Skeletal Variation and Adaptation in Europeans: Upper Paleolithic to the Twentieth Century

A comprehensive analysis of changes in body form and skeletal robusticity from the Terminal Pleistocene through the Holocene, leading to the modern European human phenotype.

Skeletal Variation and Adaptation in Europeans: Upper Paleolithic to the Twentieth Century brings together for the first time the results of an unprecedented large-scale investigation of European skeletal remains. The study was conducted over ten years by an international research team, and includes more than 2,000 skeletons spanning most of the European continent over the past 30,000 years, from the Early Upper Paleolithic to the 20th century. This time span includes environmental transitions from foraging to food production, small-scale to large-scale urban settlements, increasing social stratification and mechanization of labor, and climatic changes.  Alterations in body form and behavior in response to these transitions are reconstructed through osteometric and biomechanical analyses.

Divided into four sections, the book includes an introduction to the project and comprehensive descriptions of the methods used; general continent-wide syntheses of major trends in body size, shape, and skeletal robusticity; detailed regional analyses; and a summary of results. It also offers a full data set on an external website.

• Brings together data from an unprecedented large-scale study of human skeletal and anatomical variations
• Includes appendix of specific information from each research site
• Synthesizes data from spatial, temporal, regional, and geographical perspectives

Skeletal Variation and Adaptation in Europeans will be a valuable resource for bioarchaeologists, palaeoanthropologists, forensic anthropologists, medical historians, and archaeologists at both the graduate and post-graduate level.

• Language: English
• Publisher: Wiley
• Release date: Dec 27, 2017
• ISBN: 9781118628027

Book preview

Preface

Human body form is subject to a complex array of environmental influences, both natural and cultural in origin, acting throughout the lifespan. One way to approach this issue is through comparative studies of different populations exposed to varying environments. Adding temporal depth to such comparisons allows one to assess how populations respond through time to changing environmental conditions. In anthropology, this usually involves the use of skeletal material from historical, archaeological, or paleontological contexts. Many such studies have concentrated on fairly broad characteristics based on standard skeletal metrics, or have focused on specific areas of inquiry such as paleopathological indicators. More can be learned from skeletal remains, and the people represented by them, by incorporating other methods of analysis, including those based on engineering principles. The information obtained can shed new light on both the behavior and biology of past populations, and provide context for understanding modern human variation. This is what we have attempted to do here, for a large, representative sampling of European populations spanning the past 30,000 years and several major environmental transitions.

The investigations reported in this volume are the result of a collaboration that began in 2007 between Christopher Ruff, Brigitte Holt, Markku Niskanen, Vladimir Sládek, and Margit Berner. Subsequent data collection was primarily funded by the National Science Foundation (BCS‐0642297 and BCS‐0642710), with additional support from the Grant Agency of the Czech Republic (206/09/0589) and the Academy of Finland and Finnish Cultural Foundation. A symposium reporting some preliminary results was held at the annual meeting of the American Association of Physical Anthropologists in Portland, Oregon, in 2012.

The project expanded to include many co‐authors from a number of different countries, who are represented among the chapters of the present volume. Many other people also helped with various aspects of the project. Those who assisted in collecting or processing of data include: Trang Diem Vu, Sarah Reedy, Quan Tran, Andrew Merriweather, Juho‐Antti Junno, Anna‐Kaisa Salmi, Tiina Väre, Rosa Vilkama, Jaroslav Roman, and Petra Spevackova. For access to skeletal collections and otherwise facilitating data acquisition, we thank Andrew Chamberlain, Rob Kruszynski, Jay Stock, Mercedes Okumura, Jane Ellis‐Schön, Jacqueline McKinley, Lisa Webb, Jillian Greenaway, Alison Brookes, Jo Buckberry, Chris Knüsel, Horst Bruchhaus, Ronny Bindl, Hugo Cardoso, Sylvia Jiménez‐Brobeil, Maria Dolores Garralda, Michèle Morgan, Clive Bonsall, Adina Boroneant, Alexandru Vulpe, Monica Zavattaro, Elsa Pacciani, Fulvia Lo Schiavo, Maria Giovanna Belcastro, Alessandro Riga, Nico Radi, Giorgio Manzi, Maryanne Tafuri, Pascal Murail, Patrice Courtaud, Dominique Castex, Fredérik Léterlé, Emilie Thomas, Aurore Schmitt, Aurore Lambert, Sandy Parmentier, Alessandro Canci, Gino Fornaciari, Davide Caramella, Jan Storå, Anna Kjellström, Petra Molnar, Niels Lynnerup, Pia Bennike, Leena Drenzel, Torbjörn Ahlström, Per Karsten, Bernd Gerlach, Lars Larsson, Petr Veleminsky, Maria Teschler‐Nicola, and Anna Pankowska. We also thank Erik Trinkaus, Steven Churchill, and Trent Holliday for generously making available data for Upper Paleolithic and Mesolithic specimens included in the study. Erin Whittey produced all of the maps included throughout the volume.

Christopher B. Ruff

Baltimore, Maryland

1

Introduction

Christopher B. Ruff

Center for Functional Anatomy and Evolution, Johns Hopkins University School of Medicine, Baltimore, MD, USA

The modern human body is the product of a long evolutionary history stretching back millions of years (Aiello and Dean, 1990). A basically modern body plan is usually considered to have been achieved with Homo erectus about 1.5 million years ago (Walker and Leakey, 1993), although with some significant subsequent modifications and regional variation en route to Homo sapiens (Ruff, 1995; Weaver, 2003; Larson et al., 2007). Early ‘anatomically modern humans’ (EAM), that is, H. s. sapiens, appeared in the late Middle Pleistocene and were the only form to survive until the end of the Pleistocene (Klein, 2009). With the appearance of EAM humans, it is often assumed that …fundamental evolutionary change in body form ceased (ibid.: 615). However, some systematic changes in body size and shape continued through the terminal Pleistocene and early Holocene (Frayer, 1980, 1984; Jacobs, 1985; Ruff et al., 1997; Meiklejohn and Babb, 2011). In addition, changes in skeletal robusticity, or bone strength relative to body size, occurred during this time period, again with regional variation (Holt, 2003; Ruff, 2005; Ruff et al., 2006b; Shackelford, 2007; Marchi et al., 2011).

Studies of such trends within EAM humans, however, have generally been carried out on a relatively limited number of samples and/or within narrow geographic areas or time periods. There has also been a tendency to divide work between academic disciplines, with paleoanthropological studies more focused on the Pleistocene, and bioarchaeological studies on more recent variation within the Holocene. However, as demonstrated in some of the studies cited above, morphological changes in the human skeleton form a continuum across the Pleistocene–Holocene boundary, as populations adapted to various environmental changes – climatic, technological, and ecological – that also bridged this time period. Behavioral changes within the Holocene can also be viewed as extensions of those initiated earlier in the terminal Pleistocene, for example, increasing sedentism. Characterization of long‐term trends extending from the Late Pleistocene through very recent populations should shed light on both the full adaptability of the modern human skeleton, as well as the effects of a number of major transitions occurring over this time range, including from foraging to early food production, the intensification of agriculture, and increasing urbanization, mechanization, and social complexity.

The present volume is an attempt to do this, for one broad but well‐defined geographic region: Europe. In terms of available late Pleistocene and Holocene skeletal material, Europe is very well sampled (e.g., Jaeger et al., 1998; Roberts and Cox, 2003). It also has a rich archaeological and expanding genetic record (Milisauskas, 2002; Brandt et al., 2013), as well as much historic evidence for the latest periods. Thus, there is both abundant primary material as well as considerable context within which to interpret temporal and regional variation in skeletal morphology.

The overall intent of the study was to sample adult skeletal material from as broad a representation of European populations as possible, beginning at about 30,000 years ago through the 20th century. Although ontogenetic analyses would have provided additional potentially quite interesting data, it was felt that this initial sampling should be limited to adults, to ensure feasibility of the study within a reasonable time frame. Also, for the same general reason, the study samples were concentrated primarily in Western, Central, and northern Europe – that is, not including Greece and the southern Balkans or much of Eastern Europe (see Figure 1.1).

Map of Europe illustrating the period of location sites in European populations with circle markers for EUP, EUP, LUP, Mesolithic, Neolithic, Bronze Age, Iron Age/Roman, Early Medieval, and Late Medieval etc.

Figure 1.1 Location of sites included in the study.

The primary focus of the study is behavioral reconstruction within these past populations – that is, to document changes in behavior as reflected in skeletal morphology. There are many possible ways to do this (Larsen, 2015), but our approach here has been through the assessment of strength characteristics of the major limb bones, derived from diaphyseal cross sections. As discussed in more detail in Chapter 3, there is much theoretical and empirical evidence tying variation in long bone diaphyseal structure to variation in mechanical loadings on the limbs, and thus to behavioral use during life (Ruff et al., 2006a; Ruff, 2008). By sampling both the upper and lower limbs, we can assess both locomotor behavior (e.g., mobility) and manipulative behavior (e.g., tool use). A number of studies have applied this approach to European skeletal samples (Holt, 2003; Marchi et al., 2006; Ruff et al., 2006b; Sládek et al., 2006a,b, 2007; Trinkaus and Svoboda, 2006; Marchi, 2008; Sparacello and Marchi, 2008; Marchi et al., 2011; Sparacello et al., 2011; Macintosh et al., 2014). However, as noted earlier, these have all been limited in scope in various ways. The present study is the first to apply this method to a broad temporal and geographic sampling of populations across Late Pleistocene and Holocene Europe.

The other major focus of the study is the reconstruction of body size and shape in these populations. Both body size and shape must be considered when assessing long‐bone strength, because each influences the mechanical loadings on the limbs (Ruff, 2000; Shaw and Stock, 2011; also see Chapter 3). Thus, these factors must be controlled in order to infer behavior from long‐bone structure. Body size and shape are also informative characteristics in themselves, as they are related to other systemic environmental effects such as climate and diet (see Chapter 4). In this regard, this study overlaps with a number of other studies of Holocene populations that examined temporal trends in body size, usually stature (Cohen and Armelagos, 1984; Steckel and Rose, 2002; Roberts and Cox, 2003; Cohen and Crane‐Kramer, 2007). However, the primary emphasis of those studies was health status – that is, using variation in stature to assess variations in health between populations. Although we consider implications of our findings in the context of changes in health and nutrition, we did not assess other skeletal indicators of health and disease in our samples, although we did eliminate individuals showing clear pathologies that may have affected bone structure (see below). How overall body form was assessed from skeletal dimensions is described in Chapter 2.

Another purpose of gathering the present study data was to create a general reference source within which to evaluate other samples, including those of living populations. For example, it has been suggested that archaeological skeletal samples may represent good ‘baselines’ to address clinical issues such as the increasing incidence of osteoporosis in very recent Western populations (Eaton and Nelson, 1991; Lees et al., 1993; Ekenman et al., 1995; Mays, 1999; Brickley, 2002; Sievanen et al., 2007). However, we actually know relatively little about broad patterns of variation in skeletal structure across time and space within the Holocene. The lifestyle and behavior of specific past populations or individual specimens are also more easily interpreted within a general temporal and geographic context (e.g., Ruff et al., 2006b; Marchi et al., 2011). The large database generated in this study should serve as a good starting point for comparisons of this kind.

Finally, several techniques of body size reconstruction and structural analysis were refined or developed specifically for this study. New formulae for reconstructing stature and body mass from skeletal dimensions, developed using the present study data set, have been published (Ruff et al., 2012), and further extensions of these techniques are described in Chapter 2. Methods for both noninvasively extracting and analyzing cross‐sectional diaphyseal properties were also advanced during the course of the study (see Chapter 3). It is hoped that these experiences and new results will help future researchers in designing and carrying out similar investigations.

1.1 Study Sample

Material from a total of 2179 individual skeletons was included in the study. All individuals were adult and possessed enough pelvic or cranial material to sex the individual (see below for sexing and aging methods). The other criteria for inclusion were possession of a relatively well‐preserved femur, tibia, or humerus, and an intact femoral head for body mass estimation. In a few cases without a measurable femoral head, body mass could be estimated using pelvic bi‐iliac breadth and stature, as described in Chapter 2. Some 95% of the individuals in the sample fulfilled these criteria. The relatively small number of individuals without body mass estimates were included in analyses of bilateral asymmetry, stature, or other body proportions not requiring body mass.

Given the emphasis of the study on diaphyseal structural properties, an important consideration for inclusion in the sample was the state of preservation of the cortex of the major long bones. Because of the sensitivity of cross‐sectional properties to small changes in the outer contour of the section (see Chapter 3), no diaphyseal cross sections with significant wear on the periosteal surface were included in the study. This was a major constraint on the selection of specimens. Cross‐sectional data were obtained on one or more long bones of 1955 individuals (90% of the total sample), with the femur the best represented (1830; 84%), followed by the tibia (1652; 76%) and humerus (1578; 72% with at least one humerus). The other 10% of the sample was used in various body size and shape analyses; the majority of these (80%) were from Medieval Scandinavia or Bronze Age Central Europe.

As noted earlier, we did not include bones showing clear signs of pathology that could affect bone structure or mechanical loading, for example, healed fractures, rickets, osteomyelitis or periostitis. However, we did not exclude individuals with arthritis, first, because some degree of arthritis was a near‐ubiquitous feature of older individuals in most of the samples, and thus could be considered a ‘normal’ condition in this age range, and second, because arthritis does not itself directly affect bone structure in the diaphysis. (There may be indirect effects from changes in mechanical loading of the limb; however, again it could be argued that this is a ‘normal’ feature of aging, i.e., a characteristic of the individual and population.)

The geographic distribution of study sample sites is shown in Figure 1.1, subdivided by temporal periods (see below). For regional analyses, the samples were divided into seven groups: 1) British Isles; 2) Scandinavia and Finland; 3) North‐Central Europe (including Germany, Czech Republic, Austria, and Switzerland); 4) France; 5) Italy; 6) Iberian peninsula (Spain and Portugal); and 7) Balkans (Bosnia‐Herzegovina and Romania [Iron Gates region]). Samples from Sardinia and Corsica were assigned to Italy and France, respectively. Sunghir 1, from Russia (Early Upper Paleolithic), was assigned to Scandinavia/Finland, although a good case could be made for grouping it with North‐Central Europe as well. All regions are represented by at least 142 individuals, except for the Balkans, with 71 (divided between only two sites – the Late Medieval Mistihalj and Mesolithic Schela Cladovei). The best represented regions are North‐Central Europe and Scandinavia/Finland.

The distribution of individuals by region within 10 archaeologically/historically defined temporal periods is shown in Table 1.1. Note that dates for the Mesolithic, Neolithic, and Bronze Age overlap slightly due to variable transition times in different regions of Europe (Milisauskas, 2002). Some of the analyses in this volume use slightly modified temporal groupings, for example, separating the Early Neolithic Copper Age from other Neolithic samples, and Iron Age from Roman samples, or combining the two Upper Paleolithic or two most recent samples. Where this is done it is explicitly noted. So‐called Neolithic Foragers from Scandinavia (the Pitted Ware culture; Linderholm, 2011) are also distinguished from other Neolithic cultures for certain analyses. Sample sizes vary temporally, with predictably more individuals in the Holocene time periods. The best‐represented periods are the Neolithic through Late Medieval. All periods include data from several regions, although regional representation is understandably sparse in the earliest periods.

Table 1.1 Study sample sizes by temporal period and region.

a Date of death, calibrated ¹⁴C.

A complete listing of the individual sites included in the study is given in Appendix 1. A data file with all individual measurements and derived variables is given in a on‐line file that can be accessed at: http://www.hopkinsmedicine.org/fae/CBR.html.

1.2 Osteological Measurements

Skeletal elements included in the study are indicated in Figure 1.2. One femur and one tibia were selected from each individual, choosing the best‐preserved side or, in cases where both sides were equally well preserved, a right or left side at random. Bilateral asymmetry in the lower‐limb bones is quite small (Auerbach and Ruff, 2006). Both right and left humeri and radii, when available, were included. Many individuals possessed bones from both upper limbs, allowing direct assessments of bilateral asymmetry. The other bones measured in the study included other elements used to reconstruct anatomical stature (Fully, 1956; Raxter et al., 2006; see Chapter 2) – the talus and calcaneus (either right or left), sacrum, vertebrae, and cranium; and elements used in analyses of body mass and body proportions – the articulated pelvis and both clavicles.

Image described by caption and surrounding text.

Figure 1.2 Skeleton showing elements measured in the study. Red: elements included in biomechanical analyses (note that only the right femur and tibia are highlighted, to indicate that only one side was included in these analyses; however, as explained in the text, either a right or a left side was chosen for measurement). Black: elements only included in stature, body mass, and body proportions analyses (including all presacral vertebrae inferior to C1).

A list of all of the linear dimensions included in the study is given in Table 1.2. All of these are standard dimensions (Martin, 1928) or have previously been defined and illustrated in Ruff (2002) or Raxter et al. (2006). Abbreviations used throughout the volume are also listed. Body size and shape variables derived from these dimensions are defined in Chapter 2, and in individual chapters. Structural variables derived from bone cross sections are defined in Chapter 3.

Table 1.2 Linear osteological measurements.

a M: Martin, R. (1928) Lehrbuch der Anthropologie. Fischer, Jena. Ruff, C.B. (2002) Long bone articular and diaphyseal structure in Old World monkeys and apes, I: Locomotor effects. Am. J. Phys. Anthropol., 119, 305–342. Raxter, M.H., Auerbach, B.M., and Ruff, C.B. (2006) A revision of the Fully technique for estimating statures. Am. J. Phys. Anthropol., 130, 374–384.

Maximum lengths of the femur, tibia, humerus, radius, and clavicle were taken; in addition, femoral bicondylar and tibial lateral condylar lengths were measured for use in anatomical stature reconstruction. The length′ (‘biomechanical length’) dimensions of the femur, tibia, and humerus were taken as distances parallel to the long axis of the diaphysis, as described by Ruff (2002): for the femur, from the average distal projection of the condyles to the superior surface of the femoral neck; for the tibia, from the average proximal projection of the midpoint of the tibial plateaus to the midpoint of the talar articular surface; and for the humerus, from the superior surface of the head to the lateral lip of the trochlea. Bones were first ‘leveled’ and positioned as illustrated by Ruff (2002), using a special osteometric board and small wedges positioned at the relevant landmarks (see Fig. 3 in Ruff and Hayes, 1983). Length′ was used to define locations for taking anteroposterior (A‐P) and mediolateral (M‐L) diaphyseal breadths and cross‐sectional measurements: at 50% of length′ in the femur and tibia, and at 35% of length′ from the distal end of the humerus (to avoid the deltoid tuberosity). Diaphyseal breadths at 50% of radius length were also taken for a subset (about one‐third) of the sample. Cross‐sectional dimensions at these sites were determined from radiographically or computed tomography (CT)‐derived images. Chapter 3 describes these methods and the cross‐sectional parameters in more detail. Other linear measurements of long bones included superoinferior (S‐I) breadths of the femoral and humeral heads, and M‐L articular breadths of the distal femur, proximal tibia, and distal humerus, the last three as defined by Ruff (2002).

Bi‐iliac (maximum M‐L) breadth of the pelvis was taken when available. The two coxal bones and sacrum were articulated and held together by several large rubber bands. No additional material was placed between the bones. Bi‐iliac breadth could be measured or estimated on about 55% of the total sample. About two‐thirds of these were taken on complete pelves; the remainder included either some estimation of missing portions or reconstruction from hemipelves. The means, standard deviations and ranges of values of bi‐iliac breadth for complete and incomplete pelves were almost identical, indicating that no systematic bias was introduced through the estimation process.

All of the other dimensions used for ‘anatomical’ reconstruction of stature were measured as described and illustrated by Raxter et al. (2006). The height of the articulated talus and calcaneus was measured in the physiological position, with space under the anterior calcaneus. Missing talocrural heights (about 20% of the sample) were estimated from femoral and tibial lengths as described in Ruff et al. (2012). With these estimations, about 87% of the total sample had this dimension. Vertebral body heights were taken as the maximum S–I height of the body anterior to the pedicles (or in the case of the first sacral vertebra, the sacral alae). Heights of missing vertebrae were estimated as described by Auerbach (2011) and Ruff et al. (2012). With these estimations, about 55% of the sample had vertebral column (S1 through C2) lengths. In about 40% of these the columns were complete, 20% had single nonadjacent missing vertebrae, and the remainder had multiple adjacent missing vertebrae in the cervical and/or thoracic regions, requiring estimation of total column length using either lumbar + thoracic or lumbar region lengths. Means and distributions of column lengths in the different categories of completeness were very similar (means within 2%). First sacral body height was estimated from combined presacral column height in a small percentage (14%) of the sample.

Basion‐bregma height was measurable in only 795 individuals (36%) of the study sample. In our previous methodological study (Ruff et al., 2012), following Auerbach (2011), we did not attempt to estimate missing cranial heights. However, for the present study we developed and tested a new method for doing this from total skeletal height inferior to the cranium, which was then applied to 527 additional individuals, increasing the sample with skeletal heights to 1322 individuals, or 61% of the sample. The technique is described in more detail in Chapter 2. Missing long bone lengths were not estimated for any individuals, because this presupposes specific linear length proportions, which vary between populations and were a subject of investigation here.

All data for Neolithic through Very Recent samples were collected by the contributors to this volume. Some Mesolithic data and much of the Upper Paleolithic data used in the study were made available from other colleagues, in particular Erik Trinkaus, Steven Churchill, and Trent Holliday, or were obtained from the literature (Matiegka, 1938). Details are given in the on‐line data set.

1.3 Other Variables

The primary features used in assigning sex were pelvic, including ischio/pubic length proportion, shape of the sciatic notch, subpubic features, and the position of the auricular surface relative to the sciatic notch (‘composite arch’) (Phenice, 1969; Buikstra and Ubelaker, 1994; Brůžek, 2002; Walker, 2005). Cranial features such as development of supraorbital ridges, mastoid processes, and other characteristics (Buikstra and Ubelaker, 1994) were used as secondary indicators when diagnostic pelvic regions were not available. Because these features also show populational variation (e.g., Garvin and Ruff, 2012), they were always first assessed in combination with pelvic morphology, when available, within the same population, and then applied to individuals without pelvic indicators. The sample as a whole was slightly male‐biased (56%).

Adult age status was defined as epiphyseal fusion of the long bones, except that almost complete fusion of the proximal humerus was allowed. Beyond this, the main purpose of estimating age was to factor that variable into the calculation of anatomical stature (see Chapter 2). Individuals were placed into age ranges based on a number of indicators, including fusion of epiphyses and pseudoepiphyses of the coxal bones, vertebrae, and medial clavicle (Krogman, 1962; Buikstra and Ubelaker, 1994); pubic symphyseal and auricular surface changes (Brooks and Suchey, 1990; Buckberry and Chamberlain, 2002; Falys et al., 2006); and dental wear (Brothwell, 1972), the latter ‘calibrated’ within populations using skeletal indicators. Depending on the age and number of indicators present, age ranges could be as narrow as a year or two (e.g., 20–21 years) or as wide as a decade, or for the oldest individuals, simply 50+ or 60+ years. Age range midpoints were calculated (using 55 and 65 years for 50+ and 60+ categories), and used in the stature estimation procedure. Almost all of the Very Recent (≥1900 AD) individuals had known ages at death, along with 63 of the Early Modern (1600–1850 AD) individuals. The mean age for the entire age‐able sample was 38 years, with mean ages for all temporal periods varying within ±5 years from this. About 5% of the sample could not be assigned even a broad age because of a lack of diagnostic elements (most were from the Upper Paleolithic or Mesolithic); these were assigned the mean age of 38 years.

In addition to temporal period and region, sites were further categorized in two other ways: by local terrain, and as either rural or urban. Both of these factors were predicted to have potential effects on body size and mechanical loading of the skeleton. As described in detail in Chapter 5, terrain was subdivided into three categories ranging from ‘flat’ to ‘hilly’ to ‘mountainous,’ based on topographic relief within a radius of 10 km from the site. ‘Urban’ sites were those characterized by large agglomerated settlements with evidence for economic specialization (i.e., towns or cities), while ‘rural’ sites did not show these characteristics. A site was considered ‘urban’ even if some of its occupants engaged in non‐urban occupations, for example they walked out to fields. It is recognized that these are very broad categories with much overlap, and that to some extent the definitions of what constitutes each is relative to the time periods considered; for example, an Iron Age city may not be equivalent to a 19th or 20th century city. However, as a first approximation this categorization was still felt to be useful, particularly for making comparisons within similar temporal ranges. The distribution of rural and urban sites by time period is shown in Table 1.3. The first urban sites in our sample appear in the later Iron Age/Roman period (the Romano‐British site of Poundbury and two Roman sites in Italy); there is only one Very Recent rural site (Sassari in Sardinia). More details can be found in the on‐line data set.

Table 1.3 Rural and urban samples.

1.4 Organization of the Book

This book is organized into three major sections. The first section presents the methodologies employed in the rest of the studies (an exception is the terrain variable, which is described in Chapter 5). As noted earlier, the development of new or modified methodologies for both body size/shape and bone structural analyses was one of the major outcomes of the study. Because body size and shape are important in mechanical analyses, that topic is presented first in Chapter 2, followed by the techniques used to derive and statistically evaluate diaphyseal structural properties in Chapter 3.

The second section of the book includes four chapters on general pan‐European variation in body size and shape (Chapter 4), long‐bone robusticity (Chapter 5), sexual dimorphism in body size/shape and robusticity (Chapter 6), and upper‐limb bone bilateral asymmetry (Chapter 7). Temporal variation across Europe as a whole is the main focus of these chapters, although the effects of terrain and rural/urban distinctions are noted when appropriate, and some cross‐region comparisons are also carried out.

The third section considers changes in body size/shape and bone structural properties within six specific regions (see Table 1.1), including more local archeological and historical context. Comparisons of region‐specific trends with those for Europe as a whole are also carried out. Most of these chapters include some anthropometric data for recent living populations from the regions, to allow further assessment of very recent secular changes in body size. Methods summaries are given in each chapter, although detailed methods are presented in Chapters 1–3. Finally, a conclusions chapter draws together findings from earlier in the book and highlights some of the major themes identified in both general and regional analyses.

As noted in the Preface this study was the result of collaborations and the joint efforts of a large group of researchers. Authorship of the general chapters in Section 2 is limited to the primary collaborators and other personnel involved in developing specific techniques utilized in those chapters, even though many other people contributed data incorporated in those analyses. Authorship of the regional chapters includes all people who worked on the collection and/or analysis of data for that region.

References

Aiello, L. and Dean, C. (1990) An Introduction to Human Evolutionary Anatomy. Academic Press, London, 596 p.

Auerbach, B.M. (2011) Methods for estimating missing human skeletal element osteometric dimensions employed in the revised fully technique for estimating stature. Am. J. Phys. Anthropol., 145, 67–80.

Auerbach, B.M. and Ruff, C.B. (2006) Limb bone bilateral asymmetry: variability and commonality among modern humans. J. Hum. Evol., 50, 203–218.

Brandt, G., Haak, W., Adler, C.J., et al. (2013) Ancient DNA reveals key stages in the formation of central European mitochondrial genetic diversity. Science, 342, 257–261.

Brickley, M. (2002) An investigation of historical and archaeological evidence for age‐related bone loss and osteoporosis. Int. J. Osteoarch., 12, 364–371.

Brooks, S. and Suchey, J.M. (1990) Skeletal age determination based on the os pubis: a comparison of the Ascádi‐Nemeskéri and Suchey‐Brooks methods. Hum. Evol., 5, 227–238.

Brothwell, D.R. (1972) Digging Up Bones. British Museum of Natural History, London.

Brůžek, J. (2002) A method for visual determination of sex, using the human hip bone. Am. J. Phys. Anthropol., 117, 157–168.

Buckberry, J.L. and Chamberlain, A.T. (2002) Age estimation from the auricular surface of the ilium: a revised method. Am. J. Phys. Anthropol., 119, 231–239.

Buikstra, J.E. and Ubelaker, D.H. (1994) Standards for Data Collection from Human Skeletal Remains (ed. J. Haas), Arkansas Archaeological Survey, Fayetteville, Arkansas.

Cohen, M.N. and Armelagos, G.J. (1984) Paleopathology at the Origins of Agriculture. Academic Press, New York.

Cohen, M.N. and Crane‐Kramer, G.M.M. (eds) (2007) Ancient Health: Skeletal Indicators of Agricultural and Economic Intensification. University Press of Florida, Gainesville.

Eaton, S.B. and Nelson, D.A. (1991) Calcium in evolutionary perspective. Am. J. Clin. Nutr., 54 (Suppl. 1), 281S–287S.

Ekenman, I., Eriksson, S.A., and Lindgren, J.U. (1995) Bone density in medieval skeletons. Calcif. Tissue Int., 56, 355–358.

Falys, C.G., Schutkowski, H., and Weston, D.A. (2006) Auricular surface aging: worse than expected? A test of the revised method on a documented historic skeletal assemblage. Am. J. Phys. Anthropol., 130, 508–513.

Frayer, D.W. (1980) Sexual dimorphism and cultural evolution in the late Pleistocene and Holocene of Europe. J. Hum. Evol., 9, 399–415.

Frayer, D.W. (1984) Biological and cultural change in the European Late Pleistocene and Early Holocene. In: The Origins of Modern Humans: A World Survey of the Fossil Evidence (eds F.H. Smith and F. Spencer), Wiley‐Liss, New York, pp. 211–250.

Fully, G. (1956) Une nouvelle méthode de détermination de la taille. Ann. Med. Legale, 35, 266–273.

Garvin, H.M. and Ruff, C.B. (2012) Sexual dimorphism in skeletal browridge and chin morphologies determined using a new quantitative method. Am. J. Phys. Anthropol., 147, 661–670.

Holt, B.M. (2003) Mobility in Upper Paleolithic and Mesolithic Europe: evidence from the lower limb. Am. J. Phys. Anthropol., 122, 200–215.

Jacobs, K.H. (1985). Evolution in the postcranial skeleton of Late Glacial and early Postglacial European hominids. Z. Morph. Anthropol., 75, 307–326.

Jaeger, U., Bruchhaus, H., Finke, L., Kromeyer‐Hauschild, K., and Zellner, K. (1998) Säkularer trend bei der körperhöhe seit dem Neolithikum. Anthropol. Anz., 56, 117–130.

Klein, R. (2009) The Human Career, 3rd edn, University of Chicago Press. Chicago.

Krogman, W.M. (1962). The Human Skeleton in Forensic Medicine. C.C. Thomas, Springfield, Illinois.

Larsen, C.S. (2015) Bioarchaeology: Interpreting Behavior from the Human Skeleton, 2nd edn. Cambridge University Press, Cambridge.

Larson, S.G., Jungers, W.L., Morwood, M.J., Sutikna, T., Jatmiko, Saptomo, E.W., Due, R.A., and Djubiantono, T. (2007) Homo floresiensis and the evolution of the hominin shoulder. J. Hum. Evol., 53, 718–731.

Lees, B., Molleson, T., Arnett, T.R., and Stevenson, J.C. (1993) Differences in proximal femur bone density over two centuries. Lancet, 341, 673–675.

Linderholm, A. (2011) The genetics of the Neolithic transition: New light on differences between hunter‐gatherers and farmers in southern Sweden. In: Human Bioarchaeology of the Transition to Agriculture (eds R. Pinhasi and J.T. Stock). Wiley‐Blackwell, New York, pp. 385–402.

Macintosh, A.A., Pinhasi, R., and Stock, J.T. (2014) Femoral and tibial cross‐sectional morphology reflects complex change in sex roles, mobility, and division of labor across similar to 6200 years of agriculture in Central Europe. Am. J. Phys. Anthropol., 153, 173–173.

Marchi, D. (2008) Relationships between lower limb cross‐sectional geometry and mobility: the case of a Neolithic sample from Italy. Am. J. Phys. Anthropol., 137, 188–200.

Marchi, D., Sparacello, V., and Shaw, C. (2011) Mobility and lower limb robusticity of a pastoralist Neolithic population from north‐western Italy. In: Human Bioarchaeology of the Transition to Agriculture (eds R. Pinhasi and J.T. Stock). Wiley‐Blackwell, New York, pp. 317–346.

Marchi, D., Sparacello, V.S., Holt, B.M., and Formicola, V. (2006) Biomechanical approach to the reconstruction of activity patterns in Neolithic Western Liguria, Italy. Am. J. Phys. Anthropol., 131, 447–455.

Martin, R. (1928) Lehrbuch der Anthropologie. Fischer, Jena.

Matiegka, J. (1938) Homo Predmostensis: Fosilní Clovek z Predmostí na Morave II. Ostatní Cásti Kostrové. Nákaladem Ceské Akademie Ved e Umení, Prague.

Mays, S.A. (1999) Osteoporosis in earlier human populations. J. Clin. Densitom., 2, 71–78.

Meiklejohn, C. and Babb, J. (2011) Long bone length, stature and time in the European Late Pleistocene and Early Holocene. In: Human Bioarchaeology of the Transition to Agriculture. (eds R. Pinhasi and J.T. Stock). Wiley‐Blackwell, Chichester, pp. 153–175.

Milisauskas, S. (ed.) (2002) European Prehistory: A Survey. Kluwer Academic/Plenum, New York.

Phenice, T.W. (1969) A newly developed visual method of sexing the os pubis. Am. J. Phys. Anthropol., 30, 297–301.

Raxter, M.H., Auerbach, B.M., and Ruff, C.B. (2006) A revision of the Fully technique for estimating statures. Am. J. Phys. Anthropol., 130, 374–384.

Roberts, C. and Cox, M. (2003) Health and Disease in Britain: From Prehistory to the Present Day. Sutton Publishing, Stroud, UK.

Ruff, C.B. (1995) Biomechanics of the hip and birth in early Homo. Am. J. Phys. Anthropol., 98, 527–574.

Ruff, C.B. (2000) Body size, body shape, and long bone strength in modern humans. J. Hum. Evol., 38, 269–290.

Ruff, C.B. (2002) Long bone articular and diaphyseal structure in Old World monkeys and apes, I: Locomotor effects. Am. J. Phys. Anthropol., 119, 305–342.

Ruff, C.B. (2005) Mechanical determinants of bone form: Insights from skeletal remains. J. Musculoskelet. Neuronal. Interact., 5, 202–212.

Ruff, C.B. (2008) Biomechanical analyses of archaeological human skeletal samples. In:. Biological Anthropology of the Human Skeleton, 2nd edn (eds M.A. Katzenburg and S.R. Saunders). John Wiley & Sons, Inc., New York, pp. 183–206.

Ruff, C.B. and Hayes, W.C. (1983) Cross‐sectional geometry of Pecos Pueblo femora and tibiae – a biomechanical investigation. I. Method and general patterns of variation. Am. J. Phys. Anthropol., 60, 359–381.

Ruff, C.B., Holt, B.H., and Trinkaus, E. (2006a) Who’s afraid of the big bad Wolff? Wolff's Law and bone functional adaptation. Am. J. Phys. Anthropol., 129, 484–498.

Ruff, C.B., Holt, B.M., Niskanen, M., et al. (2012) Stature and body mass estimation from skeletal remains in the European Holocene. Am. J. Phys. Anthropol., 148, 601–617.

Ruff, C.B., Holt, B.M., Sládek, V., et al. (2006b) Body size, body proportions, and mobility in the Tyrolean ‘Iceman’. J. Hum. Evol., 51, 91–101.

Ruff, C.B., Trinkaus, E., and Holliday, T.W. (1997) Body mass and encephalization in Pleistocene Homo. Nature, 387, 173–176.

Shackelford, L.L. (2007) Regional variation in the postcranial robusticity of Late Upper Paleolithic humans. Am. J. Phys. Anthropol., 133, 655–668.

Shaw, C.N. and Stock, J.T. (2011) The influence of body proportions on femoral and tibial midshaft shape in hunter‐gatherers. Am. J. Phys. Anthropol., 144, 22–29.

Sievanen, H., Jozsa, L., Pap, I., et al. (2007) Fragile external phenotype of modern human proximal femur in comparison with medieval bone. J. Bone Miner. Res., 22, 537–543.

Sládek, V., Berner, M., and Sailer, R. (2006a) Mobility in Central European Late Eneolithic and Early Bronze Age: Femoral cross‐sectional geometry. Am. J. Phys. Anthropol., 130, 320–332.

Sládek, V., Berner, M., and Sailer, R. (2006b) Mobility in Central European Late Eneolithic and Early Bronze Age: Tibial cross‐sectional geometry. J. Archaeol. Sci., 33, 470–482.

Sládek, V., Berner, M., Sosna, D., and Sailer, R. (2007) Human manipulative behavior in the Central European Late Eneolithic and Early Bronze Age: humeral bilateral asymmetry. Am. J. Phys. Anthropol., 133, 669–681.

Sparacello, V. and Marchi, D. (2008) Mobility and subsistence economy: a diachronic comparison between two groups settled in the same geographical area (Liguria, Italy). Am. J. Phys. Anthropol., 136, 485–495.

Sparacello, V.S., Pearson, O.M., Coppa, A., and Marchi, D. (2011) Changes in skeletal robusticity in an iron age agropastoral group: the Samnites from the Alfedena necropolis (Abruzzo, Central Italy). Am. J. Phys. Anthropol., 144, 119–130.

Steckel, R.H. and Rose, J.C. (eds) (2002) The Backbone of History: Health and Nutrition in the Western Hemisphere. Cambridge University Press, Cambridge.

Trinkaus, E.T. and Svoboda, J. (2006) Early Modern Human Evolution in Central Europe: The People in Dolni Vestonice and Pavlov. Oxford University Press, Oxford.

Walker, A. and Leakey, R. (eds) (1993) The Nariokotome Homo erectus Skeleton. Harvard University Press, Cambridge.

Walker, P.L. (2005) Greater sciatic notch morphology: sex, age, and population differences. Am. J. Phys. Anthropol., 127, 385–391.

Weaver, T.D. (2003) The shape of the Neandertal femur is primarily the consequence of a hyperpolar body form. Proc. Natl Acad. Sci. USA, 100, 6926–6929.

2

Body Size and Shape Reconstruction

Markku Niskanen¹ and Christopher B. Ruff²

¹ Department of Archaeology, University of Oulu, Oulu, Finland

² Center for Functional Anatomy and Evolution, Johns Hopkins University School of Medicine, Baltimore, MD, USA

2.1 Introduction

The reconstruction of body size and shape of past European populations is one of the major focuses of this study. The considerable importance of body size and shape reconstruction of past populations is addressed in Chapters 1 and 4.

When this study commenced it utilized the most recent methods to estimate body mass (Ruff et al., 2005) and stature (Raxter et al., 2006) available at that time. Based on the new data collected during the study, new regression equations to predict stature from long bone lengths and body mass from femoral head breadth were developed and published (Ruff et al., 2012). The development of additional methods for body size and shape estimation has continued. In this chapter, we review, describe, and evaluate previously published and new methods used in reconstructing body size and body shape in this study.

2.2 Body Size and Shape Estimation

Most linear (vertical) and lateral dimensions of the body are largely determined by underlying skeletal dimensions, and can thus be estimated from these dimensions with different levels of precision. Linear and lateral dimensions combined (e.g., stature and trunk breadth) determine the skeletal frame size, which is an important determinant of body mass (e.g., body mass estimation from stature and bi‐iliac breadth; Ruff et al., 2005). Proportional relationships of different linear (e.g., trunk length, limb length) and lateral (e.g., biacromial breadth, bi‐iliac breadth) dimensions represent body shape. For example, stature is often used as a denominator in determining body proportions, such as relative body segment lengths (e.g., relative sitting height) and relative body breadths (e.g., biacromial and bi‐iliac breadths relative to stature).

Stature estimation has a long history. Estimation methods are divided into ‘anatomical’ and ‘mathematical’ methods (Lundy, 1985). The anatomical method was pioneered by Dwight (1894), simplified by Fully (1956), and most recently modified by Raxter et al. (2006) and Niskanen et al. (2013). This method takes into account differences in body proportions because it is based on skeletal components of stature from the skull to the heel. It also provides more accurate stature estimates than the mathematical method. In the most recent versions of this method (Raxter et al., 2006; Niskanen et al., 2013), stature is estimated from the sum of contributing skeletal elements – from the so‐called skeletal height (here abbreviated as SKH) – with a regression equation. In this study, we have used Equation 1 of Raxter et al. (2006), which includes an age term.

The mathematical method refers to a method of estimating stature from a particular skeletal dimension (most often long bone lengths) generally using a sex‐ and population‐specific regression equation. The most often and widely used equations to estimate stature of past and present Europeans are those of Trotter and Gleser (1952, 1958), developed for Euroamericans. These equations may provide reasonably accurate stature estimates for the most recent Europeans, but not necessarily for earlier Europeans (Formicola and Franceschi, 1996; Maijanen and Niskanen, 2006; Vercellotti et al., 2009). Partly due to this reason, population‐specific equations have been developed more recently for both European and non‐European skeletal samples utilizing a ‘hybrid’ approach in which long bone lengths are regressed against stature estimates provided by the anatomical method (e.g., Sciulli et al., 1990; Formicola and Franceschi, 1996; Sciulli and Hetland, 2007; Raxter et al., 2008; Vercellotti et al., 2009; Auerbach and Ruff, 2010; Maijanen and Niskanen, 2010). This ‘hybrid’ approach was also used to develop the equations used in this study (Ruff et al., 2012).

The most precise regression equations to estimate stature from long bone lengths are naturally those that are designed for specific samples. Unfortunately, this requires larger sample sizes than are generally available. A sample size of N > 40 is needed to be confident that distributions for x (e.g. bone length) and y (stature) are reasonably representative of the population, and that the regression line plotted through the data is neither too steep nor shallow (on sample sizes, see Maijanen, 2011). For this reason, Ruff et al.’s (2012) equations provide the most viable option for stature estimation for European skeletal samples for which developing sample‐specific equations is not feasible. Some temporal and regional effects on stature estimation precision are naturally expected when applying long bone regression equations to skeletal samples representing different temporal and regional groups.

Body mass estimation from skeletal dimensions has a much shorter history than stature estimation (Ruff et al., 2012 and references therein). These estimations can also never rival stature estimations from skeletal dimensions in accuracy due to considerable variation in the skeletal size–soft tissue mass relationship.

The two main categories of body mass estimation are the ‘morphometric’ approach and the ‘mechanical’ approach (Auerbach and Ruff, 2004). Morphometric body mass estimation is analogous to anatomical stature estimation. The body mass estimate is based on estimated or even measured body dimensions, that is, stature and bi‐iliac breadth as in the stature/bi‐iliac breadth method (Ruff 1994, 2000; Ruff et al., 1997, 2005). These estimates exhibit little or no size‐related estimation error (M. Niskanen, unpublished), but biacromial shoulder breadth relative to bi‐iliac breadth (Ruff, 2000; Ruff et al., 2005) and relative sitting height (Niinimäki and Niskanen, 2015) affect estimation precision.

This ‘morphometric’ stature/bi‐iliac breadth method has been applied in many studies to estimate body mass of archaeological specimens (Ruff, 1994; Arsuaga et al., 1999; Rosenberg et al., 2006; Ruff et al., 2006; Kurki et al., 2010). We use the most recent version of this method (Ruff et al., 2005) in this study.

Mechanical body mass estimation is analogous to mathematical stature estimation. Body mass is estimated from skeletal dimensions (e.g., joint surface size) that mechanically support body mass. Femoral head breadth is often used because it is so often preserved, and demonstrably correlates positively with body mass (Ruff et al., 2012 and references therein). Four different studies have provided regression equations to estimate body mass of recent human samples from femoral head breadth (Ruff et al., 1991, 2012; McHenry, 1992; Grine et al., 1995). We use sex‐specific equations introduced in Ruff et al. (2012) generated by a ‘hybrid’ approach in which morphometric body mass estimates provided by the stature/bi‐iliac breadth method (Ruff et al., 2005) are regressed against femoral head breadth. We assess if our body mass estimates from femoral head breadth provided by Ruff et al.’s (2012) equation exhibit any directional prediction error and its possible effects on results (on size‐related error; see Auerbach and Ruff, 2004).

Only a few published attempts have been made to estimate body dimensions other than stature from skeletal dimensions. Olivier and Pineau (1957) used a cadaveric sample to generate regression equations to estimate upper arm length from humeral length, forearm length from radial length, thigh length from femoral length, and lower leg length from tibial length. Olivier and Tissier (1975) used the same cadaveric material and generated equations to estimate long bone lengths from upper limb and lower limb segment lengths. Piontek (1979) used measured biacromial breadths and X‐ray images of living subjects to generate regression equations to estimate biacromial shoulder breadth from clavicular lengths. Ruff et al. (1997) introduced a regression equation to estimate living bi‐iliac breadth from skeletal bi‐iliac breadth, based on a few population means of living bi‐iliac breadth and skeletal bi‐iliac breadth.

We are not aware of any published methods to estimate sitting height from skeletal dimensions, but skeletal trunk height represented by the summed dorsal body heights of the thoracic and lumbar vertebrae plus sacral anterior length has been used to represent trunk length (Franciscus and Holliday, 1992; Holliday, 1995, 1997). Crural and brachial indices have been used for decades to represent intra‐limb ratios, and the clavicular length–humeral length ratio has been used to indicate relative shoulder breadth for many decades (Martin, 1928; Martin and Saller, 1957; Trinkaus, 1981; Holliday, 1995, 1997).

In this chapter, all methods used in the study to reconstruct body size and shape are described and their precision evaluated. New, previously unpublished, methods are discussed and tested more thoroughly.

2.3 Materials and Methods

Skeletal samples included in this study are discussed in Chapter 1. More detailed information on regional and temporal samples is provided in regional chapters presented in Section III of this book.

In addition to applying previously published methods to estimate stature (Raxter et al., 2006), body mass (Ruff et al., 2006, 2012), and living bi‐iliac breadth (Ruff et al., 1997), we have estimated sitting height, subischial lower limb length, and biacromial breadth) and their commonly used ratios (e.g., relative sitting height, relative shoulder breadth, etc.) from skeletal dimensions to compare body size and shape of past Europeans with living Europeans. Procedures used to generate these new estimations are discussed in detail.

Linear osteological measurements taken on skeletal specimens are defined in Table 1.2 in Chapter 1, which also gives abbreviations. Dimensions and ratios computed using these measurements and estimated body dimensions are provided in Table 2.1. Some additional dimensions used in comparisons with living populations in Chapters 4 and 12 were also derived and are listed in the table; these are defined further below.

Table 2.1 Variables computed from skeletal dimensions.

* Anthropometric‐equivalent dimension or index.

† Basic linear measurements defined in Chapter 1; Table 1.2.

BBH: basion‐bregma height; FBICL: femoral bicondylar length; TLCL: tibial lateral condylar length; TCH: Talar‐calcaneal height; CML: clavicle maximum length; RML: radius maximum length; HML: humerus maximum length.

2.3.1 Estimation of Missing Elements

Some estimation of missing skeletal dimensions was done to increase sample sizes. In a few cases, particular long bone length dimensions were missing and were estimated from other length dimensions of the same bones, using equations generated from the study sample (Table 2.2). As correlation coefficients and standard error of estimation values presented in this table demonstrate, estimation errors are very small in these missing value estimations.

Table 2.2 Equations to estimate missing femoral and tibial lengths from available lengths.

In the case of missing vertebral heights, the total vertebral column length was estimated from existing vertebral heights by procedures described in Auerbach (2011) and Ruff et al. (2012). In the case of 53 Upper Paleolithic, Mesolithic, and Neolithic specimens, the total vertebral column length was estimated with sex‐specific regression equations from existing posterior midline heights of vertebra, anterior midline heights of vertebra and/or from maximum heights of vertebra anterior to pedicles (see Appendix 2(a)). In the case of two Upper Paleolithic and seven Mesolithic specimens, maximum midline vertebral heights provided by Vincenzo Formicola were converted to corresponding maximum heights anterior to pedicles using mean ratios of these measurements (see Appendix 2(b)). Missing S1 heights were estimated from the total presacral vertebral column height with a pooled‐sex regression equation (S1 height = 0.0458 × presacral height + 10.2) provided in Ruff et al. (2012). In all of the above cases, estimation error is so small that it has little or no effect on results.

Talocrural height (TCH) was estimated from femoral bicondylar and tibial lateral condylar lengths, abbreviated as FBICL and TLCL, respectively, (see Table 1.2) for 172 individuals using sex‐specific equations from Ruff et al. (2012):

This skeletal ankle height has such a small absolute contribution to the total skeletal height and stature that estimation error affects stature estimation by only a few millimeters in the vast majority of cases.

2.3.2 Statistical Procedures

Reduced major axis (RMA) regression is used widely in this study because least squares (LS) equations tend to underestimate stature and body mass at the high end of variation and underestimate them at the low end of variation (Ruff et al., 2012 and references therein). We converted LS slopes to RMA slopes by dividing LS slopes by the correlation coefficient value. After that, we recalculated y‐intercepts by subtracting from the mean of y variable (e.g., anatomical stature) the mean of x variable (e.g., partial skeletal height) multiplied by the RMA slope value (on this procedure, see Hofman, 1988; Hens et al., 1998).

In those cases where the actual living dimension is unknown (e.g., sitting height, subischial lower limb length), correction factors based on anatomical data (e.g., fresh bone length including cartilage versus dry bone length without cartilage) are used instead of regression equations to convert skeletal dimensions to corresponding living dimensions. Estimation precisions of these estimates cannot be evaluated by comparing observed and predicted values of matched individuals (e.g., observed value minus predicted value) because there are no observed values available for these comparisons. Instead, we have compared values estimated from skeletal dimensions with observed anthropometric mean values of the same‐sex samples representing the same population (e.g. the Euroamerican males).

Residual values (observed value minus predicted value) or the percentage prediction error (PPE or %PE), computed as [(observed – predicted)/predicted] × 100 (Smith, 1984), were used in assessing estimation error and its direction. The stature estimated using the anatomical method and the body mass estimated using the morphometric stature/bi‐iliac breadth method represent ‘observed’ stature and body mass values, respectively. These assessments of estimation error were performed to determine possible effects of temporal period and geography, as well as sex and age. Analyses were done separately for both sexes and for pooled‐sex samples, depending on the case. Results are not compared with those obtained by applying previous stature and body mass estimation equations because this was already done in Ruff et al. (2012).

All statistical analyses were performed using SPSS (Version 22). Stature is expressed in centimeters and body mass in kilograms. All dimensions used in stature estimation equations are expressed in centimeters, and also in applying the morphometric stature/bi‐iliac breadth method in body mass estimation. Femoral head breadth is expressed in millimeters in body mass estimation.

2.4 Estimating Body Size and Shape from Skeletal Dimensions

2.4.1 Stature

Stature was estimated anatomically from skeletal height (SKH) using Equation 1 (Stature = 1.009 × SKH − 0.0426 × age + 12.1; r = 0.956; SEE = 2.22; N = 119) of Raxter et al. (2006), based on the Terry Collection reference sample, whenever possible. A partial skeletal height (PSKH) was used if an individual had all other essential skeletal elements of the SKH except the basion‐bregma height. We generated an RMA‐equation (Stature = 1.045 × PSKH + 18.911; r = 0.996; SEE = 0.707; N = 537) by regressing stature estimates provided by Equation 1 of Raxter et al. (2006) against PSKH, so that statures derived from SKH and PSKH would be as directly comparable as possible. (Note that the correlation is extremely high here because stature is based on total skeletal height.) Stature can be estimated with about equal precision from SKH and PSKH in the Terry Collection reference data used in Niskanen et al. (2013), but temporal and geographic differences in basion‐bregma height naturally have some effect in our European data set. This directional bias is assessed in this study.

If neither SKH nor PSKH was available, stature was estimated from lower limb long bone lengths using the region‐ and sex‐specific equations in Ruff et al. (2012) (see Table 2.3 here). The one exception was that statures of all Early Upper Paleolithic individuals, including those from northern Europe, were estimated using equations designed for South Europeans because these individuals have elongated limbs, especially distal segments, relative to stature and there are no obvious latitudinal differences in this small sample (see Chapter 4). Figure 2.1 demonstrates a lack of latitudinal differences in relative lower leg length in the Early Upper Paleolithic sample, some suggestion of a trend in the Late Upper Paleolithic sample (with the exception of one high latitude outlier [Oberkassel 1]), and clear latitudinal differences in the Mesolithic sample.

Table 2.3 Equations to estimate stature from lower limb long bone lengths reproduced from Ruff et al. (2012; their Table 3).

Image described by caption.

Figure 2.1 Relative lower leg length (z‐score) plotted against latitude in Early Upper Paleolithic (crosses), Late Upper Paleolithic (triangles), and Mesolithic (circles) individuals.

In all cases, regardless of sample or period, combined maximum femoral and maximum tibial length (FMAXL + TMAXL) was used whenever possible. For the remaining individuals, stature was estimated from either maximum femoral or maximum tibial length. Long bone regression equations from Ruff et al. (2012) used here are presented in Table 2.3.

Stature was estimated for a total of 2132 individuals (1204 males, 928 females). Of these, anatomical stature (ASTA) was estimated from SKH for 536 individuals (290 males, 246 females) and from PSKH for 527 individuals (317 males, 210 females), providing a total sample of 1063 individuals (607 males, 456 females) whose stature is estimated using at least a partial anatomical method. Stature was estimated from lower limb long bone lengths for a total of 1069 individuals (597 males, 472 females). Combined