99 years). They were all right-handed and able to perform first serves. None of the participants played tennis outside the timetable for data collection during the research. All the participants provided informed consent according to the Declaration of Helsinki. The Extremadura University Ethical Committee ref 1 approved the procedure. Measures Product variables analyzed were stroke accuracy, measured by radial error (Robins et al., 2006), variable error, which represents serve errors made in respect of deviation from the serve target area, and the ball speed. Process variables (Table 1) were measured over the trajectory of the hand holding the racket along the antero-posterior (X), the transverse (Y), and the longitudinal (Z) axes.
With respect to non-linear variables, these give information about the structure and characteristics of the variability present in the time series. These time series were derived from the position of the hand holding the racket during its trajectory, from the beginning of the movement until the moment the racket hit the ball. Table 1 Dependent variables analyzed in the research. In each instant kinematic variable the standard deviation (SD) and the variation coefficient (CV) was analyzed Tasks, material and measurements Each tennis player performed 20 first serves. They were instructed to hit the ball with as much power and accuracy as they could, and to avoid sending the balls into the area known in tennis slang as the ��T�� (the line intersection which divides both service boxes from their respective service lines).
The ball bounce on the tennis court surface was video recorded in every serve (Sony HDR- HC3E). The video camera was set at a height of 3 meters and was positioned at the back of the court. In order to measure accuracy, a Visual Basic 5.0 application was developed (Menayo, 2010). This facilitated the calculation of real-space Cartesian coordinates for the ball bounces through a digitization process from the video recording of the serves. Non-linear kinematic variables were analyzed by using a software application created with Visual Basic 5.0, from an algorithm for calculating Approximate Entropy (Pincus, 1991). To measure ball speed, a radar gun (Sports Radar SR3600) was used. This radar device, which records the speed of moving objects with an accuracy of +/? 1 km/h, was positioned behind the tennis player, facing the direction of the stroke (Figure 1).
An electromagnetic motion tracking system Polhemus Fastrak? was used to record and analyze kinematic variables and this was connected to a computer (Toshiba Satellite 1900). This tracking system has 6 Degree-of-Freedom motion tracking sensors, with an accuracy of 0.08 cm for position (X, Y and Z Cartesian space coordinates) and 0.15 degrees for angular orientation (azimuth, elevation, and roll), and records at a frequency Brefeldin_A of 120 Hz. Figure 1 Automated measurement system.