Professional Documents
Culture Documents
To cite this article: Phil Hayes & Nicholas Caplan (2012): Foot strike patterns and ground contact times during high-calibre
middle-distance races, Journal of Sports Sciences, 30:12, 1275-1283
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Journal of Sports Sciences, August 2012; 30(12): 1275–1283
Department of Sport and Exercise Sciences, School of Life Sciences, Northumbria University, Newcastle-upon-Tyne, UK
Abstract
The aims of this study were to examine ground contact characteristics, their relationship with race performance, and the time
course of any changes in ground contact time during competitive 800 m and 1500 m races. Twenty-two seeded, single-sex
middle-distance races totalling 181 runners were filmed at a competitive athletics meeting. Races were filmed at 100 Hz.
Ground contact time was recorded one step for each athlete, on each lap of their race. Forefoot and midfoot strikers had
significantly shorter ground contact times than heel strikers. Forefoot and midfoot strikers had significantly faster average
race speed than heel strikers. There were strong large correlations between ground contact time and average race speed for
the women’s events and men’s 1500 m (r ¼ 70.521 to 70.623; P 5 0.05), whereas the men’s 800 m displayed only a
moderate relationship (r ¼ 70.361; P ¼ 0.002). For each event, ground contact time for the first lap was significantly
shorter than for the last lap, which might reflect runners becoming fatigued.
Correspondence: P. Hayes, Department of Sport and Exercise Sciences, School of Life Sciences, Northumbria University, Newcastle-upon-Tyne NE1 8ST,
UK. E-mail: phil.hayes@northumbria.ac.uk
ISSN 0264-0414 print/ISSN 1466-447X online Ó 2012 Taylor & Francis
http://dx.doi.org/10.1080/02640414.2012.707326
1276 P. Hayes & N. Caplan
date, however, there is no information on ground forefoot, midfoot, and heel strikers to determine
contact times or point of contact during either 800 m whether they differed. To assess the time course of
or 1500 m races. Furthermore, few studies have changes, we compared the ground contact time on
examined the time course of changes in running each lap of each event, which was further broken
mechanics, in particular the ground contact phase. down by forefoot, midfoot, and heel strikers.
Gazeau et al. (1997) reported that the ground
contact time was ‘modified in fatigue conditions’, as
Participants
it only increased in the latter half of a run to
exhaustion. Kyrolainen et al. (2001) suggested that a Altogether, 181 runners completed the 800 and
shorter, more rapid stretch of the muscle will elicit a 1500 m races; there were 34, 24, 71, and 52 athletes
more effective use of the stored musculo-tendon in the women’s 800 m, women’s 1500 m, men’s
elasticity in the concentric action of a stretch– 800 m, and men’s 1500 m, respectively. No descrip-
shortening cycle. Increases in ground contact times tive data were collected from the athletes, and
could reduce the effectiveness of the stretch–short- runners only competed in one race. The study
ening cycle, thereby contributing to fatigue through received institutional ethics approval. All runners
increasing the energy cost at any given running competing were aware before registration that the
speed. The ground contact phase is a critical point in races were being filmed, both for Internet broadcast
the running gait. In long-distance events, a relation- and our research.
ship has been shown between ground contact time
and performance. Moreover, during high-intensity
Data collection
running, changes in ground contact time have been
associated with an inability to maintain performance. A high-speed video camera (A602fc-2, Basler,
To date, there is a lack of evidence regarding this Ahrensburg, Germany) was placed 15 m from the
aspect of middle-distance running. The aims of this finishing line on a tripod at a height of 0.15 m. The
study, therefore, were to examine ground contact camera was positioned at right angles to the track, in
characteristics, their relationship with race perfor- the outside lane (lane 8). The shutter speed was
mance, and the time course of any changes in ground 0.002 s and the filming rate was 100 Hz. The camera
contact time during competitive 800 m and 1500 m was connected directly to a laptop PC via FireWire,
races. We hypothesized that (a) midfoot and forefoot and all video images were saved in Vicon Motus
strikers would have shorter ground contact times software (Vicon, Oxford, UK). A second, hand-held
than heel strikers, (b) midfoot and forefoot strikers camera (HVR-A1V, Sony, Tokyo, Japan) was used
would have faster average race speeds than heel to record the athletes on the start line to assist with
strikers, and (c) there would be a relationship matching athletes to their ground contact images.
between ground contact time and running speed.
Data analysis
Methods Each athlete was identifiable on every lap of his or
her race. Only one foot strike per athlete on each lap
Design
was analysed, for a total of 514 foot strikes. Using the
This field-based study analysed middle-distance images from the trackside camera and those of the
runners at a British Milers’ Club Grand Prix meeting athletes on the start line, foot strike data and
Ground contact times in middle-distance races 1277
nical error of measurement for 4.72 and 6.39 m s71 were different from all other events (F3,177 ¼ 330.037;
was 0.02 ms and 0.04 ms, respectively. When P 0.001). The average 1500 m speed was the same
runners changed foot strike point during the race, proportion of the average speed in the 800 m (91.5%)
the predominant position was used to classify them. for both males and females, suggesting that runners in
If there were an equal number of laps with different both events were of a comparable standard.
foot strike positions, the predominant position was
determined from their first lap contact. Ground
Ground contact characteristics
contact time was determined by recording the
number of frames of video data where the foot was Thirty-one percent of all male and female runners
in contact with the ground. From knowledge of the were forefoot strikers, 42% midfoot strikers, and 27%
frame rate, each frame equated to 0.01 s. heel strikers. Table I shows the values for each
individual event. There was less variation between the
races for the females. Chi-square analysis revealed no
Statistical analysis
significant differences between events and within
Before statistical analysis, average race speed and events the only significant difference was the low
average ground contact times were checked for proportion of heel strikers in the men’s 800 m.
normal distribution using the Kolmogorov-Smirnov The mean ground contact times (see Table II) were
test and Q-Q plots and deemed to be normally 171 + 16 ms, 180 + 14 ms, 162 + 12 ms, and
distributed. All values are expressed as means and 173 + 16 ms for women’s 800 m, women’s
standard deviations. Statistical analyses were per- 1500 m, men’s 800 m, and men’s 1500 m races,
formed using SPSS version 16.0 for Windows (SPSS respectively. The men’s 800 m had shorter ground
Inc., Chicago, IL). A one-way analysis of variance contact times than all other events (F3,177 ¼ 16.174;
(ANOVA) was used to compare mean ground P 5 0.001), while the women’s 800 m was different
contact times for the three different foot strike from the women’s 1500 m. No difference was evident
patterns in each event. Where significant differences between the men’s 1500 m and women’s 800 m.
occurred, a Tukey post-hoc test was employed. Analysis of variance revealed that when all runners
Differences in ground contact time between laps were combined, both forefoot and midfoot strikers
were determined by a paired samples t-test for the had shorter average ground contact times than heel
800 m races (two foot strikes) and a one-way strikers (F2,178 ¼ 25.933; P 5 0.001). Furthermore,
ANOVA with repeated measures for the 1500 m these differences were of a large magnitude (heel
races (four foot strikes). Where significant differ- strikers vs. forefoot strikers, ES ¼ 1.3; heel strikers
ences occurred for the 1500 m races, a Bonferroni
post-hoc test was employed. Pearson’s product– Table I. Proportions of forefoot, midfoot, and heel strikers during
moment correlations were used to establish the men’s and women’s middle-distance races.
relationships between average race speed and average
Event Forefoot Midfoot Heel
ground contact time. Statistical significance was set
at P 0.05. Effect sizes (ES) were calculated using Women’s 800 m (n ¼ 34) 27 41 32
the method of Cohen (1998) and interpreted using Women’s 1500 m (n ¼ 24) 25 42 33
the scale of Lipsey (1990). The interpretation of Men’s 800 m (n ¼ 71) 35 50 15
Men’s 1500 m (n ¼ 52) 37 37 26
correlation coefficients was performed using the
All runners (N ¼ 181) 31 42 27
scale of Hopkins (2006).
1278 P. Hayes & N. Caplan
Table II. Effect size, mean difference between, and 95% confidence interval of, the differences for ground contact time between differing
points of ground contact.
All runners Women’s 800 m Women’s 1500 m Men’s 800 m Men’s 1500 m
different foot strike positions. A consistent pattern the effect sizes for comparisons between midfoot and
was found for the differences across each event: both forefoot strikers were small (ES 5 0.2). However,
midfoot and forefoot strikers showed a difference of the effect sizes comparing forefoot with heel strikers
at least 0.11 m s71 compared with heel strikers, were all moderate, ranging from 0.6 (women’s
while there was a difference of less than 0.04 m s71 800 m, women’s 1500 m, men’s 800 m) to 0.8
between midfoot and forefoot strikers. For all events, (men’s 1500 m). The comparisons for midfoot and
heel strikers were more variable, ranging from 0.3 for
the women’s 1500 m to 0.8 for the men’s 1500 m.
Significant relationships were found between
ground contact time and average race speed for
each event, and all runners combined (see Table III).
Figure 4 shows the relationship between ground
contact time and average race speed for all runners
combined. It is notable that similar correlations were
achieved for all analyses with the exception of the
men’s 800 m, for which a moderate correlation was
Downloaded by [North Carolina State University] at 15:36 28 March 2013
attained.
Discussion
The main findings were that runners made initial
ground contact with predominantly either their
forefoot or midfoot, with heel strikers constituting
just over a quarter of runners. As hypothesized, the
forefoot and midfoot strikers had shorter ground
contact times than heel strikers, which was to some
extent reflected in faster race times. There were
large, negative relationships between average ground
contact times and average race speed across events.
Changes in ground contact times were evident with
increases between the first and last lap in each event.
Foot strike patterns ground contact time between positions for the
To date, two large field studies have examined foot women’s 800 m or 1500 m were found, although
strike characteristics in long-distance runners. Kerr the numbers in each foot strike position were low.
and colleagues assessed recreational runners Batterham and Hopkins (2006) highlighted that the
(N ¼ 624) over 10 km and competitive runners magnitude of effect is of greater importance than
(N ¼ 125) at two points during a marathon (Kerr, statistical significance, with the latter depending
Beauchamp, Fisher, & Neil, 1983). They found 81% upon, among other things, sample size. The magni-
of runners to be heel strikers. Hasegawa et al. (2007) tude of effect for each women’s event supports this
found approximately 75% of high-calibre half mara- pattern of evidence, even where no statistically signi-
thon runners (N ¼ 415) were heel strikers, with ficant differences occurred. Furthermore, the 95%
midfoot and forefoot strikers being associated with confidence intervals did not include 0, suggesting a
faster running speeds. Novacheck (1998), by con- real difference; however, the lower confidence inter-
trast, suggested that sprinters were likely to adopt a val was below the measurement error in our study.
forefoot pattern. As expected, the results from this Hasegawa et al. (2007) reported longer ground
study fall between these two extremes with no more contact time during heel strike (199.8 ms) compared
than one-third of the runners in any event adopting a with our data for combined midfoot and forefoot
Downloaded by [North Carolina State University] at 15:36 28 March 2013
heel strike pattern. The female runners showed a strikers (187.4 ms). The results from this study are in
remarkably similar pattern for both the 800 and accord with this, showing that forefoot and midfoot
1500 m. In the male runners, the proportion of strikers tended towards shorter ground contact time.
forefoot strikers was similar in the two events, Williams and Cavanagh (1987) found that heel
although there were considerably more midfoot strikers had a more extended leg at the point of con-
strikers in the 800 m and therefore concomitantly tact, a longer time spent at maximum knee flexion,
fewer heel strikers. To the best of our knowledge, and longer ground contact time than those runners
there are no comparable field-based data for 800 m landing further forward on their foot. These findings,
or 1500 m races. combined with those of Ardigo and colleagues
(Ardigo, Lafortuna, Minetti, Mognoni, & Saibene,
1995), who found longer acceleration times during
Foot strike pattern and ground contact time
ground contact in heel strikers, led Hasegawa et al.
Ground contact times in this study were shorter than (2007) to propose that heel striking required more
those previously reported for longer distance races, time for knee joint movement and possibly muscle
and longer than those recorded for shorter distances. activation to accelerate their leg. While this may
Notwithstanding the one participant reported who be correct, it is also possible that the shorter ground
completed a 1500 m (Williams et al., 1991) but for contact time enables greater use of stored elastic
whom values were not reported separately, this study energy (Ardigo et al., 1995; Bosco, Viitasalo, Komi,
appears to be the first to record ground contact time & Luhtanen, 1982).
during 800 m or 1500 m races. Elliott and Roberts
(1980) reported a mean value of 190 ms for 3000 m,
Point of ground contact and average race speed
while for 5000 m there are reported values of
between 203 and 207 ms (Nummela et al., 2008; In the present study, average race speed was
Paavolainen et al., 1999; Williams et al., 1991). significantly different depending on the point of
During 400 m races, ground contact times of foot contact. When all runners were combined,
between 110 ms and 145 ms have been reported forefoot and midfoot strikers were significantly faster
(Bates & Haven, 1973; Bates et al., 1977; Hobara than heel strikers; however, with one exception, there
et al., 2010). In all of these studies, foot strike were no differences within each event. While there
position was not recorded. Compared with previous were no differences within events, the effect sizes
competitive races of varying distances, the ground tended to follow the trend shown for all runners
contact times in the present study are in accordance combined. Overall, these results show some support
with the notion that ground contact time decreases in for the notion that midfoot and forefoot strikers have
line with increasing running speed (Luhtanen & a faster average race speed than heel strikers.
Komi, 1986; Salvidge & Hayes, 2010). Hasegawa et al. (2007) showed that shorter
Similar to Hasegawa et al. (2007), we found differ- ground contact time was related to a higher finishing
ences in average ground contact time between fore- position and that forefoot and midfoot strikers
foot, midfoot, and heel strikers, with heel strikers tended to have shorter ground contact times than
having longer contact time for all runners combined heel strikers. Taken as a whole, our findings would
and each of the men’s events. There were large effect support this with shorter ground contact times and
sizes for both midfoot and forefoot strikers compared faster average race speed in forefoot and midfoot
with heel strikers. No significant differences in strikers compared with heel strikers. Based on
Ground contact times in middle-distance races 1281
findings from several studies (Ardigo et al., 1995; the fastest and slowest runners. This density of the
Heise & Martin, 2001; Kyrolainen et al., 2001; sample may have contributed to the moderate
Spurrs, Murphy, & Watsford, 2003; Williams & correlation observed. In contrast to this, the coeffi-
Cavanagh, 1987), Hasegawa et al. (2007) hypothe- cients of variation (CV) for average race speed for
sized that midfoot and forefoot strikers may be more each event were similar across the events (CV ¼ 3.0–
economic than heel strikers. By contrast, Kram and 3.7%). In absolute terms, however, this event had the
Taylor (1990) proposed that the energetic cost of smallest range of race times and the largest number
running relative to body weight is inversely propor- of competitors. The fastest runner was 1.08 s quicker
tional to the rate of force application. They used the than any other competitor, leaving 70 competitors
reciprocal of ground contact time as a proxy measure spread over 13.38 s (CV ¼ 2.86%), an even more
for the rate of force application. On this basis, at any densely grouped sample. Further research is required
given speed, reducing ground contact time would to identify whether the weaker relationship in the
increase the oxygen cost, thereby making the runner men’s 800 m is real or a statistical artefact.
less economic.
For many years, coaching folklore (Watts et al.,
Time course of changes
1972) has advocated the heel strike first pattern for
Downloaded by [North Carolina State University] at 15:36 28 March 2013
middle- and long-distance runners. The results of A main finding of the present study is that in all of
this study and those of Hasegawa and colleagues the races the first lap had significantly shorter average
clearly show that high-level runners do not necessa- ground contact time than the last lap in each event.
rily follow this pattern. These data cannot validate An increased ground contact time is consistent with
claims in recent publications (Dreyer, 2003; Roma- fatigue-associated changes seen in competitive races
nov, 2002; Yessis, 2000) that advocate a midfoot over 400 m (Bates et al., 1977; Hobara et al., 2010)
strike is best. To substantiate these claims, further and treadmill running to exhaustion (Gazeau et al.,
intervention studies are required. 1997; Williams et al., 1991). Concomitant with
increased ground contact time is often an increase in
step length and reduction in running speed (Bates
Relationship between average ground contact time and
et al., 1977; Hobara et al., 2010; Kadono, Ae,
average race speed
Enomotot, Sugita, & Morioka 2008). Longer dis-
One of the most striking findings of this study was tance races, however, have not always shown
the similarity in the strength of the relationship increases in stride length (Elliott & Ackland, 1981;
between ground contact time and average race speed Elliott & Roberts, 1980; Williams et al., 1991).
across all runners combined and in each of the During competitive 800 m races, using a group of 49
events, with the exception of the men’s 800 m. The male runners of similar performance capability to the
similarity of these relationships could suggest a present study, Kadono et al. (2008) showed a
systematic effect, although further research is re- progressive decrease in both running speed and
quired to confirm this. This is, however, consistent stride length from 200 m onwards. They did not,
with Luhtanen and Komi (1986), who found a however, record ground contact time and it is not
reduction in ground contact time with increasing therefore possible to determine if the decrease in
running speed. Traditional explanatory models of stride length was related to increased ground contact
endurance running performance (Coyle, 1995; Di time. The ground contact phase is the only point of
Prampero et al., 1986; Joyner, 1991) are essentially the running gait where propulsive force can be
comprised of metabolic and morphological compo- generated by the runner. It is likely that any
nents, whereas Paavolainen and colleagues’ (1999) mechanism causing fatigue during this phase will
model incorporated both metabolic and neuromus- negatively impact upon performance. Paavolainen
cular factors. The findings for ground contact time and colleagues’ (1999) model, however, proposed
from this study would fit within the neuromuscular that fatigue can potentially be caused by a number of
aspects of Paavolainen’s model. In terms of model- different mechanisms, including changes in ground
ling endurance running performance, the strength of contact characteristics. Further study is required to
our correlations, while not exclusively explaining clarify how the increases in ground contact time
performance, suggest that ground contact time affect middle-distance performance.
should be incorporated in future models. The increase in ground contact time over the
The men’s 800 m exhibited a weaker relationship course of a race might, on first viewing, suggest that
between average race speed and average ground some element of fatigue has occurred during the
contact time. Correlations are affected by both races. There may be an alternative explanation
sample size and the spread of scores (Bates, Zhang, whereby athletes are slowing down and that ground
Dufek, & Chen 1996). In the men’s 800 m event, contact time is simply related to running speed
there were 71 competitors with only 14.46 s covering (Luhtanen & Komi, 1986; Salvidge & Hayes, 2010),
1282 P. Hayes & N. Caplan
although the impact of fatigue on the relationship speeds. There were strong correlations between
between running speed and ground contact time is ground contact time and average race speed for all
unknown. For all 22 races observed in the current events except the men’s 800 m, which had a
study, there was a pacemaker for at least the first moderate correlation. During both 800 and
50% of the race. In each race, the pre-determined 1500 m races, ground contact time increased from
pace was faster than the final average race speed. the first to last lap. Two plausible explanations are:
Split times from each race were not available, the increase in ground contact time is reflective of
although experienced pacemakers were used in fatiguing processes or changes in ground contact
each race. Our subjective observation was that the time are due to changes in running speed.
pre-determined split times were invariably met or
exceeded by the pacemakers and competitors alike,
References
thereby suggesting that the runners slowed down.
This would agree with previous work looking at Anderson, T. (1996). Biomechanics and running economy. Sports
pacing strategies in men’s elite (Tucker, Lambert, & Medicine, 22, 76–89.
Ardigo, L. P., Lafortuna, C., Minetti, A. E., Mognoni, P., &
Noakes 2006) and sub-elite (Kadono et al., 2008) Saibene, F. (1995). Metabolic and mechanical aspects of foot
800 m runners. Noakes and colleagues (Noakes,
Downloaded by [North Carolina State University] at 15:36 28 March 2013
Heise, G. D., & Martin, P. E. (2001). Are variations in running Nummela, A. T., Paavolainen, L. M., Sharwood, K. A., Lambert,
economy in humans associated with ground reaction force M. I., Noakes, T. D., & Rusko, H. K. (2006). Neuromuscular
characteristics? European Journal of Applied Physiology, 84, 438– factors determining 5 km running performance and running
442. economy in well-trained athletes. European Journal of Applied
Hobara, H., Inoue, K., Gomi, K., Sakamoto, M., Muraoka, T., Physiology, 97, 1–8.
Iso, S. et al. (2010). Continuous change in spring-mass Paavolainen, L., Hakkinen, K., Hamalainen, I., Nummela, A., &
characteristics during a 400 m sprint. Journal of Science and Rusko, H. (1999). Explosive-strength training improves 5-km
Medicine in Sport, 13, 256–261. running time by improving running economy and muscle
Hopkins, W. (2006). http://sportsci.org/resource/stats/index.html. power. Journal of Applied Physiology, 86, 1527–1533.
Humphreys, J., & Holman, R (1985). Focus on middle-distance Romanov, N. (2002). Pose method of running. New York: Dorling
running. London: A&C Black. Kindersley.
Joyner, M. J. (1991). Modeling: Optimal marathon performance Salvidge, C., & Hayes, P. R. (2010). Changes in muscle tendon
on the basis of physiological factors. Journal of Applied unit stiffness with increasing running speed. British Journal of
Physiology, 70, 683–687. Sports Medicine, 44, i27.
Kadono, H., Ae, M., Enomotot, Y., Sugita, M., & Morioka, Y. Saunders, P. U., Pyne, D. B., Telford, R. D., & Hawley, J. A.
(2008). The racing patterns of male 800 m runners of different (2004). Factors affecting running economy in trained distance
recorded levels. Japanese Journal of Physical Education, Health runners. Sports Medicine, 34, 465–485.
and Sport Science, 53. Siler, W. L., & Martin, P. E. (1991). Changes in running pattern
Kerr, B. A., Beauchamp, L., Fisher, V., & Neil, R. (1983). during a treadmill run to volitional exhaustion: Fast versus
Downloaded by [North Carolina State University] at 15:36 28 March 2013
Footstrike patterns in distance running. In B. M. Nigg & B. A. slower runners. International Journal of Sports Biomechanics, 7,
Kerr (Eds.), Procedings of the International Symposium on the 12–28.
Biomechanical Aspects of Sports Shoes and Playing Surfaces (pp. Slocum, D. B., & James, S. L. (1968). Biomechanics of running.
135–142). Calgary, Alberta: University of Calgary Press. Journal of the American Medical Association, 205, 721–728.
Kram, R., & Taylor, C. R. (1990). Energetics of running: A new Spurrs, R. W., Murphy, A. J., & Watsford, M. L. (2003). The
perspective. Nature, 346, 265–267. effect of plyometric training on distance running performance.
Kyrolainen, H., Belli, A., & Komi, P. V. (2001). Biomechanical European Journal of Applied Physiology, 89, 1–7.
factors affecting running economy. Medicine and Science in Tucker, R., Lambert, M. I., & Noakes, T. D. (2006). An analysis
Sports and Exercise, 33, 1330–1337. of pacing strategies during men’s world-record performances in
Lipsey, M. W. (1990). Designing sensitivity: Statistical power for track athletics. International Journal of Sports Physiology and
experimental research. Newbury Park, CA: Sage. Performance, 1, 233–245.
Luhtanen, P., & Komi, P. V. (1986). Mechanical factors Watts, D., Wilson, H., & Horwill, F. (1972). The complete middle
influencing running speed. In E. Asmussen & K. Jorgensen distance runner. London: Stanley Paul.
(Eds.), Biomechanics VI-B (pp. 23–29). Baltimore, MD: Williams, K. R., & Cavanagh, P. R. (1987). Relationship between
University Park Press. distance running mechanics, running economy, and perfor-
Noakes, T. D., Lambert, M. I., & Hauman, R. (2009). Which lap mance. Journal of Applied Physiology, 63, 1236–1245.
is the slowest? An analysis of 32 world mile record perfor- Williams, K. R., Snow, R., & Agruss, C. (1991). Changes in
mances. British Journal of Sports Medicine, 43, 760–764. distance runners’ kinematics with fatigue. International. Journal
Novacheck, T. F. (1998). The biomechanics of running. Gait and of Sports Biomechanics, 7, 138–162.
Posture, 7, 77–95. Yessis, M. (2000). Explosive running. Colombus, OH: McGraw-
Nummela, A. T., Heath, K. A., Paavolainen, L. M., Lambert, M. Hill.
I., St. Clair Gibson, A., Rusko, H. K. et al. (2008). Fatigue
during a 5-km running time trial. International Journal of Sports
Medicine, 29, 738–745.