Journal Logo


Lower-Limb Mechanics during the Support Phase of Maximum-Velocity Sprint Running


Author Information
Medicine & Science in Sports & Exercise: April 2008 - Volume 40 - Issue 4 - p 707-715
doi: 10.1249/MSS.0b013e318162d162
  • Free


The study of joint kinetics can improve the understanding of the underlying causes of a movement (22). In sprinting, where the goal is to cover the allotted distance in the least possible time, power production in the muscles of the leg is crucial to high performance. The anatomy of the leg, with a combination of mono- and biarticular muscles, has been shown to facilitate the transmission of power from the larger, proximal muscles to the smaller, distal muscles, and, therefore, to the track (6).

Although there have been several studies published regarding joint kinetics in sprinting, these studies have been conducted on varying phases of the sprint, and in athletes of varying ability. Moment data are available from sprinting during support from the second step after the start (6), the acceleration phase (4,7), and the maximum-velocity phase (1,9,10,13), as well as during the swing phase at maximum velocity (17). Comparisons between the studies are interesting, although because of the differences in technique between the different phases (5,15), these comparisons must be treated with caution. Furthermore, power values have only been reported in three studies (1,7,17), and, to the knowledge of the authors, "work" has not yet been reported in any phase of a sprint run. It appears that the joint kinetics of maximum-velocity sprinting have yet to be fully explained, and that the study of well-trained sprinters during the support phase will provide useful information. The overall aim of this study was to improve the understanding of sprint technique in well-trained sprinters in the maximum-velocity phase. By improving the understanding of the muscular contributions to sprint performance, the potential for assessment and adaptation of sprint performance in an applied setting will be improved.


Four male sprinters gave written informed consent to participate in this study. Subject details are given in Table 1. Approval for the study was obtained from the local regional ethics committee.

Subject information.

Data were collected in an indoor track and field hall in late November when the athletes were performing maximum-velocity training. A force plate (9287BA, Kistler Instruments Ltd.) was placed in a customized housing in the center of a lane on the infield of the track, and it was covered with a secured piece of the synthetic track surface (Mondo) to preserve ecological validity. The force plate was isolated from the track foundations in accordance with the manufacturer's guidelines, and the covering of the force plate was separate from the surrounding covering. A high-speed camera (MotionPro HS-1, Redlake) was placed perpendicular to the direction of the sprint, 25.0 m from the center of the running lane and 1.0 m above the track surface, with its 3.0-m field of view centered on the force plate. The high-speed camera was set up with a frame rate of 200 Hz, a shutter speed of 1/600 s, an open iris with no gain, and was manually focused. The resolution of the image was 768 × 604 pixels. A 2.5 × 2.0-m vertical calibration plane containing six reference points was located in the center of the high-speed camera's field of view, in the center of the running lane in the sagittal plane. A 50-Hz digital video camera (DCR-TRV 900E, Sony) was located 3.5 m above the track surface, 6.3 m away from the center of the running lane, and 1.5 m before the center of the force plate, to give a field of view of 6.5 m in the direction of the running lane. This camera was used to measure the horizontal velocity, step length, and frequency of the step from the force plate. The 50-Hz camera was set up with a shutter speed of 1/600 s and an automatic iris, and it was manually focused. A second vertical calibration plane (6.0 × 2.0 m), also comprising six reference points, was located in the center of the 50-Hz camera's field of view, in the center of the running lane in the sagittal plane. Images of each calibration plane were captured with the respective cameras before the commencement of the running trials. A single synchronization unit was used to link the high-speed camera with the force plate. Data collection on both pieces of equipment was initiated with a common external trigger. The synchronization system was simultaneously linked to a bar of 20 LEDs, which were used to synchronize the 50-Hz camera, as described by Kerwin and Trewartha (8). The area around the force plate was illuminated with a total of 7600 W of floodlighting to provide a sufficiently bright image on the high-speed camera.

A photocell timing system (Speed-Trap 2, Brower Timing Systems) was placed along the lane of interest with each beam located 15.0 m to either side of the center of the force plate. This provided the athletes with split-time feedback immediately after each run, to encourage competitiveness throughout the session. A starting check mark was used for each athlete, to aid striking the force plate without the need to alter technique in the immediately preceding steps (targeting). The check mark was located approximately 45 m before the force plate, based on pilot work carried out at a previous training session. The athletes performed their own warm-up, prescribed by the coach. During the warm-up, the athletes adjusted their starting check mark after each run when necessary, based on the coach's advice. Once the researcher and each of the athletes were happy that the starting check mark was correctly located, testing began. Each athlete performed six 60-m sprints, consisting of a 30-m build-up followed by a timed "flying 30 m," within which the force plate was centered. A trial was deemed successful if the athlete was able to strike the force plate at maximum velocity without noticeably or consciously altering his stride pattern. In an attempt to maximize the number of successful trials and simplify the two-dimensional analysis, the check marks were located such that the contact with the force plate would always occur with the right foot. Each athlete achieved two successful trials from the six runs.

Video data from the 50-Hz camera were imported into Target (Loughborough Innovations Limited) and digitized using a 20-point model, comprising shoulder, elbow, wrist, fingertip, hip, knee, ankle, head of the second metatarsal and toe on each side of the body, and top of the head and base of the neck. The 20-point model was used to determine mass center location in the calculation of velocity. Video data from the high-speed camera were imported into Peak Motus (v8.1.4.0, Peak Performance Technologies, Inc.) and digitized using a five-point model, comprising head of the second metatarsal, and the ankle, knee, hip, and shoulder joint centers on the side of the support (right) leg on the force plate. This experimental setup enabled the five-point model to be used in the calculation of joint kinematics and kinetics, which would be representative of a single-support phase in sprinting, albeit not necessarily for the dominant foot. Trial sequences were digitized for all fields in which all points of the model were within the camera view. Calibration sequences were digitized for 10 consecutive fields to attenuate the influence of individual digitization errors. One trial (1A) was digitized three times on separate days to determine the amount of random error introduced to the calculation by the digitizing process. All digitized data were exported from the digitizing software and reconstructed using the 2D-DLT (18), with lens correction included.

Vertical and horizontal coordinates of each of the five digitized points from the high-speed camera for each successful trial were padded by reflection with 10 extra data points (16). These coordinates and the anteroposterior and vertical ground-reaction forces for each individual trial, which had been sampled at 1000 Hz, were subjected to a residual analysis, using a fourth-order Butterworth filter, to determine the optimum cutoff frequency (22). Raw and filtered ground-reaction forces (absolute and normalized to body weight) from trial 1A are shown in Figure 1. Once filtered at the optimum cutoff frequency based on this residual analysis, the ground-reaction force data were matched to a video frame from the high-speed camera and were extracted at 200 Hz. However, the instant of touchdown was identified using the 1000-Hz force data, being the first data point in which the vertical force value increased to and remained greater than two standard deviations above the zero-load level. Consequently, the first video field in each contact did not necessarily occur at time-zero. Body segment inertia parameters were taken from de Leva (2), with the exception of the foot segment, for which data were taken from Winter (22). The mass of a typical sprinting shoe (200 g) was added to the mass of the foot segment (4).

Raw (A) and filtered (B) horizontal and vertical ground-reaction forces (GRF) for trial 1A.

Moments were calculated by standard inverse dynamics equations, as presented by Winter (22). Extension was denoted as positive at each of the three leg joints. Power was calculated by multiplying the moment by the joint angular velocity. The amount of work done at each joint was calculated for each power phase as the time integral of the power curve. A power phase was defined as a period of continuous positive or negative work at a joint (21), with the exception of the hip in early stance, where an occasional brief dissipation burst was included in the dominant generation phase (Fig. 2). The power phases (areas) that occurred in the support phase of a sprint step are identified in Figure 2, and these are labeled A1 and A2 at the ankle, K1, K2, …, etc. at the knee, and H1 and H2 at the hip. Moment, power, and work variables were scaled to body weight and height in accordance with the recommendations of Hof (3); moment and work were divided by body weight and height, while power was divided by body mass, root gravitational acceleration (g1/2), and height3/2, meaning that all joint kinetic variables were thus dimensionless. Variables were scaled to height rather than leg length, to allow comparison with previous sprinting studies that had only reported athlete heights. Leg length values for subjects in this study, measured from the greater trochanter to the floor during normal standing, are provided in Table 1 to allow the reader to scale values to leg length for future cross-subject comparisons. Joint angular velocity, moment, and power data sets for each trial were interpolated using a cubic spline (MathCad 13, Adept Scientific) from 0 to 100 points, to facilitate the calculation of group mean and standard deviation values for each variable. The mean RMS differences in moment and power between the three separate digitized sequences of trial 1A were calculated. This was done by calculating the mean of the RMS differences between each digitized sequence and the mean value from the other two digitized sequences. The RMS differences in moment and power were also calculated between trials 1A and 1B for comparison with the RMS differences attributable to redigitization.

Identification of power phases at the hip (A), knee (B), and ankle (C) in trial 1A.

Horizontal velocity, step length, and step frequency of one step in each trial were calculated using the information taken from the 50-Hz camera. The step cycle was defined as beginning with the instant of touchdown on the force plate, and finishing with the subsequent contact of the contralateral foot. Velocity was calculated as the change in horizontal displacement of the whole body mass center from one contact to the next (where the displacement at each individual contact was taken as the mean value from the last field of flight and the first field of support) divided by the time between the two contacts. Step length was calculated as the displacement between the toe points in the first field after touchdown in two consecutive steps, and step frequency was calculated as the velocity divided by the step length. Step lengths were normalized by dividing absolute values by height, to allow for cross-subject comparisons.


Horizontal velocity, step length, step frequency, and normalized step length for each successful trial are presented in Table 2. The results reveal that two subjects (1 and 2) ran both trials in excess of 10.00 m·s−1 in velocity, while the remaining two subjects ran both trials between 9.00 and 10.00 m·s−1. Across all subjects, step lengths ranged between 1.94 and 2.35 m (or 1.14 and 1.28 when normalized). On an intrasubject basis between trials, differences were between 0.01 and 0.09 m (or 0.00 and 0.05 when normalized). Step frequency ranged between 4.10 and 4.68 Hz across all subjects, and on an intrasubject basis between trials it varied between 0.06 and 0.21 Hz.

Horizontal velocity, step length (absolute and normalized to body height), and step frequency for the complete step cycle measured from the instant of contact with the force plate in each successful trial.

Angular velocities and normalized net moments and powers from the three joints of the support leg during contact are presented for all athletes as the means ± standard deviations in Figure 3A-I. Figure 4A-L shows angular velocity, moment, and power from both successful trials for each athlete at the knee joint only. See Figure 3 for comparison with SI unit data. All subjects exhibited a dorsiflexion followed by plantar flexion movement pattern at the ankle and flexion-extension pattern at the knee throughout contact. The hip joint extended for the duration of the support phase. The highest velocities, in both flexion and extension, were seen in the ankle joint in all athletes, with dorsiflexion velocity peaking at −15.9 ± 3.4 rad·s−1, and plantar flexion velocity peaking at 26.9 ± 4.2 rad·s−1, across all athletes. Peak knee flexion and extension velocities reached −8.0 ± 1.5 and 10.5 ± 2.5 rad·s−1, respectively. Hip extension velocity peaked at 15.2 ± 1.7 rad·s−1.

Joint angular velocity, moment, and power at the ankle (A-C), knee (D-F), and hip (G-I) for both trials for all subjects during the support phase (mean ± standard deviation). Vertical dashed line corresponds to the transition from the braking to the propulsive phase of support.
Joint angular velocity, moment, and power at the knee for both trials for subjects 1 (A-C), 2 (D-F), 3 (G-I), and 4 (K-L) during the support phase.

The ankle moment was predominantly plantar flexor throughout the majority of stance, although in some trials, a small dorsiflexor moment was present shortly before take-off. Peak plantar flexor moment ranged from 217 N·m in trial 3A to 429 N·m in 4B (mean = 336 ± 83 N·m). Once normalized to body weight and height, the equivalent range was 0.201-0.289, with a mean value of 0.251 ± 0.042. All subsequent joint kinetic variables have been presented as normalized values only. Intrasubject variation in normalized moment values was generally low. In three subjects, the difference in peak plantar flexion moment between steps was no greater than 0.013, while for subject 4 the difference was 0.030. Knee moment values were generally lower in magnitude than either ankle or hip, and they shifted more frequently between flexor and extensor. The typical pattern of the knee moment throughout stance was flexor-extensor-flexor-extensor-flexor, with trial 1B having one further extensor moment just before take-off. In all trials, there was a relatively long-duration extensor moment during midstance (Fig. 3E), which peaked between 0.031 and 0.134 (mean = 0.092 ± 0.033). Hip moment was predominantly extensor through approximately the first two thirds of stance, becoming flexor for the remainder. In some steps, there was a short burst of flexor moment dominance either shortly after touchdown or during midstance (Fig. 3H). The extensor moment throughout the first part of contact was double peaked, with a large peak shortly after touchdown (mean = 0.476 ± 0.114) and a smaller peak around midstance (mean = 0.190 ± 0.063). The magnitude of the hip flexor moment tended to increase throughout the latter part of support.

The ankle dissipated power through approximately the first half of stance, and then it generated power for the remainder of the support phase (Fig. 3C). Magnitudes of both peaks were large, reaching −6.911 ± 1.317 and 5.561 ± 1.170 for dissipation and generation, respectively. Power values at the knee were generally small in magnitude and constantly, although consistently, changing in direction, much like the moment (Fig. 3F). The largest peak power on average for all subjects at the knee was a generation peak shortly after touchdown (mean = 1.986 ± 0.898). Hip power generally exhibited two large generation peaks at early and midstance, occasionally separated by a short-duration dissipation burst, and followed by a dissipation peak before take-off (Fig. 3I). Magnitudes of the two power-generation peaks were 4.443 ± 1.304 and 3.909 ± 1.264, respectively, while the dissipation peak reached −5.067 ± 1.739. Table 3 displays the RMS differences between the three digitized sequences of the same trial. When expressed as a percentage of the range of the variable from the original sequence, RMS values were 4.4% or less for all kinetic variables other than hip power, for which it was 8.4%. Differences in kinetic variables between two trials for one subject were approximately six times greater (range 3.2-11.5) than the RMS difference between three digitizations of the same trial. The relative magnitude of these differences shows that the errors introduced by the manual digitizing process were sufficiently small that differences between trials were not masked.

Results of error analysis.

Table 4 shows the mean, maximum, and minimum amounts of positive or negative work done during each of the power phases at the ankle, knee, and hip joints. Overall, the ankle joint was a net dissipater of energy, with the magnitude of the negative work phase (A1) being greater than the subsequent positive work phase (A2). Conversely, the hip joint was a net generator of energy, as the magnitude of the initial positive extensor work (H1) was greater than the subsequent negative extensor work (H2). The net knee work performed was not consistent, although in six of the eight trials, the knee joint was a net dissipater. Values measured at the knee during each power phase were typically much lower than those measured at the ankle and hip. The largest-magnitude positive work phase at the knee was contributed by the flexors, which, across all trials, was 0.006 ± 0.003. This was followed by the largest-magnitude dissipation phase, with negative work of −0.010 ± 0.005. Although the magnitudes of work across all phases at the knee were small, both inter- and intrasubject variations were relatively large.

Magnitude of work (SI units and normalized) performed in each power phase at the ankle, knee, and hip during support.


The overall results for this study reveal a generally large magnitude and consistent pattern of joint kinetics at the ankle and hip. However, values at the knee joint for angular velocity and moment, power, and work were generally lower than those measured at the ankle and hip, especially from midstance onward. There was also less consistency in the timing of the changes in kinetic actions at the knee. The kinematic variables measured in this study in the faster subjects (horizontal velocity, step length, and step frequency) were consistent with those reported previously in the men's Olympic 200-m final (12) and in sprinters at supramaximal velocity (14). Velocities were greater than in previous studies of joint kinetics in sprinting (1,10). However, those studies (1,10) do not present step length and step frequency values. The main finding of this study was the generally low magnitudes of the kinetic variables measured at the knee joint, which were noticeably different from previous sprinting studies (1,7,11,13).

Previous studies of joint kinetics in sprinting have not normalized kinetic variables using athletes' body heights or weights. However, height and mass (converted to weight) data presented within each study have allowed the calculation of normalized values, referred to as "adjusted" values in all comparisons with data in this study. The mean peak normalized ankle plantar flexor moment measured in this study of 0.251 ± 0.042 was slightly lower in magnitude than that adjusted from the study of Johnson and Buckley (7) in six sprinters at the 14-m mark of a maximal sprint (0.270 ± 0.073), and higher than the adjusted values of approximately 0.20 from other studies in various phases of a sprint (4,9,10). The plantar flexor power-generation peak of 6.098 ± 1.170 was similar to the adjusted values of 5.930 ± 1.636 (7) and approximately 5.50 (1). However, the preceding plantar flexor power-dissipation peak was greater in this study (−6.911 ± 1.317) than the data adjusted from Johnson and Buckley (7) and Belli et al. (1), where the values were −4.266 ± 1.570 and approximately −1.40, respectively. No previous studies have presented results for work at any lower-limb joints in sprinting. Mean results for this study were −0.093 ± 0.011 for power phase A1 and 0.053 ± 0.010 for power phase A2. Across the published literature, the results for joint kinetics at the ankle have generally been consistent, and the results published in this study are in agreement with those. The one main difference, the magnitude of the plantar flexor power dissipation in early stance, was substantially greater in this study than any other reported values. Of the two studies with which these results were compared, Johnson and Buckley (7) measured joint kinetics in the acceleration phase of the sprint, and Belli et al. (1) measured distance runners during maximum-velocity sprints. The acceleration phase of a sprint has a greater proportion of horizontal relative to vertical force production than the maximum-velocity phase (4,11), because of the need to rapidly develop horizontal velocity. Furthermore, the maximum vertical ground-reaction forces documented by Belli et al. (1) were approximately 1 BW less than in this study (Fig. 1). It therefore appears that the greater plantar flexor power dissipation noted in this study is attributable to a greater vertical loading than had been seen in previous studies. It has previously been shown that the magnitude of the maximum vertical force is positively related to maximum sprinting velocity (19), which provides further evidence for the differences between data presented here and those from previous studies (1,7).

The knee joint results presented in this study differ from those available in the previously published literature, perhaps because of the higher velocity of the athletes than in other studies (1,10) and the different phase of the sprint studied (4,6,7). The knee joint angular velocity for all subjects was flexor for approximately the first half of stance and extensor for the remainder. Only the results of Belli et al. (1) show a similar pattern, although in that study knee extension began after approximately one third of stance. Mann (10) found that knee flexion recommenced before take-off occurred. In the current study, knee flexion did not begin until the first or second field after take-off (at 200 Hz), in three subjects. The exception was subject 1, where knee flexion did not occur until the third field after take-off in both trials. The continuation of extension at the knee joint until after take-off seen in this study is contradictory to the action of beginning flexion before take-off that has been advocated by Mann (11) as being beneficial to performance. It is possible that in extending the knee for longer, the subjects in this study may have been executing an action that was detrimental to their performance.

The time histories of the knee moments and powers have also shown inconsistencies across the published literature, and there were further differences in the results presented in this study. In the literature, the knee moment exhibited a large extensor peak around midstance (1,4,7,10). This was sometimes preceded by a short flexor peak (6,10), which, at times, was itself interrupted by a very brief extensor peak soon after touchdown (4,7). Results from this study followed the pattern described by Johnson and Buckley (7) and Hunter et al. (4), and towards the end of stance there was a flexor moment before take-off. This final flexor moment was matched in the published results only by the data of Johnson and Buckley (7). The flexor moment before take-off was eccentric, which acted to halt the extension described in the previous paragraph. However, it was relatively small in magnitude, and, therefore, it had not acted to change extension to flexion before take-off, which would have aided in the recovery for the next contact.

The magnitude of the major normalized knee extensor moment peak was lower in this investigation than in previously reported studies, with the mean value being 0.092 ± 0.033. Mean adjusted values from other studies include approximately 0.20 (11), 0.220 ± 0.060 (1), and 0.190 ± 0.074 (7), while the typical adjusted value for one elite subject presented by Mann and Sprague (13) was approximately 0.25. Interestingly, adjusted peak knee extensor moment calculated from Kuitunen et al. (9) decreased from 0.222 to 0.174 as running velocity increased from 70% to maximum (9.73 m·s−1). The knee extensor peaks seen in previous research have been attributed to the need to increase horizontal velocity (7,10), although it has also been shown that at maximum velocity, the emphasis of sprint technique switches from horizontal to vertical in nature (11). Therefore, the need for a large knee extensor moment may be diminished, and the smaller knee extensor moment in this study, compared with previous research, shows that the knee joint played a lesser role for these subjects in propulsion in maximum-velocity sprinting than has previously been shown (9,10,13).

The reduced propulsive role of the knee joint in this study compared with previous research is shown more clearly in the power patterns. In the case of the previously published data, the peak knee extensor moment occurred in the latter half of stance, by which time the knee joint was extending (1). This means that there was a period of extensor power generation seen at the knee that, when adjusted, has been measured at approximately 2.718 (1) or 2.987 ± 0.991 (7). In this study, the peak extensor moment at the knee occurred relatively early in stance, while the knee was still flexing. This led to a peak extensor power dissipation of −1.986 ± 0.898, which has not been documented in other studies. The early timing of this action meant that it was responsible for halting the collapse of the limb during the weight-acceptance part of the support phase. When the knee joint began to extend, the extensor moment remained dominant, although the peak value of this power-generation phase was only 0.376 ± 0.310 in magnitude. This phase corresponded to those highlighted in the referenced literature, but it was an order of magnitude lower than previously reported values. This shows that, for the subjects in this study, the knee did not play a substantial power-producing role.

The action at the hip joint measured in this study was generally consistent with data from previous research. There was power-generating extensor action for the first part of stance, followed by power-dissipating flexor action for the remainder of support. However, the pattern displayed in this study reveals a double peak in the extensor moment, something not commonly documented. The first peak, soon after touchdown, reached 0.476 ± 0.114. A peak this early in stance also occurred in only one other study (10) and was approximately 0.34 when adjusted. The second peak, which corresponds to that in the remainder of the literature, averaged 0.190 ± 0.062. Adjusted values from other studies include 0.194 ± 0.046 (1) and 0.310 ± 0.028 (7). The adjusted peak extensor power generation of 6.273 ± 2.101 (7) and 3.188 ± 1.415 (1) occurred at the same stage of contact as the second peak measured in this study of 3.909 ± 1.264. Despite the lower values of moment and power in this study compared with the corresponding actions reported by Johnson and Buckley (7), there was an additional powerful action of the hip joint immediately after touchdown (4.445 ± 1.304). This contributed substantially to the production of positive work at the hip in this study, although no values for comparison are available. The differences seen between the results presented for the hip joint early in support in this study and those presented by Johnson and Buckley (7) might at least in part be attributable to the different sampling rates used for the video data, being 200 and 50 Hz, respectively.

Winter (21) studied moments and powers during jogging and found consistent patterns of power phases at both ankle and knee joints. However, power patterns at the hip were small and inconsistent, which was said to indicate the dual role of the hip flexors and extensors during the gait cycle. That is, along with contributing to support during the early part of stance, those muscles were also responsible for maintaining a stable upper body. Winter suggests that any fine adjustments of the trunk would have masked the major patterns that might otherwise have been evident at the hip (21). The results presented in this study provide some evidence, discussed below, to suggest that a similar dual action might be present in the knee flexors and extensors during maximum-velocity sprinting. Winter's finding (21) is in agreement with the previously developed concept of the support moment (20), which measured the total extension moments of the three leg joints and was found to be relatively consistent despite larger changes in the individual moment values. The support moment has been shown to be a meaningful measure in both walking and jogging (20,21), although, to date, the authors are not aware of any published data regarding the support moment in sprinting. Despite the fact that joint kinetics profiles change considerably between jogging and sprinting (1), it is reasonable to assume that the concept of the support moment is applicable to sprinting, and that the three joints of the leg are capable of compensating for each other. The dominant action of the hip extensors early in stance, as shown in this study, has previously been shown to be a crucial variable to sprint performance (13). The major difference in joint kinetics between jogging and sprinting is the large increase in hip extensor power generation in early stance (1), which therefore leads to a substantial increase in the amount of positive work performed at the hip in sprinting. Because, compared with jogging, the amount of power generation at the hip is greatly increased, it is possible that another joint needs to take on a compensatory role.

The low magnitudes of power and work, and the inconsistency between trials of the same athlete in the duration of the power phases, is characteristic of a compensatory function and is clearly shown for the knee joint in Figure 4. There were major periods of both positive and negative work at the ankle and hip joints, and even when the absolute timings of the switch from dissipation to generation at the ankle and generation to dissipation at the hip varied between the two trials of each athlete, the relative timing of those two events within each trial remained constant. In contrast, the timings of the changes between power phases at the knee were more inconsistent for all subjects other than subject 4. When this is combined with the low magnitude of moment and power values measured at the knee joint, there is evidence that, in the subjects measured in this study, the knee had taken on a compensatory role. It should be highlighted that the sequence of the power phases at the knee were consistent across all eight trials in this study, which differs from the inconsistent findings of Winter (21) at the hip joint during jogging. Winter (21) found the sequence of power phases to be so inconsistent at the hip that it was not possible to label them as had been done at the ankle and knee.

In summary, the profiles of moment, power, and work at the ankle, knee, and hip have been presented in detail, and they generally comprise a powerful extension at the ankle and hip, and low-power, compensatory actions at the knee. Within these general patterns, several key findings have been identified that have helped to characterize maximum-velocity sprinting in well-trained athletes. The foremost finding was the generally low magnitude of the kinetic variables measured at the knee during the contact phase. This differs from previous findings of a powerful extension phase before take-off, but it may be attributable to the high velocity of the subjects and the phase of the sprint studied. Further investigation of this area, with more subjects, is necessary to confirm the reasons for the differences between studies. The durations of power phases at the knee joint were inconsistent both within and between subjects, although the sequence of the power phases was the same in all trials. These characteristics of the knee joint action reveal a compensatory action that has not been documented previously in sprinting. However, a similar role has been identified at the hip joint during jogging (21). The powerful extension of the hip joint during early stance was greater than has typically been reported. Normalized hip moment peaked at 0.476 ± 0.114 soon after touchdown. As a result of this, the magnitude of work performed across all subjects during the power-generation phase (H1) was 0.064 ± 0.018. The large amount of work done at the hip most likely allowed the reduced propulsive action and compensatory role seen at the knee joint, discussed above. Finally, the magnitude of plantar flexor power dissipation measured at the ankle during early stance was considerably higher than previously reported values. This was mainly attributed to differences between the level of the athletes and the phase of the sprint run that was studied in this and in earlier studies.

The authors would like to thank UK Athletics Ltd. for financial support for this work, the coach and athletes for giving up their time to participate in the study, and Neil Bezodis, Tom Loney, Ollie Peacock, and Paulien Roos for their assistance with data collection.


1. Belli A, Kyröläinen H, Komi PV. Moment and power of lower limb joints in running. Int J Sports Med. 2002;23:136-41.
2. de Leva P. Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters. J Biomech. 1996;29:1223-30.
3. Hof AL. Scaling gait data to body size. Gait Posture. 1996;4:222-3.
4. Hunter JP, Marshall RN, McNair PJ. Segment-interaction analysis of the stance limb in sprint running. J Biomech. 2004;37:1439-46.
5. Ingen Schenau GJv, de Koning JJ, de Groot G. Optimisation of sprinting performance in running, cycling and speed skating. Sports Med. 1994;17:259-75.
6. Jacobs R, Ingen Schenau GJv. Intermuscular coordination in a sprint push-off. J Biomech. 1992;25:953-65.
7. Johnson MD, Buckley JG. Muscle power patterns in the mid-acceleration phase of sprinting. J Sports Sci. 2001;19:263-72.
8. Kerwin DG, Trewartha G. Strategies for maintaining a handstand in the anterior-posterior direction. Med Sci Sports Exerc. 2001;33(7):1182-8.
9. Kuitunen S, Komi PV, Kyröläinen H. Knee and ankle joint stiffness in sprint running. Med Sci Sports Exerc. 2002;34(1):166-73.
10. Mann RV. A kinetic analysis of sprinting. Med Sci Sports Exerc. 1981;13(5):325-28.
11. Mann RV. Biomechanical analysis of the elite sprinter and hurdler. In: Butts NK, Gushiken TT, Zarins B, editors. The Elite Athlete. Jamaica (NY): Spectrum Publications, Inc.; 1985. p. 43-80.
12. Mann RV, Herman J. Kinematic analysis of Olympic sprint performance: men's 200 meters. Int J Sports Biomech. 1985;1:151-62.
13. Mann RV, Sprague P. A kinetic analysis of the ground leg during sprint running. Res Q Exerc Sport. 1980;51:334-48.
14. Mero A, Komi PV. Effects of supramaximal velocity on biomechanical variables in sprinting. Int J Sports Biomech. 1985;1:240-52.
15. Mero A, Komi PV, Gregor RJ. Biomechanics of sprint running: a review. Sports Med. 1992;13:376-92.
16. Smith G. Padding point extrapolation techniques for the Butterworth digital filter. J Biomech. 1989;22:967-71.
17. Vardaxis V, Hoshizaki TB. Power patterns of the leg during the recovery phase of the sprinting stride for advanced and intermediate sprinters. Int J Sport Biomech. 1989;5:332-49.
18. Walton JS. Close-Range Cine-Photogrammetry: A Generalised Technique for Quantifying Gross Human Motion [doctoral dissertation]. State College (PA): The Pennsylvania State University; May 1981.
19. 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;89:1991-9.
20. Winter DA. Overall principle of lower limb support during stance phase of gait. J Biomech. 1980;13:923-7.
21. Winter DA. Moments of force and mechanical power in jogging. J Biomech. 1983;16:91-97.
22. Winter DA. Biomechanics and Motor Control of Human Movement. Hoboken (NJ): John Wiley and Sons, Inc.; 2005.


©2008The American College of Sports Medicine