The stability and consistency of the tennis serve motion is considered to be determinant in the player's performance (Forti, 1995; Bollettieri, 2001; RFET, 2003). According to Bahamonde (2000) and Girard et al. (2005) the tennis serve is the most important stroke, and also the most complex, when compared with other movements in tennis. Within the scope of motor variability, the tennis serve has recently been attracting some investigative interest (Menayo et al. 2012; Mendes et al., 2012; Reid et al., 2010). In this perspective, Elliott et al. (2009) extol some constant features in the serve such as the timing, the magnitude peak of the knee flexion or the stability of the ratio of the player height: impact height (1:1.5).
The stabilization of the z axis (vertical) in the ball toss is another relevant invariable characteristic of the movement. Davids et al. (1999), Bennett et al. (2001) and Handford (2006) found that in volleyball serve, this stabilization resulted from a process of compensatory variability of the x (side to side) and y (back to front) axes. These results contradict the idea that the consistency of the motor pattern should be assumed as a primary goal in organizing and practicing schedule (e.g., Forti, 1995).
Given the similarity between volleyball and tennis serves, the analysis of the ball toss in the tennis serve is justified by the influence it has on the performance level of this movement (Elliott et al., 2009; Fuentes and Menayo, 2009).
Although this is the movement that most depends on the player-when compared with other tennis movements-it is also influenced by other types of constraints throughout a tennis match, from which emerge environmental factors such as the wind or rain.
In this perspective, bearing in mind Karl Newell's (1986) theoretical environmental constraints model, according to the characteristics of the practitioner and the task, they take a leading role in the serve performance.
The relevance of the wind in the tennis player's performance is enhanced by several authors (Elliott et al., 2009; Faulkner, 1997; Flanagan, 1983; Hoskins, 2003; Loehr, 1996) with no scientifically based opinion. In this context, Brody (1987), Scott and Randy (2000) and the American Sport Education Program (2009) analyzed the importance of the three wind directions in tennis serve: favorably (behind the server), against and lateral or cross. In this regard, the previous study carried out by Mendes et al (2011a) with experienced tennis coaches concluded that the wind is the most important environmental constraint in the tennis serve performance. On the other hand, there are few publications that analyze the variability of the ball toss in tennis serve (Reid et al., 2010). Therefore, we considered relevant to conduct this type of analysis with experienced tennis players under the effect of crosswind.
The present work aimed to verify if the invariant characteristic, vertical dimension of the ball throw in volleyball serve, evidenced by Davids et al. (1999), Bennett et al. (2001) and Handford (2006), was also confirmed in the tennis serve. Moreover, we intended to analyze if this invariant was observed under the influence of a relevant extrinsic constraint in tennis serve, the crosswind. For that purpose, we simulated this environmental factor with an industrial ventilator in the tennis court. Another objective of this study was to analyze the ratio between the player's height and the height of the player's impact point, with and without the environmental constraints (i.e., Induced Aerodynamic Flow or artificial crosswind).
Twelve male, right-handed players participated in this study with an average of 25.2 [+ or -] 3.9 years old. The anthropometric characteristics of this group of players were as follows: height 1.77 [+ or -] 0.06 m, wingspan of 1.81 [+ or -] 0.05 m and body mass of 72.3 [+ or -] 4.2 kg.
All players have been practicing tennis for 16.3 [+ or -] 5.6 years, from which 13.7 [+ or -] 4.3 years were on competitive national tennis. The study was conducted according to the Ethics code of the University of Coimbra and the recommendations of the Helsinki Declaration on Research with Human Beings.
The movement required was the flat serve from behind the base line of the tennis court, on the right-hand side and 80 cm away from the central mark. The indoor tennis court had the regulation dimensions for a singles game, 2377 cm long and 823 cm wide. All the participants were asked to serve at maximum speed and accuracy targeting the point of intersection of the centre line and service line ("T" point).
Experimental set up
All the tennis players performed 20 free serves (without instructional or wind constraints), called IAF0 (a control condition), and then performed four sets of 20 serves under different practice conditions: (1) minimum IAF speed of 2.4 m x [s.sup.-1] (called IAF1); (2) medium IAF speed of 4.3 m x [s.sup.-1] (called IAF2), 3); (3) maximum IAF speed of 5.8 m x [s.sup.-1] (called IAF3) and; (4) random IAF speed with random sequences of all three IAF speeds (called IAFr). Therefore, there was a total of 100 serves.
In this study, we analyzed the variation of the three points (Initial, peak an impact) on the ball toss, and both were measured based on the position of the ball in the 3 axes (3D analysis).
Induced Aerodynamic Flow device
The production of the IAF device was adapted from an industrial helical ventilator METEC--HCT--45-4T. The speed of the engine was set up using an electronic device (SEW Eurodrive) installed in the ventilator coupled with an 11 positions potentiometer. In order to regulate the air flow, a steel mesh of 0.45 cm and a conduct of 120 cm length and 45 cm diameter (see also Mendes et al., 2011b) were placed in the ventilator output.
The players' height varied. Therefore, they threw the ball at different heights. The ventilator had a diameter of 45 cm, so a telescopic lift GUILE ELC--506 was used to adjust the height of the ventilator up to a maximum of 520 cm. The calibration of the ventilator for each tennis player was made based on a preliminary study which consisted in analyzing the average of the highest point of the ball in 20 serves and determining its impact point (see details on "Device for analysis of the ball toss" described below).
The ventilator had a stable air flow rate of 60 cm diameter, independently of the position of the potentiometer. It was stipulated that the upper edge of the conduct would be positioned at the average level of the highest point reached by the ball during the toss.
Device to analyze the ball toss
The recording of the initial (I), peak (PP) and impact point (IP) were obtained from two cameras: (1) a camera in the sagittal plane of the tennis player: Casio Exilim Pro EX-F1, shooting at 210 Hz, positioned 700.5 cm away from the service mark and fixed on a tripod 206.5 cm high and (2) a camera in the frontal plane, positioned behind the player: Casio EX-FH25, shooting at 210 Hz, positioned 363 cm away from the tennis player and fixed on a tripod 263cm high. The timing of the beginning of the footage...