AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 149:181–192 (2012)

Estimating Age From Adult Occlusal Wear:

A Modification of the Miles Method
Cassandra C. Gilmore* and Mark N. Grote

Department of Anthropology, University of California, Davis; Davis, CA 95616

KEY WORDS tooth wear; age estimation; hunter-gatherers

ABSTRACT The Miles method of age estimation with age categories estimated from traditional methods;
relies on molar wear to estimate age and is widely used in this suggests that the modifications have not hampered
bioarcheological contexts. However, because the method the ability of the Miles method to estimate age even in
requires physical seriation and a sample of subadults to heterogeneous samples. As expected, the error increases
estimate wear rates it cannot be applied to many samples. with age and in populations with smaller sample sizes.
Here, we modify the Miles method by scoring occlusal These modifications allow the Miles method to be applied
wear and estimating molar wear rates from adult wear to skeletal samples of adult crania that were previously
gradients in 311 hunter-gatherers and provide formulae only amenable to cranial suture age estimation, and
to estimate the error associated with each age estimate. A importantly, provide a measure of uncertainty for each
check of the modified method in a subsample (n 5 22) age estimate. Am J Phys Anthropol 149:181–192, 2012.
shows that interval estimates overlap in all but one case C 2012
V Wiley Periodicals, Inc.

The ability to estimate adult age-at-death from skele- 1971; Griffin et al., 1979; Ruff, 1981; Kieser et al., 1983;
tal remains is critical to bioarcheological studies (e.g., Levers and Darling, 1983; Lovejoy, 1985; Whittaker
paleodemography and paleopathology) and in forensic et al., 1985, 1987; Harper, 1994; Visser, 1994; Borrman
contexts. As such, a staggering volume of literature is et al., 1996; Sullivan, 2004, 2005; Papageorgopoulou
devoted to methods of age estimation; commonly used et al., 2009; Sagona et al., 2010; Cole and Waldron, 2011)
methods include scoring the degeneration of the pubic and tests of the method routinely suggest that it per-
symphysis (e.g., Acsádi and Nemeskéri, 1970; Meindl et forms as well or better than other methods of skeletal
al., 1985; Brooks and Suchey, 1990) and auricular sur- age estimation (Nowell, 1978; Kieser et al., 1983; Love-
face (e.g., Lovejoy et al., 1985b; Buckberry and Chamber- joy, 1985; Lovejoy et al., 1985a; but see Santini et al.,
lain, 2002; Igarashi et al., 2005), cranial suture fusion 1990). Practically, the Miles method requires that the
(e.g., Meindl and Lovejoy, 1985), and dental wear (e.g., entire collection be seriated by tooth wear (Helm and
Miles, 1962; Brothwell, 1981; Walker et al., 1991). Given Prydso, 1979) and that it include roughly 20 juveniles to
the superior preservation of dental tissues and the his- make reliable wear rate estimates (Nowell, 1978).
toric preference of early anthropologists to collect skulls Unfortunately, many skeletal collections are curated
over postcranial elements (Walker et al.1991), the ability by separate institutions and therefore cannot be seriated
to estimate age from the dentition alone can be particu- together according to wear. Furthermore, juvenile
larly useful for researchers studying past populations. remains are more fragile than those of adults and are of-
Age estimation from dental wear is further distinguished ten less well preserved in archaeological assemblages.
from other methods in that it is based on an indicator This study was necessitated by just such a situation:
(enamel) that does not have the potential to remodel while collecting data to compare antemortem tooth loss
(Boldsen, 2005), and that it can be calibrated to the pop- frequencies in hunter-gatherers, one of us (CCG) was
ulation in question, rather than requiring the applica- obliged to use collections composed primarily (sometimes
tion of aging standards from a reference population exclusively) of adult crania and curated by diverse repo-
(Lovejoy et al., 1985a). This combination of qualities
likely explains why dental wear was found to be the best
single age indicator when compared with pubic symphy-
sis, auricular surface, femoral radiographs, and cranial Grant sponsors: The L.S.B. Leakey Foundation, the University of
suture closure in a known-age sample (Lovejoy et al., California, Davis, Department of Anthropology Summer Fellow-
1985a). ships, and the UC Davis Institute of Governmental Affairs.
Though researchers have long noted the value of den-
tal wear to classify individuals into broad age categories, *Correspondence to: Cassandra Gilmore, Department of Anthro-
Miles (1962, 1963, 1969, 2001) was the first to use molar pology, University of California, Davis, One Shields Avenue, Davis,
wear to estimate age systematically by progressively CA 95616, USA.
extrapolating wear rates from younger individuals—for E-mail: [email protected]
whom age can be estimated with relative certainty—to
older individuals where the determination of age is more Received 1 December 2011; accepted 22 May 2012
problematic. The Miles method, or the aging chart
accompanying the original publication, is commonly used DOI 10.1002/ajpa.22106
Published online 5 July 2012 in Wiley Online Library
to estimate age in bioarcheological contexts (Wolpoff, (wileyonlinelibrary.com).

C 2012
V WILEY PERIODICALS, INC.
 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
182 C.C. GILMORE AND M.N. GROTE
TABLE 1. Sample composition molar had been in functional occlusion. For example, if
Population N the second molars are just erupting then the individual is
roughly 12 years old, and the first molars have been in
Afalou 23 occlusion for about 6 years. The wear on the first molars is
Alaska 41 thus equivalent to 6 years of wear. Miles seriated the sam-
Australian/Tasmanian 70 ple of adults and projected the wear rates observed in the
Buriat 27
Chukchi 16 subadult sample to estimate the functional ages of young
Sami 13 adult molars. The functional age based on wear stage plus
Fuegian 9 the age of eruption results in an estimate of the age of an
Khoi-San 38 individual. Using the estimated functional ages of young
Native Californian 74 adult molars, Miles was then able to identify stages of
Total 311 wear at older functional ages. This process repeats, build-
ing upon successively older individuals to estimate func-
tional ages of progressively more worn molars. The accu-
racy of the method is expected to diminish with each
sitories across North America and Europe. The ages of extrapolation. During the course of his investigation,
individuals in these collections are unknown, but a com- Miles observed a consistent relationship between wear
parison of antemortem tooth loss between populations rates of the first, second, and third molars; he estimated
without considering sample age structures would ignore that it would take the second molar 6.5 years and the
a potentially large source of variation. third molar 7 years to reach a wear stage that took the
The purpose of the present study is to suggest a prac- first molar only 6 years to achieve.
tical modification of the Miles method using a widely Miles (1962) states two primary assumptions of this
accepted method of estimating wear rates from wear gra- method: 1) individuals belonging to a single biological
dients in adults (e.g., Smith, 1972; Lunt, 1978; Benfer population have relatively consistent wear rates; 2)
and Edwards, 1991; Constandse-Westermann, 1997; Lev- ‘‘. . .the rates of wear of the molars remain[ed] constant
Tov and Smith, 2006; Bulbeck and O’Connor, 2011). We throughout the life of the dentition’’ (884). In practice,
will call it the modified Miles method. The modification Miles did not require assumption 2, because he updated
relies on data routinely collected during skeletal analysis molar-specific wear rates in a step-wise manner as he
and can be applied to samples where juveniles and post- extrapolated to successively older individuals. However,
cranial elements are under-represented. Unlike many Miles does require a slightly weaker assumption: that
other methods of dental age estimation, it does not the ratio of wear rates among molar regions M1, M2,
require destruction or disarticulation of the remains and and M3 remains constant through the lifespan. We will
is applicable to archaeological collections. Here, we call this assumption 3. Please see Appendix A for a
describe the application of the modified Miles method to mathematical treatment of this assumption and its rela-
a sample of hunter-gatherers and further develop a tionship to the assumptions of the modified method pro-
method of quantifying the error associated with each age posed here.
estimate. The modified Miles method makes two changes to the
Miles method. First, we score occlusal wear following
MATERIALS Scott (1979) in lieu of physically seriating the sample.
Second, we use adult wear gradients and eruption sched-
The sample is composed of individuals from hunter- ules from nonindustrialized populations to estimate wear
gatherer and pastoralist populations from a variety of rate. We require all three assumptions, 1, 2, and 3, dis-
latitudes with diverse population histories and diets cussed in the context of Miles’ method above, along with
(Table 1). The Khoi-San, Australian aborigine, Fuegian, one additional assumption: 4) that molars in the same
and North Eurasian (Buriat, Chukchi, Sami) collections individual wear at the same rate. This is essentially a
all derive from modern exploratory expeditions, while more restrictive version of assumption 3; in addition to
the Afalou, Alaskan, and Native Californian samples are requiring molars to wear at a constant ratio, the modi-
archaeological. Afalou is a Late Pleistocene cave site in fied Miles method requires that this ratio be 1:1:1. We
Algeria, dating to roughly 13,000 years ago (Hachi, further describe each change and discuss each assump-
1996), while the Alaskan and Californian remains are tion below.
significantly younger. The Alaskan sample comes from
Paleo-Eskimo remains from the Point Hope site in
Alaska, dating from about 2,360 to 1,619 BP (Rainey and Modification 1: Scoring occlusal wear
Ralph, 1959), and the Californian remains are from a
We chose to seriate individuals by quantifying molar
collection of sites in Central California ranging from
wear with Scott’s (1979) method of scoring occlusal wear.
about 5,000 to 500 BP (Milliken et al., 2007). Each skele-
Ordinal scoring methods like Scott’s represent the least
tal population conforms to a general geographic area and
dramatic departure from Miles’ assessments of occlusal
broadly similar diet and lifestyle, but none represent the
wear and are routinely used by researchers to estimate
ideal small biological populations existing at a single
wear rates from molar gradients (see section on Modifi-
time and place for which Miles created his method of
cation 2, below). They are quickly and easily applied and
age estimation.
have the advantage of being exceptionally nondestruc-
tive in that they do not require sectioning, disarticulat-
METHODS ing, or even coming into physical contact with a speci-
The Miles method and its assumptions men. Scott’s method is recommended by Standards of
Data Collection from Human Skeletal Remains (Buikstra
Miles (1962, 1963, 1969, 1978) relied on molar eruption and Ubelaker, 1994) and is commonly recorded by
schedules to calculate the number of years that a child’s researchers documenting human remains, suggesting

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ESTIMATING AGE FROM OCCLUSAL WEAR 183
that this method may even be applicable post hoc to age suggests that this type of data may be an excellent
repatriated remains. Scott’s system improves on other candidate for the modified Miles method.
ordinal wear scoring methods in that it scores wear on a
scale of 1 to 10 for each quadrant; the quadrant scores Modification 2: Estimating wear rates from adults
can then be summed for a score from 4 to 40 for each
tooth. The result is a score per tooth approaching a con- Instead of estimating wear rates based on subadults of
tinuous scale (Benfer and Edwards, 1991). We follow Li known age, a population wear rate can be estimated
and Ji (1995) in averaging the quadrant scores, as based on wear gradients in the entire adult sample.
opposed to summing (so that differential preservation of Because the second molars erupt roughly 6 years after
the samples would not affect the total wear scores), but the first molars, the difference in wear between the first
this procedure has no effect on the resulting quasi-con- and second molars will correspond to 6 years of wear
tinuous nature of the scores. even in adults. This observation is not novel—the utility
The accuracy of the modified method depends on the of wear gradients was first recognized by Smith (1972)
correlation of Scott’s scores with age. As interindividual and then employed to estimate wear rates by Lunt
variation in wear increases, the correlation between (1978), Benfer and Edwards (1991), Constandse-Wester-
wear and age decreases, as does our ability to correctly mann (1997), Lev-Tov and Smith (2006), and Bulbeck
predict age from wear. All methods of age estimation and O’Connor (2011), in each case in conjunction with or-
from tooth wear must therefore rely on the assumption dinal scoring methods. We will refer to the use of the dif-
that individuals belonging to a single biological popula- ference between molar wear scores to estimate wear
tion have relatively consistent wear rates (assumption rates as the difference method. Benfer and Edwards
1). Previous work suggests that this assumption is vio- (1991) used Scott’s (1979) system to score wear in adult
lated to some degree. Interindividual variation in wear skeletal samples from coastal Peru; the resulting cul-
rate could result from bruxism (e.g., Colquitt, 1987), tural patterns in wear rate were consistent with those
occupational hazards (e.g., Pöllmann et al., 1987), estimated from crown-height measurements and support
enamel thickness (e.g., Molleson and Cohen, 1990), mal- the combined use of Scott’s (1979) system and the differ-
occlusion/pathology, individual variation in diet, use of ence method to estimate wear rates.
teeth as tools, or other factors, and can be substantial While Miles’ method of serial extrapolation uses wear
within a single population (McKee and Molnar, 1988; rates in younger individuals to estimate ages of older
Boldsen, 2005; Cuozzo et al., 2010). Though interindivid- individuals, our use of the difference method effectively
ual variation decreases accuracy, tests of the Miles averages wear rates across all individuals, thus taking
method still suggest that it performs as well or better advantage of all of the information in the sample. How-
than other methods of skeletal age estimation in high ever, the use of the difference method to estimate wear
rate does require the assumption that molars in the
wear populations (Nowell, 1978; Kieser et al., 1983;
same individual wear at the same rate (assumption 4).
Lovejoy, 1985) and often also in low wear populations
Some research (Akpata, 1975; Helm and Prydso, 1979;
(Lovejoy et al., 1985a; Constandse-Westermann, 1997;
Santini et al., 1990), including Miles’ original (1962) pub-
but see Santini et al., 1990).
lication of his method, suggests that the molars do wear
The modified Miles method further relies on the line-
at slightly different rates, with the first molar wearing
arity of the relationship between Scott’s wear scores
more rapidly than the other two. Other studies find no
and age (a reflection of the constancy in wear rate
difference in wear rates between molars (Nowell, 1978;
through life). The validity of this assumption (2) is
Dreier, 1994; Mays, 2002). Miles (2001) suggests that the
unclear. Molleson and Cohen’s (1990) experiments with
differences in rates are so small as to not make much
mechanically ground molars demonstrate that some or-
practical difference in age estimation.
dinal scoring systems (e.g., Brothwell, 1963) do not
encompass equivalent amounts of wear with each score,
and suggest that they may not be linear. In contrast, The modified Miles method
Kieser et al.’s (1985) investigation of Smith’s (1972) Wear was scored following Scott (1979) and Buikstra
scoring system in a known-age sample showed an and Ubelaker (1994) for each quadrant of every molar on
approximately linear relationship with age from 19 to the left side in adult individuals, except in cases of
39 after which the wear rate was reduced. Though obvious asymmetry, where both sides were scored, or
Scott’s system has twice as many wear scores per quad- lack of preservation on the left side, in which case the
rant as does Smith’s system, the two systems are other- right side was scored. An individual was considered an
wise similar. Regardless, it is clear that further explora- adult if at least one third molar was in occlusion; if it
tion of the relationship between Scott’s wear scoring was considered likely the third molars were congenitally
method and age is needed. If wear rates decrease with absent and attainment of adulthood was questionable,
age, the modified method will systematically underesti- complete fusion of the spheno-occipital synchondrosis
mate age in older adults. from the ectocranial surface (occurring between approxi-
The modified method proposed here is compatible with mately 13 and 20 years of age in females and 16 and 23
virtually unlimited methods of quantifying wear. Dreier years of age in males (Shirley and Jantz, 2011)) was
(1994) offers an occlusal scoring method based on obser- used to distinguish adults. Wear scores of each quadrant
vations of mechanically ground teeth, and there are a of a tooth were then averaged (Li and Ji, 1995); this av-
variety of other methods of recording occlusal wear that erage will hereafter be referred to as a tooth’s ‘‘wear
represent more dramatic departures from Miles’ method score’’ for simplicity. When wear was scored for both left
but have the advantage of recording truly continuous and right sides the wear scores were averaged before
data. For example, the close correlation (Tomenchuck estimating wear rates as below.
and Mayhall, 1979; Walker et al., 1991; Mays, 2002; but
see Benazzi et al., 2008) and approximately linear Estimating wear rate. The Miles method combines
(Mays, 2002) relationship of molar crown-heights with males and females and maxillae and mandibles to calcu-

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
184 C.C. GILMORE AND M.N. GROTE

Fig. 1. Average molar wear score for each molar class plotted against the others for every individual in every population in the
sample. A) The relationship between the M1 and M2; B) the relationship between the M1 and M3; C) the relationship between
the M2 and M3. The M1 and the M2 are more closely correlated with each other than either is with the M3, demonstrating the
increased variance in the wear scores of the latter molar class.

late wear rates for a skeletal population. When calcu- TABLE 2. Eruption ages and standard deviations from studies
lated separately, our data suggested a slightly higher of nonindustrialized populations that reported sample sizea
wear rate in the maxillary than the mandibular molars; Tooth Eruption age (yr) Standard deviation (yr) Covariance
results from some previous work agree (Murphy, 1959;
Lunt, 1978; Kieser et al., 1983; Mo and Peng, 1983; M1 6.02 0.85 –
McKee and Molnar, 1988), though others have found the M2 11.31 1.13 –
M3 17.87 1.16 –
opposite pattern (Molnar et al., 1983; Lovejoy, 1985;
M1, M2 – – 0.717
Mays, 2002), or no difference at all (Dreier, 1994; Li and M1, M3 – – 1.018
Ji, 1995; Mays et al., 1995). Reports on the effect of sex M2, M3 – – 1.362
on wear rate are similar; sexual dimorphism seems to be
a
significant in some populations (Tomenchuck and May- Data are averaged from Steggerda and Hill (1942), Fried-
hall, 1979; e.g., Benfer and Edwards, 1991), and unim- lander and Bailit (1969), Brown (1978), and Mayhall et al.
portant in others (Lunt, 1978; Richards and Brown, (1978).
1981; Lovejoy, 1985; Walker et al., 1991; Mays, 2002).
Ultimately, we elected to follow Miles’ procedure and
combine maxillary and mandibular arcades as well as approach zero and will not be representative of the true
the sexes. By combining the maxilla and mandible of wear rate. Therefore, if any quadrant of individual i has
each individual to estimate a wear rate, we increase the a wear score of 10 (the maximum), the average wear
molar sample size used for each estimate of individual score for that molar class in individual i is excluded
wear rate and decrease the influence of ‘‘noise’’ from from calculating the population average wear rate.
intraindividual variation. For example, antemortem After creating average molar class wear scores for
tooth loss and dental disease have been predicted to ei- each individual, we calculated individual wear score dif-
ther reduce the wear rate of the occluding molar (Mays, ferences: Di 5 wearM1i 2 wearM2i. Across the sample,
2002), to increase wear rate in non-occlusal partners most such differences are positive. The difference in
(Miles, 1962), or both, though Mays (2002) found no evi- wear of the M1-M3 and M2-M3 could also be used to cal-
dence of a change in wear rate of the occlusal partner in culate wear rate. Though Mays (2002) finds that the
a low wear rate sample. One could also form distinct third molar is equally well correlated with age as the
prediction equations for males and females, but the loss first and second molars, Figure 1 demonstrates that M3
of precision (resulting from splitting the sample by sex) wear is less closely correlated with M1 or M2 wear than
is compounded with the uncertainty of assigning sex cor- M1 and M2 are with each other in these data, probably
rectly. Our procedures arguably introduce bias, because because third molars tend to be more variable in erup-
distinct categories of teeth and sex are combined, but tion time (see Table 2) and morphology than first or sec-
also decrease variance by taking full advantage of sam- ond molars. For this reason, we use only the M1-M2 dif-
ple degrees of freedom. ference to calculate wear rate.
To calculate wear rates we average (equally weighted) Figure 2 plots M1-M2 differences, D, against M1 wear
the wear scores of the maxillary and mandibular molars scores, wearM1i, and shows that variation in D appears
of individual i to create a first molar wear score in Scott to increase with wear. This pattern likely results from
units (wearM1i), a second molar wear score (wearM2i), the accumulation of interindividual variation over a life-
and a third molar wear score (wearM3i). We refer to time of wear and supports previous observations that
these as average molar class wear scores. At some point, error in age estimation based on wear increases with age
each tooth will reach a maximum wear stage and cease (Miles, 1962; Kieser et al., 1983; Mays, 2002). It is possi-
to change with time. The M1 is likely to be the first ble that the lack of variation at very low wearM1i is in
molar class to reach this maximum stage, while the M2 part an artifact of the method—maximum differences in
and M3 continue to wear (Benfer and Edwards, 1991). wear are constrained by the relation Di
wearM1i 2 1,
As the M1 ceases to wear and the other molar classes and this constraint is especially severe when wearM1i is
continue to wear, the difference in wear between the small. The finite property of the scoring system suggests
first and second molars or first and third molars will that very lightly worn teeth could skew estimates of the

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ESTIMATING AGE FROM OCCLUSAL WEAR 185
ond molars (TM2 2 TM1). Research on dental eruption
shows that genetics (Garn et al., 1965, 1973), and to a
lesser extent socioeconomic level (Clements et al., 1953),
can have significant effects on the timing of eruption of
permanent teeth, often resulting in significantly earlier
eruption times than observed in children of European
ancestry (Suk, 1919; Steggerda and Hill, 1942; Dahlberg
and Menegaz-Bock, 1958; Chagula, 1960; Houpt et al.,
1967). We recommend using eruption standards from a
reference population as closely related to the study popu-
lation as possible. As the populations of concern in this
study are composed of hunter-gatherers, many without
obvious modern descendants, we chose to pool averages
and standard deviations of M1, M2, and M3 eruption
times from studies of nonindustrialized populations that
reported sample size (Steggerda and Hill, 1942; Fried-
lander and Bailit, 1969; Brown, 1978; Mayhall et al.,
1978); we then use these sample-size-weighted average
eruption times and standard deviations (Table 2) for every
human population in this study. When reported sepa-
rately, right/left and maxillary/mandibular eruption times
have been averaged (weighted equally, not by sample size)
for consistency with the method of combining wear scores
Fig. 2. M1-M2 differences (D) plotted against average M1 described above. This procedure combines cross-sectional
molar class wear score for each individual in every population and longitudinal studies (see Dahlberg and Menegaz-
in the sample, showing the increasing variation in M1-M2 dif-
Bock, 1958 for discussion) and introduces bias for those
ferences with wearM1 (a proxy for age). The dotted line shows
the maximum possible D as imposed by the method; this ceiling groups that do have a close relationship to a single mod-
may be truncating variation, particularly at the smallest wear ern comparator, but has the advantage of pooling informa-
scores, and contributing to the appearance of an increasing tion. This method and our results are very similar to those
trend. To alleviate the potential effect of truncated variation on of Ubelaker (1989) except that we excluded all modern
the estimation of wear rate, we recommend excluding M1-M2 populations, regardless of ancestry, from our averages and
differences from very lightly worn dentitions in the calculation found an earlier age of second molar eruption.
of population average wear rate. Filled black circles indicate the
individuals removed from the calculation of average wear rate Estimating age. We follow Miles in using the wear rate
when procedures outlined in the text are followed.
to calculate a functional age and then adding the age of
eruption to estimate age-at-death. To estimate the age in
actual wear rate and should properly be excluded when years of individual i in population p, wearM1i is divided
estimating a population wear rate. We dealt with this by the population wear rate Rp ¼ Dp =ðTM2  TM1Þ (in
issue (as discussed below and presented mathematically Scott’s units per year) and added to the age of molar
in Step 3 of Appendix B) by excluding individuals with eruption (e.g. TM1). A separate age estimate can be
the M1 so slightly worn that their M1-M2 difference, Di, made for each molar using this method. The M1 is least
would be severely truncated. variable in eruption time and therefore is preferred for
For each population p we averaged all M1-M2 differ- age estimation. Age is estimated from the M1 unless:
ences and calculated the standard deviation to produce
the uncorrected statistics D
p and sd(Dp). We then identi- 1. The M1 is not present
fied individuals in population p having severely con- 2. Any M1 quadrant has reached the maximum wear
strained M1-M2 differences as follows. We reasoned that score of 10.
M1 wear scores for which the maximum possible M1-M2
difference is smaller than D
p 1 sd(Dp) would produce If either of the above is true of the M1, the M2 is used
artificially constrained differences. These M1 wear scores for age estimation. The M3 is used for age estimation
are identified by the equation wearM1i 2 1 \ D
p 1 only if the M1 and M2 are missing.
sd(Dp), making use of the upper bound for the difference
on the left hand side. Individuals in population p meet- Error in age estimation. Like other skeletal indicators
ing this condition will be excluded from the calculation of age, human variation in wear rates (and therefore the
of wear rate (though their ages will still be estimated). error inherent in the Miles method and this modifica-
After excluding these individuals from population p, we tion) is substantial. Most methods of skeletal age estima-
then calculated the corrected statistics for the M1-M2 tion can only quantify error by comparison with a
difference Dp and standard deviation sd(Dp). Formal known-age reference population, whose standards may
equations for each step are provided in Appendix B. This or may not be applicable to other populations. Sources of
procedure ultimately removed only 10 individuals with error in age estimation from dental wear, on the other
very slightly worn first molars from the sample; each hand, can be identified and quantified with population-
individual removed is noted in Figure 2. The corrected specific measures. If our assumptions are valid, there
statistics are carried forward to estimate age as are three primary sources of variation to account for
described below. when using the modified Miles method. The first is inter-
To calculate the average wear per year for population p, individual variation in eruption times; the second is
each population average difference Dp is divided by the interindividual variation in wear rate discussed above;
number of years between the eruption of the first and sec- and the third is uncertainty in estimation of population

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
186 C.C. GILMORE AND M.N. GROTE
TABLE 3. Population summary statistics and wear ratesa
Population np Dp Rp var(Dp )
Afalou 18 1.97 0.37 0.046
Alaska 23 1.12 0.21 0.034
Australian aborigine 40 1.52 0.29 0.018
Buriat 14 1.30 0.25 0.034
Chukchi 7 0.84 0.16 0.177
Sami 7 0.73 0.14 0.084
Fuegian 7 1.23 0.23 0.082
Khoi-San 29 1.76 0.33 0.012
Native Californian 27 1.73 0.33 0.023
a
np is the sample size used to estimate wear rate; Dp is the av-
erage difference in first and second molar wear scores; Rp is the
estimated wear rate; varðDp Þ is the variance of Dp . Each term is
defined mathematically in the equations in Appendices B and C.

Table 3 provides summary statistics for each popula-

tion. Compared with modern and preindustrial agricul-
tural populations, hunter-gatherers generally have tough
diets, minimal food processing, and high rates of tooth
wear (Smith 1984; Larsen, 1995). The hunter-gatherer
and pastoralist populations in this study also show rapid
rates of wear, progressing through a single Scott wear
Fig. 3. Point age estimates 6 one standard deviation (black) stage every 3 to 7 years (Table 3). Interestingly, the pop-
and the age categories from pelvic age indicators (grey) for each ulations with the highest rates of wear are those from
individual in the subsample for which comparative ages were dry to temperate climates, including the Afalou, Califor-
available. Arrows indicate unbounded age categories. Specimens nian, Khoi-San, and Australian aborigine populations.
are from the Phoebe A. Hearst Museum of Anthropology (Uni- These populations tend to incorporate a higher propor-
versity of California, Berkeley). tion of plant foods in their diets relative to the popula-
tions in this study from high latitudes and very cold cli-
means due to sampling. The formulas below account for mates, the Alaskans, Buriats, Chukchi, and Sami (Mur-
variation in eruption times using published variances dock, 1967). An increased reliance on pastoralism and
(Steggerda and Hill, 1942; Friedlander and Bailit, 1969; dairy products in the Eurasian groups, particularly the
Brown, 1978; Mayhall et al., 1978) and covariances Buriats and Sami, may also contribute to their relatively
(Parner et al., 2002). Interindividual variation in wear low wear rates.
and sampling uncertainty enter the quantification of Figure 4 shows the distribution of all estimated ages in
error via sd(Dp) (see Step 4 in Appendix B) and the the sample: they range from 11.6 to 71.2 years with a
related term varðDp Þ (see Appendix C). We obtain varian- mean of 30.4 years. In addition to other biases, the Miles
ces for estimated ages using the Delta Method (Cramér, method and most other standard methods of age estima-
1945, section 27.7; see also Stuart and Ord, 1994, sec- tion regularly underestimate age in older individuals
tions 10.5–10.7), in which smooth functions of random (Aykroyd et al., 1999); it is not expected that the modifica-
variables are approximated by first-order Taylor series. tions suggested here will significantly improve upon the
In Appendix C are equations for estimated variances of general inadequacies of skeletal age estimation. Though
age estimates from each molar (AgeM1, AgeM2, and individuals with ages [40 years do seem particularly
AgeM3) and an explanation of the method. sparse, supporting the expectation of bias in older individ-
uals, it is important to remember that the samples here
consist of individuals from a variety of time periods and
RESULTS populations, many with unclear provenience and collec-
tion biases, which do not represent a true random sample
Age estimates from other skeletal age indicators, pri- of a living human population, or a realistic mortality dis-
marily pubic symphysis and auricular surface (but also tribution. Furthermore, the strongly age progressive na-
including cranial suture fusion in one case when pelvic ture of antemortem tooth loss may mean that the oldest
remains were unavailable), are available for a subset (n individuals no longer possess molars to estimate from,
5 22) of the Californian sample (Bartelink, 2006); we and would not be represented in this graph.
used these as a check to ensure that the modifications to Estimates of the error in age estimation are given in
the Miles method have not severely affected the meth- the form of a standard deviation accompanying each age
od’s accuracy. Like many methods of age estimation in estimate (square roots of the variances in Appendix C);
bioarcheology (e.g., Todd, 1920; Lovejoy et al., 1985b), the standard deviations are 5.4 years on average, but
the methods used for pelvic and cranial indicators pro- vary from 1.3 to 31.4 years. As expected, standard devia-
duce age categories (e.g., between 20 and 30 years old) tions increase with age (Fig. 5) and in populations with
rather than point age estimates. Roughly half of the larger values of varðDp Þ. All standard deviations larger
point estimates from the modified Miles method fall than 13 years come from one of two populations, the
within the age category based on pelvic indicators (Fig. Chukchi or Sami. These two populations are among the
3) and in all but one case (125811) the point estimate 6 smallest—decreasing confidence in the characterization
1 standard deviation range overlaps with the age cate- of wear rate—and have the lowest wear rates, so very
gory, suggesting the modified Miles method is still effec- large standard deviations in these cases are not unex-
tively tracking age estimated by traditional methods. pected. Otherwise, the standard deviations from this

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ESTIMATING AGE FROM OCCLUSAL WEAR 187

Fig. 5. Estimated age versus standard deviation of each age

Fig. 4. The distribution of ages in years estimated for all pop- estimate (a measure of the error in age estimation) in years by
ulations combined using the modified Miles method. Most ages population. The relationship between standard deviation and age
fall between 20 and 40 years with very few older individuals. It is is different for each population, as wear rate Rp and varðDp Þ vary,
not expected that these populations accurately represent demog- but standard deviations increase with age in all populations.
raphy given the unknown collection biases in the samples; though Note that most specimens in a population fall on a single trend
the rarity of older individuals (>40 years) does suggest that, like line formed by the linear relationship between estimated age and
the Miles method and other skeletal aging methods, this modifica- standard deviation for the M1. This relationship differs depend-
tion may underestimate age in older individuals. ing on the molar used for age estimation; individuals that appear
to depart from their respective population trend line do so
because their ages were estimated from the M2 or M3.
method are very similar to those from pubic symphysis
(Brooks and Suchey, 1990) and cranial suture closure
(Meindl and Lovejoy, 1985) methods when tested in their tion in wear rates between sexes, individuals, or even
reference populations. maxillary and mandibular molars of the same individual,
though research shows that variation in each category
DISCUSSION AND CONCLUSIONS does exist in some populations. Although we cannot con-
trol for this variation in each age estimate, our contribu-
We have presented a modification of the Miles method tion of formulae to calculate error acknowledges varia-
of estimating age from molar wear. The primary change tion and allows the corresponding uncertainty to be car-
to the Miles method is that we estimated wear rate based ried forward into statistical analyses. Error increases
on molar wear gradients in adults (Benfer and Edwards, when our confidence in the estimation of average popula-
1991; Constandse-Westermann, 1997) instead of basing tion wear rate is low (i.e., when samples are small).
wear rate estimates on subadults. Given the application Error in the method also increases when the average
of the method to hunter-gatherers, we also elected to use population wear rate itself is low; in this case, large
eruption times reported for nonindustrialized societies changes in estimated age result from small changes in
following Ubelaker (1989) rather than using the eruption wear score and thus magnify error.
times for American children of European ancestry that The age estimates from this method still track age
Miles used (Garn et al., 1959; Gleiser and Hunt, 1955). estimates from pelvic indicators in a small sample of the
None of our modifications are novel in and of themselves, dataset (n 5 22), especially when one standard deviation
though they had not been applied to the Miles method from the point estimate is considered, and suggest that
previously. With these changes instituted, the method the modification has not adversely affected the Miles
becomes applicable to a much wider variety of skeletal method’s ability to estimate age, even in samples that
samples, especially those collections that consist primar- are less than ideal. Each population in this article con-
ily of adult skulls. Before this description the only other sisted of samples from the same general geographic
method for estimating age in these types of collections region, but often spanning long time periods and
that did not require sectioning or disarticulation was undoubtedly encompassing some differences in lifestyle
based on cranial suture closure—a method often and diet. As with the Miles method, the accuracy of this
regarded as unreliable when applied to samples other method should improve in larger and more homogenous
than the reference population, due to the considerable samples that represent a population from a single time
interpopulation variation in the aging process of cranial and place. Both the Miles method and this modification
sutures (Key et al., 1994; Hershkovitz et al., 1997). give point age estimates, which, when associated with a
Interindividual variation in wear is the primary chal- standard deviation as described here, give more informa-
lenge for estimating age from dental wear, whether by tion about an individual’s age than do the age categories
the Miles method or the modification described here. returned by commonly applied methods (e.g., Todd, 1920,
Both methods implicitly assume that there is no varia- 1921; Lovejoy et al., 1985b; Meindl and Lovejoy, 1985;

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
188 C.C. GILMORE AND M.N. GROTE

Brooks and Suchey, 1990). The Miles method and this We therefore suppose that molar wear scores are real-
modification are also able to estimate age without refer- valued, differentiable functions of t 5 age, denoted as
ence to standards from a separate population and there- M1(t), M2(t), M3(t). Our development is concerned with
fore avoid mimicking the age distribution of the refer- M1(t) and M2(t)—the wear scores on which the modified
ence population (Bocquet-Appel and Masset, 1982; Aiello Miles method focuses—but the other pairs (M1(t) and
and Molleson, 1993) or making incorrect assumptions M3(t), M2(t) and M3(t)) could be treated analogously.
about the variation in the target population. Though the Taking ‘‘rate’’ as a synonym for derivative with respect
modification does not appear to improve upon the tend- to t, assumption #3 translates as
ency of commonly applied methods of skeletal age esti-
mation to underestimate age in older adults (Molleson dM1ðtÞ=dt
et al., 1993; Aykroyd et al., 1999), a large, known-age ¼ constant: ð1Þ
dM2ðtÞ=dt
sample is needed to confidently characterize the biases.
Methods of age estimation from molar wear are expected
to perform more poorly in recent, low wear rate popula- A necessary and sufficient condition for Eq. (1) is that
tions than in prehistoric populations, as interindividual there exist real numbers l and b (b = 0) such that
variation in wear may overwhelm signals of age when
wear rate is low (Molnar et al., 1983; Walker et al., M2ðtÞ ¼ l þ bM1ðtÞ: ð2Þ
1991; Mays, 2002). Nevertheless, we encourage further
testing to explore the utility of the modified Miles
method, particularly in cases where age is known and The precise form of M1(t) in Eq. (2) does not need to be
wear rates are high (rarely available, as ages are typi- specified in order for Eq. (1) to hold. Equation (2) implies
cally only known for relatively recent samples) or per- that the difference between first and second molar wear
haps in a high wear population with ages estimated by scores at age t is
tooth cementum annuli methods.
Although it still suffers from some of the same limita- M1ðtÞ  M2ðtÞ ¼ ð1  bÞM1ðtÞ  l: ð3Þ
tions as other methods of age estimation, the modifica-
tion of the Miles method we present here is a viable We understand Eqs. (1)–(3) to be consequences of Miles’
option for estimating age in skeletal samples that might assumption #3, for numerical wear scores that vary con-
otherwise be aged only by cranial suture closure. This tinuously with t.
modification is an improvement over cranial suture age In the modified Miles method, a population average
estimation in that aging standards are determined from wear score difference D, and subsequently a wear rate R,
the population in question and that point-age estimates are obtained by assuming that
with error, effectively ranking individuals, are produced.
M1ðtÞ  M2ðtÞ ¼ constant; ð4Þ
ACKNOWLEDGMENTS which is equivalent to requiring b 5 1 in Eq. (3). Thus,
The first author offers gratitude to curators and col- where Miles (1962) finds that 6 years of wear on the M1
leagues at the following museums for their permission to is roughly equivalent to 6.5 years of wear on the M2, the
study their collections: American Museum of Natural modified Miles method assumes the ratio is 1:1 (assump-
History, Institut de Paléontolgie Humaine, Musée de tion #4). The wear rate R ¼ D=ðTM2  TM1Þ, where TM1
l’Homme, National Museum of Natural History (Smith- and TM2 are respectively the mean ages of M1 and M2
sonian), Peabody Museum at Harvard University, eruption, is understood to approximate the average wear
Phoebe A. Hearst Museum of Anthropology at the Uni- rate over the portion of the lifespan for which the modi-
versity of California, Berkeley, and the Anthropology fied method is intended (broadly speaking, early-adult
Department at the Universität Wien. The authors thank through mid-adult, with decreasing accuracy for older
Dr. Eric Bartelink for his willingness to share his age adults).
estimates of the Native Californians at the Phoebe An estimated age (based on M1 wear) is obtained
Hearst Museum and Drs. Eric Bartelink, Henry under the modified method using the linear equation
McHenry, Teresa Steele, and Tim Weaver for their com- wearM1
ments on this manuscript. Finally, the authors thank Age ¼ TM1 þ ; ð5Þ
R
the Editor-in-Chief, Associate Editor, and two anony-
mous reviewers for valuable suggestions on an earlier
version that led to substantial revisions. where wearM1 is the M1 wear score of an individual of
unknown age. Thus assumption #2, that molar wear
rates remain constant throughout the life of the denti-
APPENDIX A: CONNECTING THE ASSUMPTIONS tion, is implied by the use of Eq. (5). Although Miles
OF THE MILES AND MODIFIED MILES METHODS (1962) states assumption #2, he does not seem to use it
in practice; in contrast, the modified method requires
The relationship between the Miles and modified Miles that assumption #2, and consequently Eq. (5), hold to a
methods rests on assumption #3: that the ratio of wear reasonable approximation across the middle lifespan.
rates among molar regions M1, M2, and M3 remains Figure A1 presents a scheme that is compatible with
constant through the lifespan. Although Miles (1962) the Miles method and shows the linear approximations
scored molar wear qualitatively and cautioned against implicit in the modified method. This scheme has wear
overt mathematization of his method, a mathematical scores increasing with age, though other schemes
translation seems necessary here for concreteness, and (including those for which ‘‘scores’’ decrease monotoni-
for comparison with the numerical approach of the modi- cally with age) are compatible with the two methods and
fied method. could be depicted analogously.

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ESTIMATING AGE FROM OCCLUSAL WEAR 189

wearM1i >1 þ Dp þ sdðDp Þ

Step 4
Calculate corrected average and standard deviation,
where np is population sample size using only individu-
als from Step 3 and Di is as in Step 1,
Pnp sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pnp ffi
2
i¼1 Di i¼1 ðDi  Dp Þ
Dp ¼ sdðDp Þ ¼ ;
np np  1

for individuals i satisfying Step 3.

Step 5
Calculate wear rate per year for population p, where
Dp is defined by the equation in Step 4, and TM1, TM2,
and TM3 are the mean ages of M1, M2, and M3 erup-
tion, respectively, in years (Table 2):

Dp
Rp ¼
TM1  TM2

Step 6
Estimate age of each individual i in population p:
Fig. A1. A molar wear scoring scheme compatible with the wearM1i
Miles and modified Miles methods. M1(t) and M2(t) are hypotheti- AgeM1i ¼ TM1 þ
cal wear score functions compatible with Miles’ assumption #3. The Rp
tangent lines indicated by dashes have slopes conforming to the wearM2i
6:6.5 relationship noted by Miles, and tangent lines at all common AgeM2i ¼ TM2 þ
time-points of M1(t) and M2(t) have this property. The modified
Rp
Miles method assumes that the ratio of slopes is 1:1 at all common wearM3i
points, and further approximates wear score functions by straight AgeM3i ¼ TM3 þ
Rp
lines (linear M1 and M2 in the figure). Given a particular molar
wear score, age is estimated by reflecting the score across a corre-
sponding wear function to a value on the ‘‘age’’ axis. The modified APPENDIX C: FORMULAE FOR VARIANCE
method is expected to perform best where M1(t) and linear M1 are ESTIMATION
close in value, shown here in the mid-adult years. The modified
method may notably underestimate the ages of older adults. TM1 The function for estimation based on M1 wear can be
and TM2 are eruption times of M1 and M2, respectively. expressed as
APPENDIX B: FORMULAE FOR AGE ESTIMATION TM2  TM1
AgeM1i ¼ TM1 þ wearM1i  ¼ f ðTM1; TM2; Dp Þ:
Each step in estimating age with this method is detailed Dp
below, including the steps involved in estimating popula-
tion wear rates discussed in the methods section above. Estimates based on M2 or M3 have similar forms. The ap-
proximate variance of AgeM1i by the Delta-Method is then:
Step 1
 2
For each individual i, subtract the average M2 class @f
varðAgeM1i Þ   varðTM1Þ
wear score from the average M1 class wear score: @TM1
 2 !2
Di ¼ wearM1i  wearM2i @f @f
þ  varðTM2Þ þ  varðDp Þ
@TM2 @Dp
Step 2   
@f @f
Average the differences for all individuals in popula- þ2  covðTM1; TM2Þ;
@TM1 @TM2
tion p and calculate the standard deviation, where Np is
uncorrected sample size and Di is as in Step 1:
in which sample statistics, variances, and covariances from
PNp sPffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tables 2 and 3 are substituted at appropriate places. In deriv-
Np 2
i¼1 Di i¼1 ðDi  Dp Þ ing the approximate variance, we have assumed that wearM1i
Dp ¼ sdðDp Þ ¼
Np Np  1 is measured without error, and that covðTM1; Dp Þ ¼ cov
ðTM2; Dp Þ ¼ 0. Delta-Method approximations are under-
Step 3 stood to be more accurate in large samples. Similar calcu-
lations produce variances for age estimates based on M2
Remove lightly worn individuals from the dataset by or M3, though in the latter case, the function f depends on
keeping only specimens that fit the requirements of the four variables: TM1, TM2, TM3, and Dp . After substitu-
following equation: tion, the Delta-Method variances are:

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
190 C.C. GILMORE AND M.N. GROTE
!2 Akpata ES. 1975. Molar attrition in a selected group of Niger-
Dp  wearM1i ians. Commun Dent Oral Epidemiol 3:132–135.
varðAgeM1i Þ   varðTM1Þ
Dp Aykroyd RG, Lucy D, Pollard AM, Roberts CA. 1999. Nasty, bru-
!2 tish, but not necessarily short: a reconsideration of the statisti-
wearM1i cal methods used to calcluate age at death from adult human
þ  varðTM2Þ skeletal and dental age indicators. Am Antiq 64:55–70.
Dp Bartelink E. 2006. Resource intensification in pre-contact cen-
!2 tral California: a bioarchaeological perspective on diet and
wearM1i  ðTM1  TM2Þ health patterns among hunter-gatherers from the lower Sac-
þ 2
 varðDp Þ ramento Valley and San Franscisco Bay. College Station:
Dp Texas A&M University. Published Dissertation. Ann Arbor,
! !
Dp  wearM1i wearM1i MI: University Microfilms.
þ2  covðTM1; TM2Þ; Benazzi S, Bonetti C, Cilli E, Gruppioni G. 2008. Molar crown
Dp Dp height: now always a reliable method for the evaluation of
age-at-death. J Archaeol Sci 35:2371–2378.
!2 Benfer RA, Edwards DS. 1991. The principle axis method for
Dp þ wearM2i measuring rate and amount of dental attrition: estimating ju-
varðAgeM2i Þ   varðTM2Þ
Dp venile or adult tooth wear from unaged adult teeth. In:
!2 Larsen CS, Kelley MA, editors. Advances in dental anthropol-
wearM2i ogy. New York: Wiley-Liss, Inc. p 325–340.
þ  varðTM1Þ Bocquet-Appel J-P, Masset C. 1982. Farewell to paleodemogra-
Dp phy. J Hum Evol 11:321–333.
!2 Boldsen JL. 2005. Analysis of dental attrition and mortality in
wearM2i  ðTM1  TM2Þ the Medieval village of Tirup, Denmark. Am J Phys Anthropol
þ 2
 varðDp Þ 126:169–176.
Dp Borrman H, Engström EU, Alexandersen V, Jonsson L, Gerdin AL,
!
Dp þ wearM2i Carlsson GE. 1996. Dental conditions and temporo-mandibular
þ2 joints in an early mesolithic bog man. Swed Dent J 20:1–14.
Dp Brooks S, Suchey JM. 1990. Skeletal age determination based
! on the os pubis: a comparison of Acsádi-Nemeskéri and
wearM2i Suchey-Brooks methods. Am J Phys Anthropol 5:227–238.
  covðTM1; TM2Þ;
Dp Brothwell DR. 1981. Digging up bones. Oxford: Oxford Univer-
sity Press.
!2 Brown T. 1978. Tooth emergence in Australian aboriginals. Ann
wearM3i Hum Biol 5:41–54.
varðAgeM3i Þ  varðTM3Þ þ  varðTM2Þ Buckberry JL, Chamberlain AT. 2002. Age estimation from the
Dp auricular surface of the ilium: a revised method. Am J Phys
!2 Anthropol 119:231–239.
wearM3i Buikstra JE, Ubelaker DH, editors. 1994. Standards for data
þ  varðTM1Þ
Dp collection from human skeletal remains. Fayetteville: Arkan-
!2 sas Archeological Survey.
wearM3i  ðTM2  TM1Þ Bulbeck D, O’Connor S. 2011. The Watinglo mandible: a second
þ 2
 varðDp Þ terminal Pleistocene Homo sapiens fossil from tropical Sahul
Dp with a test on existing models for the human settlement of
! the region. Homo 62:1–29.
wearM3i Chagula WK. 1960. The age at eruption of 3rd permanent
þ2  covðTM3; TM2Þ
Dp molars in male East Africans. Am J Phys Anthropol 18:77–82.
! Clements EMB, Davies-Thomas E, Pickett KG. 1953. Time of
wearM3i eruption of permanent teeth in British children in 1947–8. Br
þ2  covðTM3; TM1Þ Med J 1:1421–1424.
Dp Cole G, and Waldron T. 2011. Apple Down 152: a putative case
!2 of syphilis from sixth century AD Anglo-Saxon England. Am J
wearM3i Phys Anthropol 144:72–79.
2  covðTM1; TM2Þ;
Dp Colquitt T. 1987. The sleep-wear syndrome. J Prosthet Dent
57:33–41.
Constandse-Westermann TS. 1997. Age estimation by dental
for every individual i in population p, where var(TM1), attrition in an independently controlled early 19th century
var(TM2), and var(TM3) are the squares of the standard sample from Zwolle, The Netherlands. Hum Evol 12:269–285.
deviations of age of M1, M2, and M3 eruptions, respec- Cramér H. 1945. Mathematical methods of statistics. Princeton:
tively, from Table 2; cov(TM1,TM2), cov(TM3,TM1) and Princeton University Press.
cov(TM3,TM2) are the covariances from Table 2; and Cuozzo FP, Sauther ML, Gould L, Sussman RW, Villers LM,
Lent C. 2010. Variation in dental wear and tooth loss among
known-aged, older ring-tailed lemurs (Lemur catta): a compar-
ðsdðDp ÞÞ2 ison between wild and captive individuals. Am J Primatol
varðDp Þ ¼ :
np 72:1026–1037.
Dahlberg AA, Menegaz-Bock RM. 1958. Emergence of the perma-
nent teeth in Pima Indian children. J Dent Res 37:1123–1139.
LITERATURE CITED Dreier FG. 1994. Age at death estimates for Proto-historic Ari-
kara using molar attrition rates: a new quantification method.
Acsádi G, Nemeskéri J. 1970. History of human lifespan and Int J Osteoarchaeol 4:137–147.
mortality. Budapest: Akademiai Kiado. Friedlander JS, Bailit HL. 1969. Eruption times of the decidu-
Aiello LC, Molleson T. 1993. Are microscopic aging techniques ous and permanent teeth of natives on Bougainville Island,
more accurate than macroscopic aging techniques? J Archaeol Territory of New Guinea: a study of racial variation. Hum
Sci 20:689–704. Biol 41:51–65.

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ESTIMATING AGE FROM OCCLUSAL WEAR 191
Garn SM, Lewis AB, Kerewsky RS. 1965. Genetic, nutritional, Meindl RS, Lovejoy CO, Mensforth RP, Walker RA. 1985. A re-
and maturational correlates of dental development. J Dent vised method of age determination using the os pubis, with a
Res 44:228–242. review and tests of accuracy of other current methods of pubic
Garn SM, Lewis AB, Polacheck DL. 1959. Variability of tooth symphyseal aging. Am J Phys Anthropol 68:29–45.
formation. J Dent Res 38:135–148. Miles AEW. 1962. Assessment of the ages of a population of
Garn SM, Sandusky ST, Nagy JM, Trowbridge FL. 1973. Negro- Anglo-Saxons from their dentitions. Proc R Soc Med 55:
Causcasoid differences in permanent tooth emergence at a 881–886.
constant income level. Arch Oral Biol 18:609–615. Miles AEW. 1963. The dentition in the assessment of individual
Gleiser I, Hunt EEJ. 1955. The permanent mandibular first age in skeletal material. In: Brothwell D, editor. Dental an-
molar: its calcification, eruption, and decay. Am J Phys thropology. New York: Pergamon Press. p 191–209.
Anthropol 13:253–283. Miles AEW. 1969. Dentition of the Anglo Saxons. Proc R Soc
Griffin CJ, Powers R, Kruszynski R. 1979. Incidence of osteoar- Med 62:1311–1315.
thritis of the temporo-mandibular joint in various cultures. Miles AEW. 1978. Teeth as an indicator of age in man. In: But-
Aust Dent J 24:94–106. ler PM, Joysey KA, editors. Development, function, and evolu-
Hachi S. 1996. L’Ibéromaurusien, découvertes des fouilles tion of teeth. New York: Academic Press Inc. p 455–464.
d’Afalou. L’Anthropologie 100:55–76. Miles AEW. 2001. The Miles method of assessing age from tooth
Harper C. 1994. A comparison of medieval and modern denti- wear revisited. J Archaeol Sci 28:973–982.
tions. Eur J Orthodont 16:163–173. Milliken R, Fitzgerald RT, Hylkema MG, Groza R, Origer T, Bieling
Helm S, Prydso U. 1979. Assessment of age-at-death from man- DG, Leventhal A, Wiberg RS, Gottsfield A, Gillette D, Bellifemine
dibular molar attrition in medieval danes. Scand J Dent Res E, Strother E, Cartier R, Fredrickson DA. 2007. Punctuated cul-
87:79–90. ture change in the San Francisco Bay Area. In: Jones T, Klar K,
Hershkovitz I, Latimer BM, Dutour O, Jellema LA, Wish-Baratz S, editors. California prehistory: colonization, culture and complex-
Rothschild S, Rothschild BM. 1997. Why do we fail in aging the ity. Lanham, MD: AltaMira Press. p 99–123.
skull from the sagittal suture? Am J Phys Anthropol 103:393–399. Mo S, Peng S. 1983. Attrition of upper and lower molars with
Houpt MI, Adu-Aryee S, Grainger RM. 1967. Eruption times of relation to age in Southern Chinese skulls. Acta Anthropol
permanent teeth in the Brong Ahafo region of Ghana. Am J Sinica 2:368–374.
Orthod 53:95–99. Molleson T, Cox M, Waldron AH, Whittaker DK. 1993. The Spi-
Igarashi Y, Useu K, Wakebe T, Kanazawa E. 2005. New method talfields report , Vol.2. The anthropology: the middling sort.
for estimation of adult skeletal age at death from the mor- New York: Council for British Archaeology.
phology of the auricular surface of the ilium. Am J Phys Molleson TI, Cohen P. 1990. The progression of dental attrition
Anthropol 128:324–339. stages used for age assessment. J Archaeol Sci 17:363–371.
Key CA, Aiello LC, Molleson T. 1994. Cranial suture closure Molnar S, McKee JK, Molnar I. 1983. Measurement of tooth
and its implications for age estimation. Int J Osteoarchaeol wear among Australian aborigines. 1. Serial loss of the
4:193–207. enamel crown. Am J Phys Anthopol 61:51–65.
Kieser JA, Preston CB, Evans WG. 1983. Skeletal age at death: an Murphy T. 1959. Gradients of dentine exposure in human molar
evaluation of the Miles method of aging. J Archaeol Sci 10:9–12. tooth attrition. Am J Phys Anthropol 17:179–186.
Lev-Tov N, Smith P. 2006. Variation in occlusal dental wear of Nowell GW. 1978. An evaluation of the Miles method of ageing
two Chalcolithic populations in the Southern Levant. Am J using the Tepe Hissar dental sample. Am J Phys Anthropol
Phys Anthropol 130:471–479. 49:271–276.
Levers BGH, Darling AI. 1983. Continuous eruption of some Papageorgopoulou C, Xirotiris NI, Iten PX, Baumgartner MR,
adult human teeth of ancient populations. Arch Oral Biol Schmid M, Rühli F. 2009. Indications of embalming in Roman
28:401–408. Greece by physical, chemical, and histological analysis. J
Li C, Ji G. 1995. Age estimation from the permanent molar in Archaeol Sci 36:35–42.
northeast China by the method of average stage of attrition. Parner ET, Heidmann JM, Kjær I, Væth M, Poulsen S. 2002.
Forensic Sci Int 75:189–196. Biological interpretation of the correlation of emergence times
Lovejoy CO. 1985. Dental wear in the Libben population: its of permanent teeth. J Dent Res 81:451–454.
functional pattern and role in the determination of adult skel- Pöllmann L, Berger F, Pöllmann B. 1987. Age and dental abra-
etal age at death. Am J Phys Anthopol 68:47–56. sion. Gerodontics 3:94–96.
Lovejoy CO, Meindl RS, Mensforth RP, Barton TJ. 1985a. Multi- Rainey F, Ralph E. 1959. Radiocarbon dating in the Arctic. Am
factoral determination of skeletal age at death: a method and Antiq 24:365–374.
blind tests of its accuracy. Am J Phys Anthropol 68:1–14. Richards LC, Brown T. 1981. Dental attrition and age relation-
Lovejoy CO, Meindl RS, Pryzbeck TR, Mensforth RP. 1985b. ships in Australian aboriginals. Archaeol Ocean 16:94–98.
Chronological metamorphosis of the auricular surface of the Ruff C. 1981. A reassessment of demographic estimates for
ilium: a new method for the determination of adult skeletal Pecos Pueblo. Am J Phys Anthropol 54:147–151.
age at death. Am J Phys Anthropol 68:15–28. Sagona A, Nikolaishvili V, Sagona C, Ogleby C, Pilbrow V,
Lunt DA. 1978. Molar attrition in Danes. In: Butler PM, Joysey Briggs C, Giunashvili G, Manjegaladze G. 2010. Bridging two
KA, editors. Development, function, and evolution of teeth. continents: renewed investigations at Samtavro, Georgia.
New York: Academic Press Inc. p 465–482. Tuba-Ar-Turk Acad Sci 13:314–334.
Mayhall JT, Belier PL, Mayhall MF. 1978. Canadian Eskimo Santini A, Land M, Raab GM. 1990. The accuracy of simple or-
permanent tooth emergence timing. Am J Phys Anthropol dinal scoring of tooth attrition in age assessment. Forensic
49:211–216. Sci Int 48:175–184.
Mays S. 2002. The relationship between molar wear and age in Scott EC. 1979. Dental wear scoring technique. Am J Phys
an early 19th century AD archaeological human skeletal se- Anthropol 80:11–24.
ries of documented age at death. J Archaeol Sci 29:861–871. Shirley NR, Jantz RL. 2011. Spheno-occipital synchondrosis
Mays S, de la Rue C, Molleson T. 1995. Molar crown height as a fusion in modern Americans. J Forensic Sci 56:580–585.
means of evaluating existing dental wear scales for estimating Smith P. 1972. Diet and attrition in the Natufians. Am J Phys
age at death in human skeletal remains. J Archaeol Sci Anthropol 37:233–238.
22:659–670. Steggerda M, Hill TJ. 1942. Eruption time of teeth among
McKee JK, Molnar I. 1988. Measurements of tooth wear among whites, Negroes, and Indians. Am J Orthodont Oral Surg
Australian aborigines. II: Intrapopulational variation in pat- 28:361–370.
terns of dental attrition. Am J Phys Anthropol 76:125–136. Stuart A, Ord JK. 1994. Kendall’s advanced theory of statistics,
Meindl RS, Lovejoy CO. 1985. Ectocranial suture closure: a revised Vol.1: Distribution theory. London: Edward Arnold.
method for the determination of skeletal age at death based on Suk V. 1919. Eruption and decay of permanent teeth in Whites
the lateral-anterior sutures. Am J Phys Anthropol 68:57–66. and Negroes. Am J Phys Anthropol 2:351–388.

American Journal of Physical Anthropology

 10968644, 2012, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ajpa.22106 by Egyptian National Sti. Network (Enstinet), Wiley Online Library on [22/09/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
192 C.C. GILMORE AND M.N. GROTE
Sullivan A. 2004. Reconstructing relationships among mortality, Visser EP. 1994. Skeletal evidence of Kava use in prehistoric
status, and gender at the Medieval Gilbertine Priory of St. Fiji. J Polynesian Soc 103:299–317.
Andrew, Fishergate, York. Am J Phys Anthopol 124:330–345. Walker PL, Dean G, Shapiro P. 1991. Estimating age from tooth wear
Sullivan A. 2005. Prevalence and etiology of acquired anemia in in archaeological populations. In: Kelley MA, Larsen CS, editors.
Medieval York, England. Am J Phys Anthopol 128:252–272. Advances in dental anthropology. New York: Wiley-Liss. p 169–178.
Todd TW. 1920. Age changes in the pubic bone. I: The male Whittaker DK, Molleson T, Daniel AT, Williams JT, Rose P,
white pubis. Am J Phys Anthropol 3:285–334. Resteghini R. 1985. Quantitative assessment of tooth wear, al-
Todd TW. 1921. Age changes in the pubic bone. Am J Phys veolar-crest height and continuing eruption in a Romano Brit-
Anthropol 4:1–70. ish-population. Arch Oral Biol 30:493–501.
Tomenchuck J, Mayhall JT. 1979. A correlation of tooth wear Whittaker DK, Ryan S, Weeks K, Murphy WM. 1987. Patterns
and age among modern Igloolik Eskimos. Am J Phys Anthro- of approximal wear in cheek teeth of a Romano-British popu-
pol 51:67–78. lation. Am J Phys Anthopol 73:389–396.
Ubelaker DH. 1989. Human skeletal remains. Washington, Wolpoff MH. 1971. Interstitial wear. Am J Phys Anthropol
D.C.: Taraxacum Press. 34:205–227.

American Journal of Physical Anthropology