om
m
dam
d Ki
, De
Keywords:
Load sharing
EMG
d sh
men
ine
chi
and lateral head of triceps) were measured simultaneously during a subset of the above conditions.
rable m
ing ph
idually
a muscle, namely the activation dynamics (Ca restoring in the
sarcoplasmatic reticulum) and the contraction dynamics (detach-
ment of cross-bridges). Given the assumption that total energy
consumption might be minimised in a given task, it is not clear
to what extent changes in task conditions will influence load shar-
ing. It is obvious that changing joint angle will influence both the
moment arm and muscle length of a muscle crossing that joint,
approximate energy consumption through mechanical variables
such as muscle stress (assuming a direct correlation between en-
ergy consumption and muscle stress). Validations of these cost
functions have been generally done by EMG (Kaufman et al.,
1991a; Happee, 1994; van der Helm, 1994; Buchanan and Shreeve,
1996; Raikova and Prilutsky, 2001).
Praagman et al. (2006) performed direct measurements of the
energy consumption of individual muscles using Near InfraRed
Spectroscopy (NIRS). Twomechanical cost functions (the stress cost
function and a newly proposed energy-related cost function) were
compared with in vivo measured muscle oxygen consumption. In
* Corresponding author.
Journal of Electromyography and Kinesiology xxx (2010) xxx–xxx
Contents lists availab
gr
.e ls
ARTICLE IN PRESSE-mail address: m.praagman@fbw.vu.nl (M. Praagman).load sharing occurs between muscles, while this does not appear to
be mechanically necessary. Although frequently studied (Hardt,
1978; Dul et al., 1984; Buchanan et al., 1989; Kaufman et al.,
1991b; Challis, 1997), the principles behind this load sharing phe-
nomenon are still unknown. A leading thought is that the total en-
ergy cost of the particular activity is minimised by keeping
individual muscle contributions low and thus preventing fatigue
and the detrimental effects on efficiency related to the occurrence
of fatigue (Alexander, 1997). The energy cost of a muscle can be as-
sumed to depend on the two major energy-consuming processes in
2+
advantageous moment arm and optimum length would require
less muscle force and activation to exert the required moment,
one could also assume that the activation and energy consumption
of that particular muscle might be lower.
Musculoskeletal models heavily depend on valid cost functions
to estimate load sharing. A considerable number of cost functions
has been used (see Tsirakos et al. (1997) for an overview). Most
cost functions are based on muscle force, often scaled by physio-
logical cross sectional area or maximal force. Since predictions of
energy consumption are difficult, most cost functions attempt toNIRS
Muscle oxygen consumption
Elbow muscles
Moment arm
Muscle length
1. Introduction
Humans appear to follow compa
principles in movement tasks. A strik
a particular force task, an inter-indiv1050-6411/$ - see front matter  2010 Published by
doi:10.1016/j.jelekin.2010.04.003
Please cite this article in press as: Praagman M
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.0Joint angle and therefore both moment arm and muscle length influenced both EMG amplitude and the
load sharing between muscles. The principles behind load sharing, however, were difficult to quantify,
since it was impossible to distinguish between all individual aspects that affect muscle activity. We found
a linear relationship between EMG and _VO2, while joint angle had no major effect. Although in general
subjects showed comparable muscle activation patterns, there were also considerable inter-individual
differences, which might be explained by the use of different optimisation strategies or differences in
morphology.
 2010 Published by Elsevier Ltd.
usculoskeletal control
enomenon is that, given
compatible pattern of
but it is absolutely obscure what effect that might have on load
sharing. It can be argued that muscles with larger moment arms
and near to their optimum length are the most advantageous mus-
cles to be used and therefore their activation and energy consump-
tion would be highest in this position. On the other hand, since theAvailable online xxxx
and anconeus) was recorded in all combinations of FE and PS moments at three force levels and four
elbow angles (50, 70, 90 and 110). Second, (n = 4) EMG and _VO2 of three muscles (both heads of bicepsThe effect of elbow angle and external m
M. Praagman a,*, E.K.J. Chadwick b, F.C.T. van der Hel
aResearch Institute MOVE, Faculty of Human Movement Sciences, VU University Amster
bDepartment of Sport and Exercise Science, Aberystwyth University, Aberystwyth, Unite
cMan Machine Systems and Control, Department of Design, Engineering, and Production
a r t i c l e i n f o
Article history:
Received 20 August 2008
Received in revised form 12 March 2010
Accepted 2 April 2010
a b s t r a c t
To study elbow muscle loa
nation–supination (PS) mo
Two data sets were obta
heads of biceps brachii, bra
Journal of Electromyo
journal homepage: wwwElsevier Ltd.
et al. The effect of elbow angl
03ent on load sharing of elbow muscles
c, H.E.J. Veeger a,b
, The Netherlands
ngdom
lft University of Technology, The Netherlands
aring we investigated the effect of external flexion–extension (FE) and pro-
ts and elbow angle on muscle activation and oxygen consumption ( _VO2).
d. First, (n = 6) electromyography (EMG) of elbow flexors (long and short
oradialis, brachialis) and extensors (long and short heads of triceps brachii
le at ScienceDirect
aphy and Kinesiology
evier .com/locate / je lek ine and external moment on load sharing of elbow muscles. J Electromyogr

et al., 1998). An inter-optode distance of 4 cm was used and the
ogra
ARTICLE IN PRESSthat study it was shown that the energy-related cost function
showed better results than the stress cost function. Comparisons
between modelling results and experimental data for muscle oxy-
gen consumption indicated, especially for the stress cost function,
a large number of ‘‘false negatives”: the model estimated NO mus-
cle activity where a muscle was active, based on experimental
observations. The use of an energy-related cost function led to few-
er false negatives. It was, therefore, concluded that the latter cost
function appeared to be a promising improvement (Praagman
et al., 2006). In this previous study, we based our comparisons on
data for only one arm position and limited combinations of elbow
flexion–extension and forearm pronation–supination moments.
However, from results by Jamison and Caldwell (1993) it can be in-
ferred that, despite the complexity of measurements, these results
cannot readily be extrapolated to all possible combinations of joint
moments. Also, the muscle force–length relationship could not yet
be accounted for, while, as said before, it is likely that load-sharing
is influenced by joint angle.
Energy consumption and activation might differentiate and do
not show the same relationship for each joint angle. It is, therefore,
not certain that muscle activation, as measured with (surface) EMG
and energy consumption will show a linear relationship under all
conditions, although this relationship was shown to exist for a sin-
gle joint position (Praagman et al., 2006). As a consequence, the
most appropriate method to quantify muscle energy consumption
is still NIRS.
Muscle function is often studied around a single joint or degree-
of-freedom, ignoring the interaction with adjacent joints. For
example, the activation of m. biceps brachii not only influences el-
bow flexion, but also forearm supination and glenohumeralante-
flexion. To perform a simple elbow flexion torque, it is, therefore,
inevitable that the role of a- and antagonistic wrist, forearm and
shoulder muscles should be taken into account (Buchanan et al.,
1989; Jamison and Caldwell, 1994). A study by Jinha et al. (2006)
has shown that muscle activity predictions using a one or two de-
grees of freedom modelling approach did not lead to valid predic-
tions when more degrees of freedom were present in the system.
This implies that a valid study on in vivo muscle coordination
should:
1. involve all relevant degrees of freedom and
2. involve all relevant combinations of external force conditions.
The consequence of the above should be that validation can
only take place using models of sufficient reality and experimental
data with sufficient information. This study describes the collec-
tion of the data necessary for model cost function validation.
The purpose of the current study was to investigate the princi-
ples of load sharing between muscles in a multiple degrees of free-
dom joint system, i.e., the elbow. An extensive data set was
obtained, including variation in muscle length and moment arms,
which can be used for validation of the previously introduced cost
functions. To account for possible differences between activation
dynamics and energy consumption, both EMG (indication of mus-
cle activation) and muscle oxygen uptake (as indication for energy
consumption) were measured for selected muscles. To study the
mechanism of load sharing, sufficient combinations of flexion/
extension and pro/supination moments around the elbow and
forearm were measured. Inclusion of these moments would enable
to study the effect of the interaction between different degrees of
freedom. It was investigated in what way the activation of elbow
flexors and extensors was influenced by the different external mo-
ments and whether this was influenced by elbow angle. We further
2 M. Praagman et al. / Journal of Electromyinvestigated the influence of elbow angle on both the load sharing
between muscles and the relationship between _VO2 and EMG
within muscles.
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003differential path length factor was set to 4.0. Measurements were
done during arterial occlusion, applied by inflating a thin cuff,
placed around the upper arm, to a pressure of at least 230 mm
Hg. _VO2 values were determined by taking the slope of the linear
part of the [O2Hb] decrease immediately after occlusion. For a de-
tailed description see Praagman et al. (2003).
2.2. Set-up
Subjects were seated on a chair with their elbow flexed at aOne hypothesis is that joint angles for which a certain muscle
will have a larger moment arm and are closer to optimal muscle
length will lead to a decrease in EMG amplitude for that muscle.
However, an alternative hypothesis is that a larger moment arm
and proximity to optimal length would favour this muscle above
others, which would result in an increase of EMG amplitude.
According to the relationship between EMG and _VO2 we expect
that the previously described linear relationship (Praagman et al.,
2006) will not hold for different angles, due to changes in load
sharing.
2. Methods
Two related experiments were performed. In experiment I, sub-
jects had to perform a full set of 49 combinations of flexion/exten-
sion and pro/supination moments. EMG measurements were
performed on four elbow flexors: m. biceps brachii caput breve
(BB), m. biceps brachii caput longum (BL), m. brachialis (BA) and
m. brachioradialis (BR), and four elbow extensors: m. triceps bra-
chii caput longum (TR), m. triceps brachii caput laterale (TL), m. tri-
ceps brachii caput mediale (TM) and m. anconeus (AC).In
experiment II, EMG and muscle oxygen consumption ( _VO2) were
measured for three elbow muscles (BB, BR and TL). _VO2 was mea-
sured using Near InfraRed Spectroscopy (NIRS). Measurements
with NIRS took place during arterial occlusion of the upper arm
and, therefore, required relatively long periods of rest after each
period of force production. Measurements of experiment II were,
therefore, much more time-consuming than measurements of
experiment I, in which EMG was measured only. To ensure a pro-
tocol of acceptable length a selection of the moment combinations
performed in the first experiment was studied. Even using these
selected conditions, experiments still took 4 days per subject.
In experiment I, six subjects (four females, two males, age
19.2 years (SD 2.3), height 1.71 m (SD 0.07), body mass 64.3 kg
(SD 2.7)) participated. Experiment II comprised four male subjects
(age 29.4 years (SD 7.3), height 1.75 m (SD 0.06), body mass
70.2 kg (SD 6.9)). Prior to the experiments, all subjects were in-
formed on the intent, procedures and risks of the experiments
and then signed an informed consent review form. The protocols
of both experiments were separately reviewed and approved by
the local ethical committee.
2.1. Data collection
EMG was recorded during periods of force-production only (in-
ter-electrode distance of 2 cm, analogous band-pass filter of 5–
400 Hz, sample frequency 1000 Hz) and digitally rectified. Changes
in concentration of oxyhaemoglobin (O2Hb) and deoxyhaemoglo-
bin (HHb) of the muscles were recorded continuously with two
continuous-wave, near infrared spectrophotometers (OXYMON,
Artinis Medical Systems, Arnhem, The Netherlands) (van der Sluijs
phy and Kinesiology xxx (2010) xxx–xxxfixed angle and their forearm horizontal and in a neutral position
(Fig. 1). There was no elbow or arm support. Subjects had to gen-
erate pure moments around the elbow joint (flexion (FL) and
e and external moment on load sharing of elbow muscles. J Electromyogr

ogra
ARTICLE IN PRESSM. Praagman et al. / Journal of Electromyextension (EX)) and radio-ulnar joint (pronation (PR) and supina-
tion (SU)), as well as combinations of these moments (flexion-supi-
nation (FS), flexion-pronation (FP), extension-supination (ES) and
extension-pronation (EP)). The subjects held a special tool with
their right hand, consisting of a stick with a horizontal bar on top
to which on several positions weights (0.75, 1.5, 3 or 4.5 kg) could
be applied (directly or through a pulley), enforcing the external
moments the subject had to withstand (Fig. 1). This resulted in
flexion/extension moments around 5, 10 and 15 Nm and pro/supi-
nation moments around 1, 2 and 3 Nm. Subjects were instructed to
hold the tool in a fixed position keeping the bar horizontal. Feed-
back was given by means of a horizontal cord in front of the
subject.
During experiment I, the periods of force production lasted 5 s,
with 30-s rest intervals. A full set of 49 flexion/extension and pro/
supination moment combinations was measured (Fig. 2). The mo-
Fig. 1. Experimental set-up. Subject was sitting on a chair with the elbow flexed
and forearm in a horizontal and neutral position. The subject had to hold the tool
with his right hand, keeping the bar on top horizontal (a). Visual feedback on the
position of the tool was given by a horizontal cord. Flexion moments were enforced
by hanging weights right under the stick while extension moments were enforced
by loads applied to the middle of the bar using a pulley system. Pro/supination
moments were imposed by hanging weights on different distances left or right from
the stick (b).
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003ment combinations protocol was repeated at four different elbow
angles: 50, 70, 90 and 110 of flexion (where 0 is full elbow
extension), leading to a total of 196 trials per subject. The correct
elbow angle was checked using a goniometer. Between each elbow
angle force recording series, the subjects had a rest period of
20 min. To minimize the effect of fatigue, the order of the elbow
angles was randomly defined. The order of the elbow angles was
randomly defined. Since the forearm stayed horizontal, a decrease
of flexion angle in the elbow led to an increase of anteflexion angle
of the glenohumeral joint. For each subject, all measurements were
performed on one single day. The 3D co-ordinates of bony land-
marks on the thorax, clavicle, scapula, humerus and forearm, were
recorded for each elbow angle, using a 3D digitizer (Veeger et al.,
1993). A head rest together with a pointer at the angulusacromialis
and the epicondyluslateralis were used to ensure the subject
stayed in the same position.
During experiment II, the periods of force production and espe-
cially the periods of rest between sessions had to be longer due to
the NIRS measurements. The periods of force production varied be-
tween 20 and 30 s and were followed by a recovery period varying
from 3 to 5 min depending on the force level. The arm of the sub-
ject was resting during the recovery period. A selection of the mo-
ment combinations measured in the first experiment was used,
Fig. 2. Schematic overview of all different moment combinations the subjects had
to perform. All loads could be applied at three different force levels (level 1, 2 and 3)
leading to a total of 49 combinations. In the first experiment, this complete set of
moment combinations was measured at one single day. During the second
experiment, a selection of these conditions, represented by the dark grey cells,
was measured, divided over three different days (boxes A, B and C). TL was
measured during all three sets, BB was measured during set A and B and BR was
measured during set A only.phy and Kinesiology xxx (2010) xxx–xxx 3which was divided into three sets (A, B and C, respectively, 15, 9
and 12 trials, see Fig. 2) measured on three separate days. TL was
measured during all three sets, BB during sets A and B and BR dur-
ing set A only. Since with the NIRS device only two muscles could
be measured at once, set A was performed twice (on two separate
days), once for BB and BR and once for TL. Each day the relative set
was repeated four times, once for each elbow angle (50, 70, 90
and 100 of flexion). In total each subject had to perform 204 trials.
The order of the elbow angles was randomly defined. EMG of all
these muscles was measured on all 4 days. The position of the sub-
ject was monitored at 25 Hz by recording the 3D co-ordinates of
bony landmarks on the thorax, clavicle, scapula, humerus and fore-
arm, using an automated video based recording system (Opto-
trakTM, Northern Digital Inc., Canada). During the experiment,
the actual elbow angle was controlled on-line using markers on
top of the acromion, the epicondyluslateralis and the processussty-
loideus ulnae.
2.3. Data processing
Orientations of the body segments of the subjects were calcu-
lated from the measured 3D co-ordinates of the bony landmarks
e and external moment on load sharing of elbow muscles. J Electromyogr

diately after occlusion. The slope of the decrease was taken as the
_ 1 _
[0.05] of the R . Next, angle specific moment variables were de-
fined and it was investigated whether a substantial increase could
ogra
ARTICLE IN PRESSbe achieved including these variables in our model. In fact this
means that interaction effects between angle and the external mo-
ments were included in the considerations.
The analyses were carried out for the data of experiment I. Sub-
sequently, the resulting model was used to predict EMG results for
the second experiment and correlations between predicted and
measured EMG were determined.
Concerning the relationship between _VO2 and EMG, a stepwise
regression was performed with _VO2 ð75Þ as dependent and EMG(75)
and EMG2ð75Þ as independent variables. To investigate whether el-
bow angle significantly influences the relationship between EMG
and _VO2 a second regression model was used in which EMG(75)
was redefined into four elbow angle categories (115–100–80–
60). The subdivision in elbow angles was accepted when R2 im-
proved with more than 0.05.
3. Results
3.1. Position dataVO2 (micromoles O2s ) of the muscle. VO2 was corrected for rest
metabolism by subtracting the _VO2 measured during rest from the
_VO2 measured during force production.
The measurements were standardised for each subject and each
muscle. To prevent the detrimental effect of outliers, we standard-
ised our results relative to the 75% percentile and not to the max-
imum value obtained during the experiment. So with EMGmean the
measured value and EMGmean(75) the 75% percentile of all of the
EMGmean measurements for that subject and that muscle, our
standardised data became
EMGð75Þ ¼ 0:75  EMGmean=EMGmeanð75Þ ð1Þ
and similar
_VO2ð75Þ ¼ 0:75  _VO2mean= _VO2meanð75Þ ð2Þ
2.4. Statistics
A multiple regression model was used to test to which extent
the EMG amplitude was dependent on external moment and elbow
angle. Since muscle force can only be positive, or zero, four mo-
ment variables were defined: flexion (Mf), extension (Me), prona-
tion (Mp) and supination (Ms). If for a particular condition the
moment direction was not requested it was set to zero. In other
cases, it was set to the – coded – moment level (1–2–3, for flex-
ion/extension force level 5, 10 and 15 Nm and pro/supination mo-
ments 1, 2 and 3 Nm).
With the natural logarithm of EMG(75) as the dependent vari-
able, and these external moments as independent variables, step-
wise regression was used. (The logarithmic transformation was
suggested by the data: residuals after regression turned out to be
rather skewed if the original data were used. After this transforma-
tion, they show a normal distribution.) External moment variables
were included into the model if they led to a substantial increase
2following the protocol described in van der Helm (1997). Mean
EMG (EMGmean) values were calculated from rectified EMG over
the period of force production. Muscle _VO2 was determined by per-
forming regression on the linear part of the [O2HB] decrease imme-
4 M. Praagman et al. / Journal of ElectromyThe actual elbow angles differed somewhat from the imposed
elbow angles: 55, 80, 100 and 120 for experiment I and 60,
80, 100 and 115 for experiment II, where 0 is full extension.
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.0033.2. EMG data (experiment I)
EMG amplitude of all four flexor muscles increased with flexion
moment (Fig. 3). Regression analysis (Table 1) showed that this
influence of flexion moment depended on the elbow angle: the
EMG level increased with decreasing flexion angle. For both heads
of biceps brachii (BB and BL) the EMG level also increased with
supination moment whereas the EMG signal of BR increased with
pronation moment (Table 1 and Fig. 3). In contrast to expectations,
the EMG signal of the mono-articular elbow flexor BA was also
influenced by supination moment as well as extension moment
(Table 1).
As expected, for all extensor muscles a linear relationship with
extension moment was found (Fig. 4). Regression showed that this
relationship was not influenced by elbow angle except for AC: the
EMG level decreased with decreasing flexion angle (Table 1). There
was also a significant influence of supination moment found for all
four extensor muscles. It was also shown that activity of AC and TL
was influenced by pronation moment and that the EMG signal of
TL was linearly related to flexion moment. As was found for the el-
bow flexors, the effect of flexion moment was influenced by elbow
angle: EMG amplitude of TL increased with decreasing elbow
angle.
3.3. EMG data (experiment II)
Application of the regression model obtained in experiment I
produced a good prediction of EMG amplitude for the EMG data
of experiment II (R2 of 0.72 and 0.68 and 0.55 for, respectively,
BB, BR and TL).
3.4. Residual analysis
Analysis of variance showed that 25–46% of the residual sum of
squares could be explained by the effect of subject and subject-mo-
ment (such as FL, FP, FS) interaction. More specific analysis of the
residuals showed that for some of the moment conditions residuals
were indeed relatively large compared to other conditions and that
a large part of these residuals could be explained by subject vari-
ance. This was especially the case for the data of TL during FL in
experiment I and for BB during ES in experiment II.
3.5. Load sharing (experiment I)
Relative EMG contributions between BB and BR changed over
the different moment combinations (Fig. 5), which indicated a
change in load sharing. As was seen before in Fig. 4, supination
led to an increase of BB and pronation led to an increase of BR. Sec-
ond, relative EMG amplitude (i.e., load sharing) was also influenced
by elbow angle, which was in turn different for the different mo-
ment combinations: during FP the EMG amplitude of BR was larger
than that of BB, though the relative contribution of BB increased
with elbow extension. For the extensors TR and TL, load sharing
was only influenced by moment and not by elbow angle (Fig. 6).
During extension tasks the contribution of TR and TL was compa-
rable, whereas during flexion tasks some activity of the mono-
articular TL was found, but not, or hardly for the bi-articular TR. El-
bow angle has no influence on the distribution between the two
muscles (Fig. 6).
Looking at the load sharing between flexor (BB) and extensor
(TL) muscles (Fig. 7) it was seen that during the supination tasks
(ES and FS) this load sharing changed over elbow angle. As ex-
pected, TL was especially active during EP and ES. During EP the
phy and Kinesiology xxx (2010) xxx–xxxactivity of TL decreased with elbow extension and there was no
activity of BB. During ES on the other hand, there was a rather large
contribution of BB activity, which increased with elbow extension.
e and external moment on load sharing of elbow muscles. J Electromyogr

ogra
ARTICLE IN PRESSBB BL
M. Praagman et al. / Journal of ElectromyAs seen before TL was also active during FP and FS. During FP, the
contribution of TL was even higher than that of BB. During FS
the contribution of BB was much larger than that of TL, however,
the contribution of TL increased with elbow extension.
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
E
M
G
(7
5)
1
20
°
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
E
M
G
(7
5)
1
00
°
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
E
M
G
(7
5)
8
0°
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
flexion
pronation
E
M
G
(7
5)
5
5°
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
flexion
pronation
Fig. 3. Normalised EMG values of the four measured flexor muscles (experiment I) averag
for each muscle and each elbow angle.
Table 1
Results of stepwise regression. Regression coefficients and R2 for the relationship betwe
extension and pro/supination) and elbow angle. External moment variables were included
moment, angle had a significant contribution (increase of the R2P 0.05), angle specific va
R2 Constant SU120 SU100 SU80 SU55 PR112 PR100 PR8
BB .73 2.5 .76
BL .72 2.4 .66
BR .73 1.7 .31
BA .65 1.9 .49
TR .72 1.9 .29
TL .64 1.8 .28 .24
TM .73 2.3 .37
AC .62 1.7 .41 .41
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003BA BR
phy and Kinesiology xxx (2010) xxx–xxx 53.6. Relationship between _VO2 and EMG (experiment II)
Corresponding to previous results (Praagman et al., 2003) it was
found that the relation between EMG and _VO2 could be described
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
−3−2−1 0 1 2 3
3210−1−2−3
0
1
2
flexion
pronation
−3−2
−10
1 2
33 2 1 0−1−2−3
0
1
2
−3−2
−10
1 2
33 2 1 0−1−2−3
0
1
2
−3−2
−10
1 2
33 2 1 0−1−2−3
0
1
2
−3−2
−10
1 2
33 2 1 0−1−2−3
0
1
2
pronationflexion
ed over subjects (n = 6) plotted against the flexion moments and pronation moment
en the natural logarithm of EMG(75) (experiment I) and external moment (flexion/
into the model if they led to a substantial increase (0.05) of the R2. If, for a particular
riables were defined for this moment. (See text for further explanation.)
0 PR55 EX120 EX100 EX80 EX55 FL120 FL100 FL80 FL55
.41 .48 .62 .79
.46 .59 .67 .82
.52 .53 .58 .66
.26 .28 .34 .45 .61
.68
.44 .08 .15 .30 .48
.80 .24
.39 .34 .26 .17
e and external moment on load sharing of elbow muscles. J Electromyogr

233 233
ogra
ARTICLE IN PRESS0
1
2
E
M
G
(7
5)
1
00
°
0
1
2−3−2−10121
0−1
−2−3
0
1
2
TR
E
M
G
(7
5)
1
20
°
−3−2−10121
0−1
−2−3
0
1
2
TL
6 M. Praagman et al. / Journal of Electromyby a linear relationship. For BB and BR, the model only marginally
improved when a quadratic term was added.
For BB and TL, the linear relationship was not influenced by
elbow angle (Fig. 8). For both muscles, angle specific variables
did not give an improvement of fit (Table 2). For BR angle specific
variables increased the R2 from 0.63 to 0.71.
4. Discussion
Inverse dynamic models use cost functions in order to predict
the individual contribution of muscles to a particular external mo-
ment. The exact principles behind this load sharing are (yet) un-
known, making it difficult to find the right cost function. In the
current study, data on muscle activation of elbow muscles was col-
lected which can be used for model cost function validation. Not all
of the functioning of elbow muscles is yet understood. For studying
elbow muscles it is necessary to collect experimental data that in-
cludes sufficiently detailed information to account for the effect of
combination of flexion/extension (FE) and pro/supination (PS) mo-
−3−2−101233
21
0−1
−2−3
−3−2−101233
21
0−1
−2−3
0
1
2
E
M
G
(7
5)
8
0°
−3−2−101233
21
0−1
−2−3
0
1
2
pronationflexion
E
M
G
(7
5)
5
5°
−3−2−101233
21
0−1
−2−3
−3−2−101233
21
0−1
−2−3
0
1
2
−3−2−101233
21
0−1
−2−3
0
1
2
pronationflexion
Fig. 4. Normalised EMG values of the four measured extensor muscles (experiment I) av
moment for each muscle and each elbow angle.
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003−3−2−101233
21
0−1
−2−3
0
1
2
TM
0
1
2
−3−2−101233
21
0−1
−2−3
0
1
2
AC
0
1
2
phy and Kinesiology xxx (2010) xxx–xxxments as well as for the effect of elbow angle. In addition, these
data should be sufficiently accurate to be used in an inverse dy-
namic model, such that muscle moment arm and length can be ac-
counted for as well.
Previous studies on elbow function often focussed on flexion–
extension (FE) tasks only, neglecting or not controlling the effect
of pro/supination (PS) (Soechting and Lacquaniti, 1988; Leedham
and Dowling, 1995; van Bolhuis and Gielen, 1997; Kasprisin and
Grabiner, 2000). Investigating the function of m. biceps brachii ca-
put breve, the effect around the shoulder is often accounted for
(Soechting and Lacquaniti, 1988; van Bolhuis and Gielen, 1997),
indicating that shoulder moment plays an important role as well,
whereas, unfortunately, the supination moment is neglected. As
can be learned from studies on compensatory motion in spastic
children (Kreulen et al., 2006), it is, however, evident that the supi-
nation moment of m. biceps brachii is also of large significance and
interacts strongly with its elbow flexion function.
The results of elbow studies are not unambiguous. Differences
could possibly be due to inadequate control of degrees of freedom.
Studies in which pro/supination was included (Buchanan et al.,
−3−2−101233
21
0−1
−2−3
−3−2−101233
21
0−1
−2−3
0
1
2
−3−2−101233
21
0−1
−2−3
0
1
2
pronationflexion
−3−2−101233
21
0−1
−2−3
−3−2−101233
21
0−1
−2−3
0
1
2
−3−2−101233
21
0−1
−2−3
0
1
2
pronationflexion
eraged over the subjects (n = 6) plotted against the flexion moments and pronation
e and external moment on load sharing of elbow muscles. J Electromyogr

33
ogra
ARTICLE IN PRESS0 1 2 3
0
1
2
3
120°
FS
0 1 2 3
0
1
2
3
FP
2
3
S
0 1 2
0
1
2
3
100°
0 1 2
0
1
2
3
2
3
G
(7
5)
B
ra
ch
io
ra
di
al
is
FS
FP
SG
(7
5)
B
ra
ch
io
ra
di
al
is
M. Praagman et al. / Journal of Electromy1986, 1989; van Zuylen et al., 1988; Caldwell and van Leemputte,
1991; Hebert et al., 1991; de Serres et al., 1992; Jamison and Cald-
well, 1993, 1994; Bechtel and Caldwell, 1994), however, also do
not show explicit muscle activation patterns. Experimental results
show:
1. Load sharing during elbow/flexion is (of course) influenced by
the presence or absence of (external) pro- or supination
moment and vice versa (Buchanan et al., 1989; Hebert et al.,
1991; de Serres et al., 1992; Jamison and Caldwell, 1993,
1994; Bechtel and Caldwell, 1994) and also by elbow angle
(van Zuylen et al., 1988; de Serres et al., 1992),
2. Muscle activation patterns seem to be task dependent as well as
subject specific (Buchanan et al., 1989; Bechtel and Caldwell,
1994; Buchanan and Lloyd, 1995),
0 1 2 3
0
1E
0 1 2 3
0
1
2
3
E
P
0 1 2 3
0
1
0 1 2 3
0
1
2
3
EMG
(75)
B
E
M E
E
P
E
M
Fig. 5. Load sharing. EMG(75) values of m. brachioradialis (BR) plotted against m. biceps c
80 and 55 of flexion) and the rows show the results of the four different moment com
0 1 2 3
0
1
2
3
120°
FS
0 1 2 3
0
1
2
3
FP
0 1 2 3
0
1
2
3
E
S
0 1 2 3
0
1
2
3
E
P
0 1 2 3
0
1
2
3
100°
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
EMG
(75)
Tric
E
M
G
(7
5)
T
ric
ep
s
ca
pu
t l
at
er
al
e
Fig. 6. Load sharing. EMG(75) values of the m. triceps caput laterale (TL) plotted against
angles (120, 100, 80 and 55 of flexion) and the rows show the results of the four dif
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.0030 1 2 3
0
1
2
3
80°
0 1 2 3
0
1
2
3
2
3
0 1 2 3
0
1
2
3
55°
0 1 2 3
0
1
2
3
2
3
phy and Kinesiology xxx (2010) xxx–xxx 73. Muscles sometimes show ‘seemingly inappropriate muscle
actions’ (Buchanan et al., 1989), presumably meant to compen-
sate for unwanted additional muscle moments of other muscles
(van Zuylen et al., 1988; Buchanan et al., 1989; Jamison and
Caldwell, 1993).
The question remains how load sharing is organised and
whether the differences in results presented in the literature are
task specific, or subject specific, or both.
In the current study, an extensive data set on activation of el-
bow muscles, including variations in elbow angle and a wide range
of moment combinations as well as different force levels, was col-
lected. The use of EMG and NIRS allowed for the measurement of
indicators for both muscle activation and muscle energy consump-
tion. Since the local _VO2 measurements were extremely time
iceps caput breve
0 1 2 3
0
1
0 1 2 3
0
1
2
3
0 1 2 3
0
1
0 1 2 3
0
1
2
3
aput breve (BB). The columns represent the four different elbow angles (120, 100,
binations (FP, FS, ES and EP).
eps caput longum
0 1 2 3
0
1
2
3
80°
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
55°
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
the m. triceps caput longum (TR). The columns represent the four different elbow
ferent moment combinations (FP, FS, ES and EP).
e and external moment on load sharing of elbow muscles. J Electromyogr

3ogra
ARTICLE IN PRESS0 1 2 3
0
1
2
3
120°
FS
1
2
3
FP
0 1 2
0
1
2
3
100°
1
2
3
la
te
ra
le
8 M. Praagman et al. / Journal of Electromyconsuming, these measurements had to be restricted to a limited
set of moment combinations as well as to a limited number of
muscles and subjects. Therefore, another experiment was per-
formed in which only EMG was measured enabling measurements
of a larger number of muscles and subjects and a complete set of
external moments.
4.1. Influence of external moment
The aim of imposing a broad set of combinations of elbow flex-
ion–extension and forearm pro/supination was to quantify the ef-
0 1 2 3
0
0 1 2 3
0
1
2
3
E
S
0 1 2 3
0
1
2
3
E
P
0 1 2 3
0
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
EMG
(75)
Bice
E
M
G
(7
5)
T
ric
ep
s
ca
pu
t
Fig. 7. Load sharing. EMG(75) values of the m. triceps caput laterale (TL) plotted against t
different elbow angles (120, 100, 80 and 55 of flexion) and the rows show the resul
0 2 4
0
1
2
3
4
Biceps caput breve
V
O
2(
75
)
EMG
(75)
0 1
0
0.5
1
1.5
2
Brachioradi
EMG
(75)
Fig. 8. Normalised _VO2 values plotted against normalised EMG values for the m. biceps
Table 2
Regression equations and R2 for the relationship between _VO2 ð75Þ and EMG(75), defined by:
_VO2 ð75Þ = constant + b1  EMG(75)115 + b2  EMG(75)110 + b3  EMG(75)80 + b4  EMG(75)60.
Muscle R2 Constant EMG(75)115
BB 0.83 0.18
0.85 0.11
0.83 0.18 0.64
BR 0.63 0.06
0.64 0.07
0.71 0.05 0.74
TL 0.60 0.008
0.60 0.001
0.62 0.002 1.03
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.0030 1 2 3
0
1
2
3
80°
1
2
3
0 1 2 3
0
1
2
3
55°
1
2
3
phy and Kinesiology xxx (2010) xxx–xxxfect of external load and elbow angle on the relative contribution of
muscles (at varying lengths). Results of the current study showed,
that muscle activity (EMG) was, as expected, strongly influenced
by external moment.
Although the relationship between EMG and external moment
often can be satisfactorily described by a linear relationship (Praag-
man et al., 2003), the EMG data in the current study seemed to be
better described by a logarithmic relationship, in the sense that
residuals showed a better normal distribution.
The major part of the recorded EMG activity could be explained
mechanically, though in different ways. First, muscles are directly
ps caput breve
0 1 2 3
0
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
0 1 2 3
0
0 1 2 3
0
1
2
3
0 1 2 3
0
1
2
3
he EMG75 values of the m. biceps caput breve (BB). The columns represent the four
ts of the four different moment combinations (FP, FS, ES and EP).
2
alis
0 2 4
0
1
2
3
4
Triceps caput laterale
EMG
(75)
60°
80°
100°
115°
caput breve (BB), the m. brachioradialis (BR) and the m. triceps caput laterale (TL).
_VO2 ð75Þ = constant + b  EMG(75) or _VO2 ð75Þ = constant + b1  EMG(75) + b2  EMG2ð75Þ or
EMG(75)100 EMG(75)80 EMG(75)60 EMG2ð75Þ
0.65
0.90 0.12
0.67 0.71 0.62
0.88
1.32 0.32
0.85 0.93 1.05
0.92
0.94 0.02
0.93 0.86 0.75
e and external moment on load sharing of elbow muscles. J Electromyogr

ogra
ARTICLE IN PRESSinfluenced by the in- or decrease of a particular external moment if
that muscle can actually contribute to that particular moment (e.g.,
the increase of biceps activity with increasing supination). Second,
it can also be influenced indirectly by the activity of another mus-
cle that contributes to the particular moment (synergistic- as well
as antagonistic load sharing). If a muscle produces moments
around more than one joint axis, a change in activity due to a
change in the moment required on the one axis automatically leads
to a change in the generated moment around the other axis as well.
The latter leads to an in- or decrease of one (or more) of the syner-
gists for the second axis (synergistic load sharing). However, if the
moment generated around the second axis is not wanted at all, this
moment needs to be compensated by an antagonistic muscle
(antagonistic load sharing) (Veeger and van der Helm, 2007).
As expected, activity of all flexor muscles increased with
increasing flexion moment. Since BB and BL generate not only a
flexion- but also a supination moment, it is not surprising that
the activity during flexion further increased with increasing supi-
nation moment and decreased with increasing pronation moment.
The latter influences the contribution of other elbow flexors such
as BR as well: due to the decrease of biceps activity during flex-
ion-pronation an increase of BR can be seen, since the flexion mo-
ment that had to be generated remained the same (synergistic load
sharing). In spite of the fact that extensor muscles cannot produce
any pro- or supination moment, their activity was influenced by
supination moment. There was an increase in activity with increas-
ing supination moment. This could be explained by a compensa-
tion for the flexion moment generated by BB and BL which is
generated when these muscles contribute to the required supina-
tion moment. If this additional flexion moment is either unwanted
(i.e., during extension-supination or pure supination) or larger than
the required flexion moment, a compensating extension moment is
needed. The latter can be generated by the m. triceps brachii,
resulting in co-contraction (antagonistic load sharing).
A small part of the experimental data could not be explained by
one of the above mentioned load sharing principles. During exper-
iment I, co-contraction of TL was also seen during flexion. The
activity of TL was not only influenced by the extension moment
and supination moment but also by the flexion moment (Table
1). This type of activity cannot be explained by the load sharing
principles mentioned above. TL does not contribute to the re-
quested flexion moment and there is no side-effect that needs to
be compensated for. It is possible that this type of co-contraction
was caused by stability requirements of the task. The activity of
an antagonist can be important for maintaining the integrity of
the joint (Hirokawa et al., 1991; Gabriel et al., 2006). We also found
that there were large differences in the amount of co-contraction
between the different subjects (residual analysis), suggesting that
strategies leading to co-contraction are quite subject-specific in-
stead of general. This finding is not unique. In a study on the activ-
ity of back muscles, van Dieën (1996) showed that 25% of the
subjects showed a diverged pattern which was due to a higher le-
vel of co-contraction of the abdominal muscles. Baratta et al.
(1988) showed that co-contraction is reduced in specialised ath-
letes compared to normal subjects.
The differences found in co-contraction between the subjects in
our study might be caused by the fact that some subjects paid
more attention to the instruction keeping the bar horizontal than
others, introducing more effort for stability.
This difference in co-contraction due to differences in task
interpretation and performance will not be predicted by minimisa-
tion of energy or any other cost for that matter. For inverse
dynamic biomechanical modelling this means that stability con-
M. Praagman et al. / Journal of Electromystraints should be included in the model to enforce co-contraction,
though a correct stability analysis also requires inclusion of the
effect of muscle proprioceptive feedback.
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.0034.2. Influence of elbow angle
Elbow angle had a significant effect on EMG level and relative
muscle activation. Variation in elbow angle (from 55 to 120 of el-
bow flexion), changed both muscle lengths and moment arms of
the elbow muscles, which both can influence relative and absolute
muscle activation. It is well known that when a muscle has to pro-
duce a given force at a length shorter than optimum length, the
activation needed to generate this force increases due to the in-
creased overlap of sarcomeres. It is also obvious that the force that
a muscle has to generate to produce a given moment is inversely
related to its moment arm. Both factors might influence the rela-
tive muscle activity. It seems reasonable to expect that if a muscles’
length becomes sub-optimal or its moment arm decreases, the en-
ergy consumption increases since more activation/force is needed
to produce the same force/moment. However, this is seen from a
single muscle perspective, assuming that the power produced by
the muscle stays the same and that load sharing would not change
over the different elbow angles. Taking a multiple muscle system
as starting point, a more optimal length or larger moment arm will
make a muscle ‘cheaper’ or more economic, which could possibly
lead to a change in force sharing: an increase in activity of the par-
ticular muscle and a simultaneous decrease of activity of one (or
more) of the synergists.
The results of the current study showed that elbow angle in-
deed influenced the activity of some muscles (Table 1). But results
cannot be easily related to the actual changes in muscle length or
moment arm, as both can vary in a different way. A distinct opti-
mum (highest or lowest EMG) as was expected at angles close to
optimal length or maximum moment arm could not be found.
For instance, increase of BR activity with decreasing elbow angle
could be related to a decreasing moment arm. The maximal mo-
ment arm is found between 100 and 115 (Murray et al., 1995;
Veeger et al., 1997). Length, however, becomes more favourable
with decreasing elbow angle at first and has its optimum around
80. For BB both optimum length and moment arm are found
around 80 of elbow flexion. Nevertheless, results of the current
study show that BB activity during flexion increased with decreas-
ing flexion angle (Table 1), and no optimum (maximum or mini-
mum) was found at 80. It has to be noticed that as the
anteflexion angle of the glenohumeral joint increased with
decreasing elbow angle, the increase in BB activity might as well
be attributed to the increasing anteflexion angle. Further by influ-
encing the activity of one muscle, a change in moment arm or
length likely indirectly influences its synergists as well.
In line with previous findings (de Serres et al., 1992), our results
show that load sharing changes over elbow angle (Figs. 5 and 7).
However, it cannot be derived from our data whether the contribu-
tion of a muscle increases or decreases if it becomes more favour-
able, since it is impossible to distinguish all the individual aspects
that affected the muscle activity. To unravel the exact effects a
musculoskeletal model would be needed, in order to gain better in-
sight into the specific effects of the different factors that influence
load sharing.
4.3. Subject specific patterns
Although overall results were quite comparable between
subjects, for some specific conditions there seemed to be large
differences between subjects in activation patterns of particular
muscles. These inter-individual differences were found for pure
antagonistic muscle activity (TL during FL) as well as for antago-
nistic load sharing (BB during ES). As mentioned before, this find-
phy and Kinesiology xxx (2010) xxx–xxx 9ing was in line with previous studies (Buchanan et al., 1989;
Bechtel and Caldwell, 1994). If it is assumed that people use the
same load sharing principle and use the same task interpretation,
e and external moment on load sharing of elbow muscles. J Electromyogr

ment was found by adding a quadratic term (Table 2). Introducing
elbow-specific EMG variables also did not lead to a better fit, ex-
The relationship between EMG and VO2 could be described by a
linear relationship, it is yet unknown whether all processes lying in
ogra
ARTICLE IN PRESSbetween are linear processes as well.
Although, on average, subjects show comparable muscle activa-
tion patterns, during conditions in which more than one moment
was required and the particular muscle counteracted to one of
these moments, there are some striking inter-individual differ-
ences as well. These inter-individual differences might be the con-cept for BR (Table 2). Based on morphological data (Klein Breteler
et al., 1999) it can be concluded that within the range of elbow an-
gles (60–115 of elbow flexion) used, the corresponding change in
muscle fibre length was by far the largest for BR (almost 60% com-
pared to about 30% for BB and TL). It is, therefore, possible that the
relationship between _VO2 and EMG is influenced by change in
muscle fibre length but only at major length changes.
In musculoskeletal modelling it was shown that cost functions
linearly related to muscle force (i.e., minimisation of sum of muscle
forces or sum of muscle stress) resulted in sequential muscle
recruitment (Dul et al., 1984). Using linear cost functions it is most
advantageous to use the muscle with the largest moment arm to
produce the required muscle moment, since the least muscle force
is required. The linear relationship between _VO2 and EMG suggests
that muscle activation and muscle energy consumption are linearly
related. Since muscle force is linearly related to energy consump-
tion as well, this would mean that minimisation of energy con-
sumption (which is in general expected to be the principle
behind load sharing) would lead to sequential recruitment.From
experimental evidence it is clear that load sharing does not lead
to sequential recruitment. It can, therefore, be argued that energy
consumption is indeed minimised or that the linear relationship
between _VO2 and EMG is based upon two opposing non-linear
relationships; EMG-activation and activation- _VO2, leading to a lin-
ear relationship between EMG and _VO2 .
5. Conclusions
Joint angle, and, therefore, moment arm and muscle length
influence both the activation level of the muscle as well as the load
sharing between muscles.
The principles behind load sharing, however, are difficult to
quantify, since it is impossible to distinguish all the individual as-
pects that affect muscle activity. To solve this complex problem a
biomechanical model is needed at least.
_such individual muscle activation patterns could possibly be due
to a high sensitivity to inter-individual differences in mor-
phology. It is, however, known that training can reduce antagonist
co-contraction (Baratta et al., 1988), which can be seen as a signal
that the assumption of a general load sharing principle is too
simple.
4.4. Relationship between EMG and _VO2
Another research question of this study was whether the linear
relationship between _VO2 and EMG as was found earlier (Praag-
man et al., 2003) for a constant elbow angle would also hold for dif-
ferent elbow angles and a larger range of moment combinations.
The current study confirmed our earlier finding (Praagman
et al., 2003) that the relationship between _VO2 and EMG can be sat-
isfactorily described by a linear equation. No substantial improve-
10 M. Praagman et al. / Journal of Electromysequence of two different factors: (1) subjects use different
optimisation strategies or (2) they might reflect the role of inter-
individual differences in morphology.
Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003Acknowledgements
The authors would like to thank J. Praagman for his advice on
the statistical analyses.
References
Alexander RM. A minimum energy cost hypothesis for human arm trajectories. Biol
Cybern 1997;76(2):97–105.
Baratta R, Solomonow M, Zhou BH, Letson D, Chuinard R, D’Ambrosia R. Muscular
coactivation. The role of the antagonist musculature in maintaining knee
stability. Am J Sports Med 1988;16(2):113–22.
Bechtel R, Caldwell GE. The influence of task and angle on torque production and
muscle activity at the elbow. J Electromyogr Kinesiol 1994;4(4):195.
Buchanan TS, Almdale DP, Lewis JL, Rymer WZ. Characteristics of synergic relations
during isometric contractions of human elbow muscles. J Neurophysiol
1986;56(5):1225–41.
Buchanan TS, Rovai GP, Rymer WZ. Strategies for muscle activation during isometric
torque generation at the human elbow. J Neurophysiol 1989;62(6):1201–12.
Buchanan TS, Lloyd DG. Muscle activity is different for humans performing static
tasks which require force control and position control. Neurosci Lett
1995;194(1–2):61–4.
Buchanan TS, Shreeve DA. An evaluation of optimization techniques for the
prediction of muscle activation patterns during isometric tasks. J Biomech Eng
1996;118(4):565–74.
Caldwell GE, van Leemputte M. Elbow torques and EMG patterns of flexor muscles
during different isometric tasks. Electromyogr Clin Neurophysiol
1991;31(7):433–45.
Challis JH. Producing physiologically realistic individual muscle force estimations
by imposing constraints when using optimization techniques. Med Eng Phys
1997;19(3):253–61.
De Serres SJ, Hebert LJ, Arsenault AB, Goulet C. Effect of pronation and supination
tasks on elbow flexor muscles. J Electromyogr Kinesiol 1992;2(1):53.
Dul J, Townsend MA, Shiavi R, Johnson GE. Muscular synergism–I. On criteria for
load sharing between synergistic muscles. J Biomech 1984;17(9):663–73.
Gabriel DA, Kamen G, Frost G. Neural adaptations to resistive exercise: mechanics
and recommendations for training practices. Sports Med 2006;36:133–49.
Happee R. Inverse dynamic optimization including muscular dynamics, a new
simulation method applied to goal directed movements. J Biomech
1994;27(7):953.
Hardt DE. Determining muscle forces in the leg during normal human walking: an
application and evaluation of optimization methods. J Biomech Eng
1978;100:72–8.
Hebert LJ, De Serres SJ, Arsenault AB. Cocontraction of the elbow muscles during
combined tasks of pronation-flexion and supination-flexion. Electromyogr Clin
Neurophysiol 1991;31(8):483–8.
Hirokawa S, Solomonow M, Luo Z. Muscular co-contraction and control of knee
stability. J Electromyogr Kinesiol 1991;1:199–208.
Jamison JC, Caldwell GE. Muscle synergies and isometric torque production:
influence of supination and pronation level on elbow flexion. J Neurophysiol
1993;70(3):947–60.
Jamison JC, Caldwell GE. Dual degree of freedom tasks: flexion effect on supination/
pronation response. J Electromyogr Kinesiol 1994;4(3):143.
Jinha A, Ait-Haddou R, Binding P, Herzog W. Antagonistic activity of one-joint
muscles in three-dimensions using non-linear optimisation. Math Biosci
2006;202(1):57–70.
Kasprisin JE, Grabiner MD. Joint angle-dependence of elbow flexor activation levels
during isometric and isokinetic maximum voluntary contractions. Clin Biomech
(Bristol, Avon) 2000;15(10):743–9.
Kaufman KR, An KN, Litchy WJ, Chao EY. Physiological prediction of muscle forces–
II. Application to isokinetic exercise. Neuroscience 1991a;40(3):793–804.
Kaufman KR, An KW, Litchy WJ, Chao EY. Physiological prediction of muscle forces–
I. Theoretical formulation. Neuroscience 1991b;40(3):781–92.
Klein Breteler MD, Spoor CW, van der Helm FC. Measuring muscle and joint
geometry parameters of a shoulder for modeling purposes. J Biomech
1999;32(11):1191–7.
Kreulen M, Smeulders MJ, Veeger HE, Hage JJ. Movement patterns of the upper
extremity and trunk before and after corrective surgery of impaired forearm
rotation in patients with cerebral palsy. Dev Med Child Neurol
2006;48(6):436–41.
Leedham JS, Dowling JJ. Force-length, torque-angle and EMG-joint angle
relationships of the human in vivo biceps brachii. Eur J Appl Physiol
1995;70(5):421–6.
Murray WM, Delp SL, Buchanan TS. Variation of muscle moment arms with elbow
and forearm position. J Biomech 1995;28(5):513.
Praagman M, Veeger HEJ, Chadwick EKJ, Colier WNJM, van der Helm FCT. Muscle
oxygen consumption, determined by NIRS, in relation to external force and
EMG. J Biomech 2003;36(7):905–12.
Praagman M, Chadwick EKJ, van der Helm FCT, Veeger HEJ. The relationship
between two different mechanical cost functions and muscle oxygen
consumption. J Biomech 2006;39(4):758–65.
phy and Kinesiology xxx (2010) xxx–xxxRaikova RT, Prilutsky BI. Sensitivity of predicted muscle forces to parameters of the
optimization-based human leg model revealed by analytical and numerical
analyses. J Biomech 2001;34(10):1243.
e and external moment on load sharing of elbow muscles. J Electromyogr

Soechting JF, Lacquaniti F. Quantitative evaluation of the electromyographic
responses to multidirectional load perturbations of the human arm. J
Neurophysiol 1988;59(4):1296–313.
Tsirakos D, Baltzopoulos V, Bartlett R. Inverse optimization: functional and
physiological considerations related to the force-sharing problem. Crit Rev
Biomed Eng 1997;25(4–5):371–407.
van Bolhuis BM, Gielen CC. The relative activation of elbow-flexor muscles in
isometric flexion and in flexion/extension movements. J Biomech
1997;30(8):803–11.
van der Helm FCT. Analysis of the kinematic and dynamic behavior of the shoulder
mechanism. J Biomech 1994;27(5):527–50.
van der Helm FCT. A standardized protocol for motion recordings of the shoulder.
In: Veeger HEJ, van der Helm FCT, Rozing PM, editors. First Conference of the
International Shoulder Group, Delft University of Technology. The
Frans C. T. van der Helm is professor in Biomecha-
tronics and Bio-robotics (0.8 fte at Delft, and 0.2 fte at
the University of Twente, until 2007), and is chair of the
Department of Biomechanical Engineering at the TU
Delft. He has a M.Sc in Human Movement Science
(1985), and a Ph.D in Mechanical Engineering (1991). He
was member of the board of the International Society of
Biomechanics (ISB), the board of the Technical Group of
Computer Simulation (TGCS) and the International
Shoulder Group (ISG). He is Principal Investigator in the
TREND research consortium, which focus on patients
with Complex Regional Pain Syndrome as a neurological
disorder. He is heading the NeuroSIPE program (System
Identification and Parameter Estimation of Neurophysiological systems), funded by
the Netherlands Organisation of Scientific Research. He has published over 100
papers in international journals on topics as biomechanics of the upper and lower
extremity, neuromuscular control, eye biomechanics, pelvic floor biomechanics,
human motion control, posture stability, haptic control and humanoid robots.
H.E.J. Veeger after obtaining his masters in Human
Movement Sciences in 1984, he continued his studies in
London, where he obtained his M.Sc in Ergonomics from
University College London. Since 1986, he has been
affiliated to the Department of Human Movement Sci-
ences, where he received his Ph.D in 1992 on Biome-
chanics of Wheelchair Propulsion. His main research
interest lies in the field of musculoskeletal mechanics,
and especially the upper extremity. Most profound
research subject is the relationship between structure
and function, which has been applied to the shoulder in
wheelchair propulsion, the effect of surgical interven-
tions such as tendon transfers on muscle function and
ADL. He combines his appointment in Amsterdam with a part-time appointment in
M. Praagman et al. / Journal of Electromyography and Kinesiology xxx (2010) xxx–xxx 11
ARTICLE IN PRESSMarit Praagman studied Human Movement Sciences at
the Vrije Universiteit in Amsterdam, The Netherlands,
and got her M.Sc. degree in 1998. She obtained her Ph.D.
degree at the faculty of Human Movement Science, Vrije
Universiteit, The Netherlands, in 2008 on her thesis:
‘Muscle load sharing, an energy-based approach’.
Edward Chadwick received a B.Eng. degree in
mechanical engineering from The University of Not-
tingham, U.K., in 1993, and was awarded his Ph.D. in
bioengineering for work on biomechanical modelling of
the upper limb by the University of Strathclyde, U.K., in
1999. From there he went on to complete a four-year
post-doc in the area of shoulder biomechanics at the
Technical University of Delft in the Netherlands. From
2003-8 he was a Senior Research Associate in the Bio-
medical Engineering Department of Case Western
Reserve University in Cleveland, Ohio. He is currently a
lecturer in the Department of Sport and Exercise Science
at Aberystwyth University in the U.K. His research
interests are in biomechanical modelling of the shoulder and upper limb and the
restoration of function in people with neuromuscular deficit.van der Sluijs MC, Colier WNJM, Houston RJF, Oeseburg B. A new highly sensitive
continuous wave near-infrared spectrophotometer with multiple detectors.
SPIE 1998;3194:63–72.
van Dieën JH. Asymmetry of erector spinae muscle activity in twisted postures and
consistency of muscle activation patterns across subjects. Spine
1996;21(22):2651–61.
van Zuylen EJ, Gielen CC, Denier van der Gon JJ. Coordination and inhomogeneous
activation of human arm muscles during isometric torques. J Neurophysiol
1988;60(5):1523–48.
Veeger HEJ, Yu B, An KN, Rozendal RH. Parameters for modeling the upper
extremity. J Biomech 1997;30(6):647–52.
Veeger HEJ, van der Helm FCT. Shoulder function: the perfect compromise between
mobility and stability. J Biomech 2007;40(10):2119–29.
Veeger HEJ, van der Helm FCT, Rozendal RH. Orientation of the scapula in a
simulated wheelchair push. Clin Biomech 1993;8:81–90.Netherlands: Shaker Publishers BV; 1997. p. 7–12.Please cite this article in press as: Praagman M et al. The effect of elbow angl
Kinesiol (2010), doi:10.1016/j.jelekin.2010.04.003Delft since 2000. He now also serves as Full Professor at the Biomedical Engineering
Department of Delft University of Technology, where he is responsible for research
in the musculoskeletal mechanics, especially within the area of shoulder biome-
chanics. He is chair of the International Shoulder Group, a technical group of the
International Society of Biomechanics and member of the editorial board of Clinical
Biomechanics.e and external moment on load sharing of elbow muscles. J Electromyogr