Averaged EMG profiles in jogging and running at different speeds.
ABSTRACT EMGs were collected from 14 muscles with surface electrodes in 10 subjects walking 1.25-2.25 ms(-1) and running 1.25-4.5 ms(-1). The EMGs were rectified, interpolated in 100% of the stride, and averaged over all subjects to give an average profile. In running, these profiles could be decomposed into 10 basic patterns, 8 of which represented only a single burst. Muscles could be divided into a quadriceps, hamstrings, calf and gluteal group, the profiles of which were composed of the same basic patterns. The amplitude of some bursts was constant, but other ones varied with running speed. This speed dependency was generally different between muscles of the same group. Many muscles show a similar profile in running as in walking. The most notable exception is the calf group, which shows activation in early stance (86-125%), together with quadriceps, instead of in late stance (26-55%) as in walking. This is also visible in low-speed running, 'jogging', where stance extends to 46% or 57%, instead of 30-37% as in normal running. Jogging shows some additional differences with normal running, related to this prolonged stance phase.
- [Show abstract] [Hide abstract]
ABSTRACT: The purpose of this study was to analysis the effect of form of outsole on lower leg Electromyography (EMG) during gait. The subject were 11 women(mean age years, height cm, weight , body mass index ), who had no serious foot musculoskeletal disease within3years prior to study. The vastus medialis, gastrocnemius, semitendinosus, tibialis anterior were compared during gait with 2 types of shoe(complex function double rocker sole, negative-heel rocker sole) by using pair t-test(pJournal of the Korea Academia-Industrial cooperation Society. 01/2012; 13(1).
- [Show abstract] [Hide abstract]
ABSTRACT: The purpose of this study was to analyse scientific according to period of rehabilitation training of ACL patients. ACL patients seven subjects participated in this study. Gait (1.58 m/sec) analysis was performed by using a 3-D Cinematography, a Zebris system and a electromyograph system. The data were analyzed by paired t-test. The joint angles were recorded from the ankle, knee, hip joints. Peak max dorsi-flexion and peak max plantar-flexion identified significant differences (pKorean Journal of Sport Biomechanics. 09/2013; 23(3).
- [Show abstract] [Hide abstract]
ABSTRACT: Human activity recognition based on the computer vision is the process of labelling image sequences with action labels. Accurate systems for this problem are applied in areas such as visual surveillance, human computer interaction and video retrieval. The challenges are due to variations in motion, recording settings and gait differences. Here we propose an approach to recognize the human activities through gait. Activity recognition through Gait is the process of identifying an activity by the manner in which they walk. The identification of human activities in a video, such as a person is walking, running, jumping, jogging etc are important activities in video surveillance. We contribute the use of Model based approach for activity recognition with the help of movement of legs only. Experimental results suggest that our method are able to recognize the human activities with a good accuracy rate and robust to shadows present in the videos.International Journal of Interactive Multimedia and Artificial Intelligence. 07/2014; 2(7).
Averaged EMG profiles in jogging and running at different speeds
Marnix G.J. Gazendama, At L. Hofa,b,*
aCenter for Human Movement Sciences, University of Groningen, PO Box 196, 9700 AD Groningen, The Netherlands
bLaboratory for Human Movement Analysis, Center for Rehabilitation, University Medical Center Groningen, The Netherlands
Received 9 December 2005; received in revised form 10 May 2006; accepted 29 June 2006
EMGs were collected from 14 muscles with surface electrodes in 10 subjects walking 1.25–2.25 m s?1and running 1.25–4.5 m s?1. The
EMGs were rectified, interpolated in 100% of the stride, and averaged over all subjects to give an average profile. In running, these profiles
could be decomposed into 10 basic patterns, 8 of which represented only a single burst. Muscles could be divided into a quadriceps,
hamstrings, calf and gluteal group, the profiles of which were composed of the same basic patterns. The amplitude of some bursts was
constant, but other ones varied with running speed. This speed dependency was generally different between muscles of the same group.
Many muscles show a similar profile in running as in walking. The most notable exception is the calf group, which shows activation in
early stance (86–125%), together with quadriceps, instead of in late stance (26–55%) as in walking. This is also visible in low-speed running,
‘jogging’, where stance extends to 46% or 57%, instead of 30–37% as in normal running. Jogging shows some additional differences with
normal running, related to this prolonged stance phase.
# 2006 Elsevier B.V. All rights reserved.
Keywords: EMG; Electromyography; Human running; Velocity
In a previous paper  a method was presented to
walking in a range of walking speeds. It turned out that the
timing of the profiles, when expressed as a fraction of the
strideduration, wasusuallyinvariable,while their amplitude
could vary with speed. The profiles of each muscle could be
composed intoa limitedset of basic patterns, which theyhad
in common within their functional group: calf, quadriceps,
hamstringsandgluteal.Theaimofthe present paperistosee
if a similar analysis can be made for running. The data so
obtained may serve as a database for EMG analyses of
running, adding to the data of Nilsson et al.  on a more
quantitative basis. An additional point of interest is the
neural control of locomotion: which changes in the
activation pattern effect the switch between the walking
or running modes of locomotion? . On the basis of the
results to be presented, it will be shown that it is practical to
define an additional ‘jogging’ mode of human locomotion.
2.1. Subjects and protocol
Ten healthy male subjects took part in this study, all of
whom were physically active men with no health problems.
The experiments were in accordance with the guidelines of
the local Medical Ethical Committee. The subject’s mean
and standard deviation for age, body mass, stature and leg
length were 20.8 ? 1.2 years, 71.3 ? 6.3 kg, 1.84 ? 0.07 m,
0.99 ? 0.05 cm,respectively.EMGsof14legmusclesofthe
right leg were recorded, see Table 1. They wore sporting
Subjects walked and ran on a motor-driven treadmill
(ENRAF Entred, belt size 0.48 m ? 1.60 m) at speeds
of 1.25, 1.5, 1.75, 2.0, 2.25, 2.5, 3.0, 3.5, 4.0 and 4.5 m s?1.
Gait & Posture 25 (2007) 604–614
* Corresponding author. Tel.: +31 50 363 2645; fax: +31 50 363 3150.
E-mail address: email@example.com (A.L. Hof).
0966-6362/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
The records were made during 30 s of ‘steady state’ walking
or running, after the subjects had been accustomed to the
new speed for some 30 s. Between each recording one
minute of rest was allowed. The subjects were first asked to
walk and then to run at speeds up to 2.25 m s?1. At speeds
from 2.5 up to 4.5 m s?1the subjects were asked to run only.
2.2. EMG recording
Surface EMGs were recorded bipolarly by Conmed
disposable surface electrodes (10 mm ? 10 mm electrode
area, inter electrode distance 24 mm, Conmed Corporation,
Utica, NY, USA). All electrodes were placed in the
lengthwise direction of the muscle on the right leg. The
positions of the electrodes were according to the SENIAM
recommendations , summarized in Table 1. All 14
muscles were recorded simultaneously. The electrodes and
electrode wires were wrapped on thigh and shank with an
elastic bandage do prevent dislocation during fast running.
a 32-channel PORTI ambulatory recording system (Twente
Medical Systems, Enschede, The Netherlands). Amplifier
specifications were: 100 dB common mode rejection,
2 mV pp noise level and 1 GV input impedance. Sampling
frequency was 800 Hz. Vertical force was recorded by force
transducers mounted underneath the treadmill belt. From this
signal the timing of the foot contact could be calculated .
2.3. Data processing
After recording, the EMG signals were high-pass filtered
with a 20 Hz fourth order Butterworth high-pass filter to
remove electrode artifacts, rectified and smoothed with a
24 Hz fourth order Butterworth low-pass filter, all with
Matlab software. It was assumed that, because of noise from
amplifier and electrodes and of cross-talk from adjacent
muscles, EMG levels below 10 mV could be considered
The smoothed rectified EMGs were linearly inter-
polated to 100 points p per stride, triggered by the right
heel contact. The recorded steps were screened to exclude
any obvious EMG artefacts or incorrect foot contacts. The
number of excluded steps in a 30 s recording did not
exceed two or three steps. Depending on the stride time,
therefore between 25 and 45 steps were averaged for each
For every subject i at every speed v the average
eðp;m;v;iÞ was calculated for the 14 leg muscles m, to
get an individualmean. For the n = 10 subjects a grand mean
was obtained at every normalised speed v for every muscle
In a similar way as described for walking, the EMG
patterns of all 14 muscles at all running speeds above
2.25 m s?1could be described by a limited set of basic
patterns FF(p, k) times a speed dependent amplitude factor
Dðm;k; ˆ vÞ. The profile of a certain muscle could be
composed of one to three basic patterns. It will turn out
that 10 basic patterns are sufficient. The speed dependence
could be modelled as a linear, in some cases a quadratic
relation. The complete description of all patterns thus
becomes in matrix form:
E?ðp;m; ˆ vÞ ¼ fD0ðm;kÞ þ D1ðm;kÞˆ v
þ D2ðm;kÞˆ v2g?FFðp;kÞT
in which D0, D1and D2are 14 ? 10 matrices and FF(p, k) a
100 ? 10 matrix, containing 10 basic patterns in a 100-point
percentage scale. The ‘*’ denotes matrix multiplication. In
order to be compatible to previous work and to accommo-
date differences in leg length, speed in (2) is expressed as
normalised speed  ˆ v ¼ v=
length and g the acceleration of gravity (Table 2).
, in which l is average leg
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614605
Muscles and electrode positions
Name Electrode position PD
Biceps femoris, long head
Medial and anterior from achilles tendon
Middle of muscle bulge
Middle of muscle bulge
On line between head of fibula and lateral malleolus
Ventral side of lower leg, just lateral from tibia
Anteromedial muscle bulge thigh
Anterolateral muscle bulge thigh
Between VM an VL
Dorsolateral side of thigh
Dorsomedial side of thigh
In fossa poplitea, between tendon of BF and ST
On line between greater trochanter and sacrum
On line between greater trochanter and crista iliaca
On line between tuberculum pubis and medial epicondylus
List of muscles used in this study, with electrode position. PD, approximate location of the electrodes as a proportion of proximo-distal length. For details see
Freriks et al.
First, gain factors Gðm; ˆ v;kÞ were obtained by linear
Gðm; ˆ v;kÞ ¼
p½Eðp;m; ˆ vÞFFðp;kÞ?
and then the speed dependence of each gain factor was fitted
by least squares to obtain
Gðm; ˆ v;kÞ ¼ D0ðm;kÞ þ D1ðm;kÞˆ v þ D2ðm;kÞˆ v2
The whole fitting procedure was done for running speeds
of 2.25 m s?1ðˆ v ¼ 0:72Þ and higher. Each basic pattern
FF(k) was selected outof a representativeEMG profile at the
intermediate speed ˆ v ¼ 1 (3.0 m s?1).
The walking recordings at 1.25, 1.5, and 1.75 m s?1were
comparedtothe predictionsaccording topreviousresults,
obtained at the same speeds but on a different group of
subjects. The patterns were identical within the limits of this
group  but there was some difference in amplitude of the
in the previous results. The present data have been corrected
for this ratio, so that EMG amplitudes from this paper can be
that the predictions for walking were still correct in the range
of 1.75–2.25 m s?1. Stride time and stance duration as a
percentage of stride time have been given in Table 2.
3.1. Walking, running and jogging
Fig. 1. In walking it shows for each leg an M-shaped pattern,
running it isunimodal, with maximumforce atmidstance. At
the higher speeds there may be an airborne phase, with force
zero. Wewill call this case ‘running’, and the casewithout an
airborne phase ‘jogging’. So the proposed division is as
follows, walking: ground reaction force minimum at
midstance, jogging: ground reaction force maximum at
maximum at midstance with aerial phase.
3.2. Grouping of muscles
In running the muscles couldbegroupedinto a number of
functional groups on the basis of their EMG profiles, similar
to walking: (1) a quadriceps group: VM, VL and RF; (2) a
hamstring group: BF, STand SM; (3) a calf group: SO, GM,
GL and PL; (4) a gluteal group: GX and GD. TA and AM are
to be classified separately. Wewill present the results group-
wise, first describe the findings for running and then report
differences with walking and jogging.
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614606
SpeedStride time (s)Stance (% of stride)
MeanS.D.Mean S.D.MeanS.D.Mean S.D.
Normalised speed is calculated as ˆ v ¼ v=
jogging at 1.75 and 2.0 m s?1. In fact the range was between 32% and 60%.
, with l the average leg length of 0.99 m. Data for ‘jogging’ are italic. Note the large S.D. for relative stance in
Fig. 1. Vertical component of ground reaction force, recorded with the
instrumented treadmill, sum of left and right feet. The subject changed from
walkingintojoggingat1.25 m s?1(above)andfromwalkingintorunningat
2.25 m s?1(below). The lower and upper horizontal bars indicate foot
contactof each leg.Thevertical dotted lines correspond with midstance,the
horizontal dotted line equals body weight. In walking, midstance corre-
with a maximum. In jogging the ground reaction force is minimum in
double stance, but it does not reach zero, so there is still ground contact.
3.3. Quadriceps group
This group consists of VM, VL and RF muscles. Vastus
surface EMG. The profiles in running were all very similar,
FF(2), Fig. 3. It starts before foot contact (80%) and ends at
aboutmidstance (15%or115%)witha maximumaround7%
of stance, Fig. 3. While the profiles were identical, the speed
dependence was not: in VM the amplitude hardly changed
more substantial increase, about two-fold. This is reflected in
the entries in D0, D1, and D2(Table 3).
In the vasti the EMG profile was essentially the same for
walking and for running, Fig. 2b. The minor peak around
40% is unique for walking. In jogging the profile was
different from both walking and running. The period of
activity lasted longer, up to 35%, and peak height was lower
than running, see Fig. 2c.
RF has an additional peak from about 40–70%, FF(3). In
fact the onset shifts somewhat earlier, from 47% at
2.25 m s?1to 37% at 4.5 m s?1. The amplitude increases
quadratically with speed, with the result that at 4.5 m s?1it
is higher than the first peak. This peak corresponds to a
similar pattern in walking. At the same speed, the peak in
walking is considerably higher.
3.4. Hamstring group
EMG profiles in the hamstring group, BF, ST and SM,
show two peaks, one in the second half of swing, 70–100%,
and a double-peaked activity in stance, 6–30% (Fig. 4a).
These two basic patterns are represented in FF(4) and FF(5).
The speed dependence is quite different for the three
hamstring muscles, see Fig. 4c. In SM both peaks are
constant, the first peak increased in BF and ST, as did the
second peak in ST. In BF the second peak showed maximum
activity at 3 m s?1, and decreased at higher speeds. The
latter finding is reflected in a negative entry in the quadratic
term D2(Table 3).
Walking showed largely the same two-peaked pattern as
running, but some 10% later, Fig. 4b. The jogging profile
has essentially the timing of walking, but at a higher
amplitude.There is not a gradual shift intiming withspeed;
at 2.25 m s?1the profile just switches to the timing of
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614607
Fig. 2. (a) Quadriceps group. Profiles of rectified and smoothed EMG of VM for running at speeds from 2.25 m s?1(dashed line) to 4.5 m s?1(thick line)
plotted as a percentage of stride time, 0%–100%–50%. The temporal pattern, FF(2), is identical over the range of speeds, in VM the amplitude is constant
as well. Vertical dotted lines show foot contact (0/100%) and the range of toe-off, which is between 37% at 2.25 m s?1and 28% at 4.5 m s?1. (b) EMG
profiles for VM for walking at 2.25 m s?1(dashed line), jogging at 1.25 m s?1(dotted) and running at 2.25 m s?1(drawn line). Profiles for walking and
running are identical. In jogging activity extends over most of stance. Vertical dotted lines correspond to toe-off for running (37%), walking/jogging
(57%), and to foot contact (100%), respectively. (c) EMG amplitude factor as a function of speed for VM in walking (dashed), jogging (dotted) and
running (drawn line). The amplitude increases in walking and jogging, but is largely constant in running. The amplitude in jogging and running is always
higher than in walking.
3.5. Calf group
All four calf muscles, SO, GM, GL and PL, showed a
single peak starting shortly before stance (86%) and ending
before toe-off (125%), represented as FF(1), see Fig. 5a. The
form of this peak is closely alike the quadriceps peak FF(2),
but shifted about 10% later. The amplitudes of SO and PL
were about constant,while GM and GL increased some 40%
over the speed range 2.25–4.5 m s?1, see Table 3.
In this muscle group there were major differences
between walking and running, Fig. 5b. Walking has a major
peak, increasing with speed, in late stance (26–55%) and a
lower and speed-independent activity in early stance, 86–
126%. At the highest walking speed, 2.25 m s?1, the peak
amplitude in walking was higher than in running. Jogging
showed a profile in between: starting at 86% with a similar
the whole stance period, up to about 150%.
3.6. Gluteal group
Both glutei showed a profile with two peaks, Fig. 6a. The
first peak FF(6) is identical, it runs from 88% to 118%. In
GD it is constant in amplitude, while it linearly increases
with speed in GX, Table 3. In GX the second peak FF(7) is at
midswing, 60–84%, in GD it is at the transition from stance
to swing, 30–50%, FF(8). Both increase with speed.
running, both in profile as in amplitude, with the reservation
that the second peak is rather low, and hard to discern. In GD
the profiles of walking and running are identically timed as
well, Fig. 6b, but the amplitude is considerably higher in
jogging and running. In walking and in jogging there is an
additional third peak present, corresponding to FF(7).
3.7. Tibialis anterior
TA activity FF(10) extended over the complete swing
phase, starting slightly before toe-off, around 27%, and
ending abruptly at heel contact (100%), Fig. 7a. The
prominent peak is at 90% and there is a moderate increase
with running speed. In the first half of stance, 0–15%, there
is a minor activity.
The profile in walking also starts just before swing, which
is later at about 60%. The peak activity in walking is later,
corresponding with foot contact, and sharply descending at
10%. At higher walking speeds the TA peak is higher than in
running.Forjoggingat1.25 m s?1activitystartsagainattoe-
no prominent peak. At higher jogging speeds TA activity
always corresponds with toe-off and thus starts earlier.
3.8. Adductor magnus
90%,Fig.8a.Thesepeaksbecomeprominentonlyat3 m s?1
low and irregular, even to the extent that it is hard to see a
periodicity. In walking the pattern is very different from
running, with peaks at foot contact (0%) and toe-off (60%),
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614 608
No. mMuscle Profile (FF) kD0
D-matrices, with coefficients of speed dependent gain factors. Gain factor
g ¼ D0þ ˆ vD1þ ˆ v2D2. Only entries in which at least one of the D’s was
nonzero are included. The muscle from which the profile was taken is
printed bold. When D2= 0, the dependence of profile amplitude with speed
was linear, when both D1= 0 and D2= 0 it is constant.
number k of the FF function is given on the left, the name on the right.
Vertical dotted lines show foot contact (0/100%) and the range of toe-off,
which is between 37% at 2.25 m s?1and 28% at 4.5 m s?1.
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614609
Fig. 4. (a) Hamstrings group. Profiles of rectified and smoothed EMG of BF for running at speeds from 2.25 m s?1(dashed line) to 4.5 m s?1(thick line)
plotted as a percentage of stride time, 0%–100%–50%. The profile shows two peaks, corresponding to FF(4) and FF(5). In BF the first peak increases with
speed, the second is maximal at 3 m s?1. Vertical dotted lines show foot contact (0/100%) and the range of toe-off, which is between 37% at 2.25 m s?1and
28% at 4.5 m s?1. (b) EMG profiles for BF for walking at 2.25 m s?1(dashed line), jogging at 1.25 m s?1(dotted) and running at 2.25 m s?1(drawn line).
Profiles for walking and running are similar, but in running both peaks are about 10% earlier. Vertical dotted lines correspond to toe-off for running (37%),
walking/jogging (57%), and to foot contact (100%), respectively. (c) EMG amplitude factor for pattern FF(4), drawn lines, and pattern FF(5), dashed lines,
for the hamstring muscles as a function of speed for jogging (thin dotted lines) and running. Note that the gain-speed relations are quite different between
muscles and patterns.
Fig. 5. (a) Calf group, triceps surae and peroneus. Profiles of rectified and smoothed EMG of SO for running at speeds from 2.25 m s?1(dashed line) to
4.5 m s?1(thick line) plotted as a percentage of stride time, 0%–100%–50%. The temporal pattern is identical over the range of speeds, in SO there is a minor
increase in amplitude with speed. Vertical dotted lines show foot contact (0/100%) and the range of toe-off, which is between 37% at 2.25 m s?1and 28% at
4.5 m s?1. (b) EMG profiles for SO for walking at 2.25 m s?1(dashed line), jogging at 1.25 m s?1(dotted) and running at 2.25 m s?1(drawn line). Note the
major differences between walking and running: in running peak activity in early stance, in walking in late stance. In jogging the main activity is identical with
running, but the activity is extended to cover most of stance, which is up to 57% at this low speed. Vertical dotted lines correspond to toe-offfor running (37%),
walking/jogging (57%), and to foot contact (100%), respectively.
4.1. Walking, running and jogging
reaction force as the feature which discriminates running
from walking. The conventional criterion is the occurrence
of an airborne phase. As has been observed earlier, this
criterion is not generally valid. When a subject walking on a
treadmill at a moderate walking speed (up to 1.75 m s?1) is
asked tochange into a run,the durationof the stance phase is
so there is no aerial phase. Only at higher speeds stance
decreases below 50%.
The ground reaction force gives a more generally valid
criterion. In walking it shows for each leg an M-shaped
pattern, with a minimum at midstance, while the pattern in
running is unimodal, with maximum force at midstance .
This effect is related to the trajectory of the body centre of
mass, which at midstance shows a maximum in walking and
a minimum in running . In our research we used a
treadmill with built-in force transducers, which simply
enables walk–jog–run detection on the basis of the ground
reaction force, Fig. 1.
Jogging or running at slow speeds can easily be
performed on a treadmill, but the question is whether it
corresponds to a real-life situation. Minetti et al.  found
with a speed-controlled treadmill that subjects do not
voluntarily choose high walking and low running speeds. He
found a maximum walking speed of 2.0 ? 0.2 m s?1and a
minimum running speed of 2.3 ? 0.3 m s?1. This suggests
that running at speeds below 2.25 m s?1is more or less
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614610
Fig.6. (a)Glutealgroup.ProfilesofrectifiedandsmoothedEMGofGDforrunningatspeedsfrom2.25 m s?1(dashedline)to4.5 m s?1(thickline)plottedasa
percentage of stride time, 0%–100%–50%. The GD profile shows two peaks, corresponding to FF(6) and FF(8). In GD the first peak is constant, the second
the range of toe-off, which is between 37% at 2.25 m s?1and 28% at 4.5 m s?1. (b) EMG profiles of GD for walking at 2.0 m s?1(dashed line), jogging at
1.25 m s?1(dotted) and running at 2.25 m s?1(drawn line). Profiles for walking, jogging, and running are the same, with the addition that a minor third peak,
correspondingto FF(7),is visiblein walkingandjogging.Awalking speedof2.0 m s?1was chosen, becausethe profileat2.25 m s?1wasnotrepresentativefor
the lower speeds. Vertical dotted lines correspond to toe-off for running (37%) walking/jogging (57%), and to foot contact (100%), respectively.
Fig.7. (a)Tibialisanterior.ProfilesofrectifiedandsmoothedEMGofTAforrunningatspeedsfrom2.25 m s?1(dashedline)to4.5 m s?1(thickline)plottedas
a percentage of stride time, 0%–100%–50%. EMG activity extends over the completeswing phase, 27–100%,with a peak in final swing at 90%. Vertical dotted
lines show foot contact (0/100%) and the range of toe-off, which is between 37% at 2.25 m s?1and 28% at 4.5 m s?1. (b) EMG profiles of TA for walking at
2.25 m s?1(dashed line), jogging at 1.25 m s?1(dotted) and running at 2.25 m s?1(drawn line). In walking TA activity starts later, in correspondence with the
to foot contact (100%), respectively.
artificial. In this paper we have called it ‘jogging’ (admitting
that this term is not sharply defined) and have considered it
separately from walking and running. This separate
treatment was justified by the deviant EMG patterns, which
reflected the longer stance phase.
As the above suggests, there are a number of
disadvantages in using a treadmill in studies of running.
These do not outweigh the practical advantages. To record a
considerable number of EMGs of steady running for 30 s
would have required an outdoor track and data logging,
although in principle this would have been possible with the
available apparatus. Next to this, it is quite difficult to make
is mechanically equivalent to overground, as long as it is
ensured that the treadmill speed does not fluctuate. In our
brand of treadmill speed was constant within 5%. A second
argument is that the profiles in walking of the present study
were found identical to those of a previous one, which were
recorded on an indoor track.
4.2. EMG profiles
a set of no more than 10 basic patterns, most of which have a
constant timing, and the amplitude of which varies in a
simple way with running speed. This representation
resembles closely the corresponding procedure for walking.
The differences are that walking required 16 basic patterns,
and that walking patterns either were constant or increased
ðˆ v ? 0:16Þ2. In running several intermediate cases were
seen, amplitudes could be constant, increase slowly, or even
decline with speed. Within the calf quadriceps and ham-
string groups, general patterns are closely identical, but the
relative amplitude of the peaks could vary considerably with
speed and within muscles of the group.
ðˆ v ? 0:16Þ
Several of the basic patterns in running (FF) correspond
with those for walking as found earlier (F0, F1 and F2).
Quadriceps FF(2) is identical with F1(2), RF pattern FF(3)
is equal to F2, gluteals FF(6) with F1(4). For the hamstrings
peak 1, FF(5) is very much like F1(3), but activity starts
slightly earlier in running, at 70% instead of 77%.
Hamstrings peak 2, FF(6), is in stance when running (6–
30%), but considerably earlier in walking, F0(3) 91–19%.
The protocol of these experiments is closely equal to that
of Nilsson and Thorstensson . Differences are the
number of muscles studied, 14 versus 7, and the more
quantitative description in the present work. Nilsson et al.
were able to measure at speeds up to 9 m s?1, well into the
sprinting range. As far as could be seen, the results are in
good agreement with the present, also at the highest speeds.
One difference is that Nilsson et al. observed a monophasic
activity in GX, while we saw a biphasic pattern.
A disadvantage of using profiles averaged over a sample
of subjects is that neither step-to-step variations nor
The latter problem has be studied for walking . As to the
step-to-step variability, only a global impression can be
the temporal pattern of the average profile, while the
amplitudes vary some 15–25% per step. Where the average
or cross-talk. In running this on-off contrast is generally
better than in walking, where muscles like PL or GX could
show irregular background activity. AM as an exception,
shows a rather irregular EMG throughout, both in walking
as in running.
4.3. EMGs of hip flexors
In our study we were restricted to surface EMG
recording. This excludes several muscles, especially the
functionally important hip flexors. For these we have to
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614 611
Fig. 8. (a) Adductor magnus. Profiles of rectified and smoothed EMG of AM for running at speeds from 2.25 m s?1(dashed line) to 4.5 m s?1(thick line)
plottedas a percentage ofstride time, 0%–100%–50%. At higherspeeds three peaks canbe discerned,at low speeds activity is low and irregular. Vertical dotted
lines show foot contact (0/100%) and the range of toe-off, which is between 37% at 2.25 m s?1and 28% at 4.5 m s?1. (b) EMG profiles of AM for walking at
2.25 m s?1(dashedline),joggingat 1.25 m s?1(dotted)andrunningat3.0 m s?1(drawnline).Thewalkingprofileshows peaksatfootcontact(0%)andtoe-off
(57%). A running speed of 3.0 m s?1was chosen, because the profile at 2.25 m s?1was too low to show a distinct pattern, see (a).
refer to studies where indwelling (fine-wire) electrodes
have been used.
Nene et al.  investigated RF activity in walking with
fine-wire electrodes. Their results clearly showed that the
second peak FF(3)/F2, in early swing, was RF activity
proper. The first ‘quadriceps’ peak FF(2)/F1(2), as found in
surface EMG, did not correspond to activity of RF, but was
probably cross-talk from the underlying m. vastus inter-
An extensive study of the hip flexors, iliacus, psoas,
sartorius, rectus femoris and tensor fasciae latae, was
undertaken by Andersson et al. . In running all hip
flexors were active from about 30–65%, RF starting slightly
later, 45–65%. Psoas has a second burst in late swing, 80–
100%. Tensor fasciae latae seems not a hip flexor, showing
iliacus and psoas sharply increased with running speed,
about three-fold over the range of 2–4 m s?1. Both timing
and amplitude agreewith our findings for the second peak of
RF. The results of Andersson and Nene, taken together,
(knee extensor) but as a hip flexor.
4.4. Similarities and differences between walking,
jogging, and running
The EMG profiles in walking and running show many
similarities. GX shows essentially the same profile and
amplitude in walking as in running. The quadriceps group
and GD have identical profiles, but a considerably higher
amplitude in running (Figs. 2c and 6b). The hip flexors are
reported to have almost identical profiles but lower
amplitudes at the same speed in running. The walking
and running profiles for the hamstrings show minor
differences. TA timing is adapted to the longer swing
duration in running and the final peak is now before foot
The major difference between walking and running is
found in the calf muscles, with the main peak 26–55% in
walking and much earlier, 86–125% in jogging and running.
In all cases muscles from this group are active over the
complete stance, so there is additional activity 0–26% in
walking and 25–50% in jogging, Fig. 5b. In fast running
stance duration is so short that no additional activity is
4.5. Muscle function in running versus walking
Muscle function in walking has been discussed exten-
sively in the past [15–17] and to a lesser degree for running
to a few issues.
Two major actions can be discriminated in running: a
stance action and an alternating hip flexion–extension
action. The stance action encompasses a more or less
simultaneous activation burst of all leg extensors: glutei,
quadriceps and calf, FF(1, 2, 6). With this simultaneous
activation, and with hip, knee and ankle all slightly flexed,
of the trunk and in the second half it extends again, so the
major energy exchange is a storage and release of elastic
energy [20,21]. Due to active muscle work the energy
increases in the course of the process .
The activation bursts are not completely simultaneous.
Quadriceps is first, followed by the calf muscles. This is
probably related to the kinematics: maximum knee flexionis
earlier than maximum ankle dorsiflexion, which can be
explained by the combination of leg shortening–lengthening
and forward progression movement. The activations all start
before foot contact, to achievea landing with a stiff leg. This
part of the extensor activation goes along with a co-
contraction of the hamstrings for the knee, FF(4) and of TA
to 100% may serve the same purpose. All extensor bursts
end before toe-off, but muscle force continues for sufficient
time after the end of activation to cover the complete stance
phase, see Appendix B in Ref. . The EMG amplitudes in
calf and quadriceps muscles are high, but increase only little
with speed, in GX somewhat more.
The major difference between walking and running
profiles was present in the calf group. In running the
activation bursts of quadriceps and calf are more or less
simultaneous, in walking the major calf muscle burst is at
the end of stance. The quadriceps action is more or less the
same, both in walking as in running there is a knee
flexion–extension movement together with quadriceps
activity at initial stance. In walking the activation is less,
however, so that no aerial phase results, cf. Fig. 2c. This is
reflected in much lower knee extension moments, 30 N m–
40 N m–80 N m in slow–normal–fast walking , com-
pared to running , 210 N m at 2.72 m s?1. The ankle
moment is also low at that time, because the ground
reaction force is close to the ankle. After midstance the
knee remains straight, so knee moment can be low, but
now the ankle moment increases to a maximum at push-
off. Shortly, in running quadriceps and calf work together
in absorbing and generating energy, by way of a single
elastic bounce, while in walking impact absorption and
push-off are wider separated in time and done separately
by quadriceps and calf. A consequence is that in running
quadriceps and calf both produce upward force to the
trunk. In walking calf muscle force is directed more
horizontally, contributing power not only to the trunk but
also to the swing leg.
by psoas, iliacus and RF, around the instant of maximal hip
flexion at 50%, and extension by the glutei, which starts at
bothinthis action andin the stance action.Incontrast tothe
stance actions of quadriceps and calf, the amplitudes of
the hip flexors and GX show a major increase with speed:
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614 612
two-fold in GX and four-fold in psoas and iliacus between
speeds of 2.25 and 4.5 m s?1. This suggests that speed
increase in running is mainly accomplished by a larger leg
swing due to increased hip flexor and extensor action.
Next to these major actions, there are the actions of
hamstrings and TA. Hamstrings act late in swing, and
probably their action in running is the same as in walking,
viz. the arrest of leg swing. TA activity is needed to keep the
foot in the neutral position against gravity and passive
triceps surae elasticity. The final peak at terminal swing
(90%) is a co-contraction against triceps surae. Similarly,
final hamstring activity is simultaneous with the onset of
For the whole stance duration the extensors must be
active to prevent collapse of the leg. In fast running, the
stance burst is sufficiently long to achieve this effect. In
walking and jogging the stance period is considerably
longer. In these cases we see a continuous activity over
stance of the calf muscles, especially SO, Fig. 5b. In
jogging this holds also of quadriceps, Fig. 2b. In walking
the latter activity is less needed, because the knee is almost
straight. The stance activity of hamstrings (FF5) serves
possibly hip extension, but this has to be investigated in
more detail. The same holds for the low TA activity in
4.6. Central pattern generator
The finding of invariant EMG profiles is in agreement
with current ideas on a central pattern generator (CPG) for
human locomotion . Supportivefindings are a.o. the pre-
activations in FF(1, 2, 6) before foot contact. Next, there is
the remarkable finding that the timing of the EMG profiles is
constant,whiletoe-offgoesfrom37%at2.25 m s?1downto
27%at4.5 m s?1.Thesameeffectwas found inwalking, but
the change in relative stance duration is less in that case.
Exceptions are TA and RF (FF(3)), the onsets of which are
clearly linked to the start of swing. The deviant patterns in
jogging, compared to running, are an other exception. In
jogging stance is so much prolonged, that a prolonged
activity in quadriceps and calf muscles is needed to ensure
originate from reflexes, e.g. TA during swing and the
continuous activity in the calf muscles during early
(walking) or late (jogging) stance, which can be ascribed
to a reflex in response to foot loading (extensor reinforcing
reflex) . The start of swing seems to be initiated by hip
needed anyway to ensure that the CPG remains in pace with
the movement . To determine which actions are reflexes
and which are fully programmed is a research program in
itself, for which the present data can only give very indirect
support. The major outcome of the present research so may
just be to provide an accessible database for EMG data in
Appendix A. Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.gaitpost.
 Hof AL, Elzinga H, Grimmius W, Halbertsma JPK. Speed depen-
dency of averaged EMG profiles in walking. Gait Posture 2002;16:
in humans. Acta Physiol Scand 1985;123:457–75.
 Sasaki K, Neptune RR. Differences in muscle function during walking
and running at the same speed. J Biomech 2006;39:2005–13.
 Freriks B, Hermens H, Disselhorst-Klug C, Rau G. The recommenda-
tions for sensors and sensor placement procedures for surface
electromyography. In: Hermens HJ, editor. European recommenda-
tions for surface electromyography. Enschede: Roessingh Research
and Development; 1999. p. 15–53.
 Verkerke GJ, Hof AL, Zijlstra W, Ament W, Rakhorst G. Determining
the centre of pressure during walking and running using an instru-
mented treadmill. J Biomech 2005;38:1881–5.
 Hof AL. Scaling gait data to body size. Gait Posture 1996;4:
 Hof AL, Elzinga H, Grimmius W, Halbertsma JPK. Detection of non-
standard EMG profiles in walking. Gait Posture 2005;21:171–7.
of human walking and running. Acta Physiol Scand 1989;136:217–
 Alexander RM. Walking and running. Am Sci 1984;72:348–54.
 Minetti AE, Boldrini L, Brusamolin L, Zamparo P, McKee T. A
feedback-controlled treadmill (treadmill-on-demand) and the sponta-
neous speed of walking and running in humans. J Appl Physiol 2003;
of overground versus treadmill locomotion. Med Sci Sports Exercise
 NilssonJ, Thorstensson A.Adaptability in frequency andamplitudeof
leg movements during human locomotion at different speeds. Acta
Physiol Scand 1987;129:107–14.
 Nene A, Byrne C, Hermens H. Is rectus femoris really a part of
quadriceps?Assessmentof rectus femorisfunctionduring gait in able-
bodied adults. Gait Posture 2004;20:1–13.
 Andersson EA, Nilsson J, Thorstensson A. Intramuscular EMG from
the hip flexor muscles during human locomotion. Acta Physiol Scand
 Inman VT, Ralston HJ, Todd F. Human walking Baltimore: Williams
and Wilkins; 1981.
 Zajac FE, Neptune RR, Kautz SA. Biomechanics and muscle coor-
dination of human walking. Part II. Lessons from dynamical simula-
 Zajac FE, Neptune RR, Kautz SA. Biomechanics and muscle
coordination of human walking. Part I. Introduction to concepts,
power transfer, dynamics and simulations. Gait Posture 2002;16:
 Nilsson J. On the adaptation to speed and mode of progression in
human locomotion. Thesis. Stockholm; 1990.
 Farley CT, Ferris DP. Biomechanics of walking and running: center of
mass movements to muscle action. Exercise Sports Sci Rev 1998;
 Alexander RM, Bennett-Clark HC. Storage of elastic strain energy in
muscle and other tissues. Nature 1977;265:114–7.
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614613
 Cavagna GA. Storage and utilization of elastic energy in skeletal
muscle. Exercise Sports Sci Rev 1977;5:89–129.
 Hof AL, van Zandwijk JP, Bobbert MF. Mechanics of human triceps
surae muscle in walking, running and jumping. Acta Physiol Scand
 Hof AL. Muscle mechanics and neuromuscular control. J Biomech
 Winter DA. The biomechanics and motor control of human gait, 2nd
ed., Waterloo, Canada: University of Waterloo Press; 1991.
 Winter DA. Moments of force and mechanical power in jogging. J
 Duysens J. Human gait as a step in evolution. Brain 2002;125:2589–
 Hiebert GW, Whelan PJ, Prochazka A, Pearson KG. Contribution of
hind limb flexor muscle afferents to the timing of phase transitions in
the cat step cycle. J Neurophysiol 1996;75:1126–37.
 Kuo AD. The relative roles offeedforward and feedback in the control
of rhythmic movements. Motor Contr 2002;6:129–45.
M.G.J. Gazendam, A.L. Hof/Gait & Posture 25 (2007) 604–614 614