Perceptua[and Motor Skills, 2003,97,867-872.

O Perceptual and Motor Skills 2003

UPPER AND LOWER BODY STRENGTH IN RELATION

T O BALL SPEED DURING A SERVE BY MALE
COLLEGIATE TENNIS PLAYERS '

STEVEN F. PUGH, JOHN E. KOVALESKI,

ROBERT j. HEITMAN, WILLIAM F. GILLEY

University of Sotrd~Alaba~na

Srrmmay.-The relation of leg, shoulder, and grip strength to ball speed in the
tennis serve was investigated. For 15 collegiate male tennis players, leg and shoulder
strength were measured using a Lido Active isokinetic dynamometer, grip strength
with a handgrip dynamometer, and ball speed with a radar gun. Regression analysis
showed no significant ( p e . 0 5 ) relationship among the strength variables and ball
speed. Results were explained in terms of strength not being the only factor involved
in producing ball speed during the tennis serve.

A vital aspects of the tennis serve is abLty to exert high muscular force
and power. Since the body is interconnected as a kinetic chain, a successive
summation of forces must be transferred from the lower to the upper body
for effective power production. A key determinant of success in the tennis
serve is ball speed. A h i t e d number of studies have examined the relation-
ship between the ball speed of the tennis serve and upper body strength.
Cohen, Mont, Campbell, Vogelstein, and Loewy (1994) found moderate but
significant correlations ( r = .47 and r = 3 4 , p I .05) between ball speed in the
tennis serve and isokinetic extension torque measured at the elbow at 60
and 180 deg . set.-'. Ellenbecker (1991) is~kineticall~ tested shoulder inter-
nal and external rotation, wrist flexion and extension, and forearm prona-
tion at speeds of 90, 210, and 300 deg . set.-' and reported no significant
correlations between measures of strength and ball speed.
Strength training has been shown to increase functional gains and ath-
letic performance in tennis. A s ~ ~ n d ~ cincrease
ant in ball speed of the tennis
serve was reporred after four \\leeks of strength training using theraband and
dumbbell training of the shoulder using collegiate players as subjects (Trei-
ber, Lott, Duncan, Slavens, & Davis, 1998). Ellenbecker, Davies, and Rowin-
slu (1988) had collegiate tennis players train isokmeticdy for a period of six
weeks. Only the concentrically trained group showed a significant increase in
serving speed; the eccentrically trained group showed no significant increases
in ball speed with over half the group actually showing a decrease in ball
speed. Given these findings, it might be hypothesized that muscle strength
measurement could predict in part the ball speed during the tennis serve.

'Please address corres ondence to Dr. Robert Heitman, De artment of Health, Physical Educa-
tion, and Leisure ~tu&es,University of South Alabama, ~ o i i l eAL
, 36688-0002.
868 S. F. PUGH, ET AL.

Analysis of the tennis serve indicates that both upper and lower extrem-
ity muscular strength is important in generating maximal racket velocity. The
primary joint and pattern of movement tested in the tennis player is shoul-
der rotation (Cohen, et a/., 1994). Testing of the lower extremity for knee
extension strength in tennis players has also been reported (Ellenbecker &
Roetert, 1995). However, no studies of tennis players' lower extremities
showing the relationship between tennis serve ball speed and muscle strength
were identified. It appears that upper-body strength may be associated with
ball speed during the tennis serve, but the exact contribution of the lower
body musculature has yet to be determined. The finding of a relationship
between lower body strength and ball speed could better define the role of
the lower body musculature in the tennis serve. The purpose of this study
was to assess the relative contribution of upper body, lower body, and grip
strength to ball speed during the tennis serve in the male collegiate tennis
player.

Subjects
Fifteen male collegiate NCAA Division I tennis players ( M age=20.8
yr., SD=2.0; M height = 181 cm, SD=6.6; M weight=79.9 kg., SD=8.2) were
subjects. Thirteen reported being right-hand dominant and two left-hand
dominant. All subjects reported the right leg as dominant. According to in-
stitutional guidehes, before subjects gave written consent, they were inform-
ed of the nature, purpose, and possible risks involved in this study as ap-
proved by the University Institutional Review Board.
Isokinetic Strength Testing
Strength testing of the dominant leg and shoulder was performed using
the LIDO Active Isokinetic Dynamometer (Loredan Biomedical, sacramen-
to, CA). Isokmetic dynamometry provides for maximal muscle force through-
out the joint range-of-motion at preselected velocities with an accommodat-
ing resistance. ~ e i a b h t ymeasures of isokinetic knee extension and shoulder
strength have been reported to be high (.94 to .98) (Brown, Whitehurst, &
Bryant, 1992). For the knee extension tests, the subject was seated with a hip
trunk angle of approximately 10O0, and the dynamometer shaft aligned with
the lateral knee joint h e . The ipsilateral thigh was secured with a bolster,
and two large velcro straps secured the pelvis and torso. Arms were crossed
over the chest during testing to prevent the arms from contributing to knee
extension force. The lever arm pad was positioned immediately superior to
the medial malleolus of the leg. Maximal concentric knee extension contrac-
tions were performed through 90" range-of-motion (90" to 0" of flexion) at a
speed of 300 deg . set.?. Ellenbecker and Roetert (1995) showed no signifi-
 STRENGTH AND BALL SPEED 869

cant bilateral difference when analyzing knee extension/flexion peak torque

at 180 and 300 deg . set.-' speeds. These results indicate bilateral symmetry
between knees in tennis players; therefore, only the dominant knee was test-
ed.
The shoulder rotation test was conducted on the serving arm with the
subject positioned supine on the testing table. The actuator shaft of the dy-
namometer was positioned on the testing side at a 90" angle to the table and
aligned with the longitudinal axis of the humerus. The elbow was flexed to
90"; the shoulder abducted to 90" and the hand grasped - -
a handle attached
to the actuator shaft. Maximal concentric reciprocal shoulder rotations were
performed through 160" range of motion (0" to 90" external rotation and 0"
to 70" internal rotation) at a speed of 300 deg . set.-l. With the shoulder in
the initial position of full internal rotation, the first motion was external ro-
tation.
All knee and shoulder strength tests consisted of each subject perform-
ing five submaximal and three near-maximal repetitions for warm-up, follow-
ed by a 60-sec. rest. The subject then performed five maximal contractions
through the full range of motion. Limb testing was randomized to reduce
order effect. Isokinetic peak torque (N. m) was measured and recorded us-
ing a computer interfaced with the dynamometer.
Grip Strength Testing
Dominant hand grip strength was measured using a hand grip dyna-
mometer (Lafayette Instruments, Lafayette, IN). Prior to testing, finger grip
was adjusted so that the third finger's middle ~halanxwas ~ositionedap-
proximately at a right angle. Testing was performed standing, elbow extend-
ed, and the arm positioned with the dynamometer parallel to the subject's
side. The dynamometer was maximally squeezed for three trials, with the
maximum force recorded in kilograms (kg). A reliabLty coefficient of .95
was established by calculating the intraclass correlation among trials.
Ball Speed Testing
Tennis serve ball speed was tested on a different day from the muscle
testing to eluninate the possibility of fatigue. Testing was conducted on a
standard-sized tennis court, and ball speed was measured over a distance of
27.4 m using a Jugs Radar Gun (Decatur Electronics, Decatur, IL; Model
620-c). The radar gun was mounted on a tripod at a height of 76.2 cm and
positioned 3 m behind the baseline and midway between the singles sideline
and the center mark of the court. Calibration was done using a tuning fork
prior to each test. Only new yellow tennis balls (Wilson Sporting Goods,
Chicago, IL) were used.
AU subjects performed a serving warm-up protocol used during the nor-
mal practice session. Subjects served from a position midway between the
870 S. F. PUGH, ET clL.

singles sideline and the center mark of the court on the opposite side of the
court from the radar gun. Subjects were encouraged to hit the ball as hard
as possible using normal tennis serve form. Not counted were net serves and
serves outside the boundaries of the singles sidelines, center mark, and back
basehne of the singles senling area. Ball speed was measured in kilometers
per hour (kph), with the median score of five trials used as the datum. The
reliability of the ball speed using an intraclass correlation was .89 over trials.
Statzitical Analysis
Multiple regression analysis was conducted to evaluate the relationship
among the strength measures and tennis serve ball speed. The criterion vari-
able was ball speed, and the predictors were the three strength measures:
peak torque of internal shoulder rotation, knee extension, and handgrip
strength (kg). An a priori level of significance was set at a I .05 for all com-
parisons.
RESULTS
In Table 1 are means and standard deviations for the predictor and cri-
terion variables. Table 2 shows the intercorrelation and coefficient of deter-

TABLE 1
MEANSA N D STANDARDDEVIATIONS
FORBALLSPEED(KPH)AND MEASURES OF MAXIMUM
~SOK~NET~C
FORCEPRODUCED BY DOMINANT (N. M)A N D GRIPSTRENGTH
KNEEA N D SHOULDER (KG)( N =15)
Variable IM SD
Knee 145.73 34.78
Shoulder 44.60 9.93
Grip 55.07 9.74
Ball Speed 173.78 6.34

mination matrix between the variables. A summary of the regression analysis

is found in Table 3. The linear combination of strength measures was not
significantly related to b d speed (F,,,, = .83, ns). The multiple correlation co-
efficient was .43, indicating that approximately 19% of the variance of b d
TABLE 2
COGFF~CIENTS
~NTERCORRELATIONS AND O F DETERM~NAT~ON* AMONG LEE EXTENSION,
SHOULDER (N . M), GIUPSTRENGTH
ROTATION (KG),A N D BALLSPEED(KM/HR.)

Variable Knee Shoulder Grip Speed

f

Knee .5 1 .66 36
Shoulder .26 .52 .29
Grip .44 .27 .4 1
Ball Speed .13 .08 .17
r2
*Coefficient of Determination ( r 2 )is below the diagonal.
 STRENGTH A N D BALL SPEED 87 1

speed was accounted for by the linear combination of the strength measures.
Relative strength indices of the indvidual predictors showed no significant
bivariate or partial correlations between the strength measures and ball
speed or among any of the predictor variables. One may infer that the
strength indices are not significant predictors of ball speed during the tennis
serve.

TABLE 3
SUMMARY
OF REGRESSION
ANALYSIS
FOR STRENGTHMFASURESPREDICTING
BALLSPEED( N = 15)
Variable B SE Beta I P
Constant 92.77 10.42
Knee .02 .06 ,133 .35 ns
Shoulder .05 .2 1 ,080 .24 ns
Grip .18 .25 ,280 .74 ns

Drscusslo~
Previous investigations using the tennis serve yielded moderate to low
correlations between b d speed and isokinetic strength of the upper body
(Ellenbecker, 1991; Cohen, et a/., 1994). No studies were identified in which
lower body strength associated with ball speed during the tennis serve was
examined. Present results support previous studies indicating a moderate re-
lationship between strength and ball speed in the tennis serve. It may be
that an absolute level of strength is necessary but not sufficient for ultimate
ball speed. Ball speed may be a combination of several factors such as tech-
nique, coordination, flexibLlity,and strength. Cohen, et a/. (1994) found flex-
ibility measures including wrist flexion, shoulder flexion, and shoulder inter-
nal rotation to be related to ball speed ( r = .31 to .40, p I ,051. An integral
component of the tennis serve could be exertion of muscular force at a high
speed while at the same time transferring force up through the legs and
trunk to the arms in proper sequence and timing (Roetert, McCormick,
Brown, & Elenbecker, 1996). Groppel (1992) noted that the timing of the
transfer of forces is crucial to the success and efficiency of the tennis serve.
The results of this study imply that strength, as tested, appears to be
moderately associated with ball speed in s u e d male tennis players. Ellen-
becker (1991) stated the lack of statistical significance between ball speed
and measured strength is not surprising when considering the neurornuscu-
lar and biological complexity of the tennis serve. The consistency of low to
moderate correlations between ball speed and measured strength in all the
reviewed studies suggests that more than just strength of the involved mus-
culature is involved in performing the tennis serve. These results do not
diminish the importance of strength. Wrigley (2000) stated that the lack of a
correlation between a given strength assessment and performance for a par-
872 S. F. PUGH. ET AL.

ticular sport may be the result of several factors. He cited factors such as
lack of relevance of strength to that performance; the insensitivity of the
strength assessment method to performance-specific strength, and a require-
ment for only a threshold level of strength above which an increase in
strength is not correlated with increased performance. Codine, Bernard, Po-
cholle, Benaim, and Brun (1997) reported isokinetic rotator cuff strength of
12 nonathletes, 12 runners, 15 tennis players, and 12 baseball players. The
comparison between dominant and nondominant extremities of tennis play-
ers showed no significant difference. Also, no significant difference was
found between mean peak torque values comparing tennis players with non-
athletes at the 300 deg . set.-' speed, although the tennis players' mean scores
were higher. Therefore, the range of isokinetic strength values might be
more restricted than that expected for nonathletes. A restriction in range
could result in a reduction in strength of the correlation with b d speed.
Further, studies should be focused on factors such as different modes and
speeds of strength testing, slull, and the influence of training.
REFERENCES
BROWN.L. E., WHITEHURST, M., &BRYANT,J. R. (1992) Reliability of the Lido Active isokinetic
dynamometer concentric mode. Isokinetic and Exerczre Science, 2, 191-194.
CODINE.I?. BERNARD, l? L., POCHOLLE, M., BENAIM, C., & BRUN,V. (1997) Influence of sports
discipline on shoulder rotator cuff balance. Medicine and Scze~~ce in Sports and Exercise,
29, 1400-1405.
COHEN,D. B., MONT,M. A,, CAMPBELL, K. R., VOCEUTEIN, B. N., &LOEW, J. W. (1994) Upper
extremity physical factors affecring tennis serve velocity. American Joz(ritaf of Sports Medi-
czne, 22, 746-750.
ELLENBECKER, T. S. (1991) A total arm strength isokinetic profile of highly skilled tennis play-
ers. Isokinetic and Exercise Science, 1, 9-21.
ELLENBECKER. T. S.. DAVIES.G. J., & ROWINSKI. M. J. (1988) Concentric versus eccentric isoki-
netic strengthening of the rotator cuff: objective data versus funcrional test. American Jour-
nal of Sports Medicine, 16, 64-68.
ELLENBECKER, T. S., & ROETERT,E. I? (1995) Concentric isokinetic quadriceps and hamstring
strength in elite junior tennis players. Isokinetic and Exernlre Science, 5 , 3-6.
GROPPEL, 1. L. (1992) High tech tennis. Champaign, IL: Human Kinetics.
ROETERT,E. I?, MCCORMICK, T. J.. BROWN, S. W., &ELLENBECKER, T. S. (1996) Relationship be-
tween isokinetic and functional trunk strength in elite junior tennis players. Isoki~zeticand
Exerase Sae~zce,6 , 15-20.
TREIBER,F. A,, LOTT,J., DUNW. I., SLAVENS, G., &DAVIS.H. (1998) Effects of theraband and
lightweight dumbbell training on shoulder rotation torque and serve performance in col-
lege tennis players. American Jozrrizaf of Sporls Medicine, 26, 510-515.
WRICLEY. T. V (2000) Correlations with athletic performance. In L. Brown (Ed.), Isokirzetics in
human performance. Champaign, IL: Human Kinetics. P p . 42-73.

Accepted September 25, 2003.