Content uploaded by Robert N. Marshall
Author content
All content in this area was uploaded by Robert N. Marshall
Content may be subject to copyright.
GRF and Kinematics of Sprint Running
31
31
JOURNAL OF APPLIED BIOMECHANICS, 2005, 21, 31-43
© 2005 Human Kinetics Publishers, Inc.
1
Dept. of Sport and Exercise Science, The University of Auckland, Auckland, New
Zealand;
2
Faculty of Health and Sport Science, Eastern Institute of Technology, Hawkes
Bay, NZ;
3
Physical Rehabilitation Research Centre, School of Physiotherapy, Auckland
University of Technology, Auckland, NZ.
Relationships Between Ground Reaction
Force Impulse and Kinematics
of Sprint-Running Acceleration
Joseph P. Hunter
1
, Robert N. Marshall
1,2
, and Peter J. McNair
3
1
The University of Auckland;
2
Eastern Institute of Technology;
3
Auckland University of Technology
The literature contains some hypotheses regarding the most favorable ground
reaction force (GRF) for sprint running and how it might be achieved. This
study tested the relevance of these hypotheses to the acceleration phase of a
sprint, using GRF impulse as the GRF variable of interest. Thirty-six athletes
performed maximal-effort sprints from which video and GRF data were col-
lected at the 16-m mark. Associations between GRF impulse (expressed rela-
tive to body mass) and various kinematic measures were explored with simple
and multiple linear regressions and paired t-tests. The regression results showed
that relative propulsive impulse accounted for 57% of variance in sprint ve-
locity. Relative braking impulse accounted for only 7% of variance in sprint
velocity. In addition, the faster athletes tended to produce only moderate mag-
nitudes of relative vertical impulse. Paired t-tests revealed that lower magni-
tudes of relative braking impulse were associated with a smaller touchdown
distance (p < 0.01) and a more active touchdown (p < 0.001). Also, greater
magnitudes of relative propulsive impulse were associated with a high mean
hip extension velocity of the stance limb (p < 0.05). In conclusion, it is likely
that high magnitudes of propulsion are required to achieve high acceleration.
Although there was a weak trend for faster athletes to produce lower magni-
tudes of braking, the possibility of braking having some advantages could not
be ruled out. Further research is required to see if braking, propulsive, and
vertical impulses can be modified with specific training. This will also pro-
vide insight into how a change in one GRF component might affect the others.
Key Words: braking impulse, propulsive impulse, vertical impulse
The acceleration of the center of mass of a sprinter is determined by three
external forces: ground reaction force (GRF), gravitational force, and wind resis-
Hunter, Marshall, and McNair32
tance (Figure 1). Of these three forces, the athlete has by far the most influence
over the GRF. For analysis purposes, the GRF can be broken down into its three
orthogonal components. In the case of sprint running, the horizontal (anterior-
posterior) and vertical components are typically of most interest. Also of interest
are the two subcomponents of the anterior-posterior horizontal GRF: a braking
GRF acts posteriorly and usually occurs early in the stance phase, while a propul-
sive GRF acts anteriorly and usually occurs later in the stance phase.
The literature on sprint running contains a number of hypotheses regarding
the various GRF components. It has been recommended that sprinters should mini-
mize the braking GRF (Mero & Komi, 1986; Mero, Komi, & Gregor, 1992; Wood,
1987) and maximize the propulsive GRF (Mero et al., 1992). Furthermore, it has
been suggested that, at maximal sprint velocity, the ability to produce a high, aver-
age vertical GRF in a short stance time is of advantage (Weyand, Sternlight, Bellizzi,
& Wright, 2000).
The literature also contains hypotheses on how a sprinter can reduce the
braking GRF and increase the propulsive GRF. The braking GRF is thought to be
reduced by the following: using a highly active touchdown (i.e., minimizing the
forward horizontal velocity of the foot, relative to the ground, immediately before
ground contact) (Hay, 1994, pp. 407-408; Mann & Sprague, 1983; Wood, 1987);
ensuring a high extension velocity of the hip joint and a high flexion velocity of
the knee joint at the instant of touchdown (Mann, Kotmel, Herman, Johnson, &
Schultz, 1984; Mann & Sprague, 1983); and minimizing touchdown distance (i.e.,
the distance the foot is placed in front of the center of mass at the instant of touch-
down) (Mero et al., 1992). In contrast, the propulsive GRF is thought to be maxi-
mized by the following: ensuring a high angular velocity of the stance-limb hip
joint (Mann & Sprague, 1983; Mann et al., 1984; Wiemann & Tidow, 1995); and
fully extending the stance-limb hip, knee, and ankle joints at takeoff (see Hay,
1994, pp. 408-409 for discussion).
Figure 1 — The three external forces that determine the acceleration of a sprinter’s
center of mass: ground reaction force (GRF), gravitational force (equivalent to body
weight, BW), and wind resistance.
GRF and Kinematics of Sprint Running
33
These hypotheses, of which many have yet to be fully tested, were probably
intended to have most relevance to the maximal-velocity phase of a sprint. In con-
trast, we were interested in their possible relevance to the acceleration phase. Fur-
thermore, instead of focusing on the magnitude of the GRF as such, we wanted to
focus on GRF impulse. GRF impulse is an informative measure because, when
expressed relative to body mass, it reflects the change in velocity of the athlete (if
the effect of wind resistance is ignored).
Consequently, there were two purposes to this study. First, to determine the
relationships between relative GRF impulse (“relative” indicating relative to body
mass) and sprint velocity during the acceleration phase of a sprint. Second, to test
the above stated hypotheses regarding the techniques to minimize braking and
maximize propulsion, but from a GRF impulse perspective.
Methods
A total of 36 participants (31 M, 5 F) were tested for this study. All the men partici-
pated in sports involving sprint running (e.g., track and field, soccer, touch rugby)
and all the women were track-and-field athletes. Mean ± SD for age, height, and
body mass of the entire group were 23 ± 5 yrs, 1.76 ± 0.07 m, and 72 ± 8 kg,
respectively. However, for the purpose of investigating group relationships, the
entire group was considered too heterogeneous. Subsequently only the 28 fastest
men (intended to represent a population of athletic men of average to very good
sprint ability) were used for all regression analyses. The mean ± SD for age, height,
and body mass of these 28 men were 22 ± 4 yrs, 1.77 ± 0.06 m, and 74 ± 6 kg,
respectively. Note, though, that in other analyses in which each athlete was com-
pared to him/herself (via paired t-tests), the entire group of 36 athletes was consid-
ered. Approval to undertake the study was given by The University of Auckland
Human Subjects Ethics Committee. Written informed consent was obtained from
each athlete.
A detailed description of the data collection and data treatment is provided
elsewhere (Hunter, Marshall, & McNair, 2004). Therefore only an overview is
provided here. After warming up and being prepared with joint markers, each ath-
lete performed maximal-effort sprint-running trials, 25 m in length, on a synthetic
track in which a force plate (Bertec 6090s; Bertec Corp., Columbus, OH) was
embedded. The sprints were performed from a standing start, and the athletes wore
spiked track shoes. EVa 6.15 data collection system (Motion Analysis Corp., Santa
Rosa, CA) was used to collect sagittal-plane video data (sampled at 240 Hz) and
GRF data (sampled at 960 Hz) of a stride at the 16-m mark of the sprints. Success-
ful trials were those in which the athlete clearly contacted the force plate without
adjusting his or her natural running pattern. For this to occur, the sprint start line
was adjusted by no more than 1 m. The foot to contact the force plate was the foot
that was placed forward during the standing start. Typically, each athlete performed
about 7 or 8 sprints which usually resulted in 4 or 5 successful trials (the range was
3 to 6 successful trials). There was a rest period of about 4 minutes between sprints.
The human body was modeled as 12 segments: feet, shanks, thighs, trunk,
head (including neck), upper arms, and lower arms (including hands). Segment
inertia parameters were obtained from de Leva (1996), with the exception of the
foot’s center of mass location which was obtained from Winter (1990). The data
were filtered with a low-pass Butterworth digital filter (Winter, 1990). Kinematic
Hunter, Marshall, and McNair34
data were filtered with cutoff frequencies ranging from 7 Hz for upper-trunk markers
to 12 Hz for foot markers. GRF data were filtered with a cutoff frequency of 75 Hz.
The instants of touchdown and takeoff from the force plate were defined as
when the vertical GRF first rose above 10 N (touchdown) and reduced to 25 N
(takeoff). The instant of touchdown for the first ground contact beyond the force
plate was assumed to occur at the instant of peak vertical acceleration of the head
of the 2nd metatarsal (Hreljac & Marshall, 2000).
The following variables were calculated from the kinematic data: (a) Sprint
velocity: mean horizontal velocity of the center of mass during the step at the 16-m
mark. (b) Hip joint kinematics: angular velocity at the moment of touchdown,
mean and maximum angular velocities during stance, and the angle at the moment
of takeoff (see Figure 2). (c) Knee joint kinematics: angular velocity at the moment
of touchdown, and angle at the moment of takeoff (Figure 2). (d) Ankle joint kine-
matics: just one measurement, the angle at the moment of takeoff (Figure 2). For
this measurement the foot was represented as a link from the posterior surface of
the calcaneous to the head of the 2nd metatarsal. (e) Horizontal velocity of the foot
before touchdown: horizontal velocity, relative to the ground, of the head of the
2nd metatarsal, four frames (0.017 s) before touchdown. The lower the horizontal
velocity of the foot, the more active the touchdown. That is, the athlete actively
attempts to move the foot backward as fast as he or she is moving forward. (f) Leg
angle at touchdown: measured between horizontal and a line passing through the
stance ankle and center of mass, at the moment of touchdown (Figure 2). This
angle was used as a measure of the horizontal distance the foot was placed in front
of the center of mass at the moment of touchdown (i.e., touchdown distance).
In addition, four measures of GRF impulse (hereafter referred to as impulse)
were calculated from the force-plate data: relative vertical impulse, relative hori-
zontal impulse, relative braking impulse, and relative propulsive impulse. The term
Figure 2 — Angles measured at touchdown and takeoff. The leg angle at touchdown
(TD) was measured between horizontal and a line passing through the stance ankle
and the body’s center of mass at the moment of touchdown. Hip (h), knee (k), and
ankle (a) joint angles of the stance-limb were measured at the moment of takeoff.
GRF and Kinematics of Sprint Running
35
“relative” has been used to indicate that the impulses were expressed relative to
body mass. Relative propulsive impulse was based on all horizontal positive force
data during stance, and relative braking impulse was based on all horizontal nega-
tive force data during stance. Further details of how the relative impulses were
calculated are given in Figure 3.
Statistical analyses were performed in SPSS (release 10.0.5, SPSS Inc., Chi-
cago). For all regression analyses, the means of the fastest three trials were used.
For all paired t-tests, data from individual trials were used.
To determine the relationships between sprint velocity and the relative im-
pulses, we performed four simple (bivariate) linear regressions with sprint veloc-
ity as the dependent variable and each relative impulse as the independent variable.
From the resulting regression equations, the influence of each relative impulse on
Figure 3 — Ground reaction force (GRF) impulses are shown as areas under the GRF
curves. (a) p is the propulsive impulse, b is the braking impulse. Propulsive impulse was
based on all horizontal positive force data during stance, and braking impulse was based
on all horizontal negative force data during stance. Horizontal impulse was calculated as
propulsive impulse less the absolute value of braking impulse. (b) v is the area under the
vertical GRF curve, and BW impulse is the impulse due to body weight. Vertical impulse
was calculated as v – BW impulse. When horizontal, braking, propulsive, and vertical
impulses are expressed relative to body mass, they reflect the change in velocity of the
center of mass (ignoring the effects of wind resistance) during the respective periods and in
the respective directions.
Hunter, Marshall, and McNair36
sprint velocity was assessed by calculating the predicted increase in sprint velocity
associated with a one standard deviation increase in relative impulse. In addition,
we performed a stepwise multiple linear regression with sprint velocity as the de-
pendent variable and relative vertical, braking, and propulsive impulses as the
independent variables. The criterion for entry into the multiple regression model
was p < 0.05, and the criterion for removal was p > 0.10. Alpha was set at 0.05 for
all other statistical tests.
The hypotheses regarding techniques to minimize braking (see introductory
section) were assessed with paired t-tests. This involved selecting, from each ath-
lete, two trials that clearly differed with regard to the magnitude of relative brak-
ing impulse, and then using paired t-tests on these two trials to detect differences
in variables related to the braking hypotheses. That is, we wanted to know if a
difference in relative braking impulse was associated with a difference in other
variables of interest. For this analysis we were aware of two main requirements:
(a) we needed a sample size large enough to ensured acceptable statistical power;
and (b) we needed to exclude athletes for which relative braking impulse did not
clearly differ (i.e., there was no point in testing for differences in the other vari-
ables if relative braking impulse itself did not differ).
These two requirements were met by using the following method. From the
fastest three trials of each athlete, the trial with the greatest magnitude of relative
braking impulse was named the High Braking Trial, and the trial with the lowest
magnitude was named the Low Braking Trial. If the difference between these two
trials was less than 0.010 m/s, then that athlete was excluded from the analysis.
According to this criterion, 6 athletes were excluded, thereby leaving 28 athletes
(24 M and 4 F) included in the analysis. Paired t-tests were then used to contrast
the High vs. Low Braking Trials for the following variables: sprint velocity, rela-
tive impulses, hip and knee joint angular velocities at touchdown, horizontal ve-
locity of the foot before touchdown, and leg angle at touchdown.
The hypotheses regarding techniques for maximizing propulsion were also
assessed with paired t-tests. The within-subject variation of relative propulsive
impulse (coefficient of variation of 4%) was smaller than that of relative braking
impulse (coefficient of variation of 14%); however, the requirements of an accept-
able sample size, and exclusion of athletes for which relative propulsive impulse
did not clearly differ, could still be met using the following method. From all trials
of each athlete (not the fastest three, as used for the braking analysis), the trial with
the greatest magnitude of relative propulsive impulse was named the High Propul-
sion Trial, and the trial with the lowest magnitude was named the Low Propulsion
Trial. If the difference between these two trials was less than 0.015 m/s, then that
athlete was excluded from the analysis. According to this criterion, 6 athletes were
excluded, thereby leaving 28 athletes (25 M and 3 F) in the analysis. Paired t-tests
were then used to contrast the High vs. Low Propulsion Trials for the following
variables: sprint velocity, relative impulses, mean and maximum hip joint exten-
sion velocities during stance, and hip, knee, and ankle angles at takeoff.
According to the methods of Cohen (1977), 28 athletes provided 80% power
in detecting a correlation coefficient of 0.50. Also, paired t-tests with 28 athletes
and an expected test-retest correlation of 0.80 provided more than 70% power in
detecting an effect size of 0.3.
GRF and Kinematics of Sprint Running
37
Results
Table 1 shows the means and standard deviations of sprint velocity and relative
impulses of the 28 male athletes used in all regression analyses. Figure 4 shows
the results of the four simple linear regression analyses. The strongest predictor of
sprint velocity was relative horizontal impulse (R
2
= 0.61, p < 0.001). A one stan-
dard deviation increase (0.04 m/s) in relative horizontal impulse resulted in a pre-
dicted increase of 0.26 m/s in sprint velocity. The next strongest predictor of sprint
velocity was relative propulsive impulse (R
2
= 0.57, p < 0.001). A one standard
deviation increase (0.04 m/s) in relative propulsive impulse also resulted in a pre-
dicted increase of 0.26 m/s in sprint velocity (i.e., an amount identical to the previ-
ous example). The linear relationship between relative vertical impulse and sprint
velocity was comparatively weak but significant (R
2
= 0.17, p < 0.05). However,
this relationship showed possible departure from linearity. The 4 fastest athletes
had only moderate magnitudes of relative vertical impulse (ranging from 0.96 to
1.03 m/s). Nonetheless, if using the linear regression equation, a one standard de-
viation increase (0.10 m/s) in relative vertical impulse resulted in a predicted in-
crease of 0.14 m/s in sprint velocity (i.e., approx. half of the previous two examples).
The simple linear regression between relative braking impulse and sprint velocity
was not statistically significant (R
2
= 0.04, p > 0.05).
The multiple linear regression to predict sprint velocity resulted in relative
propulsive impulse being the first variable to enter the model, and explained 57%
(R
2
= 0.57, p < 0.001) of the variance in sprint velocity. Relative braking impulse
was the next variable to enter the model and explained a further 7% (R
2
increase =
0.07, p < 0.05) of the variance. That is, the total variance in sprint velocity ex-
plained by relative propulsive impulse and relative braking impulse was 64% (to-
tal R
2
= 0.64, p < 0.001). Relative vertical impulse did not explain any further
variance in sprint velocity and thus was not included in the model. The regression
equation to predict sprint velocity (in m/s) was… velocity = 7.15·p + 4.12·b + 6.18
…where p is the relative propulsive impulse and b is the relative braking impulse,
both measured in m/s. Note that relative braking impulse was quantified as a nega-
tive value to indicate its posterior direction. Therefore, the multiple linear regres-
Table 1 Sprint Velocity and Relative Impulses of the 28 Male Athletes Used in All
Regression Analyses
Variables Mean SD
Sprint velocity (m/s)
a
8.29 0.34
Relative vertical impulse (m/s) 0.99 0.10
Relative horizontal impulse (m/s) 0.25 0.04
Relative braking impulse (m/s) –0.10 0.02
Relative propulsive impulse (m/s) 0.35 0.04
a
The range in sprint velocity was 7.44 to 8.80 m/s.
Hunter, Marshall, and McNair38
Figure 4 — Simple linear regressions between sprint velocity and relative impulses.
The relationship between sprint velocity and relative vertical impulse (a) shows possible
departure from linearity. A nonlinear function, such as a quadratic equation (dashed
line, R
2
= 0.24), is arguably a better description of this relationship. *p < 0.05, ***p <
0.001
sion revealed that greater magnitudes of relative braking impulse had a negative
effect on sprint velocity, but greater magnitudes of relative propulsive impulse had
a positive effect.
Table 2 shows the results of the paired t-test used to analyze the hypotheses
regarding techniques to minimize braking. Sprint velocity did not differ signifi-
cantly, suggesting that fatigue was an unlikely cause of any of the other significant
differences. The Low Braking Trials had a significantly lower relative vertical
impulse, but relative propulsive impulse did not differ significantly. Of the hy-
potheses regarding minimization of braking, use of an active touchdown (i.e., low
horizontal velocity of the foot before touchdown), and a small touchdown distance
(i.e., large leg angle at touchdown) were statistically supported.
Table 3 shows the results of the paired t-test used to analyze the hypotheses
regarding techniques for maximizing propulsion. Sprint velocity did not differ sig-
GRF and Kinematics of Sprint Running
39
Table 2 Paired t-test Results for Assessment of Hypotheses Regarding
Minimization of Braking
High Braking Trials Low Braking Trials
Mean SD Mean SD
Sprint Velocity and Impulses
Sprint velocity (m/s) 8.12 0.44 8.11 0.44
Relative vertical impulse (m/s)*** 1.07 0.14 0.98 0.12
Relative braking impulse (m/s)***
a
–0.12 0.03 –0.09 0.03
Relative propulsive impulse (m/s) 0.34 0.03 0.35 0.04
Hypotheses
Hip joint extension velocity at
touchdown (deg/s) 388 93 396 75
Knee joint flexion velocity at
touchdown (deg/s)
b
–147 118 –150 88
Horizontal velocity of foot before
touchdown (m/s)*** 2.43 0.83 2.12 0.76
Leg angle at touchdown (deg)** 80 3 81 3
Note:
a
Negative value
indicates
that the impulse acted against the direction of progression
of the sprinter;
b
negative value indicates flexion. 24 male and 4 female athletes were
included in this analysis. **p < 0.01; ***p < 0.001
nificantly, suggesting that fatigue was an unlikely cause of any of the other signifi-
cant differences. The High Propulsion Trials had a significantly lower magnitude
of relative braking impulse, but relative vertical impulse did not differ signifi-
cantly. Of the hypotheses regarding maximization of propulsion, a greater mean
hip joint extension velocity during stance was the only hypothesis that had statis-
tical support.
Discussion
The GRF acting on a sprinter is obviously a major determinant of sprint running
performance. However, as discussed in the introductory portion, there are numer-
ous hypotheses regarding the relative importance of the various GRF components,
and how they might be altered to achieve a better sprint performance. Many of
these hypotheses were probably originally intended to be most applicable to the
maximal-velocity phase of a sprint. We were interested, however, in their applica-
bility to the acceleration phase. Furthermore, we chose to focus not on the magni-
tude of the GRF as such, but instead on GRF impulse.
To investigate these hypotheses, we measured sprint velocity and GRF at
the 16-m mark of a sprint. Sprint velocity at the 16-m mark is a product of sprint
performance over that entire distance, whereas the GRF at the 16-m mark is that of
a single stance phase. Nonetheless, the GRF at that mark is likely to be somewhat
Hunter, Marshall, and McNair40
representative of an athlete’s ability to apply GRF during at least some of the
previous stance phases. The result of a strong relationship between relative hori-
zontal impulse and sprint velocity supported this notion.
The first purpose of this study was to determine the relationships between
the relative impulses and sprint velocity during the acceleration phase of a sprint.
To examine these relationships we used both simple and multiple linear regres-
sions. The following paragraphs contain discussion on the relationships of relative
propulsive, braking, and vertical impulses with sprint velocity.
Both the simple and multiple regression results showed a relatively strong
trend for faster athletes to produce greater magnitudes of relative propulsive im-
pulse. This finding was expected because athletes with good acceleration ability
would likely undergo larger increases in horizontal velocity during each stance
phase. This finding is also in agreement with the research of Mero and Komi (1986),
who reported a positive relationship between average resultant GRF during pro-
pulsion and sprint velocity between the 35-m and 45-m marks (r = 0.84, or r = 0.65
when the resultant GRF was expressed relative to body weight).
A simple linear regression did not support the existence of a relationship
between sprint velocity and relative braking impulse. In contrast, the multiple lin-
ear regression showed a weak trend for faster athletes to produce lower magni-
tudes of relative braking impulse. These conflicting results, we believe, highlight a
weakness in using simple linear regression alone to explore relationships among
Table 3 Paired t-test Results for Assessment of Hypotheses Regarding
Maximization of Propulsion
High Propulsion Trials Low Propulsion Trials
Mean SD Mean SD
Sprint Velocity and Impulses
Sprint velocity (m/s) 8.15 0.47 8.12 0.45
Relative vertical impulse (m/s) 1.01 0.11 1.00 0.14
Relative braking impulse (m/s)*
a
–0.10 0.03 –0.11 0.03
Relative propulsive impulse (m/s)*** 0.37 0.03 0.33 0.04
Hypotheses
Mean hip joint extension velocity
during stance (deg/s)* 570 61 558 58
Max. hip joint extension velocity
during stance (deg/s) 767 72 759 72
Hip joint angle at takeoff (deg) 198 4 198 5
Knee joint angle at takeoff (deg) 163 5 162 5
Ankle joint angle at takeoff (deg) 116 5 116 6
a
Negative value indicates that the impulse acted against the sprinter’s direction of
progression. 25 male and 3 female athletes were included in this analysis.
*p < 0.05; ***p < 0.001
GRF and Kinematics of Sprint Running
41
variables. In the multiple linear regression, relative propulsive impulse suppressed
irrelevant variance (Tabachnick & Fidell, 2001, pp. 148-149), thereby exposing
the relationship between relative braking impulse and sprint velocity. The practi-
cal significance of this finding, however, is arguable.
The hypothesis stating that braking should be minimized is reasonably popular
with some researchers (Mero & Komi, 1986; Mero et al., 1992; Wood, 1987), and
is based on the premise that a lower magnitude of braking would result in a smaller
loss in horizontal velocity early in the stance phase. However, lack of evidence has
led some researchers to advise caution. Putnam and Kozey (1989), for instance,
warned that the braking force might be related to other important mechanical fac-
tors of performance. For example, the braking force could be involved in the stor-
age of elastic energy (Cavagna, Komarek, & Mazzoleni, 1971). In summary,
although we did find that relative braking impulse accounted for a small propor-
tion (7%) of variance in sprint velocity, we do not know if the faster athletes actu-
ally minimized their magnitude of braking, and we cannot rule out that braking
might also have some advantages. Further research is required to examine these
issues.
The relationship between relative vertical impulse and sprint velocity showed
signs of nonlinearity. The fastest athletes tended to produce only moderate magni-
tudes of relative vertical impulse, about 1 m/s. We speculate that, during the accel-
eration phase of a sprint, the most favorable magnitude of relative vertical impulse
is one that creates a flight time only just long enough for repositioning of the lower
limbs. If the athlete can reposition the limbs quickly, then a lower relative vertical
impulse is sufficient, and all other strength reserves should be applied horizon-
tally. However, if an athlete cannot achieve or maintain a high step rate, such as
when fatigued, then a greater relative vertical impulse becomes more important.
This speculation differs somewhat from the view of Weyand et al. (2000),
who proposed that a faster maximal velocity is achieved by applying a greater
vertical GRF. We agree that faster sprinters have less time to apply the GRF, and
therefore will require a greater force than usual. However, we believe that during
the acceleration phase of a sprint, a large relative vertical impulse (i.e., notably
greater than 1 m/s) will not be advantageous. Actually, a long flight time, deter-
mined by a large relative vertical impulse, may be a disadvantage. This would be
due to a decrease in the percentage of time spent in contact with the ground. An
athlete can only influence his or her sprint velocity when in contact with the ground.
This topic remains an interesting area for future research.
The second purpose of this study was to test the hypotheses regarding the
techniques for minimizing braking and maximizing propulsion. With regard to the
hypotheses stating how braking can be minimized, use of a high hip extension
velocity and high knee flexion velocity at touchdown were not supported. In con-
trast, the use of a more active touchdown (i.e., smaller horizontal velocity of the
foot before touchdown) and a smaller touchdown distance (larger leg angle at touch-
down) were supported. An active touchdown and small touchdown distance have
long been thought to play a role in determining the magnitude of braking. For
example, Hay (1994) suggested, “the horizontal velocity of the foot is the sole
determinant of whether there is a braking…effect” (p. 408). In contrast, Mero et
al. (1992) suggested, “The primary reason for the decrease in running velocity is
the horizontal distance between the first contact point and the centre of gravity of
the body at touchdown” (p. 382).
Hunter, Marshall, and McNair42
Possibly, the use of a more active touchdown is part of the cause of lower
braking, and a smaller touchdown distance is a technique adjustment required to
maintain balance. However, further research involving an intervention is required
to test for causation. In addition, the possible effect that a decrease in braking
might have on vertical impulse (see Table 2), and the consequences, should be
examined.
With regard to the hypotheses stating how propulsion can be maximized,
extra extension of the stance limb at takeoff was not supported. However, the use
of a greater hip extension velocity was partially supported (i.e., mean, but not
maximum, hip extension velocity during stance was associated with greater pro-
pulsion). The notion that the hip extensor musculature is the main determinant of
thigh kinematics during stance, and therefore the main determinant of propulsion,
is popular with some researchers and many coaches. For example, Wiemann and
Tidow (1995) stated, “the hamstrings in particular, together with a muscle rein
consisting of gluteus maximus and adductor magnus, supply the energy needed for
forward propulsion, by providing a high back-swing velocity of the support leg”
(p. 47). However, even if the angular velocity of the stance thigh is important in
producing propulsion, we feel that further research is required to see if the hip
extensor musculature is the major determinant. Putnam (1991) showed that the
kinematics of the swing-limb segments in running are determined by a combina-
tion of resultant joint moments and segment interactions. It is possible that the
situation is similar for the stance limb. The possible indirect contribution of the
swing limb to propulsion is also largely unknown.
Before concluding, we must provide some cautionary notes regarding the
results of this study. First, the results are not necessarily applicable to other phases
of a sprint, for example the maximal-velocity phase. Second, we cannot predict
with any confidence the relationships that might exist outside the caliber of ath-
letes we tested. Third, we quantified components of the GRF. We emphasize that
these components are of a single entity, the GRF, and therefore are interrelated. It
is possible that a change in one component will result in a change in another com-
ponent. Fourth, for all results in this paper, causation cannot automatically be as-
sumed.
In conclusion, relative propulsive impulse accounted for 57% of the vari-
ance in sprint velocity (greater magnitudes of relative propulsive impulse were
associated with faster sprint velocities). Relative braking impulse accounted for
only 7% of the variance in sprint velocity (lower magnitudes of relative braking
impulse were associated with faster sprint velocities). It is likely that high magni-
tudes of horizontal propulsion are required to achieve high acceleration. However,
the practical significance of the weak relationship between braking and sprint ve-
locity is arguable. Furthermore, the possibility that braking might have some ad-
vantages (e.g., storage of elastic energy) could not be ruled out. The faster athletes
tended to produce only moderate magnitudes of relative vertical impulse. We specu-
lated that, during the acceleration phase, the most favorable magnitude of relative
vertical impulse is one that creates a flight time just long enough to allow reposi-
tioning of the lower limbs; all other strength reserves should be directed horizon-
tally. Further research is required to see if braking, propulsive, and vertical impulses
can be modified with specific training. This would provide insight into how a
change in one GRF component might affect the others, and ultimately how these
changes affect sprint velocity.
GRF and Kinematics of Sprint Running
43
References
Cavagna, G., Komarek, L., & Mazzoleni, S. (1971). The mechanics of sprint running. Jour-
nal of Physiology, 217, 709-721.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York: Aca-
demic Press.
de Leva, P. (1996). Adjustments to Zatsiorsky-Seluyanov’s segment inertia parameters. Jour-
nal of Biomechanics, 29, 1223-1230.
Hay, J.G. (1994). The biomechanics of sports techniques (4th ed.). London: Prentice Hall
International.
Hreljac, A., & Marshall, R. (2000). Algorithms to determine event timing during normal
walking using kinematic data. Journal of Biomechanics, 33, 783-786.
Hunter, J., Marshall, R., & McNair, P. (2004). Interaction of step length and step rate during
sprint running. Medicine and Science in Sports and Exercise, 36, 261-271.
Mann, R., & Sprague, P. (1983). Kinetics of sprinting. Track and Field Quarterly Review,
83(2), 4-9.
Mann, R.V., Kotmel, J., Herman, J., Johnson, B., & Schultz, C. (1984). Kinematic trends in
elite sprinters. In J. Terauds et al. (Eds.), Proceedings of the International Sympo-
sium of Biomechanics in Sports (pp. 17-33). Del Mar, CA: Academic Publ.
Mero, A., & Komi, P.V. (1986). Force-, EMG-, and elasticity-velocity relationships at
submaximal, maximal and supramaximal running speeds in sprinters. European Jour-
nal of Applied Physiology, 55, 553-561.
Mero, A., Komi, P.V., & Gregor, R.J. (1992). Biomechanics of sprint running. Sports Medi-
cine, 13, 376-392.
Putnam, C.A. (1991). A segment interaction analysis of proximal-to-distal sequential seg-
ment motion patterns. Medicine and Science in Sports and Exercise, 23, 130-144.
Putnam, C.A., & Kozey, J.W. (1989). Substantive issues in running. In C.L. Vaughan (Ed.),
Biomechanics of sport (pp. 1-33). Boca Raton, FL: CRC Press.
Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics (4th ed.). Boston: Allyn &
Bacon.
Weyand, P., Sternlight, D., Bellizzi, M.J., & Wright, S. (2000). Faster top running speeds
are achieved with greater ground forces not more rapid leg movements. Journal of
Applied Physiology, 89, 1991-1999.
Wiemann, K., & Tidow, G. (1995). Relative activity of hip and knee extensors in sprint-
ing—Implications for training. New Studies in Athletics, 10(1), 29-49.
Winter, D.A. (1990). Biomechanics and motor control of human movement. New York:
Wiley & Sons.
Wood, G. (1987). Biomechanical limitations to sprint running. In B. van Gheluwe & J. Atha
(Eds.), Medicine and sport science (Vol. 25: Current Research in Sports Biomechan-
ics, pp. 58-71). Basel: Karger.
Acknowledgments
Thanks to the late James G. Hay for his expert advice and encouragement. His pres-
ence is sorely missed. Thanks also to Rene Ferdinands for assisting with data collection.
