Dental Age - Accuracy Tests of Tooth Formation Age Estimations … · 2014. 9. 14. · The resulting dental age distributions for the skeletal sample were com-pared to documented

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 92:173-188 (1993)

Accuracy Tests of Tooth Formation Age Estimations for Human Skeletal Remains

SHELLEY SAUNDERS, CAROL DEVITO, ANN HERRING, REBECCA SOUTHERN, AND ROBERT HOPPA Department of Anthropology, McMaster University, Hamilton, Ontario, Canada L8S 4L9

KEY WORDS Juvenile, Aging, Dental development

ABSTRACT Estimations of age from tooth formation standards for a large (n = 282) sample of subadult skeletal remains from a 19th century historic cemetery sample were analyzed. The standards of Moorrees et al. (1963a,b) for the permanent and deciduous teeth, and Anderson et al. (1976) for the formation of permanent dentition were employed in a variety of combi-nations to calculate mean dental ages. Tests of accuracy and bias were made on a small sample (n = 17) of personally identified individuals, and age of attainment scores were compared to age of prediction scores for each individ-ual. The resulting dental age distributions for the skeletal sample were com-pared to documented burial records for the cemetery to determine the repre-sentativeness of the skeletal sample. These comparisons showed little difference between age of attainment versus age of prediction methodologies. The standards of Moorrees et al. (1963a,b) were observed to provide the most accurate estimates of age with a standard deviation of one-half year. The standards of Anderson et al. (1976), while easier to use and more extensive, are problematic in that the original reference sample begins at three years of age, while the sample used by Moorrees and colleagues begins at birth. The skeletal age distributions compare well to the overall chronological age distri-bution for the cemetery. These results affirm that tooth formation age esti-mates for subadult skeletal remains from archaeological or forensic samples provide accurate assessments of age at both the individual and population level. © 1993 Wiley-Liss, Inc.

The dentition is the best physiological in-dicator of chronological age in juveniles (Holtz, 1959; Green, 1961; Demirjian, 1986; Smith, 1991b). Further, tooth formation, or more specifically, chronologies of tooth min-eralization as visualized radiographically, are now recognized to be the best means of estimating chronological ages of immature human skeletons from archaeological sites and for forensic cases (El-Nofely and Iscan, 1989; Merchant and Ubelaker, 1977; Saun-ders, 1992; Ubelaker, 1987, 1989).

The formation of tooth crowns and roots has been shown to be much less affected by hormonal influences, local and general envi-ronmental factors, and nutritional and so-cial factors than tooth emergence, skeletal

development, weight or height (Demirjian, 1986; El-Nofely and Iscan, 1989; Smith, 1991b). Evidence for this assertion includes relatively small variations in the timing of dental development in children with major abnormalities affecting maturation (Garn et al., 1965a; Kuhns et al., 1972; Niswander and Sujaku, 1965), lower coefficients of vari-ation in the stages of tooth formation than skeletal development (Lewis and Garn, 1960), and low correlations between tooth formation and weight, fatness, stature or

Address reprint requests to Shelley Saunders, Department of Anthropology, McMaster University, 1280 Main Street West, Hamilton, Ontario, Canada L8S 4L9.

Received October 12, 1992; accepted May 13, 1993.

© 1993 WILEY-LISS, INC.

174 S. SAUNDERS ET AL.

bone age (Anderson et al., 1976; Garn et al., 1965b and Gron, 1962, Gyulavari, 1966; Lacey et al., 1973; Prahl-Andersen and Roede, 1979; Steel, 1965). In addition, tooth formation has been found to be highly heri-table (Garn et al., 1965b; Moorrees and Kent, 1981).

A further advantage to using tooth forma-tion rather than tooth emergence in the esti-mation of chronological age is that it pro-vides a continuous series of developmental changes from before birth to about 20 years (Demirjian, 1986). While stages of tooth mineralization in dry bones are directly comparable to radiographic standards from living samples, living population standards for tooth emergence through the gingiva are not the same as observations of tooth emer-gence through the alveolus on dry bones (Demirjian, 1986; Merchant, 1973; Trodden, 1982). In addition, only a few published studies are suitable for use in age prediction because they include data for a number of teeth, consist of large reference samples and have an applicable methodology for report-ing ages (Anderson et al., 1976; Demirjian et al., 1973; Demirjian and Goldstein, 1976; Haavikko, 1970; Kataja et al., 1989; Moor-rees et al., 1963a,b; Nystrom et al., 1986). Other studies that use maturity scales (Demirjian et al., 1973; Demirjian and Gold-stein, 1976) in reporting tooth formation chronologies are difficult for osteological re-searchers to apply. Only the standards of Moorrees et al. (1963a,b) include data for both deciduous and permanent teeth from the same series of children. In addition, sub-jects in all of the above studies are of Ameri-can or European derivation with almost no data available for other geographic areas of the world.

While these difficulties are widely known and accepted, skeletal biologists working with archaeological material generally have no recourse to test the accuracy of dental age estimates because the true chronologi-cal ages of their samples are rarely known.1

"Some exceptions to this situation include certain historic cem-etery samples (Adams and Reeve, 1987; Bowman et al., 1990; Cunha, 1993).

In fact, the question of age accuracy is laid aside in growth-related studies of past popu-lations by invoking the assumption that the most useful comparisons of any group are between the two physiological indicators of maturation, dental age, which is more sta-ble, and skeletal age, which is more sensi-tive to environmental insults. This observa-tion has been the basis for a number of growth-related studies over the last twenty years (Hoppa, 1992; Hummert, 1983a,b; Jantz and Owsley, 1984a,b; Lallo, 1973; Lovejoy et al., 1990; Mensforth, 1985; Mer-chant and Ubelaker, 1977; Owsley and Jantz, 1985; Saunders and Melbye, 1990; Sundick, 1978; Walker, 1969; Wall, 1991). While the logic is justified, this assumption ignores the possibility that the dental age distribution of a skeletal sample may not even be representative of the dental age dis-tribution in the living population because of sampling error and/or a bias in morality) —

Only a few researchers (Bowman et al., 1990) have examined the accuracy of dental age estimates in archaeological or forensic samples. Tests of age prediction using chil-dren of known age have been reported from studies of living children (for a review of this literature see Smith, 1991b). In a report on a small sample of 23 children, Crossner and Mansfield (1983) found 70% of tooth forma-tion estimates for permanent mandibular and anterior maxillary teeth fell within ± 3 months of true age and discrepancies of no more than 6 months were found for age esti-mates when tested against the standards of Liliequist and Lundberg (1971) and Gustaf-son and Koch (1974). Another comparison of several dental formation standards using permanent teeth (Haag and Matsson, 1985) found standard deviations of difference be-tween dental and chronological age to be ap-proximately 10% of age. The tooth formation standards of Demirjian et al. (1973) esti-mated a subject's chronological age (from 2 to 20 years) within 15-25 months of 85% confidence. Later, tests of these Canadian standards on a sample of French children in Lyons found a mean advancement of nine months in the dental development of the French children (Proy et al., 1981). A further test of the methods developed by Demirjian

TOOTH FORMATION AGE ESTIMATION

175

and co-workers on Finnish children found that the models predicted dental age reli-ably (Kataja et al., 1989).

Smith (1991b) points out that none of the tested systems are particularly suited to age prediction. In fact, she has provided a series of recommended formation values for age prediction based on her determination of the appropriate method for constructing chro-nologies of growth stages. She argues that for age prediction it is more appropriate to assign an age that is the midpoint between the mean age of attainment of a subject's current stage of formation and the subse-quent one, since at the time of observation, the subject is in-between the attainment of one stage and the next. Smith's own test of age prediction accuracy utilized four Cana-dian children of British origin who were used to test dental age standards in the study by Anderson et al. (1976). The chrono-logical ages of these children were compared to her calculated values for predicting age from the stages of permanent mandibular tooth formation derived from the data of Moorrees et al. (1963a) (Smith, 1991b). The results were "remarkably accurate" (Smith, 1991b), differing by a maximum of 0.2 years when schedules for the correct sex were ap-plied and even when schedules for both sexes were combined and applied. Within-individual inaccuracy' based on a single tooth yielded a standard deviation of ± 0.56 years while mean values for five or more teeth decreased the standard deviation to -± 0.09 years, suggesting that dental age can be estimated to within 2 months for young children Smith expresses some sur-prise that the few earlier empirical tests of dental age prediction show good success de-spite her enumeration of considerable theo-retical and methodological difficulties asso-ciated with existing dental development standards. As yet, no one has tested her rec-ommended values for age prediction based on stage of development.

2lnaccuracy, expressed numerically, indicates the amount of error of difference between the known age and estimated age. Bias, expresses the direction or sign of the difference (Lovejoy et al., 1985).

The difficulties of dental age estimation of immature human skeletons would appear to be magnified with archaeological samples. These samples tend to be fragmentary and the number of teeth available for mean den-tal age estimates varies considerably. Con-sequently, examinations of both the accu-racy of individual dental age estimates of personally identified skeletons (from ar-chaeological or forensic circumstances) and the accuracy of skeletal sample distribu-tions of dental age against known burial age distributions would help address many of the concerns cited above. The present study reports on dental formation age estimates of a small sample (n = 17) of personally identi-fied children's skeletons from a nineteenth century historic cemetery from Upper Can-ada. In addition, distributions of dental age derived from several standards and meth-ods were compared for the total juvenile sample of dentitions (n = 241) to the ex-pected distribution of chronological ages-at-death based on a complete series of burial registers for the same cemetery. The study considers differences in age derived from combinations of three age estimation sys-tems: the standards of Anderson et al. (1976) for permanent teeth, the standards of Moorrees et al. for the permanent (1963a) and deciduous (1963b) teeth. In addition, the variation between the results for these systems is examined as well as the variation in dental age based on different combina-tions of teeth and the relative confidence that can be expected for age predictions from archaeological cases.

MATERIALS AND METHODS

The St. Thomas' Church cemetery skeletal sample

In 1989, St. Thomas' Anglican Church, Belleville, Ontario was given legal permis-sion to close and partially disinter human remains from a nineteenth century ceme-tery located on land adjacent to the church property. The church hired a private archeo-logical contract firm, Northeastern Archaeo-logical Associates, to carry out the disinter-ments of skeletal remains and associated artifacts. A total of 579 burial shafts and


some 576 individuals were excavated over four months from three-fourths of a hectare or an area of almost one and three-fourths acres. While the true area of the total ceme-tery is not known, the excavated area is be-lieved to represent approximately one-third of the original cemetery grounds (McKillop et al., 1989; Saunders et al., 1991). Tomb-stones or grave markers had long since been disturbed by historical events (i.e., two church fires, one in each century) and only a few remained standing on the property.

The excavation area was cleared of soil overburden and individual grave shafts were identified by shovel shining and then carefully excavated by hand. Additional test pitting was used to check for graves not identified by soil changes. The well-drained sandy soil promoted easy and rapid removal of the remains as well as excellent bone preservation. The skeletons were all studied initially at the church and then in the De-partment of Anthropology, McMaster Uni-versity, so that full scale collection of data was carried out for one year prior to reburial on church property.

The cemetery was known from parish reg-isters and church vestry minutes to have been used from August 30, 1821 to April 14, 1874 whereafter all subsequent town buri-als book place at the municipal cemetery (Bellestedt, 1969; Diocese of Ontario Ar-chives [DOA] Series I, part 35A). Based on comparisons to the parish registers, the skeletal sample represents 37% of the total number of 1,564 interments during that pe-riod (Saunders et al., 1991). While a sub-sample of individuals (n = 80) is personally identified from legible coffin plates or coffin fastenings, no cemetery plan or map sur-vives, probably because of its loss during the earlier church fires.

The source of documentation: quality of the records

Records of burials in St. Thomas' ceme-tery, as well as marriages and baptisms in the church, were kept by the church's minis-ters and are available for the entire period during which the cemetery was in use. Data include name, age, death date, name of the registrar, burial date, and occasional notes on family relationships as well as cause of

death. All records were transcribed to a da-tabase management program (Borland In-ternational, 1990) and checked twice by dif-ferent individuals for transcription errors.

A total of 1,564 individuals were recorded as having been interred in the cemetery. An additional 17 individuals were identified as having been buried in other locations be-sides St. Thomas's cemetery, all of which indicate the attention to detail adhered to by the recorders. Consequently, a series of his-torical demographic tests were applied to evaluate the quality of the parish registers (Drake, 1974). These involve determining the adequacy of sample size (a minimum of 100 registry entries per year); whether there are obvious and serious gaps in the regis-ters; evidence of under registration; that persons buried in other churchyards are so indicated in the records; and that recorded sexes and ages are not themselves esti-mates. The analysis of St. Thomas' parish registers (Rogers, 1991) showed that some individuals were buried without record but that these are rare, isolated cases. Reduc-tions in the number of parish events oc-curred between 1831 and 1835, but the num-ber of burials was unaffected. No 'significant gaps in the registers were detected and pa-rishioners buried elsewhere were so noted in the register. Individuals of unknown sex represent a nonsignificant proportion of the total sample (less than 2%) and those of un-recorded age appear to be randomly distrib-uted throughout the age range. Conse-quently, it was concluded that the register data can be confidently treated as a reliable source for comparison to skeletally derived sex and age profiles.

Of the total burials registered, 712 indi-viduals were listed as having died under the age of 15. The ages of 637 of these children (89%) were clearly identified in the burial register, leaving 75 cases that could not be included in the sample. However, by cross-checking the latter against the recorded baptisms in the parish registers, it was pos-sible to identify the specific ages of 45 addi-tional individuals, increasing the registry sample of subadults up to 15 years to 682, or 96%. This is a good example of how record linkage techniques improve the quality of parish record data.

<

A c.

TOOTH FORMATION AGE ESTIMATION 177

Nineteenth century settlers in the Upper Canada town of Belleville were comprised mainly of immigrants from the British Isles, Ireland and western Europe, as well as de-scendants of United Empire Loyalists from the United States (Boyce, 1967, 1991; Mika and Mika, 1986). It is also possible to iden-tify the country of origin for many of the buried individuals by reference to baptism or marriage records. In addition, early cen-suses and assessment records for the town summarize the geographic origins for Belleville's inhabitants (Canada (Province) Board of Registration and Statistics, 1853; Canada Bureau of Agriculture and Statis-tics, 1878). While it might be expected that all burials in St. Thomas's cemetery would be Anglican or Church of England, in fact, the church sold burial plots to some other denominations such as Methodists and Presbyterians throughout the century, thereby increasing some of the religious and ethnic variability of the burial sample (Her-ring et al., 1992). One Mohawk Indian and two "persons of colour" were identified in the burial register as having been buried in the cemetery. However, these three individuals were all adults at the time of their deaths.

Age estimation of subadults

Subadult skeletons were first identified on the basis of active tooth eruption and skeleton epiphyseal development and fu-sion. A total of 282 juvenile or nonmature individuals were initially identified in the skeletal sample. Of these, it was possible to evaluate tooth formation for 241 cases (86%) up to 15 years using X-rays (both panoramic and lateral) and macroscopic observations. Seventeen of these individuals (7%) had been personally identified on a separate oc-casion by reference to coffin plates (Saun-ders et al., 1992a).

The 14 developmental stages proposed by Moorrees, Fanning and Hunt (MFH) (1963a) were determined for all present and observ-able teeth. A single observer was trained to recognize the stages of crown and root for-mation and double determinations were made from the X-rays. Interpolation charts for the mean dental age of three deciduous teeth (mandibular canine, first and second molars) (MFHd) and ten permanent teeth

(MFHp) were prepared from the charts published by Moorrees et al. (1963a,b). These values were then calcu-lated by cross comparisons using a database management program to the tooth forma-tion stages recorded for the sample. In addi-tion, mean dental ages for 16 permanent teeth were taken from the tables provided by Anderson, Thompson and Popovich (ATP) (1976) and the same individual tooth mean ages calculated by cross comparisons.

Although Moorrees et al. (1963a) divide permanent mandibular root formation stages into mesial and distal roots, a sepa-rate test of the Belleville data found no sig-nificant differences between the two roots for degrees of calcification. Consequently, these two root stages were averaged after data collection. Also, since it is not possible to determine the elapsed period since an observed tooth has attained stage 14 (root apex —1st, complete), these means were not included in --- the overall age assessment unless theTo7iii Was deemed transitionallietween stages 13 and 14, after which a midpoint between the two means was taken.

Overall individual mean age estimates (values for the two sexes pooled) were calcu-lated from the sums of individual mean tooth ages based on the published values for each of the three standards. These values represent "mean age of attainment" esti-mates. Next, three combinations of these standards were also calculated by averaging the values for each method: (1) ATP (perma-nent) and MFH (permanent); (2) ATP (per-manent) and MFH (permanent) and MFH (deciduous), and (3) MFH (permanent) and MFH (deciduous). First, case-by-case com-parisons for the personally identified "knowns" in the skeletal sample were made using the three individual standards and the three combinations of age estimation techniques and compared to documented age at death. In addition, the absolute and mean differences between these standards and documented age were calculated. Sim-ple standard deviations for the overall mean dental age estimates for each individual in the entire sample were also calculated as a gauge of relative between-tooth agreement. Next, within-individual coefficients of varia-tion based on the variation contributed by

■,k 7_,TEAr , r

230 346 88 93 91 62

352 428 B 113 147 463 318 288 398 109 174 148

Burial no.

4"1

4't

't'i

'l'1

44

44

44

t41,T

178

S. SAUNDERS ET AL.

TABLE 1. Estimated age at death from tooth formation standards (pooled sexes) versus documented age of death for 17 personally identified individuals from St. Thomas' Cemetery skeletal sample'

Documented MFHpd MFHp MFHd ATP ATP/MFHp ATP/MFHpd

0.01 0.00 0.00 0.00 0.17 0.39 0.39 0.39 0.69 0.86 1.29 0.71 3.50 2.40 1.38 0.75 0.78 0.78 0.78 0.83 1.06 1.00 1.08 3.40 2.20 1.53 0.86 0.87 1.00 0.83 3.60 2.73 1.65 1.50 1.52 2.17 1.39 3.70 2.94 1.83 2.00 2.55 3.73 1.57 3.71 3.72 2.86 2.26 2.41 3.28 1.55 3.81 3.66 3.21 2.33 1.51 2.05 1.15 3.71 3.34 2.79 2.33 1.90 2.72 1.28 3.79 3.47 2.84 2.75 2.81 3.65 1.83 4.09 3.91 3.37 3.85 3.19 4.11 2.26 4.37 4.30 3.86 3.94 3.10 3.72 1.88 3.75 3.74 3.44 4.13 3.28 4.07 2.02 3.94 3.99 3.62 8.00 7.22 7.22 7.25 7.60 7.46 7.45 8.42 6.28 6.28 6.22 6.24 6.24

'ATP, Anderson, Thompson and Popovich (1976); MFH, Moorrees, Fanning and Hunt (d)eciduous (1963a), and (p)ermanent (1963b).

different teeth to the overall mean age esti-mate were calculated for all individuals aged by more than one tooth. Finally, the sample age distributions calculated from the dental remains were compared to the sam-ple juvenile age distribution for the total cemetery, derived from the parish registers.

A second set of age estimates were calcu-lated using "age of prediction" tables derived from the MFH permanent tooth standards interpolated for this study and compared to the "mean age of attainment" estimates as well as to the known chronological ages for the personally identified sample. In all, cal-culations of mean dental ages were weighted by the number of teeth used in each method for any one individual. The ab-solute difference between the estimated age of attainment or prediction and documented age can be used as a gauge of accuracy (Smith, 1991b) or corresponding inaccuracy (Lovejoy et al., 1985) of the dental age esti-mates. The sign of the difference will indi-cate bias of direction in the method of age estimation. Mean levels and standard devia-tions of accuracy can also be calculated de-pending on the number of teeth used to de-rive an estimate within any one individual.

RESULTS

Table 1 lists the dental age estimates for the three sets of standards (ATP, MFHp and MFHd) alone as well as the three combina-tions of standards as described in the Mate-

rials and Methods for the seventeen individ-uals personally identified from coffin plates and burial records. The absolute differences between these six estimates for each case and the documented age of each individual as well as the mean differences and stan-dard deviations for the overall sample are listed in Table 2. While considerable varia-tion exists between the estimates for any one individual, the two combination meth-ods of ATP + MFHp + MFHd and MFHp + MFHd produce the best overall accuracy in age estimates as well as the lowest stan-dard deviations. The average difference in age estimation for these two combination methods is approximately one-half year.

Individuals dying under one year of age are, not surprisingly, more closely aged by the MFH deciduous standards but overall sample estimates favour the two combina-tion methods identified above. In addition, very young infants dying at less than one month often cannot be aged by any of these methods because of the lack of sufficiently formed deciduous teeth. In the case of the St. Thomas skeletal sample, a number (n = 25) of such individuals required assessment based on a combination of studies of early deciduous development (Deutsch et al., 1985; Kraus and Jordan, 1965; Lunt and Law, 1974; Prahl-Andersen and van der Linden, 1972).

The proportions of individuals represent-ing annual age cohorts for each of the three


179

TABLE 2. Absolute differences between estimated age at death (years) from tooth formation standards (pooled sexes) versus documented age at death for 17 personally identified individuals from St. Thomas' Cemetery skeletal sample'

Burial no. Documented MFHpd MFHp MFHd ATP ATP/MFHp ATP/MFHpd

230 0.01 0.010 0.010 0.010 346 0.17 0.220 0.170 0.220 0.170 0.220 88 0.69 0.165 0.600 0.020 2.810 1.705 0.694 93 0.75 0.027 0.750 0.027 0.750 0.027 91 0.83 0.233 0.170 0.253 2.570 1.370 0.700 62 0.86 0.006 0.140 0.028 2.740 1.873 0.787

352 1.50 0.020 0.670 0.110 2.200 1.435 0.331 428 B 2.00 0.552 1.734 0.433 1.710 1.723 0.861 113 2.26 0.152 1.017 0.713 1.550 1.405 0.951 147 2.33 0.820 0.285 1.177 1.380 1.010 0.463 463 2.33 0.434 0.393 1.055 1.460 1.140 0.513 318 2.75 0.061 0.900 0.918 1.340 1.159 0.617 288 3.85 0.662 0.263 1.587 0.520 0.450 0.014 398 3.94 0.838 0.225 2.065 0.190 0.203 0.497 109 4.13 0.849 0.060 2.112 0.190 0.143 0.507 174 8.00 0.779 0.782 0.750 0.400 0.542 0.548 148 8.42 2.139 2.139 8.420 2.200 2.178 2.178 Absolute Mean Difference 0.47 0.64 1.17 1.39 1.17 0.58 s.d. 0.52 0.57 1.93 0.91 0.61 0.49

1ATP, Anderson, Thompson and Popovich (1976); MFH, Moorrees, Fanning and Hunt (d)eciduous (1963a), and (p)ermanent (1963b).

individual standards alone: (1) ATP (stip-pled); (2) MFH permanent (grey), and (3) MFH deciduous (black) are shown in Figure 1. This figure illustrates the differ-ences in the distributions which reflect the sampling biases and truncations of the ref-erence samples from which the original standards were derived. Particularly note-worthy are: the inability of deciduous stan-dards to age beyond nine years, and the omission of cases under three years in the ATP for permanent teeth. The original growth sample used by Anderson et al. (1976) does not include individuals younger than three years of age, whereas the Moor-rees et al. (1963a,b) sample begins at birth. As a result, the distribution of age estimates based on the Anderson et al. (1976) stan-dards gradually approaches a minimum of three years but does not extend below this age. This trend is more clearly shown in Figure 2. When individual age estimates us-ing the ATP standards are plotted against corresponding estimates using the MFH permanent standards, the effect of the trun-cation of the original ATP sample is evident by the over-aging of children below five years.

Variation in age estimations also in-creases with increasing age of the subjects when within-individual standard deviations are plotted against mean ages. The calcu-

lated correlation between the two variables is significant (see Fig. 3) (r = 0.694, P < 0.001). However, a certain proportion of individuals falling in the 2-to-5-year-old age categories seem to have high standard deviations be-yond what would be expected if a regression line was calculated. In fact, when these cases were examined, it was determined that the permanent MFH maxillary incisor age estimates were high compared to esti-mates for other teeth. Since the published formation charts for these teeth in the Moor-rees, Fanning and Hunt publication begin at a relatively late stage of development-crown complete-it might be expected that these values would have a biasing effect on age estimations of young children much like the Anderson, Thompson and Popovich standards. This difference is attributed to the Moorrees et al. (1963a) values for the anterior and lateral teeth being drawn from two separate samples.

Calculated within-individual coefficients of variation of age estimates ranged from less than 1 to 52, a broad range of variation. These coefficients were plotted against the number of teeth contributing to a mean den-tal age (Fig. 4). There is an inverse relation-ship between variation in age and the num-ber of teeth contributing to a mean dental age. This effect does not occur until at least six teeth contribute to the age estimate but


St. Thomas Church Skeletal Sample Dental Age Distribution by Method

50%

45%

40%

35%

30% 0 0 25% 2 a.

20%

15% 0

10%

iiili 5%

-1 . - ' 1.44 1171 91F1I + 1 --.I I 0% 1 1 1 1 1 i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Age Cohorts (up to but excluding)

MHF (dec) MFH (perm) 0 ATP (perm)

Fig. 1. Proportion of individuals per one-year age cohort for each of three dental standards; ATP: Anderson, Thompson and Popovich (1976); MFHp: Moorrees, Fanning and Hunt, permanent (1963a); MFHd: Moorrees, Fanning and Hunt, deciduous (1963b). •

most of the St. Thomas' cases preserve five or fewer teeth.

Consequently, the combination method of the permanent and deciduous MFH stan-dards, which had been identified as the best combination method for age estimation, were compared to the 682 burials of individ-uals up to 15 years listed in the burial records, with the permanent maxillary inci-sors included and excluded in the estimates (Fig. 5). The comparisons of the proportions of individuals in each age category by year for the dental age estimates are compared to the records' proportions. It can be seen that both dental age methods are quite close to the records' values but the method which includes only permanent mandibular teeth from the MFH standards fares slightly better.

To test this observation, documented ages for the cemetery sample were treated as ex-pected values and chi square tests of good-ness of fit were run using the two methods

(Table 3). The MFHpd (mandibular and maxillary) age estimates were compared to the burial records and returned a likelihood ratio chi square value of 19.10 at a signifi-cance level of P = 0.024. The calculation of standardized residuals' showed that the 6-to-7-year-old frequencies were significantly overrepresented (standardized residual = 3.21) in the skeletal sample and signifi-cantly underrepresented in the documents sample (standardized residual = —1.91). When the likelihood ratio chi square was run for the comparison of the MFHpd (man-dibular teeth only) dental age estimates to the burial records, a value of 15.77 resulted with a probability level of P = 0.072. The calculated standardized residuals again

3Standardized residuals in excess of ± 1.64 (the 95th percen-tile of the standard normal distribution) indicate a significant deviation in the observed cell values from those, expected. (Rey-nolds, 1977).


St. Thomas Church Skeletal Sample Dental Age Estimation by Method

F-

16 —

15 —

14 —

13 —

12 —

11 —

10 —

9 —

8 —

7 —

6-

5 --

4 —

3 —

2-

1 —

0

0

0 0

? of

0% 0 0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

MFH (perm)

Fig. 2. Estimates of age from tooth formation stan-dards. MFHp plotted against ATP, demonstrating the effects on age estimation of the truncation at three years of age in the ATP reference sample.

yielded a significant value for the 6-to-7-year age category in both the skeletal sam-ple (standardized residual = 2.99) and the burial records sample (standardized resid-ual = —1.78). This finding, again, indicated an overrepresentation of 6-to-7-year-olds in the skeletal sample. Since cell sizes in the 9 years or over cohorts of the skeletal sample are small, it was necessary to com-bine the 9-14 year cohorts in the chi square calculation. However, a further assessment of the ratios of 10-to-14-year-olds versus all other cohorts showed no significant differ-ences between the skeletal and documents sample.

Finally, for the ten personally identified individuals with observable developing per-manent teeth, determinations were made of the absolute differences between estimates using the MFH permanent mandibular tooth standards, the ATP permanent tooth standards and finally, ages of prediction based on the correct sex and incorrect sex (Table 4). Prediction values were derived fol-lowing the method described by Smith (1991b) from our originally interpolated MFH permanent mandibular standards.

These results, presented in Table 4 show, first of all, that age estimates using the cor-rect sex standards only, are not necessarily closer to the documented ages than are the standards used from the incorrect sex. In addition, the estimates based on age of pre-diction standards for Moorrees et al. (1963a,b) are not substantially different from age of attainment standards. While it can be seen that for most individual cases, the age of prediction values move upwards towards an older age estimate, as indicated by the numerical sign of the reported differ-ences, the absolute mean differences (i.e., inaccuracy) for the two types of standards are almost identical.

DISCUSSION

Comparisons of mean dental age esti-mates using several combinations of differ-ent dental formation standards, permanent teeth, deciduous teeth, and two different source samples, show very clearly that the combination of permanent and deciduous standards based on the sample of children published by Moorrees et al. (1963a,b) is the best possible method to use in estimating mean dental age from skeletal samples of juvenile skeletons, at least when general population background of the samples is comparable. In the present study, it was as-sumed that family origins based in the Brit-ish Isles, Ireland, and western Europe are similar enough to those of the American white children examined by these research-ers to have comparable dental development schedules. While slight differences in dental maturity have been found between modern North American and European samples of children (Demirjian, 1986; Demirjian and Lamarch, 1979; Prahl-Andersen et al., 1979; Proy et al., 1981), the absence of maturity schedules reported in the same way as the MFH standards precludes our examination of such differences. The use of, or addition of, the standards published by Anderson et al. (1976) is not recommended because of the truncation of their reference sample at three years of age. This course is recommended even though the ATP standards are easier to apply (the availability of published means and standard deviation values in tabular form for age of prediction) than the MFHpd

Sta

ndar

d D

evia

tion

0.50

0.00

2.00

2.50

1.00

1.50

182

S. SAUNDERS ET AL.

Variability of Estimates

Against Mean Age

Fig. 3. Standard deviations of individual dental age estimates (MFHp) plotted against overall mean age, demonstrating an age progressive increase in variability of estimates.

standards which require interpolation from graphs of the means and variation of dental formation. Here then, is a troublesome methodological problem, since each re-searcher seems to interpolate the MFH ages each time they are required or create their own charts of interpolation.'

The need for the inclusion of deciduous tooth standards is made more evident by the fact that most skeletal samples contain large proportions of infants and young chil-dren. However, there is still a distinct short-age of detailed standards for the early for-mation of deciduous crowns that can be applied to fetal and neonatal skeletons (Skinner and Goodman, 1992) which cannot be included in MFHpd mean age estimates.

Within-individual calculated coefficients of variation (CV) for the total sample of per-manent dentitions span a wide range (1 to

4The original mean and standard deviation values from the study are not available (Moorrees, personal communication).

52) with an average value of 20. This is in contrast to the low range of CVs (2-16) re-ported by Smith (1991a) for a small sample of humans. The average CV in her study is 10. In fact, the total range of coefficients of variation that Smith reports for humans along with several Pan and fossil hominid specimens reaches a maximum of 36. There are probably several explanations to explain the wider range in the St. Thomas' sample. These include the fact that most of the St. Thomas' sample of dental ages are based on five teeth or less, whereas Smith's sample of humans all have six or more teeth included in their estimates. Maxillary teeth tend to have higher CVs than mandibular teeth (Demirjian, 1986; Smith, 1991a). Our exam-ination of the St. Thomas cases with CVs over 40 suggests that variability in age esti-mates is often increased because of large dis-crepancies between the maxillary incisors (aging upward) and mandibular teeth (aging downward). In addition, some of the highest


183

Variability of Estimates Against Number of Teeth

60

50

40 5

I 30

20

10

Fig. 4. Calculated within-individual coefficients of variation for estimated age (MFHp) plotted against the number of teeth utilized to calculate the estimate.

standard deviations of within-individual age estimates are found among individuals under five years of age. These factors are produced by an additional sample bias effect from the use of the permanent MFH maxil-lary incisor standards which only include values from approximately 4 years and be-yond, and which are derived from a different reference group that the other values. These observations of coefficients of variation should serve to emphasize that studies of small samples of fossils and source cases may not necessarily present the full picture of population and sample variability (Cope and Lacy, 1992; Plavcan, 1989).

When both deciduous and permanent maxillary and mandibular tooth formation stages are included in mean dental age esti-mates for this group of young children, the total sample of dental ages using the MF-Hpd combination is significantly different from the chronological age-at-death distri-bution provided by the parish registers. However, exclusion of the MFH permanent

maxillary incisor standards from estimates improves the goodness of fit between the skeletons and records as shown by a de-crease in the chi square value even though the 6-to-7-year-old cohort of skeletons still appears overrepresented. Investigation of the absolute numbers of individuals repre-sented at the 6-to-7-year age level in the two samples showed that these ages must be reasonable accurate. It is unlikely that esti-mations of age would be any more problem-atic in one age cohort versus all others when the tooth development standards are de-rived from a longitudinally studied refer-ence sample. In addition, further examina-tion of the mean dental ages of the 15 individuals placed in this cohort finds them normally distributed over the age range.

Rather, the authors speculate that the ex-cavation of St. Thomas' cemetery produced a biased selection of burials of children at this age, and suspect that there was selec-tive burial in the excavated portion of the cemetery resulting from several children dy-

MFHpd Imand&max) fl Documented

184

S. SAUNDERS ET AL.

St. Thomas Church Skeletal Sample Dental Age Distribution by Method

50%

45%

40%

35%

30%

D. 25%

2

20%

15%

10%

5%

0%

Fig. 5. Sample comparisons of two methods of estimating age (MFHpd) one with the permanent maxillary teeth included in the estimate of age, the second with the maxillary incisors excluded from the age estimate, and compared to the overall age distribution of the documented burial records for individu-als under 15 years of age.

ing from a common cause, possibly drown-ing.' While the recorded causes of death in the parish registers are not numerous, those burials which do have a cause of death listed show an unusually high occurrence of drowning deaths, a problem for nineteenth century Belleville where the fast flowing Moira River which runs through the town experienced dramatic spring runoffs (Moodie, 1853; Herring et al., 1992). Conse-quently, it appears that any apparent dis-crepancies between the skeletal sample and the documented burials are due to sampling problems and not to problems of age estima-tion. In fact, when a further sample of 64 subadult skeletons not ageable by the Moor-rees, Fanning and Hunt tooth formation cri-teria are added into the skeletal sample, the

'Two out of the fifteen 6-to-7-year-old children listed in the parish registers have circumstances of death described and both of these drowned in the Moira River.

proportion of infants under one year in- creases dramatically and differs significantly from the documented data. This dif-

TABLE 3. Standardized residuals for likelihood ratio chi square calculations comparing skeletal sample (n= 241)

dental age estimates to documented burials (n= 682)

Age cohort Skeletal' Documents Skeletal' Documents

1 .71 -.42 .71 -.42 2 -1.35 .80 -.48 .28 3 -.47 .28 -.31 .18 4 .84 -.50 -.70 .42 5 -.63 .37 -.63 .37 6 -.11 .06 -.11 .06 7 3.213 -1.913 2.993 -1.783 8 .02 -.01 .34 -.20 9 -.60 .36 -.91 .54 9-15 -1.04 .62 -1.04 .62

'MFH permanent and deciduous tooth age estimates with maxillary standards included, compared to documented burials from parish reg-isters. 2 MFH permanent and deciduous tooth age estimates with maxillary standards excluded, compared to documented burials parish registers. 'Standardized residuals in excess of ± 1.64 indicating significant de-viations in the observed cell values.


TABLE 4. Differences and absolute overall mean differences between estimated and documented age for age of attainment versus age of prediction schedules for the permanent dentition only

B109 B113 B148 B352 B428b B147 B174 B288 B318 B463 Mean s.d.

MFH, correct sex -0.74 -0.05 -2.62 0.67 0.50 -0.29 -1.24 0.26 0.79 0.26 0.74 0.71 MFH, incorrect sex -0.71 -0.22 -2.74 0.67 0.59 -0.21 -0.73 0.35 -0.38 0.17 0.68 0.72 ATP, correct sex -0.01 1.52 -1.99 2.20 1.77 1.27 -0.82 0.43 1.13 1.34 1.25 0.65 ATP, incorrect sex -0.37 1.37 -2.60 1.65 1.27 -0.10 0.91 1.63 1.57 1.15 0.77 MFH Prediction, correct sex -0.31 0.39 -2.08 0.92 0.42 0.13 -0.67 0.63 1.14 0.63 0.73 0.53 MFH Prediction, incorrect sex -0.33 0.29 -2.35 0.96 0.46 0.11 -0.14 0.77 0.81 0.59 0.68 0.62

ference may be attributable to biased sampling of the cemetery during excavation (Herring et al., 1992).

The use of age prediction tables versus age of attainment tables for estimating age does not seem to make much of a difference in the average level of accuracy of age esti-mation for the Belleville sample of person-ally identified children, even though it has been argued that age prediction standards should be closer to true age (Smith, 1991b). The calculation of prediction values for the Moorrees et al. (1963a) standards requires two stages, first to interpolate the attain-ment values from the graphs, and second to calculate the prediction values from these interpolated values. Given the problem of interobserver error in interpolation from these graphs, any errors will be further translated into the prediction values. While it is clear that the differences between esti-mated and documented age for the MFH age of prediction values are generally positive, indicating higher estimated ages as ex-pected, this is not always the case so that the absolute mean value of the difference is al-most identical to that for the age of attain-ment estimates. The fact that attainment values do not prove to be any less accurate than secondarily calculated prediction val-ues for the St. Thomas' Cemetery sample would suggest that the added step provides no added benefit. Consequently, the use of age of prediction standards is not necessar-ily likely to improve dental age estimation in archaeological and forensic cases as indi-cated by this study. Other methodological problems as identified above are probably more pressing.

The lack of a clear difference in the accu-racy of correct sex versus incorrect sex esti-mates seems puzzling. However, this is likely a result of the fact that most of this

sample of ten individuals is under five years of age when there is greater similarity be-tween girls and boys in development timing. In fact, Demirjian and Levesque (1980) found no difference in the timing of dental development between girls and boys up to five years of age. The number of teeth used in age estimation does make a difference as Smith (1991b) showed for the four cases she tested. Dental age estimates in the St. Tho-mas' sample based on a single tooth have a standard deviation of ± 0.94, while the av-erage standard deviation when all possible teeth are used is ± 0.38 years. However, es-timates that include all available teeth, de-ciduous as well as permanent, are better in-dicators of overall accuracy so, that the standard deviation of error for the MFHpd method, excluding maxillary incisors, as re-ported in Table 2 which is ± 0.53 years, should be taken as the gauge of how closely one might expect to estimate the dental age of an archaeological or forensic case.

In summary, it is recommended that os-teological researchers use the tooth forma-tion standards of Moorrees, Fanning and Hunt for permanent and deciduous teeth as has been recommended by previous re-searchers (El-Nofely and Iscan, 1989; Mer-chant and Ubelaker, 1977; Ubelaker, 1987,1989). The standards of Anderson, Th-ompson and Popovich should not be used for individuals less than five years of age to avoid problems of sample truncation. This study shows that there is no reason to add the ATP estimates to MFH estimates in older children since dental age estimates for these latter standards alone closely approxi-mate chronological age distributions. How-ever, values for permanent maxillary inci-sors published by Moorrees et al. (1963a) should be excluded from age estimates be-cause of their biasing effect especially on in-


dividuals in the 2-to-5-year-old range. There does not seem to be a problem with using the MFH "age of attainment" schedules for age prediction although this observation may change with larger samples of tested cases. Researchers can be optimistic about their dental age estimates for archaeological and forensic cases but only as long as skeletal and dental preservation is good, as was the case for the St. Thomas' sample, and obser-vation error is controlled for as is the ever present problem of population variability.

ACKNOWLEDGMENTS

The authors thank St. Thomas' Anglican Church, Belleville, for giving its permission to study the skeletal remains and to tran-scribe parish records. We also thank the fol-lowing individuals for their assistance in the completion of this project: Gerry Boyce, Heather McKillop, Loralyn Geise, Andrea Lagan, Laurie Dailey, and Warren Pollett. The following institutions provided facilities and technical assistance: Chedoke Hospital, Hamilton, McMaster University Medical Centre Radiology Department and McMas-ter University Faculty of Social Sciences. The authors also thank four anonymous re-viewers for their comments on an earlier draft of this manuscript. This research was funded in part by the Ontario Heritage Foundation, McMaster University Arts Re-search Board and the Social Sciences and Humanities Research Council of Canada (Grant 410-92-1493).

LITERATURE CITED Adams M, and Reeve J (1987) Excavations at Christ

Church, Spitalfields 1984-6. Antiquity 61:247-256. Anderson DL, Thompson GW, and Popovich F (1976)

Age of attainment of mineralization stages of the per-manent dentition. J. For. Sci. 21:191-200.

Bellestedt I (1969) St. Thomas' Anglican Church 1818-1968, The Parish Story. Belleville, Ontario.

Borland International (1990) Paradox Relational Data-base Version 3.5, Borland International, Scotts Val-ley, CA.

Bowman JE, Maclaughlin SM, and Scheuer JL (1990) The relationship between biological and chronological age in the juvenile remains from St. Brides's Church, Fleet Street. Poster paper presented at the 8th An-nual European Meeting of the Paleopahtology Assoca-tion, Cambridge, U.K.

Boyce GE (1967) Historic Hastings. Belleville, Ontario: Belleville Hastings County Council.

Boyce GE (1991) Belleville's ethnic mosaic. Paper pre-sented to the Ethnic Festival Committee: President's Dinner, January 23,1991. Belleville, Ontario.

Canada (Province) Board of Registration and Statistics (1853) Census of the Canadas 1851-52. Presented to both houses of Parliament by command of His Excel-lency. Quebec: John Lovell, Volume 1.

Canada Bureau of Agriculture and Statistics (1878) Census of Canada 1870-71. Volume 1.

Cope DA, and Lacy MG (1992) Falsification of a Single Species Hypothesis Using the Coefficient of Variation: A Simulation Approach. Am. J. Phys. Antropol. 89:359-378.

Crossner CG, and Mansfield L (1983) Determination of dental age in adopted non-European children. Swed. Dent. J. 7:1-10.

Cunha E (1993) Testing Identification Records: Evi-dence from the Coimbra Identified Skeletal Collec-tions (19th-20th centuries). Paper presented to the American Association of Physical Anthropology Meet-ings, Toronto, Canada.

Demirjian A (1986) Dentition. In F Falkner and JM Tan-ner (eds.): Human Growth, A Comprehensive Trea-tise, 2nd ed. New York: Plenum, pp. 269-298.

Demirjian A, and Goldstein H (1976) New systems for dental maturity based on seven and four teeth. Ann. Hum. Biol. 3:411-421.

Demirjian A, Goldstein H, and Tanner JM (1973) A new system of dental age assessment. Hum. Biol. 45:211-227.

Demirjian A, and Lamarche C (1979) Dental develop-ment of British and French-Canadian children. In: Proceedings of the First International Congress of Auxology, 12-16, April 1977, Centro Auxologico Ital-iano, Rome. p. 227.

Demirjian A, and Levesque GY (1980) Sexual differ-ences in dental development and prediction of emer-gence. J. Dent. Res. 59:1110-1122.

Deutsch D, Tam 0, and Stack MV (1985) Postnatal changes in size, morphology and weight of developing postnatal deciduous anterior teeth. Growth 49:202-217.

Drake M (1974) Historical Demography: Problems and Prospects. Milton Keynes, Eng: The Open University Press.

El-Nofely A, and Iscan MY (1989) Assessment of age from the dentition in children. In MY Iscan (ed.): Age Markers in the Human Skeleton. Springfield, Ill.: Charles C. Thomas, pp. 237-254.

Garn SM, Lewis AB, and Blizzard RM (1965a) Endo-crine factors in dental development. J. Dent. Res. 44:243-248.

Garn SM, Lewis AB, and Kerewsky S (1965b) Genetic, nutritional, and maturational correlates of dental de-velopment. J. Dent. Res. 44:228-242.

Green LJ (1961) The interrelationships among height, weight, and chronological, dental and skeletal ages. Anlge Orthod. 31:189-193.

Gran AM (1962) Prediction of tooth emergence. J. Dent. Res. 41:573-585.

Gustafson G, and Koch G (1974) Age estimation up to 16 years of age based on dental development. Odont. Rev. 25:297-306.


187

Gyulavari 0 (1966) Dental and skeletal development of children with low birth weight. Acta Paeidiatr. Acad. Sci. Hung. 7:301-310.

Haag U, and Matsson L (1985) Dental maturity as an indicator of chronological age: The accuracy and preci-sion of three methods. Eur. J. Orthod. 7:25-34.

Haavikko K (1970) The formation and the alveolar and clinical eruption of the permanent teeth. Proc. Finn. Dent. Soc. 66:101-170.

Herring DA, Saunders SR, and Boyce GE (1992) Bones and burial registers: Reconstructing a 19th century pioneer community from skeletal evidence and ceme-tery records. Northeast Historical Archaeology (in press).

Holtz R (1959) The relation of dental calcification to chronological and skeletal age. Eur. Orthod. Soc. 1959:140-149.

Hoppa RD (1992) Evaluating human skeletal growth: An Anglo-Saxon example. Int. J. Osteoarch. 2: 275-288.

Hummert JF (1983a) Childhood Growth and Morbidity in a Medieval Population from Kulubnarti in the "Batn El Hajar" of Sudanese Nubia. Ph.D. Disserta-tin, University of Colorado at Boulder.

Hummert JF, and Van Gerven DP (1983a) Skeletal growth in a medieval population from Sudanese Nu-bia. Am. J. Phys. Anthropol. 60:471-478.

Hummert JF (1983b) Cortical bone growth and dietary stress among subadults from Nubia's Batn el Hajar. Am. J. Phys. Anthropol. 62:167-176.

Jantz RL, and Owsley DW (1984a) Temporal changes in limb proportionality among skeletal samples of Arikara Indians. Ann. Hum. Biol. / /:157-164.

Jantz RL, and Owsley DW (1984b) Long bone growth variation among Arikara skeletal populations. Am. J. Phys. Anthropol. 63:13-20.

Kataja M, Nystrom M, and Aine L (1989) Dental matu-rity standards in southern Finland. Proc. Finn. Dent. Soc. 85:187-197.

Kraus BS, and Jordan RE (1965) The Human Dentition Before Birth. Philadelphia: Lea and Febiger.

Kuhns LR, Sherman MP, and Poznanski AK (1972) De-termination of neonatal maturation on the chest ra-diograph. Radiology /02:597-603.

Lacey ICA, Parkin JM, and Steel GH (1973) Relationship between bone age and dental development. Lancet 2:736-737.

Lallo J (1973) The Skeletal Biology of Three Prehistoric American Indian Societies from Dickson Mounds. Ph.D. Dissertation, Department of Anthropology, University of Massachusetts, Amherst.

Lewis AB, and Garn SM (1960) The relationship be-tween tooth formation and other maturational fac-tors. Angle Orthod. 30:70-77.

Liliequist B, and Lundberg M (1971) Skeletal and tooth development. Acta Radiol. / / :97-112.

Lovejoy CO, Meindl RS, Pryzbeck TR,and Mensforth RP (1985) Multifactorial determination of skeletal age at death: A method and blind tests of its accuracy. Am. J. Phys. Anthropol. 68:1-14.

Lovejoy CO, Russell KF, and Harrison ML (1990) Long bone growth velocity in the libben population. Am. J. Human Biol. 2:533-541.

Lunt RC, and Law DB (1974) A review of the chronology of calcification of deciduous teeth. J. Am. Dent. Assoc. 89:599-606.

McKillop H, Marshall S, Boyce GE, and Saunders SR (1989) Excavations at St. Thomas' Church, Belleville, Ontario: A nineteenth century cemetery. Paper pre-sented to the Ontario Archaeology Society Sympo-sium, London, Ontario.

Merchant VL (1973) A Cross-sectional Growth Study of the Protohistoric Arikara from Skeletal Material As-sociated with the Mobridge Site (39WW1), South Da-kota. M.A. thesis, Department of Anthropology, Amer-ican University,

Merchant VL, and Ubelaker DH (1977) Skeletal growth of the protohistoric Arikara. Am. J. Phys. Anthropol. 46:61-72.

Mensforth RP (1985) Relative tibia long bone growth in the Libben and Bt-5 prehistoric skeletal populations. Am. J. Phys. Anthropol. 68:247-262.

Mika N, and Mika H (1986) Belleville, the Seat of Hast-ings County. Belleville: Mika Publishing Co.

Moodie S (1853) Life in the Clearings Versus the Bush. New Canadian Library Edition 1989. Toronto: Mc-Clelland and Stewart.

Moorrees CFA, and Kent RL (1981) Interrelations in the timing of root formation and tooth emergence. Proc. Finn. Dent. Soc. 77:113-117.

Moorrees CFA, Fanning EA, and Hunt EE (1963a) Age variation of formation stages for ten permanent teeth. J. Dent. Res. 42:1490-1502.

Moorrees CFA, Fanning EA, and Hunt EE (1963b) For-mation and resorption of three deciduous,teeth in chil-dren. Am. J. Phys. Anthropol. 21:205-213.

Niswander JD, and Sujaku C (1965) Permanent tooth eruption in children with major physical defect and disease. J. Dent. Child 32:266-268.

Nystrom M, Haataja J, Kataja M, Evalahti E, Peck L, and Kleemola-Kujala E (1986) Dental maturity in Finnish children, estimated from the development of seven permanent mandibular teeth. Acta Odontol. Scan. 44:193-198.

Owsley DW, and Jantz RL (1985) Formation of the per-manent dentition in Arikara Indians: Timing differ-ences that affect dental age assessments. Am. J. Phys. Anthropol. 61:467-471.

Plavcan M (1989) The coefficient of variation as an indi-cator of interspecific variability in fossil assemblages. Am. J. Phys. Anthropol. 78:285.

Prahl-Andersen B, Kowalski CJ, and Heyendael PHJM (1979) A Mixed Longitudinal Interdisciplinary Study of Growth and Development. New York: Academic Press.

Prahl-Andersen B, and Roede MJ (1979) The measure-ment of skeletal and dental maturity. In B Prahl-Andersen, CJ Kowalski, and PHJM Heyendael (eds.): A Mixed Longitudinal Interdisciplinary Study of Growth and Development. New York: Academic Press, pp. 491-519.

Prahl-Andersen B, and Van der Linden FPGM (1972) The estimation of dental age. Eur. Orthod. Soc. 1972:535-541.

Proy E, Sempe M, and Ajacques JC (1981) Etude corn-paree des maturations dentaire et squelettique chez


des enfants et adolescents frangais. Rev. Orthop. Dentofac. 15:3.

Reynolds HT (1977) The Analysis of Cross Classifica-tions. London: Collier MacMillan.

Rogers TL (1991) Sex Determination and Age Estima-tion: Skeletal Evidence from St. Thomas' Cemetery Belleville, Ontario. M.A. Thesis, Department of An-thropology, McMaster University, Hamilton, Ontario.

Saunders SR (1992) Subadult skeletons and growth re-lated studies. In SR Saunders and MA Katzenberg (eds.): The Skeletal Biology of Past Peoples: Advances in Research Methods. New York: Wiley-Liss Inc., pp. 1-20.

Saunders SR, Herring DA, and Boyce GE (1991) Testing theory and method in paleodemography: the St. Tho-mas' Anglican Church Cemetery. Paper presented to the Canadian Archaeological Association Meetings, St. John's, Nfld.

Saunders SR, Fitzgerald C, Rogers TL, Dudar C, and McKillop H (1992a) A test of several methods of skele-tal age estimation using a documented archaeological sample. Can. Soc. Forens. Sci. J. 25:97-118.

Saunders SR, DeVito C, Herring DA, and Southern R (1992b) An examination of subadult skeletal age esti-mation methods. Paper presented to the American Academy of Forensic Sciences Meetings, New Or-leans, La.

Saunders SR, and Melbye FJ (1990) Subadult mortality and skeletal indicators of health inLate Woodland On-tario Iroquois. Can. J. Arch. 14:61-74.

Skinner M, and Goodman AH (1992) Anthropological uses of developmental defects of enamel. In SR Saun-

ders and MA Katzenberg (eds.): Skeletal Biology of Past Peoples: Research Methods. New York: Wiley-Liss, pp. 153-174.

Smith BH (1991a) Dental development and the evolu-tion of life history in Hominidae. Am. J. Phys. Anthro-pol. 86:157-174.

Smith BH (1991b) Standards of human tooth formation and dental age assessment. In MA Kelly and CS Larsen (eds.): Advances in Dental Anthropology. New York: Wiley-Liss, Inc., pp. 143-168.

Steel GH (1965) The relation between dental maturation and physiological maturity. Dent. Pract. / 6:23-34.

Sundick RI (1978) Human skeletal growth and age de-termination. Homo 29:228-249.

Trodden BJ (1982) A Radiographic Study of the Calcifi-cation and Eruption of the Permanent Teeth in Inuit and Indian Children. Ottawa: National Museum of Man, Mercury Series Paper No. 112.

Ubelaker DH (1987) Estimating age at death from im-mature human skeletons: An overview. J. For. Sci. 32:1254-1263.

Ubelaker DH (1989) The estimation of age at death from immature human bone. In MY I§can (ed.): Age Mark-ers in the Human Skeleton, Springfield, IL: Charles C Thomas, pp. 55-70.

Walker PL (1969) The linear growth of long bones in Late Woodland Indian children. Proc. Ind. Aca. Sci. 78:83-87.

Wall CE (1991) Evidence of weaning stress and catch-up growth in the long bones of a central California Amer-indian sample. Ann. Hum. Biol. /8:9-22.