Sprint Mechanics in World-Class Athletes: A New Insight into the Limits of Human Locomotion

Abstract

The objective of this study was to characterize the mechanics of maximal running sprint acceleration in high-level athletes. Four elite (100-m best time 9.95–10.29 s) and five sub-elite (10.40–10.60 s) sprinters performed seven sprints in overground conditions. A single virtual 40-m sprint was reconstructed and kinetics parameters were calculated for each step using a force platform system and video analyses. Anteroposterior force (FY), power (PY), and the ratio of the horizontal force component to the resultant (total) force (RF, which reflects the orientation of the resultant ground reaction force for each support phase) were computed as a function of velocity (V). FY-V, RF-V, and PY-V relationships were well described by significant linear (mean R2 of 0.892 ± 0.049 and 0.950 ± 0.023) and quadratic (mean R2 = 0.732 ± 0.114) models, respectively. The current study allows a better understanding of the mechanics of the sprint acceleration notably by modeling the relationships between the forward velocity and the main mechanical key variables of the sprint. As these findings partly concern world-class sprinters tested in overground conditions, they give new insights into some aspects of the biomechanical limits of human locomotion.

Full text

Sprint mechanics in world-class athletes: a new insight into the
limits of human locomotion
G. Rabita1, S. Dorel2, J. Slawinski3, E. Sàez-de-Villarreal4, A. Couturier1, P. Samozino5, J-B. Morin6
1Research Department, National Institute of Sport, INSEP, Paris, France, 2Laboratory “Movement, Interactions, Performance”
(EA 4334), University of Nantes, Nantes, France, 3Research Center in Sport and Movement (EA 2931), University of Paris Ouest
Nanterre La Défense, Paris, France, 4Faculty of Sports, University Pablo de Olavide, Seville, Spain, 5Laboratory of Exercise
Physiology (EA 4338), University of Savoie, Le Bourget-du-Lac, France, 6Laboratory of Human Motricity, Education Sport and Health
(EA 6309), University of Nice Sophia Antipolis, Nice, France
Corresponding author: Giuseppe Rabita, PhD, Research Department, National Institute of Sport, INSEP, 11 Avenue du Tremblay,
75012 Paris, France. Tel: +33 (0)1 41 74 44 71, Fax: +33 (0)1 41 74 43 35, E-mail: giuseppe.rabita@insep.fr
Accepted for publication 13 November 2014
The objective of this study was to characterize the
mechanics of maximal running sprint acceleration in
high-level athletes. Four elite (100-m best time 9.95–
10.29 s) and five sub-elite (10.40–10.60 s) sprinters per-
formed seven sprints in overground conditions. A single
virtual 40-m sprint was reconstructed and kinetics
parameters were calculated for each step using a force
platform system and video analyses. Anteroposterior
force (FY), power (PY), and the ratio of the horizontal
force component to the resultant (total) force (RF, which
reflects the orientation of the resultant ground reaction
force for each support phase) were computed as a func-
tion of velocity (V). FY-V, RF-V, and PY-V relationships
were well described by significant linear (mean R2of
0.892 ±0.049 and 0.950 ±0.023) and quadratic (mean
R2=0.732 ±0.114) models, respectively. The current
study allows a better understanding of the mechanics
of the sprint acceleration notably by modeling the
relationships between the forward velocity and the main
mechanical key variables of the sprint. As these findings
partly concern world-class sprinters tested in
overground conditions, they give new insights into
some aspects of the biomechanical limits of human
locomotion.
Elite athletes specialized in the short sprint (60 or 100 m)
constitute a unique model to investigate some aspects of
human limits (Denny, 2008). Their specific muscular,
physiological, and biomechanical features have fre-
quently been investigated as they are considered the
fastest human runners on Earth (Ben Sira et al., 2010;
Slawinski et al., 2010). Regarding world-class sprinters
in particular, their 100-m race characteristics have been
systematically described in official events such as World
Championships or summer Olympic Games (Taylor &
Beneke, 2012; Krzysztof & Mero, 2013). Such an
approach has the obvious advantage of allowing the
characterization of the best 100-m performance ever and
the few fastest men who present at the given time an
extreme motivation and theoretically optimal technique
and physical shape. In return, because of the constraints
inherent to such international events, these conditions
usually have the drawback of allowing the measurement
or estimation of only a few basic variables [e.g., forward
velocity, flight time, and contact time (Taylor & Beneke,
2012) or step length and step frequency (Krzysztof &
Mero, 2013)] from video footage generally provided by
television stations.
When considering elite athletes, the lack of experi-
mental data still prevents us from thoroughly under-
standing the determinants of sprint performance (Morin
et al., 2012). Recent experimental measures have been
previously reported, but analyses have generally focused
either around a single mark of the sprint distance during
acceleration [e.g., starting phase (Slawinski et al., 2010;
Debaere et al., 2013a; Kawamori et al., 2014; Otsuka
et al., 2014), the 8-m mark (Kawamori et al., 2013), the
16-m mark (Hunter et al., 2004), the 45-m mark
(Bezodis et al., 2008)] or on the constant-maximal speed
phase of the sprint (Weyand et al., 2000, 2010;
Bergamini et al., 2012). Thus, it would be of great inter-
est to have the opportunity to describe from a mechanical
standpoint the sprint acceleration phase: in particular,
this would allow the characterization of (a) individual
force−velocity (F-V) and power–velocity (P-V) relation-
ships and (b) key variables of the sprint technique.
Numerous studies have investigated the F-V relation-
ships in functional tasks (Rahmani et al., 2001;
Samozino et al., 2010). Classically, the following theo-
retical parameters have been derived from the linear F-V
relationships: (a) the maximal velocity (V0) calculated
Scand J Med Sci Sports 2015: 25: 583–594
doi: 10.1111/sms.12389
© 2015 John Wiley & Sons A/S.
Published by John Wiley & Sons Ltd
583

by extrapolation to zero force; (b) the maximal force
(F0) calculated by extrapolation to zero velocity; and (c)
the maximal power output (Pmax), maximal value of the
product of F and V variables. Regarding the sprint
running task in particular, experimental investigations
have been developed using instrumented treadmills to
perform measurements throughout the entire accelera-
tion phase. The subject’s forward velocity was given by
the speed of the belt and the anteroposterior force pro-
duction can be directly obtained by force sensors posi-
tioned under the motorized treadmill frame (Morin et al.,
2010). Characterizing F-V relationships by means of
such methodologies allows to assess the ability of the
neuromuscular system to generate maximal power. This
ability, influenced by interrelated factors [e.g., muscle
fiber composition, architectural characteristics, anatomi-
cal joint configuration, levels of neural activation, and
technical aspects (for review, see Cormie et al., 2011)],
was shown to be correlated with sprint performance
(Morin et al., 2012) as previously demonstrated in other
sport activities (Dorel et al., 2005, 2010). Moreover, it
helps determine the individual mechanical profile of the
athlete in his specific sprint running task (Morin et al.,
2012).
On the other hand, the measurement of the ground
reaction force (GRF) for each step of the acceleration
phase allows not only to quantify the horizontal compo-
nent of this GRF but also its contribution to the total
GRF. Indeed, previous investigations (Morin et al., 2011;
2012), by transposing the concept of effectiveness used
in pedaling mechanics, calculated the ratio of force (RF)
during treadmill sprint running. This RF parameter
reflects the orientation of the resultant GRF at each step
and then the athletes’ ability to apply force effectively
onto the ground (i.e., in the forward direction). Interest-
ingly, these authors showed that the mean RF parameter
obtained on the treadmill was highly correlated to the
track 100-m performance. They also proposed an index
of force application technique (DRF) computed as the
slope of the linear RF-horizontal velocity relationship
from sprint start until top speed. This DRF parameter
reflects the capability to limit the systematic decrease in
RF as speed increases. In other words, it reveals the
athlete’s aptitude to maintain an efficient sprint tech-
nique despite increasing speed. DRF has also been shown
to be highly related to 100-m performance (Morin et al.,
2011, 2012). Noticeably, these studies also highlighted
that the resultant GRF produced over sprint acceleration
was not correlated to sprint performance.
The above-mentioned literature shows that the use of
F-V, P-V, and RF-V relationships because of the close
link between their associated variables and field perfor-
mance have undoubtedly contributed to a better under-
standing of the mechanical determinants of sprint
running; however, they have only been recorded and
studied in sprint treadmill conditions. Consequently,
even if these data were collected in elite athletes (Morin
et al., 2012), this limitation prevents real and definitive
insights into the human determinants of sprint accelera-
tion performance. Firstly, in treadmill conditions, sub-
jects started in a crouched position instead of using
starting blocks, which is a substantial issue considering
the importance of this phase in 60- and 100-m perfor-
mance (Harland & Steele, 1997; Slawinski et al., 2010).
Secondly, the maximal velocities reached by the athletes
on the treadmill were substantially lower than those
recorded in overground conditions (Morin et al., 2012).
For example, Chelly and Denis (2001) and Morin et al.
(2012) measured maximal treadmill velocity around 6.5
and 7.5 m/s, respectively, which is far from the maximal
speed reached by elite sprinters during field sprints (e.g.,
12.5 m/s for Usain Bolt’s 100-m World Record).
In the present study, we had the unique opportunity to
study the mechanical determinants of sprint performance
in nine high-level sprinters, including four world-class
athletes (among them was the 2013 60-m indoor Euro-
pean champion in Göteborg, Sweden, see Materials and
methods section). The second main originality of this
study was that the acceleration phase was studied for the
first time in field conditions (starting blocks and tartan
track equipped with force plates) during a quasi-standard
training session.
The aims of this study were (a) to describe the sprint
acceleration mechanics (i.e., spatiotemporal patterns,
kinetics, F-V, P-V, and RF-V relationships) in elite and
sub-elite sprinters in order to give a new mechanical
insight into the limits of human sprint acceleration; and
(b) to analyze the correlations between these parameters
and sprint performance in order to further investigate the
determinants of sprint acceleration performance.
We hypothesized that, as already reported on recent
treadmill studies, the mechanical profile of elite and
sub-elite sprinters maximally accelerating on a track can
be characterized (a) by the F-V relationship using a
linear model and (b) by the P-V relationship using a
quadratic model. We also hypothesized that the higher
acceleration of world-class sprinters compared with sub-
elite athletes is related to a higher maximal power asso-
ciated with a better (i.e., more forward) orientation of the
force onto the ground.
Materials and methods
Participants
Nine elite or sub-elite sprinters (age: 23.9 ±3.4 years; body mass:
76.4 ±7.1 kg; height: 1.82 ±0.69 m) gave written informed
consent to participate in this study, conducted according to the
declaration of Helsinki and in accordance with the local ethical
committee. The elite group (E) was composed of four athletes who
had been medalists in a sprint event during the (a) 2011 outdoor
World championships (Daegu, South Korea) and/or (b) 2012
outdoor (Helsinki) and/or 2013 indoor (Göteborg, Sweden) Euro-
pean championships (including the 60-m champion) and/or final-
ists in the 2012 Summer Olympics (London). Their personal
100-m best times ranged from 9.95 to 10.29 s. The sub-elite group
Rabita et al.
584

(SE, n=5) was composed of French national-level sprinters
(100-m records range from 10.40 to 10.60 s).
Material and experimental protocol
Participants were tested at the indoor stadium of the French Insti-
tute of Sport (INSEP) during a quasi-standard sprint training
session. After a 45-min warm-up managed by their personal coach,
the athletes performed seven sprints: 2 ×10 m, 2 ×15 m, 20 m,
30 m, and 40 m with 5 min of rest between each trial. During these
sprints, vertical, anteroposterior, and lateral components of the
GRF were measured by a 6.60-m long force platform system
(natural frequency ≥500 Hz). This system consisted of six indi-
vidual force plates (1.2 ×0.6 m) connected in series and covered
with a tartan mat that was level with the stadium track. The second
force plate is turned by 90 degrees to allow the athletes’ hands
positioning during the start. Each force plate was equipped with
piezoelectric sensors (KI 9067; Kistler, Winterthur, Switzerland).
The force signals were digitized at a sampling rate of 1000 Hz.
Based on the original work of Cavagna et al. (1971), the pro-
tocol was designed in order to reconstruct the characteristics of a
single virtual 40-m sprint for each athlete. This distance corre-
sponds to the sprint acceleration. For example, it was shown in
world-class athletes that the velocity measured between 30 and
40 m is greater than 95% of the maximal velocity reached during
the 100-m sprint (Krzysztof & Mero, 2013). To this end, the
starting blocks, initially placed over the first platform for the 10-m
sprints, were progressively installed remotely for the subsequent
trials (15 to 40 m) so that 18 foot contacts from the block to the
40-m mark could be evaluated. Indeed, the measurement area
made it possible to record the force of five foot contacts (four
steps) for the 10-m trials (including the pushing block phase), four
contacts (three steps) for the 15-m trials, and three contacts (two
steps) for the three other trials (20-, 30-, and 40-m). In addition,
an optical measurement system (Optojump Next, Microgate,
Bolzano, Italy) measured the spatiotemporal characteristics of the
first steps of all seven sprints in order to assess the consistency of
the first seven steps pattern (see Statistical analysis section).
Running velocity was measured with both the force platform
system and a digital camera (Exilim EX-F1, Casio, Tokyo, Japan)
that filmed the athletes in the sagittal plane as they entered the
force platform area (see Data processing section).
Data processing
Sprint mechanics
For all measured variables, unless otherwise specified, the
mechanical parameters described in the succeeding text represent
the average of instantaneous data recorded during each contact.
Regarding the parameters recorded twice (i.e., for the 10-m and
15-m runs), only the values of the trial corresponding to the best
performance (best time over the distance) were selected. More-
over, for several mechanical variables, all the values measured
during all the contacts over the 40 m were averaged (see Table 1,
averaged FZ, averaged FY, averaged PYand averaged RF).
Force platform signal was low-pass filtered (200 Hz cutoff,
third order Butterworth applied forward and backward for zero
phase shifting) and instantaneous data of vertical [FZ, Fig. 1(a)]
and horizontal [anteroposterior, FY, Fig. 2(b); mediolateral, FX,
Fig. 1(c)] components of ground reaction forces as well as the
resultant GRF (FTOT) were averaged for each contact phase (FZ
above 10 N) during the best 10-m and 15-m trials and during the
20-, 30-, and 40-m trials and expressed in N and body weight
(BW). They are noted as FZmean,F
Ymean,F
Xmean, and FTOTmean, respec-
tively. The anteroposterior acceleration of the center of mass
(COM) (AY) was calculated as follows
AFm
YY
=
with mas the body mass.
This expression was integrated once over time to provide
instantaneous anteroposterior velocity of the COM (VY) at time t:
VV Adt
YY Y
t
=+
∫
0
with initial velocity conditions V0Y taken as an integration
constant.
Apart from the 10-m tests, for which V0Y10 =0 m/s as the start-
ing blocks were placed over the force platform area, V0Y15,V
0Y20,
V0Y30, and V0Y40 were evaluated by high-speed video. A digital
camera was placed in a fixed position perpendicularly to the sag-
ittal plane. Focus and zoom were adapted in order to be able to
follow an entire step cycle of the subject before the entrance in the
force platform area. The motion in the sagittal plane of three
retroreflective markers placed on the anterior superior iliac spine,
posterior superior iliac spine, and the great trochanter was
recorded at 300 Hz (resolution: 512 ×384). Anteroposterior posi-
tion of the markers was determined by digitizing them with a video
analyzing system (Dartfish, Fribourg, Switzerland). Then, after
classically calibrating the camera view (3.80 m, ratio length/pixel:
0.0073 m/pixel), the horizontal distance covered by each marker
was measured between two key positions: the beginning and the
end of the last entire step cycle (i.e., contact +aerial phases) pre-
ceding the first foot contact on the force platform. The coordinates
between the beginning and the end of this cycle, as they were
measured in the same body position, were used to calculate a mean
horizontal velocity of the subject. As V0Y corresponds to the veloc-
ity during the aerial part of this phase, the mean velocity over the
step was corrected taking into account the variation that occurred
during the preceding contact on the basis of the measurement of
instantaneous variation of the horizontal velocity in the subsequent
two or three steps performed on the force plate.
VYmean was calculated for each recorded step and corresponds to
the averaged anteroposterior velocity during the contact phase of
the best 10- and 15-m trials and during the 20-, 30-, and 40-m
trials. Instantaneous PYwas obtained by the product of instanta-
neous FYand VY. The net power output in the anteroposterior
direction (PYmean in W and W/kg) was calculated for each contact
phase by averaging the instantaneous PYvalues during the contact.
In order to quantify the contribution of the anteroposterior com-
ponent of the GRF to the total force, a RF was calculated as the
ratio of FYto FTOT for each contact period (Morin et al., 2011;
2012). For this purpose, the ratio was done instantaneously and
then averaged for each contact phase. Furthermore, to our best
knowledge, RF parameter has only been assessed in two dimen-
sions yet (i.e., vertical and horizontal anteroposterior axes). Then,
in order to assess the part of the lateral force produced by the
sprinters with respect to the resultant total force, we quantified the
difference between RF variables expressed with respect to FTOT,
the sum of the force vectors measured on x,y, and zaxes, and to
FTOTyz, the sum of the forces measured in the sagittal plan only (y
and zaxes).
Before pooling the data to reconstruct a complete 40-m dataset
for each subject, running pattern repeatability was quantified on
the sixth step (around 8-m distance), which was the last step
common to all subjects before the end of the shorter sprint
(10-m distance). High repeatability was observed for all variables
– time performance at the end of the sixth step [mean =1.309 s,
coefficient of variation (CV) =1.84%, intraclass correlation
coefficient (ICC) =0.921], total distance at the end of the sixth
step (mean =7.88 m, CV =1.88%, ICC =0.958), step length
(mean =1.60 m, CV =2.15%, ICC =0.946), contact time
(mean =0.136 s, CV =3.76%, ICC =0.686) – showing that all
Sprint mechanics in elite athletes
585

bouts were performed with the same maximal involvement regard-
less of the total distance (10- to 40-m). After computation, all data
were then pooled from the seven sprints and used to draw overall
F-V, P-V, and RF-V relationships.
Using linear and second-order polynomial regressions from
these individual relationships, theoretical parameters were quanti-
fied: (a) FY0, maximal anteroposterior force theoretically pro-
duced over one contact phase at null velocity; (b) V0, maximal
anteroposterior velocity reached when the anteroposterior force is
equal to zero; and (c) PYmax, maximal anteroposterior power iden-
tified as the apex of the P-V relationship. Regarding RF-V rela-
tionships, RF0 represents the theoretical maximal contribution of
anteroposterior force to the total force produced over one contact
phase at null velocity. An index of force application technique
(DRF) was computed for each participant as the slope of the linear
RF-V relationship. For details, see Morin et al. (2011, 2012),
EJAP.
Given the bilateral nature of the start phase, F-, P-, and RF-V
regressions were quantified excluding the values measured in the
blocks. The maximal theoretical values must be differentiated
from the maximal values measured from the 18 recorded contact
phases and noted as FYpeak,V
Ypeak, and PYpeak. The start was specifi-
cally analyzed and the parameters measured in the block phase
were noted as FYblocks,P
Yblocks, and RFblocks.
The following step spatiotemporal variables were measured
using the force platform system: contact time (tc, in s), aerial time
Table 1. Mean values (SD) of the main sprint performance and spatiotemporal parameters, force-, power-, and RF-velocity relationships for the whole
group (n= 9) and for both subgroups (elite, E: n= 4; sub-elite, SE: n=5)
Variable Mean all (SD)
(n=9)
Mean E (SD)
(n=4)
Mean SE (SD)
(n=5)
Difference
(%E)
ES (Cohen’s d)
Sprint performance
10 m time (s) 1.85 (0.10) 1.79 (0.02) 1.90 (0.12) −6.1 1.10
15 m time (s) 2.50 (0.10) 2.40 (0.02) 2.58 (0.09) −7.5 1.80
20 m time (s) 3.05 (0.13) 2.94 (0.02) 3.13 (0.13) −6.5 1.46
30 m time (s) 4.08 (0.18) 3.93 (0.03) 4.20 (0.15) −6.9 1.50
40 m time (s) 5.10 (0.25) 4.90 (0.07) 5.27 (0.21) −7.6 1.48
40 m maximal velocity (m/s) 9.78 (0.52) 10.24 (0.19) 9.33 (0.31) 8.9 3.64
Spatiotemporal parameters
Contact time
tc blocks (ms) 396 (33) 376 (13) 412 (36) −9.6 1.09
Maximal contact time (ms) 193 (28) 191 (18) 193 (30) −1.0 0.07
Minimal contact time (ms) 94 (4) 94 (5) 94 (4) 0.0 0.0
Aerial time
ta blocks (ms) 75 (20) 81 (13) 70 (25) 13.6 0.55
Maximal aerial time (ms) 124 (7) 120 (6) 128 (5) −6.7 1.14
Minimal aerial time (ms) 50 (13) 42 (13) 56 (10) −33.3 1.04
Step frequency
Sfblocks (Hz) 2.14 (0.17) 2.20 (0.12) 2.09 (0.21) 5.0 0.64
Minimal frequency (Hz) 3.92 (0.34) 3.94 (0.44) 3.90 (0.44) 1.0 0.11
Maximal frequency (Hz) 4.87 (0.23) 4.95 (0.12) 4.80 (0.30) 3.0 0.65
Step length
Slblocks (m) 0.99 (0.11) 0.96 (0.16) 1.01 (0.06) −5.2 0.45
Minimal step length (m) 1.11 (0.12) 1.18 (0.07) 1.06 (0.14) 10.2 1.00
Maximal step length (m) 2.19 (0.11) 2.22 (0.10) 2.17 (0.12) 2.3 0.45
Force- and power-velocity parameters
Averaged FZ(N/kg) 17.3 (0.5) 17.2 (0.4) 17.5 (0.5) −2.0 0.59
Averaged FY(N/kg) 3.3 (0.3) 3.5 (0.6) 3.1 (0.2) 9.7 1.75
Averaged PY(W/kg) 20.8 (2.2) 22.5 (1.1) 19.4 (1.9) 13.9 1.99
Averaged RF (% FTOT) 19.2 (1.3) 20.3 (0.7) 18.3 (1.0) 9.7 2.31
V0 (m/s) 11.38 (0.84) 11.90 (0.23) 10.99 (0.97) 7.6 0.12
FY0 (N) 776 (93) 855 (60) 744 (90) 13.0 1.19
Relative FY0 (N/kg) 9.77 (0.84) 9.95 (0.67) 9.62 (1.06) 3.3 0.39
Theoretical PYmax (W) 2328 (295) 2550 (283) 2150 (158) 15.7 1.35
Measured PYpeak (W) 2421 (321) 2695 (244) 2201 (158) 18.3 1.53
Relative PYmax (W/kg) 29.3 (2.3) 31.1 (0.8) 27.8 (2.2) 10.6 1.43
Relative PYpeak (W/kg) 30.5 (2.9) 32.9 (1.2) 28.5 (2.1) 13.4 1.51
RF0 (%) 70.6 (5.4) 71.6 (2.6) 70.1 (7.3) 2.1 0.02
DRF −0.067 (0.007) −0.064 (0.003) −0.069 (0.009) −7.8 0.71
Mean difference RFxyz −RF
xy (% RFxyz) 0.25 (0.06) 0.29 (0.03) 0.21 (0.03) 27.6 1.33
Measured block parameters
Block clearing anteroposterior velocity (m/s) 3.37 (0.27) 3.61 (0.08) 3.17 (0.19) 12.1 1.60
FYblocks (N) 679 (110) 783 (59) 596 (46.9) 23.9 1.70
Relative FYblocks (N/kg) 8.56 (1.18) 9.59 (0.53) 7.74 (0.82) 19.2 1.56
PYblocks (W) 1156 (270) 1415 (118) 949 (124) 32.9 1.72
Relative PYblocks (W/kg) 14.5 (3.0) 17.3 (1.3) 12.3 (1.9) 28.8 1.64
RFblocks (%) 58.9 (5.4) 63.0 (2.6) 54.9 (4.3) 12.8 1.46
Minimal and maximal parameters referred to the values recorded on the whole acceleration phase (0–40 m) excluding the block phase. ES, effect size;
RF, ratio of force.
Rabita et al.
586

(ta, in s), step frequency (Sf) and step length (Sl). Regarding the
block phase, the following spatiotemporal parameters were com-
puted:
tc blocks, contact time in the blocks, measured during the stabili-
zation phase just before the start (where FZ=body weight), rep-
resents the period from the time corresponding to a change higher
than 10 N of FZto the block clearance (defined as the moment the
forward foot takes off from the block and where FZ=0).
ta blocks, aerial time from the blocks, period between the block
clearance and the beginning of the following foot contact.
Sfblocks, step frequency from the blocks, calculated as follow:
Sfblocks c blocks a blocks
tt=+
()
1
Slblocks, step length from the forward block to the contralateral
foot landing position.
Statistical analysis
Results were expressed as mean (±SD). To assess whether the
running pattern was similar for each of the seven sprints, the
repeatability of the sixth step (for details, see Data processing
section) was quantified for time performance, total distance, step
length, and contact time using the ICC and CV (Hopkins, 2000). In
order to compare elite and sub-elite athletes, considering the small
population inherent to the selection of high-level athletes, only the
differences (expressed in percentage of elite group values) and the
effect size (ES, Cohen’s d) were calculated. For this effect size
calculation, the results were interpreted as negligible, small,
medium, or large for ES lower than 0.2, between 0.2 and 0.5,
between 0.5 and 0.8, or higher than 0.8, respectively. When they
were suitable (P<0.05, least chi-square method), linear and
quadratic regression models were used to fit the relationship
between the different mechanical variables and the running veloc-
ity. Pearson’s correlation coefficients were used to examine the
relationships between spatiotemporal or mechanical variables and
the performance. The critical level of significance was set at
P<0.05.
Results
Forty-meter performance and
spatiotemporal parameters
Table 1 summarizes the mean values (SD) of the track
performances recorded during each sprint trial (from 10-
to 40-m). The 40-m was run in 5.10 ±0.25 s, with a top
speed of 9.78 ±0.52 m/s. Table 1 also presents the
values during or immediately after the first propulsion
from the starting blocks together with the minimal and
maximal values recorded during different sprints. The
Fig. 1. Typical signals of instantaneous vertical (FZ), anteroposterior (FY), and lateral (FX) component of the ground reaction force (in
N) obtained during the 10-, 15-, 20-, 30-, and 40-m trials that correspond to the following location of the force-plate area: (a) 10-m
trials: Zone 1 (−1.4 to 5.2 m; 0 corresponding to the starting line); (b) 15-m trials: Zone 2 (4.6–11.2 m); (c) 20-m trial: Zone 3
(12.4–19.0 m); (d) 20-m trial: Zone 4 (22.4–29.0 m); (e) 20 m trial: Zone 5 (32.4–39.0 m).
Sprint mechanics in elite athletes
587

changes in the main spatiotemporal characteristics of the
step when velocity increases are presented in Fig. 2 for
all subjects. The contact time decreased on average by
about 50% between the first (blocks) and second steps
and the same relative decrease occurred between the
second step and the 40-m minimal contact time (last
step) (Table 1). Mean values of aerial time just after
the start (between the blocks clearing and the contact of
the second step) were about five times lower than the
propulsion time in the starting blocks. This aerial time
decreased by around one-third on average after the
second step and increased linearly with the velocity to
reach the same values as contact time. While step fre-
quency was smaller by half at the first step compared
with the second step, step length did not change.
Between the second and last steps, step frequency
and step length increased by about 20% and 50%,
respectively.
F-V, P-V, and RF-V relationships
The parameters FYmean and PYmean as a function of the
horizontal velocity were well described by significant
linear (mean R2=0.892 ±0.049) and quadratic relation-
ships (mean R2=0.732 ±0.114), respectively (Fig. 3).
The parameter FTOT showed an increase in velocity that
was best described by a linear model (mean
R2=0.815 ±0.023). Regarding the orientation of force
onto the ground, the RF decreased with velocity and this
relationship was linear for all subjects (mean
R2=0.950 ±0.023). Table 1 also presents the theoretical
maximal values calculated from these relationships (FY0,
V0, Pymax, RF0) and the slope of the RF-V relationship
(DRF index). On average, the athletes reached PYmax at
around the sixth step (6.22 ±1.2 steps).
In order to assess the part of the lateral force produced
by the sprinters, we quantified the difference between RF
variables expressed with respect to (a) FTOT, the resultant
force measured on x,y, and zaxes, and (b) FTOTyz, the
resultant force measured in the sagittal plan only (yand
zaxes). The mean difference, expressed in percentage of
FTOT was 0.25 (±0.03) %.
Correlations between mechanical and
performance variables
The correlation analyses of spatiotemporal parameters,
force, and power are presented in Table 2.
Discussion
The aims of this study were, first, to characterize the
sprint acceleration mechanics (i.e., spatiotemporal pat-
terns, kinetics, F-V, P-V, and RF-V relationships) in elite
and sub-elite sprinters and, second, to analyze the corre-
lations between these parameters and sprint performance
in order to further investigate the determinants of sprint
acceleration performance. The reconstruction of a virtual
40-m sprint allowed us to describe for the first time the
mechanical characteristics of the acceleration phase in
Fig. 2. Spatiotemporal characteristics of the step (a) during (contact time, in ms) or immediately after [aerial time (ms), step frequency
(Hz), step length (m)] the first propulsion in the starting blocks and (b) during the acceleration phase (second to last steps) depicted
against the forward velocity (m/s).
Rabita et al.
588

world-class and sub-elite sprinters. The main findings of
this study show that (a) while step length increases regu-
larly during the acceleration phase, step frequency is
almost instantaneously leveled at the maximal possibil-
ity of elite athletes; (b) F- and P-V relationships during
sprints performed in realistic field conditions were well
described by linear and quadratic models, respectively;
and (c) the effectiveness of force application greatly
accounts for the difference in the performance between
highly trained athletes.
Spatiotemporal parameters have shown that the step
frequency very quickly reached the maximal values
(80% at the first step and about 90% after the third
step) and then remained constant throughout the accel-
eration phase (Fig. 2). These results are consistent with
the recent study by Debaere et al. (2013b) who
reported that high-level sprinters are capable of devel-
oping in the starting phase (0–10 m) a step frequency
higher than 95% of the step frequency reached at
maximal speed during a 60-m sprint. The acceleration
of high-level athletes is afterwards quasi-exclusively
related with the increase in step length (Debaere et al.,
2013b). The mechanical mechanisms leading to such
spatiotemporal patterns are partly described by kinetics
data.
The individual F-V relationships are aptly described
by linear models (Fig. 3; mean R2=0.89). These results
were expected from previous treadmill findings
(Jaskolska et al., 1999; Morin et al., 2011, 2012) despite
the fact that some of the sprint mechanics variables
Fig. 3. (a) Anteroposterior component of the ground reaction force (FYmean, in % theoretical anteroposterior maximal force, FY0); (b)
anteroposterior power output (PYmean, in percentage of theoretical maximal power, PYmax); (c) ratio of force (RF, in % of the total force)
and (d) total force (in N/kg) in relation to the anteroposterior velocity (in % V0, theoretical maximal velocity). For each relationship,
the mean R2are presented on graphs. Thin (individual models) and thick lines (mean trend curves) are shown for information and clarity
purposes.
Sprint mechanics in elite athletes
589

differ between treadmill and overground running
(McKenna & Riches, 2007). Regarding the individual
P-V relationships, the quadratic models present lower
coefficients of determination (Fig. 3; mean R2=0.73).
This could be explained by (a) the very few number of
steps in the ascending part of the relationship inherent in
the high acceleration capability of the sprinters who
produce their maximal power after about six steps; (b)
possible asymmetry between right and left legs consid-
ering that the actions of both lower limbs were taken into
account in the models; and (c) interstep variability
because of the great muscle coordination complexity of
the sprint start phase. Nevertheless, the quadratic model
is still acceptable (mean R2=0.73), as shown by the
strong correlation between the predicted theoretical
maximal power (Pmax) and the maximal measured power
(Ppeak)(R=0.91; P<0.001). To our knowledge, the
present study is the first to describe these biomechanical
relationships in elite athletes from sprint exercises per-
formed in the field and almost similar to competition
conditions. This novelty offers new insight into sprint
mechanics in humans as it allows for the first time
linking the horizontal force or power output to the asso-
ciated realistic velocity. The fact that the population was
partially composed of world-class athletes, including the
2013 60-m indoor European Champion, strongly sug-
gests that the modeled F-V and P-V relationships
mechanically characterize the current human limits of
the sprint acceleration. From a practical viewpoint, the
use of these relationships should be very useful for train-
ing as it allows the specific and individual determination
of a range of power by means of a simple variable: the
velocity expressed relatively to the theoretical maximal
speed, i.e., independently of the athlete’s performance.
In this context, the use of current motorized pulling
devices (e.g., Kawamori et al., 2014) adequately leveled
should be very useful in order to specifically calibrate the
training sessions.
Regarding the correlations between sprint perfor-
mance and the measured mechanical variables, the
analyses strengthen the previous findings of Morin et al.
(2011, 2012). Firstly, performance parameters of the
acceleration phase were highly related to theoretical
maximal and averaged velocity and power measured in
the forward direction and obtained from F-V and P-V
relationships. Secondly, theoretical maximal anteropos-
terior ground reaction force (FY0) was not significantly
correlated with these performance parameters, whereas
the averaged anteroposterior force was. These results
suggest that the ability of the athletes to generate high
net anteroposterior (horizontal) GRF, especially at high
velocity, seems more essential to improve sprint accel-
eration and thus overall performance than their capabil-
ity to produce very high levels of resultant GRF. The
latter represents the total resulting force output from all
lower limbs muscle actions (i.e., how much total force is
produced), whereas the former represents the horizon-
tally oriented component of this resultant GRF (i.e.,
what part of this total force production is oriented in the
forward direction). Thirdly, neither the resultant GRF,
nor its vertical component was significantly related to
any of the sprint acceleration performance parameters.
The present findings, in conjunction with previous
studies (Kyröläinen et al., 2001; Nummela et al., 2007;
Table 2. Pearson’s correlation coefficient between performance and spatiotemporal or mechanical variables
Block clearing speed (m/s) Measured 40-m maximal speed (m/s) 40-m performance (m/s)
Spatiotemporal parameters
Mean step frequency 0.120 0.424 0.236
Maximal step frequency 0.070 0.236 0.176
Mean step length 0.631 0.518 0.544
Maximal step length 0.394 0.467 0.444
Force components
Averaged FZ(N/kg) −0.241 −0.216 −0.388
Averaged FY(N/kg) 0.775* 0.904*** 0.816**
Averaged FTOT (N/kg) −0.021 −0.137 −0.228
Force- and power-velocity relationship variables
Theoretical FY0 (N/kg) −0.129 0.108 0.079
Theoretical V0 (m/s) 0.878*** 0.819** 0.803**
Theoretical PYmax (W/kg) 0.868** 0.932*** 0.932***
Averaged PY(W/kg) 0.919*** 0.958*** 0.903***
RF-velocity relationship variables
RF0 0.135 0.071 0.149
Averaged RF 0.821** 0.899*** 0.933***
DRF 0.634 0.470 0.408
Block parameters
FY blocks (N/kg) 0.873** 0.836** 0.850**
PY blocks (W/kg) 0.964*** 0.909*** 0.902***
RF blocks (%) 0.858*** 0.741** 0.829***
*, **, and *** denote a significant correlation at P<0.05, P<0.01 and P<0.001, respectively. RF, ratio of force.
Rabita et al.
590

Brughelli et al., 2011), both confirm the mechanical
logic that horizontal net GRF is paramount to accelerate
the body forward. It seems important to notice that
the vertical GRF has a major influence on the sprint
mechanics as the athletes have to produce the vertical
force needed to overcome the negative vertical accelera-
tion because of gravity. However, our results support the
argument that the vertical component of the GRF is not
by itself a determinant of performance in high-level ath-
letes during the sprint acceleration phase. These results
also clearly confirm that the biomechanical determinants
of the sprint acceleration phase differ from those related
to the final top-speed phase. Indeed, the study by
Weyand et al. (2000), who analyzed a heterogeneous
population of subjects (overground maximal 100-m
speed ranged from 6.2 to 11.1 m/s) during a treadmill
top-speed test, concluded that human runners reached
faster top speeds by applying greater support forces to
the ground. More recently, using a subtle one-legged and
backward moving protocol, the same research group
(2010) brought the specification that the mechanical
limit for top speed was in the maximal vertical GRF
possibly produced within the stance phase duration, not
in the absolute maximal level of vertical GRF subjects
could produce, should movement conditions allow it.
This importance of vertical GRF production in top-speed
performance is in line with some results of Morin et al.
(2011, 2012) who also found during treadmill tests in a
population with different sprint levels – including
moderate-level physical education students together with
national-level sprinters and one world-class sprinter
(maximal speed ranging from 7.8 to 11.2 m/s) – that the
only sprint performance parameter significantly related
to the vertical GRF production was the field 100-m top
speed.
As mentioned earlier, in optimal motivation and
physical shape conditions, the capability of an athlete to
generate maximal power during a sprint acceleration
mainly depends on (a) his neuromuscular characteristics
and musculoskeletal mechanical properties and (b) his
sprint technical ability to move his body mass forward.
This technical ability could be reflected overall by the
way an athlete orients horizontally the ground reaction
force vector (e.g., Hunter et al., 2005; Morin et al.,
2011, 2012). Firstly, our results revealed the negligible
part of mediolateral forces confirming, for the entire
acceleration phase, similar findings related to the start-
ing phase in well-trained athletes (Debaere et al.,
2013a). Secondly, the fact that averaged RF is one of
the mechanical variables most highly correlated with
the 40-m performance (R2=0.93; P<0.001) strength-
ens the hypothesis that the orientation of the total force
that high-level sprinters applied to the ground during
sprint acceleration is more important to performance
than its magnitude (Morin et al., 2011, 2012). These
findings were well summarized in Fig. 4, which differ-
entiates elite from sub-elite athletes: (a) F-V relation-
ships clearly show that elite sprinters are able to
produce higher horizontal force at any given velocity
than sub-elite sprinters; (b) the RF-V relationships show
that this higher horizontal force production was caused
by a better (i.e., more forward) orientation of the force
onto the ground given that (c) the sub-elite sprinters did
not show lower resultant force production but on the
contrary, a tendency to produce higher FTOT than elite,
especially at high velocities. The fact that elite athletes
are able to orient their GRF vector more effectively than
sub-elite athletes explained part of the difference
between these two groups in terms of anteroposterior
power production (Fig. 4). Moreover, this increased
effectiveness in elite sprinters is due to specific neuro-
muscular properties (coordination, muscle-tendon
mechanical properties, joint moment-angle relation-
ships, etc.) that lead in fine to their better technical
ability of force application. Indeed, more powerful spe-
cific muscle groups (e.g., hip extensors) and muscle
groups involved in the foot–ground interaction (e.g.,
ankle joint stabilizers) could also be involved in an
appropriate coordination and high joint moments that
finally result in a better orientation of the resultant force
produced. Further research focusing on the determi-
nants of a high RF is therefore requested.
The DRF variable reflects the ability of runners to
produce and maintain high levels of RF over the entire
acceleration despite the overall straightening up of their
body with increasing speed (Morin et al., 2011, 2012).
As such, this index of force application technique is
closely linked to the sprinter’s training background.
These authors demonstrated that this parameter was
highly related to performance in a heterogeneous popu-
lation; however, this was not the case in the present
study, which could be explained by the fact that the
present population was exclusively composed of highly
trained athletes. Briefly, the fact that the mean RF param-
eter was significantly related to performance and not the
DRF variable simply reflects that the slowest well-trained
sprinters, who orient their ground reaction forces less
effectively than the speediest athletes, do not, however,
deteriorate this ability more than them with increasing
speed.
Finally, our results show that all the variables signifi-
cantly related to the mean and maximal 40-m speed were
also highly related with the block clearing speed
(Table 2). These results are in agreement with previous
studies that showed the essential contribution of the
starting phase in the sprint performance (Harland &
Steele, 1997; Slawinski et al., 2010; Debaere et al.,
2013a). However, despite the very good linear adjust-
ment of the F- and RF-V relationships, theoretical values
extrapolated to zero velocity axis (FY0 and RF0) showed
no significant correlation with sprint performance. In
contrast, both the anteroposterior force and the ratio of
force measured or computed from the starting block
phase did. These results suggest that (a) FYblocks and
Sprint mechanics in elite athletes
591

RFblocks represent specific muscular and technical abili-
ties that give complementary information with respect to
F- and RF-V relationships (adjusted without taking into
account the block phase) and that (b) the start specificity
cannot be characterized either by the extrapolated data of
F-V or those of RF-V relationships.
An obvious limitation of the present study is that
mechanical data for virtual 40-m sprint acceleration
were reconstructed from several sprints. Beyond the
fact that no force-plate system allowing such long-
distance measurements currently exists to the authors’
knowledge, the high reproducibility shown between
sprints (see Materials and methods section) supports
the realistic feature of our virtual reconstruction
approach.
Perspectives
The methods used here might not be easily reproduced
by coaches in their typical training practice. This could
limit the use of F-V and P-V relationships as done
here, when seeking to analyze and improve sprint
mechanics on an individual basis. One of the practical
perspectives of this work is to validate a simple method-
ology to quantify these variables. Indeed, very recently,
Samozino et al. (2013) proposed a simple field method
to compute force, velocity, and power in field conditions
from only split times or running velocity measurements
by radar (e.g., Mendiguchia et al., 2014).
The findings of the present study allowed us to char-
acterize the mechanics of the sprint acceleration phase
Fig. 4. (a) Anteroposterior component of the ground reaction force (FYmean, in N/kg; (b) anteroposterior power output (PYmean, in W/kg);
(c) ratio of force (RF, in %) and (d) total force (in N/kg) were presented against the anteroposterior velocity (in m/s) for the elite
(E; dark gray) and sub-elite (SE; light gray) groups. For each group, the linear or polynomial adjustments were computed excluding
the block values (presented for information).
Rabita et al.
592

in world-class and sub-elite sprinters. As such, they
yielded new insights into some aspects of biomechanical
limits in human locomotion, notably by modeling the
relationships between the realistic velocity reached by
athletes on overground conditions and three key
variables of the sprint: the anteroposterior force, the ratio
of force, and the anteroposterior power output. The
opportunity to tests such sprinters allowed further
insights into the mechanical determinants of high-level
sprint performance. Beyond the expected ability to
produce high-power output or forward force component,
the effectiveness of force application during the accel-
eration phase, represented by the averaged ratio of force,
is one of the abilities most closely linked to performance.
This was not the case for the total amount of force that a
sprinter can produce. Furthermore, the muscular and
technical athletes’ capability during the starting phase,
represented by anteroposterior force and the ratio of
force recorded in the blocks, is essential for performance
and would complement the F-, P-, and RF-V models to
characterize the mechanics of the sprint acceleration.
From a practical viewpoint, the results of the present
study should be useful for coaches to determine the
strengths and weaknesses of their elite sprinters.
Key words: Force orientation, performance, power
output, running, elite sprinters.
Acknowledgements
We are very grateful to Guy Ontanon and the elite athletes of the
National Institute of Sport (INSEP) for their involvement in the
experimentations. We also thank Dimitri Demonière and Michel
Gilot and their athletes for participating in this study. Authors
want to thank Dr. Gaël Guilhem for his general comments on
this investigation. We are grateful to Dr Caroline Giroux, Stevy
Farcy, and Virha Despotova for their collaboration during the
experimentations.
References
Ben Sira D, Amir R, Amir O, Yamin C,
Eynon N, Meckel Y, Sagiv M, Sagiv
M. Effect of different sprint training
regimes on the oxygen
delivery-extraction in elite sprinters.
J Sports Med Phys Fitness 2010: 50:
121–125.
Bergamini E, Picerno P, Pillet H, Natta F,
Thoreux P, Camomilla V. Estimation
of temporal parameters during sprint
running using a trunk-mounted inertial
measurement unit. J Biomech 2012: 45:
1123–1126.
Bezodis IN, Kerwin DG, Salo AI.
Lower-limb mechanics during the
support phase of maximum-velocity
sprint running. Med Sci Sports Exerc
2008: 40: 707–715.
Brughelli M, Cronin J, Chaouachi A.
Effects of running velocity on running
kinetics and kinematics. J Strength
Cond Res 2011: 25: 933–939.
Cavagna GA, Komarek L, Mazzoleni S.
The mechanics of sprint running.
J Physiol 1971: 217: 709–721.
Chelly SM, Denis C. Leg power and
hopping stiffness: relationship with
sprint running performance. Med Sci
Sports Exerc 2001: 33: 326–333.
Cormie P, McGuigan MR, Newton RU.
Developing maximal neuromuscular
power: Part 1-biological basis of
maximal power production. Sports Med
2011: 41: 17–38.
Debaere S, Delecluse C, Aerenhouts D,
Hagman F, Jonkers I. From block
clearance to sprint running:
characteristics underlying an effective
transition. J Sports Sci 2013a: 31:
137–149.
Debaere S, Jonkers I, Delecluse C. The
contribution of step characteristics to
sprint running performance in
high-level male and female athletes.
J Strength Cond Res 2013b: 27:
116–124.
Denny MW. Limits to running speed in
dogs, horses and humans. J Exp Biol
2008: 211: 3836–3849.
Dorel S, Couturier A, Lacour JR,
Vandewalle H, Hautier C, Hug F.
Force-velocity relationship in cycling
revisited: benefit of two-dimensional
pedal forces analysis. Med Sci Sports
Exerc 2010: 42: 1174–1183.
Dorel S, Hautier CA, Rambaud O,
Rouffet D, Van Praagh E, Lacour JR,
Bourdin M. Torque and power-velocity
relationships in cycling: relevance to
track sprint performance in world-class
cyclists. Int J Sports Med 2005: 26:
739–746.
Harland MJ, Steele JR. Biomechanics of
the sprint start. Sports Med 1997: 23
(1): 11–20.
Hopkins WG. Measures of reliability in
sports medicine and science. Sports
Med 2000: 30: 1–15.
Hunter JP, Marshall RN, McNair PJ.
Interaction of step length and step rate
during sprint running. Med Sci Sports
Exerc 2004: 36: 261–271.
Hunter JP, Marshall RN, McNair PJ.
Relationships between ground reaction
force impulse and kinematics of
sprint-running acceleration. J Appl
Biomech 2005: 21: 31–43.
Jaskolska A, Goossens P, Veenstra B,
Jaskolski A, Skinner JS. Treadmill
measurement of the force–velocity
relationship and power output in
subjects with different maximal running
velocities. Sports Med Train Rehab
1999: 8: 347–358.
Kawamori N, Newton RU, Hori N,
Nosaka K. Effects of weighted sled
towing with heavy versus light
load on sprint acceleration ability.
J Strength Cond Res 2014: 28:
2738–2745.
Kawamori N, Nosaka K, Newton RU.
Relationships between ground reaction
impulse and sprint acceleration
performance in team sport athletes.
J Strength Cond Res 2013: 27:
568–573.
Krzysztof M, Mero A. A kinematics
analysis of three best 100 m
performances ever. J Hum Kinet 2013:
36: 149–160.
Kyröläinen H, Belli A, Komi PV.
Biomechanical factors affecting running
economy. Med Sci Sports Exerc 2001:
33: 1330–1337.
McKenna M, Riches PE. A comparison of
sprinting kinematics on two types of
treadmill and over-ground. Scand J
Med Sci Sports 2007: 17: 649–655.
Mendiguchia J, Samozino P,
Martinez-Ruiz E, Brughelli M,
Schmikli S, Morin JB,
Mendez-Villanueva A. Progression of
mechanical properties during on-field
sprint running after returning to sports
from a hamstring muscle injury in
soccer players. Int J Sports Med 2014:
35: 690–695.
Morin JB, Bourdin M, Edouard P, Peyrot
N, Samozino P, Lacour JR. Mechanical
determinants of 100 m sprint running
Sprint mechanics in elite athletes
593

performance. Eur J Appl Physiol 2012:
112: 3921–3930.
Morin JB, Edouard P, Samozino P.
Technical ability of force application as
a determinant factor of sprint
performance. Med Sci Sports Exerc
2011: 43: 1680–1688.
Morin JB, Samozino P, Bonnefoy R,
Edouard P, Belli A. Direct
measurement of power during one
single sprint on treadmill. J Biomech
2010: 43: 1970–1975.
Nummela A, Keränen T, Mikkelsson LO.
Factors related to top running speed
and economy. Int J Sports Med 2007:
28: 655–661.
Otsuka M, Shim JK, Kurihara T,
Yoshioka S, Nokata M, Isaka T.
Effect of expertise on 3D force
application during the starting block
phase and subsequent steps in sprint
running. J Appl Biomech 2014: 30:
390–400.
Rahmani A, Viale F, Dalleau G, Lacour
JR. Force/velocity and power/velocity
relationships in squat exercise.
Eur J Appl Physiol 2001: 4:
227–232.
Samozino P, Morin JB, Dorel S,
Slawinski J, Peyrot N, Saez-de-
Villareal E, Rabita G A simple method
for measuring power, force and
velocity properties of sprint running.
International Society of Biomechanics
Congress, 2013. Convention Center,
Natal, Brazil.
Samozino P, Morin JB, Hintzy F, Belli A.
Jumping ability: a theoretical
integrative approach. J Theor Biol
2010: 264: 11–18.
Slawinski J, Bonnefoy A, Ontanon G,
Leveque JM, Miller C, Riquet A,
Chèze L, Dumas R. Segment-
interaction in sprint start: analysis of
3D angular velocity and kinetic energy
in elite sprinters. J Biomech 2010: 43:
1494–1502.
Taylor MJ, Beneke R. Spring mass
characteristics of the fastest men on
Earth. Int J Sports Med 2012: 33:
667–670.
Weyand PG, Sandell RF, Prime DN,
Bundle MW. The biological limits to
running speed are imposed from the
ground up. J Appl Physiol 2010: 108:
950–961.
Weyand PG, Sternlight DB, Bellizzi MJ,
Wright S. Faster top running speeds
are achieved with greater ground
forces not more rapid leg movements.
J Appl Physiol 2000: 89: 1991–1999.
Rabita et al.
594