Histomorphological and geometric properties of human femoral cortex in individuals over 50: Implications for histomorphological determination of age-at-death

AMERICAN JOURNAL OF HUMAN BIOLOGY 6:65%667 (1994)

Histomorphological and Geometric Properties of Human Femoral Cortex in Individuals Over 50: Implications for Histomorphological Determination of Age-At-Death

ROBERT A. WALKER’, C . OWEN LOVEJOY2,3,4, AND RICHARD S . MEINDLz ‘Department of Ana.tomv, New York Chiropractic College, Seneca Falls, N e b York 13148; ‘Depa-rtment ofAnthropb1og.y and Bi&ogical Anthropology Program, Division. of Biomedical Sciences, Kent State Uniuersity, Kent, Ohio 44242; 3Cleueland Museum of Natural History and 4Cuyahoga County Coronerts Office. Cleveland, Ohio 44106

ABSTRACT Cortical area, area moment of inertia, and polar moment of inertia were determined from the midshafts of a series of 62 femurs (34 female and 28 male) from a U. S. white population, ages 51-95. The density of osteons and osteon fragments (per mm2) was also determined. Neither osteon nor osteon fragment density was significantly correlated with age. These variables were correlated, however, with normalized cortical and endosteal areas, normalized area moment of inertia, and polar moment of inertia. Osteon fragment numbers alone are not highly correlated with cross-sectional properties. This research suggests that osteon density and osteon fragment density are significantly related to cortical mass and distribution in older people, but are not a direct function of age in persons over 50 years of age. Histological age estimates in older individuals must, therefore, be used with extreme caution. o 1994 Wiley-Liss, Inc.

Histomorphometric features of cortical bone have often been advanced as age indicators for unidentified modern and archeo- logical skeletal material. Such use requires a regular relationship with chronological age that is not compromised by other factors that may influence skeletal homeostasis. When observed in large samples over wide age ranges (Kerley, 1965, 1969; Kerley and Ubelaker, 1978; Thompson, 1978, 1979; Stout and Simmons, 1979; Simmons, 1985; Walker, 1989), a demonstrable relationship between histomorphometric features and chronological age has routinely been observed. Walker (1992) demonstrated that between 18 and 95 years, low but significant correlations between osteon densities in the anterior cortex of the midshaft femur and age are observed (r = 0.44, P < .001, n = 175). However, when this sample is truncated at age 50, there is no significant correlation between age and osteon density for those >50 (r = -0.02, P = 341, n = 80) (there is a low but significant correlation between osteons per mm2 and age for those 4 0 (r = 0.37, P < .001, n = 95)). Similar results are obtained for osteon fragments

(entire age range: r = .63, P < .001, n = 175; 50 + years: r = .18, p = .loo, n = 80; under 50: r = .25, P = .016, n = 95). The group over 50 years appears to form a separate cluster with distinctly higher osteon and fragment densities, but without any further age-related trends. Chronologi- cal age does not appear to be the most impor- tant factor affecting histomorphology of the femur in this age group.

It has long been recognized that there is a strong relationship between mechanical loading and bone maintenance (Frost, 1973; see also Carter, 1987; Wong et al., 1988; Carter et al., 1989, 1991; for reviews). The research presented here examines the relationships among age, osteons, and osteon fragments per mm2, and cross-sectional properties of the midshaft femur in a population of older modern American whites.

Received February 17,1994; accepted April 23,1994. Address reprint requests to Robert A Walker, Department of

Anatomy, New York Chiropractic College, 2360 State Route 89, Seneca Falls, NY 13148.

Q 1994 Wiley-Liss, Inc.

660 R.A. WALKER ET AL.

MATERIALS AND METHODS

Cortical area, area moments of inertia, and polar moments of inertia were estimated from the midshafts of a series of femurs from an American white population, aged 51-95 years. The femur was chosen because it has been the focus of clinical and paleopathological studies of human cortical bone (Broman et al., 1958; Smith and Walker, 1964; Atkinson and Weatherell, 1967; Armelagos, 1969; Van Gerven et al., 1969; Carlson et al., 1976; Ericksen, 1976; Martin and Armelagos, 1979; Simmons, 1985; Ruff and Hayes, 1988; Carter et al., 1989; Gordon et al., 1989; Beck et al., 1990; Slemenda and Johnston, 1990). Its midshaft exhibits the thickest cortex in the body, and thus presents a wide microscopic field for study. Additionally, the histomorphology of the femur has frequently been championed as a means of determining skeletal zge at death (Kerley, 1965; Ahlqvist and Damsten, 1969; Singh and Gunberg, 1970; Bouvier and Ubelaker, 1978; Kerley and Ubelaker, 1978; Thompson, 1978,1979,1980; Pfeiffer, 1980, 1985; Uytterschaut, 1985; Ericksen, 1991).

The present sample includes 28 males (45%) and 34 females (55%). Females varied from 51-93 years and males from 53-95 years. Mean age at death does not significantly differ between the two groups (F = 1.263, P = 0.266). This sample was obtained from cadavera of known age following dissection in two medical schools (Northeast Ohio Universities College of Medicine and the University of Arkansas School of Medi- cal Sciences). Only cadavera that succumbed to diseases not directly related to bone turnover were used for this study.

Neither sex nor age was known t o the principal investigator (R.A.W. j during data collection and neither was revealed until all ages had been assigned and the remainder of the analysis carried out. This mimics ar- chaeological and forensic conditions and is the only valid means of assessing the accuracy and biases of any method of skeletal age determination.

A 1"-long cylinder of bone was removed from the midshaft of the right femur of each specimen. In a few specimens the left femur was used. Cortical areas, area moments of inertia, and polar moments of inertia were estimated for these specimens using the GEOXRAY program (Ohman, 1993a;

Walker, 1993). The program estimates geometric properties of long bone cross sections based on both concentric and eccentric elliptical models (Young, 1989). Only the results based upon the eccentric models were used because they are more accurate (the assumption that periosteal and endosteal pe- rimeters of a long bone cross section are concentric is not necessarily valid; this program does not rely on this assumption). The method is highly accurate and has been ex- tensively tested by comparison with direct calculations from CT scans. For 193 specimens the simplified method employed here for femoral cross sections differed from that obtained directly from CT scans by only 4% (Ohman, 1993bj. Furthermore, those data were estimated from orthoganal conven- tional X-rays. The data in the present report were obtained directly from each section and should, therefore, be considerably more accurate.

Undecalcified thin sections were also pre- pared for all specimens. Transverse sections were cut at a thickness of 100 Fm using a Beuhler Isomet saw. Specimens were mounted on 25 x 75 mm standard microscope slides and covered with no. 1 thickness coverslips. Osteons and osteon fragments were counted in four fields around the pe- rimeter of the midshaft femur (Kerley, 1965). Counts were made at 1 0 0 ~ power with a Bausch and Lomb binocular microscope using l o x widefield oculars. This yielded a field size of 3.03 mm2. Osteons and osteon fragments were counted according to the procedures outlined by Kerley (1965 j and are reported as osteons/mm2 or osteon fragmentdmm'. Two independent counts were made of each section as a test of reliability (Table 1). All areal data (total cortical area and total endosteal area) were normalized by division by the second power of femoral length, and all geometric data were normalized by the fourth power of femoral length. A general summary of all data obtained from these sections is provided in Ta- ble 2.

RESULTS Correlations between histological fea-

tures and geometric properties of the midshaft femur are presented in Table 3, and the relationship between age and histological features in Table 4. In this population of older white Americans, neither osteon den-

HUMAN FEMORAL CORTICAL PROPERTIES 661

TABLE 1. General data summary

Observation X u Min Max

Osteondmm 14.68 Osteon fragdmm 16.44 Norm. I, ant-post 0.61 Norm. I, med-lat 0.68 Norm. J 1.29 Norm. tot. cort. area 2.02 Norm. tot. end. area 1.09

Osterondmm 15.32 Osteon fragdmm 15.56 Norm. I, ant-post 0.64 Norm. I, med-lat 0.7 Norm. J 1.34 Norm. tot. cort. area 2.08 Norm. tot. end. area 1.05

Osteondmm 14.14 Osteon fragdmm 17.17 Norm. I, ant-post 0.59 Norm. I, med-lat 0.66 Norm. J 1.25 Norm. tot.. cort. area 1.97 Norm. tot. end. area 1.12

Males only (N = 28)'

Females only (N = 34)'

2.95 5.28 0.15 0.18 0.31 0.54 0.5

2.43 5.37 0.16 0.18 0.32 0.54 0.51

3.23 5.17 0.14 0.18 0.3 0.53 0.49

7.3 5.28 0.23 0.21 0.44 0.54 0.43

11.68 5.28 0.27 0.22 0 48 0.62 0.48

7.3 7.59 0.23 0.21 0.44 0.54 0.43

20.38 26.57 0.95 1.13 1.85 2.92 2.43

19.85 26.57 0.95 1.04 1.85 2.77 2.26

20.38 26.41

0.95 1.13 1.82 2.92 2.43

'Total sample, N = 62

TABLE 2. Reliability ofosteon counts

Field Dosition and count X u Min Max

Field 1: first count Field 1: second count Field 2: first count Field 2: second count Field 3: first count Field 3: second count Field 4: first count Field 4: second count

41.5 12.4 20 72 42.6 13.2 10 74 47.6 11.2 29 78 48.2 11.2 18 73 39.3 10.6 17 59 39.6 10.4 14 64 48 12.7 24 84 49 15.9 18 98

sity nor osteon fragment density is significantly correlated with age (Figs. 1 and 2). However, osteon density is generally correlated very significantly with normalized cortical area (NCA) and normalized endosteal area (NEA) (Figs. 3 and 41, normalized area moments of inertia about the anteroposte- rior (NIap) and mediolateral axes (N1,J (Figs. 5 and 6), and normalized polar moment of inertia (NJ) (Fig. 7). Osteon density does not generally correlate significantly with normalized total cross-sectional area (NTA).

Among females (Tables 3 and 4), osteon density is correlated with NCA, NI,,, NI,.,.,], and NJ. Among males (Tables 3 and 41, geometric properties are also significantly correlated with most measures of osteon and osteon fragment density. As a means of assessing the relationship of total osteonal

bone to geometric properties, correlations between the sum of both osteons and osteon fragments/mm2 with geometrical properties of the femur were also calculated (Table 3). This sum is not correlated with age or NTA (Table 41, but most other geometric properties were correlated significantly with this sum.

Osteon density and age determination from published regression techniques

Quantitative age estimates were made by means of the modified Kerley method (Ker- ley and Ubelaker, 1969). This method does not differentiate between males and females, which is of prime importance to ar- chaeological applications, where the sex is often unknown. In addition, this technique uses histological features that are easily quantified with a minimum of equipment and technology and is thus widely applica- ble. Although it has been applied in paleode- mographic and forensic situations, it has not been subject to rigorous tests of accuracy. Age estimates using the Kerley technique were calculated by summing osteons/mm2 across all four fields and using the equation:

AGE = 2.278 + 0.178 [[Z osteond mm2112.061 + .00226[[2 osteons/mm212.0612.

Osteon counts were multiplied by a factor of 2.06 to accommodate Kerley's original field size of 2.06 mm2 (Kerley and Ubelaker, 1978). A least-squares regression equation for age determination had also been generated in a previous study from a sample of 173 specimens ranging in age from 18-95 years (Walker, 1989). That sample included the 62 specimens examined here. This equation was used to estimate the age of the specimens in the present sample from average osteons/mm2 €or each section. The regression equation used (modified from Walker, 1989) is as follows:

AGE = 2.031[[2 osteons/mm21/41 + 25.962.

As expected from the lack of correlation of histomorphological features with age, the correlation and regression statistics for these histological age estimates (Kerley and sample-specific regression equations) (Table 5) are not significantly different from zero, and slopes of the regression lines are also not significantly different from zero.

662 R.A. WALKER ET A1

TABLE 3. Correlations between histornorphornetric features and cross-sectional propehes o f the rnidshafl femur

NTA' NCA' NISn3 NI...,~ NJ5

Osteons/mm2

Osteon Fragmentdmm'

Osteons + Fragments/mm2

Males (N = 28F Osteondmm2

Osteon Fragmentdmm'

Osteons + Fragments/mm*

Females (N = 34)6 Osteons/mm2

Osteon Fragments/mm2

Osteon + Fragmentdmm'

.03Oa7 P = ,406

,2048 P = ,055

,1690 P = .095

-.0267 P = .446

,4794 P = ,005

,3793 P = .023

.0245 P = ,445 -.0108

P = ,476 ,0032

P = .493

.6052 P < ,001

.2227 P = ,041

,4269 Pi ,001

,5391 P = ,002

,2290 P = ,121

,3868 P = ,021

,6476 P i . 001

.2550 P = ,073

,4689 P = .003

,4048 P = ,001

,3110 P = ,007

,4089 P < ,001

.2552 P = .095

,4730 P = ,006

,4791 P = .005

,4833 P = .002

,2264 P = ,099

.3755 P = ,014

.4685 P < ,001

.I863 P = .074

,3411 P = ,003

,4518 P = ,008

.1880 P = ,169

,3210 P = ,048

.4692 P = ,003

,2231 P ,102

.3669 P = ,016

.4730 P < ,001

,2607 P = ,020

,3997 P = ,001

.3866 P = ,021

,3422 P = ,037

,4219 P = .013

,5143 P = ,001

.2423 P = ,084

.4Offi P = ,009

'Normalized total area. 'Normalized cortical area. 3N~rmalized area moment of inertia along the AP axis. 4N~rmalized area moment of inertia along the ML axis. 'Normalized polar moment of inertia. 'Total sample, N = 62. 'Pearson's Carrelation CoefficientJl-tailed significance.

TABLE 4. Correlations between age and histornorphornetric features

Osteons/mm' Frmentdmm2 Fragments/mm* Osteon Osteons + t

- ,0450 ,0769 ,0394 P = ,364 P = ,276 P = ,380 Males (N = 28)l -.1246 ,1965 .1131 P = .264 P = ,158 P = ,283 Females (N = 34)' ,0509 - ,0642 -.0230 P = ,387 P = .359 P = ,449

18

2 16 E E E 14 0

. -

12 8

'Pearson's Correlation CoefficientJl-tailed significance, total sample, N = 62.

Inaccuracy and bias Previous studies of the accuracy of skele-

tal age determination techniques have re- lied heavily upon the use of Pearson's correlation coefficient (Boucquet-Appel and Masset, 1982). Although Pearson correlations may be one means of demonstrating the usefulness of a technique, it is certainly not the only, or even the best, method. A high correlation coefficient simply indicates that there is a strong correspondence between paired values of two variables. In the case of an age estimate, this provides no in- dication of accuracy or probable error. A bet- ter means of determining the usefulness of

c

Fig. 1. Plot of osteondmm' with age. Straight line represents the least-squares regression line of osteons/ mm2 regressed against age. Curved lines represent the 95% confidence limits of the regression line.

an aging technique is the assessment of inaccuracy and bias (Lovejoy et al., 1985). Bias may be defined as:

estimated age - actual age.


. . I . '

1- . ' . . . . . - .

30

25 1 Y)

c - 15

!l LL C

,2 0" 10 I I

I 0 . I I I I I

40 50 60 70 80 SO 100

Age

I .

6 8 10 12 14 16 16 20 22

Osteons / mm sq.

~ i ~ , 2, plot of osteon fragments/mmz with age, Straight line represents the least-squares regression line of osteon fragmentdmmz regressed against age. Curved lines represent the 95% confidence limits ofthe regression line.

Fig. 4. Plot of NEA with osteaons/mm2. Straight line represents the least-squares regression line of normalized endosteal area regressed against osteondmm2. Curved lines represent the 95% confidence limits of the regression line.

. . m

6 8 10 12 14 16 18 20 22 6 8 10 12 14 16 18 20 22

Osteons 1 rnm sq. Osteons / mm sq

Fig. 5. Plot of NI,, with osteons/mm2. Straight line represents the least-squares regression line of NI,, regressed against osteons/mmZ, Curved lines represent the 95% confidence limits of the regression line.

Fig. 3. Plot of NCA with osteons/mm2. Straight line represents the least-squares regression line of NCA regressed against osteaons/mrn2. Curved lines represent the 95% confidence limits of the regression line.

Bias thus includes the sign of the difference between the two variables (age and age indi- cator). If an age estimate is below known age, then bias is negative, and if an age esti-

mate is greater than known age, then bias is positive. Inaccuracy is the absolute value of the difference between actual age of the specimen and estimated age. It is defined as:

664 R.A. WALKER ET AL

TABLE 5. Correlations of Kerley osteonal age estimates and sample-specifx age estimates with known age

I .

o / I I I , I I

6 8 10 12 14 16 18 20 22

Osteons / mm sa.

Fig. 6. Plot of NI,, with osteondmm'. Straight line represents the least-squares regression line of NI,, regressed against osteondmni2. Curved lines represent the 95% confidence limits of the regression line.

2

I . A

6 8 10 12 14 16 16 20 22

Osteons / mm SQ

Fig. 7. Plot of NJ with osteonslmm'. Straight line represents the least-squares regression line of NJ regressed against osteons/mmz. Curved lines represent the 95% confidence limits of the regression line.

lestimated age - actual agel.

Useful age estimators should demonstrate low values of both bias and inaccuracy. The analysis of age indicators presented by

Total sample r Slope P

Kerley osteonal p.040 ,759 -.026

Sample-specific p.045 ,728 p.084 age estimates

age estimates Males r Slope P

Kerley osteonal -.130 ,509 -.094 age estimates

age estimates Sample-specific -.125 ,528 - ,280

Females r Slope P

Kerley osteonal ,068 .704 .040

Sample-specific ,051 .775 .086 age estimates

age estimates

TABLE 6. Biases and inaccuracies of KerZey and sample-specific osteonal age estimates

Total sample Bias Inaccuracy

Kerley osteonal -15.15 20.95 age estimates

age estimates Sample-specific - 18.63 19.09

Males Bias Inaccuracy

Kerley osteonal -9.84 18.62

Sample-specific -15.56 16.57 age estimates

age estimates Females

Bias Inaccuracy Kerley osteonal -19.52 22.87

Sample-specific -21.16 21.16 age estimates

age estimates

Lovejoy et al. (1985) demonstrated that age indicators with low bias and low inaccuracy could produce age distributions that did not differ significantly from the actual age distribution of the sample.

Bias and inaccuracy for the histological age estimates in this sample are presented in Table 6 and in Figures 8 to 11. When males and females are combined, the biases of both the sample-specific regression equation and Kerley's regression equation based on femoral osteon counts are high (Table 6). Both demonstrate a bias of more than a de- cade. Bias for the Kerley technique is -15.15 years, whereas bias for the sample- generated regression equation is - 18.63 years. Thus the sample-generated regression equation is, in fact, more biased than Kerley's equation. However, when the inac-


2o m

g -10 1 / I r-ml- L1_1 Female ll

-40 4

-50 I I I I

50-59 60-69 70-79 80-89 90+

Age Group

Fig. 8. Bias of Kerley femoral osteonal age estimates. Solid bars represent males, stippled bars represent females.

50

40

30 L?

f

20

10

0

- Male U Female

Fig. 9. Inaccuracy of Kerley femoral osteonal age estimates. Solid bars represent males, stippled bars represent females.

curacy figures for these two techniques are considered, the trend is reversed. For the sample-specific regression equatiqn, the inaccuracy is 19.09 years, meaning an average specimen was misaged by this amount. For the Kerley estimates, this figure is approxi- mately 20.95 years. Thus neither technique is accurate for individual >50 years of age.

0 U -10 -

c >. -20 - I

Female

-50 50-59 80-69 70-79 80-89 Bo+

Age Gmup

Fig. 10. Bias of sample-specific femoral osteonal age estimates. Solid bars represent males, stippled bars represent females.

50

40

30 I >

20

10

0

I Male 0 Female n

d 5 9 80-69 7d79 8Ok9 1 'O+

Age Gmup

Fig. 11. Inaccuracy of sample-specific femoral osteonal age estimates. Solid bars represent males, stippled bars represent females.

Estimated vs. actual age distributions It is necessary to know how well an age

estimator will reconstruct the known age distribution of a population. For this pur- pose, the distributions of the two histological age estimates were compared to the distributions of known ages by means of two-

666 R.A. WALKER ET AL.

sample Kologomorov-Smirnov goodness-of- fit tests (Sokal and Rohlf, 1981). Both comparisons (Kerley age vs. known age, and sample-specific estimates vs. known age) differ significantly a t the 0.05 level of proba- bility. Hence, neither method produces an age distribution that reflects the actual distribution of ages in the sample.

DISCUSSION Stout and Stanley (1991) suggest that os-

teon density, rather than percent osteonal bone (e.g., Ahlqvist and Damsten, 19691, should be the variable of choice in develop- ing histological methods of age prediction. However, as demonstrated here, osteon density in the femoral midshaft is positively correlated with cortical cross-sectional area and geometric properties, but not with age. There is a significant positive correlation between osteon density and cortical area (Fig. 3) and a significant negative correlation between osteon density and endosteal area (Fig. 4). These results suggest that, at least among white individuals over 50 years of age, osteon density is related to cortical area and cortical porosity. In highly osteoporotic individuals (i.e., those demonstrating large endosteal areas and thin cortices), osteon density is low. This relationship between cortical area and osteon density or fragment density, together with a plateau in osteon density beyond 60 years of age (Walker, 1989), indicates that the development of useful quantitative histological features for age determination among older adults may prove problematic.

Haversian structures have a significant positive correlation with cortical area, measures of cortical distribution (i.e., moments of inertia), and a significant negative correlation with endosteal area. The number of osteon fragments increases with continued remodeling and bone turnover. Concur- rently, the number of adjacent osteons that will overlap or encroach upon one another will also increase. Eventually, as older osteons are eroded away and replaced by younger ones, plateaus of osteon and osteon fragment densities are reached by a maxi- mum of 60 years of age (Walker, 1989). In young adult specimens, the overall amount of cortical bone is generally stable, reaching a peak of thickness and overall amount at -35 years of age (Goldsmith et al., 1973; Goldsmith and Johnston, 1975). From 35 years onward, the amount of cortical bone

gradually diminishes, particularly among females. Eventually, this thinning of cortical bone in the midshaft femur is accompa- nied by a decreasing density of osteons/mm2 in the cortex that remains, both processes perhaps being symptomatic of a systemic de- crease in bone metabolism. The net result is that among older individuals there exists no systematic relationship between the densities of osteons or osteon fragments and chronological age. There may, therefore, be an age plateau beyond which there exists no useful relationship between bone histomorphology and chronological age. Our results indicate that this plateau may occur as early as 50 years of age.

These results create two problematic is- sues. The simplest is that age estimates based on bone histomorphology are probably very unreliable for specimens over 50 years. The second is more difficult to deal with; specifically, if the age of the specimen is unknown, how can this failure of age determination of signaled, that is, without age data, how can the unreliability of age estimation procedures in specimens older than 50 years be recognized? This specific issue will need to be addressed before histomorphological age determination can be made more reli- able than is possible using current methods.

ACKNOWLEDGMENTS Thanks are due to B. Shea, J. Rose, S.

Ward, J. Bernard, N. Taslitz, J. Ohman, T. Greiner, C. Terranova, B. Davis, J. DeCicco, S. Firkins, N. Vijayashankar, R. Schass- burger, T. Wiant, C. Jagos, and all a t the KSU Biological Anthropology Laboratory.

LITERATURE CITED Ahlqvist J , Damsten 0 (1969) A modification of Kerley’s

method for the microscopic determination of age in human bones. J. Forens. Sci. 14:205-217.

Armelagos G (1969) Disease in ancient Nubia. Science 163:255-259.

Atkinson PJ, Weatherell J A (1967) Variation in the density of the femoral diaphysis with age. J. Bone Jt. Surg. 49Bt781-788.

Beck TJ, Ruff CB, Warden KE, Scott J r WW, Gopala U (1990) Predicting femoral neck strength from bone mineral data: A structural approach. Invest. Radiol. 25%-18.

Bouquet-Appel JP, Masset C (1982) Farewell to paleo- demography. J. Hum. Evol. 11:321-333.

Bouvier M, Ubelaker DH (1977) A comparison of two methods for the microscopic determination of age at death. Am. J. Phys. Anthropol. 46:391-394.

Broman CE, Trotter M, Peterson RR (1958) The density of selected bones of the human skeleton. Am. J. Phys. Anthropol. I6:197-211.


Carlson DS, Armelagos GJ, Van Gerven DP (1976) Pat- terns of age-related cortical bone loss (osteoporosis) within the femoral diaphysis. Hum. Biol. 48:295-314.

Carter DR (1987) Mechanical loading history and skeletal biology. J. Biomech. 20:1095-1109.

Carter DR, Orr RE, Fyhrie DP (1989) Relationships between loading history and femoral cancellous bone architecture. J. Biomech. 22:231-244.

Carter DR, Wong M, Om TE (1991) Musculoskeletal on- togeny, phylogeny, and functional adaptation. J Bio- mech. 24:3-16.

Ericksen MF (1976) Cortical bone loss with age in three Native American populations. Am. J. Phys. Anthro- pol. 45:443452.

Ericksen MF (1991) Histological estimation of age at death using the anterior cortex of the femur. Am. J. Phys. Anthropol. 84r171-179.

Frost HM (1973) Bone Remodeling and Its Relationship to Metabolic Bone Diseases. Springfield, IL: C.C. Tho- mas.

Goldsmith NF, Johnston JO (1975) Bone mineral: Ef- fects of oral contraceptives, pregnancy, and lactation. J. Bone Jt. Surg. 57A:657-668.

Goldsmith NF, Johnston JO, Picetti G, Garcia C (1973) Bone mineral in the radius and vertebral osteoporosis in an insured population. J. Bone Jt. Surg. 55A:127f% 1293.

Gordon KR, Reese SR, Knecht PA (1989) Changes in the architecture of cancellous bone in the femora of devel- oping mice as a result of short duration hypergravity. Am. Zool. 2Yr205-219.

Kerley ER (1965) The microscopic determination of age in human bone. Am. J. Phys. Anthropol. 23r149-164.

Kerley ER (1969) Age determination of bone fragments. J . Forens. Sci. 14:5!+67.

Kerley ER, Ubelaker DH (1978) Revisions in the microscopic method of estimating age at death in human cortical bone. Am. J. Phys. Anthropol. 4Y:545-546.

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

Martin DL, Armelagos GJ (1979) Morphometrics of com- pact bone: An example from Sudanese Nubia. Am. J. Phys. Anthropol. 51.571-578.

Ohman JC (1993a) Computer software for estimating cross sectional geometric properties of long bones with concentric and eccentric elliptical models. J Hum.

Ohman JC (199313) Cross section geometric properties from biplanar radiographs and computed tomogra- phy: Functional application t o the humerus and femur in hominoids. Doctoral dissertation, Kent State University, Kent, Ohio.

Pfeiffer S (1980) Bone-remodelling age estimates compared with estimates by other techniques. Current Anthropol. 21:793-794.

Pfeiffer S (1985) Comparison of adult age estimation

EvoI. 25:217-227.

techniques, using an ossuary sample. Canad. Rev. Phys. Anthropol. 4r13-17.

RuiT CB, Hayes WC (1988) Sex differences in age-related remodeling of the femur and tibia. J. Orthop. Res. 6:886-896.

Simmons DJ (1985) Options for bone aging with the microscope. Yrbk. Phys. Anthropol. 28:249-263.

Singh IJ, Gunberg DL (1970) Estimation of age a t death in human males from quantitative histology of bone fragments. Am. J. Phys. Anthropol. 33r373-381.

Slemenda CW, Johnston JR CC (1990) Osteoporotic fractures. In: Simmons DJ (ed.): Nutrition and Bone Development. New York: Oxford University Press, pp.

Smith RW Jr, Walker RR (1964) Femoral expansion in ageing women: Implications for osteoporosis and fractures. Science 145:156-157.

Sokal RR, Rohlf FJ (1981) Biometry, 2nd ed. San Fran- cisco: Freeman.

Stout SD, Simmons DJ (1979) Use of histology in ancient bone research. Yrbk. Phys. Anthropol. 22:22% 249.

Stout SD, Stanley SC (1991) Percent osteonal bone ver- sus osteon counts: The variable of choice of estimating age at death. Am. J . Phys. Anthropol. 86:515-519.

Thompson DD (1978) Age-related changes in OsteonRe- modeling and Bone Mineralization. Doctoral dissertation, University of Connecticut, Storrs.

Thompson DD (1979) The core technique in the determination of age at death in skeletons. J. Forens. Sci.

Thompson DD (1980) Age changes in bone mineralization, cortical thickness, and haversian canal area. Calcif. Tis. Internat. 313r5-11.

Uytterschaut HT (1985) Determination of skeletal age by histological methods. Z. Morph. Anthropol. 75r331- 340.

Van Gerven DP, Armelogos GJ, Bartley MH (1969) Roentgenographic and direct measurement of femoral cortical involution in a prehistoric Mississippian population. Am. J. Phys. Anthropol. 31:23-38.

Walker RA (1989) Assessments of human cortical bone dynamics and skeletal age at death from femoral cortical histomorphology. Doctoral dissertation, Kent State University, Kent, Ohio.

Walker RA (1992) Human variation, cortical bone histology, and the determination of skeletal age at death. Am. J. Phys. Anthropol. Suppl. 14:169 (Abstract).

Walker RA (1993) Relationship of cross-sectional biome- chanical properties of cortical bone to histomorphology in white Americans over 50. Am. J . Phys. Anthro- pol., Suppl. 16:202 (Abstract).

Wong M, Carter DR (1988) Mechanical stress and mor- phogenetic endochondral ossifgication of the sternum. J. Bone Jt . Surg. 70A:992-1000.

Young WC (ed.) (1989) hark’s Formulas for Stress and Strain, 6th ed. New York McGraw-Hill.

131-163.

24:902-915.

Documents

Histomorphological and geometric properties of human femoral cortex in individuals over 50: Implications for histomorphological determination of age-at-death