AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 144:625632 (2011)

Body Mass and Stature Estimation Based on the First Metatarsal in Humans
Isabelle De Groote1,2* and Louise T. Humphrey1
1 2

The Natural History Museum, Palaeontology Department, London, SW7 5BD, United Kingdom University College London, Department of Anthropology, Gower Street, London, WC1E 6BT, United Kingdom KEY WORDS body mass; stature; metatarsal; human variation; regression urements show a small increase in accuracy. Direct estimations of body mass (calculated from measured femoral head diameter using previously published equations) have an error of just over 7%. No direct stature estimation equations were derived due to the varied linear body proportions represented in the sample. The equations were tested on a sample of 35 individuals from Christ Church Spitalelds. Percentage differences in estimated and measured femoral head diameter and length were less than 1%. This study demonstrates that it is feasible to use the rst metatarsal in the estimation of body mass and stature. The equations presented here are particularly useful for assemblages where the long bones are either missing or fragmented, and enable estimation of these fundamental population parameters in poorly preserved assemblages. Am J Phys Anthropol 144:625632, 2011. V 2011 Wiley-Liss, Inc.
C

ABSTRACT Archaeological assemblages often lack the complete long bones needed to estimate stature and body mass. The most accurate estimates of body mass and stature are produced using femoral head diameter and femur length. Foot bones including the rst metatarsal preserve relatively well in a range of archaeological contexts. In this article we present regression equations using the rst metatarsal to estimate femoral head diameter, femoral length, and body mass in a diverse human sample. The skeletal sample comprised 87 individuals (Andamanese, Australasians, Africans, Native Americans, and British). Results show that all rst metatarsal measurements correlate moderately to highly (r 5 0.620.91) with femoral head diameter and length. The proximal articular dorsoplantar diameter is the best single measurement to predict both femoral dimensions. Percent standard errors of the estimate are below 5%. Equations using two metatarsal meas-

The estimation of body mass and stature are fundamental steps in the evaluation of skeletal remains in palaeoanthropology and osteoarchaeology and important components of individual identication in forensic cases. Body mass and stature are usually estimated mathematically using independent variables, such as long bone lengths, in equations that reect the linear relationship between the skeletal variable and body mass or stature (Feldesman and Lundy, 1988; Henry, 1992; Holliday and Ruff, 1997; Ruff et al., 1997; Auerbach and Ruff, 2004; Ruff et al., 2005; Sciulli and Blatt, 2008 and others). The accuracy of the estimation is reected in the strength of the correlation between the independent and dependent variables and the standard error of the estimate and varies according to which part of the skeleton is represented and the completeness of the bones. Intact long bones are not always available for measurement in forensic and archaeological contexts, where bones are often found incomplete or in poor condition. In particular, the articular surface of long-bones may be damaged or incomplete such that long bone length or articular dimensions cannot be determined. At some archaeological sites, small compact bones such as those of the hands and feet may be found in a more complete condition than larger skeletal elements. At Catalhoyuk (Konia, Turkey, 7000 BC), where most long bones were severely fractured due to trampling of the sediment overlying the burials, the hand and foot remains were relatively well represented and complete (Bello, 2006). At many sites, however, the hand and foot bones are underrepresented in skeletal assemblages due to taphonomic processes, burial practice, and differential archaeological
C V 2011

recovery (Bello et al., 2003; Bello et al., 2006). Bell and Cox (1999) show a recovery frequency of 40% of metatarsals in a Roman, early-medieval and modern forensic assemblage. Reasons for low rate frequency of certain skeletal elements need to be considered on a site-by-site basis. Depositional disturbance such as carnivore activity, root damage, and erosion are more likely to cause a physical relocation of the small bones, such as those in the foot, than larger bones (Andrews, 1990; Waldron, 2008). In cases of secondary burial the smaller less obvious bones are more likely to be left behind when the more obvious skeletal elements are gathered up for reburial. Where a burial is disturbed or truncated due to intercutting graves or other factors the foot bones may be at greater risk of disturbance since they are often located at the periphery of the grave (Mays, 1998; Waldron, 2008). When they are present though, foot bones are generally well preserved (Guthrie, 1967; Deeur et al., 1993; Bello et al., 2002). Of the metatarsal

Grant sponsor: The Leverhulme Trust. *Correspondence to: Isabelle De Groote, Department of Palaeontology, The Natural History Museum, London, SW7 5BD, United Kingdom. E-mail: i.degroote@nhm.ac.uk Received 30 September 2010; accepted 27 October 2010 DOI 10.1002/ajpa.21458 Published online 13 January 2011 in Wiley Online Library (wileyonlinelibrary.com).

WILEY-LISS, INC.

626

I. DE GROOTE AND L.T. HUMPHREY used to predict stature using linear regression. The associated errors of estimates were higher than those for regressions using complete long bones (Byers et al., 1989) but similar to or lower than those for regressions using fragmentary long bones (Byers et al., 1989; Wilbur, 1998). More recently, predictive equations have been developed for South Africans using skeletons from the Raymond A. Dart Collection (Bidmos, 2008). In this study the standard error of estimate for the metatarsal equations was lower than that obtained for equations using measures of the skull or long bone fragments for stature estimation but higher than for equations using intact long bones. Regression equations for estimating stature from the lengths of the rst and second metatarsals have also been determined for a Portuguese population using bones from documented cadavers of known stature (Cordeiro et al., 2009). In this article we explore the use of the rst metatarsal for body mass and stature estimation based on a sample of individuals with diverse body shape and population origins. To do this, we evaluate the use of metatarsal dimensions as a proxy for the femoral dimensions that are typically used for body mass and stature estimation since body mass and stature are unavailable for this broader series of populations. Specically, we explore whether metatarsal dimensions can be used to estimate femoral head diameter, which accurately predicts body mass (Ruff et al., 1991; Auerbach and Ruff, 2004) and femoral length, which is the most accurate long bone measurement for stature estimation (Steele, 1970). Estimated femoral length can then be used in population-specic regression formulae for stature estimation. This indirect method for stature estimation is necessary since population vary substantially in body shape and proportions (Ruff, 1994) whereas the equation for femoral head diameter could be converted to enable the direct estimation of body mass.

bones, the rst metatarsal tends to be the best preserved (Bello and Andrews, 2006; Waldron, 2008). Although the geometric mean of metatarsal measurements has been used as a proxy for body size (Grifn et al., 2008), body mass estimation using rst metatarsal measurements has, as far as we know, not been reported in the literature. Body mass is important from bioarchaeological and palaeoanthropological perspectives as it facilitates overall size comparison of different populations and enables measures of post-cranial strength to be standardized for body mass (Ruff et al., 1993; Pearson, 2000; Ruff and Trinkaus, 2000; Stock, 2002; Stock and Pfeiffer, 2004; Stock and Shaw, 2007). Mechanical methods of body mass estimation depend on the functional association between a weight bearing skeletal element and body mass (Auerbach and Ruff, 2004). Adult body mass is commonly estimated from femoral head diameter, which is highly correlated with body mass (McHenry, 1991; Ruff et al., 1991; Grine et al., 1995; Auerbach and Ruff, 2004). Adult size of the femoral head is established by the age at which the femoral head fuses with the metaphysis of the proximal femur and is not subject to remodeling (Ruff, 1988; Ruff et al., 1991). The size of other load bearing articulations is also expected to scale with body mass. Of the portions of the foot that have contact with the ground, approximately 50% of the load is borne by the heel and 50% is transmitted to the metatarsal heads (Nordin and Frankel, 1989). The rst metatarsal carries twice as much load as the individual load of the other metatarsals during normal stance. During the latter stance phase of the gait cycle, the load is transferred from the heel to the second metatarsal and then to the rst metatarsal during toe-off (Nordin and Frankel, 1989; Vereecke, 2003). Therefore it is predicted that the articular joint size of the rst metatarsal will also reect body size. Stature estimation is important for both bioarchaeology and in forensic cases. There has been more interest in the estimation of stature than body mass in forensic application as body mass can vary substantially throughout adult life whereas stature shows only a small decline with age (Ruff et al., 1991). The most accurate estimations of stature are derived from the length of intact long bones, in particular femoral and tibial length, and are available for a range of populations (Trotter and Gleser, 1952, 1958; Steele, 1970; Trotter and Gleser, 1977; Sciulli et al., 1990; Radoinova et al., 2002; Duyar and Pelin, 2003; Brown et al., 2004; Bidmos and Asala, 2005; Dayal et al., 2008; Raxter et al., 2008; Auerbach and Ruff, 2010 among others). In recognition of the fact that intact long bones are not always present or well preserved, predictive equations have been developed using partial long bones (Steele, 1970; Bidmos, 2008a; Bidmos, 2008c) and other measurement of the skull and the post-cranial skeleton (Byers et al., 1989; Meadows and Jantz, 1992; Holland, 1995; Bidmos and Asala, 2005; Bidmos, 2006; Ryan and Bidmos, 2007; Cordeiro et al., 2009). Measurements of the postcranial skeleton generally yield more accurate estimates of stature than measurements of the skull (Ryan and Bidmos, 2007). Several studies have derived predictive equations for the estimation of stature from metatarsal measurements based on samples of documented stature. Byers (1989) found signicant correlations between metatarsal lengths and stature for European and African Americans from the Terry Collection and Maxwell Museum of Anthropology and showed that metatarsal length can be American Journal of Physical Anthropology

MATERIALS AND METHODS

Metric data were collected on the femur and rst metatarsal of 87 skeletons representing a wide range of body sizes. The sample consists of 33 individuals from Abingdon (UK), 12 individuals from Newark Bay (UK), 9 individuals of African origin, 21 Andamanese, 7 Australasians, and 5 Native Americans from Santa Cruz (USA). An additional 35 individuals from Christ Church Spitalelds (London, UK) were measured and used to test the regression equations. All skeletons are held at the Natural History Museum, London, UK. Femoral length was measured as maximum length of the femur (Martin n81). Femoral head diameter refers to the superoinferior diameter of the femoral head. The rst metatarsal was oriented with the dorsal diaphyseal surface horizontal for all measurements (see Fig. 1). All measurements were taken with sliding calipers. Maximum length (MTL) was measured from the proximal articular surface of the head, to the tip of the styloid process. The size of the proximal articular facet was measured to obtain the greatest dorsoplantar diameter (DPP) with the arms of the calipers oriented parallel to the diaphysis. The mediolateral diameter (MLP) of the proximal articular facet was measured from the most medial point on the medial side to the most medial point on the concavity of the lateral side of the articular facet. The size of the rst metatarsal head (distal articular facet) was measured as the dorsoplantar height (DPD) and the

FIRST METATARSAL, BODY SIZE, AND STATURE

627

Fig. 1. Measurements taken on the rst metatarsal. MTL, metatarsal length; DPP, dorsoplantar diameter of the proximal articulation; MLP, mediolateral diameter of the proximal articulation; DPD, dorsoplantar diameter of the distal articulation; MLd, mediolateral diameter of the distal articulation.

technique in this instance as the x-variable is used to predict values of y such that the relationship between the variables is asymmetric (Smith, 2009). It is unlikely that predicted values for a target sample of modern humans will fall outside the range of variation of the reference sample since this sample encompasses both large and small-bodied individuals (Ruff, 2007). The standard error of estimate (SEE) is the usual measure of the expected accuracy of a regression equation in the estimation of a particular variable for an individual from the same population group from which the equation was originally derived. The percent standard error of estimate (%SEE) is the standard error of estimate (SEE) divided by the mean of the dependent variable. The advantage of the %SEE over the SEE is that it can be compared over different dimensional units, as are used here, and it allows for comparison across studies (Ruff, 2007). The Spitalelds sample was used to test the regression equations. The distributions of estimated femoral head diameter and estimated femoral length derived from the regression equations were compared to the actual distributions of femoral head diameter and femoral length respectively. A Students t-test was used to determine any signicant differences.

mediolateral width (MLD), measured on the plantar side of the head. All rst metatarsals are right bones. For the femur, the best preserved bone was measured. Analyses were carried out on a combined sex sample. Intraobserver error was evaluated by re-measuring three individuals over a period of two weeks and calculating the percentage difference between trials for each measurement (White and Folkens, 1999). The mean percentage error between repeats of the same measurement was 0.96%. Interobserver error was evaluated using measurements of three individuals taken by another observer. The mean percentage difference between observers was 2.69%. All statistical analyses were carried out using SPSS v.15. Pearsons Product Moment Correlations were determined between all metatarsal and femoral variables. The rst metatarsal measurement with the highest correlation with femoral head diameter was used to create a predictive equation for femoral head diameter. Additional equations with lower correlations are also included in the results. A second equation to infer body mass directly from the metatarsal measurements was derived using estimates of body mass determined from femoral head diameter. Body mass was calculated as the average of estimates based on three separate pooled-sex regressions (Ruff et al., 1991; McHenry, 1992; Grine et al., 1995) since the sample consisted of individuals of both large and small body size. An additional regression formula was calculated to predict femur length from the rst metatarsal measurements, which can then be used to estimate stature. Because there are marked differences in the relationship between femur length and stature between populations (Trotter and Gleser, 1952; Sciulli et al., 1990; Radoinova et al., 2002; Duyar and Pelin, 2003; Ruff et al., 2005; Dayal et al., 2008) we did not develop a further equation to estimate stature directly from metatarsal measurements. Estimation equations were generated using ordinary least squares regression. This is the most appropriate

RESULTS
Table 1 represents the descriptive statistics of the femur and rst metatarsal by population. As might be expected, the data show the variability of our samples reecting differences in stature and overall size in these populations. The Africans have the longest femora but are more gracile in all other dimensions than Abingdon, the second tallest population. The Abingdon sample shows the highest average femoral head diameter. The Andamanese group has the smallest measurements out of the entire sample. Correlations between the variables are presented in Table 2 and indicate a signicant correlation between all femoral and rst metatarsal measurements. Proximal dorsoplantar diameter is most highly correlated with femoral head diameter (r 5 0.911) (see Fig. 2) followed by the measures of metatarsal head size (DPP (r 5 0.832) and MLD (r 5 0.831). Dorsoplanar diameter of the proximal articulation is also highly correlated with femoral length (r 5 0.786) with a correlation that is slightly higher than that between femur length and rst metatarsal length (r 5 0.770) (see Fig. 3). Linear regression formulae in the form of y 5 c1mc 6 standard error of estimate, for the estimation of femoral head diameter, body size and femoral length are presented in Table 3. The most accurate regression equation is indicated by the lowest % standard error of estimate. The equations for both the femoral head and femoral length using metatarsal dorsoplantar diameter have a %SEE below 5%. Additional equations based on single metatarsal articular dimensions have a %SEE between 6% and 7%. The equations for estimating body mass directly from metatarsal measurements have a higher %SEE of 7.01% using two measurements and 7.19% using just one. The equation for estimation of femoral length from metatarsal length only had a %SEE of 5.05%. The population-specic equation for femoral length derived from the UK sample has a %SEE lower than 4%. American Journal of Physical Anthropology

628

I. DE GROOTE AND L.T. HUMPHREY

TABLE 1. Descriptive statistics for the study sample Population Abingdon Mean S.D. African Mean S.D. Andamanese Mean S.D. Australasian Mean S.D. Santa Cruz Mean S.D. Newark Bay Mean S.D. Spitalelds Mean S.D.

MT length 60.70 3.86 60.93 3.57 54.84 3.48 60.08 1.87 55.43 4.94 59.87 3.96 58.58 4.67 MT ML 50% 13.05 1.34 12.57 1.18 11.14 1.08 12.21 1.14 11.75 2.06 13.27 1.26 11.72 1.63 MT DP 50% 13.49 1.66 12.54 1.13 10.70 0.97 12.54 1.09 11.97 1.37 13.60 1.19 12.41 1.17 MT Prox ML diam 20.14 1.96 18.21 1.95 16.45 1.82 17.71 0.71 17.06 2.10 19.78 1.60 19.04 2.25 MT Prox DP diam 28.57 2.14 27.06 2.39 24.19 1.34 26.84 0.95 26.07 2.88 28.17 2.00 27.80 2.38 MT Dist ML diam 21.78 1.79 20.21 2.07 19.05 1.81 20.42 1.40 19.55 3.02 22.12 1.97 20.94 2.43 MT Dist DP diam 20.66 1.99 18.03 2.04 16.49 1.82 19.62 1.65 19.51 2.58 20.48 1.63 19.56 1.93 Femur length 433 24 439 23 381 20 428 37 421 32 427 29 425 35 BodyMassRuff91 67.34 6.83 58.79 5.13 49.92 5.19 60.22 5.06 59.31 8.79 64.72 6.95 62.39 7.46 BodyMassMcHenry92 63.37 7.86 53.51 5.90 43.31 5.98 55.16 5.83 54.12 10.12 60.35 8.00 57.66 8.60 BodyMassGrine95 68.11 7.97 58.12 5.98 47.79 6.06 59.80 5.90 58.74 10.25 65.05 8.11 62.32 8.71 Body Mass Average 66.27 7.55 56.81 5.67 47.01 5.75 58.39 5.60 57.39 9.72 63.37 7.69 60.79 8.26 All measurements are in mm. MT 5 1st metatarsal.

TABLE 2. Correlations for rst metatarsal and femoral measurements N 5 87 MT length (MTL) MT ML 50% MT DP 50% MT Prox ML diam (MLP) MT Prox DP diam (DPP) MT Dist ML diam (MLD) MT Dist DP diam (DPD) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Femur length 0.770(**) 0.000 0.702(**) 0.000 0.708(**) 0.000 0.624(**) 0.000 0.786(**) 0.000 0.666(**) 0.000 0.688(**) 0.000 Femur Head diameter 0.743(**) 0.000 0.743(**) 0.000 0.764(**) 0.000 0.801(**) 0.000 0.911(**) 0.000 0.831(**) 0.000 0.832(**) 0.000

** Correlation is signicant at the 0.01 level (2-tailed).

Equations were tested on an unsexed sample of 35 individuals from Christ Church, Spitalelds (Table 4). The Spitalelds sample is of average size (mean femoral length 5 425 mm) but has relatively high variability (S.D. 5 35 mm) and falls well within the size range of the sample used to develop the regression formulae (Table 1). Femoral head diameter was estimated using the proximal dorsoplantar diameter, a combination of the proximal dorsoplantar diameter and the distal mediolateral diameter. Additional regressions for single measurements are also reported (Table 3). The difference between the estimated mean femoral head diameter for the Spitalelds sample and the actual sample mean was 0.37 mm using proximal dorsoplantar diameter (equation a) and 0.23 mm using the combination of proximal dorsoplantar diameter and distal mediolateral diameter (equation b) resulting in an improvement in percentage difference from 0.84% to 0.53% (Table 4). Body mass estimates showed a higher percentage error than estimates of femoral head diameter (Table 4). The difference between body mass estimations based on metatarsal dimensions and those calculated from the femoral head measurements was less than 1 kg for both regressions. The mean % difference was 1.33% for proximal dorsoplantar diameter only, and 0.75% for proximal dorsoplantar diameter and distal mediolateral diameter. Femur length was estimated using the equations based on all individuals in the sample and using the equations based on the British samples from Newark American Journal of Physical Anthropology

Bay and Abingdon only. The difference between the estimated mean femoral length for the Spitalelds sample and the actual sample mean was 1.4 mm using the equation derived from all individuals and 2.37 mm using the equation derived from other British samples (Table 4). Neither estimate was improved by the addition of the metatarsal length to the proximal dorsoplantar diameter measurement. Estimation error was less than 0.6% for all estimates and there were no signicant differences between estimated sample and actual sample means (paired t-test: all comparisons P [ 0.369).

DISCUSSION
Overall, the results for the use of the rst metatarsal as an estimate of body mass are particularly encouraging. The high correlation between the diameters of the femoral head and measurements of the proximal articulation of the rst metatarsal reects the weight bearing properties of these skeletal elements and is consistent with the expectation that weight bearing articular surfaces should have a functional association with body mass (Jungers, 1988; Ruff et al., 1991; Lieberman et al., 2001; Auerbach and Ruff, 2004). Epiphyseal union of the proximal articulations of the femur and rst metatarsal takes place at a similar stage of development suggesting similar developmental constraints relating to the attainment of adult size and function in these skeletal elements (Scheuer and Black, 2004). Equations for the estimation of femoral head

FIRST METATARSAL, BODY SIZE, AND STATURE

629

Fig. 2. Correlation between femoral head diameter and the dorsoplantar diameter of the rst metatarsal.

Fig. 3. Correlation between femur length and rst metatarsal length.

diameter based on the proximal dorsoplantar diameter or a combination of the proximal dorsoplantar diameter and the distal mediolateral diameter had a low percentage error of estimate. This result indicates that measurements of the proximal end of the rst metatarsal can be used either independently as a measure of body mass, or to estimate femoral head diameter when the femoral head is absent or poorly preserved. Other single measurements on the epiphyses of the rst metatarsal were able to estimate femoral head diameter with some accuracy and could be used to estimate femoral head diameter and hence body mass from partially preserved rst metatarsals.

The predictive value of the regression equation for proximal dorsoplantar diameter was tested on a population from Spitalelds that had not been included in the series used to develop the predictive equation. The percentage difference between the actual and predicted mean value for femoral head diameter was very low indicating that the regression equation is well suited to individuals of varied but average size. There was no obvious bias in the patterning of residuals indicating that the equation predicts femoral head diameter equally well for small and large individuals within the size range of the Spitalelds series. A further regression equation to predict body mass directly from the proximal dorsoplantar diameter of the rst metatarsal was derived based on mean estimated body mass estimates from three published equations using femoral head diameter (McHenry, 1991; Ruff et al., 1991; Grine et al., 1995). The regression formula for body mass should be used with a degree of caution as it has a relatively high estimation error due to the cumulative error results from the multiple steps needed to predict body mass. The exact body mass estimation errors are not calculated here because the equation is the average of the three published equations that are based on different samples and methods (Ruff, 2010). It is therefore recommended that, when the actual body mass of the individual is not a priority, dorsoplantar diameter of the proximal articulation is used as a proxy for body mass in analyses of the relationship between body mass and other factors. As part of this study the use of the rst metatarsal in the estimation of stature was also assessed. Previous work on stature estimation using intact long bones or other complete or partial skeletal elements has shown that the relationship between skeletal measurements and stature varies between populations and between males and females (Trotter and Gleser, 1952; Sciulli et al., 1990; Ruff, 1991; Holliday and Ruff, 1997; Radoinova et al., 2002; Ruff et al., 2005; Dayal et al., 2008; Raxter et al., 2008; Auerbach and Ruff, 2010). Where available, an appropriate population and sex-specic formula for the estimation of stature from a skeletal dimension should always be used. For this reason, we only present regression formulae for the estimation of femoral length. Previous research has reported a high correlation between metatarsal length and stature within specic population groups (Byers et al., 1989; Meadows and Jantz, 1992; Bidmos, 2008; Cordeiro et al., 2009). Contrary to expectations femur length was more highly correlated with the proximal dorsoplantar diameter of the rst metatarsal (r2 5 0.617) than with metatarsal length (r2 5 0.592). The combination of these two measurements gives the most accurate regression formula (r2 5 0.669). The %SEE is less than 5.05% for all regressions. All three regressions equations were demonstrated to yield very low estimation error for mean femoral length in our test population (less than 1.8%). The accuracy of the estimations was improved (estimation error less than 0.5%) using regressions based on a UK sample only, suggesting that population-specic regressions are most suitable for the estimation of femur length from metatarsal dimensions. The distribution of residuals revealed that femur length was overestimated for shorter individuals and underestimated for taller individuals from Spitalelds regardless of which predictive equations was used. This effect has been previously reported in other stature estiAmerican Journal of Physical Anthropology

630

I. DE GROOTE AND L.T. HUMPHREY

TABLE 3. Regressions for femoral head diameter, body size, and femur length SEE r 0.917 0.922 0.801 0.832 0.831 0.911 0.917 0.786 0.818 0.770 0.745 0.775 0.683 r2 0.841 0.850 0.637 0.692 0.686 0.830 0.840 0.617 0.669 0.592 0.555 0.601 0.467 P 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 %SEE* 4.60% 4.50% 6.70% 6.22% 6.23% 7.19% 7.01% 4.90% 4.58% 5.05% 3.94% 3.78% 4.32%

Femoral head diameter a) 1.652 3 Prox DP diameter 2 1.981 b) (1.349 3 Prox DP diameter) 1 (0.424 3 Dist ML Diameter) 2 2.578 c) 1.589 3 Prox ML diameter 1 13.117 d) 1.559 3 Dist DP diameter 1 12.812 e) 1.795 3 Dist ML diameter 1 5.411 Body mass a) 3.553 3 Prox DP diameter 2 37.167 b) (2.900 3 Prox DP dameter) 1 (0.911 3 Dist ML Diameter) 2 38.450 Femoral Length all individuals a) 9.870 3 Prox DP diameter 1 151.759 b) (5.926 3 Prox DP diameter) 1 (2.861 3 Length) 1 90.024 c) 5.683 3 MT length 1 84.298 Femoral Length UK only a) 8.642 3 Prox DP diameter 1 184.486 b) (6.089 3 Prox DP diameter) 1 (1.956 3 Length) 1 139.046 c) 4.356 3 MT length 1 167.807 * SEE/mean

1.955 1.912 2.867 2.660 2.667 4.251 4.144 20.520 19.201 21.176 17.012 16.304 18.621

TABLE 4. Estimations for the test sample from Spitalelds Directional difference* Actual mean Estimated mean Raw (mm) Femoral head diameter Reg. a) 43.57 43.94 Reg. b) 43.57 43.80 Reg. c) 43.57 43.37 Reg. d) 43.57 43.31 Reg. e) 43.57 43.01 Body mass (kg) Reg. a) 60.79 61.60 Reg. b) 60.79 61.24 Femoral Length all individuals Reg. a) 424.72 426.12 Reg. b) 424.72 422.35 Reg. c) 424.72 417.21 Femoral Length Europeans only Reg. a) 424.72 424.71 Reg. b) 424.72 422.89 Reg. c) 424.72 422.98 * None are signicant (t-test). 0.37 0.23 20.20 20.26 20.56 0.81 0.45 1.40 22.37 27.51 20.01 21.83 21.74 % 0.84% 0.53% 20.46% 20.60% 21.29% 1.33% 0.75% 0.33% 20.56% 21.77% 0.00% 20.43% 20.41% Fig. 4. Correlation between femur length and the dorsoplantar diameter of the rst metatarsal. Crosses and the solid line represent the regression sample (b 5 9.870; intercept 5 151.759; r2 5 0.617); circles and the dashed line are the Spitalelds test sample (b 5 10.039; intercept 5 145.659; r2 5 0.453).

mation studies and interpreted as an artifact of the regression analysis (Trotter and Gleser, 1952; Sjvold, 1990; Formicola and Franceschi, 1996; Duyar and Pelin, 2003). Applying a Reduced Major Axis regression instead did not alter the bias in the estimation, indicating that it is not an artifact of the regression equation. The bias reects a difference in the relationship between metatarsal dimensions and femur length between the Spitalelds sample and both the original series and other UK populations. A regression line tted to the Spitalelds series has a shallower slope and lower intercept at y 5 0 than regression lines for either reference sample (see Fig. 4). Since the regression line for Spitalelds intersects regression lines for the original and UK series differences in estimated femur length for individuals in the middle range are small, whereas those at the extremes of the range are under- or overestimated. Further research is required to understand the nature of variation in the relationship between femoral length and metatarsal dimensions among geographically diverse populations and between urban and rural populations from a restricted geographical region. American Journal of Physical Anthropology

CONCLUSION
This study is a preliminary evaluation of the use of the rst metatarsal as a proxy for body mass and stature in fragmentary and poorly preserved archaeological assemblages. The results presented demonstrate that epiphyseal dimensions of the rst metatarsal are highly correlated with femoral head diameter and hence body mass. The correlations vary across the different measurements but all single articular dimensions can be used in the estimation of femoral head diameter. This is consistent with the expectation that weight bearing articular surfaces should have a functional association with body mass. First, the rst metatarsal dimensions can be used independently as a measure of body mass in studies where it is necessary to correct for body mass or where the remains are comingled and the researcher wants to assess the range of body sizes represented in the sample. Second, the rst metatarsal dimensions can be used to estimate femoral head diameter to allow comparisons across a range of samples where the femoral

FIRST METATARSAL, BODY SIZE, AND STATURE head is missing for some individuals but where associated partial or complete metatarsals are present. Third, the regression equations for femoral head diameter presented here allow body mass (weight) to be estimated in a two stage process by estimating femoral head diameter from measurements of the rst metatarsal and using published equations for the estimation of body mass from femoral head diameter. In this situation it is important to keep in mind the cumulative error of this estimate when making comparisons across populations. Correlations between metatarsal size and femur length were lower than their equivalents for femoral head diameter for all metatarsal measurements apart from length. Previous studies have demonstrated the usefulness of metatarsal length for stature estimation but the results of this study indicate that proximal dorsoplantar diameter may ultimately be a better predictor of femoral length, and potentially living stature. Although this study has not tested the regression on actual living stature, the femoral length estimates produced by the equations presented in this paper can be used in published population-specic regressions to estimate living stature, but again, bearing in mind the accumulative error in the use of multiple estimations and the need for population-specic regressions. The regression equations presented here are particularly suited for archaeological and fossil assemblages where the long bones are either missing or too fragmentary to use for the estimation of body mass or stature. They enable femoral length and body mass to be estimated from both complete and incomplete rst metatarsals. The encouraging results of the analyses presented above warrant further investigation and the renement of these regression equations on a sample of known body mass and stature. Such a study would circumvent the problem of cumulative error and allow for a comparison of estimation error with other skeletal elements.

631

ACKNOWLEDGMENTS
The authors thank Robert Kruszynski at the Natural History Museum London for collections advice and Laura Buck for help with collecting data and carrying out the interobserver error tests.

LITERATURE CITED
Andrews P. 1990. Owls, caves and fossils: predation, preservation and accumulation of small mammal bones in caves, with an analysis of the Pleistocene cave faunas from WestburySub-Mendip, Somerset, U.K. Chicago: University Of Chicago Press. Auerbach BM, Ruff C. 2004. Human body mass estimation: a comparison of morphometric and mechanical methods. Am J Phys Anthropol 125:331342. Auerbach BM, Ruff CB. 2010. Stature estimation formulae for indigenous North American populations. Am J Phys Anthrop 141:190207. Bell L, Cox M. 1999. Recovery of human skeletal elements from a recent UK murder inquiry: preservational signatures. J Forensic Sci 44:945950. Bello S. 2006. Lutilite des collections osteologiques en taphono mie et anthropologie: la collection ideale nest pas necessaire ment la mieux conservee. In: Ardagna Y, Bizot B, Boetsch G, Delestre X, editors. Les collections osteologiques humaines: gestion, valorisation et perspectives. Supplement to Bulletin Archeologiques de Provence. p 145151.

Bello S, Andrews P. 2006. The intrinsic pattern of preservation of human skeletons and its inuence on the interpretation of funerary behaviours. In: Knusel C, Gowland R, editors. The social archaeology of funerary remains. Oxford: Oxbow Books. p 113. Bello S, Thomann A, Lalys LMS, Rabino-Massa E, Dutour O. 2003. Calcul du Prol theorique de survie osseuse le plus probable et son utilisation dans linterpretation des processus taphonomiques pouvant determiner la formation dun echan tillon osteologique humain. Br Archaeological Rep - Int Series 1145:2130. Bello S, Thomann A, Signoli M, Dutour O, Andrews P. 2006. Age and sex bias in the reconstruction of past population structures. Am J Phys Anthropol 129:2438. Bello S, Thomann A, Signoli M, Rabino-Massa E, Dutour O. 2002. La conservation differentielle des os humains et le Prol theorique de survie osseuse. Archeologie et Prehistoire 113:105120. Bidmos M. 2006. Adult stature reconstruction from the calcaneus of South Africans of European descent. J Clin Forensic Med 13:247252. Bidmos MA. 2008. Metatarsals in the estimation of stature in South Africans. J Forensics Legal Med 15:505509. Bidmos M, Asala S. 2005. Calcaneal measurement in estimation of stature of South African blacks. Am J Phys Anthropol 126:335342. Byers S, Akoshima K, Curran B. 1989. Determination of adult stature from metatarsal length. Am J Phys Anthropol 79:275 279. Cordeiro C, Munoz-Barus JI, Wasterlain S, Cunha E, Vieira DN. 2009. Predicting adult stature from metatarsal length in a Portuguese population. Forensic Sci Int 193(13):e1e4. Dayal MR, Steyn M, Kuykendall KL. 2008. Stature estimation from bones of South African whites. S Afr J Sci 104:124128. Deeur A, Dutour O, Valladas H, Vandermeersch B. 1993. Cannibals among the Neanderthals? Nature 362:214214. Duyar I, Pelin C. 2003. Body height estimation based on tibia length in different stature groups. Am J Phys Anthropol 122:2327. Feldesman MR, Lundy JK. 1988. Stature estimates for some African Plio-Pleistocene fossil hominids. J Hum Evol 17:583 596. Formicola V, Franceschi M. 1996. Regression equations for estimating stature from long bones of early Holocene European samples. Am J Phys Anthropol 100:8388. Grine FE, Jungers WL, Tobias PV Pearson OM. 1995. Fossil Homo femur from Berg Aukas. Northern Namibia. Am J Phys Anthropol 97:151185. Grifn NL, Gordon AD, Richmond BG, Anton SC. 2008. Crosssectional geometric analysis of a foot bone assemblage from Mangaia. Cook Islands. HOMO 59:2740. Guthrie RD. 1967. Differential preservation and recovery of Pleistocene large mammal remains in Alaska. J Paleontol 41:243246. Henry MM. 1992. Body size and proportions in early hominids. Am J Phys Anthropol 87:407431. Holland TD. 1995. Estimation of adult stature from the calcaneus and talus. Am J Phys Anthropol 96:315320. Holliday TW, Ruff CB. 1997. Ecogeographical patterning and stature prediction in fossil hominids: comment. Am J Phys Anthropol 103:137140. Jungers WL. 1988. Relative joint size and hominoid locomotor adaptations with implications for the evolution of hominid bipedalism. In: Strasser E, Dagosto M, editors. The primate postcranial skeleton: studies in adaptation and evolution. London: Academic Press. p 247265. Lieberman DE, Devlin MJ, Pearson OM. 2001. Articular area responses to biomechanical loading: effects of exercise, age and skeletal location. Am J Phys Anthropol 116:266277. Mays SA. 1998. The archaeology of human bones. New York: Routledge. McHenry HM. 1991. Petite bodies of the robust australopithecines. Am J Phys Anthropol 86:445454.

American Journal of Physical Anthropology

632

I. DE GROOTE AND L.T. HUMPHREY

Ryan I, Bidmos MA. 2007. Skeletal height reconstruction from measurements of the skull in indigenous South Africans. Forensic Sci Int 167:1621. Scheuer L, Black S. 2004. The juvenile skeleton. New York: Elsevier Academic Press. Sciulli PW, Blatt SH. 2008. Evaluation of juvenile stature and body mass prediction. Am J Phys Anthropol 136:387393. Sciulli PW, Schneider KN, Mahaney MC. 1990. Stature estimation in prehistoric Native Americans of Ohio. Am J Phys Anthropol 83:275280. Sjvold T. 1990. Estimation of stature from long bones utilizing the line of organic correlation. Hum Evol 5:431447. Smith RJ. 2009. Use and misuse of the reduced major axis for line-tting. Am J Phys Anthropol 140:476486. Steele DG. 1970. Estimates of stature from fragments of long bones. In: Stewart TD, editor. Personal identication in mass disasters. Washington, D.C.: National Museum of Natural History. p 8597. Stock JT. 2002. Climatic and behavioural inuences on postcranial robusticity among Holocene foragers. PhD dissertation. University of Toronto. Stock JT, Pfeiffer SK. 2004. Long bone robusticity and subsistence behaviour among Later Stone Age foragers of the forest and fynbos biomes of South Africa. J Archaeol Sci 31:999 1013. Stock JT, Shaw CN. 2007. Which measures of diaphyseal robusticity are robust? A comparison of external methods of quantifying the strength of long bone diaphyses to cross-sectional geometric properties. Am J Phys Anthropol 134:412423. Trotter M, Gleser GC. 1952. Estimation of stature from long bones of American Whites and Negroes. Am J Phys Anthropol 10:463514. Vereecke E, DAout K, De Clercq D, Van Elsacker L, Aerts P. 2003. Dynamic plantar pressure distribution during terrestrial locomotion of bonobos (Pan paniscus). Am J Phys Anthropol 120:373383. Waldron T. 2008. Palaeopathology. Cambridge: Cambridge University Press. White TD, Folkens PA. 1999. Human osteology. San Diego: Academic Press.

McHenry HM. 1992. Body size and proportions in early hominids. Am J Phys Anthropol 87:407431. Meadows L, Jantz RL. 1992. Estimation of stature from metacarpal lengths. J Forensic Sci 37:147154. Nordin M, Frankel VH. 1989. Basic biomechanics of the musculoskeletal system. Philadelphia: Lea and Febiger. Pearson OM. 2000. Activity, climate and postcranial robusticity: implications for modern human origins and scenarios of adaptive change. Curr Anthropol 41:569607. Radoinova D, Tenekedjiev K, Yordanov Y. 2002. Stature estimation from long bone lengths in Bulgarians. Homo 52:221232. Raxter MH, Ruff CB, Azab A, Erfan M, Soliman M, El-Sawaf A. 2008. Stature estimation in ancient Egyptians: a new technique based on anatomical reconstruction of stature. Am J Phys Anthropol 136:147155. Ruff C, Niskanen M, Junno JA, Jamison P. 2005. Body mass prediction from stature and bi-iliac breadth in two high latitude populations, with application to earlier higher latitude humans. J Hum Evol 48:381392. Ruff CB. 1988. Hindlimb articular surface allometry in Hominoidea and Macaca, with comparisons to diaphyseal scaling. J Hum Evol 17:687714. Ruff CB. 1991. Climate and body shape in hominid evolution. J Hum Evol 21:81105. Ruff CB. 1994. Morphological adaptation to climate in modern and fossil hominids. Am J Phys Anthropol 37(S19):65107. Ruff CB. 2007. Body size prediction from juvenile skeletal remains. Am J Phys Anthropol 133:698716. Ruff C. 2010. Body size and body shape in early hominins implications of the Gona Pelvis. J Hum Evol 58:166178. Ruff CB, Scott WW, Liu AYC. 1991. Articular and diaphyseal remodeling of the proximal femur with changes in body mass in adults. Am J Phys Anthropol 86:397413. Ruff CB, Trinkaus E. 2000. Lifeway changes as shown by postcranial skeletal robustness. Am J Phys Anthropol Suppl 30:266266. Ruff CB, Trinkaus E, Holliday TW. 1997. Body mass and encephalization in Pleistocene Homo. Nature 387:173176. Ruff CB, Trinkaus E, Walker A, Larsen CS. 1993. Postcranial robusticity in Homo I: temporal trends and mechanical interpretation. Am J Phys Anthropol 91:2153.

American Journal of Physical Anthropology