ORIGINAL ARTICLE
Anastassios Philippou Æ Gregory C. Bogdanis
Alan M. Nevill Æ Maria Maridaki
Changes in the angle-force curve of human elbow flexors
following eccentric and isometric exercise
Accepted: 30 May 2004 / Published online: 4 August 2004
Springer-Verlag 2004
Abstract The aim of this study was to explore and
compare the magnitude and time-course of the shift in
the angle-force curves obtained from maximal voluntary
contractions of the elbow flexors, both before and 4
consecutive days after eccentric and isometric exercise.
The maximal isometric force of the elbow flexors of
fourteen young male volunteers was measured at five
different elbow angles between 50 and 160. Subjects
were then divided into two groups: the eccentric group
(ECC, n=7) and the isometric group (ISO, n=7). Sub-
jects in the ECC group performed 50 maximal voluntary
eccentric cont ractions of the elbow flexors on an isoki-
netic dynamometer (30
.
s
)1
), while subjects in the ISO
group performed 50 maximal voluntary isometric mus-
cle contractions with the elbow flexors at a length ened
position. Following the ECC and ISO exercise proto-
cols, maximal isometric force at the five angles, muscle
soreness, and the relaxed (RANG) and flexed (FANG)
elbow angles were measured at 24 h interv als for 4 days.
All results were presented as the mean and standard
error, and a quadratic curve was used to model the
maximal isometric force data obtained at the five elbow
angles. This approach not only allowed us to mathe-
matically describe the angle-force curves and estimate
the peak force and optimum angle for peak force gen-
eration, but also enabled us to statistically compare the
shift of the angle-force curves between and within
groups. A large and persistent shift of the angle-force
curve towards longer muscle lengths was observed 1 day
after eccentric exercise (P<0.01). This resulted in a 16
shift of the optimum angle for force generation, which
remained unchanged for the whole observation period.
A smaller but also persistent shift of the angle-force
curve was seen after isometric exercise at long muscle
length (P<0.05; shift in optimum angle 5). ECC
exercise caused more muscle damage than ISO exercise,
as indicated by the greater changes in RANG and rat-
ings of muscle soreness (P<0.05). It was suggeste d that
the shift in the angle-force curve was proportional to the
degree of muscle damage and may be explained by the
presence of overstretched sarcomeres that increased in
series compliance of the muscle.
Keywords Muscle damage Æ Force-length
relationship Æ Contractile function Æ Long muscle length
Introduction
Vigorous and unaccustomed exercise may lead to muscle
soreness due to structural disruption of myofibrils and
damage to the excitation–contraction coupling system
(Friden an d Lieber 1998; Proske and Morgan 2001).
Many studies have shown that the highest degree of
muscle damage occurs after eccentric exercise, where the
muscles are being lengthened while generating active
tension (Hunter and Faulkner 1997; Morgan and Allen
1999). Although the extent of the damage can be
quantified histologically (Friden and Lieber 1998), this is
not always practicable and, therefore, changes in the
mechanical properties of the muscle are commonly used
to assess damage. One widely used indirect measure of
muscle damage is the decrease in maximal voluntary
isometric force (MIF) which remains depressed for sev-
eral days following eccentric exercise (Clarkson et al.
1992; Cleak and Eston 1992).
In most of the studies that have examined the decline
in MIF after eccentric exercise, force was measured at a
single muscle length (Ingalls et al. 1998) or a single joint
angle (Jones et al. 1989; Clarkson et al. 1992; Nosaka
and Sakamoto 2001). However, there is evidence from
A. Philippou Æ G. C. Bogdanis (&) Æ M. Maridaki
Department of Sports Medicine and Biology of Physical Activity,
Faculty of Physical Education and Sports Science,
41 Ethnikis Antistasis Street, Dafni, 172 37 Athens, Greece
E-mail: gbogdanis@phed.uoa.gr
Tel.: +302-10-7276043
Fax: +302-10-9027840
A. M. Nevill
The School of Sport, Performing Arts and Leisure,
University of Wolverhampton, Walsall, WS1 3BD, UK
Eur J Appl Physiol (2004) 93: 237–244
DOI 10.1007/s00421-004-1209-z
both in vitro and in vivo studies that strength loss is
greater when force is measured at short versus optimal
or long muscle lengths (Wood et al. 1993; Saxton and
Donnelly 1996; Byrne et al. 2001; Proske and Morgan
2001). According to the ‘‘popping’’ sarcomere hypoth-
esis proposed by Morgan (1990), lengthening of active
muscle does not occur by uniform lengthening of all
sarcomeres, but by a non-unif orm distribution of sar-
comere length change, causing some weak sarcomeres to
over-extend (‘‘pop’’) beyond filament overlap. This
would increase the series compliance of the muscle,
leading to a shift of the length-tension (or angle-force)
curve to the right, i.e. towards longer muscle lengths,
following eccentric exercise (Morgan and Allen 1999;
Proske and Morgan 2001).
There is a growing body of evidence that the shift in the
angle-force curve is a more sensitive and more reliable
indicator of muscle damage, as compared to force mea-
surement at a single muscle length or joint angle (Talbot
and Morgan 1998; Brockett et al. 2001). This is because
the shift in the angle-force curve is not confounded by
fatigue and it also avoids the problem of uncertainty over
the optimum length for a contraction which occurs when
force is measured at the same joint angle before and after
damaging exercise (Proske and Morgan 2001). Further-
more, the magnitude of the shift seems to be proportional
to the degree of muscle damage (Wood et al. 1993; Jones
et al. 1997; Morgan and Allen 1999).
Although the phenomenon of the shift in the angle-
force curve towards longer muscle lengths following
muscle-damaging exercise is well established in single
fibres and motor units from animal muscle (Wood et al.
1993; Lynn and Morgan 1994; Lynn et al. 1998; Talbot
and Morgan 1998; Brockett et al. 2002), very few studies
have quantified this shift in humans (Jones et al. 1997;
Whitehead et al. 1998; Brockett et al. 2001). With the
exception of the study by Brockett et al. (2001), who
used intense, but not maximal, eccentric contractions of
the hamstrings (12 sets of 6 repetitions of ‘‘hamstring
lowers’’, i.e. lowering the upper body from a kneeling
position), all the other human studies have used much
lower exercise intensities to cause muscle damage (i.e.
walking backwards down an inclined treadmill for
1–2 h; Jones et al. 1997; Whitehead et al. 1998). There-
fore, the first aim of the present study was to provide
information about the magnitude and the time-course of
the possible shift in the angle-force curve of human elbow
flexors when the eccentric contractions are performed
with maximal voluntary effort. The second purpose of
this study was based on findings of a recent study by our
group (Philippou et al. 2003), where maximal isometric
exercise of the elbow flexors from a long muscle length
caused a large and sustained decrease in maximum iso-
metric force which was more pronounced at the more
acute elbow angles, indicating a possible shift of the
angle-force curve. Our protocol caused significantly
more muscle damage compared with isometric protocols
reported in previous studies, as indicated by the much
larger drop in force and by the several-fold greater
changes in indirect markers of muscle damage, such as
creatine kinase and relaxed and flexed elbow angle (e.g.
Jones et al. 1989; Nosaka et al. 200 2). Therefore, the
second aim of the present study was to examine if these
pronounced changes in contractile function caused by
maximal isom etric exercise at long muscle leng th are
accompanied by a shift of the angle-force curve, and to
compare the magnitude and duration of this shi ft with
that caused by maximal eccent ric exercise.
Methods
Subjects
Fourteen male volunteers [age 26.4 (1.6) years, height
175.5 (1.1) cm, mass 77.1 (2.8) kg] gave their informed
consent an d participated in this study, which was ap-
proved by the Athens University Ethics Committee. The
subjects were free of musculoskele tal disorders and had
not been involved in any type of resistance training for
at least 6 months before the study. Subjects were not
allowed to perform any vigorous physical activities
during the experimental period.
Preliminary measurements
The subjec ts were familiarized with the procedures of the
isometric force measurements of the elbow flexors dur-
ing at least two visits to the laboratory. All measure-
ments and testing protocols were performed with the
elbow flexors of the non-dominant arm. During the
preliminary measuremen ts (day 0), muscle shortening
ability and spontaneous muscle shortening were evalu-
ated first by measuring flexed (FANG) and relaxed
(RANG) elbow angle, respectively (Clarkson et al.
1992). Anatomical reference points were marked with
semi-permanent ink on the acromion, the epicondylus
lateralis of the humerus, the processi styloidei of the
radius and the point halfway between the processi sty-
loidei of the radius and ulna. A hand-held electronic
goniometer (Guymon, Lafayette Instruments, Ind.,
USA) was fixed on the arm using the anatomical refer-
ence points. FANG was defined as the elbow angle when
the subject tried to fully flex the forearm, with the
humerus held on the side and the palm at the supine
position. RANG was defined as the elbow an gle when
the subject kept his arm relaxed on the side.
Following these measurements, subjects were seated
upright on a Kin-Com isokinetic dynamometer (Chatta-
nooga, Tenn., USA) switched to the isometric mode. The
trunk was immobilized by straps, the shoulder joint was
stabilized at the neutral position (with the humerus par-
allel to the trunk), the forearm was at the supine position
and the wrist was placed against the lever arm. The MIF of
the elbow flexors was measured at five different elbow
angles, i.e. 50,70,90, 140 and 160 in random order
(180 represents full elbow extension). The elbow angles
238
used for MIF testing were set using the Kin-Com visual
display unit after entering a reference datum elbow angle
of 90. This reference angle was measured with a goni-
ometer. Each subject performed two maximal voluntary
isometric contractions of 3 s duration at each angle, and
the best trial was taken as the MIF of the angle. A resting
period between 45 s and 60 s was allowed between repe-
titions. Strong verbal encouragement was given to the
subjects during all trials.
Eccentric and isometric exercise protocols
Three days after the preliminary measurements, the
subjects were randomly divided into two groups [the
eccentric exercise group (ECC, n=7), and the isometric
exercise group (ISO, n=7)]. Subjects in the ECC group
performed 50 maximal voluntary eccentric contractions
of the elbow flexors of the non-dominant arm on the
isokinetic dynamometer at an angular velocity of 30 .s
)1
(2 sets of 25 eccentric muscle actions with a 5 min break
between sets). This slow angular velocity was chosen in
order to maximally load the muscles involved from the
start of the range of motion. The body position was
standardized as described above (shoul der at ne utral
position) and the range of elbow joint motion was 120
(from an elbow angle of 50 to 170). Each muscle action
lasted 4 s and a 15-s rest was allowed between repetitions.
Subjects in the ISO group also performed 50 maximal
voluntary isometric muscle cont ractions of the elbow
flexors of the non-dominant arm on the isokinetic
dynamometer switched to the isometric mode (2 sets of
25 isometric muscle actions with a 5-min break between
sets). The body position was standardized as described
above but the shoulder was held at 45 extension from
the neutral position (i.e. humerus behind the perpen-
dicular level of the torso) and the elbow joint at 140.
This position was chosen to make the elbow flexors
contract from a lengthened position during the isometric
exercise protocol (Jones et al. 1989). Each maximal
isometric contraction was performed for 10 s, with 20 s
of intervening rest.
Post-exercise measurements
Following the ECC and ISO exercise protocols subjects
visited the laboratory at 24-h intervals for 4 days (days
1–4). Muscle soreness was evaluated by a visual ana-
logue scale that had a continuous line of 100 mm with
‘‘no pain’’ on one end and ‘‘extremely sore’’ on the
other. Instructions had been given to the subjects to rate
soreness levels during one repetition of flexing and
extending the elbow joint throughout the entire range of
motion and upon light palpation of the elbow flexors
area with the arm at rest (Nosaka and Clarkson 1996).
The average of these two values for each subject was
used as the criterion score of the day. FANG and
RANG were then evaluated. FANG, RANG and rat-
ings of muscle soreness were used as indirect markers of
muscle damage (Clarkson et al. 1992). Finally, maximal
isometric force of the elbow flexors was measured as
described above at the five different elbow angles in
random order.
Statistical methods
The angle-MIF profiles of the ECC and ISO groups
were analysed separately using two one-way analyses of
covariance (ANCOVAs) with repeated measures. Each
analysis incorporated ‘days’ (the baseline day plus the 4
recovery days post-exercise) as a within-subject factor
and ‘angle’ and ‘angle
2
’ as two covariates. The effect of
angle was entered as a quadratic polynomial (Force =
a+bA+cA
2
, where a, b and c are the fitted polynomial
parameters and A is elbow angle) to accommodate the
likelihood that isometric force will peak somewhere be-
tween 50 and 160, according to the angle-force rela-
tionship. The parameters ‘b’ and ‘c’ of the polynomial
correspond to the covariate terms ‘angle’ and ‘angle
2
’,
while ‘a’ is the constant term of the polynomial. Both
covariate terms were also allowed to vary by day (by
incorporating a day-by-angle and a day-by-angle
2
interaction term). These interactions were introduced to
assess whether these quadratic polynomial curves varied
significantly during the recovery days, providing evi-
dence that a shift in the entire angle-force curve had
occurred.
In order to confirm that the angle-MIF profiles of the
two groups were similar at baseline (day 0), a one way
ANCOVA with replications (i.e. at 50,70,90, 140
and 160) was performed. The analysis incorporated a
between subjects factor ‘group’ (ECC versus ISO group)
and, as befo re, adopted ‘angle’ and ‘angle
2
’ as two co-
variates. Both covariate terms were also allowed to vary
by group (by incorporating a group-by-angle and a
group-by-angle
2
interaction term). Similarly, differences
between the angle-MIF profiles of the two groups on
each recovery day (day 1 to day 4) were examined using
this type of analysis.
Changes in RANG, FANG and muscle soreness were
assessed using two-way analyses of variance with
repeated measures, with the two modes of exercise
‘group’ being the between-subject factor and the change
over time, ‘days’ being the within-subject factor. Where
significant F ratios were found for main effects or
interaction (P<0.05), the means were compared using
Tukey’s post hoc tests. Results are presented as the mean
and standard error (SE).
Results
The ANCOVA with replications was unable to detect
any differences in the angle-MIF quadratic curves
between the ISO and ECC group at baseline (day 0), i.e.
the interaction terms group-by-angle
2
and group-by-
239
angle were not significant, and no difference was
detected between the two groups’ fitted constants (all
>0.05). Thus the two groups had similar angle-force
curves on day 0 (Table 1).
The ANCOVA comparing the daily changes in angle-
force curves within the ECC group identified significant
day-by-angle and a day-by-angle
2
interaction terms. The
fitted quadratic polynomial parameters and the calcu-
lated optimum angle and peak force by day for the ECC
group are given in Table 1. The quadratic polynomial
angle-force curves of all 4 recovery days were signifi-
cantly different from day 0 (P<0.01 to 0.05).
In cont rast, the ANCOVA comparing the daily
changes in angle-force curves within the ISO group
identified no significant day-by-angle and a day-by-an-
gle
2
interaction terms. However, by observing the simi-
larity between the fitted angle-force curves (i.e. the
constant, angle and angle
2
parameters) for days 2, 3 and
4, the data from these three days were combined to take
the same ‘day’ indicato r level for a subsequent re-a nal-
ysis. The resulting ANCOVA now identified a significant
day-by-angle interaction term (P<0.05, see Table 1).
Although not as dramatic as the ECC group, this finding
confirms that the quadratic polynomial curves for day 1
and days 2, 3 and 4 combined (day 234) changed sig-
nificantly from the baseline measurement (day 0).
The magnitude of changes of the angle-force curves
was greater for the ECC group compared to the ISO
group on each of the 4 recovery days (P<0.01). This is
evident in Fig. 1 and it is also reflected in the calculated
parameters of peak isometric force and optimum angle.
Figure 2 shows the percent changes in peak isometric
force and the shift of the optimum angle towards greater
elbow angles that is greater in the ECC group. The
interaction terms group-by-angle
2
and group-by-angle
were not significantly different between the post-exercise
days (days 1–4) in each of the two groups, indicating
that the shape of the angle-MIF curve remained
unchanged for all recovery days in both groups and thus
the shift persisted.
The changes in relaxed elbow angle were significa ntly
larger for the ECC group compared with the ISO group
(P<0.05, Fig. 3). On the other hand, changes in flexed
elbow angle were not significantly different between the
two groups (no main effect for group, Fig. 3), but the
group by day interaction marginally failed to be statis-
tically significant (P=0.07). When examining the abso-
lute values compared to the baseline, RANG and
FANG in the ISO group returned to their baseline val-
ues by day 3, whereas in the ECC gro up both RANG
and FANG remained depressed for the whole of the
recovery period (P<0.01). Ratings of muscle soreness
peaked on day 2 in both conditions [53 (10) for ECC and
28 (7) for ISO] and were significantly higher for the ECC
compared with the ISO group (main effect for group,
P<0.05).
Discussion
The aim of the present study was to examine and com-
pare the time-course of the shift in the angle-force curve
of the elbow fle xors after two types of maximal volun-
tary contractions: eccentric and isometric from a long
muscle length. This was accomplished by employing a
specialized curve fitting proce dure for the angle-force
data of the elbow joint which not only allowed the cal-
culation of peak force and optimum angle for each day,
but also enabled the statistical comparison of the angle-
force curves between and within groups. To our
knowledge, the studies that have examined the shift in
optimum angle in animal (Wood et al. 1993; Lynn and
Morgan 1994; Lynn et al. 1998; Talbot and Morgan
1998; Brockett et al. 2002) and human muscle (Jones
et al. 1997; Whitehead et al. 1998; Brockett et al. 2001)
have all fitted Gaussian curves to the force values close
to the optimum (above 75%–90% of the peak tension).
The approach used in the present study has the advan-
tage of describing not only the shift of the optimum
angle but the changes of the entire angle-tension curve
Table 1 The fitted quadratic polynomial parameters, calculated
optimum angle and peak force by day in the ECC and ISO groups
(constant, angle and angle
2
correspond to the parameters ‘a’, ‘b’
and ‘c’ of the fitted polynomial model: Force =a+bA+cA
2
, where
A refers to elbow angle). r
2
·100 (%) Coefficient of determination
for the fitted polynomial equations
ECC group
Day Constant** Angle** Angle
2
** Optimum angle () Peak force (N) r
2
·100
0 )173 (45) 8.2 (0.9) )0.036 (0.005) 113.8 292.3 (11.2) 81.8%
1 )48 (35) 3.1 (0.8) )0.012 (0.004) 129.9 153.5 (8.6) 80.3%
2 )83 (30) 4.0 (0.6) )0.015 (0.003) 131.7 181.5 (7.4) 89.1%
3 )50 (44) 3.7 (0.9) )0.014 (0.004) 131.3 190.4 (10.7) 83.5%
4 )96 (43) 5.0 (0.9) )0.020 (0.004) 126.1 218.9 (10.7) 86.7%
ISO group
Day Constant** Angle* Angle
2
Optimum angle () Peak force (N) r
2
·100
0 )206 (53) 8.6 (1.1) )0.038 (0.005) 111.7 272.4 (13.0) 74.7%
1 )298 (48) 8.9 (1.0) )0.038 (0.005) 116.5 222.7 (11.9) 78.7%
234 )281 (24) 9.0 (0.5) )0.038 (0.002) 116.9 243.7 (10.1) 83.8%
*P<0.05,
**P<0.01 for day-by-parameter interaction
240
across the functional range of motion of the joint, by
examining the changes in the quadratic polynomial
parameters.
One main finding of this study was the large and
persistent shift of the angle-force curve towards a longer
muscle length after repeated maximal eccentric con-
tractions. The magnitude of the shift in optimum an gle
observed in the present study (16–18) is the highest
reported in the literature for human muscles in vivo. The
other main findin g of our study was that a smaller but
long lasting shift of the angle-force curve was also
observed after isometric exercise with the elbow flexors
contracting from a long muscle length. A possible
explanation for the shift of the angle-force curve after
both eccentric and isometric exercise at long muscle
length is the presence of ‘‘overstretched’’ sarcomeres in
the fibres of the muscles involved. The ‘‘popping’’ sar-
comere hypothesis has been described in detail (Morgan
1990; Proske and Morgan 2001) and has been supported
by a number of studies in both single fibres and human
muscles in vivo (Jones et al. 1997; Whitehead et al.
1998). This hypothesis is based on the potential insta-
bility of half-sarcomere lengths in a muscle contracting
on the descending limb of the angle-force relationship,
i.e. beyond the optimum length (Morgan and Allen
1999). Earlier work has shown that the more ‘‘disad-
vantaged’’ sarcomeres in this respect are located in the
middle part of the muscle fibres, because sarcomere
spacing is greater compared to that near the ends (Lieber
and Baskin 1983; Friden and Lieber 1992). During
contraction beyond the optimum muscle length, these
sarcomeres may be stretched more than their neigh-
bouring sarcomer es and thus become disrupted or
‘‘overstretched’’. This will increase the series compliance
of the muscle, leading to a shift of the angle-force curve
to the right (Proske and Morgan 2001). Work on animal
muscle fibres by Macpherson et al. (1996) has provided
electron micrographs whic h support this hypothesis by
Fig. 1 Angle-force curves reconstructed using the fitted quadratic
polynomial parameters for each day for the ISO and ECC groups.
Crosses indicate the optimum angle for each curve. Standard errors
for the reconstructed force data were calculated from the residual
mean-square error term from the corresponding ANCOVAs
Fig. 2 Percent changes in peak isometric force compared to day 0
(upper panel) and shift in optimum elbow angle (lower panel),
calculated from the fitted quadratic polynomial parameters
presented in Table 1. Data for peak force and shift in optimum
angle for the combined days 234 in the ISO group are shown as
repeated points (dotted line). Standard errors for the peak force
data were calculated from the residual mean-square error term
from the corresponding ANCOVAs
241