-backed totalitarian regime realigned Romanian scholarship with the East and resulted in the imposition of a dogmatic Marxism as the official doctrine (Anghelinu 2003: 288). Three main consequences of these political changes to Palaeolithic archaeology are particularly noteworthy. First, the modernizing agenda of the communist regime, which involved massive urbanization, education, and industrialization projects (Anghelinu 2003: 284), increased available funding, particularly for Palaeolithic research8. Secondly, archaeological research was centralized through the Institute of Archaeology, created in 1956 (Babeş 2007: 327), which in turn resulted in a “radical homogenization of archaeological discourse” (Anghelinu 2003: 285). This led to a shift in Palaeolithic research from the natural sciences to history, and whereas early Palaeolithic researchers had been trained in the natural sciences, the new generations were, for the most part, trained in the humanities. Finally, Romanian researchers became effectively isolated from the international community, in particular from the west. As Babeş points out, the Romanian academy in Rome was closed in 1947, and very few archaeologists managed to attend international conferences in the west, there being zero participation of Romanian prehistorians until 1962 (Babeş 2007: 325). In this new context Palaeolithic research thrived, mainly due to the creation of a Palaeolithic section at the new institute and the appointment of Nicolăescu-Plopşor as its director, a position that he held until his death in 1968. He began a program of intense systematic research throughout Romania which necessitated the establishment of close, if hegemonic, relations with regional museums and institutes. 7  Nicolăescu-Plopşor’s thesis reveals that he subscribed to a multi-evolutionist perspective. Thus, of the Chellean culture, for example, he notes that it emerged in different regions, such as North America, South Africa, and Europe, independently “due to identical reasons” (Nicolăescu-Plopşor 1938: 99 – my translation), although he does not elaborate on what these reasons might be. 8 As noted by Anghelinu, the official Marxist doctrine did not differentiate between historic and prehistoric periods, and whereas before the war the majority of funds were allocated to archaeological research concerned with the classical period, following the Second World War an almost equal share was available for Palaeolithic research (Anghelinu 2003: 286).  11  An integral part of Nicolăescu-Plopşor’s agenda was the training of a new generation of professional Palaeolithic archaeologists, although he managed to recruit but a few specialists concerned, to some degree at least, with the Middle Paleolithic period (e.g. Alexandru Păunescu, Maria Bitiri, and Florea Mogoşanu). These new professionals were entrusted with regional research projects (e.g. Florea Mogoşanu with the region of Banat, and Maria Bitiri with north-western Romania), which aimed to study the Palaeolithic period in its regional geographic, geological, and environmental context. Due in part to the isolation of young scholars from the international scene (all of which largely discouraged the development of new, critical approaches by young researchers), and in part to his tireless efforts, Nicolăescu-Plopşor’s agenda was to dominate Paleolithic scholarship until at least 1970 (Anghelinu 2004: 193). The direction that Nicolăescu-Plopşor set for Palaeolithic research must therefore be carefully considered here. In the post-war period his writings show significant continuity with, but also some differences from, his earlier works. His empirical focus and concern with the development and application of rigorous multidisciplinary methodologies to the delineation of local taxonomies and chronologies in particular continued to develop, while his evolutionism, although still strong, was somewhat diminished. In his latter works Nicolăescu-Plopşor still relied on type fossils, as is well evidenced both in his general discussions of assemblage classification and in his very methodology, which relied on individual proveniencing of “typical” artifacts and bulk proveniencing of “atypical” ones (Doboş 2005: 241), but their definition was increasingly based on contextual and not always clearly defined criteria in which technological considerations were given particular weight. Thus, while in France François Bordes was attempting to devise a standard typology based on observable and easily identifiable landmarks, Nicolăescu-Plopşor was advocating the need to move in the opposite direction, noting: French researchers … devised a complicated mathematical system for distinguishing different facies of the moustero-levalloisian-tayacian complex, based on a statistical analysis of the knapping technique and the typology of pieces. Armed with such pseudo-scientific methods, western science will struggle through the darkness of numbers and graphs, without being able to ever reach a positive scientific result, since [classification] is not a matter of establishing biometric indexes, based on live species with constant and transmissible characteristics. Western lithic typology fails because the raw materials used, fine or coarse grained, the hand of the knapper, virtuous or clumsy, the needs specific to the conditions of life of humans from this period, are so many unknowns that cannot be easily incorporated into the proposed formulas (Nicolăescu-Plopşor 1956: 70, qtd. in Anghelinu 2004: 195 – my translation). In print at least, Nicolăescu-Plopşor argued that archaeologists should look towards the east instead, to the research conducted by Soviet scholars (see Doboş 2005: 241). Indeed, at times he cites Soviet 12  scholarship extensively, even excessively (Anghelinu 2004: 194), as was the case with his 1957 paper (Nicolăescu-Plopşor 1957). Anghelinu contends that this was most likely due to political reasons (Anghelinu 2004: 195), but given the similarities between his multidisciplinary approach and that of Soviet archaeologists (see Vasil’ev 2002 for a brief discussion of early Soviet Palaeolithic archaeology), it is likely that his interest in Soviet archaeology was at least partially genuine. Be as it may, NicolăescuPlopşor failed to develop a viable alternative to the flawed “Western lithic typology,” as his approach remained rather subjective and somewhat hermetic, based on the belief that an intimate knowledge of the materials somehow automatically led well-trained researchers to proper identification of patterns and artifact types. Although he stressed the need to understand the specific conditions in which a tool was created, Nicolăescu-Plopşor failed, as Bordes did (see Rolland and Dibble 1990; Sackett 1981,1991; Bisson 2000), to devise analytical units that are behaviourally meaningful; his continued focus on “typical” artifact types, which were not clearly defined and whose discreteness and cognitive reality remain unclear, was useful in describing at least one aspect of variability, but less so in interpreting it. Nicolăescu-Plopşor’s empiricism and strong tendency towards the natural sciences, as well as the almost exclusive focus on constructing a “map” of the Palaeolithic period resulted in few and rather unsophisticated cultural interpretations. Anghelinu (2006: 145) notes that, although at the core of Nicolăescu-Plopşor’s multidisciplinary approach lay a concern with the influences exerted by ecological conditions on material culture, ecological adaptation never constituted an important causal argument. In fact, Nicolăescu-Plopşor made a clear distinction between the work of Palaeolithic archaeologists, who concerned themselves with interpreting lithics, and the work of social scientists, arguing that it is within the purview of the latter to interpret the archaeological finds in terms of cultural behaviour (Anghelinu 2003: 290; see also Anghelinu 2004: 194). Moreover, the analysis of faunal and botanical remains, as well as soil characteristics was left to specialists in other fields, who worked more or less autonomously towards generating data that was cross-correlated, but never fully integrated, with the results produced by experts in other fields, and consequently the potential of multidisciplinary research was never fulfilled. As was the case with other archaeologists of this period, Nicolăescu-Plopşor did not engage in high- or mid-level theorizing, and as a result questions of behaviour, function, and even postdepositional processes were only superficially broached. A significant political change in 1965 - the rise to power of Nicolae Ceauşescu - had a major impact on archaeological research. The new government sought to distance itself from Moscow and cultivated an increasingly fervent nationalism. For about a decade, a relative warming of relationships with the West and an easing of government restrictions allowed a certain degree of scientific information flow, and at 13  this time some students were given bursaries to study in Germany and France. For Palaeolithic archaeology at least, this resulted in the adoption of Bordes’ typological framework (Anghelinu 2003: 294). However, Ceauşescu’s modernizing agenda displayed little regard for the preservation of national heritage, and several important sites, such as Ripiceni-Izvor and Mamaia-Sat, were destroyed during the implementation of massive infrastructure development programs. Moreover, the new regime was concerned with elucidating and exalting the history of the Romanian people and, because the interest of officials did not extend beyond the Dacian period, fewer funds were made available for Palaeolithic research. 3.2.  Research after Nicolăescu-Plopşor  The adoption of the Bordesian system did not lead to any major interpretive advances, although it did allow for more systematic descriptions of assemblages and for more ambitious efforts to situate Romanian sites in the context of European “cultural” maps. Thus Florea Mogoşanu, for example, compared the various indices and cumulative graphs of Romanian assemblages with those obtained in other parts of Europe (e.g. Austria and Germany), in an effort to explore the links between Romanian Palaeolithic “cultures” and those of other parts of the continent (Mogoşanu 1978: 121-130). The usefulness of such comparisons was questioned by other researchers, however, who continued to show strong affinities with the earlier work of Nicolăescu-Plopşor. Maria Bitiri (1972), for example, stressed the unity of the Mousterian “culture” over the territory of present day Romania and downplayed, though never altogether rejected, the role of external influences. Whereas many assemblages were clearly distinct according to Bordesian systematics, Bitiri (1972) argued that they represented local variations of the same cultural tradition, caused by differences in environment, raw material availability, and the duration of habitation episodes. Thus, differences between the quartz/quartzite assemblages of the western Romanian Carpathians and sites such as Ripiceni-Izvor were understood to reflect the type of raw materials available in the immediate vicinity of the sites, not to differences in cultural traditions. Unfortunately, Bitiri’s arguments are difficult to follow, as they are based on comparative criteria that she fails to justify; for example, the presence of bifacials, no matter how few, is regarded as an important criterion in determining cultural affiliation, but differences in core types, or the presence or absence of the Levallois technique is not. Indeed, even differences between bifacial types are of secondary importance, as Bitiri contends that workmanship evolved over time, leading to more efficient and aesthetically superior types, so differences indicated chronological, rather than chorological distinctions.  14  In one form or another, the focus on “typical” pieces continued well after the death of NicolăescuPlopşor, as did the general understanding that aesthetics and efficiency were reliable indicators of cultural evolution (Anghelinu 2003: 206). More importantly, the focus of archaeological enquiry continued to be the identification and mapping of the cultures represented by material remains. An important consequence of this is that excavations generally proceeded until enough “typical” pieces were discovered, which usually led to the removal of very large quantities of sediment, not all of which was properly examined9. In the hunt for representative pieces atypical ones were given only secondary importance, and were at times discarded on site (e.g. at Mitoc, according to Boroneanţ [2011: personal communication], Cârciumaru [2011: personal communication], Bitiri [2011: personal communication]) due to logistical problems associated with their transportation to, and curation at, museums or research institutes. The practice of conducting large excavations as quickly as possible was not only acceptable, but to a certain degree also encouraged, in the context of the Romanian socialist system. In Romania and elsewhere in the Eastern Block, centralized planning was a core feature of the socialist regime from the beginning, and its principal objective was production rather than efficiency and high-quality results. In other words, the inner logic of the system was centered on the quantity, rather than the quality, of products (Verdery 1996). The criteria used to distribute funds, according to this quantitative focus, tended to reward large excavations, impressive discoveries, and unequivocal and rapid identification of assemblages, and to treat inconclusive finds, small and slowly executed excavations, and negative results with suspicion. One way to secure future funding, then, was to excavate as much as possible, or at the very least to report findings in a way that underscored their importance, even if no major discoveries were made10. The most extreme and also the most unfortunate example of the pace at which excavations proceeded, and the problems with the type of multidisciplinary research conducted at this time, is that of RipiceniIzvor, the largest and most complex site with Middle Palaeolithic levels discovered in Romania to date. Under the direction of Alexandru Păunescu almost 4000m2 were excavated, mostly to a depth of between 8 and 11m, in 20 campaigns (1961-1981); in other words, an average of ca. 1600-2200 cubic meters of sediment, much of which was rich in archaeological material, were removed per excavation season (see Păunescu 1993: 12-16). The remarkable finds, which included a large number of lithic 9  Indeed, archaeologists conducted excavations with the aid of modestly trained, local workers, who were at times given daily quotas for sediment removals (Bitiri 2011: personal communication) 10 Cârciumaru suggests that archaeologists left clues in their publications pertaining to the actual relevance of sites, not reporting the number of lithics, for example, when discoveries were not particularly significant (Cârciumaru 2011: personal communication), but whether this was in fact the case is difficult to verify.  15  workshops, hearths, shelters, and habitation structures (Păunescu 1993), certainly earned Păunescu international fame, but the rapid pace at which excavations proceeded also resulted in problems of stratigraphy and context (Păunescu 1993: 23; Cârciumaru 1995) that render at least some of these finds highly questionable. Moreover, the insights provided by specialists in other fields were not integrated into a holistic interpretation of the assemblages; for example, malacological studies conducted by Grossu indicated that the lower levels (I-III) were likely periodically flooded by the nearby river (Prut), as these levels contained aquatic and terrestrial gastropods that were mixed throughout this part of the sequence (Păunescu 1993: 21). Păunescu never seriously considered the effects of such flooding on the distribution of archaeological materials, which in some cases were interpreted as evidence of simple shelters. In the few cases when the effects of flowing water were considered, these were called upon to explain the lack of evidence for what Păunescu thought should be there. Thus, of some finds in the Gravettian 1b level, he notes: “the presence of a few small charcoal fragments around and on top of limestone blocks suggests the existence of a hearth, the remains of which were washed away at a later date by water” (Păunescu 1993: 24 – my translation). Many if not most of Păunescu’s finds are undoubtedly genuine, but the poor quality of his overall analysis and the fact that this site has been effectively destroyed, means that the most important Middle Palaeolithic site in Romania remains poorly understood and that many of the important finds will remain questionable in the foreseeable future. In any case, in addition to the adoption of Bordesian systematics by most researchers and the continuation of research projects initiated by Nicolăescu-Plopşor, the most significant research developments in the last two decades of the Communist regime pertained to advances in dating. In the 1970s, Marin Cârciumaru developed a chronological framework for the Romanian Palaeolithic based on palynological analyses (Cârciumaru 1980), while in the 1980s Kenneth Honea, a researcher from Northern Illinois University, obtained a series of mostly conventional 14C dates from a variety of Palaeolithic sites (e.g. Honea 1981; 1984), a work which he continued in the early 1990s. These dates still constitute a sizable proportion of currently available absolute dates, and Cârciumaru’s framework continues to be used today without any major modifications. It must be noted that Cârciumaru’s framework is based on analyses of sediments recovered mainly from Palaeolithic (cave) sites, and that it extends only as far back as MIS 5e (~130,000 to ~116,000 BP – see Shackleton et al. 2003), which until recently was understood to be the lower limit for Middle Palaeolithic habitation in Romania. The nomenclature employed by Cârciumaru retains local names for the various observed climatic oscillations, but in a more recent, easily accessible publication Cârciumaru et al. (2000: 95) correlate these with Pleistocene sub-divisions better known in other parts of Europe. 16  3.3.  Research after socialism  The collapse of the regime in 1989 resulted in a degree of unprecedented openness, which slowly translated into the adoption of new theoretical frameworks by Romanian archaeologists, as well the rise of international collaborative projects, particularly with French and German institutions. Theoretically, Palaeolithic research evolved in two directions: some scholars adopted North American processualist and post-processualist perspectives, while others opted to remain within the French school of thought, finding inspiration in the works of scholars such as André Leroi-Gourhan, Marcel Otte, and Eric Boëda. These theoretical shifts resulted in new questions being asked of the record, questions for which available data was ill-suited to provide an answer. Doboş (2010), for example, attempted to investigate site function in terms of Binford’s models of residential and logistic mobility, Anghelinu and Nita (2010) discussed the role of children and women in the Palaeolithic, and Elena-Cristina Niţu is applying a techno-functional approach to the study of Carpathian assemblages. Therefore, much of the Middle Palaeolithic research conducted over the course of the last decade has focused on the re-examination of relatively well-known assemblages in an attempt to answer behavioural questions. Unfortunately, most of these analyses have focused on the lithic components, and the faunal remains recovered at most of these sites continue to be ignored. 3.4.  Summary  This chapter has provided a brief and general introduction to the history of Romanian Middle Palaeolithic research. Although by no means comprehensive (for a comprehensive analysis the reader is directed to the work of Mircea Anghelinu [2004]), this historical review has established that: a) systematic research onto the Palaeolithic period of present day Romania began relatively late, b) the development of the discipline was curtailed to a large extent by political factors, c) that the methodologies and interpretive frameworks employed throughout the twentieth century favoured large rather than carefully executed excavations, as well as the preferential analysis (and perhaps recovery) of ‘typical’ pieces, d) that although multidisciplinary research was emphasized from the 1930s onward it largely failed to deliver on its early promise due to a clear separation of roles, an almost exclusive archaeological focus on the lithic components, and the overall goals of identifying or mapping cultures (when and where they existed) rather than explaining them, and e) that the research of Márton Roska should be re-evaluated. The following section is devoted to a critical assessment of the products of this research tradition.  17  4. The Romanian Middle Palaeolithic record as it is known today The aim of this chapter is to provide a qualitative assessment of the Romanian Middle Palaeolithic record in order to highlight problems that should be addressed by future research. I focus here specifically on: a) The distribution of known versus well-researched sites so as to identify underrepresented areas of high potential; b) The chronology of the Romanian Middle Palaeolithic so as to convey an urgent need for more reliable dates; c) The various industries that have been identified to date, to highlight, in particular, the need for more work on raw material selection and utilization at Carpathian sites; and d) The faunal record and Homo discoveries, to highlight the need for re-analyses of old collections. 4.1.  General overview  According to Păunescu’s comprehensive repertory of Palaeolithic and Mesolithic sites (Păunescu 1998, 1999a, 1999b, 2000, 2001), 240 locations with potential Middle Palaeolithic finds had been discovered in Romania by the end of the twentieth century. The lithic finds at 117 of these sites were assigned by Păunescu with confidence to the Middle Palaeolithic11, although only at 60 locations were these recovered in situ, through either systematic excavations or sondajes. Some of these sites had been identified by their original excavators as exclusively Upper Palaeolithic (e.g. Tincova – Selişte 2, Românesti – Dumbrăviţa), but were reinterpreted by Păunescu as containing Middle Palaeolithic components based on his study of a portion of the recovered material, and it is not clear whose interpretation is correct. In the case of some other sites various researchers generally agreed on the classification, but the number of recovered lithics was extremely small (<10). Finally, some of these sites were systematically excavated and yielded a large number of lithics attributable with a fair degree of confidence to the Middle Palaeolithic, but they have yet to be properly analyzed and published (e.g. several sites around Gurahonţ [Iosăşel], as well as Peştera Caprelor). In short, the number of systematically excavated sites generally recognized as having Middle Palaeolithic component(s), which have yielded more than a handful of lithics, and which have been analyzed and published in some detail was, by the end of the twentieth century, no greater than 20 (Figure 1). Some of these sites have been re-excavated since12, and several old lithic collections have been wholly or partially re-analyzed in an attempt to answer behavioural questions13, but few new Middle 11  The remaining 123 locations yielded what Păunescu considered probable Middle Palaeolithic lithics, which often consisted of a single, or in some cases a handful, of surface find(s). 12 These include Mitoc – Valea Izvorului, Boroşteni – Peştera Cioarei, Cladova, Zăbrani, La Adam, Moieciu – Peştera Coacăzii, and Ohaba Ponor – Bordul Mare.  18  Palaeolithic sites have been identified following the publication of Păunescu’s repertory14. The focus of recent investigations has been, for the most part, on sites that had already been relatively wellpublished (e.g. Ohaba Ponor), and with the exception of the open-air site of Zăbrani (Tuffreau et al. 2007) and the cave site of La Adam (Dobrescu et al. 2008), sites with known but poorly published Middle Palaeolithic components have yet to be re-investigated. While some significant new discoveries have been made, such as the evidence of bitumen use at Râşnov – Gura Cheii (Cârciumaru et al. 2012), the map of Middle Palaeolithic finds within present day Romania has not changed significantly over the last two decades. This is unfortunate not only because it perpetuates regional biases, as detailed below, but also because the re-excavation of already (relatively) well-known sites, especially without proper funding, endangers the small amounts of sediments that remain in these. The spatial distribution of the 120 locations (117 identified by Păunescu and three newly discovered ones) where lithics attributable with some confidence to the Middle Palaeolithic were found (Figure 1) reveals the presence of two dense clusters, one in western Romania and one in Dobrogea in the southeastern extremity of the country. The first cluster represents a series of open-air sites in the vicinity of the town of Gurahonţ, several of which were excavated by Roska at the beginning of the twentieth century and by Roşu in the 1970s. These excavations yielded thousands of lithics, which represent a considerable portion of the total number of lithics recovered in Romania to date (see Chapter 5), but Roska’s ability to properly identify stone tools has been questioned repeatedly (e.g. Nicolăescu-Plopşor 1938, Păunescu 2001), and Roşu’s work has been dismissed by Păunescu as deeply flawed (Păunescu 2001: 218). Thus, while the presence of Mousterian industries at these locations has not been questioned, a re-analysis of the old collections is critical, as only a small portion of the excavated lithics have been analysed following the original reports. More importantly, given the large number of lithics recovered from Roska’s relatively small excavations, and the fact that a large number of lithics can be seen eroding from hillsides at several locations today, systematic excavations in this region are likely to enhance our understanding of the Romanian Middle Palaeolithic record considerably. Also, they may play an important role in understanding the transition from the Middle to the Upper Palaeolithic, as it  13  The lithics from Bordul Mare – Ohaba Ponor have been re-analyzed by Elena-Cristina Niţu as part of her PhD work. Other collections that have been re-analyzed include those from several sites in Dobrogea, as well as two cave sites: Peştera Muierii and Peştera Cioarei (Doboş 2010; see also Doboş et al. 2010). These new analyses sought to answer questions pertaining to issues such as site function (Doboş 2010), the management of raw materials (Dobos 2010) and strategies of raw material exploitation (Niţu 2011: personal communication), and the interplay between production systems and environmental factors (Niţu 2011: personal communication). 14 Recently discovered sites with Middle Palaeolithic components include Slava Rusă, possibly Porţ, and Silvaşu de Jos, but they have yet to be published in any detail.  19  has been argued that the various sites around Gurahonţ (Iosăşel) contain important Szeletian assemblages (Horvath 2009). As it stands, however, this region is poorly documented. The second cluster, in south-eastern Romania, represents 33 open-air locations, including the newly discovered site at Slava Rusă, and two cave sites. The small number of cave sites in a region characterized by abundant karst formations (see Figure 1) can be explained in terms of research bias, as can the density and tight clustering of open-air locations. Păunescu (1999b: 28) notes in this regard that of the 85 caves and rock shelters that had been identified by the 1990s, only 11 had been excavated15, despite the fact that many others appeared to be very promising. Moreover, he states that the disproportionate number of sites discovered in the county of Constanţa, relative to the number of sites discovered in the neighbouring county of Tulcea at the north, reflects the unequal intensity of research across the region (Păunescu 1999b: 28). Unfortunately, the majority (n=27, or 82%) of open-air locations identified in Dobrogea represent surface finds, where no test pits or systematic excavations were conducted. While 11 of these locations yielded ten or less lithics, others yielded a relatively large number of finds, concentrated over a relatively small area (e.g. 201 lithics at Peştera – Dealul Peşterica, and 215 lithics at Saligny – La Ghiol; see Păunescu 1999b), and it is likely that excavations at these would have revealed in-situ deposits. Other sites in the region yielded a large number of lithics (e.g. a total of 1207 at Cuza Vodă E; see Păunescu 1999b: 112-119), some of which were recovered through very limited excavations16, but no information on the actual number of lithics found in situ is given, and it is likely that the vast majority of pieces were recovered from a secondary context. The unsystematic nature of research in this region is difficult to explain, but a lack of proper funding and the inability to conduct excavations at some locations are possible causes. It is in any case clear that this region is, despite the intensity of research carried out over the course of several decades, also poorly understood. The remaining Middle Palaeolithic finds cluster in smaller groups dispersed throughout most of the country. The majority of small clusters in the southern Carpathians have yielded at least one wellpublished site, as did the clusters located around the northern periphery of the country, but in the Carpathian basin, where a series of open-air and cave sites were identified, and in the Romanian plain at the south, the purportedly Middle Palaeolithic finds are either questionable or poorly published. In the northern Carpathians and much of the low-lying areas in central/eastern Romania however, reports of Middle Palaeolithic finds are conspicuously absent. In the case of the first region this may be due to research bias, as less intense investigations were carried out there than in other parts of the country, 15  Six of these yielded Palaeolithic implements, while five yielded only faunal remains. These consisted of nothing more than the cleaning of natural profiles, such as the sides of a ravine that cut through the site of Castelu). 16  20  but in the case of the second region thick loess deposits, which were rapidly deposited during the late Pleistocene, likely cover most traces of Middle Palaeolithic habitation. To summarize, given the distribution of adequately researched and well-published sites relative to the distribution of all known locations with confirmed finds, it is evident that the Middle Palaeolithic record of Romania is poorly documented despite a history of research that spans over a century. The view that emerges from these sites is therefore very partial, and patterns evidenced in the available data should be treated with caution. In particular, differences in the spatial distributions of open-air and cave sites should be carefully considered, as they affect the interpretation of the record and likely reflect research bias rather than underlying archaeological phenomena (e.g. in Dobrogea research focused on open-air sites despite the fact that promising cave sites were known). Presently, all well-published cave sites (with the exception of Cheia – La Izvor) are located at various elevations in the southern Carpathians, in zones that relatively poor in high-quality flint and other fine-grained raw materials, while open-air sites are located in low-lying areas around the periphery of the country.  21  Figure 1: Composite map of Romania showing a kernel density map of the 120 locations where lithics attributable with some confidence to the Middle Palaeolithic have been found, as well as the location of the 21 systematically excavated and reasonably well-published sites discussed in detail here. The kernel density map is overlaid on a DEM of the country (GTOPO30, tile E020N90), and is clipped with the country border vector file provided by GADM (www.gadm.org); Karst formations (after Goran 1981) are also shown here in black. The listed sites are: 1) Remetea Somos I, 2) Remetea Somos 2; 3) Buşag; 4) Zăbrani; 5) Cladova; 6) Nandru – Peştera Curată; 7) Nandru – Peştera Spurcată; 8) Ohaba Ponor – Bordul Mare; 9) Gornea; 10) Livadiţa; 11) Băile Herculane – Peştera Hoţilor; 12) Boroşteni – Peştera Cioarei; 13) Baia de Fier – Peştera Muierii; 14) Peştera – Peştera Mare; 15) Peştera – Peştera Valea Coacăzii; 16) Râşnov – Gura Cheii; 17) Cheia – Peştera Gura Cheii; 18) Mamaia-Sat; 19) Mitoc – Valea Izvorului; 20) Ripiceni-Izvor; 21) Boineşti.  4.2.  Chronology  As is the case in many other parts of the world, the chronology of the Romanian Middle Palaeolithic is poorly understood (Horvath 2009: 139); only some 60 absolute dates, many obtained by conventional radiocarbon methods and representing only a dozen or so sites, are available for the entire record17.  17  Cârciumaru (2007: 105-132) provides a list of absolute dates that were available by 2007; very few additional absolute dates have been published since (e.g. Tuffreau et al. 2009; Doboş et al. 2010). The Middle Palaeolithic sites for which absolute dates are available include Ohaba-Ponor – Bordul Mare, Boroşteni – Peştera Cioarei, Nandru – Peştera Curată, Nandru – Peştera Spurcată, Râşnov – Gura Cheii, Peştera – Peştera Mare, Peştera – Peştera Valea Coacăzii, Ripiceni – Izvor, Peştera Muierii, Mitoc – Valea Izvorului, and Cheia – La Izvor, and Vârtop.  22  Given the scarcity of reliable dates18 and the various problems with the classification of lithic assemblages (discussed below), no attempt is made here to analyze the development of the various Middle Palaeolithic industries through time; such attempts have been, in any case, the focus of previous studies (e.g. Cârciumaru et al. 2007; Cârciumaru 1999). Rather, I focus here on some salient issues of chronology relevant to understanding the Romanian Middle Palaeolithic record as a whole. The available absolute dates are, in general, very young relative to dates obtained elsewhere in Europe. When considered on a global scale, the chronological boundaries of the Middle Palaeolithic can roughly be placed at between 300,000 to 200,000 years ago (e.g. Hopkinson 2007; Porat et al. 2002) and 40,000 to 35,000 14C BP (e.g. Roebroeks 2008; Pinhasi et al. 2011; Maroto et al. 2012). In Romania, however, only six of the dated sites, Mitoc – Valea Izvorului, Ripiceni-Izvor, Boroşteni – Peştera Cioarei, Nandru – Peştera Curată, Ohaba Ponor – Bordul Mare, and Peştera Muierii yielded uncalibrated dates older than 40,000 14C BP (see Cârciumaru et al. 2007: 105-132). The site of Vârtop also yielded a human footprint dated by U-Th to >62,000 BP, but no associated archaeological materials have been discovered there (Onac et al. 2005). The ambitious chronological framework developed by Cârciumaru (1980), based on palynological analyses but in general agreement with the various micro-vertebrate and malacological fauna studies conducted at several sites, provides some context for the development of the Romanian Middle Palaeolithic, and continues to be used today (e.g. Cârciumaru et al. 2007; Cârciumaru et al. 2012). Based on this chronological framework, and to a lesser extent on the available 14C dates, the earliest Middle Palaeolithic assemblages in Romania were, until recently, understood to date to no earlier than the Eemian interglacial (128,000 to 117,000-110,000 BP – see Forsström 2001 and Shackleton et al. 2003), or MIS 5e (Cârciumaru et al. 2007; Cârciumaru 1999).Newly obtained absolute dates, however, have shown that this is most likely not the case, as OSL (IRSL) dating of the site of Mitoc – Valea Izvorului (Tuffreau et al. 2009) suggests that the oldest lithic material (Mousterian) can be dated to MIS 6, between 177,000 and ~130,000 years). According to Doboş (2010: 39), the dated level was previously thought to be contemporary with two mid-sequence levels (IV and V) from the nearby site of RipiceniIzvor, dated through conventional radiocarbon techniques to approximately 44,000 BP. It is therefore possible that many undated sites, assumed on the basis of stratigraphic correlation to be relatively young, are in fact much older.  18  Many of the available radiocarbon dates are highly problematic; in fact, Doboş et al. (2010: 13) consider only 33 C dates, from 6 sites, to be minimally reliable.  14  23  The youngest dates obtained so far are likely just as problematic, since they overlap entirely with the oldest dates obtained for Upper Palaeolithic assemblages, although no sites have been identified with interstratified Middle and Upper Palaeolithic levels (Horvath 2009: 141). Thus, the earliest reliable dates for Upper Palaeolithic assemblages come from the site of Mitoc-Malu Galben, with at least three samples from the Aurignacian level yielding ages greater than 30,000 BP (Cârciumaru et al. 2007: 124), while the Mousterian levels from Ohaba Ponor – Bordul Mare, Nandru – Peştera Spurcată, and Râşnov – Gura Cheii are considered to have extended beyond 30,000 BP (Cârciumaru et al. 2008; Horvath 2009: 141; Cârciumaru et al. 2012; Cârciumaru et al. 2007). If proven reliable, these dates raise a series of questions ranging from the accuracy of the identification of lithic assemblages as Mousterian - it has certainly been suggested that the site of Nandru – Spurcată may in fact be Szeletian (e.g. Horvath 2009) – to the association of Mousterian industries with Neanderthals, as there are no reliable dates for Neanderthal fossils postdating ~35,000 14C BP (Joris and Street 2008; Pinhasi et al. 2011). Indeed, the sites of Ohaba Ponor and Nandru – Peştera Spurcată are located in close proximity to the site of Cioclovina, where early anatomically modern human (AMH) remains have been dated to ca. 28,500 BP (Soficaru et al. 2007), and roughly within the same geographic region as the site of Peştera cu Oase, where the earliest AMH in the country, and one of the earliest in Europe, was dated to ca. 35,000 BP (Trinkaus et al. 2003). Unfortunately, only one of these sites, namely Ohaba Ponor – Bordul Mare, retains undisturbed sediments that may be useful for future dating. 4.3.  Fauna and Homo remains  Faunal material has been recovered from Middle Palaeolithic levels at 13 of the 21 sites identified above as relatively well-published and systematically excavated, including the open-air sites of Ripiceni-Izvor and Zăbrani. Generally, the faunal remains include both herbivore and carnivore species, but it is unclear to what extent the accumulation of faunal material in the various assemblages reflects human activities. For the vast majority of sites the presence or absence of species is the only reliable faunal information that is available in the literature, as Romanian Palaeolithic archaeologists were, for the most part, interested only in the lithic components. In accordance with the strict separation of roles that characterized multidisciplinary research, the study of faunal remains was left to palaeontologists who were interested in paleontological rather than archaeological questions, and the data produced by these specialists were generally used to reconstruct environmental conditions rather than behaviours. In some cases basic measures of abundance such as MNI or NISP can be extracted from the paleontological literature, but in general the available publications are of limited use in answering archaeological questions. 24  This lack of archaeological interest in fauna had two serious and unfortunate consequences. First, in many cases, including that of Ripiceni-Izvor, faunal remains have simply not been subject of any thorough analysis, paleontological or otherwise. Secondly, at many sites the association of faunal remains with archaeological levels is problematic. At Nandru – Peştera Curată, for example, the stratigraphy used by Păunescu in his analysis of the lithic components is fundamentally different from that used by other archaeologists and paleontologists who studied and reported on the faunal material (the exception being the microfaunal analyses conducted by Alexandra-Cristina Păunescu). At Bordul Mare a similar situation exists, as the stratigraphy was subject to revision over the years but the faunal data was not properly re-integrated into the new stratigraphic sequences, which resulted in a series of inconsistencies (see Cârciumaru and Niţu 2008). In short, the available faunal data is of extremely poor quality, and a study of the relationship between the faunal and lithic components is impossible to conduct at this stage, except in a few isolated cases. Given the low number of sites which have yielded faunal remains and the fact that, as discussed in the following sections, few of these appear to have been intensely inhabited during the Middle Palaeolithic (notable exceptions are Ohaba Ponor – Bordul Mare and particularly Ripiceni – Izvor), it is not surprising that Neanderthal remains are uncommon in the Romanian record. In fact, only four phalanges have been identified to date: at Ohaba Ponor Istvan Gaál (Gaál 1928, 1943) reported the discovery of three phalanges in the material recovered by Roska in 1923 and 1924, while at Livadiţa Elena Terzea reported the discovery of a Neanderthal phalanx associated with Mousterian lithics in 1979 (Terzea 1979: 114; see also Cârciumaru 1999: 70). These human remains have not been subjected to any modern analyses, and neither have the faunal collections from which they were recovered. Considering the precarious state of faunal research, it is entirely possible that small and/or fragmentary Neanderthal remains held in museum collections still await identification. 4.4.  Lithic Industries  The classification of assemblages following the widespread adoption of Bordesian systematics resulted in the identification of several distinct Mousterian facies, but also in the disregard of some previously observed variability and commonalities between and within assemblages now classified as distinct types. The different Mousterian facies are discussed below. 4.4.1. Typical Mousterian By the end of the twentieth century, assemblages classified as Typical Mousterian of Levallois debitage had been identified in various parts of the country, namely in the lower levels (I-III) at Ripiceni-Izvor in 25  the north-east, at Cheia-La Izvor and Castelu in the south-east, at Gornea in the south-west, and at Boineşti and Remetea Somoş in the north-west (Păunescu 1989; Păunescu 1999a, Păunescu 1999b; Păunescu 2001). Typical Mousterian of non-Levallois debitage, on the other hand, had been identified in the south-east, at the site of Mamaia-Sat and Cuza-Vodă19, and has more recently been identified also at the open-air site of Peştera, and the cave sites of Peştera Cioarei, and Peştera Muierii (Doboş 2010). However, the classification of industries with predominantly coarse-grained raw materials such as quartzite (as encountered at Peştera Cioarei and Peştera Muierii) according to Bordesian systematics is problematic, as discussed below. Be that as it may, by virtue of the number and geographical distribution of assemblages classified as Typical Mousterian, it can be said that, after the Eastern Charentian or Cave Mousterian (discussed below), this facies is the most commonly represented one in the country. Moreover, it also appears to be one of the most long-lived facies, as the oldest levels from Ripiceni-Izvor are thought to date to ca. 60-65ky BP (Păunescu 1989: 135), and are likely much older (Păunescu 1993: 30; Doboş 2010: 45), while the Typical Mousterian levels from Remetea Somoş and Boineşti are considered to be very late due to the presence of Upper Palaeolithic types (Bitiri 1970), although their presence in these levels may well be due to intermixing of sediments caused by the strong erosional processes observed in the area. Regardless, technological variability in assemblages classified as Typical Mousterian, as well as links between assemblages thus classified in the north-western part of the country and Eastern Charentian assemblages in the southern Carpathians, has been recognized for a long time (e.g. Bitiri 1972). 4.4.2. Mousterian of Acheulean Tradition (MTA) Assemblages classified as MTA have been identified at Mitoc-Valea Izvorului and in levels IV-V at Ripiceni-Izvor in the north-east; they had also been recognized at other sites (e.g. Peştera Muierii – see Doboş et al. 2010), but those assemblages have been re-evaluated since as belonging to a different facies. The classification of the Middle Palaeolithic level from Mitoc as MTA is controversial, however, as the stratigraphy of Mitoc was complicated (Păunescu 1999a: 127; Bitiri- Ciortescu 1987: 208) and poorly understood, and it has been shown that the recovered lithics represent a mix of Middle and Upper Palaeolithic tools that in fact originated from different levels (Tuffreau et al. 2009). The MTA levels from Ripiceni-Izvor, on the other hand, show more affinities with the Micoquian, containing Pradnik type 19  It must be noted, however, that at the site of Castelu and Cuza-Vodă, many (perhaps the vast majority) of the lithics were recovered from a secondary context following industrial works in the area, and therefore the degree to which they are inter-associated is not immediately clear. The actual number of lithics recovered in-situ is not specified. Limited excavations, which in fact consisted of nothing more than the cleaning of natural profiles (e.g. the sides of a ravine that cut through the site of Castelu), were conducted at both locations, but these appear to have been directed mostly at identifying the stratigraphic succession of the deposits and covered a small area.  26  bifacials and a generally high incidence of bifacial leaf points (Doboş 2010; Mertens 1996; Cârciumaru et al. 2007: 136-139). These levels from Ripiceni-Izvor are, in any case, quite unique: first, they are by far the richest assemblages identified in the country; secondly these are the only open-air levels that appear to have been deposited in the context of a predominantly cold climate and, lastly, these levels are characterized by the presence of various structures that have been interpreted as representing three distinct types of shelters (Păunescu 1993). 4.4.3. Denticulate Mousterian The Denticulate Mousterian facies was identified in the upper Mousterian level at Ripiceni-Izvor, as well as at the sites of Saligny, Peninsula, and Ovidiu-Nazarcea in Dobrogea, south-eastern Romania (Păunescu 1999a; Păunescu 1999b; Doboş 2010). Păunescu originally classified the Mousterian from Mamaia-Sat as Denticulate also, but by 1999 he argued that it represented, in fact, a Typical Mousterian of non-Levallois debitage (Păunescu 1999b), an interpretation confirmed by a recent re-analysis of the lithic material (Doboş 2010). In regards to the other sites classified as Denticulate Mousterian, it must be noted that at Saligny and Ovidiu-Nazarcea no artifacts were recovered in situ, and no test pits or excavations were conducted at all; at the site of Peninsula, on the other hand, limited excavations produced 19 lithics found in situ, but these were analyzed together with out of context artifacts found on the surface (Păunescu 1999b). 4.4.4. Eastern Charentian or “Quartzitic Mousterian” Assemblages characterized by the extensive use of coarse-grained raw materials are mostly found in cave sites and are all situated in a relatively well defined cluster which spans the whole length of the southern Carpathians, but extends to the north of these in the western part of the country to include the open-air sites of Zăbrani and Cladova. Partially because of this association with a particular geographic area and with cave sites, and partly due to technological and typological similarities among the various coarse raw material assemblages on the one hand, and notable differences between these and assemblages characterized by the predominant use of fine-grained materials on the other, assemblages characterized by the extensive use of coarse-grained materials have traditionally been lumped together and treated as a distinct industry. Known originally as “Cave Mousterian” or “Quartzitic Mousterian”, after the adoption of the Bordesian system and the publication of Gábori-Csánk et al.’s (1968) seminal work on the quartzitic industry from the Hungarian site of Erd, it was generally identified as a regional variant of the Eastern Charentian of Pontinian technique (Păunescu 1989; Cârciumaru 1999: 111) and was viewed, within the dominant culture-historical framework, as representing a distinct and long-lived culture (Mogoşanu 1978: 130-136). Sites with assemblages assigned to this facies include 27  Peştera Hoţilor, Ohaba Ponor – Bordul Mare, Nandru – Peştera Curată, Nandru – Peştera Spurcată, Peştera Muierii, Boroşteni – Peştera Cioarei, Râşnov – Gura Cheii, and Peştera – Mare (Dobos 2010: 46). According to Păunescu, this variant of the Eastern Charentian is characterized technologically by the common use of the Pontinian split-pebble technique and the low incidence of Levallois technology and, typologically, by the high incidence of scrapers, denticulates, and notches, and by the general absence of bifaces (Păunescu 1989). There are serious problems with this characterization, however. First, the fact that the Levallois technique is largely absent is expected given the flaking characteristics of non-isotropic and/or coarsegrained materials (e.g. Rolland and Dibble 1990: 484; but see Eren et al. 2011), and as such it cannot be considered a diagnostic behavioural indicator (e.g. a lack of interest in or knowledge of the Levallois technique), except in the general sense that the hominins responsible for these industries were aware that different raw materials require different processing techniques. In fact, the majority assemblages with non-negligible quantities (>20%) of coarse-grained raw materials (n=21, or 61.7% of the 34 such levels included in the database analyzed in Chapter 5) do contain technologically defined Levallois implements, albeit in small quantities. Secondly, it does not make much sense to compare assemblages characterized by the use of different raw materials in terms of how common certain tools are relative to others, if one type of assemblage (those with coarse-grained raw materials) is expected to have a zero or very low incidence of certain tool types (e.g. technologically defined Levallois flakes) due to the constraints imposed by the intrinsic properties of the raw materials. Moreover, differences in knapping techniques are expected to result in different flake forms, and this in turn is expected to affect the results of a classificatory system that is largely based on tool morphology. Finally, another serious problem with the grouping of assemblages characterized by the considerable use of coarse-grained raw materials into a single facies is the fact that many of these assemblages contain, in fact, a sizable proportion of fine-grained materials. Romanian archaeologists were certainly aware of these issues since at least the 1930s (see NicolăescuPlopşor 1938), and Bitiri, for example, explicitly argued that: The use of quartzite […] during the Mousterian at Baia de Fier [Peştera Muierii], Nandru, and Ohaba-Ponor is not indicative of the cultural affiliations of the inhabitants of these caves, as suggested by Gábori […], but something specific to all human groups of the Middle Palaeolithic, which did not travel for the sole purpose of collecting raw materials of a superior quality; that is a behaviour that manifested at a large scale only during the Upper Palaeolithic (Bitiri 1972: 38 my translation).  28  However, they were largely unable to address the problems posed by these assemblages within the dominant culture-historical paradigm in which they laboured, and generally either tried to argue for cultural continuity between Carpathian assemblages and those present in other parts of the country, as was the case with Bitiri, or for a common cultural tradition evidenced at these sites (e.g. a local variant of the Eastern Charentian). Cârciumaru (1999), on the other hand, has argued that these assemblages should be treated as two distinct, chronologically separated groups: an early one which corresponds to the Charentian of Pontinian technique, and a later one, which he simply labeled “Carpathian facies,” which he understood to be quite late and to represent a transitional phase (of Szeletian influence) to the Upper Palaeolithic, showing commonalities with industries in other parts of the country (Cârciumaru 1999: 112). To the best of my knowledge, the effects of raw material properties on assemblage variability (e.g. the incidence of shatter, the sizes and morphology of flakes, number of flakes obtainable per core, and so on) were never explored in a scientific manner, although Cârciumaru et al. (2000) did attempt to address issues of raw material variability, criteria for nodule selection, and technological differences in the working of different materials at Peştera Cioarei. 4.5.  Summary  This chapter has established that many aspects of the Romanian Middle Palaeolithic are poorly understood. First and foremost, the number of well-researched sites relative to the number of locations thought to have yielded reliable traces of Middle Palaeolithic habitation is very low, and it is clear that some regions are underrepresented. Furthermore, the sample constituted by the well-researched sites is likely biased, since research in some regions has mostly involved the investigation of cave sites, while in other regions open-air sites were preferentially investigated. Finally, some regions have failed to yield any reliable trace of Middle Palaeolithic habitation, and it is at present unclear to what degree this is a reflection of research bias or an underlying archaeological phenomenon. Secondly, the chronology of the record is highly problematic, as insufficient absolute dates are available, and many of these appear to be unreliable. Third, the available faunal data is of extremely poor quality and does not allow for inferences regarding subsistence behaviours. Fourth, the uncritical adoption of Bordesian typology has resulted in: a) the classification of all Middle Palaeolithic assemblages as Mousterian, b) the disregard of previously observed variability and the classification of assemblages into facies defined elsewhere in Europe, and c) the disregard for the effects of raw material variability on assemblage composition. In the chapter that follows, I provide a brief quantitative analysis of data from the well-researched sites.  29  5. A quantitative assessment of the Romanian Middle Palaeolithic record To the best of my knowledge, no exploratory statistical analyses of the Romanian Middle Palaeolithic record have been conducted to date, despite the fact that exploratory analysis can reveal problems with, as well as previously unrecognized patterns in the available data. The goal of this chapter is to present and explore the data compiled from the 21 sites identified above as relatively well-researched and well-published. The scope of this thesis prevents an exhaustive treatment, but the full database used here has been made available in electronic format so as to facilitate further exploration (see Appendix A). The analyses presented below do not aim to provide conclusive proof pertaining to the interpretation of any of the identified patterns; rather, they aim to discriminate between those patterns which are likely of archaeological significance and those which are not, and to suggest potential explanations for the former that should be explored in depth in future studies. The chosen level of significance for all analyses is .05. Archaeological levels, as these are identified in the Romanian literature, constitute the basic analytical units investigated here20. These are time-averaged aggregates deposited in most cases over an unknown period of time, and may or may not reflect periods of relatively continuous habitation. While time averaging is a serious problem, particularly if the averaged time span is not known (Vaquero 2008; Shott 2008), it is simply a reality of Middle and Lower Palaeolithic research all over the world, and in the case of the Romanian data at least, there is no viable alternative to grouping artifacts. Archaeological levels are investigated in terms of their most thoroughly studied components, namely the lithic materials, and three main grouping variables, namely the type of site where the levels were excavated (cave or open-air, as no rock-shelter sites are included in the database), the average climatic conditions that prevailed during their deposition, and the types of raw materials used in lithic manufacture. These three variables, defined in Appendix A, were chosen because data is available for all or most of the levels and they are expected to be relevant in understanding assemblage variability. Open-air and cave sites, for instance, have been noted to show different levels of variability in assemblage composition (e.g. Rolland and Dibble 1990: 488), and some Middle Palaeolithic industries “are tied to specific site types, independent of climate” (Patou-Matis 2000: 386). This is the case of the so-called Quartzitic or Eastern-Charentian Mousterian of Pontinian Technique, which in Romania is 20  One exception is the Mousterian assemblage from Peştera Muierii. Multiple discrete archaeological levels are known to have existed at that site, but all Middle Palaeolithic remains were analyzed as a single group (see Doboş et al. 2010 and Păunescu 2000: 310-324), and as such are treated here as belonging to a single “level” for lack of a better alternative.  30  associated almost exclusively with cave sites. Beyond this, however, cave and open-air sites often require different research strategies, as spatial layouts, as well as depositional and taphonomic regimes differ; this makes it difficult to rule out research methodologies as a potential cause for some of the observed variability, particularly since most Romanian sites were excavated several decades ago with modestly trained workers. Palaeoclimate, on the other hand, is important to consider because it constitutes one of the few ways in which the relationship between the environment and the occupational episodes recorded at the various Romanian Middle Palaeolithic sites may be investigated, as the quality of the faunal data precluded their inclusion in the database. Climate is expected to have had a rather dramatic effect on European Middle Palaeolithic hominins, and changes in climate may have been reflected in different settlement patterns and mobility strategies (e.g. Rolland and Dibble 1990: 488), as has in fact been suggested for sites from the Romanian Carpathians by Riel-Salvatore et al. (2008), or in regional extinctions (see Hublin and Roebroeks 2009). The available palaeoclimatic data allowed for many of the assemblages to be classified, according to the average climatic conditions that prevailed during their deposition, in one of three classes, one indicating very harsh, cold and dry climates, a second one indicating milder, cold and wet climatic regimes, and a third one indicating temperate climates. Given the comparatively low resolution of the data and the small sample size for each palaeoclimatic class, however, the role of climate is explored here only in terms of the distribution of assemblages and sites. Finally, raw materials are important to consider because Romanian assemblages are characterized by a considerable degree of variability in the types and quality of rocks that were exploited. Raw material quality is expected to influence lithic procurement, production, maintenance and use (e.g. Dibble 1991; Webb and Domanski 2008; Andrefsky 1994; Wilson 2007; Braun et al. 2009), and may also limit the ability of researchers to identify and classify tools. Driscoll (2010), for example, has shown that even among researchers with ample experience in the study of “quartz” assemblages, the ability to identify tools and even cores is limited, which is noteworthy because many Romanian Middle Palaeolithic assemblages reflect a rather extensive use of materials labeled in the literature as “quartz”. Although only rock types (e.g. chert, quartzite, sandstone) are listed consistently in Romanian publications, it is possible to roughly estimate one raw material characteristic that can serve as a proxy to raw material quality, namely granulometry. Consequently, raw materials are grouped here in two categories, “fine” and “coarse”, with “coarse” raw materials referring to all rock types with grain sizes greater than that of chert (a crypto- or micro-crystalline material), or about 20 microns (Hesse 1989: 257). On the basis of this distinction assemblages are divided into three raw material groups, namely 31  “fine” if the incidence of fine-grained materials is equal to or greater than 80%, “coarse” if the incidence of coarse-grained raw materials is equal to or greater than 80%, and “mixed” if the composition falls somewhere in between (see Appendix A). The 80% cut-off point was chosen because it allows for a reasonable sample size for all raw material groups, while ensuring that levels where one type of raw material is clearly dominant are still visible. What are considered here as “mixed” and “coarse” raw material levels correspond to assemblages traditionally grouped together under labels such as “Quartzitic Mousterian” (but see Cârciumaru 1999). While the distinctions made here are essentially arbitrary, they are analytically useful because they serve to illustrate certain important patterns that are difficult to observe otherwise with the available data. 5.1.  General overview  Of the 21 sites included in the database, 11 are located in caves and 10 in open-air locations21. Although cave sites are more commonly represented in the database and have yielded a larger number of distinct archaeological levels than open-air sites (n=32 versus n=18; see Figure 2)22, the total number of lithics recovered from open-air sites (n=62727) is far greater than the number of lithics recovered from cave sites (n=12183)23. This is mainly due to two extremely rich levels from Ripiceni-Izvor (IV and V), which have yielded 35890 and 16064 lithics respectively, or ca. 83% of the lithics recovered from open-air sites and ca. 69% of all the lithics included in the database. Open-air sites have also yielded a larger number of formal tools24 (n=7803) than cave sites (n=1330), although proportionally the difference is negligible: approximately 12% of lithics recovered at open-air sites are formal tools, while the percentage of formal tools recovered from cave sites is approximately 11%. Unfortunately, a reasonable estimate of the volume of excavated sediment per level, and therefore of the density of lithic finds, is either difficult or impossible to determine for many of the sites. Furthermore, some excavations covered such large areas (e.g. nearly 4000m2 in the case of RipiceniIzvor) that without detailed information on the spatial patterning of finds, overall lithic densities are a meaningless metric. Without such basic contextual information, country-wide comparisons are difficult  21  Contrary to Patou-Matis’ claim to the contrary (Patou-Matis 2000: 380), Romanian open-air sites are not rare, at least not in relation to the number of Romanian cave sites. 22 Note that at Peştera Muierii two Middle Palaeolithic levels were identified in two different galleries; however, the actual number of levels is unknown (Doboş et al. 2010: 35). Since the Middle Palaeolithic artifacts have been analyzed as a single group, they are considered here as belonging to a single “Mousterian” level. 23 These counts include lithics from Mamaia – Sat and Ohaba Ponor – Bordul Mare which could not be associated to one of the levels discussed here and are therefore not included in the analyses that follow. Note also that total lithic counts include not only cores, tools, and debitage, but also manufacturing by-products (e.g. shatter). 24 These are retouched pieces and technologically defined tools listed in one of the numbered Bordesian types (163) or in the OtherTools database variable (see Appendix A).  32  to make. Nevertheless, two important observations can be made at this point: the number of lithics recovered from individual levels is, overall, quite low (median = 115, IQR = 321), and so is the total number of lithics included in the database (n=74910). In fact, the total number of lithics recovered from all of the sites under study combined is lower than the number of lithics recovered from a single French Middle Palaeolithic rock shelter site, Pech de l’Azé IV, where approximately 92000 lithics had been recovered by the end of the twentieth century (McPherron and Dibble 1999). Even acknowledging the fact that not all excavated lithics were collected, and that not all collected lithics are included in the database25, the low number of lithics suggests that the territory of present day Romania was likely far less intensely occupied during the Middle Palaeolithic than other regions of Europe (e.g. France).  Figure 2: Distribution of lithics and levels across the 21 systematically excavated and reasonably well-published Middle Palaeolithic sites discussed here. Hashed bars indicate cave sites, while grey bars indicate open-air sites. The numbers presented above the bars indicate the total number of lithics recovered from a given site; in the case of Ohaba Ponor – Bordul Mare and Mamaia – Sat, Middle Palaeolithic artifacts that could not be assigned to one of the levels discussed here are also included in this count. Note that at Peştera Muierii at least two Middle Palaeolithic levels were identified, and more likely existed (see Doboş et al. 2010) but, due to uncertainties pertaining to the stratigraphy of the site, Middle Palaeolithic artifacts have been analyzed as a single group, and therefore included here in a single “level”.  Before proceeding further, it must be noted that a series of statistically significant associations involving site type and variables such as the site altitude, climatic regimes, and raw material components complicate the interpretation of the data, making it extremely difficult to isolate the effects of any one variable on assemblage variability. First, it has been noted in the preceding chapter that the well-  25  The number of lithics not included in the database, either because they have not been published in sufficient detail (e.g. at sites around Gurahonţ) or because it is unclear whether they represent surface finds or in-situ discoveries (e.g. at Cuza-Vodă E) is unlikely to be greater than 20,000, and is probably less than 10,000.  33  published cave sites are located mostly in the Carpathians, while open-air sites are located in low-lying areas around the periphery of the country. This suggests differences with respect to the absolute elevations at which the two site types are found, which are confirmed by a Mann-Whitney nonparametric test (U (21) = 20.5, p = .015). While this association is most likely the result of research bias (most research conducted in low-lying areas focused on open-air sites), it becomes difficult to distinguish between the potential effects of site type and local topography or environment (both of which vary with altitude) on assemblage variability. Second, a statistically significant difference (exact probability [Fisher’s exact test - FET] under the null hypothesis < .001 [n=47]) is evidenced between levels excavated in caves and open-air locations with respect to the raw materials from which artifacts were made (Table 1), with open-air sites being associated with predominantly fine-grained raw materials, and cave sites being associated with a lack thereof. This association is also likely the result of research bias, since few caves were investigated in regions with plentiful sources of fine-grained raw materials such as Dobrogea. As with site altitude, it becomes difficult to ascertain to what extent the type of site or the raw materials used are related to the variability observed in assemblage composition. Finally, a statistically significant difference exists between levels excavated in caves and open-air sites with respect to the climatic conditions in which these were deposited (exact probability [FET] under the null hypothesis = .017 [n=29]) with open-air sites being associated with milder climates and cave sites with generally harsher climatic regimes. Unlike the other two associations, this likely reflects a genuine archaeological phenomenon, since Cârciumaru, whose works were used to extract palaeoclimatic data, took into account climatic differences caused by local topography and variations in altitude. Still, in this case too it is difficult to distinguish whether assemblage variability is related to the type of site (and any biases introduced by differences in excavation techniques) or to the potential effects of climate, if any. It should also be noted that, as expected given the associations between site types and climate on the one hand, and site types and raw materials on the other, a marked difference exists between levels characterized by the predominant use of different raw material types with respect to the climatic conditions in which these were deposited, further complicating interpretation. Indeed, coarse-grained raw materials are associated mostly with levels deposited during the harshest climatic regimes, while fine-grained raw materials are associated mostly with levels deposited during temperate climates (Table 2). Although this association is not statistically significant (exact probability [FET] under the null hypothesis = .065 [n=28]), it should be kept in mind given the small sample sizes and the likelihood of a type 2 error. 34  Table 1: Distribution of different raw material types across levels excavated at open-air and cave sites.  Predominant raw material type Fine Site type  Open-air  Mixed 12  3  1  16  1  9  21  31  13  12  22  47  Cave  Total levels  Total levels  Coarse  Table 2: Distribution of predominant raw material types across levels deposited in different climatic contexts.  Predominant raw material type Fine  Coarse  Total levels  Cold – dry  0  1  8  9  Cold – wet  3  1  3  7  Mild  6  2  4  12  Total levels  9  4  15  28  Climate  5.2.  Mixed  Climate and the distribution of sites  Relevant palaeoclimatic data is available for 11 sites and 29 levels, or 52.4% of the 21 sites and 58% of the 50 levels under analysis (Table 3)26. These data indicate that while caves were inhabited in all climatic contexts, more traces of cave habitation are evidenced during harsher climates than during milder ones and, more importantly, that open-air sites are associated almost exclusively with mild climates. This association between site types and climatic conditions, already shown to be statistically significant, suggests that during harsh climatic regimes Middle Palaeolithic populations were restricted to areas that offered natural shelters such as caves, and that therefore much of the territory of present day Romania was sparsely inhabited, or perhaps even uninhabited, during such cold periods. Moreover, the low number of lithics recovered from cave assemblages (n=14) known to have been deposited during harsh or relatively harsh climates (1960 lithics, of which 389 are classified as formal tools) suggests that even in these areas the population densities were low27. Levels IV and V from Ripiceni – Izvor, the only open-air levels that appear to have been deposited in the context of a relatively harsh climatic regime, constitute an important exception. These levels, which  26  Some information is available in 16 other cases, but it is either of too imprecise or it could not be reliably associated with individual archaeological levels, as in the case of Nandru – Peştera Curată. 27 It should be noted, however, that many if not most of the lithics from Ohaba Ponor – Bordul Mare which could not be associated with the archaeological levels discussed here were likely recovered from levels deposited during harsh climatic conditions. Even considering these, however, the number of lithics remains low (n = 6179, assuming all unaccounted for lithics were recovered from such levels).  35  yielded the richest collections of lithics excavated in a Middle Palaeolithic context in Romania, also contained the remains of various shelters and habitation structures (mostly in level IV), the largest and most complex of which measured 8m in length and 5m in width (Păunescu 1993). No conclusions can be drawn from such isolated cases, especially given the low resolution of the palaeoclimatic data, the many problems with the excavations at Ripiceni-Izvor, and the fact that other levels at that site also yielded structures identified as shelters. Still, it is possible that the hominins responsible for the material culture found in these levels were less susceptible to climatic conditions and associated changes in resource distribution and availability by virtue of a more complex behaviour, reflected in the presence of artificial shelters, and larger group sizes. Caves appear to have been frequented less often during mild climates, as only three cave sites and six cave levels (50% of cave sites and 30% of cave levels with valid data) show traces of habitation during mild climates, while four cave sites and nine levels (67% of cave sites and 45% of cave levels with valid data), show traces of habitation during the harshest climates. If the palaeoclimatic data is grouped into “cold” and “mild”, the differences become more obvious, as five of the six cave sites (~83%) , and 14 of the 20 cave levels with valid data (~70%) show traces of habitation during “cold” periods. Moreover, the median number of lithics recovered from cave levels deposited during mild climatic conditions (88) is appreciably lower than the median number of lithics recovered from levels deposited during relatively harsh (121) and harsh (154) climatic regimes. It was in fact suggested long ago that “Mousterians” abandoned caves during periods of mild climates (e.g. Nicolăescu-Plopşor 1961: 15), presumably so as to exploit the more varied and plentiful resources of open landscapes. While this may well have been so, the available data is insufficient to draw any conclusions in this regard, as the sample size is too small, and six of the nine cave levels that were deposited during harsh, cold and dry climates represent a single site, namely Ohaba Ponor – Bordul Mare.  36  Table 3: The distribution of climatic conditions across the different levels and sites for which valid data is available.  Site Name  Cold - dry  Cold - wet  Temperate  Total Levels  Ohaba Ponor - Bordu Mare  6  1  2  9  Boroşteni - Peştera Cioarei  1  3  2  6  Râşnov - Peştera Gura Cheii  0  0  2  2  Nandru - Peştera Spurcată  1  0  0  1  Peştera Hoţilor  1  0  0  1  Cheia - La Izvor  0  1  0  1  Boineşti  0  0  1  1  Gornea - Dealul Căuniţei  0  0  1  1  Remetea Somoş I  0  0  1  1  Ripiceni-Izvor  0  2  3  5  Mamaia-Sat  0  0  1  1  Total Levels  9  7  13  29  Total Sites  4  4  8  Cave  Open-Air  5.3.  Climate  Raw materials and assemblage variability  Data on raw materials is available for 47 of the 50 levels included in the database, and 18 of the 21 sites28. Most Middle Palaeolithic assemblages (11 sites and 34 levels, or ~72% of the levels and 61% of sites with valid data) are characterized by the considerable use of coarse-grained raw materials, and are classified here as either “mixed” or “coarse”. Levels characterized by the predominant (>80%) use of coarse-grained raw materials are the most frequent (n=22, or ~47% of 47 valid cases), although they account for only 8158 lithics (median = 62; IQR = 160) and 858 formal tools (median = 17; IQR = 25). With the exception of the Middle Palaeolithic level from the open-air site of Cladova, the predominant use of coarse-grained raw materials is associated exclusively with cave sites (n=7). Levels with “mixed” raw materials (n=12, or 25.5% of valid cases), which account for 1265 lithics (median = 76.5; IQR = 88) and 196 tools (median = 10.5; IQR = 22), are also found most frequently in caves (n=9), the sole exception being the three Middle Palaeolithic levels from the open-air site of Zăbrani. Levels with predominantly (>80%) fine-grained raw materials (n=13, or 28% of valid cases), on the other hand,  28  I did not have access to information on the raw materials used at the site of Livadiţa, and the information available for the sites of Gornea – Dealul Căuniţei and Remetea Oaşului II was insufficient to classify the assemblages into one of the raw material categories with reasonable certainty.  37  account for 60175 lithics (median = 853; IQR = 1792) and 7435 formal tools (median = 343; IQR = 584), and are found at 6 open-air sites as well as a cave site in Dobrogea29. The richness of “fine” raw material levels relative to “coarse” and “mixed” raw material levels in terms of overall lithic yields deserves further discussion, since it is also evidenced in the number of cores (n=1737, median = 45, versus n=160 [median = 2] for “mixed” raw material levels and n=470 [median = 3] for “coarse” raw material levels) and the number of platform-remnant bearing flakes (PRBs) that were produced (n=11422 , median = 542.5, versus n=317 [median = 31] for “mixed” raw material levels and n=2182 [median = 29] for “coarse” raw material levels). This relative richness may be explained by a series of factors, such as differences in the length of habitation episodes, differences in technological traditions, population densities, or in mobility or even artifact discard patterns. It is difficult to ascertain which factor(s) may have played a role, however, due to the poor understanding of the chronology of the deposits and the absence of volumetric data for most open-air sites. It may indeed be that this relative richness does not reflect an underlying archaeological phenomenon at all, but rather differences in how open-air and cave sites were researched. As has already been noted, the statistically significant association between site types (cave or open-air) and different raw material components makes it difficult to isolate the effect(s) of either variable, and it is entirely possible that the observed richness of “fine” raw material levels simply reflects the richness of open-air levels relative to levels excavated at cave sites30. There is, in any case, also an appreciable difference between levels with predominantly coarse-grained and predominantly fine-grained raw material components in terms of the percentage of lithics classified as formal tools, although it is not immediately clear whether this reflects interpretive bias (e.g. differences in the ability of researchers to identify tools made of quartz or quartzite versus tools made of flint) or an underlying archaeological phenomenon. Box-plots of the incidences of formal tools across levels classified in different raw material groups (Figure 3a) reveal that levels with predominantly coarse-grained raw materials show both lower spread and lower central values than levels characterized by the use of predominantly fine-grained raw materials and, indeed, than levels with “mixed” raw materials; in terms of spread at least, “mixed” raw material levels resemble “fine” raw material levels  29  Also found at La Adam, but that site is not included in this analysis due to the poor quality of the available publications. 30 In general, much larger surfaces were excavated at open-air sites than in caves, which would be expected to result, all else being equal, in the recovery of more lithics from open-air levels. Furthermore, there may have been some differences in the ability of researchers to identify distinct levels at caves versus open-air sites. It has been suggested, for example, that Păunescu artificially subdivided the Mousterian level from the site of Râşnov – Peştera Gura Cheii (Cârciumaru et al. 2010: 110-111).  38  more closely. The “coarse” raw material group shows a remarkably tight clustering of values around the median, but also two minor outliers (the Mousterian Level II from Râşnov – Gura Cheii and the Mousterian IIC level from Nandru – Peştera Curată) which display unexpectedly high values that are difficult to explain with the available data. While the tight clustering, slight positive skew, and the presence of outliers may simply reflect the overrepresentation of levels from the site of Ohaba Ponor – Bordul Mare (that site contributes 10 levels, or over 45% of the levels in the “coarse” raw material sample), it is clear from the plots that “coarse” raw material levels are different from both “fine” and “mixed” raw material levels in terms of the incidence of formal tools. It also appears that, at least in terms of the central value of its distribution, the “mixed” raw material group falls somewhere in between the “fine” and “coarse” raw material groups. As expected, when all three raw material groups are considered a statistically significant difference between their variances (Levene’s statistic = 7.734 (2, 44), p = .001) and mean rank values, (x2 (df=2, n=47) = 6.507, p = .039) is revealed. Pairwise comparisons with Bonferroni adjustments (alpha level = .017) indicate a statistically significant difference in variance between levels with predominantly finegrained and predominantly coarse-grained raw materials (Levene’s statistic = 9.111 (1, 33), p = .005), with “coarse” raw material levels showing less variability, and no difference between “fine” and “mixed” raw material levels (Levene’s statistic = .381 (1, 23), p = .543). A difference was also observed between “mixed” and “coarse” raw material levels (Levene’s statistic = 17.225 (1,32), p < .001), with “mixed” raw material levels showing a higher degree of variance. While “mixed” and “coarse” raw material levels do not differ significantly in terms of their mean rank values (U = 131, p = .971), it is nevertheless clear that within the broader group of assemblages characterized by the extensive use of coarse-grained raw materials (“mixed” and “coarse” levels), the incidence of formal tools is correlated with the incidence (%) of fine-grained materials (Rs(30) = .414, p = .023), defined as those materials listed in the ChertEtc variable (see Appendix A). The observed differences in the percentage of lithics that are classified as formal tools (retouched lithics or technologically defined tools) suggest that differences in the diversity of tools may also exist. Diversity is not easy to define, however, since one assemblage may be more diverse than another in terms of the richness or number of tool types it has yielded, but less diverse in terms of the evenness of the distribution of tools across those types (or vice versa). Various measures of diversity which combine these two properties (richness and evenness) exist, the two most popular being Simpson’s index of diversity (D) and Shannon-Weaver’s entropy statistic (H), both of which have been applied in lithic analysis (e.g. Shott et al. 1989; Eerkens et al. 2007; see also Odell 2004 and Andrefsky 2005). As noted 39  by Nagendra, “[t]he Shannon index stresses the richness component and rare […] types, whilst the Simpson index lays greater emphasis on the evenness component and on the dominant […] type” (Nagendra 2002: 178). With the assemblages at hand the concept of richness is problematic, since the proportion of tool types evidenced by a single artifact is high (median = 21%, IQR = 9.3% for assemblages with more than 100 formal tools [n = 10])31, and it is unclear to what extent the presence of these presumably unique artifacts is behaviourally meaningful. Aside from general questions regarding the behavioural significance of Bordesian types (e.g. Bisson 2000), classification according to Bordesian systematics is ambiguous enough that different researchers will seldom reach perfect agreement on the classification of lithics32, and it is therefore unclear whether all unique artifacts are in fact distinct from those classified in other types – they may well reflect researcher bias. Moreover, given the differences in the workability of quartzite and flint, for example, it is expected that certain tool types (e.g. Levallois flakes) will be very rare or entirely absent in assemblages dominated by quartz or quartzite. Within the Bordesian system33, which has a set and well-defined number of tool types, the fact that certain tool types are not expected to occur in coarse raw material levels implies that, all else being equal, these levels will manifest a lower number of tool types. For these reasons, then, variability in levels characterized by the use of different raw material types in terms of tool diversity is evaluated here through the more conservative Simpson index, calculated according to a modified formula proposed by Pielou (Pielou 1969; qtd. in Peet 1974: 291): ∑  where S is the number of distinct tool types identified in a given assemblage, ni is the number of tools classified in the ith tool type, and N is the total number of tools that have been identified. The value of D denotes the probability that two randomly selected tools belong to the same tool type, subtracted from the maximum possible probability (1) so as to vary directly with heterogeneity (see Peet 1974: 291). Assemblages with low D values (high probability of randomly selected tools belonging to the same tool type) will therefore tend to show more clustering of tools into particular tool types, indicating perhaps a higher degree of specialization, while assemblages with high values will tend to show a more even distribution of tools across tool types.  31  The proportion of tool types evidenced by either one or two artifacts is much higher (median = 36%, IQR = 26% for assemblages with more than 100 formal tools (n = 10). 32 There is disagreement, for example, between the results of the classifications conducted by Doboş (2010) and those conducted by Păunescu. 33 See Debénath and Dibble (1994) for a good introduction to Bordesian systematics.  40  Diversity indices could be calculated in 43 cases, or 86% of the 50 levels. Box-plots of diversity values across different raw material types (Figure 3b) reveal relatively symmetrical distributions with slight negative skews in the “mixed” and “coarse” raw material levels, and one minor outlier with an unusually low diversity value, namely Level J from Boroşteni – Peştera Cioarei in the “mixed” raw material group34. The different raw material groups show similar variances (Levene’s statistic = .073 (2, 39), p = .929) but significant differences in mean values (F (2, 39) = 9.141, p = .001). Bonferroni post-hoc tests shows a highly significant difference between “coarse” and “fine” raw material levels (p < .001), and no significant difference between “fine” and “mixed” levels (p = .640) or “mixed” and “coarse” levels (p = .104). In other words, levels with “fine” and “coarse” raw materials differ significantly in terms of tool diversity, with “fine” raw material levels showing more clustering (lower D values) and “coarse” raw material levels showing a more even distribution of tools (higher D values) or higher diversity; levels with “mixed” raw materials, on the other hand, are situated somewhere in between (Figure 3b).  Figure 3a,b: Boxplots showing the distribution of the incidence of formal tools (a) and of the diversity of tool types (b), as calculated according to Simpson’s index of diversity, across levels characterized by the use of different raw material types. The outliers in a are: 1 = Mousterian II, Râşnov – Gura Cheii; 2 = Mousterian IIC, Nandru – Peştera Curată. The outlier in b (1) is Level J, Boroşteni – Peştera Cioarei.  It is important to note that, unlike the incidence of formal tools, tool diversity in “mixed” and “coarse” raw material levels is not correlated with the incidence of any particular rock type35. Still, the observed 34  The number of tools is extremely low (n = 4, or 3.3% of the total lithics), and therefore calculations of diversity for this level are not particularly meaningful. 35 This should not be interpreted as meaning that diversity indices are not correlated with raw material properties, since there is a great deal of variability within each of the rock types considered here, as well as overlap between the categories in terms of raw material quality (durability, suitability for knapping, etcetera). Some quartzites, for example, are easier to flake than some low quality cherts. In fact, even with the available data a partial correlation, accounting for the observed volumetric density of lithics in each level (log transformed density values) indicates that the log transformed Simpson diversity indices show a moderate correlation with the percentage of lithics made of raw materials classified in the Quartz_Quartzite and OtherMat variables (see Appendix A), although, possibly due to the small sample size (n=15), the null hypothesis of no association cannot be rejected at the chosen alpha level (R(15) = .454, p = .067).  41  trends (Figure 3a,b) suggest that raw materials play a role in both the incidence of formal tools and on the diversity of toolkits, at least when lithics are classified according to Bordesian typology. In particular, the presence of coarse-grained materials seems to limit the incidence of formal tools and to result in more diverse assemblages with more evenly represented tool types. Whether the observed trends reflect the constraints imposed by the characteristics of the raw materials36, differences in the ability to recognize and classify tools made of fine- versus coarse-grained rocks, or some other phenomenon, is unclear at this point. In terms of the incidence of retouch, expressed as the ratio of retouched pieces37 to platform-remnant bearing flakes (a proxy for the minimum number of flakes produced), there is no statistically significant difference between levels with predominantly fine-grained raw materials and those characterized by the extensive use of coarse-grained materials, regardless of whether “mixed” and “coarse” levels are amalgamated into a single category (t = -.878 (33), p = .386; see Figure 4a) or are considered separately (x2 (df=2, n=35) = 3.808, p = .149; see Figure 4b)38. Nevertheless, box plots of the incidence of retouch across different raw material groups (Figure 4b) do show an interesting trend39, as the median of the distribution of “mixed” raw material levels is much higher than that of either “fine” or “coarse” raw material levels, lying at the very top of the interquartile range for “fine” raw material group and below only one extreme outlier in the “coarse” raw material sample40. The unusually high incidence of retouch in the “mixed” raw material group suggests that, in general, these levels tend to be associated with what may be broadly defined as curated assemblages, characterized by a more intense and efficient utilization of raw materials, the maintenance of tools before discard, and perhaps tool transport. This interpretation is consistent with the trends observed above with “mixed” and “coarse” raw material levels in terms of the incidence and diversity of formal tools.  36  It is expected, for example, that if final tool morphology is more difficult to control when working with coarsegrained raw materials, there will be less clustering of tools into what may be recognized as well-defined tool types in assemblages rich in coarse-grained raw materials. 37 Defined as Bordes’ essential count minus pseudo-Levallois points and naturally-backed knives (see Debénath and Dibble 1994). 38 A somewhat lower incidence of retouch in the broader group of levels characterized by a low incidence of finegrained materials (“mixed” and “coarse” levels) is expected given the difficulties involved in retouching some coarse-grained raw materials (e.g. Webb and Domanski 2008: 556). 39 It should also be noted that the incidence of retouch in levels characterized by the considerable use of coarsegrained raw materials is correlated with the incidence of fine-grained materials (Rs(18) = .558, p = .016). 40 The presence of the extreme outlier (level IIC from Nandru – Peştera Curată) in the coarse raw material group is likely an artifact of the arbitrary cut-off point (>20% incidence of fine-grained materials for classification in the “mixed” group), as an additional lithic made of chert or other fine-grained material would have resulted in the classification of this assemblage in the “mixed” group.  42  Figure 4a,b: Boxplots showing the ratio of retouched pieces to platform bearing remnants (PBRs) across levels characterized by the use of different raw materials. In figure a “mixed” and “coarse” raw material levels are amalgamated into a single “coarse” category, while in figure b each raw material category is considered separately. The outliers represent: 1) the Mousterian IIIf level from Ohaba Ponor – Bordul Mare, and 2) the F level from Peştera Cioarei – Boroşteni; 3) the Mousterian IVb level from Ohaba Ponor – Bordul Mare; 4) the Mousterian IIC level from Nandru – Peştera Curată.  Evidence that “mixed” raw material levels represent curated assemblages is also found in the observed relationships between the incidence of retouch, the volumetric density of lithics, and the raw material characteristics of the various levels classified here as “mixed” or “coarse”. In their analysis of potential behavioural changes across the transition from the Middle to the early Upper Palaeolithic in Romania, Riel-Salvatore et al. (2008) argue that the relationship between the incidence of retouch (expressed as percentage of the total lithics that are retouched) and the volumetric density of lithics in the original deposits can serve as a proxy indicator for curation behaviours and, in turn, mobility patterns. Their method (WABI) postulates a strong negative correlation between these two variables, with higher incidences of retouch and lower volumetric densities indicating more curated assemblages, and lower incidences of retouch and higher volumetric densities indicating a more expedient use of raw materials (Riel-Salvatore et al. 2008: 401). Utilizing the lithic data that I compiled, which differs from that used by Riel-Salvatore et al.41, a linear regression analysis reveals a strongly negative relationship between the incidence of retouch and lithic volumetric density (R (18) = -.923, p < .001)42, if both are log transformed and three statistically significant outliers, namely levels E, F, and H from Peştera Cioarei, are removed43.  41  Riel-Salvatore et al. (2008) seem to have used Bordes’ essential count as a proxy for retouch, but said count includes pseudo-Levallois points and naturally backed knives, which are not by definition deliberately retouched (see Debénath and Dibble 1994). A revised essential count, which excludes the aforementioned types, is used here as a proxy for retouch. Some other differences between Riel-Salvatore et al.’s data and my own are due to the use of different primary sources. 42 Note that the correlation remains strong and highly significant (R(20) = -.911, p < .001) if the Mousterian level from the site of Gornea – Dealul Căuniţei is included; this level was excluded from the regression analysis because the information on raw materials was insufficient to assign it to one of the three raw material groups. 43 Note that the correlation remains statistically significant at alpha level .05 if the outliers are left in place and a more conservative non-parametric test is conducted (Rs (22) = -.659, p = .001). The presence of these outliers may  43  This result is in agreement with that obtained by Riel-Salvatore et al. (2008). More importantly, a scatterplot of the variables (Figure 5a) shows that “mixed” and “coarse” raw material levels tend to cluster at different points along the regression line, with “mixed” raw material levels (represented by three sites) showing a generally higher incidence of retouch and lower volumetric densities than “coarse” raw material levels (represented by five sites; see Figure 5b), supporting the view that they represent more curated assemblages.  Figure 5a,b: Scatter plots showing the relationship between the percentage of retouched pieces and the volumetric density of lithics in the original deposit, when both variables are log transformed. Figure a shows the placement of mixed and coarse raw material levels along the continuum represented by the regression line, while figure b shows the placement of levels from different sites along the same continuum. Red circles in indicate levels with predominantly (>80%) coarse raw materials, while blue circles indicate levels with mixed raw materials. Statistically significant outliers (1 = Level E; 2 = Level F; 3 = Level H; all from Peştera Cioarei) are also shown in figure a, but they were excluded from all calculations, including the plotted least squares regression line.  Indeed, in these levels (“mixed” and “coarse”) a strong and positive association (Figure 6a) is observed between the incidence of fine-grained materials and the incidence of retouch when these variables are log transformed (R (15) = .787, p < .001)44. As expected given the strong correlation between the incidence of retouch and the volumetric density of lithics on the one hand, and the similarly strong correlation between the incidence of fine-grained materials and the incidence of retouch on the other, the incidence of fine-grained materials is also negatively correlated (R(16) = -.729, p = .001) with the be explained by the fact that data for Peştera Cioarei was compiled from a different publication (Doboş 2010), although Riel-Salvatore et al. also identified outliers from Peştera Cioarei using data published, apparently, by Cârciumaru et al. (2000). 44 Only levels with information on lithic densities are considered in this analysis, since these are the levels being discussed. The levels from the open-air sites of Cladova and Zăbrani, for which information is available on the incidence of retouch and chert but not on lithic densities, are statistically significant outliers. They possibly follow a similar but parallel trend (the sample size [n=3] is too small to tell), explainable perhaps by the fact that these are levels from open-air sites while the other levels are all found in caves. In any case, if the two assemblages from Zăbrani are removed in order to reduce the error in the regression analysis to acceptable levels, the correlation remains strong and highly significant (R(18) = .709, p = .001).  44  volumetric density of lithics (Figure 6b). However, it should be noted that the incidence of fine-grained raw materials does not have a significant influence on the relationship between the incidence of retouch and the volumetric densities of lithics, as a partial correlation between these variables, controlling for the effects of the incidence of fine-grained raw materials, shows a strong (albeit slightly weaker) negative correlation between retouch intensity and lithic volumetric density (R (13) = -.794, p = .001). Importantly, if controlling for the lithic volumetric density, no statistically significant association could be detected between the incidence of fine-grained raw materials and the incidence of retouch (R(13) = .441, p = .1).  Figure 6a,b: Scatter plots showing a) the relationship between the percentage of fine-grained raw materials and the percentage of lithics that are retouched, when both variables are log transformed, and b) the relationship between the incidence of fine-grained raw materials and lithic volumetric density (also log transformed).  This suggests that, in the case of these cave assemblages at least, the incidence of retouch is a factor of mobility (which in turn may be related to shifts in climate – see Riel-Salvatore et al. 2008), rather than the incidence of fine-grained raw materials; fine-grained materials were likely collected and used opportunistically by more mobile populations when and where these were accessible, but the presence or absence of better raw materials does not seem to have had a significant effect on mobility strategies, assuming that curated assemblages do reflect a higher mobility. This, in turn, suggests that the finegrained raw materials found in “mixed” raw material levels may have been collected at some distance from the sites, implying by their transport a limited preference for them, and rendering “mixed” raw material assemblages particularly promising for future sourcing studies. That chert and other finegrained rocks were preferred is also supported, to a certain extent, by the observed distribution of different raw material levels across sites: only three (cave) sites, namely Boroşteni – Peştera Cioarei, Nandru – Peştera Curată, and Râşnov – Peştera Gura Cheii have yielded both “coarse” and “mixed” raw material levels, and no sites have yielded both “mixed” and “fine” nor “fine” and “coarse” ones, as would be expected if accessibility was not an issue and coarse-grained raw materials were preferred by some groups, as suggested, for example, by Mogoşanu (1978). 45  Although not surprising, these are important observations since they suggest that “mixed” raw material levels do not reflect an almost exclusive use of raw materials available in the immediate vicinity of sites (the median proportion of fine-grained raw materials in the 10 “mixed” raw material levels with valid data is 31.71%, with an IQR of 25.62, and a maximum of 44.44%45), and that the scarcity of chert in many of the assemblages is a reflection of lower accessibility rather than any preference for coarse raw materials. It must be noted at this point that chert is locally available at many of the sites that are characterized by the extensive use of coarse-grained materials (e.g. Nandru – Peştera Curată, Nandru – Peştera Spurcată, and Ohaba Ponor – Bordul Mare), but it is mostly of a very low quality, and it appears to have been used only occasionally. To what degree local cherts were used, and to what degree “mixed” raw materials levels do reflect transport from relatively distant sources can only be clarified through future sourcing studies, however. 5.4.  Summary  In this chapter I have presented and discussed the data available for the 21 adequately published and well-researched Romanian Middle Palaeolithic sites from a quantitative perspective. It has been established that the available sample is most likely biased, and that the evidenced patterns are therefore difficult to interpret and evaluate. Nevertheless, several important observations on the characteristics of the data and of the record itself have been made. First, the Romanian Middle Palaeolithic record appears to be quite poor in terms of the number of lithics it has produced, which suggests Romania was less intensely occupied than other parts of Europe throughout the Middle Palaeolithic. Second, the hominins responsible for the majority of assemblages appear to have been illequipped to deal with adverse climatic conditions, and consequently much of the territory of present day Romania appears to have been sparsely inhabited, or perhaps even uninhabited, during harsh climatic regimes. Third, most of the assemblages are characterized by the extensive use of coarsegrained raw materials, but the grouping of these assemblages into a single category obfuscates important differences. A more detailed analysis reveals that the lower incidence of formal tools and the higher tool diversity of these assemblages likely reflect raw material constraints on lithic production, while the association of these assemblages with cave sites may be explainable in terms of climate; both observations bring into question the reality of the so-called Cave or Quartzitic Mousterian facies. Fifth, the incidence of fine-grained raw materials in assemblages characterized by the extensive use of coarsegrained materials appears to reflect mobility patterns, with more mobile groups evidencing a higher degree of utilization of fine-grained raw materials, which were likely transported from some distance.  45  Note that these numbers include only those materials classified in the ChertETC variable.  46  6. Conclusions The aim of this thesis was to present and critically assess the information available on the Romanian Middle Palaeolithic record. Through an analysis of the history of research, an evaluation of the current state of knowledge, and a limited quantitative analysis of the data available from adequately researched and well-published sites, I have shown that, notwithstanding the impressions one may get from the existing literature, the known Middle Palaeolithic record of Romania is neither particularly rich nor extensively studied, even from a culture-historical perspective. In fact, the total number of Middle Palaeolithic lithic artifacts recovered to date is likely less than 100,000 (including manufacturing byproducts), and of the approximately 120 locations thought to have yielded reliable traces of Middle Palaeolithic habitation only half have been excavated to some degree. Moreover, if only sites which have yielded more than a handful of lithics are considered, the number of sites that can be said to have been reasonably well-researched and well-published is no greater than 21. It is therefore clear that, despite a history of research that spans over a century, the Middle Palaeolithic record of present day Romania is poorly known today. I have also shown that, unfortunately, the data available from the 21 reasonably well-researched sites is problematic due to the generally poor quality of the excavations conducted in the twentieth-century, the almost exclusive focus on the lithic components, and the lack of scientific evaluation of the effects of raw material variability on the variability evidenced in the composition of the different assemblages, most of which are characterized by the extensive use of coarse-grained raw materials. More importantly, I have shown that the sample constituted by these sites is most likely highly biased, and that the interpretation of patterns evidenced in the data is difficult. In short, I have shown that given the current state of knowledge, the available data on the Romanian Middle Palaeolithic record should be approached with caution. As has been discussed throughout this thesis, several factors contributed to the poor understanding of the record. First, while multidisciplinary research was advocated by Romanian Palaeolithic archaeologists since the beginning of the twentieth century, it largely failed to deliver on its promise. This was mainly due to the clear separation of roles and the autonomy of the various specialists that worked in multidisciplinary teams, but also to the fact that following the imposition of socialism, Palaeolithic researchers were mostly trained in the humanities and lacked the scientific expertise required to continue the work of the earliest scholars. Systemic factors, such as the isolation of Romanian researchers from the international scene for much of the twentieth century, inconsistent 47  funding opportunities, the centralization of research, and the inner logic of the socialist regime, which focused on quantity of products rather than their quality, also played a major role. Given the focus of this thesis and the limitations of the available data, I did not attempt to investigate in detail the adaptive capacities, lifeways, and behaviours evidenced in the Middle Palaeolithic record of present day Romania. Nevertheless, on the basis of the available data I have argued that much of the territory appears to have been sparsely inhabited, or even uninhabited, during harsh climatic regimes, and that the hominins responsible for the majority of Romanian Middle Palaeolithic assemblages appear to have been ill-equipped to deal with adverse climatic conditions. I have also argued that the incidence of fine-grained raw materials in assemblages characterized by the extensive use of coarse-grained raw materials is correlated with mobility patterns, and that higher incidences can likely be explained in terms of the increased accessibility of more distant sources by more mobile groups. As noted by Maria Bitiri (1972) among many others, there is little evidence to suggest that Middle Palaeolithic hominins travelled with the specific purpose of collecting high quality raw materials from non-local sources. However, the available data suggest that high quality fine-grained raw materials were preferred and were consequently exploited when and where available, even by hominins accustomed to processing coarse-grained raw materials on a regular basis. More importantly, the data suggest that many Romanian Middle Palaeolithic assemblages do not reflect the almost exclusive use of raw materials available in the immediate vicinity of sites. Beyond these observations, it must be noted that the overall poverty of the record stands in stark contrast with the many outstanding discoveries made to date, some of which include evidence for complex behaviours that are puzzling given the low population densities implied by the available data. This contrast indicates that Romania has a high potential for future research, and many promising avenues have been identified throughout this thesis. First, new and properly funded excavations in regions with rich but poorly studied Middle Palaeolithic sites (e.g. the areas near Gurahonţ, as well as caves in Dobrogea) have the most potential to rectify the biases in the available data. In this regard, the Lower Danube Survey for Palaeolithic Sites project (http://www.lodans.wordpress.com), which started in 2010, is very promising. Secondly, the re-analysis of old faunal collections is critical, not only because there is currently very little information available on Middle Palaeolithic subsistence in the territory of present day Romania, but also because such analyses could reveal the presence of hominin remains that have heretofore escaped the attention of archaeologists. Third, the known Neanderthal remains should be re-analyzed, as they have not been subject to any modern investigation. Fourth, there is an urgent need to obtain more reliable absolute dates, particularly at sites where the late (post 30,000 BP) survival 48  of Mousterian industries has been postulated. Fifth, sourcing studies of fine-grained materials from assemblages characterized by the extensive use of coarse-grained raw materials are likely to reveal more complex landscape utilization patterns than previously thought. Lastly, the re-analysis of lithic materials recovered during the early decades of the twentieth century, particularly by Roska, are likely to considerably enhance our understanding of the Romanian Middle Palaeolithic record by expanding the well-published lithic collections by a considerable amount.  49  Works Cited Andrefsky, William Jr. 1994 Raw-Material Availability and the Organization of Technology. American Antiquity 59(1): 21-34. 2005 Lithics: Macroscopic Approaches to Analysis, Second Edition. Cambridge University Press, Cambridge. Anghelinu, Mircea 2003 Note Privind Impactul Marxismului în Teoria şi Practica Arheologiei Preistorice din România. Cercetări Arheologice XII: 275-303. 2004 Evoluția gândirii teoretice în arheologia Românească: Concepte şi modele aplicate în preistorie. Cetatea de Scaun, Târgovişte. 2006 O Paleontologie a “Omului Etern”: Arheologia Paleoliticului din România. Cercetări Arheologice XIII: 135-158. Anghelinu, Mircea, and Loredana Niţă 2010 Cei Uitati: Femeile şi copiii în cercetarea epocii paleolitice. In Arta Antropomorfă feminină în preistoria spaţiului Carpato-Nistrean, edited by Vasile Chirica and George Bodi. Pp. 34-108. Editura PIM, Iaşi. Babeş, Mircea 2007 Arheologia Românească şi congresele internationale în epoca “războiului rece.” Studii şi cercetări de istorie veche şi arheologie 58(3-4): 321-344. Behrens, J. T., and Chong-Ho Yu 2003 Exploratory Data Analysis. In Handbook of Psychology, Volume 2: Research methods in Psychology, edited by J. A. Schinka and W. F. Velicer, pp. 33-64. Wiley and Sons, NJ. Bisson, Michael S. 2000 Nineteenth Century Tools for Twenty-First Century Archaeology? Why the Middle Paleolithic Typology of Franҫois Bordes Must Be Replaced. Journal of Archaeological Method and Theory 7(1): 1-48. Bitiri, Maria 1970 Cercetări paleolitice în nord-vestul României. Revista Muzeelor 4: 293-298. 1972 Evoluţia culturii materiale în paleoliticul din depresiunea Oaş. Satu Mare, Studii şi comunicări II: 35-49. Bitiri-Ciortescu, Maria 1987 Paleoliticul de la Mitoc – Valea Izvorului: Probleme privind începutul paleoliticului superior pe teritoriul României. Studii şi Cercetări de Istorie Veche şi Arheologie 38(3): 207-223. Braun, David R., Thomas Plummer, Joseph V. Ferraro, Peter Ditchfield, and Laura C. Bishop 2009 Raw material quality and Oldowan hominin toolstone preference: evidence from Kanjera South, Kenya. Journal of Archaeological Science 36(7): 1605-1614. Breuil, H. 1925 Stations Paléolithiques en Transylvanie. Bulletin de la Société des Sciences de Cluj II(2): 193-217. Brudiu, M. 1997 Deux Amis: Nicolae N. Moroşan et Nicolae Zaharia (une page de l’histoire de l’archéologie Roumaine). Studia Antiqua et Archaeologica III-IV: 21-26. Cârciumaru, Marin 1980 Mediul Geografic în Pleistocenul Superior şi Culturile Paleolitice in România. Editura Academiei Republicii Socialiste România, Bucharest. 50  1989  Contexte stratigraphique, paléoclimatique et géochronologique des civilisations du Paléolithique moyen et supérieur en Roumanie. L’anthropologie 93(1): 99-122. 1995 Review of “Alexandru Paunescu, Ripiceni-Izvor. Paleolitic şi Mezolitic.” Studii şi Cercetări de Istorie Veche şi Arheologie 46(1): 73-80. 1999 Le Paléolithique en Roumanie. Jérôme Millon, Grenoble. Cârciumaru, Marin, Mircea Anghelinu, Maria Bitiri-Ciortescu, Dana-Dierna Cârciumaru, Radu Cârciumaru, Jean Chaline, Ovidiu Cîrstina, Marian Cosac, Rodica Dincă, Roxana Dobrescu, Erika Gál, Eugen Kessler, Dan Iulian Mărgărit, Marie Hélène Moncel, Marcel Otte, Romică Pavel, Monica Sandu, Marian Şeclăman, Elena Terzea, Marguerite Ulrix-Closset, Pompiliu Vasilescu 2000 Peştera Cioarei Boroşteni: Paleomediul, cronologia şi activitățile umane în Paleolitic. Editura Macarie, Târgovişte. Cârciumaru, Marin, and Elena-Cristina Niţu 2008 Considérations stratigraphiques et géochronologiques concernant le depot de la grotte Bordul Mare de Ohaba Ponor (Roumanie) - Regard rétrospectif et conclusions interdisciplinaires. Annales d’Université ‘Valahia’ Târgovişte X.1: 119-146. Cârciumaru, Marin, Elena-Cristina Niţu, Jean-Guillaume Bordes, George Murătoreanu, Marian Cosac, and Radu Ştefănescu 2008 Le paléolithique de la grotte Gura Cheii – Râşnov, edited by Marin Cârciumaru. Valahia University Press, Târgovişte. Cârciumaru, Marin, Elena-Cristina Niţu, Roxana Dobrescu, and Radu Ştefănescu 2010 Paleoliticul din judeţul Brasov, edited by Marin Carciumaru and Elena-Cristina Nitu. Valahia University Press, Targoviste. Cârciumaru, Marin, Marie-Hélène Moncel, Mircea Anghelinu, and Radu Cârciumaru 2002 The Cioarei-Borosteni Cave (Carpathian Mountains, Romania): Middle Palaeolithic finds and technological analysis of the lithic assemblages. Antiquity 76: 681-690. Cârciumaru, Marin, Marie-Hélène-Moncel, and Radu Cârciumaru 2000 Le Paleolithique moyen de la grotte Cioarei-Boroşteni (commune de Pestisani, department de Gorj, Roumanie). l’Anthropologie 104: 185-237. Cârciumaru, Marin, and Minodora Ţuţuianu-Cârciumaru 2009 L’ocre et les recipients pour ocre de la grotte Cioarei. Annales d’Université Valahia Targoviste XI(1): 7-19. Cârciumaru, Marin, Mircea Anghelinu, Elena-Cristina Niţu, Marian Cosac, and George Murătoreanu 2007 Geo-Archéologie du Paléolithique Moyen, Paléolithique Supérieur, Epipaléolithique et Mésolithique en Roumanie. Editura Cetatea de Scaun, Târgovişte. Cârciumaru, Marin, Rodica-Mariana Ion, Elena-Cristina Niţu, and Radu Ştefănescu 2012 New evidence of adhesive as hafting material on Middle and Upper Palaeolithic artefacts from Gura Cheii-Râşnov Cave (Romania). Journal of Archaeological Science 39: 1942-1950. Debénath, André, and Harold L. Dibble 1994 Handbook of Paleolithic Typology, Vol. 1. University of Pennsylvania, Philadelphia. Dibble, Harold L. 1991 Local Raw Material Exploitation and its Effects on Lower and Middle Paleolithic Assemblage Variability. In Raw Material Economies among Prehistoric Hunter-Gatherers, edited by A Montet-White and S. Holen, pp. 217-230. University of Kansas, Lawrence. Dobrescu, Roxana, Adrian Doboş, Gabriel Popescu, Roxana Carp, Alain Tuffreau, Alexandru Petculescu, Sandra Bălescu, Emilie Goval, Bertrand Lefevre, David Hérisson, Laurent Locheron, Alexine Pahaut, and Ionuţ Cătălin Dobrinescu 2008 Târguşor, com. Târguşor, jud. Constanţa. Punct: Peştera La Adam. In Cronica cercetărilor arheologice din România – campania 2007, edited by Mircea Victor Angelescu and Florea Vasilescu, pp. 317-318. CIMEC, Bucharest. 51  Doboş, Adrian 2005 C. S. Nicolăescu-Plopşor şi arheologia paleoliticului. Studii de Preistorie 2: 233-247. 2010 Paleoliticul mijlociu de pe teritoriul de la sud şi sud-est de Carpați (Oltenia, Muntenia, Dobrogea). Unpublished PhD dissertation, Institutul de Arheologie “Vasile Pârvan”, Bucharest. Doboş, Adrian, Andrei Soficaru, and Erik Trinkaus 2010 The Prehistory and Paleontology of the Peştera Muierii (Romania). ERAUL, Liège. Doize, Renée-L., and Nicolae N. Moroşan 1931 Effets de quelques Agents physiques, sur les roches employées par le Préhistoriques et spécialement sur l’Outillage. Bulletin de la Société Préhistorique Française XXVIII: 411427. Driscoll, Killian 2010 Understanding quartz technology in early prehistoric Ireland. Unpublished PhD dissertation. School of Archaeology, University College Dublin, Dublin. Eerkens, Jelmer W., Jeffrey R. Ferguson, Michael D. Glascock, Craig E. Skinner, and Sharon A. Warchter. 2007 Reduction Strategies and Geochemical Characterization of Lithic Assemblages: A Comparison of Three Case Studies from Western North America. American Antiquity 72(3): 585-597. Eren, Metin I., Stephen J. Lycett, Christopher I. Roos, and C. Garth Sampson 2011 Toolstone constraints on knapping skill: Levallois reduction with two different raw materials. Journal of Archaeological Science 38: 2731-2739. Forsström, Lars 2001 Duration of interglacials: a controversial question. Quaternary Science Reviews 20(15): 1577-1586. Gaál, Stefan 1928 Der erste mitteldiluviale Menschenknochen aus Siebenbürgen. Die palaeontologischen und archaeologischen Ergebnisse der in Ohabaponor ausgeführten Hohlenforschungen. Publicaţiile Muzeului Judeţului Hunedoara XXV-XXVI: 61-112. 1943 Újabb ember- és emlıscsontleletek Erdély moustérijéböl. Közlemények III(1): 1-46. Gábori-Csánk, V., I. Dienes, M. Kretzoi, P. Kriván, E. Krolopp, and J. Stieber 1968 La station du Paléolithique Moyen d’Érd-Hongrie, edited by Gábori-Csánk. Akadémiai Kiadó, Budapest. Gáll, Erwin 2010 Márton Roska (1880-1961) şi cercetarea arheologică a secolelor X-XI. Studii şi Cercetări de Istorie Veche şi Arheologie 61(3-4): 281-306. Good, I. J. 1983 The Philosophy of Exploratory Data Analysis. Philosophy of Science 50: 283-295. Goran, C. 1981 Catalogul Sistematic al Peşterilor din România. Institutul de Speologie Emil Racoviţa, Bucharest. Harvati, Katerina, Philipp Gunz, and Dan Grigorescu 2007 Cioclovina (Romania): affinities of an early modern European. Journal of Human Evolution 53: 732-746. Hesse, Reinhard 1989 Silica Diagenesis: Origin of Inorganic and Replacement Cherts. Earth-Science Reviews 26: 253-284. Hodder, Ian (editor) 1991 Archaeological theory in Europe: the past three decades. Routledge, NY.  52  Honea, Kenneth 1981 New Romanian Radiocarbon Dates: Middle Palaeolithic, Mesolithic, Neolithic. American Journal of Archaeology 85: 483-486. 1984 Cronologia Paleoliticului mijlociu şi superior din România. Implicaţiile rezultatelor actuale ale datării cu carbon radioactiv. Revista Muzeelor şi Monumentelor XXI(3): 5169. Hopkinson, Terry 2007 The transition from the Lower to the Middle Palaeolithic in Europe and the incorporation of difference. Antiquity 81: 294-307. Horvath, Ildiko 2009 The Early Upper Palaeolithic in Romania: past and current research. In The Mediterranean from 50,000 to 25,000 BP: Turning Points and New Directions, edited by Maria Camps and Carolyn Szmidt, pp. 137-162. Oxbow Books, Oxford. Hublin, Jean-Jacques, and Wil Roebroeks 2009 Ebb and flow or regional extinctions? On the character of Neandertal occupation of northern environments. Comptes Rendus Paleovol 8: 503-509. Jöris, Olaf, and Martin Street 2008 At the end of the 14C time scale – the Middle to Upper Paleolithic record of western Eurasia. Journal of Human Evolution 55(5): 782-802. Lascu, Cristian, Florian Baciu, Mihai Gligan and Serban Sarbu 1996 A Mousterian cave bear worship site in Transylvania, Roumania. Journal of Prehistoric Religion 10: 17-30. Maroto, Julià, Manuel Vaquero, Álvaro Arrizabalaga, Javier Baena, Enrique Baquedano, Jesús Jordá, Ramon Julià, Ramón Montes, Johannes Van Der Plicht, Pedro Rasines, and Rachel Wood 2012 Current issues in late Middle Palaeolithic chronology: New assessments from Northern Iberia. Quaternary International 247: 15-25. McPherron, Shannon, and Harold L. Dibble 1999 The lithic assemblages from Pech de l’Azé (Dordogne, France). Préhistoire Européenne 15: 9-43. Mellars, Paul 2004 Neanderthals and the modern human colonization of Europe. Nature 432(25): 461-465. Mertens, Steven B. 1996 The Middle Paleolithic in Romania. Current Anthropology 37(3): 515-521. Mogoşanu, Florea 1978 Paleoliticul din Banat. Editura Academiei Române, Bucureşti. Moroşan, Nicolae N. 1931 Existe-t-il du Micoquien en Bessarabie et quelle serait sa place dans la chronologie du pléistocène? Bulletin de la Société préhistorique Française 28(4): 223-233. 1934 La plus jolie “Pointe en feuille de laurier solutréenne” de la Roumanie. Son importance archéologique et géologique. Bulletin de la Société préhistorique Française 31(10): 451457. Murray, Tim, and Christopher Evans (editors) 2008 Histories of Archaeology: A reader in the History of Archaeology. Oxford University Press, Oxford. Nagenda, Harini 2002 Opposite trends in response for the Shannon and Simpson indices of landscape diversity. Applied Geography 22: 175-186. Nicolăescu-Plopşor, Constantin S. 1938 Le Paléolithique en Roumanie. Dacia 5-6: 41-107. 1957 Cercetări asupra paleoliticului timpuriu. Materiale 3: 281-290. 53  1961 Geochronology of the Palaeolithic in Rumania. Dacia, NS 5: 5-19. Odell, G. H. 2004 Lithic Analysis. Springer, New York. Onac, Bogdan P., Iosif Viehmann, Joyce Lundberg, Stein-Erik Lauritzen, Chris Stringer, and Vasile Popiţă 2005 U-Th ages constraining the Neanderthal footprint at Vârtop Cave, Romania. Quaternary Science Reviews 24: 1151-1157. Patou-Matis, Maryléne 2000 Neanderthal Subsistence Behaviours in Europe. International Journal of Osteoarchaeology 10: 379-395. Păunescu, Alexandru 1989 Le Paléolithique et le Mésolithique de Roumanie (un bref apercu). L’anthropologie 93(1): 123-158. 1993 Ripiceni-Izvor, Paleolitic şi Mezolitic: Studiu Monografic. Editura Academiei Române, Bucureşti. 1998 Paleoliticul şi epipaleoliticul de pe teritoriul Moldovei cuprins între Carpaţi şi Siret, Vol. I/1. Editura Satya Sai, Bucureşti. 1999a Paleoliticul şi mezoliticul de pe teritoriul Moldovei cuprins intre Siret şi Prut, vol. 2. Editura Satya Sai, Bucureşti. 1999b Paleoliticul şi mezoliticul de pe teritoriul Dobrogei, vol. II. Editura Satya Sai, Bucureşti. 2000 Paleoliticul şi mezoliticul din spațiul cuprins între Carpaţi şi Dunăre. Editura Agir, Bucureşti. 2001 Paleoliticul şi mezoliticul din spațiul transilvan. Editura Agir, Bucharest. 2003 Din istoria arheologiei Româneşti pe baza unor documente de arhivă. Editura Agir, Bucharest. Peet, Robert K. 1974 The Measurement of Species Diversity. Annual Review of Ecology and Systematics 5: 285-307. Pinhasi, Ron, Thomas F. G. Higham, Liubov V. Golovanova, and Vladimir B. Doronichev 2011 Revised age of late Neanderthal occupation and the end of the Middle Paleolithic in the northern Caucasus. PNAS 108(21): 8611-8616. Porat, N., M. Chazan, H. Schwartz, and L. K. Horwitz 2002 Timing of the Lower to Middle Paleolithic boundary: New dates from the Levant. Journal of Human Evolution 43: 107-122. Riel-Salvatore, Julien, Gabriel Popescu, and C. Michael Barton 2008 Standing at the gates of Europe: Human behavior and biogeography in the Southern Carpathians during the Late Pleistocene. Journal of Anthropological Archaeology 27: 399-417. Roebroeks, Wil 2008 Time for the Middle to Upper Paleolithic transition in Europe. Journal of Human Evolution 55: 918-926. Rolland, Nicolas, and Harold L. Dibble 1990 A New Synthesis of Middle Palaeolithic Variability. American Antiquity 55(3): 480-499. Roska, Márton 1912 Traces de l’homme diluvien dans la caverne “Cholnoky” à Csoklovina. Dolgozatok az Erdélyi Nemzeti Múzeum Érem – és Régiségtárából 3(2): 238-249. 1924 Recherches préhistoriques pendant l’année 1924. Dacia I: 297-313. 1933 Recherches Paléolithiques en Transylvanie, en 1927. Dacia 3-4: 8-23. Roşu, Lucian 2004 With C.S. Nicolăescu-Plopşor through the ages. Mica Valahie, Bucureşti. 54  Sackett, James R. 1981 From the Mortillet to Bordes: a century of French Palaeolithic research. In Towards a History of Archaeology, edited by Glyn Daniel, pp. 85-99. Thames and Hudson, New York. 1991 Straight Archaeology French Style: The Phylogenetic Paradigm in Historic Perspective. In Perspectives on the Past: Theoretical Biases in Mediterranean Hunter-Gatherer Research, edited by Geoffrey A. Clark, pp. 109-139. University of Pennsylvania Press, Philadelphia. Shackleton, Nicholas J., Maria Fernanda Sánchez-Goñi, Delphine Pailler, and Yves Lancelot 2003 Marine Isotope Substage 5e and the Eemian Interglacial. Global and Planetary Change 36: 151-155. Shanks, Michael, and Christopher Tilley 1989 Archaeology into the 1990s. Norwegian Archaeological Review 22(1): 1-12. Shott, Michael J. 1989 Diversity, Organization, and Behavior in the Material Record: Ethnographic and Archaeological Examples. Current Anthropology 30(3): 283-315. Shott, Michael J. 2008 Lower Paleolithic Industries, Time, and the Meaning of Assemblage Variation. In Time in Archaeology: Time Perspectivism Revisited, edited by Simon Holdaway and LuAnn Wandsnider, pp. 46-60. University of Utah Press, Salt Lake City. Soficaru, Andrei, Adrian Doboş, and Erik Trinkaus 2006 Early modern humans from the Peştera Muierii, Baia de Fier, Romania. PNAS 103(46): 17196-17201. Soficaru, Andrei, Catalin Petrea, Adrian Doboş, and Erik Trinkaus 2007 The Human Cranium from the Peştera Cioclovina Uscată, Romania. Current Anthropology 48(1): 611-619. Terzea, Elena 1979 Mamiferele cuaternare din unele peşteri şi adăposturi sub stîncă din zona Porţile de Fier. In Speologia, edited by Traian Orghidan and Ştefan Negrea, pp. 104-139. Editura Academiei Române, Bucureşti. Trigger, Bruce 2006 History of Archaeological Thought, 2nd Ed. Cambridge University Press, New York. Trinkaus, Erik, Oana Moldovan, Ştefan Milota, Adrian Bîlgăr, Laurenţiu Sarcina, Sheela Athreya, Shara E. Bailey, Ricardo Rodrigo, Gherase Mircea, Thomas Higham, Christopher Bronk Ramsey, and Johannes van der Plicht 2003 An early modern human from the Peştera cu Oase, Romania. PNAS 100(20): 1123111236. Tuffreau, Alain, Vasile Boroneant, Emilie Goval, Bertrand Lefevre, Adina Boroneant, Adrian Dobos, and Gabriel Popescu 2007 Le Gisement Paléolithique Moyen de Zăbrani (Département d’Arad). Materiale şi Cercetări Arheologice, Seria Noua III: 2007: 1-17. Tuffreau, Alain, Vasile Chirica, Sanda Bălescu, and Paul Haesaerts 2009 Nouvelles recherches sur le gisement paléolithique de Mitoc Valea Izvorului (dép. de Botoşani): Fouilles 2003-2004. Materiale şi Cercetări Arheologice, NS V: 21-31. Tukey, J. W. 1977 Exploratory data analysis. Addison-Wesley, Reading, MA. Vasil’ev, Sergey A. 2002 Palaeolithic studies in Russia: Retrospect and prospects. In Archaeologies of Europe: History, Methods and Theories, edited by Peter F. Biehl, Alexander Gramsch, and Arkadiusz Marciniak, pp. 255-270. Waxmann Verlag, Münster. 55  Vaquero, Manuel 2008 The history of stones: behavioural inferences and temporal resolution of an archaeological assemblage from the Middle Palaeolithic. Journal of Archaeological Science 35: 3178-3185. Verdery, Katherine 1996 What was Socialism, and What Comes Next? Princeton University Press, NJ. Webb, J. A., and M. Domanski 2008 The relationship between lithology, flaking properties and artefact manufacture for Australian silcretes. Archaeometry 50(4): 555-575. Wilson, Lucy 2007 Understanding Prehistoric Lithic Raw Material Selection: Application of a Gravity Model. Journal of Archaeological Method and Theory 14: 388-411.  56  Appendix A: A partial database of the Romanian Middle Palaeolithic record Database goals and availability: The scope of this thesis and the overall goal of reviewing the Middle Palaeolithic record of Romania as a whole required data to be compiled from the available literature, as re-analysis of collections from more than a couple of sites would have been impossible given the available resources. In accord with the primary goals of the thesis, data collection was aimed at producing a database that would be: a. Informative in regard to the basic properties of the record. b. Useful for investigating assemblage variability on a country-wide scale. c. Suitable for conducting statistical analyses Compiling data suitable for statistical analyses was a non-trivial and error-prone task. First, the information found in different publications is often incompatible; for example, a different number of lithics may be reported for a single assemblage, the site stratigraphy that is presented may be different, or the focus may be on different variables. Second, inconsistencies in the information provided within individual publications were often found (e.g. the numbers that are given on different pages or in different tables may not add up). Third, in many cases information is provided through lengthy descriptions that vary in considerably in detail, even within individual publications. Moreover, the information is often not reported in a systematic way – raw material and retouch characteristics may be reported for some lithics but not all, without any real indication so as to why this is so (e.g. there are no indications that such information is reported only for a particular type of artifacts). Fourth, different analysts often arrive at slightly different classifications, and may focus only on a sub-set of artifacts to answer specific questions; since only summary descriptions (either quantitative or qualitative) are available in the literature, in other words, information is not tied to a specific artifact ID, it was often impossible to combine data reported by different researchers. To minimize inconsistencies and errors data was extracted, when and where possible, from Păunescu’s comprehensive synthesis and repertory of Romanian Middle Palaeolithic finds (see the Data and Methods chapter), and the information was checked for consistency. Despite this, however, it is virtually impossible to guarantee that the database is free of errors. While I take responsibility for such errors, I do so only in the context of this thesis; the information contained in this database, which has been made available in SPSS (v19) and CSV format at http://hdl.handle.net/2429/44312 so as to facilitate further exploration, should not be used without consulting and citing the sources from which it was extracted (listed in the PrimaryRef variable – see below) and verifying the accuracy of the data. 57  Variable selection criteria: To fulfill the goals outlined above, the Romanian literature was reviewed and a comprehensive list of potential variables was created for all well-researched Middle Palaeolithic sites. These variables were then reviewed and a sub-set was selected for inclusion into the database according to the following criteria: a. Quantifiability: Only those variables which could be quantified reliably and with relative ease were considered. b. Frequency and compatibility across sites: Only those variables for which data could be compiled for multiple sites were included. Moreover, only data reported according to the same standards were considered. c. Compatibility within sites: Only those variables which could be reliably associated with the most intensively studied materials, namely the lithic components, were considered. d. Non-redundancy: Variables which could be computed from other, primary variables already present in the database (e.g. various typological indices) were not included. e. Relevance: Variables which were not considered useful for investigating assemblage variability on a country-wide scale were excluded. A large number of variables were thus excluded from the database. Examples include inconsistently reported variables such as retouch types and the mass of the lithic materials, but also rare or unique finds such as the ochre containers identified at Peştera Cioarei (Cârciumaru et al. 2002), the Neanderthal remains from Ohaba Ponor – Bordu Mare (Gaál 1928; Gaál 1943) and Livadiţa (Terzea 1979: 114), or the artificial shelters from Ripiceni-Izvor (Păunescu 1993). More importantly, data on faunal remains and absolute dates were not included due to their general unreliability and the difficulties of associating them with the lithic components. Data on the spatial distribution of lithics was also excluded due to their poor resolution and the difficulties of quantifying these variables in ways that are meaningful for country-wide analyses. Most of the selected variables pertain to the lithic components, since these constitute the most intensively studied aspect of the record and quantitative data is readily available. The only consistently reported, lithic-related variables that are not included in the database (due to time constraints) are those related to core types. Only a few variables related to other aspects of the record could be selected (see below).  58  Database structure and contents: The database is structured in terms of, and contains information on, 53 assemblages from 21 sites. The majority (50) of these assemblages represent archaeological levels (see Chapter 5), but one assemblage from the site of Mamaia-Sat and two assemblages from the site of Ohaba Ponor – Bordu Mare consist of poorly provenienced artifacts which could not be assigned to one of the archaeological levels identified in the reference publications (listed in the PrimaryRef variable, defined below). In the case of MamaiaSat this is because some lithics were recovered from a secondary context, but in the case of Ohaba Ponor – Bordul Mare this was caused by revisions to the original stratigraphy (see Păunescu 2001: 264298). One of the assemblages (“Mousterian III (no level)”) from this latter site (Ohaba Ponor – Bordu Mare) includes artifacts from all levels belonging to the original “Mousterian III” level, while the other assemblage (“Mousterian discovered by Roska”) includes artifacts from all Mousterian levels excavated at the site. Data is organized in 98 primary variables as well as 13 auxiliary variables derived from these and identified by the aux_ prefix. Most of the primary variables (79) provide information on the technological and typological attributes of the lithic components. The remaining primary variables provide general information on site (6 variables) and assemblage (6 variables) attributes, palaeoclimatic regimes (2 variables), and raw materials (5 variables). Although computable from the primary variables, the auxiliary variables are included because they are either directly referenced, or used to compute variables which are referenced in the text. All variables are defined in section 4 below, in the order in which they appear in the database file. Missing data is identified by a value of -1, in cases where no information could be found in the reviewed literature, or a value of -2 for cases where some information was available but could not be reliably encoded in a given variable. This latter value (-2) applies to cases where, for example, data is only available for a group of assemblages, or cases where the available information does not match the scale of the encoded variable (e.g. it is stated in the literature that most of the lithics are made of fine-grained raw materials, but no actual numbers are given and the encoding variable is on a ratio scale). Variable definitions: Primary variables: 1. AssemblageID: This nominal scale, numeric variable is used to store a unique identifier for each assemblage.  59  2. AssemblageName: This nominal scale, string variable identifies assemblages by their given name. 3. AssemblageType: This nominal scale, numeric variable identifies the type of the assemblage. A value of 1 represents assemblages with relatively well-provenienced artifacts which correspond to archaeological levels, while a value of 2 represents aggregates of artifacts with poor provenience information. 4. SiteName: This nominal scale, string variable identifies the name of the site at which a given assemblage was excavated. 5. SiteType: This nominal scale, numeric variable encodes the type of site at which a given assemblage was excavated. A value of 1 represents open-air sites while a value of 2 represents cave sites. No rock-shelter sites are included in the database, and values are known for all sites. 6. PrimaryRef: This nominal scale, string variable lists the main reference(s) used to extract data for a given assemblage. The publications are given in order of importance, and in cases of conflicting information data was taken from the first of the listed publications. 7. CoordinateMet: This nominal scale, numeric variable encodes the method used to obtain data on site coordinates and absolute elevations. A value of 1 indicates that the coordinates and elevations were recorded with the aid of a GPS unit (Garmin Oregon 450), a value of 2 indicates that these were extracted from the literature, and a value of 3 indicates that the coordinates and corresponding elevations were estimated, on the basis of descriptions provided in the literature, with the aid of Google Earth and/or topographic maps. 8. SiteLatitude: This interval scale, numeric variable encodes the latitude of the site where a given assemblage was excavated, in decimal degrees and using WGS84. In order to conceal the exact location of archaeological sites the values have been rounded to two decimal places. Precise coordinate information is available upon request. 9. SiteLongitude: This interval scale, numeric variable encodes the longitude of the site where a given assemblage was excavated, in decimal degrees and using WGS84. In order to conceal the exact location of archaeological sites the values have been rounded to two decimal places. Precise coordinate information is available upon request. 10. AbsoluteElevation: This ratio scale, numeric variable encodes the absolute elevation of the site (at the top of the sediments) above (current) sea level. It should be noted that in some cases the values recorded by GPS and confirmed with topographic maps differ from those available in previous publications. When available, values recorded by GPS are given.  60  11. ExcavatedVolume: This ratio scale, numeric variable provides information on the volume of sediments (in cubic meters) excavated from each level, as estimated by Riel-Salvatore et al. (2008). 12. Climate: This ordinal scale, numeric variable provides an estimate of the general climatic conditions that prevailed during the deposition of a given level. A value of 1 indicates very harsh, dry and cold climates, a value of 2 indicates milder wet and cold climatic regimes, and a value of 3 denotes temperate climates. Data was compiled from the literature listed in the PrimaryRef variable, more specifically from Cârciumaru’s works, in which the climatic conditions that prevailed during the deposition of the various levels are discussed. It should be noted that the process of synthesizing data for this variable was difficult and error prone. Values were derived on the basis of a detailed examination of the stratigraphic and palynological profiles from the different sites and of the available descriptions. A preliminary examination of the data revealed that the information could best be summarized by using the three categories identified above. A re-examination of the data then allowed for each assemblage to be classified in one of the aforementioned categories. The classification was later confirmed by a final review of the available information. 13. ClimateNotes: This nominal scale, string variable provides some additional information regarding the classification of the assemblages in one of the three categories encoded by the Climate variable. 14. RawMatType: This ordinal scale, numeric variable provides information on the type of raw materials present in a given assemblage. A value of 1 (“Fine”) denotes a high incidence (>= 80%) of fine-grained raw materials, a value of 3 (“Coarse”) denotes the low incidence (<=20%) of finegrained raw materials, and a value of 2 (“Mixed”) identifies assemblages where the incidence of fine-grained materials is greater than 20% but lower than 80%. Most values were assigned based on the proportion of total lithics (TotalLithics) listed in the ChertEtc variable. There were two exceptions to this rule, however: a. For the site of Peştera Cioarei the incidence of fine- and coarse-grained raw materials is listed in the literature in addition to the incidence of different rock types. Although the criteria used to classify the materials as “fine-grained” are not clear, the distinctions made by archaeologists who examined the lithic materials in detail were considered to be authoritative. Consequently, several assemblages from Peştera Cioarei were assigned a value of 2 (“Mixed”) despite the absence of raw 61  materials included in the ChertEtc variable, as the incidence of materials noted in the literature to be “fine-grained” was greater than 20% of the recovered lithics. b. In cases where the precise number of lithics made of particular raw material types is not given in the literature, values of 1 (“Fine”) or 3 (“Coarse”) were assigned if, based on the available descriptions, it could be established with reasonable certainty that the incidence of fine-grained raw materials was either very high (>80%) or very low (<20%) respectively. 15. ChertEtc: This ratio scale, numeric variable lists the number of lithics made of rock types with a micro-/crypto-crystalline or amorphous structure and composed mostly of silica (e.g. chert, jasper, opal). The values for this variable were extracted from the literature listed in the PrimaryRef variable. 16. Quartz_Quartzite: This ratio scale, numeric variable lists the number of lithics made of quartzite or what is referred to in the Romanian literature as “quartz”. The values for this variable were extracted from the literature listed in the PrimaryRef variable. 17. Sandstone: This ratio scale, numeric variable lists the number of lithics made of sandstone (most often quartzitic or siliceous sandstone). The values for this variable were extracted from the literature listed in the PrimaryRef variable. 18. OtherMat: This ratio scale, numeric variable lists the number of lithics made of rock types that cannot be classified according to one of the other three raw material variables. These include, for example, basalt, limestone, diorite, andesite, and rhyolite. The values for this variable were extracted from the literature listed in the PrimaryRef variable. 19. TotalLithics: This ratio scale, numeric variable lists the sum total of lithics in a given assemblage. This includes shatter, debitage, cores, formal tools, and other stone implements used in lithic manufacture (e.g. hammerstones). Note that in some cases, such as Level E from Peştera Cioarei, this value is smaller than the sum of the different raw material types (155 versus 167). This is because of inconsistencies in the reviewed literature – in the specific case of Peştera Cioarei, data on raw materials was extracted from Cârciumaru et al. 2000 while information on the number of cores, shatter, and so on was extracted from Doboş 2010. 20. Shatter: This ratio scale, numeric variable lists the total number of shatter or manufacture byproducts listed in the literature which cannot be classified as debitage, cores, tools, or “Other” (see below). 21. Other: This ratio scale, numeric variable lists the total number of hammerstones and whole or broken cobbles/pebbles in a given assemblage, as listed in the reviewed literature. 62  22. Cores: This ratio scale, numeric variable encodes the total number of cores listed in the reviewed literature for a given assemblage. 23. OtherTools: This ratio scale, numeric variable encodes the number of tools listed in the reviewed literature but not included in one of the original numbered Bordesian types (1-63 – see Debénath and Dibble 1994). These tools include bifacials, which are not always listed by individual types and are therefore considered here together in a single category. 24. Debitage: This ratio scale, numeric variable encodes the number of unretouched, non-Levallois flakes and blades which are not included in types 5 and 38. 25. TotalPRB: This ratio scale, numeric variable encodes the number of lithics for which platform characteristics are listed in the literature (the PrimaryRef variable) or which are identified in the reviewed publications as proximal or complete flakes. It should be noted that not all lithics included in this count have intact platforms, and that in some cases this is a minimum estimate derived from lengthy textual descriptions. 26. TotalNumberedTypeTools: This ratio scale, numeric variable lists the total number of tools that are classifiable in Bordes’ types 1-63 (see Debénath and Dibble 1994), as identified in the reviewed literature. 27. IntentionalRetPieces: This ratio scale, numeric variable lists the number of pieces identified in the reviewed literature as intentionally retouched. Note that this number does not necessarily equate with the aux_RevEssentialNumber; the criteria used to classify tools in this variable are not explicitly stated in the literature. 28. LPlainPlat: This ratio scale, numeric variable encodes the number of Levallois flakes or blades (both tools and debitage) identified in the literature as having plain (single facet) platforms. Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 29. LFacetedPlat: This ratio scale, numeric variable encodes the number of Levallois flakes or blades (both tools and debitage) identified in the literature as having faceted (>= 2 facets) platforms. Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 30. LCorticalPlat: This ratio scale, numeric variable encodes the number of Levallois flakes or blades (both tools and debitage) identified in the literature as having cortical platforms. Note that platform characteristics are not always available for all flakes known to be complete or 63  proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 31. LOtherPlat: This ratio scale, numeric variable encodes the number of Levallois flakes or blades (both tools and debitage) with platforms that cannot be classified as plain, faceted, or cortical (e.g. punctiform, broken). Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 32. NLPlainPlat: This ratio scale, numeric variable encodes the number of non-Levallois flakes or blades (both tools and debitage) identified in the literature as having plain (single facet) platforms. Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 33. NLFacetedPlat: This ratio scale, numeric variable encodes the number of non-Levallois flakes or blades (both tools and debitage) identified in the literature as having faceted (>= 2 facets) platforms. Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 34. NLCorticalPlat: This ratio scale, numeric variable encodes the number of non-Levallois flakes or blades (both tools and debitage) identified in the literature as having cortical platforms. Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 35. NLOtherPlat: This ratio scale, numeric variable encodes the number of non-Levallois flakes or blades (both tools and debitage) with platforms that cannot be classified as plain, faceted, or cortical (e.g. punctiform, broken). Note that platform characteristics are not always available for all flakes known to be complete or proximal, so for assemblages where the sum of the values of the platform variables is less than the value of the TotalPRB variable, this number should be treated as a minimum estimate. 36-98. BType1-63: These ratio scale, numeric variables list the number of lithics included in the  64  respective numbered Bordesian types (see Debénath and Dibble 1994 for a good review of Bordesian systematics) according to the reviewed publications. Auxiliary variables: 99. aux_BordesToolDiversity: This auxiliary, ratio scale numeric variable provides a count of numbered tool types (BType1-Btype64) which are represented in a given assemblage by at least one tool; it provides a basic measure of tool diversity, but is very sensitive to overall assemblage size. 100. aux_TotalToolDiversity: This auxiliary, ratio scale numeric variable provides a count of tool types which includes the OtherTools variable. As with aux_BordesToolDiversity, this variable provides a basic measure of tool diversity. 101. aux_TotalTools: This auxiliary, ratio scale numeric variable provides a total tool count for a given assemblage; it represents a sum of values from the OtherTools and TotalNumberedTypeTools variables. 102. aux_PercentageTools: This auxiliary, ratio scale numeric variable lists the percentage of the total lithics represented by the sum of values from the TotalNumberedTypeTools and OtherTools variables. 103. aux_SimpsonIndexofDiversity: This auxiliary, ratio scale numeric variable encodes the Simpson index of diversity (D) for a given assemblage. Higher values of D, which ranges between 0 and 1, indicate a lower probability that two randomly selected tools belong to the same tool type or, in other words, more diverse and evenly distributed assemblages. D is calculated according to the formula proposed by Pielou (1969; qtd. in Peet 1974: 291):  ∑  where S is the number of distinct tool types identified in a given assemblage (aux_TotalToolDiversity), Ni is the number of tools classified in the ith tool type, and N is the total number of tools that have been identified. 104. aux_EssentialCount: This ratio scale, numeric variable lists the total number of tools classifiable in types 1-63 (TotalTools), minus those tools classified in types 1, 2, 3, 45, 46, 47, 48, 49, and 50 (see Debénath and Dibble 1994). 105. aux_RevEssentialCount: This ratio scale, numeric variable encodes a revised essential tool count (aux_EssentialCount), which further excludes technologically defined (unretouched) types 65  5 and 38 (see Debénath and Dibble 1994). This variable is expected to provide the best estimate of intentionally retouched lithics. 106. aux_RetouchtoPRB: This auxiliary, ratio scale numeric variable encodes the ratio of retouched lithics (aux_RevEssentialNumber) to the minimum number of flakes (TotalPRB) present in an assemblage. 107. aux_PercentageRetouch: This auxiliary, ratio scale numeric variable encodes the percent of the total lithics (TotalLithics) that are retouched (aux_RevEssentialNumber). 108. aux_LithicDensity: This auxiliary, ratio scale numeric variable encodes the number of lithics per cubic meter of sediment. The values were calculated by dividing the sum total of lithics (TotalLithics) by the volume of sediments that were excavated in each level (ExcavatedVolume). 109. aux_PercChertEtc: This auxiliary, ratio scale numeric variable encodes the percentage of the total number of lithics recovered from a given level (TotalLithics) that is listed in the ChertEtc variable. 110. aux_BroadRawMatType: This auxiliary, ordinal scale numeric variable encodes the type of raw materials in two categories, namely “Coarse” and “Fine”. A value of 1 indicates assemblages where the RawMatType variable has a value of either 2 or 3, while a value of 2 indicates assemblages where the RawMatType variable has a value of 1. 111. aux_TechDefTools: This ratio scale, numeric variable lists the total number of technologically defined (non-retouched) tools classified in types 1, 2, 3, 5, and 38 (see Debénath and Dibble 1994).  66  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0073841/manifest

Comment

Related Items