Comparative 3D Quantitative Analyses of
Trapeziometacarpal Joint Surface Curvatures Among
Living Catarrhines and Fossil Hominins
M.W. Marzke,1* M.W. Tocheri,2 B. Steinberg,3 J.D. Femiani,4 S.P. Reece,5 R.L. Linscheid,6
C.M. Orr,1,7 and R.F. Marzke8
1School of Human Evolution and Social Change, Arizona State University, Tempe, AZ 85287-2402
2Human Origins Program, Department of Anthropology, National Museum of Natural History,
Smithsonian Institution, Washington DC 20013-7012
3Northwest Alliance for Computational Science and Engineering, Oregon State University, Corvallis, OR 97331-5501
4Division of Computing Studies, Arizona State University, Tempe, Arizona 85287-0180
51836 NE Noble Avenue, Corvallis, Oregon 97330
6Mayo Clinic, Rochester, MN 55905
7Institute of Human Origins, Arizona State University, Tempe, AZ 85287-2402
8Department of Physics and Astronomy, Arizona State University, Tempe, AZ 85287-1504
KEY WORDS thumb; morphology; stereophotogrammetry; laser scanning; geometric modeling
ABSTRACT Comparisons of joint surface curvature
at the base of the thumb have long been made to discern
differences among living and fossil primates in func-
tional capabilities of the hand. However, the complex
shape of this joint makes it dif?cult to quantify differen-
ces among taxa. The purpose of this study is to deter-
mine whether signi?cant differences in curvature exist
among selected catarrhine genera and to compare these
genera with hominin1 fossils in trapeziometacarpal cur-
vature. Two 3D approaches are used to quantify curva-
tures of the trapezial and metacarpal joint surfaces: (1)
stereophotogrammetry with nonuniform rational B-
spline (NURBS) calculation of joint curvature to compare
modern humans with captive chimpanzees and (2) laser
scanning with a quadric-based calculation of curvature
to compare modern humans and wild-caught Pan, Go-
rilla, Pongo, and Papio. Both approaches show that
Homo has signi?cantly lower curvature of the joint sur-
faces than does Pan. The second approach shows that
Gorilla has signi?cantly more curvature than modern
humans, while Pongo overlaps with humans and African
apes. The surfaces in Papio are more cylindrical and ?at-
ter than in Homo. Australopithecus afarensis resembles
African apes more than modern humans in curvatures,
whereas the Homo habilis trapezial metacarpal surface
is ?atter than in all genera except Papio. Neandertals
fall at one end of the modern human range of variation,
with smaller dorsovolar curvature. Modern human to-
pography appears to be derived relative to great apes
and Australopithecus and contributes to the distinctive
human morphology that facilitates forceful precision
and power gripping, fundamental to human manipula-
tive activities. Am J Phys Anthropol 141:38?51, 2010.
V
C 2009 Wiley-Liss, Inc.
The perennial debate about the role of prehistoric tool
use and tool making in the evolution of the human hand
has focused especially on thumb morphology and, to a
large extent, speci?cally on the joint at the base of the
thumb between the trapezium and the ?rst metacarpal,
hereafter referred to as the trapeziometacarpal (tmc)
joint. This focus is justi?ed by the functional importance
of an opposable thumb for manipulating objects with the
hands. The articular surfaces of the tmc joint are typi-
cally saddle-shaped in humans and other catarrhines
(Rose, 1992), facilitating opposition of the thumb pad to
the pads of the ?ngers, a function compatible with one-
handed manipulation of food and other objects.
Interest in fossil hominin tmc joint topography was
stimulated by the discovery of hand bones (O.H. 7),
along with other hominin elements, at Olduvai in 1960
(Leakey, 1960). These hand remains, which include a
right trapezium, were described by Napier (1962) and
subsequently were assigned as part of the type specimen
of Homo habilis (Leakey et al., 1964). Napier (1962)
described the O.H. 7 trapezium as having a saddle sur-
face for the ?rst metacarpal. Lewis (1977, 1989) noted
that this surface recalls the human one in having a
broader dorsovolar extent than in chimpanzees, although
he observed that some gorilla specimens also share this
1The term ??hominin?? refers to members of the tribe Hominini,
which includes modern humans and fossil species that are related
more closely to modern humans than to extant species of chimpan-
zees, Wood and Lonergan (2008). Hominins are in the family Homi-
nidae with great apes.
Grant sponsor: National Science Foundation; Grant number: IIS
998016; Grant sponsor: National Institutes of Health, Division of
Research Resources; Grant number: U42RR15090-01; Grant spon-
sors: The Smithsonian Institution Fellowship Program, The ASU
College of Liberal Arts and Sciences.
*Correspondence to: Mary W. Marzke, School of Human Evolution
and Social Change, Arizona State University, Tempe, AZ 85287-
2402. E-mail: mary.marzke@asu.edu
Received 11 November 2008; accepted 7 May 2009
DOI 10.1002/ajpa.21112
Published online 19 June 2009 in Wiley InterScience
(www.interscience.wiley.com).
V
C 2009 WILEY-LISS, INC.
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 141:38?51 (2010)
feature. Susman and Creel (1979, p. 329) considered the
surface to be ??very human in appearance.?? Day (1978)
inferred from the surface morphology a facility for a
wide range of thumb movements associated with both
power and precision thumb postures (including rotation).
Trinkaus (1989) conducted the ?rst quantitative study
of the articular morphology of this joint in modern
humans, Neandertals, and O.H. 7. He quanti?ed the cur-
vature of the trapezial facet for the thumb metacarpal
using caliper chord and subtense measurements and
found that the O.H. 7 surface is signi?cantly ?atter than
those of modern humans. His measured radioulnar cur-
vature of the surface in O.H. 7 is between one and two
SD below the mean for his modern human sample, and
at the lower limit of the modern human range. The
mean dorsovolar curvature is more than two SD below
the recent human mean, lying outside the range alto-
gether. Similar dorsovolar ?atness was also found in his
sample of Neandertals. His functional inferences from
this topography were that axial pronation may have
been enhanced by the relatively ?at ?rst metacarpal
saddle surface on the trapezium, and large axial joint
reaction forces could have been accommodated by the
surface.
TRAPEZIOMETACARPAL JOINT MOTION,
STRENGTH AND STABILITY
Formulating hypotheses about the functional implica-
tions of variability among genera in joint surface curva-
ture requires data on joint motion, strength, and stabil-
ity in living species (Hamrick, 1996). The following
review of studies on human and nonhuman catarrhine
genera reveals a limited knowledge base, but one that
can be applied to general predictions of functional vari-
ability from morphological variability.
Joint motion
The human tmc saddle joint has two degrees of free-
dom (movement around a ?exion/extension axis and
around an abduction/adduction axis) (see Fig. 1), but
conjunct motion about these axes produces pronation/
supination (Cooney et al., 1981). That is, pronation of
the metacarpal during opposition of the thumb to the
?ngers is a function of combined rotation about the ?ex-
ion/extension axis (in the trapezium) and the abduction/
adduction axis (in the metacarpal base) (Hollister et al.,
1992). Because these axes are not perpendicular to one
another or to the bones and are offset from the anatomi-
cal planes (Hollister et al., 1992; Brand and Hollister,
1999), the metacarpal comes into a position of opposition
to the remaining metacarpals as ?exion and abduction
occur, with the amount of pronation ?xed by the degrees
of ?exion and abduction (Hollister et al., 1992) (see Fig.
2). Validation of this kinematic model of tmc movements
by an in vivo analysis using an optoelectronic system
has been reported recently by Cerveri et al. (2008).
Relative motion of the mutual tmc joint surfaces has
been found to occur by sliding rather than by rolling
(Napier, 1955; Pieron, 1973; Martin et al., 1998). In vivo
measurements of active motion at the human tmc joint
were made by Cooney et al. (1981) using biplanar X-
rays. The metacarpal moved through a median ?exion/
extension range of 538 and abduction/adduction range of
428. The accompanying pronation/supination was 178.
Rose (1992) obtained comparable measurements for
human metacarpal abduction/adduction (46.38) using a
goniometer to record thumb metacarpal bone movement
relative to the trapezium in a skeletal collection. How-
ever, his measurement of ?exion/extension range (47.68)
was substantially lower than the range reported by Coo-
ney et al. (1981). Using the same approach for great
apes, Rose (1992) found mean ?exion/extention of 32.88
and abduction/adduction of 34.98, both lower than in his
human sample. Cercopithecine ranges were still lower,
Fig. 1. Trapeziometacarpal joint axes of motion after (A)
Napier (1955) and (B) after Hollister et al. (1992). Note that in
contrast to (B) the ?exion/extension and abduction/adduction
axes of (A) are perpendicular to one another and in the same
plane, and a longitudinal axis is assumed for metacarpal axial
rotation at mid-position on the trapezium.
Fig. 2. Dorsovolar and radioulnar curvature directions of
the trapeziometacarpal joint surfaces on the trapezium and ?rst
metacarpal. The ??set?? of the trapezium within the wrist varies
between individuals and taxa; therefore, in a comparative con-
text, the radioulnar axis will not always be consistently oriented
relative to the radius and ulna. However, in this work, we use
the term radioulnar by convention. The dorsal aspect of the
bone is ?xed by developmental homology.
39TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
with 21.48 ?exion/extension and 18.18 abduction/adduc-
tion.
Unfortunately, until more sophisticated studies are
made of nonhuman catarrhine tmc movement capabil-
ities, estimates of differences among catarrhine genera
will remain primarily qualitative and based upon com-
parative topography of the joint surfaces. However, it
does seem reasonable to conclude from Rose?s (1992)
measurements that overall the ranges of motion will
prove to be lower in great apes and Papio than in
modern humans.
Joint strength
A classic 3D study of static forces in the thumb
revealed that for an applied load of 1 kg by the thumb
and index ?nger during pinch, the tmc joint contact load
averages 12 kg (Cooney and Chao, 1977). For an applied
load of 10 kg in grasp (with the thumb opposing all the
?ngers), average tmc joint contact load is 120 kg (ibid.).
Accommodation of these loads depends upon the extent
to which orientation of the surfaces is normal to the
axial loads (Sarmiento, 1988; Hamrick, 1996) and the
degree of joint surface congruence in various grips.
Measurements of joint contact areas in humans have
been made in vitro for grasp, in which the thumb meta-
carpal is ?exed, abducted, and pronated to oppose the
four ?ngers, and for the lateral pinch posture, in which
the metacarpal is held against the side of the index ?n-
ger in adduction-?exion. Momose et al. (1999) found that
contact area was largest in opposition, with 53% of the
mutual trapezium and ?rst metacarpal surfaces in
contact. During lateral pinch, contact is in the palmar
compartment of the joint (Pellegrini, 2005).
Ateshian et al. (1992) and Xu et al. (1998) found that the
human female tmc joint is less congruent than in males
and concluded that joint contact areas in females are
smaller than in males and thus subject to greater stress.
They consider that this difference may be a factor in the
higher occurrence of degenerative joint disease in females.
The ?nding by Guthrie (1991) of a larger radius of cur-
vature of the tmc surfaces in humans compared with
chimpanzees indicates that a proportionately larger
surface of the human joints is normal to axial loads.
Measurements of joint contact have not been made in
nonhuman catarrhines, but behavioral observations of
chimpanzee and Hamadryas baboon manipulative behav-
ior indicate that maximum axial loading of the joint is
lower in these genera (Guthrie, 1991; Jude, 1993;
Marzke and Wullstein, 1996).
Joint stability
Stability of a joint is associated with its capacity to
resist displacement in a given direction (Hamrick, 1996).
Relatively large joint surface contact areas in grasp
(Momose et al., 1999) contribute to stability of the human
joint. However, there is less joint surface contact and a
tendency for the metacarpal to sublux dorsally during
pinch by the thumb and index ?nger (Eaton and Dray,
1982). Thus stability is dependent primarily upon liga-
ments (Bettinger et al., 1999, 2000; Bettinger and Berger,
2001; Pellegrini, 2005; Colman et al., 2007; Leversedge,
2008) It is suspected that laxity of the tmc ligaments, par-
ticularly, the deep anterior oblique or beak ligament (Pelle-
grini et al., 1993; Pellegrini, 2001), and the dorsoradial lig-
ament (Bettinger et al., 2000; Colman et al., 2007) can
lead to incongruent loading of the trapezial and metacar-
pal surfaces and to shear forces, as the unstable metacar-
pal slides and damages the articular cartilage (Ateshian et
al., 1994, 1995; Imaeda et al., 1994; Bettinger et al., 2000;
Pellegrini, 2005; Koff et al., 2006; Colman et al., 2007).
GOALS OF THE STUDY
The purpose of the study is to ?rst determine whether
signi?cant differences in tmc joint curvature exist among
?ve extant catarrhine genera (Homo, Pan, Gorilla,
Pongo, and Papio), and then to compare these data with
those of various hominin fossils.
To reliably compare living and fossil primates in tmc
joint shape and to make reasonable functional interpre-
tations, metrics that accurately describe the curvatures
at this joint in 3D are required. In this study, we use
two 3D approaches to quantify and compare the curva-
tures of the mutual trapezial and ?rst metacarpal joint
surfaces. The ?rst approach used stereophotogrammetry
with B-spline surface analysis and was applied to a
small sample of humans and chimpanzees while the sec-
ond used laser scanning with quadric surface analysis,
and was applied to a much larger catarrhine sample in
order to broaden the comparative scope of the study and
increase the statistical power of the results. It was
equally important to determine whether the two quanti-
tative approaches lead to similar results, because they
involve different and complex steps in the estimate of
joint surface curvature.
The present study quantitatively tests a null hypothe-
sis of shape equivalence in tmc joint curvature among
these ?ve extant catarrhine genera, and several fossil
hominin taxa are examined against this background.
Based upon previous studies in the literature, our own
qualitative observations of this joint in living and fossil
primates, and current understanding of functional corre-
lates to structural variability, we made the following pre-
dictions (all of which, we note, violate the null hypothe-
sis to a greater or lesser extent):
1. The mutual surfaces at the tmc joint in modern
humans should have less curvature than in African
apes and Pongo, but more curvature than in Papio. If
true, then this suggests that the modern human joint
topography is most parsimoniously interpreted as
derived from a more curved condition like that
observed in the great apes. Furthermore, this would
suggest that the human joint is less stable, joint con-
tact areas are primarily in the transverse plane
(normal to axial loads), and tmc stability is more
dependent upon ligament constraints.
2. Neandertals should have joint surface curvatures
more similar to modern humans than to the great
apes, but with relatively less curvature dorsovolarly
than in modern humans (Trinkaus, 1989). This would
suggest an ability to accommodate large axial loads
on the joint associated with muscle contraction during
strong pinch and grasp, but it also indicates possible
instability of the joint because of the lack of skeletal
constraints on dorsal sliding of the metacarpal on the
trapezium during pinch.
3. The O.H. 7 trapezium should show less curvature dorso-
volarly and radioulnarly than in modern humans, but
similar dorsovolar curvature to Neandertals (Trinkaus,
1989). This would suggest an unstable joint prone to sub-
luxation but capable of withstanding large axial loads.
40 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
4. The A. afarensis trapezium and ?rst metacarpals
should be more similar to those of African apes than
humans in surface curvature (Guthrie, 1991). This
would indicate a joint that was stable in pinch grips
against dorsal sliding of the metacarpal. Proportion-
ately less of the mutual joint surfaces would be ori-
ented normal to axial loads.
5. Thumb metacarpals from Swartkrans (SK 84 and
SKX 5020) will differ in joint surface curvature, judg-
ing by differences between them in other aspects of
morphology described by Susman (1988a,b).
These predictions were tested in three stages. First,
the topography of the mutual trapezial and ?rst meta-
carpal joint surfaces was quanti?ed, using the two previ-
ously mentioned approaches to 3D data acquisition and
analysis. Second, comparative statistical analyses of the
3D topography in the extant sample were performed.
Finally, the distribution of fossil hominin trapezial and
?rst metacarpal joint surface curvatures was examined
relative to those of the extant comparative sample.
MATERIALS AND METHODS
Stereophotogrammetry approach
Fifty-eight trapezia and 58 ?rst metacarpals from ?ve
human groups were photographed at the Smithsonian
Institution?s National Museum of Natural History
(NMNH) (Table 1). The groups represented were Aleuts,
South Dakota Native Americans, Chinese, from a group
of immigrant workers in Alaska, and African Americans,
and Euro-Americans from the Terry Collection. Mutual
trapezial and ?rst metacarpal joint surface images were
also obtained for 13 individuals in the Primate Founda-
tion of Arizona chimpanzee skeletal collection and for
casts of bones from A. afarensis (A.L. 333-80, trapezium;
A.L. 333w-39, ?rst metacarpal) and H. habilis (O.H. 7,
trapezium). Each joint surface was examined for evi-
dence of osteoarthritis. Criteria used in judging this
included signs of eburnation and osteophyte develop-
ment. No pathology was found in the chimpanzee speci-
mens, and only minor signs of osteoarthritis were found
in seven of the human joint surfaces.
Images of the mutual trapezial and ?rst metacarpal
joint surfaces were obtained following methods similar to
those of Ateshian et al. (1992). A Pixera digital camera
was used, operated with software run from a standard
PC (Pixera Camera Suite model 120 es camera, version
2.5). This camera took the place of the large format pho-
tographic Sinar ?lm cameras used by Ateshian et al.
(1992). The mutual joint surfaces were photographed
together in pairs, in most cases, each bone being posi-
tioned on clay in a plastic (Delrin) frame with a circular
front opening 1.5 in. in diameter. Calibration markers
were inserted into the front face of the frame at ?xed
positions surrounding the opening, providing both a
length scale and an accuracy check for the 3D stereopho-
togrammetric joint surface digitizing procedure. A grid of
intersecting lines was projected onto the joint surfaces,
TABLE 1. Comparative samples
Genus Trapezia Metacarpals Human population Trapezia Metacarpals
SPG APPROACH
Homo 58 58 Aleut 8 8
South Dakota Native American 13 13
Chinese 9 9
African American 14 14
Euro-American 14 14
Pan 13 13
LS APPROACH
Homo 113 121 Aleut and Pre-Aleut 11 15
South Dakota Native American 8 9
Chinese 23 25
African Bantu 4 4
Australian Aborigine 7 9
African American 30 29
Euro-American 30 30
Pan 47 46
Gorilla 44 47
Pongo 21 19
Papio 20 19
FOSSILS
Species Trapezia Metacarpals
Australopithecus afarensis A.L. 333-80 A.L. 333-58
A.L. 333w-39
Paranthropus robustus? O.H. 7 SK 84
Homo habilis
Homo sp.? SKX 5020
Homo neanderthalensis Kebara 2 Kebara 2
La Ferrassie 1 La Ferrassie 1
La Ferrassie 2 La Ferrassie 2
Regourdou 1 Regourdou 1
Amud 1
La Chapelle-aux-Saints 1
Shanidar 3
Shanidar 4
41TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
using a standard slide projector and a 35-mm slide made
by photographing an accurately drawn rectangular grid
(see Fig. 3). Three digital photographs were then taken,
along three directions with respect to an axis approxi-
mately normal to the frame?s front facial plane. The cam-
era angles used were 
308 to the left of this normal axis,
308 to the right, and along the axis itself. Digitizing and
3D modeling of the three photographs for each surface
was done using Photomodeler software (EOS Systems,
Vancouver, B.C.). This approach produced accurate
results, with digitizing errors rarely exceeding 0.5%.
The datasets obtained from Photomodeler, consisting
of 100 or more digitized points in 3D, were exported to a
C11 program (Steinberg, 1999), which ?tted each data-
set to NURBS (nonuniform rational B-spline) functions.
For this purpose, a subset of points representing the
boundary of each joint surface had to be separately digi-
tized, from the original photographs. Generation of each
boundary point set for a surface was straightforward,
using the original images that had been previously
marked for the full surface?s stereophotogrammetry.
The NURBS program calculated total surface area and
point-by-point root mean square (RMS) curvature. This
curvature is considered by Ateshian et al. (1992) to pro-
vide the most unambiguous quantitative index of overall
?atness of a surface. It is de?ned in Eq. (4) below and is
the inverse of a radius of curvature. The effects of over-
all joint size on curvature values were normalized follow-
ing the procedure of Ateshian et al. (1992), dividing the
radius of curvature for each joint by a radius represent-
ing its effective size, namely the square root of the joint
surface area. Figure 4 shows an example of a grayscale
display of the RMS curvatures of mutual metacarpal and
trapezial surfaces. Regions of lowest RMS curvature are
solidly ?lled and those with highest curvature are lightly
?lled. Finally, normalized RMS curvature values were
averaged over each metacarpal and trapezium surface,
leading to a single average value for curvature of each
surface as a whole.
Statistical analyses of the signi?cance of the observed
differences between groups, particularly in average nor-
malized RMS curvatures, were performed using stand-
ard two-sample resampling or randomization testing
techniques (Simon, 1974?1997; Mooney and Duval, 1993;
Bruce et al., 1997; Edgington, 1995).
Laser scanning approach
Joint surfaces on 252 metacarpals and 249 trapezia of
?ve extant catarrhine genera from the NMNH and the
Cleveland Museum of Natural History were examined
(Table 1). Pan is represented by a sample of P. troglo-
dytes, Gorilla by G. gorilla and G. beringei, and Pongo
by P. pygmaeus and P. abelii. Papio is represented by a
combined sample of P. anubis, P. cynocephalus, P. hama-
dryas, and P. ursinus. The sample of H. sapiens includes
individuals of several populations. The fossil specimens
(Table 1) were all casts, with the exception of Shanidar
3. Joint surface pathology or osteoarthritis was minimal
throughout the laser scanned sample.
All bones were scanned using the Cyberware Model 15
desktop laser digitizer. Each 3D model is a high-resolu-
tion triangular mesh consisting on average of more than
1,000 points per square centimeter. The mesh of each 3D
model was digitally segmented into articular and nonar-
ticular areas using commercial software. In most cases,
individual articular areas were segmented while refer-
ring visually to the actual bone. Following segmentation,
curvatures were calculated by ?tting modeled quadric
surfaces to the segmented joint surfaces (Christie and
Ridley, 1990; Tocheri et al., 2006; Tocheri, 2007; Tocheri
and Femiani, in press). In 3D, a quadratic surface has
the following equation:
z ? ax2 ? by2 ? 2cxy? 2dx? 2ey? f : ?1?
If the following rigid body transform is used,
^x
^y
^z
2
4
3
5
? R
x x0
y y0
z z0
2
4
3
5
?2?
Fig. 3. Photograph of mutual human metacarpal and trape-
zial joint surfaces, mounted on clay in a calibration frame (a
color version of this ?gure can be found in the on-line version).
A grid is projected onto the surfaces, and the intersections of its
lines provide points that are shown digitized on the trapezium
at the right. [Color ?gure can be viewed in the online issue,
which is available at www.interscience.wiley.com.]
Fig. 4. RMS curvature maps of mutual joint surfaces on a
human ?rst metacarpal and trapezium from the right hand,
viewed along the long (z) axis of the metacarpal. Gradations
from black to white re?ect the range from lower to higher
curvature.
42 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
where R is a 3 3 3 matrix with elements expressed in
terms of the constants a . . . f, the surface?s equation may
be put into the following quadric form:
^z ? A^x2 ? B^y2 ?3?
where A and B summarize the shape of the surface and
are functions of the parameters a through f. The A and
B coef?cients determine the principal curvatures (kmax
and kmin) of the ?tted quadric surface along the principal
directions at the origin of x, y, z space. Thus, A 5 kmax
and B 5 kmin. Figure 5 gives an example of a surface
that has a quadric shape given by (3), with A 5 2B.
This trapeziumlike surface has a convex curve in the y
(?exion/extension or dorsovolar) direction and a concave
curve in the x (abduction/adduction or radioulnar) direc-
tion. Figure 6 shows the shapes of the quadric surfaces
generated by varying signs and magnitudes of A and B.
If A and B have the same sign, the surface is of an ellip-
tical or parabolic form, whereas if A and B are different
in sign, the surface is of hyperbolic or saddle form. The
relative magnitudes of A and B thus re?ect the degree of
curvature in each of the principal coordinates of the sur-
face. The effects of overall joint size on A and B were
normalized by the square root of the joint surface area.
These normalized A and B values were used to calculate
the RMS curvature of the ?tted surface, where this
curvature is de?ned as
kRMS ?
??????????????????????????????????
k2max ? k
2
min
 
=2
q
?4?
As noted in the previous section on SPG, RMS curva-
ture provides what Ateshian et al. (1992) consider to be
??an unambiguous quantitative index of ?atness.?? A and
B were also used to calculate absolute, Gaussian, and
mean surface curvatures. These are de?ned as follows:
Absolute curvature (kabs):
Kabs ? kminj j ? kmaxj j ?5?
Gaussian curvature (kgauss):
Kgauss ? kmin3kmax ?6?
Mean curvature (kmean):
Kmean ? kmin ? kmax? ? ?7?
In all cases, it was assumed that the curvatures in the
surface along the principal directions reasonably approx-
imate the curvatures in the dorsovolar (?exion/exten-
sion) and radioulnar (abduction/adduction) directions,
thus giving numerical values for these two directions.
Numerical measures of the degree of congruence of
surfaces (e.g., metacarpal and trapezial) in (near) contact
with one another have been developed from differential
geometry and were applied to our data, following the
approach of Ateshian et al. (1992) to his sample of
human tmc joints. These measures are based upon maxi-
mum and minimum curvatures kmax and kmin. The basic
approach consists of introducing a surface that effec-
tively represents the difference between any two sur-
faces in contact at a point, that is, a surface whose nu-
merical values, when added to the lower of the two sur-
faces, yield the upper one. This difference surface, called
the effective surface, is represented as a curved surface
in contact with a plane, and its maximum and minimum
curvatures, denoted as kemax and k
e
min, are called congru-
ence indices. Their use is appropriate, because for sur-
faces that are perfectly congruent the effective surface is
Fig. 5. Example of a surface that has a quadric shape given
by (3), with A 5 2B.
Fig. 6. Shapes of the quadric surfaces generated by varying signs and magnitudes of A and B (left). All quadric surfaces that
are equidistant from the origin have the same amount of curvature (middle), which increases as one moves further from the origin
(right).
43TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
a plane, and its principal curvatures vanish. Also, the
less congruent the surfaces, the larger the equivalent
surface curvatures. Thus lower indices describe joint sur-
face pairs that are more congruent than those with
higher indices. Formulae for the equivalent surface cur-
vatures in terms of the separate curvatures of the two
surfaces in contact are given by Ateshian et al. (1992).
For our study, we applied these formulae to normalized
curvatures.
Statistical analyses of the observed curvature differen-
ces between groups were performed using the bootstrap
(Efron and Tibshirani, 1993; Manly, 1997). With 20 pair-
wise comparisons for each curvature variable, a standard
Bonferroni correction was applied to evaluate the statis-
tical signi?cance of each comparison (P \ 0.0025 (0.05/
20)). Differences in congruence index were analyzed with
two-tailed t-tests.
In the text, tables and ?gures regarding tmc joint
surfaces, ??metacarpal?? (??met??) refers to the surface on
the metacarpal for the trapezium and ??trapezial??
(??trap??) refers to the surface on the trapezium for the
metacarpal.
RESULTS
Comparison of Pan and Homo
(SPG and LS approaches)
The results from both approaches show that, as pre-
dicted, the human mean RMS curvature of the metacar-
pal surface is signi?cantly lower than the chimpanzee
mean (Table 2). Signi?cant differences in curvatures
along the dorsovolar and radioulnar directions parallel
the RMS surface curvature difference (Table 3).
SPG and LS RMS curvatures of the trapezial surface
vary signi?cantly among species in the same manner as
for metacarpal surface curvatures (Table 2), as does the
LS dorsovolar curvature (Table 3). However, humans
and chimpanzees are not signi?cantly different in trape-
zial surface radioulnar curvature (Table 3).
Comparisons among the extant genera
(LS approach)
Pairwise comparisons of genus means for normalized
dorsovolar, radioulnar, RMS, absolute, Gaussian, and
mean curvature of the ?rst metacarpal and trapezial
surfaces are given in Table 3. Table 4 summarizes the
rankings of genera in these curvature means. Bivariate
plots show the relation between dorsovolar and radio-
ulnar curvatures (Figs. 7 and 8), RMS and mean curva-
tures (Figs. 9 and 10), and absolute and mean curva-
tures for the metacarpal and trapezium surfaces (Figs.
11 and 12).
In all genera, the metacarpal is less curved than the
trapezium in the dorsovolar direction and more curved
than the trapezium in the radioulnar direction. At the
proximal surface of the ?rst metacarpal, the African
apes are characterized by marked RMS curvature,
whereas Papio is characterized by a much ?atter surface
(Fig. 9; Table 3). Homo falls between these two extremes,
with signi?cantly less RMS curvature than the African
apes but signi?cantly more than Papio. This pattern of
differences is the same in dorsovolar curvature. Particu-
larly striking are the marked RMS and dorsovolar curva-
tures in Gorilla compared with all the other genera.
Radioulnar curvature is greatest in the great apes and
least in Papio, Homo falling between with signi?cant
differences from both groups (Table 4).
The surface on the trapezium parallels that on the
metacarpal closely, with RMS curvature high in the Afri-
can apes, low in Papio, and intermediate in Homo (Fig.
10; Table 3). There are signi?cant differences in all pair-
wise comparisons, except the one between Pan and Go-
rilla. Dorsovolar curvature follows the same distribution
as RMS curvature (see Fig. 8), with the exception that
the difference between Homo and Papio is smaller and
not signi?cant (Table 3). In radioulnar curvature, the
TABLE 2. Normalized mean joint surface RMS curvature
values: signi?cant differences in Homo and Pan
Genus Approach
Joint
surface N
RMS
curvature SD P
Homo SPG Metacarpal 58 1.25 0.18 \0.000
Pan 13 1.62 0.38
Homo SPG Trapezium 58 1.37 0.29 \0.004
Pan 13 1.85 0.72
Homo LS Metacarpal 121 0.93 0.13 \0.001
Pan 46 1.29 0.14
Homo LS Trapezium 113 0.96 0.19 \0.001
Pan 47 1.41 0.25
TABLE 3. Pairwise comparisons of normalized genus mean curvatures1
Genus
Dorsovolar
curvature
Radioulnar
curvature
RMS
curvature
Absolute
curvature
Gaussian
curvature
Mean
curvature
Met. Trap. Met. Trap. Met. Trap. Met. Trap. Met. Trap. Met. Trap.
a. Homo 0.39b,c,e 20.83b,c 20.82ALL 0.43e 0.93ALL 0.96ALL 1.22ALL 1.27b,c,e 20.31b,c,e 20.35b,c,e 20.21b,c,d 20.20b,c,e
b. Pan 0.52a,c,e 21.32a,d,e 21.14a,e 0.41e 1.29a,c,e 1.41a,d,e 1.69a,c,e 1.75a,d,e 20.58a,c,e 20.52a,e 20.31a,c,e 20.46a,d
c. Gorilla 0.94ALL 21.37a,d,e 21.11a,e 0.41e 1.47ALL 1.45a,d,e 2.05ALL 1.45a,d,e 21.04ALL 20.56a,e 20.09ALL 20.48a,d,e
d. Pongo 0.45c,e 20.98b,c,e 21.15a,e 0.44e 1.26a,c,e 1.11ALL 1.61a,c,e 1.42b,c,e 20.52c,e 20.40e 20.35a,c,e 20.27b,c
e. Papio 0.22ALL 20.78b,c,d 20.64ALL 0.10ALL 0.68ALL 0.80ALL 0.86ALL 0.90ALL 20.14ALL 20.07ALL 20.21b,c,d 20.34a,c
1 Superscripts indicate signi?cant curvature differences among genera a?e at a\ 0.0025.
TABLE 4. Ranking of genera in descending order of
mean curvatures
Dorsopalmar Radioulnar RMS AbsoluteGaussian Mean
Metacacarpal
Gorilla Pongo Gorilla Gorilla Gorilla Pongo
Pan Pan Pan Pan Pan Pan
Pongo Gorilla Pongo Pongo Pongo Papio/Homo
Homo Homo Homo Homo Homo Gorilla
Papio Papio Papio Papio Papio
Trapezium
Gorilla Pongo Gorilla Gorilla Gorilla Gorilla
Pan Homo Pan Pan Pan Pan
Pongo Pan/Gorilla Pongo Pongo Pongo Papio
Homo Papio Homo Homo Homo Pongo
Papio Papio Papio Papio Homo
44 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
trapezium departs from the metacarpal in exhibiting
similar values among the great apes and humans (Fig.
8; Table 3). However, as in the metacarpal, each of these
genera has signi?cantly higher radioulnar curvature
values than Papio (Table 3).
Both surfaces are signi?cantly more curved in Pongo
than in Papio, but otherwise Pongo does not consistently
parallel either the African apes or humans. The metacar-
pal surface for the trapezium is similar to that of Pan in
RMS and to the Homo surface in dorsovolar curvature,
while radioulnar curvature falls with Pan and Gorilla,
signi?cantly above Homo. RMS and dorsovolar curvature
of the trapezial surface are intermediate between the
African apes and humans (and signi?cantly different
Fig. 7. Bivariate plot showing the relation between dorsovolar and radioulnar curvatures of the surface on the metacarpal
(modern humans, open squares; Pan, closed triangles; Gorilla, open diamonds; Pongo, closed circles; Papio, X?s; Neandertals, gray
N?s; Australopithecus, A?s; SK 84, 1; SKX 5020, *). Specimens closer to the origin exhibit less curvature than those further from the
origin along each axis.
Fig. 8. Bivariate plot showing the relation between dorsovolar and radioulnar curvatures of the surface on the trapezium
(modern humans, open squares; Pan, closed triangles; Gorilla, open diamonds; Pongo, closed circles; Papio, X?s; Neandertals, gray
N?s; Australopithecus, A; O.H. 7, O). Specimens closer to the origin exhibit less curvature than those further from the origin along
each axis.
45TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
except for Homo/Pongo dorsovolar), whereas radioulnar
curvature falls close to the Pan, Gorilla, and Homo
means.
Genus rankings of metacarpal and trapezium absolute
curvature values are the same as for RMS curvature
(Table 4). This is not surprising, because both measures
re?ect relative curvature magnitudes. Gaussian curva-
ture, which re?ects overall surface shape, shows again
the same ranking in both surfaces. However, there are
differences in rankings for mean curvature of both the
metacarpal and trapezium surfaces. For saddle surfaces,
mean curvature re?ects the extent to which the principal
curvatures differ from one another in magnitude: the
more similar the principal curvatures are to one another,
the lower the mean curvature value and vice versa. In
other words, a mean curvature of zero indictates that
the saddle surface shows the same magnitude of curva-
ture in both directions.
Fig. 9. Bivariate plot showing the relation between RMS and mean curvatures of the surface on the metacarpal (symbols as in Fig.
7). Greater values of RMS curvature indicate greater overall surface curvature; mean curvature values closer to zero indicate more
evenly curved surfaces while negative values indicate radioulnar curvature is greater than dorsovolar curvature (and vice versa).
Fig. 10. Bivariate plot showing the relation between RMS and mean curvatures of the surface on the trapezium (symbols as in Fig.
8). Greater values of RMS curvature indicate greater overall surface curvature; mean curvature values closer to zero indicate more
evenly curved surfaces while negative values indicate dorsovolar curvature is greater than radioulnar curvature (and vice versa).
46 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
The joint congruence indices (Table 5) indicate signi?-
cantly greater RMS and radioulnar congruence of the joint
surfaces in Homo than in the great apes and Papio. The
human joint in our sample is more congruent in the radio-
ulnar than in the dorsovolar direction, as Napier (1955)
and Ateshian et al. (1992) also found for humans. Pan sim-
ilarly is more congruent in the radioulnar direction, but is
striking in its very marked dorsovolar incongruence.
Comparison of fossil hominins with
the extant sample
Analyses of SPG and LS data for the early hominin
fossils have similar results (Table 6). There is a tendency
toward greater RMS curvature of the mutual tmc sur-
faces in A. afarensis compared with humans and much
lower RMS curvature of the O.H. 7 trapezial surface (see
Fig. 11. Bivariate plot showing the relation between absolute and mean curvatures of the surface on the metacarpal (symbols
as in Fig. 7). Greater values of absolute curvature indicate greater overall surface curvature; mean curvature values closer to zero
indicate more evenly curved surfaces, while negative values indicate radioulnar curvature is greater than dorsovolar curvature
(and vice versa).
Fig. 12. Bivariate plot showing the relation between absolute and mean curvatures of the surface on the trapezium (symbols as
in Fig. 8). Greater values of absolute curvature indicate greater overall surface curvature; mean curvature values closer to zero
indicate more evenly curved surfaces, whereas negative values indicate dorsovolar curvature is greater than radioulnar curvature
(and vice versa).
47TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
Table 3). In the SPG analysis, the A.L. 333-w39 meta-
carpal surface RMS curvature (1.59) falls close to the
chimpanzee mean (1.62), and the A.L. 333-80 trapezial
surface is coincident with the chimpanzee mean (1.85).
LS places the metacarpal (1.13) more than one SD unit
below the chimpanzee RMS mean (1.29) and more than
1 SD unit above the human mean (0.93). However, the
LS value for the other A. afarensis metacarpal, A.L.
333-58 (1.46), is more than four SD units higher than
the human mean (0.93) and close to the gorilla mean
(1.47). The LS dorsovolar curvatures of the A. afaren-
sis metacarpal and trapezial surfaces differ in the
same general directions as the RMS curvatures (Figs.
7 and 8). Radioulnar curvature follows similar trends,
but A.L. 333-58 shows higher curvature than the
means of the living hominids2, and a similar result is
observed for radioulnar curvature of the A.L. 333-80
trapezium.
RMS curvature of the SK 84 metacarpal surface (Table
6) is near the Pan mean, although the dorsovolar and
radioulnar curvatures result in this fossil clustering to-
ward one end of the Pan range of variation (see Fig. 9).
The dorsovolar value is below the means of humans and
all the great ape genera while radioulnar curvature is
higher than the means of all the living genera and quite
similar to that of A.L. 333-58. In contrast, all the SKX
5020 curvature values are close to those of Homo, and
the metacarpal falls well within the modern human clus-
ter (see Fig. 9). It should be noted that there is not
agreement regarding the af?nities of either the SK 84 or
the SKX 5020 specimens (see Susman, 1988a; Trinkaus
and Long, 1990; Tocheri et al., 2008).
In contrast to A. afarensis, the H. habilis O.H. 7 trape-
zial SPG mean RMS curvature is one SD below the SPG
human mean, and the LS mean is outside the means of
all genera, including Papio (see Fig. 10). Dorsovolar cur-
vature is also below all LS sample means, whereas
radioulnar curvature is below that of all LS genera
except Papio (see Fig. 8).
The Neandertal tmc surfaces are typically ?atter dor-
sovolarly than in most of the modern human sample (Ta-
ble 6; Figs. 7 and 8) with radioulnar curvature values
that fall within the modern human range. Several Nean-
dertals display RMS curvatures of the metacarpal sur-
face close to the modern human mean (see Fig. 9),
whereas the Neandertal trapezial RMS and dorsovolar
curvatures fall toward one end of the modern human
range (Figs. 8 and 10), recalling the condition seen in
O.H. 7. However, the high-radioulnar curvature distin-
guishes the Neandertals from H. habilis (see Fig. 8).
DISCUSSION
Curvature variability
No predictions were falsi?ed by the analysis. Human
tmc joint surfaces are less curved than in the great apes
in all directions. The only exception is trapezial radioul-
nar curvature, which is approximately the same in all
the hominids. Neandertal dorsovolar and O.H. 7 dorsovo-
lar and radioulnar curvatures are lower than in modern
humans, australopith curvatures recall those of African
apes, and the two Swartkrans metacarpals differ in
curvature as predicted.
Implications of curvature variability for joint
mobility, strength and stability
Joint mobility. Hamrick (1996) found an association
between high male joint mating surface curvature and
increased joint mobility in carpal joints of strepsirhine
primates. In contrast, our study shows higher trapezial
dorsovolar curvature and higher metacarpal radioulnar
curvature in great apes compared with humans, but
there is a lower range of motion in both directions in
great apes according to Rose (1992). As Hamrick (1996)
noted, conclusions from his analysis may not apply to
different joint types, movements, and taxa, and in our
study, this appears to be the case at least for hominids.
Humans share with the other genera a metacarpal male
surface for radioulnar deviation associated with a less-
curved female trapezial surface. This pattern may sug-
gest a potential for considerable mobility in this plane,
as suggested by Hamrick (1996) for strepsirhine carpal
movements. However, variation among the genera in
joint congruence must be taken into account in assessing
full potential ranges of motion.
Joint strength. Human RMS and dorsovolar curvatures
of the trapezium and metacarpal are lower than those of
the great apes, providing more surface normal to axial
loads. In addition, humans have greater RMS and radio-
ulnar congruence of the mutual tmc joint surface curva-
tures than the great apes. This pattern is favorable to
accommodation of the large axial loads associated with
forceful pinch and grasp of objects.
Cartilage thickness, joint shape, and subchondral bone
may be sensitive to loading history (Carter and Beaupre?,
2001). It would be interesting to compare humans with
nonhuman primates in tmc subarticular trabecular bone
properties, because these have been found by Rafferty
and Ruff (1994) in the humeral and femoral heads and
by Patel and Carlson (2007) at the distal radius to corre-
spond to differences in magnitude of loads.
Joint stability. The greater dorsovolar curvature of the
African ape metacarpal surface creates a long volar
beak. Attempts to slide the metacarpal dorsally on the
trapezium in cadaver and skeletal specimens are
resisted by abutment of the beak against the convex
volar trapezial surface. Lower curvature in humans is
associated with a less-projecting beak and less stability
of the joint. The shallower human joint has the advant-
age of accommodating relatively more axial load, but at
the potential expense of resistance to dorsal subluxation
of the metacarpal at tmc joints with lax ligaments, when
objects are pinched by the thumb and index ?nger (Pel-
legrini et al., 1993; Pellegrini, 2001).
Noteworthy is the exceptionally high-dorsovolar curva-
ture of the trapezium and metacarpal in the gorillas,
re?ected also in their high RMS and Gaussian curva-
tures. In addition, the joint is the most dorsovolarly con-
TABLE 5. Congruence indices1
Genus Dorsovolar Radioulnar RMS
a. Homo 0.44b 0.39ALL 0.29ALL
b. Pan 0.81ALL 0.74a,e 0.55a,c,e
c. Gorilla 0.43b 0.70a,e 0.41a,b
d. Pongo 0.53b 0.71a 0.44a
e. Papio 0.57b 0.55a,b,c 0.40a,b
1 Superscripts indicate signi?cant differences among genera a?e
at a\ 0.0025.
2See footnote 1.
48 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
gruent of the great ape genera. This pattern, favoring
tmc joint stability, should be investigated in connection
with the forceful pulling and processing of vegetation by
gorillas observed in videotapes taken by R. Byrne [cited
by Marzke (2006)].
The baboons are clearly distinguished from the homi-
nid sample by the low curvature of the mutual joint sur-
faces in the dorsovolar direction, and even lower values
in the radioulnar direction. This latter feature often
gives the joint an almost cylindrical appearance, particu-
larly on the trapezial surface. Low RMS and absolute
curvature values all indicate relatively ?at surfaces in
the baboons, but negative Gaussian curvature values
lend support to the conclusions of Napier (1961), Lewis
(1977), and Rose (1992) that baboons have saddle sur-
faces, albeit shallow ones. Furthermore, our low baboon
radioulnar curvature values con?rm the observations by
Rose (1992) that cercopithecines differ most clearly from
hominoids in radioulnar curvature and associated abduc-
tion range. He suggests that the lower range in the cer-
copithecines might limit grasp of relatively large objects
and affect opposition of the thumb to the ?ngers,
because abduction is an important component of thumb
opposition. Also, because opposition of the thumb meta-
carpal appears to occur without obvious constraints on
displacement of the metacarpal from the trapezium, this
may explain the observations of Guthrie (1991) and Jude
(1993) that manipulation of objects by Hamadryas
baboons does not involve the kinds of strong pinch and
grasp forces that would tend to displace the metacarpal.
Phylogenetic implications of the comparative
joint curvature evidence
The results of this study present compelling evidence
that the overall pattern of metacarpal and trapezial sur-
face curvature is derived in humans relative to their last
common ancestor with Pan. Although humans approach
baboons in magnitude of curvature more closely than do
the great apes, the topographic pattern is different. The
baboon surfaces are almost ?at in the radioulnar direc-
tion, and only slightly more curved in the dorsovolar
direction, forming a pattern that approaches a cylinder
in shape (Figs. 7 and 8). The human saddle-shaped pat-
tern more strongly recalls the great ape saddle, differing
primarily in the magnitude of curvature rather than in
overall shape. The cylinderlike morphology in baboons,
with metacarpal movement occurring primarily in ?ex-
ion/extension around the radioulnar axis, may underlie
differences in joint biomechanics and functional capabil-
ities that should be investigated.
Evolution of modern human tmc joint functions
and manipulative behavioral capabilities
The australopiths recall African apes most strongly in
RMS and dorsovolar curvatures, indicating a joint that
was stable against dorsal displacement in thumb/index
?nger pinch grips and capable of accommodating
stresses from several directions. It is interesting to note
that while these australopiths also share curved pha-
langes with the great apes, suggestive of arboreal loco-
motor capabilities (Stern, 2000), and proportionately
short ?ngers relative to thumb length (Alba et al., 2003)
recalling humans, they differ from both in having more
radioulnar curvature at the tmc joint (Figs. 7 and 8).
This combination of features suggests locomotor and
manipulative behaviors that may have differed to some
extent from those of both living great apes and humans.
The earliest evidence for reduction in tmc curvature is
in one of two Swartkrans thumb metacarpals (SKX
5020), and in the O.H. 7 trapezium, both from 
1.75
mya (Vrba, 1982; Walter et al., 1991; Blumenschine et
al., 2003). The SKX 5020 curvature falls within the mod-
ern human range of variation, whereas the O.H. 7 trape-
zium has a signi?cantly ?atter joint surface overall than
modern humans. This morphology suggests that the tra-
pezium could have accommodated large axial loads, as
predicted by Trinkaus (1989). Neandertals are similar to
the O.H. 7 hominin in dorsovolar ?atness of the joint;
however, they display radioulnar curvature values
within the modern human range. As Niewoehner (2000,
2001) noted, the primary shape change at the Neander-
tal tmc joint is the low degree of metacarpal volar beak
development, from which he infers an adaptation to
large axial reaction forces. It is likely that the joint was
TABLE 6. Normalized trapeziometacarpal joint surface curvature values in hominin fossils
Specimen Approach Joint surface RMS Dorsovolar Radioulnar Absolute Gaussian Mean
A.L. 333w-39 SPG Metacarpal 1.59
0.52 21.0 1.52 20.52 20.24LS 1.13
A.L. 333-58 LS Metacarpal 1.46 0.60 21.34 1.94 20.8 20.37
A.L. 333-80 SPG Trapezium 1.85
LS 1.05 20.92 0.51 1.43 20.47 20.2
O.H.7 SPG Trapezium 1.09
20.47 0.33 0.80 20.16 20.07LS 0.57
SKX 5020 LS Metacarpal 0.94 0.44 20.83 1.28 20.37 20.19
SK 84 LS Metacarpal 1.32 0.32 21.28 1.6 20.41 20.48
Kebara 2 LS Metacarpal 1.08 20.01 21.08 1.09 0.01 20.55
Kebara 2 LS Trapezium 0.63 20.43 0.46 0.89 20.2 0.02
La Ferrassie 1 LS Metacarpal 0.94 0.2 20.92 1.12 20.19 20.36
La Ferrassie 1 LS Trapezium 0.75 20.54 0.53 1.07 20.28 0.0
La Ferrassie 2 LS Metacarpal 1.11 0.3 21.07 1.37 20.32 20.38
La Ferrassie 2 LS Trapezium 0.89 20.61 0.65 1.26 20.39 0.02
Regourdou 1 LS Metacarpal 0.87 20.07 20.87 0.94 0.06 20.47
Regourdou 1 LS Trapezium 0.74 20.52 0.53 1.04 20.27 0.0
Amud 1 LS Metacarpal 0.93 0.18 20.91 1.09 20.16 20.36
La Chapelle 1 LS Metacarpal 1.18 20.38 21.12 1.49 0.42 20.75
Shanidar 3 LS Trapezium 0.73 20.49 0.54 1.02 20.26 0.03
Shanidar 4 LS Trapezium 0.59 20.32 0.49 0.82 20.16 0.09
49TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology
relatively unstable in both Neandertals and O.H. 7 in
comparison with modern humans, in the absence of a
projecting metacarpal beak to check dorsal subluxation
with pinch by the thumb and index ?nger. Stability
would have depended upon the tmc ligaments.
Modern human tmc joint topography appears to be a
morphological compromise: it is ?at enough to allow
accommodation of large axial forces associated with
forceful manipulative gripping, but curved enough to
resist subluxation in strong pinch (if the ligamentous ap-
paratus is intact). Biomechanical experiments and exam-
ination of diseased human tmc joints have shown that
initial cartilage destruction at the joint occurs near the
attachment of the beak ligament and is accompanied by
degeneration of the ligament (Pellegrini, 2005). The
more curved African ape-like ancestral hominin joint
would have been vulnerable to stress as prehistoric tool-
making and tool-using behaviors focused increasingly
large axial and shear stresses on the joint with forceful
precision and power squeeze grips. Thus, the evolution-
ary reduction of metacarpal surface curvature and short-
ening of the volar metacarpal beak may have been ad-
vantageous as hominins became more dependent on
tool-related behaviors for survival.
The loss of ancestral morphology in favor of derived
morphology that likely has performance advantages for
object manipulation may re?ect an evolutionary commit-
ment to behaviors involving tool making and tool use
that occurred in the hominin lineage leading to modern
humans and Neandertals (Tocheri, 2007; Tocheri et al.,
2007, 2008). The clinical prevalence today of osteoarthri-
tis at joint contact points stressed by forceful manipula-
tive grips and of metacarpal subluxation associated with
habitual forceful thumb/index ?nger pinching of objects,
noted earlier in the section on joint stability, is sugges-
tive of the kinds of functional liabilities that may have
channeled the evolution of hand joint structure during
hominin cultural and morphological evolution.
ACKNOWLEDGMENTS
We thank L. Berglund, Mayo Clinic Orthopedic Biome-
chanics Laboratory, for his help in the development of
our system for photographing the specimens. Equipment
and software for laser scanning and quadric surface
analysis were provided by the Partnership for Research
in Spatial Analysis (PRISM) at ASU. Special thanks to
A. Razdan and G. Farin (ASU) for encouraging the
development of this research. Access to specimens was
kindly provided by D. Hunt, R. Potts, L. Gordon, and R.
Thorington (NMNH), B. Latimer and L. Jellema (CNH),
E. Trinkaus (WUSTL), and W. Kimbel and D. Johanson
(IHO-ASU). We are also grateful to the Primate Founda-
tion of Arizona for access to their chimpanzee skeletal
collection. C. Linscheid, K. Stotesbury, and A. de Sousa
provided invaluable assistance in data collection and
digitization.
LITERATURE CITED
Alba DM, Moya-Sola S, Kohler M. 2003. Morphological af?nities
of the Australopithecus hand on the basis of manual propor-
tions and relative thumb length. J Hum Evol 44:225?254.
Ateshian GA, Ark JW, Rosenwasser MP, Pawluk RJ, Soslowsky
LJ, Mow VC. 1995. Contact areas in the thumb carpometacar-
pal joint. J Orthop Res 13:450?458.
Ateshian GA, Kwak S, Soslowsky LJ, Mow VC. 1994. A stereo-
photogrammetric method for determining in situ contact
areas in diarthroidial joints, and a comparison with other
methods. J Biomech 27:111?124.
Ateshian GA, Rosenwasser MP, Mow VC. 1992. Curvature char-
acteristics and congruence of the thumb carpometacarpal
joint: differences between female and male joints. J Biomech
25:591?607.
Bettinger PC, Berger RA. 2001. Functional ligamentous anat-
omy of the trapezium and trapeziometacarpal joint (gross and
arthroscopic). Hand Clin 17:151?168.
Bettinger PC, Linscheid RL, Berger RA, Cooney WP III, An
K-N. 1999. An anatomic study of the stabilizing ligaments of
the trapezium and trapeziometacarpal joint. J Hand Surg A
24:786?798.
Bettinger PC, Smutz P, Linscheid RL, Cooney WP III, An K-N.
2000. Material properties of the trapezial and trapeziometa-
carpal ligaments. J Hand Surg A 25:1085?1095.
Blumenschine RJ, Peters CR, Masao FT, Clarke RJ, Deino AL,
Hay RL, Swisher CC, Stanistreet IG, Ashley GM, McHenry
LJ, Sikes NE, van der Merwe NJ, Tactikos JC, Cushing AE,
Deocampo DM, Njau JK, Ebert JI. 2003. Late Pliocene Homo
and hominid land use from Western Olduvai Gorge. Tanzania.
Science 299:1217?1221.
Brand PW, Hollister AM. 1999. Clinical mechanics of the hand,
3rd ed. Mosby: St. Louis.
Bruce P, Simon J, Oswald T. 1997. Resampling stats users
guide. Arlington: Resampling Stats.
Carter DR, Beaupre? GS. 2001. Skeletal function and form.
Cambridge: Cambridge University Press.
Cerveri P, De Momi E, Marchente M, Lopomo N, Baud-Bovy G,
Barros RML, Ferrigno G. 2008. In vivo validation of a realis-
tic kinematic model for the trapezio-metacarpal joint using an
optoelectronic system. Ann Biomed Eng 36:1268?1280.
Christie PW, Ridley JN. 1990. Classi?cation and mathematical
representation of synovial articular surfaces. In: Sperber GH,
editor. From Apes to angels: essays in Anthropology in honor
of Philip V. Tobias. New York: Wiley-Liss. p 111?117.
Colman M, Mass DP, Draganich LF. 2007. Effects of the deep
anterior oblique and dorsoradial ligaments on trapeziometa-
carpal joint stability. J Hand Surg A 32:310?317.
Cooney WP III, Chao EYS. 1977. Biomechanical analysis of
static forces in the thumb during hand function. J Bone Joint
Surg Am 59:27?36.
Cooney WP III, Lucca MJ, Chao EYS, Linscheid RL. 1981. The
kinesiology of the thumb trapeziometacarpal joint. J Bone
Joint Surg A 63:1371?1381.
Day MH. 1978. Functional interpretations of the morphology of
postcranial remains of early African hominids. In: Jolly CJ,
editor. Early hominids of Africa. London: Gerald Duckworth
& Company Ltd. p 311?345.
Eaton RG, Dray GJ. 1982. Dislocations and ligament injuries in
the digits. In: Green DP, editor. Operative hand surgery,
Vol. 1. New York: Churchill. p 637?668.
Edgington E. 1995. Randomization tests. New York: Marcel
Dekker.
Efron B, Tibshiriani RJ. 1993. An introduction to the bootstrap.
Boca Raton, FL: CRC Press.
Guthrie EA. 1991. Variability of the primate trapeziometacarpal
articulation: description and functional evolution signi?cance,
M.A. thesis, Arizona State University, Arizona.
Hamrick MW. 1996. Articular size and curvature as determi-
nants of carpal joint mobility and stability in strepsirhine
primates. J Morph 230:113?127.
Hollister A, Buford W, Myers L. 1992. Axes of rotation of the
thumb carpometacarpal joint. J Orthop Res 10:454?460.
Imaeda T, Niebur G, Cooney WP III, Linscheid RL, An K-N.
1994. Kinematics of the normal trapeziometacarpal joint. J
Orthop Res 12:197?204.
Jude J. 1993. Manipulative behavior of hamadryas baboons,
Senior thesis, Arizona State University, Arizona.
Koff MF, Shrivastava N, Gardner TR, Rosenwasser MP,
Mow VC, Strauch RJ. 2006. An in vitro analysis of ligament
reconstruction or extension osteotomy on trapeziometacarpal
joint stability and contact area. J Hand Surg A 31:429?
439.
50 M.W. MARZKE ET AL.
American Journal of Physical Anthropology
Leakey LSB. 1960. Recent discoveries at Olduvai Gorge. Nature
188:1050?1052.
Leakey LSB, Tobias PV, Napier JR. 1964. A new species of the
genus Homo from Olduvai Gorge. Nature 202:7?9.
Leversedge FJ. 2008. Anatomy and pathomechanics of the
thumb. Hand Clin 24:219?229.
Lewis OJ. 1977. Joint remodeling and the evolution of the
human hand. J Anat 123:157?201.
Lewis OJ. 1989. Functional morphology of the evolving hand
and foot. Oxford: Clarendon Press.
Manly BFJ. 1997. Randomization, bootstrap and Monte Carlo
methods in biology, 2nd edition. Boca Raton, FL: CRC Press.
Martin RB, Burr, DB, Sharkey NA. 1998. Skeletal tissue
mechanics. New York: Springer-Verlag.
Marzke MW. 2006. Who made stone tools? In: Roux V, Bril B,
editors. Stone knapping: the necessary conditions for a
uniquely hominin behaviour. University of Cambridge, Cam-
bridge: McDonald Institute for Archaeological Research.
p 243?255.
Marzke MW, Wullstein KL. 1996. Chimpanzee and human
grips: a new classi?cation with a focus on evolutionary mor-
phology. Int J Primatol 17:117?139.
Momose T, Nakatsuchi Y, Saitoh S. 1999. Contact area of the
trapeziometacarpal joint. J Hand Surg A 24:491?495.
Mooney CZ, Duval RE. 1993. Bootstrapping. Newbury Park:
Sage.
Napier JR. 1955. The form and function of the carpo-metacarpal
joint of the thumb. J Anat 89:362?369.
Napier JR. 1961. Prehensility and opposability in the hands of
primates. Symp Zool Soc Lond 5:115?132.
Napier JR. 1962. Fossil hand bones from Olduvai Gorge. Nature
196:409?411.
Niewoehner WA. 2000. The functional anatomy of Late Pleisto-
cene and recent human carpometacarpal and metacarpopha-
langeal articulations, Ph.D. dissertation, University of New
Mexico, New Mexico.
Niewoehner WA. 2001. Behavioral inferences from the Skhul/
Qafzeh early modern human hand remains. PNAS 98:2979?
2984.
Patel BA, Carlson KJ. 2007. Bone density spatial patterns in
the distal radius re?ect habitual hand postures adopted by
quadrupedal primates. J Hum Evol 52:130?141.
Pellegrini VD Jr. 2001. Pathomechanics of the thumb trapezio-
metacarpal joint. Hand Clin 17:175?184, vii-viii.
Pellegrini VD Jr. 2005. The ABJS. 2005. Nicolas Andry Award:
osteoarthritis and injury at the base of the human thumb:
survival of the ?ttest? Clin Orthop Relat Res 438:266?276.
Pellegrini VD Jr, Olcott CW, Hollenberg G. 1993. Contact pat-
terns in the trapeziometacarpal joint: the role of the palmar
beak ligament. J Hand Surg A 18:238?244.
Pieron AP. 1973. The mechanism of the ?rst carpometacarpal
(CMC) joint. An anatomical and mechanical analysis. Acta
Orthop Scand Suppl 148:1?104.
Rafferty KL, Ruff CB. 1994. Articular structure and function in
Holobates, Colobus, and Papio. Am J Phys Anthropol 94:395?
408.
Rose ME. 1992. Kinematics of the trapezium-1st metacarpal
joint in extant anthropoids and Miocene hominoids. J Hum
Evol 22:255?266.
Sarmiento EE. 1988. Anatomy of the hominoid wrist joint: its
evolutionary and functional implications. Int J Primatol
9:281?345.
Simon JL. 1974?1997. Resampling stats: the new statistics, Ver-
sion 4.0.7. Resampling Stats Inc.
Steinberg B. 1999. The use of B-spline sufaces for modeling and
analysis of bone shapes, M.S. thesis, Arizona State University,
Arizona.
Stern JT. 2000. Climbing to the top: a personal memoir of
Australopithecus afarensis. Evol Anthropol 9:113?133.
Susman RL. 1998a. Hand of Paranthropus robustus from Mem-
ber 1. Swartkrans: fossil evidence for tool behavior. Science
240:781?784.
Susman RL. 1988b. New postcranial remains from Swartkrans
and their bearing on the functional morphology and behavior
of Paranthropus robustus. In: Grine FE, editor. Evolutionary
history of the ??Robust?? Australopithecines. New York: Aldine
de Gruyter. p 149?172.
Susman RL, Creel N. 1979. Functional and morphological af?n-
ities of the subadult hand (OH 7) from Olduvai Gorge. Am J
Phys Anthropol 51:311?331.
Tocheri MW. 2007. Three-dimensional riddles of the radial
wrist: derived carpal and carpometacarpal joint morphology
in the genus Homo and the implications for understanding
stone tool-related behaviors in hominins, Ph.D. dissertation,
Arizona State University, Arizona.
Tocheri MW, Femiani JC. In press. The ??bare bones?? of 3D: an
introduction to 3D modeling concepts for the physical anthro-
pologist. In: Hoppa RD, Nelson AJ, editors. 3D imaging in
biological anthropology and bioarchaeology. Cambridge: Cam-
bridge University Press.
Tocheri MW, Femiani J, Orr CM, Marzke MW. 2006. Quadric-
based metrics for shape analysis of three-dimensional osteo-
logical surfaces. Am J Phys Anthropol 129:177?178.
Tocheri MW, Orr CM, Jacofsky MC, Marzke MW. 2008. The evo-
lutionary history of the hominin hand since the last common
ancestor of Pan and Homo. J Anat 212:544?562.
Tocheri MW, Orr CM, Larson SG, Sutikna T, Jatmiko, Saptomo
EW, Awe Due R, Djubiantono T, Morwood MJ, Jungers WL.
2007. The primitive wrist of Homo ?oresiensis and its implica-
tions for hominin evolution. Science 317:1743?1745.
Trinkaus E. 1989. Olduvai hominid 7 trapezial metacarpal 1
articular morphology: contrasts with recent humans. Am J
Phys Anthropol 80:411?416.
Trinkaus E, Long JC. 1990. Species attribution of the Swartk-
rans Member 1 ?rst metacarpals: SK 84 and SKX 5020. Am J
Phys Anthropol 83:419?424.
Vrba ES. 1982. Biostratigraphy and chronology, based particu-
larly on Bovidae, of southern hominid-asociated assemblages:
Makapansgat, Sterkfontein, Taung, Kromdraai, Swartkrans;
and also Elandsfontein (Saldanha), Broken Hill (now Kabwe)
and Cave of Hearths. In: DeLumley F, DeLumley MA, editors.
L?Homo erectus et la place de l?homme de Tautavel parmi les
hominide?s fossils. Pre?tirage du 1er congre`s international pale?-
ontologie humaine, Vol. 2. Nice: Louis-Jean. p 707?752.
Walter RC, Manega PC, Hay RL, Drake RE, Curtis GH. 1991.
Laser-fusion 40Ar/39Ar dating of Bed I. Olduvai Gorge, Tanza-
nia. Nature 354:145?149.
Wood B, Lonergan N. 2008. The hominin fossil record: taxa,
grades and clades. J Anat 212:354?376.
Xu L, Stauch RJ, Ateshian GA, Pawluk RJ, Mow VC, Rose-
nwasser MP. 1998. Topography of the osteoarthritic thumb
carpometacarpal joint and its variations with regard to gen-
der, age, site, and osteoarthritic stage. J Hand Surg A
23:454?464.
51TMC JOINT SURFACE 3D CURVATURES
American Journal of Physical Anthropology