Three-dimensional kinematic analysis of elite javelin throwers at the 1999 IAAF World Championships in Athletics

By José Campos, Gabriel Brizuela, Victor Ramón

Join the Newsletter

Three-dimensional kinematic analysis of elite javelin throwers at the 1999 IAAF World Championships in Athletics
By José Campos, Gabriel Brizuela, Victor Ramón


    Dr. José Campos is the Director of the Sport Performance Unit in the Department of Physical Education and Sport at the University of Valencia (Spain). He was formerly responsible for the Javelin Throw in the Spanish Athletic Federation and was Coordinator for Biomechanical studies. 

    Dr. Gabriel Brizuela is a member of the Department of Physical Education and Sport at the University of Valencia where he lectures in Sport Biomechanics.

    Victor Ramón is a research fellow of the Department of Physical Education and Sport at the University of Valencia.



    A biomechanical analysis of the javelin throw at the 1999 IAAF World Athletics Championships in Seville was carried out by the University of Valencia (Department of Physical Education and Sport). This paper presents the results of a study of the male finalists. The methodology used is based on a 3D Video Programetry at 50Hz. The results show the characteristics of the throwers individual model at the event, which for practical purposes can be compared with the performance of the same throwers in other competition. Detailed information on the kinematic parameters is provided. The most significant differences between the patterns used by the throwers are located in the kinematic chain in the Preparatory and Final Delivery phases, in the instant of javelin release and the vertical and horizontal velocity combinations of the javelin at delivery.

Gymboss Timers

    A description of the technique used by elite throwers gives insight into individual forms of organisation used to obtain high performance. These models eventually become references that help coaches and athletes to develop their own strategies to achieve maximum mechanical efficiency.

    The pattern of motion used in the javelin throw is similar to other movements used when striking or throwing an object. These are characterised by the fact that the body segments act sequentially to attain the maximum speed in the most distal segment of the system at the instant when the object is struck or thrown (Atwater, 1979; Menzel, 1987). Javelin throwing technique has been described in many studies, including those by Hay (1993), Whiting et al (1991), Best et al (1993). Mero et al (1994) and Bartlett et al (1996).

    The present paper describes the technical models used by the finalists in the men's javelin at the 1999 IAAF World Championship in Athletics in Seville. The aim of the study is to compare the throwers' individual models in the light of the documented data available on the biomechanical analysis of javelin throw.


    3D photogrammetric analysis was used. All throws in the final were filmed and the best attempts of each athlete were subsequently analysed. The cameras were phase-locked and aligned with their optical axis at approximately 900 (side and back views):

■     Two synchronized SVHS Panasonic video cameras, operating at 50 fps.

■     Modulated reference system (Two integrated cubes of 2x2m. each) for spatial calibration.

■     Kinescan 8.3 (lBV) software for the digitising process.

    All coordinates were smoothed using quintic spline. The DLT (Direct linear transformation) algorithm was used to calculate the 3D marker coordinates (Abdel-Azir & Karara, 1971). The kinematic parameters obtained on the marker coordinates (x,y,z) were transformed as variables of the study.

Analysis procedures
    The biomechanical analysis of each athlete focused on the Preparatory and Final Delivery phases. The most important factors for javelin release occur during these decisive periods, which therefore offer the best conditions for comparing athletes' techniques.
    The main time points were the following: 

■     t1: right foot lands (support leg for right-handed thrower) on the ground (single-support) at the beginning of the Preparatory Phase.

■     t2: left foot lands (braking leg for right-handed throwers) on the ground (double-support or Power Position) at the beginning of the Final Delivery Phase.

■     t3: javelin is released (instant of release).


    All the throwers except the Cuban Emeterio Gonzalez were right-handed. Each thrower's best attempt was analysed, except for the German thrower Hecht whose second best throw was studied. None of the Norwegian athlete Fagernes' throws were analysed due to image recording problems.
    The results for 7 of the male finalists are shown in Table 1.


Phase Timing 

Duration of Preparatory Phase and Final Delivery Phase
    Based on the reference instants mentioned above (t1, t2 and t3) the throw was split into the following two sub-phases:

■     Preparatory Phase: the period between t1 and t2.

■     Final Delivery Phase: the period between t2 and t3.

    The results show that the greatest differences between the athletes occur in the Preparatory Phase. Times recorded for the duration of the Preparatory Phase (t1-t2) ranged from 140 to 260 milliseconds, and from 100 to 140 milliseconds for the Final Delivery Phase (t2 -t3).
    Two models can be distinguished for the Preparatory Phase (t1-t2). One is the model used by throwers Parvianen, Gonzalez and Henry, who base their throwing tempo on an extended Preparatory Phase (over 200 milliseconds), while throwers Gatsioudis, Zelezny and Hecht base their tempo on a shorter Preparatory Phase.
    There is less difference between the throwers in the Final Delivery Phase (t2-t3) with values ranging from 100 to 140ms.


Duration between maximum peak joint speed and the instant of release
    A factor that influences the quality of energy transfer to the javelin is the coordinated motion of the upper limb, starting from the acceleration-deceleration of the sequences in the upper kinetic chain. These sequential motions from the proximal to the distal segments are one of the fundamental keys to performance in over-arm throwing (Atwater, 1979; Whiting et aI., 1991; Mero et aI., 1994). Hip, shoulder, elbow, hand and javelin velocities are taken into account to analyse these power transmission sequences in the Final Delivery Phase. Figure 3 shows hip, shoulder, elbow, and javelin velocities in the Finnish thrower Parvianen's winning 89.52m throw and it can be seen that the general throwing model is repeated (Menzel, 1987).

    The analysis of how the maximum peak velocities for each marker are reached at the instant of release (t3) provides a more detailed description of the timing used by the throwers to structure their individual motion models for the upper limb.
    Table 2 show the data of time duration from maximum peak hip, shoulder and elbow velocities to release, with average times of 130ms for the time from maximum peak hip velocity to release, 90ms from maximum peak shoulder velocity to release and 60ms from maximum peak elbow velocity to release.


    The data confirm that throwers' velocity variability is greater than 10% in all cases. The highest level for coefficient of variation is for the period from the maximum peak shoulder velocity to release (16%), while variability for hip and elbow velocities are very similar with 11% and 10% respectively.

    The tendency observed for hip motion may be worthy of note. Taking the period between t1 and t2 as a reference, it is shown that all throwers, except Backley, reach maximum hip velocity before t2. Advance times are quite variable ranging from 10 to 80ms.
    In case of Backley, the thrower reaches maximum hip velocity 20ms after t2. These differences in starting hip motion confirm findings by Best (1993) to the effect that this parameter depends on individual technique and its effect on performance should be considered in relative terms.

Velocity Variables

Release Velocity

    Release velocity is known to be the parameter that bears most relation to distance (Ikegami (1981), Mero (1993), Menzel (1987), Morris, Barlett 8: Fowler (1997). The linear velocity of the javelin at release depends on the quality of power transmission from the body to the upper limb and then to the javelin.
    The results show release velocities that range from 28.1 m/s in Henry's 85.43m throw and 29.71 m/s in Parvianen's 89.52m throw. With regard to the relation between distance and release velocity, the correlation index was high (r: .714) but not statistically significant (p: .072).
    Horizontal (Vy) and vertical (Vz) velocity components of the Javelin at the instant of release (t3)
    The magnitudes of the two javelin velocity components at release have also been considered in order to interpret the final throwing action and its influence on javelin behaviour in the airborne phase. Figure 4 shows the values of these horizontal (Vy) and vertical (Vz) components.


    The differences between the two components in each of the throws under study ranged from Hecht's 3.4m/s to Gonzalez's 12.27m/s respectively. In absolute values the horizontal component in the men's throws ranged from Hecht's 21.54m/s to Gonzalez's 25.88m/s, and the vertical component from Gonzalez's 13.61 m/s to Hecht's 18.14m/s. Gonzalez was the athlete with the highest vertical component and the lowest horizontal component, whereas Hecht had the lowest horizontal component (18.14m/s) and the highest vertical component (18.54m/s) of all the athletes.
    It is interesting to note that in throws like those made by athletes Parvianen and Gatsioudis where both distance and release velocity are similar, different models are used to direct forces to the javelin.

Release conditions (release height; release angle and angle of attack)
    Release height is a measure of ballistic efficiency and is conditioned by the thrower's height, lateral bending of the trunk and front leg knee angle at the instant of release. Throwers should aim to throw as high as their height allows while maintaining foot contact on the ground. The results show release heights that range from 1.80m to 2.14m in throws by Zelezny and Parvianen respectively.
    The parameters relative to the position of the javelin at release should include javelin position angle, release angle and, as a consequence of these, angle of attack. Angle of attitude is the angle formed by the velocity vector and the horizontal, and the angle of attack is formed by the difference between angle of attitude and release angle. Theoretical references suggest that the release angle should be 32° - 37° and the attack angle not over + 8° for an effective throw. In a study based on a simulation, Hubbard and Alway (1987) reported that optimum conditions for throws with velocities of 23-35m/s require an angle of attack of 0-2.5°.
    Hecht used the largest release angle (40.1°) and Menendez the smallest (27.7°). Menendez had the largest resulting angle of attack (8.8°) and Parvianen (0.9°) and Hecht (1.6°) the smallest. Table 5 shows that the athlete who came closest to the reference values was the World Champion Parvianen, who was capable of generating a release velocity of over 29.5m/s with a release angle of 36.6°, resulting in a negative angle of attack of approximate -1°.


Knee Angle of the braking and support legs (Final Delivery Phase t2 - t3)
    The bracing and blocking action of the braking leg must also be taken into account in order to reach maximum release velocity, as it greatly reduces the horizontal velocity of the thrower-plus-javelin system. The knee angle of the braking leg is an indicator of the athlete's ability to transfer kinetic energy to the javelin. This blocking action favours kinetic energy transfer from the upper part of the body to the javelin (Morris, Bartlett. Navarro, 2001). It seems evident that this action is decisive, considering that in elite throwers 60% of the javelin's kinetic energy is generated in the last 50ms before release (Morris, Bartlett, 1995).
    Theoretical principles for an effective throw state the need to maintain a flexion-extension angle of 160°-180°, so that the largest degree of extension occurs at the instant of release. Figure 5 shows braking leg knee angle values at t2, t3 and maximum flexion in the Final Delivery Phase (t2-t3).


    All the finalists except Backley and Gatsioudis showed increasing extension of the braking leg knee in the Final Delivery Phase. Therefore, braking leg knee extension at release is higher than the maximum for the whole of the Final Delivery Phase. In Backley's case, however, the knee does not return to extension because maximum knee flexion (137°) is reached at release, showing a behaviour of progressive flexion that leads to a loss of support at javelin release. The same behaviour is seen in Gatsioudis' performance, but in this case extension is higher (152° - 153°). In short, Parvianen, Zelezny and Henry were the most orthodox throwers in this action and Gatsioudis, Hecht and Backley were less effective in relation to support.
    Support leg knee behaviour is not a frequently used parameter in reported studies, but support leg knee flexion-extension is decisive to drive the action and the thrower-plus-javelin system forward and direct it "against" the braking leg. The results are shown in Figure 6.


    The results show knee extension in all the throwers in the t1-t2 phase, i.e. between support leg foot contact and braking leg foot contact respectively. However, differentiated patterns were found between t2 and release (t3). One group (Parvianen, Zelezny, Gonzalez and Backley) tends to extend the support leg knee and the other group (Gatsioudis, Hecht and Henry) tends to do the opposite.
    The authors consider that this kinematic measure should not be studied separately but together with the bracing and blocking action performed by the braking leg and hip rotation on the horizontal plane.

Hip and Shoulder axis rotation on the horizontal plane
    Rotation of the hip axis and shoulder axis on the horizontal plane are two important measures that show the thrower's ability to make a wide and continuous movement in the Final Delivery Phase and help throw the javelin further. Table 6 shows the measurements recorded for each athlete at t1 and t2. The 90° position is the athlete's anatomical position facing the throw area and 180° is the position where the axes would be parallel with the Y axis, i.e. the position at t1 where the hip and shoulder axes are in maximum rotation and aligned with the release axis.


    Gatsioudis and Zelezny were the athletes with the highest hip axis rotation values at t1 and Gonzalez and Backley those with the most advanced hip axis. Gatsioudis and Zelezny were also the throwers with most hip movement from t1 to t2. Conversely, Hecht, Henry and Gonzalez showed the most passive hip action.
    With regard to shoulder motion, most of the athletes kept the shoulder axis at an angle of about 140° in double-support (t2), which is in line with a study by Morris and Bartlett (1996) on elite throwers. In addition, there was greater variability in the difference between shoulder and hip axes angles at t1 than at t2. There were differences ranging from 18° to 32° between the two axes at instant of double-support (t3), except for Henry who had a difference of only 1° that shows early advance of the shoulder axis.

Throwing arm elbow angle
    Elbow angle is another kinematic measure frequently reported in the literature. From the technical viewpoint, the throwing arm should be extended as much as possible until double-support (t2) in order to attain maximum javelin acceleration in the approach run.
    All the throwers except Zelezny held the elbow quite extended at t1. The greatest differences between throwers were found at t2, although they all bent the elbow in relation to the position held at t1, with differences that ranged from 19° to 44°. Gatsioudis was the thrower with the highest elbow angle and Gonzalez and Backley had the lowest angle change between t1 and t2 (19° and 20° respectively). Lastly, all the throwers had similar elbow angles at t3, ranging from 151° to 160°.


Acceleration Path
    The javelin's acceleration path has been used as a performance measure in the javelin throw. A long approach run facilitates optimum application of the forces and enables better use of the stretch-shortening muscle cycle. The importance of a longer approach is stressed in the literature, including two proposals for approach run assessment under the heading "Acceleration Path". This is defined by Bartlett et al (1996) as the horizontal distance from the right hip to the centre of mass of the javelin at the start of the delivery stride. Mero et al (1994) define the acceleration path as the horizontal distance from the grip of the javelin to the right hip at final foot strike.
    As both proposals refer to two of the instants used in this kinetic study, it was decided to analyse both versions of the acceleration path, i.e. the horizontal distance from the javelin's centre of mass to the hip at instants t1 and t2.

    All the athletes have a longer acceleration path at t1, which is understandable because at that instant the hip is advanced as a consequence of the cross-over step. The range was 0.73-0.96m. Parvianen had the longest acceleration path and Gatsioudis the shortest. 

    T2 values ranged from 0.65 to 0.91 m and Parvianen again had the longest acceleration path. As an overall criterion it was observed that the acceleration path decreased just before release, which means a loss of power in the most decisive phase. Zelezny (14cm), Henry (15cm), and Gonzalez (19cm) showed the largest losses. In any case, it must be noted that the acceleration path is conditioned by other factors like elbow angle. Thus, the acceleration path value is the result of several actions which should all be aimed at achieving the maximum efficacy possible. The analysis of acceleration path loss between t1 and t2 should therefore call for a review of all of the parameters analysed.


    As noted in the conclusions of previous studies it has been observed that each thrower maintains an individual throwing pattern in relation to timing and in the values obtained in the different kinematic parameters under study.
    Nevertheless, these individual patterns are conditioned by what could be called efficiency filters. These are the minimum requirements needed to throw the javelin at a long distance which affect the position of the kinetic chain in the Final Delivery Phase as well as the coordination of the body segments for ballistic movement.
    The aspects that distinguished Parvianen from the rest of the throwers were that his throw was more rectilinear and he throws from a higher position, with a longer acceleration path and more favourable release conditions (release velocity: 29.62m/s; angle of attack: -0.9°; release height: 2.14m).
    Athletes' individual models are an example of motor complexity and numerous methodologies are required to analyse them. Descriptive studies such as the present work help to understand the dimensions involved in achieving performance in sport but their repeatability is relative.
    However, we hope the information presented herein will be useful for javelin throw coaches and throwers and that it will contribute to the understanding of this sport.


Recommended Resources

. . . 

Recommended Resources

. . . 

Join Altis 360