Anatomical physiological and epidemiological correlates of the aging process A confirmation of multifactorial age determination in the Libben skeletal population.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 68:87-106 (1985) Anatomical, Physiological, and Epidemiological Correlates of the Aging Process: A Confirmation of Multifactorial Age Determination in the Libben Skeletal Population C. OWEN LOVEJOY Departments of Anthropology and Biology, Kent State University, Kent, Ohio 44242 (C. 0.L., R.P. M.) and Department of Orthopaedic Surgery, Case Western Reserve University, Cleveland, Ohio 44106, Cleveland Museum of Natural History, Cleveland, Ohio 44106, Department of Human Anatomy, Northeast Ohio Universities College of Medicine, Rootstown, Ohio 44272, and Cuyahoga County Coroner’s Office, Cleveland, Ohio 44106 (C. 0.L.1 ROBERT P. MENSFORTH AND KEY WORDS Paleodemography, Cortical bone dynamics, Fracture epidemiology, Skeletal age at death ABSTRACT Paleodemographic analyses based on estimates of skeletal age at death consistently report high levels of young adult mortality with few individuals living in excess of 50 years. Critics assert these data indicate systematic underaging of adults and justifiably remark that 1) criteria for estimating skeletal age a t death may be unreliable, 2) age determinations are too frequently based on one or two criteria alone, and 3) adult paleodemographic age profiles often mimic the age distributior, of the modern population from which a n age indicator’s standards were originally derived. This study reports a series of tests based on well-documented biological aging phenomena that can be used to investigate potential effects of systematic underaging in adults, assuming the skeletal population is of sufficient size to permit such tests. These include patterns of third decade sternal clavicular epiphyseal fusion, multiple age and sex criteria associated with cortical bone dynamics, and fractures known to occur throughout the entire adult ages range. These phenomena are examined here for the Libben site skeletal population where adult age at death was determined by the multifactorial summary age technique. None of the biological criteria reported here were used in the Libben summary age analysis and thus serve as a n independent test of accuracy in age determination. In addition, the summary age method has recently been applied to a series of modern skeletons of known age (Todd samples 1 and 21. Age standards for criteria employed with Libben and Todd 1 were identical. Since Todd 1displayed underaging in older adults, a second Libben age distribution adjusted for Todd 1bias was generated for comparison. A third Libben adult survivorship profile based on a Coale and Demeny West level 3 mortality experience, considered by some to be a more realistic model for skeletal populations, was produced for comparison. For all criteria examined, original Libben summary ages provided superior concordance with known patterns of biological aging in human populations. While Libben ages adjusted for Todd 1 bias were slightly better in the third decade, both Todd 1 adjusted and Coale and Demeny West level 3 age distributions produced unrealistic patterns of biological aging for individuals greater than 35 years. Implications of these results are discussed. Received January 19, 1984; revised January 2, 1985; accepted January 3, 1985. G 1985 ALAN R. LISS, INC 88 R.P. MENSFORTH AND C.O. LOVEJOY INTRODUCTION One of the most significant features to emerge from recent comparisons of archaeological (“extinct”) and ethnographic (“extant”) populations is their marked divergence with respect to survivorship or mortality (Lovejoy et al., 1977; Bocquet-Appel and Masset, 1982).Recently Howell (1982),using the Libben population as a n example, presented an important paper detailing the social and economic effects of the higher adult mortality seen in archaeological groups. Many of these effects are dramatic and almost surely require revision of our biological and social reconstruction of preurban human populations. As Howell points out: “we now have empirical evidence . . . that life was far more difficult for prehistoric people in North America than we have observed it to be anywhere in the world during the 19th and 20th centuries” (1982:268; emphasis added). She also points out, however, that a second interpretation of skeletal data must be considered: that they are the result of “interaction of the usual mortality pattern with selective biases in the preservation of skeletal materials, errors introduced by the necessarily indirect methods of aging and sexing skeletons, perhaps complicated by cultural practices of the living population that produced the cemetery concerning burial of persons of various ages” (1982264). Howell is addressing two issues. First is the degree to which a cemetery adequately censuses the living population that created it. Second is the accuracy of age determination. We will deal with the first of these two questions in a separate communication. The present paper is an attempt to investigate the second. The reader might immediately question how such a check of the accuracy of age assignments can be carried out, since age at death can never be “known” for a n extinct group. The Libben life table and survivorship curve were obtained by means of the multifactorial (“summary”) age method (Lovejoy et al., 1977). This technique employs multiple anatomical age indicators. In the discussions that follow, the population age distribution, derived from these individual age assignments, will be compared to a series of populational phenomena of physiological, pathological, andor developmental character that bear a strong ineasureable relationship with age. That is, the time sequence of a series of individual anatomical events (changes in age indicators) will be compared to the time sequence of biological events that are well documented from the study of living populations of known age. Although the methods by which the Libben population was aged have already been presented (Lovejoy et al., 1977, 19851, some comment is appropriate here concerning the age estimates used for the oldest segment of the population. Because so few individuals at Libben were found to have ages at death exceeding 50 years, the oldest category used for life table calculations was 50 + (which for calculation was assumed to be of 10 years duration). In fact, individuals clearly older than 60 years were present at Libben, but were so few in number as not to be formally set into higher age categories. During the assignment of values for each age criterion (to be entered into multifactorial age calculation by principal components analysis), a maximum age was set a t 55+ years. This value simply indicated a very old individual. Several individuals had all age indicators set independently a t 55 years. This meant that the specimen was clearly very old, but that there was no need to estimate age more exactly for the purposes of life table calculations. MEDIAL CLAVICULAR FUSION The human sternal clavicular epiphysis is a secondary center of ossification for which the age a t onset and process of closure are TABLE 1. Medlal clavicular fusion with age Nonfusion Sample - ( 0) Active Completed fusion fusion (1-2) (3) Libben (males and females) r\r X SD Modern white’ (males) N xSD Modern black’ (males) N ~ X SD 19 22.3 2.1 15 25.4 2.2 9 26.7 1.9 11 21.8 2.5 24 24.8 2.1 11 27.8 1.2 17 21.4 1.6 42 24.1 2.4 11 26.6 2.2 ‘Data from Todd and D’Errlco (1928) SKELETAL CORRELATESOF THE AGING PROCESS delayed (Stevenson, 1924; Todd and D’Errico, 1928). Fusion usually begins by the twentyfirst year and proceeds slowly, often taking up to 4 years for completion. Though unfused epiphyses were occasionally noted in individuals as old as 27 years, anomalous acceleration and retardation rarely varied beyond 2 years. Owing to the restricted age range for sternal clavicular fusion, it was not incorporated into multifactorial age determination for the Libben population. Therefore, population patterns of clavicular fusion with age can be used to test for any systematic bias in Libben third decade age assignments. All Libben clavicles recovered from burials ranging in age between 20 and 30 years were assessed for nonfusion (O), active fusion (1-21, and completed fusion (3). Table 1 and Figure 1 summarize the mean age and standard deviation for stages of clavicular fusion in Libben males and females compared to modern American blacks and whites studied by Todd and D’Errico (1928). Because their female sample was unusually small, only their male data are given here. They point out however, that no significant sex differences were observed for patterns of clavicular fusion with age. Results show that clavicular fusion patterns for Libben and the modern samples are ‘9 35 ,Ot A Whlto mslo 0 BIsck mslo 0 Llbbon msle h lomalo 0 1-2 3 STAGE OF CLAVICULAR FUSION Fig. 1. Mean age for stages of sternal clavicular fusion comparing modern American black and white males to Libben males and females. 0, nonfusion; 1-2, active fusion; 3, completed fusion. 89 remarkably concordant. In fact, the age a t onset for Libben clavicular fusion is 5 months later than modern whites and only 10 months later than modern blacks. Similarly, Libben clavicular fusion is completed 14 months earlier than modern whites and l month later than modern blacks. Hence, the Libben data indicate that fusion of the sternal clavicular epiphysis would both commence and terminate within the same years as the modern sample. Moreover, variation in mean age for Libben stages of fusion fall well within the range of the modern groups. We interpret these findings as confirmation that Libben third decade age assignments are reasonably accurate and we regard the analysis as constituting a n independent check €or systematic bias in young adult age determination. AGE-RELATED CORTICAL BONE DYNAMICS Cortical Area and Bone Mineral Density We recently completed a study of cortical bone dynamics for a sample of 156 Libben adults. Two variables used in this study were cortical area and intracortical porosity of the femoral midshaft. Since radiographic involution of the proximal femur was a criterion originally used in Libben age determinations, a new summary age was generated for each specimen included in the sample. This assured complete separation of dependent and independent variables. To obtain cortical area, 78 male and 78 female femurs were sectioned at midshaft, cleaned, and photographed. A radial grid was superimposed over enlarged photographs of each section, periosteal and endosteal intersects were recorded (77 of each), and a scaling factor for each photographic enlargement was computed. Actual cortical area was obtained using the GEOBONE program (Lovejoy and Burstein, 1977; Lovejoy and Barton, 1980). Each midshaft area was then divided by femoral length to normalize for size (Dewey et al., 1969b). In addition, each cross section was examined and scored for presencelabsence and degree of involvement (slight, moderate, heavy) of intracortical porosity of the haversian envelope. Van Gerven et al. (1969) demonstrated that radiographic insensitivity to intracortical porosity may introduce substantial error in measurements of cortical thickness and crosssectional area. Similar errors are possible with GEOBONE measurement (in this program only endosteal and periosteal intersects are recorded and area calculated from 90 R.P. MENSFORTH AND C.O. LOVEJOY TABLE 2. Libben midshaft femur cortical area (mm2)/femurlength (mm) Age (years) 18-19 20-24 25-29 30-34 35-39 40-44 45-49 50 + Total Female X Male ~ N 4 8 7 7 16 21 10 5 78 0.8156 0.7630 0.7202 0.8379 0.7528 0.6981 0.6598 0.5882 SD N 0.065 0.089 0.060 0.079 0.087 0.112 0.116 0.141 2 9 7 16 34 10 their spatial distribution). A Dietzen compensating polar planimeter was therefore used to calculate area evacuated by porosity and this total area was then subtracted from the cortical area of each specimen. Thus, all cortical area measurements reported here have been adjusted for intracortical porosity. From a modern clinical perspective, bone mineral density (BMD) and cross-sectional area are among the most useful measurements of age- and sex-related skeletal homeostasis in adults. Such studies demonstrate a slow rate of bone loss in males and females beginning between 35 and 40 years of age (Garn, 1970; Goldsmith et al., 1973; Arnold et al., 1966; Atkinson et al., 1962).The onset of bone loss in otherwise healthy subjects may be related to a decrease in calcium absorption that begins in middle life, affects both sexes, and becomes more pronounced with advancing age (Gallagher and Riggs, 1978). Superimposed on this slow decrease in skeletal mass is a 10 to 15-year period of accelerated bone loss in females following the onset of menopause (45 to 50 years) and consequent ovarian insufficiency (Davis et al., 1966; Meema and Meema, 1976; Cohn et al., 1976, Horsman et a]., 1977). Two persistent limitations restrict the utility of modern clinical studies of bone loss for comparisons with skeletal data. First, clinical surveys often sample attenuated age ranges (many include few or no individuals under 40 years). Second, even fewer studies provide data normalized for size effects, a methodological consideration shown to be important (Dewey et al., 1969b; Cohn et al., 1976). One comprehensive study that overcomes these limitations is available, however, and will be used here for evaluation of Libben - X SD 0.8080 0.8546 0.8673 0.8962 0.8969 0.8291 0.095 0.089 0.059 0.075 0.086 0.061 78 age-related cortical bone dynamics. Goldsmith et al. (1973) report bone mineral density (BMD) in a sample of 3,515 American blacks, whites, and orientals. BMD was determined by photon absorptiometry of the distal radius, and all measurements were normalized for size effects. Age ranged from 18 to 80+ years. It is instructive that patterns of BMD reported in the modern study of Goldsmith et al. (1973) strongly parallel Libben age- and sex-related cortical bone dynamics. These data are summarized for Libben in Table 2 and Figures 2 and 3. The Goldsmith et al. (1973) data are illustrated in Figure 4. Specifically: 1.Among 18-to 20-year-old individuals, the Libben sample shows no sex differential in cortical area just as the modern sample lacks a sex differential in BMD. 2. Both the Libben and modern samples display parabolic relationships of cortical area and BMD with age, respectively. However, the Libben male curve (Fig. 2, 3a) is clearly truncated only 5 years after peak cortical area is achieved. Thus, both Libben males and those observed in the modern study reach peak density a t age 35. In the modern sample, bone loss progresses after that age a t a slow and fairly uniform rate until age 65 at which time it accelerates. The only inconsistency between the Goldsmith et al. (1973) and Libben graphs is the more dramatic slope of bone loss in Libben males between ages 37.5 and 42.5. Since the Libben male midshaft femoral sample is truncated a t this point, these are the oldest males in the Libben population represented here. If this particular group were systematically underaged by about 5 to 10 years, then the rate of bone loss in Libben and modern groups would be similar. However, it should also be 91 SKELETAL CORRELATES OF THE AGING PROCESS 1.0, I I- P W A a a 5 W Y n W U a 3 5: v) a W LI 3 1 - e---.-.3 Female Male 18 2 0 25 30 35 40 45 50 55 AGE IN YEARS Fig. 2. Age- and sex-specific means and standard deviations for Libben midshaft femur cortical area (millimeters squared) normalized by femur length (millimeters). pointed out that Libben males between the ages of 35-40 show no evidence of loss (in fact, this group has the greatest amount of femoral cortical area), whereas bone loss in this group is seen in the modern study. This would suggest slight overaging of these Libben males. We will return to these points later. 3. The Libben and modern female samples also reach peak area and density a t age 35 and the onset of bone loss is apparent in both samples by age 40.Furthermore, there is a n acceleration of bone loss following menopause in both samples (which is superimposed on the normal rate of loss). In the Libben females, however, bone loss appears slightly earlier (about 2.5 years). A more rigorous method of determining the age at which bone loss accelerates in Libben females is discussed below. smith et al. (1973) in their modern survey, several studies have documented a 2 to 6% bone mineral loss from the cortex of normal healthy lactating females (Sorenson and Cameron, 1967; Atkinson and West, 1970; Goldsmith and Johnston, 1975). In Figure 5 Libben female cortical area has been graphed with the modal fertility curve of Weiss (1973). It is of interest that these two curves are the inverse of one another. We suggest, as have others (Dewey et al., 1969a; Armelagos, 1969; Martin and Armelagos, 1979; Stout and Simmons, 1979), that this age-specific bone loss may represent calcium stress imposed by the increased nutritional demands of pregnancy and lactation. Further support of a sex-specific pregnancy stress at Libben is provided by observations on the frequency of unremodeled porotic hyperostosis among third decade Libben females. Porotic hyperostosis is a descriptive Lactation and Pregnancy Stress: The Modal term that refers to macroscopically discerniFertility Curve ble cranial lesions appearing most frequently We have observed a 6 to 12% reduction of on the anterior portion of the supraorbital midshaft femoral cortex in Libben females plate (cribra orbitalia) and on the external during the third decade (Fig. 2). While no table of the frontal, parietal, and occipital similar phenomenon is reported by Gold- bones (cribra cranii, spongy hyperostosis, 92 R.P MENSFOKTH AND C.O. LOVEJOY ... . .44 18 20 30 25 a 1.0- 35 40 50 45 55 AGE IN YEARS .. O .6. . ... . * . m m . .* :.. : . .. . 0 . . \ .4. 18 2 0 b 25 30 35 40 45 50 55 AGE IN YEARS Fig. 3. a) Relationship between Libben male midshaft femur cortical area (millimeters squared) normalized by femur length (millimeters) and age. b) Relationship between Libben female midshaft femur cortical area (millimeters squared) normalized by femur length (millimeters) and age. 93 SKELETAL CORRELATES OF THE AGING PROCESS females melee 4 20 50 40 30 i 70 00 80 AGE IN YEARS Fig. 4. Patterns of age- and sex-related bone mineral density (grams per square centimeter) normalized by bone width (centimeters) as measured by photon absorptiometry of the distal radius in a sample of 3,515 modern American black, white, and oriental adults. Adapted from Goldsmith et al. (1973) 2 .o 1.o I l- 0 z y .9 1.5 a 3 B t B a .8 1.0 a K(x) 3 s: .7 v) a w .5 t; - .6 A o---o I cortical area fertility i0 i5 30 35 40 45 50 55 AGE IN YEARS Fig. 5. Adult Libben female midshaft femur cortical area (millimeters squared) in relation to the human archetypal fertility curve K(x) and age. Fertility data adapted from Weiss (1973). 94 R.P. MENSFORTH AND ‘2.0. LOVEJOY modern studies showing iron deficiency anemia to be more common in adult females of all ages compared to adult males, but particularly during the child-bearing period (Witts, 1966; Finch, 1968; Yusufji et al., 1973). The bone loss observed for third decade Libben females occurs at a time when nutritional demands for iron and calcium would be markedly elevated, as would the incidence of iron deficiency anemia. Indeed, the latter disorder has been shown to contribute to malabsorption of dietary calcium and iron (leading to bone loss) (Aksoy et al., 19661, impaired immune response (Prasad, 19791, and a more pronounced anemic response (Yusufji et al., 1973). 100 Female Male 80 so % 40 20 - 18-29 30-39 40-55 AGE IN YEARS Fig. 6. Frequency of occurrence of unremodeled porotic hyperostosis for male and female Libben adults. symmetrical osteoporosis)(Angel, 1966; Carlson et al., 1974; Henschen, 1961; Hengen, 1971; El-Najjar et al., 1976). The lesions are a response to erythroid marrow hyperplasia and exhibit a coral, cribriform, or sieve-like porosity associated with marginal hypervascularity and variable osseous tissue hypertrophy. These skeletal changes are common in iron deficiency anemia and a number of other relatively infrequent disease states (Moseley, 1974; Mensforth et al., 1978). The age and sex distribution of unremodeled porotic hyperostosis (the acute phase) in Libben adults is presented in Figure 6. Libben females are more frequently affected than males throughout the adult years. The only significant difference occurs in the third decade where Libben females show a much higher frequency than males of the same age (x2 = 8.55; p < 0.01). This is in accord with The Age of Menopause As previously noted, rapid bone loss is clearly evident in Libben females above the age of 35. To isolate as closely as possible the actual age a t which bone loss begins in the Libben sample, we used the method of serial splines. This method of analysis is most appropriate because it more accurately approximates the age at which bone loss in older Libben females becomes greater than third decade lactationallpregnancy variation and fourth decade normal bone loss. The female sample was progressively divided into only two age classes at 5-year increments (for example, those above and below 30;those above and below 35; and so forth). For each bisected sample a linear regression was generated for bone loss and age for each of the two age groups (above and below the bisection point). The intersection of the two linear regression slopes was then determined graphically. In all cases, the intersection occurred between 44 and 52 years (Fig. 7), with the best fit to the actual population data at 46 years. In a previous investigation by Frommer (19641, 45 years was found to be the modal age of menopause prior to recent secular trends toward later ages (i.e., up to 50 years). Intracortical Porosity Intracortical porosity specifically affecting the endosteal and haversion envelopes of femoral cortical bone is another useful criterion for estimating the age at onset of normal bone loss in males and females and menopausal loss in females. While some degree of porosity is normal in cortical bone at any age, a definite increase in the number and size of resorption spaces in femoral cortex becomes evident in both sexes by the end of 95 SKELETAL CORRELATES OF THE AGING PROCESS 1.o I I- pw .9 4 a a 3 .a a a J t Ba $ v) a -6 w c w p -I 4 5 .5 *- - -* 20-39/40-55 C-. 20-44/43-55 *-,-*20-49/50-53 25 30 s yr. spilnes i Yr. 8PhO8 Yr. 8Pfh08 3s 40 45 50 4 55 AGE IN YEARS Fig. 7. Serial spline analyses of adult Libben female midshaft femur cortical area with age. The splines intersect in the age range of 44 to 52 years, indicating a n approximate age at onset of menopausal bone loss at Libben. the fourth decade (Jowsey, 1960; Atkinson, 1965; Arnold et al., 1966; Van Gerven et al., 1969). Furthermore, females experience a more rapid and extensive femoral intracortical trabecularization, resulting in a sex differential apparent by the end of the fifth decade (Arnold et al., 1966; Atkinson et al., 1962). Figure 8 illustrates a continuum of midshaft femur intracortical porosity a t Libben classified from slight to marked. Resorption spaces follow a n endosteal to periosteal gradient, with greater concentrations of POrosity occurring along the endosteal envelope (Atkinson, 1965). More specifically, intracortical porosity affecting the haversian envelope at midshaft femur first aligns along the anteroposterior axis of the bone and progressively spreads to the mediolateral cortex (Van Gerven et al., 1969; Mensforth and Lovejoy, 1978). The ultimate result of continued endosteal and intracortical bone resorption is a cortical shell (Fig. 8, specimen h). It is worthy of comment here that only two Libben females aged 55 + exhibit cortical shells. Thus, only 1.3% of the Libben femoral sample show dramatic bone loss. Libben intracortical porosity data are summarized in Table 3. The mean age a t onset for slight macroscopically observable concentrations of porosity is 39 years in Libben males and females. This finding is concordant with modern observations reporting a n increase in the number and size of resorption spaces in both sexes by the end of the fourth decade of life (Atkinson, 1965; Arnold et al., 1966). In addition, both Libben and modern data demonstrate a sex differential in femoral intracortical porosity that becomes evident by the end of the fifth decade. Only 20.5%of Libben males over 35 display intracortical porosity compared to 46% of Libben females (x2 = 11.54; significant at the 0.001 level of probability). Furthermore, females show a significantly greater degree of moderate and heavy involvement compared to males (x2 = 7.28; p < 0.05). The results presented above are consistent with termination of the Libben male sample at 45 years. Libben females would thus have greater opportunity to be affected by intracortical porosity. Therefore, the analysis was adjusted to compare males and females lim- 96 R.P. MENSFORTH AND C.O. LOVEJOY Fig. 8. Series of Libben midshaft femur cross sections displaying 1) slight (a, h), moderate (c, d, el, and heavy (f, g) concentrations of intracortical porosity; 2) age progressive endosteal bone resorption; and 3) a cortical shell 01)exhibiting substantial endosteal and intracortical bone resorption. ited to the 35-45 age interval. The results are similar. Here 65% of females and 34% of males displayed intracortical porosity (x2 = 7.62; p < 0.01). Again, females show a significantly greater degree of moderate and heavy involvement compared to males (x2 = 4.19; p < 0.01). Thus, Libben patterns of age- and sex-related intracortical porosity are in full agreement with modern clinical observations that fifth decade females show more rapid and extensive cortical trabecularization at midshaft femur compared to males (Arnold et al., 1966). Additional Evidence From Bone Dynamics There are, in addition, a variety of qualitative indications that the Libben cemetery population contained minimal numbers of individuals over the age of 50 years. The first concerns the phenomenon of periosteal expansion with age, an effect observed primarily in tubular bones of the appendicular skeleton. Periosteal expansion has variously been interpreted to reflect secular trends, continuing bone growth, or increase in sec- tion modulus related to biomechanical demands of aging bone tissue (Smith and Walker, 1964; Smith and Frame, 1965; Garn et al., 1972; Epker et al., 1965; Trotter et al., 1968).Nonetheless, data show that periosteal expansion is common in both sexes over 50 years of age and is usually more pronounced in osteoporotic females (Smith and Walker, 1964). Libben males and females show no discernible trend for midshaft femur periosteal expansion with age and this indicates that few individuals survived beyond 50 years. Second, the typical deceleration of postmenopausal bone loss, which normally occurs following 10 to 15 years of accelerated bone loss (Goldsmith et al., 1973), is not evident in the Libben female sample. Third, our female sample terminates with an average bone loss of 29.8% at the femoral midshaft. From a modern clinical perspective radiographic diagnosis of osteoporosis requires a loss of 30 to 60% bone mineral density. This does not usually become apparent until the age of 60 years when 10 t o 15 years of cumu- 97 SKELETAL CORRELATES OF THE AGING PROCESS TABLE 3. Qualitative assessment of Libben adult midshaft femur intracortical porosity Males Intracortical porosity N Mean age (years) Slight Moderate Heavy Total 14 1 1 16 38.9 46.0 36.0 37.3 % Of % Of total sample (N = 78) porosity sample (N = 16) N 17.92 1.28 1.28 20.51 87.50 6.25 6.25 100.00 15 9 12 36 Females % Of Mean total age sample (years) (N = 78) 39.0 43.9 44.4 42.0 19.23 11.54 15.38 46.15 % Of porosity sample (N = 36) 41.67 25.00 33.33 100.00 TABLE 4. Libben population fracture data' Age (years) dx Person years at risk 5-10 11-15 16-20 21-25 26-30 31-35 36-40 41-45 46-50 f50 117 94 92 63 78 115 154 97 50 33 2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5 47.5 Observed fractures 0 1 3 4 (N) 4 14 20 11 8 6 Total person years at risk Fractures per person years at risk x lo5 293 705 1,150 1,103 1.755 31163 5,005 3,638 2,125 1.568 0 142 260 363 230 443 400 302 376 382 'From Lovejoy and Heiple (1981) lative postmenopausal bone loss has become a significant factor in fracture risk (Arnold, 1960; Meema and Meema, 1963; Baylink et al., 1964; Saville, 1965; Chalmers and Weaver, 1966). FRACTURE PATTERNS Test of Sequence Accuracy Previous discussions have dealt with the question of age bias in the Libben life table. There remains the question of an outside test for proper age sequence. That is, while previous examples showed a clear correspondence between Libben age-related biological events and those in modern populations of known age, we have yet to apply a test to determine the accuracy of the rank ordering of Libben individuals. Such a test is made possible by the collection of data on long bone fractures. Recently, Lovejoy and Heiple (1981) have reported a review of all skeletons and the incidence of long bone fractures with age for the Libben population. All intact bones were observed, normalized to a population size of 1,000, and adjusted for missing (unrecovered) bones (for methods see Lovejoy and Heiple, 1981).Table 4 presents the results of this survey. It was found that virtually no fractures at Libben were the result of interpersonal aggression and that most were clearly the result of accident. If this is the case, then the longer a person lives, the more likely the individual will bear a fracture. We tested this hypothesis by comparing the number of fractures in an age class with the total years of risk for that age class, which yielded a product moment correlation of +0.97. Thus, years at risk accounts for 94% of the variance in the frequency of occurrence of fractures in individual members of the population. We regard this as a useful demonstration that the Libben lustral age sequence is almost free of error. Fractures of Old Age In addition to information regarding proper age sequence, the study of fracture incidence in modern populations provides several criteria from which the probable lifespan of a 98 R.P. MENSFORTH AND C.O. LOVEJOY &-.-A ..-.-A M 50 I Female dlital forearm Male distal forearm Female proximal femur Male proximal femur 500 400 40 Y 8 v) v, 4 8 4 300 0 30 8 P P 200 20 I d 10 /-.- x. - -A-.- 100 __ -. ‘AT” X I 30 40 50 60 70 80 + I AGE IN YEARS Fig. 9. Distal forearm and proximal femur fracture incidence in a contemporary population from England and Wales. Annual fracture rates are represented as the number of cases per million. The left vertical margin shows the distal forearm fracture rates, and the right vertical margin shows the proximal femur fracture rates. Adapted from Buhr and Cooke (1959). skeletal population may be inferred. These include the age progressive shift in sex ratio for fractures of the distal radius (Colle’s fractures) (Bauer, 1960) and the age progressive shift in the total fracture sex ratio (Knowelden et al., 1964).The biological phenomenon of old age or “J” type fracture patterns (illustrated in Fig. 9) are of interest to us here for two reasons. First, old age fractures (primarily distal radius and proximal femur) result almost exclusively from accidental falls precipitated or exacerbated by the degenerative effects of aging (Buhr and Cooke, 1959).Cultural and technological factors therefore play only a minor role in their etiology (Knowelden et al., 1964). Second, though different populations vary in overall fracture rates, they nonetheless tend t o display similar age and sex patterns in old age fractures (Alffram and Bauer, 1962). Figure 10 illustrates the age and sex distribution for distal radius fractures in Libben adults. In general, very few old age fractures have been recovered from Libben. Those observed are given as a percentage of all individuals displaying fractures of the same type. These data show that Libben fracture patterns strongly mimic those reported for modern populations (Buhr and Cooke, 1959). Under 40 years of age, Libben males and females have a comparably low incidence of distal radius fractures. By 45 years, Libben females have experienced a substantial increase, while males retain a low profile throughout adulthood. Figure 9 demonstrates that in modern populations there is a 20-year period between the marked acceleration in distal radius fractures and those of the proximal femur. The precipitous rise in fractures of the distal radius is present a t 99 SKELETAL CORRELATES OF THE AGING PROCESS 100- - &.-A Female distal radius Male distal radius .-.- 80- 60- % P 40- ; I’ I’ 20- 20 I‘ I‘ I ‘ 30 40 50 i 60 AGE IN YEARS 20 -0 AGE IN YEARS Fig. 10. Age- and sex-specific frequency of Occurrence for Libben fractures of the distal radius. Fig. 11. Age-progressive shift in female/male ratio for distal forearm fracture rates in a contemporary population from Malmo, Sweden. Adapted from Bauer (1960). Libben and its age a t onset is identical to that of the modern population. On the other hand, only one Libben female (aged 5 5 + ) (and no males) displayed a transcervical femoral fracture. This strongly suggests that no significant portion of the population survived beyond the age of 60 years. of Libben individuals displaying fractures of the distal radius is 50.5 years. This figure compares well with the age of 52.5 years that this ratio would predict in the population reported by Bauer (1960). A substantially higher or lower ratio would imply that older Libben individuals had been significantly over- or underaged. Fracture Sex Ratio of the Distal Radius Figure 11 presents the age-progressive sex ratio for distal radius fractures reported by Bauer (1960)for a modern European population. The sigmoid curve demonstrates the acceleration and age-specific sensitivity of the sex ratio between 40 and 60 years. Age- and sex-related characteristics of this pattern have repeatedly been observed in modern groups (Buhr and Cooke, 1959; Bauer, 1960; Alffram and Bauer, 1962; Knowelden et al., 1964). The Libben female/male distal radius fracture ratio for individuals aged 40 to 55+ is 4:l. This is represented by the horizontal and vertical intersect in Figure 11. The mean age Total Fracture Sex Ratio Age-related changes in the total fracture sex ratio (TFSR) of a population are another criterion that can be used to assess the accuracy of adult skeletal age assignments. Knowelden et al. (1964) reported fracture incidences for both men and women aged 35 to 85 years in a large contemporary population of the Oxford and Dundee areas of Great Britain. Fracture patterns in these two areas were similar and the TFSR data from both areas have been combined for the present study. They are summarized in Table 5 . The change in TFSR shows a strong relationship with age; Pearson’s r is +0.98. While frac- 100 R.P. MENSFORTH A N D C.O. LOVEJOY tures are more common in males under the age of 50 years, female fractures become progressively more prevalent in successive age categories. The absolute values of the TFSR for any age are obviously determined not only by biological factors (bone loss, body weight) but also by sex-specificrisk factors related to cultural practices and habits. Though most of the fractures seen in the Libben population are clearly of the accidental type, they occurred at very high frequencies (Lovejoy and Heiple, 1982), and female fracture rates must clearly be expected to be greater than those of a modern British group because of the more active role Libben females must have played in the group's economic activities. Nonetheless, the rate of change in the TFSR TABLE 5. Change in total fracture sex ratio (TFSR) with age' Age (years) Femalehale TFSR 0.37 0.68 1.18 1.71 2.02 1.89 40 50 60 70 80 85 'Adapted from Knowelden et al. (1964) is a more direct consequence of the physiological effects of aging. We therefore obtained a linear regression for the TFSR and age from data in Table 5 (TFSR = -1.132 + 0.038(age)),and used this equation to predict the TFSRs for Libben. Because so few burials were available for observation over the age of 45 years, the Libben data were compressed into two groups as presented in Table 6. The observed modern and Libben fracture ratios are listed in Table 7. The critical observation here is that rate of change in TFSR between the two age groups is virtually identical for Libben and the modern group. These data suggest that Libben fourth, fifth, and sixth decade age assignments are reasonably accurate given expected standard error associated with skeletal age criteria and their performance in this time range. Vertebral Compression Fractures Subsequent to age determinations made on the Libben material, a complete survey of spinal pathology was conducted Wacoska, n.d.). Included in this study was a compilation of various forms of vertebral fracture, including senile compression fractures (Lonergan, 1961). These are prevalent in older females suffering postmenopausal osteoporosis. Four vertebral compression fractures (1.7%)were found in Macoska's survey (n.d.1 of all adult Libben vertebral columns. One TABLE 6. Libben adult total fracture sex ratio (TFSR) data Age group (years) 30-44 45-60 Mean age of Libben sample N F(x) % N F(x) % Female/ male TFSR 37.3 49.3 122 13 23 3 18.9 23.1 95 32 14 9 14.7 28.1 0.78 1.22 Male Female TABLE 7. Total fracture sex ratio (TFSR) change with age Age (years) 37.3 49.3 ATFSR Modern' Libben summary age Libben bias revised Coale-Demeny Libben expected 0.29 0.74 +0.45 0.78 1.22 +0.44 0.38 1.17 +0.79 0.91 1.68 +0.77 'Modern rate of change in TFSR was derived from the investigation of Knowelden et al. (1964) of contemporary fracture epidemiology. A linear regression was obtained from their data to generate comparable ratios for the Libhen mean ages sampled. The equation is given here as y = -1.132 t O.O38(x), x = age in years, y = TFSR. Comparable ages for the Libben bias-revised survivorship profile are 39.8 and 60.7, respectively. Comparable ages for the Coale-Derneny Libhen expected survivorship profile are 53.8 and 73.9, respectively. 101 SKELETAL CORRELATES OF THE AGING PROCESS . . ........... Libben Actual Libben Bias-Revised .-._.-.- C-D Age in West 3 Years Fig. 12. Comparison of the original Libben population survivorship curve t o 1) Libben survivorship with adult ages revised for Todd sample 1 bias, and 2) the Coale and Demeny (1966)West level 3 model life table age distribution. occurred in a 31-year-old male who otherwise showed no evidence of osteoporosis and in whom the defect may possibly have been the consequence of trauma or a congenital defect. The three remaining cases were found in females aged 55 +. No senile compression fractures were found in any females aged younger than 55 years, and because traumatic injuries (including fractures) were found to occur at a very high rate a t Libben (Lovejoy and Heiple, 1982),this serves as further evidence that a major portion of the population had died before high rates of bone loss were manifested in the form of spontaneous fractures. DOES THE LIBBEN SURVIVORSHIP CURVE REQUIRE REVISION? We have seen thus far that a variety of developmental, pathological, and physiological phenomena, which have a strong relationship with age, prove strikingly concordant to the Libben survivorship curve. However, two important issues must now be addressed. First, the Libben population was aged by means of the multifactorial (“summary”) method, which has been subjected to a comprehensive blind test for accuracy and bias (Lovejoy et al., 1985).The first test (Todd sample 1) showed a tendency to underage individuals over 40 years. Since the age standards used in this test were the same as those used to age the Libben population, one might expect that the Libben life table was affected by this bias. To test this possibility we carried out a systematic revision of all adult Libben age estimates. A third order polynomial’ was fit to the observed bias by decade found in Todd test 1 (Lovejoy et al., 1985). This polynomial was then used to revise the age of each adult Libben burial, and a new life table and survivorship curve were constructed. The latter is shown in Figure 12 (along with the original curve). Second, to generate conclusions about the social and economic effects of the Libben mortality profile, Howell (1982) compared Libben survivorship data to a series of Coale and Demeny’s model life tables (1966), which, as she points out, are extrapolated “from populations with better mortality conditions” (1982:264).Therefore, for the discussions that follow, we have constructed a third survivorship curve (also illustrated in Fig. 12) based on a Coale and Demeny West level 3 popula‘Polynomial regression adjustment for Todd sample 1 underaging bias. x = Todd sample 1 age; y = Libben bias-revisedage; and y = 5.24859 + 0.5922054(x) + 0.0033306(x2) + 0.00015182(x3). 102 R.P. MENSFORTH AND C.O. LOVEJOY tion with the same total adult percent survivorship as the Libben population; that is, in both the Libben and the Coale and Demeny curves, individuals over 20 years comprise 45% of the total population. A third order polynomial' was then fit to the Libben mortality data such that age-specific survivorship would yield the Coale and Demeny model population. Both the Coale and Demeny and the revised Libben curves were extremely close fits to their respective data. Table 8 presents the results of these two formulas in 5-year age intervals for comparison to the original Libben age estimates. We may now examine whether the Libben bias-revised curve or the Coale and Demeny curve provide a more parsimonious fit to observed biological trends in the Libben population. For comparison with modern biological age standards, age reassignments for each Libben burial have been carried out, and the data have been summarized in Tables 7, 8, and 9, which show that in virtually every category of biological phenomena discussed earlier, the original Libben age assignments are preferable to either the biasrevised or Coale and Demeny adjustments. Therefore, several points require discussion. It may be noted that the bias-revised ages are only minimally different from the original Libben age assignments. For example, there is only a + 1.6 year difference after revision for burials in the 30-40 age category and a +3.4 year difference for those in the 35-45 year age range. No significant change occurs until ages 45 and above are revised. As previously noted, however, in the original Libben age assignments older individuals were simply assigned a n age of 55 years. Because such assignments were rarely required, revision would have virtually no effect on vital statistics of the population (fertility rate, population growth, and soforth). There were, without question, individuals at Libben who survived into relatively old age, but such individuals were few in number. A comparison of the original and revised Libben survivorship curves clearly demonstrates this fact. However, one significant change is possibly required as a consequence of the bias tests carried out by Lovejoy et al. (1985). In that study (Todd sample l),the greatest error occurred in the age determination of late middle-aged males (35-50 years), who tend to demonstrate a quiescence of age-related anatomical change during this period. Thus, it TABLE 8. Comparison of Libben adult age distributions Original Libben summary age 20 25 30 ~~ 35 40 45 50 55 Libben Bias revised Coale-Demeny expected 19.6 24.5 30.1 36.6 44.0 52.5 62.2 73.2 20.0 28.9 38.9 49.2 59.1 67.8 74.7 79.1 In the first column, Libben adult age distribution is simply represented as reported for 5-year age intervals. In column two is the Libben adult survivorsbip adjusted for Todd sample 1 underaging bias in age indicator performance. Column three lists Libben adult age distribution fit to the Coale and Demeny (1966)West level 3 model life table. Age revisions reported here were derived from polynomial regression equations (see text footnotes 1 and 2). is possible that the oldest males a t Libben are underaged slightly (approximately 5 years). However, comparisons between Libben and modern males in the 35-45-year age category show remarkable consistency in all the variables discussed above, and the Libben population contained virtually no males beyond the age of 55 years. In fact, most were dead by age 40, and revision of older males therefore has little effect on the vital statistics of the population. In addition, the comparisons listed in Table 9 show a slight overaging of Libben males under 40 years, although this is quite possibly attributable to sampling error. Since the bias observed by Lovejoy et al. (1985) was substantial in older age categories, it is important to consider why revision of the Libben age distribution did not have a more substantial effect. Two reasons appear paramount. First, most of the bias detected in the study was observed in both sexes over 40 years, and especially over 45 years. Most Libben individuals were dead by these ages, and their original age assignments therefore reflect the general accuracy of the earlier years. Second, and equally important, was the central role played by dental attrition in the original aging of the Libben population. Lovejoy et al. (1985)found 2Polynomial regression distributing adult Libbeu burials according to the Coale and Demeny West level 3 mortality experience. x = Libben original summary age; y = Libben expected age according to Coale and Demeny model; and y = 9.9984 1.00768(x) + 0.095053(x2)- 0.00098(X3). 'Coale and Demeny (1966). menopause in females V. Fracture epidemiology a. Age at onset for dramatic rise in female distal radius fractures b. Mean age at which the F N distal radius fracture ratio is 4:l IV. Age range for onset of Garn (1970);Goldsmith et al. (1973) 111. Age at onset of bone loss a. Total cross-sectional area and bone mineral density b. Age at onset for increase in intracortical porosity 27.2 35.0 35.0 35.0 35-40 35-40 35-40 35-40 35-40 35-40 45-52 Stage 3 Female Male Total Female Male Total Female Male Total 40-50 52.5 Bauer (1960) 1-2 21.6 24.5 Stage 0 Stages Buhr and Cooke (1959); Bauer (1960) Frommer (1964) (1966) Jowsey, (1960);Atkinson (1965);Arnold et al. Goldsmith e t al. (1973) 11. Peak bone mineral density and peak cortical area Reference Todd and D'Errico (1928) (mean ages for black and white males) I. Stage of clavicular fusion Biological age criteria Age (years) 50.5 45.0 26.7 32.5 37.5 35.0 37.5 42.5 40.0 38.9 39.0 39.0 44-52 22.3 25.4 Libben summary age 63.2 52.5 26.3 33.2 40.1 36.6 40.1 49.7 44.0 42.3 42.4 42.4 51-64 21.8 24.9 Libben bias revised 74.7 67.8 32.2 44.0 54.2 49.2 54.2 64.1 59.1 56.9 57.i 57.0 66-76 23.9 29.7 C-D' Libben expected TABLE 9. Age and sex relationships for biological phenomena as they occur in modern populations and as reported for the Libben skeletal population 104 R.P. MENSFORTH AND C . O . LOVEJOY almost no bias in this age indicator, despite having been applied to a population with minimal and irregular attrition. At Libben there was consistent, progressive dental wear (Lovejoy, 1985).This clearly provided reasonably accurate and unbiased age determinations for the majority of the older members of the population. That such is the case is demonstrated by the very high loading dental wear achieved in the original intercorrelation matrix used to generate the principal components weightings for summary age (Lovejoy et al., 1977). One additional point is worthy of mention. The tests carried out on Todd sample 1 (Lovejoy et al., 1985) also indicated a slight oueraging of individuals in the beginning of the third decade. It may be noted here that the bias-revised Libben ages, as seen in relation to the onset and progress of clavicular fusion, provide a slightly better fit than the original age assignments. The error involved is only minimal, but results of the correction due indicate the sensitivity of tests for bias in this age range. Turning to the revised age distribution which results from Coale and Demeny modeling, we find the results far more dramatic. New age assignments made after revision by this model are completely inconsistent with known biological phenomena. Even in the earliest years, as reflected in clavicular fusion, the Coale and Demeny ages are greatly in excess of those associated with biological events in a modern population, and this de. monstrable discordance becomes magnified in older age groups (Table 9). DISCUSSION that bone loss and porotic hyperostosis associated with pregnancy and lactation occur at the peak of the modal fertility curve; that menopause occurs a t a n almost identical age in the archaeological sample; and that a variety of advanced degenerative changes of osteopenia, old age fractures, and age-progressive fracture ratios are absent from the Libben population, which was originally reported to contain few individuals over 55 years of age. Last, we must address the basic question of rate of change in anatomical indicators of skeletal age a t death. If these indicators supposedly age at different rates in earlier populations, what are the supposed differences? On what a priori grounds would, for example, metamorphosis of the pubic syphysis occur a t higher or lower rates in an archaeological sample? It is difficult to generate a hypothesis to explain why metamorphosis of the pubic symphysis should be delayed. It is possible to presume that higher levels of musculoskeletal stress would accelerate changes in the pubis, but this assumes that pubic changes are dependent on and sensitive to stress. No such demonstration has ever been presented, save for distortions produced by successive pregnancies, which are known to occur in modern specimens, thus allowing appropriate age correction (Gilbert and McKern, 1973; Suchey, 1979; Meindl et al., 1985). Sex differences in changes of the auricular surface are especially important in this regard. Both Lovejoy et al. (1985) and Sashin (1930) found that female auricular surfaces change less rapidly than those of males, despite the traumatic effects of pregnancy wherein the sacroiliac joint possibly is subjected to the same estrogen-induced changes as the pubis. In this case stress does not accelerate alteration of the age indicator whatsoever. The paramount feature of the pubis for the third and early part of the fourth decades is the “delayed epiphysis” in the form of the ventral rampart. For what reason would we expect musculoskeletal stress to alter the metamorphosis of this feature? It is, in fact, quite likely that differences in rate between modern and skeletal populations are only minor and that classic anatomical indicators of age are equally applicable to prehistoric and modern populations. Two primary criticisms are frequently leveled at mortality profiles derived from essentially complete archaeological sites. The first is that skeletal age indicators are not accurate. That question is considered separately (see Lovejoy et al., 1985). The second criticism is that even given the accuracy of age indicators for modern populations, we have no assurance that the rates of change in these various indicators are the same in archaeological and modern test populations from which they were derived. This paper has proposed several tests of internal consistency that can be applied to a population for comparison with similar CONCLUSION events in modern (“known age”) populations. We regard the paper presented by Howell The results appear definitive. We have found that clavicular fusion patterns show the same (1982) as a major contribution to the study of age distribution as in modern populations; paleodemography. For far too long paleode- SKELETAL CORRELATES OF THE AGING PROCESS mographers have failed to consider the social and economic consequences of the survivorship and mortality data generated by the study of skeletal populations. The present review of the Libben mortality data, furthermore, confirms the dramatic nature of Howell’s social and economic projections. The accuracy and bias tests presented here (as well as additional ones yet to be devised) should be applied to other major skeletal samples in a n effort to determine whether the pronounced differences in mortality between those seen a t Libben and those seen in extant primitive groups are localized or truly represent a major demographic shift that occurred following development of large urban population centers whose contact effects became worldwide. Not until this problem is resolved will we have a complete understanding of the demographic history of human populations and the nature of those forces of natural selection that operated on human populations of the late Pleistocene. 105 Baylink, DJ, Vose, GP, Dotter,WE, and Hurxthal, LM (1964) Two methods for the study of osteoporosis and other metabolic bone disease. Part 11. Vertebral bone densitometry. Lahey Clin. Foundation Bull. 13:217224. Bocquet-Appel, JP, and Masset, C (1982) Farewell to paleodemography. J. Hum. Evol. 11:321-333. Buhr, AJ, and Cooke, AM (1959)Fracture patterns. Lancet 1~531-536. Carlson, DS, Armelagos, GJ, and Van Gerven, DP (1974) Factors influencing the etiology of cribra orbitalia i n prehistoric Nubia. J. Hum. Evol. 3~405-410. Chalmers, J , and Weaver, J K (1966) Cancellous bone: Its strength and changes with aging and an evaluation of some methods for measuring its mineral content. 11. An evaluation of some methods for measuring osteoporosis. J. Bone Joint Surg. 48At299-308. Coale, AJ, and Demeny, P (1966) Regional Model Life Tables and Stable Populations. Princeton: Princeton University Press. Cohn, SH, Vaswani, A, Zanzi, I, and Ellis, K J (1976) Effect of aging on bone mass in adult women. Am. J. Physiol. 230:143-148. Davis, ME, Strandjord, NM, and Lanzl, LH (1966) Estrogens and the aging process. The direction, prevention, and retardation of osteoporosis. JAMA 196~219-224. Dewey, JR,Armelagos, GJ, and Bartley, MH (1969a) Femoral cortical involution i n three Nubian archaeological populations. Hum. Biol. 41:13-28. ACKNOWLEDGMENTS Dewey, JR, Bartley, MH, and Armelagos, G J (1969b) Rates of femoral cortical bone loss in two Nubian popThe research reported in this paper was ulations. (Utilizing normalized and non-normalized funded by the National Science Foundation, data). Clin. Orthop. 6561-66. Division of Behavioral and Neural Sciences, El-Najjar, MY, Ryan, DJ, Turner, CG, and Lozoff, B (1976) The etiology of porotic hyperostosis among the award No. BNS-77-07958. The Cleveland prehistoric and historic Anasazi Indians of SouthwestMuseum of Natural History generously ern United States. Am. J. Phys. Anthropol. 44:477loaned skeletal material from the Hamann488. Todd Collection. Epker, BN, Kelin, M, and Frost, HM (1965) Magnitude and location of cortical bone loss in human rib with LITERATURE CITED aging. Clin. Orthop. 41:198-203. Finch, CA (1968) Iron deficiency anemia in the United Aksoy, M, Camli, N, and Erdem, S (1966) RoentgenoStates. JAMA 203:407-412. graphic bone changes in chronic iron deficiency aneFrommer, D J (1964)Changing age of the menopause. Br. mia (a study of twelve patients). Blood 27:677-686. Med. J. 2~349-351. Alffram, P, and Bauer, GCH (1962) Epidemiology of fracGallagher, JC, and Riggs, BL (1978) Current concepts in tures of the forearm. A biomechanical investigation of nutrition. (Nutrition and bone disease.) N. Engl. J. bone strength. J. Bone Joint Surg. 44A:105-114. Med. 298:193-195. Angel, J L (1966) Porotic hyperostosis, anemias, malarias, and marshes in prehistoric Eastern Mediterra- Garn, SM (1970) The Earlier Gain and Later Loss of Cortical Bone. Springfield, IL: Charles C. Thomas. nean. Science 153:760-763. Garn, SM, Frisancho, AR, Sandusky, ST, and McCann, Armelagos, G J (1969) Disease in ancient Nubia. Science MB (1972)Confirmation of the sex difference i n contin163t255-259. uing subperiosteal bone apposition. Am. J. Phys. AnArnold, JS (1960) Quantitation of mineralization of bone thropol. 36t377-380. as a n organ and tissue in osteoporosis. Clin. Orthop. Gilbert, BM, and McKern, TW (1973)A method for aging 17r167-175. the female 0s pubis. Am. J. Phys. Anthropol. 38:31-38. Arnold, JS, Bartley, MH, Tont, SA, and Jenkins, JP Goldsmith, NF, and Johnston, J O (1975) Bone mineral: (1966) Skeletal changes in aging and disease. Clin. Effects of oral contraceptives, pregnancy, and lactaOrthop. 49t17-38. tion. J. Bone Joint Surg. 57A:657-668. Atkinson, PJ (1965) Changes in resorption spaces in Goldsmith, NF, Johnston, JO, Picetti, G, and Garcia, C femoral cortical bone with age. J. Pathol. Bacteriol. (1973) Bone mineral in the radius and vertebral osteo89:173-178. porosis i n a n insured population. J. Bone Joint Surg. Atkinson, PJ, Weatherell, JA, and Weidmann, SM (1962) 55A:1276-1293. Changes in density of the human femoral cortex with Hengen, OP (1971) Cribra orbitalia: Pathogenesis and age. J. Bone Joint Surg. 44B:496-502. probable etiology. Homo 2257-75. Atkinson, PJ, and West, RR (1970) Loss of skeletal calHenschen, F (1961) Cribra cranii, a skull condition said cium i n lactating women. J. Obstet. Gynecol. 77:555to be of racial or geographical nature. Pathol. Micro560. biol. (Basel)24~724-729. Bauer, GCH (1960) Epidemiology of fracture i n aged Horsman, A, Nordin, BEC, Gallagher, JC, Kirby, PA, persons. Clin. Orthop. 17t219-225. 106 R.P. MENSFORTH AND C.O. LOVEJOY Milner, RM, and Simpson, M (1977) Observations of Mensforth, RP, and Lovejoy, CO (1978) Femoral cortical sequential changes in bone mass in postmenopausal bone involution in a Late Woodland skeletal population: Epidemiological and biomechanical observations women: A controlled trial of estrogen and calcium therapy. Calcif. Tissue Res. 22:217-224. [Abstr.]. Am. J. Phys. Anthropol. 48:419. Howell, N (1982)Village composition implied by a paleo- Mensforth, RP, Lovejoy, CO, Lallo, JW,and Armelagos, demographic life table: The Libben site. Am. J. Phys. GJ (1978) The role of constitutional factors, diet, and infectious disease in the etiology of porotic hyperostoAnthropol. 59t263-269. sis and periosteal reactions in prehistoric infants and Jowsey, J (1960) Age changes in human bone. Clin. Orchildren. Med. Anthropol. 2:l-59. thop. 17:210-217. Knowelden, J, Buhr, AJ, and Dunbar, 0 (1964) Incidence Moseley, JE (1974) Skeletal changes in the anemias. Semin. Roentgenol. 9:169-184. of fractures in persons over 35 years of age. Br. J. Prev. Soc. Med. 18t130-141. Prasad, JS (1979) Leukocyte function in iron-deficiency anemia. Am. J. Clin. Nutr. 32550-552. Lonergan, RC (1961) Osteoporosis of the spine. Am. J. Nursing 61:79-81. Sashin, D, (1930) A critical analysis of the anatomy and the pathologic changes of the sacro-iliac joints. J. Bone Lovejoy, CO, (1985) Dental wear in the Libben PopulaJoint Surg. 12391-910. tion: Its functional pattern and role in the determination of adult skeletal age at death. Am. J. Phys. Saville, PD (1965) Changes in bone mass with age and Anthropoi. 68:47-56. alcoholism. J. Bone Joint Surg. 47A:492-499. Smith, RW, and Frame, B (1965) Concurrent axial and Lovejoy, CO, and Barton, TJ (1980) A simple, rapid appendicular osteoporosis. N. Engl. J. Med. 273:73-78. method of obtaining geometrical properties from sections or laminograms of long bones. J. Biomech. 13:65Smith, RW, and Walker, RR (1964) Femoral expansion 67. in aging women: Implications for osteoporosis and fractures. Science 145156-157. Lovejoy, CO, and Burstein, AH (1977) Geometrical properties of bone sections determined by laminography Sorenson, JA, and Cameron, JR (1967)A reliable in vivo measurement of bone mineral content. J. Bone Joint and physical section. J. Biomech. 1Ot527-528. Surg. 49A:481497. Lovejoy, CO, and Heiple, KG (1981) The analysis of fractures in skeletal populations with an exampie from the Stevenson, PH (1924) Age order of epiphyseal union in man. Am. J. Phys. Anthropol. 7t53-93. Libben site, Ottawa County, Ohio. Am. J. Phys. Anthropol. 55:529-541. Stout, S, and Simmons, DJ (1979) Use of histology in ancient bone research. Yearbook Phys. Anthropol. Lovejoy, CO, Meindl, RS, Mensforth, RP, and Barton, T J 22:228-249. (1985) Multifactorial determination of skeletal age at death A method and blind tests of its accuracy. Am. J. Suchey, JM (1979) Problems in the aging of females Phys. Anthropol. 68t1-14. using the 0s pubis. Am. J. Phys. Anthropol. 51:467470. Lovejoy, CO, Meindl, RS, Przybeck, TR, Barton, TJ, Heiple, KG, and Kotting, D (1977) Paleodemography of Todd, TW, and D’Errico, J (1928) The clavicular epithe Libben site, Ottawa County, Ohio. Science 198:291physes. Am. J. Anat. 41t25-50. 293. Trotter, M, Peterson, RR, and Wette, R (1968) The secular trend in the diameter of the femur of American Macoska, J (n.d.1 A survey of spinal morphology and whites and negroes. Am. J. Phys. Anthropol. 28:65-74. pathology in the Libben population. Kent State University Honors Thesis; Kent, Ohio, pp. 1-45. Van Gerven, DP, Armelagos, GJ, and Bartley, MH (1969) Martin, DL, and Armelagos, GJ (1979)Morphometrics of Roentgenographic and direct measurement of femoral compact bone: An example from Sudanese Nubia. Am. cortical involution in a prehistoric Mississippian population. Am. J. Phys. Anthropol. 31:23-38. J. Phys. Anthropol. 51t571-578. Meema, HE, and Meema, S (1963)Measurable roentgen- Weiss, KM (1973) Demographic models in anthropology. ologic changes in some peripheral bones in senile osAm. Antiquity 38(2). teoporosis. J. Am. Geriatr. Soc. 11:1170-1182. Witts, L J (1966) Anemia as a world health problem. Meema, S, and Meema, HE 11976) Menopausal bone loss International Society of Hematology, 11th Congress of the University of Sydney, Plenary Sessions, pp. 85and estrogen replacement, Israel J. Med. Sci. 12t601606. 102. Meindl, RS, Lovejoy, CO, Mensforth, RP, and Walker, Yusufii, D, Mathan, VI, and Baker, SJ (1973) Iron, folate, RA (1985) A revised method of age determination usand vitamin BIZ nutrition in pregnancy: A study of ing the 0s pubis with a review and tests of accuracy of 1000 women from southern India. WHO Bull. 48t1522. other current methods of pubic symphyseal aging. Am. J. Phys. Anthropol. 68:29-45.