INVESTIGATING BIOLOGICAL DISTANCE AND SKELETAL STRESS IN A LATE 
ANTIQUE AND EARLY MEDIEVAL TUSCAN TOWN 

By 

Alexis Cameron Goots 

A DISSERTATION 

Submitted to 
Michigan State University 
in partial fulfillment of the requirements   
for the degree of 

Anthropology – Doctor of Philosophy 

2025 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ABSTRACT 

The Late Antique and Early Medieval eras in the Mediterranean (5th-10th centuries AD) 

were an extremely turbulent time. In the centuries following the fall of the Western Roman Empire 

(c. 476 AD), Italy experienced the Gothic War, the plague of Justinian, the Lombard invasion, the 

rise of the Carolingian dynasty, and the spread of Christianity. The nature of this transitional period 

has garnered considerable debate in archaeological and historical literature, resulting in some 

scholars advocating for a “Dark Age” of isolation, disease, and cultural stagnation, while others have 

suggested these eras were marked by adaptation, interconnectivity, and resilience. Notably, there has 

been relatively little focus on the ways in which bioarchaeology can contribute to this conversation. 

At present, no studies have used a diachronic, intracemetery analysis with paleopathological and 

biological distance (biodistance) approaches to understand the effects of this time period on the 

residents of Tuscany, Italy. 

In order to address these scholarly gaps, this dissertation focuses on the site of Rusellae 

(Tuscany, Italy) and its cemetery, which was in use between the 6th and 12th centuries AD. The 

present study focuses on a sample of 160 adults excavated in a collaboration between the Art and 

Archaeology Museum of the Maremma and the Archaeological Superintendence of Tuscany between 

1987 and 1991. The burials comprise individuals from two archaeologically-defined phases: Phase I 

(6th-7th centuries AD) and Phase II (8th-12th centuries AD). The primary foci of this study are: an 

examination of biological distance within Rusellae and among Rusellae and contemporary sites using 

craniometrics; a craniometric population affinity analysis within Rusellae; an exploration of 

craniometric variation over time; and the frequency of skeletal stress indicators (cribra orbitalia, 

porotic hyperostosis, periosteal reaction, and linear enamel hypoplasias) between cemetery phases 

and population affinities. 

 
 
The results showed that, despite exhibiting an overall greater affinity with contemporary 

European samples than African references, Rusellae was a biologically heterogeneous site 

comprising several craniofacially distinct groups. Further, the presence of several individuals 

consistent with African affinity was preliminarily attested at the site. Notably, when compared with 

individuals showing greater European affinity, the individuals with a greater affinity to African 

reference samples showed no significant difference in frequencies of the skeletal stress markers 

analyzed, with the exception of linear enamel hypoplasia of the maxillary central incisor. In addition, 

there were no statistically significant differences in either skeletal stress indicators or cranial 

measurements between the earlier and later phases of the cemetery. 

Broadly, these results suggest genetic continuity between Rusellae and other European 

populations, but not to the exclusion of some contribution from the African continent. These findings 

are consistent with continued circum-Mediterranean trade and migration, including exchange 

between Tuscany and North Africa, during this period. Additionally, the lack of diachronic change in 

skeletal stress indicators and craniometrics suggests a degree of biological, ecological, 

environmental, and psychosocial stability throughout Late Antiquity and the Early Middle Ages at 

Rusellae. Further, the findings indicate a generally similar experience between affinities, at least with 

respect to the stress indicators analyzed, possibly due to a social structure that prioritized religious 

cohesion over ancestral divisions. Thus, the totality of the evidence supports a view of the 

transitional period between the 6th and 12th centuries, not as a “Dark Age” characterized by collapse, 

but as an era of continuity, connectivity, and resilience for Rusellae. 

 
 
 
 
This dissertation is dedicated to my family. 
I couldn’t have done it without you.

iv 

 
ACKNOWLEDGMENTS 

This dissertation would not have been possible without support and encouragement from 

so many people in my life. First and foremost, my utmost gratitude goes to my advisor, Dr. Todd 

Fenton. Thank you for believing in me from the very beginning and reminding me to believe in 

myself over the years, even when the going got tough. You gave me such gifts with the 

opportunity to study Rusellae, to get involved in your trauma research, and to gain lots of 

casework experience while at MSU. I am endlessly grateful to you for everything you have done 

to get me over the finish line—this accomplishment is as much yours as it is mine. To my 

committee members—Dr. Joe Hefner, Dr. Sandro Sebastiani, Dr. Gabe Wrobel, and Dr. Lindsey 

Jenny—I am indebted to you all for the many hours you spent meeting with me, offering 

wisdom, and revising drafts. I would never have even known where to begin on this project 

without the advice and expertise each of you offered. 

I am also grateful for the colleagues, institutions, and funding sources that supported this 

project. Dr. Mariagrazia Celuzza and Dr. Elena Chirico provided invaluable data and background 

on Rusellae. Thank you to the Art and Archaeology Museum of the Maremma, who graciously 

hosted our research team over many summers. Thank you to Dr. Jen Vollner, who graciously 

provided her dissertation data to supplement my reference samples. I am also grateful for the 

funding I received from the Department of Anthropology and the College of Social Sciences at 

Michigan State University for travel and research. 

To the members and faculty of the MSUFAL over my years there, thank you for being so 

brilliant, wise, and endlessly generous with me. I feel so lucky that we all get to be colleagues for 

years to come. I’m especially thankful to Dr. Rhian Dunn for holding me accountable with 

weekly writing groups over the last few months. To Dr. Mari Isa and Dr. Elena Watson, you are 

v 

 
the best field buddies a girl could ask for. I’ll never forget our Italian adventures, from hiking 

until we passed out, to wine nights, and accidentally pickaxing medieval walls, there was never a 

dull moment! And to Dr. Kelly Kamnikar and Dr. Amber Plemons—I don’t have the words to 

express what your friendship has meant to me over the years. I could not have chosen two better 

humans to go through a PhD program with. 

To my family—your unwavering belief, encouragement, and steadfast support have given 

me the foundation I needed to chase after my (admittedly weird) dream of being a forensic 

anthropologist. Finally, to Sarah, thank you for the years of love, adventures, and belly laughs. 

You’ve pushed me beyond what I thought I was capable of, and your proprietary blend of tough 

love and tireless encouragement are exactly what I needed to make it through the ups and downs 

of the last nine years. 

vi 

 
 
 
 
 
TABLE OF CONTENTS 

CHAPTER ONE: INTRODUCTION ............................................................................................. 1 

CHAPTER TWO: BIOARCHAEOLOGICAL APPROACHES TO SKELETAL STRESS AND 
BIOLOGICAL DISTANCE .......................................................................................................... 11 

CHAPTER THREE: ARCHAEOLOGY AND HISTORY OF LATE ANTIQUE AND EARLY 
MEDIEVAL ITALY ...................................................................................................................... 51 

CHAPTER FOUR: RESEARCH QUESTIONS AND EXPECTATIONS ................................... 79 

CHAPTER FIVE: MATERIALS AND METHODS .................................................................... 87 

CHAPTER SIX: RESULTS ........................................................................................................ 112 

CHAPTER SEVEN: DISCUSSION AND CONCLUSION ...................................................... 161 

REFERENCES ........................................................................................................................... 195 

vii 

 
 
 
 
 
 
 
 
CHAPTER ONE: INTRODUCTION 

The Late Antique and Early Medieval eras (5th-10th centuries AD) are tumultuous periods 

in Mediterranean history. These periods are characterized by human mobility, sociopolitical 

shifts, changing settlement structures, and multicultural interactions. At the beginning of Late 

Antiquity, which roughly coincides with the fall of Rome in AD 476, the Italian peninsula fell 

into an amalgamation of kingdoms under the rule of various Germanic peoples (S. Mitchell, 

2007). The following centuries saw the Gothic War, the plague of Justinian, the Lombard 

invasion of Italy, the rise of the Carolingian dynasty, and the spread of Christianity, all of which 

have been well-examined in a historical and archaeological context and all of which inevitably 

impacted the individuals who lived through these events (Brogiolo, Marazzi, & Giostra, 2017; 

Wickham, 1981). Italy was situated at the core of the Western Roman Empire and was a diverse 

center of contact and trade within the broader Mediterranean during the Classical period; this 

position as a nucleus, however, made Italy all the more vulnerable to the effects of the end of the 

Western Roman Empire. 

This transformative time in Italy has garnered considerable debate as to its nature: was it 

a “Dark Age” wherein trade ceased, disease was rampant, and cultural progress halted? Or, 

perhaps, was this period marked by cultural continuity, reconfiguration of settlement structures, 

and the maintenance of trade networks? Further, what was the nature and extent of mobility and 

migration during this period, and what were its biological and social impacts? Though these 

debates are ongoing, archaeological and historical studies have increasingly revealed that this 

period was resilient, marked by adaptations in subsistence, settlement, trade, and political 

systems, rather than an outright collapse (Riccomi, 2021). 

1 

 
There has been comparatively little focus on the ways in which bioarchaeology can 

contribute to these debates. A small number of studies have analyzed skeletal indicators of stress 

between the Classical and post-Classical periods, showing that a disastrous transition was not felt 

ubiquitously throughout Italy. Instead, studies have shown regional variation in health outcomes 

during this time (Belcastro et al., 2007; Giannecchini & Moggi-Cecchi, 2008; Manzi, Salvadei, 

Vienna, & Passarello, 1999; Riccomi, 2021; Viva, 2020). Additionally, ancient DNA and stable 

isotope analyses of skeletal remains have been used to understand the extent of mobility in Italy 

after the fall of Rome, generally concluding that Late Antiquity and the Early Middle Ages 

represent a period of coexistence between genetically diverse groups, including Lombards, 

Romans, Central Europeans, and even North Africans (Amorim et al., 2018; Antonio et al., 2019; 

Coia et al., 2023; Francisci et al., 2020; Tian et al., 2024; Viva, Lubritto, Cantini, & Fabbri, 

2022). 

Few of these studies focus on Tuscany and still fewer use a diachronic, intra-cemetery 

approach to skeletal stress analyses. No studies to date have undertaken a biological distance 

analysis at any scale to explore biodiversity, and none have integrated craniometric and stress 

data to examine how skeletal stress might vary by population affinity. Consequently, 

perspectives surrounding health and migration in Late Antiquity and the Early Middle Ages, 

especially in Tuscany, are largely drawn from historical and archaeological data and lack a 

biological component. This dissertation aims to fill these gaps by contributing novel data towards 

a better understanding of the human experience of the social, economic, and political shifts of 

Late Antique and Early Medieval Italy through biological distance and skeletal stress analyses at 

the site of Rusellae, in southern Tuscany, Italy. 

2 

 
Rusellae is an ancient city that was occupied from the Etruscan times through the Middle 

Ages. Rusellae was founded during the 9th century BC and an enormous fortification wall was 

constructed between the 8th and 7th centuries BC (Sebastiani, 2017). The Etruscan occupants of 

Rusellae fought the Romans from approximately 302-294 BC, when the city was finally besieged 

and captured for Rome (Grant, 1980). Rusellae seems to have briefly recovered until the 

Etruscans were finally displaced and a Roman colony was established at Rusellae in the early 

first century BC (Barker & Rasmussen, 1998). In the 1st century AD, construction began on a 

forum, amphitheater, temple, and domus (Campana, 2018; Sebastiani, Chirico, Colombini, & 

Cygielman, 2015). 

Between the 4th and 5th centuries AD, coincident with the fall of Rome, much of Tuscany 

experienced a substantial decline in the number of sites and occupants. Rusellae was not immune 

to this regional experience, and portions of the site were abandoned, including the forum and 

most public buildings (Sebastiani, 2017). Despite this general decline, Rusellae remained 

occupied, even while many contemporary sites were abandoned; as one of the few remaining 

inhabited cities in the area, Rusellae was chosen as the center of a diocese in the late 5th century 

AD (Celuzza, Cencetti, & Pacciani, 2002; Sebastiani, 2017). Soon thereafter, a church and 

cemetery complex were constructed on the remains of a Hadrianic-era bathhouse. Additional 

construction of a new perimeter wall and fortification of the amphitheater around this time 

provide further evidence for the resilience of Rusellae (Celuzza et al., 2002; Sebastiani, 2017). 

Importantly, it was during this period of resurgence when the Lombards conquered Rusellae in 

approximately AD 593 (Sebastiani et al., 2015). However, construction and craftsmanship 

maintained their importance at Rusellae; this is attested through continued refurbishment of the 

3 

 
church in the late 8th and early 9th centuries AD as well as the construction of a bell tower in the 

late 11th century AD (Celuzza et al., 2002; Sebastiani, 2017). 

The cemetery at Rusellae was excavated in a collaboration between the Art and 

Archaeology Museum of the Maremma and the Archaeological Superintendence of Tuscany 

between 1987 and 1991. Approximately 170 tombs have been excavated, containing the remains 

of roughly 200 individuals (Celuzza et al., 2023). Excavations revealed that the cemetery was 

used between the 6th and 12th centuries and comprised two general archaeological phases: Phase I 

(6th-7th centuries AD) and Phase II (8th-12th centuries AD). The layout of the cemetery ranges 

from neatly arranged tombs in Phase I to earthen inhumations nearing the abandonment of the 

cemetery. A few graves from Phase I (6th and 7th centuries) contain grave goods, including 

personal adornments and bone combs. One grave contains Lombard-style funerary objects, 

including a knife and a belt dating to the 7th century; these grave inclusions and this date are 

consistent with the Lombard occupation of Rusellae (Celuzza et al., 2002). The following phase 

does not contain grave goods and the burials are arranged more opportunistically with reuse of 

graves occurring frequently. The latest recorded use of the cemetery was in the 12th century AD, 

when the bishopric was transferred to nearby Grosseto in 1138 AD (Celuzza et al., 2002). 

Rusellae is an especially important site for understanding the transition between Classical 

and Post-Classical Italy for several reasons. Rusellae was located in an advantageous position 

near the Tyrrhenian Sea and adjacent to the Via Aurelia, which connected the city to Rome, Pisa, 

and the greater Mediterranean region through networks of trade and exchange (Sebastiani et al., 

2015). Its location on a hilltop was also likely a key to its survival amongst many other declining 

sites. As the Tuscan settlement structure shifted from the Roman-era villa system to the 

Medieval structure, fortified hilltop sites were favored, and Rusellae rose to prominence in the 

4 

 
 
 
Tuscan landscape (Francovich, 2007). There is archaeological and historical evidence of contact 

with the Lombard people from Central Europe whose cultural influence is widely attested in the 

literature but whose influence on biological diversity in Italy is only just beginning to be 

understood (Amorim et al., 2018). In addition, the construction and refurbishment of the church 

and cemetery made Rusellae a religious and civic center in the territory. This would have 

attracted people from throughout the region and Christianity likely formed the basis for social 

cohesion in the city. Studying Rusellae has the potential to contribute to a better understanding of 

regional population history and health outcomes. Specifically, skeletal analysis can contribute 

information about environmental and ecological influences, shifting settlement structures, the 

spread of Christianity, human mobility, and a changing sociopolitical landscape between the 

Classical and post-Classical periods. 

However, limited bioarchaeological analysis of the site has previously been undertaken, 

with the exception of a 2023 publication by the research team from Michigan State University 

and our Italian collaborators. This chapter applied a paleopathology approach to the Rusellae 

sample in order to examine the effects of malaria on the site; results indicated that malarial 

infection was present at Rusellae potentially in up to 66% of the individuals with similar 

frequencies across phases, sexes, and age cohorts. Notably, many of the lesions showed signs of 

healing, and their presence in older individuals indicates a degree of endemicity in the sample, 

wherein malarial infection was not acutely fatal (Celuzza et al., 2023). While the malaria study 

provided important insights into past health at Rusellae, it also highlighted how much was left to 

be explored. In fact, it was in the process of analyzing these remains with a focus on malaria that 

this dissertation came into being. 

5 

 
While recording data on pathologies and the biological profile, one skull stood out from 

the rest (Figure 1.1). When viewed through a forensic anthropology lens, this skull exhibited 

features consistent with African affinity, such as the shape and width of the nasal aperture and 

interorbital region, facial prognathism, and a post-bregmatic depression. The possible presence 

of an African individual at Rusellae was exciting to both our research team and our Italian 

colleagues, and it led us to wonder about the extent of craniometric variation at the site and its 

implications for understanding population dynamics in the past. 

Figure 1.1. Rusellae 82-1103, anterior and lateral views. Photos by Dr. Todd Fenton. 

To explore this and related questions, this dissertation uses a biocultural approach to 

situate biodistance and skeletal stress analyses within the broader historical framework of 

Rusellae. The biocultural approach accounts for the complex and reciprocal relationship between 

biology and culture, and is the foundational theory of bioarchaeological studies (Zuckerman & 

Martin, 2016). Craniometric data and non-specific skeletal stress indicators can provide insight 

6 

 
 
into the population history and health outcomes of Medieval Tuscans, as well as the ways in 

which they adapted to social, political, economic, and environmental changes during a period of 

major transition. The craniometric data is used as a proxy within a biodistance framework to 

understand the degree of genetic variation within Rusellae, as well as how Rusellae relates to 

geographically and temporally proximate samples. These data will contribute to better 

understanding how Rusellae’s nodal position within the Mediterranean trade network might have 

contributed to biological diversity at the site and how the Lombard invasion of Italy may have 

influenced craniometric variation at Rusellae. Additionally, within a biocultural framework, 

skeletal stress indicators including cribra orbitalia, porotic hyperostosis, periosteal reaction, and 

linear enamel hypoplasia can provide insight into the biological impact of Germanic subsistence 

strategies, the influence of collapsing Roman settlement system and consequent coalescence of 

people in a fortified settlement, and how the spread of Christianity may have influenced the 

system of social stratification. Therefore, by examining craniofacial variation and skeletal stress 

at Rusellae, these questions derived from the gaps in current historical and archaeological 

knowledge can be addressed through direct skeletal observation of individuals who lived during 

this time. 

Research Goals 

This study aims to integrate several lines of evidence, including craniometric data, non-

specific skeletal stress indicators, and the historical and regional context of Rusellae to provide a 

more in-depth view of life in Late Antique and Early Medieval Tuscany. The specific research 

goals are as follows: 

•  To understand how craniofacial variation at Rusellae resembles or differs from that of 

geographically and temporally proximate reference samples; 

7 

 
 
•  To understand which variables drive craniometric variation within Rusellae, and whether 

there is heterogeneity or homogeneity in overall craniometric variation at the site; 

•  To ascertain whether some individuals at Rusellae show a closer craniometric affinity to 

African reference samples as compared to European reference samples; 

•  To understand whether differences in affinity correlate with different frequencies of 

skeletal stress indicators; 

•  To examine whether craniofacial variation changes over time at Rusellae; 

•  And to explore diachronic changes in the frequency of skeletal stress indicators. 

An Outline of the Present Study 

In order to achieve these goals, the dissertation is organized into seven chapters. The 

present chapter, Chapter One: Introduction, begins with a brief discussion of the relevant 

historical context of Late Antiquity and Early Medieval Europe. Then, the current archaeological 

and historical debates regarding this time frame are discussed, as well as several studies that 

begin to address these questions. The study site, Rusellae, is introduced and its relevance to the 

major debates and capacity to address gaps in the literature using a biocultural approach are 

discussed. Finally, the research goals are outlined. 

Chapter Two: Bioarchaeological Approaches to Skeletal Stress and Biological 

Distance, provides insight into the utility of bioarchaeology and the biocultural approach for 

addressing broad questions of identity, ethnicity, and health. It then describes how skeletal stress 

indicators provide a framework for examining patterns of physiological disruptions and detail the 

four specific indicators used in this study: linear enamel hypoplasias, cribra orbitalia, porotic 

hyperostosis, and periosteal reactions. Next, the chapter provides an overview of biological 

distance analyses, including current practices, different avenues of research, statistical analyses, 

8 

 
 
 
and methodological advancements and critiques. It concludes with a discussion of population 

affinity and its relevance to the present study. 

Chapter Three: Archaeology and History of Late Antique and Early Medieval Italy 

begins with a brief outline of the major historical events in the Mediterranean between AD 400 

and AD 1200 that are relevant to this dissertation. Then, the chapter introduces Rusellae and its 

cemetery. Next, Rusellae is situated in the context of the Tuscan countryside in order to better 

understand its position in the regional system. Subsequently, the chapter provides an overview of 

the research agendas in modern Tuscan archaeology, followed by a discussion of the major 

archaeological debates surrounding this time period. The chapter closes with a survey of 

Mediterranean bioarchaeology, including studies on migration and skeletal stress. 

Chapter Four: Research Questions and Expectations outlines the four research 

questions guiding this dissertation. Each question is followed by a hypothesis and a justification 

for the expected outcome using historical and archaeological evidence. 

Chapter Five: Materials and Methods presents an overview of the skeletal samples 

included in the study, including the Rusellae cemetery sample and the six reference groups 

included in the biodistance analyses. It then discusses the limitations of bioarchaeological 

samples. Next, this chapter details the data collection methods employed, including those for sex 

estimation, age estimation, cranial measurements, and skeletal stress indicators. Finally, this 

chapter outlines the analytical methods and variables used for each research question. 

Chapter Six: Results presents the results of the study, organized by research question. 

Finally, in Chapter Seven: Discussion and Conclusion, each research question and its key 

findings are reiterated before providing an in-depth explanation for the findings. The discussion 

also provides possible alternative explanations for the results, if necessary. It also discusses the 

9 

 
 
 
 
 
limitations of each research question in turn. Additionally, this chapter presents the broader 

implications of the research and the potential impacts of the study. The discussion includes 

avenues of future research based on the findings and limitations of the present study. The chapter 

concludes with remarks that connect the overarching findings to the archaeological and historical 

debates that inspired this research. 

10 

 
 
 
CHAPTER TWO: BIOARCHAEOLOGICAL APPROACHES TO SKELETAL STRESS 
AND BIOLOGICAL DISTANCE 

Bioarchaeology is the study of skeletal remains within their corresponding historical and 

archaeological context with the goal of interpreting past lifeways and gaining a broader 

understanding of the human condition throughout time and space (Larsen, 2015). Beginning as a 

field primarily focused on osteological description and typology, modern bioarchaeology now 

benefits from a multidisciplinary approach, incorporating theory and methods not only from 

osteology and archaeology, but also from biology, chemistry, epidemiology, and population 

genetics (Baker & Agarwal, 2017; Buikstra, 2006; Buikstra et al., 2022; Zuckerman & Martin, 

2016). Further, advancements in statistical methods have allowed for more accurate 

interpretations of small samples, better handling of missing data, and more rapid assessment of 

complex datasets (Buikstra et al., 2022). These theoretical and methodological foundations allow 

for a wide variety of research programs within bioarchaeology, including studies of health, 

intersectional identities, migration, marginalization, and population history, among many others 

(Baker & Agarwal, 2017; Buikstra et al., 2022; Larsen, 2002). 

A common thread of bioarchaeological inquiry, no matter the topic, is the biocultural 

approach, which accounts for the complex and reciprocal relationship between biology and 

culture with respect to a given human phenomenon (Zuckerman & Martin, 2016). Notably, the 

cultural and historical facets of biological anthropology were not always at the forefront of early 

scholarship in the field; it was not until the 1950s and Washburn’s “new physical anthropology” 

that a paradigm shift occurred towards synthetic, population-level research with a basis in theory 

and hypothesis testing (Zuckerman & Martin, 2016). Modern biocultural scholarship has 

diversified its focus to include questions of social and economic status, identity, embodiment, 

gender, and structural violence, among many other avenues of research. However, these higher-

11 

 
order questions are only answerable by contextualizing the skeletal data with the corresponding 

archaeological and historical information. It is this context that the biocultural approach 

provides. Bioarchaeology, then, is the application of the biocultural approach to questions of past 

populations. 

In this dissertation specifically, the biocultural approach provides an interpretive 

framework for biological distance (biodistance) and skeletal stress analyses. These analyses will 

be used to look at changes in skeletal stressors throughout time as well as population history and 

population affinity at Rusellae. Bioarchaeology and the biocultural approach are relevant to this 

study because these topics are not purely biological in nature. Instead, examining shifting 

skeletal stress over time involves an understanding of how ancient social structures and 

environment influence the experience of nutrition and disease. Further, exploring population 

history involves social constructs such as identity and ethnicity through the ways in which 

concepts of self, otherness, and group membership influence mate choice and migration. In order 

to situate this dissertation within relevant research, this chapter will first outline the overarching 

biocultural concepts of identity, ethnicity, and health. Then, I will discuss the study of skeletal 

indicators of stress and their specific application to this research. Finally, I will provide an 

overview of biological distance analyses, including their application to ancestry estimation in 

forensic anthropology. 

Identity: Ethnicity and Health 

In an effort to critically examine the current status of bioarchaeology, a group of 

international scholars participated in a workshop in early 2020 and established topics of current 

importance to bioarchaeology, as well as identifying several avenues for growth in the field 

(Buikstra et al., 2022). One of the theoretical frameworks identified by the workshop participants 

12 

 
as useful for linking skeletal data with cultural factors is identity, or the embodiment of 

biological attributes such as age, sex, ancestry, and health outcomes and their relationship with 

social constructs such as social age, gender, ethnicity, and social status (Buikstra et al., 2022; 

Gowland, 2017; Knudson & Stojanowski, 2008, 2020; Zuckerman, 2020). In bioarchaeology, 

identity theory refers to the ways in which individuals and groups produce, maintain, and 

manipulate their identities through aspects of biology, material culture, mortuary customs, and 

social organization (Buikstra et al., 2022). The general consensus, within bioarchaeology at least, 

is that identity is flexible, but not infinitely changeable. Identity is seen as being constructed both 

by the self and by others. In this way, identity is both situational and relational (Hu, 2013). 

Gowland elegantly summarizes how identity manifests skeletally, noting that life is an embodied 

experience; interactions with the local environment, diet, migration, activity, experiences of 

gender, and social structure are all undertaken with the physical body, which thereby records 

these experience in the skeletal and dental tissue (2017). An emphasis is also placed on 

understanding the intersections and interactions between various aspects of identity (Buikstra et 

al., 2022; Knudson & Stojanowski, 2020; Torres-Rouff & Knudson, 2017). An additional 

consideration of research involving ancient identities is cultural relativism. That is, scholars 

should avoid extrapolating modern Western beliefs and social constructs to past populations 

(Beck & Quinn, 2022; Buikstra et al., 2022; Quinn & Beck, 2016). Two specific aspects of 

identity will be further explored in this dissertation: ethnicity and health status. A brief 

discussion of each follows. 

Ethnicity 

Ethnicity, as it is currently conceptualized, is a social construct that involves the 

complicated interplay between geographic origin, ancestry, skin color, religion, language and 

13 

 
traditions (Gowland, 2017). Contrary to the culture-historical approach, which defined discrete 

ethnic groups by their material culture, the modern biocultural approach considers the 

relationship between biological factors like genetic ancestry and cultural elements such as grave 

goods, burial patterns, or body modification (Zakrzewski, 2011). Ethnicity is simultaneously 

distinct from and overlapping with shared genetic ancestry; geographic origin or ancestry is a 

component of ethnic identity, but the two do not have a direct correlation. Despite the somewhat 

nebulous nature of ethnicity as a construct, the lived experience of ethnicity has biological 

implications due to the social relationship between ethnic identity and access to resources, social 

roles in society, and proximity (or lack thereof) to power (Gowland, 2017; Gowland & 

Thompson, 2013; Zuckerman, 2020). Therefore, although ethnicity is a cultural construct, it has 

undeniable biological consequences that can be studied in the skeleton (Gowland, 2017; Sofaer, 

2006). Further, migrants and their descendants whose ethnicity differs from local individuals 

may manifest their journeys and subsequent adaptations skeletally, potentially allowing for 

identification of differing ethnic cohorts in ancient contexts (Gregoricka, 2021). 

In bioarchaeology, ethnicity is studied through a combination of biological and material 

factors; genetic ancestry ascertained through ancient DNA, stable isotopes, and biodistance 

analyses are all used to explore population affinity and relatedness in bioarchaeological settings 

(Nikita, 2024). Comparisons of these biological data with archaeological variables such as 

material culture and mortuary practices or other biological variables such as sex, age, or health 

status therefore reflect cultural or social ethnic affiliations that may or may not align with 

biological population affinity. Integrating the two lines of evidence allows bioarchaeologists to 

understand the degree to which biological relatedness or genetic ancestry influences cultural 

practices. Ethnicity has been effectively studied through the comparison of ancient DNA with 

14 

 
 
funerary customs in 6th-century Hungary and Northern Italy (Amorim et al., 2018); with an 

intersectional approach to understand the experiences of enslaved African women in colonial 

Peru (Maass, 2023); and by evaluating craniometric variation in a medieval Islamic cemetery in 

Spain (Zakrzewski, 2011). 

Studying ethnicity in a modern context is fraught with challenges, and making inferences 

about ethnicity in the past is no different. Ethnicity is situated firmly in a particular time and 

place; ethnic identity changes throughout time, which complicates the endeavor of studying 

ethnicity in the past. Care should be taken not to extrapolate modern constructs of ethnicity to 

past societies (Buikstra et al., 2022). Notably, ethnicity is but one component of identity, and the 

lived experience of any ethnic identity is inextricably linked to other aspects of identity, 

including social status, gender, or social age. Further, understanding ethnicity in ancient samples 

is predicated upon the use of strong methods for estimating population affinity or geographic 

origin in the past; these methods (e.g. ancient DNA, stable isotopes, biodistance), although often 

used effectively, come with their own set of issues and assumptions. In addition, material 

markers of ethnic identity may have been mimicked by those with less power in order to socially 

connect with or show allegiance to groups of greater power, which may skew interpretations of 

the construction of ancient ethnicity if biological or biogeochemical factors are not also taken 

into account (Gregoricka, 2021). 

This dissertation uses biodistance analyses to contribute to the conversation of ethnicity 

in Late Antique and Early Medieval Italy. Specifically, cranial measurements are used to 

examine broad relationships between the study site and contemporary reference samples as well 

as craniometric variation within the study site. This variation will then be explored through the 

lens of ethnicity; that is, if heterogeneity is identified in cranial measurements in the Rusellae 

15 

 
 
 
sample and if that heterogeneity shows any geographic patterning, there may exist similarly 

variable ethnicities at Rusellae. However, as ethnicity is not directly correlated with population 

affinity, this variation will be further explored by understanding skeletal stress at Rusellae; the 

skeletal stress markers utilized (e.g. cribra orbitalia, porotic hyperostosis, periostitis, and linear 

enamel hypoplasias) can provide information about diet, health, disease, climate and activity 

throughout the life course. By comparing patterned craniometric variation with cultural variables 

(and their skeletal correlates) such as these, preliminary conclusions surrounding ethnicity at 

Rusellae may be drawn. 

Health 

Now we turn to a discussion of another aspect of identity examined in this dissertation: 

health. In bioarchaeology, health refers to an individual or population’s experience of illness or 

wellness within the context of their specific culture, which can be explored, in part, by analyzing 

skeletal lesions that represent instances of “stress” or disease during life (Larsen, 2018; Reitsema 

& McIlvaine, 2014). Notably, these stress markers do not provide an overall “health index,” as 

they do not directly correlate with all aspects of health such as mental and social well-being; 

instead, they point to periods during which physiological homeostasis was disrupted. Therefore, 

health and stress are related, but not interchangeable (Pilloud & Schwitalla, 2020; Reitsema & 

McIlvaine, 2014; Temple & Goodman, 2014). Health is inextricably linked to identity because 

access to resources and care, and therefore wellness, is historically patterned by intersectional 

identities and marginalization based on age, gender, ethnicity, or social status (DeWitte & 

Kowaleski, 2017; Gowland, 2017; Larsen, 2018; Zuckerman, 2020). Health differences that fall 

along lines of identity, such as sex or ancestry, might consequently be indications of larger social 

structures and systemic identity-based discrimination in both modern and ancient societies (de la 

16 

 
 
Cova, 2011, 2012; Harrod, Thompson, & Martin, 2012; Zuckerman, 2020). Therefore, exploring 

ancient experiences of health can also provide insight into these other aspects of identity, such as 

gender, age, or ethnicity. 

Health is studied in bioarchaeology, in part, through examining skeletal indications of 

stress, disrupted growth, infection or disease, trauma, or degenerative changes (Reitsema & 

McIlvaine, 2014). Most of these features can be identified macroscopically by a trained 

investigator, although differential diagnosis of pathological conditions can often be challenging 

(Ortner, 2012). Recent methodological improvements have allowed for greater specificity when 

understanding ancient health, such as isotopic analyses for exploring ancient diet, nutrition, and 

metabolic stress, as well as analyzing pathogen DNA for more accurate diagnoses of illnesses in 

the past (Klaus, 2014; Larsen, 2018). A biocultural approach to health further links these skeletal 

markers to cultural dimensions such as subsistence, labor, and social status (Larsen, 2018). 

Within the biocultural approach, a political-economic perspective emphasizes power differentials 

and their corollaries such as class and access to resources and medical care within different 

cultural contexts (Zuckerman & Armelagos, 2011). In addition, the intersection of multiple 

forms of identity-based discrimination have been shown to influence health (de la Cova, 2011; 

Harrod et al., 2012). For example, different types of labor can lead to occupational markers in the 

skeleton or greater degenerative changes than expected for an individual of a particular age 

(Harrod et al., 2012). Dietary differences linked to different subsistence strategies or climate 

change may lead to varying degrees of malnutrition, which can produce stunting or linear enamel 

hypoplasias if these nutritional insults occur during childhood or adolescence (Temple & 

Goodman, 2014). Social stratification may lead to unequal resource distribution and unsanitary 

living conditions, potentially causing malnutrition, illness, or evidence of nutrient deficiency in 

17 

 
 
marginalized groups (de la Cova, 2011; Harrod et al., 2012). Psychosocial stress associated with 

marginalization has also been shown to have negative health consequences, and poor health in 

Roman Britain has been hypothesized to result from the increasingly hierarchical social structure 

imposed on its residents (Gowland, 2017). 

Similarly to the study of ethnicity, the study of ancient health comes with a variety of 

challenges and ethical considerations. First, methodological and theoretical issues leave room for 

uncertainty when interpreting past disease; the Osteological Paradox has shown that skeletal 

indicators of stress are not necessarily a direct indicator of illness or wellness in a skeletal sample 

(Buikstra et al., 2022; Wood, Milner, Harpending, & Weiss, 1992). Similarly, “poor” health is 

not an experience exclusive to the marginalized; high status in the Roman Empire has been 

linked to increased caries and rickets due to a sweeter diet and more time spent indoors 

(Gowland, 2017). Issues with differential diagnosis in paleopathology are caused by the fact that 

many illnesses either leave similar lesions to one another or no lesions at all; therefore, a 

skeleton that appears free of lesions may or may not be a “healthy” individual (Wood et al., 

1992). In addition, negative health outcomes can be associated with intergenerational trauma as a 

result of epigenetic factors (Gowland, 2017; Klaus, 2014). Importantly, health is experienced 

through the lens of a given time and place, so care must be taken to avoid extending modern 

assumptions about health and disability to past populations (Buikstra et al., 2022). Finally, health 

is inextricably tied to marginalization, so researchers must be cautious not to retraumatize or 

stigmatize vulnerable communities. 

This dissertation will contribute to the conversation surrounding health at Rusellae by 

examining several non-specific skeletal stress markers: cribra orbitalia, porotic hyperostosis, 

linear enamel hypoplasia, and periosteal reaction. As noted previously, these lesions cannot 

18 

 
 
 
provide a full picture of ancient health since they do not provide information about aspects of 

health such as self-perception or mental wellness. They do, however, provide insight into 

specific experiences of stress, malnutrition, or infection that are relevant to the interests of this 

dissertation: that is, they contribute data to understand how physiological insults, malnutrition, 

and infection may have changed throughout time or varied by population affinity. Within a 

biocultural framework, the osteological data will be integrated with demographic information, 

archaeological evidence from Rusellae and the broader region of Tuscany, and historical 

information to illustrate the potential consequences of structural inequity and transitions of 

power on health in the past. 

Skeletal Stress Indicators 

Non-specific stress indicators are the embodied markers of physiological disruption, 

malnutrition, or infections that are recorded and can be observed in the skeleton (Beatrice, Soler, 

Reineke, & Martínez, 2021; Pilloud & Schwitalla, 2020). These indicators have been identified 

and studied in a wide variety of contexts, ranging from modern cases of undocumented migrants 

(Beatrice et al., 2021) to medieval Italian cemeteries (Belcastro et al., 2007; Viva, 2020) in order 

to reconstruct past health, social structure, environment, and living conditions. As noted 

previously, stress indicators cannot and do not give an overall picture of health. Much like sex is 

a biological component of the socially constructed concept of gender, stress is a biological 

component of the socially constructed concept of health. The bioarchaeological study of health 

was discussed in the previous section; this section will outline the model used to interpret 

skeletal stress, detail challenges with the evaluation of stress markers in bone, and end with an 

overview of the specific stress markers used in this dissertation. 

19 

 
Goodman and colleagues outlined a stress model in order to interpret stress in skeletal 

samples; according to their model, the impact of stress on an individual is a product of their 

physical environment, the cultural system in which they live, and the resistance of the host to the 

stressor (Goodman, Martin, & Armelagos, 1984). Importantly, this model emphasizes that stress 

cannot be understood from a purely physiological perspective; the model is inherently biocultural 

in nature, as it takes into account the effects of sociopolitical processes and psychosocial stress 

on biology, as well as the influence of cultural elements on mitigating stress (Goodman, Thomas, 

& Swedlund, 1988). In cases where stress is not sufficiently mitigated, the stress model suggests 

that the consequent biological stress response can leave a lasting and identifiable impact on an 

individual’s bones and teeth. 

Although stress models provide a valuable framework for examining patterns of 

physiological disruptions, the implementation of these models relies on accurately interpreting 

stress indicators; due to the complex and multifactorial nature of these indicators, this accurate 

interpretation has historically been challenging (Pilloud & Schwitalla, 2020). In addition, debates 

surrounding the relationship between health and stress still feature prominently in 

anthropological literature (Pilloud & Schwitalla, 2020; Reitsema & McIlvaine, 2014; Temple & 

Goodman, 2014). Recent publications have called for a better definition of “stress” within the 

context of bioarchaeology, arguing that the term has lost its meaning and that the complicated or 

unknown etiologies of some of these indicators leave little to be gleaned from their presence 

(Edinborough & Rando, 2020; Hillson, 2014). The prevailing idea in this perspective is that 

little, if anything, is being done to ascertain the true causative agents of these pathologies, and 

that a blanket diagnosis of “stress” is minimally informative (Edinborough & Rando, 2020). 

Similarly, the osteoblastic and/or osteoclastic activity that causes these pathologies may be 

20 

 
initiated by a variety of factors, and it is important to consider that these defects may indicate 

periods of metabolic disruption throughout an individual’s life, instead of simply a snapshot of 

their frailty at death (Pilloud & Schwitalla, 2020). 

Though the concerns raised surrounding stress analysis are legitimate, they do not 

preclude useful analyses of stress indicators, as several solutions and recommendations have 

been presented. For example, using multiple indicators–as this dissertation does–is recommended 

in order to provide the most holistic understanding of an individual’s stress response (Agarwal, 

2016). Certain indicators are related to stress during specific phases of growth and development; 

by analyzing a variety of these factors, a picture of an individual’s life history emerges, as 

opposed to a snapshot of stress during a limited period of time (Larsen, 2015). Further solutions 

to this issue include differentiating between healed and active lesions, as well as framing results 

in the context of the osteological paradox (Pilloud & Schwitalla, 2020). Finally, utilizing a life 

course approach and incorporating embodiment theory to evaluate the influence of intersecting 

identities on stress response at the individual level–in addition to population-level analyses–

facilitates a view of the skeleton as the product of a lived experience (Agarwal, 2012, 2016; 

Knudson & Stojanowski, 2008; Sofaer, 2006). 

This dissertation focuses on four non-specific skeletal stress indicators: linear enamel 

hypoplasias, cribra orbitalia, porotic hyperostosis, and periosteal reactions. These lesions have 

been clinically identified in modern populations, and their presence in the past can provide 

insight into facets of life such as trauma, infection, and disease (Beatrice et al., 2021; O'Donnell, 

Hill, Anderson, & Edgar, 2022; Vercellotti et al., 2014). A discussion of each indicator follows. 

21 

 
 
 
Linear Enamel Hypoplasias 

Linear enamel hypoplasias are linear defects of decreased enamel thickness, running 

roughly parallel to the cemento-enamel junction (Ritzman, Baker, & Schwartz, 2008). They are a 

signature of arrested enamel development attributed to three causes: genetic anomalies, dental 

trauma, or illness, disease, or malnutrition that causes systemic metabolic distress (Larsen, 

2015). However, it is systemic metabolic stress that causes a vast majority of linear enamel 

hypoplasias in both bioarchaeological and modern skeletal populations. Clinical studies have 

identified several causes of metabolic stress that can result in linear enamel hypoplasias, such as 

disease, health issues during the newborn phase, and dietary deficiencies. Malnutrition and 

undernutrition are particularly problematic for the development of healthy enamel; research on 

the dentition of individuals born during famine in the People’s Republic of China shows that 

enamel developed during starvation times was defective in comparison with enamel developed 

before and after the famine (Zhou & Corruccini, 1998). Given the wide range of potential causes 

of linear enamel hypoplasias, they are considered a non-specific indicator of stress (Larsen, 

2015). Importantly, enamel cannot be remodeled, so linear enamel hypoplasias are permanent 

records of developmental insults (M. Lewis & Roberts, 1997). 

Enamel formation follows a relatively consistent pattern across teeth, although the timing 

differs by tooth. Cuspal enamel is first laid down in dome-like layers, beginning with the cusp of 

the tooth and proceeding cervically. When cuspal enamel has completed formation, enamel is 

then deposited on the walls of the crown in a tile-like pattern; this is called the lateral enamel. 

When this pattern is disrupted, a linear enamel hypoplasia is formed. However, the hypoplasia is 

only visible macroscopically in the lateral enamel. Defects in the cuspal enamel can only be 

viewed via histological sections (Ritzman et al., 2008). Tooth enamel begins development in 

22 

 
 
 
utero, first with the deciduous incisors, and completes development once the crowns of the third 

molars have finished forming, at approximately twelve years old (Larsen, 2015). 

Enamel defects do not affect all teeth equally. Instead, they are more commonly observed 

on the anterior dentition–the incisors and canines–as well as the cervical two thirds of the crown 

(Larsen, 2015). This pattern is related to enamel development, so linear enamel hypoplasias are 

considered to reflect developmental stressors, as opposed to those encountered later in life. This 

is due to the fact that the defects are formed while the tooth enamel is undergoing development, 

which varies in age by tooth. Therefore, based on the tooth on which the defect is present, as 

well as its distance from the cemento-enamel junction, it may be possible to roughly ascertain the 

age at which an individual experienced some sort of stress or insult (Larsen, 2015). Some 

researchers have argued, however, that traditional macroscopic methods (e.g. measuring the 

distance between the cemento-enamel junction and the linear enamel hypoplasia), underestimate 

the age of insult, as they do not take into account cuspal enamel formation time or the varying 

rates of enamel deposition by tooth and throughout the dentition (Ritzman et al., 2008). Further, 

these methods falsely assume that the rate of tooth development in a bioarchaeological 

population is the same as in a modern healthy population, from which these methods were 

derived (M. Lewis & Roberts, 1997). 

Cribra orbitalia and Porotic Hyperostosis 

Two notable porosities are among the most commonly observed in bioarchaeological 

material: porosity of the outer table of the cranial vault (porotic hyperostosis) and porosities of 

the superior aspect of the eye orbits (cribra orbitalia).  Both lesions were initially subsumed 

under the umbrella of porotic hyperostosis, with J. Lawrence Angel having coined the term in 

1966. Prior to 2009, both conditions had long been linked to iron deficiency anemia or a 

23 

 
 
response to overwhelming pathogen load in the environment (Stuart-Macadam, 1992). The 

connection seemed a reasonable one; both iron deficiency anemia and these porotic lesions were 

widespread in the global population. This hypothesis led bioarchaeologists to connect the 

presence of porotic hyperostosis and/or cribra orbitalia to conditions causing iron deficiency, 

such as diarrheal disease or intestinal parasites (Walker, Bathurst, Richman, Gjerdrum, & 

Andrushko, 2009). 

However, iron deficiency anemia alone is not a sufficient explanation for these 

commonly observed lesions (Walker et al., 2009). In fact, the marrow expansion characteristic of 

porotic hyperostosis and cribra orbitalia requires such an intense proliferation of red blood cell 

production that iron deficiency anemia alone cannot cause. As a result, hemolytic and 

megaloblastic anemias are more likely causes for porotic hyperostosis, while the etiology of 

cribra orbitalia likely differs, possibly linked to sub-periosteal bleeding associated with 

nutritional deficiencies (Walker et al., 2009). Therefore, it is no longer advisable to associate 

either of these lesions exclusively with iron deficiency anemia. Instead, these porosities can be 

more accurately linked to chronic disease and malnutrition. 

Walker and colleagues propose a constellation of variables that can lead to porotic 

hyperostosis and cribra orbitalia, including inadequate nutrition, poor sanitation, and infectious 

disease (2009). On the other hand, porotic hyperostosis and cribra orbitalia may also be caused 

by infection, injury, or a nutrient deficiency such as scurvy. Macroscopically, etiologies like 

these may produce a similar appearance to anemia-induced porotic hyperostosis, but through an 

activation of the periosteum instead of enlargement of the diploe. In cases such as these, the 

characteristic “hair on end” radiographic appearance is not observed. Given the possibility that 

the same macroscopic appearance can result from a number of different disease processes, it is 

24 

 
recommended that porotic hyperostosis and cribra orbitalia be considered non-specific 

indicators of stress, as opposed to a specific diagnosis (Ortner, 2003). Although some scholars 

have argued that orbital and vault lesions have the same etiology (Stuart-Macadam, 1987, 1989), 

it should not be assumed that both lesions are caused by anemic marrow hypertrophy. Instead, it 

has been argued that cribra orbitalia is more likely to be caused by subperiosteal inflammation 

due to vitamin deficiencies or trauma-related hematomas (Larsen, 2015). 

On a biological level, the marrow expansion observed in cases of porotic hyperostosis is 

a response to the body’s attempt to maintain homeostasis with respect to red blood cell count. 

That is, the body should produce red blood cells at a rate similar to that with which they are 

destroyed. When hemoglobin levels drop, the body begins producing red blood cells more 

quickly. Therefore, in conditions of chronic anemia and low hemoglobin, red blood cells are 

overproduced. In extreme cases, red blood cells are overproduced in the diploe of the cranial 

vault, causing it to expand while the outer table resorbs. This process produces the characteristic 

lesion known as porotic hyperostosis. When viewed radiographically, the diploe undertakes a 

“hair on end” appearance. Notably, the sustained and massive marrow hypertrophy that creates 

porotic hyperostosis is most consistent with hemolytic or megaloblastic anemias; in contrast, iron 

deficiency anemia causes a decrease in red blood cell production, due to the lack of hemoglobin 

in the blood (Walker et al., 2009). 

Osteologically, the bones of the cranial vault undergo a multi-stage process that results in 

visible porosities. First, very fine pitting of the vault occurs in small, circumscribed areas, 

typically on the parietal bones and on the occipital bones, parallel to the lambdoidal suture. 

Further, the trabeculae of the diploe begin to orient perpendicular to the inner and outer tables of 

the vault. Second, the areas of porosity grow larger, and the pits themselves also expand due to a 

25 

 
conglomeration of the initial, smaller pits. The affected regions of the vault also expand to 

include the occipital squama. During this phase, the vault thickens and the traditional “hair on 

end” appearance can be identified; this is most readily observed radiographically. The third 

phase involves a marked thickening of the skull vault, giving way to a sponge-like, porous 

appearance. Remodeling can occur throughout this process, producing rounded edges and an in-

filling of the pores (Ortner, 2003). 

Porotic hyperostosis and cribra orbitalia can be observed in both adults and subadults, 

and can also undergo healing if and when an anemic or inflammatory episode ends. However, it 

has been argued that these conditions are most likely to occur in subadults, as the location of red 

blood cell production shifts from the diploe of the cranial vault and medullary cavities of the 

long bones in subadults, to the thoracic skeleton in adults. Further, most active lesions are 

observed in subadults, while healed lesions are more commonly associated with adult individuals 

(Stuart-Macadam, 1985; Walker et al., 2009). 

Periosteal Reaction 

Periosteal reactions are caused by inflammation of the periosteum, or the connective 

tissue lining surrounding bone. This inflammation stimulates osteoblast activity, causing new 

bone formation on the cortex of the injured or infected bone. This commonly observed condition 

can be caused by a variety of factors, including bacterial infection, traumatic injury, or any other 

disruption of the periosteum. Periosteal lesions appear similar to tree bark, generally with 

demarcated borders and a plaque-like or raised appearance. Similarly to porotic hyperostosis and 

cribra orbitalia, periosteal lesions can appear healed or unhealed; the unhealed lesion tends to 

appear “woven” in texture, with loosely organized new bone. On the contrary, healed lesions 

generally have a more dense, raised, and undulating appearance. Periosteal reactions can appear 

26 

 
 
localized, or can be widespread and involve several skeletal elements, depending on the severity 

of the infection or injury. Periostitis is considered a non-specific indicator of health, as its 

presence cannot be attributed to a particular disease, bacteria, or injury (Larsen, 2015). 

The presence and prevalence of periosteal reactions has been used widely in 

bioarchaeological research. The frequency of periostitis has been compared between males and 

females in order to understand sex-based differences in the non-specific stress indicator among a 

wide variety of groups. Further, this type of analysis has been used in conjunction with other 

variables to attempt to identify status differences between individuals in a sample (Larsen, 2015). 

Methodological Limitations of Skeletal Stress Analyses 

The last three decades of bioarchaeological inquiry have involved extensive introspection 

and have been characterized by the need to grapple with important methodological 

considerations, some seemingly the death knell of paleodemography (Bocquet-Appel & Masset, 

1982) and others simply a stern but cautionary note on careful interpretation of evidence (Wood 

et al., 1992). In 1982, Bocquet-Appel and Masset “bid farewell” to the field of paleodemography 

by elucidating their concerns about age estimation in demographic samples. The authors brought 

to light three issues—age mimicry, secular change, and poor correlation of aging methods to 

skeletal indicators—that necessarily bias the results of a paleodemographic inquiry into a 

cemetery assemblage (Bocquet-Appel & Masset, 1982). 

Publishing an additional critique of bioarchaeology, without the inherent pessimism of 

Bocquet-Appel and Masset, were Wood and colleagues with their seminal paper entitled “The 

Osteological Paradox” (1992). The three fundamental conceptual issues with bioarchaeological 

inference raised by the authors are demographic nonstationarity, selective mortality, and 

heterogeneity in risk of disease and death. Unlike Bocquet-Appel and Masset, Wood and 

27 

 
 
 
colleagues are less eager to call for the abandonment of bioarchaeology and paleodemography. 

Instead, they urge researchers to continue to press forward, but to take these conceptual issues to 

heart when making assessments about skeletal samples (1992).  

Bioarchaeologists and paleodemographers have attempted to address and relieve the 

concerns raised by Bocquet-Appel and Masset and Wood and colleagues, with varying degrees 

of success, in recent years (Wright & Yoder, 2003). Improvements in sex estimation methods, 

particularly using DNA for sex estimation in subadults, as well as the expansion of age-

estimation methods to improve their utility in cases of poor preservation, and transition analysis 

for more accurate age estimation within a statistical framework, have allowed researchers to 

create a better demographic profile of the dead. In an effort to elucidate hidden heterogeneity in 

risk, researchers have used isotopes and biodistance analyses to identify subgroups, such as 

migrants, within a cemetery sample. Additionally, studies of paleodiet have provided information 

about childhood nutritional status that can illustrate hidden frailty in particular individuals. 

Another way to strengthen bioarchaeological conclusions is to consider several lines of 

evidence at once, such as comparing adult stature with paleodiet for an indication of nutrition 

status, as well as evidence of childhood stress such as linear enamel hypoplasias to ascertain 

whether childhood stress affects adult stature. Finally, the recent use of pathogen DNA to better 

connect skeletal indicators of disease to the specific disease process has the potential to allow 

researchers to determine whether an individual was affected by a disease, regardless of whether 

or not any lesions are present in the skeleton. Therefore, we might begin to understand whether 

skeletons with extensive healed lesions may, in fact, be “healthier” than those without lesions at 

all (Wright & Yoder, 2003). 

28 

 
 
Biological Distance Analysis 

Biological distance, in its most basic form, refers to the degree of similarity or 

dissimilarity between groups or populations (Auerbach, 2023; Hefner, Pilloud, Buikstra, & 

Vogelsberg, 2016). Biological distance analyses, as they are employed in bioarchaeology and 

forensic anthropology, have traditionally explored population relatedness based on continuous 

and discrete traits of the bones and teeth within a multivariate statistical framework (Hefner et 

al., 2016). Historically, studies using craniometry were undertaken by typologists intent on 

reifying “racial” differences based on measurements of the skull and evaluating deviations from 

“normal types” (Hefner et al., 2016). These groups were seen as fixed and associated with 

distinct origins (Auerbach, 2023). Today, biodistance studies using skeletal and dental 

measurements and traits focus on characterizing human variation and interpreting that variation 

within a biocultural context using model-free and model-bound population genetic approaches 

(Auerbach, 2023). Improvements in biogeochemistry, DNA analysis, and statistical methods in 

recent decades have expanded the types of analyses anthropologists use to evaluate relatedness in 

past populations, as well as population affinity in modern forensic samples (Hefner et al., 2016). 

This section will first provide a brief discussion of important population genetics 

concepts. Then, it will outline the goals, underlying assumptions, and levels of analysis for 

biodistance studies. Next, it will describe current practices in biodistance research, including 

those using genetic data, biogeochemical methods, statistical methods, cranial and dental metrics 

and morphology, and the relationships between these avenues of analysis. This section will 

conclude with a critical discussion of ancestry and population affinity within forensic 

anthropology and its implications for the present study. 

29 

 
 
 
Population Genetics 

Several population genetics concepts are useful for this dissertation, as they provide a 

potential explanation for the observed results. Population genetics refers to the genetic 

composition of a population and how it changes over time (Relethford, 2012). Broadly, 

population genetics theory is a form of evolutionary theory that can be applied to questions of 

human population history, ancient human migrations, and the impact of geography on mate 

choice. An equilibrium model of evolution, one that assumes no selection, gene flow, genetic 

drift, or mutation, often forms the basis of population genetics studies. When populations deviate 

from this equilibrium, researchers can hypothesize that non-random genetic events have 

occurred. Two of these non-random processes that influence genetic variation—gene flow and 

assortative mating—are important to this dissertation and will be discussed briefly here. 

Gene flow is the exchange of genetic material between populations (Relethford, 2012). 

Gene flow is often the result of migration, either temporary or long-term. It introduces new 

alleles to a population and reduces genetic differentiation between populations. In other words, 

gene flow increases within-population genetic variation and decreases between-population 

variation. The degree of gene flow in humans is expected to be proportionate to the geographic 

distance between populations. This is known as the isolation-by-distance model. Nearby 

populations should exhibit more gene flow than those that are more distant because humans are 

more likely to choose a mate from nearby than from far away. The result of this model is that 

geographically proximate groups should be more genetically similar to one another than they are 

to geographically distant groups (Relethford, 2012). 

A second non-random event that affects genetic variation is assortative mating. This is a 

form of sexual selection wherein individuals choose to mate with others based on similar 

30 

 
 
phenotypic characteristics (Relethford, 2012). Assortative mating is culturally mediated and 

involves the social influence of group membership and phenotypic preferences specific to a time 

and place (Robinson et al., 2017). The selection of a partner phenotypically similar to oneself 

influences overall genetic variation by producing subgroups with less within-group variation and 

more between-group variation. In a sample of individuals with different geographic ancestries all 

living in the same place, assortative mating based on ancestry may produce genetic subsamples 

within the overall population. 

The influence of non-random genetic events such as gene flow or assortative mating will 

influence the degree of variation at Rusellae and can help explain the dynamics that may have 

led to the observed variation. 

Distance Studies 

Distance studies are a form of population genetic analysis that seek to understand 

biological relatedness between pairs of populations in order to explore their population structure 

and history (Relethford, 2016). Distance studies use one of two types of data: genetic 

information, such as nuclear or mitochondrial DNA, or phenotypic data, such as cranial or dental 

metrics or morphology. Phenotypic distance studies, which are commonly used in 

bioarchaeology, are also known as biological distance, or “biodistance” studies (Relethford, 

2016). In contrast to genetic distance, wherein genetic data is explored directly, biodistance 

studies use phenotypic data as a proxy for genetic relationships, following the assumption that 

phenotypic similarity generally aligns with genetic similarity based on theoretical population-

genetic models (Relethford, 2016). Importantly, phenotypic expression is a result of both genetic 

and environmental influences, and is therefore not directly interchangeable with genetic data. 

However, studies have routinely shown that genetic data and skeletal trait expression are 

31 

 
 
correlated highly enough to reveal biological relationships using phenotypic data and that much 

of global human cranial morphology is shaped by selectively neutral evolutionary forces 

(Herrera, Hanihara, & Godde, 2014; Martínez-Abadías et al., 2009; Relethford, 1994, 2001, 

2004; Roseman, 2004; Roseman & Weaver, 2004; von Cramon-Taubadel, 2014). Further, metric 

and nonmetric traits have reliably shown non-zero heritabilities (Carson, 2006a, 2006b; 

Martínez-Abadías et al., 2009; Šešelj, Duren, & Sherwood, 2015). 

Biodistance studies have been employed at varying levels of analysis, including at the 

global, regional, and intra-site levels (Hefner et al., 2016; Paul & Stojanowski, 2017). Global-

level biodistance studies have been used to explore peopling of the Americas through dental 

nonmetric traits (Scott et al., 2023) and microevolutionary processes that shape the human 

craniomandibular complex (von Cramon-Taubadel, 2016). On a regional scale, studies have 

examined migration in the Middle Bronze Age Nile Delta (Maaranen et al., 2021), population 

history in medieval Nubia (Streetman, 2018; Vollner, 2016), and mobility in Roman Phoenicia 

(Mardini, Badawi, Zaven, Gergian, & Nikita, 2023). Intra-site biodistance analyses include 

studies of kinship and cemetery structure, post-marital residence, sample aggregate phenotypic 

variability, temporal microchronology, and age-structured phenotypic variation. These analyses 

treat a site as if it comprises distinct subgroups, as opposed to assuming that the site is a 

homogenous unit (Stojanowski & Schillaci, 2006). Intra-site analyses have explored inter-

individual dental morphology in Neolithic and Iron Age Germany using Mahalanobis distance 

(Rathmann, Lismann, Francken, & Spatzier, 2023) and post-marital residence and biological 

kinship in an Alameda, California shellmound (Broehl, 2023). 

32 

 
 
 
 
Current Practices in Biodistance 

Current practices in biodistance research involve several analytical methods and 

measures of relatedness that merit discussion. These include genetic data, skeletal and dental 

metric and nonmetric analyses, biogeochemical methods, and advances in statistical methods. 

Each of these analytical avenues can be used alone or in conjunction with the other methods to 

illuminate ancient relationships. A discussion of each–as well as their interactions with one 

another–follows. 

Distance Studies in Bioarchaeology: Ancient DNA 

In bioarchaeology, the genetic information used in distance studies is ancient DNA 

(aDNA) derived from skeletal and dental tissue. Its use has revolutionized archaeology by 

providing information about the origin and evolution of ancient pathogens, and it has helped 

clarify methods of domestication of crops and animals. Importantly for this dissertation, ancient 

DNA has also been used to contribute to debates surrounding population structure, migration, 

and group relatedness (T. A. Brown, 2023). By its nature, genetic data provides the most direct 

measure of relatedness, as it does not involve parsing out the additional, complicating layer of 

phenotypic expression used in biodistance studies. Since aDNA was first used to sequence 

genetic data from 150-year-old faunal remains 1980s (Higuchi, Bowman, Freiberger, Ryder, & 

Wilson, 1984), molecular anthropologists and geneticists have refined their methods to allow for 

data extraction from much older specimens, such ancient Egyptian mummies (Hefner et al., 

2016; Pääbo, 1985a, 1985b). Further, advances in DNA amplification, including polymerase 

chain reaction (PCR) and single primer extension (SPEX), have facilitated more accurate 

detection of postmortem DNA alteration or contamination (Brotherton et al., 2007; Hefner et al., 

2016; Mullis & Faloona, 1987), while next-generation DNA sequencing (NGS) has allowed 

33 

 
 
scientists to make complex inferences about population structure using mass quantities of DNA 

data (T. A. Brown, 2023). 

Several types of DNA can be used in genetic distance studies, including mitochondrial 

DNA (mtDNA), which traces the maternal lineage, Y-chromosome DNA, which traces the 

paternal lineage, and autosomal DNA, which is individual-specific. Since nuclear aDNA tends to 

be more fragmentary, studies of the mitochondrial genome are particularly prevalent; a greater 

number of copies of mtDNA exist in each cell, so mtDNA is more readily recovered (Smith, 

Hulsey, (Pack) West, & Cabana, 2016). Notably, in a study of cranial and dental traits, mtDNA, 

and Y-chromosome DNA, Herrera and colleagues found that mtDNA was more highly 

correlated with cranial measurements, while Y-chromosome data is more strongly associated 

with cranial nonmetric traits (Herrera et al., 2014). 

Although ancient DNA has been considered the “gold standard” by which all other 

distance studies are measured, the destructive, expensive, and specialized analyses required 

preclude many researchers from utilizing this analytical tool. Additionally, DNA preservation is 

not guaranteed in archaeological samples, and is highly contingent upon the taphonomic 

environment and curation process. Finally, aDNA can provide information about genetic 

ancestry, but does not directly measure mobility. Therefore, although it may be useful in 

migration studies, other distance measures should also be employed to produce a fuller picture of 

ancient movements (Nikita, 2024). 

Biodistance in Bioarchaeology: Skeletal and Dental Metrics and Nonmetrics 

In cases where ancient DNA is either unavailable or unable to be used, phenotypic 

proxies can be employed in place of genetic data in order to evaluate distance. Traditional 

biodistance studies use metric and morphological (nonmetric) data from the skull and dentition; 

34 

 
these metric or nonmetric traits are the phenotypes used as genetic proxies to explore relatedness. 

In general, nonmetric traits are discrete traits scored in a categorical way, typically as a binary 

“present” vs. “absent” or using an ordinal scoring system (Buikstra & Ubelaker, 1994; Pink, 

Maier, Pilloud, & Hefner, 2016). Dental nonmetric traits, specifically, refer to the shape and 

form of teeth (Pilloud, Edgar, George, & Scott, 2016). One of the most commonly-used systems 

for nonmetric dental analysis is the Arizona State University Dental Anthropology System 

(ASUDAS), which presents standardized plaques and definitions for dental traits that have been 

consistently defined (Irish, 2015; Pilloud et al., 2016; Turner, Nichol, & Scott, 1991). Further, 

datasets such as Ossenberg’s (2013) cranial nonmetric trait database can be used for global 

comparisons and Konigsberg’s online publication of statistical code for evaluating biodistance 

have improved the utility of cranial nonmetric data for biodistance research (Pink et al., 2016). 

Nonmetric traits are especially useful in fragmentary or incomplete assemblages, as are 

common in bioarchaeological settings, because small fragments that may not provide useful 

metric data can still contain scorable nonmetric traits. Further, these traits are relatively simple to 

identify, the data is straightforward to collect, and there are a variety of statistical methods 

available for analysis, including Smith’s mean measure of divergence (Pink et al., 2016). 

However, asymmetry in trait expression, inter-trait correlations, and the relationship between 

traits and biological variables such as age and sex all require consideration prior to undertaking 

nonmetric analyses (Pink et al., 2016). Cranial and dental nonmetric traits have been used for 

large-scale studies of biodistance that have identified geographically clustered dental complexes 

(Hanihara, 1967), regional migration in Hungary and Italy using dental nonmetric traits (Piccirilli 

et al., 2023), population affinity in the United States and India (Edgar, 2014; Hefner, 2009; 

35 

 
Pillai, Parmar, Babu, & Vyas, 2023), and genetic diversity in Classic period Teotihuacan (Meza‐

Peñaloza, Zertuche, & Morehart, 2021), among a variety of other research avenues. 

Metric methods for biodistance analysis use continuous distance data, which are regularly 

applied to a multivariate statistical framework in order to explore human diversity (Dudzik & 

Kolatorowicz, 2016). Dental metric analyses specifically utilize dental dimensions, typically 

crown length and width, but sometimes involving additional measures such as crown height, 

diagonal or cervical crown measurements, or measures of enamel thickness (Kieser, 1990; 

Pilloud & Kenyhercz, 2016). Craniometric analyses rely on standardized landmark and 

interlandmark distance definitions; many of the craniometric definitions used today originated 

with Martin (1914), were refined by Howells (1973), and collated for modern use by Fleischman 

and Crowder (Fleischman & Crowder, 2018). Howells’ worldwide reference craniometric dataset 

is available online (1996), which has arguably formed the foundation for modern craniometric 

biodistance studies. Generally, cranial and dental metric methods tend to be cost-effective and 

non-destructive, allowing for biodistance analyses with as little as a pair of calipers (Dudzik & 

Kolatorowicz, 2016). However, in cases of fragmentary samples or poor preservation, the 

number and type of possible measurements may be limited. Further, issues of sexual 

dimorphism, age, inter-trait correlation, and inter-tooth correlations must be incorporated into 

research design (Pilloud & Kenyhercz, 2016). 

Critiques of Phenotypic Proxies 

Despite the observed utility of biodistance studies, there are several modern critiques of 

phenotypic proxies for genetic data that merit discussion. These include concerns relating to 

developmental plasticity, including recent work on epigenetic influences, as well as potential 

36 

 
issues with the isolation-by-distance model that underpins much of the bioarchaeological 

biodistance research to date. 

The baseline assumption for biodistance research is that metric and nonmetric human 

variation mirrors variation in neutrally evolving genetic markers, driven by genetic drift, 

mutation, and gene flow. As noted previously, studies have routinely shown the heritability of 

cranial metric and nonmetric traits (Carson, 2006a, 2006b; Martínez-Abadías et al., 2009; Šešelj 

et al., 2015) and that the global variation in these traits generally conforms to an isolation-by-

distance model (Relethford, 1994). However, the environment encountered during development 

also plays a crucial role in structuring the craniofacial complex; this environmental influence 

results in the phenomenon of cranial plasticity, or the capacity for phenotypic alterations due to 

the impacts of the environment and non-neutral selective forces. The fetal environment and the 

role of intergenerational inheritance of matrilineal physiological and psychosocial stress has also 

been reported as influencing the adult craniofacial complex through epigenetic effects, which 

provides an additional framework for understanding how environment influences phenotype 

(Kuzawa, 2005; Mulligan, 2016; F. Ross, Pentecost, & Moll, 2023; Thayer & Non, 2015). 

Epigenetics refers to modifications to the genome that do not alter DNA sequence, but still 

impact gene expression (Mulligan, 2016). Research with non-human mammals has shown 

epigenetic effects on craniofacial morphogenesis (Hallgrímsson, Lieberman, Liu, Ford‐

Hutchinson, & Jirik, 2007). Craniofacial variation results from variation in brain growth as well 

as developmental constraints in basicranial and facial dimensions; these constraints result from 

epigenetic processes, thereby providing evidence for an epigenetic component of craniofacial 

plasticity (Hallgrímsson et al., 2007). Recent data has also shown that epigenetic regulation of a 

gene responsible for craniofacial morphogenesis may cause the globular skull shape of 

37 

 
anatomically modern humans, as compared with archaic hominins (Di Pietro et al., 2021). 

Further, epigenetic changes have been linked to phenotypic differences between human and 

chimpanzee craniofacial structure (Prescott et al., 2015; Selleri & Rijli, 2023). Overall, 

epigenetic modification has been consistently shown to influence human craniofacial structure 

(Selleri & Rijli, 2023). 

Since Boas argued for the impact of the environment on changing cranial form in the 

descendants of immigrants (1912), the importance of cranial plasticity on human variation has 

garnered debate (Gravlee, Bernard, & Leonard, 2003a, 2003b; Sparks & Jantz, 2002, 2003) with 

the general conclusion that plasticity has a small, but significant, effect on cranial variation that 

mostly exists as “noise” when exploring population-level research questions (Relethford, 2004; 

Winburn, Yim, & Stock, 2022). Further, recent studies have shown that most developmental 

plasticity in the human skull is related to the functionality of the masticatory apparatus (Katz, 

Grote, & Weaver, 2016; von Cramon-Taubadel, 2009; Winburn et al., 2022), which retains its 

utility for elucidating population history despite the environmentally-induced variability of this 

region, as the effects of mastication are significantly smaller than the influence of neutral 

evolutionary processes (Collard & Wood, 2007; Katz et al., 2016; von Cramon-Taubadel, 2009). 

Therefore, cranial plasticity does not inhibit the use of craniometric methods for population 

history analyses (Winburn et al., 2022). 

An additional concern for biodistance research is the utility of the isolation-by-distance 

(IBD) model, which is routinely cited when justifying the use of craniometrics as a phenotypic 

proxy for genetic information. Relethford’s IBD model states that geographically proximate 

populations should be more genetically similar than geographically distant populations, due to 

the limitation of geographic distance on gene flow. Relethford showed that this model holds true 

38 

 
on a global scale and that craniometric traits mirror genetic information derived from red blood 

cell polymorphisms and microsatellite DNA markers (2004). However, critiques of and 

alternatives to the IBD model have been presented in recent literature (Border et al., 2024; 

Fletcher Jr. et al., 2022; Petkova, Novembre, & Stephens, 2016). IBD models assume a 

homogenous physical landscape in which distance is the only deterrent to mate selection. 

However, in a non-homogenous landscape (which is more applicable to the real world), 

movement of people, and thus gene flow, is limited by physical barriers. This model, called 

isolation by resistance, integrates physical landscape with genetic models to predict genetic 

differentiation (Petkova et al., 2016). Similarly, new statistical methods such as spatial absorbing 

Markov chains have been recommended to account for impediments to migration, directional 

migration, and landscape-induced mortality during migration, all of which influence gene flow 

(Fletcher Jr. et al., 2022). In addition, social and behavioral dynamics such as non-random 

assortative mating also influence genetic differentiation and deviate from expectations of an 

isolation-by-distance model (Border et al., 2024). 

Another potential issue with the isolation-by-distance model is how scale of analysis 

influences the model. As previously described, biodistance can be explored on scales ranging 

from intra-cemetery to global. Therefore, an important consideration in biodistance research is 

how to handle analyses on a small scale and with a small number of interlandmark distances. 

When attempting to use craniometrics as a proxy for genetics at small geographic and temporal 

scales, it is important to use the most complete data set possible. In cases where few individuals 

and landmarks are used, the resolution of the analysis decreases significantly and craniometrics 

may correlate weakly with genetic data (Smith et al., 2016). Due to the fact that biodistance 

studies using craniometric data will always involve fewer variables than genetic studies, as well 

39 

 
as the fact that morphological differentiation of human crania is constrained by trait correlations, 

there will always be less differentiation between groups in a craniometric analysis than expected 

in a genetic analysis (Strauss & Hubbe, 2010). 

Advances in Statistical Methods 

Methodological advancements for biodistance analysis have come in the form of 

improved statistical classification methods and techniques. Machine learning (ML) models are 

computationally-intensive classification methods that use an iterative approach with few 

assumptions that can circumvent issues with traditional approaches, such as linear discriminant 

function analysis (LDFA) (Ousley, 2016). LDFA approaches require normal distributions and 

are prone to overfitting, whereas machine learning models “learn” based on a training set of data 

with no a priori assumption of group membership and iteratively fine-tune themselves until they 

establish the best criteria for group separation. Machine learning methods have the advantage of 

combining various data types and they produce high classification accuracies. They also 

incorporate resampling and simulation in order to handle issues of small sample size that 

regularly occur in skeletal samples (Ousley, 2016). The utility of several machine learning 

models have been demonstrated for biodistance applications, including neural networks, random 

forests, and support vector machines (Hefner & Ousley, 2014; Ousley, 2016). Although LDFA is 

broadly useful for craniometric classification (Ousley, 2016), ML methods have shown greater 

accuracies for cranial morphoscopic data (Hefner & Ousley, 2014), mixed craniometric and 

morphoscopic data (Hefner, Spradley, & Anderson, 2014), and dental morphology data (Edgar & 

Ousley, 2013). In addition to machine learning models, probabilistic clustering algorithms and 

finite mixture analysis have shown promise for assigning group affinity probability and assessing 

40 

 
overlap within samples, instead of assuming discrete group membership (Algee-Hewitt, 2016; 

Algee-Hewitt, Hughes, & Anderson, 2018; Algee‐Hewitt, 2017a, 2017b; Winburn et al., 2022). 

An additional craniometric analytical method that has gained popularity with biodistance 

researchers is geometric morphometric (GM) analysis. Geometric morphometric methods utilize 

three-dimensional landmark data to analyze shape variation across a study sample (Mitteroecker 

& Schaefer, 2022). Traditionally, a Generalized Procrustes Analysis (GPA) is used, which uses 

superimposition in order to standardize position, orientation, and scale to facilitate comparison 

across a dataset and produce a newly derived set of shape variables (Mitteroecker & Schaefer, 

2022; Rohlf & Slice, 1990; Urbanová & Ross, 2016). GM methods have been used to explore 

population history and ancestral origin since shape variables tend to be correlated with 

population affinity, whereas size variables are more highly correlated with sex (Dudzik, 2019; 

Urbanová & Ross, 2016). GM models have been shown to illustrate more phenotypic variation 

than is present in traditional craniometric data using interlandmark distances (Dudzik, 2019; 

McKeown & Jantz, 2005; Spradley & Jantz, 2016). In fact, non-traditional landmarks and GM 

approaches have exhibited more discriminating power than the traditional metric approach 

(Spradley & Jantz, 2016). Although GM methods offer a more sophisticated analysis of cranial 

data than traditional interlandmark distance analyses, the sheer volume of data, landmarks, and 

semi-landmarks used in GM approaches may be challenging to connect to developmental and 

evolutionary processes, even with advancements in multivariate statistics and machine learning 

models (Mitteroecker & Schaefer, 2022). 

Measures of Mobility: A Biogeochemical Approach 

In addition to the aforementioned distance approaches, biogeochemical methods such as 

stable isotope analysis have also contributed to the exploration of origins and mobility in skeletal 

41 

 
samples. Although not a direct measure of distance, per se, stable isotopes can be integrated with 

genetic and biodistance methods to refine estimates of geographic origin and study past 

migrations (Nikita, 2024).  Stable isotope analysis, very simply, refers to the study of the 

chemical composition of bones and teeth with respect to certain elements that can provide 

information about diet, demography, and life history (Katzenberg, 2008). Strontium and oxygen 

isotopes, specifically, provide information about origin and residence throughout life, as they are 

tied to local drinking water and geology, which are then incorporated into skeletal and dental 

tissue through eating and drinking (Bentley, 2006; Nikita, 2024; Pederzani & Britton, 2019). 

Analysis of dental enamel, which is formed during childhood, provides information about an 

individual’s early years of residence (Montgomery, 2010), whereas isotope ratios in bone can 

illustrate mobility patterns in the later years of an individual’s life (Pederzani & Britton, 2019). 

Therefore, comparing tissues within the same individual can therefore illustrate a person’s 

movement (or lack thereof) during life. Further, comparison of an individual’s childhood isotopic 

signature to that of their burial location allows for a broad assignment of “local” vs. “non-local” 

(Pederzani & Britton, 2019). Although isotope analysis is excellent at providing direct data about 

ancient mobility, the analyses are destructive, can be costly, and requires specialized laboratories 

(Nikita, 2024). In addition, isotopic signatures are not unique to a specific region; instead, the 

same isotopic signature can be observed in various regions (Nikita, Mardini, Mardini, & 

Degryse, 2022; Prowse, 2016). Further, the resolution of isotope analyses are somewhat limited. 

While they are useful for comparing childhood and adult residence, they may not be as useful for 

understanding short-term or short-distance migrations (Nikita, 2024). Finally, human behavior 

can influence isotopic signatures; for example, drinking non-local water that was brought in via 

aqueducts can influence oxygen isotope values (Bruun, 2010). 

42 

 
Integrating (Bio)distance Methods 

Studies that integrate several–or all–of the aforementioned methods have proliferated 

within bioarchaeology, displaying the utility of theoretically-grounded, hypothesis-driven, and 

interdisciplinary endeavors. For example, Stark and colleagues employed strontium and oxygen 

isotopes as well as nonmetric traits to explore mobility at the imperial Roman sites of Isola Sacra 

and Velia, revealing a diverse Roman population with evidence of migration, especially by 

females and children (Stark, 2016; Stark et al., 2020; Stark et al., 2021). Additional work at the 

Roman site of Vagnari using isotopic and DNA analyses found a small but significant contingent 

of non-local individuals, as well as some individuals with genetic affinity with sub-Saharan 

Africa and Asia (Emery, 2018; Emery, Duggan, et al., 2018; Emery, Stark, et al., 2018; Prowse, 

2016; Prowse, Barta, von Hunnius, & Small, 2010). More recently, isotopes, dental morphology, 

and ancient DNA were integrated and used to identify an African-born migrant in the catacombs 

of Saints Peter and Marcellinus in Rome. Two individuals in the assemblage were identified as 

distinctive based on dental morphology; one was identified as having African origin through 

subsequent isotopic and genetic analyses, reiterating the idea that long-distance migration was a 

component of Imperial Roman life (Salesse et al., 2021). Additionally, an analysis of 

craniometric data and strontium and oxygen isotopes in Roman York, England established the 

likely presence of non-local individuals and an overall ancestrally diverse population (Leach, 

Eckardt, Chenery, Müldner, & Lewis, 2010; Leach, Lewis, Chenery, Müldner, & Eckardt, 2009). 

Similar distance studies incorporating macromorphoscopic traits, stable isotopes, and ancient 

DNA have identified several possible circum-Mediterranean migrants from Africa–and possibly 

Asia–in Roman-era and Medieval London (Redfern et al., 2016; Redfern & Hefner, 2019). In 

addition, stress variables and mortality were explored with relation to ancestry; higher hazards of 

43 

 
 
death from the plague were identified in individuals with African affinity (Redfern, DeWitte, 

Hefner, & Kim, 2023). 

As has hopefully been established, modern biodistance measures can be used to explore 

patterns of variation and relationships within and between groups. The present study uses 

biological distance as an analytical tool to explore craniofacial variation within Rusellae and 

between Rusellae and relevant global reference samples in order to contribute to the conversation 

surrounding the population history and demographic structure of the site throughout its 

occupation. The results of these analyses will be integrated with known archaeological and 

historical data, as is recommended in order to develop a more nuanced understanding of mobility 

(Buikstra et al., 2022; Gregoricka, 2021), in order to better understand population structure and 

population history in Late Antique and Early Medieval Tuscany. 

Methodological Limitations of Biodistance Research 

As previously discussed, biodistance analyses use a phenotypic proxy—in this case, 

cranial measurements—to illuminate underlying genetic relationships. Therefore, these analyses 

rely on using appropriate reference samples and biologically meaningful craniometric variables. 

Although the actual underlying relationships within and between groups do not change based on 

which reference groups are included or excluded from the analysis, perceptions and 

interpretations of these relationships can vary due to methodological choices (Key, 1982). 

Therefore, it is crucial to include the most temporally and geographically relevant reference 

samples in biodistance analyses; any deviations from the hypothetical “most accurate” reference 

sample will necessarily introduce error. This issue, coupled with the forced classification of 

methods like Linear Discriminant Function Analysis, will cause instances of misclassification in 

the study. 

44 

 
Another important consideration in biodistance research is how to handle analyses on a 

small scale and with a small number of interlandmark distances. The overall completeness of the 

sample will be the primary driver of which data points are included in the analysis. However, 

whenever it is possible, preference should be given to collecting data from regions of the skull 

with higher heritability (Dudzik & Kolatorowicz, 2016). When fewer individuals and landmarks 

are used, the resolution of the analysis decreases significantly to the point where cranial 

morphology corresponds weakly with genetic data from the same sample and may no longer be 

informative of biological distance (Smith et al., 2016). Due to the fact that biodistance studies 

using craniometric data will always involve fewer variables than genetic studies, as well as the 

fact that morphological differentiation of human crania is constrained by trait correlations, there 

will always be less differentiation between groups in a craniometric analysis than expected in a 

genetic analysis (Strauss & Hubbe, 2010). 

Ancestry & Population Affinity 

Although this dissertation is bioarchaeological in nature, it employs classification 

methods commonly used in forensic anthropology to estimate the ancestral origin of several 

unknown individuals from the Rusellae cemetery using measurements of the cranium. Therefore, 

this dissertation requires a brief discussion of biodistance methods as they are applied to forensic 

ancestry estimation. Following guidelines outlined by the Scientific Working Group for Forensic 

Anthropology (SWGANTH), “ancestry” refers to the ancestral origin or geographic region 

associated with a particular population (2013). As it is used in modern forensic anthropology, 

ancestry estimation refers to the classification of an unknown individual as most similar to a 

reference group or likely region of geographic origin using classification statistics (Dunn, Spiros, 

Kamnikar, Plemons, & Hefner, 2020) and ultimately results in the prediction of a probable social 

45 

 
 
race based on cranial metrics and/or morphology (Stull et al., 2021). Social race refers to the 

culturally-constructed and peer-perceived category derived from an individual’s phenotype 

(Dunn et al., 2020). Accurate ancestry estimation relies on the non-zero correlation between 

social race, skeletal morphology, and geographic origin that exists in the United States as a 

byproduct of assortative mating and institutionalized racism (Dunn et al., 2020; Ousley, Jantz, & 

Freid, 2009). These methods also rely on the heritability of the craniofacial complex and its use 

as a proxy for genetic data (Dunn et al., 2020). Fundamentally, ancestry estimation methods 

utilize a biodistance approach to inform forensic models, moving from a population-wide scale 

to an individual one (Dunn et al., 2020). 

Before diving further into ancestry estimation, it is important to make some 

terminological clarifications. Historically, the term “race” was used to describe an individual’s 

classification into one of three broad groups: African, Asian, or European (A. H. Ross & 

Williams, 2021). In light of incontrovertible evidence against the existence of biological races, 

the term “ancestry” has largely subsumed the term “race” in the literature over the past several 

decades (Pilloud et al., 2021; Tallman, Parr, & Winburn, 2021). However, many argue this has 

been a change in name only, with methods for ancestry estimation remaining largely the same as 

those used to assess “race” (Albanese & Saunders, 2006; Pilloud et al., 2021; Tallman et al., 

2021). More recently, the term “population affinity” has been adopted by some scholars, who 

argue that contemporary methods are ultimately testing the similarity (or affinity) of an unknown 

individual to a known reference sample using biodistance methods (Spradley & Jantz, 2021). I 

argue that this dissertation utilizes a population affinity approach within a biodistance 

framework, and it is this term that I have adopted for use when referring to the present study. 

However, this section primarily uses the term “ancestry,” as it is this term (and the associated 

46 

 
 
methods) that have been the center of most of the debates and discussions within forensic 

anthropology over the past two decades. 

Ancestry can be estimated using one or several of the following types of data: cranial 

measurements, cranial morphoscopic traits, dental measurements, and dental nonmetric traits 

(Dunn et al., 2020). The most commonly-used analytical methods in craniometric ancestry 

estimation are linear discriminant function analysis (LDFA) for classification and Mahalanobis 

distance for a measure of similarity. LDFA was first introduced in this context by Giles and 

Elliot for discrimination between American Black, American White, and Native American 

groups (1962). However, the relatively low accuracies reported by the authors led to a search for 

improved methods of classification, which culminated in the development of FORDISC, a 

software that facilitates sex, stature, and ancestry estimation based on statistical models and a 

robust reference database (Dunn et al., 2020; Jantz & Ousley, 2005). Additional software exists 

for ancestry estimation using macromorphoscopic cranial traits (Hefner, Ousley, & Richardson, 

2024), metric and morphological analysis of the mandible (Berg & Kenyhercz, 2017), and 

geometric morphometric methods 3D-ID (Slice & Ross, 2009). 

The history of ancestry estimation in forensic anthropology has garnered shame from 

nearly all modern practitioners due to the overtly racist misapplications of the science for 

centuries (Stull et al., 2021). Typological studies using anthropometrics and morphology have a 

long and infamous history in biological anthropology; they have been used to document 

differences between “racial types” and were associated with non-biological traits, including 

intelligence, and then injected into a hierarchical structure devolving from the apical position of 

the White male (Dunn et al., 2020; Spradley & Weisensee, 2017). This unpalatable history has 

led to critical discussions within forensic anthropology regarding the ethics of estimating 

47 

 
 
ancestry, and for good reason (Bethard & DiGangi, 2020; DiGangi & Bethard, 2021; Stull et al., 

2021). Biological anthropology’s foundation in colonialism and eugenics, historical reification of 

typological and racist thinking, and perpetuation of scientific racism and pseudoscience all 

warrant considerable reflection and amelioration, where possible (Stull et al., 2021). 

In light of these considerations, a number of reflexive questions have arisen as scholars 

have grappled with the complex nature of ancestry estimation and its implications. These debates 

surround the state of ancestry estimation in modern practice, the terminology that should be used 

in reference to patterned human variation, the representativeness and applicability of modern 

reference samples, the influence of genetic data on ancestry estimation and, importantly, whether 

forensic anthropologists should estimate ancestry at all (Edgar & Pilloud, 2021). On this last 

question, the field remains divided; some scholars believe that human biological variation is 

structured and that phenotypic traits are correlated with social categories, including ancestry, 

making that structure useful and detectable (Edgar & Pilloud, 2021; Stull et al., 2021). Others 

have argued that patterned variation does not align with social races, that these social categories 

do not have a biological meaning, that the traits used in ancestry estimation are poorly 

understood and typological, that ancestry research has been co-opted by White Nationalists for 

nefarious purposes, and that ancestry estimation in casework is harmful and may hinder personal 

identification due to institutionalized racism within the American criminal justice system 

(Adams & Pilloud, 2021; Albanese & Saunders, 2006; Bethard & DiGangi, 2020; DiGangi & 

Bethard, 2021; Edgar & Pilloud, 2021). In extreme cases, some have called for immediate 

cessation of ancestry estimation within forensic anthropology, citing its use as unethical and 

harmful (Bethard & DiGangi, 2020; DiGangi & Bethard, 2021). 

48 

 
Despite the ongoing discussion, contemporary scholars on both sides of this debate 

overwhelmingly reject the typological approach to ancestry and the existence of discrete, 

biological races (Dunn et al., 2020). Therefore, ancestry estimation remains one of the four 

pillars of the biological profile (Edgar & Pilloud, 2021). As such, a variety of justifications for 

ancestry estimation and solutions for the above-cited problems have been proposed. First, 

proponents of ancestry estimation have pointed to methodological accuracy as providing more 

than enough justification for its use (Sauer, 1992). Accuracy rates have consistently been used to 

justify the continued practice of ancestry estimation (Ousley et al., 2009; Sauer, 1992) and 

studies have shown relatively low error rates for current methods (Hughes, Juarez, & Yim, 2021; 

Thomas, Parks, & Richard, 2017; Winburn & Algee-Hewitt, 2021). Notably, however, these low 

error rates may be the product of bias, as only identified cases are included in the studies; this 

possible artificial inflation of accuracy rates deserves further study (Parsons, 2021). Additionally, 

recent studies have addressed Bethard and DiGangi’s (2020) criticism of morphoscopic traits, 

beginning to unpack the relationships between morphoscopic traits, genes, and climate. The 

results suggest that morphoscopic traits can serve as genetic proxies in biodistance analyses and, 

by extension, forensic anthropology (Plemons, 2022). Additional recommendations for 

improving ancestry estimation include: contextualizing results using evolutionary theory and the 

concepts of selection, genetic drift, and gene flow (Edgar & Pilloud, 2021; A. H. Ross & Pilloud, 

2021; Winburn et al., 2022); clarifying terminology and defining terms when they are used to 

avoid ambiguity (Pilloud et al., 2021; Tallman et al., 2021); avoid using terms that imply “racial 

purity” or reify racial typology or hierarchy (e.g. “admixture” suggests mixing of pure “types”) 

(Adams & Pilloud, 2021); engage in meaningful conversation with stakeholders within the 

49 

 
medicolegal system (Stull et al., 2021); and improve and diversify reference samples to better 

capture extant variation (Edgar & Ousley, 2021; Go, Yukyi, & Chu, 2021). 

Admittedly, it is unsettling to conclude that even the most prolific scholars in the field 

don’t have an answer for many of the issues that arise with respect to ancestry estimation, and 

even more unsettling to realize that the field may be wrestling with these issues in perpetuity, 

particularly as American society continues to reckon with the undeniable and insidious 

implications of institutionalized racism in our country. Ultimately, forensic anthropologists 

should ensure their analyses are rooted in cultural context and should continuously reflect on the 

implications of their work, especially when it is uncomfortable (Edgar & Pilloud, 2021). 

This chapter has presented the broader theoretical and methodological underpinnings of 

the dissertation. The next chapter will provide the requisite regional and historical context for 

this dissertation and forms the basis for the specific research questions guiding this study. 

50 

 
 
 
CHAPTER THREE: ARCHAEOLOGY AND HISTORY OF LATE ANTIQUE AND 
EARLY MEDIEVAL ITALY 

The Late Antique and Early Medieval periods in Italy, commonly considered to last from 

the 5th through the 10th centuries AD, are characterized by social change, migration, and cultural 

and religious transitions. This period of upheaval has been widely studied in archaeology and 

history, leading to a number of richly debated questions in these fields. Notably, it has taken until 

relatively recently for bioarchaeology, or the biocultural analysis of human skeletons, to be 

included in these debates. Therefore, the aim of this dissertation in general, and this chapter in 

particular, is to provide the historical and archaeological context of Rusellae and the surrounding 

region, to characterize the existing archaeological and historical debates, and to show how 

skeletal analysis and bioarchaeology can and has contributed to these larger research questions. 

Historical Context 

It is impossible to catalogue the entirety of the historical events that are relevant to this 

dissertation. Doing so would extend more than a millennium prior to Late Antiquity, from the 

origins of the Etruscan civilization, through the formation of the Roman Empire, to its 

expansion, eventual end, and beyond. A vast majority of the goings-on throughout the 

Mediterranean are important and impact Rusellae but are outside the scope of this project. 

However, certain consequential periods warrant discussion, as they provide context for the 

research questions asked in this dissertation. These questions investigate how craniometric 

variation at Rusellae was related to that of contemporary sites, what degree of craniometric 

variation was present at Rusellae and what that might tell us about possible Late Antique and 

Early Medieval migration, and how the people of Rusellae experienced life at the site from 

approximately the 5th century AD through its abandonment in the 12th century AD. It is 

important to understand the major migration events occurring in Tuscany throughout the study 

51 

 
 
period, as well as major sociopolitical shifts that may have impacted health, diet, and disease, as 

this information provides vital context for the results of this dissertation. The historical context is 

broken into two sections that roughly correspond to the Late Antique and Early Medieval eras: 

the first section will discuss the period between AD 400 and AD 800, and the second section will 

outline the important events from AD 800 through AD 1200. 

Late Antiquity (circa AD 400-800) 

The Roman empire was already in decline in the west prior to its collapse. Climate 

change, endemic and epidemic disease, decreasing military strength, and economic change all 

weakened the empire from the inside (Holmes, 1988). The Alemanni, Franks, and Goths began 

raiding portions of the Roman territory during the mid-fourth century AD, capitalizing on the 

internal destabilization of Rome. The Visigoths crossed the Danube into the Empire and settled 

in Roman territory in the late fourth century AD. In AD 410, Alaric, the king of the Visigoths, 

sacked Rome. Following the death of Alaric, the Visigoths withdrew to Gaul, but this invasion 

would not be the last to fall upon Rome. Odoacer, a commander of Germanic mercenaries, 

deposed Romulus Augustulus in AD 476, thereby ending the western Roman empire (Holmes, 

1988). The increasingly frequent incursions of barbarian forces destabilized Rome to the degree 

that it could no longer maintain its footing in the west, falling instead into an amalgamation of 

kingdoms under the rule of various Germanic peoples. 

After the fall of Rome, Germanic kingdoms began to emerge in the West, replacing the 

former western provinces of the Roman Empire. These included the Visigoths in southwest Gaul 

and Spain, the Burgundians in the upper Rhone valley, the Merovingian Franks in northern and 

Central Gaul, and the eastern Goths in Pannonia (S. Mitchell, 2007). At the same time, the Huns 

were destabilized due to the death of Attila, leaving a space for the Bavarians to move into the 

52 

 
upper Danube and the Avars to move into Hungary (Holmes, 1988). These kingdoms became 

independent but primarily maintained the Roman legal and political systems with a mixed 

Germanic and Roman culture. The Germanic tribes did not force the Roman inhabitants out of 

these kingdoms; instead, it is hypothesized that the populations mixed (S. Mitchell, 2007). 

In Italy, the Ostrogoths were first ruled by Odoacer beginning in AD 476, and then his 

successor Theodoric beginning in AD 493. Theodoric maintained Roman customs and politics, 

and brought nearly 100,000 Goths into Italy, who settled in the Po valley in the north, as well as 

near Ravenna and in Naples and Sicily (S. Mitchell, 2007). Following the death of Theodoric in 

AD 526, a struggle ensued for power. Justinian seized on this instability, eventually retaking the 

Italian peninsula in the Gothic War between 535-554 AD but decimating it in the process. The 

siege of urban and rural areas put pressure on precious resources, interrupting trade and 

communication and causing widespread famine. The plague of Justinian in AD 542 also 

triggered enormous demographic decline. The destabilization caused by the Gothic War left Italy 

weak and ripe for subsequent invasions (Brogiolo et al., 2017). 

Following the Gothic War, the Lombards launched a catastrophic invasion of Italy, 

moving from Pannonia and occupying northern Italy by AD 569. Eventually, the Lombards 

controlled most of Italy, with the exception of the Duchy of Rome and a few southern Byzantine 

ports (Wickham, 1981). By the year AD 605, the Lombard kingdom in Italy consisted of three 

components: northern Italy and Tuscany, and the southern duchies of Spoleto and Benevento, 

with the capital at Pavia (Wickham, 2006). The Lombard invasion was especially successful 

because Byzantium was at war with Persia in the east as well as struggling with a Muslim 

onslaught, leaving few forces to defend Italy. In AD 643 the Lombards issued a code of law 

under Rothari, and the Lombard kingdom was recognized by the empire in AD 680. The reign of 

53 

 
the Lombards in Italy lasted until AD 774, when they were defeated by the Frankish king 

Charlemagne at the behest of the Catholic church, ushering in a Frankish rule that would last for 

another century (Holmes, 1988). However, the Lombards remained in control of the Duchy of 

Benevento until AD 1077 (Brogiolo et al., 2017). 

Early Medieval Italy (circa AD 800-1200)  

Following the deposition of the Lombards in AD 774, Italy fell under the rule of 

Charlemagne and the Carolingian line. Charlemagne ascended to the role of King of the Franks 

in AD 768 after his father's death and became the sole ruler of the Franks after the death of his 

brother in AD 771. Charlemagne’s alliance with the papacy and the Catholic church motivated 

his movement against the Lombards in AD 774, thereby initiating Frankish rule in the northern 

Italian peninsula and making Charlemagne King of the Lombards. Charlemagne was crowned 

Holy Roman Emperor in AD 800 by Pope Leo III and, thereby, the king of Italy. His rule was 

characterized by the reunion and consolidation of much of Western and Central Europe for the 

first time since the fall of the Roman Empire in the West (Wickham, 2016). Carolingian kings 

would rule after Charlemagne, but infighting within the dynasty and the subsequent division of 

territory produced weakness in the Carolingian line (Holmes, 1988). Substantive Carolingian 

rule ended in Italy in AD 887 when Charles the Fat was deposed (Wickham, 2016). By this time, 

the former Roman Empire comprised three main political segments: Islamic Spain and North 

Africa, the Byzantine Empire, and the kingdoms of the West (Holmes, 1988). 

The next notable ruler in Italy was Otto I of Germany, who was crowned king in AD 961 

and emperor in AD 962, marking the beginning of a stable period during which the Holy Roman 

Emperor and the King of Italy were typically the same (Wickham, 2016). In the 10th century AD, 

however, upheaval began again in Italy with raiding Muslims from North Africa (Holmes, 1988). 

54 

 
Stability wavered further when divisions between the papacy and the Holy Roman Empire were 

ignited in the late 11th century AD, resulting soon thereafter in the formation of Republican city-

states such as Florence, Pisa, Lucca, and Siena, which were independent of the Holy Roman 

Emperor. Also, in the 11th century came the Norman conquest of southern Italy and Sicily, 

including previously Byzantine, Arab, and Lombard territories. In 1130 AD, these territories 

became known as the Kingdom of Sicily (Wickham, 2016). 

Rusellae 

Situated on a hilltop overlooking the ancient Lake Prilus in Tuscany, Italy, the site of 

Rusellae was undoubtedly subject to the complex sociopolitical developments of the Late 

Antique and Early Medieval periods (Figure 3.1). The occupation of Rusellae can be dated from 

the Etruscan era through the Middle Ages; the Etruscans founded the city during the 9th century 

BC and was then besieged and captured by the Romans in 294 BC, finally becoming a Roman 

colony in the 1st century BC (Grant, 1980). In the 1st century AD, the city underwent a revival 

with the construction of the forum, amphitheater, and sacred areas, while domus are recorded 

both in the center of the city and on the Northern hill (Campana, 2018; Sebastiani et al., 2015). 

As with other Tuscan towns at the time, Rusellae experienced a substantial decline in the 

4th and 5th centuries AD, including abandoning the forum and most public buildings (Sebastiani, 

2017). The transition of the Domus of the Mosaics from an opulent residence to a workshop and 

its eventual abandonment attests to the overall transformation of Rusellae during this period 

(Michelucci, 1985). However, since Rusellae was one of the few inhabited cities during the 

decline of the 4th-5th century AD, it was chosen as the center of a diocese in the late 5th century 

AD (Celuzza et al., 2002; Sebastiani, 2017). Rusellae’s survival was partly due to its 

55 

 
advantageous position near the Tyrrhenian Sea and its location adjacent to the Via Aurelia vetus, 

which connected the city to Rome (Sebastiani et al., 2015). 

Figure 3.1. Map of Italy highlighting the location of Rusellae (red bubble). 

56 

 
 
 
 
After Rusellae was chosen as the seat of the diocese, the church/cemetery complex was 

rebuilt in the late 5th and early 6th centuries AD, and an overall resurgence of the site began in the 

6th century, including the construction of a new perimeter wall and the fortification of the 

amphitheater (Celuzza et al., 2002; Sebastiani, 2017). During this period, craftsmanship grew in 

importance, becoming a primary resource of the area. Additionally, the Lombards took Rusellae 

in approximately AD 593 (Sebastiani et al., 2015). The cathedral appears to have undergone a 

period of refurbishment and decoration during the late 8th-early 9th century AD, and a bell tower 

was constructed in the late 11th century, just before the movement of the bishopric to Grosseto 

(Celuzza et al., 2002; Sebastiani, 2017). 

The widespread appearance of North African pottery, African red slip (ARS) wares, on 

the north hill of Rusellae suggest a circum-Mediterranean trade relationship between Rusellae 

and Tunisia during the 2nd-late 6th centuries AD. Although ARS wares were consistently used 

throughout this period, their use fluctuated in response to the Gothic Wars, the demographic 

decline in Tuscany, and the Vandal conquest of Carthage during the 4th and 5th centuries AD. The 

Lombard conquest of Rusellae in approximately 593 AD effectively severed trade ties between 

Rusellae and North Africa, at least for the following century (Vaccaro, 2011). 

Although excavations have been ongoing at Rusellae for over half a century, the cemetery 

itself was excavated in coordination with the Art and Archaeology Museum of the Maremma and 

in collaboration with the Archaeological Superintendence of Tuscany between 1987 and 1991. 

Presently, approximately 170 tombs have been unearthed, but the complete extent of the 

cemetery is as yet unknown.  Importantly, the cemetery at Rusellae contains some of the earliest 

excavated burials in the region, making its contribution to the anthropological and historical 

understanding of Late Antique Tuscany quite substantial. The cemetery at Rusellae is located on 

57 

 
the slopes of the northern hill, near the Hadrianic-era bathhouse. The baths, abandoned in the 

fourth century AD, were repurposed as a church complex between the sixth and seventh 

centuries AD. The cemetery was subsequently used between the 6th and 12th centuries AD 

(Celuzza et al., 2023). 

Phase I Burials 

The earliest burials are characterized by a more organized layout and are dated to the 

sixth and seventh centuries AD based on the personal ornaments contained therein, including 

earrings, bracelets, necklaces, and combs. The graves tend to be constructed from reused 

travertine from Roman-age paving stones; they are organized in terraces and neatly arranged in a 

northwest orientation. One grave in particular contains Lombard-style funerary objects, including 

a sax and a belt dating to the 7th century AD, which is consistent with the Lombard occupation of 

Rusellae during this period (Celuzza et al., 2002). It has been suggested that these early burials 

represent an elite group who chose burial near the church for its “privileged” location (Vaccaro, 

2011). Additionally, a Lombard inhumation cemetery was discovered southwest of Rusellae, 

producing about twenty-five burials with datable grave goods and stone coffins; scholars have 

suggested this represents a cemetery for a local middle-class family during the Lombard period 

(Campana, 2023). 

Phase II Burials 

The second phase of burials ranges from the 8th century AD until the abandonment of the 

site in the 12th century AD. Aside from dating to a later period, these burials also differ 

substantially from the Phase I burials in several ways. Unlike the earlier graves, the Phase II 

burials do not include personal adornments and are less organized than the Phase I burials. Not 

only are the Phase II burials greater in number, they are arranged more opportunistically, 

58 

 
 
constructed using thin slabs in whatever space and orientation was available. In some cases, 

graves were reused, with the earlier remains pushed to one end of the grave to make room for the 

new interment. 

In the latest stage of Phase II, nearing the abandonment of the site, a number of 

commingled and earthen burials are associated with the construction of a tower, tentatively dated 

to the 11th century AD. The construction of this tower had interrupted earlier burials, which were 

disinterred, moved to an area near the church, and buried in the ground. These interrupted burials 

generally cannot be segregated into individuals and appear instead as a mass secondary burial. A 

bronze pin and shoe buckle associated with the commingled remains suggests that at least some 

of the disturbed burials were from earlier stages of Phase II. Several primary burials can also be 

dated to this latest phase of interment. Finally, the cemetery was no longer used in the 12th 

century when the bishopric was transferred to nearby Grosseto in AD 1138 (Celuzza et al., 

2002). 

Rusellae in Context: The Tuscan Countryside 

Life at Rusellae can only be fully understood or studied by situating it in the context of 

the surrounding region, as the distinctions between Rusellae and contemporary sites may help 

explain some of the demographic shifts and health outcomes seen in this dissertation. In Tuscany, 

the height of rural settlement occurred between the 2nd century BC and the 1st century AD, with a 

substantial number of villas and surrounding settlements concentrated along the coast and near 

rivers. During the 2nd-3rd centuries AD, settlement density markedly declined, with an even more 

substantial decrease in the number of rural sites between the early 4th and mid-5th centuries. 

However, the presence of ARS ware during this period suggests that a circum-Mediterranean 

economy was in operation at this time, despite the decline in the number of rural sites. In the late 

59 

 
5th-mid 6th centuries, there is evidence for the continuity of some villages, the reuse of Roman 

settlements, some dispersed settlements, and even cave dwellings (Vaccaro, 2008). 

During this period, craft production overcame agriculture as the primary economic driver 

of the region; the site of Spolverino functioned as a glass and bone workshop, and metalworking 

is also attested archaeologically in the area (Sebastiani, 2016).  Notably, the coastal area of the 

ager Rusellanus represented an interconnected settlement network in the region of Rusellae, 

articulated by the via Aurelia vetus, the via Aemilia Scauri, the Ombrone river, and the 

Tyrrhenian Sea. These routes, both on land and over water, connected several settlement areas, 

including Alberese, Bagni di Rusellae, and Castiglione della Pescaia, which were likely supply 

centers for Rusellae. Further, this network connected the hinterland to the coast and, 

consequently, the broader Mediterranean. Tyrrhenian ports accepted imports of common goods 

from regions throughout the Mediterranean, including North Africa (Chirico & Sebastiani, 2023). 

However, the Gothic Wars likely decimated many of these settlements and their interconnected 

routes by the mid-6th century AD (Vaccaro, 2008). 

In fact, the 6th century AD can be considered a turning point for Tuscany, representing a 

transition between Roman urban structure and the Early Medieval layout of the landscape. The 

archaeological sites of Scoglietto, Spolverino, and Umbro Flumen in Alberese, Italy, conform to 

these trends, declining between roughly the 3rd and 5th centuries AD and having been abandoned 

by the first half of the 6th century AD. These sites were never reoccupied, suggesting that many 

Early Medieval settlements were probably developed de novo in rural Tuscany (Sebastiani, 

2017). Rusellae, however, maintained its occupation throughout this otherwise tumultuous period 

in the region. 

60 

 
The 6th and 7th centuries also mark a decline in building materials and craft production 

sites. During this time, many residences were built of perishable materials, including wood and 

thatched roofs, as opposed to brick and stone (Ward-Perkins, 2005). Additionally, sites of craft 

production such as Spolverino were deserted after the 6th century AD (Sebastiani, 2016). 

The collapse of the Roman settlement system between the 4th and 7th centuries AD in 

rural Tuscany was characterized by overall demographic decline and the concentration of people 

in fortified, nucleated hilltop villages (Francovich, 2007). By the 7th and 8th centuries AD, 

lowland settlements had decreased in number, and these hilltop villages were almost ubiquitous 

(Francovich, 2008). A stabilization in the number of sites occurred until the 8th century AD, 

followed by a slight increase during the 11th and 12th centuries (Bertoldi, Castiglia, & Castrorao 

Barba, 2019). At this time, hilltop castles were constructed through the process of 

incastellamento (Vaccaro, 2008).  The 8th century AD also appears to represent a turning point in 

the social structure of many sites in Tuscany, showing evidence of a transition from a more 

egalitarian society to social differentiation under an increasingly powerful Carolingian 

aristocracy. This shift is attested in the archaeological and historical literature through prominent 

dwellings inhabited by a lord, workshops, storage areas, and defensive fortifications (Barbiera & 

Dalla-Zuanna, 2009). 

Overall, the demographic decline in Tuscany of the 5th century AD and the abandonment 

of many lowland village sites in the 6th century in favor of fortified hilltop villages, such as 

Rusellae, by the 7th and 8th centuries AD is clear. This shift in settlement practice is a response 

to the drastic decline of the Mediterranean economy, the interaction with invading Germanic 

groups, and the destruction caused by the Gothic Wars and the plague of Justinian. Tuscany in 

general, and Rusellae in particular, are deeply interwoven with the economy and society of 

61 

 
western Europe, and the outcomes attested in the history and archaeology of the region show that 

the impacts made by these large transitions clearly reverberated on a small scale. 

Case Study: Cosa 

Although the present dissertation is focused on Rusellae, an understanding of 

contemporary sites can provide important historical and regional context. One key comparative 

case is the hilltop village, Cosa, which overlaps temporally in occupation with Rusellae and is 

located approximately 65 km to its south. Notably, Cosa has provided some of the only other 

evidence of African presence in Late Antique and Early Medieval Tuscany (Fentress, 2003); as 

this knowledge formed part of the impetus for the present study, a brief discussion of Cosa’s 

history and cemeteries is warranted. 

Cosa was established as a Roman colony in 273 BC on land that had been parceled out 

from the territory of the defeated Etruscan city of Vulci. Many of the original inhabitants of this 

colony were drawn from Rome proper; the colony was laid out systematically mimicking the 

organization of Rome, and probably consisted of about 9,000 people in the city of Cosa and the 

surrounding farms (Barker & Rasmussen, 1998). The collapse of the original colony required a 

new injection of colonists in 197 BC; these colonists built most of the domestic and public works 

in the town, but the settlement again collapsed in approximately 70 BC. In fact, Cosa was settled 

no fewer than six times, each resulting in decline or collapse (Fentress, 2003). However, a more 

recent reevaluation of the excavation data, particularly the numismatics, suggests minimal 

continuity of the site during Late Antiquity. Two phases of its settlement—in the fifth-sixth and 

the tenth-fourteenth centuries AD—contain burials that overlap with the cemetery of Rusellae 

and deserve special consideration here (De Giorgi, Hobart, Ludke, & Scott, 2023; Fentress, 

2003). 

62 

 
 
The fortified settlement of the sixth century was small and civilian in nature, consisting 

of a bread oven, a cemetery, a few houses, a granary, stables, a barn, and a church. The skeletons 

of seven individuals, two adults and five children, contemporary with this occupation suggest 

relatively poor health. The presence of ARS wares, north African inscriptions, and eastern 

Mediterranean amphorae is consistent with long-distance trade despite the humble size and 

ephemeral nature of the settlement. This settlement appears to have been abandoned at the end of 

the 6th century, beginning a period of relative stagnation at the site until the tenth century 

(Fentress, 2003). 

The Early Medieval cemetery (primarily 11th century AD) appears to have been used in a 

number of phases, vacillating between organized burial and random interments with almost no 

grave goods. Anthropologically, the individuals in this cemetery appeared to have undergone 

some sort of anemic response, potentially indicating the presence of malaria at the site. 

Importantly, the tentative presence of sub-Saharan African individuals was reported at this 

cemetery based on morphological features of the skeletons (Fentress, 2003). 

Archaeology of Tuscany 

It should be clear by now that the rural landscape of Tuscany provides a rich opportunity 

for understanding the transition from Roman to Early Medieval eras. However, much of this 

landscape has historically been understudied because traditional archaeological survey methods, 

including field walking, are insufficient to detect “negative features” such as ditches and pits that 

might provide evidence for rural settlements. The regions of time and space that have been 

deemed “less visible” archaeologically can be considered “emptyscapes” (Campana, 2017). It 

has not been until relatively recently that consideration has been given to these “emptyscapes,” 

63 

 
 
 
thereby allowing a more complete picture of a time and place that has long been considered 

archaeologically invisible (Campana, 2018). 

Specifically, the use of geophysical methods and a transition from a site-based to a 

landscape-based research perspective has allowed for more focus on environment, 

communication and trade, agriculture, and settlement patterns with the ultimate goal of 

illuminating the “archaeological continuum” with evidence from all time and space in a defined 

area (Campana, 2017). This method has recently been applied to rural Tuscany, revealing a shift 

from dispersed lowland villages to nucleated hilltop settlements; this evolution is now seen as 

characteristic of the transition between the Roman and Early Medieval periods in central Italy 

(Hodges & Whitehouse, 1983). 

A similar framework is used by the nEU-Med Project, a large, interdisciplinary research 

endeavor aiming to understand how a sample region of southern Tuscany experienced the 

upheaval of the seventh through twelfth centuries AD through archaeological, historical, and 

environmental investigations (Dallai & Marasco, 2023). This project, organized by the 

University of Siena, addresses historical questions relating to the Mediterranean economy 

between the 7th and 12th centuries AD, which Wickham (2023) has argued deserve special 

regard. Remarkably, the project focuses on the shift in settlement patterns, landscapes, farming, 

exploitation of resources, and the political structure of the period. The project is focused in the 

northern Maremma region, including the coast of the Tyrrhenian Sea, which is considered a 

representative sample of much of the Western Mediterranean. This region is also helpful for 

study as many of its Early Medieval structures, settlements, and production sites have been 

surveyed, excavated, and well-documented. The cemetery at Vetricella has also provided skeletal 

and bioarchaeological data to the nEU-Med project (Bianchi & Hodges, 2018). 

64 

 
 
Another important Tuscan archaeological Project is the Alberese Archaeological Project, 

which studies settlement patterns and trade networks between the mid-Republican period and the 

Late Antique period in Tuscany. This project, established in 2009, focused primarily on three 

sites: the temple and sanctuary of Diana Umbronensis at Scoglietto, the manufacturing district of 

Spolverino, and the settlement of Umbro flumen. Excavation of these sites provided evidence for 

a significant transition in settlement patterns in the mid-6th century AD, including the 

abandonment of several once-bustling cultural and economic centers (Sebastiani, 2017). 

The Alberese region is also studied within the Impero (Interconnected Mobility of People 

and Economy Along the River Ombrone) Archaeological Project, organized by the University at 

Buffalo. Launched in 2017, the Impero Project focuses on economic and settlement trends in 

southern Tuscany between roughly the 2nd and 9th centuries AD. More specifically, the Impero 

Project aims to understand how the Ombrone River connects the micro-level Tuscan settlements 

and landscapes to the macro-level Mediterranean economy. This goal is achieved, in part, 

through the excavation of a Roman to Late Antique villa and a medieval castle, as well as field 

surveys of the surrounding region. Importantly, the Impero Project seeks to understand the 

relationship between the coast, the hinterland, and the urban center of Rusellae (Sebastiani, 

2017). 

Major Archaeological Debates 

The end of the Western Roman Empire, traditionally marked by the deposition of 

Romulus Augustulus in AD 476, ushered in a complex period characterized by population 

mobility, shifting economies, and changing settlement structures. Between the 5th and 10th 

centuries AD, this transitional period has been studied extensively by historians and 

archaeologists alike. This research has led to a number of notable debates within archaeological 

65 

 
 
and historical scholarship. Importantly, these debates, in part, inspired the research questions 

guiding this dissertation, and the goal of the present research is to use a biocultural approach to 

skeletal data to contribute to the ongoing historical and archaeological conversations. 

The Nature of Late Antiquity 

The nature of the “fall” of the Roman Empire has been the subject of intense debate, with 

two primary schools of thought dominating the literature: the “Late Antiquity” cultural continuity 

model and the “decline and fall” model. These models provide a lens through which to view 

historical events, archaeological finds, and material culture. 

For many historians of Europe, the end of the Roman Empire was considered 

cataclysmic. When Rome fell in AD 476, it was thought that civilization had ended in the West 

and Europe had plunged into a “Dark Age” from which it would not recover until at least the 10th 

century. Further, the fall of Rome was seen to mark the end of Classical Antiquity and herald the 

arrival of the Medieval period. Although many scholars had debated the veracity of this 

viewpoint for at least a century, most notably Pirenne, it was not until Brown’s “The World of 

Late Antiquity” was published in 1971 that scholarly perspectives on the fall of the empire 

shifted. Brown popularized the term “Late Antiquity,” which refers to the period between the rise 

of Diocletian in the 3rd century AD and the rise of Islam in the east in the 8th century AD. The 

term “Late Antiquity” is meant to imply a continuity with Classical Antique thought and culture. 

Brown primarily used the rise of the Abrahamic religions (Christianity, Islam, and Judaism) 

during this period as evidence that civilization did not, in fact, end after AD 476. Instead, the rise 

of monasticism, the development of major world religions, and the proliferation of art suggested 

a continuation of civilization and culture (P. Brown, 1971). Brown’s viewpoint is a product of 

cultural relativism, which rejects the notion of “civilizations” as more or less complex. Instead, 

66 

 
cultural relativism posits that different cultures exist equally, irrespective of their complexity. 

Thus, in Brown’s perspective, the Roman civilization did not end; instead, the Roman culture 

continued until it was eclipsed by the rise of Islam in the east. 

Whereas Brown envisions cultural and religious continuity until the 8th century AD, 

Horden and Purcell argue that there is no natural “break” at which point classical antiquity ends 

and Medieval culture begins. They use the notions of connectivity and redistribution as a proxy 

for the economic health of the Mediterranean. Additionally, they view the Mediterranean as a 

complex aggregation of variable microregions. Using evidence from ship logs, material culture, 

and piracy, Horden and Purcell argue that the circum-Mediterranean trade was ongoing during 

the so-called “Dark Ages.” They also propose that, while this connectivity and redistribution may 

have slowed down and become more regional in nature, it never entirely ceased. This suggests 

that the Mediterranean economy never fully collapsed, as is implied by the “decline and fall” 

model (Horden & Purcell, 2000). 

A third perspective, quite distinct from Brown and Horden and Purcell, is that of Ward-

Perkins. Much like his predecessor, Gibbon, Ward-Perkins ascribes to the idea of the decline and 

fall of the Roman empire. His work is a response to the notion, espoused primarily by Brown, 

that the end of the Roman empire was a period of “transformation and accommodation.” Using 

economic history, Ward-Perkins describes a “demise of comfort.” This refers to a decline in the 

quality of goods and building materials, including a decreased number and sophistication of 

ceramics, limited coinage in the 5th and 6th centuries AD, and a decrease in graffiti, which Ward-

Perkins equates to a decline in literacy (Ward-Perkins, 2005). For Ward-Perkins, the fall of the 

Roman empire in AD 476 marks the beginning of Medieval Europe, with no intermittent “Late 

Antiquity.” 

67 

 
 
 
Importantly, Tuscany in general, and Rusellae in particular, can be examined through 

these lenses in order to understand which of these three models, if any, best fits the 

archaeological data. In general, Tuscany follows the expectations set forth by Ward-Perkins in 

terms of substantial economic decline, decreased sophistication of material culture, and lesser 

quality of building materials. However, I argue that the “decline of comfort” described by Ward-

Perkins did not occur in AD 476 in Tuscany. Instead, it seems that Rusellae experienced some 

continuity with classical antique culture until the Lombard invasion in AD 569. This continuity 

was facilitated by the Ostrogoths, who took great pains to maintain Roman culture and political 

structure. They appointed Roman civil servants to political office and oversaw the construction 

of Roman-style castles, palaces, theaters, baths, and fortifications. This strategy aimed to gain 

legitimacy by reproducing “Roman-ness” through monumental construction (Brogiolo, 1999; 

Brogiolo et al., 2017; Revell, 2008). However, the Gothic Wars and Lombard invasion of the 6th 

century AD would spell the end of cultural continuity and the acceleration of economic decline. 

The substantial demographic constriction in Tuscany following the Lombard invasion and 

the disappearance of ARS wares from the archaeological context in the 6th century AD suggests 

that Ward-Perkins’ interpretation has some merit. Additionally, the reuse of Roman sites and the 

occupation of cave sites during the 5th and 6th centuries, after the fall of Rome and during the 

Gothic Wars, implies a decrease in construction that is to be expected based on Ward-Perkins’ 

theory of “the disappearance of comfort” (2005). However, the continual occupation of Rusellae, 

even while most other sites in the area disappeared, suggests at least some level of continuity 

throughout this era that falls most in line with Horden and Purcell’s notion. As a coastal, 

fortified, hilltop site, Rusellae likely acted as a node in the more regional trade networks 

proposed by Horden and Purcell, leading to at least a low level of sustained craft production and 

68 

 
trade. Additionally, Brown might argue that the centering of Christianity at Rusellae is a 

significant factor for its resilience, citing the importance of Christianity in determining which 

sites remained viable during Late Antiquity. 

Although archaeological evidence can and does contribute to an understanding of how 

Rusellae might have experienced the period between the late 5th and 10th centuries AD, there is 

a gap in the literature regarding skeletal analysis. The human skeleton can record evidence of 

stress and disease that might have impacted the inhabitants of Rusellae, which can thereby 

facilitate an understanding of the Late Antique environment there. For example, high levels of 

non-specific stress indicators may indicate a degree of malnutrition, dietary deficiency, or 

infection within the population; this finding could corroborate the “decline and fall” model and 

suggest changes in traditional subsistence strategies or interruptions in trade that forced a 

transition to reliance on different sources of food or income. On the other hand, low levels of 

skeletal stress in the population could suggest an adherence to the continuity model, wherein 

Rusellae maintained its local and regional socioeconomic connections, and overall health at the 

site was not significantly impacted by the transition between Imperial and medieval times. This 

dissertation, therefore, aims to add to the ongoing conversation about the nature of Late 

Antiquity through the use of skeletal analysis. 

The Lombards: Invasion or Assimilation? 

A second debate involves the nature of migration and the impact of shifts in power 

between the Lombard tribes and local Romans during Late Antiquity. This period undoubtedly 

saw substantial movement of people, particularly of central European groups, into the Italian 

peninsula. However, it is not always clear how the influx of migrants impacted the 

autochthonous Romans culturally, whether these migrants or their descendants existed in great 

69 

 
 
enough numbers or assimilated with locals enough to influence genetic variation in their new 

homes, and how the transition of power might have affected overall health and stress in the 

existing population. Rusellae provides an opportunity to explore this debate through an analysis 

of craniometric–and, by proxy, genetic–variation and health outcomes during and after the 

Lombard period at a site of known Lombard occupation. 

When the Lombards began to enter northern Italy in AD 568, they encountered a 

politically, socially, and geographically fragmented region. The two decades of the Gothic War 

(AD 535-554) had destroyed the Italian peninsula, causing famine, plague, and massive 

demographic decline. Fiscal resources were limited, and economic relationships with other parts 

of the empire were severed, leaving the Byzantines unable to defend the Italian peninsula from 

the invaded Lombards (Brogiolo et al., 2017). 

Historical texts do not provide a complete or necessarily accurate account of what 

happened to the Romans when the Lombards moved through Italy. It is assumed that the 

Lombard invasion was violent and that the Romans were either enslaved or forced out of their 

homes. On the contrary, archaeological evidence suggests an assimilation of the groups and a 

much less radical Lombard invasion than originally hypothesized (Wickham, 1981), resulting in 

local communities with substantial cultural communication and crossover between the two 

groups. For example, the Lombards used Latin and copied the design of Roman gravestones and 

sarcophagi (S. Mitchell, 2007). The Lombards also made an effort to maintain existing 

fortifications and infrastructure but were not always able to prevent the decline of these 

structures over time (Brogiolo et al., 2017). This is a product of the breakdown of the taxation 

system during the Lombard era, which led to a less wealthy government than the preceding 

Byzantines. The lack of wealth also likely caused the decline in imported wares during this 

70 

 
period, particularly African red slip ceramics, as the Lombard government could no longer afford 

to import these goods (Wickham, 1994). 

Despite the archaeological evidence of assimilation by a small group of Lombard elites, 

there is some suggestion that the effect of the Lombard invasion differed based on geography; 

the frontiers of battle were probably devastated, while the remaining imperial territories in Italy 

were moderately affected by disruptive political change. Finally, the new Lombard regions 

probably experienced such a rapid invasion that the lives of Roman peasants were not drastically 

altered (Holmes, 1988).  Therefore, Rusellae’s location in Tuscany may be more consistent with 

a less devastating and disruptive invasion than what was experienced in the towns and cities 

along the northern Italian “frontier” between the Romans and the Lombards. Importantly, this 

dissertation aims to contribute to the conversation surrounding the nature of the Lombard 

“invasion” by analyzing evidence of skeletal stress at Rusellae during the Lombard period. 

A secondary debate surrounding the Lombards involves their demographic makeup and 

to what degree their presence influenced the genetic variation within the in situ Roman 

population. The concept of who, exactly, constituted the “Lombard” people is somewhat abstract 

(Brogiolo, 2000). The inhabitants of Lombard Italy are generally referred to as Lombards; 

however, the great majority of people in Italy during this period would have been Roman, with 

ethnic Lombards possibly comprising no more than 8% of the population (Wickham, 1981). The 

Lombards were probably biologically heterogeneous, primarily united by a common culture. 

However, the people the Lombards encountered as they moved throughout Italy were also 

heterogeneous; although Roman from a cultural perspective, these people were Byzantines, 

Eurasians, Middle Easterners, North Africans, and Western Europeans who had either come from 

71 

 
or migrated to the Italian peninsula in a religious, trade, or military capacity (Brogiolo et al., 

2017). 

In sum, the hypothesis presented by historians is that the Lombard people were 

genetically heterogeneous and probably made up only a tiny percentage of the overall population 

of Italy during their occupation. However, there is renewed archaeological interest in further 

exploring the demographics and genetic ancestry of the Lombard people, particularly at central 

European and northern Italian sites such as Szolad and Collegno (Alt et al., 2014; Amorim et al., 

2018; Tian et al., 2024). Rusellae offers an opportunity to contribute to the scholarly 

conversation by exploring craniometric variation in a time and place of known Lombard 

occupation and by contextualizing that variation within a broader Mediterranean context. 

Therefore, this dissertation will offer further data to better understand who the Lombard people 

were and how their presence influenced the occupants of Rusellae. 

Mediterranean Bioarchaeology 

The Mediterranean, particularly in the classical period, has been studied extensively with 

respect to history and archaeology. However, the utility of skeletal analysis for answering 

archaeological questions has only sometimes been obvious to historians of the Mediterranean. It 

has only been in recent decades that interdisciplinary and collaborative conversations between 

archaeologists, historians, and biological anthropologists have given rise to well-contextualized 

and interesting bioarchaeological research questions in the Late Antique and Early Medieval 

periods. Anthropologists have the unique power to provide skeletal data, which historians can set 

within the appropriate historical framework, thereby producing nuanced studies on topics such as 

paleopathology, health, diet, disease, stress, and mobility, among others. Further, newly-

developed and rapidly-improving methods such as stable isotope analysis and ancient DNA 

72 

 
techniques can provide robust information to the existing historical narrative. Additionally, 

integrating anthropological, archaeological, and historical research programs has shifted focus 

from site-specific case studies to landscape and population-based approaches (Sperduti, Bonditi, 

Craig, Prowse, & Garnsey, 2018). 

Migration 

One of the most prevalent research topics within Mediterranean bioarchaeology is 

migration. Specifically, skeletal data have been used to contribute to conversations surrounding 

the demographic composition and movement of people from the Imperial era through the Middle 

Ages using stable isotopes, ancient DNA, and craniometric analyses. 

Roman-Era Migration 

Overall, the genetic composition of Europe during the Imperial period showed a shift in 

ancestry toward the eastern Mediterranean but maintained a heterogeneity in ancestral 

composition and included individuals of North African ancestry (Antonio et al., 2019). The 

presence of non-local individuals in Roman-era cemeteries has been attested using strontium, 

oxygen, and carbon isotopes in the Imperial Roman cemeteries Casal Bertone and Castellaccio 

Europarco (Killgrove & Montgomery, 2016) and Isola Sacra (Prowse et al., 2007). However, 

each of these cemeteries is located in a well-connected area surrounding Rome, where migrants 

are expected. Interestingly, a more rural, southern Italian necropolis at Vagnari (1st-4th centuries 

AD) showed through oxygen and strontium isotopes that only 7% of the individuals at the site 

were migrants. Even at this site, however, researchers found isotopic and mtDNA evidence of 

North African migrants and non-local subadults at the cemetery (Emery, Stark, et al., 2018). 

In a study of cranial morphological variation during the Roman era, Hens and Ross used 

geometric morphometric analyses in three cemeteries dating from the 1st-3rd centuries AD 

73 

 
 
(2017). Two of the cemeteries (Isola Sacra and Velia) were significantly distinct from one 

another with respect to cranial morphology, while the third (Castel Malnome) was intermediate 

between the two. The authors suggest that Castel Malnome may be indistinguishable from the 

other two sites because it exhibited significant heterogeneity in cranial morphology. This 

variation, they argued, was because Castel Malnome represents a population of freed slaves with 

a wide variety of geographic origins (Hens & Ross, 2017). 

Evidence of migration has also been identified more broadly throughout Europe using the 

skeleton; Shaw and colleagues found evidence of individuals of North African origin in Imperial-

era London through lead and strontium isotope analysis of tooth enamel (Shaw, Montgomery, 

Redfern, Gowland, & Evans, 2016). Using strontium and oxygen isotopes in tandem with cranial 

measurements, Leach and colleagues found evidence of African migrants and their descendants 

in Roman York, England (2009). One young female (4th century AD) was buried in a stone coffin 

with rich grave goods. Her cranial measurements indicate an affinity with African reference 

groups and her oxygen isotope content suggests a childhood outside of York, signifying long-

distance migration and subsequent acculturation during the Roman era (Leach et al., 2010). In an 

additional study of Roman London, a combination of macromorphoscopic analysis and oxygen 

isotopes also showed the likely presence of individuals of African ancestry within Roman Britain 

(Redfern et al., 2016). Further, one late Roman cemetery (AD 330-400) from Bavaria comprised 

approximately 30% non-local individuals, many of whom had migrated as children (Schweissing 

& Grupe, 2003). Based on these results, long-distance migration was a significant component of 

Roman life, and the skeletons can provide essential data about how people moved throughout the 

Roman Empire. 

74 

 
 
Late Antique and Early Medieval Periods 

Skeletal analysis has also contributed to a better understanding of the demographic 

makeup of the Lombard people and their interactions with the autochthonous inhabitants they 

encountered. In a study of two 6th- 8th century AD Lombard cemeteries, one in Hungary (Szólád) 

and one in Northern Italy (Collegno), ancient genomic researchers found that each site consisted 

of at least two separate genetic groups–one with northern/central European ancestry and another 

with southern European ancestry. The two groups each exhibited distinct burial practices, 

potentially suggesting the coexistence of Lombard migrants and indigenous Romans within the 

cemeteries. However, the authors are cautious to note that neither of these genetic groups is 

homogenous. However, the degree of variation is significant enough to identify groups with 

differing genetic ancestries in two sites of known Lombard contact (Amorim et al., 2018). 

Additional isotopic analyses at Szólád indicate a wide heterogeneity in the geographic origin of 

individuals interred there; approximately 31% of the sample was non-local according to 

strontium isotope ratios. These results confirm the notion that migration was common during the 

Lombard period (Alt et al., 2014). Similarly, more recent isotopic and genomic research at 

Collegno concluded that the site comprised individuals with diverse genetic ancestries and 

isotopic origins, suggesting a historical model of integration of new members of the community 

(Tian et al., 2024; Viva et al., 2022). 

Contemporary sites throughout Italy have yielded similar results. For example, 

paleogenomic research in South Tyrol revealed a complex genetic history of individuals in the 

cemetery of Malles Burgusio Santo Stefano (4th-7th centuries AD), with the primary genetic 

ancestry deriving from southern Europe, but with additional contributions from southwestern, 

western, and northern Europe. This heterogeneous ancestry data points to hybridization of both 

75 

 
culture and genetics in northern Italy during the Late Antique and Early Medieval periods (Coia 

et al., 2023). Additionally, strontium isotope analysis at the site of Povegliano Veronese (Verona, 

Italy, 6th-8th century AD) has identified at least nine individuals whose geochemical signatures 

match a region in Hungary, hypothesized to be the origin of the Lombard people. In contrast, the 

remaining individuals in the cemetery appeared to be of local origin (Francisci et al., 2020). 

Similarly to other studies, this evidence points to an integration of non-local Lombard peoples 

with autochthonous groups in Italy. The research described here shows that anthropological 

scholarship throughout the last decade has moved towards using the human skeleton to identify 

trends in migration, demographics, and ancestry in Italy throughout Late Antiquity and the Early 

Middle Ages. 

Skeletal Health Studies 

Another significant research program within Mediterranean bioarchaeology uses skeletal 

stress markers to study health outcomes over time and throughout the landscape. As noted 

previously, a particularly popular archaeological and historical debate involves the nature of Late 

Antiquity–was there a “decline and fall” of the Roman Empire, leading to a “Dark Age,” or was 

there cultural continuity resulting in a long Late Antiquity? Historical evidence suggests that 

disease and stress were widespread during the sixth century AD as a result of the Gothic Wars, 

the plague of Justinian, leprosy, famine, malaria, and economic decline during this time 

(McCormick, 2001). Additionally, the population of Italy declined by half between the 3rd and 6th 

centuries AD (Belcastro et al., 2007). Therefore, one might expect to observe skeletal 

manifestations of poor health during this transitionary period. 

However, when studying skeletal data, it becomes clear that a disastrous transition was 

not felt ubiquitously throughout Italy. Instead, various studies have shown differing health 

76 

 
 
outcomes during this time. For example, a study of stature as a proxy for health showed an 

increase in stature between the Roman and Medieval periods, indicating a generally positive 

health trend over time (Giannecchini & Moggi-Cecchi, 2008). An additional study comparing the 

frequency of linear enamel hypoplasias, cribra orbitalia, and periostitis in Molise, Italy, shows 

that poor health was actually consistent from the Roman through the Early Medieval period 

(Belcastro et al., 2007). An analysis of carbon and nitrogen dietary isotopes at the site of Cosa in 

Tuscany also found no evidence of a substantial shift in diet between the Early Medieval and 

Medieval occupations (Scorrano et al., 2014). 

On the contrary, other studies have shown a decline in health using the frequency of 

dental lesions in Roman and Early Medieval cemeteries in Latium (Manzi et al., 1999). In a 

study attempting to ascertain the nature of the transition between Late Antique and Early 

Medieval Croatia, a comparison of cribra orbitalia, linear enamel hypoplasia, non-specific 

periostitis, and trauma between various sites showed a continuity in health outcomes in 

continental Croatia, but a worsening of health in Adriatic Croatia (Slaus, 2008). 

With respect to the Middle Ages, skeletal evidence from the site of Vetricella (10th-11th 

century AD; Scarlino, Grosseto, Tuscany, Italy) suggests high levels of childhood stress (between 

the ages of approximately 2.5 and 4 years) in the form of linear enamel hypoplasia in nearly 90% 

of the studied sample. Further, 76.9% of the study sample with at least one orbit present 

exhibited cribra orbitalia, while 52.6% of the sample with at least one parietal bone present 

exhibited parietal hyperostosis; the incidence of both these lesions remains high even when age 

is considered, but the incidence of both lesions is significantly higher in females than males in 

the study sample (Viva, 2020). These results suggest an environment in which health outcomes 

varied by sex, and the frequency of non-specific stress indicators is more similar to the Roman 

77 

 
era than the Late Antique period in this part of Tuscany. Interestingly, the high prevalence of 

porotic hyperostosis and cribra orbitalia was attributed, through isotope studies, to congenital 

anemia, as opposed to nutritional deficiency anemia (Viva et al., 2021). 

This seemingly conflicting research actually shows that the transition from the Late 

Roman to Early Medieval periods was experienced differently in different regions of the 

Mediterranean. Bioarchaeology has given us a substantially more nuanced perspective on what 

some historians and archaeologists call a “Dark Age.” By using the human skeleton, 

bioarchaeologists have the power to explore individual experiences and aggregate their data to 

make site- and region-wide comparisons. 

This chapter has provided an overview of the historical and archaeological background of 

Rusellae and the surrounding Tuscan countryside. The next chapter will use this data to 

contextualize the four research questions guiding this study as well as their respective 

hypotheses. 

78 

 
 
 
CHAPTER FOUR: RESEARCH QUESTIONS AND EXPECTATIONS 

The present research focuses on understanding the lived experience of individuals 

interred in the cemetery at Rusellae, Tuscany, Italy during the Late Antique and Early Medieval 

periods (5th-12th centuries AD). These eras represent a time of intense social, political, religious, 

and economic change throughout Europe, characterized by population mobility and cultural 

transitions. This study aims to understand the relationship between craniofacial variation, 

physiological stress, and biocultural factors throughout these widely debated and well-studied 

periods. These data can provide insight into how individuals navigated differing identities, 

political climates, and health statuses during a time of great change in Italy. 

Bioarchaeology is a more recent avenue of research in Italy, which can contribute greatly 

to the numerous archaeological and historical debates surrounding the Late Antique and Early 

Medieval periods. Due to historical underrepresentation of skeletal analyses in Italian 

bioarchaeology, there are a number of gaps in understanding of ancient life that can be addressed 

by direct skeletal observation, particularly with regard to lived experiences. By studying 

skeletons in separate phases of a cemetery, anthropologists can also gain an understanding of 

how health and burial practices changed over time. 

The present study focuses specifically on the site of Rusellae, which serves as an example 

of a Late Antique and Early Medieval town in Tuscany. Although Rusellae followed the general 

trajectory of many Italian towns during this time period (Sebastiani, 2017), it was also chosen as 

the center of a diocese, suggesting it held a place of importance in the landscape (Celuzza et al., 

2002). Although the results of this dissertation will be informative for understanding Rusellae’s 

place in history, they may not be generalizable to all contemporary sites. However, this research 

79 

 
 
 
is valuable to inform an understanding of ancient life in a particular place and over a period of 

time that remains under debate by historians and archaeologists alike. 

This research builds on questions such as: (1) What is the nature of Mediterranean 

migration, and what are the range of homelands represented by migrants throughout the former 

Roman Empire? (Horden & Purcell, 2000; Killgrove & Montgomery, 2016; Leach et al., 2010; 

Leach et al., 2009; McCormick, 2001; Scheidel, 2004, 2005); (2) Is there evidence for African 

migrants in Medieval Europe? (Redfern et al., 2016; Shaw et al., 2016); and (3) How did health 

outcomes change during major shifts of power during and after the Roman Imperial Era? 

(Belcastro et al., 2007; Giannecchini & Moggi-Cecchi, 2008; Manzi et al., 1999). These research 

questions have been explored in recent literature using craniometric, macromophoscopic, stable 

isotope, ancient DNA, and pathological analyses. I propose to use similar analyses to understand 

craniofacial variation, health, and identity during the Late Antique and Early Medieval periods in 

Italy. 

The following research questions represent an initial step in understanding the daily lives 

of people living at Rusellae. Each research question will be presented, followed by the 

hypothesized results and a justification for these expectations. The research questions guiding 

this dissertation are as follows: 

Research Question 1: How does craniofacial variation at Rusellae resemble or differ from that 

of geographically and temporally proximate reference samples, and what might these 

relationships suggest about population continuity, migration, or gene flow? Further, what 

variables drive craniometric variation within Rusellae, and does there appear to be heterogeneity 

or homogeneity in overall craniometric variation at the site? 

80 

 
 
 
Hypothesis 1: I expect that Rusellae will more closely resemble medieval European 

craniometric reference samples and differ most from African reference samples, suggesting 

greater genetic continuity with the European continent. I also expect to find evidence of 

significant craniofacial variation within Rusellae, suggesting the potential presence of multiple 

craniofacially distinct groups at the site. 

First, and most obviously, as Rusellae is in Europe, I expect to find more similarity 

between Rusellae and other European groups due simply to geographic proximity. Despite the 

fact that human migration was a substantial component of Roman and Late Antique life, and 

even knowing that contact with Africa regularly occurred, the largest proportion of that 

migration was likely within the European continent. Therefore, gene flow is primarily expected 

between European groups, with a lesser (but non-zero) degree of gene flow with African groups. 

Further, central European groups such as those studied here are subject to similar environmental 

influences as one another, especially when compared with African groups. These environmental 

factors, such as temperature and humidity, are known to impact cranial traits and craniometrics 

(Plemons, 2022; Relethford, 2004, 2010). Therefore, it is expected that samples with similar 

environmental pressures should exhibit a more similar phenotypic trait expression. 

Although I hypothesize that Rusellae as a whole will cluster most closely with other 

European groups, indicating greater genetic similarity with Europeans as compared with African 

samples, I also expect to find substantial craniofacial variation within Rusellae. Like much of the 

former Western Roman Empire, Rusellae likely would have been composed of a number of 

different population groups. First, Rusellae occupies a coastal location, serving as a node in a 

trade network connecting the sea and inland sites as part of the “Ombrone system,” a commercial 

network in the Ombrone valley (Bertoldi et al., 2019). Further, Rusellae’s location adjacent to 

81 

 
the Via Aurelia vetus connected the city to Rome (Sebastiani et al., 2015). As a node on the 

coast, Rusellae would have attracted traders, migrants, and members of the army from the rest of 

the Empire as they traveled throughout Italy. Inevitably, some would have settled in Rusellae, 

produced offspring, and died, thus leaving their remains in the cemetery. In fact, merchants often 

moved throughout the empire as they sold their wares, as well as members of the Roman army, 

who sometimes settled in the lands they conquered. Often, these migrants were assimilated into 

the Roman world, which was relatively open to foreigners (Horden & Purcell, 2000). This 

migration would have presented the opportunity for gene flow between native Romans and 

migrant groups, thereby introducing more craniofacial variation into the skeletal sample. Further, 

the presence of individuals of diverse origins, or their descendants, also increases expected 

craniofacial variation within the sample. 

In addition, Rusellae was occupied for an extended period of time, having initially been 

founded by the Etruscans in the 9th century BC (Sebastiani, 2017) and not having been 

abandoned until the 12th century AD (Celuzza et al., 2002). During this approximately 2,000-

year occupation, there is both ample time for and ample evidence that several groups of people 

migrated to Rusellae and made it their home. Most notably, Rusellae’s history of having been 

conquered by the Romans and then the Lombards corroborate this hypothesis. Rusellae was 

established as a Roman colony in the early first century BC (Barker & Rasmussen, 1998), at 

which point the Etruscans were displaced and a cultural shift occurred at Rusellae. Much later, 

the Lombards moved into Rusellae in approximately AD 593 (Sebastiani et al., 2015), initiating 

a further change in both the population and culture of the site. When the temporally aggregated 

cemetery assemblage is viewed at once, as is the case here, it is expected that the shifting 

demographics of Rusellae throughout its occupation will manifest as observed craniofacial 

82 

 
variation. Overall, Rusellae is expected to exhibit a degree of variation that indicates the 

presence of a number of groups with distinct craniofacial features throughout the duration of its 

occupation. 

Research Question 2: Do any individuals at Rusellae show a closer affinity to African reference 

samples than European reference samples (Phases I and II)? 

Hypothesis 2: I expect to find that some individuals at Rusellae show a closer affinity to African 

reference samples than European reference samples. 

As previously discussed, Rusellae was a cosmopolitan city that was deeply integrated in 

the trade networks of Late Antique and Early Medieval Tuscany, which were in turn connected 

to the Mediterranean economy as a whole. Evidence of North African migration throughout the 

Roman Empire has been identified broadly throughout Europe; Shaw and colleagues found 

evidence of individuals of North African origin in Imperial-era London through lead and 

strontium isotope analysis of tooth enamel (Shaw et al., 2016). Using strontium and oxygen 

isotopes in tandem with cranial measurements, Leach and colleagues found evidence of African 

migrants and their descendants in Roman York, England (2009). One young female (4th century 

AD) was buried in a stone coffin with rich grave goods. Her cranial measurements indicate an 

affinity with African reference groups and her oxygen isotope content suggests a childhood 

outside of York, signifying long-distance migration and subsequent acculturation during the 

Roman era (Leach et al., 2010). In an additional study of Roman London, a combination of 

macromorphoscopic analysis and oxygen isotopes also showed the likely presence of individuals 

of African ancestry within Roman Britain (Redfern et al., 2016). Based on these results, it is 

clear that long-distance migration, particularly between Africa and Europe, was a significant 

component of Roman life. 

83 

 
Further, Cosa, a site contemporary and near to Rusellae, may provide evidence for 

African migration into Italy. The tentative presence of sub-Saharan African individuals was 

reported at this cemetery based on morphological features of the crania. Additionally, the 

presence of African Red Slip (ARS) wares, north African inscriptions, and eastern Mediterranean 

amphorae is consistent with long-distance trade despite the humble size and ephemeral nature of 

the settlement at Cosa (Fentress, 2003). 

Based on evidence from other sites similar to and contemporary with Rusellae, as well as 

substantial textual and archaeological evidence of long-distance trade involving Africa and Italy, 

there is reason to believe that Rusellae might have been the home of individuals who show 

greater craniofacial affinity to African reference samples as compared with European reference 

samples. 

Research Question 3: Is there evidence of differential life experience, evaluated using skeletal 

stress indicators, between individuals with African affinity as compared with individuals with 

European affinity in Phase II? 

Hypothesis 3: I expect to find similar frequencies of stress indicators in individuals with African 

affinity when compared to individuals with European affinity in Phase II. 

The relevant literature suggests that acculturation was a critical part of maintaining 

cohesion in the Roman and post-Roman world. Archaeological and historical evidence suggests 

that the Ostrogoths, who reigned in Italy during the early 6th century AD, took great care to 

maintain and imitate many features of the Roman empire, including government and architecture 

in an effort to establish their legitimacy (Brogiolo et al., 2017). The Lombards, who ruled Italy in 

the late 6th century AD and early 7th century AD, likely also assimilated with the native people, 

invading so rapidly that the lives of the Roman peasants they encountered were not drastically 

84 

 
changed (Holmes, 1988). Therefore, I anticipate that individuals who lived together and were 

buried in the same cemetery likely experienced similar physiological stressors in life. 

Research Question 4: Do detectable craniofacial differences exist between Phase I (6th-7th 

centuries AD) and Phase II (8th-12th centuries AD) burials at Rusellae, suggesting either shifting 

population demographics or relative population continuity? Further, do differences exist in the 

frequency of cribra orbitalia, porotic hyperostosis, periosteal reaction, and linear enamel 

hypoplasia between phases, suggesting changing environmental stressors, diet, disease, or other 

cultural factors over time? 

Hypothesis 4: I do not expect to find significant craniofacial and health differences between 

individuals interred during Phase I and Phase II at Rusellae. 

I expect to find minimal craniofacial and stress indicator differences between Phase I and 

Phase II individuals, representing general population and health stability over time at Rusellae. 

On a regional scale, the 6th-7th centuries AD (Phase I) represent a turning point in Tuscany, 

between Roman urban structure and the Early Medieval layout of the landscape. During this 

period, many contemporary sites and workshops were abandoned and never reoccupied 

(Sebastiani, 2017). Rusellae, however, maintained its occupation during what was otherwise a 

period of regional upheaval, indicating a degree of population continuity and socioeconomic 

stability within the site. In fact, when many lowland and rural sites were being abandoned, 

people concentrated in fortified, nucleated hilltop villages such as Rusellae (Francovich, 2008). 

Therefore, I expect Rusellae to represent a relatively diverse population comprising people from 

throughout the region, as well as potential central European migrants–such as the Lombards–or 

their descendants during Phase I. 

85 

 
During Phase II, from roughly the 8th-12th centuries AD, Rusellae maintained its 

importance as a craft production site, and its occupants continued construction and 

improvements on the church and cemetery complex, even constructing a tower near the 

abandonment of the site (Celuzza et al., 2002; Sebastiani, 2017; Sebastiani et al., 2015). This 

suggests a degree of socioeconomic stability at Rusellae, despite a transition in power from 

Lombard rule to Carolingian aristocratic rule during Phase II. 

Further, changes in burial practices, from earlier rock-lined tombs with grave goods to 

earthen inhumations without decoration, are consistent with a shift in religious practice from pre-

Christian to Christian burial traditions. Although some grave goods from Phase I suggest the 

presence of a high-status, non-local Lombard individual, the overwhelming majority of graves 

are consistent with the expected tradition for Romans throughout time. Since the distinction in 

burial type is consistent with expectations for change over time and not with an invading group 

or massive cultural shift, I expect the inhabitants to represent a diverse, but relatively stable, 

population. 

Given that Rusellae was a fortified hilltop site throughout Phases I and II with continued 

construction and increasing craft production, especially during and after a Tuscan demographic 

decline, I expect that its inhabitants experienced some degree of overall population continuity 

and were also buffered from some of the diet and environmental stressors that may have affected 

smaller, isolated, lowland settlements. Therefore, I do not expect to find significant differences 

in craniometrics or stress indicators between Phase I and Phase II individuals. 

Now that the research questions have been presented, the next chapter will discuss in 

detail the materials and methods used to answer them. 

86 

 
 
 
 
CHAPTER FIVE: MATERIALS AND METHODS 

This chapter introduces the analytical and reference skeletal samples that are used in this 

dissertation. Additionally, this chapter describes each of the different types of data, data 

collection protocols, and analytical methods utilized in this study. The first section details the 

Rusellae skeletal sample, for which novel craniometric and skeletal stress data were collected for 

this dissertation. This section also discusses features of the Rusellae skeletal sample and the 

cemetery from which it was derived, including demographic information, burial chronology, and 

general layout of the cemetery. Further, the first section describes the comparative reference 

samples used in this study, including Howells’ samples of African and European crania and 

samples from Mis Island in modern-day Sudan. The second section in this chapter describes the 

types of data and data collection methods used in this dissertation, including sex and age 

indicators, craniometric data, and skeletal health data. Lastly, this chapter details the analytical 

methods used to answer the previously outlined research questions. 

The Skeletal Samples 

The materials for the present research include the skeletal sample from Rusellae, 

Tuscany, Italy, as well as three European and three African craniometric reference samples. 

Rusellae Sample 

The overall Rusellae skeletal sample represents all individuals excavated from the 

Rusellae cemetery for which age and sex data could be ascertained. A total of 202 individuals 

excavated from Rusellae were sufficiently complete to permit estimation of age and sex. The 

assemblage consists of 42 subadults (20.8%) and 160 adults (79.2%). Subadults were further 

divided into age cohorts as follows: 16 infants (7.9%), 17 children (8.4%), and 9 adolescents 

(4.5%). Adults were subdivided into the following age cohorts: 42 young adults (20.8%), 71 

87 

 
 
middle adults (35.1%), 18 old adults (8.9%), and 29 adults over 20 years (14.4%). Adult 

individuals were further subdivided into males and females. Of the 160 adults in the sample, 72 

are female (45%), 81 are male (50.6%), and 7 are of indeterminate sex (4.4%). Table 5.1 outlines 

the demographic breakdown of the cemetery. 

Table 5.1. Demographic breakdown of Rusellae sample 

Male  Female  Indeterminate  Total 

Infant (0-3 years) 
Child (3-12 years) 
Adolescent (12-20 years) 
Young Adult (20-35 years) 
Middle Adult (35-50 years) 
Old Adult (>50 years) 
Adult (>20 years) 
Total 

0 
0 
0 
19 
43 
8 
11 
81 

0 
0 
0 
23 
24 
10 
15 
72 

16 
17 
9 
0 
4 
0 
3 
49 

16 
17 
9 
42 
71 
18 
29 
202 

The cemetery at Rusellae was in use between the 6th and 12th centuries AD and comprises 

two general phases of use (Celuzza et al., 2023). Phase determinations were initially made by the 

archaeologists who excavated the site, with additional data added recently through radiocarbon 

dating of several bone samples (Table 5.2). 

The earlier burials, designated Phase I, date to the sixth and seventh centuries AD based 

on grave inclusions. The Phase I graves are generally orderly, laid out in a northwest orientation, 

and tend to be constructed from stone. These graves roughly correspond to the “Lombard 

Period,” or the time during which Germanic tribes moved from Central Europe into portions of 

Italy, including Rusellae. 

The later burials, designated Phase II, date from the 8th century AD until the 

abandonment of the site in the 12th century AD. These burials are less organized than the Phase I 

burials and include several instances of secondary burials and reuse of graves. Further, the Phase 

88 

 
 
 
II burials are generally lacking in grave goods, which is consistent with the Christian burial 

tradition of the time (Celuzza et al., 2002). 

Phase data was available for 138 individuals; 27 individuals were dated to Phase I 

(19.6%) and 111 individuals were dated to Phase II (80.4%). 

Table 5.2. Skeletal samples and their associated radiocarbon dates 

Skeleton 
1174 
1185 
1216 
1219 
1220 
3514 
1263 
1081 
1091 
1094 
5103 
5102 
5523 
2006 
2009 
1103 

Dates (AD) 
993-1175 
990-1171 
820-992 
1020-1165 
976-1163 
1022-1175 
1122-1273 
1033-1216 
978-1159 
1155-1279 
578-679 
771-992 
661-777 
890-1048 
771-977 
1038-1169 

Howells’ Datasets 

Phase (Radiocarbon and Archaeology) 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 2 
Phase 1 
Phase 1 
Phase 1 
Phase 2 
Phase 2 
Phase 2 

Several reference samples from W. W. Howells’ craniometric dataset (1973, 1989) are 

also included in this study in order to situate the Rusellae sample within a broader global 

context.  W.W. Howells collected craniometric data for a variety of samples across the globe in 

order to analyze the degree of craniofacial variation among and between groups around the world 

(1973). The four subsamples of the Howells dataset used in this dissertation are the Berg, 

Zalavar, and Norse samples from Europe and the Dogon sample from Africa (Howells, 1973). 

The specific makeup of each skeletal sample and its relevance to the present study are outlined 

below and in Table 5.3. The geographic locations of each sample are illustrated in Figure 5.1. 

89 

 
 
 
Africa: Dogon (West Africa) 

The Dogon sample comprises 48 males and 53 females from the burial caves of the 

Dogon tribe in Mali and dating to AD 1055-1750 (Howells, 1973; Willett, 1971). The sample 

was collected by Professor Marcel Griaule in 1934 and comprises an earlier and later phase 

which are pooled into one sample due to relatively little difference in cranial form between the 

groups. Howells estimated sex for these individuals using cranial morphology during the process 

of measurement (1973). 

The Dogon sample was selected for use as a reference due to its origin in northwest 

Africa and its temporal overlap with portions of the Rusellae cemetery. Although Mali is not 

specifically a hypothesized origin of individuals at Rusellae, Howells notes that this sample is a 

generally good representation of a West African series (1973), and the sample provides a good 

temporal comparison; therefore, the Dogon sample is an effective reference group for this 

dissertation. 

Europe: Medieval Norse (Northern Europe) 

The Medieval Norse sample is a collection of 54 male and 54 female crania from several 

medieval parish graveyards in the vicinity of Oslo, Norway. According to Howells, these people 

were relatively isolated and the sample likely includes primarily local individuals, representing a 

relatively homogenous sample (1973, 1989). In fact, 85% of the sample derives from one 

cemetery near St. Nicolaus Church (Howells, 1989). Howells does not specify the dates of this 

sample beyond classifying it as “Medieval.” Sex estimation was based on cranial and postcranial 

skeletal traits, when present (Howells, 1989). 

This sample was selected for use as a reference in this dissertation due to both its time 

period and its location. Despite the lack of specific dates, the Medieval era is generally 

90 

 
 
 
 
considered to span between AD 500 and AD 1500, which is concurrent with the cemetery at 

Rusellae. Further, Norse incursions are historically documented in Tuscany in the late 9th 

century (Haywood, 2015), providing evidence for at least some degree of contact between 

Medieval Norway and Medieval Italy. Due to the overlapping time frame and a possible 

historical connection between the reference region and the study region, the Norse sample 

provides an effective comparative group. 

Europe: Zalavar (Central Europe) 

The Zalavar sample comprises 53 males and 45 females and comes from the cemeteries 

at Zalavar in western Hungary. The remains date from the 9th to 11th centuries AD and represent 

a heterogeneous population, comprising Avars, Romanized and newly arrived Germanic and 

Slavic individuals, and a few of the earliest Magyars. Howells estimated sex for this sample 

using traits of the postcranial skeleton (Howells, 1973, 1989). 

The Zalavar sample was selected for comparison in this dissertation due to its time frame 

and ancestral makeup. In fact, the aforementioned ethnic groups are among the Germanic tribes 

that are historically attested to have been present in Italy during the time period under 

investigation in this dissertation. Therefore, the Zalavar reference group is the most similar 

sample to the somewhat nebulous “Lombard” people who are known to have occupied Rusellae 

for at least a century, beginning in the late 6th century (Sebastiani et al., 2015), making this a 

valuable reference sample. 

Europe: Berg (Central Europe) 

The Berg sample is made up of 56 males and 53 females from a small mountain village in 

Carinthia, Austria. The remains were recovered from a charnel house in an isolated village and 

contain almost all the crania from the village’s population over the course of five generations. 

91 

 
 
 
Therefore, the dates for this sample are challenging to reconstruct, although they are likely to 

range from AD 1700 to AD 1900. Sex estimation in this sample was particularly challenging, 

according to Howells, and sex was solely determined based on visual traits of the skulls 

(Howells, 1973). 

The Berg sample was selected for comparison to Rusellae primarily due to its geographic 

origin. Although the Berg sample is later in time than the Rusellae cemetery, it represents the 

nearest geographic reference sample to Rusellae. In fact, the modern-day region of Carinthia 

borders the northeast Italian region, Friuli-Venezia Giulia. Notably, Carinthia would have been 

part of the Roman Empire and its occupants would have been considered “Romans” until the 

region fell to the Ostrogoths in the late 5th century (Arnold, Bjornlie, & Sessa, 2016). Therefore, 

the sociopolitical trajectory of this region mimics that of Rusellae, making the Berg sample an 

interesting comparative group in this study. 

Nubian Datasets 

In order to develop a more temporally and geographically representative reference 

sample, two medieval Nubian datasets are also used for comparative purposes. Craniometric data 

collected by Dr. Jennifer Vollner for individuals from Mis Island cemeteries 3-J-10 and 3-J-11, 

located in present-day Sudan, are used to bolster the reference sample with North African 

individuals corresponding to the Late Antique and Early Medieval periods (Figure 5.1, Table 

5.3) (Vollner, 2016). The samples contain adult individuals with no apparent cranial pathology or 

cranial modification. 

Mis Island is located upstream from the fourth cataract of the Nile River in present-day 

Sudan. The individuals from Mis Island were interred in two cemeteries from the Christian 

period, 3-J-10 and 3-J-11. Cemetery 3-J-10 was in use from AD 1100-1500 and contained 13 

92 

 
 
 
males and 11 females with sufficiently complete crania for analysis, totaling 24 individuals. 

Cemetery 3-J-11 was in use from AD 300-1400 and was much larger than 3-J-10, containing 33 

males and 42 females with sufficiently complete crania, totaling 75 individuals. Sex was 

estimated for this sample using pelvic morphology (Vollner, 2016). 

The two Mis Island cemetery samples were selected for comparison in this dissertation 

due to their temporal and geographic parameters. Their location in northeast Africa provides a 

generally good representation of the possible origin of African migrants to Tuscany. Further, the 

time periods of the two cemeteries together entirely overlap with the cemetery at Rusellae, 

creating a good temporal comparative sample. Notably, the two cemeteries will be treated 

separately in this dissertation due to their distinct time frames, with Cemetery 3-J-11 having a 

longer and earlier span of usage compared to Cemetery 3-J-10. 

Table 5.3. Craniometric reference samples 

Group 

Continent 

Region 

Time Period 

Male  Female 

Berg 

Europe 

Carinthia, 
Austria 

Unknown; likely 
AD 1700-1900 

Norse 

Europe 

Oslo, Norway 

Medieval 

56 

53 

54 

53 

54 

45 

AD 800-1000 

AD 1055-1750 

47 

52 

AD 1100-1500 

13 

11 

AD 300-1400 

33 

42 

Zalavar 

Europe 

Dogon 

Africa 

Mis Island 3-J-10 

Africa 

Mis Island 3-J-11 

Africa 

Zalavar, 
Hungary 

Mali, West 
Africa 

Nubia, 
Northeast 
Africa 

Nubia, 
Northeast 
Africa 

93 

 
 
 
 
NORSE 

BERG 

ZALAVAR 

RUSELLAE 

DOGON 

MIS ISLAND 

Figure 5.1. Map of craniometric reference samples. 

94 

 
 
Sample Limitations 

Bioarchaeological skeletal samples necessarily differ from modern skeletal samples in a 

number of ways that need to be addressed. The cemetery sample is a result of decisions made by 

the living, including variables such as how the corpses are disposed (inhumation vs. cremation), 

whether the burials are temporary or permanent, or where in the cemetery people can be buried 

(Weiss‐Krejci, 2011). For example, in Medieval Italy females tend to be underrepresented in 

cemeteries compared to the Roman period and Late Antiquity (Barbiera, Castiglione, & Dalla-

Zuanna, 2017). This may be the result of different burial practices for females, such as 

inhumation in a separate part of the cemetery or burial directly in the ground instead of in a rock-

lined tomb. Similarly, in the present study, individuals from Phase I are particularly sparse. This 

is due in part to the fact that many of the Phase I graves were re-used during Phase II, and the 

remains originally interred were commingled and reburied, meaning that fewer Phase I 

“individuals” were preserved (Celuzza et al., 2002). 

Following deposition of the remains, taphonomic processes can either preserve or destroy 

the quality of the osseous material. Factors such as soil acidity, burial environment, exhumation 

and reburial, and construction can all influence the preservation of the remains, which in turn 

affects the depth of analysis that is possible. As a result of these taphonomic processes, skeletal 

samples are regularly fragmentary. Skeletal trauma or disease near the time of death may also 

make the skeleton more fragile, resulting in poor preservation. Curation or handling during 

analysis further increases the possibility of postmortem damage. This fragmentation can 

substantially restrict the number and type of cranial landmarks available for study. This directly 

impacts the granularity with which phenotypic, and subsequently genotypic, variation can be 

understood. 

95 

 
Sample bias is also introduced during excavation of the remains, where the research 

interests, time, money, or political constraints of the archaeologists dictates what gets excavated, 

what gets recorded, and how it is curated. This can lead to incomplete cemetery recoveries or a 

focus on monumental burials instead of cemeteries containing common folks. Although 

examining elite burials is important and interesting, they are often not representative of the true 

lived experience for most people in a particular context. Further, the interests of Italian 

archaeologists and classicists have not always involved the human skeleton (Gowland, 2017; 

Killgrove, 2005). In some of the most egregious cases, grave goods were carefully excavated, 

while human remains were subsequently disposed of without further analysis. Fortunately, this 

practice is no longer legal. Excavations rarely included skeletal analysis beyond reporting body 

position within a grave. When excavated, these skeletons were often subject to poor curation 

with limited associated contextual information, making future research challenging (Killgrove, 

2005). Working with skeletal collections that were excavated decades earlier is challenging due 

to the loss of taphonomic and archaeological context (Nikita, 2024). 

When it comes to a statistical analysis of the remains, a sampling strategy is typically 

used to ensure that the sample under study is representative of the site or population at large. It is 

this representativeness that allows for larger conclusions to be drawn from a smaller sample 

group. When working with skeletal samples, researchers do not always have the luxury of 

randomly selecting specimens from a larger population. Skeletal samples are often relatively 

small, requiring the use of the complete, excavated assemblage as opposed to a statistically-

derived study sample. The results of such investigations are informative, but they may be limited 

in their scope and generalizability. 

96 

 
Another avenue through which bias is introduced in the Rusellae sample is temporal 

aggregation, which refers to the grouping of individuals who lived over a broad time scale and 

considering them to be one population. Individuals who lived several hundred years apart (such 

as those in the earlier vs. later part of Phase II) would have likely led quite different lives. In this 

study, however, they are considered to represent one time period. This approach subsequently 

masks some of the variation inherent within each phase by lumping together several generations 

into one group. 

Each of the aforementioned issues with bioarchaeological samples contributes to a trend 

of small sample sizes in bioarchaeological research. Although limited samples should not deter 

research, there are considerations or concessions that need to be made. In some cases, when data 

are too sparse to allow for a statistical analysis, an osteobiography approach can be used to 

examine the lived experience of a single individual (Hosek & Robb, 2019; Robb, 2024). Specific 

statistical methods for small samples, such as the Fisher’s exact test and data imputation 

methods, can be used to maximize the sample and apply statistical rigor (Buikstra et al., 2022). 

Overall, sample limitations must be acknowledged, and results presented with the appropriate 

caveats. 

Data Collection Methods 

Several types of data were collected or generously provided by colleagues for the present 

research: skeletal indicators of sex, age, and stress, and cranial measurements. All possible data 

points were collected for each skeleton, depending on the level of completeness and preservation 

in each case. The data collection procedures for each type of data are described below. 

97 

 
 
 
 
Sex Estimation 

For this dissertation, sex was estimated by morphological analyses of the pelvis and skull, 

as well as measurements of the postcranium. The three primary pelvic features used in sex 

estimation are the ventral arc, subpubic concavity, and the medial aspect of the ischiopubic 

ramus, which can be used to estimate sex with approximately 95% accuracy (Phenice, 1969). 

These three traits, in addition to the morphology of the greater sciatic notch and preauricular 

sulcus, were scored using the system outlined in Buikstra and Ubelaker (1994). Generally, the 

presence of a ventral arc, subpubic concavity, a narrow ischiopubic ramus, and a wide greater 

sciatic notch are consistent with female, while the lack of a ventral arc, no subpubic concavity, a 

broad ischiopubic ramus, and a narrow greater sciatic notch are consistent with male. 

Similarly to the pelvis, the skull also exhibits sexual dimorphism in several features, five 

of which are traditionally used in sex estimation: the mental eminence, orbital margin, glabellar 

area, nuchal area, and mastoid process. These traits were also scored following Buikstra and 

Ubelaker (1994), with a more gracile mental eminence, glabella, nuchal area, mastoid process, 

and sharp orbital margins scored as consistent with female. Conversely, a robust mental 

eminence, glabella, nuchal area, mastoid process, and blunt orbital margins are consistent with 

male. 

Finally, when present, the humeral and femoral heads were measured using spreading 

calipers and the measurements were recorded in millimeters. These measurements were 

compared with Stewart’s sectioning points for sex estimation (1979). For the humeral head, a 

vertical diameter less than 43 mm indicates female, 43-44 mm indicates probable female, 45 mm 

is indeterminate, 46-47 mm indicates probable male, and greater than 47 mm indicates male. For 

the femoral head, a maximum diameter less than 42.5 mm indicates female, 42.5-43.5 mm 

98 

 
 
indicates probable female, 43.5-46.5 is indeterminate, 46.5-47.5 mm indicates probable male, 

and greater than 47.5 mm indicates male (Stewart, 1979). 

When present and observable, morphological features of the pelvis were prioritized in sex 

estimation. However, all the aforementioned features were used in combination to assess sex 

overall, using the categories of undetermined, female, probable female, ambiguous, probable 

male, and male. For the purposes of this dissertation, the categories of female and probable 

female are lumped together as female, probable male and male are lumped together as male, and 

individuals with undetermined and ambiguous sex are not included in the sex-specific analyses 

of this study. 

Age Estimation 

Adult age is estimated using predictable degenerative changes in the skeleton that can be 

correlated to chronological age. The primary methods of adult age estimation in this dissertation 

include the fusion of the medial clavicle and the morphologies of the pubic symphysis, auricular 

surface, and sternal end of the rib. 

The fusion of the medial clavicle epiphyses, when observable, is used as an indicator of 

adult versus subadult age. That is, when the medial clavicle epiphysis is fused, the individual is 

considered to have reached skeletal maturity, which allows for the use of the following sex 

estimation methods. This epiphysis generally fuses between the ages of 18 and 32 years 

(Buikstra & Ubelaker, 1994). 

Features of the face of the pubis bone were utilized for age estimation in the adult 

skeleton. With time, the pubic symphysis transitions from a sturdy, billowing surface with little 

delineation of the face, to a flat, finely grained face with a complete rim, and finally to a 

depressed, porous face with lipping and other irregularities (Suchey & Katz, 1998). The Suchey-

99 

 
Brooks method, which outlines six phases of changes in pubic symphysis morphology, employs 

descriptions and casts to aid practitioners in its use. Males and females were analyzed separately, 

as they undergo age-related changes at different rates and exhibit some different features. 

Similarly to the pubic symphysis, the morphology of the auricular surface of the ilium 

was also used to estimate adult age. Over time, the auricular surface undergoes a transition from 

a youthful, billowy surface to a poorly-organized, irregular surface with marked apical activity. 

Features evaluated for aging include the transverse organization, surface texture, microporosity, 

macroporosity, and apical changes of the auricular surface. For the present research, these 

features are taken in combination and scored using a phase method to estimate age (Lovejoy, 

Meindl, Pryzbeck, & Mensforth, 1985). As with the pubic symphysis, males and females were 

analyzed separately as the age-related changes vary slightly by sex. 

The sternal rib end also undergoes predictable changes with age, transitioning from a 

relatively flat and billowy surface to a thin-walled surface with irregular projections. Ribs were 

sorted into phases from 0-8, each corresponding to an age range with an associated 95% 

confidence interval, based on traits such as scalloping of the margins, the shape and depth of the 

pit at the sternal end, and the presence of any irregular claw-like projections (Iscan, Loth, & 

Wright, 1984, 1985). Males and females were scored separately in this method as well, as they 

undergo these changes along a slightly different time scale. 

After using all possible methods to estimate age, a consolidated age estimate was 

produced using the data gleaned from each method. The pubic symphysis data was weighed most 

heavily, when present, and the age range was adjusted as necessary using data from the auricular 

surface and sternal rib ends. After the age range was established, each individual was placed into 

an age category as follows: young adult (20-35 years), middle adult (35-50 years), old adult (50+ 

100 

 
years). If age indicators are limited due to lack of skeletal completeness or poor preservation, a 

broader age category of subadult (< 20 years) or adult (> 20 years) was assigned. 

Craniometrics 

Coordinate data and interlandmark distances were recorded for each cranium using a 

Microscribe 3D digitizer and the program 3Skull (Ousley, 2014). Cranial landmarks are defined 

points that are theoretically homologous across all individuals, allowing for a comparison of 

landmark locations across several samples. This study utilized the 108 landmarks prompted in 

the 3Skull software package and defined in the Harris County Standard Operating Procedure 

(Fleischman & Crowder, 2018). All possible landmark data were recorded for each cranium, 

even if the cranium was incomplete. 

Coordinate data were collected using a Microscribe 3D digitizer, which records x, y, and 

z coordinates for each landmark. Before a cranium is digitized, several instrumentally-

determined landmarks are measured with sliding or spreading calipers and marked with a pencil, 

including ectoconchion, upper orbital border, cheek height superior point, cheek height inferior 

point, zygion, eurion, frontotemporale, maximum frontal point, glabella, and ectomolare. 

Additionally, several points are also marked for ease of identification during the digitizing 

process, including sphenion, krotaphion, stephanion, asterion, opisthion, and basion. After 

locating and marking these landmarks, the cranium is set on three clay pedestals and secured to 

prevent any movement during the digitizing process. The 3Skull software prompts the analyst to 

collect each landmark, at which point the Microscribe’s stylus is placed on the landmark and a 

foot pedal is depressed to collect the coordinate data. If a landmark is not present due to 

pathology or poor preservation, a note is recorded in 3Skull and data for that landmark is not 

recorded. 

101 

 
Finally, these coordinates are imported through the 3Skull software program into a 

Microsoft Excel spreadsheet. 3Skull also calculates interlandmark distances, or the distance 

between two landmarks in space, from these coordinates; these data are recorded in millimeters 

and exported to a separate Microsoft Excel spreadsheet. 

Skeletal Stress Indicators 

Linear Enamel Hypoplasia 

All present teeth were observed for evidence of linear enamel hypoplasias. When 

necessary, directed lighting was used to facilitate observation of enamel defects. These traits 

were scored according to Buikstra and Ubelaker (1994). If present, the type of defect was scored 

as either linear horizontal grooves, linear vertical grooves, linear horizontal pits, a nonlinear 

array of pits, or single pits. The location of each lesion was then measured in millimeters from 

the cemento-enamel junction using digital calipers. The defects were recorded by tooth for each 

tooth present. 

Skeletal Porosities 

Porosities of the cranium, including porotic hyperostosis and cribra orbitalia, were 

identified and scored following the standardized scoring system outlined in Buikstra and 

Ubelaker (1994). When observed, these porosities were scored in terms of expression and 

activity. The expression of the observed lesions was characterized as either barely 

discernible/very indistinct porosity, true porosity/porosity only, coalescing pores, or coalescing 

pores and expansive changes. Activity was scored as either active, healed, or a mixed reaction of 

the two. Specifically, active lesions exhibit sharp lesion margins and a sieve-like appearance, 

while healed lesions exhibit a more smooth texture with evidence of infilling of the pores 

(Mensforth, Lovejoy, Lallo, & Armelagos, 1978). 

102 

 
 
 
Periosteal Reactions 

The long bones were observed for evidence of periosteal reactions and data were 

recorded regarding the presence, location, extent, and vascularization of the lesions, as outlined 

in Cook (1976) and Weston (2008). The reactions were also characterized as focal vs. diffuse 

and woven vs. sclerotic vs. mixed. The extent of the reaction was scored as either normal, 

isolated on <⅓ of the surface, isolated on ⅓-⅔ of the surface, isolated on > ⅔ of the surface, or > 

⅔ of the surface with elevation > 2-3 mm. The vascularization was scored as normal, multiple 

small striae, multiple small foramina, multiple large striae, multiple large foramina, or mixed 

abnormal vascularization. Further, focal reactions exhibited clearly defined boundaries, while 

diffuse reactions occupied a larger area with less clear boundaries. Finally, woven bone appears 

fibrous and unconsolidated, while sclerotic bone appears as woven bone that has been filled in, 

exhibiting a more dense texture (Cook, 1976; Weston, 2008). 

Analytical Methods 

Research Question 1 

The first research question seeks to understand the craniometric variation at Rusellae on 

two levels: between Rusellae and geographically and temporally relevant craniometric reference 

samples and within Rusellae as a whole. Each of these levels of variation will be explored using 

specific statistical methods, as outlined below. All analyses for this research question are 

performed in R (v. 4.4.1), a freely-available open-source programming language used primarily 

for statistical computing and graphics (R Core Team, 2024). 

Twenty-one craniometric variables were selected for analysis. These variables were 

selected because a) they correlate well with genetic data (Relethford, 1994); b) they capture the 

morphology of the midface and the cranial vault; c) they are commonly used in forensic 

103 

 
 
 
anthropology population affinity estimation as well as biodistance analyses; d) they have 

standardized definitions and measurement protocols (Fleischman & Crowder, 2018); and e) they 

maximize the sample size in the present study as they are better preserved than other variables. 

The twenty-one variables under study are listed in Table 5.4. Measurements and abbreviations 

are adapted from Fleischman and Crowder (2018). 

Table 5.4. Craniometric variables used in Research Question 1 

Abbreviation 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

Measurement 
Maximum cranial length 
Basion-nasion length 
Basion-bregma height 
Maximum cranial breadth 
Bizygomatic breadth 
Biauricular breadth 
Biasterionic breadth 
Basion-prosthion length 
Nasion-prosthion height 
Nasal height 
Nasal breadth 
Mastoid height 
Orbital height 
Orbital breadth 
Interorbital breadth 
Bimaxillary breadth 
Biorbital breadth 
Frontal chord 
Parietal chord 
Occipital chord 
Foramen magnum length 

Prior to analysis, all datasets included in both the reference and study sample must be 

complete. That is, there can be no missing measurements in any of the individuals for any of the 

21 variables under analysis. Although the Howells reference data is complete, the Rusellae study 

sample and the Mis Island 3-J-10 and Mis Island 3-J-11 cemeteries include individuals with an 

incomplete set of observations. This fragmentation is common in bioarchaeological samples as a 

result of poor preservation or postmortem damage caused by taphonomic processes or during 

104 

 
 
excavation, curation, or analysis. Therefore, data imputation will be used to complete these 

datasets. Kenyhercz and Passalacqua (2016) showed that accurate classifications can still be 

achieved when up to 50% of data is missing. Consequently, data is imputed for individuals 

missing ten or fewer data points out of the 21 total craniometric variables used in the study. 

Missing data for the Rusellae and Mis Island samples are imputed using the Multivariate 

Imputation by Chained Equations (MICE) method using the ‘mice’ package in R (Van Buuren & 

Groothuis-Oudshoorn, 2011). The MICE method is effective for imputing incomplete 

multivariate data simultaneously on a variable-by-variable basis (Van Buuren & Groothuis-

Oudshoorn, 2011). Each dataset is imputed separately by population (e.g. each Mis Island 

cemetery and the Rusellae sample were treated separately); this conservative strategy is 

recommended by Kenyhercz and Passalacqua (2016) to minimize the difference between 

imputed and actual values that may occur by pooling datasets prior to imputation. The specific 

method of imputation is ‘predictive mean matching’, which evaluates the range of realistic 

values for each variable and fills in missing data while preserving the distribution of the original 

dataset (Van Buuren & Groothuis-Oudshoorn, 2011). This analysis employed 50 iterations to 

produce five imputed datasets. The first imputed dataset was selected for further analysis for all 

three imputed datasets. 

Following imputation, all seven groups used in this analysis are combined: Berg, Zalavar, 

Norse, Dogon, Mis Island 3-J-10, Mis Island 3-J-11, and Rusellae. Then, the data are 

standardized by scaling and centering. Scaling and centering sets the mean of each variable to 

zero and the standard deviation to one, which prevents variables with large magnitudes (in this 

case, larger size) and variables with large ranges (as in the range between male and female 

105 

 
 
values) from biasing the results (Kabacoff, 2022). This approach allows males and females to be 

pooled in order to maximize sample size. 

In order to compare the overall craniometric variation at Rusellae to temporally and 

geographically relevant reference samples, a Mahalanobis distance (MD) statistic is employed, 

followed by a Principal Components Analysis (PCA) to visualize intergroup distances in 

multivariate space. The Mahalanobis distance can be employed in both the original variable and 

the Principal Components (PC) spaces; here, the MD is applied to the original scaled data. The 

MD is a distance between a point and a distribution; one of the most commonly-used distance 

measures in modern research, MD outperforms simple Euclidean distance as it is a model-free 

approach that takes into account the variance within the dataset and the correlations between 

variables, such as those between interlandmark distances of the cranium (De Maesschalck, 

Jouan-Rimbaud, & Massart, 2000; Relethford, 2016). Mahalanobis distance using cranial 

measurements correlates with genetic distances in studies using matched datasets (Relethford, 

2004). The present analysis measures similarity and dissimilarity between the groups in order to 

understand the degree of relatedness among the samples. For this analysis, the Mahalanobis 

distance between all group centroids is calculated. Permutation testing is used to test for 

significant differences. Then, a PCA is performed on the complete dataset and group centroids 

are plotted against PC1 and PC2 to visualize group relatedness in multivariate space. 

In order to understand within-group variation at Rusellae, a Principal Components 

Analysis (PCA) is employed on the Rusellae craniometric data alone. Like the Mahalanobis 

Distance analysis, PCA requires a complete dataset; therefore, the imputed Rusellae dataset 

produced using the MICE method is analyzed here. The data are again scaled and centered to 

allow for pooling of sexes. 

106 

 
PCA is an analytical method that reduces dimensionality in multivariate data to maximize 

the variance within the data. PCA takes a large number of correlated variables (in this case, 

interlandmark distances) and transforms them into a smaller and more manageable set of 

uncorrelated variables, known as principal components (PCs). The output of this analysis is a 

series of PCs, ranked in order of how much variation they capture from the original dataset. 

Typically, the first few PCs explain a large proportion of data variation, which allows for 

dimensionality reduction while maintaining as much variance in the dataset as possible 

(Kabacoff, 2022). Further, PCA does not require a grouping variable; therefore, all individuals at 

Rusellae can be evaluated against one another without any a priori assumptions about group 

membership (sex, age, ancestry, phase, etc.). 

The PCA is performed in R version 4.4.1 using the prcomp function in base R (R Core 

Team, 2024). The PCA is performed on the scaled dataset and the principal components are 

extracted. The PCs with an eigenvalue greater than 1.0 are retained, per the Kaiser-Harris 

criterion; eigenvalues greater than 1.0 suggest the principal component explains more variance 

than a single variable alone (Kabacoff, 2022). The results of this analysis quantify combinations 

of craniometric variables most significantly driving the craniometric variation within the 

cemetery at Rusellae. 

In order to further understand the within-cemetery craniometric variation, a hierarchical 

cluster analysis is performed on the retained PCs with eigenvalues greater than 1.0. The resulting 

dendrogram is visualized and the tree is cut at the number of clusters established from the 

dendrogram. Finally, a scatterplot is produced of individuals in the cemetery based on the first 

two PCs to visualize the clusters in multivariate space. 

107 

 
 
 
 
Research Question 2 

The second research question aims to understand whether individuals of African ancestry 

might have been present at Rusellae throughout the occupation of the cemetery. This question 

will be addressed using linear discriminant analysis (LDA). LDA is a form of classification 

statistics that relies on building a model using a known reference sample, followed by classifying 

unknowns into one of the reference groups (Zhao, Zhang, Yang, Zhou, & Xu, 2024). All 

analyses for this research question are performed in R (v. 4.4.1) (R Core Team, 2024). 

The same 21 craniometric variables from Research Question 1 are also used in this 

analysis. As with MD and PCA, LDA requires complete datasets; therefore, the Rusellae and 

Mis Island datasets were again completed using the ‘mice’ package. Following data imputation, 

the two Mis Island cemetery reference datasets are added to the four Howells datasets (Berg, 

Norse, Zalavar, and Dogon) to produce the overall reference sample used in this analysis. Then, 

each dataset (the combined reference sample and the Rusellae sample) is scaled and centered in 

order to minimize the effects of size and to allow for pooling of sexes. 

After the reference and study datasets are clean, a linear discriminant analysis (LDA) is 

performed. LDA is a form of discriminant function analysis that maximizes between-group 

variance and minimizes within-group variance by producing the linear combination of variables, 

in this case cranial measurements, that best separates each group (Zhao et al., 2024). The LDA is 

performed using the MASS package in R (Venables & Ripley, 2002), where the model is built to 

classify observations into one of the six reference groups based on the 21 craniometric variables 

in the dataset. Effectively, the reference groups act as a training sample used to generate the 

model. The model applies equal prior probabilities to each of the six groups in the reference 

dataset. The overall correct classification rate of the model is calculated, as well as the correct 

108 

 
 
classification rate for each group. Then, the model is used to make predictions about group 

membership for the individuals in the unknown dataset; here, the Rusellae individuals act as the 

holdout or testing set for the model. Therefore, the model is cross-validated using separate 

testing and training datasets. 

Research Question 3 

The third research question seeks to understand whether skeletal stress differed between 

individuals of estimated African and European ancestry. In order to evaluate this question, the 

frequency of each non-specific stress indicator is calculated by dividing the number of instances 

of each indicator by the total number of observable instances in the assemblage. Both affinity 

groups are divided by sex and differences in frequency are compared between males and females 

using a Fisher’s exact test. If no significant difference is found between males and females 

within each affinity group, the sexes are pooled and frequencies are compared between African 

and European affinity, again using a Fisher’s exact test. 

A Fisher’s exact test assesses the independence of categorical variables, especially in 

cases where presence or absence counts for a given variable are less than five (Kim, 2017). This 

statistical method is used to understand whether there is a relationship between any health/stress 

indicators (porotic hyperostosis, periosteal reaction, cribra orbitalia, and LEH) and estimated 

affinity or whether the frequency of a stress indicator is independent of affinity. 

Research Question 4 

In order to examine craniofacial variation over time at Rusellae, individuals from Phase I 

are compared with individuals from Phase II with respect to cranial measurements. This is 

achieved through a multivariate analysis of variance (MANOVA) performed on a standardized 

craniometric dataset grouped by phase. MANOVAs test for the significance of group differences 

109 

 
 
over multiple variables at once (Kabacoff, 2022). All analyses for this research question are 

performed in R (v. 4.4.1) (R Core Team, 2024). 

Since phase data is only available for a small subset of individuals with crania present, 

the sample size for this research question is quite limited. To mitigate some issues associated 

with statistical analysis of small samples, the number of craniometric variables used in this 

analysis is also limited, with the aim of preserving all four individuals from Phase I in the 

dataset. Therefore, craniometric variables with data for at least three of the four Phase I 

individuals were retained, leaving 12 variables (Table 5.5). To impute missing data following the 

recommendation of Kenyhercz and Passalacqua (2016), case deletion was performed for all 

individuals not having data for at least 6 of the 12 variables. Missing data are imputed using the 

MICE method for each phase separately in order to maintain any variation that exists between 

phases in the data. The resulting dataset comprises 12 craniometric variables for four individuals 

from Phase I and 20 individuals from Phase II.  

Table 5.5. Craniometric variables used in Research Question 4 

Abbreviation 
WFB 
NPH 
NLH 
JUB 
NLB 
OBH 
OBB 
DKB 
ZMB 
FMB 
UFBR 
UFHT 

Measurement 
Minimum frontal breadth 
Nasion-prosthion height 
Nasal height 
Jugal breadth 
Nasal breadth 
Orbital height 
Orbital breadth 
Interorbital breadth 
Bimaxillary breadth 
Bifrontal breadth 
Upper facial breadth 
Upper facial height 

Following imputation, the complete Phase I and Phase II datasets are combined and 

scaled and centered to minimize the effect of size and to allow for pooling of sexes. Then, the 

110 

 
 
 
two phases are compared using a MANOVA to determine if there are significant differences in 

craniometric variables over time. Following the MANOVA, univariate one-way analyses of 

variance (ANOVAs) are performed on each variable to determine whether and where significant 

differences exist between phases on any of the 12 variables included in the analysis. 

The final research question also seeks to understand whether skeletal stress indicators 

were more frequent in the earlier phase (Phase I) or later phase (Phase II) of the cemetery. All 

individuals with phase data who had at least one scorable stress indicator are included in the 

analysis. In order to evaluate this question, the frequency of each non-specific stress indicator is 

calculated by dividing the number of instances of each indicator by the total number of 

observable instances in the assemblage. Each phase is divided by sex and differences in 

frequency are compared between males and females using a Fisher’s exact test. If no significant 

difference is found between males and females within each phase, the sexes are pooled and 

frequencies are compared between Phase I and Phase II, again using a Fisher’s exact test. 

This chapter provided the specific skeletal samples and methods used to answer the 

research questions. The next chapter will present the results of these analyses. 

111 

 
 
 
CHAPTER SIX: RESULTS 

First, I will report summary data for the craniometric datasets as well as illustrations of 

missingness and the effects of imputation on the datasets. This chapter will then present results 

organized by each research question, as outlined in Chapter Four: Research Questions and 

Expectations. 

Summary Data 

Summary data for the seven craniometric datasets are presented in Tables 6.1 through 

6.7. These show details of the data prior to imputation. Missingness is illustrated in Figures 6.1 

through 6.3. Following imputation, a series of graphics were generated to ensure no unrealistic 

values were imputed. Figure 6.4 provides an example of the pre- and post-imputation distribution 

of GOL in the Rusellae sample. 

112 

 
 
 
 
Table 6.1. Descriptive statistics, Rusellae 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 
WFB 
JUB 
FMB 
UFBR 
UFHT 

n 
24 
22 
22 
23 
13 
23 
21 
19 
25 
27 
27 
23 
28 
28 
28 
23 
25 
28 
27 
24 
21 
29 
25 
29 
29 
26 

Mean (mm) 
184 
98.8 
128 
139 
128 
121 
113 
95.3 
63.8 
50.1 
25.1 
29.3 
32.6 
38.8 
22.5 
92.2 
96.4 
109 
115 
95.9 
37.2 
94.9 
112 
97.3 
103.1 
66.8 

SD 
7.56 
5.59 
5.97 
6.29 
6.35 
7.78 
5.46 
4.36 
3.44 
3.05 
1.67 
2.85 
2.09 
1.85 
2.16 
5.28 
3.52 
6.22 
5.88 
5.29 
2.79 
4.28 
4.61 
3.72 
3.83 
3.69 

Min (mm)  Max (mm) 

172 
84 
117 
127 
112 
103 
102 
85 
58 
45 
22 
24 
28 
34 
18 
79 
90 
97 
107 
84 
33 
87 
104 
91 
96 
61 

198 
111 
136 
152 
137 
134 
123 
104 
72 
57 
28 
37 
36 
41 
27 
104 
102 
123 
134 
110 
43 
102 
119 
105 
110 
78 

113 

 
 
 
 
Table 6.2. Descriptive statistics, Mis Island 3-J-10 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
23 
23 
23 
23 
19 
23 
23 
22 
23 
23 
24 
23 
23 
23 
24 
23 
22 
24 
24 
23 
23 

Mean (mm) 
181.2 
98.3 
129.2 
126.7 
122.3 
112.3 
104.8 
98.0 
65.0 
47.3 
25.4 
27.3 
33.0 
37.2 
21.8 
91.3 
93.4 
108.5 
117.3 
92.5 
36.3 

SD 
4.85 
4.15 
5.25 
4.51 
4.78 
4.31 
5.00 
6.12 
4.02 
2.74 
1.41 
4.08 
1.85 
1.93 
1.49 
4.12 
2.77 
4.62 
5.66 
4.34 
3.10 

Min (mm)  Max (mm) 

172 
90 
119 
116 
114 
103 
96 
86 
56 
41 
23 
18 
29 
32 
19 
85 
87 
101 
105 
84 
30 

191 
105 
141 
135 
131 
120 
116 
109 
75 
52 
29 
36 
37 
41 
25 
101 
99 
117 
130 
102 
44 

114 

 
 
 
 
Table 6.3. Descriptive statistics, Mis Island 3-J-11 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
74 
74 
74 
74 
60 
73 
74 
71 
72 
72 
74 
72 
72 
72 
74 
71 
69 
75 
75 
75 
75 

Mean (mm) 
182.1 
99.3 
128.7 
126.9 
120.9 
112 
104.6 
98.1 
65.1 
47.1 
26.6 
26.6 
32.5 
37.4 
23.1 
93.0 
94.2 
109 
115.6 
93.3 
35.5 

SD 
6.17 
4.66 
4.71 
4.80 
5.42 
4.17 
4.24 
4.95 
4.67 
3.43 
1.90 
3.09 
2.05 
1.68 
1.96 
4.66 
3.12 
3.98 
5.06 
4.59 
2.57 

Min (mm)  Max (mm) 

171 
89 
117 
113 
111 
100 
93 
88 
54 
40 
22 
20 
28 
33 
19 
83 
86 
101 
102 
83 
30 

195 
114 
138 
138 
133 
121 
119 
112 
77 
56 
32 
34 
37 
41 
30 
107 
102 
120 
137 
104 
43 

115 

 
 
 
 
 
Table 6.4. Descriptive statistics, Berg 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 
109 

Mean (mm) 
175.6 
95.8 
127.4 
144.1 
131.1 
124 
111 
91.9 
65.8 
50 
25.2 
27 
33.3 
39.3 
22.5 
91.5 
97 
108.7 
107.7 
92.8 
37.6 

SD 
8.45 
5.03 
5.34 
6.22 
6.37 
6.1 
5.34 
5.72 
4.63 
3.41 
1.86 
3 
1.89 
1.59 
2.35 
4.47 
3.41 
4.60 
5.70 
5.21 
3.10 

Min (mm)  Max (mm) 

155 
84 
115 
131 
117 
110 
98 
81 
52 
42 
22 
18 
28 
36 
18 
80 
89 
96 
95 
79 
31 

198 
110 
141 
161 
148 
140 
127 
108 
79 
58 
31 
36 
39 
44 
32 
104 
105 
122 
120 
105 
50 

116 

 
 
 
 
Table 6.5. Descriptive statistics, Dogon 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 
99 

Mean (mm) 
173.6 
96.6 
130 
134.6 
125.1 
112.2 
101.8 
97.9 
63.1 
46.9 
28.0 
27.0 
33.2 
38.8 
22.8 
94.6 
97.1 
107.8 
109.8 
94.2 
34.1 

SD 
6.94 
4.02 
4.92 
5.17 
5.69 
4.64 
4.61 
4.70 
4.17 
2.65 
1.68 
3.22 
1.85 
1.79 
2.30 
4.38 
3.81 
4.71 
6.08 
5.07 
2.62 

Min (mm)  Max (mm) 

157 
86 
118 
121 
112 
98 
89 
84 
52 
41 
25 
20 
29 
35 
17 
84 
88 
98 
97 
79 
29 

194 
107 
152 
148 
138 
123 
115 
108 
72 
54 
32 
35 
37 
43 
28 
105 
107 
118 
128 
106 
42 

117 

 
 
 
 
Table 6.6. Descriptive statistics, Norse 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 
108 

Mean (mm) 
184.2 
99.5 
128.9 
139.1 
129.5 
121.2 
109.3 
95.5 
66.5 
50.6 
24.8 
27.8 
33.5 
39.8 
21.4 
92.1 
96.8 
110.6 
112 
96.3 
35.7 

SD 
6.55 
4.22 
5.45 
5.26 
6.31 
5.25 
4.63 
4.97 
4.38 
2.98 
1.80 
3.48 
2.15 
1.62 
2.40 
5.15 
3.72 
4.99 
5.79 
4.41 
2.51 

Min (mm)  Max (mm) 

170 
88 
117 
127 
118 
108 
96 
86 
57 
44 
19 
19 
27 
36 
16 
79 
88 
99 
99 
86 
30 

201 
112 
141 
152 
142 
134 
123 
114 
76 
59 
30 
37 
38 
44 
27 
109 
106 
123 
128 
107 
43 

118 

 
 
 
 
Table 6.7. Descriptive statistics, Zalavar 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

n 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 
98 

Mean (mm) 
181.1 
99.1 
132.1 
139.3 
129.5 
121.4 
109.5 
94.8 
66.0 
50.1 
25.0 
28 
32.4 
39.4 
21.1 
92.6 
96.6 
110.3 
113.2 
95.3 
36.3 

SD 
7.24 
4.74 
5.91 
4.74 
5.32 
4.77 
4.68 
6.07 
4.71 
3.09 
1.60 
3.22 
1.89 
1.57 
2.39 
4.46 
3.22 
4.85 
5.60 
4.76 
2.64 

Min (mm)  Max (mm) 

165 
90 
120 
127 
114 
110 
97 
83 
54 
42 
22 
21 
28 
34 
16 
82 
88 
98 
100 
85 
30 

196 
109 
144 
149 
139 
133 
122 
110 
79 
57 
30 
35 
36 
43 
27 
105 
105 
120 
127 
108 
43 

119 

 
 
 
 
Figure 6.1. Missing data by variable, Rusellae. 

Figure 6.2. Missing data by variable, Mis Island 3-J-10. 

120 

 
 
 
 
 
Figure 6.3. Missing data by variable, Mis Island 3-J-11. 

Figure 6.4. Pre- and post-imputation distribution of GOL, Rusellae. 

Research Question 1 

A subset of sufficiently complete crania from Rusellae (n=26) were evaluated using 

Mahalanobis Distance (MD) and Principal Components Analysis (PCA) to explore craniofacial 

variation between Rusellae and global reference samples, as well as within the Rusellae sample 

itself. Of the 26 individuals from Rusellae, 13 are female (50%) and 13 are male (50%). Phase 

data is available for 22 of the individuals; 2 are from Phase I (9.1%) and 20 are from Phase II 

121 

 
 
 
 
(90.9%). The demographic breakdown of the six reference samples (Berg, Zalavar, Norse, 

Dogon, Mis Island 3-J-10, and Mis Island 3-J-11) is outlined in Chapter Five: Materials and 

Methods. 

In order to understand how Rusellae compares with geographically and temporally 

relevant reference samples, a Mahalanobis Distance test is performed. This test allows for 

understanding of intergroup similarity and dissimilarity in multivariate space between the Berg, 

Zalavar, Norse, Dogon, Mis Island 3-J-10, Mis Island 3-J-11, and Rusellae groups using the 21 

craniometric variables described in Chapter Five: Materials and Methods. The resulting 

Mahalanobis distances are depicted in Table 6.8. A permutation test for significance found 

significant differences at the α=0.05 level for all pairs of groups except between the two Mis 

Island cemeteries (p=0.113). 

Table 6.8. Mahalanobis distances between Rusellae and the six reference samples 

Berg  Dogon 

Berg 
Dogon 

0 
5.89 
Mis Island 3-J-10  7.88 
Mis Island 3-J-11  7.22 
3.49 
5.04 
2.99 

Norse 
Rusellae 
Zalavar 

5.89 
0 
7.13 
5.95 
5.67 
8.30 
5.25 

Mis 
Island 
3-J-10 
7.88 
7.13 
0 
1.58 
5.72 
6.00 
5.25 

Mis 
Island 
3-J-11 
7.22 
5.95 
1.58 
0 
4.70 
6.64 
4.75 

Norse 

Rusellae  Zalavar 

3.49 
5.67 
5.72 
4.70 
0 
4.26 
1.48 

5.04 
8.30 
6.00 
6.64 
4.26 
0 
4.14 

2.99 
5.25 
5.25 
4.75 
1.48 
4.14 
0 

A Principal Components Analysis on the complete Howells, Nubian, and Rusellae dataset 

can visualize the major contributing variables to the craniometric variation between the seven 

groups (Figure 6.5). Generally, the Dogon and Mis Island cemeteries cluster most closely and are 

separated from the cluster that contains the Norse, Zalavar, and Rusellae samples based on the 

first two PCs. The Berg sample is closer to the European centroids than the African centroids, 

122 

 
 
 
 
 
but is generally distant from all groups. Notably, the 95% confidence ellipse for Rusellae is 

larger than that of the other groups, and is the only one that encompasses the centroids of all 

groups, suggesting Rusellae exhibits more craniometric variation than the other samples included 

in the analysis. 

Figure 6.5. Scatterplot of Principal Components 1 and 2 with associated 95% confidence ellipses. 

123 

 
 
 
In order to evaluate within-cemetery craniometric variation at Rusellae, a Principal 

Components Analysis was performed using the same 21 craniometric variables used in the MD. 

The PCA resulted in seven PCs with eigenvalues greater than one (Table 6.9); these seven PCs 

each represent more variation than a single craniometric variable alone. Together, they explain 

78% of the craniofacial variation within the cemetery. A Scree plot of all PCs shows the 

percentage of variance explained by each PC, with a red dashed line separating those PCs with 

eigenvalues greater than one from the remaining PCs (Figure 6.6). 

Table 6.9. Variable loadings for first seven PCs 

Measurement 
GOL 
BNL 
BBH 
XCB 
ZYB 
AUB 
ASB 
BPL 
NPH 
NLH 
NLB 
MDH 
OBH 
OBB 
DKB 
ZMB 
EKB 
FRC 
PAC 
OCC 
FOL 

PC1 
-0.30989 
-0.10075 
-0.19478 
-0.23126 
-0.24729 
-0.15659 
-0.21717 
0.027653 
-0.30611 
-0.31013 
-0.09139 
-0.21179 
-0.09735 
-0.19232 
-0.19519 
-0.21447 
-0.24173 
-0.24031 
-0.2937 
-0.16202 
-0.25776 

PC2 
-0.19032 
-0.41491 
-0.17129 
0.341286 
0.214291 
0.247669 
0.215467 
-0.45243 
-0.01822 
-0.03473 
-0.25151 
-0.0062 
0.269546 
-0.06643 
-0.13264 
-0.04607 
0.049198 
-0.03549 
0.035017 
-0.34578 
0.036735 

PC3 
0.166309 
0.241611 
0.097424 
0.120018 
0.306868 
0.389715 
-0.00281 
0.041557 
-0.14961 
-0.10931 
-0.05112 
0.179922 
-0.22442 
-0.35411 
-0.20851 
0.23645 
-0.25822 
-0.38664 
-0.16196 
0.105232 
0.210613 

PC4 
0.05782 
-0.16011 
0.088396 
-0.17997 
0.120946 
-0.21223 
-0.09293 
-0.16661 
0.013714 
0.142414 
0.229791 
0.363026 
-0.1193 
-0.46318 
-0.01524 
-0.27727 
-0.33528 
0.367566 
0.152358 
-0.18314 
0.139805 

PC5 
-0.00026 
0.237165 
0.470154 
0.085946 
-0.21028 
-0.08353 
0.382177 
-0.18641 
-0.23966 
-0.21957 
-0.34542 
0.112131 
0.039423 
-0.10448 
0.400852 
-0.07592 
-0.1365 
0.036197 
0.069146 
0.012577 
-0.2049 

PC6 
0.218944 
0.082573 
-0.31719 
-0.14171 
-0.08604 
0.029023 
0.110742 
-0.04152 
0.210768 
0.092342 
-0.55552 
0.023016 
0.016722 
0.120076 
-0.31681 
-0.15694 
-0.36785 
0.167469 
0.131685 
0.342968 
-0.04705 

PC7 
0.059634 
-0.03839 
0.029472 
0.137811 
0.215344 
-0.30747 
0.308816 
0.10423 
-0.23547 
-0.2553 
0.017364 
-0.1302 
-0.51568 
0.222219 
-0.34125 
-0.07987 
0.122389 
0.053638 
0.333506 
-0.11537 
0.097779 

124 

 
 
 
Figure 6.6. Scree Plot of Rusellae Principal Components. 

The first two Principal Components explain the majority of the variation within the 

sample. The craniometric variables with the highest absolute loading on PC1 are maximum 

cranial length (-0.31), nasal height (-0.31), nasion-prosthion height (-0.31), and the parietal chord 

(-0.29). Overall, these measurements contribute to the height and length of the cranium in the 

sagittal plane. The variables that contribute most to PC2 are basion-nasion length (-0.41), basion-

prosthion length (-0.45), occipital chord (-0.35), and maximum cranial breadth (0.34); these 

variables primarily relate to cranial base length and cranial vault breadth. The loadings of the 

length variables on PC2 are all negative while the loading on the breadth variable is positive; this 

may represent a general shape component wherein cranial length and breadth are inversely 

related. Figure 6.7 is a scatterplot showing all 21 craniometric variables and their loadings on 

PC1 and PC2. 

125 

 
 
 
Figure 6.7. Scatterplot of interlandmark distances on PCs 1 and 2. 

The remaining PCs explain smaller proportions of overall variance. The variables with 

the highest positive loading on PC3 are bizygomatic breadth (0.31) and biauricular breadth 

(0.39), while the variables with the highest negative loading are frontal chord (-0.39) and orbital 

breadth (-0.35). In general, PC3 may represent a component of facial breadth. PC4 is heavily 

weighted by orbital breadth (-0.46), with moderate contributions from biorbital breadth (-0.34), 

bimaxillary breadth (-0.28), mastoid height (0.36), and frontal chord (0.37). PC4 may be 

capturing the relative dimensions of the eye orbits to the rest of the face. PC5 has the highest 

contributions from basion-bregma height (0.47), interorbital breadth (0.40), and biasterionic 

breadth (0.38); this may generally relate to a cranial base shape component. PC6 is weighted 

heavily by nasal breadth (-0.56), with moderate contributions from basion-bregma height (-0.32), 

interorbital breadth (-0.32), biorbital breadth (-0.37), and the occipital chord (0.34), representing 

a relationship between cranial base shape and midfacial structure. Finally, PC7 is primarily 

driven by orbital height (-0.52), with moderate contributions from biauricular breadth (-0.31), 

126 

 
 
 
biasterionic breadth (0.31), interorbital breadth (-0.34), and parietal chord (0.34); this PC appears 

to represent a shape variable connecting midfacial structure with cranial breadth. 

The Rusellae sample is plotted with each individual’s scores for PC1 and PC2 to 

visualize craniometric variation throughout the cemetery (Figure 6.8). In general, PC 1 separates 

males on the left from females on the right, suggesting a size component to PC1.  

Figure 6.8. Rusellae individuals plotted against PCs 1 and 2. 

In order to establish whether the variation present in the cemetery produces statistically 

distinct clusters of individuals, a hierarchical cluster analysis is performed on the seven retained 

PCs with eigenvalues greater than 1.0. The resulting dendrogram is shown in Figure 6.9. 

Although there is some overlap, the major cluster on the left primarily comprises females, while 

the major cluster on the right is mostly composed of males. 

127 

 
 
 
 
Figure 6.9. Hierarchical clustering dendrogram. 

The hierarchical clustering dendrogram shows three general clusters, so the dendrogram 

is cut at k=3. The Rusellae individuals are plotted again based on their PC1 and PC2 scores, but 

this time color coded by cluster (Figure 6.10). 

Figure 6.10. Rusellae individuals plotted against PC 1 and 2, color coded by cluster. 

128 

 
 
 
 
Research Question 2 

The same subset of crania as in Research Question 1 (n=26) were tested using the linear 

discriminant analysis (LDA) to explore population affinity throughout the cemetery. The model 

produced from the LDA using six reference datasets (Berg, Norse, Zalavar, Dogon, Mis Island 3-

J-10, and Mis Island 3-J-11) had an overall correct classification rate of 77.2%. The correct 

classification rates for each reference group are outlined in Table 6.10. 

Table 6.10. Correct classification rates for groups used to build LDA model 

Group 
Berg 
Norse 
Zalavar 
Dogon 
Mis Island 3-J-10 
Mis Island 3-J-11 

Correct Classification 
81.7% 
69.4% 
67.3% 
94.9% 
75.0% 
72.0% 

Of the 26 sufficiently complete individuals analyzed, 11 individuals (42.3%) showed 

closer affinity to one of the African groups (Dogon, Mis Island 3-J-10, and Mis Island 3-J-11), 

while 15 individuals (57.7%) showed closer affinity to one of the European groups (Berg, Norse, 

and Zalavar) (Table 6.11). 

129 

 
 
 
 
 
 
 
 
 
 
Table 6.11. Affinity estimates for Rusellae individuals 

Skeleton ID 

Sex 

Phase 

Group 

Posterior Probability  Continent 

82-1084 

Male  Unknown 

Norse 

82-1088 c.2  Female  Unknown  Mis Island 3-J-11 

82-1094 

Male 

Phase II  Mis Island 3-J-11 

82-1103 

Female  Phase II 

82-1116 

Male  Unknown 

82-1118A 

Female  Phase II 

82-1124 

Male 

Phase II 

82-1152 

Male 

Phase II 

82-1174 

Male 

Phase II 

82-1181 

Female  Phase II 

82-1185 

Female  Phase II 

82-1186 

Male 

Phase II 

82-1187 

Female  Phase II 

82-1188 

Male 

Phase II 

82-1189 

Male 

Phase II 

82-1220 

Female  Phase II 

82-1263 

Female  Phase II 

82-1308 

Female  Unknown 

82-5045 

Female 

Phase I 

82-5104 

Male 

Phase II 

Dogon 

Berg 

Zalavar 

Dogon 

Zalavar 

Zalavar 

Norse 

Norse 

Norse 

Norse 

Norse 

Berg 

Berg 

Dogon 

Dogon 

Berg 

Norse 

82-5112 

Female  Phase II  Mis Island 3-J-10 

82-5115 

Male 

Phase II 

Zalavar 

82-5142 

Female  Phase II  Mis Island 3-J-11 

82-5193 

Male 

Phase II  Mis Island 3-J-10 

82-5199 

Male 

Phase II 

82-5523 

Female 

Phase I 

Dogon 

Dogon 

130 

0.5095 

0.8165 

0.6712 

0.9933 

0.9338 

0.6541 

0.7760 

0.5540 

0.7089 

0.5058 

0.4219 

0.8069 

0.4956 

0.6938 

0.4301 

0.9605 

0.3990 

0.9963 

0.9964 

0.8689 

0.6058 

0.3817 

0.9479 

0.7878 

0.9984 

0.7223 

Europe 

Africa 

Africa 

Africa 

Europe 

Europe 

Africa 

Europe 

Europe 

Europe 

Europe 

Europe 

Europe 

Europe 

Europe 

Europe 

Africa 

Africa 

Europe 

Europe 

Africa 

Europe 

Africa 

Africa 

Africa 

Africa 

 
 
The values for the unknowns in the first two linear discriminant functions are plotted to 

visualize their relationship with the centroids of the six reference groups (Figure 6.11). The first 

linear discriminant axis separates the European and African groups. The second axis roughly 

separates the Berg and Dogon samples from the Norse, Zalavar, and Mis Island reference groups. 

Figure 6.11. Rusellae unknowns (black dots) and reference group centroids (red dots) plotted 
against the first two linear discriminant functions. 

Research Question 3 

In this section, I will refer to individuals of African affinity and individuals of European 

affinity for the sake of brevity. Please note, this shorthand should be taken to mean individuals 

showing closer statistical affinity to the African or European reference samples used in the 

craniometric analysis. 

Cribra orbitalia 

Of the total sample of 160 adults, 20 (12.5%) had an estimated affinity and at least one 

scorable eye orbit in order to evaluate presence or absence of cribra orbitalia. Of these, eight 

(40%) show greater African affinity and 12 (60%) show greater European affinity. Of the 

131 

 
 
 
 
 
individuals with African affinity, four are male (50%) and four are female (50%). Seven of the 

individuals with European affinity are male (58.3%), while five (41.7%) are female (Table 6.12). 

Table 6.12. Individuals with data for affinity and cribra orbitalia 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
5 
9 

4 
7 
11 

8 
12 
20 

In individuals with a closer affinity to the African reference samples, cribra orbitalia is 

present in zero out of four males (0%) and one of four females (25%) (Table 6.13). A Fisher’s 

exact test showed no significant differences in the frequency of cribra orbitalia between males 

and females with African affinity (p=1.00).  

Table 6.13. Counts for cribra orbitalia by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
4 
3 
7 

0 
1 
1 

4 
4 
8 

In individuals with European affinity, cribra orbitalia is present in one out of seven 

males (14.3%) and zero out of 5 females (0%) (Table 6.14). A Fisher’s exact test showed no 

significant difference in the frequency of cribra orbitalia between males and females with 

European affinity (p=1.00). Consequently, males and females were pooled for the overall 

analysis comparing frequency of cribra orbitalia between individuals of African and European 

affinity. 

Table 6.14. Counts for cribra orbitalia by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
6 
5 
11 

7 
5 
12 

1 
0 
1 

132 

 
 
 
 
 
 
 
In the African affinity pooled sexes sample, cribra orbitalia is present in one out of eight 

individuals (12.5%). In the European affinity pooled sexes sample, cribra orbitalia is present in 

one out of 11 individuals (9.1%) (Table 6.15). A Fisher’s exact test showed no significant 

differences in the frequency of cribra orbitalia between individuals with African and European 

affinity (p=1.00). 

Table 6.15. Counts for cribra orbitalia by affinity, pooled sexes 

African Affinity 
European Affinity 

Porotic Hyperostosis 

Present  Absent  Total 
7 
11 

8 
12 

1 
1 

Of the total sample of 160 adults, 20 (12.5%) had an estimated affinity and at least one 

parietal bone present in order to evaluate presence or absence of porotic hyperostosis. Of these, 

eight (40%) show greater African affinity and 12 (60%) show greater European affinity. Of the 

individuals with African affinity, four are male (50%) and four are female (50%). Seven of the 

individuals with European affinity are male (58.3%), while five (41.7%) are female (Table 6.16). 

Table 6.16 Individuals with data for affinity and porotic hyperostosis 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
5 
9 

4 
7 
11 

8 
12 
20 

In individuals with African affinity, porotic hyperostosis is present in two out of four 

males (50%) and one of four females (25%) (Table 6.17). A Fisher’s exact test showed no 

significant differences in the frequency of porotic hyperostosis between males and females with 

African affinity (p=1.00).  

133 

 
 
 
 
 
 
 
 
Table 6.17. Counts for porotic hyperostosis by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
2 
3 
5 

4 
4 
8 

2 
1 
3 

In individuals with European affinity, porotic hyperostosis is present in three out of seven 

males (42.9%) and zero out of 5 females (0%) (Table 6.18). A Fisher’s exact test showed no 

significant difference in the frequency of porotic hyperostosis between males and females with 

European affinity (p=0.2045). Consequently, males and females were pooled for the overall 

analysis comparing frequency of porotic hyperostosis between individuals of African and 

European affinity. 

Table 6.18. Counts for porotic hyperostosis by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
4 
5 
9 

7 
5 
12 

3 
0 
3 

In the African affinity pooled sexes sample, porotic hyperostosis is present in three out of 

eight individuals (37.5%). In the European affinity pooled sexes sample, porotic hyperostosis is 

present in three out of 12 individuals (25%) (Table 6.19). A Fisher’s exact test showed no 

significant differences in the frequency of porotic hyperostosis between individuals with African 

and European affinity (p=0.6424). 

Table 6.19. Counts for porotic hyperostosis by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
5 
9 

8 
12 

3 
3 

134 

 
 
 
 
 
 
 
 
 
 
Periosteal Reaction 

Of the total sample of 160 adults, 16 (10%) had an estimated affinity and at least one tibia 

or femur present in order to evaluate presence or absence of periosteal reaction. Of these, six 

(37.5%) show greater African affinity and 10 (62.5%) show greater European affinity. Of the 

individuals with African affinity, three are male (50%) and three are female (50%). Six of the 

individuals with European affinity are male (60%), while four (40%) are female (Table 6.20). 

Table 6.20. Individuals with data for affinity and periosteal reaction 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
3 
4 
7 

6 
10 
16 

3 
6 
9 

In individuals with African affinity, periosteal reaction is present in three out of three 

males (100%) and three out of three females (100%) (Table 6.21). A Fisher’s exact test showed 

no significant differences in the frequency of periosteal reaction between males and females with 

African affinity (p=1.00).  

Table 6.21. Counts for periosteal reaction by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
0 
0 
0 

3 
3 
6 

3 
3 
6 

In individuals with European affinity, periosteal reaction is present in all six males 

(100%) and all four females (100%) (Table 6.22). A Fisher’s exact test showed no significant 

difference in the frequency of periosteal reaction between males and females with European 

affinity (p=1.00). Consequently, males and females were pooled for the overall analysis 

comparing frequency of periosteal reaction between individuals of African and European 

affinity. 

135 

 
 
 
 
 
 
Table 6.22. Counts for periosteal reaction by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
0 
0 
0 

6 
4 
10 

6 
4 
10 

In the African affinity pooled sexes sample, periosteal reaction is present in all six 

individuals (100%). In the European affinity pooled sexes sample, periosteal reaction is present 

in all ten individuals (100%) (Table 6.23). A Fisher’s exact test showed no significant 

differences in the frequency of periosteal reaction between individuals with African and 

European affinity (p=1.00). 

Table 6.23. Counts for periosteal reaction by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
0 
0 

6 
10 

6 
10 

Linear Enamel Hypoplasia: Maxillary Canine 

Of the total sample of 160 adults, 17 (10.6%) had an estimated affinity and at least one 

scorable maxillary canine in order to evaluate presence or absence of linear enamel hypoplasia. 

Of these, eight (47.1%) show greater African affinity and nine (52.1%) show greater European 

affinity. Of the individuals with African affinity, four are male (50%) and four are female (50%). 

Five of the individuals with European affinity are male (55.6%), while four (44.4%) are female 

(Table 6.24). 

Table 6.24. Individuals with data for affinity and maxillary canine LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
4 
8 

8 
9 
17 

4 
5 
9 

136 

 
 
 
 
 
 
 
 
In individuals with African affinity, LEH of the maxillary canine is present in three out of 

four males (75%) and three of four females (75%) (Table 6.25). A Fisher’s exact test showed no 

significant differences in the frequency of LEH of the maxillary canine between males and 

females with African affinity (p=1.00).  

Table 6.25. Counts for maxillary canine LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
1 
1 
2 

3 
3 
6 

4 
4 
8 

In individuals with European affinity, LEH of the maxillary canine is present in four out 

of five males (80%) and all four females (100%) (Table 6.26). A Fisher’s exact test showed no 

significant difference in the frequency of LEH of the maxillary canine between males and 

females with European affinity (p=1.00). Consequently, males and females were pooled for the 

overall analysis comparing frequency of LEH of the maxillary canine between individuals of 

African and European affinity. 

Table 6.26. Counts for maxillary canine LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
1 
0 
1 

5 
4 
9 

4 
4 
8 

In the African affinity pooled sexes sample, LEH of the maxillary canine is present in six 

out of eight individuals (75%). In the European affinity pooled sexes sample, LEH of the 

maxillary canine is present in eight out of nine individuals (88.9%) (Table 6.27). A Fisher’s 

exact test showed no significant differences in the frequency of LEH of the maxillary canine 

between individuals with African and European affinity (p=0.5765). 

137 

 
 
 
 
 
 
 
 
Table 6.27. Counts for maxillary canine LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
2 
1 

8 
9 

6 
8 

Linear Enamel Hypoplasia: Maxillary Lateral Incisor 

Of the total sample of 160 adults, 14 (8.8%) had an estimated affinity and at least one 

scorable maxillary lateral incisor in order to evaluate presence or absence of linear enamel 

hypoplasia. Of these, seven (50%) show greater African affinity and seven (50%) show greater 

European affinity. Of the individuals with African affinity, three are male (42.9%) and four are 

female (57.1%). Four of the individuals with European affinity are male (57.1%), while three 

(42.9%) are female (Table 6.28). 

Table 6.28. Individuals with data for affinity and maxillary lateral incisor LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
3 
7 

7 
7 
14 

3 
4 
7 

In individuals with African affinity, LEH of the maxillary lateral incisor is present in two 

out of three males (66.7%) and one of four females (25%) (Table 6.29). A Fisher’s exact test 

showed no significant differences in the frequency of LEH of the maxillary lateral incisor 

between males and females with African affinity (p=0.4857).  

Table 6.29. Counts for maxillary lateral incisor LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
1 
3 
4 

3 
4 
7 

2 
1 
3 

In individuals with European affinity, LEH of the maxillary lateral incisor is present in 

two out of four males (50%) and all three females (100%) (Table 6.30). A Fisher’s exact test 

138 

 
 
 
 
 
 
 
 
showed no significant difference in the frequency of LEH of the maxillary lateral incisor 

between males and females with European affinity (p=0.4286). Consequently, males and females 

were pooled for the overall analysis comparing frequency of LEH of the maxillary lateral incisor 

between individuals of African and European affinity. 

Table 6.30. Counts for maxillary lateral incisor LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
2 
0 
2 

4 
3 
7 

2 
3 
5 

In the African affinity pooled sexes sample, LEH of the maxillary lateral incisor is 

present in three out of seven individuals (42.9%). In the European affinity pooled sexes sample, 

LEH of the maxillary lateral incisor is present in five out of seven individuals (72.4%) (Table 

6.31). A Fisher’s exact test showed no significant differences in the frequency of LEH of the 

maxillary lateral incisor between individuals with African and European affinity (p=0.5921). 

Table 6.31. Counts for maxillary lateral incisor LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
4 
2 

3 
5 

7 
7 

Linear Enamel Hypoplasia: Maxillary Central Incisor 

Of the total sample of 160 adults, 15 (9.4%) had an estimated affinity and at least one 

scorable maxillary central incisor in order to evaluate presence or absence of linear enamel 

hypoplasia. Of these, seven (46.7%) show greater African affinity and eight (53.3%) show 

greater European affinity. Of the individuals with African affinity, four are male (57.1%) and 

three are female (42.9%). Five of the individuals with European affinity are male (62.5%), while 

three (37.5%) are female (Table 6.32). 

139 

 
 
 
 
 
 
 
Table 6.32. Individuals with data for affinity and maxillary central incisor LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
3 
3 
6 

7 
8 
15 

4 
5 
9 

In individuals with African affinity, LEH of the maxillary central incisor is present in 

three out of four males (75%) and all three females (100%) (Table 6.33). A Fisher’s exact test 

showed no significant differences in the frequency of LEH of the maxillary central incisor 

between males and females with African affinity (p=1.00).  

Table 6.33. Counts for maxillary central incisor LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
1 
0 
1 

3 
3 
6 

4 
3 
7 

In individuals with European affinity, LEH of the maxillary central incisor is present in 

one out of five males (20%) and one out of three females (33.3%) (Table 6.34). A Fisher’s exact 

test showed no significant difference in the frequency of LEH of the maxillary central incisor 

between males and females with European affinity (p=1.00). Consequently, males and females 

were pooled for the overall analysis comparing frequency of LEH of the maxillary central incisor 

between individuals of African and European affinity. 

Table 6.34. Counts for maxillary central incisor LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
4 
2 
6 

5 
3 
8 

1 
1 
2 

In the African affinity pooled sexes sample, LEH of the maxillary central incisor is 

present in six out of seven individuals (85.7%). In the European affinity pooled sexes sample, 

LEH of the maxillary central incisor is present in two out of eight individuals (25%) (Table 

140 

 
 
 
 
 
 
 
 
6.35). A Fisher’s exact test showed a significant difference in the frequency of LEH of the 

maxillary central incisor between individuals with African and European affinity (p=0.04056, 

α=0.05), with a greater frequency in LEH of the maxillary central incisor in individuals with 

African affinity. 

Table 6.35. Counts for maxillary central incisor LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
1 
6 

7 
8 

6 
2 

Linear Enamel Hypoplasia: Mandibular Canine 

Of the total sample of 160 adults, 17 (10.6%) had an estimated affinity and at least one 

scorable mandibular canine in order to evaluate presence or absence of linear enamel hypoplasia. 

Of these, seven (41.2%) show greater African affinity and ten (58.8%) show greater European 

affinity. Of the individuals with African affinity, three are male (42.9%) and four are female 

(57.1%). Six of the individuals with European affinity are male (60%), while four (40%) are 

female (Table 6.36). 

Table 6.36. Individuals with data for affinity and mandibular canine LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
4 
8 

7 
10 
17 

3 
6 
9 

In individuals with African affinity, LEH of the mandibular canine is present in two out 

of three males (66.7%) and all four females (100%) (Table 6.37). A Fisher’s exact test showed 

no significant differences in the frequency of LEH of the mandibular canine between males and 

females with African affinity (p=0.4286).  

141 

 
 
 
 
 
 
 
 
Table 6.37. Counts for mandibular canine LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
1 
0 
1 

3 
4 
7 

2 
4 
6 

In individuals with European affinity, LEH of the mandibular canine is present in four 

out of six males (66.7%) and all four females (100%) (Table 6.38). A Fisher’s exact test showed 

no significant difference in the frequency of LEH of the mandibular canine between males and 

females with European affinity (p=0.4667). Consequently, males and females were pooled for 

the overall analysis comparing frequency of LEH of the mandibular canine between individuals 

of African and European affinity. 

Table 6.38. Counts for mandibular canine LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
2 
0 
2 

6 
4 
10 

4 
4 
8 

In the African affinity pooled sexes sample, LEH of the mandibular canine is present in 

six out of seven individuals (85.7%). In the European affinity pooled sexes sample, LEH of the 

mandibular canine is present in eight out of ten individuals (80%) (Table 6.39). A Fisher’s exact 

test showed no significant difference in the frequency of LEH of the mandibular canine between 

individuals with African and European affinity (p=1.00). 

Table 6.39. Counts for mandibular canine LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
1 
2 

7 
10 

6 
8 

142 

 
 
 
 
 
 
 
 
 
 
Linear Enamel Hypoplasia: Mandibular Lateral Incisor 

Of the total sample of 160 adults, 15 (9.4%) had an estimated affinity and at least one 

scorable mandibular lateral incisor in order to evaluate presence or absence of linear enamel 

hypoplasia. Of these, eight (53.3%) show greater African affinity and seven (46.7%) show 

greater European affinity. Of the individuals with African affinity, four are male (50%) and four 

are female (50%). Three of the individuals with European affinity are male (42.9%), while four 

(57.1%) are female (Table 6.40). 

Table 6.40. Individuals with data for mandibular lateral incisor LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
4 
8 

8 
7 
15 

4 
3 
7 

In individuals with African affinity, LEH of the mandibular lateral incisor is present in 

two out of four males (50%) and one out of the four females (25%) (Table 6.41). A Fisher’s 

exact test showed no significant differences in the frequency of LEH of the mandibular lateral 

incisor between males and females with African affinity (p=1.00).  

Table 6.41. Counts for mandibular lateral incisor LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
2 
3 
5 

4 
4 
8 

2 
1 
3 

In individuals with European affinity, LEH of the mandibular lateral incisor is present in 

one out of three males (33.3%) and one out of four females (25%) (Table 6.42). A Fisher’s exact 

test showed no significant difference in the frequency of LEH of the mandibular lateral incisor 

between males and females with European affinity (p=1.00). Consequently, males and females 

143 

 
 
 
 
 
 
were pooled for the overall analysis comparing frequency of LEH of the mandibular lateral 

incisor between individuals of African and European affinity. 

Table 6.42. Counts for mandibular lateral incisor LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
2 
3 
5 

3 
4 
7 

1 
1 
2 

In the African affinity pooled sexes sample, LEH of the mandibular lateral incisor is 

present in three out of eight individuals (37.5%). In the European affinity pooled sexes sample, 

LEH of the mandibular lateral incisor is present in two out of seven individuals (28.8%) (Table 

6.43). A Fisher’s exact test showed no significant difference in the frequency of LEH of the 

mandibular lateral incisor between individuals with African and European affinity (p=1.00). 

Table 6.43. Counts for mandibular lateral incisor LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
5 
5 

8 
7 

3 
2 

Linear Enamel Hypoplasia: Mandibular Central Incisor 

Of the total sample of 160 adults, 12 (7.5%) had an estimated affinity and at least one 

scorable mandibular central incisor in order to evaluate presence or absence of linear enamel 

hypoplasia. Of these, seven (58.3%) show greater African affinity and five (41.7%) show greater 

European affinity. Of the individuals with African affinity, three are male (42.9%) and four are 

female (57.1%). Three of the individuals with European affinity are male (60%), while two 

(40%) are female (Table 6.44). 

144 

 
 
 
 
 
 
 
 
 
 
Table 6.44. Individuals with data for affinity and mandibular central incisor LEH 

African Affinity 
European Affinity 
Total 

Male  Female  Total 
4 
2 
6 

7 
5 
12 

3 
3 
6 

In individuals with African affinity, LEH of the mandibular central incisor is present in 

one out of three males (33.3%) and one out of the four females (25%) (Table 6.45). A Fisher’s 

exact test showed no significant differences in the frequency of LEH of the mandibular central 

incisor between males and females with African affinity (p=1.00).  

Table 6.45. Counts for mandibular central incisor LEH by sex and African affinity 

African Affinity Male 
African Affinity Female 
Total 

Present  Absent  Total 
2 
3 
5 

3 
4 
7 

1 
1 
2 

In individuals with European affinity, LEH of the mandibular central incisor is present in 

one out of three males (33.3%) and one out of two females (50%) (Table 6.46). A Fisher’s exact 

test showed no significant difference in the frequency of LEH of the mandibular central incisor 

between males and females with European affinity (p=1.00). Consequently, males and females 

were pooled for the overall analysis comparing frequency of LEH of the mandibular central 

incisor between individuals of African and European affinity. 

Table 6.46. Counts for mandibular central incisor LEH by sex and European affinity 

European Affinity Male 
European Affinity Female 
Total 

Present  Absent  Total 
2 
1 
3 

3 
2 
5 

1 
1 
2 

In the African affinity pooled sexes sample, LEH of the mandibular central incisor is 

present in two out of seven individuals (28.6%). In the European affinity pooled sexes sample, 

LEH of the mandibular central incisor is present in two out of five individuals (40%) (Table 

145 

 
 
 
 
 
 
 
 
6.47). A Fisher’s exact test showed no significant difference in the frequency of LEH of the 

mandibular central incisor between individuals with African and European affinity (p=1.00). 

Table 6.47. Counts for mandibular central incisor LEH by affinity, pooled sexes 

African Affinity 
European Affinity 

Present  Absent  Total 
5 
3 

7 
5 

2 
2 

Research Question 4 

The results for Research Question 4 comprise two parts. In the first part, differences in 

the craniometric variables will be compared between Phase I and Phase II. Then, the frequency 

of each skeletal stress variable (cribra orbitalia, porotic hyperostosis, periostitis, linear enamel 

hypoplasia) will be compared between Phases I and II. 

Craniometric Variation 

In order to examine craniofacial variation over time at Rusellae, individuals from Phase I 

are compared with individuals from Phase II with respect to cranial measurements using 

MANOVA and ANOVA. The MANOVA testing for craniometric differences between Phase I 

and Phase II did not show significant differences (

=0.05, p=0.8827). The one-way ANOVAs 

comparing measurements for each variable by phase also showed no significant differences 

𝛼𝛼

(

=0.05) between phases for any of the 12 variables included in the analysis. 

𝛼𝛼
Cribra orbitalia 

Of the total sample of 160 adults, 51 (31.9%) had data for Phase and at least one scorable 

eye orbit in order to evaluate presence or absence of cribra orbitalia. Of these, six (11.8%) date 

to Phase I and 45 (88.2%) date to Phase II. Of the individuals from Phase I, three are male (50%) 

and three are female (50%). Twenty-six of the 45 Phase II individuals are male (57.8%), while 

19 (42.2%) are female (Table 6.48). 

146 

 
 
 
 
 
Table 6.48. Individuals with data for Phase and cribra orbitalia 

Phase I 
Phase II 
Total 

Male  Female  Total 
3 
19 
22 

6 
45 
51 

3 
26 
29 

In Phase I, cribra orbitalia is present in one out of three males (33.3%) and zero of three 

females (0%) (Table 6.49). A Fisher’s exact test showed no significant differences in the 

frequency of cribra orbitalia between males and females in Phase I (p=1.00). 

Table 6.49. Counts for cribra orbitalia by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
2 
3 
5 

3 
3 
6 

1 
0 
1 

In Phase II, cribra orbitalia is present in six out of 26 males (23.1%) and one out of 19 

females (5.3%) (Table 6.50). A Fisher’s exact test showed no significant difference in the 

frequency of cribra orbitalia between males and females in Phase II (p=0.2108). Consequently, 

males and females were pooled for the overall analysis comparing frequency of cribra orbitalia 

between Phases I and II. 

Table 6.50. Counts for cribra orbitalia by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
20 
18 
38 

26 
19 
45 

6 
1 
7 

In the Phase I pooled sexes sample, cribra orbitalia is present in one out of six 

individuals (16.7%). In the Phase II pooled sexes sample, cribra orbitalia is present in seven out 

of 45 individuals (15.6%) (Table 6.51). A Fisher’s exact test showed no significant differences in 

the frequency of cribra orbitalia between Phase I and Phase II (p=1.00). 

147 

 
 
 
 
 
 
 
 
Table 6.51. Counts for cribra orbitalia by Phase, pooled sexes 

Phase I 
Phase II 

Porotic Hyperostosis 

Present  Absent  Total 
5 
38 

6 
45 

1 
7 

Of the total sample of 160 adults, 79 (49.4%) had data for Phase and at least one parietal 

bone present in order to evaluate presence or absence of porotic hyperostosis. Of these, 12 

(15.2%) date to Phase I and 67 (84.8%) date to Phase II. Of the individuals from Phase I, five are 

male (41.7%), six are female (50.0%), and one is of indeterminate sex (8.3%). Thirty-six of the 

67 Phase II individuals are male (53.7%), 30 (44.8%) are female, and one is of indeterminate sex 

(1.5%) (Table 6.52). 

Table 6.52. Individuals with data for Phase and porotic hyperostosis 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

5 
36 
41 

6 
30 
36 

1 
1 
2 

12 
67 
79 

For the by-sex analysis, the individuals of indeterminate sex are left out. In Phase I, 

porotic hyperostosis is absent in all five males and all six females. Therefore, no statistical 

analyses are employed to evaluate significant differences between the sexes in Phase I. In Phase 

II, porotic hyperostosis is present in 13 out of 36 males (36.1%) and six out of 30 females (20%) 

(Table 6.53). A Fisher’s exact test showed no significant difference in the frequency of cribra 

orbitalia between males and females in Phase II (p=0.1803). Consequently, males and females 

were pooled for the overall analysis comparing frequency of cribra orbitalia between Phases I 

and II. 

148 

 
 
 
 
 
 
 
Table 6.53. Counts for cribra orbitalia by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
23 
24 
47 

36 
30 
66 

13 
6 
19 

In the Phase I pooled sexes sample, porotic hyperostosis is present in none of the 12 

individuals (0%). In the Phase II pooled sexes sample, porotic hyperostosis is present in 19 out of 

67 individuals (28.4%) (Table 6.54). A Fisher’s exact test showed no significant differences in 

the frequency of porotic hyperostosis between Phase I and Phase II (p=0.06001). However, due 

to the complete absence of porotic hyperostosis in Phase I, these results should be interpreted 

cautiously. 

Table 6.54. Counts for porotic hyperostosis by Phase, pooled sexes 

Phase I 
Phase II 

Periosteal Reaction 

Present  Absent  Total 
12 
48 

0 
19 

12 
67 

Of the total sample of 160 adults, 93 (58.1%) had data for Phase and at least one tibia or 

femur present in order to evaluate presence or absence of periosteal reaction. Of these, 15 

(16.1%) date to Phase I and 78 (83.9%) date to Phase II. Of the individuals from Phase I, six are 

male (40.0%), seven are female (46.7%), and two are of indeterminate sex (13.3%). Forty-five of 

the 78 Phase II individuals are male (57.7%), 31 (39.7%) are female, and two are of 

indeterminate sex (2.6%) (Table 6.55). 

Table 6.55. Individuals with data for Phase and periosteal reaction 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

6 
45 
51 

7 
31 
38 

2 
2 
4 

15 
78 
93 

149 

 
 
 
 
 
 
 
For the by-sex analysis, the individuals of indeterminate sex are left out. In Phase I, 

periosteal reaction is present in five out of six males (83.3%) and six of seven females (85.7%) 

(Table 6.56). A Fisher’s exact test showed no significant differences in the frequency of 

periosteal reaction between males and females in Phase I (p=1.00). 

Table 6.56. Counts for periosteal reaction by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
1 
1 
2 

6 
7 
13 

5 
6 
11 

In Phase II, periosteal reaction is present in 41 out of 45 males (91.1%) and 23 out of 31 

females (74.2%) (Table 6.57). A Fisher’s exact test showed no significant difference in the 

frequency of periosteal reaction between males and females in Phase II (p=0.05972). 

Consequently, males and females were pooled for the overall analysis comparing frequency of 

cribra orbitalia between Phases I and II. 

Table 6.57. Counts for periosteal reaction by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
4 
8 
12 

45 
31 
76 

41 
23 
64 

In the Phase I pooled sexes sample, periosteal reaction is present in 12 of the 15 

individuals (80%). In the Phase II pooled sexes sample, periosteal reaction is present in 66 out of 

78 individuals (84.6%) (Table 6.58). A Fisher’s exact test showed no significant differences in 

the frequency of periosteal reaction between Phase I and Phase II (p=0.7036). 

Table 6.58. Counts for periosteal reaction by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
3 
12 

15 
78 

12 
66 

150 

 
 
 
 
 
 
Linear Enamel Hypoplasia: Maxillary Canines 

Of the total sample of 160 adults, 63 (39.4%) had data for Phase and at least one scorable 

maxillary canine in order to evaluate presence or absence of linear enamel hypoplasia. Of these, 

nine (14.3%) date to Phase I and 54 (85.7%) date to Phase II. Of the individuals from Phase I, six 

are male (66.7%) and three are female (33.3%). Twenty-seven of the 54 Phase II individuals are 

male (50%), while 27 (50%) are female (Table 6.59). 

Table 6.59. Individuals with data for Phase and maxillary canine LEH 

Phase I 
Phase II 
Total 

Male  Female  Total 
3 
27 
30 

9 
54 
63 

6 
27 
33 

In Phase I, linear enamel hypoplasia of the maxillary canine is present in five out of six 

males (83.3%) and two of three females (66.6%) (Table 6.60). A Fisher’s exact test showed no 

significant differences in the frequency of LEH of the maxillary canine between males and 

females in Phase I (p=1.00).  

Table 6.60. Counts for maxillary canine LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
1 
1 
2 

6 
3 
9 

5 
2 
7 

In Phase II, linear enamel hypoplasia of the maxillary canine is present in 21 out of 27 

males (77.8%) and 15 out of 27 females (55.6%) (Table 6.61). A Fisher’s exact test showed no 

significant difference in the frequency of LEH of the maxillary canine between males and 

females in Phase II (p=0.148). Consequently, males and females were pooled for the overall 

analysis comparing frequency of LEH of the maxillary canine between Phases I and II. 

151 

 
 
 
 
 
 
Table 6.61. Counts for maxillary canine LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
6 
12 
18 

21 
15 
36 

27 
27 
54 

In the Phase I pooled sexes sample, LEH of the maxillary canine is present in seven out 

of nine individuals (77.8%). In the Phase II pooled sexes sample, LEH of the maxillary canine is 

present in 36 out of 54 individuals (66.7%) (Table 6.62). A Fisher’s exact test showed no 

significant differences in the frequency of maxillary canine LEH between Phase I and Phase II 

(p=0.7064). 

Table 6.62. Counts for maxillary canine LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
2 
18 

9 
54 

7 
36 

Linear Enamel Hypoplasia: Maxillary Lateral Incisors 

Of the total sample of 160 adults, 58 (36.3%) had data for Phase and at least one scorable 

maxillary lateral incisor in order to evaluate presence or absence of linear enamel hypoplasia. Of 

these, seven (12.1%) date to Phase I and 51 (87.9%) date to Phase II. Of the individuals from 

Phase I, six are male (85.7%) and one is female (14.3%). Twenty-five of the 51 Phase II 

individuals are male (49%), while 24 (47.1%) are female and two are of indeterminate sex 

(3.9%) (Table 6.63). 

Table 6.63. Individuals with data for Phase and maxillary lateral incisor LEH 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

6 
25 
31 

1 
24 
25 

0 
2 
2 

7 
51 
58 

152 

 
 
 
 
 
 
 
 
In Phase I, linear enamel hypoplasia of the maxillary lateral incisor is present in four out 

of six males (66.7%) and absent in the female (Table 6.64). A Fisher’s exact test showed no 

significant differences in the frequency of LEH of the maxillary lateral incisor between males 

and females in Phase I (p=0.4286).  

Table 6.64. Counts for maxillary lateral incisor LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
2 
1 
3 

6 
1 
7 

4 
0 
4 

For the sex-specific analysis, individuals of indeterminate sex are left out. In Phase II, 

linear enamel hypoplasia of the maxillary lateral incisor is present in six out of 25 males (24%) 

and eight out of 24 females (33.3%) (Table 6.65). A Fisher’s exact test showed no significant 

difference in the frequency of LEH of the maxillary lateral incisor between males and females in 

Phase II (p=0.538). Consequently, males, females, and individuals of indeterminate sex were 

pooled for the overall analysis comparing frequency of LEH of the maxillary lateral incisor 

between Phases I and II. 

Table 6.65. Counts for maxillary lateral incisor LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
19 
16 
35 

6 
8 
14 

25 
24 
49 

In the Phase I pooled sexes sample, LEH of the maxillary lateral incisor is present in four 

out of seven individuals (57.1%). In the Phase II pooled sexes sample, LEH of the maxillary 

lateral incisor is present in 14 out of 51 individuals (27.5%) (Table 6.66). A Fisher’s exact test 

showed no significant differences in the frequency of maxillary lateral incisor LEH between 

Phase I and Phase II (p=0.1874). 

153 

 
 
 
 
 
 
Table 6.66. Counts for maxillary lateral incisor LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
3 
37 

7 
51 

4 
14 

Linear Enamel Hypoplasia: Maxillary Central Incisors 

Of the total sample of 160 adults, 57 (35.6%) had data for Phase and at least one scorable 

maxillary central incisor in order to evaluate presence or absence of linear enamel hypoplasia. Of 

these, eight (14%) date to Phase I and 49 (86%) date to Phase II. Of the individuals from Phase I, 

five are male (62.5%) and three are female (37.5%). Twenty-four of the 49 Phase II individuals 

are male (49%), while 24 (49%) are female and one is of indeterminate sex (2%) (Table 6.67). 

Table 6.67. Individuals with data for Phase and maxillary central incisor LEH 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

5 
24 
29 

3 
24 
27 

0 
1 
1 

8 
49 
57 

In Phase I, linear enamel hypoplasia of the maxillary central incisor is present in one out 

of five males (20%) and one of three females (33.3%) (Table 6.68). A Fisher’s exact test showed 

no significant differences in the frequency of LEH of the maxillary central incisor between males 

and females in Phase I (p=1.00).  

Table 6.68. Counts for maxillary central incisor LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
4 
2 
6 

1 
1 
2 

5 
3 
8 

For the sex-specific analysis, individuals of indeterminate sex are left out. In Phase II, 

linear enamel hypoplasia of the maxillary central incisor is present in nine out of 24 males 

(37.5%) and 12 out of 24 females (50%) (Table 6.69). A Fisher’s exact test showed no 

154 

 
 
 
 
 
 
 
significant difference in the frequency of LEH of the maxillary central incisor between males and 

females in Phase II (p=0.5612). Consequently, males, females, and individuals of indeterminate 

sex were pooled for the overall analysis comparing frequency of LEH of the maxillary central 

incisor between Phases I and II. 

Table 6.69. Counts for maxillary central incisor LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
15 
12 
27 

9 
12 
21 

24 
24 
48 

In the Phase I pooled sexes sample, LEH of the maxillary central incisor is present in two 

out of eight individuals (25%). In the Phase II pooled sexes sample, LEH of the maxillary central 

incisor is present in 21 out of 49 individuals (42.9%) (Table 6.70). A Fisher’s exact test showed 

no significant differences in the frequency of maxillary central incisor LEH between Phase I and 

Phase II (p=0.4533). 

Table 6.70. Counts for maxillary central incisor LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
6 
28 

8 
49 

2 
21 

Linear Enamel Hypoplasia: Mandibular Canine 

Of the total sample of 160 adults, 72 (45%) had data for Phase and at least one scorable 

mandibular canine in order to evaluate presence or absence of linear enamel hypoplasia. Of 

these, ten (13.9%) date to Phase I and 62 (86.1%) date to Phase II. Of the individuals from Phase 

I, six are male (60%) and four are female (40%). Thirty-three of the 62 Phase II individuals are 

male (53.2%), while 28 (45.2%) are female and one is of indeterminate sex (1.6%) (Table 6.71). 

155 

 
 
 
 
 
 
 
 
Table 6.71. Individuals with data for Phase and mandibular canine LEH 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

6 
33 
39 

4 
28 
32 

0 
1 
1 

10 
62 
72 

In Phase I, linear enamel hypoplasia of the mandibular canine is present in six out of six 

males (100%) and two out of four females (50%) (Table 6.72). A Fisher’s exact test showed no 

significant differences in the frequency of LEH of the mandibular canine between males and 

females in Phase I (p=0.1333). 

Table 6.72. Counts for mandibular canine LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
0 
2 
2 

6 
4 
10 

6 
2 
8 

For the sex-specific analysis, individuals of indeterminate sex are left out. In Phase II, 

linear enamel hypoplasia of the mandibular canine is present in 23 out of 33 males (69.7%) and 

22 out of 28 females (78.6%) (Table 6.73). A Fisher’s exact test showed no significant difference 

in the frequency of LEH of the mandibular canine between males and females in Phase II 

(p=0.562). Consequently, males, females, and individuals of indeterminate sex were pooled for 

the overall analysis comparing frequency of LEH of the maxillary central incisor between Phases 

I and II. 

Table 6.73. Counts for mandibular canine LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
10 
6 
16 

23 
22 
45 

33 
28 
61 

In the Phase I pooled sexes sample, LEH of the mandibular canine is present in eight out 

of 10 individuals (80%). In the Phase II pooled sexes sample, LEH of the mandibular canine is 

156 

 
 
 
 
 
 
 
 
present in 46 out of 62 individuals (74.2%) (Table 6.74). A Fisher’s exact test showed no 

significant differences in the frequency of mandibular canine LEH between Phase I and Phase II 

(p=1.00). 

Table 6.74. Counts for mandibular canine LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
2 
16 

8 
46 

10 
62 

Linear Enamel Hypoplasia: Mandibular Lateral Incisor 

Of the total sample of 160 adults, 62 (38.8%) had data for Phase and at least one scorable 

mandibular lateral incisor in order to evaluate presence or absence of linear enamel hypoplasia. 

Of these, nine (14.5%) date to Phase I and 53 (85.5%) date to Phase II. Of the individuals from 

Phase I, six are male (66.7%) and three are female (33.3%). Twenty-seven of the 53 Phase II 

individuals are male (50.9%), while 25 (47.2%) are female and one is of indeterminate sex 

(1.9%) (Table 6.75). 

Table 6.75. Individuals with data for Phase and mandibular lateral incisor LEH 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

6 
27 
33 

3 
25 
28 

0 
1 
1 

9 
53 
62 

In Phase I, linear enamel hypoplasia of the mandibular lateral incisor is present in zero 

out of six males and one out of three females (33%) (Table 6.76). A Fisher’s exact test showed 

no significant differences in the frequency of LEH of the mandibular lateral incisor between 

males and females in Phase I (p=0.3333).  

157 

 
 
 
 
 
 
 
 
Table 6.76. Counts for mandibular lateral incisor LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
6 
2 
8 

0 
1 
1 

6 
3 
9 

For the sex-specific analysis, individuals of indeterminate sex are left out. In Phase II, 

linear enamel hypoplasia of the mandibular lateral incisor is present in 12 out of 27 males 

(44.4%) and ten out of 25 females (40%) (Table 6.77). A Fisher’s exact test showed no 

significant difference in the frequency of LEH of the mandibular lateral incisor between males 

and females in Phase II (p=0.7852). Consequently, males, females, and individuals of 

indeterminate sex were pooled for the overall analysis comparing frequency of LEH of the 

mandibular lateral incisor between Phases I and II. 

Table 6.77. Counts for mandibular lateral incisor LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
15 
15 
30 

12 
10 
22 

27 
25 
52 

In the Phase I pooled sexes sample, LEH of the mandibular lateral incisor is present in 

one out of nine individuals (11.1%). In the Phase II pooled sexes sample, LEH of the mandibular 

lateral incisor is present in 23 of 53 individuals (43.4%) (Table 6.78). A Fisher’s exact test 

showed no significant differences in the frequency of mandibular lateral incisor LEH between 

Phase I and Phase II (p=0.1353). 

Table 6.78. Counts for mandibular lateral incisor LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
8 
30 

9 
53 

1 
23 

158 

 
 
 
 
 
 
 
 
Linear Enamel Hypoplasia: Mandibular Central Incisor 

Of the total sample of 160 adults, 53 (33.1%) had data for Phase and at least one scorable 

mandibular central incisor in order to evaluate presence or absence of linear enamel hypoplasia. 

Of these, seven (13.2%) date to Phase I and 46 (86.8%) date to Phase II. Of the individuals from 

Phase I, four are male (57.1%) and three are female (42.9%). Twenty-four of the 46 Phase II 

individuals are male (52.2%), while 21 (45.7%) are female and one is of indeterminate sex 

(2.2%) (Table 6.79). 

Table 6.79. Individuals with data for Phase and mandibular central incisor LEH 

Male  Female  Indeterminate  Total 

Phase I 
Phase II 
Total 

4 
24 
28 

3 
21 
24 

0 
1 
1 

7 
46 
53 

In Phase I, linear enamel hypoplasia of the mandibular central incisor is present in zero 

out of four males and zero out of four females (Table 6.80). A Fisher’s exact test showed no 

significant differences in the frequency of LEH of the mandibular central incisor between males 

and females in Phase I (p=1.0). 

Table 6.80. Counts for mandibular central incisor LEH by sex in Phase I 

Phase I Male 
Phase I Female 
Total 

Present  Absent  Total 
4 
3 
7 

4 
3 
7 

0 
0 
0 

For the sex-specific analysis, individuals of indeterminate sex are left out. In Phase II, 

linear enamel hypoplasia of the mandibular central incisor is present in five out of 24 males 

(20.8%) and six out of 21 females (28.6%) (Table 6.81). A Fisher’s exact test showed no 

significant difference in the frequency of LEH of the mandibular central incisor between males 

and females in Phase II (p=0.7302). Consequently, males, females, and individuals of 

159 

 
 
 
 
 
indeterminate sex were pooled for the overall analysis comparing frequency of LEH of the 

mandibular central incisor between Phases I and II. 

Table 6.81. Counts for mandibular central incisor LEH by sex in Phase II 

Phase II Male 
Phase II Female 
Total 

Present  Absent  Total 
19 
15 
34 

24 
21 
45 

5 
6 
11 

In the Phase I pooled sexes sample, LEH of the mandibular central incisor is present in 

zero out of seven individuals (0%). In the Phase II pooled sexes sample, LEH of the mandibular 

central incisor is present in 12 of 46 individuals (26.1%) (Table 6.82). A Fisher’s exact test 

showed no significant differences in the frequency of mandibular central incisor LEH between 

Phase I and Phase II (p=0.3291). 

Table 6.82. Counts for mandibular central incisor LEH by Phase, pooled sexes 

Phase I 
Phase II 

Present  Absent  Total 
7 
34 

7 
46 

0 
12 

160 

 
 
 
 
 
CHAPTER SEVEN: DISCUSSION AND CONCLUSION 

This study had two primary goals: 1) to explore biodistance within Rusellae and between 

Rusellae and contemporary sites using cranial measurements; and 2) to examine variation in 

skeletal stress indicators over time and between affinities. As compared with archaeological and 

historical studies, osteological research from this region and period are relatively sparse. 

Therefore, these research avenues contribute to a better understanding of the demography and 

environment of a Late Antique and Early Medieval Italian city. 

Four research questions were developed to address these goals. Overall, the results 

showed that, despite exhibiting an overall greater affinity with contemporary European samples 

than African references, Rusellae was a heterogeneous site comprising several craniometrically 

distinct groups. Further, the presence of individuals of African affinity was preliminarily attested 

at the site. Notably, the individuals with a greater affinity to African reference samples showed 

no significant difference in frequencies of the skeletal stress markers analyzed, with the 

exception of the maxillary central incisor. In addition, there were no statistically significant 

differences in either skeletal stress indicators or cranial measurements between the earlier and 

later phases of the cemetery. This chapter will address each research question in turn, including 

the key findings, their interpretations, implications, and potential alternate explanations. Then, I 

will discuss the broader impact of this research, acknowledge its limitations, and recommend 

future research directions. Finally, I will briefly conclude by describing the key takeaways from 

this project. 

Research Question 1 

The first research question asked: how does craniofacial variation at Rusellae resemble or 

differ from that of geographically and temporally proximate reference samples, and what might 

161 

 
 
 
these relationships suggest about population continuity, migration, or gene flow? Further, what 

variables drive craniometric variation within Rusellae, and does there appear to be heterogeneity 

or homogeneity in overall craniometric variation at the site? 

This research question has three key findings. First, based on the Mahalanobis distance 

analysis, Rusellae is statistically distinct from the Berg, Zalavar, Norse, Mis Island 3-J-10, Mis 

Island 3-J-11, and Dogon samples. It is closest to the other European references (Zalavar, Norse, 

and Berg), and more distant from the African reference samples (the two Mis Island cemeteries 

and the Dogon sample). Second, in the seven-group analysis, the first two principal components 

separate the African from European samples, indicating that craniometric variation across the 

groups is patterned geographically, with more geographically proximate groups clustering 

together. Third, in the intra-cemetery principal components analyses at Rusellae, seven PCs were 

identified as contributing most to the craniometric variation within the site. Hierarchical cluster 

analysis on these seven PCs identified three phenotypically distinct subgroups, indicating 

craniometric heterogeneity in the study sample. 

I argue that the totality of these findings suggests Rusellae experienced overall population 

continuity with central, northern, and eastern Europe during Late Antiquity and the Early Middle 

Ages. This means that most gene flow was occurring between Rusellae and other geographically 

proximate individuals on the European continent. The results also indicate Rusellae interacted 

with diverse groups, including some from Africa, through its position in the Tuscan landscape 

and its connection with the broader Mediterranean region. Using a biocultural approach, this 

argument situates biological variation within the archaeological and historical context through an 

exploration of how migration and invasion, trade and economic networks, urbanization, and 

sociocultural mating practices may have influenced the observed craniometric heterogeneity. 

162 

 
 
 
From a population genetic perspective, these results are generally consistent with the 

expectations of an isolation-by-distance model with consistent gene flow. Gene flow refers to the 

transfer of genetic material between populations. In humans, gene flow is typically caused by 

migration and the resulting exchange of genes, which increases genetic variation in the receiving 

population. In this study, cranial measurements are used as a phenotypic proxy for the underlying 

genetic data. One explanation for the craniofacial variation observed at Rusellae is the migration 

of non-local groups to the site, including through invasion or conquest. Notably, the first phase 

of the cemetery overlaps with the “Migration Period,” also known as the “Barbarian Invasions,” 

which occurred between roughly AD 300 - AD 700 (Rubini, Libianchi, Gozzi, & Zaio, 2022). 

The Migration Period refers generally to the movements of Germanic groups from central and 

Eastern Europe into the Italian peninsula during and after the fall of the Roman Empire in the 

West (Rubini et al., 2022). The Lombards moved into Rusellae and overtook the city in 

approximately AD 593 (Sebastiani et al., 2015), roughly coinciding with the earliest dates of the 

cemetery. The presence of Lombard-style funerary objects such as a sax and a belt dating to the 

7th century AD at Rusellae suggest the presence–or at least the imitation–of Lombard culture at 

Rusellae (Celuzza et al., 2002). Therefore, the archaeological and historical evidence indicates 

the presence of Germanic migrants at Rusellae; this non-local genetic contribution to Rusellae 

would have increased the genetic–and thereby craniometric–variation within the site, potentially 

contributing to the observed craniofacial variation. 

The results are also congruent with the participation of Rusellae in trade and economic 

networks connecting far-flung regions throughout the Mediterranean and facilitating movement 

of people and goods. In addition to the archaeological indication of Rusellae being a 

multicultural site, it is also unsurprising that a site with proximity to a system of Roman roads, 

163 

 
particularly the via Aurelia, which connected Rome to Pisa (Sebastiani, 2017), would be 

composed of several cultural groups, as opposed to being isolated and homogenous. Previous 

biodistance studies have shown the construction of Roman roads reduced geographic barriers to 

migration (such as the Apennine mountains), thereby increasing gene flow within the Italian 

peninsula, as well as southern and central Europe (Rubini et al., 2022). Similarly, a biodistance 

analysis of nonmetric traits in the Abruzzo region of Italy also showed greater trait variation in 

sites along the Roman road system, in contrast to more biological isolation in sites distant from 

the roads (Cilli, 2021). Rusellae’s connection–by the via Aurelia–to the famously cosmopolitan 

city of Rome was, of course, a direct route for migrants from across the Mediterranean to travel 

to Rusellae, but Rusellae’s connection to Pisa is also noteworthy. During the Late Antique and 

Early Medieval periods, Pisa was an active port city, which, like Rusellae, was also under control 

of the Lombards during the 7th century AD. During the 9th century AD, Pisa also became a 

major defense outpost against Muslim raids from North Africa, with fleets navigating between 

Pisa and Tunisia (Alberti, 2023), thereby facilitating gene flow and leading to greater genetic 

diversity. The phenomenon of coastal cities in Italy exhibiting greater diversity than inland cities, 

has been demonstrated in the Roman era (Emery, Stark, et al., 2018) and appears to continue 

throughout the occupation of Rusellae. Additionally, recent aerial and magnetometer surveys 

have revealed the remains of a road roughly 200 meters from the city walls, connecting Rusellae 

to the surrounding countryside (Campana, 2021). Therefore, Rusellae would inevitably have 

been visited by merchants and migrants using the ports and roads that connected Rusellae to the 

rest of the Mediterranean, further increasing the potential for gene flow and genetic variation at 

the site.  

164 

 
Additionally, the transition between the Roman to Medieval settlement structure of 

Tuscany would have influenced the genetic variation at the site. Between the early 4th and mid-

5th centuries, settlement density in the region markedly declined, including a substantial 

decrease in the number of rural sites (Vaccaro, 2008). By the mid-6th century AD, sites near 

Rusellae, including Scoglietto, Spolverino, and Umbro Flumen, in Alberese, Italy, had 

experienced demographic decline and were ultimately abandoned (Sebastiani, 2016, 2017; 

Sebastiani et al., 2015). A declining Mediterranean economy, Germanic invasions, the Gothic 

Wars, and the plague of Justinian caused further population decline and abandonment of rural 

sites between the 5th and 7th centuries AD (Brogiolo et al., 2017; Francovich, 2007). By the 7th-

8th centuries AD, lowland settlements in Tuscany were infrequent and fortified hilltop 

settlements such as Rusellae were favored (Francovich, 2008). Rusellae maintained its 

occupation despite the decline of the surrounding Tuscan countryside, which may have instigated 

a convergence of diverse peoples in the urban center. This constriction of the population would 

have thereby increased the genetic effect of any migration on the overall variation at the site. In 

sum, the concentration of people from the surrounding region at Rusellae may be responsible for 

introducing genetic variation into the sample, producing the observed craniometric 

heterogeneity. 

Sociocultural factors such as kinship, status, and ethnicity may help explain the 

persistence of diverse subpopulations within the sample. Assortative mating based on status, 

ethnicity, or religion is a common human phenomenon (Diggelmann, 2016; Sebro, Peloso, 

Dupuis, & Risch, 2017). Genealogies within Medieval Europe were important for support 

networks, models of inheritance, establishing alliances, legitimizing power and status (Pohl, 

2016). At Rusellae, the importance of recognizing kinship and maintaining and consolidating 

165 

 
status could have led to endogamous subpopulations of individuals who primarily interbred 

within their group. Kinship was particularly important for Lombards. Studies have shown that 

Lombards in Hungary and northern Italy structured their cemeteries around pedigrees and that 

biological relatedness was an important component of their society (Amorim et al., 2018). For 

example, marriage between elite individuals is attested in Lombard society (Amorim et al., 

2018). It is hypothesized that the Phase I burials at Rusellae may comprise elite individuals 

buried closer to the church due to their higher social status. These burials are also the ones that 

include Lombard-style grave goods (Celuzza et al., 2002). Genetic isolation of this elite cultural 

group within the site may have contributed to the variation seen within the sample. Therefore, if 

different ethnic or social groups were identified by occupants of Rusellae based on different 

geographic origins, cultural practices, status, language, or religion, assortative mating may have 

occurred along these lines. If assortative mating occurred, this could maintain genetic diversity, 

in contrast to a system of random mating which would drive genetic homogeneity at the site over 

time. 

In general, the results of this research question are consistent with the findings of 

previous distance and mobility studies in the Late Antique and Early Medieval Mediterranean. 

As a reminder, the key finding for this question is that the cemetery generally comprised 

individuals most similar to other European groups, but maintained heterogeneity within the site 

that could be attributed to migration, invasion, trade, shifts in regional settlement structure, or 

sociocultural barriers to mating. Distance studies have identified Lombard cemeteries in Hungary 

and Italy wherein people with central and eastern European ancestry coexisted with those of 

southern European ancestry (Amorim et al., 2018). Further isotopic analysis of the Hungarian 

cemetery at Szólád indicates a wide heterogeneity in the geographic origin of individuals interred 

166 

 
there; approximately 31% of the sample was non-local according to strontium isotope ratios (Alt 

et al., 2014). 

The results of the present study are also consistent with recent genetic findings. In a 

large-scale analysis of genetic ancestry for individuals from Rome with known historical and 

archaeological contexts, researchers found that Late Antique individuals have an average 

ancestry most similar to modern central European populations. This was probably facilitated by a 

reduction in Roman population combined with moderate amounts of immigration, primarily from 

Germanic tribes from central and Eastern Europe, that would have influenced the average genetic 

ancestry (Antonio et al., 2019). However, Antonio and colleagues also found high inter-

individual heterogeneity in Late Antique Roman individuals with respect to their geographic 

ancestries. They attributed this “persistent genetic diversity” to the remnants of trade, migration, 

slavery, and conquest in the Imperial era, along with the maintenance of trade networks within 

the western Mediterranean and the movement of Germanic tribes into the Italian peninsula 

(Antonio et al., 2019). 

Researchers also found that Medieval individuals can be modeled as a two-way 

combination of Late Antique Romans and a central and northern European component, including 

Lombards from Hungary, Saxons from England, and Vikings from Sweden (2019). They 

hypothesize that this result is likely due to the connection between Medieval Rome and mainland 

Europe through the Holy Roman Empire, which connected a loose confederation of states under 

a Christian emperor, beginning with the coronation of Charlemagne in AD 800 (Antonio et al., 

2019). Their summary suggests that the overall ancestry of Roman individuals generally 

coincided with the changes in the political affiliations of Rome; however, individuals maintained 

highly diverse genetic ancestries, including contributions from the Near East, Europe, and North 

167 

 
Africa through the medieval period, making Rome a “genetic crossroads” of peoples from 

Europe and the greater Mediterranean (Antonio et al., 2019). 

I argue that the results from Rusellae show that it, too, represented a “genetic 

crossroads.” The overall biodistance analysis showed the greatest affinity between Rusellae and 

other medieval European references, as expected within an isolation-by-distance model. 

However, the interconnectedness of Rusellae with its hinterlands and large port cities, its 

conquest by the Lombards in the late 6th century AD, and its use as an urban center facilitated a 

degree of genetic diversity that is identifiable from the crania in its cemetery. 

Research Question 2 

The second research question asked: do any individuals at Rusellae show a closer affinity 

to African reference samples than European reference samples (Phases I and II)? 

The key finding from this research question is that, based on the LDFA and the reference 

samples used in this analysis, 11 individuals (42.3%) in the Rusellae sample showed greater 

affinity with African craniometric references, while 15 individuals (57.7%) showed greater 

affinity with European craniometric references. When the Rusellae individuals are plotted 

against the reference group centroids for the first two linear discriminant functions, it is clear that 

the crania exhibit overall heterogeneity. However, the model appears to be separating the African 

from the European reference samples well on the first linear discriminant axis, while the second 

linear discriminant axis appears to be separating the group based on temporal variation, with the 

earlier reference samples (Norse, Zalavar, and Mis Island) having positive values and the later 

reference samples (Berg and Dogon) having negative values. 

I argue that these findings suggest the presence of individuals with African affinity at 

Rusellae, supporting the existing historical, archaeological, genetic, and anthropological 

168 

 
 
 
 
evidence of African migration to Italy during the Roman Imperial era, Late Antiquity, and the 

Early Middle Ages. These results likely reflect a much more complex biological and cultural 

reality than the binary affinity classification can illuminate. Therefore, I posit that the greater 

affinity of some individuals from Rusellae to African references is likely either a product of 

genuine gene flow caused by migration and admixture through historical population interactions, 

methodological limitations of craniometric classification statistics, or both. 

The first possible explanation for several individuals from Rusellae showing a greater 

affinity with African references that individuals either of African origin or African descent lived 

at Rusellae. Given Italy’s geographic and cultural position, it is expected that the results would 

primarily indicate affinity with European samples. However, there is still some degree of African 

influence within Rusellae. The results indicate at least a small degree of gene flow from Africa 

into the site, which may be influencing the observed patterns of population affinity in the 

cemetery. A biocultural approach can be used to account for the relationship between human 

behavior and mobility and the observed biological variation in the sample.  

This gene flow is expected based on extensive historical and archaeological 

documentation. Italy was undeniably a crossroads of trade, migration, and invasion for the entire 

Mediterranean, including North Africa, before, during, and after the use of the cemetery at 

Rusellae (Holmes, 1988; Wickham, 2006). 

Overseas trade was particularly important during the Imperial era, and there is clear 

evidence of trade relationships between Italy and North Africa during this period as well. One of 

the wrecks at the island of Giglio on the Tuscan coast carried North African amphorae, 

indicating trade between North Africa and the North Mediterranean (Bound, 1987). Small 

merchant ships and large vessels traversed the open seas between Carthage and Rome, and 

169 

 
 
 
Roman wrecks at the Skerki Bank between Tunisia and Sicily reveal varied cargo, including 

goods from throughout the Mediterranean. Shipwreck archaeology provides clear evidence of 

trade relationships between North Africa, particularly Carthage, and the western coast of Italy 

during the Roman era (McCann, 2008). 

African Red Slip, or ARS, wares played a crucial role in trade between Egypt, North 

Africa, and Italy, so much so that the density, location, and dating of these wares is a useful 

proxy for trade due to their widespread distribution and almost exclusive use in some parts of the 

empire (S. Mitchell, 2007; Wickham, 2023). Importantly, ARS wares have been identified in 

Tuscany during the 4th-5th centuries AD, suggesting an ongoing trade relationship between 

Tuscany and North Africa during this time (Vaccaro, 2008). Despite a decline in trade with 

North Africa during the mid-fifth century AD, due in part to the Vandal conquest of Carthage in 

AD 439, by the 9th and 10th centuries AD, coastal cities in southwest Italy show clear evidence 

of resumed trade with North Africa (Horden & Purcell, 2000; Wickham, 1994, 2006). These 

examples provide evidence for a circum-Mediterranean economy involving Italy and North 

Africa, which would have naturally required human agents to move the goods, thereby providing 

a vector for gene flow between the continents. Additionally, other cultural variables such as 

occupation, military membership, or enslavement may have contributed to migration patterns 

(Redfern et al., 2016). 

The results of this research question also align well with previous skeletal studies, which 

have consistently identified the presence of individuals with African affinity in Imperial, Late 

Antique, and medieval European cemeteries using several distance measures and biogeochemical 

methods (Emery, 2018; Emery, Duggan, et al., 2018; Emery, Stark, et al., 2018; Fentress, 2003; 

Leach et al., 2010; Leach et al., 2009; Redfern et al., 2023; Redfern et al., 2016; Redfern & 

170 

 
 
 
Hefner, 2019; Salesse et al., 2021). Individuals hypothesized to have African affinity were 

identified in Roman York through craniometric discriminant function and stable isotope analysis, 

suggesting African origins for roughly 10% of their study sample (Leach et al., 2010; Leach et 

al., 2009). Crania with macromorphoscopic traits consistent with African affinity were identified 

in Roman and Medieval London; non-White European ancestry was hypothesized in as much as 

29% of the 41-person sample (Redfern et al., 2016; Redfern & Hefner, 2019). 

In Italy, several studies have identified the possible presence of African individuals in 

cemeteries contemporary with Rusellae. In the rural southern Italian necropolis at Vagnari (1st-

4th centuries AD), oxygen and strontium isotope analysis revealed that 7% of the individuals at 

the site were non-local; further analyses of the non-local individuals revealed isotopic and 

mtDNA evidence of North African origin (Emery, Duggan, et al., 2018; Emery, Stark, et al., 

2018). At the Imperial sites of Velia and Isola Sacra, isotope analysis also suggests possible 

North African origin for several individuals (Prowse et al., 2007; Stark et al., 2020). Moreover, 

Salesse and colleagues use multiple isotopes, dental morphology, and ancient DNA to identify an 

individual in the Roman catacombs of Saints Peter and Marcellinus who they suggest was likely 

born in the Nile Valley or the central Sahara Desert. The authors hypothesize this individual may 

have endured forced migration as a slave Salesse et al. (2021). Finally, the tentative presence of 

individuals with African affinity was also reported at the Early Medieval (11th century AD) 

cemetery at Cosa, near Rusellae, based on craniofacial morphology (Fentress, 2003). 

Genetic studies have also highlighted the complexity of the modern and archaeological 

Italian genome, which appears to be a product of regular multidirectional gene flow (Arauna, 

Hellenthal, & Comas, 2019). Evidence of North African influence has been detected in the 

Italian genome and hypothesized to be a result of Arab conquests, including migration to Italy 

171 

 
 
from Tunisia (Arauna et al., 2019; Busby et al., 2015; Fiorito et al., 2016). North African male 

DNA has also been identified in Iberia, Sicily, and the Italian peninsula as a result of Arab and 

North African incursions in these regions during the medieval period (Capelli et al., 2009). 

Additional genetic investigation of haplotype data found that gene flow from North Africa into 

Tuscany primarily came from the North African coast of the Mediterranean Sea as opposed to 

the Atlantic coast; the authors reported that the estimated date of admixture for the Tuscan and 

North African samples suggests movement of North Africans into Tuscany around the fall of 

Rome, in approximately the 4th century AD (Arauna et al., 2019). Botigue and colleagues further 

note that the genetic contributions to the Italian genome from North Africa are more substantial 

than those from sub-Saharan Africa (2013). 

The other possible explanation for the observed presence of individuals with African 

affinity at Rusellae is that the results are, in part, a product of the methodological limitations of 

classification statistics and issues with extrapolating forensic methods to a bioarchaeological 

context. Although the results obtained in the present study are consistent with historical, 

archaeological, anthropological, and genetic studies that show the presence of African affinity 

individuals throughout Imperial and post-Roman Europe, this dissertation shows a much higher 

proportion of African affinity in the sample than any other comparable study. Therefore, it is 

likely that at least some of the individuals with African affinity in the Rusellae sample are 

misclassified. 

This misclassification, if present, is probably due to the following causes: the clinal 

nature of human variation, missing data, complex genetic history, applying forensic methods to 

bioarchaeological samples, and poor reference samples. Human populations do not fall into 

discrete categories or types with respect to any genetic or phenotypic variable. Therefore, 

172 

 
 
overlapping craniometric variation of the reference samples is expected regardless of which 

groups are included in the analysis. This means that even if an individual shows the greatest 

affinity with one reference sample, they may still fall within the expected variation of a different 

reference group and may simply represent an atypical member of that group. The overall correct 

classification rate of the model used in this study is 77.2%, meaning that approximately 23% of 

individuals will be misclassified when the function is applied. Importantly, the concepts of 

“correct classification” and “misclassification” are relative to the reference samples used. Linear 

discriminant function analysis, by its nature, will classify every unknown individual into one of 

the reference samples used to build the model, regardless of whether or not the unknown 

individual actually belongs to one of the reference groups. This “forced classification,” so-to-

speak, will inevitably lead to misclassifications, especially if the true affinity group for an 

unknown individual is not included in the model. That is why the term “affinity” is so useful in 

this context; an unknown individual may accurately be said to show a greater affinity to a 

particular reference sample, without the assumption that the individual actually belongs to that 

population. 

Another factor that may contribute to misclassification is missing data. Using data 

imputation to complete a dataset, while useful for improving sample size, lowers classification 

accuracy because it minimizes variation within the sample (Kenyhercz & Passalacqua, 2016). 

The complexity of the Italian genome suggests regular gene flow, and it is almost certain 

that misclassification occurs as a result of this complexity, with individuals classifying as 

African when, in reality, they belong to a diverse European group. Misclassification is not 

surprising in Rusellae specifically, as several genetic studies have identified a complex and 

admixed Italian genome with contributions from throughout the Mediterranean (Arauna et al., 

173 

 
 
2019; Busby et al., 2015; Fiorito et al., 2016). The phenotypic proxy of this complex genetic data 

is therefore also expected to exhibit a heterogeneity that may be challenging to classify. 

Moreover, studies have shown issues with misclassification when applying discriminant 

function analysis to bioarchaeological samples. Using the forensic anthropology classification 

software, FORDISC 2.0, Williams and colleagues found that the program misclassified Meroitic 

Nubian individuals into a variety of the Howells’ reference samples included in the software; 

they suggest, and I agree, that this misclassification stems from the inability for a statistically-

defined sample to capture the extent of variability in a given human population. The clinal nature 

of human variation, as well as cultural and environmental factors that contribute to deviation 

from the expectations of an isolation-by-distance model, will always pose challenges to 

classification (Williams, Belcher, & Armelagos, 2005). Previous studies have also shown 

misclassification of affinity when using macromorphoscopic traits as a genetic proxy (Redfern & 

Hefner, 2019). The authors also argue that the lack of temporally and geographically 

representative reference data leads to misclassification of affinity and “false positives” of 

individuals with non-local origin or affinity (Redfern & Hefner, 2019). 

One way to improve classification accuracy, and thereby increase confidence in the 

results, is to refine reference datasets. The more representative a reference sample is of the 

potential affinity of any given unknown individual, the more likely the model will be able to 

correctly classify that unknown. This can be achieved by collecting or collating additional data 

from contemporary populations with a wider geographic range and including more samples that 

are potentially related to the unknown sample or better correspond with the “sending” 

populations when attempting to “identify” a migrant. Including more geographically or 

174 

 
 
temporally consistent samples may alter classifications, but will not affect the underlying group 

relationships and will also improve confidence in results. 

Research Question 3 

Research Question 3 asked whether there is evidence of differential life experience 

between individuals with African affinity, as compared with individuals with European affinity 

with respect to skeletal stress indicators (Phase II). 

The key finding of this research question is that no significant differences were found in 

any of the skeletal stress indicators analyzed by sex within affinity or overall affinity, except for 

linear enamel hypoplasia of the maxillary central incisor. This tooth had a higher frequency of 

LEH in the African affinity sample compared with the European affinity sample. Generally, 

cribra orbitalia and porotic hyperostosis exhibited relatively low frequencies, while periosteal 

reaction was present in all individuals, and linear enamel hypoplasias of the anterior maxillary 

and mandibular dentition varied in frequency by tooth, ranging from 25% to 88.9%. 

Before delving into this discussion, it is important to acknowledge the limitations of this 

research question up front. The dataset for this analysis is exceptionally small, severely limiting 

the power of the statistical analyses and potentially introducing sampling bias, making any 

conclusions drawn from the results tentative at best. Therefore, this discussion serves as a 

preliminary exploration of the data with the caveat that further data will be necessary to confirm 

the findings. With these limitations in mind, I argue that the observed patterns of physiological 

stress in the cemetery provisionally suggest broadly similar exposure to stressors across affinity 

groups, potentially resulting from shared sociocultural practices and environmental conditions 

among individuals with varying affinities during the later phase of the cemetery. However, there 

is a weak indication that early life developmental stressors were more frequent in the African 

175 

 
 
 
 
affinity group, reflecting the potential for a bioculturally complex life experience in Late Antique 

and Early Medieval Italy. 

The combined results for cribra orbitalia and porotic hyperostosis suggest a relatively 

low prevalence of both porosities across the sample. Caution should be taken in interpreting 

these lesions due to their complex etiologies. A constellation of variables has been proposed that 

can lead to porotic hyperostosis and cribra orbitalia, including inadequate nutrition, vitamin 

deficiencies, inflammation, poor sanitation, and infectious disease (Larsen, 2015; Walker et al., 

2009). Since it is impossible to determine the cause of these lesions based on the data collected 

here, it can simply be said that the absence of affinity-related differences suggests a relatively 

similar nutritional status, hygienic environment, and disease load across the samples. 

Similarly, lack of significant differences in frequency of periosteal reaction also suggests 

similar exposure to infection, trauma, or inflammatory conditions between affinity groups. This 

result may be indicative of similar labor conditions or a similar propensity to injury among all 

adults in the later phase. Periosteal reactions are not limited to the period of growth and 

development, and can therefore provide insight into later-in-life insults (Weston, 2011). 

With respect to linear enamel hypoplasia, results indicate a similar degree of 

developmental stress between affinity groups for most teeth, especially during the period 

between one year and six years old, during which time enamel forms and records developmental 

insults (Reid & Dean, 2006). The maxillary central incisor shows a significantly higher 

frequency of LEH in the African affinity sample. This tooth tends to begin forming slightly 

earlier than the others, at around six months old (Liversidge, 2000), tentatively suggesting a 

higher proportion of very early childhood stress in the African affinity sample. 

176 

 
 
A variety of factors could result in a higher frequency of LEH in the maxillary central 

incisor for the African affinity group. For example, weaning practices in non-local individuals in 

a Late Roman/Early Medieval Roman frontier province in Southern Bavaria differed from those 

of local individuals, as identified by isotopes. LEHs were most frequent in the post-weaning 

interval, suggesting that children were at increased risk of developing LEH after breastfeeding 

had ceased (Velte et al., 2023). It is possible that earlier weaning, mediated by variable social 

norms and diverse weaning practices among differing affinity groups, may have contributed to a 

higher frequency of LEHs in the maxillary central incisor for the African affinity sample. A 

higher proportion of childhood illness, poor diet and nutrition, or psychosocial stressors related 

to migration or integration with the local population may all lead to the observed pattern. 

Overall, these results may suggest at least some degree of inequity in life experience within the 

sample. A more likely explanation, however, is that the results are a product of sampling issues. 

As previously noted, the findings for this question should be interpreted cautiously, as the 

results may be an artifact of several underlying methodological limitations. Therefore, I will now 

discuss several potential alternate explanations for the observed results: sample size and bias, 

misclassification of affinity, variation within an affinity group, and the osteological paradox. 

First, the lack of significant differences among most of the indicators may be a product of 

small sample size or random sample bias rather than a true absence of differences. This has been 

suggested as a cause for similar results in Redfern and Hefner’s analysis of the same stress 

markers between different affinity groups in Medieval London (2019). Further, age-associated 

differences in stress indicators and age-at-death distributions of the groups may compound to 

create a false depiction of differential stress exposure (Redfern et al., 2023). For example, 

periosteal new bone formation has been shown to increase with age (DeWitte, 2014), so the high 

177 

 
 
 
frequency of periosteal reaction could indicate a greater proportion of older adults in the sample 

(Redfern et al., 2023). 

A second consideration stems from compounding issues with the results of the previous 

research question. It is almost certain that some of these individuals are misclassified with 

respect to affinity, making the affinity groupings themselves tenuous. Therefore, it is completely 

possible that the observed results are a product of misclassifications and consequent faulty data 

partitioning as opposed to being reflective of the biocultural reality of life at Rusellae. 

Third, lumping together individuals into continent-wide affinity groups flattens the 

variation within each group; importantly, there are known cultural and ancestral distinctions 

within the “European” affinity group (i.e. Germanic central European groups vs. autochthonous 

Italians) present at Rusellae. These individuals may have had heterogeneous life experiences, 

resulting in further hidden heterogeneity within each broadly-assigned affinity group that is 

masked by the present research design. 

Finally, it is important to consider alternative interpretations of the data within the 

framework of the osteological paradox. The cemetery sample represents the individuals who 

survived any insults long enough for them to leave markers in the bones; since there is some 

indication of high infectious load (high frequency of periosteal reaction) and high developmental 

metabolic stress (high frequency of LEH), I argue for at least some degree of resilience in the 

face of chronic illness, infection, and childhood developmental stress in the individuals studied. 

Similarly, the higher frequency of LEH observed in the African affinity sample may indicate 

decreased frailty and increased resilience in this group, especially since the adults included in 

this study would have lived for many years beyond whatever insult caused the LEH during their 

178 

 
early life. All of these options–or a combination thereof–should be considered as potentially 

contributing to the observed pattern of physiological stress. 

Despite these legitimate concerns, it is notable that the results obtained herein are 

consistent with the limited comparable studies comparing affinity to skeletal stress in Late 

Antique and Early Medieval contexts. Redfern and colleagues examined the relationship between 

the Black Plague, physiological stress, and population affinity (estimated using 

macromorphoscopic traits) in a 14th-century plague cemetery in London (2023). Although the 

plague cemetery contained a higher proportion of individuals with estimated African affinity as 

compared with non-plague cemetery, no statistically significant differences were found between 

African affinity and European affinity in the frequency of linear enamel hypoplasia, cribra 

orbitalia, or periosteal reaction (Redfern et al., 2023). In a separate study, Redfern and Hefner 

found a higher frequency of LEH, periosteal reaction, and cribra orbitalia in individuals with 

White European ancestry (2019). They argue that the complete absence of some stress indicators, 

particularly in Black males, may suggest an increased mortality risk in those individuals within 

the context of a racialized society. Although these results are not directly comparable to the 

present findings due to the lack of statistical testing, it is notable that the authors propose the 

influence of a racialized society as underpinning their results, supporting a worse health outcome 

in individuals with estimated African affinity. As their study site post-dates the Rusellae 

cemetery by several centuries and is located in England instead of Italy, the sociocultural context 

of their conclusions differs from the expectations of Rusellae. 

I argue that in the context of Rusellae, the lack of distinction in frequency between 

affinity groups for nearly all the stress indicators under study might be the product of a more 

expansive view of ethnicity that incorporates religious affiliation into the concepts of self and 

179 

 
 
otherness in Late Antique and Early Medieval Tuscany. Therefore, it is important to consider 

how residents of Rusellae might have constructed a concept of ethnicity that would have 

produced an environment wherein different affinity groups appear to have been exposed to 

similar degrees of infection, comparable diets, and similar sanitary conditions. Heng (2011) and 

Lopez-Jantzen (2019) argue that Late Antique and Early Medieval “race” was not necessarily 

based on skin color or origin, but that religious affiliation served as a sociocultural and 

biopolitical marker of medieval racialization. Importantly, the incorporation of Christianity into 

Roman identity inextricably linked religion with “Romanness.” In the 6th century AD, Justinian 

limited full membership in the Roman community to those identifying as Christians (Lopez‐

Jantzen, 2019). In the 7th century AD, the Byzantine Emperor Heraclius ordered all Jews to 

convert to Christianity to attain religious conformity; this decree was also enforced in North 

Africa (Bat-Sheva, 2010). Therefore, as Christianity proliferated, so did marginalization along 

religious lines, with Jews and pagans viewed as “other” (Lopez‐Jantzen, 2019).  

Beginning in the 8th century AD, the church represented not only a sacred building, but 

also the node of a district, having baptismal and burial rights as well as responsibility for 

collecting tithes. In Tuscany, churches were historically located in hilly areas and distributed in 

strategic locations near rivers or roads, representing a pole of attraction for diverse occupants of 

the region (Riccomi, 2021). Given the nature of churches as religious and civic centers in rural 

Tuscany, it is also possible the church not only attracted diverse people from throughout the 

region, but that the shared sense of Christianity, and thereby an adoption or reiteration of Roman 

identity, may be a greater unifying force than any affinity-based social division. In support of 

this assertion, Ristuccia posits that baptism and burial in the same place produced a social and 

ritual community (2018). Therefore, it is possible that individuals buried within the same 

180 

 
cemetery would have envisioned themselves as members of the same community, thereby 

constructing a social environment with similar stressors. The burial of individuals with differing 

affinities in the same cemetery is especially surprising since North Africa was predominantly 

Muslim during this period (Fenwick, 2013), but the individuals interred at Rusellae were given 

Christian burials. Given the trend at Rusellae towards Christian burial tradition in Phase II 

(Celuzza et al., 2002) it appears that individuals with differing affinities may have been 

subsumed under the “Christian Roman” umbrella, thereby forming a sense of “self” and “other” 

that was less reliant upon concepts of affinity or origin and more in line with religious affiliation. 

This argument does not, of course, exclude the possibility that marginalization occurred 

on the lines of skin color or geographic origin during this time period or in this part of the world; 

the Early Medieval era was by no means “pre-racial,” but the concepts of race, self, and 

otherness were not necessarily drawn along modern Western lines. In fact, there is a general 

consensus that socially-constructed race was an important factor in Late Antiquity through the 

Middle Ages and that the production and reproduction of race and racism from classical antiquity 

through the central Middle Ages contributes to modern concepts of racism (Lopez‐Jantzen, 

2019). However, colonization in the Mediterranean was thought to not have had as harmful a 

biocultural effect as that observed in other contexts because racism in the Mediterranean was 

considered to be less severe due to long histories of trade and interaction (Kyle, Shehi, Koçi, & 

Reitsema, 2020). Rusellae was the seat of a diocese and a fortified hilltop settlement that would 

have attracted a diverse population in the wake of the collapsed rural Roman settlement system 

in Tuscany. Given the results of this research question, it seems that the Christian church was a 

unifying sociopolitical force, potentially outweighing distinctions based on sex or affinity. 

181 

 
 
Research Question 4 

Research Question 4 asked whether detectable craniofacial differences exist between 

Phase I (6th-7th centuries AD) and Phase II (8th-12th centuries AD) burials at Rusellae, 

suggesting either shifting population demographics or relative population continuity. Further, it 

sought to understand whether differences exist in the frequency of cribra orbitalia, porotic 

hyperostosis, periosteal reaction, and linear enamel hypoplasia between phases, suggesting 

changing environmental stressors, diet, disease, or other cultural factors over time. 

The key findings of this question are twofold. First, no significant differences were found 

between Phase I and Phase II individuals with respect to any of the 12 craniometric variables 

included in the study. Second, no significant differences were found in any of the skeletal stress 

indicators analyzed by sex within Phase or overall Phase: cribra orbitalia, porotic hyperostosis, 

periosteal reaction, and linear enamel hypoplasia of the maxillary and mandibular canines and 

incisors.  Generally, cribra orbitalia and porotic hyperostosis exhibited relatively low 

frequencies, while periosteal reaction was present in a high proportion, and linear enamel 

hypoplasias of the anterior maxillary and mandibular dentition varied in frequency by tooth, 

ranging from entirely absent in the mandibular central incisor to 80% in the mandibular canine. 

The caveats from Research Question 3 also apply here with respect to small sample size, 

limited statistical power, and consequent interpretive limitations. Thus, these findings should be 

considered tentative and as forming the basis for future studies. With these constraints in mind, I 

argue that the observed diachronic craniometric variation may preliminarily indicate an overall 

population continuity at the site between the earlier and later phases of the cemetery; there is not 

strong evidence for secular change or a sudden influx of new phenotypes that would alter the 

overall craniofacial complex at Rusellae. Further, the combined results of physiological stress 

182 

 
 
 
 
indicators support the presence of a generally consistent diet, nutrition, and exposure to 

environmental, developmental, and infectious disease stressors over time. 

Although it is nearly impossible to establish a specific cause-and-effect relationship for 

the studied lesions due to their nature as primarily non-specific stress indicators, the unique 

historical, cultural, and environmental circumstances of Rusellae may help us understand the 

observed pattern of cranial variation and physiological stressors, which are characterized by a 

relative degree of continuity in the midst of an otherwise transitional period. Three primary 

drivers are likely contributing, with various degrees of importance, to the observed outcome: 

demographic and epidemiological stressors, environmental and ecological factors, and diet and 

weaning practices. Overall, these sociocultural and environmental variables can help explain the 

consistency and frequency of stress indicators found in this study. 

The context of Rusellae as a continuously occupied site within a declining Roman 

settlement system is crucial. Historical evidence suggests that disease and stress were widespread 

during the sixth century AD, as a result of the Gothic Wars, the plague of Justinian, leprosy, 

famine, malaria, and economic decline during this time (Belcastro et al., 2007; McCormick, 

2001). Additionally, the population of Italy declined by half between the 3rd and 6th centuries 

AD (Belcastro et al., 2007). This reduction in population coincided with the 6th-century 

transition from traditional Roman settlement patterns, such as the villa system, to fortified hilltop 

sites that were military and civil centers of a territory (Chirico & Sebastiani, 2023). As one of 

these sites during Late Antiquity, Rusellae was surrounded by cities either abandoned or in crisis, 

including nearby Cosa, Heba, Vetulonia, and Saturnia. In contrast, Rusellae became a bishopric 

and constructed a church from the ruins of a Hadrianic bath. Despite the abandonment of parts of 

the site, a theme of its occupation was reuse, transformation, and rebuilding. Similarly, Rusellae 

183 

 
 
 
was valued for its interconnectedness, including accessibility to Roman roads, control of the 

coast of the Tyrrhenian Sea, and proximity to the northern frontier of the territory of Rome 

(Celuzza et al., 2023). 

In combination, these factors would have led to a coalescence of people within Rusellae, 

which may have resulted in the urban penalty phenomenon, or an increased pathogen or disease 

load for individuals in urban vs. rural settings. This “urban penalty” has been identified in Late 

Antiquity and the Early Middle Ages throughout Europe as a failure to adapt to diseases 

common in urban contexts (Meinzer, Steckel, & Baten, 2019) and has been linked to 

proliferative bone reactions, such as periosteal new bone formation (Riccomi, 2021). Zoonotic 

diseases, such as those transmitted from cattle to humans, and parasites may have been common, 

despite efforts made to improve sanitation (P. D. Mitchell, 2017); parasites including whipworm 

and roundworm, fleas, lice, and bed bugs are all vectors of infection, possibly from 

contamination of food resulting from poor hand hygiene and fecal contamination during this 

period (P. D. Mitchell, 2017). Therefore, it is possible that the process of urbanization led to 

higher population density within the city walls, resulting in issues with sanitation and waste 

management, creating the conditions for outbreaks of infectious disease (Riccomi, 2021). An 

increased pathogen load may contribute to the high frequency of linear enamel hypoplasia and 

periosteal reaction observed in the sample (Larsen, 2015). Since the site is held constant in this 

study, it is unsurprising that the stressors potentially linked to urbanization would also remain 

consistent in frequency, even over time. 

In addition to shifting settlement structure and the impact of urbanization, environmental 

and ecological factors could also have contributed to the skeletal stress markers at this site. Most 

importantly, environmental conditions favorable to the spread of parasitic infections, such as 

184 

 
 
 
malaria, likely influenced the sample. As the Roman Empire declined, marshes and coastal 

swamps likely expanded due to an inability to maintain waterway infrastructure, in concert with 

rising sea levels, facilitating mosquito breeding (Celuzza et al., 2023; Fornaciari, Giuffra, 

Ferroglio, Gino, & Bianucci, 2010; Fornaciari & Mallegni, 1985). During the Early Middle 

Ages, a brackish lagoon extended northwest of Rusellae, providing ideal breeding conditions for 

the Anopheles mosquito, which spread the malaria-causing Plasmodium falciparum, P. vivax, 

and P. malariae parasites across the landscape (Bellotti et al., 2004; Celuzza et al., 2023; 

Fornaciari et al., 2010; Sallares, Bouwman, & Anderung, 2004). In fact, malaria remained 

hyperendemic in the region of Rusellae until the mid-20th century (Boccolini et al., 2012; 

Celuzza et al., 2023). 

Cribra orbitalia has been linked to malaria-induced anemia (Gowland & Western, 2012; 

Schats, 2023); although these lesions are by no means pathognomonic for malaria, their presence 

in a region of known malarial infection should be considered, as malaria cannot be ruled out as 

causing them based on historical data. Additionally, differing ecological conditions (and the 

resulting variation in parasitic infections) have been suggested as a cause for different 

frequencies of cribra orbitalia in different regions of Italy (Belcastro et al., 2007). Further, 

recent application of Smith-Guzman’s (2015) paleopathological approach to identifying malaria 

using cribra orbitalia, femoral cribra, humeral cribra, spinal porosity, and periosteal reaction was 

applied to the Rusellae sample; results indicated that malarial infection was present at Rusellae 

based on Smith-Guzman’s criteria (2015), potentially in up to 66% of the individuals with 

similar frequencies across phases, sexes, and age cohorts (Celuzza et al., 2023). Notably, many 

of the lesions showed signs of healing, and their presence in older individuals indicates a degree 

of endemicity in the sample, wherein malarial infection was not acutely fatal. Therefore, the low 

185 

 
 
frequency of cribra orbitalia lesions at Rusellae may be linked to the environment, as a result of 

exposure to parasites causing endemic, but not acutely fatal, malarial infection. The 

hyperendemicity of malaria in the region of Rusellae until relatively recently also supports the 

consistency in lesions over time. 

Another component of life at Rusellae that may provide context for the lesion frequencies 

is diet, nutrition, and weaning practices. After the fall of the western Roman Empire, diet in the 

Roman world was characterized by bread, oil, and wine, while hunting, cattle breeding, and 

fishing were common of Germanic groups (Belcastro et al., 2007). In the 5th and 6th centuries, 

there was an integration of the two dietary cultures, as a result of migration, that persisted for the 

following centuries (Belcastro et al., 2007). Early Middle Ages historical sources suggest the 

importance of cereals along with uncultivated vegetable and animal products. A typical meal for 

most people of the time, including peasants, consisted of a meat stew with beans and vegetables, 

along with bread and wine. Animal products appear to have contributed to the diet even of the 

poor class during this period, which was somewhat unusual in the context of human dietary 

history (Belcastro et al., 2007). The role of meat consumption in the Late Antique and Early 

Medieval Italian diet, largely as a result of German influence, might have increased iron 

reserves, thereby contributing to the relatively low frequency of iron-related porosities such as 

cribra orbitalia and porotic hyperostosis. 

Additionally, weaning practices and the post-weaning diet would likely have influenced 

early-life stress, thereby affecting the frequency of linear enamel hypoplasias in the sample. 

LEHs occur after the first year of life on the anterior teeth; insults prior to this are unlikely to be 

recorded in the enamel (Reid & Dean, 2000; Reid & Dean, 2006). This period corresponds to the 

weaning period for Roman and Early Middle Age groups in Italy, based on historical documents. 

186 

 
 
 
Early Middle Ages medical literature on infant feeding suggested that a diet of exclusively breast 

milk was fed until approximately 6 months, or when the infant cuts their first tooth. This phase 

was to be followed by gradually introducing solid foods such as bread and milk, cereal, and 

porridge. Weaning was complete by approximately 3 years of age; historical documentation 

suggests this general trajectory did not change between Roman and Early Middle Ages periods 

(Belcastro et al., 2007). Therefore, a possible explanation for the lack of significant difference in 

LEH is that weaning practices did not change significantly over time, with a protein-deficient 

and possibly unsanitary post-weaning solid diet leading to disturbances in enamel formation 

across much of the population (Belcastro et al., 2007). However, increased protein consumption 

at a later age may have contributed to greater resilience and a lower overall frequency of porotic 

hyperostosis and cribra orbitalia. 

Given these examples, it is possible that the aforementioned elements would have 

contributed to a relatively stable living environment over time–imperfect, but consistent. 

Therefore, the lack of significant diachronic metric change in the cranium is relatively 

unsurprising. Secular change in skeletal measurements is influenced by environmental factors 

and changes in living conditions, as well as variables such as social class, income, family size, 

urban vs. rural location, region, health and hygiene, and overcrowding (Cole, 2000). The 

consistency in environment and living conditions that appears to be influencing the stasis in 

stress markers is also likely exerting similar selection pressures across time, minimizing the 

influence of cranial secular change in the sample. Another possible explanation is that the high 

degree of craniometric heterogeneity observed in the sample is maintained over time, thereby 

masking small temporal differences. 

187 

 
 
These data are consistent with previous findings that have suggested a regional variation 

within Italy in life experience for the transition between Late Antiquity and the Middle Ages. A 

study of the frequency of LEH, cribra orbitalia, and periosteal reaction in Molise, Italy show 

similar frequencies from the Roman to the Early Medieval period (Belcastro et al., 2007). 

Further, isotopic analysis revealed no substantial shift in diet between Early Medieval and 

Medieval occupations at Cosa, a geographically and temporally contemporary site to Rusellae; 

this dietary stability may help explain the lack of difference between nutrition-related stressors, 

such as porotic hyperostosis, cribra orbitalia, and LEH (Scorrano et al., 2014). In addition, no 

difference in frequency was found for cribra orbitalia or porotic hyperostosis among one Late 

Antique and two Early Medieval sites in Tuscany (Riccomi, 2021). Notably, although their study 

was conducted in Late Antique and Early Medieval Albania, Kyle and colleagues found that 

stress indicators did not increase in frequency through time, despite historically attested waves of 

colonization and social turmoil (2020). However, there is some evidence for worsening 

conditions (using stress indicators as a proxy) from the Roman to Medieval period in Latium, 

near Rome (Belcastro et al., 2007); this finding suggests that the periphery of Rome may have 

experienced greater devastation, but the impact of the Classical to post-Classical transformation 

may have been felt less strongly with increasing distance from Rome proper, such as the Tuscan 

countryside. 

The diachronic analysis shares many of the limitations as Research Question 3 with 

respect to sample size and statistical power, so I will not go into too much detail here. 

Importantly, the sample sizes are very disparate between Phase I and Phase II, with much smaller 

samples in Phase I. Not only is the extremely small Phase I sample in the craniometric analysis 

(n=4) unrepresentative of the overall variation in the cemetery during that period, there are also 

188 

 
 
statistical concerns associated with comparing a small sample to a substantially larger one using 

MANOVA. Therefore, it is possible that the observed lack of difference is a byproduct of the 

sample and statistical limitations. Further, porotic hyperostosis and linear enamel hypoplasia of 

the mandibular central incisor are entirely absent in Phase I, making those analyses especially 

tenuous. Finally, the Phase categorization may be a semi-arbitrary delineation that does not 

adequately align with the biocultural shifts occurring at the site, consequently limiting the 

capacity for a fine-grained diachronic analysis. 

Broader Implications 

This research has a number of methodological lessons and contributions that can be 

applied to future studies. First, the results highlight the value in partitioning datasets to uncover 

hidden heterogeneity. The significant difference found in linear enamel hypoplasia between 

African and European affinity groups for the maxillary central incisor may indicate some degree 

of different life experience between affinity groups. Although methods for estimating affinity 

undoubtedly need improvement, this study provides preliminary evidence for the value of 

considering the influence of affinity on susceptibility to skeletal stressors. 

Second, this study stresses the importance of taking a regional perspective, specifically 

with regard to Late Antiquity and the Early Middle Ages. Riccomi notes the value in a regional 

approach, as it is understood that the impact of the end of the Western Roman Empire differs, in 

part, based on proximity to the core of the empire versus the periphery (2021). The present 

research, in conjunction with several similar studies, contributes unique, regional, osteological 

evidence for varied living conditions throughout the Late Antique and Early Medieval eras, 

rather than supporting a universal collapse or stagnation, which has been hypothesized by Ward-

189 

 
 
Perkins (2005). Thus, this dissertation shows that there is value in employing a regional level of 

analysis when attempting to understand the “big picture.” 

Additionally, the findings related to estimating affinity within the cemetery present a 

“cautionary tale” for the utility of classification methods; as described previously, these methods 

will always produce a classification, even if the unknown does not belong to one of the reference 

samples in the model. Consequently, it is crucial to use caution and avoid overinterpreting the 

results of classification statistics, especially when the reference samples used are unlikely to 

represent the true affinity of the unknown. In the present study, classification methods should be 

viewed as a heuristic device that provides a framework for interpreting sample variation as 

opposed to a binary classification of African or European ancestry. 

This study also has modern relevance, despite its focus on the past. In an analysis of race 

and religion in Medieval England, Lewis introduced the idea of Africans as “officially absent 

although actually present” (2016). This concept was acknowledged by Redfern and Hefner as 

guiding their study of ancestry in Medieval London, noting that anecdotal evidence of African 

presence in Medieval cemeteries was not supported by published evidence (2019). The broad 

failure of bioarchaeologists to identify and record the presence of Africans in Europe during the 

Classical and Post-Classical periods perpetuates the invisibility and erasure of these migrants and 

their descendants, contributing to further marginalization, exclusion from academic discourse, 

and ignorance of their lived experiences. Perhaps even more nefariously, a whitewashing of the 

Middle Ages has been used by the Far Right to support White supremacist ideas of national 

identity and cultural heritage; excluding the non-White “other” from history thus legitimizes 

their modern racism and xenophobia from their perspective (Elliott, 2018). For Italy–a country 

where Mussolini took great pains to instill a national identity and a connection to the Romans–

190 

 
 
populism, xenophobia, and far-right thinking are again on the rise (Dotti, 2020; Favero, 2010; 

Fazzi, 2015). Despite the modern Italian genome showing contributions from the broader 

Mediterranean, including North Africa (Antonio et al., 2019), there are still factions who 

legitimize their racism and xenophobia with concepts of a White Rome and a White Middle 

Ages. Although the present study is relatively small in scope, it is my hope that it provides an 

opportunity to record the “official presence” of Africans in Late Antique and Early Medieval 

Italy. 

Future Directions 

The limitations of the samples and methods used in this dissertation have been detailed in 

previous chapters (Chapter Two: Bioarchaeological Approaches to Biological Distance and 

Skeletal Stress and Chapter Five: Materials and Methods), and this discussion has included 

an acknowledgement of the methodological and sample-related issues specific to each research 

question and dataset that may lead to alternate interpretations. In light of these constraints, I 

recommend several avenues of future research that may help overcome these limiting factors. 

These including improving reference samples and comparative data, expanding analytical 

approaches, integrating additional lines of evidence, and providing a more in-depth analysis of 

stress indicators. 

In order to expand reference samples and comparative data, future research should 

include additional skeletal material. Excavations are currently underway in and near Rusellae, 

which have uncovered additional burials in both the urban and suburban contexts (Campana, 

2022). Incorporating this material into future biodistance and stress studies will not only improve 

sample size, potentially allowing for more rigorous statistical analysis, but also may allow for a 

comparison of urban, suburban, and rural contexts of Rusellae. Additionally, adding reference 

191 

 
 
 
samples from more contemporary datasets near Rusellae will allow for a more fine-grained 

biological distance analysis. Nearby sites, including Cosa, have shown preliminary evidence for 

individuals with African affinity (Fentress, 2003), which would provide a useful craniometric 

comparison with Rusellae. Finally, should Late Antique or Early Medieval North African 

craniometric reference samples become available, particularly from Roman Tunisia, these would 

be invaluable comparative material, providing a better approximation of the possible range of 

homelands for migrants in Tuscany. 

Additionally, future research could expand analytical approaches. For example, 

geometric morphometric methods have been utilized to study craniofacial variation in Roman 

Italy (Hens & Ross, 2017). Three-dimensional cranial shape data are available for Rusellae, as 

the craniometric variables in this study were collected using a 3-D digitizer. This avenue of 

research would allow for a more nuanced shape analysis of the crania, as opposed to the size data 

used here. This may allow for an evaluation of more subtle morphological differences between 

individuals and between phases, and the wireframe models produced through geometric 

morphometric analyses provide a good visual representation of any variation. A spatial analysis 

of the cemetery would also be interesting, as it could contribute additional data towards an 

understanding of how age, sex, affinity, stress, and phase relate to cemetery orientation. Finally, 

incorporating a paleodemography approach and including subadults in the analysis would allow 

for a finer analytical resolution regarding age-at-death distributions, sex ratios, and mortality 

curves. Some skeletal stress indicators have shown a relationship to age (Redfern et al., 2023), so 

evaluating the age distribution of the site will provide important context. Further, examining 

mortality in addition to skeletal stress indicators could illuminate hidden shifts in health patterns 

192 

 
 
and may also reveal sex- and age-specific differences that are invisible in the paleopathology 

data. 

Another way to overcome some of this study’s limitations would be to integrate 

additional lines of evidence. Adding an ancient DNA analysis, particularly for those individuals 

with estimated African affinity, could help illuminate whether the genetic data are concordant 

with craniometric information and provide further evidence for long-distance migration in Late 

Antiquity and the Early Middle Ages. Analyzing pathogen DNA could also provide insight into 

the disease environment in which these individuals lived; malarial parasite DNA has successfully 

been identified in archaeological populations (Celuzza et al., 2023). Similarly, stable isotope 

analyses for understanding both migration and diet would provide critical information to this 

study. Not only can stable isotopes help identify instances of ancient mobility, they can also help 

understand whether individuals are first-generation migrants or, perhaps, are the descendants of 

migrants. Stable isotope analyses of dietary behavior would also provide important context for 

understanding stress indicators in relation to nutrition as well as sociocultural differentiation 

based on dietary patterns. Finally, it would be useful to incorporate nonmetric data, including 

macromorphoscopic traits, to add to an analysis of population affinity; ascertaining whether the 

metric and morphological traits are concordant in this sample is broadly relevant to biodistance 

and forensic research programs. 

A further avenue of research involves a more in-depth avenue of stress indicators. First, 

additional indicators of stress and disease can be included, such as caries and antemortem tooth 

loss, which might provide more information regarding dietary behaviors and the degree of 

carbohydrates in the diet. Additionally, acknowledging and analyzing the differences between 

active, healing, and healed lesions would help understand what phase of life the stress event 

193 

 
 
 
occurred in, which would help narrow down whether most stressors are occurring during 

development or as a result of disease processes later in life. The sample size for the present study 

was too small to partition the data this way, but should more skeletal material become available, 

it will be important to analyze the pathology data using this approach. 

Conclusion 

The present dissertation contributes preliminary data towards a better understanding of 

the nature of the transition between Classical and Post-Classical Rusellae through an analysis of 

biological distance and skeletal stress indicators. The key findings ultimately paint a picture of a 

diverse and resilient town that was influenced by, but did not succumb to, conquests and political 

upheaval, migration, changing settlement structure, and environmental and ecological stressors. 

The lack of diachronic change in skeletal stress indicators and craniometric variation suggests a 

degree of biological, ecological, environmental, and psychosocial stability throughout Late 

Antiquity and the Early Middle Ages. Further, it seems that migration, both before and during 

the cemetery’s use, introduced biological diversity into the population, while a general continuity 

with Roman ideals and the growing importance of Christianity seem to have encouraged 

assimilation instead of social stratification. These outcomes are consistent with Brown’s 

conception of a “long Late Antiquity” that emphasizes cultural continuity with Roman times 

through the Middle Ages (1971), and Horden and Purcell’s notion of the Mediterranean region as 

comprising ecologically diverse microregions between which exchange and connectivity ebbed 

and flowed, but never entirely ceased (2000). Thus, the totality of the evidence supports a view 

of the transitional period between the 5th and 12th centuries, not as a “Dark Age” characterized 

by collapse, but as an era of continuity and resilience for Rusellae. 

194 

 
 
 
 
REFERENCES 

Adams, D. M., & Pilloud, M. A. (2021). The (mis) appropriation of biological anthropology in 

race science and the implications for forensic anthropology. Forensic Anthropology, 4(4), 
97.  

Agarwal, S. C. (2012). The past of sex, gender, and health: Bioarchaeology of the aging skeleton. 

American Anthropologist, 114(2), 322-335.  

Agarwal, S. C. (2016). Bone morphologies and histories: Life course approaches in 
bioarchaeology. American Journal of Physical Anthropology, 159, 130-149.  

Albanese, J., & Saunders, S. R. (2006). Is it possible to escape racial typology in forensic 

identification? In A. Schmitt, E. Cunha, & J. Pinheiro (Eds.), Forensic anthropology and 
medicine: Complementary sciences from recovery to cause of death (pp. 281-316). 
Totowa, NJ: Humana Press. 

Alberti, A. (2023). The City of Pisae from Late Antiquity to the Early Medieval Period. In R. 

Rao & A. Sebastiani (Eds.), Archaeological Landscapes of Late Antique and Early 
Medieval Tuscia: Research and Field Papers (pp. 21-38). Turnhout, Belgium: Brepols. 

Algee-Hewitt, B. F. (2016). Population Inference from Contemporary American Craniometrics. 

American Journal of Physical Anthropology, 160, 604-624.  

Algee-Hewitt, B. F., Hughes, C. E., & Anderson, B. E. (2018). Temporal, geographic and 

identification trends in craniometric estimates of ancestry for persons of Latin American 
origin. Forensic Anthropology (University of Florida), 1(1).  

Algee‐Hewitt, B. F. (2017a). Geographic substructure in craniometric estimates of admixture for 
contemporary American populations. American Journal of Physical Anthropology, 
164(2), 260-280.  

Algee‐Hewitt, B. F. (2017b). Temporal trends in craniometric estimates of admixture for a 

modern American sample. American Journal of Physical Anthropology, 163(4), 729-740.  

Alt, K. W., Knipper, C., Peters, D., Müller, W., Maurer, A.-F., Kollig, I., . . . von Freeden, U. 

(2014). Lombards on the Move--An Integrative Study of the Migration Period Cemetery 
at Szólád, Hungary. PLoS ONE, 9(11), 1-15.  

Amorim, C. E. G., Vai, S., Posth, C., Modi, A., Koncz, I., Hakenbeck, S., . . . Veeramah, K. R. 
(2018). Understanding 6th-century barbarian social organization and migration through 
paleogenomics. Nature Communications, 9, 1-11.  

Antonio, M. L., Gao, Z., Moots, H. M., Lucci, M., Candilio, F., Sawyer, S., . . . Pritchard, J. K. 

(2019). Ancient Rome: A Genetic Crossroads of Europe and the Mediterranean. Science, 
366, 708-714.  

195 

 
 
 
 
 
 
 
 
 
 
 
 
Arauna, L. R., Hellenthal, G., & Comas, D. (2019). Dissecting human North African gene-flow 

into its western coastal surroundings. Proceedings of the Royal Society B, 286(1902), 
20190471.  

Arnold, J., Bjornlie, S., & Sessa, K. (2016). A Companion to Ostrogothic Italy. Boston, UNITED 

STATES: BRILL. 

Auerbach, B. M. (2023). Biological Distance. In A. M. Pollard, R. A. Armitage, & C. A. 

Makarewicz (Eds.), Handbook of Archaeological Sciences (Second Edition ed., Vol. 1, 
pp. 387-404). Hoboken, NJ: Wiley. 

Baker, B. J., & Agarwal, S. C. (2017). Stronger Together: Advancing a Global Bioarchaeology. 

Bioarchaeology International, 1(1-2), 1-18.  

Barbiera, I., Castiglione, M., & Dalla-Zuanna, G. (2017). Missing Women in the Italian Middle 

Ages? Data and Interpretation. In S. R. Huebner & G. Nathan (Eds.), Mediterranean 
Families in Antiquity: Households, Extended Families, and Domestic Space (pp. 285-
309). Chichester: Wiley Blackwell. 

Barbiera, I., & Dalla-Zuanna, G. (2009). Population Dynamics in Italy in the Middle Ages: New 
Insights from Archaeological Findings. Population and Development Review, 35(2), 367-
389.  

Barker, G., & Rasmussen, T. (1998). The Etruscans. Oxford: Blackwell Publishers. 

Bat-Sheva, A. (2010). Christians and Jews. In T. Noble & J. Smith (Eds.), The Cambridge history 
of Christianity, vol. 3, Early Medieval Christianities, c. 600-c. 1100. Cambridge: 
Cambridge University Press. 

Beatrice, J. S., Soler, A., Reineke, R. C., & Martínez, D. E. (2021). Skeletal evidence of 

structural violence among undocumented migrants from Mexico and Central America. 
American Journal of Physical Anthropology, 176(4), 584-605.  

Beck, J., & Quinn, C. P. (2022). Balancing the scales: archaeological approaches to social 

inequality. World Archaeology, 54(4), 572-583. doi:10.1080/00438243.2023.2169341 

Belcastro, G., Rastelli, E., Mariotti, V., Consiglio, C., Facchini, F., & Bonfiglioli, B. (2007). 

Continuity or discontinuity of the life-style in Central Italy during the Roman Imperial 
Age-Early Middle Ages Transition: Diet, Health, and Behavior. American Journal of 
Physical Anthropology, 132, 381-394.  

Bellotti, P., Caputo, C., Davoli, L., Evangelista, S., Garzanti, E., Pugliese, F., & Valeri, P. (2004). 

Morpho-sedimentary characteristics and Holocene evolution of the emergent part of the 
Ombrone River delta (southern Tuscany). Geomorphology, 61(1-2), 71-90.  

196 

 
 
 
 
 
 
 
 
 
 
 
 
 
Bentley, A. R. (2006). Strontium isotopes from the earth to the archaeological skeleton: a review. 

Journal of Archaeological Method and Theory, 13(3), 135-187.  

Berg, G. E., & Kenyhercz, M. W. (2017). Introducing human mandible identification [(hu) 
MANid]: a free, web‐based GUI to classify human mandibles. Journal of Forensic 
Sciences, 62(6), 1592-1598.  

Bertoldi, S., Castiglia, G., & Castrorao Barba, A. (2019). A multi-scalar approach to long-term 
dynamics, spatial relations, and economic networks of Roman secondary settlements in 
Italy and the Ombrone Valley system (Southern Tuscany): Towards a Model? In P. 
Verhagen, J. Joyce, & M. R. Groenhuijzen (Eds.), Finding the Limits of the Limes: 
Modelling Demography, Economy, and Transport on the Edge of the Roman Empire: 
Springer. 

Bethard, J. D., & DiGangi, E. A. (2020). Moving beyond a lost cause: Forensic anthropology and 
ancestry estimates in the United States. Journal of Forensic Sciences, 65(5), 1791.  

Bianchi, G., & Hodges, R. (Eds.). (2018). Origins of a new economic union (7th-12th centuries): 
preliminary results of the nEU-Med project: October 2015-March 2017. Florence: 
Edizioni All’Insegna del Giglio. 

Boas, F. (1912). Changes in the bodily form of descendants of immigrants. New York: Columbia 

University Press. 

Boccolini, D., Toma, L., Luca, M. D., Severini, F., Cocchi, M., Bella, A., . . . Angeli, L. (2012). 

Impact of environmental changes and human-related factors on the potential malaria 
vector, Anopheles labranchiae (Diptera: Culicidae), in Maremma, Central Italy. Journal 
of medical entomology, 49(4), 833-842.  

Bocquet-Appel, J.-P., & Masset, C. (1982). Farewell to Paleodemography. Journal of Human 

Evolution, 11, 321-333.  

Border, R., Wang, J., Caggiano, C., Sankararaman, S., Schork, A. J., Turley, P., . . . Zaitlen, N. 

(2024). Simple models of non-random mating and environmental transmission bias 
standard human genetics statistical methods. bioRxiv, 2024.2010. 2016.618755.  

Botigué, L. R., Henn, B. M., Gravel, S., Maples, B. K., Gignoux, C. R., Corona, E., . . . Flores, 
C. (2013). Gene flow from North Africa contributes to differential human genetic 
diversity in southern Europe. Proceedings of the National Academy of Sciences, 110(29), 
11791-11796.  

Bound, M. (1987). Tubi fittili (vaulting tubes) from the sea--The Roman wreck at Punta del 

Fenaio, Island of Giglio. The International Journal of Nautical Archaeology and 
Underwater Exploration, 16(3), 187-200.  

197 

 
 
 
 
 
 
 
 
 
 
 
 
Broehl, K. A. (2023). Biological Distance Analysis of Identity, Postmarital Residence, and 

Population Structure at the CA-ALA-11 Shellmound.  

Brogiolo, G. P. (1999). Ideas of the Town in Italy during the Transition from Antiquity to the 

Middle Ages. In G. P. Brogiolo & B. Ward-Perkins (Eds.), The Idea and Ideal of the Town 
between Late Antiquity and the Early Middle Ages (Vol. 4, pp. 99-126). Leiden: Brill. 

Brogiolo, G. P. (2000). Towns, Forts and the Countryside: Archaeological Models for Northern 

Italy in the Early Lombard Period (AD 568-650). In G. P. Brogiolo, N. Gauthier, & N. 
Christie (Eds.), Towns and their territories between Late Antiquity and the Early Middle 
Ages (pp. 403). Leiden: Brill. 

Brogiolo, G. P., Marazzi, F., & Giostra, C. (Eds.). (2017). I Longobardi: Un Popolo che Cambia 

la Storia: Skira. 

Brotherton, P., Endicott, P., Sanchez, J. J., Beaumont, M., Barnett, R., Austin, J., & Cooper, A. 
(2007). Novel high-resolution characterization of ancient DNA reveals C> U-type base 
modification events as the sole cause of post mortem miscoding lesions. Nucleic acids 
research, 35(17), 5717-5728.  

Brown, P. (1971). The World of Late Antiquity: From Marcus Aurelius to Muhammad. London: 

Thames and Hudson. 

Brown, T. A. (2023). Archaeogenetics. In Handbook of Archaeological Sciences (Second Edition 

ed., Vol. 2, pp. 607-612). Hoboken, NJ: Wiley. 

Bruun, C. (2010). Water, oxygen isotopes, and immigration to Ostia-Portus. Journal of Roman 

Archaeology, 23, 109-132.  

Buikstra, J. E. (2006). Chapter 1: A Historical Introduction. In J. E. Buikstra & L. A. Beck (Eds.), 

Bioarchaeology: The Contextual Analysis of Human Remains (pp. 7-25). Amsterdam: 
Academic Press Elsevier. 

Buikstra, J. E., DeWitte, S. N., Agarwal, S. C., Baker, B. J., Bartelink, E. J., Berger, E., . . . 

Zakrzewski, S. R. (2022). Twenty-first century bioarchaeology: Taking stock and moving 
forward. American Journal of Biological Anthropology, 178(S74), 54-114. 
doi:https://doi.org/10.1002/ajpa.24494 

Buikstra, J. E., & Ubelaker, D. H. (Eds.). (1994). Standards for Data Collection from Human 

Skeletal Remains. Fayetteville, Arkansas: Arkansas Archeological Survey. 

Busby, G. B., Hellenthal, G., Montinaro, F., Tofanelli, S., Bulayeva, K., Rudan, I., . . . 

Karachanak-Yankova, S. (2015). The role of recent admixture in forming the 
contemporary West Eurasian genomic landscape. Current biology, 25(19), 2518-2526.  

198 

 
 
 
 
 
 
 
 
 
 
 
 
 
Campana, S. R. L. (2017). Emptyscapes: filling an 'empty' Mediterranean landscape at Rusellae, 

Italy. Antiquity, 91(359), 1223-1240.  

Campana, S. R. L. (2018). Mapping the Archaeological Continuum: Filling 'Empty' 

Mediterranean Landscapes: Springer International Publishing. 

Campana, S. R. L. (2021). Archaeology of former and historical urban landscapes in the 

Mediterranean world: current trends and future perspectives. ARCHEOLOGIE: TEMI, 
CONTESTI, MATERIALI, 2, 31-49.  

Campana, S. R. L. (2022). Archaeology of once-townscapes and rural landscapes in the 

Mediterranean. In D. Filippi (Ed.), Rethinking the Roman City: The Spatial Turn and the 
Archaeology of Roman Italy. 

Campana, S. R. L. (2023). Infrastructure, Water Management, Settlement, Agriculture, and 

Funerary Landscapes near Rusellae in the longue duree. In R. Rao & A. Sebastiani (Eds.), 
Archaeological Landscapes of Late Antique and Early Medieval Tuscia: Research and 
Field Papers (pp. 39-54). Turnhout, Belgium: Brepols. 

Capelli, C., Onofri, V., Brisighelli, F., Boschi, I., Scarnicci, F., Masullo, M., . . . Gusmao, L. 
(2009). Moors and Saracens in Europe: estimating the medieval North African male 
legacy in southern Europe. European Journal of Human Genetics, 17(6), 848-852.  

Carson, E. A. (2006a). Maximum-likelihood variance components analysis of heritabilities of 

cranial nonmetric traits. Human Biology, 78(4), 383-402.  

Carson, E. A. (2006b). Maximum likelihood estimation of human craniometric heritabilities. 

American Journal of Physical Anthropology, 131(2), 169-180.  

Celuzza, M. G., Cencetti, S., & Pacciani, E. (2002). Scavi nel cimitero della più antica cattedrale 
di Roselle (Grosseto). Vi-Xii sec. d.C. Paper presented at the Antropologia del Medioevo: 
Biologia e Cultura, Sestino. 

Celuzza, M. G., Watson, E. O., Goots, A. C., Isa, M. I., Beatrice, J. S., Streetman, E. R., & 

Fenton, T. W. (2023). Late Antique and Early Medieval Rusellae: Archaeology and the 
Anthropology of the Cemetery of the Episcopal Church. In R. Rao & A. Sebastiani 
(Eds.), Archaeological Landscapes of Late Antique and Early Medieval Tuscia: Research 
and Field Papers (pp. 65-87). Turnhout, Belgium: Brepols. 

Chirico, E., & Sebastiani, A. (2023). The Coastal ager Rusellanus between the Fourth and Sixth 

Centuries AD. In R. Rao & A. Sebastiani (Eds.), Archaeological Landscapes of Late 
Antique and Early Medieval Tuscia: Research and Field Papers (pp. 88-106). Turnhout, 
Belgium: Brepols. 

199 

 
 
 
 
 
 
 
 
 
 
 
 
Cilli, J. (2021). Influence of the Roman road network on the biological proximity of Italic 

Abruzzo populations through mean measure of divergence analysis. Archaeological and 
Anthropological Sciences, 13(12), 214.  

Coia, V., Paladin, A., Zingale, S., Wurst, C., Croze, M., Maixner, F., & Zink, A. (2023). Ancestry 

and kinship in a Late Antiquity-Early Middle Ages cemetery in the Eastern Italian Alps. 
iScience, 26(11), 108215. doi:10.1016/j.isci.2023.108215 

Cole, T. J. (2000). Secular trends in growth. Proceedings of the Nutrition Society, 59(2), 317-324.  

Collard, M., & Wood, B. (2007). Hominin homoiology: an assessment of the impact of 

phenotypic plasticity on phylogenetic analyses of humans and their fossil relatives. 
Journal of Human Evolution, 52(5), 573-584.  

Cook, D. (1976). The epidemiology of periostitis in prehistoric Illinois. American Journal of 

Physical Anthropology, 44, 171.  

Dallai, L., & Marasco, L. (2023). The nEU-Med Project: Archaeology of a Coastal District in 

Tuscany during the Early Middle Ages. In R. Rao & A. Sebastiani (Eds.), Archaeological 
Landscapes of Late Antique and Early Medieval Tuscia: Research and Field Papers (pp. 
107-125). Turnhout, Belgium: Brepols. 

De Giorgi, A. U., Hobart, M., Ludke, M., & Scott, R. T. (2023). Cosa during Late Antiquity. In 

R. Rao & A. Sebastiani (Eds.), Archaeological Landscapes of Late Antique and Early 
Medieval Tuscia: Research and Field Papers (pp. 126-140). Turnhout, Belgium: Brepols. 

de la Cova, C. (2011). Race, health, and disease in 19th‐century‐born males. American Journal 

of Physical Anthropology, 144(4), 526-537.  

de la Cova, C. (2012). Patterns of trauma and violence in 19th-century-born African American 
and Euro-American females. International Journal of Paleopathology, 2(2-3), 61-68.  

De Maesschalck, R., Jouan-Rimbaud, D., & Massart, D. L. (2000). The Mahalanobis distance. 

Chemometrics and Intelligent Laboratory Systems, 50(1), 1-18. 
doi:https://doi.org/10.1016/S0169-7439(99)00047-7 

DeWitte, S. N. (2014). Health in post‐Black Death London (1350–1538): Age patterns of 

periosteal new bone formation in a post‐epidemic population. American Journal of 
Physical Anthropology, 155(2), 260-267.  

DeWitte, S. N., & Kowaleski, M. (2017). Black Death Bodies (2017). Fragments: 

Interdisciplinary Approaches to the Study of Ancient and Medieval Pasts, 7.  

Di Pietro, L., Barba, M., Palacios, D., Tiberio, F., Prampolini, C., Baranzini, M., . . . Lattanzi, W. 
(2021). Shaping modern human skull through epigenetic, transcriptional and post-

200 

 
 
 
 
 
 
 
 
 
 
 
 
 
transcriptional regulation of the RUNX2 master bone gene. Scientific Reports, 11(1), 
21316. doi:10.1038/s41598-021-00511-3 

DiGangi, E. A., & Bethard, J. D. (2021). Uncloaking a lost cause: Decolonizing ancestry 

estimation in the United States. American Journal of Physical Anthropology, 175(2), 422-
436.  

Diggelmann, L. (2016). Marriage, Peace, and Enmity in the Twelfth Century. Common 

Knowledge, 22(2), 237-255.  

Dotti, G. (2020). Racism, xenophobia, and migration in Italy, a Post‐Catholic country. The 

Ecumenical Review, 72(1), 37-47.  

Dudzik, B. (2019). Landmark and semilandmark data collection using digitizers and data 

processing. In N. Seguchi & B. Dudzik (Eds.), 3D data acquisition for bioarchaeology, 
forensic anthropology, and archaeology (pp. 59-69): Elsevier. 

Dudzik, B., & Kolatorowicz, A. (2016). Craniometric data analysis and estimation of 

biodistance. In J. T. Hefner & M. A. Pilloud (Eds.), Biological Distance Analysis: 
Forensic and Bioarchaeological Perspectives: Elsevier. 

Dunn, R. R., Spiros, M. C., Kamnikar, K. R., Plemons, A. M., & Hefner, J. T. (2020). Ancestry 
estimation in forensic anthropology: A review. WIREs Forensic Science, e1369.  

Edgar, H. J. (2014). Dental morphological estimation of ancestry in forensic contexts. In G. E. 

Berg & S. C. Ta'ala (Eds.), Biological affinity in forensic identification of human skeletal 
remains: Beyond Black and White (pp. 191-208). Boca Raton: CRC Press. 

Edgar, H. J., & Ousley, S. (2013). New approaches to the use of dental morphology in forensic 

contexts. In G. R. Scott & J. D. Irish (Eds.), Anthropological Perspectives on Tooth 
Morphology : Genetics, Evolution, Variation. Cambridge: Cambridge University Press. 

Edgar, H. J., & Ousley, S. D. (2021). Testing the Homogeneity of" White": Dental Morphology 

in Americans and Australians of European Descent. Forensic Anthropology (University of 
Florida), 4(4).  

Edgar, H. J., & Pilloud, M. (2021). A Reassessment of Assessing Race:" Ancestry" Estimation 

and Its Implications for Forensic Anthropology and Beyond. Forensic Anthropology 
(University of Florida), 4(4).  

Edinborough, M., & Rando, C. (2020). Stressed Out: Reconsidering stress in the study of 
archaeological human remains. Journal of Archaeological Science, 121, 105197. 
doi:https://doi.org/10.1016/j.jas.2020.105197 

Elliott, A. B. (2018). Internet medievalism and the white Middle Ages. History Compass, 16(3), 

e12441.  

201 

 
 
 
 
 
 
 
 
 
 
 
 
 
Emery, M. (2018). Assessing Migration and Demographic Change in pre-Roman and Roman 
Period Southern Italy Using Whole-Mitochondrial DNA and Stable Isotope Analysis. 
(PhD), McMaster University, Hamilton, Ontario, Canada.  

Emery, M., Duggan, A. T., Murchie, T. J., Stark, R. J., Klunk, J., Hider, J., . . . Prowse, T. L. 
(2018). Ancient Roman mitochondrial genomes and isotopes reveal relationships and 
geographic origins at the local and pan-Mediterranean scales. Journal of Archaeological 
Science: Reports, 20, 200-209.  

Emery, M., Stark, R. J., Murchie, T. J., Elford, S., Schwarcz, H. P., & Prowse, T. L. (2018). 

Mapping the origins of Imperial Roman workers (1st-4th century CE) at Vagnari, 
Southern Italy, using 87Sr/86Sr and delta 18O variability. American Journal of Physical 
Anthropology, 166, 837-850.  

Favero, P. (2010). Italians, the “good people”: Reflections on national self-representation in 

contemporary Italian debates on xenophobia and war. Outlines. Critical Practice Studies, 
12(2), 138-153.  

Fazzi, L. (2015). Social work, exclusionary populism and xenophobia in Italy. International 

Social Work, 58(4), 595-605.  

Fentress, E. (Ed.) (2003). Cosa V: An Intermittent Town. Ann Arbor: University of Michigan 

Press. 

Fenwick, C. (2013). From Africa to Ifrīqiya: settlement and society in Early Medieval North 

Africa (650–800). Al-Masaq, 25(1), 9-33.  

Fiorito, G., Di Gaetano, C., Guarrera, S., Rosa, F., Feldman, M. W., Piazza, A., & Matullo, G. 
(2016). The Italian genome reflects the history of Europe and the Mediterranean basin. 
European Journal of Human Genetics, 24(7), 1056-1062.  

Fleischman, J., & Crowder, C. M. (2018). Standard Operating Procedure for MicroScribe 3-

Dimensional Digitizer and Craniometric Data. In. Houston, TX: Harris County Institute 
of Forensic Sciences, Forensic Anthropology Division. 

Fletcher Jr., R. J., Sefair, J. A., Kortessis, N., Jaffe, R., Holt, R. D., Robertson, E. P., . . . Austin, 

J. D. (2022). Extending isolation by resistance to predict genetic connectivity. Methods in 
Ecology and Evolution, 13(11), 2463-2477.  

Fornaciari, G., Giuffra, V., Ferroglio, E., Gino, S., & Bianucci, R. (2010). Plasmodium 

falciparum immunodetection in bone remains of members of the Renaissance Medici 
family (Florence, Italy, sixteenth century). Transactions of the Royal Society of Tropical 
Medicine and Hygiene, 104(9), 583-587.  

Fornaciari, G., & Mallegni, F. (1985). Le ossa umane. In A. Ricci (Ed.), Settefinestre: una villa 

schiavistica nell'Eturia romana (pp. 275-277). Modena: Edizioni Panini. 

202 

 
 
 
 
 
 
 
 
 
 
 
 
Francisci, G., Micarelli, I., Iacumin, P., Castorina, F., Di Vincenzo, F., Di Matteo, M., . . . Tafuri, 

M. A. (2020). Strontium and oxygen isotopes as indicators of Longobards mobility in 
Italy: an investigation at Povegliano Veronese. Scientific Reports, 10(1), 11678. 
doi:10.1038/s41598-020-67480-x 

Francovich, R. (2007). The hinterlands of Early Medieval towns: the transformation of the 

countryside in Tuscany. In J. Henning (Ed.), Post-Roman Towns, Trade and Settlement in 
Europe and Byzantium (Vol. Vol. 1: The Heirs of the Roman West, pp. 135-152). Berlin: 
Walter de Gruyter. 

Francovich, R. (2008). The Beginnings of Hilltop Villages in Early Medieval Tuscany. In J. R. 

Davis & M. McCormick (Eds.), The long morning of medieval Europe: new directions in 
Early Medieval studies (pp. 55-82). Aldershot: Ashgate Pub. Co. 

Giannecchini, M., & Moggi-Cecchi, J. (2008). Stature in archaeological samples for Central 
Italy: Methodological issues and diachronic changes. American Journal of Physical 
Anthropology, 135, 284-292.  

Giles, E., & Elliot, O. (1962). Race Identification from Cranial Measurements. Journal of 

Forensic Sciences, 7(2), 147-157.  

Go, M. C., Yukyi, N., & Chu, E. Y. (2021). On WEIRD Anthropologists and Their White 

Skeletons. Forensic Anthropology (University of Florida), 4(4).  

Goodman, A. H., Martin, D. L., & Armelagos, G. J. (1984). Indications of stress from bone and 
teeth. In M. N. Cohen & G. J. Armelagos (Eds.), Paleopathology at the Origins of 
Agriculture (pp. 13-49). New York: Academic Press. 

Goodman, A. H., Thomas, R. B., & Swedlund, A. C. (1988). Biocultural perspectives on stress in 
prehistoric, historical, and contemporary population research. Yearbook of Physical 
Anthropology, 31, 169-202.  

Gowland, R. (2017). Embodied identities in Roman Britain: a bioarchaeological approach. 

Britannia, 48, 177-194.  

Gowland, R., & Thompson, T. (2013). Human Identity and Identification. Cambridge: 

Cambridge University Press. 

Gowland, R., & Western, A. G. (2012). Morbidity in the marshes: Using spatial epidemiology to 
investigate skeletal evidence for malaria in Anglo‐Saxon England (AD 410–1050). 
American Journal of Physical Anthropology, 147(2), 301-311.  

Grant, M. (1980). The Etruscans. London: Weidenfeld and Nicolson. 

203 

 
 
 
 
 
 
 
 
 
 
 
 
 
Gravlee, C. C., Bernard, H. R., & Leonard, W. R. (2003a). Boas's "Changes in Bodily Form": 

The immigrant study, cranial plasticity, and Boas's physical anthropology. American 
Anthropologist, 105(2), 326-332.  

Gravlee, C. C., Bernard, H. R., & Leonard, W. R. (2003b). Heredity, Environment, and Cranial 

Form: A Reanalysis of Boas' Immigrant Data. American Anthropologist, 105(1), 125-138.  

Gregoricka, L. A. (2021). Moving forward: a bioarchaeology of mobility and migration. Journal 

of Archaeological Research, 1-55.  

Hallgrímsson, B., Lieberman, D. E., Liu, W., Ford‐Hutchinson, A., & Jirik, F. (2007). Epigenetic 
interactions and the structure of phenotypic variation in the cranium. Evolution & 
development, 9(1), 76-91.  

Hanihara, K. (1967). Racial characteristics in the dentition. Journal of Dental Research, 46(5), 

923-926.  

Harrod, R. P., Thompson, J. L., & Martin, D. L. (2012). Hard labor and hostile encounters: What 
human remains reveal about institutional violence and Chinese immigrants living in 
Carlin, Nevada (1885–1923). Historical Archaeology, 46(4), 85-111.  

Haywood, J. (2015). Northmen: the Viking Saga AD 793-1241. New York: St. Martin's Press. 

Hefner, J. T. (2009). Cranial nonmetric variation and estimating ancestry. Journal of Forensic 

Sciences, 54(5), 985-995.  

Hefner, J. T., & Ousley, S. (2014). Statistical classification methods for estimating ancestry using 

morphoscopic traits. Journal of Forensic Sciences, 59(4), 883-890.  

Hefner, J. T., Ousley, S., & Richardson, R. (2024). MaMD Analytical 1.0: A Computer Program 
for Macromorphoscopic Trait Analysis in Forensic Anthropology. Forensic Anthropology 
(University of Florida), 7.  

Hefner, J. T., Pilloud, M. A., Buikstra, J. E., & Vogelsberg, C. C. M. (2016). A Brief History of 

Biological Distance Analysis. In M. A. Pilloud & J. T. Hefner (Eds.), Biological Distance 
Analysis: Forensic and Bioarchaeological Perspectives: Elsevier. 

Hefner, J. T., Spradley, M. K., & Anderson, B. (2014). Ancestry assessment using random forest 

modeling. Journal of Forensic Sciences, 59(3), 583-589.  

Heng, G. (2011). The invention of race in the European middle ages I: Race studies, modernity, 

and the middle ages 1. Literature Compass, 8(5), 315-331.  

Hens, S. M., & Ross, A. H. (2017). Cranial Variation and Biodistance in Imperial Roman 

Cemeteries. International Journal of Osteoarchaeology, 27, 880-887.  

204 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Herrera, B., Hanihara, T., & Godde, K. (2014). Comparability of multiple data types from the 
Bering Strait region: Cranial and dental metrics and nonmetrics, mtDNA, and Y‐
chromosome DNA. American Journal of Physical Anthropology, 154(3), 334-348.  

Higuchi, R., Bowman, B., Freiberger, M., Ryder, O. A., & Wilson, A. C. (1984). DNA sequences 

from the quagga, an extinct member of the horse family. Nature, 312(5991), 282-284.  

Hillson, S. (2014). Tooth development in human evolution and bioarchaeology. Cambridge: 

Cambridge University Press. 

Hodges, R., & Whitehouse, D. (1983). Mohammed, Charlemagne, and the Origins of Europe: 
The Pirenne Thesis in Light of Archaeology. Ithaca: Cornell University Press. 

Holmes, G. (Ed.) (1988). The Oxford Illustrated History of Medieval Europe. New York: Oxford 

University Press. 

Horden, P., & Purcell, N. (2000). The Corrupting Sea: A Study of Mediterranean History. 

Malden, Mass.: Blackwell. 

Hosek, L., & Robb, J. (2019). Osteobiography: A platform for bioarchaeological research. 

Bioarchaeology International, 3(1), 1.  

Howells, W. W. (1973). Cranial Variation in Man: A study by multivariate analysis of patterns of 
differences among recent human populations. Cambridge, Massachusetts: Peabody 
Museum. 

Howells, W. W. (1989). Skull Shapes and the Map: Craniometric Analyses in the Despersion of 

Modern Homo. Peabody Museum of Archaeology and Ethnology, Harvard Univ., 79.  

Howells, W. W. (1996). Howells' craniometric data on the internet. American Journal of Physical 

Anthropology, 101(3), 441-442.  

Hu, D. (2013). Approaches to the Archaeology of Ethnogenesis: Past and Emergent Perspectives. 

Journal of Archaeological Research, 21, 371-402.  

Hughes, C. E., Juarez, C., & Yim, A. D. (2021). Forensic anthropology casework performance: 

Assessing accuracy and trends for biological profile estimates on a comprehensive 
sample of identified decedent cases. Journal of Forensic Sciences, 66(5), 1602-1616.  

Irish, J. D. (2015). Assessing dental nonmetric variation among populations. In J. D. Irish & G. 

R. Scott (Eds.), A companion to dental anthropology (pp. 265-286): Wiley. 

Iscan, M. Y., Loth, S. R., & Wright, R. K. (1984). Age estimation from the rib by phase analysis: 

White males. Journal of Forensic Sciences, 29(4), 1094-1104.  

205 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Iscan, M. Y., Loth, S. R., & Wright, R. K. (1985). Age estimation from the rib by phase analysis: 

White females. Journal of Forensic Sciences, 30(3), 853-863.  

Jantz, R. L., & Ousley, S. D. (2005). FORDISC 3. Knoxville, TN: University of Tenessee.  

Kabacoff, R. (2022). R in action: data analysis and graphics with R and Tidyverse: Simon and 

Schuster. 

Katz, D. C., Grote, M. N., & Weaver, T. D. (2016). A Mixed Model for the Relationship Between 

Climate and Human Cranial Form. American Journal of Physical Anthropology, 160, 
593-603.  

Katzenberg, M. A. (2008). Stable isotope analysis: a tool for studying past diet, demography, and 
life history. In M. A. Katzenberg & S. R. Saunders (Eds.), Biological anthropology of the 
human skeleton (Second ed., pp. 467-504): John Wiley & Sons, Inc. 

Kenyhercz, M. W., & Passalacqua, N. V. (2016). Missing Data Imputation Methods and their 
Performance with Biodistance Analyses. In M. A. Pilloud & J. T. Hefner (Eds.), 
Biological Distance Analysis: Forensic and Bioarchaeological Perspectives. San Diego: 
Academic Press. 

Key, P. J. (1982). Craniometric Relationships among Plains Indians: Culture-Historical and 

Evolutionary Implications. (PhD), University of Tennessee,  

Kieser, J. A. (1990). Human adult odontometrics: the study of variation in adult tooth size: 

Cambridge University Press. 

Killgrove, K. (2005). Bioarchaeology in the Roman World. (Master of Arts), University of North 

Carolina at Chapel Hill, Chapel Hill.  

Killgrove, K., & Montgomery, J. (2016). All roads lead to Rome: Exploring human migration to 

the Eternal City through biochemistry of skeletons from two Imperial-Era cemeteries 
(1st-3rd c AD). PLoS ONE, 11(2), 1-30.  

Kim, H.-Y. (2017). Statistical notes for clinical researchers: Chi-squared test and Fisher's exact 

test. Restorative Dentistry & Endodontics, 42(2), 152-155. 
doi:10.5395/rde.2017.42.2.152 

Klaus, H. D. (2014). Frontiers in the bioarchaeology of stress and disease: Cross‐disciplinary 

perspectives from pathophysiology, human biology, and epidemiology. American Journal 
of Physical Anthropology, 155(2), 294-308.  

Knudson, K. J., & Stojanowski, C. M. (2008). New Directions in Bioarchaeology: Recent 
Contributions to the Study of Human Social Identities. Journal of Archaeological 
Research, 16, 397-432.  

206 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Knudson, K. J., & Stojanowski, C. M. (2020). Bioarchaeology and Identity Revisited. 

Gainesville: University of Florida Press. 

Kuzawa, C. W. (2005). Fetal origins of developmental plasticity: are fetal cues reliable predictors 

of future nutritional environments? American Journal of Human Biology, 17(1), 5-21.  

Kyle, B., Shehi, E., Koçi, M., & Reitsema, L. J. (2020). Bioarchaeological reconstruction of 

physiological stress during social transition in Albania. International Journal of 
Paleopathology, 30, 118-129.  

Larsen, C. S. (2002). Bioarchaeology: The Lives and Lifestyles of Past People. Journal of 

Archaeological Research, 10(2), 119-166.  

Larsen, C. S. (2015). Bioarchaeology: Interpreting Behavior from the Human Skeleton (Second 

ed.). Cambridge: Cambridge University Press. 

Larsen, C. S. (2018). The bioarchaeology of health crisis: Infectious disease in the past. Annual 

Review of Anthropology, 47(1), 295-313.  

Leach, S., Eckardt, H., Chenery, C., Müldner, G., & Lewis, M. (2010). A Lady of York: 

migration, ethnicity and identity in Roman Britain. Antiquity, 84, 131-145.  

Leach, S., Lewis, M., Chenery, C., Müldner, G., & Eckardt, H. (2009). Migration and Diversity 

in Roman Britain: A Multidisciplinary Approach to the Identification of Immigrants in 
Roman York, England. American Journal of Physical Anthropology, 140, 546-561.  

Lewis, M., & Roberts, C. (1997). Growing pains: the interpretation of stress indicators. 

International Journal of Osteoarchaeology, 7(6), 581-586. 
doi:https://doi.org/10.1002/(SICI)1099-1212(199711/12)7:6<581::AID-
OA325>3.0.CO;2-C 

Lewis, T. E. (2016). ‘Like Devils Out of Hell’: Reassessing the African Presence in Early 

Modern England. Black theology, 14(2), 107-120.  

Liversidge, H. M. (2000). Crown formation times of human permanent anterior teeth. Archives of 
Oral Biology, 45(9), 713-721. doi:https://doi.org/10.1016/S0003-9969(00)00055-8 

Lopez‐Jantzen, N. (2019). Between empires: Race and ethnicity in the Early Middle Ages. 

Literature Compass, 16(9-10), e12542.  

Lovejoy, C. O., Meindl, R. S., Pryzbeck, T. R., & Mensforth, R. P. (1985). Chronological 
metamorophosis of the auricular surface of the ilium: A new method for the 
determination of adult skeletal age at death. American Journal of Physical Anthropology, 
68, 15-28.  

207 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Maaranen, N., Zakrzewski, S., Kharobi, A., Stantis, C., Prell, S., Bietak, M., & Schutkowski, H. 
(2021). The people of Avaris: Intra-regional biodistance analysis using dental non-metric 
traits. Bioarchaeology of the Near East, 15, 1-24.  

Maass, C. K. (2023). Race, Gender, and Intersectionality in the Bioarchaeology of the African 

Diaspora: Perspectives from Colonial Peru. Journal of Archaeological Method and 
Theory, 30(3), 805-827. doi:10.1007/s10816-022-09577-3 

Manzi, G., Salvadei, L., Vienna, A., & Passarello, P. (1999). Discontinuity of life conditions at 

the transition from the Roman Imperial Age to the Early Middle Ages: Example from 
central Italy evaluated by pathological dento-alvelolar lesions. American Journal of 
Human Biology, 11, 327-341.  

Mardini, M., Badawi, A., Zaven, T., Gergian, R., & Nikita, E. (2023). Bioarchaeological 

perspectives to mobility in Roman Phoenicia: A biodistance study based on dental 
morphology. Journal of Archaeological Science: Reports, 47, 103759.  

Martin, R. (1914). Lehrbuch der Anthropologie in systematischer Darstellung mit besonderer 

Berucksichtigung der anthropologischen Methoden. Jena: Gustav Fischer. 

Martínez-Abadías, N., Esparza, M., Sjøvold, T., González-José, R., Santos, M., & Hernández, M. 
(2009). Heritability of human cranial dimensions: comparing the evolvability of different 
cranial regions. Journal of Anatomy, 214, 19-35.  

McCann, A. M. (2008). Cosa and Deep Sea Exploration. Memoirs of the American Academy in 

Rome, 6, 37-50.  

McCormick, M. (2001). Origins of the European Economy: Communications and Commerce AD 

300-900. Cambridge: Cambridge University Press. 

McKeown, A. H., & Jantz, R. L. (2005). Comparison of coordinate and craniometric data for 

biological distance studies. In Modern morphometrics in physical anthropology (pp. 215-
230): Springer. 

Meinzer, N., Steckel, R., & Baten, J. (2019). Agricultural specialization, urbanization, workload, 

and stature. In R. Steckel, C. S. Larsen, C. Roberts, & J. Baten (Eds.), The backbone of 
Europe: Health, diet, work and violence over two millennia (Vol. 80, pp. 231). 

Mensforth, R. P., Lovejoy, C. O., Lallo, J. W., & Armelagos, G. J. (1978). Part two: the role of 
constitutional factors, diet, and infectious disease in the etiology of porotic hyperostosis 
and periosteal reactions in prehistoric infants and children. Medical anthropology, 2(1), 
1-59.  

Meza‐Peñaloza, A., Zertuche, F., & Morehart, C. (2021). Population level comparisons in central 
Mexico using cranial nonmetric traits. American Journal of Physical Anthropology, 
176(2), 237-248.  

208 

 
 
 
 
 
 
 
 
 
 
 
 
Michelucci, M. (1985). Roselle: la domus dei mosaici. Montepulciano: Editori del Grifo. 

Mitchell, P. D. (2017). Human parasites in the Roman World: health consequences of conquering 

an empire. Parasitology, 144(1), 48-58.  

Mitchell, S. (2007). A History of the Later Roman Empire: AD 284-641. Oxford: Blackwell 

Publishing. 

Mitteroecker, P., & Schaefer, K. (2022). Thirty years of geometric morphometrics: 

Achievements, challenges, and the ongoing quest for biological meaningfulness. 
American Journal of Biological Anthropology, 178, 181-210.  

Montgomery, J. (2010). Passports from the past: Investigating human dispersals using strontium 

isotope analysis of tooth enamel. Annals of Human Biology, 37(3), 325-346.  

Mulligan, C. J. (2016). Early environments, stress, and the epigenetics of human health. Annual 

Review of Anthropology, 45(1), 233-249.  

Mullis, K. B., & Faloona, F. A. (1987). Specific synthesis of DNA in vitro via a polymerase-

catalyzed chain reaction. Methods in enzymology, 155, 335-350.  

Nikita, E. (2024). Human Mobility in the Central and Eastern Mediterranean during Hellenistic 
and Roman Times: The Potential and Limitations of Bioarchaeological Research. Nature 
Anthropology, 2(2), 10005.  

Nikita, E., Mardini, M., Mardini, M., & Degryse, P. (2022). SrIsoMed: An open access strontium 
isotopes database for the Mediterranean. Journal of Archaeological Science: Reports, 45, 
103606.  

O'Donnell, L., Hill, E. C., Anderson, A. S., & Edgar, H. J. H. (2022). A biological approach to 

adult sex differences in skeletal indicators of childhood stress. American Journal of 
Biological Anthropology, 177(3), 381-401.  

Ortner, D. J. (2003). Identification of Pathological Conditions in Human Skeletal Remains 

(Second Edition ed.). San Diego, CA: Academic Press. 

Ortner, D. J. (2012). Differential diagnosis and issues in disease classification. In A. L. Grauer 
(Ed.), A companion to paleopathology (pp. 250-267). Chichester, UK: Blackwell. 

Ossenberg, N. S. (2013). Brief communication: Cranial nonmetric trait database on the internet. 

American Journal of Physical Anthropology, 152(4), 551-553.  

Ousley, S. (2014). 3Skull (Version 1.76).  

Ousley, S. (2016). Forensic classification and biodistance in the 21st century: the rise of learning 

machines. In Biological Distance Analysis (pp. 197-212): Elsevier. 

209 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Ousley, S., Jantz, R., & Freid, D. (2009). Understanding race and human variation: why forensic 
anthropologists are good at identifying race. American Journal of Physical Anthropology, 
139(1), 68-76.  

Pääbo, S. (1985a). Molecular cloning of ancient Egyptian mummy DNA. Nature, 314(6012), 

644-645.  

Pääbo, S. (1985b). Preservation of DNA in ancient Egyptian mummies. Journal of 

Archaeological Science, 12(6), 411-417.  

Parsons, H. R. (2021). Ancestry Estimation in Practice: An Evaluation of Forensic Anthropology 
Reports in the United States. Forensic Anthropology (University of Florida), 4(4).  

Paul, K. S., & Stojanowski, C. M. (2017). Comparative performance of deciduous and 

permanent dental morphology in detecting biological relatives. American Journal of 
Physical Anthropology, 164(1), 97-116.  

Pederzani, S., & Britton, K. (2019). Oxygen isotopes in bioarchaeology: Principles and 
applications, challenges and opportunities. Earth-Science Reviews, 188, 77-107.  

Petkova, D., Novembre, J., & Stephens, M. (2016). Visualizing spatial population structure with 

estimated effective migration surfaces. Nature genetics, 48(1), 94-100.  

Phenice, T. W. (1969). A newly developed visual method of sexing the Os Pubis. American 

Journal of Physical Anthropology, 30, 297-302.  

Piccirilli, E., Sorrentino, R., Lugli, F., Bortolini, E., Silvestrini, S., Cavazzuti, C., . . . Köhler, K. 

(2023). New insights on Celtic migration in Hungary and Italy through the analysis of 
non-metric dental traits. PLoS ONE, 18(10), e0293090.  

Pillai, J. P., Parmar, G., Babu, R., & Vyas, J. (2023). Variability in the dental non-metric traits in 

maxillary central incisors among the geographically distinct populations in Gujarat, 
India: a cross-sectional study using the Turner-Scott dental anthropology system. Dental 
Anthropology Journal, 36(1).  

Pilloud, M. A., Edgar, H. J., George, R., & Scott, G. R. (2016). Dental Morphology in 

Biodistance Analysis. In M. A. Pilloud & J. T. Hefner (Eds.), Biological Distance 
Analysis (pp. 109-133). San Diego: Academic Press. 

Pilloud, M. A., & Kenyhercz, M. W. (2016). Dental Metrics in Biodistance Analysis. In M. A. 
Pilloud & J. T. Hefner (Eds.), Biological Distance Analysis (pp. 135-155). San Diego: 
Academic Press. 

Pilloud, M. A., & Schwitalla, A. W. (2020). Re-evaluating traditional markers of stress in an 

archaeological sample from central California. Journal of Archaeological Science, 116, 
105102. doi:https://doi.org/10.1016/j.jas.2020.105102 

210 

 
 
 
 
 
 
 
 
 
 
 
 
 
Pilloud, M. A., Skipper, C. E., Horsley, S., Craig, A., Latham, K., Clemmons, C. M., . . . Philbin, 

C. S. (2021). Terminology used to describe human variation in forensic anthropology. 
Forensic Anthropology, 4(4), 119.  

Pink, C. M., Maier, C., Pilloud, M. A., & Hefner, J. T. (2016). Cranial Nonmetric and 

Morphoscopic Data Sets. In M. A. Pilloud & J. T. Hefner (Eds.), Biological Distance 
Analysis (pp. 91-107). San Diego: Academic Press. 

Plemons, A. (2022). The Interaction between Genetics and Climate on Craniofacial Variation: 
Examining the Causative Forces of Macromorphoscopic Trait Expression. (PhD), 
Michigan State University,  

Pohl, W. (2016). Genealogy: A Comparative Perspective from the Early Medieval West. In E. 

Hovden, C. Lutter, & W. Pohl (Eds.), Meanings of Community across Medieval Eurasia 
(pp. 232-269): Brill. 

Prescott, S. L., Srinivasan, R., Marchetto, M. C., Grishina, I., Narvaiza, I., Selleri, L., . . . 

Wysocka, J. (2015). Enhancer divergence and cis-regulatory evolution in the human and 
chimp neural crest. Cell, 163(1), 68-83.  

Prowse, T. L. (2016). Isotopes and mobility in the ancient Roman world. In L. de Ligt & L. E. 

Tacoma (Eds.), Migration and mobility in the early Roman empire (pp. 205-233). Leiden: 
Brill. 

Prowse, T. L., Barta, J., von Hunnius, T. E., & Small, A. M. (2010). Stable isotope and 

mitochondrial DNA evidence for geographic origins on a Roman estate at Vagnari (Italy). 
J. Rom. Archaeol, 78, 175-198.  

Prowse, T. L., Schwarcz, H. P., Garnsey, P., Knyf, M., Macchiarelli, R., & Bondioli, L. (2007). 
Isotopic evidence for age-related immigration to Imperial Rome. American Journal of 
Physical Anthropology, 132, 510-519.  

Quinn, C. P., & Beck, J. (2016). Essential tensions: A framework for exploring inequality 

through mortuary archaeology and bioarchaeology. Open Archaeology, 2(1).  

R Core Team. (2024). R: A Language and Environment for Statistical Computing. Vienna, 

Austria: R Foundation for Statistical Computing.  

Rathmann, H., Lismann, S., Francken, M., & Spatzier, A. (2023). Estimating inter-individual 
Mahalanobis distances from mixed incomplete high-dimensional data: Application to 
human skeletal remains from 3rd to 1st millennia BC Southwest Germany. Journal of 
Archaeological Science, 156, 105802.  

Redfern, R., DeWitte, S. N., Hefner, J. T., & Kim, D. (2023). Race, Population Affinity, and 
Mortality Risk during the Second Plague Pandemic in Fourteenth-Century London, 
England. Bioarchaeology International, 7(4), 307–327-307–327.  

211 

 
 
 
 
 
 
 
 
 
 
 
 
Redfern, R., Gröcke, D. R., Millard, A. R., Ridgeway, V., Johnson, L., & Hefner, J. T. (2016). 

Going south of the river: A multidisciplinary analysis of ancestry, mobility and diet in a 
population from Roman Southwark, London. Journal of Archaeological Science, 74, 11-
22.  

Redfern, R., & Hefner, J. T. (2019). “Officially absent but actually present”: Bioarchaeological 

evidence for population diversity in London during the Black Death, AD 1348–50. In M. 
L. Mant & A. J. Holland (Eds.), Bioarchaeology of marginalized people (pp. 69-114): 
Elsevier. 

Reid, D. J., & Dean, M. C. (2000). Brief communication: the timing of linear hypoplasias on 
human anterior teeth. American Journal of Physical Anthropology, 113(1), 135-139.  

Reid, D. J., & Dean, M. C. (2006). Variation in modern human enamel formation times. Journal 

of Human Evolution, 50(3), 329-346.  

Reitsema, L. J., & McIlvaine, B. K. (2014). Reconciling “stress” and “health” in physical 

anthropology: What can bioarchaeologists learn from the other subdisciplines? American 
Journal of Physical Anthropology, 155(2), 181-185.  

Relethford, J. H. (1994). Craniometric variation among modern human populations. American 

Journal of Physical Anthropology, 95, 53-62.  

Relethford, J. H. (2001). Global analysis of regional differences in craniometric diversity and 

population substructure. Human Biology, 73(5), 629-636.  

Relethford, J. H. (2004). Boas and beyond: Migration and craniometric variation. American 

Journal of Human Biology, 16, 379-386.  

Relethford, J. H. (2010). Population‐specific deviations of global human craniometric variation 

from a neutral model. American Journal of Physical Anthropology, 142(1), 105-111.  

Relethford, J. H. (2012). Human population genetics: John Wiley & Sons. 

Relethford, J. H. (2016). Biological Distances and Population Genetics in Bioarchaeology. In M. 

A. Pilloud & J. T. Hefner (Eds.), Biological Distance Analysis: Forensic and 
Bioarchaeological Perspectives: Elsevier. 

Revell, L. (2008). Roman Imperialism and Local Identities. Cambridge: Cambridge University 

Press. 

Riccomi, G. (2021). Bioarchaeology and dietary reconstruction across Late Antiquity and the 
Middle Ages in Tuscany, central Italy. Oxford: Archaeopress Publishing Ltd. 

Ristuccia, N. J. (2018). Christianization and commonwealth in Early Medieval Europe: a ritual 

interpretation: Oxford University Press. 

212 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Ritzman, T. B., Baker, B. J., & Schwartz, G. T. (2008). A fine line: A comparison of methods for 

estimating ages of linear enamel hypoplasia formation. American Journal of Physical 
Anthropology, 135(3), 348-361. doi:https://doi.org/10.1002/ajpa.20750 

Robb, J. (2024). Osteobiography: The Next Generation. In A. Cucina & G. D. Wrobel (Eds.), 

Mesoamerican Osteobiographies: Revealing the Lives and Deaths of Ancient Individuals 
(pp. 382-396): University Press of Florida. 

Robinson, M. R., Kleinman, A., Graff, M., Vinkhuyzen, A. A., Couper, D., Miller, M. B., . . . 
Nolte, I. M. (2017). Genetic evidence of assortative mating in humans. Nature Human 
Behaviour, 1(1), 0016.  

Rohlf, F. J., & Slice, D. (1990). Extensions of the Procrustes method for the optimal 

superimposition of landmarks. Systematic zoology, 39(1), 40-59.  

Roseman, C. C. (2004). Detecting interregionally diversifying natural selection on modern 

human cranial form by using matched molecular and morphometric data. PNAS, 101(35), 
12824-12829.  

Roseman, C. C., & Weaver, T. D. (2004). Multivariate Apportionment of Global Human 

Craniometric Diversity. American Journal of Physical Anthropology, 125, 257-263.  

Ross, A. H., & Pilloud, M. (2021). The need to incorporate human variation and evolutionary 
theory in forensic anthropology: A call for reform. American Journal of Physical 
Anthropology, 176(4), 672-683.  

Ross, A. H., & Williams, S. E. (2021). Ancestry studies in forensic anthropology: back on the 

frontier of racism. Biology, 10(7), 602.  

Ross, F., Pentecost, M., & Moll, T. (2023). Epigenetics and the anthropology of reproduction. A 

Companion to the Anthropology of Reproductive Medicine and Technology, 458-472.  

Rubini, M., Libianchi, N., Gozzi, A., & Zaio, P. (2022). The Population of the Mediterranean 

Basin in the First Millennium AD Migration Period. Austin Anthropol, 6(1), 1027.  

Salesse, K., Dufour, É., Balter, V., Tykot, R. H., Maaranen, N., Rivollat, M., . . . Brůžek, J. 

(2021). Far from home: A multi-analytical approach revealing the journey of an African-
born individual to imperial Rome. Journal of Archaeological Science: Reports, 37, 
103011.  

Sallares, R., Bouwman, A., & Anderung, C. (2004). The spread of malaria to Southern Europe in 

antiquity: new approaches to old problems. Medical history, 48(3), 311-328.  

Sauer, N. J. (1992). Forensic anthropology and the concept of race: If races don't exist, why are 
forensic anthropologists so good at identifying them? Social Science & Medicine, 34(2), 
107-111.  

213 

 
 
 
 
 
 
 
 
 
 
 
 
 
Schats, R. (2023). Developing an archaeology of malaria. A critical review of current approaches 

and a discussion on ways forward. International Journal of Paleopathology, 41, 32-42.  

Scheidel, W. (2004). Human Mobility in Roman Italy, I: The Free Population. The Journal of 

Roman Studies, 94, 1-26.  

Scheidel, W. (2005). Human Mobility in Roman Italy, II: The Slave Population. The Journal of 

Roman Studies, 95, 64-79.  

Schweissing, M. M., & Grupe, G. (2003). Stable strontium isotopes in human teeth and bone: a 
key to migration events of the late Roman period in Bavaria. Journal of Archaeological 
Science, 30, 1373-1383.  

Scorrano, G., Brilli, M., Martínez-Labarga, C., Giustini, F., Pacciani, E., Chilleri, F., . . . 

Rickards, O. (2014). Palaeodiet reconstruction in a woman with probable celiac disease: 
A stable isotope analysis of bone remains from the archaeological site of Cosa (Italy). 
American Journal of Physical Anthropology, 154, 349-356.  

Scott, G. R., Navega, D., Vlemincq‐Mendieta, T., Dern, L. L., O'Rourke, D. H., Hlusko, L. J., & 

Hoffecker, J. F. (2023). Peopling of the Americas: A new approach to assessing dental 
morphological variation in Asian and Native American populations. American Journal of 
Biological Anthropology.  

Sebastiani, A. (2016). New data for a preliminary understanding of the Roman settlement 

network in South Coastal Tuscany: The case of Alberese (Grosseto, IT). Res Antiquae, 
13, 243-272.  

Sebastiani, A. (2017). From villa to village. Late Roman to Early Medieval settlement networks 
in the ager Rusellanus. In J. Moreland, J. Mitchell, & B. Leal (Eds.), Encounters, 
Excavations, and Argosies: Essays for Richard Hodges: Archaeopress. 

Sebastiani, A., Chirico, E., Colombini, M., & Cygielman, M. (Eds.). (2015). Diana Umbronensis 
a Scoglietto Santuario, Territorio e Cultura Materiale (200 a.C.-550 d.C.). Oxford: 
Archaeopress. 

Sebro, R., Peloso, G. M., Dupuis, J., & Risch, N. J. (2017). Structured mating: patterns and 

implications. PLoS Genetics, 13(4), e1006655.  

Selleri, L., & Rijli, F. M. (2023). Shaping faces: genetic and epigenetic control of craniofacial 

morphogenesis. Nature Reviews Genetics, 24(9), 610-626.  

Šešelj, M., Duren, D. L., & Sherwood, R. J. (2015). Heritability of the human craniofacial 

complex. The Anatomical Record, 298(9), 1535-1547.  

214 

 
 
 
 
 
 
 
 
 
 
 
 
 
Shaw, H., Montgomery, J., Redfern, R., Gowland, R., & Evans, J. (2016). Identifying migrants in 
Roman London using lead and strontium stable isotopes. Journal of Archaeological 
Science, 66, 57-68.  

Slaus, M. (2008). Osteological and Dental Markers of Health in the Transition From the Late 
Antique to the Early Medieval Period in Croatia. American Journal of Physical 
Anthropology, 136(4), 455-469.  

Slice, D. E., & Ross, A. (2009). Geometric morphometric classification of crania for forensic 

scientists.  

Smith‐Guzmán, N. E. (2015). The skeletal manifestation of malaria: An epidemiological 

approach using documented skeletal collections. American Journal of Physical 
Anthropology, 158(4), 624-635.  

Smith, H. F., Hulsey, B. I., (Pack) West, F. L., & Cabana, G. S. (2016). Do biological distances 

reflect genetic distances? A comparison of craniometric and genetic distances at local and 
global scales. In J. T. Hefner & M. A. Pilloud (Eds.), Biological Distance Analysis: 
Forensic and Archaeological Perspectives: Elsevier. 

Sofaer, J. R. (2006). The Body as Material Culture: A Theoretical Osteoarchaeology. 

Cambridge: Cambridge University Press. 

Sparks, C. C., & Jantz, R. L. (2002). A reassessment of human cranial plasticity: Boas revisited. 
Proceedings of the National Academy of Sciences of the United States of America, 
99(23), 14636-14639.  

Sparks, C. C., & Jantz, R. L. (2003). Changing Times, Changing Faces: Franz Boas's Immigrant 

Study in Modern Perspective. American Anthropologist, 105(2), 333-337.  

Sperduti, A., Bonditi, L., Craig, O. E., Prowse, T. L., & Garnsey, P. (2018). Bones, teeth, and 
history. In W. Scheidel (Ed.), The science of Roman history: biology, climate, and the 
future of the past. Princeton: Princeton University Press. 

Spradley, M. K., & Jantz, R. (2021). What are we really estimating in forensic anthropological 

practice, population affinity or ancestry? Forensic Anthropology, 4(4), 309–318-309–318.  

Spradley, M. K., & Jantz, R. L. (2016). Ancestry Estimation in Forensic Anthropology: 

Geometric Morphometric versus Standard and Nonstandard Interlandmark Distances. 
Journal of Forensic Sciences, 61(4), 892-897.  

Spradley, M. K., & Weisensee, K. (2017). Ancestry estimation: The importance, the history, and 
the practice. In N. R. Langley & M. A. Tersigni-Tarrant (Eds.), Forensic anthropology: A 
comprehensive introduction (2nd Edition ed., pp. 181-192): CRC Press. 

215 

 
 
 
 
 
 
 
 
 
 
 
 
 
Stark, R. J. (2016). Ancient lives in motion: a bioarchaeological examination of stable isotopes, 
nonmetric traits, and human mobility in an imperial roman context (1st-3rd c. CE). 
(PhD), McMaster University, Hamilton, Ontario, Canada.  

Stark, R. J., Emery, M. V., Schwarcz, H., Sperduti, A., Bondioli, L., Craig, O. E., & Prowse, T. 
(2020). Imperial Roman mobility and migration at Velia (1st to 2nd c. CE) in southern 
Italy. Journal of Archaeological Science: Reports, 30, 102217.  

Stark, R. J., Emery, M. V., Schwarcz, H., Sperduti, A., Bondioli, L., Craig, O. E., & Prowse, T. L. 
(2021). Dataset of oxygen, carbon, and strontium isotope values from the Imperial 
Roman site of Velia (ca. 1st-2nd c. CE), Italy. Data in Brief, 38, 107421.  

Stewart, T. D. (1979). Essentials of Forensic Anthropology. Springfield: Charles C. Thomas. 

Stojanowski, C. M., & Schillaci, M. A. (2006). Phenotypic approaches for understanding patterns 

of intracemetery biological variation. Yearbook of Physical Anthropology, 49, 49-88.  

Strauss, A., & Hubbe, M. (2010). Craniometric similarities within and between human 

populations in comparison with neutral genetic data. Human Biology, 82(3), 315-330.  

Streetman, E. (2018). Inferring the social organization of Medieval Upper Nubia using 

nonmetric traits of the skull. (Doctor of Philosophy), Michigan State University,  

Stuart-Macadam, P. (1985). Porotic hyperostosis: Representative of a childhood condition. 

American Journal of Physical Anthropology, 66(4), 391-398. 
doi:https://doi.org/10.1002/ajpa.1330660407 

Stuart-Macadam, P. (1987). Porotic hyperostosis: New evidence to support the anemia theory. 

American Journal of Physical Anthropology, 74(4), 521-526. 
doi:https://doi.org/10.1002/ajpa.1330740410 

Stuart-Macadam, P. (1989). Porotic hyperostosis: Relationship between orbital and vault lesions. 

American Journal of Physical Anthropology, 80(2), 187-193. 
doi:https://doi.org/10.1002/ajpa.1330800206 

Stuart-Macadam, P. (1992). Porotic hyperostosis: A new perspective. American Journal of 
Physical Anthropology, 87(1), 39-47. doi:https://doi.org/10.1002/ajpa.1330870105 

Stull, K. E., Bartelink, E. J., Klales, A. R., Berg, G. E., Kenyhercz, M. W., L'Abbé, E. N., . . . 
Mariscal, C. (2021). Commentary on: Bethard JD, DiGangi EA. Letter to the Editor—
Moving beyond a lost cause: Forensic anthropology and ancestry estimates in the United 
States. J Forensic Sci. 2020; 65 (5): 1791–2. Journal of Forensic Sciences, 66(1).  

Suchey, J. M., & Katz, D. (1998). Applications of Pubic Age Determination in a Forensic 

Setting. In K. J. Reichs (Ed.), Forensic Osteology: Advances in the Identification of 
Human Remains (2nd ed., pp. 204-236). Springfield, IL: Charles C. Thomas. 

216 

 
 
 
 
 
 
 
 
 
 
 
 
 
SWGANTH. (2013). Ancestry assessment.  

Tallman, S. D., Parr, N. M., & Winburn, A. P. (2021). Assumed differences; unquestioned 
typologies: The oversimplification of race and ancestry in forensic anthropology. 
Forensic Anthropology, 4(4), 73-96.  

Temple, D. H., & Goodman, A. H. (2014). Bioarcheology has a “health” problem: 

Conceptualizing “stress” and “health” in bioarcheological research. American Journal of 
Physical Anthropology, 155(2), 186-191. doi:https://doi.org/10.1002/ajpa.22602 

Thayer, Z. M., & Non, A. L. (2015). Anthropology meets epigenetics: Current and future 

directions. American Anthropologist, 117(4), 722-735.  

Thomas, R. M., Parks, C. L., & Richard, A. H. (2017). Accuracy Rates of Ancestry Estimation by 

Forensic Anthropologists Using Identified Forensic Cases. Journal of Forensic Sciences, 
62(4), 971-974.  

Tian, Y., Koncz, I., Defant, S., Giostra, C., Vyas, D. N., Sołtysiak, A., . . . Veeramah, K. R. 

(2024). The role of emerging elites in the formation and development of communities 
after the fall of the Roman Empire. Proceedings of the National Academy of Sciences, 
121(36). doi:10.1073/pnas.2317868121 

Torres-Rouff, C., & Knudson, K. J. (2017). Integrating Identities: An Innovative 

Bioarchaeological and Biogeochemical Approach to Analyzing the Multiplicity of 
Identities in the Mortuary Record. Current Anthropology, 58(3), 381-409.  

Turner, C. G. I., Nichol, C. R., & Scott, G. R. (1991). Scoring procedures for key morphological 
traits of the permanent dentition: the Arizona State University Dental Anthropology 
System. In M. A. Kelley & C. S. Larsen (Eds.), Advances in Dental Anthropology. New 
York: Wiley-Liss. 

Urbanová, P., & Ross, A. H. (2016). Advanced Methods in 3-D Craniofacial Morphological 

Analysis. In M. A. Pilloud & J. T. Hefner (Eds.), Biological Distance Analysis (pp. 61-
90). San Diego: Academic Press. 

Vaccaro, E. (2008). An overview of rural settlement in four river basins in the province of 

Grosseto on the coast of Tuscany (200 B.C.–A.D. 600). Journal of Roman Archaeology, 
21, 225-247.  

Vaccaro, E. (2011). Sites and Pots: Settlement and Economy in Southern Tuscany (AD 300-900). 

Oxford: Archaeopress. 

Van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained 

equations in R. Journal of statistical software, 45, 1-67.  

217 

 
 
 
 
 
 
 
 
 
 
 
 
 
Velte, M., Czermak, A., Grigat, A., Neidich, D., Trautmann, B., Lösch, S., . . . Harbeck, M. 
(2023). Tracing early life histories from Roman times to the Medieval era: weaning 
practices and physiological stress. Archaeological and Anthropological Sciences, 15(12), 
190.  

Venables, W., & Ripley, B. (2002). Modern Applied Statistics with S (Fourth ed.). New York: 

Springer. 

Vercellotti, G., Piperata, B. A., Agnew, A. M., Wilson, W. M., Dufour, D. L., Reina, J. C., . . . 
Stout, S. D. (2014). Exploring the multidimensionality of stature variation in the past 
through comparisons of archaeological and living populations. American Journal of 
Physical Anthropology, 155(2), 229-242.  

Viva, S. (2020). Burials from the cemetery at Vetricella (Scarlino, Grosseto): anthropological, 
paleodemographic and paleopathological analyses. In G. Bianchi & R. Hodges (Eds.), 
The nEU-Med project: Vetricella, an Early Medieval royal property on Tuscany's 
Mediterranean (pp. 105-120). Sesto Fiorentino: All'Insegna del Giglio. 

Viva, S., Fabbri, P. F., Ricci, P., Bianchi, G., Hodges, R., & Lubritto, C. (2021). Project nEU-
Med. The contribution of isotopic analysis in the differential diagnosis of anemia, the 
case of the Medieval cemetery of Vetricella (Scarlino, GR) in Tuscany. Environmental 
Archaeology.  

Viva, S., Lubritto, C., Cantini, F., & Fabbri, P. F. (2022). Evidence of Barbarian migrations and 

interpersonal violence during the Gothic War in sixth-century Tuscany: the case of the 
Goth horseman from San Genesio (Pisa). Archaeological and Anthropological Sciences, 
14(39).  

Vollner, J. M. (2016). Examining the population history of three Medieval Nubian sites through 

craniometric analyses. (Doctor of Philosophy), Michigan State University,  

von Cramon-Taubadel, N. (2009). Revisiting the homoiology hypothesis: the impact of 

phenotypic plasticity on the reconstruction of human population history from 
craniometric data. Journal of Human Evolution, 57(2), 179-190.  

von Cramon-Taubadel, N. (2014). Evolutionary insights into global patterns of human cranial 

diversity: population history, climatic and dietary effects. J. Anthropol. Sci, 92(4), 43-77.  

von Cramon-Taubadel, N. (2016). Population Biodistance in Global Perspective: Assessing the 

Influence of Population History and Environmental Effects on Patterns 
of Craniomandibular Variation. In M. A. Pilloud & J. T. Hefner (Eds.), Biological 
Distance Analysis (pp. 425-445). San Diego: Academic Press. 

Walker, P. L., Bathurst, R. R., Richman, R., Gjerdrum, T., & Andrushko, V. A. (2009). The 

causes of porotic hyperostosis and cribra orbitalia: A reappraisal of the iron-deficiency-

218 

 
 
 
 
 
 
 
 
 
 
 
anemia hypothesis. American Journal of Physical Anthropology, 139(2), 109-125. 
doi:https://doi.org/10.1002/ajpa.21031 

Ward-Perkins, B. (2005). The Fall of Rome: And the End of Civilization. Oxford: Oxford 

University Press. 

Weiss‐Krejci, E. (2011). The formation of mortuary deposits: implications for understanding 

mortuary behavior of past populations. Social bioarchaeology, 68-106.  

Weston, D. A. (2008). Investigating the specificity of periosteal reactions in pathology museum 
specimens. American Journal of Physical Anthropology: The Official Publication of the 
American Association of Physical Anthropologists, 137(1), 48-59.  

Weston, D. A. (2011). Nonspecific infection in paleopathology: interpreting periosteal reactions. 

In A. L. Grauer (Ed.), A companion to paleopathology (pp. 492-512). 

Wickham, C. (1981). Early Medieval Italy: Central Power and Local Society, 400-1000. 

London: Macmillan. 

Wickham, C. (1994). Land and power: studies in Italian and European social history, 400-1200: 

London: British School at Rome. 

Wickham, C. (2006). Framing the Early Middle Ages: Europe and the Mediterranean, 400-800. 

Oxford: Oxford University Press. 

Wickham, C. (2016). Medieval Europe. New Haven: Yale University Press. 

Wickham, C. (2023). The Donkey and the Boat: Reinterpreting the Mediterranean Economy, 

950-1180. Oxford: Oxford University Press. 

Willett, F. (1971). A Survey of Recent Results in the Radiocarbon Chronology of Westers and 

Northern Africa. The Journal of African History, 12(3), 339-370.  

Williams, F. L., Belcher, R. L., & Armelagos, G. J. (2005). Forensic misclassification of ancient 

Nubian crania: Implications for assumptions about human variation. Current 
Anthropology, 46(2), 340-346.  

Winburn, A. P., & Algee-Hewitt, B. F. B. (2021). Evaluating population affinity estimates in 
forensic anthropology: Insights from the forensic anthropology database for assessing 
methods accuracy (FADAMA). Journal of Forensic Sciences, 66, 1210-1219.  

Winburn, A. P., Yim, A. D., & Stock, M. K. (2022). Recentering forensic anthropology within a 
multifaceted body of evolutionary theory: Strengthening method by making theory 
explicit. American Journal of Biological Anthropology, 179(4), 535-551.  

219 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Wood, J. W., Milner, G. R., Harpending, H. C., & Weiss, K. M. (1992). The Osteological 

Paradox. Current Anthropology, 33(4), 343-370.  

Wright, L. E., & Yoder, C. J. (2003). Recent Progress in Bioarchaeology: Approaches to the 

Osteological Paradox. Journal of Archaeological Research, 11(1), 43-70.  

Zakrzewski, S. (2011). Population Migration, Variation, and Identity. In Social bioarchaeology 

(pp. 183-211). 

Zhao, S., Zhang, B., Yang, J., Zhou, J., & Xu, Y. (2024). Linear discriminant analysis. Nature 

Reviews Methods Primers, 4(1), 70. doi:10.1038/s43586-024-00346-y 

Zhou, L., & Corruccini, R. S. (1998). Enamel hypoplasias related to famine stress in living 
Chinese. American Journal of Human Biology: The Official Journal of the Human 
Biology Association, 10(6), 723-733.  

Zuckerman, M. K. (2020). Identity and Health: Exploring Relationships among Health, Disease, 

and Identity in Past Populations. In K. J. Knudson & C. Stojanowski (Eds.), 
Bioarchaeology and Identity Revisited. Gainesville: University of Florida Press. 

Zuckerman, M. K., & Armelagos, G. J. (2011). The origins of biocultural dimensions in 

bioarchaeology. Social bioarchaeology, 13-43.  

Zuckerman, M. K., & Martin, D. L. (2016). New directions in biocultural anthropology: Wiley 

Online Library. 

220