Annu. Rev. Anthropol. 1998.

27:375–99
Copyright 1998 by Annual Reviews. All rights reserved

RECENT ADVANCES IN METHOD

AND THEORY IN
PALEODEMOGRAPHY
Richard S. Meindl
Department of Anthropology, Kent State University, Kent, Ohio 44242-0001;
e-mail: [email protected]

Katherine F. Russell
Department of Biology, University of Massachusetts, Dartmouth, Massachusetts
02747-2300; e-mail: [email protected]

KEY WORDS: archaeological demography, skeletal biology, sex and age estimation,
model life tables, intrinsic growth rate

ABSTRACT
Current methods in skeletal biology have improved significantly our ability
to estimate the demographic parameters of extinct populations. Gross mor-
phological and histological age indicators have been developed and tested in
a variety of contexts, revealing great variation in the levels of accuracy of age
prediction of each indicator. Primary attention is given here to the best-
performing hard-tissue indicators of age and to composite methods of recov-
ering the age and sex distribution of a cemetery. It is becoming increasingly
apparent that some cemetaries should not be used for demographic recon-
struction. Such collections have no bearing on the feasibility of paleodemo-
graphic research. Our review concludes with discussions about the role of
comparing modern mortality patterns to those of paleodemography, and the
issue and impact of departures from stationary demographic conditions dur-
ing prehistoric times.

INTRODUCTION
More than 15 years ago Bocquet-Appel & Masset (1982) raised a number of
questions concerning the analysis of vital rates from skeleton-based demogra-

375
0084-6570/98/1015-0375$08.00
376 MEINDL & RUSSELL

phies. Since that time the once-tranquil field of paleodemography has become
almost as disputatious as some other areas of anthropology. However, this
shift has resulted in constructive debates and timely reexaminations of models
and assumptions. Further critical review and new research efforts in this area
have increased dramatically, moving paleodemography from the periphery to
the forefront of prehistoric studies (Roth 1992).
Paleodemography is more than the study of mortality and fertility of ar-
chaeological populations. It also includes the estimation of the distribution,
density, and age composition of prehistoric peoples. It considers intrinsic rates
of growth or decline, and it may include migration and the age and sex struc-
ture of migration as well. The field attracts anthropologists for different rea-
sons. Like Goodman et al (1984), Weiss (1989), and Buikstra (1997), our pri-
mary interests in cemeteries have always been the health, life-history, and evo-
lutionary aspects of human biodemography. We limit our comments here to re-
cent methodological advances in estimating the vital structure of an extinct
population from the cemetery on which it was based. We discuss some of the
sampling problems in paleodemography, methods for estimating the distribu-
tion of ages by sex in skeletal series, and life-table mortality schedules and fer-
tility in the context of stable population theory. We also discuss the role of
mortality models external to paleodemography and an approach to the problem
of unknown intrinsic rates of growth in prehistory.
Space also precludes discussion of recent views concerning morbidity pat-
terns in skeletal populations (Wood et al 1992) or the health status surrounding
the adoption of agricultural systems (Cohen 1997). Limits remain to the infer-
ences we can draw from paleodemography; however, studies of living, so-
called “demographically inconvenient” populations have their own unique
challenges (Leslie & Gage 1989:20). We maintain that archaeological popula-
tions offer a special and important view of the biology, cultural evolution, and
basic demography of extinct peoples that is not available from other fields of
anthropology.

SAMPLING PROBLEMS IN PALEODEMOGRAPHY

The demographic reconstruction of an extinct society is a sampling process
that may involve several potential sources of error. Aboriginal inhumation
practices, soil and other conditions of interment including subsequent distur-
bances of the burials, and the care and skill of the excavators may all affect
how well a cemetery represents the population on which it was based. It bears
repeating that if “an unbiased, representative sample . . . cannot be assumed,
further demographic analysis is not likely to be productive” (Weiss 1973:58).
Certainly, burial customs associated with infanticide have skewed the find-
ings from many cemeteries (Scrimshaw 1984, Saunders 1992). Yet if the prac-
 PALEODEMOGRAPHY 377

tice of infanticide was restricted to such uncommon situations as genetic de-

fects, the birth of twins, and breach births there is little impact on mortality
profiles. Conditions that predispose cultures toward (typically female) infanti-
cide for less immediate reasons include very short interbirth intervals, high
costs of transporting infants over large distances, perceptions of the need to
limit family size, and the greater value of males as economic providers or com-
batants (Divale & Harris 1976, Scrimshaw 1984). While it might be widely ac-
cepted that infanticides were usually not buried with the rest of the population
(Saunders 1992), it is important to note that skewed sex ratios resulting from
frequent female infanticide would be archaeologically detectable.
In any case, such perinatal deaths may not be recovered, usually took place
within a matter of minutes of birth, and would suggest that paleodemographers
should alter their definition of the radix by approximately that amount of time.
Infanticide in human populations, both living and extinct, is a difficult demo-
graphic parameter to estimate. Although a condition of mortality by definition,
it should be regarded by anthropological demographers as a deliberate control
over fertility. That is, removing infanticide altogether from the calculation of
life expectancy at birth would make the metric more reflective of the actual
biological conditions of mortality and, therefore, more utilitarian to population
scientists across the spectrum of human ecologies and cultures.
Ethnographic evidence indicates that, in many societies, the old and the in-
firm are killed occasionally, but their remains are rarely treated differently
from those of other adult decedents. Any graves containing crippled, injured,
or malformed elderly would imply that efforts had been made to return all of
that society’s members to a common place of burial. Young adults, particularly
males, may have died violently or otherwise some distance from the habita-
tion/cemetery. A mixture of bundle burials (i.e. only crania and assorted long-
bones) with articulated ones further suggest such a general effort, since it indi-
cates return of the kinsmen’s essential remains to the cemetery.
It is not the purpose of our review to present and evaluate paleodemo-
graphic methods for sites with scattered or poorly preserved remains. Nor is it
possible to usefully estimate the levels of bias in preservation by age or sex for
any given site, when it is clear that some differential loss of skeletons has oc-
curred (Gordon & Buikstra 1981). If a portion of a cemetery cannot be aged or
sexed owing to dissolution of calcified tissues, or if evidence exists for care-
lessness in excavation and recovery, the value of that site to the study of human
demography is diminished by an unknown extent.
Our experience with skeletal sites in the eastern United States leads us to
conclude that the general claims about the effect of differential preservation by
age or sex of bone remains on the field of paleodemography (Masset 1973,
Walker et al 1988, Jackes 1992) are overstated. Any potential for this kind of
bias is obvious at once by simple inspection of the integrity of the burial pits,
378 MEINDL & RUSSELL

articulations, and periosteal bone layers. Bone and tooth survival is a function
of mechanical displacement, soil drainage, and especially pH (Gordon &
Buikstra 1981). Unfavorable conditions cause the destruction or agitation of
hard-tissue remains with such speed that sites of moderate antiquity fall pri-
marily into two groups—those with demographically useful bone assemblages
and those without. An intermediate site in which the human remains are actu-
ally in the process of disappearing would probably have such fragile contents
that even the adult sample would be in poor condition.
Both Sundick (1978) and Saunders (1992) argued that in such sites the hard
evidence required to recognize and age a subadult (i.e. primarily teeth) is pre-
served at least as well as the bones of a robust adult skeleton. Nevertheless,
some added provision may be needed in the wake of moderate mechanical dis-
turbances, such as tree roots and rodents, to estimate accurately the proportion
of infants and very young children (Lovejoy et al 1977). The issue of differen-
tial preservation of bone microstructure is no different. Although histomor-
phological data contain instances of age-related preservation bias (Garland
1987, Hanson & Buikstra 1987), the potential for bias should be recognizable,
and therefore, any impact on the feasibility of paleodemographic inference is
minor.
In nearly all instances the archaeological data themselves should demon-
strate clearly which sites are useful to paleodemographers (contra Jackes
1992:199). A case in point is the inappropriate use of the late-eighteenth cen-
tury Newton Plantation skeletons to question a method of fertility estimation.
The divergence of skeletal- and historical-based inferences failed to “promote
demographic optimism” despite a meager sample and the fact that “preserva-
tion is poor” (Corruccini et al 1989:610). A more widely cited example is the
Purisma Mission cemetery in California, again admittedly a “poorly pre-
served” skeletal collection (Walker et al 1988:184). An argument was made
from the field notes and the curated materials that biases in preservation as
well as selective archaeological recovery were sizable sources of census error.
However, it has always been clear that no demographic analysis should be un-
dertaken with culled skeletons. A companion survey of the Franciscan priests’
written burial records for La Purisma only verified the obvious archaeological
inferences.

ESTIMATION OF SEX OF ADULT SKELETONS

Sex, the most important demographic variable, is recoverable directly through
the osteology of primary and secondary sexual characteristics. These result
principally from differential growth rates during adolescence. Males exhibit a
prenatal slowing of maturation beginning early in gestation, protracted prepu-
bescent growth, a greater intensity of the growth spurt (which, because of its
 PALEODEMOGRAPHY 379

delay, acts upon an absolutely larger musculoskeletal frame), and additional

changes that occur in late and post-adolescence (Stini 1985).
A remarkable aspect of human development is its degree of canalization:
Growth trajectories tend to be self-stabilizing. In longitudinal studies, ultimate
growth is found to be largely genetic. In instances in which growth is inhibited
temporarily (e.g. hypothyroidism later corrected, or a cortisol-producing tu-
mor later removed), subsequent catch-up rates may be as high as three times
the normal (Harrison et al 1977). However, some studies on contemporary
populations tend to underestimate and obscure the potential magnitude of en-
vironmental effects. The great recuperative powers of children can be over-
whelmed when poor conditions are too prolonged, especially malnutrition and
illness. Since girls are more resistant than boys to intensive growth-inhibiting
stress, the effect is a decrease in dimorphism. [See Hamilton (1982) for an in-
formative survey of the causes of skeletal dimorphism].
Thus, most adult somatic dimensions are environmentally labile, and since
the environment most probably played a more powerful role in growth during
prehistoric times, any use of pure size measures to determine the sex of a skele-
ton is problematic. This applies to all methods not based on primary sexual
characteristics, including those of the upper limb (Holman & Bennett 1991),
radial head (Berrizbeitia 1989), tibia (Holland 1991, Kieser et al 1992), femur
and humerus (Dittrick & Suchey 1986), proximal components of the appen-
dicular skeleton (Richman et al 1979), femoral circumference (Black 1978),
talus and calcaneous (Steele 1976), sternal rib (êcan & Loth 1986), first cervi-
cal vertebra (Marino 1995), metacarpals (Scheuer & Elkington 1993, Falsetti
1995), metacarpal asymmetry (Lazenby 1994), iliac crest of the pelvis
(Fernández Camacho et al 1993), and tooth crowns (Owsley & Webb 1983),
just to sample this ongoing industry. [See Buikstra & Mielke (1985) for a com-
parison of techniques for adult sex determination. For each technique the
authors listed the method’s accuracy as well as whether it is based on a series
of known sex.] Sex determination in paleodemography depends on the true
magnitude of sexual dimorphism, the amount of within-sex variation, and the
direction of sexing error as well.
As has been recognized for some time (Hrdli¹ka 1952, Krogman 1962,
Acsádi & Nemeskéri 1970, Hamilton 1982, White 1991), the os coxae, or in-
nominate bone of the pelvis, is the most reliable osteological indicator of sex in
the adult owing to the central role this structure plays in parturition. In homi-
nids a critical disproportion between the dimensions of the female inner pelvis
and the term fetal head and shoulders developed progressively over the past
few million years (Tague & Lovejoy 1986). The dimorphism of the modern
bony pelvis is an adaptation to this disproportion. In fact, in nonhominoid pri-
mates the ischiopubic index is greater in females, and—except for a few lorises
and a species of howler monkey—strongly so (Leutenegger 1982). The index
380 MEINDL & RUSSELL

is a correlate of the proximity of neonatal cranial dimensions to those of the fe-

male adult true pelvis [African apes are too large to be subject to such selection
pressures (Tague 1991)] and has its basis in local estrogen receptor sites in the
pubis. Evidence indicates antecedent systems for this basis and for the poste-
rior portion of the ilium as well in more generalized mammals such as mice
(Uesugi et al 1992). For human females these selection pressures have resulted
in (a) altered growth trajectories that position the sacrum more posterior in the
birth canal and (b) continued elongation of the iliopubic ramus even into the
third decade of life (Budinoff & Tague 1990).
The repositioning of the female sacrum produced a wider sciatic notch (i.e.
the large indentation superior to the ischial spine) in the female. The latter ad-
aptation forms the basis of a popular and very accurate method for sex determi-
nation based on the morphology of the pubis (Phenice 1969). Covariation of
the width of the sciatic notch and the ischiopubic index can be best appreciated
in a well-known bivariate plot of Eskimo pelvises (Hanna & Washburn
1953:24). Three variables (ratio of the ischium to the pubis, sciatic notch, and
medioventral aspects of the pubis) are the primary bases of the objective meth-
ods that paleodemographers and forensic scientists have used traditionally.
They are also the basis of the discriminatory power of conventional multivari-
ate approaches to the problem (Schulter-Ellis et al 1983) as well as size-
corrected algorithms (Arsuaga & Carretero 1994). Using pelvises from skele-
tons of known sex, we and others (Richman et al 1979, Meindl et al 1985a,
Sutherland & Suchey 1991, Luo 1995) have approximated closely the high ac-
curacies first predicted by the developers of both observational and statistical
methods of sexing. Some workers have reported less satisfactory results (Lov-
ell 1989, Rogers & Saunders 1994). Efficacy of the methods appears to depend
on the aspects of the innominate bone that are evaluated.
Meindl et al (1985a) suggested that the opposing selection pressures of effi-
ciency in locomotion and success in parturition may have limited female pel-
vic variability relative to that of the male. However, Tague (1989) found no
such sex-specific difference in terms of absolute metric variability. The issue
is important in demographic reconstruction because it has some bearing on the
likely direction of error in sexing adult skeletons. While Meindl et al (1985a)
predicted that the most common error should be “male-called-female,” Weiss
(1972) has pointed to the high adult sex ratios of many archaeological popula-
tions. However, a prehistoric tendency toward heavy female subadult mortal-
ity (which may have included infanticide) would explain these observations.
Sex identification using the skull has traditionally been based on differ-
ences in the prominence of muscle attachments, the curvature of the frontal
profile, the dimensions of the cranial base, and absolute tooth and jaw size and
morphology. Although male skulls tend to be larger and more robust, the de-
gree of dimorphism can vary greatly among populations. Giles & Elliot (1963)
 PALEODEMOGRAPHY 381

derived batteries of discriminant functions based on the metrics of crania (i.e.

skulls minus mandibles) from cadaver skeletons of known sex. They predicted
error rates not much better than some of the longbone methods (see above). Pa-
leodemographic applications are even more problematic: (a) There is no provi-
sion for distorted prehistoric crania (pre- or postmortem). (b) The functions de-
rive nearly all of their power from size effects, which are population specific
(see above), resulting in reduced accuracy in collections other than the Terry or
the Hamann-Todd collections (Kajanoja 1966). (c) Older adults of both sexes
display increasingly masculine craniometrics (Meindl et al 1985a) because of
remodeling and craniofacial expansion, which occur well into old age (Israel
1973, 1977). This kind of directional error might prove to be the greatest liabil-
ity in formulating two-sex mortality models in paleodemographic contexts.
Some components of the human skull exhibit hormonally mediated sexual
dimorphism. In fact, discriminant functions that incorporate some measure of
the prominence of the superciliary arches (brow ridges) and the external oc-
cipital protuberance (at the midline of the meeting of the occipital and nuchal
planes) perform quite well (Song et al 1992). We suspect that receptor sites
may soon be discovered on the supraorbital aspect of the frontal bone and per-
haps elsewhere. If so, subjective and statistical sexing of skulls may one day
have more to offer in paleodemographic reconstruction.

AGE ESTIMATION OF SUBADULT SKELETONS

Three kinds of hard-tissue data reflect age in the developing skeleton. These
are, starting with the most accurate (and least affected by environmental condi-
tions), dental development and eruption, then longbone epiphyseal closure,
and finally linear longbone growth. Linear seriation should be used when
working with these data in order to minimize observer error and for other rea-
sons as well (see below).
Several studies have provided chronologies of deciduous and permanent
tooth formation and emergence. For all 26 pairs of teeth in the human primary
and secondary dentitions, Oliver (1969) presented schedules of stages rather
than single eruption times. Moorrees et al (1963a,b) did the same but restricted
their analyses to several teeth. Ubelaker (1989) estimated the timing of dental
development in Native-American children. Demirjian and colleagues (Demir-
jian et al 1973, Demirjian & Levesque 1980) predicted a dental age from a
composite maturity score based on the radiographs of seven permanent left
mandibular teeth. Gustafson & Koch (1974) published a schematic representa-
tion of tooth formation and eruption and claimed that comparison of the dental
developmental status of any 3 to 13 year old to the diagram will yield an age es-
timate generally within two months of actual age. Bang (1989) and El-Nofely
& êcan (1989) surveyed the literature on age assessment from developing
382 MEINDL & RUSSELL

dentition. White (1991) and Schwartz (1995) provided instructions for using
some of these methods.
Tooth formation and eruption varies among populations. For example,
some African children tend to develop more quickly than some Europeans and
especially Asians; however, while these distinctions are statistically signifi-
cant owing to the large sample sizes, the average differences are less than a
year (El-Nofely & êcan 1989). Despite these minor geographic variations,
there is consensus that dental development is the age indicator most resistant to
nutritional deficiencies, hormone imbalances, and any other environmental ef-
fect.
Saunders et al (1993) verified the precision of Moorrees’s standards with a
small sample from the St. Thomas’s Church cemetery. Demographers of con-
temporary nonindustrial societies have pointed out that for their purposes there
are no meaningful differences among populations in deciduous tooth eruption;
moreover, they now recognize that, in so-called noncounting cultures, dental
development in young children is a better predictor of a child’s age than is the
mother’s recall of the date of birth (Townsend & Hammel 1990).
Data on the union of epiphyses are frequently used to establish age at death
during the second decade of life; however, environmental stress can retard
skeletal age by as much as three years (Johnston & Zimmer 1989). General
summaries of the standards available may be found in Stewart (1979), Krog-
man & êcan (1986), and Ubelaker (1989). Cemeteries may represent the
growth of the population’s healthy children (Wood et al 1992); therefore, in
paleodemographic applications, age estimation from diaphyseal length and
epiphyseal closure may be biased downward.

ADULT AGE ESTIMATES

Postdevelopmental skeletal aging is a complex product of both genetics and
environment. In the third decade of life, variation in the aging process starts to
increase, both between individuals and even within a single skeleton. Our
work at Kent State University was the most recent in a series of studies, span-
ning several decades, that used multiple indicators of adult skeletal age in or-
der to partly compensate for this phenomenon (Krogman 1962, Acsádi &
Nemeskéri 1970, Lovejoy et al 1985a). Although the indicators of age are re-
viewed separately here, we do not imply that age assessment of adults can be
based on single age indicators. We present in this review those methods ap-
pearing to be among the most reliable estimators of age currently available.
We allude to other methods of age determination because of their historical
importance.
A simple way to reduce observer error in age estimation is to seriate all
skeletons for each age indicator (Lovejoy et al 1985a). This procedure has two
 PALEODEMOGRAPHY 383

effects: (a) The intercalation of every osteological specimen into the entire or-
dered series extracts maximum information about relative biological age, and
(b) seriation minimizes observer error by eliminating fatigue and other time-
shift effects. The arrangement of a sequence of increasing skeletal age, with
the progressive detection and reduction of error, and the final assignment of in-
dividual target ages requires long tables and consumes considerable amounts
of time and effort. However, it produces final ages for each skeleton that in ef-
fect are assigned simultaneously. We recommend this exercise to even the
most skilled osteologist for its heuristic effects. The process should be re-
peated for each age estimator independently. Phase, component, and even re-
gression methods for estimating age can benefit greatly from this approach.
Error in age estimation can be quantified only when methods are tested on
individuals of known chronological age. The amount of bias1 and inaccuracy2
of age prediction vary greatly among skeletal aging methods. In most circum-
stances inaccuracy can be minimized across the age range by including more
than one method. Bias, however, is another matter and, in paleodemographic
contexts, a more serious one. Most methods exhibit age-related levels of error,
for example, small error below a certain age but unacceptably wide error mar-
gins afterward. Application of such methods must be limited to those age
ranges for which age-related changes are regular. Otherwise, the problem of
reference population bias may become pronounced.
A critical problem in paleodemography is the bias that results from poor ac-
curacy in age estimation and its effect in skewing the calculation of demo-
graphic profiles. The reversal of the fixed independent regressor (real age)
with the “free-to-vary” dependent regressand (age indicator morphology) has
always been a problem in the field of skeletal demography (Bocquet-Appel &
Masset 1982). Age estimation of a series of skeletons may produce a distribu-
tion reflecting the skeletal reference population on which the aging method
was originally developed, rather than the age structure of the cemetery being
analyzed. Such mimicry tends to be marked when indicators correlate poorly
with actual age. Several studies have addressed this issue (Van Gerven & Ar-
melagos 1983; Buikstra & Konigsberg 1985; Mensforth & Lovejoy 1985;
Buikstra et al 1986; Siven 1991; Jackes 1985, 1992, 1993).
Bocquet-Appel & Masset (1996) have become convinced recently that the
mean age of a cemetery is indeed recoverable. Konigsberg & Frankenberg
(1992, 1994; Konigsberg et al 1997) pursue a more useful goal, the demonstra-
tion that both the mean and the shape of a cemetery age-distribution can be es-
timated with minimal bias. They have surveyed paleodemographic reports, ex-
plored the fisheries literature on the subject of calibration, and shown the supe-

1 1Bias = Σ (estimated age − real age)/number of individuals; that is, sign is considered.
2 2Inaccuracy = Σ |estimated age − real age|/number of individuals; that is, sign is not considered.
384 MEINDL & RUSSELL

riority of maximum likelihood methods of adult age estimation over Bayesian

approaches. If these methods are to be used, future developments in age deter-
mination methods will have to include distributions of morphology types by
extreme age, and cross-classification of multiple indicator morphologies
against known age as well (Konigsberg & Frankenberg 1992). However, there
will always be the need to appreciate that no statistical procedure can ever
overcome the limitations of an abysmal age indicator (i.e. one that is unrelated
to age). Even some of the traditional adult indicators of skeletal age yield virtu-
ally no additional information after a certain developmental stage has been at-
tained (see below).
Gross Morphological Indicators of Age
The gross morphological indicators of age are relatively easy to apply, require
no special equipment, and are nondestructive. However, their application re-
quires the observer to have a clear understanding of the anatomy, develop-
ment, and physiology of the soft tissues associated with the bony/dental land-
marks in order to understand normal variation in the age-related features and to
discriminate outliers, pathologies, and postmortem damage.
CRANIAL SUTURE CLOSURE Cranial suture closure has long been used in the
estimation of skeletal age. There are now formal aging methods based on both
ectocranial and endocranial closure, using different batteries of sutures and
different scoring systems (Meindl & Lovejoy 1985, Masset 1989). Employ-
ment of sutures as an aging indicator fell into disfavor during the 1950s follow-
ing widely cited criticisms (Singer 1953, Brooks 1955, McKern & Stewart
1957). The subsequent abandonment of such methods during this period was
the result of poor and variable performances of available systems in addition to
the expectation that more accurate techniques would soon emerge. However,
because no such method has been forthcoming, cranial suture closure is being
used once again, but only as one of multiple indicators of skeletal aging.
Acsádi & Nemeskéri (1970) presented one of the most widely used methods
for aging the skeletons of young and middle-aged adults (for whom suture clo-
sure was still active at the time of death). Subsequent tests of vault suture aging
of the endocranium by Masset (1989) and the ectocranium by Saunders et al
(1992) revealed poor performance. Most vault closure is simply too early to be
of any great help in estimating extreme age, for which new forensic standards
are most needed. Meindl & Lovejoy (1985) offered two ectocranial methods,
one of which appears to monitor changes at the higher ages more efficiently;
nevertheless, error may become large after age 50 (Lovejoy et al 1985a).
Schwartz (1995) fully described the application of these techniques.
PUBIC SYMPHYSIS The pubic symphyseal face (i.e. the surface of the pubic
bone where the two innominates most closely approach) has received the most
 PALEODEMOGRAPHY 385

attention of all the anatomical structures in estimating age at death. Todd first
described age-related changes in humans (1920, 1921a ) and in other mammals
(1921b). His careful characterization of pubic epiphyseal fusion and progres-
sive degenerative changes provided the basis for all such subsequent aging
methods (McKern & Stewart 1957, Gilbert & McKern 1973). When used in
one of its condensed forms (Meindl et al 1985b, Suchey et al 1986) the pubic
symphysis can be a valuable method for estimating young and middle adult
age.
The importance of the pubic symphysis in paleodemography is limited to
the aging of adults up to about age 40 years (Hanihara & Suzuki 1978; Meindl
et al 1985b; Lovejoy et al 1995, 1997). The appearance and fusion of the ven-
tral rampart of the pubic symphysis are nothing more than uniquely human, de-
layed epiphyseal events, representing the only discrete features that make the
pubic symphysis a valuable aging locus. Fusion is typically completed before
age 35, marking the end of all developmental activity at this joint. Post-
epiphyseal changes are all degenerative, highly variable, and not particularly
valuable for age estimation, as confirmed by Saunders et al (1992) and others.

AURICULAR SURFACE OF THE ILIUM Lovejoy et al (1995, 1997) examined the

senescent biology of the auricular surface of the ilium (i.e. the iliac side of the
innominate-sacrum articulation) in hominoids. They argued that age changes
in the morphology of the sacroiliac joint are a consequence of its highly un-
usual phylogenetic and embryological development and that the iliac face has
a uniquely protracted metamorphosis in both bipedal and quadrupedal homi-
noids. They demonstrated that, unlike the pubic symphysis, the auricular sur-
face reflects changes well into the sixth decade. The changes from a billowed
epiphyseal-like surface to a more coarsely granulated one and especially the
late metamorphosis from a smooth dense face to one of increasing degenera-
tive irregularity provide a series of biological events that can be used to assess
age (Lovejoy et al 1985b, Meindl & Lovejoy 1989).
This method is not without its critics: Rogers (1990) was not able to repli-
cate our results for older skeletons, and Jackes (1992) cited “constant interob-
server disagreement” over the auricular ages of apparently very fragmentary
Mesolithic skeletons. Murray & Murray (1991) confirmed that there are nei-
ther sex nor geographic differences in auricular aging but were displeased with
the method’s performance in a forensic setting. Konigsberg & Frankenberg
(1992) recognized from the Murrays’s account that the auricular surface is
valuable to paleodemography because it fits Howell’s (1976) “uniformitarian
assumption” that both extant and extinct human populations age in similar
fashions and at identical rates. Saunders et al (1992) examined a series of older
skeletons of known age but failed to adequately test the ability of the auricular
surface to predict extreme age: They chose not to seriate, and they curiously
386 MEINDL & RUSSELL

assigned unknown specimens to the youngest age class possible despite our
explicit directions to the contrary (Meindl et al 1990:353). Still, the auricular
surface predicted extreme age better than the other indicators, especially the
pubis, in this test (Saunders et al 1992) and in another (Bedford et al 1993). Fi-
nally, researchers have noted that the frail pubic symphysis is not nearly as
archaeologically durable as the auricular surface (Pfeiffer 1986, Waldron
1987).
COSTOCHONDRAL JOINTS The potential of costochondral joints (i.e. ventral
articular surfaces of ribs) to represent adult age was appreciated as early as the
1950s (McKern & Stewart 1957). êcan et al were the first to formalize an ag-
ing method (by sex) based on the fourth rib (1984a,b). Independent tests of
their method have confirmed its utility in age estimation and its acceptably low
bias before age 50 (Saunders et al 1992, Russell et al 1993, Dudar et al 1993).
The metamorphosis of the joint includes a change from a smooth billowed im-
mature surface to a more concave and irregular surface as a result of remodel-
ling (i.e. periosteal accretion and endosteal resorption). Any geographic differ-
ences in the rates of change (êcan et al 1987) cannot be very large (Russell et
al 1993).
OCCLUSAL DENTAL WEAR Occlusal dental wear is one of the most reliable
methods of estimating adult age, as long as the assumption of uniform rates of
attrition in the population can be reasonably met. Murphy (1959) and Miles
(1963) presented two of the seminal methods for estimating age from dental
wear. Murphy (1959) used an ordinal scale in which each cusp was scored in-
dependently. Miles (1963) used a single evaluation per tooth. Modifications of
such methods were fruitful and can be recommended for age estimation (Smith
1984, Kieser et al 1985, Lovejoy 1985). Even in the heterogenous Hamann-
Todd collection, occlusal dental wear was the least-biased single estimator of
extreme adult age (Lovejoy et al 1985a); therefore, it is very important for pa-
leodemographic application.
Careful assessment of adolescent tooth wear, consideration of individual
variation in dental attrition, and an appreciation of the effects of antemortem
tooth loss on the rest of the dentition must all be evaluated for successful appli-
cation of the methods. We are aware of two sites, both Green River Archaic
cemeteries, for which dental wear was too extreme to be useful in aging skele-
tons: the Carlston-Annis site (Mensforth 1990) and the Ward site (Meindl et al
1998). Certainly the rates of wear changed for the oldest members of these
populations when attrition began to involve the buccal roots of the lower mo-
lars.

CHANGES IN BONE DISTRIBUTION Many investigators have observed age-

related changes in adult cortical and trabecular (i.e. compact and spongy) bone
 PALEODEMOGRAPHY 387

distribution (Acsádi & Nemeskéri 1970, Singh 1972, Walker & Lovejoy
1985). In the macroscopic methods used for such observations, seriation of ra-
diographs is especially important in fostering the retention of biologically rele-
vant differences among individuals in both bone density and distribution. Be-
cause of the dramatic physiological and mechanical factors influencing bone
remodeling, the internal macroscopic structure of bone is not an extremely ac-
curate method of age assessment. However, the method can contribute effec-
tively to improving age estimation when used in concert with other aging sites
(Lovejoy et al 1985a).
Histological Indicators of Age
Bone growth and modeling produce a mature cortex with a characteristic struc-
ture. Remodeling occurs throughout life in long-lived adult mammals in re-
sponse to mechanical and physiological activity. The remodeling of osteons
provides a quantifiable means of assessing age-at-death from sectioned adult
cortical bone. Kerley (1965, Kerley & Ubelaker 1978) provided the first sys-
tematic means of quantifying osteons and osteon fragments, circumferential
lamellar bone (thin parallel layers of bone), and non-Haversian vascular canals
(formed as the cortex increases diameter around peripheral blood vesels) to
produce regression equations suitable for estimating age in unknown individu-
als. Other modifications followed (Ahlqvist & Damsten 1969, Singh & Gun-
berg 1970, Thompson 1979, Walker 1989).
Since bone continues to remodel under mechanical stress throughout adult
life, multiple sampling sites from more than one cross-section are necessary
(Stout 1992). Stout & Paine (1992) suggested that non–weight-bearing bones
such as the clavicle and sixth rib might be more suitable than the major long
bones because they undergo less mechanical remodeling. Frost (1987) and
Stout (1992) proposed the calculation of a mean tissue age.
Histological aging methods all use linear regression equations, which are
notoriously limited (Giles and Klepinger 1988, Maples & Rice 1979, Lucy et
al 1996). Confidence intervals are neither small nor equivalent but expand as
the regression line moves away from the mean. Histological techniques have
been reported which are averages of histological and gross estimators adequate
for measuring rib age in individuals under age 60 (Dudar et al 1993). However,
they may be no better at indicating older ages than methods using the pubic
symphysis (Aiello & Molleson 1993). Walker et al (1994) claimed that for in-
dividuals older than age 50, osteon density does not correlate as well with age
as it does with geometric properties of femoral cross-sections.
Preservation of microscopic bone structure depends heavily on soil condi-
tions, and macroscopic appearance of a bone may not be a good indicator of its
histological preservation (Garland 1987). There can be partial to complete de-
struction (particularly of periosteal regions) of the bone microstructure, even
388 MEINDL & RUSSELL

when gross bone preservation is excellent (Lovejoy et al 1977, Hanson &

Buikstra 1987, Pfeiffer 1992, Jackes 1992).
Gustafson presented a detailed method for estimating age from histological
changes in individual adult teeth. The method involves assigning ordinal
scores to six age-related changes: attrition, periodontosis, secondary dentin
formation, cementum apposition, root resorption, and root transparency (Gus-
tafson 1950). Problems with the original method have been well documented
(Maples & Rice 1979). Yet the age-related changes first described by Gustaf-
son have generated many useful methods for estimating age at death, even
when only one or a few teeth are available per individual (Maples 1978, Con-
don et al 1986, Lucy et al 1996, Russell 1996). Bias and inaccuracy have been
evaluated for a cementum annulation method (Condon et al 1986) and for a
combination of attrition, secondary dentin, root dentin transparency, and ce-
mentum thickness (Russell 1996).
A Hierarchical Method of Age Estimation
Acsádi & Nemeskéri (1970) introduced a complex method for determining
adult age, composed of estimates from the pubis, the endocranial sutures, and
the loss of trabecular bone in both proximal humerus and femur. They sug-
gested dividing the adult portion of the cemetery into three groups on the basis
of pubic symphyseal morphology (young, middle, and old). This step would
determine a specific portion of the age ranges (low, middle, or high) for each of
the three remaining indicators. The final estimate was the arithmetic average
of the four.
We have suggested calculating the final age estimates of specimens from a
cemetery as a weighted average of all indicators (called the summary age),
with the weights determined by the first eigenvector of a principal component
analysis of the intercorrelation matrix of the various age indicators used
(Lovejoy et al 1977, Meindl et al 1983). We later suggested a modification to
eliminate systematic bias in the higher ages: Mean ages should be calculated
from all but the lowest estimates, or the lowest two, for the oldest skeletons in
the series (Lovejoy et al 1985a). The correlations of the final age estimates
with real ages were very high when subjected to blind tests of accuracy. Most
important for paleodemographic reconstruction, bias was acceptably low;
therefore, unknown age distributions could be approximated quite well by
such multifactorial methods as long as actual adult life expectancy is low, that
is, e15 is less than about 30 years (Lovejoy et al 1985a).
This precision was attributed incorrectly to an artificial reduction in varia-
tion in the cadaver test population (Cleveland’s Hamann-Todd collection) by
Katz & Suchey (1986, 1989), and in turn by êcan (1989a) and êcan & Loth
(1989). We provided a large-sample analysis showing that the Hamann-Todd
collection is a superior resource when careful subselection procedures are ap-
 PALEODEMOGRAPHY 389

plied to ensure accuracy of age at death of the specimens used for analysis; in
addition, subselection did not result in reduced variation (Meindl et al 1990).
In summary, methods for estimating the age distributions of representative
cemeteries can be accurate provided that (a) seriation is applied, (b) biological
assessment of each aging site is used, rather than so-called comparison-
matching of specimens to a series of discrete morphotypes, and (c) reliance is
placed on a battery of proven adult age estimators.
At this time we would also offer the following amendment to our previous
research: As before, a weighted average or subjective combination of aging
methods should be used for all young and middle-aged adult skeletons. This
can be done according to an algorithm (as above) or in a general way, such as
the subjective estimation of clinical age (Lovejoy et al 1985a). Furthermore,
for those skeletons determined to be older than 40 years, new estimates should
be provided from dental attrition (if possible) and especially from the auricular
surface. Although these methods remain difficult to learn and apply and do not
often satisfy the specific needs of forensic specialists, few alternatives exist at
this time for estimating the ages of the oldest decedents in archaeological
populations—one of the most important methodological issues in paleode-
mography today.

MORTALITY MODELS AND REFERENCE

POPULATIONS
A compelling mathematical consistency in the pattern of adult mortality by
age has long been recognized, beginning with Gompertz’s work early in the
nineteenth century. After World War II there was a shift in approach away
from functional graduations toward reference sets of life tables (see Brass
1971 for discussion). These tables primarily supplemented incomplete demo-
graphic data from developing countries, especially those in tropical Africa.
Conventionally called models or model life tables, such reference tables are
based on the same basic assumption that governs mathematical models: A lim-
ited number of dimensions of variation exist for the shape of the “curve of
deaths.” The best known reference set was developed at Princeton’s Office of
Population Research (Coale & Demeny 1966, 1983). From more than 300 two-
sex life tables of the nineteenth and twentieth centuries, there emerged four
families of mortality profiles. For instance, life tables representing populations
still in the grasp of the European tuberculosis epidemic underlie the North se-
ries of models. An East and a South pattern also coalesced. The West family
took form as a residual collection after the three other regions were removed.
The West reference set has become the regular choice in most anthropological
applications.
390 MEINDL & RUSSELL

Many anthropologists have used the models (accessible by a continuum of

either intrinsic growth rates or gross reproductive rates) to examine and ex-
plain fundamental processes as they might apply to extinct populations (Sat-
tenspiel & Harpending 1983, Gage 1988, Horowitz & Armelagos 1988, Buik-
stra & Konigsberg 1985, Dumond 1997, Sullivan 1997). However, others have
invoked the modern patterns of longevity in efforts to dismiss certain archaeo-
logical populations from the study of human demography altogether (Howell
1982, Johansson & Horowitz 1986, Milner et al 1989, Paine 1989). Not only
are the survivorship levels of most skeleton-based demographies extremely
low, but their patterns stand in marked contrast to the lowest survivorship lev-
els of the West family. Since by modern standards the levels of childhood mor-
tality in paleodemography are low relative to adult mortality levels, some re-
searchers have inferred that paleodemographic tables must be incorrect.
Whether such conclusions are appropriate depends on the nature of the data
on which the Coale & Demeny tables were based. The historical mortality pro-
files that were used had to meet strict standards of accuracy. As a result only
7% of the database populations predated 1870, and of these not one was from
outside Europe (Coale & Demeny 1966:7). However, there remained “large
variations from the model patterns” especially in the under-10 and over-60 age
groups (Coale & Demeny 1966:12). This variation was especially evident in
the underrepresented areas of nineteenth-century peasant Europe and in the
underdeveloped countries in Asia, Africa, and Latin America. Brass (1971)
went on to examine many of the profiles not used in the regional set calcula-
tions, such as Mauritius 1942–1946 and Guyana 1945–1947, which had very
high adult death rates compared to those expected during childhood. He noted
that “social and environmental factors in particular populations could lead to
larger deviations from the model experience” (Brass 1971:95).
In any event, these deviations were lost under the mathematical restrictions
of the final two-parameter graduations (i.e. region and level). Such modeling
may have maximized demographic precision for applications to modern popu-
lations, but there remains too little variety in mortality patterns to justify using
these models for purposes of generalizing about mortality conditions in an-
thropological contexts.
Although the vital experiences of European-derived and European-accul-
turated populations during the past 100 years were recorded accurately, we
would argue that they also represented a unique period in human demographic
evolution. In particular, human longevity had never been so artificially or rap-
idly protracted. More important, the extension of the patterns derived from ta-
bles with very high life expectancies to populations with much lower values
represent extrapolations to levels well beyond the range of observed data
(Brass 1971, Weiss 1973). That is, there is no empirical basis to indicate that
the shapes of the mortality profiles for, say, Spain in 1900, Taiwan in 1921, or
 PALEODEMOGRAPHY 391

Sweden in 1851, for which life expectancy at birth exceeded 35 years, actually
approximate those of paleodemographic populations with lower life expectan-
cies (Coale & Demeny 1966:29).
Many alternatives are available for smoothing or completing life-table data.
Weiss’s (1973) tables are a little more flexible; however, they depend to some
degree on modern patterns. Gage (1988, 1989, 1998) has developed some haz-
ard models for use in anthropological contexts. These alternatives have some
valuable practical and theoretical properties, including the flexibility to
smooth life-table data without shoehorning them into modern patterns.

PREHISTORIC POPULATION GROWTH

A fundamental problem in paleodemography is that for any given prehistoric
cemetery (i.e. a distribution of ages at death), each of a continuum of stable
populations could have filled it with exactly the same age proportions. These
populations may range from high mortality/moderate fertility (with low intrin-
sic growth or decline) to moderate mortality/high fertility (with high growth),
and every gradation in between. The implications are important. For example,
what does a very young cemetery age distribution from the period of early
adoption of agriculture imply? A moderate increase in fertility signaling the
success of a cultigen-based economy and a healthy population growth rate is
one possibility. Another is exactly the opposite: a considerable increase in
childhood mortality, population decline, and the nutritional limitations of un-
diversified agricultural products. By itself, the cemetery could be interpreted
in either fashion or, more appropriately, in a whole range of fashions.
Bennett (1973), Weiss (1973), and Moore et al (1975) were the first to give
archaeologists the practical means under stable population theory for relating
the age distribution of a cemetery to a mortality profile through an intrinsic
growth rate (r, the Malthusian parameter). Weiss (1973) also outlined the cal-
culation of both an age-specific fertility function and the population’s age pro-
portions, or pyramid, under assumed conditions of stable growth. The problem
is that there is no direct cemetery evidence for fertility, age structure, or intrin-
sic growth rate. Therefore, in the parlance of elementary algebra, there are
more unknowns than equations here, and an exact solution is not possible with-
out more information. A given cemetery does not provide a solution in the
sense of one demographic profile but rather a continuum of solutions, each
point of which contains a mortality level and a fertility level, with an associ-
ated growth level.
Sattenspiel & Harpending (1983) looked at variation in life expectancy and
birthrate within several of the Coale & Demeny West models and offered an
interesting observation: For stable populations the cemetery age-distribution
determines the birthrate but makes no prediction of life expectancy. Mean age
392 MEINDL & RUSSELL

at death predicts the reciprocal of the crude birthrate nearly exactly for all (hu-
manly possible) values of the intrinsic rate of growth (i.e. −1% < r < +4.0% per
year). In other words, the cemetery fixes the crude birthrate, but both mortality
and growth are anybody’s guess. Johansson & Horowitz (1986) restated this
finding and explored more closely the small departures from linearity of the re-
lationships between mean age and life expectancy. Horowitz & Armelagos
(1988) showed that the closeness of these approximations is limited to the
kinds of conditions that only paleodemographers examine, that is, heavy early
mortality.
If fertility is fixed by the cemetery, how can demographers choose a spe-
cific mortality level on the continuum of solutions? Bennett’s (1973) analysis
of Point of Pines, Asch’s (1976) approach to Middle Woodland groups in the
Lower Illinois Valley, and Muller’s (1997) models for the dynamics of late
prehistoric populations in the Eastern Woodlands all argued for solutions
based on a hypothesized growth rate. However, archaeological support for a
single value is exceedingly difficult to find (Horowitz & Armelagos 1988).
Life expectancy at birth ( e 0 ) is a function of mortality only. By contrast, the
crude birth rate (b) depends on age-specific fertilities, the age structure of the
population, and even the mortality function. We suggest that a hypothesized
fertility (or restricted range) be used to complete a demographic reconstruction
whenever the use of a growth rate is not feasible. However, as Sattenspiel &
Harpending (1983) pointed out, the regression of b on r or on any mortality
variable is virtually horizontal. No numerical solution could emerge from a hy-
pothesized crude birth rate; moreover, only one value of b is possible for a
given cemetery.
A different fertility measure, one stripped of any influence of maternal mor-
tality or age structure, may be more appropriate. The total fertility rate (TFR) is
a measure of completed fertility performance only and has been emphasized in
general frameworks by Howell (1976), Roth (1992), Harpending (1997), and
Keckler (1997). The TFR represents the average number of live births to
women who live to the end of the reproductive span (age 50 years), that is, the
sum of the age-specific fertility rates. For purposes of estimation this measure
increases too slowly with r in the vicinity of stationarity (i.e. zero growth), but
it increases geometrically after growth rates exceed about 1%. Therefore, one
problem with this approach is that the only exact solutions possible are high-
growth solutions. Another problem is the issue of choosing a value for the
TFR, because its variance in anthropological populations is surprisingly large
(Wood 1990), an observation that may one day force a reconsideration of the
meaning of the term “natural fertility.”
An example of this approach is our analysis of the Late-Archaic Ward site
cemetery in Kentucky (15McL11), for which we found a mean age of death in
the mid-20s (Meindl et al 1998). A hypothesized value of 6.5 children for the
 PALEODEMOGRAPHY 393

TFR determined both a new life expectancy in the mid-30s and an average
nonzero growth rate of 2.5% (It was assumed that only the periods of growth
were represented in the cemetery). A value of 2.5% per annum is high but not
unusual for well-censused primitive Amerindian populations of the twentieth
century (Weiss 1975, Hill & Hurtado 1996). This analysis corresponded to a
consensus among Eastern Woodland archaeologists that the first population
explosion in Kentucky took place in the Late Archaic (Griffin 1967, Jefferies
1996).

CONCLUSION
Some of the criticisms leveled recently at osteological paleodemography have
been justified. Too many studies had been undertaken using unacceptable
skeletal samples. Some demographies were based on skeletal age indicators
whose only merit was that they correlated in some way with increasing age.
The fundamental observation by modern demographers that fertility changes
tend to have far larger impacts on age structure than mortality had gone largely
unappreciated by our field. As a result anthropologists had to begin again with
what Larry Angel called the “Bases of Paleodemography” (1969).
We have limited our review to the advances in the most fundamental ele-
ment of this field, the consideration of age-at-death distributions. And while
there have been improvements in the field, there is more work to be done. First,
skeletal aging methods for the oldest decedents in a cemetery must be further
developed and refined, and we must recognize the fact that many traditional
aging methodologies will never be able to contribute here. Second, paleode-
mographers must borrow fertility and growth parameters from living anthro-
pological populations more freely in order to complete their inferences from
skeletal age-at-death distributions. And last, paleodemography, a “valid, spe-
cialized subfield of demography” (Roth 1992:175), should now assert itself.
Its data should not be forced into modern industrialized demographic profiles
without some empirical justification. If the demographic patterns of prehistory
were fundamentally different, archaeological demographers should reserve
the opportunity to detect them.
ACKNOWLEDGMENTS
We thank Owen Lovejoy and Alan Swedlund for critical readings of this
manuscript. Henry Harpending gave helpful comments on the “Growth” sec-
tion. These scholars have been influential in the field of paleodemography and
have improved this paper.

Visit the Annual Reviews home page at

http://www.AnnualReviews.org.
394 MEINDL & RUSSELL

Literature Cited
Acsádi G, Nemeskéri J. 1970. History of Hu- arc of the human pubis. Am. J. Phys. An-
man Life Span and Mortality. Budapest: thropol. 82:73–79
Hung. Acad. Soc. 346 pp. Buikstra JE. 1997. Paleodemography: context
Ahlqvist J, Damsten O. 1969. A modification and promise. See Paine 1997, pp. 367–80
of Kerley’s method for the microscopic de- Buikstra JE, Konigsberg LW. 1985. Paleode-
termination of age in bone. J. Forensic Sci. mography: critiques and controversies.
14:205–12 Am. Anthropol. 87:316–33
Aiello LC, Molleson T. 1993. Are microscopic Buikstra JE, Konigsberg LW, Bullington J.
ageing techniques more accurate than 1986. Fertility and the development of ag-
macroscopic ageing techniques? J. Ar- riculture in the prehistoric midwest. Am.
chaeol. Sci. 20:689–704 Antiq. 51:528–46
Angel JL. 1969. The bases of paleodemogra- Buikstra JE, Mielke JH. 1985. Demography,
phy. Am. J. Phys. Anthropol. 30:427–37 diet, and health. In The Analysis of Prehis-
Arsuaga JL, Carretero JM. 1994. Multivariate toric Diets, ed. RI Gilbert, JH Mielke, pp.
analysis of the sexual dimorphism of the 359–422. Orlando, FL: Academic
hip bone in a modern human population Coale AJ, Demeny P. 1966. Regional Model
and in early hominids. Am. J. Phys. An- Life Tables and Stable Populations.
thropol. 93:241–57 Princeton, NJ: Princeton Univ. Press. 871
Asch DL. 1976. The Middle Woodland Popu- pp.
lations of the Lower Illinois Valley: A Coale AJ, Demeny P, Vaughan B. 1983. Re-
Study in Paleodemographic Methods. gional Model Life Tables and Stable Popu-
Northwest. Univ. Archaeol. Program, Sci. lations. New York: Academic, 496 pp.
Pap. No. 1. 99 pp. Cohen MN. 1997. Does paleopathology meas-
Bang G. 1989. Age changes in teeth: develop- ure community health? A rebuttal of “The
mental and regressive. See êcan 1989b, osteological paradox” and its implications
pp. 211–35 for world history. See Paine 1997, pp.
Bedford ME, Russell KF, Lovejoy CO, 242–60
Meindl RS, Simpson SW, Stuart-Maca- Condon K, Charles DK, Cheverud JM, Buik-
dam PL. 1993. A test of the multifactorial stra JE. 1986. Cementum annulation and
aging method using skeletons with known age determination in Homo sapiens. II. Es-
ages at death from the Grant Collection. timates and accuracy. Am. J. Phys. Anthro-
Am. J. Phys. Anthropol. 91:287–97 pol. 71:321–30
Bennett KA. 1973. On the estimation of some Corruccini RS, Brandon EM, Handler JS.
demographic characteristics on a prehis- 1989. Inferring fertility from relative mor-
toric population from the American South- tality in historically controlled cemetery
west. Am. J. Phys. Anthropol. 39:223–32 remains from Barbados. Am. Antiq. 54:
Berrizbeitia BA. 1989. Sex determination with 609–14
the head of the radius. J. Forensic Sci. Demirjian A, Goldstein H, Tanner JM. 1973.
34:1206–13 A new system of dental age assessment.
Black TK. 1978. A new method for assessing Hum. Biol. 45:211–17
the sex of fragmentary skeletal remains: Demirjian A, Levesque G-Y. 1980. Sexual dif-
femoral shaft circumference. Am. J. Phys. ferences in dental development and predic-
Anthropol. 48:227–31 tion of emergence. J. Dent. Res. 59:
Bocquet-Appel JP, Masset C. 1982. Farewell 1110–26
to paleodemography. J. Hum. Evol. 11: Dittrick J, Suchey JM. 1986. Sex determina-
321–33 tion of prehistoric central California skele-
Bocquet-Appel JP, Masset C. 1996. Paleode- tal remains using discriminant analysis of
mography: expectancy and false hope. Am. the femur and humerus. Am. J. Phys. An-
J. Phys. Anthropol. 99:571–83 thropol. 70:3–9
Brass W. 1971. On the scale of mortality. In Divale W, Harris M. 1976. Population, war-
Biological Aspects of Demography, ed. W fare, and the male supremacist complex.
Brass, pp. 69–110. London: Taylor & Am. Anthropol. 78:521–38
Francis Dudar JC, Pfeiffer S, Saunders SR. 1993.
Brooks ST. 1955. Skeletal age at death: the re- Evaluation of morphological and histo-
liability of cranial and pubic age indica- logical adult skeletal age-at-death estima-
tors. Am. J. Phys. Anthropol. 13:567–89 tion techniques using ribs. J. Forensic Sci.
Budinoff LC, Tague RG. 1990. Anatomical 38:677–85
and developmental bases for the ventral Dumond DE. 1997. Seeking demographic
 PALEODEMOGRAPHY 395

causes for changes in population growth Hamilton ME. 1982. Sexual dimorphism in
rates. See Paine 1997, pp. 175–90 skeletal samples. In Sexual Dimorphism in
El-Nofely A, êcan MY. 1989. Assessment of Homo sapiens, ed. RL Hall, pp. 107–63.
age from the dentition in children. See New York: Praeger
êcan 1989b, pp. 237–54 Hanihara K, Suzuki T. 1978. Estimation of age
Falsetti AB. 1995. Sex assessment from meta- from the pubic symphysis by means of
carpals of the human hand. J. Forensic Sci. multiple regression analysis. Am. J. Phys.
40:774–76 Anthropol. 48:233–40
Fernández Camacho FJ, Gómez Pellico L, Hanna RE, Washburn SL. 1953. The determi-
Fernández-Valencia Rodríguez R. 1993. nation of the sex of skeletons, as illustrated
Osteometry of the human iliac crest: pat- by a study of the Eskimo pelvis. Hum. Biol.
terns of normality and its utility in sexing 25:21–27
human remains. J. Forensic Sci. 38: Hanson DB, Buikstra JE. 1987. Histomorpho-
779–87 logical alteration in buried human bone
Frost HM. 1987. Secondary osteon popula- from the Lower Illinois Valley: implica-
tions: algorithm for determining mean tions for paleodietary research. J. Ar-
bone tissue age. Yearb. Phys. Anthropol. chaeol. Sci. 14:549–63
30:221–38 Harpending H. 1997. Living records of past
Gage TB. 1988. Mathematical hazard models population change. See Paine 1997, pp.
of mortality: an alternative to model life ta- 89–100
bles. Am. J. Phys. Anthropol. 76:429–41 Harrison GA, Weiner JS, Tanner JM, Barnicot
Gage TB. 1989. Bio-mathematical approaches NA. 1977. Human Biology. Oxford: Ox-
to the study of human variation in mortal- ford Univ. Press. 499 pp. 2nd ed.
ity. Yearb. Phys. Anthropol. 32:185–214 Hill K, Hurtado AM. 1996. Ache Life History.
Gage TB. 1998. Aggregate patterns of varia- New York: Aldine de Gruyter. 561 pp.
tion in human demography. Annu. Rev. An- Holland TD. 1991. Sex assessment using the
thropol. 27:197–221 proximal tibia. Am. J. Phys. Anthropol.
Garland AN. 1987. A histological study of ar- 85:221–27
chaeological bone decomposition. In Holman DJ, Bennett KA. 1991. Determination
Death, Decay and Reconstruction: Ap- of sex from arm bone measurements. Am.
proaches to Archaeology and Forensic J. Phys. Anthropol. 84:421–26
Science, ed. A Boddington, AC Garland, Horowitz S, Armelagos G, Wachter K. 1988.
RC Janaway, pp. 109–26. Manchester, On generating birth rates from skeletal
UK: Manchester Univ. Press populations. Am. J. Phys. Anthropol. 76:
Gilbert BM, McKern TW. 1973. A method for 189–96
aging the female os pubis. Am. J. Phys. An- Howell N. 1976. Toward a uniformitarian the-
thropol. 38:31–38 ory of human paleodemography. In The
Giles E, Elliot O. 1963. Sex determination by Demographic Evolution of Human Popu-
discriminant function analysis of crania. lations, ed. RH Ward, KM Weiss, pp.
Am. J. Phys. Anthropol. 21:53–68 25–40. New York: Academic
Giles E, Klepinger LL. 1988. Confidence in- Howell N. 1982. Village composition implied
tervals for estimates based on linear re- by a paleodemographic life table. Am. J.
gression in forensic anthropology. J. Fo- Phys. Anthropol. 59:263–69
rensic Sci. 33:1218–22 Hrdli¹ka A. 1952. Practical Anthropometry.
Goodman A, Lallo J, Armelagos GJ, Rose JC. Philadelphia, PA: Wistar Inst. Anat. Biol.
1984. Health changes at Dickson Mounds, 241 pp. 4th ed.
Illinois (A.D. 950–1300). In Paleopathol- êcan MY. 1989a. Research strategies in age
ogy at the Origins of Agriculture, ed. MN estimation: the multiregional approach.
Cohen, GJ Armelagos, pp. 271–306. New See êcan 1989b, pp. 325–39
York: Academic êcan MY, ed. 1989b. Age Markers in the Hu-
Gordon CC, Buikstra JE. 1981. Soil pH, bone man Skeleton. Springfield, IL: Thomas
preservation, and sampling bias at mortu- êcan MY, Loth SR. 1986. Estimation of age
ary sites. Am. Antiq. 46:566–71 and determination of sex from the sternal
Griffin JB. 1967. Eastern North American ar- rib. In Forensic Osteology: Advances in
chaeology: a summary. Science 156: the Identification of Human Remains, ed.
175–91 KJ Reichs, pp. 68–89. Springfield, IL:
Gustafson G. 1950. Age determination of Thomas
teeth. J. Am. Dent. Assoc. 41:45–54 êcan MY, Kennedy KAR, eds. 1989. Recon-
Gustafson G, Koch G. 1974. Age estimation struction of Life from the Skeleton. New
up to 16 years of age based on dental devel- York: Liss
opment. Odontol. Revy 25:297–306 êcan MY, Loth SR. 1989. Osteological mani-
396 MEINDL & RUSSELL

festations of age in the adult. See êcan & Patterns of dental wear in the Lengua Indi-
Kennedy 1989, pp. 23–40 ans of Paraguay. Am. J. Phys. Anthropol.
êcan MY, Loth SR, Wright RK. 1984a. Meta- 66:21–29
morphosis at the sternal rib end: a new Kieser JA, Moggi-Cecchi J, Groeneveld HT.
method to estimate age at death in white 1992. Sex allocation of skeletal material
males. Am. J. Phys. Anthropol. 65:147–56 by analysis of the proximal tibia. J. Foren-
êcan MY, Loth SR, Wright RK. 1984b. Age sic Sci. 56:29–36
estimation from the rib by phase analysis: Konigsberg LW, Frankenberg SR. 1992. Esti-
white males. J. Forensic Sci. 29:1094–104 mation of age structure in anthropological
êcan MY, Loth SR, Wright RK. 1987. Racial demography. Am. J. Phys. Anthropol. 89:
variation in the external extremity of the 235–56
rib and its effect on age determination. J. Konigsberg LW, Frankenberg SR. 1994. Pa-
Forensic Sci. 32:452–66 leodemography: “Not quite dead.” Evol.
Israel H. 1973. Age factor and the pattern of Anthropol. 3:92–105
change in craniofacial structures. Am. J. Konigsberg LW, Frankenberg SR, Walker
Phys. Anthropol. 39:111–28 RB. 1997. Regress what on what? Paleode-
Israel H. 1977. The dichotomous pattern of mographic age estimation as a calibration
craniofacial expansion during aging. Am. problem. See Paine 1997, pp. 64–88
J. Phys. Anthropol. 47:47–52 Krogman WM. 1962. The Human Skeleton in
Jackes MK. 1985. Pubic symphysis age distri- Forensic Medicine. Springfield, IL: Tho-
butions. Am. J. Phys. Anthropol. 68: mas. 337 pp.
281–99 Krogman WM, êcan MY. 1986. The Human
Jackes MK. 1992. Paleodemography: prob- Skeleton in Forensic Medicine. Spring-
lems and techniques. See Saunders & field, IL: Thomas. 2nd ed. 568 pp.
Katzenberg 1992, pp. 189–224 Lazenby RA. 1994. Identification of sex from
Jackes MK. 1993. On paradox and osteology. metacarpals: effect of side asymmetry. J.
Curr. Anthropol. 34:434–39 Forensic Sci. 39:1188–94
Jefferies RW. 1996. Hunters and gatherers af- Leslie P, Gage TB. 1989. Demography and hu-
ter the Ice Age. In Kentucky Archaeology, man population biology: problems and
ed. RB Lewis, pp. 39–77. Lexington: progress. In Human Population Biology: A
Univ. Press Kentucky Trandisciplinary Science, ed. M Little, JD
Johansson SR, Horowitz S. 1986. Estimating Haas, pp. 15–44. New York: Oxford Univ.
mortality in skeletal populations: influ- Press
ence of the growth rate on the interpreta- Leutenegger W. 1982. Encephalization and
tion of levels and trends during the transi- obstetrics in primates with particular refer-
tion to agriculture. Am. J. Phys. Anthropol. ence to human evolution. In Primate Brain
71:233–50 Evolution, ed. E Armstrong, D Falk, pp.
Johnstone FE, Zimmer LO. 1989. Assessment 85–95. New York: Plenum
of growth and age in the immature skele- Lovejoy CO. 1985. Dental wear in the Libben
ton. See êcan & Kennedy 1989, pp. 11–21 population: its functional pattern and role
Kajanoja P. 1966. Determination of Finnish in the determination of adult skeletal age at
crania by discriminant function analysis. death. Am. J. Phys. Anthropol. 68:47–56
Am. J. Phys. Anthropol. 24:29–34 Lovejoy CO, Meindl RS, Mensforth RP, Bar-
Katz D, Suchey JM. 1986. Age determination ton TJ. 1985a. Multifactorial determina-
of the male Os pubis. Am. J. Phys. Anthro- tion of skeletal age at death: a method and
pol. 69:427–35 blind tests of its accuracy. Am. J. Phys. An-
Katz D, Suchey JM. 1989. Race differences in thropol. 68:1–14
pubic symphyseal aging patterns in the Lovejoy CO, Meindl RS, Pryzbeck TR, Barton
male. Am. J. Phys. Anthropol. 80:167–72 TS, Heiple KG, Kotting D. 1977. Paleode-
Keckler CNW. 1997. Catastrophic mortality mography of the Libben site, Ottawa
in simulations of forager age at death: County, Ohio. Science 198:291–93
Where did all the humans go? See Paine Lovejoy CO, Meindl RS, Pryzbeck TR, Mens-
1997, pp. 205–28 forth RP. 1985b. Chronological metamor-
Kerley ER. 1965. The microscopic determina- phosis of the auricular surface of the ilium:
tion of age in human bone. Am. J. Phys. a new method for the determination of
Anthropol. 23:149–64 adult skeletal age at death. Am. J. Phys.
Kerley ER, Ubelaker DH. 1978. Revisions in Anthropol. 68:15–28
the microscopic method of estimating age Lovejoy CO, Meindl RS, Tague RG, Latimer
at death in human cortical bone. Am. J. B. 1995. The senescent biology of the
Phys. Anthropol. 49:545–46 hominoid pelvis: its bearing on the pubic
Kieser JA, Groeneveld HT, Preston CB. 1985. symphysis and auricular surface as age-at-
 PALEODEMOGRAPHY 397

death indicators in the human skeleton. a review and tests of accuracy of other cur-
Riv. Antropol. 73:31–49 rent methods of pubic symphyseal aging.
Lovejoy CO, Meindl RS, Tague RG, Latimer Am. J. Phys. Anthropol. 68:29–45
B. 1997. The comparative senescent biol- Meindl RS, Mensforth RP, York H. 1998.
ogy of the hominoid pelvis and its implica- Mortality, fertility, and growth in the Ken-
tions for the use of age-at-death indicators tucky Archaic: the paleodemography of
in the human skeleton. See Paine 1997, pp. the Ward site, McLean County. In
43–63 Hunters-Gatherers to Horticulturalists,
Lovell NC. 1989. Test of Phenice’s technique ed. K Vickery. Columbus, OH: Ohio Ar-
for determining sex from the Os pubis. Am. chaeological Council. In press.
J. Phys. Anthropol. 79:117–20 Meindl RS, Russell KF, Lovejoy CO. 1990.
Lucy D, Aykroyd RG, Pollard AM, Solheim T. Reliability of age at death in the Hamann-
1996. A Bayesian approach to adult human Todd collection: validity of subselection
age estimation from dental observations by procedures used in blind tests of the sum-
Johanson’s age changes. J. Forensic Sci. mary age technique. Am. J. Phys. Anthro-
41(2):189–94 pol. 83:349–57
Luo Y-C. 1995. Sex determination from the Mensforth RP. 1990. Paleodemography of the
pubis by discriminant function analysis. Carlston-Annis (Bt-5) Late Archaic skele-
Forensic Sci. Int. 74:89–98 tal population. Am. J. Phys. Anthropol. 82:
Maples WR. 1978. An improved technique us- 81–99
ing dental histology for estimation of adult Mensforth RP, Lovejoy CO. 1985. Anatomi-
age. J. Forensic Sci. 23:764–70 cal, physiological, and epidemiological
Maples WR, Rice PM. 1979. Some difficulties correlates of the aging process: a confirma-
in the Gustafson dental age estimations. J. tion of multifactorial age determination in
Forensic Sci. 24:168–72 the Libben skeletal population. Am. J.
Marino EA. 1995. Sex estimation using the Phys. Anthropol. 68:87–106
first cervical vertebra. Am. J. Phys. An- Miles AEW. 1963. The dentition in the assess-
thropol. 97:127–33 ment of individual age in skeletal materi-
Masset C. 1973. Influence du sexe et de l’âge als. In Dental Anthropology, ed. DR
sur la conservation des os humains. In Brothwell, pp. 191–209. New York: Per-
L’Homme, Hier et Aujourd’hiu: Recueil gamon
d’études en Hommage à André Leroi- Milner GR, Humpf DA, Harpending HC.
Gourhan, ed. M Sauter, pp. 333–43. Paris: 1989. Pattern matching of age-at-death
Cujas distributions in paleodemographic analy-
Masset C. 1989. Age estimation on the basis of sis. Am. J. Phys. Anthropol. 80:49–58
cranial sutures. See êcan 1989b, pp. Moore JA, Swedlund AC, Armelagos GJ.
71–103 1975. The use of life tables in paleode-
McKern TW, Stewart TD. 1957. Skeletal age mography. Am. Antiq. 40:57–70
changes in young American males. Quar- Moorrees CFA, Fanning EA, Hunt EE. 1963a.
termaster Res. and Dev. Command Tech. Formation and resorption of three decidu-
Rep. EP–45. Natick, MA ous teeth in children. Am. J. Phys. Anthro-
Meindl RS, Lovejoy CO. 1985. Ectocranial pol. 21:205–13
suture closure: a revised method for the de- Moorrees CFA, Fanning EA, Hunt EE. 1963b.
termination of skeletal age at death based Age variation of formation stages for ten
on the lateral-anterior sutures. Am. J. Phys. permanent teeth. J. Dent. Res. 42:
Anthropol. 68:57–66 1490–1502
Meindl RS, Lovejoy CO. 1989. Age changes Muller J. 1997. Native Eastern American
in the pelvis: implications for paleode- population continuity and stability. See
mography. See êcan 1989b, pp. 137–68 Paine 1997, pp. 343–64
Meindl RS, Lovejoy CO, Mensforth RP. 1983. Murphy TR. 1959. The changing pattern of
Skeletal age at death: accuracy of determi- dentine exposure in human tooth attrition.
nation and implications for human demog- Am. J. Phys. Anthropol. 17:167–85
raphy. Hum. Biol. 55:73–87 Murray KA, Murray TM. 1991. A test of the
Meindl RS, Lovejoy CO, Mensforth RP, Don auricular surface aging technique. J. Fo-
Carlos L. 1985a. Accuracy and direction of rensic Sci. 36:1162–69
error in the sexing of the skeleton: implica- Oliver G. 1969. Practical Anthropology.
tions for paleodemography. Am. J. Phys. Springfield, IL: Thomas. 330 pp.
Anthropol. 68:79–85 Owsley DW, Webb RS. 1983. Misclassifica-
Meindl RS, Lovejoy CO, Mensforth RP, tion probability of dental discriminant
Walker RA. 1985b. A revised method of functions for sex determination. J. Foren-
age determination using the os pubis, with sic Sci. 28:181–85
398 MEINDL & RUSSELL

Paine RR. 1989. Model life table fitting by Biology of Past Peoples: Research Meth-
maximum likelihood estimation: a proce- ods. New York: Wiley-Liss
dure to reconstruct paleodemographic Scheuer JL, Elkington NM. 1993. Sex deter-
characteristics from skeletal age distribu- mination from metacarpals and the first
tions. Am. J. Phys. Anthropol. 79:51–62 proximal phalanx. J. Forensic Sci. 38:
Paine RR, ed. 1997. Integrating Archaeologi- 769–78
cal Demography: Multidisciplinary Ap- Schulter-Ellis FP, Schmidt DJ, Hayek IA,
proaches to Prehistoric Population. Car- Craig J. 1983. Determination of sex with a
bondale: South. Ill. Univ. 395 pp. discriminant analysis of new pelvic bone
Pfeiffer S. 1986. Morbidity and mortality in measurements: part I. J. Forensic Sci. 28:
the Uxbridge ossuary. Can. J. Anthropol. 169–80
5:23–31 Schwartz JH. 1995. Skeleton Keys: An Intro-
Pfeiffer S. 1992. Cortical bone age estimates duction to Human Skeletal Morphology,
from historically known adults. Z. Mor- Development, and Analysis. Oxford: Ox-
phol. Anthropol. 79:1–10 ford Univ. Press. 362 pp.
Phenice TW. 1969. A newly developed visual Scrimshaw SCM. 1984. Infanticide in human
method of sexing the Os pubis. Am. J. populations: societal and individual con-
Phys. Anthropol. 30:297–301 cerns. In Infanticide: Comparative and
Richman EA, Michel ME, Schulter-Ellis FP, Evolutionary Perspectives, ed. G Haus-
Corruccini RS. 1979. Determination of sex fater, SB Hrdy, pp. 439–62. New York:
by discriminant function analysis of post- Aldine
cranial skeletal measurements. J. Forensic Singer R. 1953. Estimation of age from cranial
Sci. 24:159–67 suture closure. J. Foren. Med. 1:52–59
Rogers T. 1990. A test of the auricular surface Singh IJ. 1972. Femoral cortical trabecular
method of estimating age-at-death and a pattern of the upper end of the femur as an
discussion of its usefulness in the construc- index of osteoporosis. J. Bone J. Surg. A
tion of paleodemographic lifetables. Pre- 52:457–67
sented at 18th Annu. Meet. Can. Assoc. Singh IJ, Gunberg DL. 1970. Estimation of
Phys. Anthropol., Banff, Alberta age at death in human males from quantita-
Rogers T, Saunders S. 1994. Accuracy of sex tive histology of bone fragments. Am. J.
determination using morphological traits Phys. Anthropol. 33:373–82
of the human pelvis. J. Forensic Sci. 39: Siven CH. 1991. On estimating mortalities
1047–56 from osteological age data. Int. J. Anthro-
Roth EA. 1992. Applications of demographic pol. 6:97–110
models to paleodemography. See Saunders Smith BH. 1984. Patterns of molar wear in
& Katzenberg 1992, pp. 175–88 hunter-gatherers and agriculturalists. Am.
Russell KF. 1996. Determination of age-at- J. Phys. Anthropol. 63:39–56
death from dental remains. PhD thesis. Song H-W, Lin QL, Jia JT. 1992. Sex diagno-
Kent State Univ., Ohio sis of Chinese skulls using multiple step-
Russell KF, Simpson SW, Genovese J, wise discriminant function analysis. Fo-
Kinkel MD, Meindl RS, Lovejoy CO. rensic Sci. Int. 54:135–40
1993. Independent test of the fourth rib ag- Steele DG. 1976. The estimation of sex on the
ing technique. Am. J. Phys. Anthropol. 92: basis of the talus and calcaneous. Am. J.
53–62 Phys. Anthropol. 45:581–88
Sattenspiel L, Harpending H. 1983. Stable Stewart TD. 1979. Essentials of Forensic An-
populations and skeletal age. Am. Antiq. thropology: Especially as Developed in
48:489–98 the United States. Springfield, IL: Tho-
Saunders SR. 1992. Subadult skeletons and mas. 300 pp.
growth related studies. See Saunders & Stini WA. 1985. Growth rates and sexual di-
Katzenberg 1992, pp. 1–20 morphism in evolutionary perspective. In
Saunders SR, DeVito C, Herring DA, South- The Analysis of Prehistoric Diets, ed. RI
ern R, Hoppa R. 1993. Accuracy tests of Gilbert, JH Mielke, pp. 191–226. Orlando,
tooth formation age estimations for human FL: Academic
skeletal remains. Am. J. Phys. Anthropol. Stout SD. 1992. Methods of determining age
92:173–88 at death using bone microstructure. See
Saunders SR, Fitzgerald C, Rogers T, Dudar Saunders & Katzenberg 1992, pp. 21–35
JC, McKillop H. 1992. A test of several Stout SD, Paine RR. 1992. Histological age es-
methods of skeletal age estimation using a timation using rib and clavicle. Am. J.
documented archaeological sample. Can. Phys. Anthropol. 87:111–15
Soc. Forensic Sci. 25:97–118 Suchey JM, Wiseley DV, Katz D. 1986.
Saunders SR, Katzenberg MA. 1992. Skeletal Evaluation of the Todd and McKern-
 PALEODEMOGRAPHY 399

Stewart methods for aging the male os pu- death have been greatly exaggerated. J.
bis. In Forensic Osteology, ed. KJ Reichs, Hum. Evol. 12:353–60
pp. 33–67. Springfield, IL: Thomas Waldron T. 1987. The relative survival of the
Sullivan NC. 1997. Contact period Huron de- human skeleton: implications for paleopa-
mography. See Paine 1997, pp. 327–42 thology. In Death, Decay, and Reconstruc-
Sundick RI. 1978. Human skeletal growth and tion: Approaches to Archaeology and Fo-
age determination. Homo 29:228–49 rensic Science, ed. A Boddington, AN
Sutherland LD, Suchey JM. 1991. Use of the Garland, RC Janeway, pp. 55–64. Man-
ventral arc in pubic sex determination. J. chester, UK: Manchester Univ. Press
Forensic Sci. 36:501–11 Walker PL, Johnson JR, Lambert PM. 1988.
Tague RG. 1989. Variation in pelvic size be- Age and sex biases in the preservation of
tween males and females. Am. J. Phys. An- human skeletal remains. Am. J. Phys. An-
thropol. 80:59–71 thropol. 76:183–88
Tague RG. 1991. Commonalities in dimor- Walker RA. 1989. Assessments of human cor-
phism and variability in the anthropoid tical bone dynamics and skeletal age at
pelvis, with implications for the fossil rec- death from femoral cortical histomorphol-
ord. J. Hum. Evol. 21:153–76 ogy. PhD thesis. Kent State Univ., Ohio
Tague RG, Lovejoy CO. 1986. The obstetric Walker RA, Lovejoy CO. 1985. Radiographic
pelvis of A.L.288-1 (Lucy). J. Hum. Evol. changes in the clavicle and proximal femur
15:237–55 and their use in the determination of skele-
Thompson DD. 1979. The core technique in tal age at death. Am. J. Phys. Anthropol.
the determination of age at death in skele- 68:67–78
tons. J. Forensic Sci. 24:902–15 Walker RA, Lovejoy CO, Meindl RS. 1994.
Todd TW. 1920. Age changes in the pubic Histomorphological and geometric prop-
bone: I. The white male pubis. Am. J. Phys. erties of human femoral cortex in individu-
Anthropol. 3:285–34 als over 50: implications for determination
Todd TW. 1921a. Age changes in the pubic of age-at-death. Am. J. Hum. Biol. 6:
bone: II. The pubis of the male Negro- 659–67
white hybrid; III. The pubis of the white fe- Weiss KM. 1972. On the systematic bias in
male; IV. The pubis of the female Negro- skeletal sexing. Am. J. Phys. Anthropol.
white hybrid. Am. J. Phys. Anthropol. 4: 37:239–50
1–70 Weiss KM. 1973. Demographic Models for
Todd TW. 1921b. Age changes in the pubic Anthropology. Mem. 27 Soc. Am. Ar-
bone: V. Mammalian pubic metamorpho- chaeol.; also Am. Antiq. 38:2, Part II. 186
sis. Am. J. Phys. Anthropol. 4:333–406 pp.
Townsend N, Hammel EA. 1990. Age estima- Weiss KM. 1975. The application of demo-
tion from the number of teeth erupted in graphic models to anthropological data.
young children: an aid to demographic sur- Hum. Ecol. 3:87–103
veys. Demography 27:165–74 Weiss KM. 1989. A survey of human biode-
Ubelaker DH. 1989. Human Skeletal Remains: mography. J. Quant. Anthropol. 1:79–151
Excavation, Analysis, and Interpretation. White TD. 1991. Human Osteology. San Di-
Washington, DC: Taraxacum. 116 pp. 2nd ego: Academic. 455 pp.
ed. Wood JW. 1990. Fertility in anthropological
Uesugi Y, Taguchi O, Noumura T, Iguchi T. populations. Annu. Rev. Anthropol. 19:
1992. Effects of sex steroids on the devel- 211–42
opment of sexual dimorphism in mouse in- Wood JW, Milner GR, Harpending HC, Weiss
nominate bone. Anat. Rec. 234:541–48 KM. 1992. The osteological paradox.
Van Gerven DP, Armelagos GJ. 1983. Fare- Curr. Anthropol. 33:343–70
well to paleodemography? Rumors of its