Buscar

diferenças entre ossos masculinos e femininos - moderno e antigo

Prévia do material em texto

Christopher. Ruff 
Department of Cell Biology and 
Amtory. Johns Hopkins l’ni~errip 
School of Medicine, 725 N. Wolfe 
S&et. Baltimore, .UD 21’205. 
l..S..i. 
Received 10 March 1987 
Revision received 10 July 1987 
and accepred 11 September 1987 
Publication datr,January 1988 
Kpwords: sexual dimorphism. 
biomcchanics, lower limb 
skeleton. 
Sexual dimorphism in human lower limb 
bone structure: relationship to subsistence 
strategy and sexual division of labor 
Cross-sectional geometric properties of the human femur and tibta are 
compared in males and females in a number of recent and archaeological 
population samples extending back to the Middle Paleolithic. There is a 
consistent decline in sexual dimorphism from hunting-gathering to 
agricultural to industrial subsistence strategy levels in properties which 
measure relative anteroposterior bending strength of the femur and tibia in 
the region about the knee. This trend parallels and is indicative of 
reductions in the sexual division of labor, in particular differences in the 
relative mobility of males and females. Sexual dimorphism in medialateral 
bending strength near the hip shows no consistent temporal trend, probabb 
reflecting relatively constant sex differences in pelvic structure related to tht 
requirements of childbirth. Upper and Middle Paleolithic samples arr 
indistinguishable in terms of sexual dimorphism from modern huntcr- 
gatherers, suggesting a similar sexual division of labor. ‘l‘hr results 
tllustrate the utility of cross-sectional geometric parameters of long horn 
diaphyses in reconstructing behavioral ditlerencrs within and between past 
populations. Some variations in the accuracy of sexing techniques based on 
diaphyseal measurements of the lower limb long bones may also hc 
explained by these behavioral and structural factors. 
Journal of Human G~olur~on j 1987) 16, 391-416 
Introduction 
Thus, where for Anatomy there seemed to exist hitherto a relatively simple proposition there is 
discovered on close scrutiny a whole web ofconditions. As every other feature, so the shape ofthe 
body of a free bone has not merely a range of its normal variation, but also a tendency to 
differentiate into a number of different types, and presents a whole life history of changes 
resulting mainly from the mechanical influences that act on the bone. 
All ofwhich opens a vast and alluring new field for the human biological science of the future 
(Hrdlicka, 1934a, p. 39). 
Differences in sex roles in human societies have long interested anthropologists, particularly the 
demarcation of economic behavior along sex lines. Yet there has been practically no interest 
among paleoanthropologists as to the relationship between a rigid sexual division of labor and 
any contingent biological effects. . If specific physiques have their economic advantages in 
modern society, what sort of inferences can be made concerning morphological attributes in 
prehistoric populations? (Frayer, 1980, p. 399). 
Despite the early optimism expressed by Hrdlicka in the passage quoted above, thr 
succeeding decades have not produced a dramatic increase in our understanding of the 
relationship between environmental forces and variation in bone shape. Several factors 
have contributed to the general lack of progress in this area: greater than expected 
complexity in the “laws” of bone architectural change in response to altered environmental 
conditions (e.g., Lanyon, 1982); a continued emphasis in skeletal studies on relatively 
simple, usually linear measurements and indices which limit structural interpretations; 
and to some extent a lingering typological and essentially descriptive approach 
(paradoxically well exemplified by Hrdlicka himself) with little consideration of functional 
correlates and relevance of findings to more general issues of biological adaptation 
(Lovejoy et al., 1982). 
0047- 2484/87/05039 I + 26 $03,00/O 0 1987 .4cademic Press I.imited 
392 (:. RI-FF 
The history of the study ofsexual dimorphism in human postcranial skeletal dimensions. 
in particular lower limb long bones, provides a good example ofthis last point. M’hile s(‘s 
differences in femoral and tibia1 morphology have long been noted (Hrdlicka, 1898, 19346; 
Parsons, 1914; Holtby, 1917; Pearson & Bell, 1919; Martin, 1928). rarely has an attempt 
been made to provide a functional interpretation of these differences. The same can be said 
of the various techniques developed for determining sex from lower limb long bone remains 
(e.g., Black, 1978; Iscan & Miller-Shaivitz, 1984; MacLaughlin & Bruce, 1985). 11s 
Armelagos and coworkers ( 1982, pp. 3 18-319) have noted in an historical review of the 
field of skeletal biology: “Although a methodology for interpreting the functional 
significance ofsexual dimorphism [in postcranial dimensions] could have been developed. 
it was not. Physical anthropologists instead created statistical techniques for determining 
sex without concern for the functional factors that lead to these differences.” 
Those more recent studies which have examined relationships between postcranial 
sexual dimorphism and environmental factors have tended to concentrate on measures 01 
overall size (e.g., Frayer, 1980, 1984; H amilton, 1982; also see Armclagos Br Van Gcrven, 
1980). However, size alone, whether of the body as a whole (e.g., stature) or an individual 
bone (e.g., length, circumference), is a relatively general and imprecise indicator ot 
biological adaptation. For example, both nutritional and mechanical-behavioral factors 
may contribute to an increase or decrease in size or sexual dimorphism in size (Stini, 1974; 
Gray & Wolfe, 1980; Ruff el al., 1984). Changes in bone geometry, or shape, may be more 
informative with respect to specific environmental adaptations, in particular, adaptation to 
specific mechanical forces which are indicative of functional USC and thus behavioral 
differences (Lovejoy et al., 1976; Ruff & Hayes, 1983a,b; Ruff e/ al., 1984). 
In a previous study of skeletal material from the Pecos Pueblo, New Mexico 
archaeological site (Ruff & Hayes, 19836), several sex-related differences in lower limb 
bone structure were noted. These appeared to reflect different mechanical forces, or 
loadings, placed upon male and female lower limb bones during life. Specifically, male 
lower limb bones were adapted for relatively greater anteroposterior bending, and female 
lower limb bones for greater mediolateral bending. Two possible explanations were 
proposed for this sex difference: (1) greater participation ofmales in activities that required 
running, which should produce high A-P bending loads in the lower limb, especially 
around the knee; and (2) relatively greater pelvic breadth and consequent higher M-L 
bending loads about the hip in females. 
Th e present study expands this investigation to include a large modern U.S. population 
sample. Sexual dimorphism in lower limb bone structure of this sample is found to be 
markedly different in some respects from that of Pecos Pueblo. Using a comparative 
approach, the relationship between bone structural differences and changes in suhsistencc 
strategy and sexual division of labor is explored in these and other population samples. 
The specific structural characteristics examined here are cross-sectional geometric 
properties of the femoral and tibia1 diaphyses. In particular, variation in second moments 
of area of the compact bone cortex are investigated. Second moments of area, also referred 
to as area moments of inertia, are properties used by engineers to measure the bending and 
torsional rigidity and strength of structural beams, and reflect both the area of a cross 
section and the area1 distribution of material in the section(Timoshenko & Gerc, 1972). 
The precise meaning and interpretation of these section properties in a biomechanical 
context has been discussed in detail elsewhere (Lovejoy el al., 1976; Ruff & Hayes, 1983a). 
In the present study, only two cross-sectional geometric properties are examined-the 
SEX DIFFERENCES IN SKELETAL STRt:CTlrRE 39’ 
ratio ofmaximum to minimum second moments of area. Z,,,,,/Z,i,,, and the ratio ofsccond 
moments of area about the x (M-L) and y (A-P) axes, I,/!,. The first ratio mrasures the 
relative maximum bending strength of the bone at that cross section, while the second riiti(J 
measures the bending strength in the A-P plane rclativc to the R/I-L plane.’ Both ratios are 
indices of cross-sectional “shape”, since they reflect relative distribution of bone about 
perpendicular axes. For example, as illustrated in Figure 1. an I,//,. ratio of 14 indicates an 
equivalent distribution of bone about x and Y axes, while ratios ,greatrr or less than 1.0 
indicate a direction of greater bending strength in the A-P or M-L planes, respectively. 
The same reasoning applies to ImnxlZmi,,, except that this ratio will always be greater than 
or equal to 14, and I,,, and I+, may be oriented in anv direction. though al\~;rvx 
perpendicular IO each other. 
I& =I.0 I*/Iy~2~0 lK’IY :o?j 
Figure 1. Efkcts ofvarying cross-sectional shape on IT/I, ratio. In the present study. the x axis rcprrscn~\ 
the mediolateral axis. thus: greater values of /,/I, indicate relativeI! greater .-I-P bending strengtll. 
I,/!,. and f,,,/I,,,,, ratios were chosen for analysis fbr severa reasons. First. as noted 
earlier, structural characteristics which quantify differences in bone shape are more likely to 
reflect differences in specific mechanical loadings and behavioral patterns. These two 
ratios represent simple and readily understandable indices ofcross-sectional shape which 
nevertheless have direct mechanical relevance. Second, the use of ratios avoids 
complexities associated with the standardizing of raw data for body size differences 
bctwren the sexes or population samples (see Ruff & Hayes. 1983a; Ruff, 1984). Finally, 
these ratios are in some ways analogous to and can be compared with simple linear rxternal 
breadth ratios which have been measured in many skeletal samples, e.g., the “pilasteric” 
and “platycnemic” indices of the femur and tibia, respectively- (Martin, 1928). Thus. this 
allows inclusion ofa much larger number ofsamples in a subsequent comparati\,e analysis, 
broadening both the data base and the generality of results, as well as increasing the 
applicability offindings to samples where engineering cross-sectional propcrtirs cannot be 
measured. 
A comparison of sexual dimorphism in lower limb bone cross-sectional geometry in the 
Pccos and modern U.S. samples is presented first. Several other population samples with 
available cross-sectional geometric.data are then included in the comparison. Linear 
external breadth ratios for other modern and archaeological samples are examined and 
interpreted in li,ght of these findings. Finally, again usin,g external breadth data, sexual 
’ Note that properties measured about one axis indicatt, strmgrh in ;I plane of bending- perpndimlnr IO tht asi\. 
394 I:. RIIE‘F 
dimorphism in lower limb bone structure is dcmonstratrd to occur at least as c,arl>, ‘is the, 
Upper and Middle Paleolithic. 
Cross-sectional geometry-Pecos and modern U.S. 
The proveniencc and method of analysis of the Pecos Pueblo skeletal sample has been 
described at length e&where (Ruff & Hayes, 1983a). Briefly, Pecos Pueblo was a large late 
prehistoric and early historic settlement in north-central New Mexico with an agricultural 
subsistence base. The study sample includes 119 adults, 59 males and 60 females, each 
represented by a femur and tibia. The average age at death within each sex is about 37 
years. Cross-sectional geometric properties wcrc determined by sectioning and direct 
measurement at 11 locations: 20, 35, 50, 65, and 80% of femoral and tibia1 length, 
measured from the distal end, and the mid-femoral neck. Only the 10 diaphyseal locations 
are included in the present study. 
The modern U.S. sample was obtained from cadavers used in anatomy classes, plus a 
few specimens from a bone bank. All individuals in the sample were adult Whites. The 
complete sample includes 103 fcmora and 99 tibiae, with approximately 80% concordance 
between bones (i.e., matched specimens from the same individual) (Ruffet al., 1986; Ruff & 
Hayes, in press). In order to ensure age comparability with the Pecos sample, only younger 
individuals between 20 and 59 years arc included in the present study, reducing the sample 
size to 47 femora (23 male, 24 female) and 42 tibiae (19 male and 23 female). The average 
age at death of males is 42 years, of females, 43 years. Although a total of 20 cross section 
locations were originally measured in this sample, only the 10 locations corresponding to 
the diaphyseal sections included in the Pecos sample are analyzed here. Section properties 
were determined using the same techniques as those employed for the Pccos sample. 
Summary statistics for Zm,,xllmir, and ZJZ, arc presented for the Pccos sample in Table 1 
Table 1 Sex differences in I,., /I,;, and ZJZ, of femoral and tibia1 cross sections in the Pecos Pueblo 
sample 
~rd~t,,,~~ I,ll,' 
M&S Fern&!, MACS Fmlalcs 
(XI-F)/F ill-F]!E 
Section &an SE Mean SE x 100 XIean SE hICar SE x 100 
‘Lb. 20%” 1425 0.028 I.406 0.2U4 I.& 1.033 O,(J’L’, 14J27 w21 0.6 
35% 2.510 0.065 2.28 1 0.053 10,1** 1.41; 0.043 I.422 0.037 -0.3 
50% 3.153 0.077 2.704 WO6i ,6.7*** 1,942 0059 I.81 I 04 47 7.2 
65% 3.374 0.076 2.926 0.077 1.5.0*** 2.338 0.060 2,128 0.05 I 9.9** 
80% 2.768 0.066 2453 0,068 13.1*** I.954 0~0.5 1 I.758 001& I I.I** 
Fem. 20% 1,333 0,024 I.552 0.028 -I&l*** W826 0.0 17 0.684 NJ I L’ L’().8*** 
35% 1.356 0.029 1,177 0.018 ,5.2*** I.211 0~02.5 1.069 0.0 10 ,3.3*** 
50% 1,478 0,028 1.287 0.022 I-&,7*** 1.143 0~026 I+&7 (JW? 9.2** 
65% 1.286 0.02 1 1,355 WJ25 -.5.1* 0.964 0.022 Il.947 (1.02 1 I.8 
80% 1,898 0,036 1.926 OG36 - 1.5 0~902 0.022 U.862 0,023 .4$ 
1 Maximum/minimum second moments of arra: maximum/minimum hrndinq strerqtb 
1 Second moments of at-Pa about .x/ahout_r axrs: A-I’D-f, bending strength. 
i Percent of bone length from the distal end. 
* P < 0.05: ** P < 0.01; *** P < 0.001: t-trots bctwwn m&s and frmates. 
SEX DIFFEREh’CES IN SKELETAL STRL’(:Tl.RE 395 
and the modern U.S. sample in Table 2. Mean data for males and females and the 
percentage difference between the sexes, [(male-fcmale)/fcmale] X 100, are given. 
Statistical significance of sex differences within groups is evaluated using f-tests. l’o 
facilitate direct comparisons between groups. percentage differences between the sexes in 
the ZJZ, and ZnI.<y/Zrnir\ ratios are also plotted for each sample in Fi,gurcs 2 and 3. 
respectively. 
Table 2 Sex differences in Z ,,&Imin and Z.Jl, of femoral and tibia1 cross sections in the modern U.S. 
White autopsy sample’ 
'l'ih. 20% 143-4 0,052 I.326 OG45 &I 
35% 1.919 0.1 I2 1.775 0.081 8. I 
50% 2.116 0.111 2.036 0.078 i,.4 
65% 2,428 0, 103 2.265 0.065 7.2 
80% 2.391 WO81J 2.198 lJ.073 8.9 
FCT. 20% I.465 0~033 1.666 oG49 -1_?1** IJ.71 I (PO16 0.657 lJW7 8.2 
35 % 1.175 0.019 1.270 0.025 -7 j** I .I1211 ll.022 0.988 UWVI 1.2 
50% 1 .‘,!43 
I .k 
0,026 1.277 0.014 I.3 I ~L’O6 lJ.035 I.231 0047 - "- 1 
65 % 1).03fJ I.363 0.051 - 8.3 I.153 (W.39 I .28j O-O.5 7 -11.1 
80% I~'~53 0.039 1,399 0.034 -'j.i 1JW4 __ (HJ','~ lm77 WWI _-'(:)* 
’ Smtion proprrties nnd locations as in Tahlr I. 
* I' < WS; ** I' < O.(II: /-tests brtween malrs and fcrmles. 
As shown in Table 1 and Figures 2 and 3, in the Pecossample the sex difIerencr in 
femoral and tibia1 cross-sectional shape is highly significant at most cross section locations. 
As discussed previously (Ruff & Hayes, 19836), these shape differences indicate that Pecos 
males were adapted for relatively greater A-P bending loads throughout most of the lower 
limb, but particularly in the region around the knee, an area ofprobable high A-P bending 
moments in viva. This is shown most clearly by sex differences in the I,/!, ratio (Figure 2 j. 
The same sex difference is also reflected in the Z,.‘,/Z,,iI, ratio (Figure 3), although results 
appear more complex because of changes in orientation of greatest and least bending 
strength at different section locations (Ruff & Hayes, 1983a). In the tibia, the orientation of 
greatest bending strength is always posteromedial-anterolateral (with respect to the knee 
joint), coming closest to directly A-P in the proximal halfofthe shaft, the region ofgreatest 
difference in I ,n,JZmirr between the sexes. In the femur, the direction of greatest bending 
strength rotates along the length ofthe shaft, being more M-I, oriented in the proximal and 
most distal diaphysis and more A-P oriented in the mid-distal region of the shaft. Thus, in 
keeping with the relatively greater A-P bending strength of males and greater M-1, 
bending strength of females, ZmaxlZnnr, is greater in males in the mid-distal femur and 
greater in females in the distal and proximal femur (although not significantly in the most 
proximal femoral section). 
In contrast to these findings for the Pecos sample, the modern U.S. sample shows much 
less sexual dimorphism in cross-sectional shape of the femur and tibia at most locations 
(Table 2 and Figures 2 and 3). Concentrating first on sex diKerences in ZJZ,. (Figure 2). 
396 (a. RC’FE 
;2;,;-..._______~~_ 
VI ‘v 
E \ 
z \ 
L 
a” \ l 
-lO- ‘v- -7 
t1 1 I 1 I I I I I I 
20% 35% 50% 65% 00% 20% 35% 50% 65% 80% 
Tlbla FWltN 
Figure 2. Sexual dimorphism in /,/I,, or relative A-P to M-L bending strength of the femur and tibia m 
the Pecos (0) and U.S. White autopsy (0) samples. Cross section locations are measured as percent 
distance from thedistal end ofthe bone. Percent sex differencein IX/I,: [(Male-Female)/Female] X 100. 
Positive values indicate relatively greater A-P bending strength &I males (or greater M-L bending 
strength in females); negative values indicate relatively greater M-L bending strength in males (or 
greater A-P bending strength in females). * P < 04.5. ** P < 0.01. *** P < 0.001. t-tests betwrcn malrs 
and females. 
Tiblo FWlUC 
Figure 3. Sexual dimorphism in I,,,,lI,,,,,, or relative maximum to minimum bending strength of thy 
femur and tibia in the Pecos (0) and U.S. M’hite autopsy (v) samples. Positive values indicate relativeI> 
less equal distribution of bending strength in all planes in males (less circular shape); negative valur> 
indicate that this is more characteristic of females. Section locations. calculation of prrccnt scs 
difference, and significance levels as in Figure 2. 
SEX DIFFERENCES Ih’ SKELETAL STRL’CTtrRi? 397 
modern U.S. males show some evidence of relatively greater A-P bending strength in the 
proximal tibia and distal femur, but the sex difference is at most half that for the Pccos 
sample, and does not reach statistical significance at any location. In the mid-proximal 
femur, modern U.S. males actually exhibit relatively greater ‘W-L bending strength than 
females, reaching significance in the most proximal section. Interestingly, the pattern of’ 
charge in sex differences along the length of the femur is very similar in the Pccos and 
modern U.S. samples, as if the entire Pecos pattern of sex differences were simply 
transposed downwards. The same is true to a lesser extent of the tibia. Examination of the 
data in Tables 1 and 2 shows that relative ,4-P bending strength (I,/!,,) of the tibia and 
distal femur is lower in both sexes in the Pecos sample than in the modern U.S. sample, but 
that this reduction is relatively greater among males in the proximal tibia and distal femur. 
Sex comparisons of the ratio Zmax/Zmin in the modern U.S. sample (Figure 3) lead to the 
same conclusion. In the tibia, the modern V.S. sample again shows a “flatter” curve ofsex 
differences in maximum to minimum bending strength, with no increase in sexual 
dimorphism in the proximal half as occurs in the Pecos sample. Although sex differences in 
the most distal and proximal femoral diaphysis are similar in the two samples, scsual 
dimorphism in the mid-distal region is much reduced compared to the Pccos sample. i1gain 
(Tables 1 and 2) these sample differences are due mainly to relatively greater reductions in 
/,,,,lZ,,i,, among the modern U.S. males. 
In summary’, the modern U.S. White sample shows significantly less sexual dimorphism 
in cross-sectional shape of the lower limb bones than the Pecos sample, particularly in 
sections from the mid-femur through the mid-tibia. This is brought about mainly by a 
relatively greater reduction in bending strength in or close to the A-P plane in males in this 
region. Thus, the most marked sex difference noted in the Prcos sample-relatively greater 
A-P bending strength in males around the knee-is very much reduced in the modern 
sample. 
Taken together, these findings indicate that Pecos males and females engaged more often 
in sex-specific activities which placed different bending loads on the lower limb bones in 
the region around the knee than recent U.S. M:hites. If the Pecos sex differcncc near tht 
knee reflects more running and long distance travel by males (RufY & Hayes. 198%: also 
see Discussion), then the reduction of such a difference in the modern U.S. sample may 
indicate less sexual dimorphism in the time spent in these activities. 
If this hypothesis is correct, then there should be a general correlation between sexual 
dimorphism of lower limb bone structure and sexual dimorphism in activities invobing 
running in diGrent populations. Since greater time spent running and in lon,g distance 
travel implies greater mobility, this can be stated even more generally: the tnore 
populations difier with regard to culturally. prescribed and/or economically necessary 
differences in mobility between the sexes, the more they should differ with regard to lower 
limb bone structure, particularly around the knee. It is this proposal which is tested in the 
following sections. 
Cross-sectional geometry-other samples 
The number of skeletal samples where cross-sectional geometric properties have been 
directly measured or accurately estimated in an adequate number of individuals for sex 
comparisons (minimum of about 10 per sex) is relatively small. In addition to the Prcos 
and modern U.S. White samples, such data arc availahlc for nine other lower limb bone 
Table 3 Sex differences in Z msxlZmin and ZJZ, of femoral and tibia1 cross sections in autopsy and 
archaeological samples’ 
Subsistence technoloa\ 
Hunter-gatherers Agricultural Industrial 
Sect. Prop. Sample? M F %dif.” M F %dif. 11 F %dif. 
Tibia 
50% 
Tibia 
50% 
Femur 
50% 
Femur 
50% 
Femur 
80% 
Femur 
80% 
Eskimo 
Pews 
U.S. White 
I rnax Pews 
c U.S. White 
Japanese 
Georgia 
Tennessee R 
New Mexico 
Japanese 
Pecos 
U.S. White 
k Tennessee R 
Georgia 
New Mexico 
Pecos 
U.S. White 
% 
Georgia 
Pecos 
U.S. White 
I max Georgia 
G New Mexico 
Pecos 
U.S. White 
1.56 
l-28 l-16 10 
1.21 1.06 14 
1.28 1.18 8 
1.50 1.10 36 
1.34 
1.41 
1.66 
0.88 
2.09 2-14 -2 
I.54 1.92 -20 
1.29 21 
1.22 
l-14 
1.42 
:: 
16 
1.14 -23 
1.94 
3.15 
1.08 
Ia5 
1.24 
1.14 
1.28 
1.30 
I .40 
1.481.08 
0.90 
I .82 
I68 
1.90 
1.81 7 
2.70 17 
1.03 
I.01 2 
1.21 2 
1.05 9 
I.23 
1.31 : 
1.32 6 
1.29 15 
1.03 5 
0.86 5 
2.10 -13 
I.83 -8 
1.93 -2 
1.54 
2.15 
2.13 
1.46 6 
2.04 5 
I.91 11 
1 .U6 I ,06 0 
I.21 1.23 -2 
1.29 
@88 
I.35 
1.28 1 
0.98 - 10 
140 -3 
’ Section properties and locations as in Table 1. 
2 See Appendix for provenance of samples. 
s [(Male-Female)/Female] X 100. 
samples, listed in Table 3. These include a protohistoric Eskimo sample (Martin et al., 
1985), modern and prehistoric archaeological samples from Japan (Kimura & Takahashi, 
1982)) and paired Amerindian archaeological samples from two different temporal periods 
in western New Mexico (Brock, 1985; Brock & Ruff, in press), the Georgia coast (Ruffetal., 
1984), and the Tennessee River valley (Bridges, 1985). A more detailed description of the 
samples included here is given in the Appendix. 
Only three section locations-the tibia1 and femoral midshafts and the femoral 
subtrochanteric region (80% level in Pecos and modern U.S. studies)-can be compared 
across samples. Fortunately, these three section locations include bone regions determined 
earlier to be critical in distinguishing patterns of sexual dimorphism in the Pecos and 
modern U.S. samples, especially the femoral and tibia1 midshafts. 
For purposes of comparison, samples were grouped into three broad subsistence 
technology categories: hunting-gathering, agricultural, and industrial. The grouping of 
data in this way was carried out because it was anticipated that subsistence technology and 
relative differences between the sexes in mobility and other behavioral patterns would bc 
related (see Discussion). It is obvious that this type of categorization groups together 
SEX DIFFERENCES IN SKELETAL STRUCTURE 399 
populations with very different sociocultural and biological characteristics, e.g., modern 
Japanese and modern U.S. White; Eskimo, southeastern U.S. Amerindian and prehistoric 
Japanese. However, the aim here is to test whether in fact broad generalizations regarding 
subsistence strategy, behavior, and morphology are valid, regardless of other obfuscating 
factors. The availability ofthree pairs ofsamples with different major subsistence strategies 
but otherwise similar environments and at least partial genetic continuity (Georgia coast, 
western New Mexico, and Tennessee River valley) also provides an opportunity to test the 
hypothesis under better controlled conditions. To a lesser extent this is also true for the 
comparison between prehistoric and modern Japanese groups. 
Z,T/Z,. and Zm;ix/Zmin ratios for the tibia1 and femoral midshafts and proximal femoral 
diaphysis are listed for all available samples in Table 3. As shown here, there is relatively 
less data available for the tibia than the femur, and the femoral midshaft has been the best 
studied region. However, with the exception of I,,, /Z,i” of the midshaft tibia, there is at 
least one sample in each subsistence technology category for each property. 
In both the tibia1 and femoral midshafts, sexual dimorphism in ZmaxlZmin and Z,lZ, 
declines from hunting-gathering through agricultural to industrial groups. In the tibia, 
male ratios are always greater than female ratios, but the hunting-gathering period shows 
the most sexual dimorphism and the industrial period the least of the three groups. In the 
femoral midshaft, male ratios are greater than female ratios in all of the hunting-gathering 
samples and in all but one of the agricultural samples (Tennessee River valley, where the 
two sexes were equal). Percent sexual dimorphism in femoral midshaft I,/!, ranges from 8 
to 36% in hunter-gatherers and from 2 to 9% in agriculturalists; dimorphism in Z,,,,lZ,n,,, 
ranges from 10 to 24% in hunter-gatherers and from 0 to 15% in agriculturalists (0 to 6% 
without Pecos). The industrial period samples show essentially no sexual dimorphism in 
either Z,/Z, or Zmax/Zmin at the femoral midshaft, ranging from a -2 to 1% difference. 
Given the diverse nature of the population samples included, the general consistency, of 
the temporal decline in sexual dimorphism of these femoral and tibia1 midshaft properties 
is remarkable, with almost no overlap between subsistence categories (except for the PPCOS 
Z,,,/‘Z,i,l index at the femoral midshaft’). Furthermore, in every comparison between 
paired samples from the same region, the temporally later group shows less sexual 
dimorphism than the earlier group. 
Patterns of change in Z.Y/Z,, and Zmax/Imin in the proximal femoral diaphysis are not as 
clear as in the femoral and tibia1 midshafts. Females always show greater Zm,,/Zmi” ratios 
than males here, but there is no consistent temporal change in the ratio, either across or 
within geographical regions. The Zr/ZY ratio in the proximal femur shows even less 
consistency, with some groups showing greater values for females and some greater values 
for males. However, this is probably largely a result of the “diagonal” orientation of 
greatest bending strength in this region, tending to fall almost midway between A-P and 
M-L planes (Ruff & Hayes, 1983a), and the dependence of this orientation on the 
antetorsion angle of the femoral neck (Ruff & Hayes, 19836). These variations in 
orienration tend to confound interpretations of I,/!,; for this reason the ZmaxlZTrlln ratio is 
more informative with regard to shape differences in the proximal femur. 
2 It should be noted that the orientation of greatest bending rigidity of the femoral midshaft is not dirwtl) 
anteroposterior, but tends to fall between about 50 and 80 degrws from the mediolateral axis, as described in Ruff 
& Hayes (1983~1). Ruffe~ al. (1984) and unpublished data on the modun U.S. White sample. Thus, a high degree 
ofsexual dimorphism in I,,.lZ,,,,, at midshaft does not necessarily indicate a high degree ofsexual dimorphism in 
A--P/h{-1, bending strength per se, as also reflected in the femoral midshaft 1,/Z, index for Prcos. 
400 (:. KI’Ft 
Thus, the data for the femoral and tibia1 midshatis strongly support an association 
between subsistence strategy and sexual dimorphism in lower limb bone cross-sectional 
geometry. Reduced sexual dimorphism in relative A-P bending strength around the knee, 
noted earlier in the modern U.S. compared with Pecos samples is now seen to be part ofa 
general trend spanning three broad subsistence technologies and several geographic 
regions. The lack of consistent trends in the region about the hip, also noted in the 
Pecos-modern U.S. comparison, may be explained by other structural and mechanical 
factors discussed later. 
External breadth ratios-recent samples 
To provide an even broader range of populations in which to test these associations, ratios 
of lower limb bone external breadths were compared in males and females in a number of 
additional samples. External breadth is only a partial reflection of cross-sectional 
geometric distribution of bone and does not include variations in internal contours as well 
as more subtle variations in total shape. However, because second moments of area are 
heavily influenced by external dimensions (Ruff & Hayes, 1983a), ratios of external 
breadths measured in perpendicular planes are correlated with ratios of corresponding 
second moments of area (Jungers & Minns, 1979). Thus, the same general patterns of 
variation in cross-sectional shape should be expected for these indices as for the more 
complete Zm,,/Z,i” and I,/!, ratios. 
Osteometric external breadth indices have been most commonly calculated at four 
locations in the tibia1 and femoral diaphyses: the tibia1 midshaft, the mid-proximal tibia at 
the level of the nutrient foramen(platycnemic index), the midshaft femur (pilasteric 
index), and the proximal femur just distal to the lesser trochanter (platymeric index) 
(Martin, 1928). Although all of these indices are purportedly derived from breadths 
measured in A-P and M-L planes, in practice it is likely that the measurements for the two 
tibia1 indices and the platymeric index of the femur are actually often taken as maximurn 
and minimum breadths close to the true A-P and rvl-L planes rather than actually in these 
planes. Therefore, these indices, as derived by many investigators, may be more analogous 
to ratios between I,,, and Zmin. Because these three indices are customarily reported as 
minimum/maximum breadths, i.e., as numbers less than or equal to 1.0, they are inverted 
here to correspond directly to Z,,,lZ,;,,, and arc indicated as Dn,axID,nl,,. Breadth 
measurements for the femoral midshaft, or pilastcric index arc more likely to bc taken in or 
very close to the true A-P and M-L planes of the femur, partly because the linea aspera lies 
almost directly posterior and makes an obvious landmark for orientation. Therefore, the 
pilasteric index, D,/D,,, should be directly analogous to the femoral midshaft ZJZ,. ratio. 
The samples incorporated in these comparisons are listed in Table 4. With the exception 
of the Eskimo sample (for which there were no reported breadth data) they include all of 
the samples used in the previous cross-sectional analyses, although not all cross section 
locations with second moment of area data had available corresponding breadth data. 
Samples are more fully described in the Appendix. 
The trend of reduced sexual dimorphism in cross-sectional shape from 
hunting-gathering through agricultural and industrial subsistence technologies is largely 
borne out by the external breadth data. The results are particularly striking for the femoral 
midshaft DaplD,,,t index, in which no overlap between a relatively large number of groups 
occurs: hunter-gatherers range from 7 to 15%, agriculturalists from 3 to 6%, and industrial 
SEX DIFFERENCES IN SKELETAL STRLrCTc!RE 10 1 
Table 4 Sex differences in external breadth ratios of femoral and tibia1 diaphyses in autopsy and 
archaeological samples 
Tihi.1 
mid 
‘I‘ibia 
nut. 
Ohio R. ( 1) 
‘I-rnnessee R. 
Georgia 
,JapatW 
I_‘.%. \yhite(3) 
Ohio R. (1 ) 
Trnnessee R. 
U.S. \Vhiw (3) 
.4ustralian 
Ohio R. il ) 
Ohio R. (2) 
l‘rnnrssre R. 
Gwrgia 
Stw’ k\lcxico 
,Japam?sc 
Peco, 
British ( I, 3) 
Y.S. \Vhitc (I) 
1’.S. Cl’hitc (2) 
OhioR. (I) 
‘l‘ennessce R. 
Grorgia 
New Mexico 
British (2, 3) 
143 
1.60 
I.17 
1.18 
1.20 
1.17 
1.17 
1.18 
1.21 
1.16 1 .ZJ --:i 
1.27 1.34 -5 
1.36 1.32 3 
1.03 1.10 -b 
1.29 
I47 
145 
I fJ8 
I.32 
I.51 
I .08 
1.08 
1.1 I 
1.04 
1 TJY 
I.09 
I 48 
148 
1.57 
1 ,.5’ 
1+,1 
1 .h(J 
1.17 
I.1 I 
I.10 
I.14 
1.10 
1.28 
1.29 
1 .?Y 
1.06 
1.2i 
I.06 
1 .OY 
1.11 
I .0-1 
1.19 
1 SIT Appendix Ior- provenance of samples; Ohio R. , I): \Vebh Cy hug I 1Yljl: Ohio R. (2): Perzigtan P/ rti. 
I 1984 I: British ( 1): Krnnrd) (1985); British (2): Parsow (191 1): British 13): Holthy (1917); U.S. \Vhitv f I j’ Kull 
rt 01. I 1986); U.S. LVhitt (2): presrnt study. Terry collection: I1.S. \Vhilc (3): Hrdlicka (1898). 
’ [(~lalr-Ftmalf)/Frmairj X 100. 
period groups from - 1 to 1% sexual dimorphism. As noted above, this is also the cross 
section in which the orientation of breadth measurements is most likely to be directly X-P 
and M-L, and thus most consistently measured within and between groups. The results for 
the mid-proximal tibia1 section Dmas/Dmin index show a similar consistent temporal trend. 
Every paired temporal sequence for these two sections shows a decline from earlier to later 
samples. The tibia1 midshaft section index is somewhat less consistent, although there is a 
definite trend towards a reduction in sexual dimorphism through time over all groups. The 
presence of overlap between categories and the slight increase in sexual dimorphism 
between the first two categories in the Tennessee River valley and Georgia coast samples 
for this section may be due to greater measurement error at this location, i.e., variation in 
actual planes of’measurement (A-P/M-L versus maximum/minimum breadths). 
As with the earlier cross-sectional data, the results for the proximal femoral diaphysis do 
‘1’ ‘y 
Ij 
R 
Dap /%7l 
I_ 15 2 9 8 0 c 
1 10 $ E z 0 
B 
i 
2 
5 ; 
s 
ki 
a 
l-l 
“_________8____-__--__-_~_ o 
I I I I I I 
H-G AG IND H-G AG IND 
(4) 14) (2) (7) (7) (4) 
Subsrstence technology 
Figure 4. Sexual dimorphism in two measures of relative A-P/M-L bending strength of the femoral 
midshaft in three broad subsistence categories. Total ranges and median values of sample means are 
plotted. IX/J,: ratio of second moment of area about x (M-L) axis to that about-v (A-P) axis; D.plD,,: 
ratio of A-P external diameter to M-L external diameter. Percrnt sex difference: 
[(Male-Female)/Female] X 100. H-G: hunting-gathering; AG: agricultural; IND: industrial. 
Numbers in parentheses indicate number ofpopulation samples included in each category. See Tables 3 
and 4 and text for provenance of samples. 
not strongly support a temporal decline in sexual dimorphism in this skeletal area. Again, 
like second moment of area ratios, with the exception of the Georgia coast preagricultural 
sample, females show consistently greater D,;,,lD,i” indices in the proximal femur than 
males over all groups. There is also a suggestion of a reduction in sexual dimorphism 
between hunter-gatherers and agriculturalists in this index, except on the Georgia coast. 
Changes in the degree of sexual dimorphism in both external diameter (D,,lD,l) and 
corresponding second moment of area (I,/!,) ratios at the femoral midshaft are illustrated 
graphically in Figure 4. To reduce the effect of extreme outliers, medians rather than 
means are indicated along with the total ranges of subsistence groups (use of mean values 
increases the difference between groups). The steady decline through time of sex 
differences in both indices and the lack of significant overlap between subsistence 
categories is readily apparent. Thus, regardless of the particular population samples 
included or measurement techniques applied, the results clearly indicate that sexual 
dimorphism in relative A-P/M-L bending strength in the mid-femur (and proximal 
tibia-Tables 3, 4) is reduced in successive broad stages of subsistence technology. 
External breadth ratios-upper and Middle Paleolithic 
To add temporal depth to the analysis, as well as to test the association between skeletal 
morphology and subsistence economy in even more remotely related human populations, 
SEX DIFFERENCES II% SKELETAL STRUCTURE 403 
Table 5 Sex differences in external breadth ratios of the femoral diaphysis in early H. s. sapiens (Upper 
Paleolithic) and Neanderthals (Middle Paleolithic) 
Group” 
Midshaft DJD,,,,’ Subtroch. D,,,,lD,,,,’ 
Male Fumble (M-F)IF Male Femalr (Xl-Fl/l: 
CR) (n) x 100 (ni (n) x I(10 
Earl\ N. 5. tapirns I.180 1.077 
112) 17) 
Neanderthal 1,014 0938 
(9) (1-j 
’ Properties as in ‘l‘able 4. 
J SIT Appendix for provenance ofsamples. 
9.6 1~363 I.370 -Il.> 
11.51 161 
8.1 I.219 I.282 -.k’) 
17) 14) 
data for a series of femoral specimens from the Upper and Middle Paleolithic were 
examined (Table 5). As in the previous analysis, breadth ratios for the femoral midshaft 
(pilasteric index) and subtrochanteric region (platymeric index) were compared in male 
and female specimens. All osteometric data and taxonomic and sex assignments are takenfrom Trinkaus (1976, 1983 and personal communication). Neanderthal (Middle 
Paleolithic) and early Homo sapiens sapiens (Upper Paleolithic) samples were analyzed 
separately; sample sizes range from four Neanderthal females to 15 early H. s. sapiens males 
(suhtrochanteric section). Individual specimens included in each comparison arc listed in 
the .-\ppcndix. Sex comparisons were carried out for various combinations of specimens: 
only those with pelvic determinations of sex, with and without specimens of questionable 
sex assignment, and with and without Middle Eastern specimens. Since every type of 
comparison yielded essentially the same results, findings for the pool of all available 
specimens, including those with less certain sex assignment, arc presented in Table 5. 
Both Neanderthals and early H. s. sapiens show the same basic pattern of sex differences 
in femoral cross-sectional geometry as recent human populations. Males are relatively 
broader in the A-P plane at midshaft, with average percentage differences between the 
sexes in A-P/M-L breadth of 9.6% in early H. s. sapiens and 8.1% in Neanderthals. These 
figures are closr to the median values for the same index in recent hunter-gatherers (Figure 
1). Early H. s. sapiens females have relatively greater maximum (“M-L”) subtrochantrric 
breadths than males on average in each group, although the two sexes are almost equal in 
this index in early H. s. sapiens. Again the average 1 to 5% greater value for females at this 
location is similar to that found in recent human populations (Table 4). 
Thus, patterns of sexual dimorphism in lower limb bone structure demonstrated for 
modern humans appear to be characteristic of earlier Homo sapiens population samples as 
well. These earlier samples are most similar to modern hunter-gatherers, with whom, of 
course. they share the greatest similarity in subsistence technology. 
Discussion 
Temporal trends in lower limb bone cross-sectional geometry 
Sex differences in femoral and tibia1 external breadth indices have been previously noted 
by several authors. Interestingly, in some studies which included only autopsy material 
from modern Euro-American populations, sex differences were noted in only the femoral 
suhtrochanteric region (females more “A-P“ flattened) and not the femoral midshaft or 
404 C:. RI.I;F 
tibia (Hrdlicka, 1898; Parsons, 1914; Holtby, 1917). T‘his is consistcttt with results of the 
present study, in which sexual dimorphism in cross-sectional shape of mid-femur and tibia 
is greatly reduced in “industrial” period samples: while the subtrochanteric rc,gion shows 
little decline in sexual dimorphism through different subsistence categories. ‘l‘hr largest 
compilations of data relevant to the present study arc those of Pearson & Bell (1919) and 
Martin (1928), and both support the general findings of this study. Martin, in particular, 
noted that in most of his population samples (which included many archaeological 
samples) females had rounder cross sections in the midshaft femur and proximal tibia, and 
males were rounder in the proximal femoral diaphysis (1928, pp. 1136, 1139, 1158). 
Hrdlicka also noted a similar trend in the midshaft femur among several large autopsy and 
archaeological samples, although this argument was based on a typological rather than 
osteometric approach (19346, p. 157). None ofthese authors, however, identified temporal 
trends in sexual dimorphism or attempted to provide a mechanical basis for these 
differences. 
Temporal trends in cross-sectional shape of the lower limb bones in populations as n 
whole have been noted previously by various authors (e.g., Buxton, 1938; Lovejoy, 1970; 
Brothwell, 1972; Lovejoy et al., 1976; Ruff & Hayes, 1983a). Recent Euro-American 
populations tend to have rounder diaphyscs at all levels than earlier or preindustrial 
populations, probably a result of relatively reduced and more generalized mechanical 
loadings of the lower limb (Lovejoy et al., 1976: Ruff & Hayes, 1983n). 
To further evaluate temporal differences in the present sample, the midshaft femur 
shape indices of Figure 4---1,/I, and DnplDmr are replotted for each sex separately within 
each subsistence category in Figure 5. As expected given earlier results, the difference 
between male and female median values steadily declines from hunting-gathering through 
industrial groups. There is also successively less overlap between male and female ranges, 
from no overlap in hunter-gatherers to almost complete overlap in the industrial period. 
It is also apparent from Figure 5 that males show a much stronger temporal decrease in 
both ZJJy and D,,/D,l than females. The largest decrease in males occurs between the 
hunting-gathering and agricultural periods, with almost complete nonoverlap between 
groups. Females also show some decrease in median values between these groups, but the 
hunting-gathering range is completely encompassed by the agricultural range. Males show 
little change in either index between agricultural and industrial periods, although the 
D,,/D,l median continues to decline. Females actually increase in the median value ofZ,/Z, 
between the last two periods, although they show essentially no change in DOp/D,~ medians 
and the ranges for both indices are again almost totally overlapping. Some of the variation 
in median values between the last two groups may be due to sampling error because of the 
small number of industrial samples, particularly for IV/J,, (fr = 2). 
Taken as a whole, the data in Figure 5 suggest that the previously documented general 
temporal increase among populations in diaphyseal circularity (decrease in the present 
shape ratios) has actually been restricted primarily to males, at least in the femoral 
midshaft. This observation cannot be verified at other cross section locations, where data 
are unfortunately much more limited (Tables 3, 4) and techniques of measurement arc 
more variable, as discussed earlier. However, results for the femoral midshaft. by far the 
best represented section, indicate that females show no large changes in cross-sectional 
shape through time, while males become significantly more circular, particularly between 
hunting-gathering and agricultural periods. Stated in terms of mechanical loadings, it 
appears that relative A-P/M-I, loading of the lower limb has dcclinrd more through time 
SEX DIFFERENCES IN SKELETAL STRL’T(:TI-RF: ‘105 
I ,5 - 
1.4- 
I ,3- 
.% 
‘.. 
.* 
I.2 t 
I 
I.1 
1 
d 
,.,I 
H-G AG IND H-G IND 
(4) (4) (2) (7) (4) 
- 1. 
I ., 
-I. 
- I., 
- I.1 
- 1.1 
25 
20 
15 
c? 
.\ 
3 
Q 
IO 
05 
00 
Subslsience technology 
Fiqure 5. Male and female values for two measures ofrelative i\-Plhl-I. bending strength ofthe femoral 
midshaft in three subsistence categories. Total ranges and mrdian values of xx-specific sample means 
arr plotted. Symbols as in Figure 4. 
in males than in females. A similar conclusion was also reached based on the morr 
complete analysis of the Pecos-modern U.S. sample differences. 
The Upper Paleolithic sample means for femoral midshaft DaplDml fall exactly on the 
corresponding median values for recent male and female hunting-gathering samples 
(Table 5, Figure 5). Thus, it is apparent that Upper Paleolithic populations were, at least 
in this regard, essentially indistinguishable from recent hunter-gatherers, i.e., that little 
change in relative A-P/M-L lower limb bone loadings or sexual dimorphism in loadings 
occurred until the development of agriculture. The Neanderthal sample as a whole differs 
significantly from all other samples in this property, with much lower values for femoral 
midshaft indices in both sexes(Table 5; P < 0.05, l-tests. combined or within-sex 
comparisons with Upper Paleolithic samplcj. This difference may reflect fundamental 
structural and functional alterations of the skeleton in this group (Trinkaus, 1976). It is 
significant, however, that even with this basic difference in structure, .rexual dimorphism in 
cross-sectional shape among Neanderthals is very similar to that of Upper Paleolithic and 
recent hunter-gathcrcr samples (Table 5, Figure 4). 
Increawd mobili[y and anteroposterior bending of the lower limb 
It has been assumed here, based on a previous interpretation of male-female behavioral 
differencrs at Pecos Pueblo (Ruff & Hayes, 1983a), that increased running and long 
distance travel over uneven surfaces will lead preferentially to increased A-P bending loads 
in the region near the knee joint. The evidence in support of this assertion deiives from 
406 c:. KrFk 
several sources, including both experimental and thcorctical analvsrs of lower limb bone. 
loadings. 
Human strain gage data analyzed by Lanyon et al. (1975) and Carter (1978) indicate 
that bending stresses in the tibia increase greatly from walking to jogging. Minns PI ~1,‘s 
(1977) theoretical study of stresses during walking indicate that maximum A-P bending 
stresses in the tibia occur in the proximal half of the diaphysis, near the knee, and it seems 
likely that this would also be the case during running. Theoretical analyses of lower limb 
bone loadings under various weight bearing conditions indicate that A-P bending near the 
knee increases greatly as the knee is flexed and/or the joint center is moved progressively 
further from the body center of gravity (Pauwcls, 1980; also see Preuschoft, 1970). Both 
movements occur to a greater degree during running than in walking. The same effect 
would also occur during climbing or movement over rough surfaces (using only or 
primarily the lower limbs for support). 
The large muscles or muscle tendons crossing the knee joint-quadriceps, hamstrings, 
and gastrocnemius-are capable of generating large A-P bending loads and very little 
M-L bending of the femur and tibia. All of these muscles become more active during 
running (Alexander & Vernon, 1975; Mann, 1982), and together with changes in leg to 
body position create A-P bending moments approximately double those generated during 
walking (Alexander & Vernon, 1975). Knee flexor/extensor muscles are also much more 
active during walking on uneven surfaces, such as up and down stairs or ramps (Morrison, 
1968, 1969). Conversely, as would be expected, M-L bending about the knee is not 
significantly affected by knee flexion or relative A-P position of the knee with respect to the 
body center (Pauwels, 1980). Hip abduction-adduction is very little changed from walking 
to running (Mann, 1982)) indicating little alteration in relative M-L lower limb position. 
This is also supported by analysis of ground reaction forces, in which mediolateral shear 
shows no increase from walking to jogging, while both fore-aft shear and vertical force do 
increase, vertical force nearly doubling (Mann, 1982; also see Cavanagh & LaFortunc, 
1980). Increased speed in walking also leads to an increase in the vertical and fore-aft 
components of ground reaction force, while the mediolateral component remains 
essentially unchanged (Rohrle et al., 1984). F inally, soft tissue structures which could 
transmit significant M-L bending loads about the knee, such as the iliotibial tract and the 
lateral collateral ligament, tend to become lax during knee flexion (Morrison, 1970; 
Preuschoft, 1970). 
Thus, it is apparent from several lines of evidence that A-P bending loads in the region 
near the knee joint should increase much more than M-L bending loads from slow level 
walking to running or movement over uneven surfaces. These actions are precisely those 
which would be expected to show the greatest increase in a lifestyle which included more 
long distance travel, i.e., greater mobility. A-P/M-L bending near the hip would be 
expected to show less of a change, because of the relative constancy in position of the hip 
joint relative to the body center ofgravity and the less marked changes in flexion-extension 
of the hip during different activities (e.g., Mann, 1982). 
Sexual division of labor 
As noted by Frayer in one of the opening quotes of this paper, anthropologists have long 
been interested in the division of behavior along sex lines. If the preceding analyses are 
correct, then there should be evidence in the ethnographic literature that behavioral 
differences between the sexes, in particular differences in relative mobility, decline from 
SEX DIFFERENCES IN SKELETAL STRI:CTtTRE 407 
hunting-gathering to agricultural to industrial societies. Since much ofbehavior is centered 
around economic roles, this is largely equivalent to testing whether the sexual division of’ 
tasks requiring different degrees of mobility has been reduced across the three broad 
subsistence categories examined here. 
Murdock & Provost (1973) have carried out perhaps the most extensive cross-cultural 
statistical analysis of the sexual division of labor, including sex allocation of 50 
technological activities in 185 societies. Among those activities which are exclusivcl). or 
usually carried out by males are hunting of large aquatic and land fauna, fowling, trapping 
of small land fauna, fishing, and herding-all behaviors requiring a relatively high degree 
of mobility. Activities which are most often carried out by females include gathering and 
preparation of wild vegetal foods as well as various “camp” activities such as cookin,q and 
water carrying, none of which require great mobility. Significantly. many “intermediate” 
activities, not strongly male or female, such as manufacturing of clothing and house and 
boat building, tend to be performed by females in hunting-gathering and nomadic 
pastoralist societies, but by males in agricultural societies. Similarly, as agriculture 
intensifies and craft specialization increases. males more often tend to be assigned othc,r 
“intermediate” tasks, including soil preparation, crop planting. tending and harvestirla, 
and pottery making. All of these “intermediate” technological activities require relati\,ci\ 
little mobility. 
Thus, two general features of the sexual division of labor in preindustrial societies 
emerge: (1) tasks that require a high degree of mobility are usually if not cxclusivel>, 
assigned to males; (2) with increasing dependence on agriculture and technological 
specialization, males tend to be assigned increasingly more sedentary activities. .I\ 
discussed by Brown (1970), the allocation of relati\,ely sedentary tasks to females in mosl 
societies is probably largely determined by the direct and indirect demands ofchild care. lr 
is true that some almost exclusively male activities, such as metal, wood and stone workin,c 
(Murdock & Provost, 1973), arc also relatively sedentary, hut if tasks requiring large 
amounts of time away from the home arc present in the society, they will almost always b(, 
pcrfbrmed by males. Watanabe (1977) has also prolidcd good illustrations and discussion 
of the sexual division of labor and differences in relative mobility of the two sexes in the 
hunting-gathering Ainu and other groups. 
Therefore, it is apparent from ethnographic observations of living and recent pc~~plc.s 
that sex diff‘erences in relative mobility are indeed greatest in hunter-gatherers and dcclinc 
with the adoption and intensification ofagriculture. \Vhilr comparable data are not strict11 
available for industrial societies such as the modern U.S.. it is apparent that all sex roles, 
inclltdingthose involving differing degrees ofmobility, ha1.c been blurred in these cultures. 
Also. with increasing mechanization, travel over lorlg distances and many other acti\itics 
can be accomplished without a concomitant incrcasc in physical exertion. Thercforc, it 
seems intuitively oblrious that sex differences in lower limb loadinqs resulting from acti\.it!, 
differences tijould he least marked and consistent in industrial societies. 
Thus. variations in the degree of sexual dimorphism of bone cross-sectional geometry 
from the mid-femur through the mid-tibia do appear to correspond to real differc%nces in 
actilitv patterns in the two sexes. Relatively greater A-P bending strength around the knee 
is associated with relatively greater mobilit!-more frequent tra\.el over greater distances 
;111 d more frequent running. The largest decrease in A-P/M-L bending strcnath in this 
region occurs in males between hunting-aathering and a,qricultural subsistence Wchnolo- 
Fits i Figure 5). precisely where the greatest relative dccrcasc in long distance travc.1 mighr 
408 ,:. KI‘kF 
be expected. Relatively little o\~all change in AI-l’/hZ-I. bending strcrlgth then occurs i11 
the transition to an industrial economy, although avcragc scx diffcr~ncc~a continue to 
decline (Figure 5). 
Frayer (1980, 1984) used a similar line of rcasonin,q to explain a rclativc rc-duction 
through time of sexual dimorphism in stature (as well as dental and cranial dimrnsions) in 
samples spanning the European Upper Paleolithic, Mesolithic and Neolithic. As in the. 
present study, he found most of this reduction to be due to grcatcr dcclincs in malt 
dimensions, which he explained as a response to decreased need for large body size with the 
development ofmore advanced hunting techniques. His conclusion that “the level ofscxual 
dimorphism within a population is roughly proportional to the exclusivity ofthc division of 
labor by sex” ( 1980, p. 399) is also supported by the findings of thr present study. i LTnlikc 
the results presented here, however, Fraycr found no evidence for a continued drcrcasr in 
sexual dimorphism from the European Neolithic (i.c., agricultural) to the present da), (i.c.. 
industrial), and stature for samples as a whole incrcascd over this time period. This is most 
likely a result of the combined action of both mechanical factors and other en\.ironmcntal 
influences on general body size. For example, an incrcasc in gcncral nutritional quality, 
characteristic of the recent history of Europe, may lead to both an increase in body size and 
an increase in sexual dimorphism in body size (Stini, 1974; Eveleth & Tanner, 1976: Gra)- 
& Wolfe, 1980). The morphological characteristics cxamincd in the present study should 
be more sensitive to purely mechanical factors, and thus it would not bc surprising ifthc), 
reflected changes in activity patterns more accurately than ,general body size. ‘l’hc same 
explanation applies to measures of general skeletal size (bone length) or “robusticity” 
(midshaft breadths or circumference divided hy bone length), which do not show clear 
temporal trends in sexual dimorphism (Trinkaus, 1980). These properties and other 
measures of bone size may partly reflect the general lcvrl of applied mechanical loadings 
(Garn el al., 1972), but they do not indicate the !Y~PS of loadings bones arc suhjectcd to 
in uiuo (also see Ruff ~1 al., 1984). 
Unlike the skeletal region around the knee, the proximal femoral diaphysis shows no 
consistent change in sexual dimorphism of cross-sectional shape (Tables 3, 4). hlales arc 
almost always more circular than females here, i.e., females are adapted for rclativclk 
greater M-L bending, or more precisely, for greater bending in the antetorsion plane of the 
femoral neck (Ruff & Hayes, 19836). This is probably best explained as a conscqucncr of 
relatively wider pelves and greater interacetabular distance among tkmalcs (Van Gcr\,en, 
1972; Lovejoy ef al., 1973; Brinckmann e/al., 1981), which should lead to rclativrly grcatcr 
M-L bending of the proximal femoral diaphysis (based on calculations of relative hip 
abductor force from data presented in Lovcjoy el al., 1973, and analysts rcportcd t)) 
Rybicki et al., 1972). The lack of consistent change in sexual dimorphism of proximal 
femoral shape therefore probably reflects a lack of change in basic structural difrcrcnccs 
between the sexes related to the requirements of childbirth. Data hearing directly on this 
issue are relatively scarce, but what information there is suggests that sexual dimorphism 
in relative pelvic hrcadth does not vary significantly along temporal or suhsistcncc, stratcg:) 
lines, supporting the hypothesis (Howells (Ir Hotelling. 1936: Young 8r Inca. 1940: 
Washburn, 1948, 1949; Lovejoy e1 al., 1973; Brinckman e/ (II., 1981 ). 
Another way in which activity levels can br assessed from skrletal remains is through the 
incidence of osteoarthritis, or degencrativc joint disease [Jurmain, 1977, 1980; Larsen, 
s For another virwpoint. sce,Jacobs (1985) 
SEX DIFFERENCES IN SKEI.b:TAI. STRI’C TI’RF, IO0 
1982; Mcrhs, 1983). Variation brtwcen populations in the sex-specific incidcncc ol‘low~~~ 
liml) osteoarthritis is generally consistent with trends in hone cross-sectional gcomrtr) 
.Amc)n,g hunting-gathering and agricultural populations males have highrr ittcidcnc,rs 01’ 
osteoarthritis at almost all major appendicular joints, including those of thr lower liml) 
Uurmain, 1977. 1980: Larsen, 1982: Mrrbs. 1983). In contrast, modrrn I-IS TZ’hitc and 
Black samplrs show relatively little sexual dimorphism in incidcttcr ofostc.o~trthritis. with 
fixmalcs actual11 showing somewhat hightsr frrqucncicx of‘ occurrcncc (Jutmain. I977. 
1980). ‘l‘hc corrrspondence bctwcctt subsistence rtratc*g)’ and sexual dimorphism irt 
ostc’oarthritis is not as strong as in cross-scctirmal gromrtry, howcvrr. Again. like mvasurcs 
of‘g~~ttcral skclctal size, this is probably due at Icast part]\ to the relative nottspccific,it\ (II’ 
ostroarthritis bvith regard to ly@r rather than ,gcttcral lrv~ls of mrchanicai loadings. 
Man!. attalvscs of population diff‘crrnccs in srsual dimorphism have cottcrntratrd ott 
tlic~tar). rather than mrchanical rxplan;~tions (r.,g., Hatniltott, 1982). \Vhilc. :I> t~otvtl 
;I ho\ r, this may hc rrlrvant to general size difIirrrt1cc.s l)ct\\.rrn malrs and limalcs. it ih 
tlifIic ult to srr how a systemic factor likr dirt could ha\-c. the specific rffi,cts on iocalizc3d 
bone rrmodrlitq documcntcd hcrr. Skclrtal size and usual dimorphism in sixr \ :tr! 
sigriificantl~~ amcq thr population samplc4 iticludrd in this study. both Lvithiti and 
bc.t\vc*c‘tt subsistrttcr lr\.cls. yet srx-rclatrd difti,rrnccs in ~CIIIC shape sho~v rc.trtark;tl)l! 
consistrtit trc*ticls. 
‘1‘ttc association brtwren srxual dimorphism in ac (iLit\, patterns and skclrlal morphol- 
or, itround the kncr has implications fi)r the. rrconstruction of hrhavior ant, subsisu~rtcc~ 
stratchg)- in past populations. In populations in which thr suhsistcttcc trchnolo,q or sc~ual 
division of labor is unknown or uncrrtain, the dcgrcr of‘ scsual dimorphism in rclati\~c 
A-I’/‘YI-13 cross-srctional dimensions of the mid-li*moral to mid-tibia] diaphysrs pro\ ic1c.s 
one tttcans of trstitt,? altrrnative hypothcscs. It should h ttotcsd that adrquatc \;tmplitt~ of 
both s(‘xcs is tic’crssary to carry out such an anal\%--at Ir:ist IO ittdi\4duals of(3(~tt s(‘?i 
wrrc inc-ludcd in almost all of the samplrs sruclicd hc-rc. 
‘T‘hr rc‘sujts of this study also ha\re implications with t-cgarcl to rcchniclurs for drtrrmirtirtg 
SC’S l?om fi%moral attd tibial diaphysral hrradths or circumfivttcrs (Kimura. I !I/ I : Black. 
1978; I)iBcrmardo & ‘Tavlor, 1979, 1982; Iscan& hlillrr-Shai\.itz, 1984; RlacLaughlirt & 
Bruce,. 198.5; Dittrick & Suchcy, 1986). c’, ariations in tttc. accuracy of‘such tcchttiqurs ha\ (’ 
hcrtt provisiotiall~~ attributed to srvcral factors. irtcluditt,q ,grttrral mechanical rfli.c.ts 
(l)iBr~ntiardo & FI‘aylor, 1982; Iscan & Millrr-Shai\,itz. 1984: hlacLaughlitt & Hrurr. 
1985). ‘I‘hc rrnults of‘thc prcscnt stud\- suggest that spccilic ntcchanical factors ma!’ c~\;plairt 
at Irast somr of‘thrsr variations in rr.sults. .A striking li.aturc. ofthosc rrchrtiqucs that lta\.c, 
ittclttdcd diaphvsral breadths as wrll as circumfivnccs is that ;\-I’ brradth CtJlltritJUtCS 
ntorc IO discrimination than hl-I. hrcadth (Kimura. 197 1: Dittrick & Sucltr~. l!P~(j: 
I)iBrnttardo 6: I‘a\~lor. 1982: Iscatt Br hlillrr-Shai\-itz, 1984~. III OIIC stud>. which IW~ otll> 
th m,Gmum :\-I’ diamctcr of the femoral sltaf’t. a scxittg accuracy of91 % cotttparc~(l to 
assigttmckttts hascd on traditional nonmrtric prlvic and cranial tncthods \vas obtaittrd. 
ttighcr, thatt arty othrr rrportrd figurrs (hlacl,aughlitt & Bruce. l!j85). 
l‘hc rc.sults of‘thr prrscrtt study indicatr that A-l’ breadth ofrhc mid-li>mur to mid-tilJi;r 
is ;I b(,ttc~t- discrimittator of srx than ,21-I, hrcadth I~wausr malts tcttd to h;t\.(. grc~atc~r 
.&P/XI-I. brradttt ratios than frmalrs in this akcl(,tal rcqiott. rJ‘tlus. sitter rnal~~ ;~r(’ 
generally larger than fcmalcs, hl-L hrcadths v, ill tend to overlap hctwcen the sixes ~I(III’ 
than A-P breadths. The relatively greater A-P breadth of the micl-fi~mur in males is 1101 
caused simply by greater development of the muscle attachment areas on the linea asp(‘ra 
(MacLaughlin & Bruce, 1985), since a similar sex diffrrcncc in shape is also characteristic 
in regions without the linea aspera, i.e., from the distal femur through the mid-tibia (Figurt 
3). Rather, greater mobility of males and consequent higher A-P bending loads in the 
region near the knee arc prohahlv responsible for this sex difference, as discussed carlicr. 
Thus, variations in the accuracy of sex discrimination between difrerrnt samp1t.s ma! 
reflect at least partly the dc,yrcc of sexual dimorphism in activity patterns in thosc~ 
populations. For example, A-P diamctcr of the mid-femur may more clearI\, aid sex 
discrimination in archaeological samples ( AlacLaughlin & Bruce, 1985; IXttrick & 
Suchcy, 1986) because of temporal/subsistence strategy differences in sexual dimorphism 
(Figure 4). Any future attempt to employ thcsc methods in sex determination should take 
into account these population diffcrcnces in behavior. The present results also suggest that 
the inclusion of A-P/M-L, ratios along with bone “size” metric-s. e.g., circumti,rcncc, and 
length, might improve the accuracy of resulting sex discriminant functions> dcpcndcnt. OI 
course, on the particular sample anal!-zcd. 
Summary and conclusions 
A renewed emphasis on function and shape-related rather than simply size-related 
characteristics in the analysis of sex differences in skeletal structure has been called for b! 
several investigators (Van Gervcn, 1972; Armclagos et al., 1982; DiBrnnardo & ‘Taylor, 
1982). The present study illustrates one way in which a biomrchanical approach, stressing 
behavioral reconstruction, can be used to contribute some new insights into this arca 01 
research. This study also shows that long hone diaphyseal form, sometimes overlooked or 
de-emphasized in functional investigations (e.g., in favor of articulation surface 
morphology) may in fact contain important information regarding past mechanical 
loading history and thus behavioral patterns. Also, it should be stressed that bvhilc 
engineering theory provides a framework in which to interpret such variations in fhrm, the 
general concepts and approach involved here are more critical than the relatively more 
complex calculation ofspecific engineering properties such as second moments ofarea. For 
example, in the present study, once a basic functional hypothesis was established based on 
comparisons of engineering properties, it was possible to test the hypothesis using data 
obtained from simpler and more widely available techniques of measurement. It is 
anticipated that this general approach will be applicable to several related questions in 
hominid evolution extending beyond the analysis of sex differrnccs. 
The study began with a detailed comparison of sexual dimorphism in cross-sectional 
geometric parameters at 10 femoral and tibia1 locations in two samples: a late prehistoric 
archaeological sample from Pccos Pueblo, New hlcxico, and a modern U.S. White sample. 
Sex differences in relative A-P/M-L bending strength were found to be marked in sections 
from the mid-femur to the mid-tibia in the Pecos sample, with males shelving relative11 
greater A-P bending strength. This sex difference was greatly reduced in the U.S. LVhitc 
sample. Females showed relatively greater hl-I, bending strength in the proximal femoral 
diaphysis in both samples. Comparison of similar cross-sectional geometric properties in 
other population samples showed that the Pccos-U.S. \l:hite difference in sesual 
dimorphism in the region around the knee was part of a general trend spanning 
SEX DIFFERENCES IiY SKELETAL STRI’(:TYRE 411 
hunting-gathering, agricultural, and industrial subsistence strategies. This trend vqas 
confirmed in a larger sample of populations using less exact external breadth ratio 
measurements. Finally, lower limb bone external breadth ratios avrailable for Keandcrthal 
and Upper Paleolithic samples were compared with the recent samples. Results were then 
interpreted in light of functional anatomical and ethnographic information available for 
livin,g humans. Together these findings lead to the following conclusions: 
( I ) Sexual dimorphism in cross-sectional shape of the femur and tibia is characteristic of’ 
a wide variety of human populations, from the Middle Paleolithic through living peoples. 
The strongest difference between the sexes occurs in the region near the knee, where males 
tend to have relatively more bone distributed in the antcroposterior plane. This is probably 
caused by relatively greater mobility and more frequent running among malts, M,hich 
should create higher A-P bending loadings in this area. A less marked tendency for females 
to have rclativeiy more bone distributed in the principal plane of bending (near the 
mcdiolatcral plane) of the proximal femoral diaphysis is most likely a result of sexual 
dimorphism in pelvic breadth, with greater intcracetabular distance among females 
creating relatively higher M-L bending loads in the proximal femoral diaphysis. 
(2) There is a consistent systematic variation between populations in the degree ofscxual 
dimorphism in bone shape around the knee which is related to subsistence strategy. 
Hunting-gathering populations show the greatest sexual dimorphism here, agricultural 
populations an intermediate level, and industrial societies the least sexual dimorphism. 
This corresponds to a similar reduction in the sexual div-ision of labor through these three 
I~~rls of subsistence technology. .A concurrent general world-wide trend towards >grcatcr 
diaphyseal circularity in the lower limb is primarily a result of changes in bone shape in 
males, who have also undergone the largest changes in mobility and activity types with 
shifts in subsistence strategy. There is little systematic difl‘crcncc between populations in 
sexual dimorphism near the hip, probably because the basic requirements ofchildbirth do 
not \‘:iry si,gnificantly. 
(3) Early Homo sapiens sapiens samples arc essentially indistinguishable from modern 
hunter-gatherers in both femoral and tibia1 shape and sexual dimorphism inshape. 
Srandcrthals are different in basic shape, but are very similar with regard to sexual 
dimorphism in shape. Thus, sexual division of labor, at least with regard to relative 
mobility. appears to have been similar to modern hunter-gatherers at least as f:lr bac,k as 
the hliddle Paleolithic. 
(4) ‘Techniques for the assessment of sex from lower limb skeletal remains should 
consider these systematic population differences. Inch~sion ol’a shape index in addition to 
size measures may aid in discrimination between the sexes. 
Acknowledgements 
The collection of data for the Pccos and part of the modern LT.S. population samples v~as 
carried out in collaboration with Dr LVilson C. Hayes in the Orthopacdic Biomechanics 
L.iboratory, Beth Israel Hospital, Boston. I would like to thank Dr Erik Trinkaus for 
providing data on the Neanderthal and early Homo sapiens sapiens samples, Dr Clark Larsen 
for providin,g useful reprints, and Dr Larsen, Dr Patricia Bridges, Dr David Burr, and 
anonymous reviewers for offering helpful comments on the manuscript. Supported in part 
by NIH grants #.4MO7112, #AGO4809 and a research grant from Howmedica, Inc. 
References 
SEX DIFFERENCES In- SKELETAI. STR17CTI’RE 4 I 3 
Jungcrs. W. I,. & hlinns, R. J. (1979). Computed tomography and hiomcchnnical analysis of fossil Ione hontx. 
An. ,/. phys. .4nthrop. 50, 285-290. 
Jurmain, R. D. (1977). Stress and the etiology of osteoarthritis. Am. J. /I&J. .4nthrop. 46, 333-366. 
Jurmain. R. I). (1980). The pattrrn of involvcmenr of appcndirular dcg-cnuativc joint dcsrasc. .iw ,/. phv~. 
.+l,,throp., 53, 143-150. 
Kcnnrdy, <:. E. ( 1085). Bonr thickness in Homo vrwtus. _/. hum. Em/. 14. 69%708. 
Kimura, I’. (1971 )_ Srx determination on thr cross-action of human lo\vrr Icq hone. ,]u,b. ,/. Lqal .IJpd. 25, 
13 I- -138. 
tiimura. ‘I‘. Cyr ‘l‘akahashi. H. (1982). Mechanical prcl1)crtic.h r~f cross wrtion ~~I‘lou~~r limh long bone\ 111,Jomot~ 
m,m. ,/. anthrop. Sec. :Vippon 90, 105-l 17. 
Kotani, 1’. ( 1981). Evidencr of plant cultivation in Jomon Japan: Some m~plicarions. In (‘1‘. L’mcsn~~. Ed. I hn 
Ethnolo,@cnl Studier, pp. 201-212. Osaka: National Lluseum of Ethnology. 
Laqun, I.. E. (1982). Mechanical function and hone rcmodrlinq. In iG. Sumwr-Smith. Ed. I Hone in f.Ym~coI 
Orthopnedirr, pp. 273-304. Philadelphia: Saunders. 
I.xn)on. I.. E.. Hampson, W. C.J., Goodship. A. E. & Shah.,J. S. ( 19751. Bone drformation rrcordrrl in 1 i\o li.om 
strb gauges attached to the human tibia1 shaft. A& orthop. mnd. 46, 2.5f%268. 
Larsrn, <:. S. i 1982). The anthropology of St. Cathrrines Irland. 3. Prehistoric human hioloyic-‘11 A~ptatwn 
.4rt/hrop. f’ap. Am. Mur. nat. Hi.ct. 57, 159-270. 
I.ovcI~~. C. 0. (19701. Biomrchanical mrthods for thr analysis of skcle*tal variation with an applicdtiou b\ 
u~rnparison of thv theoretical diaphyscal strrngth ol’platycntmi( and vuricncmic tihiaa. Thesis. I’ni\cr\it\ qji’ 
.LI;iasachusetts. 
I.ovrj~~~, C. 0.. Burstrin. .I\. H. & Heiplr, K. G. i 1976). ‘l’hc hiomechanical analvsis ol’honc str-rnqth: .1 mr~thwl 
and Its application to platycncmia. Am. J. p&s. .inth,op. 44, 489-X6. 
I.mqrr)-. CZ. 0.. Hriplr. K. G. & Burstcin, .4. H. i l’Ji?i. ‘l‘hr gait of.~\ustralr~pithccus. .im. ,/. ,@I\. :lnthr~~p. 38, 
7.5;-780. 
Lr~vcjoy. (:. <I., hlwsforth. R. P. & Armelagos, G. J. i 19821. Fi1.r drradrs ofskrletal hi&q as rrllrctrti in the, 
‘;\ mcrlcan .Journal ~rf Physical Anthropology. In (F Spencu. Ed. I .l Ifi\by a/ .-lmenmn &hrc~c~~/ .-lnthrop~dq~. 
/W/L/!&?U. pp. 32%336. Nrw York: Academic Press 
.tfacI,au~hlin. S. 51. & Bruce, 51. F. (1985). A simple uni\.aritr trchni(luc tbr detrrmininq WY fr<lrn fraqrn~~rltar~ 
fcmura: Its applicatiun to a Scottish short cist population. Am. J. ph+\. .Whrop. 67, 113--117. 
.\lann. R. A. (lSR2i. Biomcchanicsofrunning. In (R. P. X\lack. Ed.) Th~L;lotnn(fI.rernRunnirrq.Sp~~rtl. pr~ I 2’1. 51. 
Lollis: Lioshy. 
Xlartin. R. f 19%). Lehrbuch der .dnthropolqle. German): Fischrr. 
Martin. R. H. & htkinsun, P. J. (1977). Agr and sex-wlatrd chanqcs in the structurr and strenqth Irftht, IIII~~II 
fcm<lral shalt. ,/. B~omech. 10, 223-231. 
.\lmns. R ,J., Brcmhlr,, CJ. R. Cu (:amphrll.~J. (1077). A hiomcchanic.tl studyrll.intcrnal liution ofth~ tibi.ll Alli 
,]. Hmrwch. 10, 569-~‘,79. 
.\lorl-ivm. .J. B. ( 196!)). Function of the kneejomt in v;wious acti\,ities. Hiorned. Lqn,q. 4, 571%iH0. 
Xlorrivm. J. B. ( 19701. ‘lk mechanics of thr knee joint in wlatiun to normal walking-. ,/. Biomerh. 3, i I -6 I. 
Alurtlock. G. P. & Provost. (1. i 1973). Factors in thr division of labor h! WI: A cn,ss-cultural analysis. Ethnoio~~ 
12, _‘1%225. 
P~~rwn~. I‘. (;. 1191 41 ‘l‘hc characters of the English ttliqh-honr. .im. ,J. phvl. :lnthrop. 48, 2X-267. 
I’.~r~wcl\, I,>. (19801. Principles of construction of the lo\\w rstrcmtty. ‘I‘hrir significancr fix the strrssinq 111thr 
\kclvr~~rl of the ICC. In (F. Pauwels, Ed.) Biomrhanicr I)[’ thr I,ommotor _ippnratur., pp. 193~201. Berlin: 
Sprirlqcr-\‘rrlaq. 
414 (:. RUFF 
Ruff, C. B. & Hayes, W. C. (19836). Cross-sectional geometry of Prcos Pueblo femora and ribl.u-~ ,I 
hiomechanical investigation. II. Sex, agr. and side difircnces. aim. J. phyr. dnthrop. 60, 383-400. 
Ruff, C. B. & Hayes. b$‘. C. (in press). Srx differences in age-rrlatcd remodeling ofrhc femur and tibia. ,I. O&~/J 
RCS. 
RUE, C. B. Hayes, W. C. & Lotz, J. (1986). Sex differrnces in age-related remodeling ofthc femur and tibia. 7;-nns. 
Orth@. Res. SIX. 11 434. 
Ruff, C. B., Larson, C. S. & Hayes, W. C. (1984). Structural changes in the femur with the transition trr 
agriculture on the Georgia coast. Am. J. phy~. Anthrop. 64, 125-136. 
Rybicki. E. F., Simonen, F. A. & Weis, E. B. Jr. (1972). On the mathematical analysis of stress in thr human 
femur. J. Biomech. 5, 203-215. 
Stini, W. A. (1974). Adaptive strategies ofhuman populations under nutritional stress. In (F. E.,Johnston & E. 5. 
Watts, Eds) &social Interrelations in Pq%ulalion Adapfation, pp. 19-40. The Hague: Moutan. 
Timoshenko, S. P. & Gere, J. M. (1972). Mechnnicr oflciatenafs. New York: Van Nostrand Reinhold. 
Trinkaus, E. (1976). The evolution ofthe hominid femoral diaphysis during the Upper Pleistocene in Europe and 
the Near East. 2. Morph. Anthrop. 67, 291-319. 
Trinkaus, E. (1980). Sexual differences in Neanderthal limb bones. J. hum. Glob. 9, 377-397. 
Trinkaus, E. (1983). The Shanidar Neandertals. New York: Academic Press. 
Van Gerven, D. P. (1972). Contribution of sizr and shape variations to patterns of srxual dimorphism of the 
human femur. Am. J. phys. Anthrop. 37, 49-60. 
Washburn, S. L. (1948). Sex differences in the pubic bone. Am. J #bps. Anthrop. 6, 199-207. 
Washburn, S. I.. (1949). Srx diffewnccs in the pubic bone of Bantu and Bushman. Am. ,/. ph_w. Anthrop. 7, 
424-432. 
Watanabe, H. (1977). The human activity system and its spatiotemporal structure. In (H. Watanabe. Ed.) 
Human Actiuicv $wtem: Its Spatiokmporal Structure, pp. 3-39. Tokyo: University of Tokyo Press. 
Webb, W. S. & Snow, C. E. (1945). The Adena people. l:nw. Ken. Rep. Anthrop. ilrch. 6. 
Young, M. & Ince, ,J. G. H. (1940). A radiographic comparison of the male and female pelvis. ,/. Ana!. 74, 
374-385. 
Appendix. Samples used in the study 
Cross-sectional geometric samples 
The categorization of samples in Table 3 by major subsistence technology follows that of 
the investigators themselves. In the samples from western New Mexico, properties of the 
hunting-gathering group are means of the data reported for “Early Villages” and 
“Abandonments” periods (A.D. 500-l 300)) while the agricultural group corresponds to 
the “Aggregated Villages” period (A.D. 1300-1540) of th’ 1s region (Cordell, 1984; Brock, 
1985; Brock & Ruff, in press). Although some agriculture was practiced during Early 
Villages and Abandonments,

Continue navegando