Received: 20 August 2021 Revised: 9 November 2021 Accepted: 22 November 2021

DOI: 10.1002/ajpa.24462

YEARBOOK OF BIOLOGICAL ANTHROPOLOGY ARTICLE

Paleodemography: From archaeology and skeletal age

estimation to life in the past

Jesper L. Boldsen1 | George R. Milner2 | Stephen D. Ousley3

1
ADBOU, University of Southern Denmark,
Campusvej 55, Odense M, Denmark Abstract
2
Department of Anthropology, The Much of paleodemography, an interdisciplinary field with strong ties to archaeology,
Pennsylvania State University, University Park,
among other disciplines, is oriented toward clarifying the life experiences of past peo-
Pennsylvania, USA
3
Department of Anthropology, University of ple and why they changed over time. We focus on how human skeletons contribute
Tennessee, Knoxville, Tennessee, USA to our understanding of preindustrial demographic regimes, including when changes

Correspondence took place that led to the world as we know it today. Problems with existing
George R. Milner, Department of paleodemographic practices are highlighted, as are promising directions for future
Anthropology, The Pennsylvania State
University, University Park, PA 16802, USA. work. The latter requires both better age estimates and innovative methods to handle
Email: [email protected] data appropriately. Age-at-death estimates for adult skeletons are a particular prob-
Funding information lem, especially for adults over 50 years that undoubtedly are mistakenly underrepre-
National Institute of Justice, Grant/Award sented in published studies of archaeological skeletons. Better age estimates for the
Number: 2014-DN-BX-K007
entirety of the lifespan are essential to generate realistic distributions of age at death.
There are currently encouraging signs that after about a half-century of intensive,
and sometimes contentious, research, paleodemography is poised to contribute much
to understandings of evolutionary processes, the structure of past populations, and
human-disease interaction, among other topics.

KEYWORDS
age and sex estimation, age-at-death distributions, bioarchaeology, paleodemography,
preindustrial populations

1 | I N T RO DU CT I O N the conditions under which they did so. While there is more to the
field than skeletons, in this article we focus on human remains.
Paleodemography is the window through which many aspects of the Even when restricting ourselves to bones, it is only possible to
past human experience can be seen, including life expectancy, fertility, highlight current problems, controversies, and directions for future
risk of dying, and survival in diverse cultural and natural settings. Our work. Doing so involves first diving into how data are obtained and
predecessors' life experiences have long been ripe for speculation. their limitations. Beyond knowing how a skeletal sample came
According to what many consider divinely inspired text, people lived about, paleodemographic work requires morphological indicators
as long as “threescore years and ten” (Psalms 90:10). Or perhaps the that accurately reflect sex and age. Unfortunately, conventional
situation was entirely different, and lives were “poore, nasty, brutish, adult age-estimation methods perform poorly, although there is
and short” (Hobbes, 1904 [1651], p. 84). Only one thing is certain. reason to be optimistic about ongoing research that promises to
There is no shortage of seemingly definitive, but ill-informed, pro- yield unbiased estimates of age throughout the lifespan. Moving
nouncements about such topics. This interest, long predating today's beyond individual skeletons, generating age-at-death distributions
paleodemography, stems in part from an inchoate recognition that and figuring out what they might mean are not straightforward
skeletons can tell us much about life in the past. tasks. Producing a distribution of deaths is heavily weighted by
Paleodemographers are principally concerned with clarifying assumptions and decisions that affect the results, including data
how people once lived, why ways of life changed over time, and comparability.

Yearbook Biol Anthropol. 2022;178(Suppl. 74):115–150. wileyonlinelibrary.com/journal/ajpa © 2021 Wiley Periodicals LLC. 115
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
116 BOLDSEN ET AL.

A balance must be struck between a need for better data and Blakely, 1971, 1977; Buikstra, 1976; Green et al., 1974; Lovejoy
innovative methods to handle them, and how that information can be et al., 1977; Owsley & Bass, 1979; Swedlund & Armelagos, 1969;
used to advance understandings of evolutionary processes, commu- Ubelaker, 1974).
nity organization and function, and human-disease interaction, among Paleodemography's origin as a coherent field, in all its various
other topics that require accurate appraisals of population structure. dimensions, owes much to a reorientation of research that accompa-
Being able to address such issues is why paleodemography is impor- nied the 1960s New Archaeology. An emphasis was placed on the
tant, although the road getting to something useful has been long and people of the past rather than mere descriptions of artifacts and archi-
tortuous. In fact, the foundations of much of what we would like to tectural remains. Researchers accordingly focused on people's lives,
know about life in past communities—estimating age, generating age with the structure of populations as recoverable through skeletons
distributions, and interpreting them—have been the subjects of con- being an important part of that work (Angel, 1971; Buikstra, 1976;
siderable work and intense debate. Green et al., 1974; Ubelaker, 1974).
With regard to the experiences of past peoples, we first turn to Initial enthusiasm about paleodemography's potential was soon
inferences that can be drawn from skeletal age distributions and single dampened by problems posed by skeletal data, their analysis, and
figure summaries of them. We then cover what can be said about the interpretation (Bocquet-Appel & Masset, 1982, 1985, 1996; Hoppa,
course of life from childhood to old age, the latter much-neglected in 2002; Hoppa & Vaupel, 2002a; Konigsberg & Herrmann, 2006;
research focused on archaeological skeletons. Skeletons also provide Masset, 1976; Petersen, 1975). Some of the issues have been
the primary means of identifying the principal characteristics of addressed, whereas others have not. Nevertheless, an early
preindustrial demographic regimes, as well as when changes took announcement of the field's death was surely premature, but in
place that led to the world as we know it today. For example, the rep- Bocquet-Appel and Masset's (1982) Farewell to Paleodemography the
resentation of juveniles in archaeological samples, interpreted as a fer- research effort suffered a direct, telling, and much-deserved hit. Four
tility indicator, provides clues concerning what took place and the decades on, their challenge coupled with issues raised by other
pace of change from hunter-gatherer to agricultural ways of life. researchers continues to inspire the conceptual and methodological
Finally, the role of paleodemographic findings in allied fields is work necessary to estimate age from skeletal remains, generate age-
highlighted through ongoing developments in paleoepidemiology, a at-death distributions, and glean information from them about life in
quantitative assessment of the disease experience of past people. the past (Frankenberg & Konigsberg, 2006; Hoppa & Vaupel, 2002a;
Milner et al., 2019, 2021; Séguy & Buchet, 2013).

2 | P A L E O D E M O G R A P HY
2.2 | Contributions
2.1 | Coverage and development
Demography is a driving force of evolution and societal development.
Human skeletons are only part of what is covered by a broad Of the many potential contributions of paleodemography to our
paleodemographic umbrella (Chamberlain, 2006; French, 2016; understanding of the human experience, several that resonate in
French et al., 2021; Hassan, 1981; Schmidt et al., 2021; Williams, today's world are of special interest. They include how different cate-
2012). Studies of population size, distribution, and growth also involve gories of people were perceived and treated in the past, our predeces-
sites, artifacts, and radiocarbon dates. These topics, in fact, are more sors' life courses, and the impact of diseases, trauma, and migration
directly addressed through archaeological material than skeletal on the processes behind major societal transformations, including the
remains. Nevertheless, skeletons have been used to estimate settle- development of more intensive subsistence systems and organization-
ment size and duration, and this work dates back to a time when there ally complex societies.
were few such estimates about excavated communities (Hooton,
1920, 1930). Demographic information for past societies with written
records has also been derived from funerary inscriptions, among other 2.2.1 | Life in past communities
source material (Floris et al., 2021; Hin, 2013; Scheidel, 2012;
Woods, 2007). These data have their own biases and interpretive Without a grasp of a group's age and sex composition, one important
ambiguity, so much so that a critique of funerary epitaph information objective of contemporary bioarchaeological work (Larsen, 2015), the
has been aptly titled Graveyards for Historians (Hopkins, 1987). characterization of what specific groups of people, up to entire com-
Prior to the mid-20th century, characterizations of the age and munities, experienced, would be severely compromised. In the
sex composition of past populations, regardless of whether data came absence of such information, one cannot hope to appreciate, let alone
from skeletons or funerary inscriptions, were few and far between measure, the challenges people faced in terms of dependents to pro-
(Angel, 1947; MacDonnell, 1913; Pearson, 1902; Todd, 1927). But ducers; the allocation of household and community labor to essential
since about 1970, paleodemography has played a central, if contro- survival tasks; and the capacity to siphon off time and effort to con-
versial, role in studies of archaeological skeletons (Acsádi & struct monuments, defensive structures, and other public works.
Nemeskéri, 1970; Angel, 1969; Asch, 1976; Bennett, 1973; Archaeological findings are testimony to what people in the past
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 117

achieved, but not the toll it took on past communities or specific seg- an individual's characteristics rather than a strict accounting of age,
ments of them. have been identified in Neolithic Europe (Bickle & Fibiger, 2014). In a
One of the best means to delineate the organizational structure late prehistoric village in the American midcontinent, people in their
of past societies is through a combination of skeletal information and late teenage years apparently assumed the roles of adults (Milner
archaeological data, including grave form and location, body treat- et al., 1991). They were ambushed by their enemies much like older
ment, and associated items. Of particular interest are age and sex- people, presumably while engaged in subsistence-related tasks that
related roles and status, as well as the nature of, and interactions exposed them to settings where they risked being killed. Behavior
among, the social groups that constructed and maintained specific associated with the transition to adulthood—something that cannot
mortuary contexts. Extracting the most information possible from be observed directly—can be inferred from an enhanced risk of dying
mortuary sites requires, as a first step, knowledge about who was bur- at the hands of their enemies. A modern example would be legal age,
ied in them, notably their age and sex. These biological characteristics specifically the attainment of adulthood, in medicolegal investigations
of the deceased, when combined with variation in body treatments, (Arge et al., 2020; Konigsberg et al., 2019; Thevissen et al., 2010). But,
grave forms, and burial artifacts, provide perspectives on how general despite its importance in enriching community-level studies, social age
categories of people were regarded in their communities. is not considered further because age-related roles and how people
In recent years, bioarchaeologists have increasingly focused were regarded by others require context-specific archaeological infor-
attention on group and individual identities, as reconstructed from mation beyond what bones alone can provide.
skeletal remains and contextual information (Hosek & Robb, 2019;
Knudson & Stojanowski, 2008; Torres-Rouff & Knudson, 2017;
Zakrzewski, 2015). The emphasis is on understanding how people, 2.2.2 | Life histories
often specific individuals, experienced their lives. Although rarely
acknowledged, or perhaps even recognized, this work owes much to The risks (probabilities) of dying experienced by once-living people, as
Goodenough's (1965, p. 7) fieldwork on the Micronesian island of estimated from contextually well-defined archaeological samples, is an
Chuuk (then Truk). He detailed how a person's multiple identities, only essential part of describing the lives of people in the past and how
some of which come into play in specific circumstances such as funer- they changed over time. For example, such information can provide a
ary proceedings, form an individual's situationally determined “social deep history to our understanding of the human lifespan (Wilmoth,
persona.” Age, chronological or otherwise, and gender are among the 2000). That includes variation in age-specific mortality, differences
identities that shape social relationships. in male and female life courses, and longevity relative to that of
Turning first to sex and gender, one can estimate the former from nonhuman primates (Colchero et al., 2016). Without this
adult skeletons, but not the latter (Garofalo & Garvin, 2020; Walker & paleodemographic information, it would be difficult, if not impossible,
Cook, 1998). An attribution of sex means the observed morphological to characterize transformations in the human experience from our
characteristics conform to a greater or lesser degree to those hunting-and-gathering predecessors to the societies of today.
expected of males or females. It is, therefore, an inherently probabilis- Age-at-death is affected by the conditions people face, with one
tic statement, although not often expressed as such. The biological readily appreciated outcome being longer life expectancies over the
and social components of gender are receiving increasing attention in past century or two that accompanied improvements in public health
anthropology (DuBois & Shattuck-Heidorn, 2021). For archaeological measures and internal medicine. Twin studies show that about a quar-
samples, doing so involves contextual information beyond bones. That ter of the variance in the lifespan is explicable in terms of genetics
is, it requires simultaneously examining both bones and other informa- (Christensen et al., 2006; Herskind et al., 1996). Notably lower esti-
tion, such as grave form, burial accompaniments, and what remains of mates have been generated from a larger and more diverse sample
garments. That work is necessarily undertaken on a case-by-case that was not based on twins (Ruby et al., 2018). Because of the strong
basis, so however much it tells us about life in the past these environmental influence, it comes as no surprise that age-at-death
individual-oriented analyses are not susceptible to quantitative ana- patterns can further understandings of the mortality consequences of
lyses of numerous individuals, the basis of paleodemographic our ancestors' diverse ways of life.
investigations. Unfortunately, and for reasons explained later, we have only a
The social significance of age is more amenable to paleodemographic poor grasp on the length of life in the past when conditions were
studies. Here the familiar distinction between biological and chrono- quite unlike those of today. Much depends on having good skeletal
logical age is augmented by evidence for the roles that people age estimates from many archaeological samples. If available, the rela-
adopted, as well as how they were perceived and treated by others tionship between life expectancy and lifespan equality, the latter cap-
(Gowland, 2006; Sofaer, 2011). Coming to grips with social age turing variation in age at death, as it is understood for relatively
involves, once again, the use of both skeletal and contextual data. recent societies (Aburto et al., 2020) could be extended farther into
From a paleodemographic perspective, investigations of social age the past. It might even be possible to determine if in the distant past
have the advantage of being amenable to quantitative analyses of there was a deceleration in the risk of dying among the elderly, which
large skeletal samples. For example, different stages in the subadult today is most notable for supercentenarians (Gampe, 2010;
years, specifically a “middle childhood” where limits were defined by Thatcher, 1999; Vaupel, 2010). Perhaps this aspect of mortality
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
118 BOLDSEN ET AL.

occurred at an earlier age in preindustrial populations when life expe- Archaeology on autochthonous change. Recently, however, increasing
riences were much different from those of today. interest has been directed toward the part played by population
mobility in societal transformations (Gregoricka, 2021; McSparron
et al., 2020; Smith, 2014).
2.2.3 | Disease experience There is indeed reason to direct more attention toward mobility
in the past, regardless of a society's structural form. To date, archaeo-
A community's experience with trauma and disease represents a criti- logical work on population mobility has emphasized the identification
cal part of characterizations of life in the past (Cohen & of migrants through the stable isotope, and to a lesser extent the trace
Armelagos, 1984; Larsen, 2015). That is particularly true when multi- element, composition of bones and teeth (Bentley, 2006; Katzenberg
ple lines of evidence, skeletal remains and much else, are combined Waters-Rist, 2019; Knudson & Price, 2007; Price et al., 2000;
when interpreting what happened and why it did so (Larsen Rasmussen et al., 2020). The reconstruction of individual life histories
et al., 2019). Although traditionally the domain of paleopathology, a through analyses of isotopes and trace elements, when combined with
largely descriptive endeavor, age and sex information is an essential good skeletal age estimates, facilitates the estimation of migration fre-
component of estimating the impact deleterious conditions had on quencies on a societal level.
past communities, not just their presence in certain times and places. Modeling the effects of migration on population parameters is a
Going beyond simple counts of fractured or diseased bones in daunting task. Of concern is not only the effect on recipient settle-
mortality samples is the objective of quantitatively rigorous ments, such as developing urban centers, but also its impact on donor
paleoepidemiology that combines indicators of risk, notably skeletal communities. The difficulty here can be readily appreciated: one could
or dental lesions but also including sex and stature, with age at death have the skeletons of migrants included in samples from the commu-
information (Boldsen, 1997, 2001, 2005b; Boldsen, Milner, & Weise, nities where they died, but not from where they originated. Identify-
2015; DeWitte, 2018; DeWitte & Hughes-Morey, 2012; DeWitte & ing and quantifying the effects of the arrival or departure of
Wood, 2008; Gamble et al., 2017; Godde et al., 2020; Godde & appreciable numbers of people, in most instances likely dominated by
Hens, 2021; Kelmelis et al., 2017; Kelmelis & Dangvard Pedersen, young adults, is a potentially important, but as yet unrealized, dimen-
2019; Milner & Boldsen, 2017; Usher, 2000). sion of paleodemographic research.
Early death from illness and injuries undoubtedly produced age-
at-death distributions unlike those of developed nations today where
adult mortality is dominated by chronic, late-onset diseases, such as 2.2.5 | Evolution and the life course
coronary disease and many cancers. At present, standard adult age-
estimation methods are not equal to the task of documenting the Evolution acts through the genetic consequences of demographic pro-
magnitude of such differences, although the means of getting suffi- cesses, including differential fertility and survival, as well as population
ciently precise information are on the horizon (Hoppa & structure, growth (or decline), movement, and interaction (Carey &
Vaupel, 2002a; Milner et al., 2021; Séguy & Buchet, 2013). Nor are Vaupel, 2005; Ward & Weiss, 1976). Knowledge about the structure
adult age estimates produced by standard methods a satisfactory of past populations furthers our understanding of human evolution,
basis for investigating the deep history of age-related mortality of including past and present geographical variation in our species.
concern to medical science, such as the lingering effects of early ill- Indeed, one early objective of paleodemography was the charac-
ness on later well-being and survival (Barker, 2004; Gluckman & terization of mortality patterns for much of human existence
Hanson, 2004; Hoffman et al., 2017). A modern world example is the (Acsádi & Nemeskéri, 1970; Hassan, 1981). This work extends back to
long-term health consequences of the Dutch Hongerwinter our distant hominin ancestors, especially with regard to expected
(Roseboom et al., 2001), quite aside from the excess mortality that lifespans and the durations of life-history stages (Bermúdez de Castro
occurred at that time (Ekamper et al., 2017). Much earlier populations et al., 2004; Bocquet-Appel & Arsuaga, 1999; Caspari & Lee, 2004;
were surely subject to periodic privations with similarly severe imme- Lovejoy, 1981; Migliano & Guillon, 2012; Monge & Mann, 2015; Riga
diate and downstream impacts on morbidity and mortality that are et al., 2019; Trinkaus, 1995, 2011; Trinkaus & Thompson, 1987). The
potentially measurable (Gowland, 2015). overwhelming amount of paleodemographic research, however,
focuses on anatomically modern humans from the mid-Holocene
onward. Populations following the development of agricultural econo-
2.2.4 | Migration mies are particularly well represented because cemeteries associated
with villages and, later, urban centers often yield large skeletal
Population movement, despite its prominence in world news today, samples.
has had a checkered history in archaeology (Adams et al., 1978; While the remains of our hominin ancestors are poorly represen-
Anthony, 1990; Burmeister, 2000; Snow, 2009). Once considered a ted in geological deposits, studies of Holocene skeletons can contrib-
primary driver of change—separate groups of people carried equally ute to knowledge about the forces that shaped human evolution.
distinctive cultural baggage—migration's significance in regional cul- They include the capacity of old individuals to transmit long-
tural trajectories was largely dismissed with the focus of New remembered knowledge that benefited their social groups
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 119

(Rosenberg, 2004). Much discussed is the Grandmother Hypothesis 3.1 | Sampling bias
in which women past reproductive age made contributions that
enhanced the well-being and survival of younger family members Turning first to skeletons as samples, the word population is com-
(Hawkes, 2006; Hawkes et al., 1998; Hawkes & Blurton Jones, 2005). monly used in studies of archaeological skeletons. But what comes to
The question naturally arises if there were enough women of post- a bioarchaeologist's attention are skeletons that have undergone a
reproductive age to make a difference (Crews & Gerber, 2003; lengthy selection process that reduces the individuals who died to the
Kennedy, 2002). If few people during the past several millennia sur- skeletons that happen to be examined (Dutour et al., 2003;
vived to old age, then surely such longevity, and any benefits it Hoppa, 1999; Knüsel & Robb, 2016; Konigsberg & Frankenberg,
afforded, could not have played a major role in the evolution of our 1994; Milner et al., 2019; Pinhasi & Bourbou, 2008; Saunders
species. Holocene skeletal samples are all but truncated at about et al., 1995; Séguy & Buchet, 2013). For the following reasons, it is
50 years of age. If published results are accepted at face value, one best to refer to groups of archaeological skeletons as samples, not
must conclude that virtually all adults in the more distant past also populations.
died early deaths. Alternatively, the remarkably consistent results are Starting with deaths, often group members are not interred in the
an artifact of widely used, but deeply flawed, age estimation methods. same way or place, so the likelihood of their skeletons being discovered
and excavated differs accordingly. Some of these remains are more
likely to have lasted to the present than others. In heavily used grave-
2.2.6 | Demographic and socioeconomic transitions yards, it is not uncommon for skeletons to have been disturbed when
later graves were dug, with bones being displaced or discarded. Much
Holocene population growth lies at the heart of what we want to the same can be said of skeletons in contexts that attracted looters
know about how we got to where we are today. Attention has accord- searching for artifacts. Burial conditions, including soils that are even
ingly focused on when and why populations increased, notably during slightly acidic, affect the preservation of bones (Bello et al., 2006; Djuri
c
the shift to an agricultural (Neolithic) way of life and later to organiza- et al., 2011; Gordon & Buikstra, 1981; Guy et al., 1997; Manifold,
tionally complex societies. As indicated above, tracking an increase in 2015; Stanton & Hefner, 2016; Walker et al., 1988; Willey et al., 1997).
regional population size and density is most effectively done with The small bones of children are more likely to be lost than the bigger
archaeological evidence, particularly through compilations of sites and bones of adults, with infant underrepresentation being the norm in
radiocarbon dates. There is, however, still a place for skeletal studies archaeological samples. This potential for bias was recognized as long
in this aspect of paleodemographic research. ago as the mid-18th century when Jefferson (1788, p. 105) commented
Where food-production systems developed independently, this on the preservation of infant bones in an excavated mound in Virginia.
transition toward an increasingly settled existence, plant cultivation, There is also the excavation process itself. Some archaeological projects
and animal husbandry spanned millennia (Smith, 2001, 2011). But, are more attentive to skeletons than others. Not all bones that survive
we do not know if the shift was gradual, involving innumerable to the point of discovery might have been deemed worth saving, a par-
incremental and individually imperceptible changes, or if it occurred ticular problem for old skeletal collections, and contextual documenta-
in a stepwise fashion. It is here where a skeletal measure, which is tion can be unclear or has been lost.
based on subadult remains relative to other skeletons, indicates All burial locations do not have the same chance of being found
that there was at least one such episode of, archaeologically speak- and excavated. Much of the archaeological field work over the past
ing, rapid change interpreted as an increase in Neolithic fertility few decades has been in response to modern land use. Construction
(Bocquet-Appel, 2011; Bocquet-Appel et al., 2008). It remains to projects, of course, encounter archaeological sites in rural areas. But
be seen how closely this skeletal indicator corresponds to shifts in such work tends to be concentrated in and around urban settings, so
morbidity and mortality, as identified through paleodemographic cemeteries near cities have a greater chance of being excavated.
and paleoepidemiological studies of multiple skeletal samples with The total excavation of cemeteries can help reduce sampling
appropriately tight temporal controls. Nevertheless, evidence for problems (Dutour et al., 2003). For example, at the Danish Tirup site,
stepwise changes is consistent with Wood's (1998) model of popu- a rural church, and graveyard encompassed by a ditch, burials were
lation growth and production strategy intensification in the nonrandomly distributed by age, sex, social position, and disease
preindustrial world. experience (Boldsen, 1995, 2002, 2005a, 2005b, 2007; Kieffer-Olsen
et al., 1986). Many babies were buried around the church choir, nota-
bly east of it. Early in the graveyard's history women were often
3 | DEATH ASSEMBLAGES interred north of the church and men south of it, although that prac-
tice was later abandoned when family or farm burial plots came into
It is best to begin with the obvious: when examining archaeological use. Graves for important parishioners were located inside the church.
skeletons, one is looking at a sample of the dead, not the living. Here Finally, people with skeletal lesions consistent with leprosy were
there are actually two issues of interest—samples and deaths—and mostly buried directly west of the church. Anything less than the
they affect all that follows from the initial collection of data to how churchyard's complete excavation would have resulted in a biased
they are interpreted. sample of the people buried in it.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
120 BOLDSEN ET AL.

The preceding pertains to what happens after death has occurred. centuries. For these accumulations of human remains, it would
But what comes to a researcher's attention is a biased sample of the behoove researchers to determine if the skeletons came from people
people who were ever alive at a particular point in time (Wood who had much the same life experiences, regardless of whether they
et al., 1992). This is true even in a situation that is never likely to were, or were not, part of a single multigenerational group. The indi-
occur: the remains of everyone who lived in a community were dis- viduals might have shared nothing more than the same place of burial.
covered, the bones were all well preserved, and every excavator was For example, the Read shell midden in the American midcontinent
highly skilled. The bias comes about because death is selective with held skeletons interred over many centuries (Milner &
regard to conditions that increase the risk of dying, such as diseases, Jefferies, 1998). These people were members of hunting-and-
injuries, and occupational hazards. The individuals who die at a partic- gathering groups that camped at the site, with some being buried
ular age tend to be the sickest, weakest, and most vulnerable of their there. Although they could be considered representative of a particu-
birth cohort, with others living longer only to die later, often when lar way of life—in this instance, there was no demonstrably meaningful
they too are among the frailest of their age-mates. In short, mortality change from start to finish—these individuals did not necessarily have
is selective, and it acts on a heterogeneous population with regard to a direct descendant relationship with one another.
individual frailty (Vaupel et al., 1979; Vaupel & Yashin, 1985; For cemeteries that lasted generations, even when there was a
Weiss, 1990). descendant connection, the skeletons might have come from people
Furthermore, it is rarely possible to identify the full range of the who experienced quite different living conditions. A settlement's
once-living population that should be sampled to obtain skeletons occupational history can also affect the age composition of the indi-
representative of the entire biologically, socially, and economically viduals buried in a cemetery. Records for an Australian graveyard
interacting community. Like all other archaeological work, the most show that the age-at-death distribution for the first half-century of its
that can be done is to recognize that inferences drawn from skeletons use, when people flocked to the community, was markedly different
must be tempered by the limits of the available sample. That involves from what it was like long after the community was well established
taking advantage of what can be learned from habitation sites as well (Jackes, 2011). In short, one must be careful about what is represen-
as mortuary contexts, augmented by historical sources when ted by the skeletons being analyzed, hence what can be learned from
available. them, including their age and sex distribution.
Even when many people died at the same time for the same rea-
son, such as in battles (Boylston et al., 2000; Holst et al., 2018; Kyle
et al., 2018; Loe et al., 2014), there is still the issue of defining the 3.3 | Sequence of deaths to survival functions
social group, or groups, that contributed to the sample (Milner
et al., 2019). One cannot exclude selection bias affecting the skeletons Once skeletons are arranged from young to old, there remains the
resulting from catastrophic events, as has mistakenly been done for issue of what to do with them. Later in this article, we address what
Herculaneum's victims of Mount Vesuvius' AD 79 eruption complicates the estimation of age and generation of age distributions.
(Capasso & Capasso, 1999). After all, in modern mass casualty inci- Here, however, it is important to draw a distinction between different
dents excess mortality is unevenly distributed by age, sex, and socio- ways of arranging skeletons by age that underscore the difference
economic status (Myung & Jang, 2011; Santos-Burgoa et al., 2018; between looking at them from the perspective of the dead as opposed
Zagheni et al., 2015). The same can be expected to have occurred in to the living.
the past. The selective effect of mortality has even been identified in The simplest way of looking at deaths is to calculate the number
Black Death victims from a medieval London plague pit. Existing of skeletons in each age interval relative to the entire skeletal sample.
health conditions affected the risk of dying, perhaps including stature In other words, the denominator is the total number of dead people
as a measure of juvenile growth deficits (DeWitte & Hughes- (all skeletons), and the numerator is the number of deaths in each age
Morey, 2012; DeWitte & Wood, 2008). Furthermore, from documen- interval (a fraction of all skeletons). It is the proportion of the sample
tary records it is possible that in the Netherlands somewhat more that corresponds to each of the age intervals.
women than men died during high-mortality late medieval plague Two other ways of plotting deaths according to age—mortality
years (Curtis & Roosen, 2017). The general point is an elementary profiles and survival functions—have the same general appearance,
one: in mass disasters, deaths are unlikely to be entirely independent although they highlight a distinction between approaches that tend to
of age, sex, physical ability, or anything else affecting exposure or be descriptive and those that provide a deeper understanding of lives
susceptibility. in the past. While paleodemographic analyses of population structure
start with tallies of skeletons organized by age, the objective is to
translate this information into a survival function.
3.2 | Skeletal accumulations over time A mortality profile, as the term is used here, differs from the pro-
portion of the total sample represented in each age interval, which is
There is another reason why collections of archaeological skeletons described above. The mortality profile is the number of people in the
are not populations in the way one normally might think about them. skeletal sample who were alive at the beginning of each age interval
In all but unusual situations, the deaths took place over years, if not relative to the number of individuals in the skeletal sample
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 121

(Boldsen, 1984, 1988). It closely resembles a survival function as the regarded as typically male to female, only permit reliable assessments
resulting curve monotonically declines with advancing age. The word- after adolescence. Sex estimates from skeletal characteristics for sub-
ing here—in the sample—is important because the individuals might, adults are not sufficiently accurate for paleodemographic samples—
or might not, have been born in the community. That is, it includes the bones of children are often poorly preserved and size-based esti-
people who moved to the community, while it excludes those who mates would be affected by uncertainty about the precise age and
moved away. The mortality profile captures the distribution of the health status of the individuals examined—despite promising work
dead as seen from the vantage point of the skeletons a researcher has currently underway (Stull et al., 2017, 2020).
on hand. To estimate an adult's sex, one must take into account the popu-
A survival function describes the chance of survival as seen from lation from which the skeletons originated (Jantz & Ousley, 2020;
the perspective of the living. Because it is a measure of survival to a Klales, 2020; Ubelaker & DeGaglia, 2020). Skeletons from some parts
given age, it captures what took place in a once-living population. of the world tend to be more gracile than those from other places,
Numerators in the risk of dying are the number of people who died in and sexual dimorphism within groups is also variable.
each of the age intervals. Denominators for each age interval, how- In practical terms, sex estimation using adult skeletal structures is
ever, are composed of the people who were alive in the community at reasonably accurate and sufficient for paleodemographic work, as long
the onset of those same intervals. Because the survival function does as enough of the right kind of bones, notably the pelvis, are present
not have to be directly tied to individual skeletons—that is, it is a con- (Meindl et al., 1985). In American forensic investigations undertaken
tinuous function—it can be portrayed, more accurately approximated, by many researchers over several decades, more than 97% of the
at 1-year intervals. Here a simplifying assumption is made that the assessments based on the pelvis and skull were consistent with DNA
population is stationary, as is defined later. results, and accuracy improved slightly when long bones were also
The mortality profile and survival function distinction is perhaps present (Thomas et al., 2016). Success declined when fewer bones
best explained by looking at a decline in the number of individuals were observable.
with advancing age. Even in the impossible situation where everyone's A designation of sex is inherently a probability statement. For
age is known without error, arranging skeletons so there are ever- archaeological skeletons, uncertainty is imperfectly captured by the
decreasing numbers of individuals in sequential age intervals is not use of M? or F?, with more question marks indicating even greater
equivalent to the hazard of dying in each of those same age intervals doubt about sex. It would be better to calculate the probability of a
(age-specific mortality rates). The latter requires one to know the skeleton being from a person of one sex or the other, as can be done
number of people alive in each of the age intervals, as well as the for forensic cases when an appropriate reference sample is available
number who died (Boldsen, 1988; Horowitz et al., 1988; Johansson & (Boldsen et al., 2015; Jantz & Ousley, 2005; Klales et al., 2020;
Horowitz, 1986). Deaths can be obtained directly from a tally of skele- Milner & Boldsen, 2012a). But for archaeological samples, most of
tons; the number alive cannot. which are from preindustrial societies, finding such a reference sample
In modern national statistics, the survival function is based on age- verges on impossible.
specific mortality (typically in 1-year intervals) during a given period of When estimating the probability a skeleton was from a male or
time (usually 1 year). There might never have been a birth cohort that female, it is also necessary to have some idea about the sex com-
had the mortality experience summarized by an empirical survival func- position of the sample from which the skeleton was drawn
tion because the type and intensity of threats to survival change over (Albanese et al., 2005; Milner & Boldsen, 2012a). That is because
time. Archaeological skeletons can likewise be regarded as cross- probabilities estimated as if there were an equal number of men
sectional data since people are not followed throughout their individual and women are not the same as those when one sex greatly out-
life courses. They are sampled at one particular time: at death. numbers the other. One example would be a battlefield mass grave
Cemetery samples also differ from modern censuses, which gen- containing mainly men. In that burial context, a small skeleton is
erally pertain to a single year, because the deaths usually took place more likely to have been from a male than if it came from a ceme-
over much longer periods, often several centuries or more. The skele- tery that held an accumulation of skeletons that might normally
tal sample might be representative of a relatively uniform way of life, occur in a community.
or it could be a composite of people who experienced quite different
activity levels, pathogen load, dietary sufficiency, and the like. Identi-
fying the conditions of everyday life is largely an archaeological 5 | AGE E STIM ATION
(or historical) issue, hence the need for contextual information in
paleodemographic studies. Estimates of age from bones and teeth track biological changes, not
the time elapsed since birth. Discrepancies between biological and
chronological age in subadults are usually of little importance in paleo-
4 | SEX ESTIMATION demography because individuals are typically grouped into multiyear
categories to accommodate small archaeological samples. That is not
When estimating sex, it is commonly understood that bony features, true of adults where changes in skeletal features are only loosely
which for any single trait can be arrayed on a spectrum from those associated with chronological age.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
122 BOLDSEN ET AL.

5.1 | Age and the skeleton Before going further, two points should be emphasized. First, the
situation for paleodemographers is somewhat the reverse of what it is
Age estimates for subadult skeletons are based on bone and tooth for researchers seeking to clarify aging at the molecular, cellular, and
development and long bone lengths. Developmental processes tend organ system levels. From a biological or clinical perspective, change
to be under genetic control and tightly canalized, so changes in the in an organism is examined in terms of the relationship of known age
dentition and skeleton progress in a roughly similar manner and pace to mortality and morbidity. For health risk assessments, a case could
from one individual to the next. Therefore, ages estimated for reason- be made that inferior biological predictors of chronological age are
ably intact immature skeletons, especially those with complete denti- potentially more useful than superior ones (Pyrkov et al., 2018). Here
tions, are sufficient for most paleodemographic purposes. That is, one might be interested in identifying individuals who are at risk
skeletal age intervals are typically no longer than a few years, and because a measurable biological characteristic departs from what is
often less. Nevertheless, the logical and analytical issues discussed for normal for a particular age. A paleodemographer, in contrast, uses
adults also pertain to subadults (Kamnikar et al., 2018; Kumagai observable skeletal features to estimate an individual's age. That is,
et al., 2018; Stull et al., 2021). one approximates chronological age through an examination of mor-
The situation changes once one switches to bony structures reg- phological structures that are responsive to biological aging processes.
arded as markers of senescence (Garvin et al., 2012; Jackes, 2000a; Second, when dealing with biological markers of aging, one must
Pignolo et al., 2021). Here senescence—a deterioration over time in reckon with distinctions between samples of the dead and the living.
the physiological functions required to maintain life and a concomitant For paleodemographers, if a skeletal structure happened to be associ-
increase in the risk of dying—is used narrowly as it applies to the ated with an increased risk of dying, no matter how indirect it might
physical appearance of bones, rather than the biological process of have been, the transition from a young to old state would occur ear-
aging. For skeletal aging, the shift to senescent changes essentially lier in a sample of the dead than the living.
takes place after the fusion of the medial clavicle epiphysis and the
formation of the pubic symphyseal ventral margin, which occur in the
20s. Precisely what lies behind the pace of age-progressive senescent 5.2 | Conventional estimation method (in)
skeletal changes from that point onward is largely unknown, beyond a adequacy
general appreciation of the influence of lifelong wear-and-tear, habit-
ual activities, childbirth (pubic symphysis), body size, and injuries The use of different procedures for estimating adult age is an obvious
(Baccino & Schmitt, 2006; Becker et al., 2010; Merritt, 2015; problem faced by paleodemographers when trying to assess whether
Ubelaker & Khosrowshahi, 2019). Different reference samples have two or more samples were dissimilar or not. But it is arguably not the
been examined to accommodate varied genetic backgrounds and non- biggest difficulty. That is because no conventional age-estimation
genetic effects. But within any group of people, no matter how tightly method for adults yields satisfactory results (Buckberry, 2015; Milner
defined, there remains considerable variation among individuals. Skel- et al., 2019). Once one begins to rely on senescent skeletal changes,
etons are no different than what common experience and gerontolog- the effectiveness of widely used age-estimation methods deteriorates
ical findings would lead us to believe: some people look younger or rapidly with advancing years.
older than their chronological age for poorly understood reasons Cranial suture closure, for example, has long been recognized as a
(Barthold Jones et al., 2019). poor means of adult age estimation (Brooks, 1955; Dwight, 1890;
The heritability of the human life span is low, estimated at less McKern & Stewart, 1957; Singer, 1953). Over a century ago,
than 10% to as much as about 25%, so nongenetic influences loom Dwight (1890, p. 389) said that “the time and order of the closing of
large in how long people can be expected to live (Christensen the cranial sutures are very uncertain, far too much so for them to be
et al., 2006; Herskind et al., 1996; Ruby et al., 2018). The latter trustworthy guides to determine the age of the skull.” Figure 1, based
includes environmental effects, but also gene–environment interac- on Dwight's descriptions of 100 male and female crania, shows transi-
tions and stochastic variation arising during development (Finch & tions from open to any ectocranial sagittal and lambdoidal suture
Haghani, 2021). Nevertheless, a multi-tissue DNA methylation “epige- obliteration, as estimated through logistic regression. Sexes are com-
netic clock” shows a 0.96 correlation with chronological age bined for the lambdoidal suture because there is no significant differ-
(Horvath, 2013). Although for blood alone the clock's correlation with ence between them. The coronal suture is not shown because when
age was lower than what was reported in the original study, it still pre- scored in this manner it yields no information about age. The transi-
dicts mortality better than a variety of environmental and individual tion curves span adulthood, so visual examinations of sutures in their
characteristics associated with increased risks of dying (Marioni entirety do not tell us much about age, precisely what Dwight con-
et al., 2015). Other biological estimators exist, including a model based cluded when examining these particular skeletons. Since his work,
on physical activity where there was a correlation of 0.75 with chro- great efforts have been made to turn cranial vault and facial sutures
nological age (Pyrkov et al., 2018). For paleodemographers, these esti- into an effective means of adult age estimation (Acsádi &
mates provide a level of accuracy that one might strive to achieve Nemeskéri, 1970; Mann et al., 1987, 1991; McKern & Stewart, 1957;
with skeletal markers of age, as long as reasonably good morphologi- Meindl & Lovejoy, 1985; Nawrocki, 1998; Todd & Lyon Jr., 1924,
cal features can be identified and analyzed appropriately. 1925a, 1925b). Unfortunately, little has been gained from such
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 123

analyses, despite the use of innovative procedures (Hens & Brooks, 1955; Brooks & Suchey, 1990; Katz & Suchey, 1986;
Godde, 2020; Lynnerup & Jacobsen, 2003; Milner & Boldsen, 2012c; McKern & Stewart, 1957; Meindl et al., 1985; Schmitt et al., 2002;
Nikolova et al., 2019). Todd, 1920, 1921). The iliac portion of the sacroiliac joint has likewise
In most studies of archaeological and forensic skeletons greater been the subject of much interest, partly because it tends to be better
reliance is placed on age-progressive changes in the pubic symphysis preserved in skeletal remains than the pubic bones (Boldsen
and sacroiliac joint (iliac auricular surface) than on cranial sutures et al., 2002; Buckberry & Chamberlain, 2002; Igarashi et al., 2005;
(Falys & Lewis, 2011; Garvin & Passalacqua, 2012). The pubic sym- Lovejoy et al., 1985; Osborne et al., 2004; Rougé-Maillat et al., 2009;
physis has attracted attention for over a century (Boldsen et al., 2002; Schmitt et al., 2002). But even with these anatomical structures stan-
dard methods fail to yield satisfactory age estimates for adults.
To illustrate this point, one of us (G.R.M.) examined 234 well-
documented skeletons in the Bass Donated Collection at the University
of Tennessee. Skeletons were selected to span the adult lifespan, but
they were examined blindly with regard to age and sex. Results for the
left pubic symphysis and iliac auricular surface are shown in Figure 2.
The means of the Suchey-Brooks pubic symphysis phases are plotted
against reported ages for the Bass Collection individuals (Brooks &
Suchey, 1990). Because male and female means for each Suchey-
Brooks stage are quite close, they are averaged to simplify this presen-
tation. For the Lovejoy et al. (1985) iliac auricular surface age-
estimation method, the midpoints of “modal age” intervals are plotted,
except for the final open-ended age category. The age assigned to indi-
viduals in the terminal 60+ category is the mean of the ages of individ-
uals in our sample whose ilia correspond to the final stage. Because the
terminal sacroiliac category in Figure 2 is tailored to the age composi-
F I G U R E 1 The transition between open and any cranial suture tion of our sample, old looking iliac auricular surfaces artificially yield
obliteration is shown for 100 individuals reported by Dwight (1890): somewhat reasonable results. In an archaeological study, one would
Sagittal suture for females (SF), sagittal suture for males (SM), and
have no idea about what an open-ended 60+ category implies, other
lambdoid suture for both sexes (LB). By the end of the 19th century, it
than individuals with decrepit sacroiliac joints were presumably old.
was clear that cranial sutures contribute little to adult age estimation,
although efforts to extract information from sutures continue to the If the Suchey-Brooks and Lovejoy methods yielded a satisfactory
present day approximation of age, then the medians should be reasonably close to

F I G U R E 2 Commonly used methods are used to estimate the ages of adults using the left iliac auricular surface and pubic symphysis in
modern Americans from the Bass Collection. The distributions of ages by stage are shown in horizontal boxplots representing recorded ages; the
shaded areas represent the interquartile range (25th–75th percentile) for recorded ages; and the identity line is indicated. Lovejoy et al. (1985)
Iliac auricular surface (left) and Brooks and Suchey (1990) pubic symphysis (right) estimates are provided for 213 and 222 individuals,
respectively. The terminal open-ended iliac auricular surface stage appears to perform better than it actually does because the age estimate used
here is the average age of people in our sample whose ilia were assigned to the last stage
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
124 BOLDSEN ET AL.

the identity (diagonal) line, and the estimated ages (dots) would not Quick impressions of age-progressive changes in the skeleton can
show a large spread. Instead, both the Suchey-Brooks and Lovejoy be obtained when examining anatomical structures, such as the pubic
methods result in a systematic underestimation of the ages of most symphysis, in their entirety. But consistently classifying specimens is
people in the Bass Collection sample. There is also considerable over- difficult because features that distinguish one stage from the next do
lap in the ages of the different stages, especially for the iliac auricular not change in lockstep with advancing age. Dividing single anatomical
surface. units into parts and scoring them separately takes into account those
The results, notably the estimate bias from middle age onward, differences in the aging process. But doing so comes with a penalty
are not surprising as they are similar to what other researchers have because individually scored changes are unlikely to be independent
found with these age-estimation methods (Hens et al., 2008; Martrille from one another.
et al., 2007). We are pleased, therefore, that researchers have pursued Whatever is done, a useful age indicator must fulfill three criteria
more sophisticated means of analyzing morphological structures, but (Milner & Boldsen, 2012b). First, there must be an unambiguous clas-
the results for the pubic symphysis have been disappointing. sification of continuous changes that take place in an anatomical
Stoyanova et al. (2015, 2017) have generated 3-D surface scans of structure. Second, the stages must progress from younger to older
the pubic symphysis using geometric morphometric methods to esti- states without sequence reversals. Third, the timing of change from
mate age. Their samples included individuals ranging in age from the one stage to the next must be reasonably consistent across all individ-
teens to over 90 years, and their estimates plateau around 40– uals, at least those within broadly defined groups, such as when they
50 years of age. In fact, the maximum estimated age was 52 for every- are separated by sex or place of origin.
one over 50, who on average had an estimated age that was 27 years The last criterion—morphological change occurs at roughly the
younger than the actual age (Stoyanova et al., 2015). A validation same chronological age for everyone—is slippery because it can be
study of their method showed the familiar pattern of underestimating assessed for modern people, but not those who lived long ago. For
age in individuals over 40 years (Kotěrová et al., 2018). archaeological skeletons, there are few individuals of known age avail-
All of the above is consistent with what has long been recog- able to test the performance of methods based on modern people
nized about archaeological samples: there is a tendency for old peo- (Buikstra et al., 2006). Nevertheless, a uniformitarian position with
ple to be underrepresented, so there is a corresponding surplus of regard to the aging process is central to paleodemographic work
young adults (Aykroyd et al., 1997, 1999; Konigsberg & (Howell, 1976). Researchers must make what Müller et al. (2002, p. 4)
Frankenberg, 1994; Milner et al., 1989). For example, a marked called an “invariant” assumption about the age progression of what-
underenumeration of individuals older than 60 years was found in a ever skeletal indicator is used: the pace of change in the past is the
comparison of skeletal age estimates and parish records for a 19th- same as it is today.
century Canadian cemetery (Saunders et al., 2002). The Canadian Researchers who have examined many known-age skeletons
age estimates were based on the same pelvic scoring systems used recognize that there can be interindividual variation in the rate of
in the Figure 2 estimates. If conventional methods produce consis- change in skeletal indicators of age. Furthermore, multiple bony
tently erroneous results—that is our position, although others do not features in the same person can appear to be ahead or behind
agree (Meindl et al., 2001, 2008; Meindl & Russell, 1998)—then the what might be expected for the individual's chronological age. Of
resulting age-at-death distributions, and inferences based on them, greater concern is the possibility that there are systematic differ-
are flawed. As discussed below, however, there is reason for opti- ences among the world's populations in skeletal markers of age. It
mism based on a more diverse suite of skeletal characteristics and is, therefore, important to look at ordinal categories to see if the
innovative analytical methods. timing of change from one stage to the next is uniform across
populations (Frankenberg, 2021; Konigsberg & Herrmann, 2006).
As indicated above, one ideally wants universally applicable age-
5.3 | Skeletal indicators of age informative skeletal traits. If that is not possible, it will be hard to
find appropriate reference samples to satisfy all the needs of
Before looking toward the future of age estimation, it is best to start forensic investigations. Doing so for archaeological samples will
with the bony features that track skeletal changes throughout adult- rarely, if ever, be possible because of the unknown mix of genetic
hood. Two general approaches have been adopted. Most commonly, backgrounds and physical activities that might have affected the
entire anatomical units, such as the pubic symphysis or the sacroiliac appearance of the skeletons of people who lived in the past.
joint, have been divided into a sequence of stages (Brooks & There are additional practical considerations when selecting age
Suchey, 1990; Lovejoy et al., 1985; Todd, 1920, 1921). The alterna- indicators. Many skeletons, preferably hundreds of them, are needed
tive, referred to as a component approach following McKern and for rigorous paleodemographic analyses, so it must be feasible to col-
Stewart (1957), involves dividing an anatomical structure into its con- lect age-informative data quickly. It would be best if that could be
stituent elements, scoring them separately, and combining the scores done without destroying specimens to examine bone and tooth
in some manner to yield an age estimate (Boldsen et al., 2002; microstructure. Destructive analyses are not always possible, they are
Buckberry & Chamberlain, 2002). expensive, and they require special equipment and training.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 125

5.4 | From skeletal indicators to age estimates Konigsberg et al., 2008, 2016; Konigsberg & Herrmann, 2006;
Milner & Boldsen, 2012c; Tangmose et al., 2015). Relying on such
Identifying age-progressive skeletal changes is only the first step. transitions addresses the age mimicry problem when age intervals and
Beyond that point, there are two issues of concern. The first will come central tendencies are derived directly from reference samples. More-
as no surprise, although its implications are rarely considered. The sec- over, it provides a way to determine if separate morphological stages
ond undermines all conventional age-estimation methods. defined in existing or future age-estimation methods should be
First, estimates of age for individual skeletons can either be point retained or combined with one or more adjoining stages
estimates or intervals, with the latter generally being favored. Age (Konigsberg & Herrmann, 2006). The transition-based framework has
intervals accommodate the fact that alterations in bony structures do involved the use of different analytical procedures, but from this point
not progress at the same rate in all individuals, and biological changes onward the approach of Boldsen et al. (2002) operationalized for bio-
are not the same as chronological ages. archaeologists is referred to as Transition Analysis (statsmachine.net/
Despite capturing uncertainty in the estimation process, age software/).
intervals derived directly from a reference sample, such as those Transition Analysis version 2, designated TA2, is based on cranial
found in frequently used methods (Brooks & Suchey, 1990; Katz & suture closure and changes that occur in the pubic symphysis and iliac
Suchey, 1986; Lovejoy et al., 1985), imply one is equally sure that all portion of the sacroiliac joint (Boldsen et al., 2002). Logistic regression
skeletons placed in a particular interval actually belong in it (Milner curves characterize the transition from one stage to the next in multi-
et al., 2019). Age estimates will be tighter for some skeletons than ple parts of the two pelvic joints and in several suture segments. To
others based on the bones observed and the perceived consistency of maximize information for morphological categories ordered from
age indicators. The problem is exacerbated by morphological stages young to old, the skeletal sample is separated into all individuals up to
with age ranges that do not overlap (Todd, 1920, 1921). Long over- a certain stage as opposed to those who were in higher ones. The var-
lapping age intervals encompass the true ages of many individuals, but ious parts of the pubic symphysis, sacroiliac joint, and cranial sutures
do so by sacrificing the precision needed for paleodemographic work. are used to generate three age estimates, one for each anatomical
The problematic second issue was first noted by Bocquet-Appel structure (separate sutures are treated here as one structure). The
and Masset (1982), later termed “age structure mimicry” (Mensforth, three anatomical structures are then combined to produce a compos-
1990, p. 91; now simply age mimicry, e.g., Konigsberg & ite estimate for each individual using an uninformative prior distribu-
Frankenberg, 1992). In conventional age-estimation methods, age tion, American homicides, or 17th-century Danish rural parish deaths.
interval limits and central tendencies for a particular expression of a It is worth emphasizing that TA2 does not, in fact, yield chrono-
skeletal feature are taken directly from the distribution of individuals logical ages. Instead, it estimates the age when a particular combina-
in a reference sample. Although doing so might at first seem reason- tion of traits are most likely to co-occur. That is, biological features—
able, it leads to biased age estimates when the method is subse- senescent changes for the most part—are scored by an observer;
quently used to age archaeological or forensic skeletons. chronological age is not assessed by, say, counting the years since
Several researchers have detailed the statistical basis for why birth.
biased results arise when age and morphological stage are treated as Useful results are not guaranteed by TA2. While there is a general
the dependent and independent variables, respectively (Aykroyd correspondence between estimates and age, maximum likelihood esti-
et al., 1997, 1999; Bocquet-Appel & Masset, 1982; Konigsberg mates based on morphological features can be far from reported ages
et al., 1997; Konigsberg & Frankenberg, 1994, 2002). Here it is suffi- (Milner & Boldsen, 2012c, fig. 2). There is little hope for cranial
cient to point out that progressing through morphological stages as sutures, included in a largely failed effort to salvage something from
defined for an anatomical structure does not cause a person to get them. For the two pelvic joints, the situation is worse for the sacroiliac
older. Treating age as the dependent variable means the age interval joint than it is for the pubic symphysis. Sacroiliac point estimates for
and central tendency for a stage reflect the age composition of the the first several decades of adulthood approximate increasing chrono-
reference sample used to develop the age-estimation method. The logical age, although beyond about 60 years nothing is gained by
resulting age mimicry in the archaeological skeletons of interest can looking at the joint because signs of senescence, as scored, do not
result in misleading interpretations. As pointed out by Konigsberg and progress further (Milner & Boldsen, 2012c, fig. 3). For both pelvic
Frankenberg (1992), in the extreme case where a bony structure is joints, age intervals—they are of primary interest, not point
uninformative about age the resulting age-at-death distribution estimates—throughout adulthood are long. In fact, they are often so
matches that of the original reference sample. While no researcher long that their potential contribution to estimating the ages of individ-
would rely on such a skeletal feature, even frequently used bony ual skeletons is severely compromised. Nevertheless, TA2 performs
structures, including the pelvic joints, are only loosely related to adult better than standard methods based on readily visible skeletal mor-
age (Figure 2). phology, although that is damning with faint praise (cf. Figure 2 to
A different estimation approach is to calculate the probability at Milner & Boldsen, 2012c, figs. 7 and 8).
each age of having made the transition from one stage to the next When dealing with skeletal features in this manner, there is a
(Boldsen et al., 2002; DiGangi et al., 2009; Fojas et al., 2018; problem of correlated traits. A study based on a different way of clas-
Frankenberg, 2021; Godde & Hens, 2012, 2015; Jooste et al., 2016; sifying morphological change in the pubic symphysis, sacroiliac joint,
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
126 BOLDSEN ET AL.

and cranial suture segments has highlighted the correlation among Konigsberg & Frankenberg, 1992; Love & Müller, 2002). Fortunately,
scores, especially those within either the pelvis or the cranium, after work is progressing on that difficult task (Gampe et al., 2012).
controlling for age (Frankenberg, 2021). In TA2 this issue remains To judge from TA2 results, the choice of a suitable prior distribu-
unresolved, except through an ad hoc procedure (Boldsen et al., 2002; tion is mainly a concern in the upper part of the lifespan where the
Fieuws et al., 2016). pelvic joints and cranial sutures are poor age indicators. In a compari-
son of estimates generated through a uniform distribution versus one
from 17th-century Danish records, there was no discernible differ-
5.5 | Prior distributions ence in maximum likelihood age estimates up to about 50 years
(Milner & Boldsen, 2012c, fig. 11). Beyond that point, estimates were
Much discussed in the literature is the issue of specifying an appropri- pulled upward or downward toward about 75 years; in other words,
ate prior distribution in a Bayesian analytical framework, regardless of toward the modal adult age-at-death in 17th century Denmark. This
whether one is interested in the ages of single skeletons or the age- tendency can have a noticeable effect on the upper portions of
at-death distribution of an entire sample (Boldsen et al., 2002; archaeological age distributions (Bullock et al., 2013). The result is an
Caussinus & Corgeau, 2010; Konigsberg & Frankenberg, 1994, 2013; adult modal age-at-death that might look reasonable, but is largely an
Lucy et al., 2002; Maaranen & Buckberry, 2018; Müller et al., 2002; artifact of the prior distribution. Skeletal traits that capture old age
Sasaki & Kondo, 2016a; Séguy et al., 2013; Séguy & Buchet, 2013). A better would dampen the informative prior's effect on the sample's
distinction in the use of prior distributions should be mentioned here, age-at-death distribution.
which is taken up again later. Prior distributions have been used when Concerns over appropriate prior distributions, however, pale to
estimating the ages of individual skeletons. The resulting ages are insignificance when compared to the inaccuracy and imprecision of
then combined to approximate the age-at-death distribution of an TA2 age estimates for individual skeletons (Milner & Boldsen, 2012c).
entire archaeological sample. This is their use as described in this sec- When looked at separately, the pubic symphysis generally produces
tion. Alternatively, a prior distribution could be applied to a sample the best estimates followed by the sacroiliac joint, with the cranial
after skeletons have already been allocated to a series of age intervals sutures finishing a distant third, similar to what another validation
according to the probability an indicator stage would occur in each study found (Maaranen & Buckberry, 2018).
age interval. This approach foregoes the need to estimate the ages of
individual skeletons as a first step.
For medicolegal work, one might use national mortality, homicide, 5.6 | A new approach to estimate adult age
or forensic caseload figures. Similar information is not available for
most paleodemographic applications, except when records exist for Perhaps only one iron-clad conclusion can be reached after more
the past few centuries. A number of studies have used the Danish his- than a century of intensive work by many researchers who have
torical distribution provided in TA2 (Ahlström, 2015; Boldsen, 2007; approached the adult age-estimation problem in various ways using
Buikstra et al., 2006; Bullock, 2013; Bullock et al., 2013; DeWitte, different samples: standard age-informative skeletal features are not
2015; DeWitte & Hughes-Morey, 2012; Ham et al., 2021; Hughes- very informative at all. That is especially true for the uppermost part
Morey, 2016; Kelmelis & DeWitte, 2021; Wilson, 2014; Zoeller of the lifespan. Something else must be done to produce better age
et al., 2021). While derived from a premodern population, its use for estimates, both at the level of the skeletal traits and in the way they
paleodemographic purposes is not entirely satisfactory because there are handled.
is no reason to believe it approximates all such distributions. An obvi- Both issues are being addressed simultaneously in an ongoing
ous example is when an archaeological sample did not come from research effort based on over 1700 skeletons from collections on four
deaths that simply accumulated over time, but was instead the prod- continents (Milner et al., 2021). The research initiative was inspired by
uct of a catastrophic event where the probability of dying was to a the success of subjectively assessing age from the general appearance
large extent independent of age. of skeletons. This initial work was undertaken by several osteologists
The problem of appropriate prior distributions for archaeological over a period spanning almost two decades when visiting collections
samples has not yet been resolved. Model life tables have been used in several countries.
by Gowland and Chamberlain (2005) and Gowland (2007), as have his- Experience-based estimates performed better than TA2 results in
torical records (Godde et al., 2020; Godde & Hens, 2021) and modern, a blind examination of well-documented skeletons in the Bass
historical, and anthropological populations characterized through Donated Collection (Milner & Boldsen, 2012c, figs. 7, 8, and 14). Such
Gompertz parameters (Sasaki & Kondo, 2016a). An approach that is estimates by one of us (G.R.M.) are plotted against known ages in
gaining traction is the use of a combined sample of many populations Figure 3 for 234 skeletons, the same ones as in Figure 2 where the
regarded as representative of preindustrial societies (Caussinus & results of standard methods of aging the pubic symphysis and iliac
Corgeau, 2010; Łukasik et al., 2021; Séguy et al., 2008, 2013; Séguy & auricular surface are shown. Also included are blind estimates of age
Buchet, 2013). What would be best as an informative prior is the age- by many observers based on photographs of 4374 individuals
at-death distribution one is striving to obtain when examining skele- (Barthold Jones et al., 2019). These data were from people who were
tons in the first place to avoid age mimicry (Hoppa & Vaupel, 2002b; at least 15 years old, and the estimate for each person was randomly
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 127

selected from multiple assessments in the database (reshare. explored, including a modified version of TA2 and machine-learning
ukdataservice.ac.uk/853684). There is reason to believe an osteolo- approaches, both featuring a new suite of bony traits, in the latest
gist who has examined closely numerous known-age skeletons from iteration of transition analysis (TA3). While the procedure is not yet
diverse collections can estimate age from bones as well as members fully developed, the work to date is encouraging in terms of what can
of the general public do when they observe soft tissue features during be done with a wide array of skeletal traits when analyzed
the course of their daily lives. But that can only be done by looking at appropriately.
a large suite of skeletal structures, many of which have not received To illustrate the value of a new set of skeletal age indicators,
much, or any, attention in previous research. Figure 4 shows ages of transition estimated through logistic regres-
Although, individually and collectively the subjective assessments sion for three femoral features, scored as dichotomous traits (absent
can be quite accurate when undertaken by members of our research or present). The development of lipping around the femoral fovea and
team, established quantitative procedures are needed to estimate the ossifications where muscles insert in two places on the proximal
uncertainty in age assessments and to ensure that the results are rep- femur are scored. Their age transitions are compared to the shift from
licable. Both are required to undertake comparative analyses, espe- Suchey-Brooks pubic symphysis Stages 5–6 (Brooks & Suchey, 1990).
cially those involving multiple investigators. Various ways of handling The same 234 Bass Collection skeletons, both males and females,
skeletal indicators to generate age estimates are currently being were used for the femoral and pelvic features. For the purposes of
age estimation, steep transitions are better than gradual ones. Of
course, the ideal—the patently absurd situation where a bony charac-
teristic changes predictably and only on one's birthday—never hap-
pens. Using the slope criterion alone, the proximal femur outperforms
the pubic symphysis during middle adulthood. The entire skeleton is
full of such age-informative features, and in combination they hold
the key for improving age estimates for adults.
Taken together, many such skeletal traits capture much of what
an experienced osteologist observes when examining a skeleton
(Figure 3). Figure 5 shows estimated ages for 537 individuals from a
worldwide sample of male and female skeletons generated through
partial least squares regression (Zelditch et al., 2012). The estimates,
plotted against reported ages, are generated from 20 ordinal and
dichotomous traits distributed widely throughout the skeleton. The
correlation between estimated and reported ages is 0.91, and the root
mean square error is 7.9 years. For this example, five traits came from
the innominate; three each from the femur, lumbar vertebrae, and
clavicle; and one each from the calcaneus, humerus, ribs, sacrum,
F I G U R E 3 Blind assessments of age for 234 Bass Collection
scapula, and sternum. Most of the traits involve the addition of bone
skeletons (dark dots) and 4374 photographs of living individuals (light
with advancing age. The loess trend line approximates the identity
dots) by multiple participants in a database compiled by Barthold
Jones et al. (2019). Estimates from skeletal remains and photographs line, except in the uppermost part of adulthood where age is under-
of people both approximate chronological age estimated. These results, presented for illustrative purposes from

F I G U R E 4 Curves showing
ages of transition for three
proximal femoral features, fovea
capitis lipping (FC) and muscle
insertion ossifications (TR, TF),
compared to the Suchey-Brooks
pubic symphysis Stage 5–6
transition in 234 modern
American Bass Collection
skeletons
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
128 BOLDSEN ET AL.

death distribution. Generating such a distribution, however, involves

more than simply counting skeletons because of the assumptions and
decisions that lie behind the process.

6.1 | Assumptions in paleodemography

6.1.1 | Stable and stationary populations

The assumption that a skeletal sample came from a stable population

has long been a cornerstone of paleodemography. It is an artificial
model insofar as a population, closed to migration, has age-specific
fertility and mortality rates that remain fixed over time. That means
F I G U R E 5 Estimates and reported ages are plotted for birth and death rates, hence the intrinsic growth rate, remained con-
537 individuals from a worldwide sample of male and female
stant during the cemetery's use. As an initial starting point, it is not an
skeletons. The estimates were generated using 20 traits from
unreasonable assumption, given the tendency of populations to
throughout the skeleton and partial least squares (PLS) regression.
The identity (solid) and loess trend lines (dashed) are shown approximate a stable state over time, as understood early in
paleodemographic studies (Weiss, 1973). It is even true for small
ongoing work, are not entirely satisfactory because age mimicry is not populations subject to stochastic variation in fertility and mortality
completely eliminated. Nevertheless, a capacity to estimate age well (Ward & Weiss, 1976; Weiss & Smouse, 1976). Under fixed fertility
into advanced years is clear. Good age estimates span the duration of and mortality rates, a population's initial age structure is reestablished
the adult lifespan likely to have been typical of the preindustrial socie- in well under a century following a disturbance, such as unevenly dis-
ties that paleodemographers tend to study. tributed excess deaths from war, famine, or disease (Paine, 2000;
When advocating for a new approach to estimate the ages of Weiss, 1975). Here researchers can find solace because most large
adult skeletons, we must abide by our own admonition that scoring skeletal samples are composed of deaths that occurred over multiple
skeletons should not take up much time. Once one is familiar with the generations.
procedure, our experience in several collection facilities has shown it An additional simplifying assumption is typically made: the skele-
takes about half an hour to get a boxed skeleton off a shelf, record tons were drawn from a stationary population. In that special case of
the age indicators of interest, repack the bones, and return the box. a stable population, intrinsic growth is zero. It has long been recog-
That is longer than it takes to score the pelvic joints with standard nized that the stationary assumption affects inferences from age-at-
methods (Brooks & Suchey, 1990; Lovejoy et al., 1985). But the death distributions (Asch, 1976; Bennett, 1973; Buikstra, 1976;
reward for spending a bit more time is much better information. Frankenberg & Konigsberg, 2006; Horowitz et al., 1988; Johansson &
In terms of the scored bony features, the step forward in this Horowitz, 1986; McCaa, 2002; Moore et al., 1975; Wood et al.,
approach is the use of age-informative skeletal variation in virtually all 2002). In a stable population, the deaths in an age interval are deter-
parts of the skeleton. Individually the traits yield little information—merely mined by age-specific mortality and the number of individuals who
early or late stages, although transitions for the skeletal features occur at reached that age. When populations are growing, with each passing
different points in adulthood—but collectively they provide much better year there are more births in each cohort than in the previous one.
age estimates than the skeletal structures that have been the cornerstone That means the birth cohorts contributing young members to the
of osteological research for over a century. Yet in spite of the advance, it cemetery were larger than the birth cohorts that are contributing
still remains to be seen which statistical methods will be best at finding older people. Because many children die, especially infants, and skele-
the age at which the observed pattern of scores for any single individual is tons accumulate over time in archaeological cemeteries, the result
most likely to occur. As the joint correlation between stage scores and would be noticeable because the skeletal sample is enriched in young
reported age at death is high, reaching the 0.9 target recommended by individuals, assuming good bone preservation. When the size of a
Bocquet-Appel and Masset (1982, 1996), differences in how one might population is neither increasing nor decreasing over time—that is, it is
find the most likely age for a suite of scores is less critical than it would be stationary—the age-at-death distribution is determined by age-
for more traditional age estimation methods that perform poorly. specific mortality. That is why the stationary assumption is so attrac-
tive, other than it simplifies calculations.
At first blush, the stationary assumption might seem reasonable
6 | S K E L E T O NS TO A G E- A T - D E A T H because population growth at continental to global levels was
DISTRIBUTIONS extremely low, approximating zero, for virtually all of human exis-
tence. That is the position advanced at the beginning of paleo-
Once reasonably good skeletal age estimates are in hand—no small demography as we know it today (Acsádi & Nemeskéri, 1970). On
task—they can be used to approximate the skeletal sample's age-at- average, growth over any appreciable period of time, such as the
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 129

lengthy forager to agricultural transition, must have been only a tiny 6.1.2 | Migration
fraction of a percent. That can be shown by one study of the shift to
agriculture from hunting-and-gathering in the Levant (Eshed Migration in the deep past—both of large numbers of people all
et al., 2004). The effect on age-at-death distributions was illustrated at once and gradual trickles of them over time—has attracted increas-
with a growth rate of 1%. Such growth could well have occurred as ing archaeological attention over the past several decades
episodic bursts within the time intervals spanned by one or both of (Anthony, 1990; Burmeister, 2000; Gregoricka, 2021; McSparron
the samples, each made up of skeletons from several sites. But it did et al., 2020; Smith, 2014; Snow, 2009). If movement was unequally
not take place over the entire ca. 2500 years that separated the two distributed by sex and age, one might detect alterations in the
samples. If there was only one couple at the outset—Adam and Eve composition of skeletal samples. But, to date, it has not been a signifi-
frolicking in an Eden free of Malthusian constraints and inbreeding cant part of paleodemography (Boldsen, 1984). In fact, in
depression—that rate of growth would yield about 127 billion people paleodemographic analyses, it is tacitly, if not explicitly, assumed
after some 2500 years. migration can be safely ignored. The problem with doing so was rec-
Simply assuming stationarity, however, is specious reasoning. ognized several centuries ago: estimates from “Bills of Mortality”
Skeletons are not drawn from enormous land masses where local require “that the People we treat of should not at all be changed, but
increases and declines averaged out to approximate zero. They gener- die where they were born, without any Adventitious Increase from
ally come from a single community, or several of them in a tightly Abroad, or Decay by Migration elsewhere” (Halley, 1693, p. 597).
defined area. These groups—people who once lived in one settlement We can be sure that there has always been some movement from
or a few nearby ones—presumably went through episodes of local one residential group to another. If in- and out-movement was essen-
population expansion or contraction. The former might have resulted tially balanced among neighboring communities, such as the exchange
in group fission and the movement of some people to other places; of marriage partners, it would be invisible to researchers. The possible
the latter could have progressed to community extinction. In that effects of asymmetric movement, however, are for the most part pas-
respect, lengthy cemetery use is arguably a desirable feature of sed over because it is difficult to estimate the magnitude of migration,
archaeological sites because short-term stochastic changes in and who took part in it. Nevertheless, it appears that selective migra-
populations are dampened. But for any specific skeletal sample, and in tion was a factor in the growth of urban centers in Scandinavia, and
the absence of corroborating archaeological information, that is more presumably elsewhere as well, and it can be identified through skeletal
of a hope than a certainty. remains (Petersen et al., 2006).
Not taking into account the intrinsic growth rate can have a dis- Turning first to immigrants, their effect on the age distribution of
quieting effect on a skeletal sample's age distribution, as illustrated by skeletal samples would be hard to detect if it is assumed that mainly
Séguy et al. (2008, fig. 2). Lacking reliable information on the direction young adults moved and they experienced the same fertility and mor-
and magnitude of population change, various values of intrinsic tality rates as locally born people (Paine, 1997). But, there could be a
growth can be used to capture the likely range within which the true discernible signature if the newcomers experienced a higher risk of
shape of the age-at-death distribution, as generated from skeletons, dying than a community's original inhabitants. It might be found if
lies. A range of possible growth rates is an honest admission of the urban settings were unhealthy enough places that they had turned
uncertainty inherent in paleodemographic estimates. Yet doing so into population sinks that drained rural communities of surplus young
means that mortality distributions from skeletons would have to be people of limited means who had few prospects of a good life at
quite different and the range of possible growth rates rather narrow home. An extreme case, unlikely in archaeological contexts, is the
before archaeological samples can be said to be truly dissimilar from shocking mortality among soldiers shipped to early 19th century garri-
one another. sons in equatorial West Africa (Curtin, 1990).
There is, however, some good news, and here we return to a dis- Even in the best of circumstances, demonstrating migration
tinction, covered earlier, between mortality profiles and survival func- through age-at-death distributions requires large skeletal samples and
tions. A mortality profile generated from an unbiased sample of accurate age estimates. Detecting the effects of migration would be
perfectly aged skeletons drawn from a stationary population is the facilitated if the individuals who had moved could be identified in
same as the survival function. If these aging and sampling criteria are archaeological samples. Fortunately, many studies have done so by
met and the burial ground was used for a long time, a practical issue finding nonlocal individuals based on the chemical composition of
comes into play that will please paleodemographers. When average bones and teeth, notably using stable isotopes (Bentley, 2006;
population growth is close, but not quite equal, to zero, it would not Katzenberg & Waters-Rist, 2019; Knudson & Price, 2007; Price et al.,
be possible to distinguish an archaeological sample from a stationary 2000). An effective marriage between chemical and paleodemographic
population. Intrinsic population growth, therefore, would not badly findings, however, lies in the future. It involves the chemical characteri-
distort information estimated directly from a skeletal sample relative zation of many, if not all, of the individuals in the large skeletal samples
to the survival function parameters that capture the experience of required for paleodemographic work. But it also involves addressing
once-living people. This statement is not an excuse for sloppy additional issues.
methods. It simply recognizes that it is possible to learn something First, the population of interest must be defined, and that could
from archaeological skeletons. involve going beyond the specific burial context or settlement of
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
130 BOLDSEN ET AL.

origin. Precisely what is sampled might be uncertain, for example, in stages defined in some standard methods inevitably leads to a heaping
early urban centers with neighborhoods consisting of people of of individuals on a limited number of ages. Furthermore, there are
diverse origins (Carballo & Fortenberry, 2015; Price et al., 2000; simply more age categories—for example, adulthood is commonly
Storey, 1992). What a nonlocal signature really indicates about physi- divided into 5 or 10-year intervals in paleodemographic life tables—
cal distance has given rise to misleading interpretations of lifetime than there are age-indicator stages in methods such as the Suchey-
movement (Frei et al., 2015; cf. Thomsen & Andreasen, 2019; Brooks classification of pubic symphysis morphology (Brooks &
Thomsen et al., 2021). Physical distance is also not the same as social Suchey, 1990; Katz & Suchey, 1986). Having only a handful of possi-
distance when identifying the population one believes is being sam- ble point estimates is of little help when characterizing mortality
pled. Depending on the geological setting, a nonlocal signature might patterns.
only indicate someone from outside a narrowly defined locality who The more reasonable alternative is to allocate individuals to the
was nevertheless part of the same interacting group, perhaps from a overall age-at-death distribution based on the age intervals assigned
neighboring village. to each skeleton (Alesan et al., 1999; Asch, 1976; Boldsen, 1988;
Identifying migrants does not tell us much about what took place Eshed et al., 2004; Storey, 1992; Ubelaker, 1974, 2001). Distributing
during the growth of urban centers other than the arrival of some individuals across multiple years acknowledges the ambiguity associ-
people. The number of nonlocal people in a mortality sample is not as ated with skeletal age estimation. There are different ways of doing
informative as the proportion of migrants among once-living commu- so, the most unrealistic being the equal allocation of a fraction of an
nity members, hence their possible influence on recipient communi- individual to each of the years between the lower and upper bounds
ties. Estimating the fraction of a settlement's living population that of a skeleton's age range. A better approach would be to apportion
originated in some other place requires good age-at-death data to individuals according to the probability of a person having died in a
account for possible differences in mortality between the new arrivals particular year, the next one, and so on. Age estimates produced by
and locally born people. Age estimates should be combined with evi- TA2, for example, might be used for that purpose.
dence of bony lesions to determine what happened to these people
after they arrived in the new community. That could be done through
an existing procedure for estimating the relative risk of dying in differ- 6.2.2 | Life tables and parametric models
ent groups of skeletons (Boldsen et al., 2015).
To date, immigrants have attracted attention simply because they Prior to the 1970s, age and sex were usually reported, if at all, in long
are potentially visible in cemetery samples through bone chemistry. lists of skeletons. Rarely were skeletons grouped into age intervals
Dealing with net migration is more difficult because the people who that yielded some appreciation of the distribution of deaths, occasion-
left a community were not buried in it. But their departure, such as in ally with idiosyncratic calculations of questionable utility (Angel, 1971;
European villages where historic period migration was mostly from Blakely, 1971; Goldstein, 1953; Hooton, 1930; Johnston &
rural to urban settings, might distort the mortality profile in relation to Snow, 1961; Snow, 1948; Stewart, 1962; Todd, 1927). Life tables
the survival function. People who emigrated were at risk of dying represented a major step forward as a way to systematize and sum-
before they left their home community, often as young adults if marize data derived from the ages of multiple skeletons (Acsádi &
centuries-long migration patterns can be extended back to a more Nemeskéri, 1970; Asch, 1976; Bennett, 1973; Blakely, 1977; Boldsen,
remote past. In such situations, the juveniles in skeletal samples would 1979; Buikstra, 1976; Green et al., 1974; Lovejoy et al., 1977;
be inflated because some survivors eventually moved away. By doing Owsley & Bass, 1979; Palkovich, 1980; Swedlund & Armelagos, 1969;
so, the adult skeletons that otherwise would have ended up in the Ubelaker, 1974). Generating them involved working backwards from
cemetery are correspondingly reduced. deaths to what was of interest, such as life expectancy (see
Frankenberg & Konigsberg, 2006). In that respect, it was much like
what Carrier (1958, p. 149) presented years earlier as a “simple guide”
6.2 | Generating mortality distributions for those interested in “bygone populations” where only deaths are
known. He pointed out the need to assume the individuals were
6.2.1 | Age-at-death distributions from tallies of drawn from a stable population, usually treated as a stationary popula-
skeletons tion, as discussed previously.
Life tables, however, are data hungry, and archaeological samples
Usually, mortality distributions from archaeological samples are gener- are usually small, rarely numbering over several hundred skeletons.
ated by assigning skeletons to their proper position according to their Age intervals often have few individuals in them, so there are irregu-
separate age estimates. That is done using point estimates or age larities and gaps that would not exist if the sample happened to be
intervals. larger. Forcing what is observable, biological age, into fixed life table
While the use of point estimates of age simplifies the process of age intervals, which typically encompass 5 or 10 years apiece, is also
producing an age-at-death distribution, it does not capture the uncer- an issue (Roksandic & Armstrong, 2011). The generation of life tables
tainty associated with estimating age from bones and teeth. The use is one thing, their interpretation is quite another (Hoppa, 1999). Per-
of age-interval midpoints or means associated with morphological haps most notable in bioarchaeological applications is the downstream
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 131

effect of juvenile underenumeration on plots of the distribution of societies based on subsistence agriculture, including those where mar-
people who died at older ages. Although sometimes overlooked, this kets are only peripheral to local economic activities (Boldsen, 2000;
problem was recognized early in the systematic examination of what Boldsen & Paine, 1995). Taken together, the three-component curves
could be done with the ages of skeletons (Green et al., 1974; Moore are sufficiently flexible to fit most mortality distributions. But despite
et al., 1975). its potential value, with few exceptions the Siler model, sometimes
Life tables still appear in the paleodemographic literature, altered by omitting the second component, has yet to find wide use in
although there have been calls to move beyond results derived solely paleodemography (Ahlström, 2015; Herrmann & Konigsberg, 2002;
from skeletons ordered by age. The desirability of doing so for archae- Nagaoka et al., 2006; Redfern & DeWitte, 2011a, 2011b; Stanton &
ological samples has long been advocated for both statistical and data Hefner, 2016).
quality reasons (Frankenberg & Konigsberg, 2006; Konigsberg & Gompertz and Gompertz-Makeham models, and others that are
Frankenberg, 2002; Séguy et al., 2008, 2013; Séguy & Buchet, 2013; used (Konigsberg & Herrmann, 2002; Wood et al., 2002), have also
Weiss, 1973; Wood et al., 2002). For example, a half-century ago been employed to approximate adult age-at-death in archaeological
Weiss (1973) developed a series of model life tables to accommodate samples (DeWitte, 2015; DeWitte et al., 2013; DeWitte & Yaussy,
anthropological, including skeletal, needs that complement those 2020; Konigsberg & Frankenberg, 2002; Milner & Ferrell, 2011; Red-
widely used by demographers (e.g., Coale et al., 1983). The objective fern & DeWitte, 2011a, 2011b; Sasaki & Kondo, 2016b; Stanton &
is to find a best fit between flawed skeletal data and a model that, in Hefner, 2016; Wilson, 2014). The emphasis on adults accommodates
turn, is thought to approximate the characteristics of the past three issues; two are practical, while the third is of more interest. First,
population. However, this is not the end of the story; for example, archaeological samples are typically deficient to some unknown
published model life tables do not capture the pattern of childhood extent in subadult skeletons. Although focusing on adults solves that
mortality in well-documented medieval Scandinavian skeletal samples particular problem, it does so by omitting a potentially informative
(Boldsen, 1996). part of the lifespan. Second, if the researcher's interest centers on dif-
Alternatively, mortality can be expressed as a function of age ferences between the sexes in mortality experience, ignoring children
through a limited number of parameters. The five-parameter Siler makes sense because their sex cannot be reliably estimated from skel-
competing hazards model captures the main elements of human mor- etal morphology. Third, the use of such models to accommodate defi-
tality, and what it misses would be difficult to characterize with the cient skeletal data can contribute to our understanding of adult
size and quality of anthropological samples (Gage, 1988, 1989, 1991; mortality by adding a temporal dimension that extends deep into
Gage & Dyke, 1986; Gurven & Kaplan, 2007; Wood et al., 2002). The the past.
Siler model consists of three components. The first describes the haz-
ard of dying early in life, expressed as exponentially declining mortal-
ity from infancy onward. Its counterpart for older ages, the third 6.3 | Age-at-death distributions bypassing
component, is an exponentially increasing senescent component that skeletons
summarizes the increasing risks of dying that accompany advancing
age. The importance of the first and third components, deaths during The typical way to proceed is for researchers to start with individual
the immature and senescence phases of life, for describing the princi- skeletons and then combine them in some fashion, as discussed to
pal elements of normal, or attritional, mortality are immediately obvi- this point, to produce an age-at-death distribution for the archaeologi-
ous. The second component is mortality's intensity independent of cal sample. That is, skeletons come first, and the patterning of deaths
age. Here mortality rates do not change during early development or by age comes second. Recently another route has been outlined to
later senescence; that is, the third component estimates background obtain an overall age-at-death distribution, as long as one is willing to
mortality. forego the ages of individual skeletons as the initial step (Caussinus &
The hazard rate at any age is the sum of the juvenile, senescent, Corgeau, 2010; Séguy et al., 2008; Séguy et al., 2013; Séguy &
and constant components. Adult mortality, for example, is a combina- Buchet, 2013).
tion of the exponentially increasing (Gompertz) and age-independent To pursue this approach, it is first necessary to know how various
(Makeham) components. The sum of the three mortality components stages of a skeletal feature are distributed across the lifespan. These
produces a U-shaped curve that is well suited to describe nearly all probabilities are estimated from a reference sample, as has been done
naturally occurring patterns for the risk of dying. for adults using the extent of cranial suture closure (Séguy &
One advantage of the Siler model for anthropological purposes is Buchet, 2013, tabs. 5.1 and 5.2). The archaeological sample's age dis-
it is relatively robust when handling small samples where data are less tribution is then estimated by allocating skeletons to the anatomical
than ideal. The general characteristics of archaeological age-at-death structure's age-progressive stages. An appropriate prior distribution—
distributions can be captured without undue weight being placed on a preindustrial prior these researchers developed is recommended—is
the irregularities and gaps that inevitably occur in small samples com- then used in a Bayesian framework to estimate the posterior distribu-
posed of poor-quality data (Gage, 1988; Wood et al., 2002). Interest- tion. The latter is the distribution of deaths that is of interest. The
ingly, the age-independent component, of minor significance in estimation process can be more complex because the originators of
modern societies, appears to account for a large fraction of deaths in this approach advocate incorporating contextual information about
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
132 BOLDSEN ET AL.

the sample, such as population growth rates, when there is reason to The distribution of deaths as they gradually occur over time can
do so. In other words, researchers should make maximum use of avail- be explained by starting with what everyone realizes is true about a
able information, not all of which is derived directly from skeletons, in population as a whole. The force of mortality is high early in life, espe-
paleodemographic research. cially during infancy; it then decreases; and in adulthood, it increases
The difficulty with this procedure as presently outlined is not with with advancing age. The shape of the resulting age-at-death distribu-
the general approach. It nicely accommodates the inexact nature of tion is determined by the hazard rate and the number of individuals
the connection between biological change and chronological age. It entering a given age interval. Many young children, especially infants,
also avoids problems stemming from the allocation of individual skele- will be represented by skeletons because the hazard rate is high and
tons to specific age intervals, such as in the life table approach as dis- many people are at risk. During the rest of childhood, there will be
cussed above. Instead, the problem is with the skeletal trait that has fewer skeletons because both the likelihood of dying and the number
been used. For the cranial sutures, each closure category can be found at risk decline. For adults, the risk of dying increases with advancing
in almost all of the possible age intervals used in this procedure years. But a correspondingly large number of skeletons will not be
(Séguy & Buchet, 2013, tabs. 5.1 and 5.2). In addition, the work has found in the oldest age categories where the hazard is highest simply
been limited to cranial sutures, so it has yet to be shown how multiple because few people survive to that point. In bioarchaeological studies
anatomical structures can be combined to improve adult age where open-ended terminal intervals are used, the distribution of
estimates. skeletons should be U-shaped (Milner et al., 1989; Paine, 1989;
Nevertheless, this approach is perhaps the only, or at least the Paine & Harpending, 1998). It is simply an artifact of lumping all of the
best, way to extract something useful from published data on cranial oldest people into a single category, such as 50+ years.
sutures and other traditionally scored skeletal structures. Doing so is In catastrophes—events when many people died at the same time
quite important because collecting new data on existing collections is and they can be distinguished from other deaths, typically by the
costly in time and money, and many skeletal samples are no longer burial context—the pattern of mortality can differ from what would be
available for examination. There is little reason, other than long famil- expected of the usual accumulation of skeletons. Once again, the dis-
iarity, to continue using conventional age-estimation methods tribution of deaths is a function of both age-specific hazard rates and
because of their proven inability to yield satisfactory information, the number of individuals in each age category. If the risk of dying
except through the entire sample (instead of individual skeleton) was more or less independent of age, there could be a distinguishable
approach outlined by Caussinus and Corgeau (2010), Séguy signature in the mortality distribution.
et al. (2008, 2013), and Séguy and Buchet (2013). One example of catastrophic mortality is a Black Death mass
grave excavated in London (Gowland & Chamberlain, 2005; Mar-
gerison & Knüsel, 2002). The distribution of deaths by age differed
7 | I N F E R EN C E S F RO M A G E - A T - DE A T H from what was expected of the skeletons that might accumulate in a
DISTRIBUTIONS normal cemetery, and in that restricted sense it was not dissimilar
from what took place several centuries later during plague outbreaks
Once mortality profiles are generated from skeletons, there remains in Italy and France (Alfani & Bonetti, 2019; Alfani & Murphy, 2017;
the issue of what they can tell us about past groups of people. To this Dutour et al., 2003; Ell, 1989; Hollingsworth & Hollingsworth, 1971;
point, what has been described is a technique in search of a problem. Signoli et al., 2002; Tzortzis & Signoli, 2016). Excess mortality in other
We now turn to inferences, some better supported than others, that epidemics, however, is concentrated among the most vulnerable
might be drawn from age-at-death information. members of society, typically the youngest and oldest people. Histori-
cal records indicate that mortality from epidemics or famines can be
difficult to distinguish from attritional age-at-death distributions
7.1 | Attritional, catastrophic, and unknown (Dutour et al., 2003; Signoli et al., 2002). Diseases are not the only
distributions cause of mass mortality incidents that might be detected through age-
at-death distributions (Williams, 2009). For example, an essentially flat
Archaeological samples could be from an ultimately inexplicable group hazard rate from late childhood onward indicates that killing was
of dead people, an accumulation of deaths that would normally occur indiscriminate with respect to age in a late prehistoric massacre that
over many years, or the result of casualties from natural disasters, bat- took place on the northern Plains of the United States (Konigsberg &
tles, or high-mortality epidemics. For convenience, mortality patterns Frankenberg, 1994).
are often characterized as having either a catastrophic or attritional Mass casualty events, even if deaths were largely independent of
origin. That is, many people died at once, perhaps largely independent age, could be difficult to identify through mortality if the victims were
of age, in contrast to the grim reaper's whittling away at a birth cohort buried in graveyards that otherwise held an accumulation of skeletons
until nobody was left standing. The two types of age-at-death distri- spanning many generations. Their detection would depend on the
butions can range from distinctively different to indistinguishable, burial context and the duration of cemetery use relative to disaster
especially when samples are small and age indicators are poor severity, frequency, and the ages of the people most affected. In one
(Bocquet-Appel & Bacro, 2008). model, however, the effect of a high-mortality epidemic on the
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 133

cumulative age-at-death distribution could still be seen a century later

(Paine, 2000).

7.2 | Making sense of attritional age distributions

If distinguishing catastrophic from attritional samples is all that can be

done, there is little to be gained from working with skeletons. Fortu-
nately, better age estimates permit much to be said about what people
experienced in the past. To illustrate this point, we turn to four skeletal
samples with experience-based age estimates, which are more accurate
than those from other methods, including TA2 (Figure 3; Milner &
Boldsen, 2012c, fig. 14). Two sites are from the American midcontinent:
an Archaic shell midden, Read, and a late prehistoric village, Norris F I G U R E 6 Mortality profiles, starting at 5 years, are shown for
Farms. The other two are a Danish medieval village, Tirup, and a parish two Danish medieval samples, Tirup (T) and Sct. Mikkel (SM), and two
from the American midcontinent, Read Mound (RM) and Norris Farms
graveyard, Sct. Mikkel, on the outskirts of a town. Three samples were
(NF). The medieval curves are distinguishable from the American ones
examined by G.R.M., and J.L.B. collected data from the fourth, Tirup.
because of higher age-independent mortality, as estimated through a
Skeletal data were fit to a Siler model using age intervals, not point esti- Siler model
mates. Here the Siler model likelihood function was generated by maxi-
mizing the product of individual age interval probabilities. Starting at
5 years eliminates uncontrolled, indeed uncontrollable, differences distinguishing them from adults. They also accommodate the error-
among the sites in the representation of young children. prone nature of adult age estimates, complicated still further by the
The medieval Danish mortality profiles differ from the American collection of data by different researchers using dissimilar methods.
ones that, to use terms loosely, were from Mesolithic and Neolithic
societies (Figure 6). A more rapid decline with age in the medieval sam-
ples reflects higher age-independent mortality probably attributable to 7.3.1 | Mean age at death
periodic high-mortality epidemics. A major difference in disease
regimes, notably epidemics that repeatedly swept through medieval The mean age-at-death has long been part of bioarchaeological stud-
Denmark, is consistent with a higher regional population density than ies, typically when trying to see if one group or sex was better off
in the prehistoric American midcontinent, more frequent contact than the other, or to determine whether a skeletal marker of ill-health
among the inhabitants of spatially discrete settlements, and local com- was associated with a shorter lifespan (Angel, 1947; Duray, 1996;
munity participation in a wider and more diverse world, largely through Eshed & Galili, 2011, 2018; Franklin & Wilson, 2020; Goldstein, 1953;
the ship-borne trade of various commodities. The Sct. Mikkel parish Goodman & Armelagos, 1988; Grauer, 1993; Hodge, 2011; Lallo
served a poor and transient community, and these people were seem- et al., 1978; Thompson et al., 2013). But it can be hard to make sense
ingly worse off than those who lived in the countryside at Tirup. of differences among archaeological samples in the mean age-at-
In all four samples, but especially the American midcontinental death.
societies, an appreciable fraction of the adults lived beyond the age of Biased adult age estimates compromise any such figure at the
50 years. Paleodemographic studies typically include few individuals outset (Hoppa & Saunders, 1998). Underrepresented age categories
of that age. This sort of adult survival, however, would likely be found affect the mean age-at-death. Undercounting young children, a com-
more often if age estimation procedures were employed that do not mon problem in archaeological samples, increases the mean age-of-
result in a significant underestimation of the ages of old adults death, whereas under aging old people has the opposite effect.
(Figure 2). The mean age-at-death is affected by the population growth rate,
and it is more sensitive to changes in fertility than mortality
(Frankenberg & Konigsberg, 2006; Johansson & Horowitz, 1986;
7.3 | Single figure summaries of age distributions Konigsberg & Frankenberg, 1994; McCaa, 2002; Sattenspiel &
Harpending, 1983). If mortality remains fixed but fertility increases or
Several summary measures have been used to draw inferences from decreases, the mean age-at-death decreases or increases accordingly.
skeletal age distributions. The most obvious single figure, the mean As explained earlier, when there is positive growth, more children rel-
age-at-death, is flawed. Two other measures of mortality are of partic- ative to those from adults would be buried in a cemetery than if the
ular value when data are gleaned from published material and there population was stationary. The opposite is true of a declining popula-
are few, if any, controls over bone preservation, excavation practices, tion where fewer people were born in each succeeding year. As noted
age-estimation methods, and the like. They take advantage of the previously, large growth rates (i.e., those of ca. 1% or more) that
accuracy of estimating the ages of immature people and the ease of would noticeably affect small archaeological samples are unlikely in
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
134 BOLDSEN ET AL.

most long-lasting cemeteries, although they would be an issue with (Buikstra et al., 1986; Horowitz et al., 1988; Johansson &
skeletal samples that accumulated over short periods, such as those Horowitz, 1986; Konigsberg & Frankenberg, 1994; McCaa, 2002;
spanning a generation or so. Milner et al., 1989; Paine, 1989; Sattenspiel & Harpending, 1983).
Many deaths normally occur among the youngest members of
preindustrial populations where upwards of one-half of the people
7.3.2 | Age independent mortality ever born might die before reproductive maturity. Deaths among
adults, however, are liberally scattered across multiple decades
One way to obtain reasonably reliable figures from many archaeologi- according to the hazard of dying and the number of individuals at risk.
cal samples for comparative purposes is to use skeletons grouped into The effect on skeletal age distributions can be readily envisioned if
two categories: 2–5 and 5–18 years (Paine & Boldsen, 2002). Assum- mortality remains unchanged but fertility does not. In this situation,
ing skeletons were drawn at random from stationary populations, the high fertility results in a greater proportion of juvenile skeletons, espe-
ratio between mortality during the 2–5 and 5–18 years periods yields cially from the very young, than when fertility is low.
a rough estimate of the relative importance of the first and second Several measures capturing the proportion of young skeletons in
Siler mortality components; that is, immature and age-independent archaeological samples have been proposed to track what took place
mortality. The d5–18/d2–5 indicator, designated the death rate ratio in the past, notably during the transition to agricultural economies
(DRR), is sufficiently insensitive to intrinsic population growth, regard- (Barbiera et al., 2018; Barbiera & Dalla-Zuanna, 2009; Bocquet-
less of whether it is positive or negative, to approximate the demo- Appel, 2002, 2008, 2011; Bocquet-Appel et al., 2008; Bocquet-
graphic parameters of interest. Appel & Naji, 2006; Buikstra et al., 1986; Downey et al., 2014; Eshed
Archived data, mostly from central Europe, have been used to et al., 2004; Jackes & Meiklejohn, 2008; Kohler et al., 2008; Kohler &
chart changes in the DRR over thousands of years when major shifts Glaude, 2008; Kohler & Reese, 2014; McFadden & Oxenham, 2018;
took place in ways of life (Paine & Boldsen, 2002). The DRR—the rela- Paine & Harpending, 1996; Robbins, 2011). This work can be traced
tive importance of the second Siler component is what is of interest— back to Bocquet-Appel and Masset (1977, 1982) who noted the rela-
increased from the Mesolithic through early Neolithic periods, tionship between one such means of characterizing the juvenile
reaching a maximum in the Bronze and Iron Ages. It subsequently fell fraction of a skeletal sample and other demographic estimates.
with increasing urbanization during the Roman and medieval periods. Figures where juveniles are compared to other age intervals, with age
Narrowing the geographical and temporal scales of analysis to medi- ranges for both variably defined, are commonly labeled a “juvenility
eval Scandinavia shows that economic developments, which were rapid index” following Bocquet-Appel and Masset (1996, p. 580). One such
relative to those much earlier in time, were associated with changes in a measure is d5–19/d5+, to use notation similar to what was introduced
functionally similar measure, d5–20/d1–5 (Boldsen, 1996). Designated the in the previous section (Bocquet-Appel, 2002, 2008, 2011; Bocquet-
mortality rate ratio (MRR), it is based on different juvenile age ranges. Appel et al., 2008; Bocquet-Appel & Naji, 2006; Downey et al., 2014;
Skeletal samples dating to AD 1000–1670 spanned a period when sub- Kohler et al., 2008; Kohler & Glaude, 2008; Kohler & Reese, 2014).
sistence agriculture gave way to one with periodic markets and fairs, as Having a satisfactory skeletal measure, regardless of what it is
well as a system where town markets became fully incorporated into the and how it is interpreted, is only the first step. If patterning is to be
agricultural sector. Enhanced intercommunity integration was accompa- detected using such a skeletal indicator—the object of the exercise—it
nied by an increase in the MRR. When a market-based economy subse- will be a central tendency in a diffuse cloud of points. In natural-
quently became fully established—villagers produced food mainly for fertility populations, the total fertility rate (number of births through
exchange rather than household consumption—the MRR fell. menopause) is quite variable, and there is considerable overlap in the
Changes in what is interpreted as largely age-independent, thus fertility rates of hunter-gatherers and subsistence agriculturalists
unpredictable, mortality likely reflects the frequency of high-mortality (Bentley et al., 1993; Campbell & Wood, 1988; Wood, 1990). Turning
epidemic diseases, assuming many of them had the capacity to kill to archaeological concerns, compilations of numerous cemeteries will
more-or-less indiscriminately (Paine & Boldsen, 2002). By modeling feature little control over the culturally, geographically, temporally,
epidemic cycles, the highest DRR was reached when such infections and situationally varied settings that influenced who died, when they
struck populations about once a generation. That would account for did so, and who was buried in the site that was excavated. Sampling
the rise and fall of the DRR over the long run and the MRR during the biases stemming from differential bone preservation and varied exca-
much shorter medieval period. Some regular intercommunity contact vation strategies muddy the picture still further, as does high age-
increased the DRR, whereas an increase in frequency, hence epi- independent mortality.
demics at shorter intervals, lowered it. In combination, these issues underscore the need for many ceme-
teries, the more the better, to detect general trends relative to major
transitions in ways of life. One would expect figures that in aggregate
7.3.3 | Fertility track what took place during the forager to agriculturalist continuum,
or any other such change in human societies, would display a great
Since the 1980s, researchers have recognized fertility's effect on mor- range of values, even within tightly defined geographical regions.
tality profiles and the mean age-at-death as calculated from skeletons After all, local groups were constantly adapting to their immediate and
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 135

ever-changing natural and social settings, and it can be expected that consistent with the cumulative effects of epidemics that varied in fre-
age-at-death distributions would be correspondingly diverse. Short- quency. It provides a way to assess the impact of diseases, including
term cemeteries might hold skeletons that accumulated during acute crowd infections that do not leave a distinguishable signature
periods of rapid community growth or decline. In that regard, long- on bones. They were nonetheless important aspects of the disease
duration cemeteries could be advantageous. So too would combining landscape, especially during the two millennia or so leading up to the
temporally similar, geographically proximate, and culturally related modern world when densely populated urban centers developed and
samples in recognition of the unknown, uncontrolled, and ultimately stronger connections among widely distributed populations were
uninterpretable variability that exists among archaeological samples, established.
as has been done for prehistoric and historic skeletal samples
(Barbiera et al., 2018; Downey et al., 2014).
8.2 | Early to middle adulthood

8 | TH E L I F E COU R S E F R OM A G E - A T- What took place during the first several decades of adulthood can
DEATH clarify the challenges people once faced and their success in doing
so. This part of life spans the female reproductive period and, for both
Clarifying life experiences is one reason to conduct paleodemographic sexes, when people have the greatest physical capacity to contribute
analyses. For convenience, the lifespan is divided into three parts. Old to the welfare of their communities. It should come as no surprise that
adults receive the most attention here because they have been largely there have been calls to look carefully at this part of adulthood and
overlooked in studies of archaeological skeletons. what it can tell us about life experiences in the past (Boldsen &
Paine, 1995; McFadden & Oxenham, 2019). For example, variation in
what is interpreted as age-independent mortality during this segment
8.1 | Childhood of the lifespan, specifically a contrast between prehistoric and medie-
val groups, demands additional attention (Figure 6).
In recent years, bioarchaeologists have become increasingly interested Data from multiple collections suggest that from the European
in the children of the past (Halcrow & Tayles, 2011; Lewis, 2007; Mesolithic through the Middle Ages young women were not as likely
Lewis & Gowland, 2007; Mays et al., 2017; Perry, 2006, 2014). Much as men to reach middle age, defined largely by cranial suture closure
of this work is directed toward pathological bony lesions where it is (Boldsen & Paine, 1995). Because much rests on cranial sutures, it is
essential to take age into account because disease and injury frequen- not known what the attainment of middle age really means, and there
cies vary from birth to reproductive maturity. A relationship between is considerable slop in age estimates compiled from existing reports.
age and skeletal lesion frequencies means that a sample's age compo- There are few skeletal samples where one cannot be suspicious
sition influences the proportion of skeletons likely to exhibit signs of at the outset that adult age estimates are biased, perhaps badly so. Of
particular pathological processes. Frequencies for children lumped those deemed reasonably reliable, how the two sexes fared relative to
together as a single group will differ from those where the same indi- one another in early through middle adulthood presents a mixed pic-
viduals are separated into age intervals spanning only a few years ture. Relatively high mortality was found among women through their
apiece. child-bearing years in a medieval Danish village (Boldsen, 1995, 1997,
It is worth repeating what should be glaringly obvious. In mortal- 2007). That would have made a small community vulnerable to extinc-
ity samples, a complete picture of health during childhood cannot tion, as indeed happened to this village since it was eventually aban-
come from simply examining the bones of immature individuals. Doing doned when its inhabitants presumably moved to a more viable, but
so omits everyone who lived to adulthood, each of whom was once a nearby, settlement. Excess female mortality during the reproductive
child. When looking at immature skeletons, one is only examining period also occurred elsewhere in medieval Europe, but not every-
non-survivors, a badly biased sample of children who were ever alive. where (Dangvard Pedersen et al., 2020; Walter & DeWitte, 2017). In
Once again, mortality's selective effect has a big role in shaping the much different kinds of societies represented by late prehistoric vil-
skeletal samples that are the basis of inferences about the past. lages in the American Midwest, death early in adulthood did not result
From an age-at-death perspective, one difficulty is the common in one sex being consistently advantaged over the other (Milner &
underrepresentation of childhood deaths, especially of infants, in Ferrell, 2011; Wilson, 2014).
archaeological samples, often attributable to differential preservation
and cultural practices. That is indeed unfortunate because infant mor-
tality is such an important measure of community well-being in public 8.3 | Old adults
health surveys of modern societies. But as mentioned previously,
what took place in childhood from the Mesolithic to the Middle Ages For the foreseeable future, some of the most interesting results will
in Europe has been approximated by gleaning data from numerous be centered on old people. There is reason to believe more people in
osteological studies (Paine & Boldsen, 2006). The DRR indicator that the past survived to their sixth decade or beyond than would be imag-
captures deaths early in life (2–5 years) and later on (5–18 years) is ined from remarkably consistent, but likely erroneous, skeletal results.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
136 BOLDSEN ET AL.

8.3.1 | Increasing age and error? demographic models or data from small-scale societies (Blurton Jones
et al., 2002; Brewis et al., 1990; Buikstra, 1976; Chamberlain, 2006;
Pursuing efforts to estimate age in the upper part of the lifespan Hawkes & Blurton Jones, 2005; Howell, 1982; Hutchinson, 2004;
would be a fool's errand if a common assumption was true: uncer- Konigsberg & Herrmann, 2006; Lovejoy et al., 1977; Meindl
tainty about age simply increases with advancing age. It is indeed the et al., 2001; Meindl & Russell, 1998; Milner et al., 1989; Paine, 1989;
view expressed by Angel (1971, p. 71), an early leader in the use of Paine & Harpending, 1998; Palkovich, 1980, 1981). The discrepancy
skeletons in wide-ranging reconstructions of past ways of life: the has led to two quite different conclusions. First, what is seen captures
“divergence between individual physiological age and actual age in what truly took place in the past: adults did not live to old age. Sec-
years appears to increase as people grow older.” Visualized on a ond, commonly used age-estimation methods result in an underesti-
scatterplot of estimated ages plotted against known ages, ever- mation of age for people who were in their sixth decade or older
increasing error produces a cone-shaped distribution. when death occurred.
Bioarchaeologists typically use an open-ended terminal interval to The essential veracity of early adult mortality has been explicitly
accommodate increasing age-estimation error. There are essentially supported by a few researchers who recognize that paleodemographic
two approaches to move beyond such open-ended intervals. One is findings typically do not conform to expectations from recent, includ-
to distribute skeletons throughout the upper part of adulthood by ing historical, population data and model estimates (Baitzel & Gold-
using demographic models, information from roughly similar kinds of stein, 2016; Brewis et al., 1990; Lovejoy et al., 1977; Meindl
societies, or whatever seems to make sense to the investigator et al., 2001; Meindl & Russell, 1998; Mensforth, 1990;
(Buikstra, 1976; Cave & Oxenham, 2016; Greene et al., 1986; Palkovich, 1980). The use of multiple skeletal age-estimation methods
Storey, 1992). However much that improves the appearance of age- is said to minimize the likelihood of directional error, in this instance
at-death distributions, it provides little of value to understanding the underestimation of the ages of old people (Kennedy, 2002). But if
actual differences in longevity. The second approach is devilishly methods share the same biases, combining the results does not
difficult—the improvement of age estimates in the upper part of the improve outcomes (Milner et al., 2019). Usually, however, skeletal
lifespan. results are simply accepted at face value with little or no concern over
Despite being widely accepted, a progressive diminution in the how such differences in adult age-at-death distributions might have
ability to estimate age throughout adulthood is decidedly not what arisen. A dearth of old people has been attributed to differential pres-
occurs, as shown in TA2 and experience-based estimates of age ervation in notably poor soil conditions (Walker et al., 1988). Early
(Figure 3; Milner & Boldsen, 2012c, figs. 7, 8, and 14). Nor is it what is selection for “immunological competence” in small-scale societies has
seen in the ongoing development of TA3 where a broad array of skel- been suggested as the reason age-at-death distributions are so differ-
etal features are used to operationalize subjective appraisals of age ent (Lovejoy et al., 1977, p. 293). Other explanations include the many
(Figure 5). hazards faced by young hunter-gatherers (Bernardi et al., 2020;
These findings indicate the ages of the oldest people in Flensborg et al., 2015), accidents suffered by middle-aged adults in
preindustrial societies are well within our grasp. In fact, the greatest American southwestern pueblos (Palkovich, 1980), as well as generally
challenge does not appear to be the very old; it is instead with the poor conditions in medieval Italy and Japan (Baldoni et al., 2016; Gnes
people who were between their mid-40s to mid-70s when they died et al., 2018; Nagaoka et al., 2006), Iron Age Spain (Alesan et al., 1999),
(Milner & Boldsen, 2012c, figs. 7 and 8). Why that should be the case and Mesolithic to medieval Sweden (Welinder, 2001). In Andean
is not known, although selective mortality probably plays a part in Tiwanaku, old people moved, or their bones were carried, back to
it. The elderly are seemingly not as variable in bony indicators of their original homeland, a contorted way to account for their apparent
senescent aging, perhaps because they survived many years during absence (Baitzel & Goldstein, 2016).
which their cohort was progressively reduced for various reasons. Pudd'nhead Wilson's maxim—“Faith is believing what you know
Skeletally speaking, the people who died in middle age appear to have ain't so”—best sums up the situation (Twain, 1897). Bioarchaeological
been a mix of those who aged quickly and slowly. That is not a sur- findings—specifically excessive mortality during the first three
prise because physical appearance, or “perceived age,” has been decades of adulthood—are simply accepted as true. If an explanation
shown to have a mortality effect in a large twin study (Christensen is deemed necessary, bad conditions and hard lives top the list for
et al., 2009). A dampening of increases in mortality in the 80s is con- why skeletal studies indicate old people were a rarity in the past.
sistent with a selective winnowing of the frailest individuals at youn- There is admittedly not as much solid information as one might
ger ages (Gampe, 2010). like on the mortality structure of recent societies roughly similar to
the preindustrial societies that dominate bioarchaeological work. Fur-
thermore, obtaining reasonably accurate ages is a problem, as
8.3.2 | Over 50 mortality acknowledged by field researchers (Diekmann et al., 2017; Hill &
Hurtado, 1996; Howell, 1979). But what is known from groups on the
It has long been recognized that paleodemographic results—numerous forager to subsistence agriculturalist spectrum raises questions about
adults died in their 20s through 40s, with vanishingly small numbers the accuracy of age-at-death distributions generated from archaeolog-
of them surviving to more advanced ages—do not square with ical skeletons (Blurton Jones et al., 1992, 2002; Gurven et al., 2007;
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 137

Gurven & Kaplan, 2007; Headland, 1989; Hill & Hurtado, 1996; few studies where some reliance can be placed on estimates of what
Howell, 1979, 2010; Neel & Weiss, 1975; Wood, 1987). It is difficult passed for old age in preindustrial societies, it is possible that women
to brush aside the notable discrepancy between ethnographic and also outlived men. That is what occurred in one medieval Danish vil-
paleodemographic mortality estimates (Blurton Jones et al., 2002; lage and several late prehistoric communities in the American Mid-
Bocquet-Appel & Masset, 1982, 1985; Chamberlain, 2006; Hawkes & west (Boldsen, 2007; Milner & Ferrell, 2011; Wilson, 2014). But it was
Blurton Jones, 2005; Konigsberg & Herrmann, 2006; Milner not true everywhere, as shown by other medieval settlements in
et al., 1989). Denmark and the United Kingdom (Dangvard Pedersen et al., 2020;
Historical records are likewise inconsistent with paleodemographic Walter & DeWitte, 2017).
findings. Information, which varies in reliability, from classical antiq- The lack of a consistent pattern, if indeed there was one, in
uity and the medieval to early modern periods in Europe indicates archaeological samples is not unexpected. Available data are few and
people regularly lived to an old age (Batrinos, 2008; Cummins, 2017; scattered among societies where life experiences were quite different.
Floris et al., 2021; Montagu, 1994). Here old merely refers to the Distinctions between the sexes in mortality during early to middle
short lifespans typical of studies of archaeological skeletons, so it is adulthood, extending to the elderly, that correspond to various types
something in excess of 50 years. Perhaps more trustworthy, if only of preindustrial societies are worth investigating further. Doing so,
because they were closely tied to practical matters, were medieval however, requires accurate age estimates for skeletons from many
exemptions from public duties, including military service, when people culturally well-characterized contexts to identify a signal in data
reached the age of 60 or 70 (Shahar, 1993). People who had survived derived from small samples influenced by local conditions over which
to advanced years were also noted in small-scale societies encoun- researchers have little control. Once such information is available, cur-
tered during the period of European colonization, although what pre- ves such as shown in Figure 6 would be an effective way to examine
cisely was meant by reported ages is questionable (Jackes, 2000b). this issue.
For American readers, perhaps the most familiar is Powhatan who, in
the first decade of the 17th century was described by Smith (2007
[1624], p. 296) as “neare sixtie; of a very able and hardy body to 8.3.4 | The oldest old
endure any labour.”
A number of researchers have raised the possibility that what It is not known how long the oldest people might have lived in the
seems to be too many young adults could be tied to a systematic bias kinds of societies bioarchaeologists study. Fortunately, the ongoing
in aging skeletons (Chamberlain, 2006; French & Chamberlain, 2021; development of TA3 based on a wide array of skeletal characteristics
Goldstein, 1953; Howell, 1982; Jackes, 2000b; Milner et al., 2019; permits estimates for the elderly whose ages approach 100 years
Paine & Boldsen, 2006). Age-at-death data from skeletal samples (Figure 5; Milner et al., 2021). Reliably going beyond that point is a
examined using experience-based assessments as well as several problem because people of such advanced age are exceedingly rare in
quantitative procedures, including TA2 based on cranial and pelvic the modern skeletal collections that form the basis of age-estimation
features, indicate that an appreciable number of adults in the distant methods. From a paleodemographic perspective, this issue is of little
past survived beyond 50 years (Ahlström, 2015; Boldsen, 1995, 2000, practical significance. Here we naturally exclude stumbling across the
2002, 2007; Bullock, 2013; Bullock et al., 2013; Caussinus & bones of one of the long-lived people who populate the pages of Gen-
Corgeau, 2010; Dangvard Pedersen et al., 2020; DeWitte, 2015; esis, with Methuselah topping the list at 969. To judge from the late
Gowland, 2007; Milner & Ferrell, 2011; Nagaoka et al., 2012; Sasaki & 19th century to the present, it would have been exceedingly unlikely
Kondo, 2016b; Séguy et al., 2008, 2013; Séguy & Buchet, 2013; Wal- for someone in a preindustrial society to attain, let alone exceed, the
ter & DeWitte, 2017; Wilson, 2014). Much the same is indicated by age of 100 (Vaupel, 2010; Wilmoth, 2000). Even if a few people lived
canine pulp reduction with advancing age (Sasaki & Kondo, 2016b). that long, the possibility they would turn up in small archaeological
The point here is simple: when something other than conventional samples stretches credulity. The maximum lifespan in preindustrial
age-estimation methods are used, adults in preindustrial societies populations can only be estimated by modeling age-at-death well into
appear to have routinely lived beyond what most skeletal studies old age (Boldsen, 1995; Thatcher, 1999). One can then project out-
would lead us to believe. ward until there would have been vanishingly small numbers of people
likely to have been still alive.
There is still the interesting, and at this point unaddressed, ques-
8.3.3 | Longest surviving sex tion of whether a dampening of the risk of dying occurs among
the oldest of the old in preindustrial societies (Thatcher, 1999). In
It is generally true that in historical and modern societies, life expec- the modern world the hazard of dying even flattens at about
tancy for females exceeds that of males, and it even occurs in high- 110 (Gampe, 2010; Vaupel, 2010). Perhaps a deceleration of what
mortality conditions (Zarulli et al., 2018). An imbalance between the had earlier been steeply increasing mortality rates took place at a
sexes among the elderly can be readily observed in old-age-homes, somewhat younger age in populations with markedly different life
and for supercentenarians, a highly selected group of people, there experiences than in modern ones. If this reduction in the hazard of
are about nine times more females than males (Gampe, 2010). In the dying is solely a feature of the uppermost reaches of the lifespan, it
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
138 BOLDSEN ET AL.

would be of little interest to paleodemographers. But maybe it is continentally, and globally. Juxtaposing the fertility-related skeletal
linked to something else, such as more extreme selection all along. indicator with summed radiocarbon samples indicates that after an ini-
Perhaps this feature of extraordinary age was shifted to an earlier tial increase the Neolithic population in Europe leveled off, or even
point in the lifespan because all but the most physiologically robust declined (Downey et al., 2014).
people had already succumbed to any number of life-threatening con- It will not be easy to sort out what took place. Pennington's (1996)
ditions. If so, then this aspect of human mortality would be more likely models, for example, show that better child survival leads to increases
to be detected in small-scale societies, not those of the medieval in population growth, precisely what is seen during the agricultural
world with higher age-independent mortality, assuming the findings transition and beyond. In a modern group whose members are under-
shown in Figure 6 hold true. going a transition from a mobile to settled existence, there appears to
be a trade-off between fertility, childhood mortality, and disease load
(Page et al., 2016). Despite many childhood deaths, sedentary women
9 | PREINDUSTRIAL POPULATION achieve a slightly higher reproductive output, as measured by off-
CHANGE AND AGE-AT-DEATH spring surviving to adulthood, than their mobile counterparts. Skele-
tons, in fact, indicate that disease had a role to play in this long
When considering the entire world's preindustrial population, the process, as indicated by change over time in age-independent mortal-
increase in numbers from our earliest ancestors was assuredly slow, ity (Boldsen, 1996; Paine & Boldsen, 2002).
approaching zero for the vast majority of human existence. Yet the
rate of increase, when averaged across enormous expanses of land up
to the entire globe, picked up ever so slightly during the Holocene. 9.2 | Stepwise, not gradual, transition?
With regard to what took place, perhaps the most interesting result to
date from skeletal remains is a rather sudden, archaeologically speak- The transition to agricultural economies spanned millennia where it
ing, increase in the relative proportions of juveniles in archaeological developed in place through niche construction and local resource
samples during the shift to agricultural economies (Bocquet- enhancement (Smith, 2001, 2011). Just such a lengthy process was
Appel, 2011; Bocquet-Appel et al., 2008). true of eastern North America where people in the midcontinent
relied heavily on native cultigens, some domesticated, long before
maize and, later, beans were introduced and grafted onto existing sub-
9.1 | Neolithic transition sistence systems (Smith, 1989, 2006). While this shift in food acquisi-
tion occurred over several millennia, plant remains and the isotopic
Fertility indicators are generally used to identify when demographic composition of human bones indicate the principal changes occurred
regimes changed during the transition to sedentary ways of life and in steps that, archaeologically speaking, were abrupt as they spanned
more intensive food-acquisition strategies (Bocquet-Appel, 2002, only a handful of generations apiece (Milner, 2004; Smith, 1989). It
2008, 2011; Bocquet-Appel et al., 2008; Bocquet-Appel & Naji, 2006; was not a gradual process of continuous incremental changes that
Buikstra et al., 1986; Downey et al., 2014; Eshed et al., 2004; eventually gave rise to distinctly different diets and ways of life.
Jackes & Meiklejohn, 2008; Kohler et al., 2008; Kohler & Glaude, Stepwise changes in food-acquisition practices are consistent
2008; Kohler & Reese, 2014). The developing consensus is that higher with the skeletal fertility-related figures. Furthermore, it is the archae-
fertility played a role in what was a virtually imperceptible, but cumu- ological signature expected of Wood's (1998) Malthus-and-Boserup
latively important, increase in the Holocene global population. It was Ratchet model, which builds on Lee's (1986) earlier work, for popula-
related to a food supply that, in general, had greater predictability, tion increase and production intensification in the preindustrial world.
was spatially concentrated, was easily stored for lean times, and was Episodic population growth with much the same basic assumptions,
richer in high-calorie cultigens. Yields could be increased through although not laid out in as much detail, has also been modeled for
greater investments of labor and technological innovations to main- medieval Europe (Komlos & Nefedov, 2002).
tain or improve cropland. Archaeological data are accumulating that are broadly consistent
While fertility and all the rest make up a tidy package, it leaves with episodic, not gradual, Holocene population change. In that sense,
mortality out of the picture. That is partly because changes in mortality they are consistent with what would be predicted by models and seen
are harder to visualize relative to fertility in age-at-death data (Milner in the skeletal fertility indicator. Turning again to eastern North Amer-
et al., 1989; Paine, 1989; Sattenspiel & Harpending, 1983). But a ica, irregular, not steadily increasing, growth is consistent with what
method's limitations—the skeletal indicator tracks fertility, not took place from the subcontinent's initial occupation to sustained con-
mortality—should not result in an undue emphasis on births over deaths tact with Europeans a half-millennium ago. Counts of archaeological
in how one explains population trends over great periods of time. sites from several states indicate an initial population expansion, a
As noted by Bocquet-Appel (2002, 2008, 2011), a major propo- long period of hunter-gatherer stasis that encompassed mobile to
nent of the fertility-driven scenario, the initial uptick in fertility must early semi-sedentary societies, and finally another period of expansion
soon have been balanced by an increase in mortality. Otherwise, beginning when several native plants first showed signs of domestica-
populations would have skyrocketed to absurd sizes regionally, tion (Milner, 2004). Coarse chronological controls and variation in site
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 139

size and visibility mean all subtlety is lost, but the overall trend is clear: among others (Glick et al., 2020; Kuperus et al., 2020; Mader
there was not a steady increase over time in population size on a sub- et al., 2009).
continental level. Subsequent work by a different research team has For several decades, the long-term impact of ill-health early in life
produced similar results (Anderson et al., 2019). The adoption of a has been a feature of paleopathological research. Obtaining samples
semi-sedentary hunter-gatherer way of life with a correspondingly large enough to be tractable with skeletal collections that rarely num-
greater focus on resources near settlements, broadly characterized as ber more than several hundred individuals means the work must be
Mesolithic, was not in itself sufficient to yield a discernible population done with commonly occurring signs of ill health, such as develop-
increase. That came later with lurching steps toward agricultural mental enamel defects (Armelagos et al., 2009; Boldsen, 2007;
economies. DeWitte, 2018; Duray, 1996; Gamble et al., 2017; Garland, 2020;
Goodman & Armelagos, 1988; Gowland, 2015; Ham et al., 2021;
O'Donnell & Moes, 2021; Roberts & Steckel, 2019). Once again, ade-
10 | PALEOEPIDEMIOLOGY AND AGE-AT- quately addressing this issue requires good adult age estimates
DEATH because the ability to assess lost years of life is for the most part
restricted to adults. Practically speaking, late childhood or adolescent
To date, most paleopathological work incorporates little demo- effects would be difficult to measure because mortality at that age is
graphic information beyond separating adult skeletons by sex. generally low, so relatively few skeletons will be found in a cemetery
Researchers tend to shy away from age, except to classify skeletons sample.
according to lengthy intervals such as subadult and adult. These
intervals are far too coarse to provide useful information about age-
related disease processes, and that is also true of young, middle, and 11 | C O N CL U S I O N
old adult. Any value these overly long age intervals might have is
eroded further by the age-estimation biases already discussed If paleodemography is to advance to the point where it provides use-
(Buckberry, 2015). ful perspectives on the past, there is no alternative other than the
Over the past quarter-century, studies have slowly accumulated development of methods that yield accurate and unbiased age esti-
that could be labeled paleoepidemiology, a decidedly quantitative mates throughout the lifespan, and the means of using them effec-
approach to paleopathology that often features a fusion with paleo- tively. Bocquet-Appel and Masset (1985, p. 111), when reflecting on
demography (Boldsen, 1997, 2001, 2005b; Boldsen et al., 2015; their original farewell to the field, made it clear that their concern was
DeWitte, 2018; DeWitte & Hughes-Morey, 2012; DeWitte & not over the importance of “research on ancient population
Wood, 2008; Gamble et al., 2017; Godde et al., 2020; Godde & dynamics,” but simply “a refusal to continue to estimate age struc-
Hens, 2021; Kelmelis et al., 2017; Kelmelis & Dangvard tures which are only vaguely related to reality.” Their criticisms, and
Pedersen, 2019; Milner & Boldsen, 2017; Usher, 2000). To date, sev- those of others, have led to considerable work, and some progress, on
eral general approaches have been pursued to estimate the impact of the difficult problems covered in this article. Today we have a better
disease and trauma on past communities. They include a three-state appreciation of how age estimates can be improved, the questions
model developed by Usher (2000) and extended by DeWitte and that can be addressed with them, and how that should be done. Like
Wood (2008) where hazard functions describe transitions between other researchers working on the issue of obtaining reasonable age-
well, ill, and dead categories. Estimates of sensitivity and specificity at-death distributions from archaeological skeletons, we remain opti-
have been developed for bony lesions that, in turn, allow estimates of mistic (Frankenberg & Konigsberg, 2006; Séguy & Buchet, 2013).
the prevalence in mortality samples of two diseases: leprosy and The direction paleodemographers must go is clear, although the
tuberculosis (Boldsen, 2001, 2005b; Dangvard Pedersen et al., 2019). precise path to be followed, with all its twists and turns, is not. For
Of greater relevance to paleodemography are studies that explic- more accurate information on age-at-death, better skeletal traits and
itly incorporate age estimates into analyses. That has been done for ways to handle them are needed, with that work already underway
medieval to early modern Danish skeletons where male survivors of (e.g., Milner et al., 2021). Better estimates by themselves, however, do
cranial vault fractures were shown to have experienced a higher risk not tell us about the experiences of past people. To do so additional
of dying than uninjured men (Boldsen et al., 2015). The healed frac- attention must be directed toward the inferences that can be reliably
tures serve as a proxy for something that cannot be directly observed: drawn from skeletal samples. That includes what can be learned from
traumatic brain injuries that can have lingering detrimental neurologi- mortality profiles for individual samples as well as many sites in a com-
cal consequences. The potential of age distributions to reveal morta- parative framework, such as identifying what happened during the
lity's selective effect has been highlighted by differences between the transition to agricultural economies. Making the best use of existing
sexes in diffuse idiopathic skeletal hyperostosis (DISH) frequencies in data represents a formidable challenge, although there are two ways
a modern American skeletal sample (Milner et al., 2018). Women with forward. One is to use figures based on the relationship between two
early onset DISH experienced a high risk of dying in their 60s. The age intervals, with error-prone adults either omitted or grouped
proliferation of bone is associated with several conditions that affect together (e.g., Bocquet-Appel, 2002; Bocquet-Appel & Naji, 2006).
mortality, including obesity, diabetes, and cardiovascular disease, The other is to salvage the limited information from existing adult
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
140 BOLDSEN ET AL.

age-estimation methods through a procedure outlined by Séguy and Ahlström, T. (2015). Paleodemography of maritime hunter-gatherers and
Buchet (2013) and colleagues. the quest for forager baseline demography. In K. Brink, S. Hydén, K.
Jennbert, L. Larsson, & D. Olausson (Eds.), Neolithic diversities: Perspec-
Although we emphasize the osteological and analytical dimen-
tives from a conference in Lund, Sweden (Acta Archaeologica Lundensia
sions of paleodemography, the research is inherently interdisciplinary 8:65) (pp. 11–20). Department of Archaeology and Ancient History,
in nature, requiring heavy doses of cultural and environmental input. University of Lund.
In fact, this aspect of paleodemographic research was recognized as Albanese, J., Cardoso, H. F. V., & Saunders, S. R. (2005). Universal method-
ology for developing univariate sample-specific sex determination
much as a half-century ago when the work first began to gain traction
methods: An example using the epicondylar breadth of the humerus.
(Angel, 1969). Although the approaches to answering questions are Journal of Archaeological Science, 32, 143–152.
often new—better ways to estimate sex and age, to identify migrants, Alesan, A., Malgosa, A., & Simo
, C. (1999). Looking into the demography of
to monitor change over time in demographic regimes, and to estimate an iron age population in the western Mediterranean. I. Mortality.
American Journal of Physical Anthropology, 110, 285–301.
the risk of dying—the questions are mostly old ones that have long
Alfani, G., & Bonetti, M. (2019). A survival analysis of the last great
interested anthropologists, historians, demographers, and medical European plagues: The case of Nonantola (northern Italy) in 1630.
scientists alike. In short, paleodemography contributes nothing Population Studies, 1, 101–118.
unless the research questions and results are folded into appropriate Alfani, G., & Murphy, T. E. (2017). Plague and lethal epidemics in the pre-
industrial world. The Journal of Economic History, 77, 314–343.
archaeological, historical, and environmental contexts to further our
Anderson, D. G., Echeverry, D., Miller, D. S., White, A. A., Yerka, S. J.,
understanding of how and why the human experience changed
Kansa, E., Kansa, S. W., Moore, C. R., Myers, K. N., Wells, J. J.,
over time. Bissett, T. G., & Smallwood, A. M. (2019). Paleoindian settlement in
the southeastern United States. In D. K. Thulman & E. G. Garrison
ACKNOWLEDGMENTS (Eds.), New directions in the search for the first Floridians (pp. 241–347).
University Press of Florida.
This article owes much to several decades of conversation and free
Angel, J. L. (1947). The length of life in ancient Greece. Journal of Gerontol-
exchange of ideas with many colleagues. It would be impossible at this ogy, 2, 18–24.
point to name them all or to disentangle their diverse contributions to Angel, J. L. (1969). The bases of paleodemography. American Journal of
skeletal age estimation and related topics as they have developed Physical Anthropology, 30, 427–438.
Angel, J. L. (1971). The people of Lerna. American School of Classical Stud-
over the years. We would, however, like to single out several individ-
ies at Athens.
uals, with sincere apologies to those we undoubtedly missed: Jo Anthony, D. W. (1990). Migration in archaeology: The baby and the bath-
Buckberry, Lyle Konigsberg, Jutta Gampe, Sara M. Getz, Darryl Hol- water. American Anthropologist, 92, 895–915.
man, Svenja Weise, Kenneth M. Weiss, and James W. Wood. Data Arge, S., Wenzel, A., Holmstrup, P., Jensen, N. D., Lynnerup, N., &
Boldsen, J. L. (2020). Transition analysis applied to third molar develop-
included in this article were generated in part through the generous
ment in a Danish population. Forensic Science International, 308, 110145.
support of the National Institute of Justice (2014-DN-BX-K007). For Armelagos, G. J., Goodman, A. H., Harper, K. N., & Blakey, M. L. (2009).
access to collections, we thank Dawnie W. Steadman and Lee Enamel hypoplasia and early mortality: Bioarcheological support for
Meadows Jantz, the Bass Donated Collection at the University of the Barker hypothesis. Evolutionary Anthropology, 18, 261–271.
Asch, D. (1976). The Middle Woodland population of the lower Illinois valley
Tennessee; Ericka L'Abbé and Marius Loots, the Pretoria Bone Collec-
(Scientific Papers 1). Northwestern University Archeological Program.
tion at the University of Pretoria; Pasuk Mahakkanukrauh, the Chiang Aykroyd, R. G., Lucy, D., Pollard, A. M., & Roberts, C. A. (1999). Nasty,
Mai Collection at Chiang Mai University; and Maria Judite Alves and brutish, but not necessarily short: A reconsideration of the statistical
Susana Garcia, the Lisbon Collection at the Museu Nacional de His- methods used to calculate age at death from adult human skeletal and

ria Natural e da Ciência. This article was improved by comments dental age indicators. American Antiquity, 64, 55–70.
to
Aykroyd, R. G., Lucy, D., Pollard, A. M., & Solheim, T. (1997). Regression
from two anonymous reviewers. The authors have no conflicts of
analysis in adult age estimation. American Journal of Physical Anthropol-
interest to declare. ogy, 104, 259–265.
Baccino, E., & Schmitt, A. (2006). Determination of adult age at death in
DATA AVAI LAB ILITY S TATEMENT the forensic context. In A. Schmitt, E. Cunha, & J. Pinheiro (Eds.),
Forensic anthropology and medicine: Complementary sciences from recov-
Data available on request subject to privacy restrictions.
ery to cause of death (pp. 259–280). Humana Press.
Baitzel, S. I., & Goldstein, P. S. (2016). No country for old people: A
ORCID paleodemographic analysis of migration dynamics in early Andean
George R. Milner https://orcid.org/0000-0002-3330-0726 states. International Journal of Osteoarchaeology, 26, 1001–1013.
Baldoni, M., Nardi, A., Muldner, G., Lelli, R., Gnes, M., Ferraresi, F.,
Meloni, V., Cerino, P., Greco, S., Manenti, G., Angle, M., Rickards, O., &
RE FE R ENC E S Martínez-Labarga, C. (2016). Archaeo-biological reconstruction of the
Aburto, J. M., Villavicencio, F., Basellini, U., Kjærgaard, S., & Vaupel, J. W. Italian medieval population of Colonna (8th-10th centuries CE). Journal
(2020). Dynamics of life expectancy and life span equality. Proceedings of Archaeological Science: Reports, 10, 483–494.
of the National Academy of Sciences, USA, 117, 5250–5259. Barbiera, I., Castiglioni, M., & Zuanna, G. D. (2018). A synthetic measure of
Acsádi, G., & Nemeskéri, J. (1970). History of human life span and mortality. mortality using skeletal data from ancient cemeteries: The d index.
Akadémiai Kiado
. Demographic Research, 38, 2053–2072.
Adams, W. Y., Van Gerven, D. P., & Levy, R. S. (1978). The retreat from Barbiera, I., & Dalla-Zuanna, G. (2009). Population dynamics in Italy in the
migrationism. Annual Review of Anthropology, 7, 583–532. Middle Ages: New insights from archaeological findings. Population
and Development Review, 35, 367–389.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 141

Barker, D. J. P. (2004). The developmental origins of well-being. Philosophi- Bocquet-Appel, J.-P., & Masset, C. (1977). Estimateurs en pal-
cal Transactions of the Royal Society of London B, 359, 1359–1366. éodémographie. L'Homme, 17, 65–90.
Barthold Jones, J. A., Nash, U. W., Vieillefont, J., Christensen, K., Bocquet-Appel, J.-P., & Masset, C. (1982). Farewell to paleodemography.
Misevic, D., & Steiner, U. K. (2019). The AgeGuess database, an open Journal of Human Evolution, 11, 321–333.
online resource on chronological and perceived ages of people aged Bocquet-Appel, J.-P., & Masset, C. (1985). Paleodemography: Resurrection
5-100. Scientific Data, 6, 246. or ghost? Journal of Human Evolution, 14, 107–111.
Batrinos, M. L. (2008). The length of life and eugeria in classical Greece. Bocquet-Appel, J.-P., & Masset, C. (1996). Paleodemography: Expectancy
Hormones, 7, 82–83. and false hope. American Journal of Physical Anthropology, 99,
Becker, I., Woodley, S. J., & Stringer, M. D. (2010). The adult human pubic 571–583.
symphysis: A systematic review. Journal of Anatomy, 217, 475–487. Bocquet-Appel, J.-P., & Naji, S. (2006). Testing the hypothesis of a world-
Bello, S. M., Thomann, A., Signoli, M., Dutour, O., & Andrews, P. (2006). wide Neolithic demographic transition: Cooroboration from American
Age and sex bias in the reconstruction of past population structure. cemeteries. Current Anthropology, 47, 341–365.
American Journal of Physical Anthropology, 129, 24–38. Bocquet-Appel, J.-P., Naji, S., & Bandy, M. (2008). Demographic and health
Bennett, K. A. (1973). On the estimation of some demographic characteris- changes during the transition to agriculture in North America. In J.-P.
tics on a prehistoric population from the American southwest. Ameri- Bocquet-Appel (Ed.), Recent advances in palaeodemography (pp. 277–
can Journal of Physical Anthropology, 39, 223–232. 292). Springer.
Bentley, G. R., Goldberg, T., & Jasien
sak, G. (1993). The fertility of agricul- Boldsen, J. L. (1979). Liv og død I middeladerens Viborg. Resultater fra
tural and non-agricultural traditional societies. Population Studies, 47, skeletud gravningen på Sct. Mikkels Kirkegård. Museerne i Viborg Amt,
269–281. 8, 76–85.
Bentley, R. A. (2006). Strontium isotopes from the earth to the archaeolog- Boldsen, J. L. (1984). Palaeodemography of two southern Scandinavian
ical skeleton: A review. Journal of Archaeological Method and Theory, medieval communities. Meddelanden Från Lunds Universitets Historiska
13, 135–187. Museum, 1983-1984(5), 107–115.
Bermúdez de Castro, J. M., Martino
n-Torres, M., Lozano, M., Boldsen, J. L. (1988). Two methods for reconstructing the empirical mor-
Sarmiento, S., & Muela, A. (2004). Paleodemography of the Atapuerca- tality profile. Human Evolution, 3, 335–342.
Sima de los Huesos hominin sample: A revision and new approaches Boldsen, J. L. (1995). Patterns of advanced age mortality in the medieval
to the paleodemography of the European Middle Pleistocene popula- village Tirup. In B. Jeune & J. W. Vaupel (Eds.), Exceptional longevity:
tion. Journal of Anthropological Research, 60, 5–26. From prehistory to the present (Monographs on Population Aging 2)
Bernardi, L., Arrieta, M. A., & Ianniello, I. L. (2020). Preliminary study of the (pp. 37–46). Odense University Press.
age-at-death profile in a prehistoric skeletal sample from Médano Boldsen, J. L. (1996). Patterns of childhood mortality in medieval Scandina-
Peroquímica site, Argentina. Journal of Archaeological Science: Reports, via. Rivista di Antropologia, 74, 147–159.
34, 102651. Boldsen, J. L. (1997). Estimating patterns of disease and mortality in a
Bickle, P., & Fibiger, L. (2014). Ageing, childhood and social identity in the medieval Danish village. In R. R. Paine (Ed.), Integrating archaeological
early Neolithic of Central Europe. European Journal of Archaeology, 17, demography: Multidisciplinary approaches to prehistoric population
208–228. (Occasional Paper 24) (pp. 229–241). Center for Archaeological Investi-
Blakely, R. L. (1971). Comparison of the mortality profiles of Archaic, Mid- gations, Southern Illinois University.
dle Woodland, and middle Mississippian skeletal populations. American Boldsen, J. L. (2000). Human demographic evolution – Is it possible to forget
Journal of Physical Anthropology, 34, 43–54. about Darwin? (Research Report 14). Danish Center for Demographic
Blakely, R. L. (1977). Sociocultural implications of demographic data from Research, University of Southern Denmark.
Etowah, Georgia. In R. L. Blakely (Ed.), Biocultural adaptation in prehis- Boldsen, J. L. (2001). Epidemiological approach to the paleopathological
toric America (Southern Anthropological Society Proceedings 11) (pp. 45– diagnosis of leprosy. American Journal of Physical Anthropology, 115,
66). University of Georgia Press. 380–387.
Blurton Jones, N. G., Hawkes, K., & O'Connell, J. F. (2002). Antiquity of Boldsen, J. L. (2002). Demografisk struktur I landsbyen Tirup. Hikuin, 27,
postreproductive life: Are there modern impacts on hunter-gatherer 233–244.
postreproductive life spans? American Journal of Human Biology, 14, Boldsen, J. L. (2005a). Analysis of dental attrition and mortality in the
184–205. medieval village of Tirup, Denmark. American Journal of Physical
Blurton Jones, N. G., Smith, L. C., O'Connell, J. F., Hawkes, K., & Anthropology, 126, 169–176.
Kamuzora, C. L. (1992). Demography of the Hadza, an increasing and Boldsen, J. L. (2005b). Leprosy and mortality in the medieval Danish village
high density population of savanna foragers. American Journal of Physi- of Tirup. American Journal of Physical Anthropology, 126, 159–168.
cal Anthropology, 89, 159–181. Boldsen, J. L. (2007). Early childhood stress and adult age mortality – A
Bocquet-Appel, J.-P. (2002). Paleoanthropological traces of a Neolithic study of dental enamel hypoplasia in the medieval Danish village of
demographic transition. Current Anthropology, 43, 637–650. Tirup. American Journal of Physical Anthropology, 132, 59–66.
Bocquet-Appel, J.-P. (2008). Explaining the Neolithic demographic transi- Boldsen, J. L., Milner, G. R., & Boldsen, S. K. (2015). Sex estimation from
tion. In J.-P. Bocquet-Appel & O. Bar-Yosef (Eds.), The Neolithic demo- modern American humeri and femora, accounting for sample variance
graphic transition and its consequences (pp. 35–55). Springer. structure. American Journal of Physical Anthropology, 158, 745–750.
Bocquet-Appel, J.-P. (2011). The agricultural demographic transition dur- Boldsen, J. L., Milner, G. R., Konigsberg, L. W., & Wood, J. W. (2002). Tran-
ing and after the agriculture inventions. Current Anthropology, 52, sition analysis: A new method for estimating age from skeletons. In
S497–S510. R. D. Hoppa & J. W. Vaupel (Eds.), Paleodemography: Age distributions
Bocquet-Appel, J.-P., & Arsuaga, J.-L. (1999). Age distributions of hominid from skeletal samples (pp. 73–106). Cambridge University Press.
samples at Atapuerca (SH) and Krapina could indicate accumulation by Boldsen, J. L., Milner, G. R., & Weise, S. (2015). Cranial vault trauma and
catastrophe. Journal of Archaeological Science, 26, 327–338. selective mortality in medieval to early modern Denmark. Proceedings
Bocquet-Appel, J.-P., & Bacro, J.-N. (2008). Estimation of an age distribu- of the National Academy of Sciences, USA, 112, 1721–1726.
tion with its confidence intervals using an iterative Bayesian procedure Boldsen, J. L., & Paine, R. R. (1995). The evolution of human longevity from
and a bootstrap sampling approach. In J.-P. Bocquet-Appel (Ed.), the Mesolithic to the Middle Ages: An analysis based on skeletal data.
Recent advances in palaeodemography (pp. 63–82). Springer. In B. Jeune & J. W. Vaupel (Eds.), Exceptional longevity: From prehistory
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
142 BOLDSEN ET AL.

to the present (Monographs on Population Aging 2) (pp. 25–36). Odense Christensen, K., Thinggaard, M., McGue, M., Rexbye, H., Hjelmborg, J. B.,
University Press. Aviv, A., Gunn, D., van der Ouderaa, F., & Vaupel, J. W. (2009). Per-
Boylston, A., Holst, M., & Coughlan, J. (2000). Physical anthropology. In V. ceived age as clinically useful biomarker of ageing: Cohort study. Brit-
Fiorato, A. Boylston, & C. Knüsel (Eds.), Blood red roses: The archaeol- ish Medical Journal, 339, b5262.
ogy of a mass grave from the battle of Towton AD 1461 (pp. 45–59). Coale, A. J., Demeny, P., & Vaughan, B. (1983). Regional model life tables
Oxbow Books. and stable populations (2nd ed.). Academic Press.
Brewis, A. A., Molloy, M. A., & Sutton, D. G. (1990). Modeling the prehis- Cohen, M. N., & Armelagos, G. J. (Eds.). (1984). Paleopathology at the ori-
toric Maori population. American Journal of Physical Anthropology, 81, gins of agriculture. Academic Press.
343–356. Colchero, F., Rau, R., Jones, O. R., Barthold, J. A., Conde, D. A., Lenart, A.,
Brooks, S., & Suchey, J. M. (1990). Skeletal age determination based on Nemeth, L., Scheuerlein, A., Schoeley, J., Torres, C., Zarulli, V.,
the os pubis: A comparison of the Acsádi-Nemeskéri and Suchey- Altmann, J., Brockman, D. K., Bronikowski, A. M., Fedigan, L. M.,
Brooks methods. Human Evolution, 5, 227–238. Pusey, A. E., Stoinski, T. S., Strier, K. B., Baudisch, A., … Vaupel, J. W.
Brooks, S. T. (1955). Skeletal age at death: The reliability of cranial and (2016). The emergence of longevous populations. Proceedings of the
pubic age indicators. American Journal of Physical Anthropology, 13, National Academy of Sciences, USA, 113, E7681–E7690.
567–597. Crews, D. E., & Gerber, L. M. (2003). Reconstructing life history of homi-
Buckberry, J. L. (2015). The (mis)use of adult age estimates in osteology. nids and humans. Collegium Antropologicum, 27, 7–22.
Annals of Human Biology, 42, 323–331. Cummins, N. (2017). Lifespans of the European elite, 800-1800. Journal of
Buckberry, J. L., & Chamberlain, A. T. (2002). Age estimation from the Economic History, 77, 406–439.
auricular surface of the ilium: A revised method. American Journal of Curtin, P. D. (1990). The end of the “white man's grave”? Nineteenth-
Physical Anthropology, 119, 231–239. century mortality in West Africa. Journal of Interdisciplinary History, 21,
Buikstra, J. E. (1976). Hopewell in the lower Illinois valley (Scientific Papers 2). 63–88.
Northwestern University Archeological Program. Curtis, D. R., & Roosen, J. (2017). The sex-selective impact of the Black
Buikstra, J. E., Konigsberg, L. W., & Bullington, J. (1986). Fertility and the Death and recurring plagues in the southern Netherlands, 1349-1450.
development of agriculture in the prehistoric Midwest. American Antiq- American Journal of Physical Anthropology, 164, 246–259.
uity, 51, 528–546. Dangvard Pedersen, D., Milner, G. R., Kolmos, H. J., & Boldsen, J. L. (2019).
Buikstra, J. E., Milner, G. R., & Boldsen, J. L. (2006). Janaab' Pakal: The age- The association between skeletal lesions and tuberculosis diagnosis
at-death controversy re-visited. In V. Tiesler & A. Cucina (Eds.), Janaab' using a probabilistic approach. International Journal of Paleopathology,
Pakal of Palenque: Reconstructing the life and death of a Maya ruler 27, 88–100.
(pp. 48–59). University of Arizona Press. Dangvard Pedersen, D., Tarp, P., Søvsø, M., Petersen, H. C.,
Bullock, M. (2013). Pathological skeletal lesions and selective mortality in Milner, G. R., & Boldsen, J. L. (2020). A millennium of population
the Postclassic population of Cholula. Estudios de Antropologia change in pre-modern Danish Ribe. Anthropologischer Anzeiger, 77,

gica, 16, 67–94.
Biolo 13–25.
Bullock, M., Márquez, L., Hernández, P., & Ruíz, F. (2013). DeWitte, S. N. (2015). Setting the stage for medieval plague: Pre-Black
Paleodemographic age-at-death distributions of two Mexican Death trends in survival and mortality. American Journal of Physical
skeletal collections: A comparison of transition analysis and Anthropology, 158, 441–451.
traditional aging methods. American Journal of Physical Anthropology, DeWitte, S. N. (2018). Stress, sex, and plague: Patterns of developmental
152, 67–78. stress and survival in pre- and post-Black Death London. American
Burmeister, S. (2000). Archaeology and migration: Approaches to an Journal of Human Biology, 30, e23073.
archaeological proof of migration. Current Anthropology, 41, 539–567. DeWitte, S. N., Boulware, J. C., & Redfern, R. C. (2013). Medieval monastic
Campbell, K. L., & Wood, J. W. (1988). Fertility in traditional societies. In P. mortality: Hazard analysis of mortality differences between monastic
Diggory, M. Potts, & S. Teper (Eds.), Natural human fertility: Social and and nonmonastic cemeteries in England. American Journal of Physical
biological determinants (pp. 39–69). Macmillian. Anthropology, 152, 322–332.
Capasso, L., & Capasso, L. (1999). Mortality in Herculaneum before volca- DeWitte, S. N., & Hughes-Morey, G. (2012). Stature and frailty during the
nic eruption in 79 AD. Lancet, 354, 1826. Black Death: The effect of stature on risks of epidemic mortality in
Carballo, D. M., & Fortenberry, B. (2015). Bridging prehistory and history London, A.D. 1348-1350. Journal of Archaeological Science, 39, 1412–
in the archaeology of cities. Journal of Field Archaeology, 40, 542–559. 1419.
Carey, J. R., & Vaupel, J. W. (2005). Biodemography. In D. L. Poston & M. DeWitte, S. N., & Wood, J. W. (2008). Selectivity of Black Death mortality
Micklin (Eds.), Handbook of population (pp. 625–658). Kluwer with respect to preexisting health. Proceedings of the National Academy
Academic. of Sciences, USA, 105, 1436–1441.
Carrier, N. H. (1958). A note on the estimation of mortality and other pop- DeWitte, S. N., & Yaussy, S. L. (2020). Sex differences in adult famine mor-
ulation characteristics given deaths by age. Population Studies, 12, tality in medieval London. American Journal of Physical Anthropology,
149–163. 171, 164–169.
Caspari, R., & Lee, S.-H. (2004). Older age becomes common late in human Diekmann, Y., Smith, D., Gerbault, P., Dyble, M., Page, A. E.,
evolution. Proceedings of the National Academy of Sciences, 101, Chaudhary, N., Migliano, A. B., & Thomas, M. G. (2017). Accurate age
10895–10900. estimation in small-scale societies. Proceedings of the National Academy
Caussinus, H., & Corgeau, D. (2010). Estimer l'âge sans le mesurer en pal- of Sciences, USA, 114, 8205–8210.
éodémographie. Population, 65, 117–146. DiGangi, E. A., Bethard, J. D., Kimmerle, E. H., & Konigsberg, L. W. (2009).
Cave, C., & Oxenham, M. (2016). Identification of the archaeological ‘invis- A new method for estimating age-at-death from the first rib. American
ible elderly’: An approach illustrated with an Anglo-Saxon example. Journal of Physical Anthropology, 138, 164–176.
International Journal of Osteoarchaeology, 26, 163–175. Djuri
c, M., Djuri
c, K., Milovanovi
c, P., Janovi
c, A., & Milenkovi
c, P. (2011).
Chamberlain, A. (2006). Demography in archaeology. Cambridge University Representing children in excavated cemeteries: The intrinsic preserva-
Press. tion factors. Antiquity, 85, 250–262.
Christensen, K., Johnson, T. E., & Vaupel, J. W. (2006). The quest for Downey, S. S., Bocaege, E., Kerig, T., Edinborough, K., & Shennan, S.
genetic determinants of human longevity: Challenges and insights. (2014). The Neolithic demographic transition in Europe: Correlation
Nature Reviews Genetics, 7, 436–448. with juvenility index supports interpretation of the summed calibrated
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 143

radiocarbon date probability distribution (SCDPD) as a valid demo- Franklin, M., & Wilson, S. M. (2020). A bioarchaeological study of African
graphic proxy. PLoS One, 9, e105730. American health and mortality in post-emancipation U.S. south. Ameri-
DuBois, L. Z., & Shattuck-Heidorn, H. (2021). Challenging the binary: can Antiquity, 85, 652–675.
Gender/sex and the bio-logics of normalcy. American Journal of Human Frei, K. M., Mannering, U., Kristiansen, K., Allentoft, M. E., Wilson, A. S.,
Biology, 33, e23623. Skals, I., Tridico, S., Louise Nosch, M., Willerslev, E., Clarke, L., &
Duray, S. M. (1996). Dental indicators of stress and reduced age at death Frei, R. (2015). Tracing the dynamic life story of a Bronze Age female.
in prehistoric Native Americans. American Journal of Physical Anthro- Scientific Reports, 5, 10431.
pology, 99, 275–286. French, J. C. (2016). Demography and the Palaeolithic archaeological
Dutour, O., Ardagna, Y., Maczel, M., & Signoli, M. (2003). Epidemiology of record. Journal of Archaeological Method and Theory, 23, 150–199.
infectious diseases in the past: Yersin, Koch, and the skeletons. In C. L. French, J. C., & Chamberlain, A. T. (2021). Demographic uniformitarianism:
Greenblatt & M. Spigelman (Eds.), Emerging pathogens: The archaeology, The theoretical basis of prehistoric demographic research and its
ecology, and evolution of infectious disease (pp. 151–165). Oxford Uni- cross-disciplinary challenges. Philosophical Transactions of the Royal
versity Press. Society B, 376, 20190720.
Dwight, T. (1890). The closure of the cranial sutures as a sign of age. Bos- French, J. C., Riris, P., Fernandéz-Lo
pez de Pablo, J., Lozano, S., & Silva, F.
ton Medical and Surgical Journal, 122, 389–392. (2021). A manifesto for palaeodemography in the twenty-first century.
Ekamper, P., Bijwaard, G., van Poppel, F., & Lumey, L. H. (2017). War- Philosophical Transactions of the Royal Society B, 376, 20190707.
related excess mortality in the Netherlands, 1944-45: New estimates Gage, T. B. (1988). Mathematical hazard models of mortality: An alterna-
of famine- and non-famine-related deaths from national death records. tive to model life tables. American Journal of Physical Anthropology, 76,
Historical Methods, 50, 113–128. 429–441.
Ell, S. R. (1989). Three days in October of 1630: Detailed examination of Gage, T. B. (1989). Bio-mathematical approaches to the study of human
mortality during an early modern plague epidemic in Venice. Reviews variation in mortality. Yearbook of Physical Anthropology, 32, 185–214.
of Infectious Diseases, 11, 128–139. Gage, T. B. (1991). Causes of death and the components of mortality:
Eshed, V., & Galili, E. (2011). Palaeodemography of southern Testing the biological interpretations of a competing hazard model.
Levantine pre-pottery Neolithic populations: Regional and American Journal of Human Biology, 3, 289–300.
temporal perspectives. In R. Pinhasi & J. T. Stock (Eds.), Human Gage, T. B., & Dyke, B. (1986). Parameterizing abridged mortality tables:
bioarchaeology of the transition to agriculture (pp. 403–428). Wiley- The Siler three-component hazard model. Human Biology, 58,
Blackwell. 275–291.
Eshed, V., & Galili, E. (2018). Agriculture and life style: A paleodemography Gamble, J. A., Boldsen, J. L., & Hoppa, R. D. (2017). Stressing out in medie-
of pottery Neolithic (8500-6500 cal. BP) farming populations in the val Denmark: An investigation of dental enamel defects and age at
southern Levant. Paléorient, 44, 93–112. death in two medieval Danish cemeteries. International Journal of
Eshed, V., Gopher, A., Gage, T. B., & Hershkovitz, I. (2004). Has the transi- Paleopathology, 17, 52–66.
tion to agriculture reshaped the demographic structure of prehistoric Gampe, J. (2010). Human mortality beyond age 110. In H. Maier, J. Gampe,
populations? Evidence from the Levant. American Journal of Physical B. Jeune, J.-M. Robine, & J. W. Vaupel (Eds.), Supercentenarians
Anthropology, 1234, 315–329. (pp. 219–230). Springer.
Falys, C. G., & Lewis, M. E. (2011). Proposing a way forward: A review of Gampe, J., Milner, G. R., Boldsen, J. L., & Weise, S. (2012) New tools for
standardisation in the use of age categories and ageing techniques in ancient populations: Estimating age-at-death distributions from skele-
osteological analysis (2004-2009). International Journal of tal remains by calibrated expert inference. Paper presented at the Pop-
Osteoarchaeology, 21, 704–716. ulation Association of America. San Francisco, CA.
Fieuws, S., Willems, G., Larsen-Tangmose, S., Lynnerup, N., Boldsen, J., & Garland, C. J. (2020). Implications of accumulative stress burdens during
Thevissen, P. (2016). Obtaining appropriate interval estimates for age critical periods of early postnatal life for mortality risk among Guale
when multiple indicators are used: Evaluation of an ad-hoc procedure. interred in a colonial era cemetery in Spanish Florida (ca. AD
International Journal of Legal Medicine, 30, 489–499. 1605-1680). American Journal of Physical Anthropology, 172, 621–637.
Finch, C. E., & Haghani, A. (2021). Gene-environment interaction and sto- Garofalo, E. M., & Garvin, H. M. (2020). The confusion between biological
chastic variations in the Gero-exposome. Journals of Gerontology: Bio- sex and gender and potential implications of misinterpretations. In
logical Sciences and Medical Sciences, 76, 1740–1747. A. R. Klales (Ed.), Sex estimation of the human skeleton (pp. 35–52). Aca-
Flensborg, G., Martínez, G., & Bayala, P. D. (2015). Mortality profiles of demic Press.
hunter-gatherer societies: A case study from the eastern Pampa- Garvin, H. M., & Passalacqua, N. V. (2012). Current practices by forensic
Patagonia transition (Argentina) during the final late Holocene. Interna- anthropologists in adult skeletal age estimation. Journal of Forensic Sci-
tional Journal of Osteoarchaeology, 25, 816–826. ences, 57, 427–433.
Floris, P., Dore, M. P., & Pes, G. M. (2021). Does the longevity of the Sar- Garvin, H. M., Passalacqua, N. V., Uhl, N. M., Gipson, D. R.,
dinian population date back to Roman times? A comprehensive review Overbury, R. S., & Cabo, L. L. (2012). Developments in forensic anthro-
of the available evidence. PLoS One, 16, e0245006. pology: Age-at-death estimation. In D. C. Dirkmaat (Ed.), A companion
Fojas, C. L., Kim, J., Minsky-Rowland, J. D., & Algee-Hewitt, B. F. B. (2018). to forensic anthropology (pp. 202–238). Wiley-Blackwell.
Testing inter-observer reliability of the transition analysis aging Glick, K., Novofastovski, I., Schwartz, N., & Mader, R. (2020). Cardiovascu-
method on the William M. Bass forensic skeletal collection. American lar disease in diffuse idiopathic skeletal hyperostosis (DISH): From the-
Journal of Physical Anthropology, 165, 183–193. ory to reality – A 10-year follow-up study. Arthritis Research and
Frankenberg, S. R. (2021). Testing for differences in senescence using Therapy, 11, 190.
score data to understand the effects of reference sample choices. In Gluckman, P. D., & Hanson, M. A. (2004). Living with the past: Evolution,
B. F. B. Algee-Hewitt & J. Kim (Eds.), Remodeling forensic skeletal age development, and patterns of disease. Science, 305, 1733–1736.
(pp. 27–46). Academic Press. Gnes, M., Baldoni, M., Marchetti, L., Basoli, F., Leonardi, D., Canini, A.,
Frankenberg, S. R., & Konigsberg, L. W. (2006). A brief history of paleo- Licoccia, S., Enei, F., Rickards, O., & Martínez-Labarga, C. (2018). Bio-
demography from Hooton to Hazards analysis. In J. E. Buikstra & L. A. archaeological approach to the study of the medieval population of
Beck (Eds.), Bioarchaeology: The contextual analysis of human remains Santa Severa (Rome, 7th-15th centuries). Journal of Archaeological Sci-
(pp. 227–261). Academic Press. ence: Reports, 18, 11–25.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
144 BOLDSEN ET AL.

Godde, K., & Hens, S. M. (2012). Age-at-death estimation in an Italian his- survival in the ancestral Pueblo southwest. American Journal of Human
torical sample: A test of the Suchey-Brooks and transition analysis Biology, 33, e23506.
methods. American Journal of Physical Anthropology, 149, 259–265. Hassan, F. A. (1981). Demographic archaeology. Academic Press.
Godde, K., & Hens, S. M. (2015). Modeling senescence changes of the Hawkes, K. (2006). Slow life histories and human evolution. In K. Hawkes &
pubic symphysis in historic Italian populations: A comparison of the R. R. Paine (Eds.), The evolution of human life history (pp. 95–16). School
Rostock and forensic approaches to aging using transition analysis. of American Research Press.
American Journal of Physical Anthropology, 156, 466–473. Hawkes, K., & Blurton Jones, N. (2005). Human age structures, paleo-
Godde, K., & Hens, S. M. (2021). An epidemiological approach to the analy- demography, and the grandmother hypothesis. In E. Voland, A. Chas-
sis of cribra orbitalia as an indicator of health status and mortality in iotis, & W. Schiefenhovel (Eds.), The evolutionary significance of the
medieval and post-medieval London under a model of parasitic infec- second half of female life (pp. 118–140). Rutgers University Press.
tion. American Journal of Physical Anthropology, 174, 631–645. Hawkes, K., O'Connell, J. F., Blurton Jones, N. G., Alvarez, H., &
Godde, K., Pasillas, V., & Sanchez, A. (2020). Survival analysis of the Black Charnov, E. L. (1998). Grandmothering, menopause, and the evolution
Death: Social inequality of women and the perils of life and death in of human life histories. Proceedings of the National Academy of Sciences
medieval London. American Journal of Physical Anthropology, 173, USA, 95, 1336–1339.
168–178. Headland, T. N. (1989). Population decline in a Philippine Negrito hunter-
Goldstein, M. S. (1953). Some vital statistics based on skeletal material. gatherer society. American Journal of Human Biology, 1, 59–72.
Human Biology, 25, 3–12. Hens, S. M., & Godde, K. (2020). New approaches to age estimation using
Goodenough, W. H. (1965). Rethinking “status” and ‘role’ toward a gen- palatal suture fusion. Journal of Forensic Sciences, 65, 1406–1415.
eral model of the cultural organization of social relationships. In M. Hens, S. M., Rastelli, E., & Belcastro, G. (2008). Age estimation from the
Banton (Ed.), The relevance of models for social anthropology (ASA Mono- human os coxa: A test on a documented Italian collection. Journal of
graphs 1) (pp. 1–24). Tavistock. Forensic Sciences, 53, 1040–1043.
Goodman, A. H., & Armelagos, G. J. (1988). Childhood stress and Herrmann, N. P., & Konigsberg, L. W. (2002). A re-examination of the age-
decreased longevity in a prehistoric population. American Anthropolo- at-death distribution of Indian knoll. In R. D. Hoppa & J. W. Vaupel
gist, 90, 936–944. (Eds.), Paleodemography: Age distributions from skeletal samples
Gordon, C. C., & Buikstra, J. E. (1981). Soil pH, bone preservation, and (pp. 243–257). Cambridge University Press.
sampling bias at mortuary sites. American Antiquity, 46, 566–571. Herskind, A. M., McGue, M., Holm, N. V., Sørensen, T. I. A., Harvald, B., &
Gowland, R. L. (2006). Ageing the past: Examining age identity from funer- Vaupel, J. W. (1996). The heritability of human longevity: A
ary evidence. In R. L. Gowland & C. Knüsel (Eds.), Social archaeology of population-based study of 2872 Danish twin pairs born 1870-1900.
funerary remains (pp. 143–154). Oxbow Books. Human Genetics, 97, 319–323.
Gowland, R. L. (2007). Age, ageism and osteological bias: The evidence Hill, K., & Hurtado, A. M. (1996). Ache life history: The ecology and demogra-
from late Roman Britain. Journal of Roman Archaeology, 65, 153–169. phy of a foraging people. Aldine de Gruyter.
Gowland, R. L. (2015). Entangled lives: Implications of the developmental Hin, S. (2013). The demography of Roman Italy. Cambridge University Press.
origins of health and disease hypothesis for bioarchaeology and the Hobbes, T. (1904 [1651]). Leviathan or the matter, forme and power of a
life course. American Journal of Physical Anthropology, 158, 530–540. common wealth ecclesiasticall and civil. Cambridge University Press.
Gowland, R. L., & Chamberlain, A. T. (2005). Detecting plague: Hodge, S. C. (2011). Population dispersal and human health at Moundville.
Palaeodemographic characterisation of a catastrophic death assem- Southeastern Archaeology, 30, 226–241.
blage. Antiquity, 79, 146–157. Hoffman, D. J., Reynolds, R. M., & Hardy, D. B. (2017). Developmental ori-
Grauer, A. L. (1993). Patterns of anemia and infection from medieval York, gins of health and disease: Current knowledge and potential mecha-
England. American Journal of Physical Anthropology, 91, 203–213. nisms. Nutrition Reviews, 75, 951–970.
Green, S., Green, S., & Armelagos, G. J. (1974). Settlement and mortality of Hollingsworth, M. F., & Hollingsworth, T. H. (1971). Plague mortality rates
the Christian site (1050 A.D.-1300 A.D.) of Meinarti (Soudan). Journal by age and sex in the parish of St. Botolph's without Bishopsgate,
of Human Evolution, 3, 297–316. London, 1603. Population Studies, 25, 131–146.
Greene, D. L., Van Gerven, D. P., & Armelagos, G. J. (1986). Life and death Holst, M. K., Heinemeier, J., Hertz, E., Jensen, P., Løvschal, M.,
in ancient populations: Bones of contention in paleodemography. Mollerup, L., Odgaard, B. V., Olsen, J., Søe, N. E., & Kristiansen, S. M.
Human Evolution, 3, 193–207. (2018). Direct evidence of a large northern European Roman period
Gregoricka, L. A. (2021). Moving forward: A bioarchaeology of mobility martial event and postbattle corpse manipulation. Proceedings of the
and migration. Journal of Archaeological Research, 29, 581–635. National Academy of Sciences, USA, 115, 5920–5925.
Gurven, M., & Kaplan, H. (2007). Longevity among hunter-gatherers: A Hooton, E. A. (1920). Indian village site and cemetery near Madisonville Ohio
cross-cultural examination. Population and Development Review, 33, (Papers 8.1). Peabody Museum of American Archaeology and Ethnol-
321–365. ogy, Harvard University.
Gurven, M., Kaplan, H., & Supa, A. Z. (2007). Mortality experience of Hooton, E. A. (1930). The Indians of Pecos pueblo: A study of their skeletal
Tsimane Amerindians of Bolivia: Regional variation and temporal remains. Yale University Press.
trends. American Journal of Human Biology, 19, 376–398. Hopkins, J. (1987). Graveyards for historians. In F. Hinard (Ed.), La mort, les
Guy, H., Masset, C., & Baud, C. A. (1997). Infant taphonomy. International morts et l'au-delà dans le monde Romain (Actes du colloque de Caen 20–
Journal of Osteoarchaeology, 7, 221–229. 22 Novembre 1985) (pp. 113–126). Université de Caen.
Halcrow, S. E., & Tayles, N. (2011). The bioarchaeological investigation of Hoppa, R. D. (1999). Modeling the effects of selection bias on
children and childhood. In S. C. Agarwal & B. A. Glencross (Eds.), Social palaeodemographic analyses. Homo, 50, 228–243.
bioarchaeology (pp. 333–360). Wiley-Blackwell. Hoppa, R. D. (2002). Paleodemography: Looking back and thinking
Halley, E. (1693). An estimate of the degrees of the mortality of mankind; ahead. In R. D. Hoppa & J. W. Vaupel (Eds.), Paleodemography: Age
drawn from curious tables of the births and funerals at the city of distributions from skeletal samples (pp. 9–28). Cambridge University
Breslaw; with an attempt to ascertain the price of annuities upon lives. Press.
Philosophical Transactions of the Royal Society of London, 17, 596–610. Hoppa, R. D., & Saunders, S. (1998). The MAD legacy: How meaningful is
Ham, A. C., Temple, D. H., Klaus, H. D., & Hunt, D. R. (2021). Evaluating life mean age-at-death in skeletal samples. Human Evolution, 13, 1–14.
history trade-offs through the presence of linear enamel hypoplasia at Hoppa, R. D., & Vaupel, J. W. (Eds.). (2002a). Paleodemography: Age distri-
Pueblo Bonito an Hawikku: A biocultural study of early life stress and butions from skeletal samples. Cambridge University Press.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 145

Hoppa, R. D., & Vaupel, J. W. (2002b). The Rostock manifesto for paleo- Katzenberg, M. A., & Waters-Rist, A. L. (2019). Stable isotope analysis: A
demography: The way from stage to age. In R. D. Hoppa & J. W. tool for studying past diet, demography, and life history. In M. A.
Vaupel (Eds.), Paleodemography: Age distributions from skeletal samples Katzenberg & A. Grauer (Eds.), Biological anthropology of the human
(pp. 1–8). Cambridge University Press. skeleton (3rd ed., pp. 469–504). Wiley-Blackwell.
Horowitz, S., Armelagos, G., & Wachter, K. (1988). On generating birth Kelmelis, K. S., & Dangvard Pedersen, D. (2019). Impact of urbanization on
rates from skeletal populations. American Journal of Physical Anthropol- tuberculosis and leprosy prevalence in medieval Denmark. Anthro-
ogy, 76, 189–196. pologischer Anzeiger, 76, 149–166.
Horvath, S. (2013). DNA methylation age of human tissues and cell types. Kelmelis, K. S., & DeWitte, S. N. (2021). Urban and rural survivorship in
Genome Biology, 14, R115. pre- and post-Black Death Denmark. Journal of Archaeological Science:
Hosek, L., & Robb, J. (2019). Osteobiography: A platform for bio- Reports, 38, 103089.
archaeological research. Bioarchaeology International, 1, 1–15. Kelmelis, K. S., Price, M. H., & Wood, J. W. (2017). The effect of leprotic
Howell, N. (1976). Toward a uniformitarian theory of human paleo- infection on the risk of death in medieval rural Denmark. American
demography. Journal of Human Evolution, 5, 25–40. Journal of Physical Anthropology, 164, 763–775.
Howell, N. (1979). Demography of the Dobe !Kung. Academic Press. Kennedy, G. E. (2002). Palaeolithic grandmothers? Life history theory and
Howell, N. (1982). Village composition implied by a paleodemographic life early Homo. Journal of the Royal Anthropological Institute, 9, 549–572.
table: The Libben site. American Journal of Physical Anthropology, 59, Kieffer-Olsen, J., Boldsen, J. L., & Pentz, P. (1986). En nyfunden kirke ved
263–269. Bygholm. Vejle Amts Årbog, 1986, 24–51.
Howell, N. (2010). Life histories of the Dobe !Kung. University of California Klales, A. R. (2020). Sex estimation using pelvis morphology. In A. R. Klales
Press. (Ed.), Sex estimation of the human skeleton (pp. 75–93). Academic
Hughes-Morey, G. (2016). Interpreting adult stature in industrial London. Press.
American Journal of Physical Anthropology, 159, 126–134. Klales, A. R., Ousley, S. D., & Passalacqua, N. V. (2020). Statistical
Hutchinson, D. L. (2004). Bioarchaeology of the Florida gulf coast: Adapta- approaches to sex estimation. In A. R. Klales (Ed.), Sex estimation of the
tion, conflict, and change. University Press of Florida. human skeleton (pp. 203–217). Academic Press.
Igarashi, Y., Uesu, K., Wakebe, T., & Kanazawa, E. (2005). New method for Knudson, K. J., & Price, T. D. (2007). Utility of multiple chemical techniques
estimation of adult skeletal age at death from the morphology of the in archaeological residential mobility studies: Case studies from
auricular surface of the ilium. American Journal of Physical Anthropol- Tiwanaku- and Chiribaya-affiliated sites n the Andes. American Journal
ogy, 128, 324–339. of Physical Anthropology, 132, 25–39.
Jackes, M. (2000a). Building the bases for paleodemographic analysis: Knudson, K. J., & Stojanowski, C. M. (2008). New directions in bio-
Adult age determination. In M. A. Katzenberg & S. R. Saunders (Eds.), archaeology: Recent contributions to the study of human social identi-
Biological anthropology of the human skeleton (pp. 417–466). Wiley- ties. Journal of Archaeological Research, 16, 397–432.
Liss. Knüsel, C. J., & Robb, J. (2016). Funerary taphonomy: An overview of goals
Jackes, M. (2000b). Ethnohistory and osteology in southern Ontario. In M. and methods. Journal of Archaeological Science: Reports, 10, 655–673.
Boyd, J. C. Erwin, & M. Hendrickson (Eds.), The entangled past: Inte- Kohler, T. A., & Glaude, M. (2008). The nature and timing of the Neolithic
grating history and archaeology (30th Chacmool Conference) (pp. 164– demographic transition in the North American southwest. In J.-P.
174). University of Calgary. Bocquet-Appel & O. Bar-Yosef (Eds.), The Neolithic demographic transi-
Jackes, M. (2011). Representativeness and bias in archaeological skeletal tion and its consequences (pp. 81–105). Springer.
samples. In S. C. Agarwal & B. A. Glencross (Eds.), Social bioarchaeology Kohler, T. A., Glaude, M. P., Bocquet-Appel, J.-P., & Kemp, B. M. (2008).
(pp. 107–146). Wiley-Blackwell. The Neolithic demographic transition in the U.S. southwest. American
Jackes, M., & Meiklejohn, C. (2008). The paleodemography of central Por- Antiquity, 73, 645–669.
tugal and the Mesolithic-Neolithic transition. In J.-P. Bocquet-Appel Kohler, T. A., & Reese, K. M. (2014). Long and spatially variable Neolithic
(Ed.), Recent advances in palaeodemography (pp. 209–258). Springer. demographic transition in the North American southwest. Proceedings
Jantz, R. L., & Ousley, S. D. (2005). FORDISC 3.1 personal computer forensic of the National Academy of Sciences, USA, 111, 10101–10106.
discriminant functions. Forensic Anthropology Center, University of Komlos, J., & Nefedov, S. (2002). A compact micromodel of pre-industrial
Tennessee. https://fac.utk.edu/fordisc-3-1-personal-computer- population growth. Historical Methods, 35, 92–94.
forensic-discriminant-functions Konigsberg, L. W., & Frankenberg, S. R. (1992). Estimation of age structure
Jantz, R. L., & Ousley, S. D. (2020). Sexual dimorphism variation in Fordisc in anthropological demography. American Journal of Physical Anthropol-
samples. In A. R. Klales (Ed.), Sex estimation of the human skeleton ogy, 89, 235–256.
(pp. 185–200). Academic Press. Konigsberg, L. W., & Frankenberg, S. R. (1994). Paleodemography: “Not
Jefferson, T. (1788). Notes on the state of Virginia. Prichard and Hall. quite dead”. Evolutionary Anthropology, 3, 92–105.
Johansson, S. R., & Horowitz, S. (1986). Estimating mortality in skeletal Konigsberg, L. W., & Frankenberg, S. R. (2002). Deconstructing death in pal-
populations: Influence of the growth rate on the interpretation of eodemography. American Journal of Physical Anthropology, 117, 297–309.
levels and trends during the transition to agriculture. American Journal Konigsberg, L. W., & Frankenberg, S. R. (2013). Bayes in biological anthro-
of Physical Anthropology, 71, 233–250. pology. American Journal of Physical Anthropology, 57, 153–184.
Johnston, R. E., & Snow, C. E. (1961). The reassessment of the age and sex Konigsberg, L. W., Frankenberg, S. R., & Liversidge, H. M. (2016). Optimal
of the Indian Knoll skeletal population: Demographic and methodolog- trait scoring for age estimation. American Journal of Physical Anthropol-
ical aspects. American Journal of Physical Anthropology, 19, 237–244. ogy, 159, 557–576.
Jooste, N., L'Abbé, E. N., Pretorius, S., & Steyn, M. (2016). Validation of tran- Konigsberg, L. W., Frankenberg, S. R., & Liversidge, H. M. (2019). Status of
sition analysis as a method of adult age estimation in a modern South mandibular third molar development as evidence in legal age threshold
African sample. Forensic Science International, 266, 580.e1–580.e7. cases. Journal of Forensic Sciences, 64, 680–697.
Kamnikar, K. R., Herrmann, N. P., & Plemons, A. M. (2018). New Konigsberg, L. W., Frankenberg, S. R., & Walker, R. B. (1997). Regress what
approaches to juvenile age estimation in forensics: Application of tran- on what? Paleodemographic age estimation as a calibration problem.
sition analysis via the Shackelford et al. method to a diverse modern In R. R. Paine (Ed.), Integrating archaeological demography: Multi-
subadult sample. Human Biology, 90, 11–30. disciplinary approaches to prehistoric population (Occasional Paper 24)
Katz, D., & Suchey, J. M. (1986). Age determination of the male os pubis. (pp. 64–88). Center for Archaeological Investigations, Southern Illinois
American Journal of Physical Anthropology, 69, 427–435. University.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
146 BOLDSEN ET AL.

Konigsberg, L. W., & Herrmann, N. P. (2002). Markov chain Monte Carlo Sea Scythians: An exploration by using Bayesian methods. American
estimation of hazard model parameters in paleodemography. In R. D. Journal of Physical Anthropology, 174, 595–613.
Hoppa & J. W. Vaupel (Eds.), Paleodemography: Age distributions from Lynnerup, N., & Jacobsen, J. C. B. (2003). Age and fractal dimensions of
skeletal samples (pp. 222–242). Cambridge University Press. human sagittal and coronal sutures. American Journal of Physical
Konigsberg, L. W., & Herrmann, N. P. (2006). The osteological evidence for Anthropology, 121, 332–336.
human longevity in the recent past. In K. Hawkes & R. R. Paine (Eds.), Maaranen, N., & Buckberry, J. (2018). Exploring age – Transition analysis as a
The evolution of human life history (pp. 267–306). School of American tool for detecting the elderly. In G. Lillehammer & E. Murphy (Eds.),
Research Press. Across the generations: The old and the young in past societies (AmS-Skrifter
Konigsberg, L. W., Herrmann, N. P., Wescott, D. J., & Kimmerle, E. H. 26) (pp. 143–154). Museum of Archaeology, University of Stavanger.
(2008). Estimation and evidence in forensic anthropology: Age-at- MacDonnell, W. R. (1913). On the expectation of life in ancient Rome, and
death. Journal of Forensic Sciences, 53, 541–557. in the provinces of Hispania and Lusitania, and Africa. Biometrika, 9,
Kotěrová, A., Velemínská, J., Cunha, E., & Brůžek, J. (2018). A validation 366–380.
study of the Stoyanova et al. method (2017) for age-at-death estima- Mader, R., Novofestovski, I., Adawi, M., & Lavi, I. (2009). Metabolic syn-
tion quantifying the 3D pubic symphyseal surface of adult males of drome and cardiovascular risk in patients with diffuse idiopathic skele-
European populations. International Journal of Legal Medicine, 133, tal hyperostosis. Seminars in Arthritis and Rheumatism, 38, 361–365.
603–612. Manifold, B. M. (2015). Skeletal preservation of children's remains in the
Kumagai, A., Willems, G., Franco, A., & Thevissen, P. (2018). Age estima- archaeological record. Homo, 66, 520–548.
tion combining radiographic information of two dental and four skele- Mann, R. W., Jantz, R. L., Bass, W. M., & Willey, P. S. (1991). Maxillary
tal predictors in children and subadults. International Journal of Legal suture obliteration: A visual method for estimating skeletal age. Journal
Medicine, 132, 1769–1777. of Forensic Sciences, 36, 781–791.
Kuperus, J. S., Hoesein, F. A. A. M., de Jong, P. A., & Verlaan, J. J. (2020). Mann, R. W., Symes, S. A., & Bass, W. M. (1987). Maxillary suture oblitera-
Diffuse idiopathic skeletal hyperostosis: Etiology and clinical rele- tion: Aging the human skeleton based on intact or fragmentary maxilla.
vance. Best Practice and Research Clinical Rheumatology, 34, 101527. Journal of Forensic Sciences, 32, 148–157.
Kyle, B., Reitsema, L. J., Tyler, J., & Fabbri, P. F. (2018). Examining the oste- Margerison, B. J., & Knüsel, C. J. (2002). Paleodemographic comparison of
ological paradox: Skeletal stress in mass graves versus civilians at the a catastrophic and an attritional death assemblage. American Journal of
Greek colony of Himera (Sicily). American Journal of Physical Anthropol- Physical Anthropology, 119, 134–143.
ogy, 167, 161–172. Marioni, R. E., Shah, S., McRae, A. F., Chen, B. H., Colicino, E., Harris, S. E.,
Lallo, J., Armelagos, G. J., & Rose, J. C. (1978). Paleoepidemiology of infectious Gibson, J., Henders, A. K., Redmond, P., Cox, S. R., Pattie, A., Corley, J.,
disease in the Dickson Mounds population. MCV Quarterly, 14, 17–23. Murphy, L., Martin, N. G., Montgomery, G. W., Feinberg, A. P.,
Larsen, C. S. (2015). Bioarchaeology: Interpreting behavior from the human Fallin, M. D., Multhaup, M. L., Jaffe, A. E., … Deary, I. J. (2015). DNA
skeleton (2nd ed.). Cambridge University Press. methylation age of blood predicts all-cause mortality in later life.
Larsen, C. S., Knüsel, C. J., Haddow, S. D., Pilloud, M. A., Milella, M., Genome Biology, 16, 25.
Sadvari, J. W., Pearson, J., Ruff, C. B., Garofalo, E. M., Bocaege, E., Martrille, L., Ubelaker, D. H., Cattaneo, C., Seguret, F., Tremblay, M., &
Betz, B. J., Dori, I., & Glencross, B. (2019). Bioarchaeology of Neolithic Baccino, E. (2007). Comparison of four skeletal methods for the esti-
Çatalhöyük revals fundamental transitions in health, mobility, and life- mation of age at death on white and black adults. Journal of Forensic
style in early farmers. Proceedings of the National Academy of Sciences, Sciences, 52, 302–307.
USA, 116, 12615–12623. Masset, C. (1976). Sur la mortalité chez les anciens Indiens de l'Illinois. Cur-
Lee, R. D. (1986). Malthus and Boserup: A dynamic synthesis. In D. Col- rent Anthropology, 17, 128–132.
man & R. Schofield (Eds.), The state of population theory: Forward from Mays, S., Gowland, R., Halcrow, S., & Murphy, E. (2017). Child bio-
Malthus (pp. 96–130). Basil Blackwell. archaeology: Perspectives on the past 10 years. Childhood in the Past,
Lewis, M. E. (2007). The bioarchaeology of children: Perspectives from biolog- 10, 38–56.
ical and forensic anthropology. Cambridge University Press. McCaa, R. (2002). Paleodemography of the Americas from ancient times
Lewis, M. E., & Gowland, R. (2007). Brief and precarious lives: Infant mor- to colonialism and beyond. In R. H. Steckel & J. C. Rose (Eds.), The
tality in contrasting sites from medieval and post-medieval England backbone of history: Health and nutrition in the Western Hemisphere
(AD 850-1859). American Journal of Physical Anthropology, 134, (pp. 94–124). Cambridge University Press.
117–129. McFadden, C., & Oxenham, M. F. (2018). The D0-14/D ratio: A new
Loe, L., Boyle, A., Webb, H., & Score, D. (2014). ‘Given to the ground’: A paleodemographic index and equation for estimating total fertility
Viking age massgrave on Ridgeway Hill, Weymouth (Monograph 22). Dor- rates. American Journal of Physical Anthropology, 165, 471–479.
set Natural History and Archaeological Society. McFadden, C., & Oxenham, M. F. (2019). The paleodemographic measure
Love, B., & Müller, H. G. (2002). A solution to the problem of obtaining a of maternal mortality and a multifaceted approach to maternal health.
mortality schedule for paleodemographic data. In R. D. Hoppa & J. W. Current Anthropology, 60, 141–146.
Vaupel (Eds.), Paleodemography: Age distributions from skeletal samples McKern, T. W., & Stewart, T. D. (1957). Skeletal age changes in young Amer-
(pp. 181–192). Cambridge University Press. ican males. (Technical Report EP-45). Quartermaster Research and Engi-
Lovejoy, C. O. (1981). The origin of man. Science, 211, 341–350. neering Command, US Army.
Lovejoy, C. O., Meindl, R. S., Pryzbeck, T. R., Barton, T. S., Heiple, J. G., & McSparron, C., Donnelly, C., Murphy, E., & Geber, J. (2020). Migration,
Kotting, D. (1977). Paleodemography of the Libben site, Ottawa group agency, and archaeology: A new theoretical model. International
county, Ohio. Science, 198, 291–293. Journal of Historical Archaeology, 24, 219–232.
Lovejoy, C. O., Meindl, R. S., Pryzbeck, T. R., & Mensforth, R. P. (1985). Meindl, R. S., & Lovejoy, C. O. (1985). Ectocranial suture closure: A revised
Chronological metamorphosis of the auricular surface of the ilium: A method for the determination of skeletal age at death based on the lateral-
new method for the determination of adult skeletal age at death. anterior sutures. American Journal of Physical Anthropology, 68, 57–66.
American Journal of Physical Anthropology, 68, 15–28. Meindl, R. S., Lovejoy, C. O., Mensforth, R. P., & Walker, R. A. (1985). A
Lucy, D., Aykroyd, R. G., & Pollard, A. M. (2002). Nonparametric calibration revised method of age determination using the os pubis, with a
for age estimation. Applied Statistics, 51, 183–196. review and tests of accuracy and of other current methods of pubic
Łukasik, S., Bijak, J., Krenz-Niebała, M., & Sinika, V. (2021). symphyseal aging. American Journal of Physical Anthropology, 68,
Paleodemographic analysis of age at death for a population of Black 29–45.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 147

Meindl, R. S., Mensforth, R. P., & Lovejoy, C. O. (2008). The Libben site: A Moore, J. A., Swedlund, A. C., & Armelagos, G. J. (1975). The use of life
hunting, fishing, and gathering village from the eastern Late Wood- tables in paleodemography. Memoirs of the Society for American
lands of North America. Analysis and implications for Archaeology, 30, 57–70.
palaeodemography and human origins. In J.-P. Bocquet-Appel (Ed.), Müller, H.-G., Love, B., & Hoppa, R. D. (2002). Semiparametric method for
Recent advances in palaeodemography (pp. 259–275). Springer. estimating paleodemographic profiles from age indicator data. Ameri-
Meindl, R. S., & Russell, K. F. (1998). Recent advances in method and the- can Journal of Physical Anthropology, 117, 1–14.
ory in paleodemography. Annual Reviews of Anthropology, 27, Myung, H.-N., & Jang, J.-Y. (2011). Causes of death and demographic char-
375–399. acteristics of victims of meteorological disasters in Korea from 1990
Meindl, R. W., Lovejoy, C. O., Mensforth, R. P., & Don Carlos, L. (1985). to 2008. Environmental Health, 10, 82.
Accuracy and direction of error in the sexing of the skeleton: Implica- Nagaoka, T., Hirata, K., Yokota, E., & Matsu'ura, S. (2006). Paleo-
tions for paleodemography. American Journal of Physical Anthropology, demography of a medieval population in Japan: Analysis of human
68, 79–85. skeletal remains from the Yuigahama-minami site. American Journal of
Meindl, R. W., Mensforth, R. P., & York, H. P. (2001). Mortality, fertility, Physical Anthropology, 131, 1–14.
and growth in the Kentucky Late Archaic: The paleodemography of Nagaoka, T., Ishida, H., Shimoda, Y., Sunagawa, M., Amano, T., Ono, H., &
the Ward site. In O. H. Prufer, S. E. Pedde, & R. S. Meindl (Eds.), Hirata, K. (2012). Estimation of skeletal adult age distribution of
Archaic transitions in Ohio & Kentucky prehistory (pp. 87–109). Kent Okhotsk people in northern Japan. Anthropological Science, 120,
State University Press. 103–113.
Mensforth, R. P. (1990). Paleodemography of the Carlston Annis (Bt-5) Nawrocki, S. P. (1998). Regression formulae for estimating age at death
Late Archaic Skeletal Population. American Journal of Physical Anthro- from cranial suture closure. In K. J. Reichs & W. M. Bass (Eds.), Forensic
pology, 82, 81–99. osteology: Advances in the identification of human remains (pp. 276–
Merritt, C. E. (2015). The influence of body size on adult skeletal age esti- 292). Charles C. Thomas.
mation methods. American Journal of Physical Anthropology, 156, Neel, J. V., & Weiss, K. M. (1975). The genetic structure of a tribal popula-
35–57. tion, the Yanomama Indians. XII. Biodemographic studies. American
Migliano, A. B., & Guillon, M. (2012). The effects of mortality, subsistence, Journal of Physical Anthropology, 42, 25–52.
and ecology on human adult height and implications for Homo evolu- Nikolova, S., Toneva, D., Georgiev, I., & Lazarov, N. (2019). Sagittal suture
tion. Current Anthropology, 53, S359–S368. maturation: Morphological reorganization, relation to aging, and reli-
Milner, G. R. (2004). The Moundbuilders. Thames and Hudson. ability as an age-at-death indicator. American Journal of Physical
Milner, G. R., Anderson, E., & Smith, V. G. (1991). Warfare in late prehis- Anthropology, 169, 78–92.
toric west-central Illinois. American Antiquity, 56, 581–603. O'Donnell, L., & Moes, E. (2021). Sex differences in linear enamel hypopla-
Milner, G. R., & Boldsen, J. L. (2012a). Humeral and femoral head diame- sia prevalence and frailty in ancestral Puebloans. Journal of Archaeolog-
ters in recent white American skeletons. Journal of Forensic Sciences, ical Science: Reports, 39, 103153.
57, 35–40. Osborne, D. L., Simmons, T. L., & Nawrocki, S. P. (2004). Reconsidering the
Milner, G. R., & Boldsen, J. L. (2012b). Skeletal age estimation: Where we auricular surface as an indicator of age at death. Journal of Forensic Sci-
are and where we should go. In D. C. Dirkmaat (Ed.), A companion to ences, 49, 905–911.
forensic anthropology (pp. 224–238). Wiley-Blackwell. Owsley, D. W., & Bass, W. M. (1979). A demographic analysis of skeletons
Milner, G. R., & Boldsen, J. L. (2012c). Transition analysis: A validation from the Larson site (39WW2) Walworth county, South Dakota: Vital
study with known-age modern American skeletons. American Journal statistics. American Journal of Physical Anthropology, 51, 145–154.
of Physical Anthropology, 148, 98–110. Page, A. E., Viguier, S., Dyble, M., Smith, D., Chaudhary, N., Salali, G. D.,
Milner, G. R., & Boldsen, J. L. (2017). Life not death: Epidemiology from Thompson, J., Vinicius, L., Mace, R., & Migliano, A. B. (2016). Repro-
skeletons. International Journal of Paleopathology, 17, 26–39. ductive trade-offs in extant hunter-gatherers suggest adaptive mecha-
Milner, G. R., Boldsen, J. L., Ousley, S. D., Getz, S. M., Weise, S., & Tarp, P. nism for the Neolithic expansion. Proceedings of the National Academy
(2021). Great expectations: The rise, fall, and resurrection of adult of Sciences, USA, 113, 4694–4699.
skeletal age estimation. In B. F. B. Algee-Hewitt & J. Kim (Eds.), Remo- Paine, R. R. (1989). Model life table fitting by maximum likelihood estimation: A
deling forensic skeletal age (pp. 139–154). Academic Press. procedure to reconstruct paleodemographic characteristics from skeletal
Milner, G. R., Boldsen, J. L., Ousley, S. D., Getz, S. M., Weise, S., Tarp, P., & age distributions. American Journal of Physical Anthropology, 79, 51–61.
Steadman, D. W. (2018). Selective mortality in middle-aged American Paine, R. R. (1997). Uniformitarian models in osteological paleo-
women with diffuse idiopathic skeletal hyperostosis (DISH). PLoS One, demography. In R. R. Paine (Ed.), Integrating archaeological demography:
13, e0202283. Multidisciplinary approaches to prehistoric population (Occasional Paper
Milner, G. R., & Ferrell, R. J. (2011). Conflict and death in a late prehistoric 24) (pp. 191–204). Center for Archaeological Investigations, Southern
community in the American Midwest. Anthropologischer Anzeiger, 68, Illinois University.
415–436. Paine, R. R. (2000). If a population crashes in prehistory, and there is no
Milner, G. R., Humpf, D. H., & Harpending, H. C. (1989). Pattern matching paleodemographer there to hear it, does it make a sound? American
of age-at-death distributions in paleodemographic analysis. American Journal of Physical Anthropology, 112, 181–190.
Journal of Physical Anthropology, 80, 49–58. Paine, R. R., & Boldsen, J. L. (2002). Linking age-at-death distributions and
Milner, G. R., & Jefferies, R. W. (1998). The Read Archaic shell midden in ancient population dynamics: A case study. In R. D. Hoppa & J. W.
Kentucky. Southeastern Archaeology, 17, 119–132. Vaupel (Eds.), Paleodemography: Age distributions from skeletal samples
Milner, G. R., Wood, J. W., & Boldsen, J. L. (2019). Paleodemography: (pp. 169–180). Cambridge University Press.
Problems, progress, and potential. In M. A. Katzenberg & A. Grauer Paine, R. R., & Boldsen, J. L. (2006). Paleodemographic data and why
(Eds.), Biological anthropology of the human skeleton (3rd ed., pp. 593– understanding Holocene demography is essential to understanding
633). Wiley-Blackwell. human life history evolution in the Pleistocene. In K. Hawkes & R. R.
Monge, J., & Mann, A. (2015). The paleodemography of extinct hominin Paine (Eds.), The evolution of human life history (pp. 307–330). School
populations. In W. Henke & I. Tattersall (Eds.), Handbook of paleoan- of American Research Press.
thropology (2nd ed., pp. 643–670). Springer. Paine, R. R., & Harpending, H. C. (1996). Assessing the reliability of
Montagu, J. D. (1994). Length of life in the ancient world: A controlled paleodemographic fertility estimators using simulated skeletal distribu-
study. Journal of the Royal Society of Medicine, 87, 25–26. tions. American Journal of Physical Anthropology, 101, 151–159.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
148 BOLDSEN ET AL.

Paine, R. R., & Harpending, H. C. (1998). Effect of sample bias on Roksandic, M., & Armstrong, S. D. (2011). Using the life history model to
paleodemographic fertility estimates. American Journal of Physical set the stage(s) of growth and senescence in bioarchaeology and pal-
Anthropology, 105, 231–240. eodemography. American Journal of Physical Anthropology, 145,
Palkovich, A. M. (1980). The Arroyo Hondo skeletal and mortuary remains. 337–347.
School of American Research Press. Roseboom, T. M., van der Meulen, J. H. P., Ravelli, A. C. J., Osmond, C.,
Palkovich, A. M. (1981). Demography and disease patterns in a protohis- Barker, D. J. P., & Bleker, O. P. (2001). Effects of prenatal exposure to
toric Plains group: A study of the Mobridge site (39WW1). Plains the Dutch famine on adult disease in later life: An overview. Molecular
Anthropologist, 26, 71–84. and Cellular Endocrinology, 185, 93–98.
Pearson, K. (1902). On the change in expectation of life in man during a Rosenberg, K. (2004). Living longer: Information revolution, population
period of circa 2000 years. Biometrika, 1, 261–264. expansion, and modern human origins. Proceedings of the National
Pennington, R. L. (1996). Causes of early human population growth. Ameri- Academy of Sciences, USA, 101, 10847–10848.
can Journal of Physical Anthropology, 99, 259–274. Rougé-Maillat, C., Vielle, B., Jousset, N., Chappard, D., Telmon, N., &
Perry, M. A. (2006). Redefining childhood through bioarchaeology: Toward Cunha, E. (2009). Development of a method to estimate skeletal age
an archaeological and biological understanding of children in antiquity. at death in adults using the acetabulum and the auricular surface on a
Anthropological Papers of the American Anthropological Association, 15, Portuguese population. Forensic Science International, 188, 91–95.
89–111. Ruby, J. G., Wright, K. M., Rand, K. A., Kermany, A., Noto, K., Curtis, D.,
Perry, M. A. (2014). Tracking the second epidemiologic transition using Varner, N., Garrigan, D., Slinkov, D., Dorfman, I., Granka, J. M.,
bioarchaeological data on infant morbidity and mortality. In M. K. Byrnes, J., Myres, N., & Ball, C. (2018). Estimates of the heritability of
Zuckerman (Ed.), Modern environments and human health (pp. 225– human longevity are substantially inflated due to assertive mating.
241). Wiley-Blackwell. Genetics, 210, 1109–1124.
Petersen, H. C., Boldsen, J. L., & Paine, R. R. (2006). Population relation- Santos-Burgoa, C., Sandberg, J., Sudrez, E., Goldman-Hawes, A., Zeger, S.,
ships in and around medieval Danish towns. In G. R. Storey (Ed.), Garcia-Meza, A., Pérez, C. M., Estrada-Merly, N., Colo
n-Ramos, U.,
Urbanism in the preindustrial world: Cross-cultural approaches (pp. 110– Nazario, C. M., Andrade, E., Roess, A., & Goldman, L. (2018). Differen-
120). University of Alabama Press. tial and persistent risk of excess mortality from hurricane Maria in
Petersen, W. (1975). A demographer's view of prehistoric demography. Puerto Rico: A time-series analysis. Lancet Planet Health, 2, E478–
Current Anthropology, 16, 227–245. E488.
Pignolo, R. J., Law, S. F., & Chandra, A. (2021). Bone aging, cellular senes- Sasaki, T., & Kondo, O. (2016a). An informative prior probability distribu-
cence, and osteoporosis. Journal of Bone and Mineral Research Plus, 5, tion of the Gompertz parameters for Bayesian approaches in paleo-
e10488. demography. American Journal of Physical Anthropology, 159, 523–533.
Pinhasi, R., & Bourbou, C. (2008). How representative are human skeletal Sasaki, T., & Kondo, O. (2016b). Maximum likelihood estimate of life
assemblages for population analysis? In R. Pinhasi & S. Mays (Eds.), expectancy in the prehistoric Jomon: Canine pulp volume reduction
Advances in human palaeopathology (pp. 31–44). John Wiley & Sons. suggests a longer life expectancy than previously thought. American
Price, T. D., Manzanilla, L., & Middleton, W. D. (2000). Immigration and the Journal of Physical Anthropology, 161, 170–180.
ancient city of Teotihuacan in Mexico: A study using strontium isotope Sattenspiel, L., & Harpending, H. (1983). Stable populations and skeletal
ratios in human bone and teeth. Journal of Archaeological Science, 27, age. American Antiquity, 48, 489–498.
903–913. Saunders, S. R., Herring, A., Sawchuk, L., Boyce, G., Hoppa, R., & Klepp, S.
Pyrkov, T. V., Slipensky, K., Barg, M., Kondrashin, A., Zhurov, B., Zenin, A., (2002). The health of the middle class: The St. Thomas' Anglican
Pyatnitskiy, M., Menshikov, L., Markov, S., & Fedichev, P. O. (2018). church cemetery project. In R. H. Steckel & J. C. Rose (Eds.), The back-
Extracting biological age from biomedical data via deep learning: Too bone of history: Health and nutrition in the Western Hemisphere (pp.
much of a good thing? Scientific Reports, 8, 5210. 130–161). Cambridge University Press.
Rasmussen, K. L., Milner, G. R., Delbey, T., Skytte, L., Lynnerup, N., Saunders, S. R., Herring, A., Sawchuk, L. A., & Boyce, G. (1995). The nine-
Thomsen, J. L., Schiavone, S., Torino, M., Larsen, L. A., & Boldsen, J. L. teenth-century cemetery at St. Thomas' Anglican church, Belleville:
(2020). Trace element distribution in human cortical bone microstruc- Skeletal remains, parish records and censuses. In S. R. Saunders & A.
ture: The potential for unravelling diet and social status in archaeologi- Herring (Eds.), Grave reflections: Portaying the past through cemetery
cal bones. Heritage Science, 8, 111. studies (pp. 93–117). Canadian Scholars' Press.
Redfern, R. C., & DeWitte, S. N. (2011a). A new approach to the study of Scheidel, W. (2012). Epigraphy and demography: Birth, marriage, family,
Romanization in Britain: A regional perspective of cultural change in and death. Proceedings of the British Academy, 177, 101–129.
late Iron Age and Roman Dorset using the Siler and Gompertz- Schmidt, I., Hilpert, J., Kretschmer, I., Peters, R., Broich, M., Schiesberg, S.,
Makeham models of mortality. American Journal of Physical Anthropol- Vogels, O., Wendt, K. P., Zimmermann, A., & Maier, A. (2021).
ogy, 144, 269–285. Approaching prehistoric demography: Proxies, scales and scope of the
Redfern, R. C., & DeWitte, S. N. (2011b). Status and health in Roman Dor- Cologne protocol in European contexts. Philosophical Transactions of
set: The effect of status on risk of mortality in post-conquest the Royal Society B, 376, 20190714.
populations. American Journal of Physical Anthropology, 146, 197–208. Schmitt, A., Murail, P., Cunha, E., & Rougé, D. (2002). Variability of the pat-
Riga, A., Mori, T., Pickering, T. R., Moggi-Cecchi, J., & Menter, C. G. (2019). tern of aging on the human skeleton: Evidence from bone indicators
Ages-at-death distribution of the early Pleistocene hominin fossil and implications on age at death estimation. Journal of Forensic Sci-
assemblage from Drimolen (South Africa). American Journal of Physical ences, 47, 1203–1209.
Anthropology, 168, 632–636. Séguy, I., & Buchet, L. (2013). Handbook of Palaeodemography. Springer.
Robbins, G. (2011). Don't throw out the baby with the bathwater: Estimat- Séguy, I., Buchet, L., & Bringé, A. (2008). Model life tables for pre-industrial
ing fertility from subadult skeletons. International Journal of populations: First application in palaeodemography. In J.-P. Bocquet-
Osteoarchaeology, 21, 717–722. Appel (Ed.), Recent advances in palaeodemography (pp. 83–117).
Roberts, C. A., & Steckel, R. H. (2019). The developmental origins of health Springer.
and disease: Early life health conditions and adult age at death in Séguy, I., Caussinus, H., Courgeau, D., & Buchet, L. (2013). Estimating the
Europe. In R. H. Steckel, C. S. Larsen, C. A. Roberts, & J. Baten (Eds.), age structure of a buried adult population: A new statistical approach
The backbone of Europe: Health, diet, work, and violence over two applied to archaeological digs in France. American Journal of Physical
millennia (pp. 325–351). Cambridge University Press. Anthropology, 150, 170–183.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BOLDSEN ET AL. 149

Shahar, S. (1993). Who were old in the Middle Ages? Social History of Med- Thevissen, P. W., Fieuws, S., & Willems, G. (2010). Human dental age esti-
icine, 6, 313–341. mation using third molar developmental stages: Does a Bayesian
Signoli, M., Séguy, I., Biraben, J.-N., & Dutour, O. (2002). Pal- approach outperform regression models to discriminate between juve-
éodemographie et démographie historique en context épidémique: La niles and adults? International Journal of Legal Medicine, 124, 35–42.
peste en Provence au XVIIIe siècle. Population, 57, 821–848. Thomas, R. M., Parks, C. L., & Richard, A. H. (2016). Accuracy rates of sex
Singer, R. (1953). Estimation of age from cranial suture closure: A report estimation by forensic anthropologists through comparison with DNA
on its unreliability. Journal of Forensic Medicine, 1, 52–59. typing results in forensic casework. Journal of Forensic Sciences, 61,
Smith, B. D. (1989). Origins of agriculture in eastern North America. Sci- 1307–1310.
ence, 246, 1566–1571. Thompson, R. C., Allam, A. H., Lombardi, G. P., Wann, L. S.,
Smith, B. D. (2001). Low-level food production. Journal of Archaeological Sutherland, M. L., Sutherland, J. D., Soliman, M. A.-T., Frohlich, B.,
Research, 9, 1–43. Mininberg, D. T., Monge, J. M., Vallodolid, C. M., Cox, S. L., el-
Smith, B. D. (2006). Eastern North America as an independent center of Maksoud, G. A., Badr, I., Miyamoto, M. I., el-Din, A. e.-H. N., Narula, J.,
plant domestication. Proceedings of the National Academy of Sciences, Finch, C. E., & Thomas, G. S. (2013). Atherosclerosis across 4000years
USA, 103, 12223–12228. of human history: The Horus study of four ancient populations. Lancet,
Smith, B. D. (2011). A cultural niche construction theory of initial domesti- 381, 1211–1222.
cation. Biological Theory, 6, 260–271. Thomsen, E., & Andreasen, R. (2019). Agricultural lime disturbs natural
Smith, J. (2007 [1624]). The generall historie of Virginia, New-England, and strontium isotope variations: Implications for provenance and migra-
the Summer Isles. In J. Horn (Ed.), Captain John Smith: Writings with tion studies. Science Advances, 5, eaav8083.
other narratives of the Roanoke, Jamestown, and the first English settle- Thomsen, E., Andreasen, R., & Rasmussen, T. L. (2021). Homogeneous gla-
ment of America (pp. 199–670). Library of America. cial landscapes can have high local variability of strontium isotope sig-
Smith, M. E. (2014). Peasant mobility, local migration and premodern natures: Implications for prehistoric migration studies. Frontiers in
urbanization. World Archaeology, 46, 516–533. Ecology and Evolution, 8, 588318.
Snow, C. E. (1948). Indian Knoll skeletons of site Oh 2, Ohio county, Kentucky Todd, T. W. (1920). Age changes in the pubic bone. I. American Journal of
(Reports in Anthropology 4[3,2]). Department of Anthropology, Univer- Physical Anthropology, 3, 285–334.
sity of Kentucky. Todd, T. W. (1921). Age changes in the pubic bone. II. III. IV. American Jour-
Snow, D. R. (2009). The multidisciplinary study of human migration: Prob- nal of Physical Anthropology, 4, 1–70.
lems and principles. In P. N. Peregrine, I. Perios, & M. Feldman (Eds.), Todd, T. W. (1927). Skeletal records of mortality. Scientific Monthly, 24,
Ancient human migrations: A multidisciplinary approach (pp. 6–20). Uni- 481–496.
versity of Utah Press. Todd, T. W., & Lyon, D. W., Jr. (1924). Endocranial suture closure its pro-
Sofaer, J. (2011). Towards a social bioarchaeology of age. In S. C. Agarwal & gress and age relationship. I. American Journal of Physical Anthropology,
B. A. Glencross (Eds.), Social bioarchaeology (pp. 285–311). Wiley-Blackwell. 7, 325–384.
Stanton, P. B., & Hefner, J. T. (2016). Preservation and population. In P. B. Todd, T. W., & Lyon, D. W., Jr. (1925a). Cranial suture closure its progress
Stanton, J. G. Douglas, & S. N. Reddy (Eds.), People in a changing land: and age relationship. II. American Journal of Physical Anthropology, 8,
The archaeology and history of the Ballona in Los Angeles: Bioarchaeology 23–45.
and paleodemography (Vol. 4, pp. 209–234). SRI Press. Todd, T. W., & Lyon, D. W., Jr. (1925b). Suture closure – Its progress and
Stewart, T. D. (1962). Comments on the reassessment of the Indian Knoll age relationship. IV. American Journal of Physical Anthropology, 8,
skeletons. American Journal of Physical Anthropology, 20, 143–148. 149–168.
Storey, R. (1992). Life and death in the ancient city of Teotihuacan. Univer- Torres-Rouff, C., & Knudson, K. J. (2017). Integrating identities: An innova-
sity of Alabama Press. tive bioarchaeological and biogeochemical approach to analyzing the
Stoyanova, D. K., Algee-Hewitt, B., Kim, J., & Slice, D. (2017). A computa-
multiplicity of identities in the mortuary record. Current Anthropology,
tional framework for age-at-death estimation from the skeleton: Sur-
58, 381–409.
face and outline analysis of 3D laser scans of the adult pubic
Trinkaus, E. (1995). Neanderthal mortality patterns. Journal of Archaeologi-
symphysis. Journal of Forensic Sciences, 62, 1434–1444.
cal Science, 22, 121–142.
Stoyanova, D. K., Algee-Hewitt, B., & Slice, D. (2015). An enhanced com-
Trinkaus, E. (2011). Lae Pleistocene adult mortality patterns and modern
putational method for age-at-death estimation based on the pubic
human establishment. Proceedings of the National Academy of Sciences,
symphysis using 3D laser scans and thin plate splines. American Journal
108, 1267–1271.
of Physical Anthropology, 158, 431–440.
Trinkaus, E., & Thompson, D. D. (1987). Femoral diaphyseal histomorphometric
Stull, K. E., Cirillo, L. E., Cole, S. J., & Hulse, C. N. (2020). Subadult sex esti-
age determinations for the Shanidar 3, 4, 5, and 6 Neandertals and Nean-
mation and KidStats. In A. R. Klales (Ed.), Sex estimation of the human
dertal longevity. American Journal of Physical Anthropology, 72, 123–129.
skeleton (pp. 219–242). Academic Press.
Twain, M. (1897). Following the equator: A journey around the world. Ameri-
Stull, K. E., Corron, L. K., & Price, M. H. (2021). Subadult age estimation
can Publishing.
variables: Exploring their varying roles across ontogeny. In B. F. B.
Tzortzis, S., & Signoli, M. (2016). Characterization of the funeral groups
Algee-Hewitt & J. Kim (Eds.), Remodeling forensic skeletal age (pp. 49–
associated with plague epidemics. Microbiology Spectrum, 4, PoH-
73). Academic Press.
0011-2015.
Stull, K. E., L'Abbé, E. N., & Ousley, S. D. (2017). Subadult sex estimation from
Ubelaker, D. H. (1974). Reconstruction of demographic profiles from ossuary
diaphyseal dimensions. American Journal of Physical Anthropology, 163,
skeletal samples: A case study from the tidewater Potomac. (Smithsonian
64–74.
Swedlund, A. C., & Armelagos, G. J. (1969). Une recherche en paléo- Contributions to Anthropology 18). Smithsonian Institution Press.
démographie: La Nubie Soudanaise. Annales, Economies, Sociétés, Civili- Ubelaker, D. H. (2001). Weighting of age interval values in life table con-
sations, 24, 1287–1298. struction. L'Anthropologie, 39, 9–13.
Tangmose, S., Thevissen, P., Lynnerup, N., Willems, G., & Boldsen, J. L. Ubelaker, D. H., & DeGaglia, C. M. (2020). Factors of population variation
(2015). Age estimation in the living: Transition analysis on developing in sex estimation methodology. In A. R. Klales (Ed.), Sex estimation of
third molars. Forensic Science International, 257, 512.e1–512.e7. the human skeleton (pp. 281–293). Elsevier.
Thatcher, A. R. (1999). The long-term pattern of adult mortality and the Ubelaker, D. H., & Khosrowshahi, H. (2019). Estimation of age in forensic
highest attained age. Journal of the Royal Statistical Society A, 162, anthropology: Historical perspective and recent methodological
5–43. advances. Forensic Sciences Research, 4, 1–9.
 26927691, 2022, S74, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.24462 by University Of Illinois At, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
150 BOLDSEN ET AL.

Usher, B.M. (2000). A multistate model of health and mortality for paleo- “health” in pre-Columbian societies. American Journal of Physical
demography: Tirup cemetery [PhD dissertation]. University Park: Penn- Anthropology, 155, 268–280.
sylvania State University. Wood, J. W. (1987). The genetic demography of the Gainj of Papua New
Vaupel, J. W. (2010). Biodemography of human ageing. Nature, 464, Guinea. 2. Determinants of effective population size. The American
536–542. Naturalist, 129, 165–187.
Vaupel, J. W., Manton, K. G., & Stallard, E. (1979). The impact of heteroge- Wood, J. W. (1990). Fertility in anthropological populations. Annual Review
neity in individual frailty on the dynamics of mortality. Demography, of Anthropology, 19, 211–242.
16, 439–454. Wood, J. W. (1998). A theory of preindustrial population dynamics:
Vaupel, J. W., & Yashin, A. I. (1985). The deviant dynamics of death in het- Demography, economy, and well-being in Malthusian systems. Current
erogeneous populations. In N. B. Tuma (Ed.), Sociological methodology Anthropology, 39, 99–135.
(pp. 179–211). Jossey-Bass. Wood, J. W., Holman, D. J., O'Connor, K. A., & Ferrell, R. J. (2002). Mortal-
Walker, P. L., & Cook, D. C. (1998). Gender and sex: Vive la difference. ity models for paleodemography. In R. D. Hoppa & J. W. Vaupel (Eds.),
American Journal of Physical Anthropology, 106, 255–259. Paleodemography: Age distributions from skeletal samples (pp. 129–168).
Walker, P. L., Johnson, J. R., & Lambert, P. M. (1988). Age and sex biases in Cambridge University Press.
the preservation of human skeletal remains. American Journal of Physi- Wood, J. W., Milner, G. R., Harpending, H. C., & Weiss, K. M. (1992). The
cal Anthropology, 76, 183–188. osteological paradox: Problems of inferring prehistoric health from
Walter, B. D., & DeWitte, S. N. (2017). Urban and rural mortality and sur- skeletal samples. Current Anthropology, 33, 343–370.
vival in medieval England. Annals of Human Biology, 44, 338–348. Woods, R. (2007). Ancient and early modern mortality: Experience and
Ward, R. H., & Weiss, K. M. (1976). The demographic evolution of human understanding. Economic History Review, 60, 373–399.
populations. Journal of Human Evolution, 5, 1–23. Zagheni, E., Muttarak, R., & Striessnig, E. (2015). Differential mortality pat-
Weiss, K. M. (1973). Demographic models for anthropology. Memoirs of terns from hydro-meterological disasters: Evidence from cause-of-
the Society for American Archaeology, 27. death data by age and sex. Vienna Yearbook of Population Research, 13,
Weiss, K. M. (1975). Demographic disturbance and the use of life tables in 4–70.
anthropology. Memoirs of the Society for American Archaeology, 30, Zakrzewski, S. (2015). “Behind every mask there is a face, and behind that
46–56. a story”. Egyptian bioarchaeology and ancient identities. In S. Ikram, J.
Weiss, K. M. (1990). The biodemography of variation in human frailty. Kaiser, & R. Walker (Eds.), Egyptian bioarchaeology (pp. 157–167).
Demography, 27, 185–206. Sidestone.
Weiss, K. M., & Smouse, P. E. (1976). The demographic stability of small Zarulli, V., Barthold Jones, J. A., Oksuzyan, A., Lindahl-Jacobsen, R.,
human populations. Journal of Human Evolution, 5, 59–73. Christensen, K., & Vaupel, J. W. (2018). Women live longer than men
Welinder, S. (2001). The archaeology of old age. Current Swedish Archaeol- even during severe famines and epidemics. Proceedings of the National
ogy, 9, 163–178. Academy of Sciences, 115, E832–E840.
Willey, P., Galloway, A., & Snyder, L. (1997). Bone mineral density and sur- Zelditch, M. L., Swiderski, D. L., & Sheets, H. D. (2012). Geometric Morpho-
vival of elements and element portions in the bones of the Crow metrics for biologists: A primer (2nd ed.). Elsevier.
Creek massacre victims. American Journal of Physical Anthropology, Zoeller, G. E., Drew, B. L., Schmidt, C. W., Peterson, R., & Wilson, J. J.
104, 513–528. (2021). A paleodemographic assessment of mortality and fertility rates
Williams, A. N. (2012). The use of summed radiocarbon probability distri- during the second demographic transition in rural central Indiana.
butions in archaeology: A review of methods. Journal of Archaeological American Journal of Human Biology, 2021, e23571.
Science, 39, 578–589.
Williams, L. L. (2009). “The dam is becoming dangerous and may possibly
go”: The paleodemography and political economy of the Johnstown
flood of 1889. In E. C. Jones & A. D. Murphy (Eds.), The political econ- How to cite this article: Boldsen, J. L., Milner, G. R., & Ousley,
omy of hazards and disasters (Society for Economic Anthropology Mono- S. D. (2022). Paleodemography: From archaeology and skeletal
graphs 27) (pp. 31–57). Altamira. age estimation to life in the past. Yearbook Biological
Wilmoth, J. R. (2000). Demography of longevity: Past, present, and future
Anthropology, 178(Suppl. 74), 115–150. https://doi.org/10.
trends. Experimental Gerontology, 35, 1111–1129.
Wilson, J. J. (2014). Paradox and promise: Research on the role of recent 1002/ajpa.24462
advances in paleodemography and paleoepidemiology to the study of