orr
Med
isto
ona
Wo
esia
niver
Article history: Several bones of the upper extremity were recovered during excavations of Late Pleistocene deposits at
Liang Bua, Flores, and these have been attributed to Homo ?oresiensis. At present, these upper limb
that upper limb elements might still exist in the walls of the
excavation. That speculation was born out in the 2004 ?eld season
during which a right humerus and incomplete right and left ulnae
were discovered in the adjacent Sector XI, Spit 58A ? the same level
as the LB1 remains recovered during the previous ?eld season
(Morwood et al., 2005). Their proximity to the other LB1 material
indicated that they were part of the same partial skeleton.
material from Flores belonged to a new species, Homo ?oresiensis.
The mix of features they observed was also found to characterize
the upper limb material described by Morwood et al. (2005). In
particular, they noted greater shaft robusticity in the LB1 humerus
and ulnae than is found in modern humans, and a markedly low
degree of humeral torsion. In addition, the skeleton exhibited
a humerofemoral index of only 85.4, putting it in the range of
australopithecines and quite unlike modern humans.
Considering the very small endocranial volume, short stature,
Contents lists availab
m
.e l
Journal of Human Evolution 57 (2009) 555?570* Corresponding author.Introduction
The announcement by Brown et al. (2004) of a new small-
bodied Late Pleistocene hominin from Flores, Indonesia, was based
on a partial skeleton (LB1) composed of a cranium, mandible, and
several lower limb elements. Included in the holotype, but not
described, were incomplete hands and feet and other fragmentary
material. The LB1 partial skeleton was found at a depth of 5.9 m in
Sector VII of the 2003 excavations and most elements came from
a small area of approximately 500 cm2 with some parts in articu-
lation (Brown et al., 2004; Morwood et al., 2004). Brown et al.
(2004) reported that the position of the partial skeleton suggested
In addition to the upper limb remains of the LB1 partial skeleton,
the 2003 and 2004 excavations resulted in the recovery of additional
upper limb elements from other individuals (Table 1). These include
a proximal ulnar fragment (LB2/1), a left radius (LB3), a radial shaft of
a child (LB4/1), a metacarpal fragment (LB5/2), and a scapula, radius,
ulna, metacarpal, and several phalanges of LB6 (Brown et al., 2004;
Morwood et al., 2005). The excavated sector and spit number (i.e.,
depth level) where each of these elements was recovered are shown
in Table 1 along with their estimated age range.
On the basis of a unique mosaic of primitive and derived traits of
the cranium, mandible, and lower limb elements, Brown et al.
(2004) concluded that the partial skeleton and other homininReceived 18 January 2008
Accepted 5 June 2008
Keywords:
Homo ?oresiensis
scapula
clavicle
humerus
radius
ulna
carpals
phalangesE-mail address: susan.larson@stonybrook.edu (S.G
0047-2484/$ ? see front matter  2008 Elsevier Ltd.
doi:10.1016/j.jhevol.2008.06.007remains have been assigned to six different individuals ? LB1, LB2, LB3, LB4, LB5, and LB6. Several of these
bones are complete or nearly so, but some are quite fragmentary. All skeletal remains recovered from
Liang Bua were extremely fragile, but have now been stabilized and hardened in the laboratory in Jakarta.
They are now curated in museum-quality containers at the National Research and Development Centre
for Archaeology in Jakarta, Indonesia. These skeletal remains are described and illustrated photo-
graphically. The upper limb presents a unique mosaic of derived (human-like) and primitive morphol-
ogies, the combination of which is never found in either healthy or pathological modern humans.
 2008 Elsevier Ltd. All rights reserved.a r t i c l e i n f o a b s t r a c tDescriptions of the upper limb skeleton
S.G. Larson a,*, W.L. Jungers a, M.W. Tocheri b, C.M. O
T. Sutikna e, Rokhus Due Awe e, T. Djubiantono g
aDepartment of Anatomical Sciences, Stony Brook University Medical Center, School of
bHuman Origins Program, Department of Anthropology, National Museum of Natural H
c School of Human Evolution and Social Change & The Institute of Human Origins, Ariz
dGeoQuEST Research Centre, School of Earth and Environmental Sciences, University of
e Indonesian Centre for Archaeology, Jl. Raya Condet Pejaten No. 4, Jakarta 12001, Indon
fArchaeology and Palaeoanthropology, School of Human and Environmental Studies, U
g The National Research and Development Centre for Archaeology, Jakarta, Indonesia
Journal of Hu
journal homepage: www. Larson).
All rights reserved.f Homo ?oresiensis
c, M.J. Morwood d,f,
icine, Stony Brook, NY 11794-8081, USA
ry, Smithsonian Institution, Washington, DC 20013-7012, USA
State University, Tempe, AZ 85287-2402, USA
llongong, Wollongong, NSW 2522, Australia
sity of New England, Armidale, New South Wales 2351, Australia
le at ScienceDirect
an Evolution
sevier .com/locate/ jhevoland relatively recent date of LB1, both Brown et al. (2004) and
al
e
e
al
al
al
e
e
e
manTable 1
Summary of Homo ?oresiensis upper limb remains excavated from Liang Bua
Sector Spit ID Element
IV 42D LB2/1 ulna, right
IV 58R LB3 radius, left
XI 42 LB4/1 radius, left
XI 46 LB5/2 metacarpal
VII 50 LB12 manual phalanx, distal
XI 51 LB6/2 radius, right
XI 51 LB6/3 ulna, left
XI 51 LB6/4 scapula, right
XI 51 LB6/5 metacarpal shaft
XI 51 LB6/7 manual phalanx, distal
XI 52 LB6/8 manual phalanx, proxim
XI 52 LB6/9 manual phalanx, middl
XI 52 LB6/10 manual phalanx, middl
XI 52 LB6/11 manual phalanx, distal
XI 52 LB6/12 manual phalanx, distal
XI 52 LB6/16 manual phalanx, proxim
XI 56B LB1/61 manual phalanx, proxim
XI 57A LB1/62 manual phalanx, proxim
XI 58A LB1/50 humerus, right
XI 58A LB1/51 ulna, left
XI 58A LB1/52 ulna, right
VII 59 LB1/5 clavicle, right
VII 59 LB1/40 manual phalanx, middl
VII 59 LB1/42 manual phalanx, middl
VII 59 LB1/44 scaphoid, left
VII 59 LB1/45 capitate, left
VII 59 LB1/46 hamate, left
VII 59 LB1/47 trapezoid, left
VII 59 LB1/48 manual phalanx, middl
VII 59 LB1/49 manual phalanx, distal
VII 59 LB1/59 metacarpal fragment
VII 59 LB1/60 lunate, left
S.G. Larson et al. / Journal of Hu556Morwood et al. (2005) considered whether or not modern human
developmental disruptions, such as insulin-like growth factor
(IGF-1), related growth retardation, pituitary dwar?sm, or
primordial microcephalic dwar?sm, could explain its unusual
morphology. However, they concluded that none of these condi-
tions matched the unique cranio-facial features, limb proportions,
and other skeletal characteristics of the Flores hominin material.
Nonetheless, alternative interpretations of the LB1 partial skeleton
have advocated various modern human developmental syndromes
to account for its distinctive morphology. These include claims that
the LB1 skeleton manifests characteristics of IGF-1 de?ciency or
insensitivity (Richards, 2006; Hershkovitz et al., 2007), endemic
cretinism (congenital hypothyroidism) (Obendorf et al., 2008), and
microcephaly with accompanying developmental abnormalities
(Jacob et al., 2006). Although a detailed rebuttal to each of these
assertions is beyond the scope of this paper, some claims have been
based on inaccurate depictions of H. ?oresiensis morphology,
perhaps due in part to the necessary brevity of the original
descriptions of the material. The following detailed descriptions of
the upper limb elements from Flores provide a more complete
characterization of this material. For more detailed and compre-
hensive comparative analyses, the reader is referred to Argue et al.
(2006), Larson et al. (2007), and Tocheri et al. (2007).
LB1 upper limb elements
Clavicle (LB1/5)
The clavicle is represented by an incomplete right element that
is missing its medial end, and was still coated with some matrix at
the time of description (Fig. 1). The fragment measures 85.9 mm
(maximum length), and is reconstructed to have been 91 mm
VII 62 LB1/55 manual phalanx, distal 1st
a ka? thousand years before present.
b For speci?c details regarding the dating samples used for the age estimates see RobAge (ka)a Age Rationaleb
w74?14/12 wLB-JR-8a
w74?14/12 wLB-JR-8a, <LBS4-32
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, >ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610,<ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
>15.7?17.1, <17.1?18.7 >ANUA-23610, <ANUA-27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
Evolution 57 (2009) 555?570(3.5 mm based on the prediction error value) in total length (see
Larson et al., 2007). The lateral end is somewhat eroded and the
articular facet for the acromion is not visible due to postmortem
damage. The lateral portion is ?attened superoinferiorly, while the
remainder of the shaft has a more rounded contour. What remains
of the medial end of the bone is broken in several places. Contrary
to Jacob et al. (2006), we can identify only one clavicle for LB1; it
seems likely that one of the ribs still in matrix was confused for
a second clavicle.
Superior view (Fig. 1) The shaft displays two prominent curves in
superior view: a lateral curve that is concave anteriorly, and
a medial curve that is convex anteriorly. The maximum ante-
roposterior width at the in?ection of the lateral curvature is
13.87 mm, and 11.28 mm at the in?ection of the medial curve. The
inner surface of the lateral curvature has a crescent-shaped smooth
facet for the attachment of anterior deltoid. This surface is bounded
by a slight ridge, and posterior to this ridge is a groove that
broadens laterally for the attachment of cranial trapezius. Medial to
the deltoid attachment facet, the bone has a more rounded contour
and is smooth for the attachment of pectoralis major.
Anterior view (Fig. 1) The inner margin of the lateral curve is
narrow and sharp and the lateral end of the bone is ?attened
superoinferiorly. A crescent-shaped facet for the anterior deltoid is
visible anteriorly. Passing from lateral to medial, the shaft becomes
wider and rounder, and the attachment area for pectoralis major
wraps around the bone from superior to anterior to inferior. The
increased width of the medial half of the bone creates an inferiorly
convex curvature, while the lateral end appears to display a slight
inferior concavity.
Inferior view (Fig. 1) A narrow depression for subclavius measures
31.6 mm long. The conoid tubercle is visible but not prominent, and
is positioned 22.06 mm from its midpoint to the lateral end of the
17.1?18.7, 17.9?19.0 ANUA-27116, 27117
erts et al. (this volume: Table 1).
manFigure 1. Composite photo of four views of the LB1/5 right clavicle. Note that becauseS.G. Larson et al. / Journal of Hubone. Some surface bone over the tubercle may have been lost
postmortem. A portion of the pectoral attachment area is visible on
the inferior surface, and a slightly raised area near the beginning of
the lateral curve may mark the most lateral extent of the pectoral
attachment area. A low ridge separates this area from the groove for
subclavius.
Posterior view (Fig. 1) The posterior surface of the lateral end of the
bone is ?attened, and the superior edge of this area rises to form
a sharp boundary with the groove for the attachment of cranial
trapezius. The conoid tubercle is opposite to this ?attened area on
the inferior surface of the bone. A slight ridge extends medially
from the medial end of the ?attened area for a few millimeters, but
soon merges with the more rounded surface of the medial half of
the shaft. Voisin (2006) has reported that modern humans are
distinct in displaying a single inferior curve of the clavicle in
posterior view. His preliminary examination of LB1/5 (based on
photographs only) indicates that it retains the primitive double
curvature seen African apes and all hominins except modern
humans (Voisin, pers. comm.)
Humerus (LB1/50)
A right humerus, LB1/50, is nearly complete except for missing
greater and lesser tubercles and adjacent parts of the articular
surface at the proximal end, and the lateral epicondyle and portions
of the capitulumat the distal end (Figs. 2?4, Table 2). The humerus is
243 mmin total length (Morwood et al., 2005),which,while short in
an absolute sense, can be matched to the lower extremes of small-
bodied African pygmies and Andaman Islanders (WLJ, pers. obs.).
Proximal end ? anterior view (Fig. 3) Both greater and lesser
tubercles are missing. The remaining portion of the
the clavicle has been rotated in place to display the four different views, the superior
aspect of the bone is facing down in the posterior view. Some matrix still covers the
specimen. The lateral end of the fragment is nearly complete, but part of the medial
end is missing. Each block on the scale bar represents one centimeter.Evolution 57 (2009) 555?570 557intertubercular sulcus is very wide and shallow with a proximal
width of 13.87 mm and a depth of 1.22 mm. The circumference of
the surgical neck just below the missing portion is 65.5 mm. The
lateral edge of the intertubercular sulcus is a ridge positioned
slightly lateral to the midline of the shaft, implying an anterior
position for the greater tubercle. Lateral to this ridge, the bone is
rounded and smooth until the beginning of the deltoid tuberosity.
A short, rough ridge that merges into a roughened elongated area
for attachment of latissimus dorsi and teres major descends from
what was the root of the lesser tubercle along the medial margin
of the anterior surface.
Proximal end ? medial view (Fig. 3) A medial view clearly reveals
that the missing region included some portion of the head as well
as the tubercles. The ridge descending from what was the root of
the lesser tubercle is positioned near the anterior margin of the
medial view of the proximal end of the humerus. Descending down
the shaft from this short ridge is the roughened area for the
attachment of latissimus dorsi and teres major.
Proximal end ? posterior view (Fig. 3) The humeral head preserves
its posterior aspect, including most of the lateral margin, although
Figure 2. Composite photo of two views of the LB1/50 right humerus. The specimen is
nearly complete except for missing greater and lesser tubercles, adjacent parts of the
humeral head, and the lateral region of the distal end. The shaft is fairly straight and
displays light muscle markings.
Figure 3. Composite photo of four views of the proximal end of the LB1/50 humerus. The medial and lateral views best show the missing regions of the tubercles and articular
surface. Each block on the scale bar represents one centimeter.
S.G. Larson et al. / Journal of Human Evolution 57 (2009) 555?570558we note some erosion in this area. The preserved portion of the
head displays a rounded contour. Immediately inferior to the lateral
margin of the head is an oblique sulcus or depression. The medial
margin of this sulcus is a small buttress for the head, which is
positioned slightly medial to the midline of the posterior surface.
The lateral edge of the sulcus is the lateral ?are that would have led
to the most posterior aspect of the greater tubercle. Passing infe-
riorly from this ?aring region are two roughened lines: one for
attachment of triceps brachii that begins laterally and becomes
more medial distally, and the second that heads more anteriorly
toward the deltoid tuberosity. Other than these ridges for muscle
attachments, most of the posterior aspect of the shaft is rounded
and smooth.
Proximal end ? lateral view (Fig. 3) The greater tubercle is missing
and the lateral margin of the head is somewhat eroded. A fracture
line is visible just above the humeral surgical neck. The lateral ?are
of the root of the greater tubercle that is visible in posterior view is
posteriorly positioned. Two roughened lines descend from this
?aring region, one heading distally and anteriorly toward the
deltoid tuberosity, and the other distally and posteriorly for the
attachment of triceps brachii.
Midshaft There is a postmortem fracture through the approximate
midpoint of the shaft. The midshaft appears rounded in cross-
section with an anteroposterior diameter of 17.44 mm and
a mediolateral diameter of 16.35 mm. The deltoid tuberosity ends
118.36 mm from the proximal end of the bone and is most visibleFigure 4. Composite photo of four views of the distal end of the LB1/50 humerus. The mis
shows that the appearance of a sharp lateral trochlear ridge on anterior view is actually due
the scale bar represents one centimeter.from a lateral perspective as a ridge 51.69 mm long. This ridge is
a continuation of a roughened line that begins at the root of the
greater tubercle, and forms the posterolateral margin of the deltoid
tuberosity. The anterior margin is formed by the ridge that extends
from the lateral lip of the intertubercular groove. The bone inferior
to the greater tubercle between these two boundaries is ?attened
and faces anterolaterally. A laterally positioned ridge leading to the
supracondylar ridge begins at about midshaft. On themedial side of
the shaft is a roughened area that is probably for the attachment of
coracobrachialis. There is no detectable spiral/radial groove on the
posterior aspect of the shaft. However, there is a slight ?attening of
the diaphysis above the midshaft fracture that might re?ect the
pathway of the radial nerve.
Distal end ? anterior view (Fig. 4) The distal end is largely complete
except for the lateral epicondyle and the lateral half of the capit-
ulum (see Table 2). Several postmortem fractures are present. There
is a vertical fracture just lateral to the trochlear groove through the
trochlea. This fracture also pierces the coronoid and olecranon
fossae. There is a horizontal fracture across the shaft approximately
30 mm from the distal end of the bone. Approximately 7?8 mm
inferior to this horizontal fracture is yet another fracture line,
located across the medial pillar of the coronoid and olecranon
fossae. A ridge develops on the anterior surface approximately two-
thirds of the way down the shaft, and then splits into two diverging
lines toward the medial and lateral margins of the distal articular
surfaces. The coronoid fossa reaches a depth of 4.65 mm. Thesing portions are best seen in the posterior and lateral views. The posterior view also
to the incompleteness of the lateral region of the distal end of the bone. Each block on
Medial epicondyle ML length (posterior view) 11.48 mm
manmedial trochlear ridge is prominent anteriorly and distally, but
there is some erosion on its medial margin. The lateral trochlear
ridge is not prominent anteriorly; there is just a slight swelling
lateral to the trochlear groove. The appearance of a sharp lateral
trochlear ridge distally in anterior view is actually due to missing
portions of the bone posterolaterally. There is a shallow groove
between the capitulum and the trochlea. The preserved portion of
the capitulum appears to be round.
Distal end ? medial view (Fig. 4) Slight posterior and anterior ridges
converge along the proximal margin of the medial epicondyle.
Some erosion has occurred on the inferior and medial surfaces of
the epicondyle reducing its apparent length. In medial view the
medial trochlear ridge is round. Between the inferior rim of the
medial trochlear ridge and the inferior margin of the medial epi-
condyle is a crescent-shaped groove.
Distal end ? posterior view (Fig. 4) The distal quarter of the shaft is
?attened and bounded by ridges on the medial and lateral sides.
The olecranon fossa reaches a depth of 5.76 mm. Missing parts
include the lateral pillar of the olecranon fossa, the lateral epi-
condyle, the posterior aspect of the capitulum, and some of the
lateral trochlear ridge. The posterior aspect of the medial epi-
condyle is eroded and this may reduce the apparent length of the
epicondyle.
Medial epicondyle ML length (distal view) 12.11 mm
Medial epicondyle angle (relative to ML axis of distal articular surface) 26Table 2
Measurements of the distal end of humerus LB1/50
Measurement LB1/50
Coronoid fossa ML diameter 12.76 mm
Coronoid fossa depth 4.70 mm
Trochlea ML width (from medial to lateral trochlear ridges) 15.75 mm
Zona conoidea ML width 4.28 mm
Zona conoidea PD length 13.15 mm
Medial trochlear ridge PD length 20.79 mm
Medial trochlear ridge AP length 20.61 mm
Lateral trochlear ridge PD height 15.76 mm
Trochlear groove AP length 12.70 mm
Trochlear groove PD length 11.92 mm
Olecranon fossa ML diameter 19.72 mm
Olecranon fossa depth 5.76 mm
Width of medial pillar of olecranon fossa (lateral pillar damaged) 9.02 mm
Medial epicondyle AP length (distal view) 10.80 mm
Medial epicondyle ML length (anterior view) 9.19 mm
S.G. Larson et al. / Journal of HuDistal end ? lateral view (Fig. 4) The supracondylar ridge is sharp
but has little ?are. Most of the lateral aspect of the capitulum is
missing, as well as the entire lateral epicondyle.
Humeral torsion in LB1
Morwood et al. (2005) estimated the torsion of the LB1/50
humerus to be 110 degrees, which is extremely low for a hominin.
However, the damage to the proximal end of the humerus makes it
dif?cult to use traditional methods to measure torsion since they
rely on being able to determine the orientation of the humeral
head. Therefore, Larson et al. (2007) remeasured torsion in LB1/50
by using two alternate landmarks for humeral head position. One
was a bisector of the intertubercular groove, which Larson (1996)
has shown can be used as an indicator of humeral head position in
other fossil humeri. The other was the position of a posterior
buttress for the humeral head, which is aligned with the humeral
head axis. The two methods yielded estimates of 119 and 121
degrees, respectively, for an average of 120 degrees. Though slightly
higher than the estimate offered by Morwood et al. (2005), this
second set of measurements veri?es that humeral torsion in LB1/50
was well below human averages. Larson et al. (2007) suggestedusing 115 degrees, which is the average of the values offered by the
two independent observers, as the best estimate of humeral torsion
in LB1/50.
Larson et al. (2007) noted that while there is variation in the
amount of humeral torsion among modern human populations,
LB1/50 falls below (or just within) the low end of that variation. In
addition, early hominins also display low humeral torsion, as does
the humerus of the early Homo erectus skeleton, KNM-WT 15000.
Shoulder con?guration in LB1
Based on the reconstructed length of 91 mm for the LB1/5
clavicle, Larson et al. (2007) estimated a claviculohumeral ratio for
LB1 of 37.5%, which is low compared to modern humans (mean
values ranging from 43.0% to 49.1% for various populations)
(Schultz, 1937). Noting that KNM-WT 15000 also displays a rela-
tively short clavicle with a claviculohumeral ratio of 40.9%, Larson
et al. (2007) examined claviculohumeral ratios across primate
species and in other fossil hominins, and their observations suggest
that a relatively short clavicle represents the primitive condition for
primates. Unfortunately, relative clavicular length is currently
unknown for early hominins, but it is possible that the short clav-
icles of KNM-WT 15000 and LB1 are primitive retentions in H.
erectus and H. ?oresiensis, respectively. The recent description of
comparatively short clavicles from Dmanisi (Lordkipanidze et al.,
2007) supports the view that a relatively short clavicle is charac-
teristic of earlyH. erectus and is probably the primitive condition for
hominins.
The Dmanisi hominins also resemble KNM-WT 15000 and LB1
in displaying a very low degree of humeral torsion (Lordkipanidze
et al., 2007). Larson et al. (2007) suggested that this combination of
a short clavicle and low humeral torsion are components of
a transitional stage in the evolution of the hominin pectoral girdle
and shoulder between the somewhat ape-like con?guration seen in
early hominins and the condition of modern humans. According to
this proposal, the descent of the shoulder from the high position of
early hominins was limited by a relatively short clavicle. As a result,
the scapula moved anteriorly as well as inferiorly, thus assuming
a more lateral position on the rib cage. This resulted in a more
anteriorly directed glenoid fossa. Therefore, a humeral head that
faced posteriorly (i.e., hadminimal torsion) wouldmaintain sagittal
plane elbow function, and the manipulatory ability of the hands
would be unimpaired. Larson et al. (2007) further speculated that
selection for increased range of motion at the shoulder led to
clavicular elongation in later stages of hominin evolution.
Ulnae (LB1/51 and LB1/52)
Portions of both the left (LB1/51) and right (LB1/52) ulnae are
preserved (Fig. 5). The left ulna is missing both the proximal and
distal ends. The right ulna is also missing the distal end, but
preserves a nearly complete proximal end.
Left ulna (LB1/51) (Fig. 5)
Both the proximal and distal ends of the bone are missing and
the remaining fragment is 167 mm long (Morwood et al., 2005)
(Fig. 5). The preserved portion of the proximal end consists of
several pieces joined together, and there are two postmortem
fractures of the shaft. One fracture is approximately 50 mm from
what is left of the proximal end of the bone, and the second fracture
is 109 mm from this same end. The shaft is gently curved ante-
roposteriorly, with the concave surface facing anteriorly. A portion
of a ridge extending from the coronoid process is preserved at the
Evolution 57 (2009) 555?570 559proximal end, and medial to this ridge is a marked depression for
the insertion of brachialis (ulnar tuberosity). Overall, the
man Evolution 57 (2009) 555?570S.G. Larson et al. / Journal of Hu560morphology of LB1/51 is very similar to that of the more complete
right ulna (LB1/52).
Right ulna (LB1/52) (Figs. 5 and 6)
The right ulna is nearly complete except for missing the distal
end of the shaft and head (Figs. 5 and 6, Table 3). The preserved
portion is 190 mm long and after comparing it to human African
Pygmies, Morwood et al. (2005) estimated that its total length
would have been approximately 205 mm. There is a shaft fracture
approximately 73 mm from the proximal end, and another fracture
approximately 17 mm from what remains of the distal end. The
cortex at the site of the more proximal shaft fracture is, on average,
3.75 mm thick. The mediolateral diameter of the medullary cavity
at this point is approximately 7 mm (the mediolateral shaft diam-
eter is 14.48 mm). The shaft displays a gentle anteroposterior curve
that is anteriorly concave. The shaft has a rounded contour, and the
interosseous crest is more of a roughened surface than an actual
crest. It does separate anterior and lateral surfaces, however. The
shaft is fairly broad proximally, and narrows only very gradually
toward the distal end. The preserved portion of the distal end
displays the beginning of an anterior ?are for the ulnar head.
Proximal end ? anterior view (Fig. 6) The trochlear notch is wide
and short. The proximal rim is broken and missing pieces both
laterally and medially. However, the central region is preserved
making it possible to measure the length of the notch (Table 3). The
articular surface of the trochlear notch is clearly divided intomedial
and lateral semilunar-shaped facets by a blunt ridge. The lateral
facet appears to be larger than the medial facet, and there is an
elongated pit (approximately 2 mm in length) in its surface. The
Figure 5. Composite photo of three views of the LB1/51 left ulna and LB1/52 right ulna.
The right ulna is the more complete of the two and includes most of the shaft and the
proximal end, except for parts of the olecranon.
Figure 6. Composite photo of four views of the proximal end of the LB1/52 right ulna.
The olecranon is missing most of its medial corner. Each block on the scale bar
represents one centimeter.
Table 3
Measurements of proximal end of ulnae LB1/52, LB6/3 and LB2/1
Measurement LB1/52 LB6/3 LB2/1
Radial notch AP length 15.00 mm 13.45 mm ?
Radial notch PD length 10.56 mm 8.65 mm ?
Radial notch angle (relative to an AP axis) 35 10 ?
Trochlear notch ML width at proximal rim
of notch
19.24 mm 17.03 mm 18.28 mm
Trochlear notch ML width at midpoint of notch 19.28 mm 15.45 mm 16.30 mm
Trochlear notch ML width at distal rim of notch 16.85 mm 12.21 mm 17.73 mm
Trochlear notch PD length 17.11 mm 16.05 mm 16.48 mm
Trochlear notch AP height (between
proximal rim and posterior surface of bone)
19.61 mm 15.53 mm 14.44 mm
Trochlear notch AP height (between base
of notch and posterior surface of bone)
12.80 mm 10.82 mm 11.80 mm
Trochlear notch AP height (height at
coronoid process)
24.67 mm 20.79 mm 21.60 mm
Trochlear notch depth 9.05 mm 7.49 mm 6.61 mm
Trochlear notch angle (line between proximal
and distal rims and long axis of bone)
13 8 20
Coronoid ML width including radial notch 17.32 mm 15.83 mm 17.00 mm
Coronoid ML width not including radial notch ? 12.52 mm 11.96 mm
Shaft ML width at distal margin of radial notch 19.37 mm 12.28 mm 16.49 mm
Shaft AP depth at distal margin of radial notch 20.59 mm 16.43 mm 15.89 mm
Olecranon process AP depth 16.51 mm 13.34 mm 9.93 mm
Olecranon process PD length (from proximal
edge of trochlear notch)
8.48 mm 2.96 mm 1.50 mm
Olecranon process PD length (from midpoint
of trochlear notch)
16.71 mm 13.33 mm 12.41 mm
Olecranon process ML width ? 17.43 mm 18.50 mm
PD length from middle of trochlea to midpoint
of ulnar tuberosity
29.02 mm 26.79 mm ?
AP diameter of shaft at level of distal border
of ulnar tuberosity
15.86 mm 13.43 mm ?
ML diameter of shaft at level of distal border
of ulnar tuberosity
16.11 mm 11.59 mm ?
mantwo facets meet at an angle of 113 degrees, with the medial side
appearing to be somewhat more oblique (relative to horizontal)
than the lateral. The medial and lateral margins of the trochlear
notch are almost parallel, i.e., the notch does not have a waisted
appearance. There is a crack across the width of the trochlear notch
at its base.
The coronoid process is blunt and is missing the medially
positioned tubercle. A prominent longitudinal ridge extends
distally from the coronoid process, and is ?anked on either side by
marked depressions. Dorsal to these depressions the bone widens
again so that the shaft distal to the trochlear notch is wide.
Proximal end ? medial view (Fig. 6) The medial corner of the
olecranon process is missing, along with part of the proximal rim
of the trochlear notch. The coronoid process projects further
anteriorly than the proximal trochlear rim (Table 3). The area for
the insertion of brachialis (ulnar tuberosity) forms a marked
depression, and is positioned between the anterior ridge, which
extends from the coronoid process, and a blunt ridge that begins
dorsal to the distal half of the trochlear notch. The latter continues
distally as the border between the anteromedial and poster-
omedial surfaces of the ulnar shaft. The olecranon process is fairly
long (Table 3).
Proximal end ? posterior view (Fig. 6) The posterior subcutaneous
surface of the proximal end of the ulna is ?at and broad. This
surface is bordered by two faint lines beginning on either side of the
olecranon, which converge distally, creating a central ridge that
forms the subcutaneous surface. The medial line marks the edge of
the ?exor carpi ulnaris attachment area and the lateral line marks
the edge of the attachment of anconeus.
Proximal end ? lateral view (Fig. 6) The radial notch occupies the
lateral surface of the coronoid process. The notch is somewhat
obliquely oriented at approximately 35 degrees to vertical. It is
fairly shallow and has an anteroposterior height of 15 mm and
a proximodistal width of 10.56 mm. Dorsal to the radial facet is
a marked ridge for the attachment of supinator, and dorsal to this
ridge is a ?attened area for the attachment of anconeus.
Carpals (LB1/44, LB1/45, LB1/46, LB1/47, and LB1/60)
The carpal remains include a scaphoid (LB1/44), a capitate (LB1/
45), a hamate (LB1/46), a trapezoid (LB1/47), and a lunate (LB1/60).
All ?ve are left carpals and articulate well with one another (Fig. 7).
Note that Jacob et al. (2006) incorrectly described the trapezoid and
capitate as being from the right side. They also list a right trapezium
that was either misidenti?ed or was not returned to the collections
in Jakarta.
A three-dimensional (3D) quantitative comparative analysis
of the scaphoid, trapezoid, and capitate of LB1 showed that
these carpals lack the suite of derived features that characterize
modern human and Neandertal carpals, and instead show a sym-
plesiomorphic pattern of features that also characterizes carpals
belonging to African apes, Homo habilis (OH 7), Australopithecus
africanus (TM 1526), and Australopithecus afarensis (A.L. 288-1, A.L.
333-40) (Tocheri et al., 2007). Here we present basic anatomical
descriptions of each carpal along with ?gures that include photo-
graphs of the original specimens, high quality casts, and images of
the 3D models acquired by laser scanning.
Scaphoid (LB1/44) (Fig. 8) The scaphoid shows slight damage to the
tip of the tubercle. As in the African ape-human clade, the os cen-
trale is completely incorporated as part of the scaphoid indicating
that the cartilaginous precursors of the scaphoid and centrale likely
coalesced in utero. It has distinct, well-preserved articular facets for
the capitate, lunate, and radius, as well as a single combined facet
for the trapezium and trapezoid. It measures approximately
S.G. Larson et al. / Journal of Hu19.7 mm from the dorsal (i.e., posterior) border of the radial artic-
ular surface to the most palmar (i.e., anterior) tip of the tubercle,and approximately 16.5 mm proximodistally. The capitate articular
surface is large, oval in outline, strongly concave, and takes up
much of the distoulnar surface of the bone. The lunate articular
surface shares the distoulnar surface of the bone, and is approxi-
mately 20% of the size of the capitate articulation. This surface has
a half-moon outline and is relatively ?at. The proximoradial side of
the bone is dominated by the articular surface for the radius. The
radial surface has a D-shape outline with the ?at edge of the D
occurring distally. Overall, this surface is moderately convex, but
much of this curvature occurs closer to the more palmar aspects of
the articulation; the more dorsal portions of the surface appear
rather ?at. The shared articular surface for the trapezium and
trapezoid is on the distal aspect of the bone. This surface is strongly
convex dorsopalmarly and displays a pronounced bump when
viewed radially or ulnarly. The trapezoid articulates primarily on
the more dorsally-slanting portion of the surface whereas the
trapezium would articulate on the more palmarly-slanting portion.
Furthermore, the articular surface does not extend to the end of the
tubercle, which is rather gracile and uniformly cylindrical.
Capitate (LB1/45) (Fig. 8) As in African apes and Australopithecus,
the neck of the capitate is highly excavated or waisted along its
radial aspect, resulting in radioulnar cross sections that are nar-
rower in the middle of the bone. The capitate head, or proximal
end, is approximately the same width as the palmar aspect of the
distal end, but the dorsal aspect of the distal end is slightly wider
than both. The maximum measurements of the capitate are
approximately 17.5 mm proximodistally, 11.1 mm radioulnarly,
and 14.1 mm dorsopalmarly, while the head measures approxi-
mately 7.5 mm radioulnarly. The lunate articular surface of the
head is broad and dominates the proximal aspect of the bone. In
contrast, the articular surface for the scaphoid is mostly oriented
radially and is only slightly convex such that little of the surface
is visible in either proximal or dorsal views. A dorsally situated
projection of the scaphoid joint surface extends distally and
forms a slight ?J-hook? in the radioulnar plane resulting in a knob-
like process projecting from the middle of the dorsoradial aspect
of the bone. Consequently, this portion of the scaphoid surface is
clearly visible in proximal view. Slightly distal to the knob-like
process lies a small, dorsally-placed articular facet for the trap-
ezoid, and immediately palmar to this facet is a distinct attach-
ment area for the trapezoid-capitate ligament. The third
metacarpal articulation is rather ?at dorsally but the more
palmar aspect becomes moderately indented along the surface,
particularly toward the ulnar side, whereas the articular facet for
the second metacarpal is sagittally-oriented along the palmar
half of the radial side of the bone. Finally, the hamate articular
surface extends along the entire dorsal and proximal aspects of
the ulnar side of the bone.
Hamate (LB1/46) (Fig. 8) The hamate is incomplete; it lacks the
entire proximal half of the bone as well as the hamulus. It measures
approximately 10.9 mm radioulnarly. The distal articular surface is
reasonably well-preserved and shows slight indentations on both
radial and ulnar aspects. On the distal aspect of the radial side of the
bone, the capitate articulation begins as a small, dorsally-placed
facet that broadens palmarly as it extends proximally. The ulnar
side of the bone preserves the distal half of the triquetrum articular
surface, which has a spiral shape and reaches to just beneath the
distal boundary of the ?fth carpometacarpal joint.
Trapezoid (LB1/47) (Fig. 8) The trapezoid has six distinct surfaces.
The largest surface is dorsal and is completely non-articular. The
remaining ?ve surfaces converge at the most proximopalmar tip of
the bone, producing an overall wedge-shaped appearance. The
maximum measurements of the trapezoid are approximately
9.9 mm proximodistally, 10.0 mm radioulnarly, and 7.6 mm dorso-
Evolution 57 (2009) 555?570 561palmarly. Large, triangular-shaped articulations for the scaphoid
and trapezium, respectively, dominate both the proximal and radial
manS.G. Larson et al. / Journal of Hu562surfaces. The articulation for the second metacarpal is /\-shaped,
with the ulnar side approximately four times larger than the radial
side. The ulnar side of this articular surface slopes considerably in
a proximal direction beginning from the dorsodistal boundary of
the joint. The remaining surface on the ulnar (i.e., medial) aspect of
the bone displays a palmarly-placed attachment area for the trap-
ezoid-capitate ligament as well as a small, dorsally-placed articu-
lation for the capitate.
Lunate (LB1/60) (Figs. 8 and 9) The lunate is incomplete, lacking
approximately 25% of its proximodorsal surface. Thus, the articular
surface for the triquetrum is missing, as is the most dorsal portion of
the articular surface for the radius. The maximum measurements of
the lunate are approximately 9.4 mm dorsopalmarly, 8.1 mm radio-
ulnarly, and 5.3 mm proximodistally (measured from the midpoints
of the capitate and radius articular surfaces). Despite the prox-
imodorsal damage, the distal articular surface for the capitate is
relatively complete because a thinportion of bone remains preserved
dorsodistally. This surface is strongly concave dorsopalmarly and
relatively ?at radioulnarly. The radial (i.e. lateral) side of the bone
preserves approximately 75% of the scaphoid articular surface,which
is placed slightly distal and dorsal to a large pit for the intercarpal
scaphoid-lunate ligament. Proximally, the articular surface for the
radius is moderately convex both dorsopalmarly and radioulnarly.
Metacarpals and phalanges
Metacarpal (LB1/59) This metacarpal is very fragmentary and lacks
both the proximal and distal ends. Approximately 30 mm of the
Figure 7. Articulated 3D models of left-sided carpal bones. In the middle is the LB1/44 scap
partial hamate (h). Comparable bones for Homo sapiens are on the left, and for Pan troglod
dorsoulnar portion of the lunate in LB1 are missing due to postmortem damage. All viewsEvolution 57 (2009) 555?570shaft is preserved. The body is curved slightly longitudinally and
faint markings for the interosseus muscles are present.
Phalanges (LB1/62, LB1/61, LB1/40, LB1/42, LB1/48, LB1/49, and
LB1/55) (Fig. 10) Two incomplete proximal manual phalanges, LB1/
62 and LB1/61 (formerly attributed to LB7, see Table 1), lack their
proximal bases and varying degrees of their shafts. LB1/62
preserves 33.4 mm of its total length, longer than any of the more
complete proximal pedal phalanges associated with LB1. LB1/61 is
very similar in overall morphology to LB1/62, but preserves just
under 28 mm of its total length. The shafts sport well-de?ned
marginal ridges for the attachments of the ?brous digital sheaths.
The shafts are expanded mediolaterally near midshaft and taper
distally to their heads. The heads end in two condylar surfaces
separated by a shallow groove, and the articular surface extends
farther onto the palmar surface than the dorsal. The shafts appear
to be relatively straight, but this is dif?cult to judge conclusively
without an articular base.
A complete intermediate manual phalanx, LB1/48, measures
25.6 mm in length, which is more than 10 mm longer than the
longest intermediate pedal phalanx attributed to LB1. The base has
two concave articular facets separated by an elevatedmedian ridge;
the dimensions of the base are 11.0 mm mediolaterally and 8.5 mm
dorsopalmarly. The shaft, which has marginal ridges for the digital
?brous sheaths, is 7.1 mm wide at midshaft and measures 5.3 mm
dorsopalmarly at the same level. The shaft tapers distally, and the
head expands into two condylar articular surfaces separated by
a shallow groove; the head measures 8.0 mm mediolaterally and
4.6 mm dorsopalmarly. LB1/40 is another intermediate manual
hoid (s), LB1/60 partial lunate (l), LB1/47 trapezoid (t), LB1/45 capitate (c), and LB1/46
ytes on the right. Note that the proximal half and the hamulus of the hamate and the
are scaled to one another using the proximodistal length of the capitate.
Figure 8. The LB1/45 left capitate, LB1/47 left trapezoid, LB1/44 left scaphoid, and LB1/46 partial left hamate (scale bar? 1 cm). The 1st and 4th columns from the left show the
original carpals; the 2nd and 5th columns show high-quality casts; the 3rd and 6th columns show 3D laser-scanned models. The capitate and scaphoid are shown in radial (on left)
and distal views (on right); the trapezoid and hamate are shown in palmar (on left) and distal views (on right).
Figure 9. The LB1/60 partial lunate and its articulating elements (scale bar? 1 cm). A. The midcarpal articular surfaces of the LB1/44 scaphoid and LB1/60 lunate (dorsal is toward
page top). B. The radiocarpal articular surfaces of the scaphoid and lunate (dorsal is toward page top); note the damage to the dorsal portion of the lunate. C. Palmar view of the LB1/
45 capitate (above) and LB1/60 lunate (below).
S.G. Larson et al. / Journal of Human Evolution 57 (2009) 555?570 563
phalanx with similar morphology, but it is damaged proximally;
what remains of it measures 19.7 mm in length. The head measures
7.6 mm by 3.82 mm, and the approximate midshaft measures
5.8 mm by 4.4 mm.
A nearly complete distal phalanx from one of the nonpollical rays,
LB1/49, is 13.4 mm long with a base that is 8.7 mm wide. The apical
tuft is eroded, but the palmar surface presents a rugose ungual
tuberosity. The ungual spine for the interosseus ligament is
preserved on one side only. It is human-like in overall osseous
anatomy. LB1/55 is a pollical distal phalanx that is eroded proximally
and distally; it is roughly 15.2 mm long, and it sports a well-de?ned
proximal pit that is associated with the attachment for the tendon of
?exor pollicis longus (Susman,1988; but see Shrewsbury et al., 2003).
LB6 upper limb elements
Radius (LB6/2) (Fig. 11)
The radius is represented by LB6/2. This complete right radius
has a highly distorted distal end (Fig. 11), which Morwood et al.
(2005) interpreted as the result of an unset, healed fracture with
compensatory remodeling and callus development. A radiograph of
the radius supports the interpretation of a healed fracture resulting
in ulnar displacement of the distal radius (Ortner, pers. comm.;
Sampson, pers. comm.). The overall length of the radius is 157 mm.
There is a shaft fracture through the radial tuberosity, and two
additional fractures 58 mm and 35 mm from the distal end. The
radial head is round (anteroposterior diameter? 15.69 mm;
mediolateral diameter? 14.78 mm) with a circular depression
(fovea) at its proximal end that is slightly offset laterally. The
anteroposterior width of the fovea is 10.07 mm and the medio-
lateral width is 11.2 mm. The mediolateral width of the medial
proximal articular surface (between the fovea and outer edge of
medial proximal surface) is 3.87 mm, and the anteroposterior
width of the anterior proximal articular surface (between the fovea
and outer edge of anterior proximal surface) is 2.35 mm. The outer
articular surface of the head is asymmetric and is widest medially
(medial width? 5.4 mm, lateral width? 3.9 mm). The bone lacks
a segment from the anterior aspect of the outer articular surface.
The radial tuberosity is not prominent and instead appears as
a roughened oval measuring 17.8 mm by 7.8 mm. It faces almost
directly medially. The anteroposterior midshaft diameter is 9.2 mm
and the mediolateral diameter is 10.86 mm, part of which is due to
the sharp and prominent interosseous crest.
An articular facet for the ulna is visible on the medial side of the
distal end, although the distal end is distorted. The carpal articular
surface is smooth and concave without any apparent ridge sepa-
rating the scaphoid and lunate surfaces. An unusual ?ange of bone
S.G. Larson et al. / Journal of Human Evolution 57 (2009) 555?570564Figure 10. A. Left to right, ventral views of LB1/61, LB1/52 (formerly LB7), LB1/48, LB1/40, and
40 are intermediate manual phalanges, and LB1/49 is a distal manual phalanx. B. Dorsal viewLB1/49. LB1/61 and LB1/52 are fragmentary proximalmanual phalanges, LB1/48 and LB1/
of same. Each block on the scale bars represent one centimeter.
manextends from the anteromedial border of the carpal surface.
Another unusual ?ange occurs along the lateral edge of the bone
near the distal end, possibly for the attachment of brachioradialis.
The anterior surface of the distal end is convex rather than concave,
and appears to be somewhat ??lumpy,?? probably as a result of callus
development and remodeling associated with the unset fracture.
The posterior side of the distal end is dominated by a prominent
swelling, perhaps representing the dorsal (Lister?s) tubercle. A
strong ridge extends proximally from this swelling. The maximum
distal mediolateral width is 20.2 mm.
Ulna (LB6/3)
The LB6/3 specimen is a left ulna with a complete proximal end
and shaft, but missing the distal end (Fig. 12, Table 3). There is
a shaft fracture just distal to the coronoid process and another
Figure 11. A. Composite photo of three views of the LB6/2 right radius. Distal view is
shown along side that of a modern human for comparison. Each block on the scale bar
represents one centimeter. B. Radiographic image of LB6/2 radius.
S.G. Larson et al. / Journal of Hufracture that is 33 mm distal to the coronoid process. In addition,
the distal portion of the shaft is composed of several pieces. The
preserved portion of the ulna is 137 mm long and displays a slight
anteroposterior curvature. The shaft is generally rounded in
contour and the interosseous crest begins at the supinator crest
and spirals slightly from dorsal to more ventral as it travels
distally. The exposed bone cortex at the distal end is 2.6?2.7 mm
thick, and the medullary cavity is round with a diameter of
4.0 mm.
Proximal end ? anterior view (Fig. 12) The trochlear notch is short
and broad and has a slightly morewaisted appearance than LB1/52.
The articular surface is divided into medial and lateral facets by
a blunt ridge, and the lateral facet is somewhat larger than the
medial. The two facets meet at an angle of 110 degrees and the
medial facet appears to be the more inclined of the two. There is
a large pit in the lateral facet.
A strong ridge extends distally for 35 mm from the lateral side of
the coronoid process toward the anterior midline of the shaft.
Another slightly less prominent ridge passes distally from the
medial side of the coronoid process and together they ?ank
a depression for the attachment of brachialis (the ulnar tuberosity).
Proximal end ? medial view (Fig. 12) The coronoid process projects
further anteriorly than does the proximal rim of the trochlear notch
(Table 3). There is a sharp ridge running distally from the lateral
side of the coronoid process that forms the lateral border of the
ulnar tuberosity. Dorsal to the distal half of the trochlear notch isa ridge that extends distally down the shaft as the border between
the anteromedial and posteromedial surfaces of the shaft. The
olecranon process appears to be short (Table 3).
Proximal end ? posterior view (Fig. 12) The posterior surface of the
proximal end is ?at. A central ridge forming the subcutaneous
surface arises just distal to the supinator crest, and then extends
down the shaft.
Proximal end ? lateral view (Fig. 12) The radial notch occupies the
lateral side of the coronoid process and is vertically oriented (with
a radial notch angle of 10 degrees). The supinator crest is a rough ridge
that begins just dorsal to the radial notch and extends for 13 mm. The
interosseous crest continues from the supinator crest dorsally to amore
ventral position as it travels distally.
Scapula (LB6/4)
A nearly complete right scapula, LB6/4, is missing only the
medial part of the spine and the supraspinous fossa (Figs. 13 and
14). It includes threemain pieces: a superolateral portion consisting
of the glenoid fossa, coracoid process, and lateral part of the blade;
Figure 12. Composite photo of three views of the LB6/3 left ulna. The proximal end is
complete but the distal end is missing. Each block on the scale bar represents one
centimeter.
Evolution 57 (2009) 555?570 565an inferomedial portion that makes up essentially the inferior half
of the blade; and the projecting part of the scapular spine. In
addition, several fractures exist within each of these large pieces.
The interface between the superolateral portion and inferomedial
portion lacks several fragments, and although they are reasonably
well aligned in the region of the axillary border, the alignment
diminishes across the infraspinous fossa due to postmortem
distortions in what is preserved of the scapular blade of each. This
misalignment creates a somewhat ??dished-in?? appearance of the
infraspinous fossa (compare Fig. 13A and 13B). The scapular spine
piece lacks fragments from the lateral tip of the acromion, and the
proximal regionwhere it attaches to the blade has several fractures
complicating exact alignment. The inferior angle of the blade is
eroded and seems to be incomplete. If so, it is unlikely that this is
due to a missing unfused epiphysis since the other elements of LB6
indicate adult age.
Spine and acromion (Figs. 13 and 14) The root of the spine emerges
dorsally and cranially from the blade. Several longitudinal fractures
occur at the root of the spine where smaller fragments have been
glued together. In addition, there are fractures in the portion of the
scapular blade where the root of the spine attaches, especially on
the superior side of the scapular spine, and there appears to have
been some distortion in this region. As a consequence, the scapular
spine fragment may be slightly misaligned so that it is a little too
ventral in position making the notch of the scapular neck very
narrow (Fig. 14C). The depth of the notch is 14.28 mm (from the
surface of the glenoid fossa). The distance between the most lateral
end of the root of the scapular spine and the axillary border of the
scapula is 15.50 mm. Although incomplete, the spine appears to
have been horizontal as it is in modern humans (see Larson et al.,
2007). A line along the base of the scapular spine forms a 45 degree
angle with the axillary border, and an angle of 102 degree with the
vertebral border.
A small round facet for articulation with the clavicle is visible at
the lateral tip of the acromion, and a short lip of bone extends
dorsally and laterally from that. Anterior to the facet there is some
erosion of the bone. The angle of the acromion is missing and there
is a shallow pit where it would have been. The acromion projects
well beyond the glenoid fossa (18 mm from the superior edge of the
glenoid). The acromion does not curve across the glenoid but
remains dorsal.
Glenoid fossa (Fig. 14A) The glenoid fossa is teardrop-shaped with
a notch in the superior ventral edge where the coracoid joins the
scapular neck. The articular surface has a height of 24.52 mm and
width of 15.59 mm. Some bone is eroded from the inferior dorsal
as lateral, inclination. The preserved portion of the coracoid is
18.26 mm long, with a 4.66 mm anteroposterior diameter and
9.32 mm mediolateral diameter. The coracoacromial projection
(Ciochon and Corruccini, 1977) is 14 mm but certainly would have
been higher with a complete coracoid. There is a prominent
dorsolateral tubercle on its superior surface (10.91 mm in length),
but since the portion of the coracoid and the superior border of the
scapula immediately medial to this tubercle is missing, the
completeness of the tubercle dif?cult to ascertain. If it is complete,
then the distance between its midpoint and the superior edge of
the glenoid fossa reaches 13.48 mm.
Axillary border (Fig. 13) The axillary border measures 82 mm in
length (Morwood et al., 2005), and the ??dishing?? of the infraspinous
fossa makes the border appear very prominent dorsally. The axil-
lary border begins at the infraglenoid tubercle as a sharp ridge.
Ventral to this ridge is a region bounded by the axillary border and
the ventral bar on the costal surface of the scapula, known as the
ventral sulcus (Stern and Susman, 1983) or subscapular lateral
expansion (Larson, 1995) (Fig. 13D). It begins as a depression
slightly inferior to the infraglenoid tubercle and faces inferiorly and
ventrally. Approximately two-thirds of the way toward the inferior
blad
S.G. Larson et al. / Journal of Human Evolution 57 (2009) 555?570566edge of the fossa, possibly reducing the maximum width
measurement. Two horizontal ?ssures extend through the middle
of the fossa. Both stem from a larger fracture that begins on the
ventral side of the midpoint of the glenoid rim, and then diverges
across the fossa toward the dorsal rim. A small fragment of bone is
missing from the common origin of these ?ssures. In addition, there
are smaller longitudinal ?ssures in the fossa. The supraglenoid
tubercle is a small bump at the most superior edge of the fossa.
There appears to have been some distortion to the scapular neck
resulting in a slight dorsal orientation to the glenoid (Fig. 14C). The
glenoid/ventral bar angle is 157 degrees, which puts it in the range
of modern humans (Stern and Susman, 1983).
Coracoid process (Fig. 14) The root of the coracoid is short (6.73 mm
measured along its inferior margin from the edge of the glenoid
fossa to the beginning of the coracoid process proper) and thick
(9.37 mm), and there are several pits in its surface. It forms an angle
of 130 degrees with the vertical axis of the glenoid fossa. The root
begins 5 mm medial to the lateral edge of the supraglenoid
tubercle. The coracoid itself is somewhat ?attened and narrow, and
it makes a sharp angle with the root (Fig. 14B). Although the tip of
the coracoid is missing, it is clear that it had a strong inferior, as well
Figure 13. Composite photo of different views of the LB6/4 right scapula. The scapular
most of the supraspinous fossa. A. Dorsal view with scapular blade lying ?at. The infraspin
dorsally. The effect of these possible distortions can be seen in B in which the blade has b
Ventral view of scapula showing missing regions of the coracoid process and superior angangle, this sulcus rotates to face more directly inferiorly and
essentially becomes the axillary border. The ridge that had been the
border continues as a dorsal margin of the sulcus. Near the inferior
angle, the sulcus narrows and twists even more dorsally so that it
merges with the dorsal surface of the inferior angle, and the ventral
bar completes the axillary border before merging with the ?ange of
bone forming the inferior angle. At its widest part the ventral
sulcus/lateral expansion is 8.98 mm wide and 53.58 mm long.
Scapular blade (Fig. 13) As noted above, there is a marked concavity
in the infraspinous fossa medial to the axillary border, perhaps due
to some distortions in the fragments and slight misalignments of
the pieces. The medial half of the inferomedial fragment ?ares
dorsally, especially near the inferior angle, also perhaps due to
postmortem distortion (compare Fig. 13A and 13B). The costal
surface of the superolateral fragment contains many fractures
surrounding the missing portions of the blade and coracoid process
(Fig. 13C). On the costal surface, the ventral bar begins at the
scapular neck as a blunt ridge, and forms the ventral border of the
ventral sulcus/lateral expansion. As the sulcus changes orientation
from facing ventral-inferior to directly inferior, and ?nally to infe-
rior and slightly dorsal, the ventral bar comes to form the axillary
e is composed of two main pieces joined across the infraspinous fossa, and is missing
ous fossa is dished in medial to the axillary border, and the vertebral border projects
een positioned so that the glenoid fossa is perpendicular to the plane of the photo. C.
le. D. Lateral view of axillary border/ventral sulcus.
raco
leno
y ha
manborder just before it merges with the bony ?ange making up the
inferior angle of the scapula. The preserved portion of the vertebral
border of the blade is very thin as is the bone forming the infra-
spinous/subscapular fossae. The vertebral border is slightly concave
on dorsal view, and straight when viewed edge-on.
Scapular dimensions Oxnard (1963) identi?ed several scapular
dimensions that were useful in differentiating the scapular form of
different primate locomotor groups. These include the ratio of
scapular length and the distance between the insertions of serratus
anterior and cranial trapezius, the angle of insertion of cranial
trapezius, the orientation of the components of the muscle couple
made up of trapezius and serratus, and the orientation of the gle-
noid fossa relative to the axillary border. For the LB6/4 scapular
fragment, scapular length (from the glenoid fossa to the vertebral
border) is 68.19 mm and the distance between the serratus anterior
and cranial trapezius insertions is 112.67 mm, yielding a ratio of
165.23 mm. However, this is at best an estimate because both
components of this ratio may be inaccurate. Scapular length is
traditionallymeasured to the point on the vertebral border where it
meets the scapular spine, and this is missing in LB6/4. In addition,
the distance between the insertions of serratus and cranial trape-
zius includes the inferior angle, some of which may also be missing.
Figure 14. Composite photo of different views of the LB6/4 scapula. A. Ventral view of co
tubercle. Each block on the scale bar represents one centimeter. B. Lateral view of the g
tubercle of the coracoid. The root of the scapular spine projects dorsally, but there ma
S.G. Larson et al. / Journal of HuThe angle of insertion of cranial trapezius in LB6/4 is 18 degrees, the
angle describing the orientation of the components of the trape-
zius/serratus force couple is 64 degrees, and the angle between the
glenoid fossa and the axillary border is 136 degrees. None of these
angles appears to be statistically different from those of modern
humans, falling within the corresponding 95% ?ducial limits of
modern human means reported by Oxnard (1963).
Metacarpal (LB6/5)
A metacarpal shaft, LB6/5, lacks the distal and proximal ends.
Morwood et al. (2005) listed it as a metatarsal; however, on reex-
amination, it appears more likely to be a metacarpal.
Phalanges (LB6/8, LB6/9, LB6/10, LB6/7, LB6/12) (Fig. 15)
A complete proximal manual phalanx, LB6/8, (Fig. 15B and 15C)
measures 31.2 mm long. Marginal ridges for the attachment of the
digital ?brous sheath are well-de?ned. It tapers gradually from its
elliptical base (10.7 mm by 8.5 mm) to midshaft (7.4 mm by
5.8 mm) and it measures 7.8 mm across the epicondyles of the
head. The paired condylar surfaces of the head are separated by
a groove, and the articular surface is muchmore extensive palmarlythan dorsally. The shaft is thick and exhibits a midline keel on the
palmar surface, not unlike those seen on LB1?s proximal phalanges.
Curvature of the shaft in the dorsopalmar plane can be estimated by
the included angle (Stern et al., 1995). At 34.5 degrees, LB6/8 falls at
the extreme upper end of the human range and overlaps with
gorillas. It is similar in this respect to A.L. 333w-4, an Austral-
opithecus afarensis specimen (Susman et al., 1984).
Intermediate hand phalanges include the nearly complete LB6/9
(Fig. 15B and 15C). LB6/9 measures 16.9 mm in length. The base has
double articular fossae, and the base itself measures 7.1 mm by
5.6 mm. As in human intermediate manual phalanges, the nar-
rowest part of the shaft is quite distal; midshaft dimensions are
4.5 mm by 3.7 mm. The head broadens to 6.1 mm. Markings for the
digital ?brous sheaths are present but faint. LB6/10 is another
intermediate manual phalanx but it is in very fragmentary
condition.
Complete distal manual phalanges are represented by LB6/12
(Fig. 15B) and LB6/7 (Fig. 15B and 15C) (save slight erosion of one
ungual spine in LB6/7). The apical tufts arewell developed, and LB6/
12 has well-preserved ungual spines for the paired interosseus
ligaments. Overall anatomy is clearly hominin. LB6/12 is 12.9 mm
long, has basal dimensions of 6.5 mm by 4.4 mm, and its apical tuft
id process region. The tip of the coracoid is missing as is the region medial to the dorsal
id fossa. C. Superior view of glenoid region showing supraglenoid tubercle and dorsal
ve been some postmortem distortion of the spine.
Evolution 57 (2009) 555?570 567is 4.9 mm across. Comparable measurements for LB6/7 are 11.4 mm
in length, 7.1 mm by 5.0 mm at the base, and 5.2 mm across the
apical tuft. The apical tuft expansion index (100 tuft breadth
divided by basal width) of both distal phalanges is greater than that
of the average modern human (Mittra et al., 2007). LB6/11 of
Morwood et al. (2005) is among the elements that could not be
located following the return of the material from Gajah Mada
University; however, with a maximum length of 10.5 mm, it cannot
be a proximal manual phalanx. This length would be more
consistent with another distal manual phalanx.
Upper limb elements from additional individuals
Ulna (LB2/1) (Fig. 16)
A proximal fragment of a right ulna, LB2/1, is from one of the
oldest sections of Liang Bua fromdeposits dated at 74 ka (Table 1). It
includes a complete trochlear notch and olecranon process and is
51.9 mm long (Fig. 16). The articular surface of the trochlear notch is
divided into medial and lateral semilunar facets by a strong ridge,
and the lateral facet is larger than the medial. The two surfaces
meet at an angle of 110 degrees. The lateral facet contains an
elongated pit. The medial edge of the notch is concave, giving the
becomes a smooth ridge continuing straight toward a medial ?are
representing the beginning of the distal end of the radius. Some
oblique striations are visible on the posterior side of the proximal
end, probably representing impressions from the supinator. At
approximately midshaft on the lateral side is a ?attened area of
roughened bone, perhaps for the attachment of pronator teres. On
t, LB6/9 (dorsal view), LB6/12 (dorsal view), and LB6/7 (palmar view). LB6/9 is a complete
lated elements of LB6/8, LB6/9, andLB6/7.
man Evolution 57 (2009) 555?570notch a waisted appearance, and the bone posterior to the notch
appears somewhat shallow (Table 3). Although the very tip of the
coronoid is eroded, the process still projects further anteriorly than
the proximal rim of the trochlear notch (Table 3). The beginning of
a broad ridge can be seen passing distally from the coronoid
process. Another low ridge extends distally from the medial edge of
the coronoid process. Dorsal to the distal half of the trochlear notch
on the medial side of the bone is a low ridge that extends distally,
and probably would have been the border between the ante-
romedial and posteromedial surfaces of the shaft. On the lateral
surface there is the proximal portion of a supinator crest beginning
at the base of the trochlear notch. Although the lateral rim of the
trochlear notch appears to be intact, the margins of the radial notch
are not visible. It is unclear whether this is due to postmortem/
Figure 15. A. Lateral view of LB6/8, a complete proximal manual phalanx. B. Left to righ
intermediate manual phalanx; LB6/12 and LB6/7 are distal manual phalanges. C. Articu
S.G. Larson et al. / Journal of Hu568depositional damage, or to some pathology during the life of the
individual. The olecranon process is short and square (Table 3). The
posterior surface of the bone dorsal to the trochlear notch is gently
rounded, and a ridge representing the subcutaneous surface begins
a little distal to the position of the coronoid process.
Radius (LB3) (Fig. 17)
A left radial shaft, LB3, lacks the head and distal end (Fig. 17). It
was found in one of the oldest sections of Liang Bua from deposits
dated at 74 k.yr. (Brown et al., 2004) (Table 1). The fragment is
164 mm in length, and Brown et al. (2004) estimate that its
complete length would have been approximately 210 mm. The
bone is quite straight except for a slight medial curve toward the
distal end. At midshaft, the mediolateral diameter is 11.47 mm and
the anteroposterior diameter is 10.81 mm. The radial neck appears
to have been long, measuring 13.52 mm from the proximal edge of
the radial tuberosity to the broken proximal end. The cross section
of the radial neck is round (its mediolateral width is 10.5 mm, and
its anteroposterior width is 11.1 mm), and the cortical bone is thin
and measures 1.3?1.7 mm. The medullary cavity has a diameter of
8.3 mm. The radial tuberosity is 18.51 mm long and 8.95 mm wide.
It is more prominent than that of LB6/2, but like LB6/2, the radial
tuberosity faces more directly medially than in modern humans,
judging from the position of the interosseous crest. The proximal
portion of the interosseous crest begins 18.09 mm from the distal
edge of the radial tuberosity as a rough line for 17.8 mm, and then Figure 16. Composite photo of two views of right proximal ulnar fragment LB2/1.
the anterior surface of the preserved portion of the distal end is
a hint of the outline for the insertion of pronator quadratus.
Radius (LB4/1) (Fig. 17)
A juvenile left radius, LB4/1, lacks its distal end (Fig. 17). The
proximal end is complete except for the epiphysis for the radial
head. It is 101.61 mm long, with a mediolateral midshaft diameter
of 8.31 mm, and anteroposterior diameter of 6.98 mm. The shaft is
broken in two places, one at about 40.57 mm from the proximal
end, and the other at 55.7 mm from the proximal end. The shaft is
straight except for a slight medial curve toward the distal end. The
radial tuberosity is prominent and is 8.76 mm long and 5.88 mm
wide. Like LB3 and LB6/2, it faces medially rather than ante-
romedially. The radial neck is 10.61 mm long (measured from the
proximal edge of the radial tuberosity to the missing epiphysis),
and round (mediolateral width is 6.74 mm, and anteroposterior
width is 6.82 mm), but ?ares slightly toward the head. The inter-
osseous crest runs straight toward the beginning of the ?are for the
distal end. At approximately midshaft on the lateral side is a ?at-
tened rough area, probably for the attachment of pronator teres.
Metacarpal (LB5/2)
Morwood et al. (2005) reported the existence of a fragmentary
adult metacarpal lacking the base and measuring 58 mm in length
(LB5/2). However, because it could not be found following the
return of the Liang Bua material from Gajah Mada University,
additional description of this specimen is not possible.
Lorraine Cornish of the Natural History Museum (London). The
S.G. Larson et al. / Journal of HumanFigure 17. Composite photo of four views of the (presumably) adult left radius LB3,
and juvenile left radius LB4/1. Both show slight medial curvature. Judging by the
position of the interosseous crest, the radial tuberosity faces more medially than is
typical in modern humans.Wenner-Gren Foundation provided funds to support the curation
and preservation of the Liang Bua fossil material. We also thank
Djuna Ivereigh (www.indonesiawild.com) for her photographs of
the carpal bones and casts.
References
Argue, D., Donlon, D., Groves, C., Wright, R., 2006. Homo ?oresiensis: Microcephalic,
pygmoid, Australopithecus, or Homo? J. Hum. Evol. 51, 360?374.
Brown, P., Sutikna, T., Morwood, M.J., Soejono, R.P., Jatmiko, Saptomo, E.W., Rokus
Awe Due, 2004. A new small-bodied hominin from the Late Pleistocene of
Flores, Indonesia. Nature 431, 1055?1061.
Ciochon, R.L., Corruccini, R.S., 1977. The coraco-acromial ligament and projection
index in man and other anthropoid primates. J. Anat. 124, 627?632.
Hershkovitz, I., Kornreich, L., Laron, Z., 2007. Comparative skeletal features between
Homo ?oresiensis and patients with primary growth hormone insensitivity
(Laron Syndrome). Am. J. Phys. Anthropol. 134, 198?208.
Jacob, T., Indriati, E., Soejono, R.P., Hsu?, K., Frayer, D.W., Eckhardt, R.B.,
Kuperavage, A.J., Thorne, A., Henneberg, M., 2006. Pygmoid Australomelanesian
Homo sapiens skeletal remains from Liang Bua, Flores: Population af?nities and
pathological abnormalities. Proc. Nat. Acad. Sci. 103 (36), 13421?13426.
Larson, S.G., 1995. New characters for the functional interpretation of primate
scapulae and proximal humeri. Am. J. Phys. Anthropol. 98, 13?35.
Larson, S.G., 1996. Estimating humeral torsion on incomplete fossil anthropoid
humeri. J. Hum. Evol. 31, 239?257.
Larson, S.G., Jungers, W.L., Morwood, M.J., Sutikna, T., Jatmiko, Saptomo, E.W., Rokus
Awe Due, Djubiantono, T., 2007. Homo ?oresiensis and the evolution of the
hominin shoulder. J. Hum. Evol. 53, 718?731.
Lordkipanidze, D., Jashashvile, T., Vekua, A., Ponce de Leo?n, M.S., Zollikofer, C.P.E.,
Rightmire, G.P., Pontzer, H., Ferring, R., Oms, O., Tappen, M., Bukhsianidze, M.,
Agusti, J., Kahlke, R., Kiladze, G., Martinez-Navarro, B., Mouskhelishvile, A.,
Mioradze, M., Rook, L., 2007. Postcranial evidence from early Homo from
Dmanisi, Georgia. Nature 449, 305?310.
Mittra, E.S., Smith, H.F., Lemelin, P., Jungers, W.L., 2007. Comparative morphometrics
of the primate apical tuft. Am. J. Phys. Anthropol. 134, 449?459.
Morwood, M.J., Soejono, R.P., Roberts, R.G., Sutikna, T., Turney, C.S.M.,
Westaway, K.E., Rink, W.J., Zhao, J.-X., van den Bergh, G.D., Rokus Awe Due,
Hobbs, D.R., Moore, M.W., Bird, M.I., Fi?eld, L.K., 2004. Archaeology and age of
a new hominin from Flores in eastern Indonesia. Nature 431, 1087?1091.
Morwood, M.J., Brown, P., Jatmiko, Sutikna, T., Saptomo, E.W., Westaway, K.E., Rokus
Awe Due, Roberts, R.G., Maeda, T., Wasisto, S., Djubiantono, T., 2005. Further
evidence for small-bodied hominins from the Late Pleistocene of Flores, Indo-Phalanx (LB12)
An immature distal phalanx, LB12, lacks the basal epiphysis and
the apical tuft is damaged.
Conclusions
The fossil material known for H. ?oresiensis includes examples
from nearly every component of the upper limb. Most of the
material is part of the LB1 partial skeletonmaking it one of themost
complete fossil skeletons currently known. While most upper limb
elements are represented by single specimens, multiple examples
are known from some regions, such as the proximal ulna. To date, in
depth comparative analyses have been completed only for the
shoulder and wrist, but much has been written about the
morphology of H. ?oresiensis in service of alternative interpreta-
tions of this material based on limited descriptive information. The
detailed documentation of the morphology of each known upper
limb element presented here, along with those for other parts of
the skeleton included in this volume, will hopefully facilitate
thoughtful consideration of this material, whatever its ultimate
interpretation.
Acknowledgments
Some of the costs associated with analysis of the Liang Bua
hominin remains in Jakarta were covered by an Australian Research
Council grant to MJM. Conservation efforts were spearheaded by
Evolution 57 (2009) 555?570 569nesia. Nature 437, 1012?1017.
Obendorf, P.J., Oxnard, C.E., Keffort, B.J., 2008. Are the small human-like fossils
found on Flores human endemic cretins? Proc. R. Soc. B 275, 1287?1296.
Oxnard, C.E., 1963. Locomotor adaptations of the primate forelimb. Symp. Zool. Soc.
Lond. 10, 165?182.
Richards, G.D., 2006. Genetic, physiologic and ecogeographic factors contributing to
variation in Homo sapiens: Homo ?oresiensis reconsidered. J. Evol. Biol. 19 (6),
1744?1767.
Schultz, A.H., 1937. Proportions, variability and asymmetries of the long bones of
the limbs and the clavicles in man and apes. Hum. Biol. 9, 281?328.
Shrewsbury, M.M., Marzke, M.W., Linscheid, R.L., Reece, S.P., 2003. Comparative
morphology of the pollical distal phalanx. Am. J. Phys. Anthropol. 121, 30?47.
Stern Jr., J.T., Jungers, W.L., Susman, R.L., 1995. Quantifying phalangeal curvature:
An empirical comparison of alternativemethods. Am. J. Phys. Anthropol. 97,1?10.
Stern Jr., J.T., Susman, R.L., 1983. The locomotor anatomy of Australopithecus afar-
ensis. Am. J. Phys. Anthropol. 60, 279?317.
Susman, R.L., 1988. New postcranial remains from Swartkrans and their bearing on
the functional morphology and behavior of Paranthropus robustus. In: Grine, F.E.
(Ed.), Evolutionary History of the ??Robust?? Australopithecines. Aldine, New
York, pp. 149?172.
Susman, R.L., Stern Jr., J.T., Jungers, W.L., 1984. Arboreality and bipedality in the
Hadar hominids. Folia Primatol. 43, 113?156.
Tocheri, M.W., Orr, C.M., Larson, S.G., Sutikna, T., Jatmiko, Saptomo, E.W., Rokus
Awe Due, Djubiantono, T., Morwood, M.J., Jungers, W.L., 2007. The primitive
wrist of Homo ?oresiensis and its implications for hominin evolution. Science
317, 1743?1745.
Voisin, J.-L., 2006. Clavicle, a neglected bone: morphology and relation to arm
movements and shoulder architecture in primates. Anat. Rec. 288A,
944?953.
S.G. Larson et al. / Journal of Human Evolution 57 (2009) 555?570570