J Appl Physiol 98: 2126 –2131, 2005;
doi:10.1152/japplphysiol.00511.2004.
Metabolic energy and muscular activity required for leg swing in running
Jesse R. Modica and Rodger Kram
Department of Integrative Physiology, University of Colorado, Boulder, Colorado
Submitted 17 May 2004; accepted in final form 28 January 2005
appear to move almost effortlessly; however, running is a complex neuromuscular task with
a high metabolic demand. We posed two specific questions
pertaining to the swing phase of running. First, what fraction of
the energetic cost of running at a moderate speed is due to
initiating and propagating leg swing? Second, what muscles
are responsible for these actions?
Energetics. Several mechanical functions are thought to
determine the energetic cost of level running, including support
of body weight, horizontal propulsion, and leg swing. The most
significant is the production of vertical force to support body
weight (18). Additionally, horizontal propulsion accounts for
⬃40% of the net energetic cost of running (8). However, the
portion of the energetic cost of running due to leg swing remains
controversial (19, 33). Leg swing in running is a combination of
active and passive mechanics (2, 6). In addition to initiating and
propagating leg swing, muscle actions are also needed for maintaining flexed knee and ankle joint positions during swing and
arresting leg motion at the end of the swing phase.
A diverse array of studies has creatively but indirectly
investigated the energetic cost of leg swing in running and
found it to be negligible. Torso loading (30) and simulated
reduced gravity (13, 27) experiments have found direct proportionality between the vertical force needed to support
weight and the energetic cost of running, implying that leg
swing does not comprise a substantial fraction of the cost of
running. If leg swing costs were substantial, then such loading
and unloading procedures would cause changes in cost that are
less than proportional. Thus Taylor et al. (30) surmised that leg
swing does not exact a significant metabolic cost. A comparison of gazelles and cheetahs, animals with the same body
weight but with light vs. massive limbs, found the metabolic
cost of locomotion to be indistinguishable. This evidence also
suggests that the cost of leg swing is small (31). Overall, these
studies indicate that leg swing comprises a negligible fraction
of the energetic cost of running.
Results from limb-loading studies, biomechanical analyses,
and blood flow experiments support the opposite view.
Whereas adding 1% of body weight to the torso causes a 1%
increase in metabolic rate, adding 1% of body weight to the
feet causes as much as a 9% increase in running humans (9, 17,
21, 25). This evidence suggests that the cost of leg swing is
more substantial than the cost of supporting body weight.
Furthermore, loads placed on the feet cause a greater increase
in metabolic rate than loads placed more proximally on the leg
(21). This suggests that the moment of inertia of leg segments
about proximal leg joints determines the energetic cost of leg
swing. Limb loading experiments on other species also indicate
a significant cost of leg swing (29). Classically, Hill (15) and
more recently van Ingen Schenau et al. (33) have reasoned that
the mechanical work involved in leg swing accounts for the
vast majority of the metabolic cost of running. Most recently,
Marsh et al. (19, 20), quantified leg muscle blood flow in
running birds and concluded that leg swing comprises 26% of
the net cost of normal running. Overall, these studies indicate
that leg swing is a substantial determinant of the energetic cost
of running.
Our first aim was to quantify the energetic cost of initiating
and propagating leg swing in running. We applied external
swing assist (ESA) forces to accelerate the swing leg forward
at the end of stance phase. We hypothesized that the energetic
cost of running would decrease when ESA was provided. We
suggest that a decrease in energy consumption with the optimal
ESA would indicate the normal cost of initiating and propagating leg swing.
Muscle activity. Remarkably, we do not have a definitive
understanding of which muscles are responsible for initiating
and propagating leg swing in running. In walking, ankle plantar
flexor muscle(s) contributes to swing phase initiation (16, 23,
26). In running, these muscles are active during the second half
of stance phase (22), suggesting that they could contribute to
leg swing initiation. The hip flexor muscles, i.e., rectus femoris
(RF), iliacus (IL), and psoas (PS), are active just before and
during early leg swing and thus are thought to contribute to leg
swing initiation and propagation (1, 22, 24, 28). However,
these inferences are based on temporal correlation rather than
Address for reprint requests and other correspondence: R. Kram, Dept. of
Integrative Physiology, Univ. of Colorado, UCB 354, Boulder, CO 803090354 (E-mail: [email protected]).
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
biomechanics; locomotion; electromyography; energetic cost
AN ELITE DISTANCE RUNNER MAY
2126
8750-7587/05 $8.00 Copyright © 2005 the American Physiological Society
http://www. jap.org
Downloaded from jap.physiology.org on July 31, 2005
Modica, Jesse R., and Rodger Kram. Metabolic energy and
muscular activity required for leg swing in running. J Appl Physiol
98: 2126 –2131, 2005; doi:10.1152/japplphysiol.00511.2004.—The
metabolic cost of leg swing in running is highly controversial. We
investigated the cost of initiating and propagating leg swing at a
moderate running speed and some of the muscular actions involved.
We constructed an external swing assist (ESA) device that applied
small anterior pulling forces to each foot during the first part of the
swing phase. Subjects ran on a treadmill at 3.0 m/s normally and with
ESA forces up to 4% body weight. With the greatest ESA force, net
metabolic rate was 20.5% less than during normal running. Thus we
infer that the metabolic cost of initiating and propagating leg swing
comprises ⬃20% of the net cost of normal running. Even with the
greatest ESA, mean electromyograph (mEMG) of the medial gastrocnemius and soleus muscles during later portions of stance phase did
not change significantly compared with normal running, indicating
that these muscles are not responsible for the initiation of leg swing.
However, with ESA, rectus femoris mEMG during the early portions
of swing phase was as much as 74% less than during normal running,
confirming that it is responsible for the propagation of leg swing.
2127
LEG SWING IN RUNNING
clear cause and effect. During the swing phase of running, the
hamstring muscles are active to maintain a flexed knee, and the
anterior tibialis (AT) presumably acts to maintain a dorsiflexed
ankle. At the end of swing phase, muscles such as the biceps
femoris (BF) are active, presumably to arrest leg motion via
eccentric actions.
Our second aim was to more definitively understand which
muscles are responsible for leg swing initiation and propagation in running. We recorded electromyographic (EMG) signals from key leg muscles during normal running and with
ESA. We hypothesized that medial gastrocnemius (MG), soleus (Sol), and RF EMG would be reduced during late stance
and/or early swing phases when ESA was provided. We
suggest that a decrease in the activation of these muscles would
indicate how much they normally contribute to leg swing.
METHODS
J Appl Physiol • VOL
Fig. 1. Schematic of external swing assist apparatus that applied small anterior
forces to the dorsum of each shoe. We used different combinations of rubber
tubing stretched over low-friction pulleys to adjust the magnitude of the
assistive force. During stance phase, as the foot moved backward, the treadmill
motor stretched the rubber tubing. At the onset of swing phase, the device
pulled the foot and leg anteriorly. Partway through the swing phase, the cable
clamp met the stopper plate and the pulling force was eliminated.
experimental running trials, we allowed 7 and 4 min, respectively, for
the subjects to reach steady state, and then we calculated the average
V̇O2 (ml O2/s) and V̇CO2 (ml CO2/s) for the subsequent 2 min. We
monitored mean respiratory exchange ratios to verify steady-state
aerobic metabolism (mean respiratory exchange ratio ⬍ 1.0). We
calculated metabolic rate, i.e., metabolic power (Pmet; W/kg), using
the subjects’ body mass and a standard equation (5):
P met ⫽ 16.58
W䡠s
W䡠s
䡠 V̇O2 ⫹ 4.51
䡠 V̇CO2
ml O2
ml CO2
To derive net metabolic rate, we subtracted the standing value from
the gross value for each running trial. The mean (⫾ SE) standing
metabolic rate was 1.51 ⫾ 0.19 W/kg.
Stride temporal variables and EMG. To determine when stance and
swing phases occurred, we detected foot-strike and toe-off events with
foot switches (B & L Engineering, Tustin, CA). We calculated swing
time, contact time, and stride frequency from the foot-switch data.
We measured EMG signals using a telemetered amplifier system
(Noraxon, Scottsdale, AZ). Before electrode placement, we shaved
and prepared the skin with fine sandpaper and alcohol. We placed
bipolar surface electrodes (1-cm diameter; silver-silver chloride), 2
cm apart (center to center) on the skin over five muscles (MG, Sol,
RF, long head of the BF, and AT) of the right leg according to the
guidelines of Cram and Kasman (11). Before the experiment, subjects
performed a series of contractions to ensure cross talk was negligible
(34). The electrodes remained in place throughout the experimental
protocol without being removed or replaced.
We sampled EMG signals at a frequency of 1,000 Hz and applied
a recursive second-order Butterworth band-pass digital filter (20 –500
Hz). We performed two separate analyses on the EMG signals:
temporal patterns and amplitude. We examined the temporal characteristics of each muscle’s EMG for normal running to define when in
the stride cycle we should evaluate the amplitude. For the temporal
98 • JUNE 2005 •
www.jap.org
Downloaded from jap.physiology.org on July 31, 2005
Overview. Subjects ran on a motor driven treadmill normally and
with anterior pulling forces applied to their feet during late stance and
early swing phases. We measured metabolic energy expenditure,
stride kinematics, and surface EMG signals of selected leg muscles.
Subjects. Ten recreational distance runners (5 men, 5 women; age:
26.1 ⫾ 3.3 yr; mass: 65.7 ⫾ 10.0 kg; height 1.74 ⫾ 0.08; mean ⫾ SD)
volunteered to participate. Subjects regularly ran at least 2 h/wk. They
were in good health and free of musculoskeletal injury. The experimental protocol was approved by the University of Colorado Human
Research Committee. All subjects gave written, informed consent.
Protocol. The experiment commenced with a standing metabolic
rate trial and a 10-min control (normal running) trial without the
external swing assist (ESA 0%). All subjects were familiar with
treadmill running before data collection. Subjects ran on a Quinton
18-60 treadmill at 3.0 m/s. The experimental trials consisted of four
randomly ordered, 7-min ESA trials (ESA 1, 2, 3, and 4% body
weight). A second 10-min ESA 0% trial concluded each experimental
session. Subjects rested for at least seven min between running trials.
ESA. The ESA device applied an independent anterior force to each
shoe, above the distal dorsum of the foot, during the swing phase (Fig. 1).
During the stance phase, as the foot moved backward, the treadmill motor
acted to stretch the rubber tubing. At the onset of swing phase, the device
pulled the foot and leg anteriorly. Shortly after midswing position, the
cable clamp on the nylon cord reached the stopper plate and the ESA
force dropped to zero. Thus, beyond midswing, the leg naturally followed
through the remainder of swing phase. In pilot experiments without the
stopper plate, ESA forces ⬎4% caused delayed onset knee flexor muscle
soreness presumably due to eccentric muscle actions during the swing
phase. So, for safety reasons and to minimize eccentric muscle actions,
we employed the stopper plate and limited the ESA setting to 4% body
weight during the experimental trials.
To set the ESA to the specified percent body weight, the subject
stood in place approximately at the middle of the length of the
treadmill bed. We adjusted the tension of the rubber tubing to the
specified force and tightened the cable clamp in place at the stopper
plate. We fixed a “reminder cord” between the treadmill railings in
front of the subject at waist height. This ensured that the subjects ran
at the same location on the treadmill so that the magnitudes of the
ESA forces were consistent. The ESA force at toe off, i.e., the onset
of swing phase, was ⬃1% of body weight greater than the specified
force level. During the swing phase, as the rubber tubing shortened,
the ESA force decreased to the specified force, at which point the
cable clamp met the stopper plate and the ESA force immediately
dropped to zero (i.e., the nylon cord went slack).
Energetics. We measured rates of oxygen consumption (V̇O2) and
carbon dioxide production (V̇CO2) using an open-circuit respirometry
system (Physio-Dyne Instrument, Quogue, NY). We measured standing metabolic rate before the running trials. For the control and
2128
LEG SWING IN RUNNING
Fig. 2. Net metabolic rate as a function of external swing assist force. Values
are means ⫾ SE. External swing assist forces of 2, 3, and 4% of body weight
significantly decreased net metabolic rate by 11, 15, and 21%, respectively.
Line is a linear least squares regression (W/kg ⫽ 9.7– 0.52 external swing
assist; R2 ⫽ 0.51). *Significant differences (Tukey’s honestly significant
difference test: P ⬍ 0.05).
RESULTS
Energetics. Net metabolic rate (W/kg) decreased with
greater ESA force (P ⬍ 0.0001; Fig. 2, Table 1). Specifically,
with ESA forces of 2, 3, and 4% of body weight, net metabolic
rate significantly decreased by 11.2, 15.6, and 20.5%, respectively.
Temporal stride variables. Stride frequency only modestly
increased with ESA forces; it was ⬍6% greater (P ⬍ 0.05)
with an ESA of 4% body weight. This change in stride
frequency was exclusively due to decreasing swing time,
which was 9.4% less with an ESA of 4% body weight (P ⬍
0.0001; Table 1; Fig. 3). Contact time did not change with ESA
(P ⫽ 0.98; Table 1).
Temporal EMG in normal running. Before EMG amplitude
calculations, we determined when each muscle was active
during a full gait cycle of normal running as described in
METHODS. On the basis of the temporal EMG patterns of normal
running (Fig. 4) and to test our hypotheses, we compared the
mEMG for each muscle between conditions, only for specific
sections of the stride cycle. We defined 0% of a stride cycle to
be the foot strike. On average, toe off occurred at 36% of the
stride cycle. For muscle activations that span late swing and
early stance, we refer to foot strike as 100% of the stride cycle
and toe off as 136%. During normal running, the MG was
active from 86 to 130% of the stride cycle (late swing through
almost all of stance phase; mean for all subjects). The Sol was
active from 92 to 127% of the stride cycle. To elucidate any leg
swing initiating functions of MG and Sol, we analyzed mEMG
beginning halfway through stance phase and ending when the
MG and Sol deactivated (MG:18 –30% of stride cycle, Sol:
18 –27% of stride cycle).
Eight of the 10 subjects exhibited biphasic activation patterns for the RF. One burst occurred from 88 to 119% of the
stride cycle (late swing to middle stance) and a more pertinent
burst occurred from 40 to 60% of the stride cycle (early to
middle swing phase). To quantify the swing propagating function of the RF, we analyzed mEMG during the early swing
phase burst of activation.
BF activity patterns were less consistent between subjects
than the other muscles studied. All subjects activated the BF in
midswing phase (average of 69% of stride cycle), and in 9 of
the 10 subjects the BF activity ceased at an average of 129%
Table 1. Metabolic and kinematic data for each external swing assist condition
ESA Force, %body weight
Metabolic power, W/kg
Stride frequency, strides/s
Leg swing time, ms
Ground contact time, ms
0
1
2
3
4
9.62⫾0.23
1.43⫾0.02
445⫾9.6
258⫾7.6
9.37⫾0.17
1.44⫾0.03
436⫾13.2
260⫾7.5
8.54⫾0.20*
1.46⫾0.03
427⫾11.9
260⫾7.6
8.12⫾0.26*
1.47⫾0.02*
420⫾9.5*
260⫾6.4
7.65⫾0.30*
1.51⫾0.03*
403⫾10.6*
262⫾8.1
Values are means ⫾ SE for 10 subjects. Metabolic power decreased with external swing assist (ESA) forces. Stride frequency increased slightly and swing
time decreased slightly with ESA. Ground contact time did not significantly change with ESA (P ⫽ 0.98). *Significant differences with ESA 0% (Tukey’s
honestly significant difference test: P ⬍ 0.05).
J Appl Physiol • VOL
98 • JUNE 2005 •
www.jap.org
Downloaded from jap.physiology.org on July 31, 2005
analyses, we full-wave rectified the data, and then we used a secondorder Butterworth low-pass digital filter (7 Hz) to yield a linear
envelope. With foot-strike and toe-off information, we synchronized
the EMG signals to the gait events. Through visual inspection of each
muscle’s normal running EMG data, we determined the time in a gait
cycle that a quiescent baseline consistently occurred in all subjects.
We calculated the means and standard deviations (SDs) of the baseline data and defined a muscle as active if the EMG linear envelope
exceeded a threshold of the baseline plus a certain number of SDs for
at least 100 ms. For the different muscles, the number of SDs ranged
from two to six, depending on the noise level of the baseline. For each
muscle, the same number of SDs was used for all subjects. To
calculate mean EMG (mEMG) amplitude, we full-wave rectified the
band-pass-filtered data, and then we calculated mEMG on the basis of
the temporal EMG characteristics of normal running.
Statistical analyses. Energetic, kinematic, and mEMG data from
this study were analyzed across all conditions using a repeatedmeasure single-factor ANOVA. When warranted, we performed a
Tukey’s honestly significant difference post hoc test to analyze differences between each ESA level. Significance was defined as P ⬍ 0.05.
Fig. 3. Swing time represented as a percentage of normal running for each
external swing assist magnitude. Values are means ⫾ SE. Swing time significantly decreased by 6 and 9% with external swing assist forces of 3 and 4%
of body weight, respectively (*P ⬍ 0.01).
LEG SWING IN RUNNING
2129
our device may have caused an increase in BF mEMG. However, BF mEMG did not significantly increase with any amount
of ESA. Although the BF activity did not increase during the
swing phase, neither did it decrease, which indicates that our
ESA device did not eliminate all the muscle activity needed for
the swing phase in general. The AT mEMG during the swing
phase did not change significantly from any of the ESA perturbations (P ⫽ 0.09) indicating that the ESA device had no effect on
the ankle dorsiflexor muscles. Again, this indicates that we did not
eliminate the need for AT activity during swing.
DISCUSSION
(late stance phase). Two subjects had brief periods of inactivity
from 100 to 116% of the stride cycle. One subject only briefly
activated the BF (70 – 87%). Thus, to evaluate the effect of
ESA on the BF during swing phase, we analyzed mEMG from
69 to 100% of the stride cycle for the other nine subjects. To
evaluate the effect of ESA on the BF during stance phase, we
analyzed mEMG from 16 to 29% of the stride cycle for 9 of 10
subjects; one subject was excluded due to the absence of
activation during stance phase.
Nine of 10 subjects maintained activation of the AT from 42
to 106% of the stride cycle (most of swing and early stance).
To evaluate the effect of ESA on the AT during swing phase,
we analyzed mEMG from 42 to 100% of the stride cycle.
Effect of ESA on EMG amplitude. Decreases in mEMG with
ESA indicate that a particular muscle contributes to swing
initiation and/or propagation during normal running. We did
not detect a significant change during late stance phase in MG
(P ⫽ 0.38) or Sol (P ⫽ 0.76) mEMG (Fig. 5). During the swing
phase, however, mEMG of the RF dramatically decreased by
38%, 63%, and 74% (P ⬍ 0.0001) in response to the 2, 3, and
4% ESA conditions, respectively (Fig. 5).
We evaluated BF mEMG during late swing phase to detect
any confounding effects of our ESA device. Because BF acts to
arrest the forward motion of the leg at the end of the swing
phase and ESA acts to pull the leg forward, we anticipated that
J Appl Physiol • VOL
Fig. 5. Mean EMG (mEMG) presented as a percentage of normal running for
each external swing assist magnitude. Values are means ⫾ SE. On the basis of
temporal muscle activations for normal running, we calculated mEMG for the
following specific portions of a stride: MG (18 –30%), Sol (18 –27%), RF
(40 – 60%), BF (69 –100%), and AT (42–100%). MG and Sol activations did
not significantly change with external swing assist (P ⫽ 0.38, P ⫽ 0.76). In
contrast, RF activation significantly decreased by 38, 63, and 74% with ESA
forces of 2, 3, and 4% of body weight, respectively (P ⬍ 0.0001). BF
activation decreased by 10% with an external swing assist force of 4% of body
weight (P ⬍ 0.05). Significant difference from normal running: *P ⬍ 0.001;
**P ⬍ 0.05.
98 • JUNE 2005 •
www.jap.org
Downloaded from jap.physiology.org on July 31, 2005
Fig. 4. Electromyographic (EMG) data for normal running. Traces are filtered,
rectified EMG data for a typical subject. One full stride (1 stance and 1 swing
phase) followed by a second stance phase are presented (0 –136% of stride
cycle). Horizontal bars, group mean (⫾ SE) onset and offset timing as a
percentage of stride cycle. Muscles depicted are medial gastrocnemius (MG),
soleus (Sol), rectus femoris (RF), biceps femoris (BF), and anterior tibialis
(AT). Dashed vertical lines, mean foot-strike time at 0 and 100% and mean
toe-off time at 36 and 136% of the stride cycle.
Energetics. As hypothesized, metabolic rate decreased with
increasing ESA forces. With the greatest magnitude of ESA,
4% body weight, net metabolic rate decreased by 20.5% from
the normal running condition. Thus our data suggest that the
metabolic cost of initiating and propagating leg swing is ⬃20%
of the net cost of running at 3.0 m/s. Although we hypothesized
that we would find an energetically optimum ESA force, we
did not find such a minimum below the 4% body weight ESA
force.
The actual total cost of swing may be more or less than 20%.
If we had applied greater ESA forces, metabolic cost may have
further decreased, but we did not apply forces ⬎4% body
weight to avoid injuring our subjects. Another factor to consider is the metabolic cost of arresting leg swing before foot
strike. Our ESA device applied a forward pulling force to the
feet that might have increased the cost of arresting leg swing
2130
LEG SWING IN RUNNING
J Appl Physiol • VOL
and found almost exactly proportional decreases in net metabolic cost. It remains unclear why reduced-gravity running
experiments suggest that swing cost is negligible, yet several
loading experiments suggest that there is some cost of swing.
Recent studies have found similar quantitative results similar
to ours using dramatically different and innovative methodology. Marsh et al. (19, 20) measured the rate of blood flow to
leg swing and stance muscles in running birds (guinea fowl).
They found a direct relationship between cardiac output and
systemic V̇O2, so they were able to estimate local oxygen
demand using local blood flow. Whereas stance muscles demand the vast majority of the leg blood flow (74%) during
running, leg swing muscles demand the remaining 26% of the
leg blood flow. Although these measurements are not from
human runners, it is intriguing that our estimate of the cost of
leg swing is so similar to that of Marsh et al.
EMG. We reject the hypothesis that activation of ankle
plantar flexors would decrease with ESA. We did not find
changes in mEMG of the MG and Sol muscles with ESA in the
second half of stance phase. These results suggest that the
plantar flexor muscle group is not responsible for leg swing
initiation in running. This finding is in contrast to the idea
raised in several walking studies that the ankle plantar flexor(s)
initiate leg swing (16, 23, 26). We attribute the activation of the
ankle plantar flexor muscles in late stance to the functions of
forward propulsion and supporting body weight. Further defining the function of this muscle group is warranted.
We reject the hypothesis that RF activation decreases during
late stance phase with ESA. However, as hypothesized, RF
activation decreased during the first half of swing phase with
ESA. The RF was not characteristically active during normal
running in the later portion of stance phase, suggesting that RF
does not contribute to leg swing initiation. However, the
mEMG of the RF during 40 – 60% of the stride cycle, decreased
progressively and dramatically with ESA. Thus our EMG results
show directly that RF muscle activation is responsible for leg
swing propagation but not leg swing initiation. Through the
elegant use of indwelling fine wire EMG electrodes, Andersson et
al. (1) were able to demonstrate that in late stance phase two other
major hip flexors, IL and PS, are activated before RF activation.
Thus we can still only infer from their study that EMG timing and
joint kinematics that IL and/or PS initiate leg swing in running.
Limitations and future study. Our ESA device is simple, is
inexpensive, and appears to nearly eliminate the need for active
leg swing initiation and propagation during running. However,
the force pattern of our ESA is not necessarily energetically
optimal, and an actively controlled ESA device might be
superior. An ideal ESA device might be activated and deactivated at exact points in the stride cycle and with programmable
force magnitudes. An ideal ESA device might also apply
separate forces to each leg segment; however, our goal was to
begin with a simple device and a simple experimental design.
Pulling on multiple leg segments might cause undesirable
intersegmental interactions and cocontractions. Finally, an
ESA device might somehow eliminate the need for the muscle
actions during swing to maintain knee flexion and/or ankle
dorsiflexion.
It is likely that the treadmill motor exclusively loaded the
elastic cords of the ESA device. We attempted to confirm this by
measuring BF muscle activation during stance phase. BF muscle
activation during late stance phase (16 –29% of stride cycle)
98 • JUNE 2005 •
www.jap.org
Downloaded from jap.physiology.org on July 31, 2005
and thus counteracted the reduction in swing initiation and
propagation costs. However, our EMG data for the BF, a hip
extensor and knee flexor muscle, did not show increased
activation with ESA in late swing phase. This suggests that the
cost of arresting leg swing did not increase due to ESA.
However, we did not eliminate the basic cost of arresting leg
motion, which properly should be considered part of the total
cost of leg swing. AT muscle activity was present during the
swing phase and invariant with ESA. The metabolic cost of AT
activity must have some, although probably small, metabolic
cost. Because we did not reduce, let alone eliminate, BF or AT
activity with ESA, our estimate of 20% could be an underestimate of the total cost of leg swing.
On the other hand, the actual cost of swing may be ⬍20% if
there is some interaction between leg swing and forward
propulsion. The ESA device pulls each leg anteriorly during
the first portion of swing phase. Although ESA primarily
reduces the body’s need to perform work required for leg
swing, it also applies a forward propulsive impulse to the
whole body. Chang and Kram (8) estimated that forward
propulsion comprises ⬃40% of the net metabolic cost of
running. Our observed decrease in metabolic cost with ESA
probably reflects both a decreased cost of leg swing and a
decreased cost of propulsion. Thus we believe that our estimate
of leg swing cost of 20% of the net cost of running is a
moderate overestimate, but further investigation is needed to
consider the interactions of propulsion and swing.
If the cost of leg swing comprises 20% of the net metabolic
cost of running, how have whole body loading and unloading
studies surmised that it is negligible (13, 30)? Recall that if leg
swing costs are substantial, then whole body loading procedures should cause less than proportional increases in cost.
Taylor et al. (30) added loads of 20% body weight and found
a proportional increase in metabolic cost, but they used only
two human subjects in their experimental design. Thus their
statistical resolving power may not have been robust enough to
detect a small cost of leg swing. Another limitation is that their
subjects carried the loads in backpacks, which probably required additional recruitment of abdominal and back postural
muscles (3). The metabolic demand of such postural muscles
could have inflated the apparent cost of supporting the additional weight and obscured the cost of leg swing. Furthermore,
Taylor et al. (30) compared loaded and unloaded gross metabolic cost. We calculated that net metabolic cost and believe
that this value gives better insight into the cost of running
itself, although that philosophy is controversial (7). More
recent studies of loaded running energetics have found less
than proportional increases in net metabolic cost due to small
(⬃10% body weight) additions of mass balanced around the
torso. Cureton and Sparling (12) found a 6.8% increase with an
average load of 7.5% body weight. Thorstensson (32) reported
a 7.5% increase in response to a 10% body weight load.
Bourdin et al. (4) observed a 4.7% increase for a 9.3% body
weight load. Cooke et al. (10) found a statistically insignificant
3.3% increase with a 10% body weight load. Obviously,
parsing out the small changes in metabolic rate due to only
10% body weight extra loads pushes the limits of accuracy in
these measurements. But overall, there is a less than proportional increase in the net metabolic rate when loads are carried
around the torso during running. However, Farley and McMahon (13) unweighted their subjects (by 25, 50, and 75%)
LEG SWING IN RUNNING
ACKNOWLEDGMENTS
We thank the members of the Locomotion Laboratory and Dale Mood for
valuable feedback and assistance.
GRANTS
This research was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant AR-44688.
REFERENCES
1. Andersson EA, Nilsson J, and Thorstensson A. Intramuscular EMG
from the hip flexor muscles during human locomotion. Acta Physiol Scand
161: 361–370, 1997.
2. Bennett MB. A possible energy-saving role for the major fascia of the
thigh in running quadrupedal mammals. J Zool 219: 221–230, 1989.
3. Bobet J and Norman RW. Effects of load placement on back muscle
activity in load carriage. Eur J Appl Physiol 53: 71–75, 1984.
4. Bourdin M, Belli A, Arsac LM, Bosco C, and Lacour JR. Effect of
vertical loading on energy cost and kinematics of running in trained male
subjects. J Appl Physiol 79: 2078 –2085, 1995.
5. Brockway JM Derivation of formulae used to calculate energy expenditure in man. Hum Nutr Clin Nutr 41: 463– 471, 1987.
6. Cavagna GA and Kaneko M. Mechanical work and efficiency in level
walking and running. J Physiol 268: 467– 481, 1977.
7. Cavanagh PR and Kram R. The efficiency of human movement—a
statement of the problem. Med Sci Sports Exerc 17: 304 –308, 1985.
8. Chang Y-H and Kram R. Metabolic cost of generating horizontal forces
in running. J Appl Physiol 86: 1657–1662, 1999.
9. Claremont AD and Hall SJ. Effects of extremity loading upon energy
expenditure and running mechanics. Med Sci Sports Exerc 20: 167–171,
1988.
J Appl Physiol • VOL
10. Cooke CB, McDonagh MJ, Nevill AM, and Davies CT. Effects of load
on oxygen intake in trained boys and men during treadmill running. J Appl
Physiol 71: 1237–1244, 1991.
11. Cram JR and Kasman GS. Electrode placements. In: Introduction to
Surface Electromyography, edited by Cram JR, Kasman GS, and Holz J.
Gaithersburg, MD: Aspen, 1998, p. 348 –383.
12. Cureton KJ and Sparling PB. Distance running performance and metabolic responses to running in men and women with excess weight
experimentally equated. Med Sci Sports Exerc 12: 288 –294, 1980.
13. Farley CT and McMahon TA. Energetics of walking and running:
insights from simulated reduced-gravity experiments. J Appl Physiol 73:
2709 –2712, 1992.
14. Fedak MA, Heglund NC, and Taylor CR. Energetics and mechanics of
terrestrial locomotion. II. Kinetic energy changes of the limbs and body as
a function of speed and body size in birds and mammals. J Exp Biol 97:
23– 40, 1982.
15. Hill AV The dimensions of animals and their muscular dynamics. Sci
Prog 38: 209 –230, 1950.
16. Hof AL, Nauta J, Van der Knaap ER, Schallig MAA, and Struwe DP.
Calf muscle work and segment energy changes in human treadmill
walking. J Electromyogr Kinesiol 2: 203–216, 1993.
17. Jones BH, Toner MM, Daniels WL, and Knapik JJ. The energy cost
and heart-rate response of trained and untrained subjects walking and
running in shoes and boots. Ergonomics 27: 895–902, 1984.
18. Kram R and Taylor CR. Energetics of running: a new perspective.
Nature 346: 265–267, 1990.
19. Marsh RL, Ellerby DJ, Carr JA, Henry HT, and Buchanan CI.
Partitioning the energetics of walking and running: swinging the limbs is
expensive. Science 303: 80 – 83, 2004.
20. Marsh RL, Ellerby DJ, and Henry HT. Energetic costs of back and limb
loading in guinea fowl. Integ Comp Biol 43: 890, 2003.
21. Martin PE Mechanical and physiological responses to lower extremity
loading during running. Med Sci Sports Exerc 17: 427– 433, 1985.
22. McClay IS, Lake MJ, and Cavanagh PR. Muscle activity in running. In:
Biomechanics of Distance Running, edited by Cavanagh PR. Champaign,
IL: Human Kinetics, 1990, p. 165–186.
23. Meinders M, Gitter A, and Czerniecki JM. The role of ankle plantar
flexor muscle work during walking. Scand J Rehabil Med 30: 39 – 46,
1998.
24. Montgomery WH 3rd, Pink M, and Perry J. Electromyographic analysis of hip and knee musculature during running. Am J Sports Med 22:
272–278, 1994.
25. Myers MJ and Steudel K. Effect of limb mass and its distribution on the
energetic cost of running. J Exp Biol 116: 363–373, 1985.
26. Neptune RR, Kautz SA, and Zajac FE. Contributions of the individual
ankle plantar flexors to support, forward progression and swing initiation
during walking. J Biomech 34: 1387–1398, 2001.
27. Newman DJ, Alexander HL, and Webbon BW. Energetics and mechanics
for partial gravity locomotion. Aviat Space Environ Med 65: 815– 823, 1994.
28. Nilsson J, Thorstensson A, and Halbertsma J. Changes in leg movements and muscle activity with speed of locomotion and mode of progression in humans. Acta Physiol Scand 123: 457– 475, 1985.
29. Steudel K. The work and energetic cost of locomotion. I. The effects of
limb mass distribution in quadrupeds. J Exp Biol 154: 273–285, 1990.
30. Taylor CR, Heglund NC, McMahon TA, and Looney TR. Energetic
cost of generating muscular force during running: a comparison of small
and large animals. J Exp Biol 86: 9 –18, 1980.
31. Taylor CR, Shkolnik A, Dmi’el R, Baharav D, and Borut A. Running
in cheetahs, gazelles, and goats: energy cost and limb configuration. Am J
Physiol 227: 848 – 850, 1974.
32. Thorstensson A. Effects of moderate external loading on the aerobic
demand of submaximal running in men and 10 year-old boys. Eur. J Appl
Physiol 55: 569 –574, 1986.
33. Van Ingen Schenau GJ, Bobbert MF, and De Haan A. Mechanics and
energetics of the stretch-shortening cycle: a stimulating discussion. J Appl
Biomech 13: 484 – 496, 1997.
34. Winter DA, Fuglevand AJ, and Archer SE. Crosstalk in surface electromyography: theoretical and practical estimates. J Electromyogr Kinesiol 4: 15–26, 1994.
98 • JUNE 2005 •
www.jap.org
Downloaded from jap.physiology.org on July 31, 2005
actually decreased in response to ESA. In a future ESA study, it
would be helpful to collect EMG data from the gluteus maximus
muscle during stance phase to further quantify the role of the
treadmill motor in loading the ESA device during stance phase.
Moreover, because we observed substantial decreases in metabolic cost, the results suggest that the runners did not expend
much, if any, additional energy to load the ESA elastic cords.
Our experiment gave insight into the leg swing functions of
MG, Sol, and RF, but EMG measurements of other hip flexor
muscles such as the IL and PS would provide further information. Another useful experiment would examine the possible
increases in activity of these muscles in response to ankle
weights that make leg swing more difficult. Moreover, a study
investigating the interaction between horizontal propulsion and
leg swing is needed to clarify our energetics findings. Specifically, an experiment that combines an applied horizontal force
at the waist for center of mass propulsion (8) and ESA would
reveal the interaction between horizontal propulsion and leg
swing mechanics in running. If the individual effects of each
device do not sum when combined, this would indicate that the
cost of leg swing initiation and propagation is less than the
20% we found here. We also plan to study the effects of ESA
for a range of running speeds because many have argued that
leg swing is more costly at faster speeds (14, 15). ESA could
also be used to evaluate the cost of leg swing in walking.
Finally, we are exploring potential clinical uses of ESA for
treadmill walking rehabilitation.
In conclusion, we find that leg swing requires ⬃20% of the
net energy consumed in running. Neither MG, Sol, nor RF
muscles contribute to leg swing initiation; however, RF does
play a significant role in leg swing propagation.
2131