The basic (two-dimensional) kinematics of the lower limb segments during instep soccer kicks have been previously reviewed (Lees, 1996; Lees and Nolan, 1998). These include examination of angular position - time and angular velocity curves during the kick as well as the linear kinematics of the joints involved (Figure 1). In this review, two characteristics of this movement will be described a) that the soccer kick is characterized by segmental and joint rotations in multiple planes b) the proximal-to-distal pattern of segmental angular velocities.

Soccer kick is characterized by segmental and joint rotations in multiple planes
Segmental rotations in multiple planes are observed throughout the kick. During the backswing phase, the kicking leg moves backwards, with the hip extending up to 29° (0° is defined as the neutral orientation with respect to hip flexion / extension, Levanon and Dapena, 1998) with a velocity of 171.9-286.5 deg·s^-1 (Nunome et al., 2002; Levanon and Dapena, 1998). The hip is also slowly adducted and externally rotated (Levanon and Dapena, 1998). The knee flexes (at an angular velocity of 745-860 deg·s^-1) and internally rotates (Nunome et al., 2002). Given that the neutral position of the ankle is 0°, the ankle is plantarflexed (10°), abducted (20°) and slightly pronated (Levanon and Dapena, 1998) reaching maximum plantarflexion velocities of 860 deg·s^-1 (Nunome et al., 2002). The back swing motion of the kicking leg is completed just after ground contact with the hip extended and the knee flexed (Levanon and Dapena, 1998).

Forward motion is initiated by rotating the pelvis around the supporting leg and by bringing the thigh of the kicking leg forwards while the knee continues to flex (Weineck, 1997). The hip starts to flex (reaching values of 20° (Levanon and Dapena, 1998) at speeds up to 745 deg·s^-1 (Nunome et al., 2002; Levanon and Dapena, 1998) and abducts while it remains externally rotated (Levanon and Dapena, 1998). In the same period, the ankle is adducted and plantarflexed whereas supination - pronation motion is minimal (Levanon and Dapena, 1998). Simultaneously, knee extension velocity is maximized (860-1720 deg·s^-1) while external / internal tibial rotation values are generally low and less than 57.3deg·s^-1 (Nunome et al., 2002). Upon impact, the hip is flexed, abducted and externally rotated and the ankle plantarflexed and adducted (approximately 12°) (Levanon and Dapena, 1998).

Proximal-to-distal pattern of segmental angular velocities
The majority of studies on soccer kick biomechanics have identified the importance of proximal-to-distal sequence of segmental angular velocities for kick performance (Dorge et al., 2002; Dorge et al., 1999; Huang et al., 1982; Levanon and Dapena, 1998; Nunome et al., 2002).

During the backswing phase, the thigh angular velocity is nearly minimal while the shank velocity is negative, due to the backward movement of the shank. During the initial part of the forward swing phase, the thigh angular velocity is positive ~286-401 deg·s^-1 (Huang et al., 1982; Lees and Nolan, 1998) whereas a negative shank angular velocity ~286-401 deg·s^-1 (Huang et al., 1982; Lees and Nolan, 1998) is observed. This is due to the instantaneous forward movement of the thigh while the shank moves backwards (until maximal knee flexion is achieved).

As the leg continues its forward movement, both thigh and shank move forward. The angular velocity of the thigh continues to increase and reaches its peak value (~516-573 deg·s^-1) just before the knee starts to extend. At this point, the thigh angular velocity equals the shank angular velocity and, thus, knee joint velocity is zero. As the knee starts to extend, the angular velocity of the thigh declines and the shank velocity increases linearly until ball impact reaching values of 1891 deg·s^-1 (Dorge et al., 1999). At ball impact, the thigh angular velocity is almost zero while the shank and the foot reach peak angular velocity and zero acceleration (Huang et al., 1982).

Joint and segmental movements are the result of moments produced during the kick. Two types of analysis have been reported in the literature: estimation of the net moments exerted around joints (Dorge et al., 1999; Nunome et al., 2002; Roberts et al., 1974) and analysis of motion-dependent moments acting on specific segments (Kellis et al., 2006; Putnam, 1991; Putnam, 1983; Sorensen et al., 1996; Dorge et al., 2002).

Research on joint kinetics during the kick has mainly focused on two issues: first, the magnitude of the moments exerted around lower limb joints and, second, the time-sequence of moment generation during the kick. With respect to the first factor, research has shown that hip flexion moments are almost twice the corresponding knee extension moments (Dorge et al., 1999; Luhtanen, 1988; Nunome et al., 2002; Putnam, 1991; Roberts et al., 1974; Zernicke and Roberts, 1978) during the kick (Table 1). Further, ankle plantarflexion moments are even smaller, reaching 20-30 Nm (Nunome et al., 2002) (Table 1).

The joint moment - time curve patterns during the kick differ between studies (Dorge et al., 1999; Nunome et al., 2002; Roberts et al., 1974). Particularly, during the initial backswing phase, some studies reported very low hip extension values (Roberts et al., 1974) whereas others reported high hip flexion moments (Dorge et al., 1999; Nunome et al., 2002). Further, some studies (Luhtanen, 1988; Nunome et al., 2002; Roberts et al., 1974) reported hip and knee moment - curves with one peak. Hip flexion moments reached maximal value at the end of the backswing whereas maximal knee extension values were observed immediately after, approximately at the end of the leg-cocking phase (Nunome et al., 2002). In contrast, Dorge et al., 1999 reported that the hip and knee moment - time curves demonstrate two peaks during the kick. Particularly, peak hip flexion moment was achieved approximately at 25-30% of kick duration, it then declined and increased again reaching an almost similar peak value just before impact. A curve with two peaks was also observed for the knee moment, with peak moments occurring immediately after the corresponding hip moment peaks. Both hip flexion and knee extension moments significantly decline immediately before impact (Dorge et al., 1999; Huang et al., 1982; Nunome et al., 2002; Roberts et al., 1974) while a recent study (Nunome et al., 2006b) reported an almost minimal hip moment at ball impact. Finally, ankle moments are generally very low during the first half of the kick duration and then increase, reaching maximal values at 70-80% of kick duration (Nunome et al., 2002; Zernicke and Roberts, 1978).

Comparison of previous findings shows a wide range of values for hip and knee joint moments mainly due to methodological differences (Table 1). For example, some studies (Nunome et al., 2002; Putnam, 1991) reported average values during the kick as opposed to instantaneous values reported by others (Dorge et al., 1999; Luhtanen, 1988; Zernicke and Roberts, 1978). Further, three-dimensional models yield higher knee extension moments compared with moments derived using two-dimensional analysis (Nunome et al., 2002; Rodano and Tavana, 1993).

Inverse dynamics models demonstrate several limitations which should also be taken into consideration when explaining soccer kick kinetics (Dorge et al., 1999; Levanon and Dapena, 1998; Nunome et al., 2002). Data processing has a significant impact on the magnitude and the patterns of estimated moments. The most important problem is data smoothing. From the start of the movement until ball impact, joint displacement data could be smoothed using an ordinary filter (i.e. Butterworth filter). However, upon impact there is a sudden change in segmental displacement and velocity values which requires further attention. Application of some filtering techniques may significantly alter the displacement signal by cutting high frequency components leading to an underestimation of the true displacement, velocity and acceleration patterns upon foot - ball impact. For example, Nunome et al. , 2002 illustrated that the use of one direction smoothing shifted the time of hip peak moment towards ball impact compared with bi-directional smoothing, thus altering interpretation of the moment-time curves during the kick. Others have shown that the smoothing routines (polynomial curve fitting) applied to the hip and knee moment data may affect the predicted hip and knee joint moment curves (Huang et al., 1982). Recent data suggest that the use of a modified time-frequency algorithm achieves better capture of segmental motion upon impact compared with traditional filtering techniques, thus improving prediction of segmental moment - time curves during the kick (Nunome et al., 2006b).

Examination of moments exerted in other than the sagital plane also provides additional insight regarding kick performance. For example, prior to ball impact a considerable (~115 Nm) hip adduction moment has been reported (Nunome et al., 2002). This emphasizes the importance of hip adductor and abductors in controlling the orientation of the whole leg (Nunome et al., 2002). Rotation moments around the knee are rather minimal whereas ankle inversion moments (15-20 Nm) are almost equal to plantarflexion moments (Nunome et al., 2002). Despite their small magnitude, ankle moments are important as they may affect the final position of the foot at ball contact which determines not only the "power" of the shot but also the path and direction of the ball after impact.

Being a swing motion, soccer kick is characterised by proximal-to-distal sequence of segment motions. For kicking, this is the action of the thigh which slows down or reverses its motion prior to full knee extension is reached. Such motion is accomplished through exertion of moments generated through the joints at the proximal end of the segment, exertion of several motion-dependent moments generated through segmental interactions as well as the moment of inertia of the segment about a transverse axis passing through its proximal end (Putnam, 1993; Nunome et al., 2006a; Dorge et al., 2002). Putnam, 1991 first quantified both joint and motion-dependent moments acting on the thigh and the shank during the kick by modelling body segments as a series of rigid links rotating about points fixed in a system. It was found that initiation of the thigh movement is achieved through a hip flexor moment. This is followed by increased angular acceleration of the thigh while the knee flexes and the whole leg is being accelerated in the forward direction. As knee extension motion is initiated, the thigh starts to decelerate due to exertion of motion-dependent moments from the shank (Putnam, 1991) as well as a hip flexion moment (Nunome et al., 2002; Putnam, 1991; Dorge et al., 2002). This contradicts previous studies (Luhtanen, 1988; Zernicke and Roberts, 1978) which attributed the backward acceleration of the thigh to exertion of hip extension moment. In a recent study, Nunome et al. (Nunome et al., 2006a) confirmed the findings by Putnam, 1991 regarding the role of the reactive moments from the shank for thigh deceleration; however, in contrast to all previous studies, Nunome et al., 2006a found that the hip flexion moment had minimal influence on thigh deceleration.

The shank angular velocity increases as the knee extends towards the ball. Shank angular velocity is the result of the moments exerted by the knee joint muscles, the moment due to angular velocity and linear acceleration of the thigh, the moment due to gravitational acceleration of the shank and the moments due to hip acceleration (Putnam, 1991). Of these, the most influential are the muscle (extensor) moment and the moment due to the angular velocity of the thigh (Kellis et al., 2006; Dorge et al., 2002; Nunome et al., 2006a). Particularly, a high knee extensor moment is observed when the forward rotation of the lower leg is initiated (Nunome et al., 2006a). After this, the knee muscle moment declines which coincides with the increase of shank angular velocity. From this point onwards and until ball impact, an interaction moment is developed which increases gradually until just prior to ball impact (Nunome et al., 2006a). Nunome et al., 2006a noticed that at the final stages prior to ball impact, the interactive (forward) moment accelerates the shank while the knee muscle moment acts in the opposite direction (backwards) as the muscular system is forced to be stretched due to the rapid segmental action of the shank. This is an important finding as it may assist us to better understand not only the kinetics of soccer kick but the associated activity of the involved musculature. The reader, however, should be aware that a limitation of the above studies is the assumption that motion-dependent moments are independent of joint moments which, in reality, is not the case (Putnam, 1991). Further, estimation is based on kinematic variables and therefore it is particularly sensitive to errors in kinematic data.

To summarize, it becomes apparent that the soccer kick is a complex movement which is driven by two types of moments: those exerted by the muscles around the joints and those exerted by the interaction of adjacent segments. To date, we have found only one study (Nunome et al., 2006a) which presents a global description of soccer kick movement based on both moments exerted. Since the initiation of human movement is normally due to forces exerted by the muscles, one may suggest that joint moment exertion should be linked to motion-dependent moments. However, based on previous simulations Mochan and McMahon, 1980 and Putnam, 1991 commented that this might not be the case. Due to movement complexity, the relationship between joint and interactive moments is non-linear thus making difficult to explain the precise role of joint moments during the movement (Putnam, 1991), although recent evidence is very promising (Nunome et al., 2006a). It is almost certain that further research is necessary to investigate the kinetics of soccer kick motion, taking into consideration moments exerted outside the sagital plane. For example, the role of hip adductors during the initial part of the movement should be explored in relation to the backward movement of the thigh, the exertion of hip extension - flexion moment and perhaps the effects of a motion-dependent moment by the shank whereas a similar type of analysis could be performed for the shank movement. This would allow a better understanding of the "optimal" soccer technique, identification of the major mechanisms that contribute to a fast or an accurate kick as well as the role of specific muscles in various phases of the kick.

Electromyography (EMG) has been used to examine muscle activation patterns to explain the role and level of muscle activation during the kick (Bollens et al., 1987; De Proft et al., 1988; Dorge et al., 1999; Kellis et al., 2004; McCrudden and Reilly, 1993; McDonald, 2002; Orchard et al., 2002). To allow comparisons between different findings, all EMG values are frequently expressed as percentage of the EMG recorded during a maximum isometric effort (MVC).

Examination of EMG activity levels reported in the literature (Table 2) indicates large variations in EMG magnitude and temporal patterns, which prevents extraction of safe conclusions regarding the role of various muscles during the kick.

It appears that joint and segmental movements during the kick are driven by simultaneous activation of a relatively large number of muscles. From an anatomical point of view, some of these muscles or muscle groups produce moments around a joint in opposite directions (antagonists). Early studies in these area have called this observation as "soccer paradox" (Bollens et al., 1987; De Proft et al., 1988) because the higher the simultaneous activity of antagonist musculature, the lower the net moment produced around the joint and less powerful the resulting segmental action. In other words if both agonist and antagonist muscles co-contract, they produce opposing forces around a joint. The result of this action is a low net joint moment. This may enhance the stability of the joint but the movement becomes inefficient. However, examination of muscle function should take into consideration several factors such as the function of each skeletal muscle (bi-articular vs uniarticular), the type of action (eccentric vs concentric), the simultaneous movement of adjacent joints and segments and the time where each muscle is activated during the movement.

Some studies examined joint kinematics and kinetics in combination with activation patterns of specific muscles with somewhat different views regarding kick kinematics (Dorge et al., 1999; Sorensen et al., 1996). Although the study by Sorensen et al., 1996 refers to martial arts kick, two observations on muscle activity are worth mentioning. First, that the thigh acceleration was accompanied by considerable levels of rectus femoris (hip flexors) and hamstring (hip extensors) EMG as well as a high motion-dependent moment from the shank (Sorensen et al., 1996). Second, that shank acceleration was accompanied by a high muscular activity of the knee extensors whereas when the shank starts to decelerate this activity decreased whereas the hamstring and gastrocnemius (antagonistic) activation increased. Based on these observations, Sorensen et al., 1996 suggested that the hip muscles play a compensatory role during the backswing phase and an active role during the forward swing phase. Furthermore, it was proposed that thigh deceleration is mainly caused by interactive moment exerted by the shank rather than a hip extension or flexion moment. The recorded rectus femoris and hamstring activation were considered as indicative of their compensatory role for thigh deceleration. Due to the position of the hip and the knee (Hoy et al., 1990) during the backswing, it was also considered that the hamstrings exerted more torque around the knee rather than the hip.

Dorge et al., 1999 supported the active role of muscles throughout the kick which is partly in contrast to Sorensen et al., 1996. Using needle electrodes, these authors (Dorge et al., 1999) suggested that the EMG activity levels correspond to the proximal-to-distal segmental movement observed during the kick. Particularly, there was a high activation of the iliopsoas during the start of the kick which was followed by a high activation of the rectus femoris during backswing. In turn, the main forward swing phase was characterized by high activation of vastus lateralis. The biceps femoris and gluteus maximus demonstrated their peaks just prior to ball impact. In terms of EMG magnitude, a high activation of m. iliopsoas was observed during the backswing phase, which is indicative of the important role of the hip flexor muscles for kick performance.

Using surface EMG electrodes, De Proft et al., 1988 found maximal activity of hip and knee muscles during the terminal stage of the backswing phase which increased again prior to ball impact. Compared with needle measurements, surface EMG activity levels were higher for all muscles throughout the kick whereas a proximal-to-distal sequence of muscle activation was not evident.

From the above descriptions, it becomes clear that the rapid knee flexion and extension is an important aspect of soccer kick performance. This movement is accompanied by a stretch of the knee extensor musculature during backswing followed by immediate shortening during forward shank movement. It has been shown that kicking speed is significantly higher when the knee extensor musculature is stretched and then shorten compared with kicks involving only concentric actions (Bober et al., 1987). For this reason, the role of stretch-shortening cycle of the knee extensors for a successful kick has been particularly emphasized (Lees and Nolan, 1998).

Muscle activity of both agonist and antagonist musculature is high at ball impact, mainly around the knee (De Proft et al., 1988; Dorge et al., 1999; Sorensen et al., 1996) (Table 2). If it is assumed that the main aim of the kicking action is to produce the highest ball speed possible, then one would suggest that antagonist (knee flexor) activity at the final stages of the kick is a limiting factor for performance. This seems to be supported by the observation that skilled players showed higher agonist and less antagonist muscle activity in the swinging phase than less skilled players (Bollens et al., 1987; De Proft et al., 1988). For example, research has shown that temporal EMG characteristics do not differ between expert and novice subjects' kicking patterns (Smith et al., 2002), thus suggesting that it is the magnitude rather than the sequence of muscle activity that characterizes soccer kicks by more skillful players.

From the above literature, it appears that only a few studies examined activation patterns during the kick with conflicting findings. This leaves many questions regarding the role of various muscles unanswered. For example, what is the activity of other muscles of the hip and the ankle? What is the role of stretch-shortening cycle of knee extensor musculature for the shank acceleration? What is the link between muscle activation patterns and sequential joint moment development until impact?

Another important issue is that the above observations mostly apply to maximal instep kicks. However, one should consider that a powerful kick is not necessarily a successful (accurate) kick. In the latter case, muscle activation patterns around different joints may be more complex in order to achieve a fine control of lower limb movement. For example, what are the differences in muscle activity when a player has to kick the ball against a high or a low target? What are the necessary adjustments in muscle activity when the player uses an almost diagonal approach relative to the target?