The natural history of human gait and posture
Part 2. Hip and thigh
C. Owen Lovejoy*
Matthew Ferrini Institute of Human Evolutionary Research, Division of Biomedical Sciences,
Kent State University, Kent, OH 44242, USA
Accepted 28 June 2004
The human fossil record is one of the most complete for any mammal. A basal ancestral species, Australopithecus afarensis, exhibits a
well-preserved postcranium that permits reconstruction of important events in the evolution of our locomotor skeleton. When compared to
hip and thigh are reviewed, including the unusual corticotrabecular structure of the human proximal femur,and our markedly elongated lower
limb. It is postulated that the latter may be more related to birthing capacity than to locomotion.
# 2004 Elsevier B.V. All rights reserved.
Keywords: Wolff’s law; Osteoporosis; Australopithecus; Cancellous bone; Hominid
While many vertebrates engage in bipedal locomotion,
the upright walking and running of modern humans are
unique because of their unusual evolutionary history; i.e.
they emerged as a terrestrial adaptation from ancestors
which were at least partially adapted to life in the arboreal
canopy. This is the second in a series of essays that discusses
elements of the genetics, comparative anatomy, and fossil
record that bear on anatomical modifications that accom-
panied the emergence of human bipedality. The issue of
what form of locomotion is most likely to have characterized
the immediate ancestor of hominid bipeds, as well as
potential reasons why bipedality was adopted in those
ancestors, will be discussed in a later contribution. The
current one is limited to several key morphological char-
acters of the human hip and thigh that differ dramatically
from their counterparts in our closest primate relatives.
While the human fossil record now extends to between 6
and 7 million years ago (MYA), its earliest portions are still
fragmentary, and substantial evidence of the ancestral
postcranium does not become available until the species,
Australopithecus afarensis, is first encountered at about
of these presentations .
afarensis. Humans differ demonstrably from our closest
living relatives, the African apes, in having long flexible
lumbar spines that permit lordosis and placement of the
head, arms and trunk (HAT) vertically above the joints ofthe
lower limb, thus allowing each to be completely extended
during upright locomotion. This was almost certainly the
initial adaptation that paved the way for the others that have
collectively yielded the modern human postcranium. A.
afarensis probably had six lumbar vertebrae, and its lumbar
spine was clearly at least as free to lordose as our own, and
far more so than any living ape .
In addition to its flexible lumbar column, the pelvis of
A. afarensis permitted effective and capable bipedality: its
sacrum was dramatically broadened, as were its ilia. The
latter were, in addition, superoinferiorly shortened and
anterolaterally angled to relocate the anterior gluteals (mini-
mus and medius) for effective abduction during single limb
Gait & Posture 21 (2005) 113–124
* Present address: Department of Anthropology, Kent State University,
Kent OH 44242, USA.
E-mail address: email@example.com.
0966-6362/$ – see front matter # 2004 Elsevier B.V. All rights reserved.
support. The kinetics of the hip of A. afarensis were largely
indistinguishable from those of modern humans , and
changes in the pelvis during the past 2–3 million years are
largely adaptations to facilitate birth of an expanding fetal
Here the hip is examined with respect to its structure in A.
afarensis and humans. These differ substantially from their
counterparts in living apes, a fact that provides interesting
clinical insights into human disease.
2. The corticotrabecular structure of the femoral neck
in higher primates
2.1. The corticotrabecular structure of the hominid femoral
Whereas the femoral neck of apes exhibits a complete
ring of cortex at its neck/shaft juncture (Fig. 1), humans bear
distinctly thickened cortex only near the inferior portion of
this interface, and often exhibit a complete absence of any
appreciable cortex in its upper (superior) portion [2,3]. This
is significant since the femoral neck of the limb in stance
phase is loaded as a cantilevered beam, supporting both the
HATand the limb in swing phase or approximately 4/5-body
weight (Fig. 2). This difference in cortical distribution
therefore signals different joint loading, different develop-
mental patterning between humans and apes, or some
combination of these two.
2.2. Two alternative explanations of the specialized
structure of the human femoral neck
One potential explanation involves the uniquely human
abductor apparatus. The primary abductors, the anterior
gluteals, as well as various ‘‘auxiliary’’ ones (e.g.
pyriformis, gemelli, obturators) produce a strong horizontal
force component along with the vertical one that prevents
pelvic tilt during single limb support. This component may
eliminate a significant portion of the tensile stress during
single support that emanates from the femoral neck’s
cantilevering. As shown in Fig. 2, abductor contraction
should result in a progressive decrease of compressive stress
along an inferomedial to superolateral transect of the neck/
shaft interface . Since chimpanzees lack an abductor
‘‘apparatus,’’ sufficient transduction ‘‘signal’’ may be
consistently produced in the upper regions of their neck/
shaft interface to encourage bone production, or more likely
to maintain it, during ontogeny.
This explanation has strong experimental and inferential
action of the abductors and body mass load while
simultaneously measuring surface strain in the femoral neck
. They found that altering knee valgus, which equally
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124 114
Fig. 1. Paracoronal sections of normal adult chimpanzee (A) and human (B) proximal femurs, along with X-rays of 1 cm slab cross-sections taken at the
locations indicated by the vertical lines (these sections are from different specimens). While the metaphyseal cortex along the inferior portion of the neck is
roughly similar in the two specimens, the distribution of cortical bone within the necks differs dramatically. The superior neck of the chimpanzee contains
abundant cortex, whereas there is virtually none in the corresponding location in the human. While the shaft metaphyses are therefore similar, only the
chimpanzee neckalso resemblesa typicalmetaphysiswithrespectto its corticaldistribution,in relationto the physisforthe femoralhead.The thickenedcortex
in the chimpanzee probably reflects maintenance and consolidation of trabeculae in the region of greatest strain (i.e. its superior surface) consequent to
cantilevering of the neck during the single support phase. Chimpanzees lack a well-developed abductor mechanism (c.f. Fig. 2) and demonstrate a
Trendelenburg gait when walking bipedally. However, the strain history responsible for such cortical robusticity is most likely generated not during bipedal
progression (which accounts for only a small proportion of their locomotor activity), but rather during arboreal climbing and terrestrial knucklewalking. For
further discussion see text and Fig. 2.
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124115
Fig. 2. Stress distribution in the human femur during single support phase (see inset). (A) Space diagram showing lines of action of body mass and shaft
reaction; these two forces constitute a couple whose separation is NL (neck length) and generate stress as shown in (C) (A = abductor force; JRFh= hip joint
reaction force; JRFk= knee joint reaction force; BW = body weight; SRF = shaft reaction force). (B) Typical bending stress distribution generated from
that also generated by simultaneous bending yields the distribution shown here, with minimal tension at the top of the femoral neck and maximum compression
at the bottom. Chimpanzees, which lack an effective abductor apparatus and which are active arboreal clamberers, apparently fail to habitually generate
sufficient neck compression to eliminate tension along the top of the femoral neck. This model is consistent with collagen fiber orientation in humans and
modifies neck/shaft angle since the femur is a rigid body,
greatly altered the surface strain pattern in the femoral neck.
Similarly, Carter and colleagues, using finite element
analysis, demonstrated that typical human loading patterns
during stance phase lead to much higher stresses in the lower
part of the neck, while an absence of abduction leads to high
bending stresses and a predicted distribution of bone much
tion, however, presumes that osteoblasts transduce primarily
signals of bone strain magnitude. However, osteoblast
response systems are likely to be exceedingly complex .
An alternative explanation is that differences in human
and chimpanzee hip structure result from dissimilarities in
pattern formation. The proximal femoral epiphysis ‘‘begins
life’’ as a single structure and the isolation of the head and
greater trochanter does not occur until later in ontogeny. The
developmental biology of the human hip is unusually
complex [9,10]. Throughout most of development the head
and greater trochanter remain connected via a distinct
‘‘intraepiphyseal physis’’ along the top of the femoral neck.
Not until this band-like physis matures and synostoses does
any opportunity occur for subperiosteal (i.e. in contrast to
subchondral) bone deposition. Indeed, during the age period
of 9–12 years this very special interepiphyseal region
‘‘develops a significant fibrocartilaginous component and
... membranous rather than endochondral bone formation
.... Extensive microvascularity is present in the superficial
layers of the fibrous tissue along the posterosuperior femoral
neck, along with a distinct ... histologic pattern’’ [10: 154].
It has been demonstrated (reviewed in ) that dramatic
differences in pelvic pattern formation accompanied the
adoption of bipedality. Similar changes are likely to have
occurred in the proximal femur. Perhaps it is not simply
roentgenographic trivia that ‘‘[n]o other joint is so much
retarded at birth as the [human] hip joint’’ .
of different hip loading patterns in chimpanzees and
humans. Kalmey and Lovejoy  used circular polarized
light to study cartilage fiber orientation [13–16] in the
femoral necks of both taxa. As predicted, the chimpanzee
superior femoral cortex was highly birefringent, i.e.
predominantly tensile loading during deposition, while that
of modern humans, and the inferior cortex of both taxa,were
less so indicating that compression predominates at these
sites during loading.
Perhaps a combination of both hypotheses is most able
to account for human-chimpanzee differences. It is not
irrelevant that Wolff developed his ideas largely from the
proximal femur, whose trabecular structure is unusually
complex and unlike those of other joints such as the distal
femur in which trabeculae are more simply oriented. Wolff
attributed proximal femoral complexity to active trabecular
reorientation to internal strain patterns; however, it is also
readily explicable by interactive growth within its three
physes—those of the head, greater trochanter, and the
To be sure, connective tissue cells are known to be
exquisitely sensitive to strain and may generate/maintain
extracellular matrix whenever even minimally stimulated
. However, if the disposition of trabeculae in the
proximal femur is determined primarily by pattern forma-
tion, and bone is simply retained wherever it experiences a
sufficient strain signal, or lost whenever it does not , the
complexity of the human proximal femur is almost entirely
obviated. Our abductor apparatus may reduce strain in the
upper portions of the femoral neck to such an extent that
trabeculae are continually lost (Fig. 2), whereas in the much
more acrobatic large bodied apes, the signal is sufficient to
superior trabeculae, so much so that they eventually appear,
at least radiologically and macroscopically, as a ring of
cortex(Fig. 1). In humans progressiveloss of some so-called
‘‘arcing trabeculae’’ leads to eventual formation of ‘‘Ward’s
Triangle,’’ [19,20] which remains more radiopaque in other
The reader will likely recognize that the three primary
thresholds in the above model broadly parallel those of
Frost’s ‘‘Mechanostat Hypothesis’’ (see  for an excellent
summary). With specific reference to its quantitative details,
the human abductor apparatus may reduce loads in the
superior cortex to levels below Frost’s ‘‘Trivial Loading
Zone’’ (50–200 me), whereas in apes lacking this apparatus,
loads probably fall within the boundaries of his intermediate
‘‘Physiologic Loading Zone’’. Similarly, as the juncture of
the ape femoral neck and shaft are approached laterally,
bending stresses may be presumed to rise until they exceed
the upper limit of this zone (i.e. >2000 me), leading to a
progressive consolidation of trabeculae along a medial to
lateral transect (c.f. Fig. 1) (see  for a discussion of the
effects of geometry on neck structure and failure).
Frost’s model is complex and beyond the scope of the
current discussion. However, a principal reason for its
complexity is its attempt to account for developmental
modeling, adult remodeling, bone maintenance, and fracture
repair by modified versions of cellulars response to strain.
Some of this burden would be lifted if such response
protocols were more directly integrated with the anabolic
guidance provided by positional information during devel-
opment. Femoral structure in apes and humans may provide
a useful tool with which to sort some of the complexities of
2.3. The dramatic differences in cortical bone
distribution in the femoral necks of humans and other
apes have been confirmed
Contradictions in the literature concerning the above
noted differences in bone distribution warrant some discus-
sion here. Susman and Stern claimed that the superior cortex
of the human femoral neck did not differ from that in
chimpanzees, but based this conclusion on a single photo-
graph of an AP radiograph of a human femur and single AP
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124116
radiographs of orang, chimp, and gorilla . Stern recently
continued to assert this view, citing Rafferty, who examined
femoral neck structure in cercopithecoids and strepsirhines
[small bodied primates such as lemurs and lorises]: ‘‘She
found the distribution of cortical bone in the femoral necks
of these two groups, most species of which are arboreal, to
be similar to that of humans‘‘ .
Closer examination contradicts this conclusion. Rafferty
reported that ‘‘the superior portion of the femoral neck is
devoid of trabecular bone’’ in her strepsirhine samples .
This is pivotal in light of mammalian trabecular patterns
. Unlike diaphyseal cortex, mammalian trabecular
dimensions do not behave allometrically, but instead
maintain relatively constant individual spicule thickness
regardless of body size. This is not unexpected, since they
‘‘begin life’’ as calcified chondrocyte columns within the
growth plate, but its implications are nevertheless far
reaching. The functional roles of trabecular compliance,
stiffness, and internal structure (e.g. the degree of
trabeculae/cortical connectivity) are substantially altered
by body mass. Primates with body masses almost an order of
magnitude less than those of large hominoids, (most of
Rafferty’s taxa weighed less than 8 kg), are therefore
especially inappropriate models for investigating trabecular
disposition in apes and humans.
Rafferty’s measurements of inferior and superior cortex
were made directly from simple AP radiographs, which are
often difficult to assess, even in large bodied primates. Case
in point: consider measuring cortical thickness in the
hominoids shown in Fig. 3. Moreover, her measurements
were made at the midpoint of the femoral neck, and included
the head as part of its overall length. This method causes
small primates with relatively short necks to be assessed
longer necks. This is important because the superior cortex
is not evenly distributed in hominoid femora. Rather it
progressively thickens laterally (see Figs. 1 and 3) as
demonstrated by high quality CT scans [27,28]. Thus, the
superior/inferior ratio also increases laterally, and must be
assessed nearer the neck/shaft interface. In addition, the
head and greater trochanter in many small primates develop
from a single asseous epiphysis. In these species cortical
thickness in the superior neck largely reflects the size and
density of that epiphysis and not consolidation of trabeculae,
which largely explains the poor correlation between cortical
thickness indices and locomotion in strepsirhines . Such
species are, once again, inappropriate analogs with which to
investigate cortical thickness in large bodied hominoids.
Even given these caveats, Rafferty’s data still contradict
Stern’s assertion. Fig. 4 shows her measurements of superior
and inferior cortical thickness compared to the cube root of
body mass for each species (see also ). Only one non-
human sample shows a significantshift inposition inthe two
graphs. Moreover, the ratio of bone in the upper and lower
portions of the femoral neck is not the datum of import; it is
the regular absence of cortical bone in the superior neck of
humans. Every primate illustrated in Rafferty’s study shows
a clear opacity indicative of substantial amounts of superior
cortex entirely homologous to that seen in the apes shown in
Fig. 3. Only humans lack such opacity (Fig. 5).
Stern concludes his recent review by referencing text-
book radiographs of the proximal femora of ambulatory
patients with cerebral palsy (CP), noting that they ‘‘appear to
illustrate the same general pattern of femoral neck bone
distribution as that found in people who walk normally’’
. This excellent observation provides substantial evi-
dence of the ‘‘developmental hypothesis’’ as discussed ear-
lier, but nevertheless conflates the primary issue. Locomotor
skills in patients with CP vary widely, and it is perhaps not
surprising that more motorically involved patients with CP
may not develop robust superior cortical bone. Femoral
strain during slow or assisted walking must be well below
normal thresholds, as some patients with CP are ambulatory
even with subluxed hips . However, while patients with
CP may induce subnormal strain signals, their diaphyseal
growth continues in a manner roughly similar to that in
children with no physical disabilities, although they almost
certainly develop subnormal bone mineral density (BMD)
[32–34]. Once again, it is not the presence of robust cortex in
the superior neck of primates that must be explicated, but its
absence even in elite human athletes.
2.4. Understanding the evolutionary biology of the human
proximal femur may have important clinical implications
‘‘convenience.’’ Its inferior part is simply the medial portion
of homologous locations of other long bones. It undergoes
typical active subperiosteal/endosteal turnover as do other
growing metaphyses. The superior portion of the neck,
however, has an entirely different developmental history, i.e.
its homologue in other joints is the subchondral region of the
metaphysis, and like these other joints is overlain by only a
thin cortical plate. Given the importance of trabecular bone
compliance in modulating articular cartilage stress  and
the high compressive loads imposed by our abductor
apparatus, it might play a significant role in stress dissipation
within the joint.This wouldbe consistent with itsextensively
thinned superior cortex, which, if thicker, would greatly
sharing with its trabeculae. The absence of superior cortex
permits such compliance, and it is not until later life, when
trabeculae have been systemically compromised by osteo-
porosis, that this otherwise potential cartilage sparing
arrangement becomes subject to traumatic failure.
2.5. The corticotrabecular patterning of australopithecine
proximal femora was human-like
How old is this specialized distribution of bone in the
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124117
such as A. afarensis and A. africanus (an early descendant
species) are rare, because radio-opaque matrix is often
embedded and mineralized in ways that make it difficult to
distinguish it from bone. There are a few exceptions (see also
below). One is specimen MAK-VP 1/1, a nearly adult femur
(the trochanters are fully fused and the head was partially
fused but lost during fossilization) from the site of Maka,
, is completely free of any included matrix, and as a
consequence both CTand conventional radiography produce
vividly clear images of its internal structure  (Fig. 6). In
addition, the femoral necks of two other specimens of A.
afarensis from another Ethiopian site (Hadar) were broken
near their neck/shaft junctions, allowing the necks to be
viewed directly. Even though they exhibit a solid surface of
cortex can be visually separated from infill, and that it thins
precipitously in their superior portions, as it does in modern
humans (for example, see A.L. 128-1 in Fig. 5) . In
addition, high-resolution CT scans of two South African
specimens of A. africanus show a clearly hominid pattern at
their neck/shaft interface .
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124118
Fig. 3. Standard AP radiographs of (A) Gorilla, (B) Human, (C) Chimpanzee, (D) Orangutan. Note the presence of dense, cortex-like bone at the neck-shaft
junctionin the non-human genera,andcompare to the physicalsectionsshownin Fig. 1. Suchbonemost likelyrepresentscoalescence oftrabeculae in response
to strain generated during single support in the absence of a strong abductor apparatus.
3. Femoral length and its role in human evolution
3.1. The femur of A. afarensis was relatively and absolutely
shorter than that of Homo sapiens
Based on a variety of ratios it is clear that the femur is
relatively longer in modern humans than in A. afarensis, and
out tibia is probably longer as well (reviewed in ). Over
the course of the Pleistocene, the hominid lower limb
elongated. Why? An attempt will be made to answer this
question shortly, but before doing so, it may be best to view
data in their raw, rather than relative form, in order to
maintain an objective understanding of the implications of
The length of the femur in A.L.288-1, the only specimen
of A. afarensis in which it can be reliably investigated, is
approximately 280 mm long. Its humerus is 236 mm. These
yield a humerus/femur ratio of 0.84. This ratio in modern
apes hovers around 1.00, in modern humans its mean is
both African apes and humans . It is therefore clear that
the femur had undergone some elongation in A.L.288-1,
which almost certainly increased stride length. It has long
been presumed that this was an adaptation to reduce energy
consumption. This is doubtful, however, since human
locomotion is generally more costly than it is in other
mammals of equivalent body mass . After reviewing
data on energy consumption in human walking and running,
Alexander opined: ‘‘perhaps we should look for some other
major advantage for the human bipedal gait’’ . But
perhaps locomotion in A. afarensis was even more costly
than in H. sapiens? This is also unlikely.
3.2. Elongation of the lower limb does not reduce energy
costs, carries substantial disadvantages, and would not
have occurred without substantial positive selection
As just noted, Stern, along with many others, has argued
that ‘‘elongation [of the lower limb in bipeds]...is
energetically advantageous because of its effect on stride
length.’’ However, elongation also increases both the mass
and radii of gyration of limb segments Kramer recently
demonstrated  that walking with A.L.288-1’s moder-
ately shorter lower limbs would have probably been less
energetically demanding than in humans. Stern critically
evaluated Kramer’s study as follows:
The assumptions underlying Kramer’s conclusion are that
Lucy had the same movement profile as a modern human,
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124119
Fig. 4. Simple scatter plots of superior and inferior cortical thickness of the
difference in position in the two plots. Humans, however, show a substantial
difference in the two graphs. The only other taxa with a substantial shift,
size for this point was small (n = 4) and represents a composite of three
different species. From ; data from .
Fig. 5. Bivariate plot of total cortical bone (cross-sectional area) present in
the superior and inferior halves of femoral neck cross-sections (from both
CT scans and physical sections) in several hominoids. Each section or CT
scan was taken at the neck/shaft boundary and each was divided into equal
halves along a transect made at the superoinferior midpoint of the section
(for more detail see . Humansandthe A. afarensis specimen(A.L.128-1)
exhibit considerably less bone in the superior halves of their cross-sections
have higher ratios than CT scans because determination of the cortex/
trabeculae boundaryin the latter was automated(see ). See also Figs. 1, 3
that the masses of Lucy’s lower limb segments were
proportionately the sameas ina modernhuman, and that it is
most appropriate to compare energy use of the two species
when Lucyiswalking atabout 80%of the speed ofa modern
human .... Therefore, it is of little moment to me if the
calculation of energy-use based on Kramer’s assumptions is
correct or not. For me, the issue is the implausibility of the
Essentially, Stern’s objection is that Kramer assumed that
Lucy walked in the same manner as modern humans, but
with a shorter lower limb. Yet given this assumption, her
results show that no energy benefit accrues to humans as a
her assumptions that negate Stern’s argument (i.e. longer
limbs reduce energy consumption), but rather her results.
other experimental data. Differentiating the energy cost of
locomotor components is complex , but there is reason-
able certainty that total vertical force (i.e. maintaining the
body’s center of mass above the ground), rather than moving
the limbs, is the principal determinant of total cost. It ranges
from 75% to 90% [42–44]. Most importantly, cost is even
independent of limb number, since the mass-specific energy
required to move a given distance is largely the same for
cockroaches, ghost crabs, mammals, and birds .
Therefore, even if the highest estimates of swing phase
cost are correct , substantial alteration of a limb’s length
would still have only minimal effects on the animal’s total
energy budget [46,47].
Taylor and colleagues experimentally examined the
‘‘limb length’’ hypothesis some years ago. They measured
oxygen consumption in cheetahs, gazelles, and goats, which
‘‘provide the opportunity to quantify the effect of limb
design on the energy cost of running .... [They] found that,
despite large differences in limb configuration, the energetic
cost of running in cheetahs, gazelles, and goats of about the
(: 848). At autopsy, the average distance to the center of
mass of the limbs from their pivot points (shoulder or hip)
was determined to be 18 cm in the cheetah, 6 cm in the goat,
and only 2 cm in the gazelle, far in excess of any that could
possibly have differed between A. afarensis and modern
humans. Taylor later concluded: ‘‘[a] general rule of running
energetics is that energy cost of running does not depend on
limb design’’(: 193). ‘‘[A]ll one needs ... to arrive at a
running] is... knowledge of the speed [and] the animal’s
body weight’’ .
Bipedality is a dangerous form of locomotion, and limb
elongation in an habitual biped carries significant locomotor
liabilities. It greatly increases the probability of injuries
common to the human hamstrings when their length/tension
phasing lacks harmony with the limb’s inertial properties as
it nears ground contact, or when the limb suffers a sudden
erratic load during dashes or rapid changes in direction.
Limb elongation substantially increases the torques gener-
ated about the knee and ankle joints and their ligaments
during such events. Traumatic injuries of the ankle and knee
are common in humans, and their debilitating effects have
been undoubtedly a far more substantial selective force on
our lower limb than has energy consumption. The above
question therefore bears repeating—what reproductive
advantage could so selectively outpace these substantial
disadvantages that it would still encourage limb elongation
in humans and their ancestors, once they were bipedal?
One possibility is running velocity. A comparison of
small and large animal species over a wide speed range
indicates that animals moveina dynamically similar fashion
when they travel at speeds that translate to equal values of
the dimensionless Froude number [51,52]. Details of the
derivation of this number can be found in Alexander ,
but here it need only be noted that it reduces to the square of
velocity divided by the product of limb length and the
gravitational constant. Thus given dynamic similarity (i.e.
Froude numbers tend not to change with body size), an
increase in limb length increases velocity but at consider-
able geometric disadvantage. Making Lucy’s femur the
same length as a small human female yields only about a
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124120
Fig. 6. Standard (Faxitron#) AP X-ray of MAK VP 1/1 from A. afarensis
(see text). The specimen has suffered postmortem loss of its femoral head,
but was developmentally adult. Note the complete absence of any cortical
bone in the upper portion of the neck as it nears the neck/shaft junction; c.f.
legend of Fig. 1. Compare to other hominoids shown in Figs. 1, 3 and 5.
10% increase in velocity. This seems a minimal gain given
the substantially increased chance of injury.
3.3. The elongated femur of A. afarensis might have been a
response to thermoregulation
Two other explanations of lower limb elongation have
been suggested. One is that elongation of the lower limb is
merely an allometric consequence of increased body size
[54–58]. What is lacking in this argument, however, is an
actual mechanism by which only hindlimb length can be
expected to differentially increase relativeto forelimb length
as a consequence of body size. ‘‘Allometry’’ is an
observation, not a mechanism.
A related suggestion is that the hominid lower limb has
selective mechanism that underlies ‘‘Allen’s Rule’’ (limb
length decreases with increasing latitude) [55,56,59,60].
Here, a known selective agency is provided, but why are
such changes not already present in A. afarensis? Allen’s
Rule is observed within contemporary species. The
antecedents of ‘‘Lucy’’ had probably been bipedal for at
least several hundred thousand years. It is possible that A.
afarensis had not yet regularly occupied areas of savanna as
open as those in use by later hominids. However, a more
recent species, A. garhi, exhibits lower limb lengths equal to
those of modern humans, but upper limbs that were as long
as those of A. afarensis . Therefore, the upper limb
underwent substantial shortening during a time period in
which regular savanna occupation was very probable for this
species and its descendants. Since lower limb elongation
carries locomotor disadvantages that do not accrue to upper
limb lengthening, it is difficult to reconcile upper limb
shortening with lower limb elongation as adaptations to
radiational cooling—unless the two limbs were under
substantially different selective regimens.
That is certainly possible, given that upper limb
shortening is probably a consequence of antebrachial
shortening—itself a probable pleiotropic effect of changes
in hand proportion that improved the power grip .
However, another selective agency, potentially much more
pressing than radiational cooling, is also known to have
faced descendants of A. afarensis.
3.4. Elongation of the human lower limb may be a
collateral consequence of adaptations to birthing strictures
Birth canal size represented a continual anatomical crisis
throughout the Pleistocene. The rate and scale of human
brain enlargement during this period were remarkable and
constituted the principal focus of anatomical change in the
proximal femur and pelvis. Modern human infants are
excessively altricial  and must be jettisoned ‘‘before
implications are simple: during the Pleistocene, selection
was making births progressively more ‘‘premature’’ in order
to ensure parturition through a birth canal made constrictive
by pelvic adaptations to bipedality [1,65]. But such pre-
mature birth would have decreased survivorship. The alter-
native—a more ample birth canal—would thereby have
been under intense natural selection.
One of the most important advances to our understanding
of evolution during the past decade has been the realization
that many Type 2 traits (non-selected collateral changes that
accompany others [3,66]) often accompany those that are the
principal targets of selection (Type 1). While it is not yet
possible to posit a definitive causal link between limb elon-
gation and birth cavity enlargement, it is nevertheless not
difficult to speculate reasonably about possible develop-
mental mechanisms that would cause such an association,
lengthening is correct, then there should be a relationship
between lower limb length and birthing capacity.
Tague has amassed a body of powerful evidence in favor
of this hypothesis. It is certainly not irrelevant to the issue of
lower limb length that obstetricians regard ‘‘a pregnant
woman’s stature as an anthropometric correlate of her
reproductive efficiency’’ , and that short females have
‘‘higher rates of cephalo-pelvic disproportion ..., cesarean
section, stillbirth and perinatal mortality than tall females’’
(: see references therein). The supporting data are robust
[68–78]. ‘‘In view of the evidence of pelvic contraction in a
high proportion of short women it is not surprising that
perinatal deaths due to birth trauma show a clear association
with maternal stature’’ . In fact, Stewart and Bernard
 reported that mechanical difficulty during birth was 14
times higher in short females than in tall ones!
The most obvious presumption, given the above observa-
tions, is that ‘‘big females [should] have big pelves’’ —
though given careful scrutiny there is no a priori reason
why this should be so. Birth canal dimensions could easily,
in theory, be evolved without simultaneous elongation of the
lower limb, and to some extent probably have. Indeed,
Tague investigated numerous pelvic and femoral dimen-
sions, and found generally positive but weak associations.
He concluded that selection may have acted primarily on a
stature threshold, i.e. females under 5 ft exhibit non-linear
However, we have learned anything about how devel-
opmental principles affect evolutionary ‘‘strategies’’ in the
past decade, it is that more often than not ‘‘adaptations’’ can
be far more complicated than they might be presumed at
first sight. Tague’s investigations of pelvic and femoral
dimensions were unusually thorough and incorporated a
number of ratios, including one of birth inlet shape
(anteroposterior diameter of the inlet/mediolateral diameter
of the inlet). The latter proved to be the most strongly
correlated with femoral length of all the characters he
investigated, though the critically important anteroposterior
dimension of the pelvic inlet also evinced an almost
equally strong positive relationship with femoral length
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124121
(P < 0.0001). Here it should be noted that the most
restrictive dimension of A. afarensis’s pelvis was, in fact, its
anteroposterior dimension [1,65,66]. Indeed, its mediolat-
eral dimension significantly exceeded that required for
passage of term fetal crania.
Sjogren et al. recently reported an unexpected conse-
quence of postcranial dimensions in mice in which the gene
for growth hormone receptor (GHR) had been inactivated
. They found not only a uniform reduction in skeletal
dimensions, but also substantial changes in the relative
lengths of some skeletal elements (such as, for example,
femoral/tibial length ratios). Could the strong relationship
between birth inlet shape and femoral length observed by
Tague be a consequence of selection on any of several
endocrinological pathways that impact differential skeletal
growth (GH, IGF1, IGF2, GHR, etc.) and pelvic form? Only
further in-depth research into these kinds of issues can
resolve the ‘‘dilemma of human lower limb length’’ and its
richly documented correlation with cephalopelvic dishar-
Lastly, the issue of why selection on female pelvic size
would also equally impact the structure of the male pelvis is
frequently noted. The ‘‘default’’ mammalian genotype is, by
definition, female. Any modifications of the underlying
positional information responsible for pelvic embryogenesis
will therefore alter its morphology similarly in both sexes.
Indeed, sexual differences in human pelvic structure appear
to be a consequence predominantly of differential hormone
expression, especially during adolescence, beforewhich it is
virtually impossible to determine the sex of a human
skeleton. Furthermore, those changes that result in the more
ample female birth canal are clearly the same processes that
produce forensically useful sexual identification characters
in females, such as the more open sciatic notch, the presence
of a ‘‘ventral arc’’ (an adductor insertion on the anterior
pubic body ), a longer superior pubic ramus, a more
obtuse sub-pubic angle, and so on.
3.5. When did elongation of the hominid lower limb and
changes in pelvic proportions first appear?
Only one early hominid fossil is available that might shed
some light on timing of the above events: KNM-WT-15000
(circa 1.8 MYA), a remarkably complete skeleton of a sub-
adult H. erectus from Kenya . Unfortunately, this
specimen tells us little about the relationship between
femoral length and pelvic form because it is male and its
proximal and femoral epiphyses are unfused. It is therefore
moot with respect to any of those changes in pelvic
proportions that accompany modern human female skeletal
maturity. In addition, the shape of its birth canal relies
almost entirely on conjecture because neither pubic bone
was preserved, its sacrum was fragmentary, and its
acetabulae are still patent.
The adult stature of KNM-WT-15000 was projected to
have been just shy of six feet (both femur and tibia were
nearly complete) . This is a marked increase relative to
A. afarensis. Ruff and Walker argue that this represents a
climatological adaptation, but this seems unlikely simply
because of the lack of similar adaptation in either the
species’ antecedents or its collateral contemporaries such as
A. robustus and A. bosei . It would seem more probable
that the limb was already at least partially elongated as a
byproduct of changes in pelvic proportion as discussed
Thelatterconclusion suggests that maximum permissible
stature may have been achieved in early H. erectus (actually
A. garhi—its antecedent) as an adaptation to increases in
term fetal cranial dimensions, and that once this statural
‘‘ceiling’’ was reached, selection for modification of birth
canal shape by other developmental mechanisms would
have become even more intense, as further elongation of the
lower limb would have resulted in a disproportionate
elevation of injury rates. Of course, once a favorable shape
had been obtained, simple overall enlargement, independent
of that associated with simultaneous lower limb elongation,
may have been favored and elicited by changes in pelvic
pattern formation. Indeed, there are clear specializations in
the growth processes of the human pubic symphysis that
provide strong support for this view [1,84,85].
These factors may therefore constitute a classic example
of stabilizing selection, in which greater divergence in either
direction from a mean phenotypeyields non-linear increases
in negative selection. In the case at hand, a shorter lower
limb reduces the risk of traumatic injury but may negatively
impact parturition. Only after emergence of more ‘‘locally
determined’’ anatomical adequacy of the birth canal, i.e. by
pattern formation rather than previously suggested endo-
crinological means, as well as world-wide occupation of an
array of vastly different climates made possible by cultural
sophistication, did H. sapiens then also evolve its current
compliance with Allen’s Rule.
As noted throughout this paper, a number of authors have
argued over the years that the locomotor anatomy and
behavior of A. afarensis were markedly distinct from those
of H. sapiens. As detailed above and previously, such claims
have consistently been grounded on adaptationist arguments
using osteological details even though their ranges of
expression are fully shared by both species. Of particular
importance in assessing such arguments is that they lack
virtually any Type 1 traits (for definition see [1,3,59]) save
those associated with either elongation of the human lower
limb or modifications of the human birth canal. Both of
these, as noted above, comprisevirtually all of the significant
biomechanical differences of the hip and thigh in the two
species, andeven thoughlimb elongation hashabitually been
argued to comprise an important adaptation to bipedality, the
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124 122
the lower limb has been elongated in later hominids; a
related to thermoregulation, birthing, or both, at least when
analyzed on purely mechanical grounds.
It should again be emphasized that in assessing the degree
to which A. afarensis was adapted to upright walking and
running, the issue in question is the degree to which it exhi-
bited a lower limb adapted to the mechanics of bipedal
progression, andnotwhetherits skeleton ‘‘looked like that of
a human.’’ Gorilla postcrania differ extensively from those of
chimpanzees, but both are arboreal climbers and terrestrial
knuckle-walkers. The skeletons of cheetahs and house cats
differ substantially, but both are digitigrade, cursorial, terr-
estrial quadrupeds. Once again, modern humans are not
simply bipeds: they are primate bipeds that gestate fetuses
extensively reshaped to this end.
The author thanks Melanie McCollum, Richard Meindl,
Phil Reno, Burt Rosenman, Maria Serrat, Gen Suwa, and
Tim White for critical readings of the manuscript and/or
valuablediscussionsabout itscontent,Linda Spurlock, Luba
Gutz, and Melanie McCollum for preparation of illustra-
tions, and the Cleveland Museum of Natural History for
access to specimens and substantial assistance in their
examination. This work supported by grants from the
National Science Foundation (BCS-9910211 and SBR-
pelvis. Gait and Posture 2004;21:95–112.
 Lovejoy CO. Evolution of human walking. Sci Am 1988;259:118–25.
femur and its bearing on the antiquity of human walking: applying
contemporary concepts of morphogenesis to the human fossil record.
Am J Phys Anthropol 2002;119:97–133.
Musculo-Skeletal System. Springfield, Ill: Charles C. Thomas; 1961.
 Carter DR. Mechanical loading history and skeletal biology. J Bio-
stress on prenatal and postnatal skeletal development. Clin Orthopaed
Rel Res 1987;219:237–50.
 Carter DR, Wong M. The role of mechanical loading histories in the
development of diarthrodial joints. J Orthopae Res 1988;6:804–16.
 Martin RB, Burr DB, Sharkey NA. Skeletal Tissue Mechanics. New
York: Springer; 1998.
 Ogden J. The development and growth of the musculoskeletal system.
In: Albright J, Brand R, editors. The Scientific Basis of Orthopaedics.
New York: Appleton Century Crofts; 1979. p. 41–103.
 Ogden J. Hip development and vascularity: relationship to chondro-
osseous trauma in the growing child. Hip 1981;9:139–87.
in Skeletal Roentgenology. New York: Grune & Stratton; 1968.
 Kalmey JK, Lovejoy CO. Collagen fiber orientation in the femoral
necks of apes and humans: do their histological structures reflect
differences in locomotor loading? Bone 2002;31:327–32.
 Boyde A, Riggs CM. The quantitative study of the orientation of
collagen in compact bone slices. Bone 1990;11:35–9.
 McMahon JM, Boyde A, Bromage TG. Pattern of collagen fiber
orientation in the ovine calcaneal shaft and its relation to locomo-
tor-induced strain. Anat Rec 1995;242:147–58.
 Riggs CM, Lanyon L, Boyde A. Functional associations between
collagen fibre orientation and locomotor strain direction in cortical
bone of the equine radius. Anat Embryol 1993;187:231–8.
 Boyde A, Riggs CM. The quantitative study of the orientation of
collagen in compact bone slices. Bone 1999;11:35–9.
 Takahashi I, Nuckolls GH, Takahashi K, Tanaka O, Semba I, Dashner
R, Shum L, Stavkin C. Compressive force promotes Sox9, type II
collagen and aggrecan and inhibits IL-1b expression resulting in
chondrogenesis in mouse embryonic limb bud mesencymal cells. J
Cell Sci 1998;111:2067–76.
 Cadet ED, Gafni RI, McCarthy EF, McCray DR, Bacher JD, Barnes
KM, Baron J. Mechanisms responsible for longitudinal growth of the
cortex: coalescence of trabecular bone into cortical bone. J Bone Jt
 Bruno RJ, Sauer PA, Rosenberg AG, Block J, Sumner DR. The pattern
of bone mineral density in the proximal femur and radiographic signs
of early joint degeneration. J Rheumatol 1999;26:636–40.
 Walker RA, Lovejoy CO. Radiographic changes in the clavicle and
proximal femur and their use in the determination of skeletal age at
death. Am J Phys Anthropol 1985;68:67–78.
 Skedros J, Mason M, Bloebaum R. Modeling and remodeling in the
developing artiodactyl calcaneus: a model for evaluating Frost’s
 Nakamura T, Turner CH, Yoshikawa T, Slemenda CW, Peacock M,
Burr DB, Mizuno Y, Orimo H, Ouchi Y, Johnston CC. Dovariations in
hip geometryexplain differences in hip fracture risk betweenJapanese
and White Americans? J Bone Miner Res 1994;9:1071–6.
 Susman RL, Stern JT, Jungers WL. Arboreality and bipedality in the
Hadar hominids. Folia Primatol 1984;443:113–56.
 Stern JT. Climbing to the top: a personal memoir of Australopithecus
afarensis. Evol Anthropol 2000;9:113–33.
 Rafferty KL. Structural design of the femoral neck in primates. J Hum
 Swartz SM, Parker A, Huo C. Theoretical and empirical scaling
patterns and topological homology in bone trabeculae. J Exp Biol
 Ohman JC, Krochta TJ, Latimer B, Lovejoy CO, Mensforth RP.
Cortical distribution in the femoral neck of hominoids. Am J Phys
 Ohman JC, Krochta TJ, Lovejoy CO, Mensforth RP, Latimer B.
Cortical bone distribution in the femoral neck of hominoids: implica-
tions for the locomotion of Australopithecus afarensis. Am J Phys
 Serrat MA, Reno PL, McCollum MA, Meindl RS, Lovejoy CO.
Multivariate comparison of divergent ossification patterns in the
mammalian proximal femur. Am J Phys Anthropol 2004;123(S38):
 Demes B, Jungers WL, Walker C. Cortical bone distribution in the
femoral neck of strepsirhine primates. J Hum Evol 2000;39:367–79.
 Metaxiotis D, Accles W, Siebel A, Doederlein L. Hip deformities in
walking patients with cerebral palsy. Gait and Posture 2000;11:86–91.
 Henderson R. Bone density and other possible predictors of fracture
risk in children and adolescents with spastic quadriplegia. Dev Med
Child Neurol 1997;39:224–7.
 Henderson R, Lark R, Gurka M, Worley G, Fung E, Conaway M,
its corollaries.Anat Rec
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124 123
adolescents with moderate to severe cerebral palsy. Pediatrics Download full-text
 Haydar A, Buyukavci M, Akcay F, Polat P, Yildiran A, et al. Bone
mineral density in children with cerebral palsy. Ped Intern
 Currey J. The Mechanical Adaptations of Bones. Princeton, NJ:
Princeton University Press; 1996.
 White TD, Suwa G, Hart WK, Walter RC, WoldeGabriel G, de
Heinzelin J, Clark JD, Asfaw B, Vrba E. New discoveries of Aus-
tralopithecus at Maka in Ethiopia. Nature 1993;366:261–5.
 White TD. Ape and hominid limb length. Nature 1994;369:194–9.
 Taylor C, Heglund N, Maloy G. Energetics and mechanics of terres-
trial locomotion. J Exp Biol 1982;97:1–21.
 Alexander RM. Characteristics and advantages of human bipedalism.
Cambridge: Cambridge University Press; 1991.
 Kramer PA. Modellingthe locomotor energetics of extinct hominids. J
Exp Biol 1999;202:2807–18.
 Heglund N. Running a-fowl of the law. Science 2004;303:47–8.
 Marsh R, Ellerby D, Carr J, Havalee T, Buchanan C. Partitioning the
energetics of walking and running: Swinging the limbs is expensive.
 Taylor CR, Heglund NC, McMahon TA, Looney TR. Energetic cost of
generating muscular force during running: a comparison of large and
small animals. J Exp Biol 1980;86:9–18.
 Taylor CR, Schmidt-Nielsen K, Raab JL. Scaling of energetic cost of
running to body size in mammals. Am J Physiol 1970;219:1104–7.
 Full R, Tu M. Mechanics of six-legged runners. J Exp Biol
 Kram R, Taylor CR. Energetics of running: a new perspective. Nature
 Fedak MA, Seeherman HJ. Reappraisal of energetics of locomotion
shows identical cost in bipeds and quadrupeds including ostrich and
horse. Nature 1979;282:713–6.
 Taylor CR, Shkolnik ADee, Razi Baharav DBA. Running in cheetahs,
gazelles, and goats: energy cost and limb configuration. Am J Physiol
 Taylor C. Relating mechanics and energetics during exercise. Adv Vet
 McMahon TA. Muscles, Reflexes and Locomotion. Princeton, NJ:
Princeton University Press; 1984.
 Donelan J, Kram R, Kuoiwa A. The effect of reduced gravity on the
kinematics of human walking: a test of the dynamic similarity
hypothesis in locomotion. J Exp Biol 1997;200:3193–201.
 Donelan J, Kram R, Kuo AD. Mechanical work for step-to-step
transitions is a major determinant of the metabolic cost of human
walking. J Exp Biol 2002;205:3717–27.
 Alexander R, Jayes A. Fourier analysis of forces exerted in walking
and running. J Biomech 1980;13:383–90.
 Ruff CB. Hindlimb articular surface allometry in Hominoidea and
Macaca, with comparisons to diaphyseal scaling. J Hum Evol
 Ruff CB. Climatic adaptation and hominid evolution: the thermo-
regulatory imperative. Evol Anthropol 1993;2:53–60.
 Ruff CB. Climate, body size, and body shape in hominid evolution. J
Hum Evol 1991;21:81–105.
 Jungers WL, Stern JTJ. Body proportions, skeletal allometry and
locomotion in the Hadar hominids. A reply to Wolpoff. J Hum Evol
 Wolpoff M. Lucy’s little legs. J Hum Evol 1983;12:443–53.
 Wheeler PE. The thermoregulatory advantages of large body size for
 Wheeler PE. The influence of bipedalism on the energy and water
budgets of early hominids. J Hum Evol 1992;23:379–88.
environments.J Hum Evol
 Asfaw B, White T, Lovejoy O, Latimer B, Simpson S, Suwa G.
Australopithecus garhi: a new species of early hominid from Ethiopia.
 Lovejoy CO, Reno P, McCollum M, Hamrick M, Cohn M. Evolution
of primate hands: growth scaling registers with posterior HOXD
expression. Am Zool 2000;40:1109.
 Gould SJ. Ontogeny and Phylogeny. Cambridge, MA: Belknap Press;
 Eccles JC. Evolution of the Brain: Creation of the Self. London:
 Tague R, Lovejoy CO. The obstetric pelvis of A.L.288-1 (Lucy). J
Hum Evol 1986;15:237–73.
 Lovejoy CO, Cohn MJ, White TD. Morphological analysis of the
mammalian postcranium: a developmental perspective. Proc Natl
Acad Sci 1999;96:13247–52.
 Tague R. Do big females have big pelves? Am J Phys Anthropol
 Adadevoh SWK, Hobbs C, Elkins TE. The relation of the true
conjugate to maternal height and obstetric performance in Ghanaians.
Int J Gynaecol Obstet 1989;28:243–51.
 Bresler JB. Maternal height and the prevalence of stillbirths. Am J
Phys Anthropol 1962;20:515–7.
 Camilleri AP. The obstetric significance of short stature. Eur J Obstet
Gynecol Reprod Biol 2000;12:347–56.
 DoughertyCR,JonesAD.Obstetricmanagementandoutcome related
to maternal characteristics. Am J Obstet Gynecol 1988;158:470–4.
 Gebbie DA. The influence of maternal height on the outcome of
pregnancy. J Trop Pediat Afr Child Health 1966;12:20–4.
 Kappel B, Eriksen G, Hansen KB, Hvidman L, Krag-Olsen B, Nielsen
J, Videbech P, Wohlert M. Short stature in Scandinavian women. Acta
Obstet Gynecol Scand 1987;66:153–8.
 Liljestrand J, Bergstrom S, Westman S. Maternal height and perinatal
outcome in Mozambique. J Trop Pediat 1985;31:306–10.
 Molloy WB. Labor in short women. Obstet Gynecol 1969;34:330–4.
and weight gain during pregnancy as risk factors for cesarean section.
Milit Med 1989;154:365–7.
 ParsonsMT, Winegar A,SiefertL,Spellacy WN.Pregnancyoutcomes
in short women. J Reprod Med 1989;34:357–61.
 Thompson AM. Maternal stature and reproductive efficiency. Eugen
 Stewart DB, Bernard RM. A clinical classification of difficult labour
and some examples of its use. Obstet Gyn Br Emp 1954;61:
 Sjogren K, Bohlooly Y, Olsson B, Coschigano K, Tornell J, Mohan S,
et al. Disproportional skeletal growth and markedly decreased bone
mineral content in growth hormone receptor ?/? mice. Biochem
Biophys Res Commun 2000;267:603–8.
 Budinoff LC, Tague RG. Anatomical and developmental bases for
the ventral arc of the human pubis. Am J Phys Anthropol 1990;82:
 Walker A, Leakey RE. The Nariokotome Homo Erectus Skeleton.
Cambridge, MA: Harvard University Press; 1993.
 Ruff CB, Walker A. Body size and body shape. In: Walker A, Leakey
RE, editors. The Nariokotome Homo Erectus Skeleton. Cambridge,
MA: Harvard University Press; 1993.
 Lovejoy CO, Meindl RS, Pryzbeck TR, Mensforth RP. Chronological
metamorphosis of the auricular surface of the ilium: a new method for
the determination of adult skeletal age at death. Am J Phys Anthropol
 Lovejoy CO, Meindl RS, Tague RG, Latimer BM. The senescent
biologyofthe hominoidpelvis:itsbearingonthe pubicsymphysisand
auricular surface as age-at-death indicators in the human skeleton,
Rivisita de Anthropologia Roma 73:31–49.
C. Owen Lovejoy/Gait & Posture 21 (2005) 113–124 124