AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 92:395-399 (1993) Brief Communication: An Update on Stature Estimation in Prehistoric Native Americans of Ohio PAUL W. SCIULLI AND MYFL4 J. GIESEN Department ofdnthropology, The Ohio State University, Columbus, Ohio 43210 KEY WORDS Height, Regression equations, Archaic period, Native Americans ABSTRACT We present regression equations to estimate skeletal height and stature for prehistoric Native Americans of Ohio. The regression equations are based on skeletal height as the dependent variable and various postcranial elements and combinations of elements as the independent variables. A total of 171 individuals, 95 males and 76 females, make up the sample. The present sample includes the 64 individuals we previously used for stature estimation (Sciulli et al.: Am. J. Phys. Anthropol. 83:275-280, 1990) and 107 additional individuals distributed more widely in time and space. The present more inclusive sample, however, shows the same proportional contributions to skeletal height of each skeletal height component as the previous sample. This result suggests that these proportions were a consistent feature of the prehistoric Native Americans of Ohio. Because the prehistoric Native Americans of Ohio were characterized by relatively long legs and distal elements of the limbs, stature estimation from regressions based on East Asian populations, which express in general relatively short legs and distal limb elements, will overestimate stature in Native Americans of Ohio and, possibly, all Eastern Woodlands Native Americans. 0 1993 Wiley-Liss, Inc. In a previous report we presented regression equations for the estimation of stature in prehistoric Native Americans of Ohio (Sciulli et al., 1990). The regression equations in that report were based on skeletal height (Fully, 1956) as the dependent variable and lower limb bone and vertebral column lengths as the independent variables. That approach allowed us to eliminate the assumption that the prehistoric populations included in our sample expressed the same stature component proportions as in any living population(s). The use of skeletal height also provided us with the best estimates of stature for skeletonized individuals. Comparisons of the best estimates of stature (from skeletal height) in our sample with stature estimates from the regression equations we calculated from the sample showed that they were virtually identical. 0 1993 WILEY-LISS. INC However, stature estimates based on regression equations from distinct but genetically related East Asian populations consistently overestimated stature in our prehistoric sample. The present report is an extension of our previous work with a much larger sample from a wider geographic area. The larger sample allows us to evaluate our previous conclusions concerning the apparent homogeneity of stature component proportions in prehistoric Native Americans from the Ohio area and allows us to include additional measures from the upper limb bones for stature estimation. Received October 26,1992; accepted June 8,1993. Address reprint requests to Paul W. Sciulli, Department of Anthropology, The Ohio State University, Columbus, OH 43210. 396 P.W. SCIULLI AND M.J. GIESEN MATERIALS AND METHODS A total of 171 adult prehistoric Ohio Valley Native Americans comprise the present sample, 76 females and 95 males. The sample includes the 64 individuals from our previous report plus 107 individuals primarily from the Late Prehistoric period. The samples represent a time span of almost 3,000 years. The earliest samples date to the Late Archaic period (ca. 3150 BP.) while the most recent date to the Late Prehistoric-Protohistoric period (ca. 800-300 BP.). The distribution of females over time periods is as follows: Late Archaic, n = 15; Late Woodland, n = 2; Late Prehistoric, n = 59. The distribution of males over time periods is as follows: Late Archaic, n = 20; Middle Woodland, n = 1; Late Woodland, n = 5; Late Prehistoric, n = 69. In addition to an increased sample size we are also able to include individuals from a wider geographical area in this report. The 64 individuals in our previous study were taken from populations located primarily in northwestern and north central Ohio. Most of the 107 individuals added in this report are from populations located in west central Ohio (n = 381, southwest Ohio (n = 30), southwest Pennsylvania (n = 181, and north central West Virginia (n = 131, the remainder coming from northern Ohio. We estimated the sex and age of each individual as well as skeletal height using the same methods as in our previous study (Sciulli et al., 1990). With the present larger sample we are able to include regression equations using the lengths of the bones of the upper extremity: maximum length of humerus and radius and the length of the ulna excluding the styloid process. Upper and lower extremity measures are the average of the left and right side if both are available; otherwise, the single side available represents the length of the element. We were unable to measure the heights of 45 vertebrae in 32 individuals because the vertebrae were either missing or damaged. However, we did estimate the heights of these vertebrae by averaging the heights of vertebrae directly superior and inferior to them (Sciulli et al., 1990). These vertebral height estimates should be accurate because the heights of individual vertebra, as a per- centage of the total vertebral column height, are almost identical not only among prehistoric Native Americans but among African and European derived populations as well (Sciulli et al., 1990; Sciulli, in press). We used skeletal height as the dependent variable and various elements and combinations of elements as independent variables to develop regression equations for estimating stature (BMDPGD: Dixon, 1985). Inspection of scatterplots indicated strong linear relationships and thus we did not use transformations. We also calculated major axis equations (Sokal and Rohlf, 1981) for the same combinations of skeletal height and the element or combination of elements for which we present regression equations. The major axis equations are not presented here as their estimates of stature are essentially the same as those from the regression equations. However, the major axis equations are available on request for those who, on philosophical grounds, prefer them to regression equations in analyses of this kind. RESULTS Table 1 contains descriptive statistics for each component of skeletal height and the elements of the upper extremity, the percent of total skeletal height of each component and the elements of the upper extremity, and the average skeletal height for males and females in the Late Archaic, Late Prehistoric, and total sample. The Late Archaic and Late Prehistoric samples are listed separately because there is a significant difference in skeletal height (and stature) between males of these periods (Sciulli, in press), raising the possibility of differences in the contribution to skeletal height by the five components. Examination of the percent of total skeletal height contributed by each component in both sexes shows that while the absolute magnitude of skeletal height may differ significantly between the Late Archaic and Late Prehistoric male samples, the proportion contributed to skeletal height by each component is virtually identical. We evaluated this observation by testing for the homogeneity of covariance matrices of the five skeletal height components (cranium height through foot height) in the Late Archaic and Late Prehistoric males (Box, 1949). We are STATURE ESTIMATION IN OHIO NATIVE AMERICANS 397 TABLE 1 , Descriptiue statistics for stature components and arm elements of the prehistoric Ohio Natiue American sample .- Measure Males Cranium Vertebral column Femur Tibia Foot Humerus Radius Ulna Skeletal height Females Cranium Vertebral column Femur Tibia Foot Humerus Radius Ulna Skeletal height N Late Archaic Z S - N 69 69 69 69 69 66 55 50 69 14.3 50.7 45.2 37.8 6.1 32.6 25.5 27.2 154.1 0.52 2.32 2.11 1.87 0.37 1.61 58 58 58 58 58 51 46 42 58 13.7 48.4 42.4 35.0 5.6 30.6 23.5 25.1 145.1 0.48 2.24 1.60 1.50 0.32 1.26 1.03 1.02 4.74 20 20 20 20 20 20 20 19 20 14.5 51.0 46.8 39.8 6.0 33.9 26.6 28.3 158.3 0.31 3.33 2.22 1.85 0.42 1.19 0.97 0.94 6.23 9.2 32.2 29.6 25.2 3.8 21.4 16.8 17.9 15 15 13.8 47.5 43.8 36.0 5.4 31.6 24.3 26.4 146.5 0.48 2.84 2.09 1.52 0.37 1.84 1.58 1.28 5.93 9.4 32.4 29.9 24.6 3.7 21.5 16.4 17.9 15 15 15 12 12 10 15 Late Prehistoric ? S % - - 1.11 1.07 5.80 . % 9.3 32.9 29.3 24.5 4.0 21.2 16.5 17.6 - 9.4 33.4 29.2 24.1 3.8 21.0 16.2 17.3 - -~ N 95 95 95 95 95 92 Total sample2 Z S % 75 95 14.4 50.9 45.6 38.3 6.1 33.0 25.8 27.5 155.3 0.48 2.55 2.22 2.06 0.39 1.89 1.17 1.14 6.12 9.3 32.8 29.4 24.7 3.9 21.2 16.6 17.6 76 76 76 76 76 65 61 54 76 13.7 48.2 42.7 35.2 5.5 30.8 23.6 25.4 145.4 0.48 2.35 1.77 1.48 0.35 1.42 1.16 1.15 4.92 9.4 33.2 29.4 24.2 3.8 21.1 16.2 17.4 80 - - ‘All measures In cm. The 92 column shows the average ratio of the measure to skeletal height. Bicondylar femur length, condyle-malleolus tibia length, ulna length minus styloid process. Homogeneity of Late Archaic and late Prehistoric Covariance matrices for stature coniponents (cranium through foot cannot be rejected for m a l e s ( M = 18.5,0.5>P>.IO)orfemalesLM= 17.7,0.5>P>.l01). unable to reject the hypothesis of homogene- combination with other elements yield the ity of the covariance matrices in the samples best estimates of skeletal height, while the (Table l),thus supporting the observations elements of the upper extremity alone or in on the similarity of the individual propor- various combinations yield useful but less tions. Females show no significant differ- accurate estimates. ence in skeletal height between groupings In preliminary trials we tested a number and, as with the males, show essentially the of combinations of the elements of the upsame proportion contributed to skeletal per extremity as independent variables height by each component. We also cannot to estimate skeletal height. However, all of reject the homogeneity of covariance matri- the upper extremity combinations-for exces of the skeletal height components of the ample, (2 x humerus length) + (2 x radius Late Archaic and Late Prehistoric females lengthkexcept humerus plus radius length (Table 1).As the stature component propor- in males yielded poorer estimates than the tions are identical in the male groupings and elements used singly. Thus the various comin the female groupings, we pooled all same binations of the upper extremity elements sexed individuals into total samples for re- are not listed. As we found in our previous study and as gression analyses. Table 2 contains the regression equations Tibbets (1981) notes, vertebral column we developed for estimating skeletal height height is a very poor estimator of skeletal in the present sample. As in our previous height. The only exception in the present report we have also included the maximum study is the combination of vertebral column length of the femur and tibia, as these are height with cranial height in females. To use the formulae in Table 2 to estimate commonly obtained measures. As we expected, because of the strong similarity in stature, insert the measure of the element or bodily proportions over time and space in the sum of the measures of elements in the Native Americans of the Ohio Valley area, appropriate formula. The result of the operthe regression equations in Table 2 are very ations indicated in the formula will yield similar to the corresponding equations of skeletal height. Stature is obtained by addour previous study. Again, as is generally ing the soft tissue corrections listed at the the case, the femur and tibia alone and in bottom of Table 2. If a living stature esti- P W SCIULLI AND M J GIESEN 398 TABLE 2 Regression formulae for estimatlon of Ziulng stature for prehistoric Ohio Native Americans ~ Formula' Elernent(s1 ~~ Males (BF + CMT + FO + LV) = LTLV (BF + CMT LV) = LPLV (XF + XT + LV) = LXLV (BF + CMT + FO) = LT (XF + X T ) = L X (BF + CMT) = LP + RF _. XF XT CMT (XH + X R ) = ARM XH XR su Females LTLV LPLV LXLV LT LX XT XH XR STJ ~~ ~ _ _ _ _ _ _SE _ _ _ _ _N_ - r SKHT = 25 421 + 1 2 5 4 (LTLV) SKHT = 29 012 + 1 296 (LPLV) SKHT = 28 776 + 1283 (LXLV) SKHT = 35 864 + 1 326 (LT) SKHT = 39 630 + 1 360 (LX) SKHT = 40 403 + 1369 (LP) SKHT = 41 444 + 2 497 IBF) SKHT = 42 805 + 2 443 (XF) SKHT = 50 721 + 2 680 (XTI SKHT = 54 529 + 2 629 (ST) SKHT = 36 592 + 2 023 (ARM) SKHT = 48.829 + 3.229 (XH) SKHT = 53.972 + 3.943 (XR) SKHT = 55.138 + 3.666 (SU) 156 159 1 62 2 25 2 31 2 32 2 61 2 66 2 74 2 85 3 22 3.42 3.92 4.19 95 95 95 95 95 95 95 95 95 95 80 92 80 75 0.967 0.966 0.965 0.931 0.927 0.926 0.905 0.902 0.896 0.886 0.848 0.833 0.764 0.709 SKHT = 22.581 + 1.269 (LTLV) SKHT = 23.662 + 1.334 (LPLV) SKHT = 22.126 + 1.332 (LXLV) SKHT = 34.130 + 1.333 (LT) SKHT = 34.189 + 1.404 (LX) SKHT = 36.730 + 1.394 (LP) SKHT = 44.253 + 2.336 (XF) SKHT = 43.697 + 2.381 (BF) SKHT = 50.764 + 2.686 (CMT) SKHT = 47.824 + 1.576 (UPRI SKHT = 49.527 + 2.668 (XT) SKHT = 62.360 + 2.706 (XHI SKHT = 73.945 + 3.033 (XR) SKHT = 76.588 + 2.717 (SU) 1.62 1.69 1.72 2.26 2.37 2.38 2.54 2.56 2.91 2.98 3.04 3.07 3.61 3.72 76 76 76 76 76 0.945 0.940 0.938 0.890 0.878 0.877 0.859 0.856 0.809 0.800 0.790 0.785 0.702 0.647 76 76 76 76 76 76 65 61 54 ' BF = bicondylar femur l e n g t h CMT = condyle-malleolus tibia length; FO = foot height; LV = sum of heights of lumbar ventebrae; = maximum tibia length, XH = maximum humerus l e n g t h XR = maximum radius length; SU = length of ulna without styloid process; CRAN = cranial height; VERT = vertebral column height; SKHT = skeletal height. All measures in centimeters. 2Formulae yield skeletal heights. To obtain stature add 10 cm to skeletal heights o f 153.5 cm or less, 10.5cm to skeletal heights of153.6165.4 cm, and 11.5 em to skeletal heights of 165.5 cm or greater. For living stature estimates subtract 0.06 em for each year over 30 years of age. XF = maximum femur length; XT mate is desired, subtract 0.06 cm for each year over 30 years of age from the stature estimate. CONCLUSIONS The results of the present analyses strongly support our prior conclusion that bodily proportions were conservative features in Native Americans of the Ohio Valley area. In our previous study we included individuals from Late Archaic through Late Prehistoric populations located primarily in the northern sections of Ohio. In this study we expanded both the geographical and temporal range of the sample by adding individuals from central and southern Ohio, southwestern Pennsylvania, and north central West Virginia and by including individuals from protohistoric populations. However, with this expanded sample we find that the percent contribution of each component of skeletal height remains virtually identical to that which we previously noted and is constant among the present temporal groupings. Even the male groupings that differ in total skeletal height do not differ in proportional contributions to skeletal height. Thus the regression equations developed for the total samples should be applicable, in general, to any geographical or temporal subset of Ohio Valley area Native Americans with similar proportional contributions to skeletal height. The generally poor performance of regression equations based on living East Asian populations for the estimation of stature in Ohio Valley Native Americans, shown in our previous report (Sciulli e t al., 1990), appears to be due to the relatively long limbs, especially the distal elements, expressed by the Ohio Valley Native Americans. It is well known that the proportionally shortest legs as well as the shortest distal elements of the limbs can be found among living East Asian populations and populations recently derived from them. However, as Table 1 STATURE ESTIMATION IN OHIO NATIVE AMERICANS shows, Ohio Valley Native Americans are characterized by relatively long legs and, as has been shown previously (Sciulli et al., 1991; Sciulli, in press), by relatively long distal elements of the limbs (for example, a brachial index of 77.8 and a crural index of 85.0).Thus, use of regressions based on East Asian populations, or other populations of similar proportions, to estimate stature in Ohio Valley Native Americans will significantly overestimate stature (Sciulli et al., 1990).Because many other prehistoric Eastern Woodlands populations show similar relatively long distal limb elements (Sciulli et a]., 1991), regression equations from living East Asian populations are probably inappropriate for stature estimation in these populations also. Additional testing should show whether the present equations or those from living East Asian populations provide better stature estimates for general 399 use among Eastern Woodlands Native Americans. LlTERATURE CITED Box GEP (1949)A general distribution theory for a class of likelihood criteria. Biometrika 36r317-346. Dixon WJ (1985) BMDP Statistical Software. Berkeley: University of California Press. Fully G (1956) Une nouvelle methode de determination de la taille. Ann Med. Legale Criminal. 35t266-273. Sciulli PW (in press) Stature and proportions in the Monongahela population of western Pennsylvania. Ann Carnegie Mus. Sciulli PW, Pacheco PJ, and Janini CA (1991) Variation in limb bones of terminal Late Archaic populations of Ohio. Midcontinental J Arch. 16:247-271. Sciulli PW, Schneider KN, and Mahaney MC (1990) Stature estimation in prehistoric Native Americans of Ohio. Am. J. Phys. Anthropol. 83:275-280. Sokal RR, and FJ Rohlf (1981)Biometry. San Francisco: Freeman. Tibbets GL (1981) Estimation of stature from the vertebral column in American Blacks. J. Forensic Sci. 26:715-723.