INTRODUCTION

The modern game of tennis has evolved from a technical-based sport to a more dynamic and explosive sport as a result of higher stroke and service velocities, bolstered by the improvement of rackets and strings (Ulbricht et al., 2016). Above all, service is the most powerful and important shot in tennis. After the toss-up, while moving the power from the lower half to the upper half of the body using the whole-body movement chain, the racket speed is increased toward impact (Kibler, 2009). This racket motion can create a mix of ball speed and spin. A clear trade-off between the ball spin and horizontal ball speed has been reported in tennis (Sakurai et al., 2013). Theoretically, the speed decreases when spin is applied and vice versa. Elite tennis players generally stroke a flat or power service first, reducing the spin rate and focusing on speed, then select a slice or spin as the second service (Douglas, 1992; Bahamonde, 2000; Whiteside et al., 2013). When the first service fails, it is common for players to slow down and increase the spin rate on the second service, which increases the likelihood of the ball successfully landing in the service box. This technique increases accuracy without losing points (Chow et al., 2003). To achieve higher speeds during service, the player must account for the net height, position of the service line, and height of the hitting point. Brody (1987) reported that a minimum height of 2.74 m on the baseline is required to eliminate gravity and ball aerodynamics concerns. For athletes who are not tall enough, spinning is essential; however, at the same time, speed is also required to some extent. As of July 2020, the average height of the top 50 players in the International Tennis Federation ranking was 188.73 cm (ITF, 2021). However, the average height of the four Japanese players in the top 100 was 176.50 cm (ITF, 2021). Japanese players, who are generally inferior in height, must overcome the technical hurdle associated with service performance to reach global elite status.

A significantly high correlation between service speed and the percentage of points won has been shown. Service reportedly influences the game outcomes for both male and female players; in par-ticular, service speed is highly correlated with an athlete’s competition level (Ulbricht et al., 2016). Fett et al. (2017) reported significant differences between junior Davis cup and regional level players in service speed and upper body strength. According to Fett (2018) and Kramer (2017), maximal service speed is the most appropriate on-court predictor of player performance. Thus, the importance of service speed in an athlete’s competition level is widely known in the tennis world. Moreover, various researchers have reported the relationship between service speed and physical factors. In a study on German National junior tennis players, Ulbricht et al. (2016) reported that the upper limb muscular strength, power, and service velocity were interrelated in both male and female players, and there was a strong relationship between service speed and level of competition. In a study emphasizing the importance of transmitting power in the kinetic chain from the lower body to the upper body, Girard et al. (2005) examined the earlier activities in the lower leg muscles in high level players and suggested that for hitting, the power of not only the upper body but also of the whole-body kinetic chain is vital. Hence, it is important to acquire a service form that links the whole-body as well as muscle activity. Many studies of these services have been historically centered on adult, male players and, consequently, the mechanics characterizing the female and junior serves is underpinned by emulation, rather than the objective data that have guided instruction of the adult male serve (Whiteside et al., 2013). Even when we consult tennis-coaching books from each country, no gender-specific coaching methods on the technical instruction of service are found. Although many researchers have reported a relationship between service velocity and physical strength factors (Ferrauti and Baestiaens, 2007; Knudson et al., 2004; Cardoso, 2005), no studies have so far explored the correlation between service spin rate and physical strength. There are some reports on the spin rate in service. However, these did not elucidate its relationship with physical strength (Murakami et al., 2016; Muramatsu et al., 2015; Muramatsu et al., 2010; Sakurai et al., 2007). As mentioned above, there is a trade-off between the speed and the spin rate of service. However, it has been reported that the world’s top-level players could hit the speed service while maintaining a high spin rate (Muramatsu et al., 2015). It is important for Japanese athletes, who are inferior in height, to learn to improve both spin rate and speed.

Although muscle strength or strength power is thought to play a role in service speed, only a few studies have clarified the effects of physical strength on service spin. This study aimed to examine the relationship between performance in tennis service, based on speed and spin rates, and physical factors among national-level male and female junior tennis players.

MATERIAL AND METHODS

Participants

Forty-one elite tennis players (30 men; mean ± standard deviation [SD] age, 15.03 ± 2.58 years; height, 170.64 ± 7.06 cm; mass, 60.39 ± 7.72 kg and 11 women; age, 16.73 ± 2.65 years; height, 163.56 ± 5.35 cm; mass, 55.56 ± 6.45 kg) participated in this study. These were either top-level junior players of each generation in Japan or candidates who participated in the Uni-versiade tournament. These players were selected by the National Federation’s coaching staff based on competitive performance. All players had at least 10 years of tennis training. Under the Institutional Review Board’s policies for the use of human participants in research in accordance with the Declaration of Helsinki, the investigator informed all participants about the benefits and possible risks associated with participation in the study. All participants signed a written informed consent document indicating voluntary participation. Self-reported medical histo-ries were received from all participants. We confirmed their medical history of injuries and determined that there was no effect on this study.

Design & Procedures

An attempt was made to examine whether there was a relationship between service performance (speed, spin, impact height when hitting a ball, impact depth), anthropometry (player’s height, mass, skeletal muscle mass), and physical condition (grip strength, isokinetic peak torque (PT), jump, sprint, medicine-ball throw). Variables were categorized as service speed, service spin, impact height, impact depth, physique, and physical strength. We set 3 independent variables of physique and 15 independent variables of physical strength. Independent physique and physical strength variables were converted into the principal component analysis (PCA) score, and the correlation with ser-vice performance (speed, spin, impact height, impact depth) was analyzed. Furthermore, we examined whether there was a difference by sex in the above correlation

Service performance test

The service tests were conducted using the TrackMan tennis radar device (TrackMan Inc., Vedæk, Denmark). This device is an advanced radar that utilizes the Doppler effect to capture the behavior of an object; in this case, the full three-dimensional flight of a tennis ball (speed, spin, spin axis). The Doppler effect is a phenomenon in which the relative velocity between a generation source and an observer result in the shifting of sound or radio wave frequencies (Murata and Takahashi, 2020). If a ball in flight reflects a radio wave, the frequency of the reflected wave shifts, depending on the velocity and spin of the ball, and is calculated by a dedicated software (Martin, 2012). Doppler effect-based measurement devices are superior to other devices in terms of immediacy and user-friendliness and are used in golf coaching and baseball (Murata and Takahashi, 2020). The accuracy of the TrackMan system is equivalent to that of conventional high-speed cameras or speed radar devices, as reported in recent studies (Martin, 2012; Murakami et al., 2016; Murata and Takahashi, 2020; Sato et al., 2017). That allows for real-time measurements and an analysis of parameters such as spin rate, ball speed, ball direction, impact location (height, depth), net clearance, and landing position. In this study, the TrackMan was set at approximately 4 m behind the center mark, which allowed the radar to visualize the service box (Fig.1).

Figure 1. Experimental set up for service performance test and service impact items.

Service impact was defined as the moment when the ball hit the racket, and the height and depth of the hitting point at that moment were measured. The impact depth was analyzed with the baseline set to zero, with the forward direction closer to the net considered as positive (+) and backward direction away from the net as negative (–) (Fig.1). The tests were conducted on an indoor tennis court. After a brief warm-up consisting of serves with increasing velocities, players were instructed to perform the first service as fast as possible and the second service with as much spin as possible, with both directed at the T of the service box. No returner was placed. They performed three sets of service tests, hitting the 1st and 2nd serve from the deuce side and then hitting the same from the advantage side. Details of a total of 12 services per player were collected. For the analysis, the average of the six services was used for each variable. Services such as fault and net were excluded, and only successful services were used.

Physical strength tests

Based on the published literature, 15 measure-ment variables related to service performance were selected (Delgado et al., 2019; Kramer et al., 2017; Signorile et al., 2005; Koya et al.,2018). In addition to height, weight, and grip, the following physical condition variables were measured using the Japan Tennis Association protocols (Koya et al., 2014; Koya et al., 2015):

1. Skeletal muscle mass: Data were obtained via impedance methods using a body composition analyzer (In Body 730, BioSpace Inc., Los Angeles, CA, USA).

2. Isokinetic peak torque (knee and hip): Isokinetic muscle strengths of the legs and trunk were measured using the Biodex (Biodex Medical Systems, Inc., Shirley, NY, USA). The knee joint was tested at 1.05 rad • s-1 and 3.14 rad • s-1 and the hip joint was tested 1.05 rad • s1 and 2.10 rad • s-1. The peak torque of extension and flexion was recorded, and both legs were measured. Total value of flexion and extension was used.

3. Squat jump: This test was performed using a mat-switch (Multi Jump Tester, DKH Inc., Tokyo, Japan). The players completed a jump test from a squat position with the hands on the waist. The best value was used.

4. Vertical jump: This test was also measured using the mat-switch. Players were required to jump using an arm action. The best value was used.

5. Rebound Jump-Index (RJ-Index): The players performed six consecutive rebound jumps with their hands on their waists. This test serves to measure ankle joint stiffness. The RJ-Index was calculated by dividing the jumping height by the landing time and was used as a criterion for the stretch-shortening-cycle (SSC) function. The best value was used.

6. 5-m sprint and 20-m sprint: The participants were instructed to sprint in a straight-line for 20 m; the times from the starting line to the 5-m and 20-m marks were measured using a photocell (Timing Systems by Brower Timing Systems, Draper, UT, USA). In the analysis, the speed was calculated (m/s) based on the time. The best value was used.

7. Medicine ball throw (MBT): Players performed three types of forward throws (overhead, right side, and left side) and backward overhead throws in an open stance, with both legs fixed whilst throwing a 2 kg medicine ball. The best value was used.

All those variables were highly related to the ability to hit the service (Delgado et al., 2019; Kramer et al., 2017; Signorile et al., 2005; Koya et al.,2018). Test reliability was confirmed by performing a retest.

Analyses

Partial correlations (controlling for age) were conducted to assess the association between the service performance variables (speed, spin rate, im-pact height and impact depth) and anthropometrics variables (physique and physical strength) for each gender (Table 1). Age was used as a control variable to exclude the influence of growth and maturity (Fett et al., 2018). We performed principal component analysis (PCA) using the physical strength test variables and then extracted the first principal component score as a comprehensive index of physical general strength (strength PCA) to comprehensively evaluate each player’s physical strength. Correlation analyses were conducted to assess the relationship between strength PCA and service performance within sex. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 27 (IBM Corp., Armornk, N.Y., USA). Statistical significance was set at P < 0.05.

 

RESULTS

Table 1 shows physique and strength test results of players in each group. The partial correlation coefficients for males and females are shown in Tables 2-1 and 2-2, respectively. Among male players, correlation analyses revealed some significant relationships of 1st and 2nd service with physique and physical strength; however, service speeds were not correlated with isokinetic PT (knee), squat jump, vertical jump and the 5-m sprints. A significant correlation was observed between the service speed and grip strength, isokinetic PT (hip), broad jump, 20 m sprint, MB throw, and physique. As for service spin, in the 2nd service, a significant correlation was observed between grip strength, isokinetic PT (hip), and broad jump. In both 1st and 2nd services, speed was more correlated than spin with physique and physical strength. As for impact, there was a significant correlation between impact height and isokinetic PT (hip) and RJ-index, but not between impact depth and the other items (Table 2-1). The results among the female players were entirely different compared to the male players. We found a significant correlation between the 1st service speed and forward overhead MB throw; however, no correlation was observed between both service speeds and the other tests. As for service spin, in both services, there were significant correlations with isokinetic PT. Regarding impact (height, depth), neither physique nor physical strength items showed any correlation (Table 2-2).

Table 3 shows the factor loading of physique PCA and strength PCA (Table 3).

Table 1. Physique and strength test results of players in each group.

		Male players (30)	Female players (11)
Physique	Age (years)	15.03 ± 2.58	16.73 ± 2.65
	Height (cm)	170.64 ± 7.06	163.56 ± 5.35
	Mass (kg)	60.39 ± 7.72	55.56 ± 6.45
	Skeletal muscle (kg)	30.96 ± 4.36	24.46 ± 3.17
Strength	Grip strength (kg)	43.23 ± 7.70	33.85 ± 4.27
	Isokinetic PT / Knee at 1.05 rad•s-1 (Nm)*	534.59 ± 113.14	389.82 ± 201.81
	Isokinetic PT / Knee at 3.14 rad•s-1 (Nm)*	406.09 ± 86.71	260.82 ± 136.37
	Isokinetic PT / hip at 1.05 rad•s-1 (Nm)**	405.47 ± 135.94	309.55 ± 162.92
	Isokinetic PT / hip at 2.10 rad•s-1 (Nm)**	369.97 ± 132.81	252.00 ± 147.75
	Squat Jump (cm)	33.57 ± 6.82	30.72 ± 3.73
	Vertical Jump (cm)	41.56 ± 8.33	37.52 ± 4.01
	RJ-index (m/sec)	1.94 ± 0.41	2.20 ± 0.32
	Broad Jump (m)	2.24 ± 0.24	2.03 ± 0.17
	5m sprint (m/sec)	4.50 ± 0.28	4.35 ± 0.23
	20m sprint (m/sec)	6.12 ± 0.31	5.81 ± 0.26
	MBT overhead backward (m)	11.71 ± 2.32	8.78 ± 1.11
	MBT overhead forward (m)	8.25 ± 1.66	6.36 ± 1.21
	MBT forehand (m)	10.17 ± 1.70	7.80 ± 0.85
	MBT back hand (m)	9.47 ± 1.73	7.50 ± 0.85
Service	1st service speed (km/h)	171.51 ± 15.10	151.68 ± 8.79
	1st service spin (rpm)	1409.58 ± 380.06	1415.02 ± 240.74
	1st service impact H (m)	2.61 ± 0.12	2.53 ± 0.09
	1st service impact D (m)	0.03 ± 0.19	0.05 ± 0.19
	2nd service speed (km/h)	131.75 ± 15.63	129.68 ± 4.32
	2nd service spin (rpm)	3514.24 ± 501.49	2200.37 ± 452.99
	2nd service impact H (m)	2.56 ± 0.15	2.54 ± 0.08
	2nd service impact D (m)	-0.18 ± 0.25	-0.11 ± 0.13

PT = peak torque; RJ = rebound jump; MBT = medicine ball throw; H = height; D = depth.

* : Total value of flexion and extention of both knee joints.

** : Total value of flexion and extention of hip joint.

Table 2-1. The partial correlation coefficient of physique and physical strength versus service performance in male players.

			1st service				2nd service
	Measured item		Speed (km/h)	Spin (rpm)	Impact H	Impact D	Speed (km/h)	Spin (rpm)	Impact H	Impact D
Physique	Height (cm)	r	0.27	-0.22	0.77	0.02	0.49	-0.19	0.76	-0.10
		p	0.16	0.26	0.00	0.90	0.01	0.32	0.00	0.61
	Mass (kg)	r	0.61	-0.26	0.33	0.23	0.42	0.40	0.30	0.14
		p	0.00	0.18	0.08	0.24	0.02	0.03	0.12	0.48
	Skeletal muscle (kg)	r	0.65	-0.26	0.52	0.21	0.51	0.29	0.50	0.14
		p	0.00	0.18	0.00	0.26	0.00	0.13	0.01	0.48

Table 2-1. The partial correlation coefficient of physique and physical strength versus service performance in male players (Continuation).

Strength	Grip strength (kg)	r	0.71	-0.41	0.24	0.14	0.43	0.39	0.22	-0.05
		p	0.00	0.03	0.22	0.46	0.02	0.04	0.25	0.78
	Isokinetic PT/Knee at 1.05 rad s-1 (Nm)	r	0.34	-0.17	0.41	0.16	0.19	0.19	0.35	0.16
		p	0.07	0.37	0.03	0.40	0.31	0.32	0.06	0.40
	Isokinetic PT/Knee at 3.14 rad s-1 (Nm)	r	0.36	-0.28	0.38	0.15	0.14	0.25	0.30	0.08
		p	0.06	0.14	0.04	0.43	0.48	0.19	0.11	0.68
	Isokinetic PT/hip at 1.05 rad s-1 (Nm)	r	0.39	-0.12	0.42	0.06	0.21	0.41	0.40	0.04
		p	0.03	0.52	0.02	0.76	0.26	0.03	0.03	0.85
	Isokinetic PT/hip at 2.10 rad s-1 (Nm)	r	0.30	-0.02	0.25	0.18	0.26	0.29	0.26	0.15
		p	0.11	0.92	0.20	0.35	0.18	0.13	0.17	0.44
	Squat Jump (cm)	r	0.29	-0.23	0.16	0.26	0.29	-0.01	0.16	0.17
		p	0.13	0.22	0.42	0.18	0.13	0.96	0.42	0.38
	Vertical Jump (cm)	r	0.32	-0.20	0.03	0.32	0.23	-0.03	0.06	0.21
		p	0.09	0.30	0.89	0.09	0.23	0.89	0.76	0.27
	RJ - index (m/s)	r	0.21	-0.37	0.50	-0.22	0.08	0.06	0.44	-0.37
		p	0.26	0.05	0.01	0.24	0.66	0.76	0.02	0.05
	Broad Jump (cm)	r	0.62	-0.34	0.38	0.11	0.37	0.38	0.31	0.00
		p	0.00	0.07	0.04	0.59	0.05	0.04	0.10	0.98
	5m sprint (m/s)	r	0.24	-0.19	0.11	-0.09	0.10	0.01	0.00	-0.08
		p	0.21	0.33	0.57	0.65	0.59	0.95	0.99	0.69
	20m sprint (m/s)	r	0.41	-0.23	0.23	0.00	0.30	0.08	0.14	0.01
		p	0.03	0.23	0.23	0.99	0.11	0.68	0.48	0.96
	MBT overhead backward (m)	r	0.65	-0.09	0.37	0.29	0.67	0.14	0.42	0.28
		p	0.00	0.65	0.05	0.13	0.00	0.48	0.02	0.14
	MBT overhead forward (m)	r	0.60	-0.06	0.25	0.14	0.52	0.17	0.26	0.16
		p	0.00	0.75	0.20	0.47	0.00	0.38	0.18	0.40
	MBT forehand (m)	r	0.58	-0.23	0.34	0.35	0.56	0.16	0.34	0.23
		p	0.00	0.23	0.07	0.07	0.00	0.41	0.08	0.23
	MBT backhand (m)	r	0.65	-0.30	0.51	0.28	0.65	0.04	0.54	0.23
		p	0.00	0.12	0.01	0.14	0.00	0.84	0.00	0.22

r : partial corelation coefficient; p<0.05

PT = peak torque; RJ = rebound jump; MBT = medicine ball throw; H = height; D = depth.

Table 2-2. The partial correlation coefficient of physique and physical strength versus service performance in female players.

			1st service				2nd service
	Measured item		Speed (km/h)	Spin (rpm)	Impact H	Impact D	Speed (km/h)	Spin (rpm)	Impact H	Impact D
Physique	Height (cm)	r	0.69	-0.21	0.76	0.74	0.47	0.44	0.75	-0.18
		p	0.08	0.65	0.05	0.06	0.29	0.32	0.05	0.70
	Mass (kg)	r	0.79	-0.02	0.06	0.35	0.19	0.70	-0.12	-0.41
		p	0.03	0.97	0.91	0.44	0.68	0.08	0.80	0.37
	Skeletal muscle (kg)	r	0.88	-0.05	0.31	0.74	0.58	0.50	0.24	-0.39
		p	0.01	0.92	0.50	0.05	0.17	0.25	0.60	0.38
Strength	Grip strength (kg)	r	0.41	0.29	0.18	0.26	0.12	0.54	0.10	-0.32
		p	0.36	0.53	0.71	0.57	0.80	0.21	0.83	0.48
	Isokinetic PT/Knee at 1.05 rad s-1 (Nm)	r	0.25	0.80	0.48	-0.08	0.19	0.50	0.36	0.44
		p	0.59	0.03	0.27	0.87	0.69	0.25	0.42	0.33
	Isokinetic PT/Knee at 3.14 rad s-1 (Nm)	r	0.53	0.46	0.43	0.46	0.55	0.34	0.41	0.06
		p	0.22	0.30	0.33	0.30	0.20	0.45	0.36	0.90
	Isokinetic PT/hip at 1.05 rad s-1 (Nm)	r	0.03	0.65	-0.06	-0.48	-0.31	0.56	-0.27	0.15
		p	0.96	0.12	0.91	0.28	0.50	0.19	0.56	0.75
	Isokinetic PT/hip at 2.10 rad s-1 (Nm)	r	0.04	0.54	0.15	-0.41	-0.46	0.77	-0.08	-0.04
		p	0.94	0.21	0.75	0.37	0.30	0.04	0.87	0.93
	Squat Jump (cm)	r	0.00	0.61	0.25	0.16	0.19	0.29	0.16	-0.08
		p	1.00	0.15	0.59	0.73	0.68	0.53	0.74	0.86
	Vertical Jump (cm)	r	0.25	0.64	0.76	0.27	0.46	0.28	0.76	0.44
		p	0.58	0.12	0.05	0.56	0.30	0.55	0.05	0.33
	RJ - index (m/s)	r	-0.24	0.23	-0.57	0.00	0.32	-0.63	-0.45	0.07
		p	0.60	0.62	0.18	1.00	0.49	0.13	0.31	0.88
	Broad Jump (cm)	r	0.36	0.28	0.69	0.58	0.45	0.32	0.72	-0.08
		p	0.42	0.55	0.08	0.18	0.31	0.48	0.07	0.87
	5m sprint (m/s)	r	-0.38	0.39	-0.04	0.11	0.39	-0.59	0.11	0.27
		p	0.40	0.39	0.94	0.82	0.39	0.17	0.82	0.56
	20m sprint (m/s)	r	-0.28	0.51	0.22	0.16	0.29	-0.23	0.31	0.14
		p	0.55	0.24	0.63	0.17	0.53	0.62	0.49	0.76
	MBT overhead backward (m)	r	0.63	-0.28	-0.08	0.22	0.33	-0.01	0.01	0.13
		p	0.13	0.54	0.87	0.64	0.47	0.99	0.99	0.78
	MBT overhead forward (m)	r	0.92	-0.28	0.01	0.50	0.41	0.39	-0.03	-0.28
		p	0.00	0.55	0.98	0.26	0.36	0.39	0.95	0.55
	MBT forehand (m)	r	0.61	-0.03	0.58	0.44	0.08	0.75	0.50	0.36
		p	0.14	0.94	0.17	0.33	0.87	0.05	0.25	0.43
	MBT backhand (m)	r	0.38	0.02	0.63	0.00	-0.20	0.69	0.54	0.08
		p	0.39	0.96	0.13	1.00	0.67	0.08	0.21	0.86

r : partial corelation coefficient; p<0.05

PT = peak torque; RJ = rebound jump; MBT = medicine ball throw; H = height; D = depth.

Table 3. Factor loading in physique PCA and strength PCA

		Factor loading in PCA
		Male players (30)	Female players (11)
Physique PCA	Height (cm)	.789	.981
	Mass (kg)	.931	.851
	Skeletal muscle (kg)	.960	.858
Strength PCA	Grip strength (kg)	.770	.551
	Isokinetic PT/Knee at 1.05 rad s-1 (Nm)	.844	.773
	Isokinetic PT/Knee at 3.14 rad s-1 (Nm)	.839	.807
	Isokinetic PT/hip at 1.05 rad s-1 (Nm)	.861	.672
	Isokinetic PT/hip at 2.10 rad s-1 (Nm)	.829	.776
	Squat Jump (cm)	.757	.084
	Vertical Jump (cm)	.776	.577
	RJ-index (m/s)	.697	-.485
	Broad Jump (m)	.907	.431
	5m sprint (m/s)	.529	-.498
	20m sprint (m/s)	.787	-.055
	MBT overhead backward (m)	.900	.617
	MBT overhead forward (m)	.775	.757
	MBT forehand (m)	.897	.910
	MBT backhand (m)	.899	.838

PCA: the first principal component analysis score.

We analyzed the correlation of service (speed, spin) with strength PCA and physique PCA as well as service speed with impact height and impact depth. Male players showed a significantly high correlation of service speed with physique PCA and strength PCA (Figure 2), but there was no significant correlation between service spin and either of the PCA scores (Figure 3). A significant correlation was found between service speed and both impact height and depth (Figure 4).

In contrast, female players showed a significant correlation between the 1st service speed and physi-que PCA, and between the 2nd service spin and physique and strength PCA. As for impact, there was no correlation between service speed and impact height. However, there was a significant correlation between the service speed and impact depth in the 1st service alone.

PCA: the first principal component analysis score.

Figure 2. Correlation between service speed and physique PCA (upper row), and between service speed and physical strength PCA (bottom row) in male and female players.

PCA: the first principal component analysis score.

Figure 3. Correlation between service spin and physique PCA (upper row), and between service spin and physical strength PCA (bottom row) in male and female players.

Figure 4. Correlation between service speed and impact height (upper row), and between service speed and impact depth (bottom row) in male and female players.