Vous êtes sur la page 1sur 29

The Osteological Paradox: Problems of Inferring Prehistoric Health from Skeletal Samples

[and Comments and Reply]

Author(s): James W. Wood, George R. Milner, Henry C. Harpending, Kenneth M. Weiss,
Mark N. Cohen, Leslie E. Eisenberg, Dale L. Hutchinson, Rimantas Jankauskas, Gintautas
Cesnys, Gintautas esnys, M. Anne Katzenberg, John R. Lukacs, Janet W. McGrath, Eric
Abella Roth, Douglas H. Ubelaker and Richard G. Wilkinson
Source: Current Anthropology, Vol. 33, No. 4 (Aug. - Oct., 1992), pp. 343-370
Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for
Anthropological Research
Stable URL: http://www.jstor.org/stable/2743861
Accessed: 14-11-2016 16:43 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

The University of Chicago Press, Wenner-Gren Foundation for Anthropological

Research are collaborating with JSTOR to digitize, preserve and extend access to Current Anthropology

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October i992
? i992 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved ooII-3204/92-3304-OOOI$2.50

i982. He has been curator of the Museum of Anthropology of the

University of Kentucky and has conducted archaeological field-

The Osteological work in Illinois. He has published "Social and Temporal Implica-
tions of Variation among American Bottom Mississippian Ceme-
teries" (American Antiquity 49:468-88), "The Late Prehistoric

Paradox Cahokia Cultural System of the Mississippi River Valley: Foun-

dations, Florescence, and Fragmentation" (Journal of World Pre-
history 4:I-43), and (with D. H. Humpf and H. C. Harpending)
"Pattern Matching of Age-at-Death Distributions in Paleodemo-
graphic Analysis" (American Journal of Physical Anthropology
Problems of Inferring Prehistoric 80:49-5 8).

HENRY C. HARPENDING is Professor in the Department of An-

Health from Skeletal Samples' thropology and the Graduate Program in Genetics at Pennsylva-
nia State University. He was born in I944 and educated at Ham-
ilton College (B.A., i964) and Harvard University (Ph.D., I972).
He taught at the University of New Mexico from I972 to I985.
by James W. Wood, Among his publications are (with L. Sattenspiel) "Stable Popula-
tions and Skeletal Age" (American Antiquity 48:489-98), (with
George R. Milner, A. Rogers and P. Draper) "Human Sociobiology" (Yearbook of
Physical Anthropology 30: I27-50), (with R. Pennington) "Age
Structure and Sex-biased Mortality among Herero Pastoralists"
Henry C. Harpending, (Human Biology 63:327-52), and (with A. Rogers) "Population
Growth Makes Waves in the Distribution of Pairwise Nucleotide
and Kenneth M. Weiss Differences" (Molecular Biology and Evolution, in press).
KENNETH M. WEISS iS Professor of Genetics and Anthropology at
Pennsylvania State University. Born in I94I, he received his B.A.
from Oberlin College in I963 and his Ph.D. from the University
of Michigan in I972. He taught at the Center for Demographic
Paleodemography and paleopathology presuppose that direct rela- and Population Genetics, Graduate School of Biomedical Sci-
tionships exist between statistics calculated from archaeological ences and School of Public Health, University of Texas, Houston,
skeletal series (e.g., skeletal lesion frequencies and mean age at from I973 to I984. His publications include Demographic Mod-
death) and the health status of the past populations that gave rise els for Anthropology (Washington, D.C.: Society for American
to the series. However, three fundamental conceptual problems Archaeology, I973), (with R. E. Ferrell and C. L. Hanis) "A New
confound the interpretation of such statistics: demographic non- World Syndrome of Metabolic Diseases with a Genetic and Evo-
stationarity, selective mortality, and unmeasured, individual- lutionary Basis" (Yearbook of Physical Anthropology 27:15 3-78),
level heterogeneity in the risks of disease and death. Using sim- "A Survey of Human Biodemography" (Journal of Quantitative
ple models of the relationship between individual "frailty" and Anthropology I :79-I 5 I), and The Genetic Causes of Human
the hazard of death at each age, this paper explores the implica- Disease (Cambridge: Cambridge University Press, in press).
tions of these problems for archaeological interpretation. One
conclusion is that the skeletal evidence pertaining to the transi- The present paper was submitted in final form 28 XI 9I.
tion from hunting-and-gathering to settled agriculture is equally
consistent with an improvement in health and a deterioration in
health resulting from the transition.
Measurement and interpretation of differences in the
JAMES W. WOOD iS Associate Professor of Anthropology and Se-
level of health in prehistoric populations are two funda-
nior Research Associate of the Population Issues Research Center mental goals of paleodemography and paleopathology.
at Pennsylvania State University (University Park, Pa. i6802, Studies of such differences are vital to our understanding
U.S.A.). Born in I949, he was educated at Columbia University of how the adaptive success of human populations has
(B.A., I97I) and the University of Michigan (M.A., I975; Ph.D.,
varied through time and space. For the purposes of these
ig80). He has taught at Michigan and at the University of Wis-
consin and has done fieldwork in demography, reproductive biol- studies, several standard indices of morbidity and mor-
ogy, and population genetics in Papua New Guinea. His publica- tality derived from skeletal series (e.g., skeletal lesion
tions include "Fertility in Anthropological Populations" (Annual frequencies, life expectancies, and average age at death)
Review of Anthropology i9), "Fertility and Reproductive Biology would seem readily interpretable. Common sense sug-
in Papua New Guinea," in Human Biology in Papua New
Guinea, edited by R. D. Attenborough and M. P. Alpers (Oxford:
gests that there should be some reasonably direct associ-
Oxford University Press, i992), and Dynamics of Human Repro- ation between these aggregate-level measures and the
duction (Hawthorne: Aldine, in press). risks of illness and death experienced by the individual
GEORGE R. MILNER is Associate Professor and Museum Curator members of past populations. For example, an increase
in the Department of Anthropology at Pennsylvania State Uni- in the frequency of a particular type of skeletal lesion
versity. He was born in I953 and received his B.A. from Beloit might be interpreted as indicating an elevation in the
College in I975 and his Ph.D. from Northwestern University in
individual risks of experiencing the conditions that in-
duce the lesion and hence an increase in the population
i. A version of this paper was presented at the I989 meetings of prevalence of those conditions. Similarly, a decrease in
the Society for American Archaeology in Atlanta, Ga. We thank the average age at death might be taken as signaling
Eve Anderson, Gillian Bentley, Anne Buchanan, Timothy Gage, an elevation in the risks of death experienced by the
Darryl Holman, Patricia Johnson, Paul Leslie, Claire McHale Mil-
individuals in the once-living population. Most interpre-
ner, and Richard Paine for comments and suggestions. We also
thank the Illinois State Museum for permission to examine the tations of paleodemographic and paleopathological data
Oneota skeletal collection. presuppose that such straightforward relationships exist.


This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
344 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

We suggest that interpretation of these standard mea- summary statistics such as life expectancies or the mean
sures may be more difficult than it first appears, espe- age at death are often effectively measures of fertility
cially when dealing with skeletons from archaeological rather than mortality.
contexts. Failure to address three important conceptual The implications of this basic fact of mathematical
issues-demographic nonstationarity, selective mortal- demography for prehistoric skeletal samples have been
ity, and hidden heterogeneity in risks-can render in- discussed elsewhere (Sattenspiel and Harpending i983,
ferences based on various demographic and epidemio- Johannson and Horowitz i986, Milner, Humpf, and Har-
logical measures meaningless. These problems are, we pending i989, Paine i989). Paleodemographers have
believe, among the most difficult currently facing recon- started to pay some attention to this problem, although
structions of the demography and health of past popula- they still tend to underrate its importance (Cohen i989).
tions, far more difficult than the issues raised by Because the issue of demographic nonstationarity is not
Bocquet-Appel and Masset (i982) in their "Farewell to new, we concentrate in this paper on the other two prob-
Paleodemography" (indeed, their points have largely lems, selective mortality and hidden heterogeneity in
been addressed by Van Gerven and Armelagos [i 9831 and This treatment is convenient because those two
Buikstra and Konigsberg [i985], among others). problems are fundamentally related: in the absence of
This paper is not, however, a counsel of despair. Con- heterogeneity, selectivity does not occur.
tinual reexamination of underlying premises is essential The selective mortality problem is easy to state but
to any vigorous field of research. In the case of paleo- difficult to solve. There is one, and perhaps only one,
pathology, studies of ancient bone have undergone sig- irrefutable fact about the cases making up a skeletal se-
nificant changes in orientation and sophistication since ries: they are dead. We never have a sample of all the
the mid-ig6os (Armelagos, Goodman, and Jacobs I978, individuals who were at risk of disease or death at a
Buikstra and Cook I980, Huss-Ashmore, Goodman, and given age, but only of those who did in fact die at that
Armelagos i982, Ubelaker i982, Martin, Goodman, and age. For example, the only 2o-year-olds we observe in a
Armelagos I985, Larsen I987, Ortner and Aufderheide skeletal sample are those who died at age 2o. While we
I99I). During this period, the field has moved from a may find the skeletons of many of the other individuals
particularistic concern with individual lesions or skele- who had been at risk of death at age 2o but who died
tons to a population-based perspective on disease pro- later, say, at age 6o, we observe their characteristics as
cesses, the evolution of humans and their pathogens, 6o-year-olds, not 2o-year-olds. As a result, the sample of
and the evaluation of past community health, especially 2o-year-olds (or any other age-group) is highly selective
as it has changed with the adoption of fundamentally for lesions that increase the risk of death at that age.2
new ways of life. Similarly, paleodemography has moved Estimates of the population prevalences of such lesions
from simple and often idiosyncratic tabulations of age from skeletal series are therefore subject to precisely the
and sex data to more rigorous analytical investigations same sort of selectivity bias as the derivation of popula-
of demographic processes in prehistoric populations tion prevalences from clinical data, because neither type
(Angel i969, Acsadi and Nemeskeri I970, Weiss I973, of data constitutes a representative sample of the entire
Ubelaker I974, Moore, Swedlund, and Armelagos I975, population at risk. In both cases, the observed frequency
Buikstra I976, Lovejoy et al. I977, Lallo, Rose, and Arm- of pathological conditions should overestimate the true
elagos I980, Buikstra and Konigsberg I985, Buikstra and prevalence of the conditions in the general population.3
Mielke I985, Buikstra, Konigsberg, and Bullington I986, This bias cannot be avoided simply by obtaining larger,
Johannson and Horowitz i986, Horowitz, Armelagos, more representative skeletal samples; it is built into the
and Wachter I988). The concerns raised in this paper very structure of the data.
would not be relevant had paleopathology and paleode- The problem of hidden heterogeneity in risks is per-
mography not become established fields of active re- haps more counterintuitive. In the present context, hid-
search. den heterogeneity means that the population from

2. This selectivity differs from that attributable to having an unrep-

Conceptual Problems resentative sample of the dead, for example, because of poor preser-
vation of infant skeletons. Our point is that all samples of the dead
are inherently unrepresentative of the original living population at
Demographic nonstationarity refers to the departure of
risk of death, even a skeletal collection that is a perfect random
a population from the stationary state-a state charac- sample of all those who died.
terized by closure to migration, constant age-specific fer-3. This generalization leaves aside the question of how sensitive a
tility and mortality, zero growth rate, and an equilib- skeletal lesion may be as an indicator of disease. Pathological
rium age distribution. In nonstationary populations, changes in bone tend to be markers of chronic conditions, and
often only a small fraction of cases develops skeletal lesions. Thus,
age-at-death distributions are extremely sensitive to as paleopathologists frequently point out, skeletal lesions may be
changes in fertility but not to changes in mortality expected to underestimate the population prevalences of their as-
(Coale I957, Keyfitz i985). Thus, if a population is not sociated conditions. At present, it is difficult to weigh the counter-
stationary-and changing populations never are-small vailing effects of the underestimation caused by low sensitivity and
the overestimation caused by selectivity bias. There is no reason to
variations in fertility have large effects on its age-at-
expect, however, that these errors will perfectly balance each other.
death distribution, while even quite large modifications Therefore, we doubt that it will ever be possible to estimate popula-
of mortality have virtually none. Paradoxically, then, tiorn nrPr7,nlPnrcPQ rplinbhlIT frl-n qrkPl.tnl Lgiryn fIrP.a11PnC

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 345

which the skeletal series is drawn was made up of an disease and typically before a bony response is elicited;
unknown mixture of individuals who varied in their un- in this group, rapid death results in few, if any, skeletal
derlying frailty or susceptibility to disease and death.4 lesions in the mortality sample. Considering the differ-
Such heterogeneity may arise from genetic causes, from ential exposure to stress experienced by these three
socioeconomic differentials, from microenvironmental groups, it seems appropriate to rank them from high to
variation, or even from temporal trends in health, since low according to their level of health, at least with re-
most skeletal series, especially large ones, represent ac- spect to the condition of interest. Judging solely from
cumulations over more or less prolonged periods of the skeletal lesions, however, there appear to be not
-time. Hidden heterogeneity in risks makes it virtually three groups but only two: one healthy (unstressed) sub-
impossible to interpret aggregate-level age-specific mor- population with no lesions and one stressed subpopula-
tality rates in terms of individual risks of death. This tion with many. The groups experiencing the least and
problem is far from restricted to archaeological samples the most stress are indistinguishable with respect to the
but is now widely acknowledged throughout demogra- distribution of this particular kind of lesion.
phy (Vaupel and Yashin i985a, b; Trussell and Rodri- This fictitious (but not implausible) example intro-
guez I990). duces several of the themes explored in this paper: the
All these problems reflect two unavoidable facts. possible presence within the population of interest of
First, it is impossible to obtain direct estimates of demo- multiple, undetected subgroups that experience varying
graphic or epidemiological rates from archaeological risks of disease and death, the difficulty of recon-
samples. Such estimates require that we know the structing population prevalences of pathological condi-
amount of exposure to the risk of illness or death, that tions from skeletal lesion frequencies, the complex rela-
is, the number of individuals exposed to the risk and the tionship between the degree of stress and the likelihood
length of their exposures, neither of which is ever of developing a lesion, and the possibility that individu-
known with sufficient precision in archaeological re- als displaying lesions may actually be healthier than at
search to permit formal estimation by standard meth- least some individuals without lesions. In what follows,
ods. Assumptions such as stationarity are ruses that we show how each of these factors can confound the
allow us to reconstruct exposure indirectly, if imper- interpretation of osteological evidence about the health
fectly. Second, although "health" (however defined) is of past populations.
a biological characteristic of the individual, inferences
about it must be based on aggregate- or population-level
statistics. Just as it is a truism in epidemiology that the Hidden Heterogeneity and the Deviant
single-case study is of limited value, it is widely recog- Dynamics of Death
nized in paleopathology that reports on single specimens
tell us little about the disease experience of ancient pop- Life-table estimates of age-specific mortality are often
ulations. However, when the population of interest is used in paleodemography to draw inferences about the
heterogeneous for factors that affect health, the relation- levels and age patterns of risks experienced by individual
ship between aggregate measures and the experience of members of the population being studied. However,
the individuals making up the aggregate can be remark- when individuals vary in frailty and thus their risk of
ably tenuous. death and this variation is "hidden" (i.e., not captured
For example, imagine a living population comprising by observed covariates), death rates computed from
three subpopulations, all of which are potentially ex- aggregate-level mortality data may provide a misleading
posed to an acquired condition reflecting some form of picture of individual-level risks. This can be shown with
stress. This condition, when it occurs, increases the risk a very simple model, one that is intended not to be par-
of death and may eventually produce a distinctive skele- ticularly realistic but simply to make a basic point about
tal lesion in survivors. Members of the first subpopula- hidden heterogeneity and what Vaupel and Yashin
tion never experience the stress; therefore, none of them (i985b) call "the deviant dynamics of death."
develops skeletal lesions. The second subpopulation ex- Imagine that we observe a population of newborns,
periences moderate stress, sufficient to cause wide- each of whom experiences an absolutely constant risk
spread sickness lasting long enough to produce bony le- of death but who vary among themselves in their risks.
sions but causing few deaths; most individuals with The hazard of death (the continuous-time analog of the
bony lesions survive the stress but join the mortality central mortality rate in the life table) experienced by
sample later when they succumb to some other cause each child might be modeled as h,(t) = zi hc, where t
of death. The third subpopulation suffers heavy stress, is the child's age, zi is a measure of its individual-level
resulting in numerous deaths soon after the onset of the frailty, and h, is a component of the hazard common to
all children in the population. While we write hi(t) as a
4. The term frailty, which has entered standard demographic usage, function of age, both zi and h, are constants; therefore,
was originally defined by Vaupel, Manton, and Stallard (I979) as the overall hazard for the ith child does not in fact
an individual's relative risk of death compared to a standardized change with age.
cohort risk. More generally, it refers to individual biological charac-
teristics associated with persistent differences among individuals
Now suppose that z1 is distributed in some regular
in susceptibility, propensity, or relative risk with respect to disease fashion among the newborns; for example, the frailty of
or death (Vaupel i990). newborns might be normally distributed as in figure i,

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
346 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

<},> 50 _

WEEK 0 40 40

:5 30 \

ZO 20 -

~~~~~~~~ 0

0 1 2 3 4 5

Age of c

zo z


- ~020 A
020 \h(t) = 0.244exp[-0.75 1 (t)1
c 015 r2 0.901 (on log scale)

* 0 .10\
*Q 01

0 1 2 3 4 5

Ag9 of c
FIG. 2. Observed age-at-death distribution up to age
five years from a late prehistoric (A.D. I300) Oneota
site in west-central Illinois (Milner and Smith I990)
FIG. I. A model of heterogeneous frailty and selective and (below) age-specific hazards of death estimated
mortality. Each child's hazard of death is assumed to from the age-at-death distribution under the
be constant and proportional to its individual-level assumption that the Oneota population was
frailty (z). Frailty in turn is assumed to be normally stationary. The smooth curve is a two-parameter
distributed among newborns (top). Deaths during the negative Gompertz function fitted to the hazards by
first week of life (dots) are selective with respect to ordinary least squares.
the frailty distribution; that is, children of high frailty
make up a disproportionately large fraction of all
deaths. As a consequence, the frailty distribution are selective with respect to the frailty distribution: the
shifts downward by the second week of life (bottom), greater a child's frailty, the more likely it is to die. Thus,
and mean frailty decreases (arrow). Since the the frailty distribution shifts with age as high-frailty in-
aggregate-level hazard of death at each age is dividuals are eliminated by death (fig. i, bottom). As a
proportional to the mean frailty of survivors at that result, the mean frailty of surviving children declines
age, the aggregate hazard declines even though the with age, causing the mean hazard of death to decline
hazards of the individual children remain constant. as well. In other words, the aggregate-level risk of death,
given by the mean hazard, drops with age even though
top. If gt(z) is the probability density function specifyingthe individual-level risks remain absolutely constant.
that distribution at age t, then the mean hazard for all Thus, where there is heterogeneity in frailty, changes in
the children alive at age t is the aggregate-level hazard do not reflect the risks experi-
enced by any of the individuals who make up the popu-
h(t) = f gt(z)zhcdz = Zh (I )
The confounding effect that heterogeneous frailty can
have on the relationship between aggregate-level mor-
In other words, the mean hazard at any given age is sim- tality and individual health can be illustrated with refer-
ply the product of the common hazard and the mean of ence to a real archaeological sample. Figure 2, top, shows
the frailty distribution at that age. the age-at-death distribution up to age five years for an
The crucial point here is that both gt(z) and Z-t change unusually well-preserved sample of late prehistoric skel-
over time, because deaths of children at any given age etons from a completely excavated Oneota, cemetery in

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 347

west-central Illinois (Milner and Smith i990). A total of Unfortunately, the aggregate-level hazards of death
264 skeletons is included in this sample, with I25.4 of may tell us disturbingly little about individual risks of
them estimated to be below age five.5 The age-at-death disease and death if the population is heterogeneous
distribution closely approximates the mortality patterns with respect to frailty, as any real population almost
expected of a traditional, preindustrial population with certainly must be. For example, assume that the Oneota
high fertility (Milner, Humpf, and Harpending I989, population that gave rise to the skeletal sample was ac-
Paine i989). Most important, there is no obvious under- tually made up of two distinct and nonoverlapping sub-
enumeration of the very young in this sample, a com- populations, denoted i and 2, with differing levels and
mon problem in archaeological collections. The age- age patterns of mortality. Of course, there is no prior
specific hazards of death estimated from this sample by reason to believe that exactly two subpopulations exist,
standard life-table techniques6 are entirely plausible for but this is the simplest form of heterogeneity imagin-
this sort of population (fig. 2, bottom), with mortality able. For further simplicity, assume that each subpopu-
during the first six months of life just above 2oo per lation was homogeneous with respect to death (i.e., ev-
i,ooo live births and a rapid and consistent decline in ery individual within a given subpopulation experienced
mortality beyond early childhood. The estimated rates the same hazard of death at each age). How might the
are fit quite satisfactorily by a two-parameter Gompertz two subpopulations be combined to yield the Gompertz
hazard function (fig. 2, bottom), a model that is widely mortality curve characteristic of the Oneota skeletal
used to describe the decline in human mortality at early sample as a whole?
ages (Gage I990).7 From this point on, we will take the The model presented in the appendix provides a way
fitted Gompertz curve as a reasonable representation of to address this question. All that has to be done is to
early childhood mortality in the Oneota population, as specify a mathematical form for the age-specific hazard
estimated at the aggregate level. function in one of the subpopulations in order to find
What do these aggregate mortality rates reveal about the corresponding function in the other. Suppose, for
the risks of morbidity and mortality experienced by indi- example, that the population contains a group of rela-
vidual children in the Oneota population? A conven- tively disadvantaged (i.e., frail) children who represent
tional interpretation, common in both the health sci- approximately half of all newborns. If the hazard of
ences and paleodemography, is that the "average" death for these children is constant and equals o.25 at
newborn infant is at a high risk of death but its risk all ages, then we have the mixture of subpopulations
declines rapidly as it grows older. This interpretation, shown in figure 3, A. The hazard experienced by the
which would arouse no controversy, is not necessarily second, less frail subpopulation now declines at early
consistent with recent individual-level studies of disease ages, but it does so much more rapidly than the aggre-
processes and specific causes of death. For example, the gate mortality curve, reaching zero by about one year of
risks of morbidity and mortality from diarrheal disease, age. In this instance, the aggregate hazard curve is not
a major killer of children in the contemporary Third an accurate reflection of the hazards experienced by in-
World, tend to increase over the first several months ofdividuals in either subpopulation. Instead, the shape of
life as exposure to pathogens increases and as maternal the aggregate hazard function is dominated by a selec-
antibodies from breastmilk decline (Long et al. i992). tion process: because of their high risk of death, children
And yet the hazard rates in figure 2 seem to speak for in subpopulation i are progressively selected out of the
themselves: the risk of death is initially high but then initial pool of newborns, and subpopulation 2 comes to
declines steadily. represent an ever-increasing fraction of all surviving
children. The aggregate hazard thus converges on the
5. Fractional individuals result from the way in which skeletons hazard of subpopulation 2 as time goes by.
assigned to various ages are distributed over adjacent six-month Next, suppose that the hazard in subpopulation i re-
intervals, a reflection of uncertainty in estimating the age of indi-
mains constant with age but drops to o. i 6 and that the
6. These hazard rates were estimated on the assumption that the fraction of newborns belonging to subpopulation i is re-
population was stationary (see Milner, Humpf, and Harpending duced to 0.4. These changes produce the curves shown
I989, Paine i989). We make this assumption here strictly for con- in figure 3, B. Now the hazard function in subpopulation
venience, since our intention is not to highlight the nonstationar- 2 starts out higher than the aggregate curve but drops
ity problem but rather to focus on the issues of heterogeneity and
selectivity. Paine (i989) applied maximum-likelihood methods to
below it at about nine months of age. Again, the aggre-
fit model stable populations to the Oneota sample and estimated gate hazards fail to capture the pattern of risks of any
that the population growth rate was approximately -0.002. This individuals in the population.
estimate suggests that, while the Oneota population may not have An immediate reaction to these subpopulation hazard
been absolutely stationary, it was probably not far from it. How-
curves is liable to be that, while mathematically possi-
ever, because the likelihood surface was quite flat near the maxi-
mum, it is impossible to rule out either positive or negative growth ble, they make little sense from either a demographic or
rates of small absolute value. Consequently, the hazards in figure a biological point of view. We suggest that this reaction
2 should not be taken as final estimates of mortality at these ages. may be incorrect, precisely because commonsense no-
7. Many applications of the Gompertz function to preadult mortal- tions about what age-specific mortality rates ought to
ity use a three-parameter form that includes a constant, age-
independent component (we are grateful to Timothy Gage for this
look like are based entirely on aggregate-level data. By
observation). For the age-range considered here, this more compli- and large, we have little idea how hazards should behave
cated specification adds little to the analysis. at the individual level. As we have already noted, there

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
348 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

0.32 I I I 0.32

0.24 024

N hl~~~~~~~~~~~~~~~~~~~~~(t)

0.08 0.08

0.12 C 0.321, I , W
0.00 0.00
0 1 2 3 4 5 0 1 2 3 4 5

0.32 0.32


0m24 4hat
the mixture assuming that h04h=(t)

= 0.16

r-4 0.16 hi(t) ~~~~~~~~~~0.08

2 2~~~~~~~~~~~~~~~~~~~~~~~~~~~(t)

0.00 .0

0.00 ~~~~~~~~~000
0 1 2 3 4 5 0 1 2 3 4 5

Age of chikld (years) Age of child (years)

FiG. 3. Different mixtures of subpopulation hazards, hl(t) an
same Gompertz function for the average age-specific hazard, h(t) (based on the two-point frailty model of
Vaupel and Yashin [i985a]). In each case, the Gompertz curve is the same as in figure 2 (bottom). A, the
mixture assuming that hl(t) = 0.25 and p (the proportion of newborns belonging to subpopulation i) = 0.5; B,
the mixture assuming that hl(t) = o.i6 and p = 0.4; C, the mixture assuming that hl(t) = 0.02 + 0.04t and p
= 0.4; D, the mixture assuming that hl(t) =0.3 - o.it and p = 0.2.

is reason to suspect that the hazard for diarrhea may a given aggregate mortality pattern can result from an
actually go up at early ages as the duration of exposure infinite number of possible combinations of subpopula-
increases, and the same is likely to be true of other infec- tion mortalities. Moreover, we rarely if ever have any
tious diseases. Figure 3, C, allows for such an increasing idea how many subpopulations actually exist. It could
hazard in subpopulation i. Although radically different in fact be two, but it could also be three, four, or more,
from the aggregate pattern, an increasing hazard in one and indeed at the limit every individual could have his
subpopulation is perfectly compatible with a decreasing or her own unique level of frailty, giving rise to the sorts
aggregate hazard so long as the hazard in subpopulation of continuous distributions of frailty shown in figure i.
2 starts higher than the aggregate curve but drops below This fact only adds to the range of possible models that
it at about one and a half years of age. are perfectly consistent with a given aggregate mortality
Finally, with a linear decline in the hazard for subpop- curve. The problem is not simply a reflection of the fact
ulation i (fig. 3, D), the curve in subpopulation 2 be- that we have arbitrarily assumed two subpopulations.
haves very oddly indeed. It roughly parallels the aggre- This is a specific instance of what is known in mathe-
gate hazard up to age two but then skyrockets as the matical statistics as the identification problem (Hsiao
aggregate curve approaches zero. I990, Lancaster I990). A model is said to be uniquely
These examples could be multiplied endlessly. How- identifiable if, when fitted to a particular set of data by
ever, the general point of the exercise should be clear. If some formal method such as maximum likelihood,
there is hidden heterogeneity in the risk of death, then there is one and only one value of each of its parameters

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 349

that is consistent with the data. The models we have ples are by definition composed of a very special subset
just reviewed are unidentifiable in that an infinite num- of individuals who were exposed to the risk of death at
ber of values of the subpopulation hazards are equally each age, namely, the nonsurvivors. By their nature, the
compatible with the observed aggregate hazards. Far dead are a biased sample of all the individuals in the
more disturbing, every other model of heterogeneous population who were alive at a given age. Because of this
frailty, whether discrete or continuous, is equally un- selectivity, it is surprisingly difficult to draw inferences
identifiable. What this means in practice is that there is about the prevalence of pathological conditions marked
no way to infer individual risks of death from aggregate by hard-tissue abnormalities in a once-living population
mortality data in the absence of prior theory that speci- from the frequency of lesions in a skeletal sample. The
fies the way in which frailty varies among individuals. reason for this difficulty is simple: if the disease state
The implications of this problem for the demographic of interest affected individuals' risks of death (and pre-
treatment of mortality are profound. Indeed, hidden het- sumably that is precisely why it is interesting), then
erogeneity in risks and the identification problem asso- the frequency of the disease must be greater among
ciated with it have become dominant concerns in de- nonsurvivors than among the living. Stated differently,
mography during the past decade (Keyfitz and Littman proportional mortality (the fraction of deaths attribut-
I980; Heckman and Singer i984b; Vaupel and Yashin able to a particular cause) overestimates population
I985a, b; Heckman and Walker I987; Manton and Stal- prevalence because mortality itself is selective.
lard I988; Lancaster I990; Trussell and Rodriguez I990). Selective mortality is a special concern when bone
Two general strategies have been developed for dealing lesions that were active at the time of death indicate an
with hidden heterogeneity. One is to model the distribu- elevated risk of death. Enhanced risk may arise directly
tion of frailty directly in a way that makes biological from the disease process that induced the bony re-
sense (Weiss i990). Often, however, there is too little sponse-that is, that particular disease may be the pri-
prior theory about how the relevant processes work to mary cause of death. Alternatively, the disease may be
permit this sort of etiologic modeling. The second ap- a contributory cause, or it may simply be correlated with
proach involves the development of more general mod- some other factor that affects mortality. In all these in-
els of heterogeneous frailty (e.g., Heckman and Singer stances, the presence of an active skeletal lesion marks
I984a, Trussell and Richards I985, Heckman and a person at elevated risk of death. Therefore, the fre-
Walker I987, Vaupel I990). Here one specifies an arbi- quency of active lesions in a skeletal sample is greater
trary probability density function for the unobserved than the fraction of affected individuals in the living
heterogeneity and then checks to see if the final numeri- population from which the sample was drawn. This is
cal results are robust for changes in the specification. true not only of the skeletal sample as a whole but also
While it is not yet certain how successful these strate- of its constituent age- and sex-groups considered indi-
gies will be, recent applications in the analysis of data vidually. Without knowledge of the risk of death associ-
on morbidity and mortality offer some hope (Manton ated with a particular condition, it is impossible to pre-
and Stallard I988, Weiss I990). dict from modern population prevalences (e.g., from
For the present, it is important for paleodemographers epidemiological studies) the proportion of a skeletal
and paleopathologists to be aware of the general prob- sample that would be expected to show a hard-tissue
lem. When hidden heterogeneity in risks exists, as we response. Despite this fact, direct extrapolations from
suspect it must in all populations, then population-level clinical and epidemiological findings to mortality sam-
mortality patterns may tell us little if anything about ples are sometimes encountered in the literature on
individual risks of death. As a consequence, inferences archaeological skeletons (Morse I978, Blakely and
from paleodemographic life tables to the health status Detweiler-Blakely I989).
of individuals in prehistoric populations are, to say the The effects of selective mortality can be illustrated
least, problematic. Even if the osteologist wishes merely with reference to the Oneota skeletal series. Numerous
to rank prehistoric populations by their general level of crania in this collection exhibit the distinctive orbital
health, the problem is still important: populations can- and cranial-vault lesions usually referred to, respec-
not be compared meaningfully if their distributions of tively, as cribra orbitalia and porotic hyperostosis. These
frailty differ in unknown ways. A solution to this prob- lesions are known to accompany various anemias and,
lem will require serious thought about the best ways to when encountered in pre-Columbian skeletons from the
model the variation in frailty that almost certainly ex- Americas, are usually attributed to iron-deficiency ane-
ists in all archaeological contexts. mia (El-Najjar et al. I976; Steinbock I976; Mensforth
et al. I978; Ortner and Putschar I985; Stuart-Macadam
I985, I989; Larsen I987; Palkovich i987).8 The presence
Selective Mortality and the Mysterious
Meaning of Morbidity 8. Deficiencies of specific nutrients are frequently part of a broader
pattern of nutritional inadequacy, often complicated by infection.
It has long been argued that the elevated mortality rates among
Heterogeneity and selectivity complicate the interpreta-
young children common in the contemporary Third World reflect
tion of morbidity as well as mortality. As has been the interaction of infection and malnutrition (Gordon, Chitkara,
pointed out (Cook and Buikstra I979; Cook 1981, 1984; and Wyon I963, Gordon, Wyon, and Ascoli I967, Scrimshaw, Tay-
Palkovich 1985; Milner and Smith 1990), skeletal sam- lor, and Gordon I968, Puffer and Serrano I973, Scrimshaw I977,

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
350 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October I992

80 - 10

60 - 08 _

0~ ~ ~ 40
o~~~ 40
D 0 6 - \eletal


04 reconstructed
40 _ NONE
_8 Ppopatmn


0.5-4 5-9 10-1415-19 20+


Age group (years) 00 L I I

FIG. 4. Numbers of Oneota crania in five age-groups 0 5 10 15 20
with active cribra orbitalia or porotic hyperostosis,
inactive (remodeled) lesions, and no lesions. While
the differentiation of active and inactive lesions is Age (years)
inexact, it is the overall age pattern, not individual FIG. 5. Age-specific proportions of Oneota crania with
crania, that is important here. cribra orbitalia or porotic hyperostosis (solid curve)
and a reconstruction of the age-specific prevalences of
those lesions in the living population (broken curve).
of these cranial lesions is thought to reflect a bony re- Because of the selective effects of mortality, lesion
sponse in early childhood to pathological expansion frequencies are higher among those who died (the
of hematopoietic tissue in restricted diploic spaces skeletal sample) than in the reconstructed population
(Stuart-Macadam I985). Although iron-deficiency ane- at risk.
mia cetainly occurs in later life as well, it is unlikely to
induce comparable bony responses at those ages. Once active at the time of death, should be higher than their
developed, however, these cranial lesions can be re- prevalence in a comparable age-group in the living popu-
tained throughout life, as is indicated by their frequent lation. This is shown in figure 5, which presents esti-
occurrence in the skeletons of adults from many parts mates of the proportion of the once-living population
of the world. at each age that may have exhibited bony lesions. The
In the Oneota skeletal collection, both cribra orbitalia reconstructed population at risk at each age was derived
and porotic hyperostosis are present on most of the juve- by summing all observed crania and then decrementing
nile crania but few adult crania (fig. 4). Lesions indica- the sum by the number of individuals that died in the
tive of an active bony response are restricted to children, previous age-group; the number of individuals with le-
principally those less than five years old. The discrep- sions at each age in the reconstructed population was
ancy in the numbers of affected juvenile and adult crania similarly estimated. This reconstruction is based on
presumably reflects ongoing selectivity that preferen- three assumptions: (i) the population was stationary, (2)
tially eliminates the most frail individuals within each the bony lesions indicate a condition that developed dur-
age interval. The remodeled lesions, in this view, iden-
ing the first five years of life, and (3) the lesions are
tify individuals who managed to survive a period of early not entirely obliterated by subsequent bone remodeling.
illness only to succumb somewhat later to stress of un- Undoubtedly, all these assumptions oversimplify real-
known origin-stress that may or may not have had any ity.9 Nonetheless, they permit a tentative reconstruc-
relation to the conditions that produced the lesion.
Since we are viewing the end result of selective entry 9. The assumption of stationarity has been discussed in n. 6. As-
into the mortality sample, the frequency of pathological sumption 2 reflects the known childhood association of these bony
hard-tissue lesions, especially those that appear to be responses (Stuart-Macadam i985). Assumption 3 is more question-
able, but lesion obliteration through progressive bone remodeling
would be expected to have its greatest effect in the adult category
Chandra and Chandra I986), and several osteologists have stressed (20 years and older). If account could be taken of this effect, it
this synergism in the etiology of cribra orbitalia and porotic hyper- would increase the reconstructed population prevalences by a con-
ostosis (Mensforth et al. I978, Walker I985, Palkovich I987, stant of proportionality but not enough to bring them up to the
Stuart-Macadam I989). frequencies of lesions in the skeletal sample.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox I 35I

tion of the population at risk and the frequency of le- 100

sions in that population and thereby reveal trends during
the juvenile years that reflect selective mortality Under CD

this pattern of selectivity acting upon heterogeneous

DO 80
susceptibility to stress, the lesion frequencies among the
dead are consistently greater than the prevalence of le- g 60
sions in the living population. CO,

It is important to emphasize that this particular re- co

construction is based on a series of assumptions (e.g., 40

stationarity) that may not be correct. When such as-
sumptions cannot be made, there is no way to recon-
struct the numbers of individuals at risk and thus no L_ %~~~~~~~~~~~~~

way to infer population prevalences from proportional
Selective mortality is also a concern in dealing with 60 80 100 120 140
more subtle markers of poor health. For example, be-
cause of its known multifactorial etiology, short stature
Stature (cm)
is often taken as an indicator of general stress. A number
FIG. 6. Effects of two different levels of mortality on
of researchers have used the length of long bones in juve-
the stature of the dead when the risk of death is
niles to identify differences in linear growth among
negatively correlated with stature in the living. Solid
groups of skeletons drawn from separate time periods
curve, distribution of stature in the living population,
and have linked the bone length data to other skeletal
assumed to be normal with a mean of IOO cm and a
evidence of ill health (Hummert and Van Gerven I983,
standard deviation of 6 cm (values typical of those
Cook I984, Goodman et al. I984, Mensforth i985).10 In
observed in five-year-old-children from rural
these studies, shorter bones are consistently interpreted
Sub-Saharan Africa [Eveleth and Tanner
as reflecting increased stress. This seemingly straight-
forward interpretation does not take into account that iggo:appendix table 38]); broken curve, distribution of
stature in dead five-year-olds when mortality is high
the differences in stature are observed strictly in individ-
(total mortality = 46%); dotted curve, distribution of
uals who died at each age. Alternative interpretations
stature among the dead when mortality is low (total
are possible when the skeletons are viewed as nonsurvi-
mortality = 8%). The mean stature of the dead is
vors who entered the sample as a result of selective mor-
95.7 cm under low mortality and 98.6 cm under high
tality acting on heterogeneous frailty. In figure 6, for
mortality. At both levels of mortality, the probability
example, we assume that the distribution of stature
of death is one when stature is less than 8o cm and
among living children at a given age is absolutely invari-
e-I(x-80) otherwise, where x is stature in cm and A =
ant between two populations or time spans, that is, that
O.I2 under low mortality and 0.04 under high
there is no differential, stress-related stunting between
these two groups. We further assume that mortality is
selective with respect to the distribution of stature-in
particular, that shorter children are at higher risk of
death, a relationship commonly observed in the contem- mortality and the relationship between stature and
porary developing world (Kielmann and McCord I978, frailty are known. Once again, proportional mortality is
Chen, Chowdhury, and Huffman I980, Heywood I982). a poor guide to population prevalence."
Under these assumptions, variation in the stature of
dead individuals may indicate different levels of mortal-
ity (and perhaps stress) but not in the direction com- Proportional Mortality and Competing
monly supposed. When mortality is high, a larger frac- Causes of Death
tion of the entire distribution of stature is represented
among dead individuals; as a result, dead individuals are One solution to the difficulties posed above is to aban-
comparatively tall on average (fig. 6). If mortality slack- don any attempt to estimate population prevalences
ens, only the most frail (i.e., short-statured) individuals from frequencies of skeletal lesions, interpreting such
die. Periods of low mortality are therefore characterized frequencies strictly as estimates of proportional mortal-
by comparatively low mean stature among the dead. In ity or the fraction of deaths attributable to a given cause.
general, the frequency of apparent "stunting" among the
dead is uninformative about the distribution of stature i i. It is worth noting that the interpretive problems posed by s
or relative health among the living unless the level of tive mortality are not restricted to comparisons between ancient
samples. In a recent review of studies on hard-tissue growth and
development, differences in long bone length between prehistoric
io. Differential susceptibility to life-threatening stress has also and contemporary samples are attributed to the "poorer environ-
been invoked to explain the smaller size of permanent teeth in ments" of earlier peoples, thus ignoring the fact that the prehistoric
juveniles when compared with their adult counterparts (Guagliardo samples are made up of dead people and the contemporary samples
I 9 82). of the living (Johnston and Zimmer i989: i8).

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
352 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

This, in effect, is the solution advocated by Cohen shows that proportional mortality from any one cause
(I989:io6), who suggests that "estimates of the inci- is determined in part by the total number of causes op-
dence [sic] of disease in skeletal populations are more erating in the population. Paradoxically, what might be
useful for comparison to one another than for direct considered an improvement in health (either a reduction
comparison to rates reported in living groups." Unfortu- in risk from cause k or a decrease in the total number
nately, estimates of proportional mortality "can give of causes) will produce an increase in the proportional
rise to misleading conclusions if used to compare [the] mortality from the remaining causes. A familiar exam-
mortality experience of populations with different dis- ple of this phenomenon is the recent increase in cancer
tributions of causes of death" (Last I988:I06; see also deaths as a fraction of all deaths in industrial societies,
Miettinen I985:260). The reason for this is that, by reflecting the virtual elimination of deaths from such
definition, proportional mortality for a single cause of infectious diseases as tuberculosis, smallpox, diphtheria,
death reflects the importance not only of that particular and typhoid. In this instance, the shift in causes of death
cause but also of all the other competing causes of death has resulted in greater longevity but also a higher preva-
experienced by the population. lence of cancers and more years of life spent disabled,
This fact is easiest to demonstrate for completely in- indicating that the very concept of "health" may be
dependent competing causes, although it is equally true more complex than is often assumed.
of interacting causes. If a population of individuals is The dependence of proportional mortality on the total
exposed to k independent causes of death, then the total distribution of causes would occasion no alarm if we
hazard of death for an individual at age t is simply the knew every single competing cause present in the popu-
sum of the hazards associated with each independent lation. This is difficult enough, and perhaps impossible,
cause (Elandt-Johnson and Johnson i98o: 273): to achieve for living populations. For prehistoric popula-
tions, we know only a tiny subset of all competing
htotal(t) = hl(t) + h2(t) + + hk(t) (2) causes, namely, those that leave skeletal traces. As a
result, it becomes extremely difficult to interpret differ-
This is the total hazard experienced by an individual at ences in proportional mortality among skeletal samples.
age t, not the total population hazard combining individ- An increase in the fraction of deaths associated with a
uals of differing frailty at that age. The number of deaths particular cause may indicate an increased hazard of
to individuals exposed to the total hazard given by equa- death from that cause, but it may also indicate a de-
tion 2 depends not only on the total hazard itself but crease in the hazard of death from some other, totally
also on the number of individuals exposed to that haz- unrelated cause. Thus, the same change in proportional
ard (N) and the duration of exposure (8t). For 5t small, mortality may signal either an improvement or a deteri-
the total number of deaths at age t is approximately oration in the population's health. Once again, we are
faced with a fundamental identification problem when
Nhtotai(t)*t = Nh1(t)8t + Nh2(t)8t + + Nhk(t)8t.
Now suppose that we are interested specifically in dealing with paleopathological and paleodemographic
cause i because it leaves an unambiguous skeletal signa- data.
ture. The proportional mortality attributable to cause i
is simply
Pathological Processes and the
Healthiness of the Healed
deaths from cause i
total deaths Paleopathologists routinely note the importance of dis-
tinguishing active from inactive or healed lesions, even
while recognizing that this simple dichotomization may
not reflect all the complexities of the pathological pro-
= Nh1(t)8t L Nhi(t)8t1
cess (Mensforth et al. I978, Martin, Goodman, and Arm-
elagos I985, Walker I985, Palkovich i987). As crucial
as this distinction may be from a biological perspective,
= hl(t) [ hi(t)1 = hl(t) hi(t) + hk(t). (3) however, it further complicates the interpretation of pa-
leopathological evidence. The active-inactive distinc-
tion represents an attempt to separate disease processes
Writing the expression this way highlights the depen- that were ongoing at the time of death, including
dence of the proportional mortality associated with chronic conditions, from those that had occurred earlier.
cause i on the hazard of death from some other cause, If this distinction is correct, the presence of an inactive
k. lesion indicates survival of a disease process earlier in
If conditions in our population change in such a way life and thus may signify an individual whose frailty is
that the hazard of death from cause k decreases or is low compared with those who died at earlier ages.
eliminated altogether while the hazard from cause i is From this perspective, the differing age patterns dis-
unaffected, the proportional mortality from cause i willplayed by various kinds of lesions take on a new and
then increase even though its hazard remains un- largely unexplored significance. In figure 8, we show the
changed. This effect is illustrated in figure 7, which also age-specific frequencies of cribra orbitalia and porotic

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 353



FIG. 7. Proportional mortality from a specific cause of death (cause i, a disease that leaves a hard-tissue
lesion) as a function of the total number of competing causes of death and of the hazard of death from the kth
of these causes, assuming that all causes are independent and that the hazard of death from each of the first
k - i causes is 0.05.

hyperostosis and of periostitis during the first fourthose yearswith active lesions. In this case, the presence of a
of life in the Oneota sample, distinguishing between ac- healed lesion may actually be evidence that the individ-
tive and healed lesions. These lesions are of interest be- ual was of lower frailty and therefore at a lower risk of
cause of their distinct etiologies: crudely speaking, peri- death; because of his or her superior condition, such an
ostitis is a lesion of infectious origin, while cribra individual was able to survive long enough to manifest
orbitalia and porotic hyperostosis reflect synergistic in- a healed lesion. Frailer individuals, in contrast, would
teractions of infection and nutrition. Bearing in mind die soon after the onset of infection, either before a bony
that the estimates in figure 8 are lesion frequencies response had time to develop or during the time when
among the dead at each age, one might suggest that cri- bone lesions were active. Thus, the presence of a healed
bra orbitalia and porotic hyperostosis represent the com- lesion may sometimes indicate a state of comparatively
mon view of nutritional stress: children are differen- good overall health.
tially susceptible early in life but once stressed remain An alternative and radically different interpretation of
frail throughout early childhood. Therefore, even skele- the data on cribra orbitalia and porotic hyperostosis is
tons with healed lesions are selectively added to the possible. In this interpretation, virtually all individuals
mortality sample. Periostitis, in contrast, shows a quite develop the lesions early in life, but the underlying con-
different pattern. Early life is again a time of elevated dition places these individuals at no elevated risk of
stress, but now there appears to be no differential ten- death whatsoever. Thus, all children survive long
dency for individuals with healed lesions to die and thus enough to recover and for their lesions to heal, unless
enter the skeletal sample. In fact, individuals with they happen to die from some unrelated cause. In this
healed lesions appear to be much less likely to die than scenario, but only in this scenario, the frequencies of

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
354 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October I992

1.0 1.0

0.8 inactive 7 ?0.8

|~ -/ E | \active
E /E

76 0.6 / -_ 0.6 -

Co Co

0.4 0 0.4

0.2 active 0.2

GL inactive

0.0 0.0 , I Incive

0 1 2 3 4 0 1 2 3 4

Age (years) Age (years)

FIG. 8. Observed proportions of crania exhibiting cribra orbitalia or porotic hyperostosis (left) and skeletons
with periostitis (right) from individuals less than four years old in a late prehistoric Oneota sample. Skeletons
are classified according to whether the lesions were active or inactive (remodeled) at the time of death.

active and healed lesions in the skeletal sample would Goodman and Armelagos (i988) have reported data on
provide unbiased estimates of the population prevalence enamel hypoplasia in the permanent dentition for three
of such lesions. Of course, if this proved to be true, cribra samples of late prehistoric skeletons from Dickson
orbitalia and porotic hyperostosis would be remarkably Mounds, Illinois.12 These data indicate that the over
uninformative about stress and the risk of death. fraction of individuals with such enamel defects in-
Clearly, this interpretation makes little sense given creases consistently from the earliest to the most recent
what is known about the etiology of these lesions. How- sample, that the overall mean age at death is lower in
ever, the general point stands: unless we have a strong the most recent sample than in the two earlier ones,
prior model of the way in which a given lesion, whether and that, except for the earliest period, skeletons with
active or inactive, relates to frailty and of the way in enamel hypoplasia within each sample had a lower
which frailty varies among individuals, skeletal lesion mean age at death than their contemporaries without
frequencies cannot be interpreted in any straightforward enamel hypoplasia. They suggest three hypotheses to ac-
fashion. count for the lower age at death in individuals with
enamel hypoplasia: (i) that the enamel defects directly
reflect frailty, lesions and early death being correlated
The Tangled Tales That Dead Folk Tell: An outcomes of an innate constitutional susceptibility to
Extended Example stress; (2) that childhood stress resulting in the forma-
tion of an enamel defect permanently damaged the indi-
Skeletal samples from archaeological contexts can be ex- vidual's capacity to mount an effective biological re-
pected to present all the problems associated with non- sponse to later stress, and this acquired susceptibility
stationarity, hidden heterogeneity in risks, and selective
mortality simultaneously. In such situations, obvious or
i2. This particular study is singled out for four reasons. First, these
intuitive interpretations are likely to be badly mis-
researchers have been instrumental in developing the data-
leading. In this section, we illustrate some of the diffi- recording protocol now commonly used for hypoplastic enamel
culties that can arise in interpreting the distribution of defects. Second, they have produced some of the most innovative
hard-tissue lesions that are relics of stress followed by osteological analyses to date. Third, their use of carefully collected
survival. For concreteness, we restrict attention to the dental data, in conjunction with skeletal-lesion and age-at-death
information, contributed in the early I98os to a reevaluation of the
developmental defect of the teeth known as enamel hy- biological effects of the adoption of agriculture (Goodman et al.
poplasia, which provides evidence of stress experienced I 984). Finally, we believe that they deserve special credit for explic-
during childhood (Goodman and Rose I990). itly considering multiple hypotheses in interpreting their results.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 355

led to premature death at a later time; and (3) tions: that (i)
The accumulation of skeletons represents a
conditions produced disproportionate exposure of se- random sample of deaths from a mixture of two stable
lected segments of the population to stress of diverse populations,'4 the advantaged and disadvantaged groups
origins, members of social groups with higher risks of of our hypothesis; the advantaged group initially makes
exposure exhibiting more stress-related enamel defects up io% of the pooled population, but the mix of the
as well as premature mortality. Although Goodman and two populations changes over time according to their
Armelagos (I988) do not rule out any of these three respective fertility and mortality patterns. (2) Both popu-
hypotheses and indeed acknowledge that others may be lations experience fairly high mortality, corresponding
possible, they lean toward the third. to an expectation of life of about 30 years for the disad-
A fourth hypothesis might also account for their find- vantaged group and 37 years for the advantaged group.'5
ings. This hypothesis is based on the fact that enamel (3) The disadvantaged group has low fertility, in particu-
hypoplasia does not occur unless the child survives the lar a total fertility rate of four, approximating levels re-
period of stress and resumes normal enamel formation ported for the !Kung San (Howell I979, Harpending and
(Suckling I989). Under this hypothesis, the population Wandsnider I982) and the Gainj of highland New
from which each skeletal sample was drawn was made Guinea (Wood, Johnson, and Campbell I985); the advan-
up of two subgroups with differing levels of fertility and taged population, in contrast, has high fertility, corre-
mortality-one relatively advantaged (for whatever rea- sponding to a total fertility rate of eight (four and eight
son) and the other disadvantaged. Members of the advan- are about equally distant from the mean total fertility
taged group did not entirely escape childhood insult but rate observed in preindustrial societies [Wood I990]). (4)
were able to survive it long enough to develop enamel Because of their comparatively poor condition, children
hypoplasia and then go on to enjoy higher fertility thanin the disadvantaged group who become seriously ill die
their disadvantaged counterparts. Thus, individuals very quickly; thus, they do not survive episodes of stress
with observable lesions were principally from the less that would otherwise have induced an enamel defect.
frail (i.e., advantaged) group, and they had a lower mean For example, the children in this group might not have
age at death than the disadvantaged group because of sufficient nutritional reserves to sustain them through
their higher fertility, not because of poor survival. If the a bout of diarrheal disease. Consequently, there are no
differences in survival and fertility between the two dental signs of stress among the survivors of childhood
groups persisted across generations, then the advantaged in the disadvantaged population simply because those
group would come to represent an increasing fraction who experience stress do not survive long enough to
of the population as a whole. This hypothesis would develop hypoplastic enamel defects. In contrast, child-
therefore explain the temporal trends reported in both hood illness is often survived in the advantaged group,
lesion frequency and mean age at death. This hypothesis resulting in numerous older children and adults who ex-
and the one preferred by Goodman and Armelagos (i 988) hibit enamel hypoplasia reflecting episodes of stress ear-
are mutually exclusive, since under our hypothesis indi- lier in life.
viduals with detectable lesions and early death are pref- Table i lists the characteristics of the two model pop-
erentially drawn from the advantaged group. ulations constructed on the basis of the above assump-
The implications of this fourth hypothesis for the dis- tions, while figure 9 shows how the expected propor-
tribution of skeletons with hard-tissue lesions can be tions of dead individuals with enamel hypoplasia change
explored using a simple demographic model that gener- over time. The better health of the advantaged popula-
ates various statistics about hypothetical populations tion is reflected in its higher survival and fertility. As a
given two input parameters, the total fertility rate (TFR) consequence of its higher fertility, the mean age of the
and the level of mortality.'3 Expected proportions of skeletal remains from the advantaged group is I 5 years,
dead individuals displaying hypoplasia at each age were while the mean age of those from the disadvantaged
derived from this model under the following assump- group is 32. We emphasize that this difference is entirely
attributable to the difference in fertility: it is uninforma-
tive about any difference in mortality. Finally, because
I3. Under this model, age-specific fertility schedules were gener-
the presence of a skeletal lesion is indicative of low
ated using the rates given by Coale and Demeny (i983) for a mean
age of women at childbearing equal to 29 years; these rates were frailty (since frail individuals die before developing a le-
multiplied by one-half the input value of the TFR, roughly equal
to the gross reproduction rate. Age-specific mortality rates were
produced by the two-parameter logit method developed by Brass I4. A stable population is one with constant age-specific fertility
(I97I, I975). Using the so-called African standard life table as a and mortality rates and an equilibrium age distribution but not
starting point, a wide variety of mortality schedules can be gener- necessarily a zero growth rate. The stationary population is a spe-
cial case of the stable population.
ated by varying eo, the expectation of life at birth (set by the param-
eter a. in the Brass model), and the ratio of childhood to adult I5. The mortality schedules for the two populations are derived
mortality (set by the parameter P). In order to accommodate the from the Brass logit model using the African standard with a. =
varying ways in which ages are pooled in published paleopathologi- 0.4 and o.2 in the disadvantaged and advantaged groups, respec-
cal sources, we have generated a smooth year-by-year survivorship tively, and P = i for both groups. Sensitivity tests of the model
curve by spline interpolation through the African standard values indicate that the qualitative results do not depend upon either the
and have used this as the standard against which variant survivor- specific levels of mortality or the standard mortality schedule
ship schedules are generated. chosen.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
356 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

TABLE I sion), all the lesions in figure 9 are contributed by the

Characteristics of Two Model Populations advantaged population. In sum, a significant health ad-
vantage is reflected in a lower mean age at death and a
higher frequency of skeletal lesions. Better health makes
Advantaged Disadvantaged
Population Population for worse skeletons.
As the intrinsic growth rates in table i indicate, our
model does not produce a stable mixture of populations:
Input into model the advantaged population is growing rapidly while the
Mortality moderate high
disadvantaged group is almost stationary and indeed is
(eo = 37 yrs.) (eo = 30 yrs.)
Fertility high low declining slightly. Therefore, given enough time, the
(TFR = 8) (TFR = 4) disadvantaged population will disappear from the mix.
Output from model As a result of this intergenerational selection process,
Mean age at death I5 32 the mean age at death will go down and the frequency
Mean age at death, individ- 44 53
uals > I5 yrs. of skeletal lesions will go up as time goes by (fig. 9), even
Crude birth rate (per i,ooo) 56 30 though the characteristics of the separate populations
Crude death rate (per i,ooo) 27 33 remain fixed. The model thus accounts for the temporal
Intrinsic rate of increase +0.029 -0.003 trends observed at Dickson Mounds (Goodman and
Armelagos I988).

Reinterpreting the Transition to Agriculture

The issues raised here have serious implications for

G) 0.6 _ bme - 100 yrs

what is arguably the most important potential contribu-
_Y flme - 160 yrs
c,, co tion of paleopathology and paleodemography to health
o 0.8 science: elucidation of the biomedical consequences of
the transition from hunting-and-gathering to agricul-
ture. In I982 a decade of osteological findings pertaining
to this transition was summarized and evaluated at the
influential conference "Paleopathology at the Origins of
Agriculture" (Cohen and Armelagos I984a). In summa-
ries of this conference, Cohen and Armelagos (i984b)
and Roosevelt (I984) point to a common but not univer-
sal pattern in which the adoption of agriculture and sed-
a) 0.4
entary settlement is associated with a reduced mean age
0.2 at death and comparatively high frequencies of skeletal
0\ lesions, including those attributable to undernutrition
E firre -50 yr
and infectious disease. These data are interpreted as in-
0.2 -
dicating increased stress and reduced survival at various
ages, both signs of an apparent deterioration in general
health among the early agriculturalists. Cohen (I989)
Ume - 0 yrs
summarizes further evidence related to this issue and
0.0 I I I interprets it as showing that agriculture and sedentism
15-24 25-34 35-44 45+ resulted in increased local environmental contamina-
tion with pathogens, more frequent person-to-person
transmission of infections, and a decline in dietary qual-
Age group (years) ity and diversity.
FIG. 9. Age- and period-specific proportions of While these interpretations may be correct, the data
skeletons with enamel hypoplasia predicted by a do not force them upon us. As is emphasized by Satten-
mixed-population model with two subpopulations, spiel and Harpending (I983) and by Milner, Humpf, and
one advantaged and the other disadvantaged. The Harpending (I989), it is at least as plausible that the
aggregate proportions of individuals with enamel shift in mean age at death reflects an increase in fertility
defects were computed assuming that the advantaged rather than an increase in mortality associated with a
subpopulation is initially io% of the total aggregate deterioration in general health. Such an elevation in fer-
but that the mixture changes over time according to tility, if it occurred, might have reflected greater avail-
the intrinsic rates of increase listed in table i. The ability of digestible weaning foods and, hence, shorter
time span of 150 years is slightly less than the periods of lactational infecundability among the early
difference in the midpoints of the intervals used by agriculturalists (see Buikstra, Konigsberg, and Bull-
Goodman and Armelagos (I988) for the earliest and ington I986). Further, the logic developed in this paper
latest Dickson Mounds samples. suggests that the higher frequencies of skeletal lesions

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 357

observed in early agricultural samples could reflect an drawn from well-documented medical collections, and
enhanced ability to survive episodes of illness and stress the continued refinement of field, laboratory, and ana-
or an amelioration of other, competing and perhaps un- lytical methods.'7 In contrast, the problems discussed in
observable causes of death. Thus, the trends summa- this paper require a complete rethinking of the relation-
rized by Cohen and Armelagos (I984b), Roosevelt (I984), ships among pathological processes, the risk of death,
and Cohen (I989) are equally consistent with an inter- and the formation of mortality samples. In the face of
pretation that is the opposite of the one they endorse. demographic nonstationarity, hidden heterogeneity in
It is important to emphasize that our reinterpretation frailty, and selective mortality, intuition can be a poor
of the health consequences of early agriculture is not guide for evaluating how demographic and epidemiolog-
necessarily more correct than previous interpretations. ical changes should be reflected in a sample of skeletons.
And, indeed, we suspect that both interpretations may Unfortunately, existing theory may be no more useful
be correct for different periods and locations. The point, as a guide because most of it is concerned with the char-
however, is not that we are right and other authors are acteristics of the living, not the dead. As a consequence,
wrong but that the data support both interpretations models for death distributions are poorly developed.
equally well. The correct model for linking data and in- What must be done to elucidate the complex relation-
terpretation is in this instance, as in so many others, ship between the health of a community and the charac-
unidentifiable. teristics of the skeletons it leaves behind? Four tasks
immediately suggest themselves.
First, we need to develop deeper insight into the likely
Is There Hope? sources of heterogeneity in frailty and the shape of the
frailty distribution in real populations, work currently
We remain convinced that osteological research will being pursued by epidemiologists, physiologists, and ge-
continue to play a central role in the study of human- neticists (Weiss I990). This research will need to pay
disease interactions. This is especially true of attempts due attention to the fact that frailty is multidimen-
to understand variation in the health of peoples with sional: different disease processes interact with each
contrasting subsistence and settlement systems, natural other and also with an individual's constitutional sus-
environments, levels of cultural complexity, and popula- ceptibility to stress in determining frailty.
tion sizes. Present characterizations of the health of pre- Second, we need to understand how a given frailty
industrial populations are based upon a complicated distribution is related to the distribution of risks of
amalgam of information from modern medical surveys, death among individuals. A considerable part of recent
the ethnographic and historical records, and archaeologi- demographic theory has been devoted to this issue (Man-
cal excavations of ancient sites (e.g., Cohen I989). Ar- ton and Stallard I984, I988; Vaupel I988, 1990), and
chaeological evidence is crucial to this effort since con- we suspect that it may be resolved during the coming
temporary preindustrial populations, especially modem decade.
hunter-gatherers, are undoubtedly a highly select sample Third, we need a better understanding of the details
of all such populations that have ever existed.'6 We will of various pathological processes at the cell, tissue, and
continue to need the perspective and time depth that organ levels. Specific aspects of pathology that need to
archaeology can provide. be addressed include how long it takes to develop partic-
Nevertheless, using the archaeological record to infer ular hard-tissue lesions, how overall health affects the
the health characteristics of a once-living population is probability of lesion development, and how frailty and
far more difficult than is commonly acknowledged. the risk of death vary by stage of lesion development,
Most of the widely recognized problems associated with including the inactive or healed stage. This is a line of
osteological materials from archaeological sites-in- research best pursued by pathologists working with re-
cluding determining the age and sex of skeletons (Buiks- cent, well-documented biological materials and by epi-
tra and Mielke I985, Iscan and Loth I989, Iscan I989, demiologists.'8
St. Hoyme and Iscan I989, Meindl, Russell, and Lovejoy These first three tasks are ones to which archaeologi-
I990), the inadequate size and unrepresentative compo- cal osteologists are unlikely to make fundamental con-
sition of many skeletal samples (Howell I982, Milner, tributions: they must remain consumers rather than
Humpf, and Harpending I989, Paine I989), variable and producers of the relevant theory and observations. There
selective preservation of bones (Gordon and Buikstra is, however, a fourth task to which all anthropologists
I98I, Waldron I987, Walker, Johnson, and Lambert can contribute, namely, development of a better under-
I988), and the differential diagnosis of lesion-producing standing of the role played by cultural context in de-
diseases (Steinbock I976, Buikstra and Cook I980, Zim-
merman and Kelley I982, Ortner and Putschar I985, I7. Another problem, currently of major concern to epidemiolo-
Larsen I987)-can probably be resolved by additional gists and demographers but receiving only passing attention from
excavation, the development of better chronological and osteologists, is that of distinguishing primary from contributory
contextual controls, reference to comparative materials causes of death (Manton and Stallard I984, Nam i990).
i8. Information on the course of disease and its relationship to
mortality in the preantibiotic era would clearly be of great interest
i6. See, for example, the current controversy over the status of the
in this connection, but pertinent data are difficult to locate and
!Kung (Solway and Lee I990, Wilmsen and Denbow i990). intepret (Alter and Riley i989).

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
358 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

termining heterogeneous frailty and the level of selec- birth to age t:

tive mortality. For example, interpretation of the Oneota
data is facilitated by the fact that these skeletons were p(t) = P(o)S1(t)/{p(o)S1(t) + [I - p(o)]S2(t)} (A.2)
drawn from a relatively simple tribal society. The ceme- = p(?)NW13/(t),
tery, which was completely excavated, is comparatively
homogeneous in composition and appears to have been
used for a relatively short period of time, spanning de- where
cades rather than generations. There is no archaeological
evidence suggesting that the sample represents a com- S(t) = p(o)51(t) + [I - p(o)jS2(t) (A. 3)
plex accumulation from distinct social groups with dif-
ferential access to basic resources or exposure to patho- is the mean survival from birth to age t in the po
gens, such as might occur in samples drawn from as a whole. Being a constant, p(o) can be written as p
complex societies or accumulated over multiple genera- with no loss of generality.
tions when either fertility or mortality was changing. Combining equations A.i to A.3 and rearranging, we
While these conditions almost certainly do not elimi- can write h2(t) as a function of p, h(t), S(t), hl(t), and
nate the problem of heterogeneous frailty entirely, they S 1 (t):
undoubtedly minimize its effects. The Dickson Mounds
skeletal series, in contrast, is more difficult to interpret h2(t) = [E(t)S(t) - ph1(t)S1(t)]/[S(t) - pS(t)]. (A.4)
because of its cultural context. The cemetery is large
and its internal organization is complex, consisting of Assuming that h(t) is a two-parameter negative Gom-
multiple, superimposed mounds (Harn I980). Much of pertz function, then
the burial area remains unexcavated, increasing the like-
lihood that the available sample is selective. The skele- h(t) = ote-t and S(t) = exp [-x(i - e-t)/r]. (A.5)
tons from Dickson Mounds were accumulated over
many generations spanning hundreds of years during a If we substitute these expressions in equation A.4 and
period of considerable cultural change. Thus, sample provide empirical estimates of ox and p, then h2(t) is
heterogeneity and selectivity are likely to be of extreme uniquely determined by p and hl(t) since Sl(t) can be
importance in their interpretation. found from hl(t) as
In sum, choosing among competing interpretations of
the osteological evidence requires tight control over cul-
tural context as well as a deeper understanding of the 51(t) = exp -f hl(y)dyl. (A.6)
biology of frailty and death. These problems deserve far
more attention than they have received to date if we are
Thus, if we specify a functional form for h1(t)
to make sense of the biomedical consequences of the
for p, we can immediately solve for h2(t).
major social and environmental changes that have oc-
curred during the course of cultural evolution.

Appendix Comments
Here we present a hazard model of mortality with a
two-point distribution of frailty, based on a model origi- MARK N. COHEN

nally developed by Vaupel and Yashin (I985a). We as- Department of Anthropology, State University of New
sume that the population of interest comprises two non- York, Plattsburgh, N.Y. 12901, U.S.A. 3I III 92
overlapping fractions, subpopulations i and 2. These
subpopulations differ from each other in the level and Wood and colleagues raise a number of significant ques-
age pattern of risk of death, but the individuals within tions about whether cemetery populations can be pre-
each subpopulation have identical risks at each age. If sumed accurately to reflect the health status of the
once-living populations from which they are derived.
hl(t) and h2(t) are the hazards of death at age t in subpop-
ulations i and 2, respectively, then the mean hazard in Specifically they challenge several of my conclusions
the population as a whole is simply the weighted av- (Cohen and Armelagos I984, Cohen I989) about the
erage, meaning of apparent trends in skeletal data related to
the origins of agriculture and the rise of civilization.
h(t) = p(t)hl(t) + [I - p(t)]h2(t), (A.i) They suggest that my conclusions are possibly, but not
necessarily, accurate because the skeletal data are ame-
where p(t) is the fraction of surviving individuals at nable
age to more than one reading.
t who belong to subpopulation i. The value of p(t) is But those conclusions come from several sources of
determined by the initial mixture of subpopulations data, not just from skeletons. Skeletal data display an
(i.e., byp(o)) and by Si(t), the probability that an individ-
apparent increase in infection associated with sedentism
ual in the ith subpopulation (i = i, 2) survives from and large population aggregates in many parts of the

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 359

world. This trend might be spurious, as Wood et al. sug- viewed as separate from related demographic consider-
gest, but it is also predicted by epidemiological theory ations.
and demonstrated, repeatedly, in ethnographic compari- Paleopathologists and bioarchaeologists excavate and
sons and transitions. Similarly, tuberculosis not only analyze human remains which are often incomplete and
blossoms with incipient urbanization in the archaeologi- poorly preserved and may fall short of being representa-
cal record but also is clearly a disease of cities in the tive of the population(s) from which they are derived.
modern world and one which is considered unlikely to Even worse, skeletal lesions can be expected to develop
have been capable of propagation in a sparsely populated only during periods of chronic illness, and of those not
preagricultural ancient world. all are known to induce quantifiable or even recogniz-
The apparent increase in anemia in skeletal popula- able bony changes. This thought-provoking article at-
tions associated with farming, sedentism, and large pop- tempts to make explicit, and therefore more objective,
ulation aggregates might also be spurious, but anemia is the multiple factors which play into what the authors
also a well-known consequence of the increasing parasit- term "selective mortality" and which color all subse-
ism and disease which predictably accompany larger and quent efforts to interpret lesion frequencies among sur-
more sedentary groups in the modern world. Moreover, vivors and nonsurvivors in a particular age-group. The
contemporary hunter-gatherers really do have low rates notion of "frailty," or susceptibility to stress, presented
of anemia as well as low rates of protein, vitamin, and here is critical to a deeper understanding of both active
mineral deficiency. A similar argument can be made and healed lesion frequencies and distributions and
about the apparent increase in frequency of enamel hy- raises the larger and more complex question of just what
poplasia of teeth associated with the Neolithic revolu- observed skeletal lesion frequencies really mean.
tion. Hypoplasia is a general stress marker mostly for The authors are in good company, because even as
the infant and toddler years. One prominent cause of early as I985, D. J. Ortner (personal communication)
hypoplasia, weanling diarrhea, is, in fact, conspicuously was convinced that the presence of healed skeletal le-
rare or absent among contemporary hunter-gatherers sions suggested that the individual manifesting such
even when common in neighboring farmers. Most diar- bony changes was immunologically capable of "fighting
rhea in fact is density- and sedentism-dependent and is back." As Wood et al. note, "the presence of an inactive
associated with levels and kinds of malnutrition rarely lesion indicates survival of a disease process earlier in
seen among hunter-gatherers. Moreover, the major epi- life and thus may signify an individual whose frailty is
demic diseases, such as measles, which are also consid- low compared with those who died at earlier ages."
ered possible causes of hypoplasia arguably did not exist Rather than being viewed as the "sick ones," Wood et
in the ancient world. It seems to me that, in the name al. believe that "individuals with detectable lesions and
of parsimony, the obvious interpretations of the skeletal early death are preferentially drawn from the advan-
trends in these cases are also the right ones. taged group." That is, the individuals with the healed
On a more general level, I am concerned with the clear bony lesions and the hypoplastic enamel defects are ac-
pro-state or procivilization bias which dominates both tually the healthier ones. I must confess that it has
popular and scientific interpretations of history (and of taken me a number of years to accept this argument, but
current events). Much of the work that is at stake in I finally have; in fact, it makes good inherent sense.
this discussion (Cohen I989) has challenged this bias, What we are looking at when we analyze prehistoric
and it disturbs me that the questions raised by Wood et skeletal populations and observe healed bony pathology
al. emerge now, just as uncivilized life-styles and the is those who, in a sense, have "made it," to succumb,
meaning of our history are being reevaluated. In giving perhaps, to a different stressor or group of stressors later
the questions raised by Wood et al. the attention they in life. This is all the more reason to insist that an ex-
merit, I hope that we will remember to be equally skep- plicit analytical distinction be made between healed and
tical of all conclusions from skeletal data (including active lesions; as intuitive as this may seem, some pa-
those already embedded in our history books) and not leopathological studies still do not do it, leaving us with
just discard some of the recent revisions to standard the impossible task of blindly teasing out the meaning
history. hidden in the numbers.
While I am convinced that the notion of heteroge-
neous frailty is an extremely important one, I am consid-
ereably less optimistic than the authors that we will
LESLIE E. EISENBERG ever be able to develop an index of frailty for either indi-
8S Curtis Place, Lynbrook, N.Y. 11563, U.S.A. viduals or populations. From their discussion, however,
24 III 92 it is clear that we should not assume that hidden hetero-
geneity is normally distributed in all populations. It is
In this cautionary yet optimistic article, the authors ad- by applying these concepts that we can begin to make
dress a number of issues fundamental to the reconstruc- sense of, or at least offer cogent competing hypotheses
tion and interpretation of prehistoric health using skele- for explaining, observed bioarchaeological phenomena
tal series derived from archaeological contexts. While such as the economic, technological, and social changes
my comments focus primarily on the impact of this ar- inherent in the transition from a more mobile hunter-
ticle on paleopathological studies, they should not be gatherer life-style to a more sedentary agricultural one.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
360 I CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

That the transition may not have been accompanied by and more general lesions, such as periostitis, could be
a deterioration in general health status is well worth attributed to other etiologies. In areas where the trepo-
examining in light of the issues raised here. nemal diseases are endemic, such as Africa, morbidity is
high but mortality is low. Consequently, the individuals
who exhibit skeletal lesions characteristic of treponema-
DALE L. HUTCHINSON tosis probably represent not all survivors of the disease
Department of Anthropology, University of Illinois, but those who have had the disease longest. In this case,
Urbana, Ill. 6I8oI, U.S.A. i9 III 92 periosteal inflammations may actually underestimate
the prevalence of treponematosis in a population. It is
How well a skeletal sample represents the society from equally as plausible that nonspecific periosteal inflam-
which it is drawn is a long-standing issue. Much of the mations have other etiologies. This is aside from other
controversy involves methods for estimating age and sex health factors which might exacerbate the infections,
from skeletal remains (Bocquet-Appel and Masset i982, such as nutrition.
I985; Buikstra and Konigsberg I985) and biases due to That Wood et al. have left me thinking about other
preservation differences (Walker, Johnson, and Lambert examples of hidden heterogeneity, selective mortality,
I988). There is general agreement among researchers and demographic nonstationarity demonstrates the util-
that inadequate techniques for estimating age and sex ity of their discussion. Consideration of a suite of possi-
could result in aberrant demographic reconstructions. ble explanations for skeletal lesions is important for the
Wood and his colleagues focus on another issue- development of more reliable interpretations of disease
inferring the health of a past population from frequen- in past populations. As they point out, however, many
cies of pathology preserved in its skeletal remains. As of the tasks here are likely to fall to researchers other
they point out, many of the problems encountered in than anthropologists. There is still much to be learned
estimating the prevalence of particular diseases are the about the interaction of disease, environment, and cul-
same for skeletal and living populations; estimates of tural factors in the overall epidemiology of particular
disease prevalence in both are dependent on the samples diseases. Finally, while anthropologists can contribute
from which they are drawn. For example, it is not sur- more toward an understanding of cultural context in the
prising that the prevalence of tuberculosis in contempo- interpretation of past disease, factors such as skeletal
rary U.S. populations is likely to be greater in a sample preservation and basic demographic methods for skeletal
from a county health clinic subsidized by state funding materials will still play a role in accurate estimates of
than in a sample from a community health-management the health of past populations.
organization that is not.
The emphasis on populations rather than individuals
in osteological studies during the past few decades has RIMANTAS JANKAUSKAS AND GINTAUTAS CESNYS
been an attempt to model studies of the dead on studies Department of Anatomy, Histology, and
of the living. One question that Wood et al. raise is when Anthropology, Faculty of Medicine, Vilnius
an aggregate is a population and how many subpopula- University, Ciurlionic 2I/27, 2009 Vilnius, Lithuania.
tions it contains. This question was a major focus in the 6 I 92
Bocquet-Appel and Masset (I982, I985) exchanges, and
their concern with the effect of migration on demo- This thought-provoking paper is of great interes
graphic reconstructions remains a problem. leoanthropologists. While the amount of skeletal mate-
Estimates of disease prevalence in skeletal popula- rial is rapidly increasing, most studies of it remain on a
tions are further hindered by the unavailability of clini- rather simple empirical level, limiting themselves to the
cal data. Do skeletal lesions accurately reflect the preva- mere statement of the facts observed. 'rhere is a growing
lence of a disease in a living population, or are they need for theoretical evaluation of these observations,
representative of the survivors of a disease who lived and Wood et al.'s paper is a nice attempt to fill this
long enough to manifest them? The reconstruction of demand. It is axiomatic although often forgotten by
disease prevalence in a past population from skeletal le- paleo-osteologists that skeletal series first of all repre-
sions is certainly problematic. Many of the lesions pre- sent dead people and this means that direct extrapola-
served in skeletal remains are nonspecific. Wood et al. tion from their data to a living population is problematic
discuss one such, enamel hypoplasia. Another, perios- (higher frequencies of active lesions is one example).
teal inflammation, may result from infectious disease or The idea of hidden heterogeneity and selective mortality
from minor injury. One group of related diseases, the in the dynamics of death is another interesting and im-
treponematoses, may produce periosteal inflammation, portant suggestion. For hunter-gatherer and early ag-
but only some of the individuals who test positive sero- ricultural societies it is perhaps more a theoretical possi-
logically exhibit skeletal involvement (Grin I956, Mur- bility, but for populations with social stratification it is
ray et al. I956). Furthermore, only a small percentage of increasing importance. The problem can be solved
of the individuals with skeletal lesions develop lesions only through close collaboration with archaeologists.
characteristic of treponemal infection (stellate cranial Another question of particular interest is the validity of
lesions, saber tibiae), usually in the secondary and ter- skeletal markers of stress. We completely agree with the
tiary stages of the disease. Consequently, the less severe authors that there can be no simple way of interpreting

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox I 36I

these markers. For example, short stature and periostitis Recent in-depth studies of conditions such as enamel
incidence can be understood better when the levels of hypoplasia (reviewed by Goodman and Rose I990 and
mortality are known. In our late-medieval materials, Skinner and Goodman I992) and porotic hyperostosis
short life-span is correlated with shorter stature, but (Stuart-MacAdam I985, I987) bring us closer to under-
these samples of shorter stature (and lower level of sex- standing the implications of finding these lesions in pre-
ual dimorphism) and shorter life-span have a higher inci- historic peoples.
dence of healed periosteal lesions. According to the hy- A problem not mentioned in this work is that of dif-
pothesis suggested by Wood et al., this might mean that ferential preservation of skeletons based on age at death
short stature, short life-span populations with high fre- (see Walker, Johnson, and Lambert I988). This com-
quencies of periosteal lesions (in our case, medieval pounds the problems pointed out by Wood and col-
town populations) were healthier. This conclusion dif- leagues in that it is usually the very young and very
fers dramatically from traditional interpretations and old who are omitted from skeletal samples because of
calls for explanation. differential preservation.
This paper is an excellent presentation of the possibil- It seems that the best approach we can take at this
ities of interpretation of osteological materials that will point is the one these authors suggest: to provide alter-
serve as a basis for future discussion and a call for better native explanations for the set of data at hand. Along
understanding of cultural context as well as of the biol- the same lines as Ortner's plea for better descriptions of
ogy of pathological processes. It should encourage re- skeletal lesions, this approach will allow others to work
search collaboration between physical and cultural an- with the data and the possible scenarios when future
thropologists. studies provide additional insights into the interpreta-
tion of past populations.

Department of Archaeology, University of Calgary, JOHN R. LUKACS

Calgary, Alberta, Canada T2N 1N4. 23 III 92 Department of Anthropology, University of Oregon,
Eugene, Ore. 97403, U.S.A. 9 iII 92
Wood, Milner, Harpending, and Weiss have very elo-
quently pointed out that we still have a long way to go In many ways this article addresses theoretical and prac-
before we can speak with confidence about the health tical concerns for interpreting the human skeletal record
of prehistoric people. They have also delimited some parallel to those expressed in the early I980S by students
problems that distort interpretations of prehistoric of vertebrate taphonomy (Behrensmeyer and Hill I980,
health from skeletal remains. While some of these prob-Shipman I980). The latter field of enquiry is essentially
lems have been brought forward in earlier works (e.g., concerned with understanding the demographic charac-
McHenry I968 on differing interpretations of Harris teristics of living populations the better to interpret the
lines and Howell i 982 on paleodemographic reconstruc- factors influencing the creation of vertebrate fossil as-
tion), this paper is unique in highlighting and ques- semblages. One wonders why it has taken ten long years
tioning specific assumptions in paleodemography and for physical anthropologists to address similar complex
paleopathology which influence larger-scale interpreta- taphonomic processes in human mortuary samples.
tions of health among earlier peoples such as those pre- Though many physical anthropologists are familiar with
sented in the volume by Cohen and Armelagos (i984a). the works just cited, paleopathology apparently has bor-
My comments focus on the problem of evaluating mor- rowed these concerns not directly from paleontology but
bidity. from demography. Nevertheless, the problematic issues
The emphasis on problem-oriented research in paleo- of sampling bias and the complicating effects of distinct
pathology, paralleling research in archaeology, has re- demographic patterns on the genesis of vertebrate faunal
sulted in a shift from describing individual lesions in asemblages directly parallel concerns expressed here re-
detail to grouping lesions with similar appearances to- garding the measurement and intepretation of pathology
gether and focusing on their frequencies. This popula-and vital statistics in prehistoric skeletal series.
tion approach, starting most recently with the work ofHuman osteologists and dental anthropologists
Armelagos (I969) and becoming widespread throughout should benefit from the cautionary tone of this paper,
the I970S and I980s, has resulted in a wealth of litera- which serves the valuable purpose of questioning ac-
ture linking diet, settlement patterns, and health (e.g., cepted methods for measurement and interpretation of
Cohen and Armelagos I984a, Rose and Rathbun I987). health status in prehistoric populations. Since the au-
Ortner (i99i) has warned researchers of the dangers thors of rely primarily upon examples from prehistoric Na-
making diagnoses without providing careful descrip- tive Americans and many readers may be unfamiliar
tions of bony lesions. He calls for a return to emphasis with these and similar issues among South Asian popu-
on detailed descriptions so that the findings of different lations, my comment will be devoted to additional ex-
investigators will be comparable. While this seems ret- amples from India and Pakistan that generally support
rogressive, a more careful consideration of the causes of their contentions.
lesions would aid in interpreting what recovery with The problems of hidden heterogeneity in frailty and
some scar on bone really means in terms of adaptation. sample selectivity are especially well illustrated in In-

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
362 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

dia, where society is clearly divided into distinct groups: sample, frail in other respects. These sex differences in
the very wealthy and the very poor. Adequate nutrition, linear enamel hypoplasia among Bronze Age Harappans
education, and access to good health care enable the have been interpreted to reflect the culturally sanc-
wealthy to live a healthier and less stressed life in a tioned differential treatment of children commonly
culturally buffered environment. Their impoverished known as "son preference" or "daughter neglect" (Miller
neighbors are routinely subjected to malnutrition, lack I985; see also Lukacs and Joshi I992).
of education, more vectors of disease, and minimal or Using natural rather than clinical samples, our inves-
inadequate health care. This visible heterogeneity in af- tigation of three living ethnic groups of northwestern
fluence among the living translates into a culturally in- India was expected to show that female Rajputs, mem-
duced variability in frailty that must be considered si- bers of a caste group known to practice son preference
multaneously with the genetic variation in frailty that and daughter neglect, displayed a higher incidence of
exists within both wealthy and poor segments of Indian linear enamel hypoplasia than male Rajputs, female trib-
society. An individual of "medium frailty" might sur- als, or low-caste females (Lukacs and Joshi I992). Gender
vive in a wealthy environment but perish in a poor one. differences proved not significant, however, and Rajput
Variations in affluence and frailty are not always clearly women had the lowest prevalence of the three groups of
discernible in prehistoric mortuary collections. females studied. These findings suggest that the osteo-
Among the Bronze Age Harappans the interpretive dif- logical paradox has its corollary in studies of the living.
ficulties raised by hidden heterogeneity and selectivity In the absence of other reasonable explanations, the hy-
are numerous and not easily resolved. Aggregate preva- pothesis was advanced that if the mortuary sample of
lence rates for dental diseases for the entire Harappan Rajput females could have been analyzed, a higher preva-
skeletal sample undoubtedly combine the prevalences lence of linear enamel hypoplasia would have been
for more than one subpopulation at this 3d-millennium found.
urban site (Lukacs I992). The series is not large enough Human osteologists and dental anthropologists whose
to permit subdivision into socioeconomic groups on the research encompasses both living and prehistoric popu-
basis of funerary goods or ornateness of body ornaments, lations are routinely confronted by the issues raised in
which would be one approach to identifying the preva- this paper, and because of the dramatic differences in
lence of pathological lesions in subgroups based on af- the composition of their study samples (living vs. dead)
fluence. Alternatively, the human skeletal samples from many of these investigators are especially cognizant of
Mesolithic sites of the Gangetic Plains are much more the problems of hidden heterogeneity in frailty and se-
likely to be culturally homogeneous than either modern lectivity in skeletal samples. While being aware of these
Indian society or Harappan society and therefore less problems does not guarantee that workable solutions
susceptible to these interpretive dilemmas (Lukacs and will be immediately forthcoming, it does demand that
Hemphill I992, Lukacs and Pal I992). conclusions regarding the health status of prehistoric
The interpretation of linear enamel hypoplasia data human skeletal series be carefully thought out and ad-
on living Indians and prehistoric Harappans is an excel- vanced with considerable caution. The authors of "The
lent example of the complexities that Wood et al. dis- Osteological Paradox" should be commended for bring-
cuss. At the most elementary level, the genders could ing these problems to the attention of a wider audience.
be considered as subgroups which occur in all human May their discussion generate a higher level of achieve-
populations. In the Harappan skeletal sample as whole, ment in paleodemography and paleopathology!
linear enamel hypoplasia prevalence is 72.2%, but sig-
nificant differences are evident between the sexes both
in this feature (males 52%, females 93%) and in dental JANET W. MCGRATH

caries (Lukacs I992, Hemphill, Lukacs, and Kennedy Department of Anthropology, Case Western Reserve
I99I). These results have been interpreted to suggest University, Cleveland, Ohio 44io6-7125, U.S.A.
that females were subjected to greater stress than males 2I III 92

during childhood and consumed a more cariogenic diet

than males as adults. While the mean age of linear "Better health makes for worse skeletons" is a state-
enamel hypoplasia formation is not significantly dif- ment with complex and interesting implications for the
ferent for males and females, females are affected study of ancient disease. Wood et al. present a well-
at younger and older ages and have more events, or reasoned article that points to what now seems obvious:
"growth disruptions," than males. These observations since skeletal lesions develop as the result of a long-term
go beyond mere subgroup prevalence rates to suggest disease process, individuals with skeletal lesions are
that females were ill or malnourished more often than likely to have been rather healthy, having survived long
males. Among the select sample of Harappan skeletons enough to develop them. The potential reinterpretation
available for analysis, females appear to have been more of paleopathological data in light of this discussion is
often affected by factors causing linear enamel hypopla- exciting to anticipate.
sia than males and may be considered frail, or less well- I confess that I have always had trouble interpreting
buffered against those factors, in this respect. Alterna- the data suggesting that the transition to agriculture re-
tively, males may be envisioned as less frail with regard duced health. There is compelling evidence that the sed-
to those factors but, in that they are part of a mortuary entism associated with agriculture increased exposure

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 363

to pathogens, thereby potentially increasing the risk of I99OS, this paper raises a number of important, far-
death. Many of these pathogens, however, cause acute, reaching points concerning morbidity and mortality
infectious disease which would not result in skeletal analyses of prehistoric skeletal populations. In con-
lesions, and therefore this increase in mortality would trast to the aforementioned work, it adopts a positive
not be reflected in the skeleton. Discussion has focused approach to the problems it addresses. Responses to
instead on change in diet, and it has always been diffi- Bocquet-Appel and Masset's attack on paleodemography
cult to explain to students why humans turned to agri- (Buikstra and Konigsberg I985, Van Gerven and Arme-
culture under such conditions. As Wood et al. point out, lagos I983) constituted critical advances in a field which
the work of Armelagos and Goodman has been exem- had been largely spared self-criticism if not external cri-
plary here in offering multiple hypotheses to explain the tiques (cf. Petersen I975, Howell I982). I hope that re-
archeological record, and they clearly demonstrate yet search responding to the issues raised in this study will
another set of explanations. further solidify paleodemography's status as a valid, al-
The next important step, as Wood et al. indicate, will beit specialized subfield of human demography.
be the clarification of patterns of frailty. If one aspect I concur with the authors, however, that the points
of frailty is the immunologic status of the individual, raised here are far more difficult than those cited by
which is influenced by both diet and infection, then Bocquet-Appel and Masset. While selectivity and demo-
frailty might have been affected by the transition to agri-
graphic nonstationarity are certainly nontrivial matters,
culture and it could be argued that the transition in- recent research has addressed them (cf. Sattenspiel and
creased mortality through increased frailty. This, how- Harpending I983, Saunders and Katzenberg n.d.). In my
ever, leaves unanswered the question why specific opinion, the most bedeviling of the basic conceptual
skeletal lesions and skeletal markers of stress (e.g., hypo-problems raised by the authors is that of hidden hetero-
plasia and Harris lines) are reported in higher frequen- geneity. While they effectively use hazards analysis to
cies in populations in transition to agriculture. Perhaps define and describe it, they also point to the identifica-
frailer individuals produce bone lesions more quickly? tion problem associated with testing the goodness-of-fit
We will not know, as Wood et al. point out, until we of these models to actual data. As they note, one way to
understand more about the way in which frailty, risk of test for goodness-of-fit is to model frailty distributions
death, and lesion development interact. on known biological patterns, but at the same time they
Although they address disease in ancient populations detail examples, both contemporary (e.g., diarrheal dis-
only, the question of "hidden heterogeneity of risks" is ease) and prehistoric (e.g., cribra orbitalia, enamel hy-
relevant for studies of living populations as well. For poplasia), in which counterintuitive mortality distri-
example, if the case fatality rate for a particular diseasebutions are either apparent or plausible. Problems of
is equal to the number dying of the disease/the number interpretation of the resulting ill fit between such data
with the disease, then differences in frailty in the sick and models based on biological patterns can only be
population will mean that not everyone with the disease exacerbated in prehistoric skeletal analyses, given the
is at equal risk of dying. Differences in frailty are not compounding problems of demographic nonstationarity
considered in the calculation of case fatality rates, and/or selectivity bias. To my mind, this is the crucial
mostly because of the difficulty of identifying the point distri-where all three problems converge.
bution of frailty. If we examine behavioral risks for peo- Highlighting these problems does not mean that pa-
ple who die of this disease, we may erroneously con- leodemography should be abandoned. Instead, Wood et
clude that a particular behavior or set of behaviors al. are to be congratulated for bringing them to the atten-
carries a higher risk of mortality when the differences tion of researchers. Consideration of them, while cer-
are really due to differences in frailty. tainly difficult, can only improve paleodemography.
Wood et al. deserve praise for the scientific rigor of
their multiple-hypothesis approach. In interpreting data
on prehistoric populations it is critical to walk a fine DOUGLAS H. UBELAKER

line between pat answers that simply reflect the biases Department of Anthropology, National Museum
of the investigators and complete dismissal of the field of Natural History, Smithsonian Institution,
of research. These authors do this well by reminding Washington, D.C. 20560, U.S.A. II III 92
us of the complexity of the matter while retaining en-
thusiasm for the effort of learning about health and dis- Wood et al. raise key issues in the interpretation of de-
ease in human populations. mographic and paleopathological data gleaned from the
analysis of human skeletal remains. One of the issues is
related to sampling. We have long known that sampling
ERIC ABELLA ROTH problems plague our interpretations of past human biol-
Department of Anthropology, P.O. Box 3050, ogy. If we have no way of knowing from archeological
University of Victoria, Victoria, B.C., Canada or other sources what inadequacies or complexities exist
V8W 3P5. 6 III 92 within the samples used for our studies, then we have
little opportunity to overcome them or to incorporate
In what is inevitably to become the "Farewell to Pa- them into a more sophisticated research design. I would
leodemography" (Bocquet-Appel and Masset I982) of the agree that selective mortality and hidden heterogeneity

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
364 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

are potential sources of error, but it is important to re- (or as a response to stress) are interesting and important.
member that sampling problems do not preclude the for- The well-known stature reduction in prehistoric Meso-
mulation of general statements about the disease ex- america, or even the unexpected differences in stature
perience, mortality, and/or fertility of the population between status groups at Monte Alban, where the seem-
represented. ingly higher-status individuals have shorter stature (Wil-
If interpreted properly, skeletal lesions usually indi- kinson and Norelli I98I), may be nicely explained by
cate that the individual survived a disease experience or differential mortality, but there is an opposite relation-
stress episode, and that individual is obviously healthier ship among the prehistoric Maya (Haviland I967) and
than someone who died from the same disease experi- at Illinois Hopewell (Buikstra I976). The relationship
ence without its having had an opportunity to leave its between stature and mortality could be investigated
mark on the skeleton. As the authors point out, frequen- with relatively bountiful historical data. Stature data on
cies of skeletal lesions cannot be equated with disease military conscripts could be compared with male mor-
frequencies among the living. Similarly, we have learned tality during the past 50o years in the United States and
that life tables developed from mortality data can be an even longer period in Europe.
influenced by changes in fertility, migration, and other The bottom line of the pessimistic prognosticators
factors that remain largely unknown to the investigator. from Pennsylvania is a serious one. They note that "pop-
Elucidation of the dynamic relationship among skeletal ulations cannot be compared meaningfully if their dis-
evidence of disease in mortuary samples, the distribu- tributions of frailty differ in unknown ways." Since het-
tion of ages at death, disease frequency among the living erogeneity of frailty is very likely and the distribution
represented by the skeletal sample, the demographic of frailty in prehistoric populations may be unknowable,
structure, and the health of the living populations re- we are in a pickle. The concept of "identification prob-
mains a worthwhile but elusive goal of our research. By lems" is a critically important one, and this is, as the
raising issues and clarifying assumptions of the research authors note, an area much in need of refinement. We
process, this article may enable us to advance. too often accept a skeletal series as a unified entity
rather than as a collection of subgroups. Many of our
large, complex skeletal series need to be reinvestigated
RICHARD G. WILKINSON for the existence of such subdivisions. Subdivisions with
Department of Anthropology, University at Albany, genetic bases should be identifiable through gross ana-
Albany, N.Y. 12222, U.S.A. io III 92 tomical comparisons and/or molecular methods. Subdi-
visions of a more genetically homogeneous population
Wood and colleagues offer practitioners of paleopopula- which differ in their relative frailty, however, would
tion (re)construction the provocative proposition that seem exceedingly difficult to identify. The existence of
"better health makes for worse skeletons." We are asked a group of weaning-age skeletons may suggest increased
to accept the idea that people with lesions-who are, as frailty relative to weanling diarrhea, but this same group
the authors note, indisputably dead-were healthier also provides evidence-by its very existence-of low
than those without lesions (who, by virtue of their inclu-frailty relative to neonatal and infant mortality. A low
sion in our samples, are equally dead). To the extent that frequency of lesions (from, say, porotic hyperostosis or
skeletal lesions may represent survived stress and the enamel hypoplasia) in this group informs us only that
absence of lesions may denote acute, lethal stress, their members avoided anemia and growth arrests earlier.
point is well taken. As they note, it has serious repercus- In the section "Is There Hope?" we are presented
sions on our interpretations of the health of prehistoric with a litany of confounding problems and some excel-
populations. lent suggestions for advancing our science, specifically
The concept of differential morbidity (and mortality), in the area of research into questions of the heterogene-
expressed in terms of heterogeneity of frailty, may not ity of frailty and its relationship with distributions of
be as broadly applicable as Wood et al. imply, and the deaths. The call for increased pathological research on
concept(s) of health could have been clarified. A lesion- hard-tissue lesions is also well reasoned, but it remains
free skeleton of 40 years at death should not suggest to to be seen whether specialists outside of anthropology
us the same thing as a lesion-free infant skeleton in and paleopathology can be coerced into dealing with dry
terms of relative health, frailty, or immunological com- bone. Finally, we need to understand the "role played
petence. Aggregated age-at-death data may be interpret- by cultural context in determining heterogeneous frailty
able as Wood et al. suggest, but the interpretation seems and the level of selective mortality." There are "good"
unlikely to be applicable to all individuals; some cau- skeletal series from short-term cemeteries, utilized by a
tionary statement about age-specific application is in or- single cultural unit itself characterized by relative equal-
der, especially since the examples given center on o-5- ity and, ideally, no migration to muddy the analytical
year age-groups. Pathological conditions also need to be waters. Then there are sites such as Dickson Mounds,
specified, as "better health makes for worse skeletons" representing relatively long temporal spans and a good
hardly applies to treponemal diseases, leprosy, and the deal of cultural complexity, both synchronically and dia-
like. chronically. We do need to concentrate on the "simple"
The implications of this study for the analysis of stat-sites before we can approach the issue of the heterogene-
ure and the use of stature as a measure of adaptability ity of frailty in the complex ones, but we cannot aban-

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox I 365

don the latter, which are the rule and not the exception. widespread. Since skeletal lesions are typically rare out-
And at some level, these complex sites are more inter- comes and skeletal samples are always small (at least by
esting in that they offer tantalizing evidence of bio- epidemiological standards), the probability of including
cultural change and of the interrelationships among a lesion in the sample is slim even if the associated dis-
populations. It now seems unlikely that "additional ex- ease was common in the living population. Third, selec-
cavation" can be pointed to as a potential solution to tive mortality may add cases with lesions to the skeletal
data-related problems, and we must continue to make sample at a rate that overrepresents their prevalence in
advances with the material at hand. the living population. This last point is the one stressed
in our paper, in large part because it has not been empha-
sized enough in the existing paleopathological literature.
We are left in a quandary. Does the frequency of skeletal
Reply lesions underestimate disease prevalence among the liv-
ing because of the sensitivity problem or overestimate
it because of the selectivity and specificity problems? At
JAMES W. WOOD, GEORGE R. MILNER, HENRY C. this stage, it is anyone's guess. But we can say this with
HARPENDING, AND KENNETH M. WEISS confidence: disease prevalence and skeletal-lesion fre-
Department of Anthropology, Pennsylvania State quency are two very different things and, as Ubelaker
University, University Park, Pa. I6802, U.S.A. and Hutchinson note, must never be equated.
27 Iv 92 One point should be reiterated for clarity's sake. The
presence of a particular kind of skeletal lesion is evi-
We thank all the respondents for their thoughtful com- dence for either persistence until death of the pathologi-
ments. While we cannot pretend that the problems we cal process responsible for it or survival long after a dis-
have flagged are easy to solve-much less that we have ease episode ended. It is more difficult to interpret the
solved them-it is encouraging to see how many of our absence of such lesions in ancient skeletons. Their ab-
readers are willing to give them the serious consider- sence may indicate either that the causal condition was
ation they deserve. Most commentators recognize that not present or that early death occurred before a distinc-
we are trying to be constructive rather than simply bid- tive skeletal response had time to develop. The first in-
ding farewell to paleodemography and paleopathology. terpretation is the traditional explanation, while the sec-
And most realize that we are urging consideration of ond is the alternative explored in our paper. We do not
multiple hypotheses rather than taking sides on difficult mean to suggest that the traditional explanation must
and complex issues. Because the comments are gener- always be wrong and the alternative correct; the point
ally positive, we confine our response to a few points is that it is generally impossible to tell.
that deserve amplification or clarification. Eisenberg suggests that we may be overly optimistic
Eisenberg and Hutchinson point out that, in addition about the possibility of developing an "index of frailty
to the issues we raise, sensitivity is a special concern for for either individuals or populations." On the contrary,
paleopathology. The sensitivity of a diagnostic criterion we share her pessimism. Still, we think it is worth try-
(e.g., presence of a particular kind of skeletal lesion) is ing to measure frailty, as long as it is acknowledged just
the proportion of truly diseased individuals in a sample how difficult a task it is. We also think an alternative
who are identified as diseased according to that crite- approach is more likely to yield results in the short run.
rion. As Eisenberg and Hutchinson emphasize, most This approach involves modeling the distribution of
skeletal lesions are insensitive indicators of their associ- frailty in a population, preferably on the basis of sound
ated disease processes in that they typically develop in biological principles but at the very least using flexible
only a small fraction of cases (e.g., advanced, chronic statistical distributions that will fit a wide variety of
cases of tuberculosis and treponemal infections). We cases (see Weiss i990). As we detail elsewhere (Wood
agree that sensitivity is a serious problem for paleopa- et al. i992), such models can then be used to adjust
thology but did not focus upon it simply because the statistically for differing frailty distributions when com
issue has already been widely discussed in the field. It paring populations, even in the absence of individual-
is important, however, to be cautious in responding to level measures of frailty. However, in view of the identi-
the sensitivity problem. For example, if we were to find fication problem that Roth and Wilkinson so rightly
a few cases of a particular lesion known to have low underscore, it will still be necessary to test as many
sensitivity, we might be tempted to infer that the dis- different specifications of the frailty distribution as pos-
ease causing the lesion must have been very widespread sible.
among the living. This inference is problematic for three Lukacs and McGrath make a point of fundamental
reasons. First, as Hutchinson notes, most skeletal le- importance; hidden heterogeneity and selectivity are not
sions are not only insensitive but also of low specificity.just problems for osteological research but occur in stud-
That is, many distinct pathological conditions induce ies of living populations as well. McGrath's example of
lesions that are morphologically similar, making accu- the case fatality rate is especially apt in this regard. We
rate diagnosis difficult or impossible. Second, absence of would only add that at least a substantial minority of
a particular type of lesion in a skeletal sample cannot researchers in demography and epidemiology are already
be taken as evidence that the associated disease was not struggling with these same problems (e.g., Keyfitz and

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
366 1 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

Littman I980, Heckman and Singer I984a, Trussell present even when social variation is minimal. None-
and Richards I985, Vaupel and Yashin I985b, Manton theless, Jankauskas and Cesnys are correct in stating
and Stallard I988, Gage i989). There is one important that social complexity only makes the heterogeneity-
respect in which studies of the living have an advantage: selectivity problem worse. In light of this fact, Wilkin-
when dealing with living subjects, it may be possible to son's advice that we concentrate on the simple sites
"uncover" some of the heterogeneity by actually mea- before tackling the complicated ones is well worth con-
suring variables associated with frailty. McGrath points sidering.
to immunocompetence as a major component of frailty. Several of the commentators remind us that it is im-
While immunocompetence cannot be measured on the portant, when considering the issues we raise, not to
dead, there are new procedures that allow it to be mea- lose sight of other problems that have plagued osteolo-
sured on the living, even in remote field settings (Shell gists for many years. Katzenberg and Hutchinson point
i992). In general, the most satisfying solution to the hid- to the differential preservation of skeletons by age at
den heterogeneity problem is to convert as much of the death, while Jankauskas, Cesnys, and Lukacs highlight
variation as possible from the "hidden" to the "mea- the need for detailed attention to the cultural contexts
sured" category. Unfortunately, it is precisely this ap- from which skeletal samples are drawn. Roth and Ube-
proach that is so difficult, perhaps impossible, to imple- laker emphasize the untoward effects of demographic
ment in archaeological studies. Still, the problem cannot nonstationarity on paleodemographic analyses, and Kat-
be solved by ignoring it. zenberg stresses the need to learn more about the patho-
Wilkinson suggests that we place too little emphasis logical processes that result in hard-tissue lesions. We
on the information provided by the age distribution of agree that all these concerns are fundamentally impor-
skeletal lesions. He is entirely correct that consideration tant and anticipate that osteologists will continue to
of the age pattern of lesion frequencies is an essential wrestle with these and other issues that confound the
part of any osteological interpretation. We would only interpretation of ancient bones. Such wrestling is an ab-
point out that an examination of age-specificity was the solute prerequisite for continued scientific progress.
very foundation of our interpretation of cribra orbitalia, Cohen apparently regards our comments about the or-
porotic hyperostosis, and periostitis in the Oneota sam- igins of agriculture as a "challenge" to his own position.
ple. It is equally essential, when trying to make sense This reaction is disappointing, because we had no wish
of the age distribution of various lesions, to distinguish to become embroiled in the debate-however popular
insofar as possible between hard-tissue responses that and politically charged-over whether agriculture and
were active at the time of death and those that were the state were good or bad developments in human his-
healed. While recognizing that it is often difficult to dif- tory. As we tried to make clear, the point "is not that
ferentiate active and inactive lesions when dealing with we are right and other authors are wrong" about the
dry bone, we concur with Eisenberg that it is important effects of agriculture but that multiple hypotheses need
to try. We further suggest that well-remodeled skele- to be entertained and means of distinguishing among
tal lesions-evidence of events occurring long before them found. To a large degree, Cohen does precisely
death-are more informative about the disease experi- what we advocate. He turns to other kinds of evi-
ence of past peoples than has previously been realized. dence-demographic, biomedical, and epidemiologi-
Using the Oneota skeletons, we show that particular cal-to supplement the skeletal remains. It is of great
kinds of healed skeletal lesions may indicate groups of importance, however, that such material be used criti-
individuals who experienced an elevated risk of death cally and with extreme care, since it is subject to many
relative to their contemporaries, while others may mark of the same biases and difficulties of interpretation dis-
individuals of comparatively low frailty. By examining cussed in our paper, as well as additional confounding
the age patterns of healed and active lesions and by com- problems of its own.
paring those patterns across lesions of differing etiolo- Cohen points out, for example, that there are many
gies, we can begin to link pathological changes in bone reports, both archaeological and ethnographic, of deteri-
to frailty in a meaningful way. For this reason, osteolo- orating health following the settling of formerly no-
gists should never attempt to draw inferences from skel- madic groups. Those reports cannot be dismissed out of
etal lesions considered individually. hand, although we note that the effects of sedentism on
Jankauskas and Cesnys make the interesting observa- morbidity and mortality in living populations are more
tion that heterogeneity and selectivity are likely to be- variable than Cohen acknowledges (see, e.g., Roth I985).
come more important as social complexity increases. More fundamental, however, we suggest that two en-
Doubtless they are right, as suggested in our brief re- tirely different meanings of the word "health" are being
marks on the archaeological contexts of the Norris confused. One meaning of health is something that ad-
Farms (Oneota) and Dickson Mounds sites. We must not versely affects an individual's relative risk of death. In
take it for granted, however, that simple, small-scale so- conformity with what has become standard demo-
cieties never contain significant variation in frailty. One graphic usage, we call this form of health "frailty." A
important lesson of the last 2o years of research in popu- newborn who experiences an unusually high probability
lation genetics is that even small, localized, inbred hu- of dying during the first year of life is thus "frail." An-
man populations retain a considerable amount of genetic other meaning of health is general physical condition,
variation, some portion of which may have important which if poor might be termed "decrepitude." A thin
phenotypic effects. Thus, biological variation may be and anemic child with chronic skin lesions and respira-

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 367

tory disease is "decrepit." We assume, with Cohen, that By presenting this example, we do not mean to imply
physical decrepitude can leave signatures in the skeleton that sedentization is always accompanied by a decline
and dentition, especially if the illness is survived (e.g., in mortality, for evidence from other areas suggests that
enamel hypoplasia) or develops into a chronic condition the opposite sometimes occurs. In general, we consider
(e.g., skeletal tuberculosis). it extremely unlikely that sedentization always has sim-
One fundamental point that we have tried to make in ple, predictable effects on mortality or, for that matter,
this paper is that these two meanings of health need fertility. In fact, we would go farther and suggest that
have no simple relationship to each other. We suggest the transition to settled life differs in important respects
that in many groups with poor nutritional status- from one location to another. Among the Northern
perhaps including many foragers-children are so frail Cree, settlement led to a rapid increase in fertility, but
that they do not have sufficient reserves to support only because the Canadian government supplied for-
chronic decrepitude, which takes time to develop. Infant mula and milk, thus shortening the duration of lacta-
and childhood death, when it occurs, happens quickly tional infecundability (Romaniuk 198I). In other loca-
without leaving any hard-tissue signature. With more tions, permanent settlement may result in a marked and
food, in contrast, children are capable of sustaining immediate rise in mortality, but only because certain
themselves in a decrepit state-a statement that seems infectious diseases such as cholera happen to be epi-
paradoxical only if decrepitude and frailty are confused. demic in the region at the time of settlement. Very pos-
If decrepitude lasts long enough before death, then sibly, the development of settled agriculture in the Old
markers of this state could show up in the archaeologi- World, especially where domesticated animals and non-
cal record. But actual hazards of death, which are propor- human milk were available, had different consequences
itonal to frailty, are lower in such a situation. for fertility and mortality than the parallel process in
For example, Harpending and Wandsnider (i982) have the New World. We will never know unless we reject
described !Kung Bushmen living in settlements around the claim that a uniform response always occurs and its
Ghanzi in Botswana. The children in these settled direction is self-evident from existing data. In sum, it is
groups were noticeably decrepit compared with their no- unrealistic to think that the transition from foraging to
madic counterparts in the bush (Truswell and Hansen agriculture must always have had the same demographic
I976): many were thin and inactive, and respiratory dis- and biomedical effects. Indeed, we believe that future
ease and skin and eye infections were common. Yet the research should abandon the concept of "sedentization"
estimated infant mortality rate in that undeniably de- as a single process with uniform biological effects and
crepit population was half that observed in the bush, concentrate instead on the specific mechanisms govern-
and mortality before age I5 years was one-third that in ing demography and health in each population.
the bush. The sedentary children were apparently more All the problems we have raised can be summed up
decrepit but less frail, despite the fact that they had no very simply: data do not speak for themselves. This is
better access to effective medical care than did nomadic an elementary lesson that has never had the impact it
children. The distinction between frailty and decrepi- deserves. The information contained in empirical data
tude helps explain a paradox first noted by Howell can be extracted and interpreted only when specific
(I979): nomadic !Kung appear to be in good condition models are applied to them. To borrow a wonderful aph-
but have mortality rates at all ages that are substantially orism from Nathan Keyfitz (I975), "No model, no un-
higher than those of surrounding settled groups. derstanding." In one form or another, everyone uses
Pennington (i992; Pennington and Harpending i992), models, even if only vague and nonmathematical ones,
using data reported by Howell (I979), finds the same when thinking about data. All too often, however, these
response to sedentization in Ngamiland. Herero pasto- models are implicit and thus unavailable for critical
ralists entered !Kung areas in Ngamiland in the mid- evaluation. It has seemed so obvious that early ages at
I950s, and many !Kung settled at Herero cattleposts, death imply high mortality or that diseased skeletons
where milk and other foods were widely available. The imply diseased populations that little thought has been
result of settlement, as Pennington shows, was a drop given to the assumptions involved in these inferences.
in infant mortality of about 5o%. The largest decline We have tried to show that even very simple alternative
was in the age-interval I-4 years, corresponding to the models can lead to conclusions that are the exact oppo-
usual age at weaning, when mortality dropped to one- site of our immediate intuitions. Of course, as several
fourth its level before arrival of the Herero. Not only of our commentators point out, this problem of overem-
were these !Kung newly sedentary, thus meeting the phasis on data at the expense of theory and analysis is by
conditions of Cohen's model, but they were living in no means restricted to archaeological osteology. Papers
close contact with cattle, a source of zoonoses such as similar to ours could have been written for demography,
tuberculosis. An archaeological osteologist of the future epidemiology, genetics, paleoanthropology, evolution-
might well find much more skeletal evidence of decrepi- ary ecology, or any other field of empirical research. But
tude following settlement and yet have no way to infer that does not lessen the importance of the message for
that this increased decripitude was accompanied by a paleodemography and paleopathology. The challenge
dramatic decline in early childhood mortality. This is now is for researchers to come up with reasonable bio-
particularly true if the age composition of the sample is logical models that will help us make sense of the way
ignored, as happens all too often in paleopathological skeletal samples are formed and, hence, what they
studies. mean.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
368 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October 1992

References Cited . I984. "Subsistence and health in the lower Illinois Val-
ley: Osteological evidence," in Paleopathology at the origins of
ACSADI, G., AND J. NEMESKERI. I970. History of human life agriculture. Edited by M. N. Cohen and G. J. Armelagos, pp.
span and mortality. Budapest: Akademiai Kiad6. 235-69. Orlando: Academic Press.
ALTER, G., AND J. C. RILEY. I989. Frailty, sickness, and death:COOK, D. C., AND J. E. BUIKSTRA. I979. Health and differen-
Models of morbidity and mortality in historical populations. tial survival in prehistoric populations: Prenatal dental defects.
Population Studies 43:25-45. American Journal of Physical Anthropology 5 I:649-64.
ANGEL, J. L. I969. The bases of paleodemography. American
Journal of Physical Anthropology 30:427-38. ZOFF. I976. The etiology of porotic hyperostosis among the
ARMELAGOS, GEORGE J. I969. Disease in ancient Nubia. Sci- prehistoric and historic Anasazi Indians of southwestern
ence i63:255-59. [MAK] United States. American Journal of Physical Anthropology
ARMELAGOS, G. J., A. GOODMAN, AND K. H. JACOBS. I978. 44:477-87.
"The ecological perspective in disease," in Health and the hu- ELANDT-JOHNSON, R. C., AND N. L. JOHNSON. I980. Sur-
man condition. Edited by M. Logan and E. E. Hunt, pp. 7I-84. vival models and data analysis. New York: Wiley.
North Scituate: Duxbury. EVELETH, P. B., AND J. M. TANNER. I990. 2d edition. World-
BEHRENSMEYER, A. K., AND ANDREW HILL. I980. Fossils in wide variation in human growth. Cambridge: Cambridge tJni-
the making. Chicago: University of Chicago Press. [JRLI versity Press.
BLAKELY, R. L., AND B. DETWEILER-BLAKELY. I989. The im- GAGE, T. B. I989. Bio-mathematical approaches to the study of
pact of European diseases in the sixteenth-century Southeast: human variation in mortality. Yearbook of Physical Anthro-
A case study. Midcontinent Journal of Archaeology I4:62-89. pology 32:i85-2I4.
BOCQUET-APPEL, J-P., AND C. MASSET. i982. Farewell to pa- . I990. Variation and classification of human age patterns
leodemography. Journal of Human Evolution I I:32I-33. of mortality: Analysis using competing hazards models. Hu-
.I985. Small populations: Demography and paleoanthro- man Biology 62:589-6I7.
pological inferences. Journal of Human Evolution I4:683-9I. GOODMAN, A. H., AND G. J. ARMELAGOS. I988. Childhood
[DLHI stress and decreased longevity in a prehistoric population.
BRASS, W. I97I. "On the scale of mortality," in Biological as- American Anthropologist 90:936-44.
pects of demography. Edited by W. Brass, pp. 69-II0. London: GOODMAN, A. H., J. LALLO, G. J. ARMELAGOS, AND J. C.
Taylor and Francis. ROSE. I984. "Health changes at Dickson Mounds, Illinois (A.D.
I975. Methods for estimating fertility and mortality 950-I300)," in Paleopathology at the origins of agriculture.
from limited and defective data. Chapel Hill: Carolina Popula- Edited by M. N. Cohen and G. J. Armelagos, pp. 27I-305. Or-
tion Center. lando: Academic Press.
GOODMAN, A. H., AND J. C. ROSE. I990. Assessment of sys-
BUIKSTRA, J. E. I976. Hopewell in the Lower Illinois Valley.
Northwestern University Archaeological Program Scientific Pa- tematic physiological perturbations from dental enamel hypo-
pers 2. plasias and associated histological structures. Yearbook of
BUIKSTRA, J. E., AND D. C. COOK. I980. Palaeopathology: An Physical Anthropology 33:59-I IO.
American account. Annual Review of Anthropology 9:433-70. GORDON, C. C., AND J. E. BUIKSTRA. I98I. Soil pH, bone
BUIKSTRA, J. E., AND L. W. KONIGSBERG. I985. Paleodemog- preservation, and sampling bias at mortuary sites. American
raphy: Critiques and controversies. American Anthropologist Antiquity 46:566-7I.
87:3i6-33. GORDON, J. E., I. D. CHITKARA, AND J. B. WYON. I963.
BUIKSTRA, J. E., L. W. KONIGSBERG, AND J. BULLINGTON. Weanling diarrhea. American Journal of Medical Science
I986. Fertility and the development of agriculture in the pre- 245:345-77.
historic Midwest. American Antiquity 5i:528-46. GORDON, J. E., J. B. WYON, AND W. ASCOLI. I967. The
BUIKSTRA, J. E., AND J. H. MIELKE. I985. "Demography, diet, second-year death rate in less developed countries. American
and health," in The analysis of prehistoric diets. Edited by Journal of Medical Science 254:357-80.
R. I. Gilbert and J. H. Mielke, pp. 359-422. Orlando: Academic GRIN, E. I. I956. Endemic syphilis and yaws. Bulletin of the
Press. World Health Organization I5:959-73. [DLH]
CHANDRA, S., AND R. K. CHANDRA. I986. Nutrition, immune GUAGLIARDO, M. F. i982. Tooth crown size differences be-
response, and outcome. Progress in Food and Nutritional Sci- tween age groups: A possible new indicator of stress in skele-
ence IO:I-65. tal samples. American Journal of Physical Anthropology 58:
CHEN L. C-, A. K. CHOWDHURY, AND S. L. HUFFMANN. 383-89. I980.
HARN, A.D. I980.
Anthropometric assessment of energy-protein malnutrition andThe prehistory of Dickson Mounds: The
subsequent risk of mortality among preschool-aged excavation. Illinois State Museum Reports of Investi-
gations 35.
American Journal of Clinical Nutrition 33:I836-43.
COALE, A. J. I957. How the age distribution of a human popula-
tion is determined. Cold Spring Harbor Symposia on Quanti- tion structures of Ghanzi and Ngamiland !Kung," in Current
tative Biology 22:83-88. developments in anthropological genetics, vol. 2, Ecology and
COALE, A. J., AND P. DEMENY. I983. 2d edition. Regional population structure. Edited by M. H. Crawford and J. H.
model life tables and stable populations. New York: Aca- Mielke, pp. 29-50. New York: Plenum Press.
demic Press. HAVILAND, W. A. I967. Stature at Tikal, Guatemala: Implica-
COHEN, M. N. I989. Health and the rise of civilization. New tions for ancient Maya demography and social organization.
Haven: Yale University Press. American Antiquity 32:3I6-25. [RGW]
COHEN, M. N., AND G. J. ARMELAGOS. Editors. I984a. Paleo- HECKMAN, J. J., AND B. SINGER. I984a. A method for min-
pathology at the origins of agriculture. Orlando: Academic imizing the impact of distributional assumptions in economet-
Press. ric models for duration data. Econometrica 52:27I-320.
. I984b. "Paleopathology at the origins of agriculture: Edi- . I984b. Econometric duration analysis. Journal of Econo-
tors' summation," in Paleopathology at the origins of agricul- metrics 24:63-I32.
ture. Edited by M. N. Cohen and G. J. Armelagos, pp. 585-60I. HECKMAN, J. J., AND J. R. WALKER. I987. "Using goodness of
Orlando: Academic Press. fit and other criteria to choose among competing duration
COOK, D. C. I98I. "Mortality, age-structure, and status in the models: A case study of Hutterite data," in Sociological meth-
interpretation of stress indicators in prehistoric skeletons: A odology 1987. Edited by C. Clogg, pp. 247-307. Washington,
dental example from the lower Illinois Valley," in The archae- D.C.: American Sociological Association.
ology of death. Edited by R. Chapman, I. Kinnes, and K. HEMPHILL, BRIAN E., JOHN R. LUKACS, AND K. A. R. KEN-
Randsborg, pp. I34-44. Cambridge: Cambridge University NEDY. 1991. "Biological adaptations and affinities of Bronze
Press. Age Harappans," in Harappa excavations I986-I990: A multi-

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
WOOD ET AL. The Osteological Paradox | 369

disciplinary approach to third millennium urbanism. Edited man remains from the Gangetic Valley; Mahadaha. Edited by
by Richard H. Meadow, pp. I37-82. Madison: Prehistory Press. K. A. R. Kennedy, J. R. Lukacs, and C. B. Burrow. Cornell Uni-
[JRLI versity South Asia Program Occasional Papers and Theses. In
HEYWOOD, P. i982. The functional significance of malnutri- press. [JRLI
tion: Growth and prospective risk of death in the highlands of LUKACS, JOHN R., AND M. R. JOSHI. I992. Enamel hypoplasia
Papua New Guinea. Journal of Food and Nutrition 39:I3-I9. prevalence in three ethnic groups of northwest India: A test of
HOROWITZ, S., G. J. ARMELAGOS, AND K. WACHTER. I988. daughter neglect and a framework for the past. Journal of Pa-
On generating birth rates from skeletal populations. American leopathology. In press. [JRLI
Journal of Physical Anthropology 76:I89-96. LUKACS, JOHN R., AND J. N. PAL. i992. Dental anthropology
HOWELL, N. I979. Demography of the Dobe !Kung. New York: of Mesolithic hunter-gatherers: A preliminary report on the
Academic Press. Mahadaha and Sarai Nahar Rai dentitions. Man and Environ-
.i982. Village composition implied by a paleodemographic ment. In press. [JRLI
life table: The Libben site. American Journal of Physical An- MCHENRY, HENRY M. I968. Transverse lines in long bones of
thropology 59:263-69. prehistoric California Indians. American Journal of Physical
HSIAO, C. I990. "Identification," in The new Palgrave: Econo- Anthropology 29: I-I 8. [MAKI
metrics. Edited by J. Eatwell, M. Milgate, and P. Newman, pp. MANTON, K. G., AND E. STALLARD. I984. Recent trends in
95-I00. London: Macmillan. mortality analysis. New York: Academic Press.
HUMMERT, J. R., AND D. P. VAN GERVEN. I983. Skeletal . i988. Chronic disease modelling. London: Charles Griffin.
growth in a medieval population from Sudanese Nubia. Ameri- MARTIN, D. L., A. H. GOODMAN, AND G. J. ARMELAGOS.
can Journal of Physical Anthropology 60:47I-78. i985. "Skeletal pathologies as indicators of quality and quan-
HUSS-ASHMORE, R., A. H. GOODMAN, AND G. J. ARME- tity of diet," in The analysis of prehistoric diets. Edited by
LAGOS. i982. Nutritional inference from paleopathology. Ad- R. I. Gilbert and J. H. Mielke, pp. 227-79. Orlando: Academic
vances in Archaeological Methods and Theory 5:395-474. Press.
ISCAN, M. Y. Editor. I989. Age markers in the human skeleton.
Springfield: Thomas. Reliability of age at death in the Hamann-Todd collection: Va-
ISCAN, M. Y., AND S. R. LOTH. I989. "Osteological manifesta-lidity of subselection procedures used in blind tests of the
tions of age in the adult," in Reconstruction of life from the summary age technique. American Journal of Physical Anthro-
skeleton. Edited by M. Y. Iscan and K. A. R. Kennedy, pp. pology 83:349-57.
23-40. New York: Alan R. Liss. MENSFORTH, R. C. I985. Relative tibia long bone growth in the
JOHANNSON, S. R., AND S. HOROWITZ. I986. Estimating mor- Libben and Bt-5 prehistoric skeletal populations. American
tality in skeletal populations: Influence of the growth rate on Journal of Physical Anthropology 68:247-62.
the interpretation of levels and trends during the transition MENSFORTH, R. C., C. 0. LOVEJOY, J. W. LALLO, AND G. J.
to agriculture. American Journal of Physical Anthropology 7I: ARMELAGOS. I978. The role of constitutional factors, diet,
233-50. and infectious disease in the etiology of porotic hyperostosis
JOHNSTON, F. E., AND L. 0. ZIMMER. I989. "Assessment of and periosteal reactions in prehistoric infants and children.
growth and age in the immature skeleton," in Reconstruction Medical Anthropology 2: I-5 9.
of life from the skeleton. Edited by M. Y. Iscan and K. A. R. MIETTINEN, O. S. I985. Theoretical epidemiology. New York:
Kennedy, pp. ii-2i. New York: Alan R. Liss. Wiley.
KEYFITZ, N. I975. How do we know the facts of demography? MILLER, BARBARA. I985. The endangered sex: Neglect of fe-
Population and Development Review i:267-88. male children in rural North India. Ithaca: Cornell University
.I985. 2d edition. Applied mathematical demography. Press. [JRLI
Berlin: Springer-Verlag. MILNER, G. R., D. A. HUMPF, AND H. C. HARPENDING. I989.
KEYFITZ, N., AND G. LITTMAN. I980. Mortality in a heteroge- Pattern matching of age-at-death distributions in paleodemo-
neous population. Population Studies 33:333-43. graphic analysis. American Journal of Physical Anthropology
KIELMAN, A., AND C. MC CORD. I978. Weight for age as an in- 80:49-58.
dex of risk of death in children. Lancet I:I247-50. MILNER, G. R., AND V. G. SMITH. I990. "Oneota human skel-
LALLO, J., J. C. ROSE, AND G. J. ARMELAGOS. I980. "An eco- etal remains," in Archaeological investigations at the Morton
logical interpretation of variation in mortality within three Village and Norris Farms 36 cemetery, pp. III-48. Illinois
prehistoric American Indian populations from Dickson State Museum Reports of Investigations 45.
Mounds," in Early Native Americans. Edited by D. L. Brow- MOORE, J. A., A. C. SWEDLUND, AND G. J. ARMELAGOS.
man, pp. 203-38. The Hague: Mouton. I975. "The use of life tables in paleodemography," in Popula-
LANCASTER, T. I990. The econometric analysis of transition tion studies in archaeology and biological anthropology: A
data. Cambridge: Cambridge University Press. symposium. Edited by A. C. Swedlund, pp. 57-70. American
LARSEN, C. S. I987. Bioarchaeological interpretations of subsis-
Antiquity Memoir 30.
tence economy and behavior from human skeletal remains. MORSE, D. I978. Ancient disease in the Midwest. Illinois State
Advances in Archaeological Methods and Theory I0:339-445. Museum Reports of Investigations i5.
LAST, j. M. I988. 2d edition. A dictionary of epidemiology. MURRAY,
York: Oxford University Press. AND D. J. DE VILLIERS. I956. Endemic syphilis in the Bak-
LONG, K. Z., J. W. WOOD, K. M. WEISS, R. WILSON, AND H. wena Reserve of the Bechuanaland Protectorate: A report on
DUPONT. I992. Hazards analysis of the effects of idiotypic mass examination and treatment. Bulletin of the World Health
breastmilk antibodies and supplemental foods on diarrhea posi- Organization I5:975-IO39. [DLHI
tive for enterotoxigenic E. coli in a cohort of Mexican infants. NAM, C. B. I990. "Mortality differentials from a multiple-
MS. cause-of-death perspective," in Measurement and analysis of
LOVEJOY, C. O., R. S. MEINDL, T. R. PRYZBECK, T. S. BAR- mortality: New approaches. Edited by J. Vallin, S. D'Souza,
TON, K. G. HEIPLE, AND D. KOTTING. I977. Paleodemogra- and A. Palloni, pp. 328-42. Oxford: Oxford University Press.
phy of the Libben site, Ottawa County, Ohio. Science I98: ORTNER, DONALD J. I99I. "Theoretical and methodological is-
29I-93. sues in paleopathology," in Human paleopathology: Current
LUKACS, JOHN R. I992. Dental paleopathology and agricultural syntheses and future options. Edited by D. J. Ortner and A. C.
intensification in South Asia: New evidence from Bronze Age Aufderheide, pp. 5-ii. Washington, D. C.: Smithsonian Insti-
Harappa. American Journal of Physical Anthropology 87: tution Press. [MAKI
I33-50. [JRL] ORTNER, D. J., AND A. C. AUFDERHEIDE. Editors. I99I. Hu-
LUKACS, JOHN R., AND BRIAN E. HEMPHILL. I992. "The den- man paleopathology: Current syntheses and future options.
tal anthropology of Mesolithic South Asians: A descriptive and Washington, D.C.: Smithsonian Institution Press.
analytic study of the Mahadaha dentition," in Mesolithic hu- ORTNER, D. J., AND W. G. J. PUTS CHAR. 1985. Revised edi-

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms
370 CURRENT ANTHROPOLOGY Volume 33, Number 4, August-October I992

tion. Identification of pathological conditions in human skele- . I989b. Porotic hyperostosis: Relationship between orbital
tal remains. Smithsonian Institution Contributions to Anthro- and vault lesions. American Journal of Physical Anthropology
pology 28. 80:I87-94. [MAK]
PAINE, R. R. I989. Model life table fitting by maximum likeli- SUCKLING, G. I989. Developmental defects of enamel: Histori-
hood estimation: A procedure to reconstruct paleodemographic cal and present-day perspectives on their pathogenesis. Ad-
characteristics from skeletal age distributions. American Jour- vances in Dental Research 3:87-94.
nal of Physical Anthropology 79:5I-6I. TRUSSELL, J., AND T. RICHARDS. I985. "Correcting for un-
PALKOVICH, A. M. I985. "Interpreting prehistoric morbidity in- measured heterogeneity in hazard models using the Heckman-
cidence and mortality risk: Nutritional stress at Arroyo Hondo Singer procedure," in Sociological methodology i985. Edited
Pueblo, New Mexico," in Health and disease in the prehistoric by N. B. Tuma, pp. 242-76. San Francisco: Jossey-Bass.
Southwest. Edited by C. F. Merbs and R. J. Miller, pp. I39-64. TRUSSELL, J., AND G. RODRIGUEZ. I990. "Heterogeneity in
Arizona State University Anthropological Research Papers 34. demographic research," in Convergent issues in genetics and
.I987. Endemic disease patterns in paleopathology: Po- demography. Edited by J. Adams, A. I. Hermalin, D. Lam, and
rotic hyperostosis. American Journal of Physical Anthropology P. E. Smouse, pp. III-32. Oxford: Oxford University Press.
74: 527-37. TRUSWELL, A. S., AND J. D. L. HANSEN. I976. "Medical re-
PENNINGTON, R. I992. Did food increase fertility? An evalua- search among the !Kung," in Kalahari hunter-gatherers. Edited
tion of !Kung and Herero history. Human Biology. In press. by R. B. Lee and I. DeVore, pp. I66-94. Cambridge: Harvard
PENNINGTON, R., AND H. C. HARPENDING. I992. The struc- University Press.
ture of an African pastoralist community: Demography, his- UBELAKER, D. H. I974. Reconstruction of demographic profiles
tory, and ecology of the Ngamiland Herero. Oxford: Oxford from ossuary skeletal samples: A case study from the Tidewa-
University Press. In press. ter Potomac. Smithsonian Institution Contributions to Anthro-
PETERSEN, W. I975. A demographer's view of prehistoric de- pology I8.
mography. CURRENT ANTHROPOLOGY I6:227-45. [EAR] 1i982. "The development of American paleopathology," in
PUFFER, R. R., AND C. V. SERRANO. I973. Patterns of mortal- A history of American physical anthropology, I930-I980. Ed-
ity in childhood. Pan American Health Organization Scientific ited by F. Spencer, pp. 337-56. New York: Academic Press.
Publication 262. VAN GERVEN, D. P., AND G. J. ARMELAGOS. I983. "Farewell
ROMANIUK, A. I98I. Increase in natural fertility during the to paleodemography?" Rumors of its death have been greatly
early stages of modernization: Canadian Indians case study. exaggerated. Journal of Human Evolution I2:353-60.
Demography I8:I57-72. VAUPEL, J. w. I988. Inherited frailty and longevity. Demogra-
ROOSEVELT, A. C. I984. "Population, health, and the evolution phy 25 :277-87.
of subsistence: Conclusions from the conference," in Paleopa- .I990. Relatives' risks: Frailty models of life history data.
thology at the origins of agriculture. Edited by M. N. Cohen Theoretical Population Biology 37:220-34.
and G. J. Armelagos, pp. 27I-305. Orlando: Academic Press. VAUPEL, J. W., K. G. MANTON, AND E. STALLARD. I979. The
ROSE, JEROME C., AND TED A. RATHBUN. I987. Preface to impact of heterogeneity in individual frailty on the dynamics
Afro-American Biohistory Symposium. American Journal of of mortality. Demography I6:439-54.
Physical Anthropology 74:I77-78. [MAK] VAUPEL, J. W., AND A. I. YASHIN. I985a. Heterogeneity's
ROTH, E. A. I985. A note on the demographic concomitants of ruses: Some surprising effects of selection on population dy-
sedentism. American Anthropologist 87:380-82. namics. American Statistician 39:I76-85.
ST. HOYME, L. E., AND M. Y. ISCAN. I989. "Determination of . Ig8sb. "The deviant dynamics of death in heterogeneous
sex and race: Accuracy and assumptions," in Reconstruction of populations," in Sociological methodology i985. Edited by
life from the skeleton. Edited by M. Y. I?can and K. A. R. Ken-
N. B. Tuma, pp. I79-2II. San Francisco: Jossey-Bass.
nedy, pp. 53-93. New York: Alan R. Liss. WALDRON, T. I987. "The relative survival of the human skele-
SATTENSPIEL, L., AND H. C. HARPENDING. I983. Stable pop- ton: Implications for palaeopathology," in Death, decay and
ulations and skeletal age. American Antiquity 48:489-98. reconstruction: Approaches to archaeology and forensic sci-
SAUNDERS, S. R., AND M. A. KATZENBERG. n.d. The skeletal ence. Edited by A. Boddinton, A. N. Garland, and R. C. Jana-
biology of past peoples: Advances in research methods. New way, pp. 55-64. Manchester: Manchester University Press.
York: Wiley. In press. [EAR] WALKER, P. L. I985. "Anemia among prehistoric Indians of the
SCRIMSHAW, N. S. I977. Effect of infection on nutrient require- American Southwest," in Health and disease in the prehistoric
ments. American Journal of Clinical Nutrition 30:I536-44. Southwest. Edited by C. F. Merbs and R. J. Miller, pp. I39-64.
SCRIMSHAW, N. S., C. E. TAYLOR, AND J. E. GORDON. I968. Arizona State University Anthropological Research Papers 34.
Interactions of nutrition and infection. Geneva: World Health WALKER, P. L., J. R. JOHNSON, AND P. M. LAMBERT. I988.
Organization. Age and sex biases in the preservation of human skeletal re-
SHELL, B. K. I992. Nutrition-related immunosuppression among mains. American Journal of Physical Anthropology 76:I83-88.
nomadic Turkana pastoralists (abstract). American Journal of WEISS, K. M. I973. Demographic models for anthropology.
Physical Anthropology, suppl., I4:I50. American Antiquity Memoir 27.
SHIPMAN, PAT. I980. The life history of a fossil. Cambridge: .I990. The biodemography of variation in human frailty.
Harvard University Press. [JRL] Demography 27:I85-206.
SKINNER, MARK, AND ALAN H. GOODMAN. I992. "Anthropo- WILKINSON, R. G., AND R. J. NORELLI. I98I. A biocultural
logical uses of developmental defects of enamel," in Skeletal analysis of social organization at Monte Alban, Oaxaca. Ameri-
biology of past peoples: Research methods. Edited by S. R. can Antiquity 46:743-58. [RGW]
Saunders and M. A. Katzenberg, pp. I53-74. New York: WILMSEN, E. N., AND J. R. DENBOW. I990. Paradigmatic his-
Wiley-Liss. [MAK] tory of San-speaking peoples and current attempts at revision.
SOLWAY, J. S., AND R. B. LEE. I990. Foragers, genuine or spuri-
ous? Situating the Kalahari San in history. CURRENT ANTHRO-
WOOD, j. W. I990. Fertility in anthropological populations. An-
POLOGY 3 I:I09-46. nual Review of Anthropology I9:2II-42.
STEINBOCK, R. T. I976. Paleopathological diagnosis and inter- WOOD, J. W., D. J. HOLMAN, K. M. WEISS, A. V. BUCHANAN,
pretation. Springfield: Thomas. AND B. LEFOR. I992. Hazards models for human population
STUART-MACADAM, P. I985. Porotic hyperostosis: Representa- biology. Yearbook of Physical Anthropology. In press.
tive of a childhood condition. American Journal of Physical WOOD, J. W., P. L. JOHNSON, AND K. L. CAMPBELL. I985.
Anthropology 66:39I-98. Demographic and endocrinological aspects of low natural fer-
. I989a. "Nutritional deficiency diseases: A survey of tility in highland New Guinea. Journal of Biosocial Science
scurvy, rickets, and iron-deficiency anemia," in Reconstruction I7:57-79.
of life from the skeleton. Edited by M. Y. Iscan and K. A. R. ZIMMERMAN, M. R., AND M. A. KELLEY. 1982. Atlas of hu-
Kennedy, pp. 20I-22. New York: Alan R. Liss. man paleopathology. New York: Praeger.

This content downloaded from on Mon, 14 Nov 2016 16:43:20 UTC
All use subject to http://about.jstor.org/terms