Bone 40 (2007) 1265 – 1275

www.elsevier.com/locate/bone

Strain redistribution and cracking behavior of human bone

during bending
Vincent Ebacher a , Cecelia Tang b , Heather McKay b , Thomas R. Oxland b ,
Pierre Guy b , Rizhi Wang a,⁎
a
Department of Materials Engineering, University of British Columbia, Vancouver, BC, Canada
b
Department of Orthopaedics, University of British Columbia, Vancouver, BC, Canada

Received 13 September 2006; revised 19 November 2006; accepted 4 December 2006

Available online 8 January 2007

Abstract

Long bones often fail due to bending loads. Understanding the fracture process during bending is of great importance to the prevention and
treatment of bone fractures. In this study, we investigated the origin of long bone’s bending strength through the study of the dynamic strain
redistribution happening during the post-yield stage of deformation and its relation to microdamage at the microstructural level. This was
accomplished by comparing the behaviors of human long bones with standard cortical bone specimens in terms of strain redistribution, Poisson’s
ratios, microdamage morphologies, and macro-scale fracture patterns. It was found that human tibia failure in bending was very similar to that of
standard beam cortical bone specimens with respect to the four previous aspects. Also, the examination of bone’s Poisson’s ratio indicated very
different inelastic deformation mechanisms under tension and compression: bone volume expanded in tension but was nearly conserved in
compression. Finally, as a result of strain redistribution, bone’s bending strength mainly depended on its compressive strength, which was
significantly influenced by the osteonal “porous” microstructure of human bone as compared to its tensile behavior. Thus, we concluded that bone
microstructure at the Haversian system level plays an important role in bone deformation and fracture.
© 2006 Elsevier Inc. All rights reserved.

Keywords: Bending fracture; Cortical bone; Inelastic deformation; Microdamage; Poisson’s ratio; Strain redistribution

Introduction the bone thickness during the post-yield stage of deformation.

Physiologically, long bones rarely fail due to pure tension or
compression, but rather fail as a result of torsion or bending
loads or a combination of the two [1–3]. Thus, bending is one of
the most common modes of loading that may lead to bone
failure. When a long bone is subjected to bending, one side is
under compressive stress, while the other side is under tensile
stress. It is well known that although bone has the same Young’s
modulus in tension and compression [4,5], it yields and
fractures at lower stresses in tension than in compression.
Because of this asymmetrical stress–strain behavior [5–9], both
stress and strain would undergo continuous redistribution across
⁎ Corresponding author. Fax: +1 604 822 3619.
E-mail address: rzwang@interchange.ubc.ca (R.Z. Wang).
8756-3282/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.bone.2006.12.065
Understanding the detailed stress and strain redistribution
process before fracture and its dependence on bone micro-
structure and geometry holds a key to the understanding of bone
strength. In spite of its obvious importance to the prevention and
treatment of bone fractures, experimental studies on such a
deformation process have been limited.
The dynamic strain redistribution in bone subjected to
bending was first studied by Burstein et al. [10] who
developed an analytical model based on force balance,
assuming idealized elastic–perfectly plastic asymmetric
stress–strain behaviors. Their analysis showed that the
discrepancy between measured bending and tensile strengths
were due to inelastic deformations and asymmetric yield
strengths of bone resulting in a shift of the neutral axis of the
beams towards the compressive surface. Simkin and Robin
[11] also demonstrated that bending moment and strain and
1266 V. Ebacher et al. / Bone 40 (2007) 1265–1275
stress distributions across a bending beam at failure could be
predicted based on force and moment balance theory and a
graphical method involving post-yield stresses. More recently,
Fondrk et al. [12] used a damage model based on the non-
linear tensile behavior of cortical bone to simulate cantilever
bending and showed that the neutral axis shifted towards the
compressive surface following tensile yielding. All these
studies emphasized the importance of post-yielding processes.
As Currey [13] concluded with his study on a wide variety of
bone types, post-yield deformations are one of the main factors
that determine the bending strength of compact bone. Hence,
an exhaustive study on post-yield deformation and its dynamic
redistribution is essential to the understanding of bone fracture
process in bending. Such a study on human cortical bone is
still lacking. More importantly for the clinical field is the lack
of such study on whole bones.
A uniaxial stress causes materials to deform not only in the
axial direction but also in the transverse directions. Thus,
transverse strains and Poisson’s ratios (ν = − εt/εl) should also be
considered when studying the post-yielding process. Unfortu-
nately, only a few papers reported Poisson’s ratio values for wet
bone [6,7,14–18] and most of these values were not in
agreement with each other [6]. The early work of Ko [14] and
the recent work of Fondrk et al. [15], Pidaparti and Vogt [16], and
Mercer et al. [17] on bovine and human cortical bone showed
that the tensile Poisson’s ratio would decrease with load or
fatigue cycle, indicating that, under tensile stress, bone inelastic
deformation would not be volume conservative. Therefore,
human cortical bone compressive and tensile Poisson’s ratios
and especially their changes during deformation would help us
to better understand bone deformation process leading to
fracture.
Inelastic strains occurring in bone are mainly associated with
microdamage [6,19–33]. It is well known that microdamage
morphologies in tension and compression are totally different
[20,21,25,29]. Compressive microcracks are relatively straight
and long [20,25,26,29,31], and oriented at approximately 30–
40° from the bone’s long axis [6,9,19,20,34] forming a typical
cross-hatched pattern. Tensile microdamage is of a diffuse nature
and consists of smaller microcracks (≈ 2–10 μm) forming flame-
like arrays [22,24,25,29,31,33] oriented normally to the tensile
stress [23,24,27]. Crack initiation has been linked to the
presence of Haversian canals [9,21,29] and even osteocyte
lacunae [32]. However, the contributions of other structural
features and variables, such as bone lamellae, geometry of
Haversian systems, and degree of mineralization have been far
less studied [35]. It is also not clear how the micro-scale cracks
are developed into a macro-scale fracture pattern such as the
well-known “butterfly” fracture pattern commonly found in long
bone fractures [21].
The purpose of this study was to investigate the deforma-
tion and fracture process of human bones in bending through
the study of strain redistribution and its relation to micro-
damage at the microstructural level. We compared the fracture
of whole bone with cortical bone specimens in terms of strain
redistribution, Poisson’s ratios, microdamage morphologies,
and macro-scale fracture patterns. The role of human bone
osteonal microstructure on microdamage development was
also examined.
Materials and methods
A total of 19 un-embalmed human cadaver long bones (15 tibiae, 4 femora)
were used in this study. They were obtained from the Department of Anatomy at
the University of British of Columbia. The tibia specimens, 5 males and 5 females
(5 pairs plus 5 right), were used for the whole bone study. Age at death ranged
from 67 to 88 years. The femoral specimens, 3 males and 1 female selected based
on thickness of the cortex, were used for the cortical bone study. Age at death
ranged from 69 to 77 years. All bones were visually examined for macroscopic
defects or pre-fracture and were stored at − 20 °C until testing. The study was
approved by the Clinical Research Ethics Review Board at the University of
British Columbia.
Mechanical tests
For the whole bone study, the distal 25% of each tibia specimen was first
removed in order to be used as part of another investigation. The remaining
proximal 75% sections of the tibial shafts were fractured in four-point bending
under wet conditions using a servohydraulic testing machine (Instron 8874,
25 kN load cell) with a crosshead speed of 6.0 mm/min. The inner loading span,
between the 50% and 66% sites from the distal end, was adjusted with respect to
specimen length by changing the positions of the load application points. The
outer support span was set consistently to be three times the length of the loading
span. Therefore, the removal of the distal 25% did not interfere with our
measurements and our goal of studying the strain redistribution. To ensure
stability and alignment, the distal end was embedded in PMMA and a PMMA
support was added at the proximal end. Prior to mechanical testing, rosette strain
gages (Omega KFG-2-120-D17-11L1M2S) were mounted at the center of the
span (∼ 58% of the bone length from the distal end) on both the lateral and
medial surfaces, which would be subjected to compression and tension,
respectively (Fig. 1b). These surfaces were defined according to the orientation
of the proximal condyle. The strain gage application procedure followed
standard guidelines (Vishay Intertechnology Inc., Malvern, USA). Briefly,
surface preparation consisted of cleaning with water, degreasing with diluted
isopropyl alcohol, lightly abrading with 400-grit silicon carbide paper, and
finally conditioning with M-Prep Conditioner A and neutralizing with M-Prep
Neutralizer 5A. Then, the strain gages were aligned and bonded to the surface
using M-Bond 200 Adhesive combined with M-Bond 200 Catalyst and coated
with a thin layer of coating (M-Coat) to protect them from moisture. Data were
collected using a data acquisition system managed by LabVIEW software
(National Instruments; Austin, USA). From these measurements, principal
strains, strain rates, longitudinal strain ratios (longitudinal strain on the tensile
surface over that on the compressive surface, −εT/εC), transverse strain ratios, as
well as Poisson’s ratios (ν = − εt/εl) were calculated. Since the cross-sections of
tibia specimens were not standard and the surfaces were not flat, the absolute
values of the strain ratios and the Poisson’s ratios would vary from bone to bone.
Hence, for comparison purposes, the ratios were normalized with respect to their
individual value at 10% of the fracture load.
For the cortical bone study, 10 cortical bone specimens were cut
longitudinally from the mid-diaphysis of the four femora using a low-speed
diamond saw (Isomet 1000, Buehler) under constant water irrigation. Two to four
samples, with final dimensions of 35 mm × 4 mm × 3 mm, were obtained from
each femur. They were first manually ground into rectangular beams and then
mechanically polished down to 1.0 μm using diamond suspension. The upper and
lower surfaces (4 mm sides) were sectioned to be roughly parallel to the radial
direction of the femoral cortex to minimize structural variations across the beam
thickness (3 mm sides) and allow a better comparison between the compressive
and tensile behaviors when subjected to bending. The specimens were placed in a
phosphate-buffered solution (0.05 PBS, pH 7.2) until rosette strain gages (Omega
KFG-1-120-D17-11L1M2S) were applied to the center of the upper and lower
surfaces, corresponding to compressive and tensile surfaces. Four-point bending
tests were conducted under wet conditions using a servohydraulic testing
machine (Instron Dynamight, 1 kN load cell). A crosshead speed of 0.5 mm/min
was chosen so that the strain rate on the specimen surface would be close to that
 V. Ebacher et al. / Bone 40 (2007) 1265–1275
Fig. 1. Typical load–strain curves: (a) Cortical bone specimen; (b) Whole tibia
specimen; Inserts: Configurations of the four-point bending on cortical bone
specimens and proximal tibia specimens, respectively; SG: Strain Gage,
measured in the long bone bending tests. The inner loading span was 15 mm and
the outer support span was 30 mm. The same calculations were performed as
described for the whole bone study. Additionally, the stress–strain curves were
calculated via the de-convolution equations [17,36].
Fracture and microdamage analyses
Following mechanical testing, whole tibia macro-scale fracture patterns
were classified by an orthopaedic surgeon (P.G.) according to the A.O.
classification [1]. The specimens were then carefully examined under a
stereomicroscope (Nikon SMZ 1000) to characterize the macro-scale damage
and to locate potential sites for further microdamage analysis. Eight of the 15
tibia specimens with typical macro-scale fracture patterns (in terms of A.O.
fracture type) and load–strain curves were selected for microdamage
observations. They were transversely cut into segments (comprising the selected
sites for microdamage observations) using a hand saw under constant irrigation
at positions that were at least 1 cm away from the main fracture surfaces.
For the observation of microdamage, both cortical bone and segments of
tibia specimens were stained using a procedure based on the work of Zioupos
and Currey [37] and Fazzalari et al. [27]. The specimens were first put into
acetone overnight for defatting and then fixed and dehydrated into a graded
series of ethanol/water solutions (80%, 90%, and 100%) for periods of 24 h per
step. They were then stained under vacuum for 24 h in a filtered saturated
solution of fluorescein (Fisher Scientific) and 70% ethanol. Finally, they were
rinsed in 100% ethanol for at least 1 h before being air-dried. The selected sites
on the lateral (compressive) and medial (tensile) surfaces of the tibia specimens
were then carefully cut using the diamond saw under constant water irrigation,
embedded in epoxy resin (Epothin, Buehler), and finally ground and polished
using a mechanical polishing machine (Isomet 1000, Buehler) to expose the
surface of the cortices. The compressive and tensile surfaces, which were either
the original upper and lower surfaces of the cortical bone specimens or the
1267
surfaces parallel to the original upper and lower surfaces of the tibia specimens,
of all the prepared specimens were ultimately examined under an optical
microscope (Nikon Eclipse E600) using both the white light and the epi-
fluorescence light (with excitation at approximately 490 nm and emission at
approximately 525 nm). Microdamage introduced during the bending tests
would be stained by the fluorescein dye and appear bright green under the
fluorescence microscope. The dye also stained some elements of the bone
microstructure such as osteocyte lacunae and Haversian systems. The angles
between the stained shear cracks and the bone’s long axis were measured on the
surfaces, assuming that the crack planes were perpendicular to the analyzed
surfaces.
As the gray levels obtained from backscattered electron (BSE) images are
sensitive to the average atomic number, this technique has often been used to
quantify bone mineral content [38–42]. Thus, to relate the occurrences of
microdamage with variations in degree of mineralization, selected specimens
were carbon coated by vacuum evaporation (JEE-4B Vacuum Evaporator,
JEOL — Japan Electron Optics Laboratory Co. Ltd., Tokyo, Japan) and further
examined under scanning electron microscope (Hitachi S-3000N, Hitachi Ltd.,
Tokyo, Japan) using the backscattered electron (BSE) detector.
Results
Strain measurements
The principal strains generally corresponded to the long-
itudinal and transverse directions of the specimen’s long axis.
Fig. 1 presents typical load–strain curves for both cortical bone
and whole tibia specimens. The curves are similar in shape
between the two groups. The longitudinal strains on both the
tensile and the compressive surfaces initially increase linearly
with load, which is followed by a non-linear stage where the
tensile strain becomes progressively higher than its compressive
counterpart. This clearly depicts a strain redistribution taking
place during the post-yield deformation process within both
cortical bone and whole bone specimens. In order to better
understand such a phenomenon, the longitudinal and transverse
strain ratios as well as the Poisson’s ratios were plotted against
the load normalized with respect to the fracture load of
individual specimens (Figs. 2a,b: strain ratios; Figs. 2c,d:
Poisson’s ratios). As the load in cortical bone specimens
increased, the longitudinal strain ratio initially remained more or
less constant to approximately 60% of the fracture load, and then
increased until fracture. The redistribution of the longitudinal
strains towards the tensile surface implies that the neutral axis in
the bending specimens moves towards the compressive surface.
The transverse strains followed an opposite trend and redis-
tributed towards the compressive surface. This is better under-
stood by analyzing the tendency of the Poisson’s ratios during
loading. The tensile Poisson’s ratio decreased from 0.35 ± 0.03
to 0.21 ± 0.04 as load increased. Assuming that human cortical
bone is transversely isotropic [7–9,15], this suggests that, in the
inelastic regime, the volume of bone would not remain constant:
bone expands under tension in the way that the transverse strain
does not increase proportionally to the increase of the
longitudinal strain. On the compressive side, starting from
0.34 ± 0.01, Poisson’s ratio increased just prior to fracture.
Although the value did not reach 0.5, the increasing trend
suggests that the deformation mechanism of bone under
compression would be nearly volume conservative. Note finally
that, although more pronounced probably due to architecture,
1268 V. Ebacher et al. / Bone 40 (2007) 1265–1275
Fig. 2. Longitudinal and transverse strain ratios for (a) cortical bone specimens, (b) whole bone specimens; tensile and compressive Poisson's ratios for (c) cortical
bone specimens, (d) whole bone specimens.
the same trends in both strain ratios and Poisson’s ratios were
observed in whole tibiae (Figs. 2b,d).
The stress–strain curves for five cortical bone specimens
(Fig. 3) show very similar tensile and compressive moduli
(23 ± 3 GPa and 23 ± 2 GPa, respectively). The tensile curves are
close to each other in both shape and stress level. The ultimate
stress reaches ∼ 110 MPa. Yield generally occurs before
Fig. 3. Tensile and compressive stress–strain curves for five cortical bone
100 MPa, which is followed by a very mild hardening and
then a slight softening up to an ultimate strain of approximately
1.0% to 1.5%. The compressive curves are more scattered and
the strength values are higher than the tensile ones, in the range
of 150–200 MPa. The compressive yield point is also higher
than the tensile yield but its ultimate strain is generally lower for
an individual specimen. Note that, due to possible strain gage
debonding at later stage of the tests, it was not possible to
calculate the curves all the way to the final fracture so that a
complete description of the fracture process in bending could
not be obtained.
Typical curves of the longitudinal strain rates during the bone
deformation process are shown in Fig. 4. The strain rates in
tension and compression are first similar but then differ, the
tensile strain rate increasing at the expense of its compressive
counterpart. Later in the process, the tensile strain rate reaches a
maximum before dramatically decreasing prior to the final
fracture. Note that the increase of the tensile strain rate
corresponds approximately to the tensile yield point. Also, the
maximum tensile strain rate is reached prior to maximum load,
meaning that the loading capability of the bone is still
increasing. This strain rate peak also suggests that the strains
would become localized away from the strain gage. Note finally
 V. Ebacher et al. / Bone 40 (2007) 1265–1275
Fig. 4. Typical load–time curves and variations of strain rates: (a) Cortical bone
that whole tibia specimens showed similar trends to cortical
bone specimens although there were some differences due to
architectural effects and influence of the set-up: the tensile strain
rate was initially higher than the compressive one and increased
significantly more than for the cortical bone specimens, both
strain rates increased after their initial steady states and prior to
the decrease in compressive strain rate.
Fig. 5. Side views of typical “butterfly” macro-scale fracture patterns: (a) Cortical bone specimen; (b) Whole tibia specimen; (c) pQCT (peripheral Quantitative
Computed Tomography) cross-section of tibia at the location of the strain gage (58%). White area shows geometry of bone cortex.
1269
Fracture and microdamage analyses
The macro-scale fracture patterns of the whole tibiae
classified according to the A.O. classification [1] were as
follows: five specimens showed an oblique fracture (A2), nine a
bending wedge or “butterfly” fracture (B2) and one a
fragmented wedge fracture (B3). The macro-scale fracture
patterns of the standard beam specimens were very similar to
the “butterfly” fractures of whole bones (Fig. 5). The fracture
path is straight and runs transversely to the beam on the tensile
side, while it is oblique on the compressive side. It suggests a
difference in failure modes when bone is subjected to different
stress states. Clinically, the “butterfly” fracture pattern usually
consists of two macro-scale oblique shear cracks on the
compressive side connected to a transverse crack on the tensile
side, thus forming a “Y” shape [1]. In this study, half of the
standard beam specimens showed only one of these shear cracks
(Fig. 5a), while most of the whole tibiae showed at least two
oblique shear cracks although all were not necessarily paths of
final failure (Fig. 5b).
At the microscopic level, a clear difference in microdamage
morphologies was also observed between the compressive and
the tensile surfaces in both the standard beam and the whole
bone specimens. On the compression surface, most of the
microdamage had typical “cross-hatching” pattern (Fig. 6a)
where the discrete microcracks are long and straight but
oblique to the specimen’s long axis. Based on 10 measurable
macro-scale cracks on 10 different cortical bone specimens,
the average angle between the crack plane and the bone’s long
axis (assuming that the crack plane is perpendicular to the
surface used for the measurement) was 27.4 ± 7.4°. Based on
59 microcracks from eight different cortical bone specimens,
the average angle was 28.4 ± 4.5°. Scale-type cracks (Fig. 6b)
were also observed at different locations on the compressive
surface. This type of damage was not stained by the
fluorescein dye and could only be seen under normal reflected
light. Note that these scale-type cracks follow a similar
curvature to the macro-scale crack on the compressive surface
(Fig. 6b). The tensile surface showed extensive diffuse
1270 V. Ebacher et al. / Bone 40 (2007) 1265–1275

microdamage (Fig. 6c) generally oriented transversely to the
longitudinal tensile stress and running into and across the
surface. Although not necessarily continuous, both cross-
hatching and diffuse microdamage seemed uniformly dis-
tributed along the entire region subjected to the maximum
bending moment (inner loading span) of the standard beam
specimens. When following the fatal macro-scale cracks from
the tensile side to the compressive side, diffuse microcracks
were observed (Fig. 7) in the fracture process zone until just
before the tensile crack merges to the shear crack on the
compressive side where the damage in the process zone is not
detectable. This clearly indicates different crack propagation
mechanisms.
Closer observation of the stained specimens under bright
field and epi-fluorescence modes revealed the important role of
bone ultrastructure in the process of fracture. Both tensile and
compressive microdamage showed initiation at stress concen-
trators such as Haversian canals (Fig. 6). However, tensile
microdamage could initiate in both osteonal and interstitial
bone. In Fig. 8, diffuse microcracks are obviously developed in
interstitial bone as they do not extend to the lamellae of
neighboring osteons. According to the BSE image, the

Fig. 6. Microdamage morphologies under bright field (left) and epi-fluorescence (right) modes: (a) Compression surface showing cross-hatching microdamage far
away from the final fracture site; (b) Compression surface showing scale-type microdamage at the final fracture site; (c) Tensile surface showing diffuse microdamage
at the final fracture site. Principal stress along length of images.
 V. Ebacher et al. / Bone 40 (2007) 1265–1275 1271

Strain redistribution during bending

Bone shows a discrepancy between its bending and tensile

strengths [5–9], similar to synthetic fiber composites [44]. It
was showed that this discrepancy was due to inelastic
deformations as well as asymmetric tensile and compressive
strengths [10–13,44–46]. In cortical bone, the inelastic strains
appearing after yielding are mainly associated with micro-
damage [23,24]. Thus, their dynamic redistribution would
determine the ultimate properties of bone in bending. In spite of

Fig. 7. Side view of tensile diffuse microdamage in the fracture process zone

interstitial area has a higher degree of mineralization, indicating

that the initiation of tensile microdamage could also be related
to bone mineral density distribution. Compressive cross-
hatched microcracks were more structure dependent as they
were frequently localized within individual Haversian systems
(Fig. 9). They did not appear to initiate within or extend easily
to the interstitial bone surrounding the osteon, a region shown to
be more highly mineralized in the BSE image. Consequently,
the cross-hatching pattern often developed into bands that were
confined within individual osteons suggesting that the presence
of Haversian canals and the size, geometry, and properties of
Haversian systems would significantly affect compressive
microcrack initiation and development.

Discussion

The strain measurements and the staining procedure

enabled a direct comparison of the failure process of whole
bone and cortical bone specimens. The combination of bright
field and epi-fluorescence optical microscopy as well as BSE
electron microscopy further allowed us to relate microdamage
development to bone microstructure. Therefore, it was
possible to characterize bone fracture process from the
material level to the structural level. When subjected to
bending, whole tibia specimens failed similarly to standard
cortical bone specimens in terms of strain redistribution,
microdamage development, and macro-scale fracture patterns.
Our observations both at the micro-scale and the macro-scale
clearly highlighted the different damage mechanisms under
tensile and compressive stress states. The similarity between
the measured shear angles formed by individual microcracks
(28.4 ± 4.5°) and macro-scale cracks (27.4 ± 7.4°) evokes a
potential relation between compressive microdamage and
macro-scale cracks in the form of microcrack growth or
coalescence as it is the case in tension [29,43]. Therefore, the
“butterfly” bending fractures observed clinically in long bones
Fig. 8. (a) Tensile diffuse microdamage in both osteonal (O) and interstitial (I)
[1] are most probably the result of a compressive failure as bone visualized under epi-fluorescence mode; (b) Corresponding bright field
well as a tensile failure of bone material starting at the mode micrograph; (c) Corresponding BSE image. Tensile stress along top–
microstructural level. bottom axis of images.
1272 V. Ebacher et al. / Bone 40 (2007) 1265–1275
Fig. 9. (a) Compressive cross-hatching microdamage localized within an individual Haversian system (H) visualized under epi-fluorescence mode; (b) Corresponding
BSE image. Compressive stress along top/right–bottom/left axis of images. Arrows mark two of the shear cracks.
this, very little attention has been given to the understanding of
this phenomenon.
In both the whole bone and the standard beam specimens,
our results clearly show that, as bone is loaded to fracture, the
inelastic tensile strains become progressively higher than the
compressive ones (Figs. 1 and 2). This strongly confirms that
strain redistribution takes place in human cortical bone
subjected to bending. These results not only strengthen the
analyses by previous researchers [10–12] but add the fact that
the tendencies are also observed in whole human tibia
specimens. This implies that the basic response to bending
loads and the deformation process taking place prior to final
failure are essentially the same even though whole bone
architecture may affect the magnitude of the response. There-
fore, bone fracture at the material level plays a fundamental role
to the structural failure.
Assuming a linear strain distribution across the beam
thickness, it was possible to calculate the tensile and
compressive stress–strain curves from a bending test
[17,36,44,46,47]. The ensuing curves (Fig. 3) reveal tendencies
that agree relatively well with the literature [5–9] and that
evidently support the strain redistribution occurring during
bending as tensile yielding happens prior to compressive
yielding. The possibility of strain gage debonding at later stage
of the tests may be one reason why the compressive strengths
are somewhat lower than reported in the literature [9], although
Reilly and Burstein [6] reported values of 131 to 159 GPa. The
elastic moduli are also relatively high as compared to the values
of 17.7 ± 3.6 GPa given in Reilly and Burstein [7]. One possible
explanation could be the stiffening effect of the strain gages and
their mounting medium due to an increased beam thickness.
However, this error was estimated to be approximately 1 GPa.
What can be learned from these curves is that the tensile
behavior seems relatively constant in shape and level from one
specimen to another as compared to the more scattered
compressive behavior. This suggests that, as opposed to
compressive microdamage, the development of tensile micro-
damage would not be very sensitive to variations in bone’s
microstructural features (e.g. porosity and Haversian systems)
occurring from one specimen to another. Additionally, since it
was observed that microdamage was uniformly distributed over
the stressed surfaces, it is safe to assume that, at least prior to the
maximum tensile strain rate, the strain measurements would be
representative of the deformation process and therefore, that the
strain rates (Fig. 4) would give very good insight on the rate of
microdamage accumulation prior to the final failure. Our results
suggest that, in contrast to compressive microdamage, tensile
microdamage would start to develop early in the fracture
process and would accumulate increasingly rapidly as a result of
the shift of the neutral axis. This is consistent with the study by
Diab and Vashishth [48] who found that diffuse microdamage
on the surface of bovine cortical beams subjected to fatigue
reached an early saturated state as compared to compressive
microdamage. Reilly and Currey [29] also reported that in
equine cortical bone, tensile microdamage appeared at very low
strains prior to compressive microcracks and increased in
density with strain but only coalescing into longer microcracks
at very high strains. Nonetheless, our results add the fact that
these trends were also observed in whole bone specimens,
implying the presence of the same damage accumulation
process.
Poisson’s ratio
Poisson’s ratio is an important parameter to the study of bone
deformation process. In spite of this, reports on human cortical
bone’s compressive and tensile Poisson’s ratios remain very
limited. Our results showed very similar values for tensile and
compressive Poisson’s ratios of human cortical bone in the
elastic regime: 0.35 ± 0.03 and 0.34 ± 0.01, respectively. How-
ever, opposite trends were found when considering the inelastic
stage: decreasing tensile Poisson’s ratio and increasing com-
pressive Poisson’s ratio. Indeed, early measurements of tensile
Poisson’s ratio of cortical bone by Ko [14] showed values
decreasing from 0.8 to 0.45 as load increased, although the
absolute values were not reliable due to experimental inaccura-
cies (a value of 0.8 being thermodynamically impossible based
on the theory of isotropic elasticity [49]). Fondrk et al. [15]
examined the trend of the volumetric strains under rapid tensile
loading of human and bovine cortical bone and found an average
 V. Ebacher et al. / Bone 40 (2007) 1265–1275
Poisson’s ratio value of 0.352 ± 0.021, which is comparable to
our result. They concluded that bone’s volume was increasing
with load and associated the phenomenon with the development
of new internal surfaces. Pidaparti and Vogt [16] tested human
cortical bone specimens in tensile fatigue and found a Poisson’s
ratio of 0.28 which was decreasing with the loading cycles as a
result of damage accumulation. Mercer et al. [17] also concluded
that bovine bone dilates under tension because the transverse
strains were negligible but that the deformations are nearly
volume conservative under compression. Under the assumption
of transverse isotropy for human cortical bone [7–9,15], our
results confirm these findings and show that whole bones also
deform according to the same deformation mechanisms. More
importantly, they emphasize the difference between the inelastic
deformation mechanisms of human bone in tension and
compression giving rise to different damage patterns.
Role of bone microstructure
The ability of bone to develop microdamage in order to
dissipate energy is the key to prevent its early failure. This
ability will depend on the properties of the different elements of
its microstructure. Yet, at least for human bone, the under-
standing of the influence of bone microstructure on the
development of microdamage has been very limited. It has
been shown that the stress concentration effect around
Haversian canals [9,21,29] and osteocyte lacunae [32] provided
sites for microdamage initiation. From our observations, tensile
microdamage could also initiate in the more highly mineralized
interstitial bone (Fig. 8). This could be explained based on the
fact that yield strain decreases as mineral content increases [50].
In Fig. 8, both the osteonal bone and the interstitial bone would
undergo the same strain because of continuity of strain, and the
zone with higher degree of mineralization (i.e. the interstitial
bone) would yield sooner through microcracking. Local
variations of bone mineral content could therefore contribute
to tensile microdamage initiation. We hypothesize that this
would make tensile microdamage less sensitive to the presence
of Haversian canals than compressive microdamage, resulting
in more homogeneous damage as seen from a lower
magnification (Fig. 6c). Note that the tensile stress–strain
curves (Fig. 3), showing little variations from one specimen to
another, also support this hypothesis.
In contrast, our observations of compressive surfaces
showed a strong influence of local Haversian systems. The
cross-hatched shear cracks were not only initiated at the
Haversian canals, they were also often confined within each
Haversian system (Fig. 9). To our knowledge, this crack
localization within individual osteons has not been reported
before. The underlying mechanism is probably related to the
fact that the degree of mineralization, and hence the elastic
modulus [13,51,52], within each Haversian system is lower than
that in the surrounding interstitial bone, as can be seen from the
BSE micrograph in Fig. 9b. Such a difference in elastic modulus
has been known to increase the cracking resistance when a crack
approaches the interface from the side with lower modulus (the
Haversian system) [53]. The weak interface at the cement line
1273
[21,54–56] could also be involved in crack arresting even
though crack deflection at the cement line was not observed in
the current study. The susceptibility of Haversian systems to
microdamage may also be related to their low shear strength. It
was reported that osteons with lower degree of mineralization
had lower shear strength [57]. Therefore, Haversian systems
may have a lower shear strength than the more highly
mineralized interstitial bone, and would be prone to early
shear cracking especially in the presence of stress concentration.
It is not clear whether microdamage localization is associated
with any specific collagen fibril organization existing inside the
Haversian systems. Ascenzi and Bonucci [34] reported that the
compressive strength of osteon samples was lower in those
having marked longitudinal spiral course of collagen fiber
bundles in successive lamellae. Variations in mineral content
and collagen fiber orientation within lamellar sublayers may
also impede the development of microcracks as suggested by
Gupta et al. [35] and contribute to the local cross-hatched
microcracks.
Our results therefore indicate that compressive cracking is
not only more sensitive to the presence of Haversian canals than
tensile cracking, but also depends on the mineralization, size,
geometry, and microstructure of the Haversian systems. The
scattering in shape and levels of compressive stress–strain
curves (Fig. 3) also supports these observations. It can be
concluded that, as a result of the strain redistribution process,
bone’s compressive behavior, which is more sensitive to
variations in bone’s microstructural features (e.g. porosity and
Haversian systems), would be the major determinant of the
resistance to fracture in bending. Since cortical bone porosity
increases with age [58–61], the cracking sensitivity to
Haversian canals (i.e. porosity) found here in the senior
population may not directly apply to a younger population.
Nevertheless, the results are highly relevant to the senior
population with the highest bone fracture risk. Bone micro-
structure at the Haversian system level plays an important role
in bone deformation and fracture. Further work is thus needed to
understand the origins of the compressive microcracks within
the secondary osteons.
Acknowledgments
This study was funded by the Canadian Institutes of Health
Research and the Michael Smith Foundation for Health
Research. R.W. is incumbent of the Canada Research Chair in
Biomaterials. V.E. is grateful for the University Graduate
Fellowship from UBC. We thank Dr. Peter Zioupos and Dr.
John D. Currey for their valuable instructions on bone crack
staining. We also thank Dr. Danmei Liu for discussions and
assistance in the experiments.
References
[1] Muller ME, Nazarian S. Classification of fractures of the femur and its use
in the A.O. index (author’s transl). Rev Chir Orthop Reparatrice Appar
Mot 1981;67:297–309.
[2] Court-Brown CM, McBirnie J. The epidemiology of tibial fractures. J
Bone Joint Surg Br 1995;77:417–21.
1274 V. Ebacher et al. / Bone 40 (2007) 1265–1275
[3] Grutter R, Cordey J, Buhler M, Johner R, Regazzoni P. The
epidemiology of diaphyseal fractures of the tibia. Injury 2000;31(Suppl
3):C64–7.
[4] Bargren JH, Andrew C, Bassett L, Gjelsvik A. Mechanical properties of
hydrated cortical bone. J Biomech 1974;7:239–45.
[5] Reilly DT, Burstein AH, Frankel VH. The elastic modulus for bone. J
Biomech 1974;7:271–5.
[6] Reilly DT, Burstein AH. Review article. The mechanical properties of
cortical bone. J Bone Joint Surg Am 1974;56:1001–22.
[7] Reilly DT, Burstein AH. The elastic and ultimate properties of compact
bone tissue. J Biomech 1975;8:393–405.
[8] Natali AN, Meroi EA. A review of the biomechanical properties of bone as
a material. J Biomed Eng 1989;11:266–76.
[9] Currey JD. Bones: Structure and Mechanics. Princeton, NJ: Princeton
University Press; 2002. p. 3–123.
[10] Burstein AH, Currey JD, Frankel VH, Reilly DT. The ultimate properties
of bone tissue: the effects of yielding. J Biomech 1972;5:35–44.
[11] Simkin A, Robin G. The mechanical testing of bone in bending. J Biomech
1973;6:31–9.
[12] Fondrk MT, Bahniuk EH, Davy DT. A damage model for nonlinear tensile
behavior of cortical bone. J Biomech Eng 1999;121:533–41.
[13] Currey JD. What determines the bending strength of compact bone? J Exp
Biol 1999;202:2495–503.
[14] Ko R. The tension test upon the compact substance of the long bones of
human extremities. J Kyoto Pref Med Univ 1953;53:503–25.
[15] Fondrk MT, Bahniuk EH, Davy DT. Inelastic strain accumulation in
cortical bone during rapid transient tensile loading. J Biomech Eng
1999;121:616–21.
[16] Pidaparti RM, Vogt A. Experimental investigation of Poisson’s ratio as
a damage parameter for bone fatigue. J Biomed Mater Res 2002;
59:282–7.
[17] Mercer C, He MY, Wang R, Evans AG. Mechanisms governing the
inelastic deformation of cortical bone and application to trabecular bone.
Acta Biomater 2006;2:59–68.
[18] Shahar R, Zaslansky P, Barak M, Friesem AA, Currey JD, Weiner S.
Anisotropic Poisson’s ratio and compression modulus of cortical bone
determined by speckle interferometry. J Biomech 2007;40:252–64.
[19] Chamay A. Mechanical and morphological aspects of experimental over-
load and fatigue in bone. J Biomech 1970;3:263–70.
[20] Currey JD, Brear K. Tensile yield in bone. Calcif Tissue Res 1974;
15:173–9.
[21] Carter DR, Hayes WC. Compact bone fatigue damage: a microscopic
examination. Clin Orthop Relat Res 1977(127):265–74.
[22] Schaffler MB, Pitchford WC, Choi K, Riddle JM. Examination of compact
bone microdamage using back-scattered electron microscopy. Bone
1994;15:483–8.
[23] Zioupos P, Currey JD, Sedman AJ. An examination of the micromechanics
of failure of bone and antler by acoustic emission tests and Laser Scanning
Confocal Microscopy. Med Eng Phys 1994;16:203–12.
[24] Zioupos P, Currey JD. The extent of microcracking and the morphology of
microcracks in damaged bone. J Mater Sci 1994;29:978–86.
[25] Boyce TM, Fyhrie DP, Glotkowski MC, Radin EL, Schaffler MB. Damage
type and strain mode associations in human compact bone bending fatigue.
J Orthop Res 1998;16:322–9.
[26] Burr DB, Turner CH, Naick P, et al. Does microdamage accumu-
lation affect the mechanical properties of bone? J Biomech 1998;
31:337–45.
[27] Fazzalari NL, Forwood MR, Manthey BA, Smith K, Kolesik P. Three-
dimensional confocal images of microdamage in cancellous bone. Bone
1998;23:373–8.
[28] Taylor D, Lee TC. Measuring the shape and size of microcracks in bone.
J Biomech 1998;31:1177–80.
[29] Reilly GC, Currey JD. The development of microcracking and failure in
bone depends on the loading mode to which it is adapted. J Exp Biol
1999;202:543–52.
[30] O’Brien FJ, Taylor D, Dickson GR, Lee TC. Visualisation of three-
dimensional microcracks in compact bone. J Anat 2000;197(Pt
3):413–20.
[31] Reilly GC, Currey JD. The effects of damage and microcracking on the
impact strength of bone. J Biomech 2000;33:337–43.
[32] Reilly GC. Observations of microdamage around osteocyte lacunae in
bone. J Biomech 2000;33:1131–4.
[33] Vashishth D, Koontz J, Qiu SJ, et al. In vivo diffuse damage in human
vertebral trabecular bone. Bone 2000;26:147–52.
[34] Ascenzi A, Bonucci E. The compressive properties of single osteons. Anat
Rec 1968;161:377–91.
[35] Gupta HS, Stachewicz U, Wagermaier W, Roschger P, Wagner HD, Fratzl
P. Mechanical modulation at the lamellar level in osteonal bone. J Mater
Res 2006;21:1913–21.
[36] Wang R, Suo Z, Evans AG, Yao N, Aksay IA. Deformation mechanisms in
nacre. J Mater Res 2001;16:2485–93.
[37] Zioupos P, Currey JD. Pre-failure toughening mechanisms in the dentine of
the narwhal tusk: Microscopic examination of stress/strain induced
microcracking. J Mater Sci Lett 1996;15:991–4.
[38] Skedros JG, Bloebaum RD, Bachus KN, Boyce TM, Constantz B.
Influence of mineral content and composition on graylevels in
backscattered electron images of bone. J Biomed Mater Res 1993;
27:57–64.
[39] Skedros JG, Bloebaum RD, Bachus KN, Boyce TM. The meaning of
graylevels in backscattered electron images of bone. J Biomed Mater Res
1993;27:47–56.
[40] Boyde A, Jones SJ, Aerssens J, Dequeker J. Mineral density quantitation of
the human cortical iliac crest by backscattered electron image analysis:
variations with age, sex, and degree of osteoarthritis. Bone 1995;16:
619–27.
[41] Bloebaum RD, Skedros JG, Vajda EG, Bachus KN, Constantz BR.
Determining mineral content variations in bone using backscattered
electron imaging. Bone 1997;20:485–90.
[42] Roschger P, Fratzl P, Eschberger J, Klaushofer K. Validation of
quantitative backscattered electron imaging for the measurement of
mineral density distribution in human bone biopsies. Bone 1998;23:
319–26.
[43] George WT, Vashishth D. Damage mechanisms and failure modes of
cortical bone under components of physiological loading. J Orthop Res
2005;23:1047–53.
[44] Steif PS, Trojnacki A. Bend strength versus tensile-strength of fiber-
reinforced ceramics. J Am Ceram Soc 1994;77:221–9.
[45] Allen HG. Stiffness and strength of two glass-fiber reinforced cement
laminates. J Compos Mater 1971;5:194–207.
[46] Laws V. Derivation of the tensile stress–strain curve from bending data. J
Mater Sci 1981;16:1299–304.
[47] Yokoyama T. A microcomputer-aided four-point bend test system
determining uniaxial stress–strain curves. J Test Eval 1988;16:198–204.
[48] Diab T, Vashishth D. Effects of damage morphology on cortical bone
fragility. Bone 2005;37:96–102.
[49] Fung YC. Foundations of Solid Mechanics. Englewood Cliffs, NJ:
Prentice-Hall; 1965. p. 252–3.
[50] Currey JD. Tensile yield in compact bone is determined by strain, post-
yield behaviour by mineral content. J Biomech 2004;37:549–56.
[51] Currey JD. The effect of porosity and mineral content on the Young’s
modulus of elasticity of compact bone. J Biomech 1988;21:131–9.
[52] Currey JD. Physical characteristics affecting the tensile failure properties
of compact bone. J Biomech 1990;23:837–44.
[53] He MY, Hutchinson JW. Crack deflection at an interface between
dissimilar elastic-materials. Int J Solids Struct 1989;25:1053–67.
[54] Burr DB, Schaffler MB, Frederickson RG. Composition of the cement line
and its possible mechanical role as a local interface in human compact
bone. J Biomech 1988;21:939–45.
[55] Schaffler MB, Radin EL, Burr DB. Mechanical and morphological
effects of strain rate on fatigue of compact bone. Bone 1989;10:
207–14.
[56] Courtney AC, Hayes WC, Gibson LJ. Age-related differences in post-yield
damage in human cortical bone. Experiment and model. J Biomech
1996;29:1463–71.
[57] Ascenzi A, Bonucci E. The shearing properties of single osteons. Anat Rec
1972;172:499–510.
 V. Ebacher et al. / Bone 40 (2007) 1265–1275 1275

[58] Thompson DD. Age changes in bone mineralization, cortical
thickness, and Haversian canal area. Calcif Tissue Int 1980;31:
5–11.
[59] Brockstedt H, Kassem M, Eriksen EF, Mosekilde L, Melsen F. Age- and
sex-related changes in iliac cortical bone mass and remodeling. Bone
1993;14:681–91.
[60] Bousson V, Meunier A, Bergot C, et al. Distribution of intracortical
porosity in human midfemoral cortex by age and gender. J Bone Miner Res
2001;16:1308–17.
[61] Thomas CD, Feik SA, Clement JG. Regional variation of intracortical
porosity in the midshaft of the human femur: age and sex differences.
J Anat 2005;206:115–25.