Journal of Sports Sciences, January **th ****; **(2): 181 188
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
The role of upper torso and pelvis rotation in driving performance
during the golf swing
JOSEPH MYERS1, SCOTT LEPHART2, YUNG-SHEN TSAI3, TIMOTHY SELL2,
JAMES SMOLIGA2, & JOHN JOLLY4
1
Department of Exercise and Sport Science, University of North Carolina, Chapel Hill, NC, USA, 2Department of Sports
Medicine and Nutrition, University of Pittsburgh, Pittsburgh, PA, USA, 3Department of Physical Therapy, National
Cheng Kung University, Tainan City, Taiwan, ROC, and 4Center for Adaptive Neural Systems, Arizona State
University, Tempe, AZ, USA
(Accepted 29 March 2007)
Abstract
While the role of the upper torso and pelvis in driving performance is anecdotally appreciated by golf instructors, their actual
biomechanical role is unclear. The aims of this study were to describe upper torso and pelvis rotation and velocity during the
golf swing and determine their role in ball velocity. One hundred recreational golfers underwent a biomechanical golf swing
analysis using their own driver. Upper torso and pelvic rotation and velocity, and torso-pelvic separation and velocity, were
measured for each swing. Ball velocity was assessed with a golf launch monitor. Group differences (groups based on ball
velocity) and moderate relationships (r ! 0.50; P 5 0.001) were observed between an increase in ball velocity and the
following variables: increased torso pelvic separation at the top of the swing, maximum torso pelvic separation, maximum
upper torso rotation velocity, upper torso rotational velocity at lead arm parallel and last 40 ms before impact, maximum
torso pelvic separation velocity and torso pelvic separation velocity at both lead arm parallel and at the last 40 ms before
impact. Torso pelvic separation contributes to greater upper torso rotation velocity and torso pelvic separation velocity
during the downswing, ultimately contributing to greater ball velocity. Golf instructors can consider increasing ball velocity
by maximizing separation between the upper torso and pelvis at the top of and initiation of the downswing.
Keywords: Golf, swing mechanics, biomechanics, kinematics
Often, teaching professionals seek to maximize
Introduction
upper torso rotation during the backswing while
The current teaching philosophy of the golf swing minimizing pelvic rotation in their students, creating
emphasizes an increase in torso coiling during the torso pelvic separation. Potentially, this creates
backswing, which theoretically results in increased resistance between the upper torso and pelvis during
impulse during the downswing, and subsequent the backswing, increasing the stored energy, which is
increased ball velocity and ball ight distance. In released during the downswing. The release of stored
pro cient golfers, the backswing is initiated by energy results in more impulse and increased club
simultaneous rotation of the upper torso, upper head speed, ball velocity, and therefore driving
extremities (arms, wrists, and hands), and club away distance. Teaching professionals often describe this
from the address position, followed immediately by separation between the upper torso and pelvis
some degree of pelvic rotation (Hogan & Wind, rotation as x-factor or segment separation,
1957; McTeigue, Lamb, Mottram, & Pirozzolo, which is speci cally de ned as the difference in axial
1994). The order of events during the downswing rotation between the upper torso and pelvis at the top
includes initiating pelvic rotation back towards the of the backswing (McLean & Andrisani, 1997). It is
impact position, immediately followed by upper believed that maximizing torso pelvic separation
torso rotation, and movement of the arms, wrists, will contribute to increased ball velocity and
hands, and club (Hogan & Wind, 1957; McTeigue driving distance and as such has recently been
et al., 1994). described as the secret power move to add 25
Correspondence: J. Myers, Department of Exercise and Sport Science, University of North Carolina, Chapel Hill, NC 27599, USA.
E-mail: abptm5@r.postjobfree.com
ISSN 0264-0414 print/ISSN 1466-447X online 2008 Taylor & Francis
DOI: 10.1080/02640410701373543
182 J. Myers et al.
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
yards (Kostis & Midland, 2006). The x-factor hypothesized that as this separation between the
stretch has been described as the maximum upper torso and pelvic rotation increases, the
torso pelvic separation that occurs during the resulting increase in concentric contraction during
downswing and is suggested to result from initiation the golf swing will increase club head speed, resulting
of the downswing with the pelvis rotating back in increased ball velocity and driving distance.
towards the impact position while the upper torso While individuals who teach and study the golf
is still rotating towards the top of the backswing, swing anecdotally appreciate the important role that
creating maximum separation between the segments the upper torso and pelvis play in increasing ball
(McLean & Andrisani, 1997). Burden and collea- velocity and driving distance, the biomechanical role
gues (Burden, Grimshaw, & Wallace, 1998) demon- of the upper torso and pelvis rotation and resulting
strated that skilled golfers (sub-10 handicap) perform driving performance characteristics has not been
this countermovement of the pelvis and upper torso scienti cally described. To date, there is little peer-
at the start of the downswing. They further describe reviewed published research that describes the role of
how the countermovement of the pelvis and upper upper torso and pelvis rotation in generating driving
torso create a summation of speed that ultimately performance. In the present study, ball velocity was
results in greater force being applied by the club to the variable we used to represent driving perfor-
the ball at impact. mance. The aims of the study were to describe upper
From a biomechanics perspective, this belief that torso and pelvis rotation and velocity during the golf
torso pelvic separation is an important contributor swing and determine their role in ball velocity. The
to increasing driving distance has merit. The action study provides golf instructors, clinicians, and
of the torso during the golf swing can be classi ed researchers with a description of the upper torso
as a stretch shortening movement (Fletcher & and pelvis during the golf swing and their role in
Hartwell, 2004). Movements that involve a stretch generating ball velocity, in the hope of applying the
shortening contraction utilize stretching active mus- results to how the swing is taught.
cles (eccentric loading) to load the muscle in order to
increase power output during the nal phase of the
movement (concentric shortening) (Komi, 1984, Methods
2000; Norman & Komi, 1979). Ultimately, a muscle
Participants
that is eccentrically loaded before a concentric
contraction results in increased force and power One hundred recreational golfers participated in the
production compared with an isolated concentric or study. All participants had a United States Golf
eccentric muscle contraction (Ettema, Huijing, & De Association registered handicap. Complete partici-
Haan, 1992; Ettema, Huijing, Van Ingen Schenau, & pant demographics are given in Table I. All
De Haan, 1990a; Ettema, Van Soest, & Huijing, participants were free of injury and had no signi cant
1990b). The increased force production is a result of history of joint injury at the time of testing. All
utilization of elastic energy within the muscle participants provided informed consent as required
tendon unit during the eccentric loading of the by the university s institutional review board.
active muscle that is released during the concentric
phase of the movement. (Finni, Ikegawa, Lepola, &
Instrumentation
Komi, 2003; Komi, 2000).
We can potentially apply these stretch shortening Kinematic data of the golf swing were collected using
principles to the golf swing. Electromyography the Peak Motus System v.8.2 (Peak Performance
studies have demonstrated that the trunk muscles Technologies, Inc., Englewood, CO). This is a
including erector spinae, abdominal obliques, rectus three-dimensional motion analysis system with eight
abdominis, latissimus dorsi, and gluteals are active optical cameras that surround the golfer, each placed
during the backswing (Horton, Lindsay, & Macintosh, at a distance of 4 m from the golf teeing area. A
2001; Pink, Jobe, & Perry, 1990; Pink, Perry, & sampling rate of 200 frames per second was used in
Jobe, 1993; Watkins, Uppal, Perry, Pink, & Dinsay, this study. Calibration was done using the wand
1996). Additionally, during the backswing, separa- calibration method according to the manufacturer s
tion between the upper torso and pelvis results in guidelines. Our laboratory has established both the
stretching (eccentric loading) of these activated trunk position and orientation error of our system, result-
ing in root mean square error of 0.002 m and 0.2548
muscles, which could ultimately contribute to the
powerful concentric trunk muscle contractions respectively.
needed to drive the ball. These activated muscles Ball ight characteristics were assessed with the
play a signi cant role in generating club head speed Flight Scope Sim Sensor (EDH, Ltd., South Africa)
during the downswing (Horton et al., 2001; Pink integrated with AboutGolf (AboutGolf Limited,
et al., 1990, 1993; Watkins et al., 1996). Thus, it is Maumee, OH) simulation software. The Flight
183
Upper torso and pelvis rotation in golf driving performance
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
Table I. Demographics of the participants (mean + s).
Age (years) Stature (m) Body mass (kg) USGA handicap index
All golfers (n 100) 45.1 + 15.9 1.80 + 0.07 86.5 + 14.0 8.1 + 7.3
Golfers with low ball velocity (n 21) 58.5 + 13.7 1.79 + 0.08 85.1 + 11.5 15.1 + 5.2
Golfers with medium ball velocity (n 65) 44.6 + 14.9 1.80 + 0.07 86.7 + 14.8 7.8 + 6.9
Golfers with high ball velocity (n 14) 33.1 + 11.4 1.82 + 0.05 87.4 + 13.4 1.8 + 3.2
Scope Sim Sensor applies three-dimensional phased-
array microwave technology that operates at 7 kHz to
track ball ight from club impact until impact with a
screen 5 m away. Ball velocity, vertical launch angle,
horizontal launch angle, spin rates, carry distance,
and total distance are derived from ball tracking data.
The variable of interest in the current study was ball
velocity given that this variable is measured directly
by the ball ight sensor from the point of ball impact
until the ball hits the protective backstop.
Procedures
Each participant attended one test session. Partici-
pants were tted with retrore ective markers
(0.025 m diameter) on the sacrum and seventh
cervical vertebra as well as bilaterally on the anterior
superior iliac spine, acromion, and lateral epicondyle
of the humerus. In addition to the lateral epicondyle
markers, two markers were placed on the golf club to
identify the phases of the golf swing (Figure 1).
Each participant was instructed to warm up before
data collection at their own discretion. Common
modes of warm-up included but were not limited to
cardiovascular warm-up on a treadmill or exercise
bike, stretching, the swinging of weighted clubs or
Figure 1. The retrore ective marker con guration utilized for
training devices, and hitting practice shots with the
motion analysis of the golf swing. Permission was obtained from
club of the participant s choosing. Data collection
the participant to reproduce this image.
consisted of each participant hitting the same brand
of golf ball (Titleist, Acushnet Co., Acushnet, MA)
with their own driver to represent the swing and ball
ight patterns experienced while playing. Partici- change direction (along all global axes) at the end of
pants hit 10 shots off an arti cial turf tee box into a the backswing. Lead arm parallel was the instant
projected practice range image on the screen (back- during the downswing where a vector connecting the
stop) while both kinematics of the golf swing and ball shoulder marker and the elbow marker is parallel to
ight characteristics were collected. the horizontal plane of the global coordinate system.
Impact was the point when the club makes contact
with the ball and was identi ed with a view camera
Data reduction
that was synchronized with the kinematic collection
While each participant hit 10 shots, only the ve cameras and veri ed with club coordinate data. The
shots with the highest ball velocity were reduced and last 40 ms point was de ned as eight frames before
analysed. The swing points of interest, including top the impact point given the sampling frequency was
of the swing, lead arm parallel during the down- 200 frames per second. Miura (2001) described the
swing, the point at the last 40 ms before impact, and last 40 ms before impact to be an important instant
impact, were determined from the position of the during the swing because much of the momentum
club and upper extremity markers. Top of the swing generated by the body is imparted on the club at this
was calculated as the point when the club markers time. The raw coordinate data collected from each
184 J. Myers et al.
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
camera were ltered using an optimized cut-off de ned based on whether individual ball velocity
frequency and used to calculate the kinematic data fell below the group mean minus the standard
(Jackson, 1979). Upper torso rotation and pelvis deviation (a group of golfers with low ball velocity),
rotation angles were calculated as the angle between above the mean plus standard deviation (a group of
the respective segment and the global x-axis. The golfers with high ball velocity) or within the window
of the mean + standard deviation (a group of
global x-axis was set up so that a neutral address
position of the upper torso and pelvis would be zero golfers with medium ball velocity). Group demo-
degrees (Figure 2). The torso pelvic separation graphics are given in Table I. Group comparisons
variable was calculated as the difference between were made using one-way analyses of variance
the pelvic rotation angle and upper torso rotation (ANOVA) with post-hoc Bonferroni correction
angle and at the top of the backswing (Figure 2). analyses. Pearson pairwise correlations were also
Maximum torso pelvic separation was de ned as used to examine further the relationships between
the maximum difference between the upper torso ball velocity and upper torso and pelvic rotation
rotation angle and pelvic rotation angle that occurred angles and velocities measured. All statistical
during the downswing and is representative of x- assumptions underlying the use of parametric
factor stretch described in the golf instruction procedures were checked and veri ed. All statistical
literature (McLean & Andrisani, 1997) . In the analyses were performed with the SPSS v.11.0
current study, the torso pelvic separation is repre- (SPSS Inc, Chicago IL) statistical software package.
sented by a negative number, since the upper torso An alpha level of 0.05 was set a priori for statistical
rotation commonly exceeds pelvic rotation. As such, analyses.
more separation between the two segments will be
represented by a more negative value. Upper torso
Results
and pelvic rotational velocity was de ned as the rate
of change of the rotation angle with respect to time. The mean ball velocity for the entire group was
64.9 m s71 (s 6.8). From these data, group
Torso pelvic separation velocity was de ned as the
rate of change of the separation with respect to time. strati cations included a group of golfers
(n 21) with low ball velocity (558.1 m s71
A negative torso pelvic separation velocity repre-
[64.9 7 6.8 58.1]), a group (n 14) with high ball
sents coiling, while a positive torso pelvic separation
velocity (4 71.8 m s71 [64.9 6.8 71.8]), and a
velocity represents uncoiling. The mean of the ve
group (n 65) with medium ball velocity within the
shots reduced for data analysis was recorded for
window of the mean + standard deviation (58.1
statistical analysis.
71.8 m s71). Strati ed group demographics are
provided in Table I. The means, standard deviations,
Statistical analysis
and notation of group statistical differences (based
At the end of data collection, the 100 golfers was on the ANOVA with Bonferroni post-hoc analyses)
strati ed according to their ball velocity. The group for each of the variables assessed are presented in
mean and standard deviation of the entire group s Table II. Group differences were observed for
ball velocity was calculated. Group strati cations torso pelvic separation at the top of the swing,
were based on descriptive data and groups were maximum torso pelvic separation, upper torso
rotation velocity at lead arm parallel and last 40 ms
before impact, maximum upper torso velocity, pelvic
rotation at the top of swing, lead arm parallel and last
40 ms before impact, maximum pelvic rotation
velocity, torso pelvic separation velocity lead arm
parallel and last 40 ms before impact, and maximum
torso pelvic separation velocity.
The correlation coef cients and level of signi -
cance for all comparisons are given in Table III. A
moderate positive correlation was observed be-
tween ball velocity and torso pelvic separation at
the top of the swing, maximum torso pelvic
separation, maximum upper torso rotational velo-
city, upper torso rotational velocity at lead arm
parallel and last 40 ms before impact, maximum
torso pelvic separation velocity, and torso pelvic
separation velocity at both lead arm parallel and
Figure 2. De nition of the upper torso rotation, pelvic rotation,
last 40 ms before impact.
and torso pelvic separation angles assessed.
185
Upper torso and pelvis rotation in golf driving performance
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
Table II. Swing mechanics descriptive statistics (mean + s).
Low ball Medium ball High ball
velocity velocity velocity P
Ball velocity (m s71) 50.0001a
55.7 + 2.7 65.6 + 3.7 75.4 + 4.4
Upper torso rotation at top of swing (8) 794.0 + 13.5 797.0 + 20.2 7104.0 + 10.3 0.264
Upper torso rotation at lead arm parallel (8) 736.1 + 9.1 739.6 + 15.9 737.4 + 9.4 0.588
Upper torso rotation at last 40 ms before impact (8) 0.0 + 10.0 70.2 + 16.2 2.3 + 7.2 0.835
Upper torso rotation at impact (8) 20.3 + 10.2 22.8 + 16.1 25.2 + 8.9 0.600
Maximum upper torso rotation (8) 795.4 + 13.2 798.3 + 19.9 7106.1 + 10.6 0.201
Pelvic rotation at top of swing (8) 749.8 + 11.4 747.5 + 17.4 744.9 + 10.3 0.660
Pelvic rotation at lead arm parallel (8) 74.3 + 7.4 71.3 + 16.6 6.4 + 7.2 0.087
Pelvic rotation at last 40 ms before impact (8) 17.8 + 8.1 22.4 + 16.9 26.8 + 6.7 0.190
Pelvic rotation at impact (8) 29.4 + 8.6 35.3 + 17.0 38.3 + 7.2 0.163
Maximum pelvic rotation (8) 753.2 + 10.6 751.3 + 16.5 750.4 + 10.1 0.210
50.0001b,c
Torso pelvic separation at top of swing (8) 744.2 + 7.7 749.5 + 9.6 759.1 + 8.2
50.0001a
Maximum torso pelvic separation (8) 745.6 + 8.0 751.7 + 10.3 761.8 + 7.8
Upper torso rotation velocity at top of swing (8 s71) 48.8 + 55.5 54.4 + 62.6 79.6 + 56.2 0.296
Upper torso rotation velocity at lead arm parallel (8 s71) 50.0001a
546.1 + 61.6 625.6 + 99.0 738.3 + 79.2
Upper torso rotation velocity at last 40 ms before impact (8 s71) 50.0001b,d
515.5 + 72.2 603.3 + 76.3 637.9 + 81.7
Upper torso rotation velocity at impact (8 s71) 498.7 + 123.5 539.1 + 98.8 520.1 + 117.1 0.321
Maximum upper torso rotation velocity (8 s71) 50.0001a
591.2 + 66.8 675.1 + 84.4 766.6 + 73.0
Pelvic rotation velocity at top of swing (8 s71) 0.032b
74.8 + 57.9 96.3 + 59.7 128.7 + 52.4
Pelvic rotation velocity at lead arm parallel (8 s71) 0.015d
348.8 + 59.9 395.4 + 67.2 401.7 + 67.5
Pelvic rotation velocity at last 40 ms before impact (8 s71) 0.021d
310.2 + 63.4 349.8 + 58.5 318.7 + 72.4
Pelvic rotation velocity at impact (8 s71) 258.8 + 65.7 277.4 + 67.0 248.5 + 82.9 0.270
Maximum pelvic rotation velocity (8 s71) 0.003b,d
357.6 + 58.3 410.4 + 66.4 433.6 + 90.9
Torso pelvic separation velocity at top of swing (8 s71) 726.0 + 27.2 741.9 + 37.1 749.1 + 31.2 0.104
Torso pelvic separation velocity at lead arm parallel (8 s71) 50.0001b,c
197.3 + 47.7 230.2 + 68.4 336.6 + 69.9
Torso pelvic separation velocity at last 40 ms before impact (8 s71) 50.0001a
205.4 + 47.0 253.4 + 66.9 319.2 + 65.6
Torso pelvic separation velocity at impact (8 s71) 239.9 + 86.6 261.7 + 70.6 271.6 + 86.0 0.416
Maximum torso pelvic separation velocity (8 s71) 50.0001b,c
278.1 + 46.6 311.8 + 60.3 389.6 + 55.6
a
All three groups signi cantly different.
b
Low ball velocity group signi cantly different vs. high ball velocity group.
c
Medium ball velocity group signi cantly different vs. high ball velocity group.
d
Low ball velocity group signi cantly different vs. medium ball velocity group.
Table III. Correlation coef cients between ball velocity and the upper torso and pelvic rotation variables assessed.
Lead arm Last 40 ms
Maximum Top of swing parallel before impact Impact
70.54 (P 5 0.001) 70.55 (P 5 0.001)
Torso pelvic separation N. A . N. A . N.A.
(P 0.056) (P 0.324) (P 0.959) (P 0.467)
70.19 70.10
Upper torso rotation N. A . 0.01 0.07
(P 0.189) (P 0.023) (P 0.041) (P 0.042)
Pelvic rotation N. A . 0.13 0.23 0.21 0.20
(P 0.025) (P 0.536)
0.59 (P 5 0.001) (P 5 0.001) (P 5 0.001)
Upper torso rotational velocity 0.23 0.06
0.61 0.50
(P 0.439) (P 0.604)
0.36 (P 5 0.001) (P 5 0.001) (P 5 0.001) 70.05
Pelvic rotational velocity 0.35 0.32 0.08
(P 0.041) (P 0.178)
0.50 (P 5 0.001) 70.21 (P 5 0.001) (P 5 0.001)
Torso pelvic separation 0.14
0.55 0.53
velocity
Note: Moderate relationships (r 4 0.5) are shown in bold font. N.A. not applicable.
upper torso and pelvic rotation position does not
Discussion
make a signi cant contribution to ball velocity;
The aims of this study were to examine the relation- rather, it is the separation between the two segments
ship between ball velocity, upper torso rotation that appears to be most important contributor given
position and velocity, pelvis rotation position the group differences and moderate relationship that
and velocity, and torso pelvic separation position was observed between torso pelvic separation (at
and velocity measured during the golf swing, and top of swing and maximum) and ball velocity. Thus,
determine whether the upper torso and pelvic if a golfer increases upper torso rotation at the top of
variables differed between groups strati ed by ball the golf swing to increase ball velocity, it would be
velocity. The results indicate that the magnitude of most effective to do so while limiting the amount of
186 J. Myers et al.
Downloaded By: [University of Pittsburgh] At: 18:51 27 November 2007
pelvic rotation, thereby increasing separation be- As in the current study, McTeigue et al. found that
tween the two segments. the torso and pelvis position were less important than
Theoretically, this greater separation could result the separation between the two segments (torso
in eccentric loading of the torso musculature through pelvic separation). Their results demonstrated that
lengthening (Fletcher & Hartwell, 2004). This professional Tour players who have long driving
eccentric loading potentially could lead to faster distances tended to have increased separation be-
uncoiling velocity during the downswing, thereby tween the upper torso and pelvic segments. It was
contributing to increased ball velocity. Additionally, hypothesized that the increased separation was a
the greater separation between segments might allow result of the pelvis starting the downswing while the
increased time for force to be applied to the club, torso continued to rotate away from the target
resulting in greater impulse. Thus, there is the (maximum torso pelvic separation). Cheetham
potential to increase momentum of the club and and colleagues (Cheetham, Martin, Mottram, &
ultimately increase club velocity. In the current Laurent, 2000) reported that touring professionals
study, both torso pelvic velocity (which represents demonstrated increased torso pelvic separation at
the how quickly the golfer uncoils during the both the top of the swing and maximum torso
downswing) and upper torso rotational velocity pelvic separation compared with amateur golfers,
during the downswing (maximum, at lead arm