By - the UWA Profiles and Research Repository · Machar Reid This Thesis is Presented for the...

LOADING AND VELOCITY GENERATION IN THE HIGH PERFORMANCE TENNIS SERVE

By

Machar Reid

This Thesis is Presented for the

Doctor of Philosophy

School of Human Movement and Exercise Science

The University of Western Australia

THE COPYRIGHT STATEMENT

Works produced under this Research project will be housed by the School as an

internal resource collection. These works are not to be reproduced or published,

either in whole or part until any known copyright material has been granted an

official clearance by the responsible authority.

ii

ACKNOWLEDGEMENTS

First and foremost, I would like to extend my gratitude to Professor Bruce Elliott and

Dr Jacque Alderson. Without their enthusiasm and patience, this dissertation would

have paralleled Richmond’s quest to finish in the AFL’s top eight. I’m similarly

indebted to Si, Bakes, Lloydy, Amity and Winbas for their assistance throughout the

compilation of this thesis.

To that end, all of those that resided in 1.5.1 must be commended for surviving the

deprivation of Vitamin D but also a certain resIdent.

Further appreciation to the high performance players for their time and willingness to

participate, and also to Milo Bradley for pointing me in the right direction.

To all of the staff and technicians in the School of Human Movement and Exercise

Science, your efforts have been invaluable, and your ‘humour’ robust.

And before exhausting the synonymistic genius of Roget’s, I’d like to acknowledge

the enormous role that the International Tennis Federation, and more particularly Dr

Miguel Crespo and Dave Miley, as well as Dr Ann Quinn, have played in the

background.

I dedicated my honours thesis to my family, and while a wonderful idea at the time,

I’ll refrain from doing as much this time round. Rather, with this document’s cob-web

collecting fate all but sealed, they – along with Amity – deserve far greater eulogising

… thanks.

iii

ABSTRACT

Shoulder injuries rank among the most prevalent and debilitating sustained by

professional tennis players. The loads, or magnitude, location, direction, duration,

frequency, variability and rate of force application, endured by tissues of the shoulder

during stroke production, and more particularly the serve, are commonly implicated

in shoulder joint injury (Chandler et al., 1992; McCann and Bigliani, 1994; Kibler,

1995). Indeed, past evidence points to these loads increasing along with serve

velocity, as well as with varied segment use (Elliott et al., 2003). This dissertation

therefore aimed to quantify hypothesised relationships between certain serve types

and techniques, and shoulder joint loading among high performance able-bodied and

wheelchair players.

In some agreement with the theoretical models of performance analysis advanced by

Mullineaux et al. (2001) and Knudson and Morrison (2002), reliable evaluation of the

serve, and by extension loading of the shoulder, required that at least three

successful service trials be considered. Interpolation of data describing the serve

from one frame pre-impact to five frames post-impact was also required to effectively

minimise end point error associated with racquet-ball impact, and thus ensure the

validity of the kinematic and kinetic analysis of the serve.

Similar shoulder joint kinetics were shown to assist the development of dichotomous

3D racquet velocities in the high performance able-bodied flat serve (FS) and kick

serve (KS). That is, higher peak horizontal, vertical and absolute racquet velocities

were developed during the FS, while higher lateral velocities characterised the KS.

The comparable shoulder joint loading conditions nevertheless point to the repetitive,

long-term performance of either serve as relevant in shoulder joint injury

pathologies.

Coordinative lower limb variation in the serve, encapsulated by able-bodied players’

ranges of front and rear knee joint extension and peak angular velocity of rear knee

joint extension, was also shown to influence the development of FS racquet velocity.

Facilitated by pronounced bilateral knee joint extension, high-performance players

generated similar absolute pre-impact racquet velocities from both foot-up (FU) and

(FB) service stances. Conversely, less dynamic engagement of their lower limbs saw

iv

players unable to generate commensurate pre-impact absolute racquet velocities.

Interestingly, comparable shoulder joint kinetics were inherent to the FS, irrespective

of the lower limb kinematic variation observed in the FU, FB and ARM (i.e. FSs hit

with minimal active ankle, knee and hip joint flexion-extension) techniques. So, with

technique-related differences in absolute racquet velocities arising from similar

shoulder joint loads but divergent lower limb drives, it is plausible that other links in

the ‘kinetic chain’, such as the trunk, may be more affected by different leg actions.

In contrast to able-bodied FS and KS performance, similar peak pre-impact absolute

racquet velocities were generated during the wheelchair FS (WFS) and KS (WKS).

Wheelchair serve tactics nevertheless necessitated that higher peak pre-impact

horizontal and lateral racquet velocities also punctuated the WFS and WKS

respectively. The shoulder joint kinetics that contributed to these differential velocity

profiles were consistent across wheelchair serve type, but specific to the individual

players; likely varying with their level and severity of spinal cord injury. When

expressed relative to absolute racquet velocity, both high-performance able-bodied

and wheelchair players experienced matching pre- and post-impact shoulder joint

loads such that both playing populations appear subject to parallel shoulder joint

injury risk.

Indications are thus that high performance players can expect to develop different

pre-impact racquet velocities depending on the type of serve they hit, and the

technique that they employ. The influence of serve type and technique on the

shoulder joint kinetics that help to produce these racquet velocities is less obvious.

More specifically, it appears that players tolerate similar pre- and post-impact

shoulder joint loads, regardless of serve performed. The prospect of high

performance players experiencing injurious loading conditions at the shoulder would

seem independent of able-bodied serve type and technique, and wheelchair serve

type.

Of final note is that prospective 3D biomechanical examinations of shoulder joint

motion in the tennis serve should consider placement of humeral triads distal to the

biceps and/or triceps muscle belly. In comparison to markers placed at the mid-point

of the humerus (i.e. as used in this thesis), these more distal triad positions appear

to alleviate the spurious effects of soft tissue artefact thereby enhancing the accuracy

v

of estimated long-axis rotation of the upper arm. Although the current representation

of 3D humeral motion did not confound the comparisons made between serve types

or techniques, it is likely that upper arm triads located just above the epicondyles of

the humerus could have offered more insightful absolute comparisons to the

literature. Further, the elaboration of a joint coordinate system at the shoulder to

provide for the more meaningful and functional expression and interpretation of

shoulder joint kinetic and kinematic data should also be central to all future, related

investigative efforts.

vi

TABLE OF CONTENTS

THE COPYRIGHT STATEMENT.................................................................................... ii

ACKNOWLEDGEMENTS..............................................................................................iii

ABSTRACT............................................................................................................... iv

TABLE OF CONTENTS...............................................................................................vii

LIST OF TABLES ...................................................................................................... xi

LIST OF FIGURES ................................................................................................... xiii

LIST OF APPENDICES............................................................................................. xvii

CHAPTER 1: THE PROBLEM ...................................................................................... 1 1.1 INTRODUCTION.................................................................................................... 1 1.2 STATEMENT OF THE PROBLEM.............................................................................. 4 1.3 JUSTIFICATION OF THE STUDY............................................................................. 5 1.4 HYPOTHESES....................................................................................................... 6 1.5 LIMITATIONS....................................................................................................... 8 1.6 DELIMITATIONS................................................................................................... 9 1.7 DEFINITION OF TERMS ........................................................................................ 9

CHAPTER 2: LITERATURE REVIEW......................................................................... 16 2.1 INTRODUCTION.................................................................................................. 16 2.2 LOADING AND MUSCLE INJURY ........................................................................... 18 2.3 EPIDEMIOLOGY OF INJURIES IN TENNIS PLAYERS................................................ 20

2.3.1 Profile of Tennis Injury.................................................................................. 21 2.3.1.1 Injury distribution .................................................................................. 21

2.3.1.1.1 Junior tennis................................................................................... 21 2.3.1.1.2 Elite adult and professional tennis .................................................... 24

2.3.2 Types of Injuries .......................................................................................... 25 2.4 SHOULDER INJURIES IN TENNIS PLAYERS............................................................ 26

2.4.1 Anatomy of the Shoulder............................................................................... 27 2.4.2 Proposed Mechanisms of Shoulder Injury........................................................ 30

2.4.2.1 Mechanisms of injury to the rotator cuff................................................... 31 2.4.2.1.1 Primary impingement ...................................................................... 31 2.4.2.1.2 Primary tensile overload (rotator cuff failure)..................................... 31 2.4.2.1.3 Instability ....................................................................................... 32 2.4.2.1.4 Secondary impingement .................................................................. 33 2.4.2.1.5 Macrotrauma .................................................................................. 33

2.4.3 Variables that may contribute to Shoulder Injury............................................. 34 2.4.3.1 Technique-related factors ....................................................................... 34 2.4.3.2 Inflexibility ............................................................................................ 35 2.4.3.3 Muscle imbalance................................................................................... 36 2.4.3.4 Improper scapula mechanics................................................................... 38

2.5 THE RELATIONSHIPS BETWEEN SERVING MECHANICS, SHOULDER LOADING AND PERFORMANCE......................................................................................................... 40

2.5.1 Serve Type .................................................................................................. 40 2.5.2 Foot Arrangement ........................................................................................ 41 2.5.3 Leg Drive ..................................................................................................... 42 2.5.4 Follow-through ............................................................................................. 44 2.5.5 Wheelchair Tennis Serve ............................................................................... 46

vii

2.6 REPEATABILITY OF KINETIC AND KINEMATIC DATA IN MOTION............................ 49 2.6.1 Selection of Data Treatment Procedures ......................................................... 51

2.7 CONCLUSION...................................................................................................... 52

CHAPTER 3: GENERAL METHODOLOGY.................................................................. 54 3.1 OVERVIEW OF 3D INTERPRETATION OF THE SHOULDER JOINT ............................ 54

3.1.1 Analysis of Upper Arm Position with Euler Z-X-Y and ISB Decompositions.......... 59 3.2 METHODS OF DATA COLLECTION ........................................................................ 66

3.2.1 Subject Information ...................................................................................... 66 3.2.2 Overview of the UWA Model .......................................................................... 66 3.2.3 UWA Marker Set ........................................................................................... 67

3.2.3.1 Marker set used for wheelchair players .................................................... 72 3.2.4 Pointer Method............................................................................................. 73 3.2.5 Joint Centre Definitions ................................................................................. 74

3.2.5.1 Shoulder ............................................................................................... 74 3.2.5.2 Elbow ................................................................................................... 75 3.2.5.3 Wrist..................................................................................................... 75 3.2.5.4 Hip ....................................................................................................... 75 3.2.5.5 Knee..................................................................................................... 76 3.2.5.6 Ankle .................................................................................................... 76

3.2.6 Segment Coordinate Definitions ..................................................................... 76 3.2.6.1 Head .................................................................................................... 76 3.2.6.2 Thorax and torso ................................................................................... 77 3.2.6.3 Upper arm............................................................................................. 77 3.2.6.4 Forearm................................................................................................ 77 3.2.6.5 Hand .................................................................................................... 78 3.2.6.6 Pelvis .................................................................................................... 78 3.2.6.7 Femur................................................................................................... 78 3.2.6.8 Lower leg .............................................................................................. 78 3.2.6.9 Foot...................................................................................................... 78 3.2.6.10 Racquet .............................................................................................. 79

3.2.7 Definition of Joint Coordinate Systems............................................................ 80 3.2.7.1 Shoulder ............................................................................................... 80

3.2.8 Calculation and Interpretation of 3D Joint Kinetics........................................... 81 3.2.8.1 Definitions of force, moment and power .................................................. 81 3.2.8.2 Bodybuilder computation of 3D joint kinetics ............................................ 82 3.2.8.3 Functional interpretation of force, moment and power .............................. 82

3.2.9 Data Collection ............................................................................................. 84 3.2.10 Data Reduction........................................................................................... 85

CHAPTER 4: QUANTIFICATION OF DATA TREATMENT AND ANALYSIS TECHNIQUES OF THE TENNIS SERVE..................................................................... 87

4.1 INTRODUCTION.................................................................................................. 87 4.2 METHODOLOGY OF DATA TREATMENT................................................................. 88

4.2.1 Subject Preparation and Performance............................................................. 88 4.2.2 Selection of the MSE for Data Smoothing........................................................ 89

4.2.2.1 Results.................................................................................................. 90 4.2.3 Assessment and Selection of Data Treatment – Negotiating Impact .................. 94

4.2.3.1 Results.................................................................................................. 95 4.2.4 Determining the Repeatability of the Tennis Serve........................................... 97

4.2.4.1 Results.................................................................................................. 98 4.3 DISCUSSION..................................................................................................... 100

4.3.1 MSE for Best Representation of Tennis Serve Motion ..................................... 101 4.3.2 Assessment and Selection of Data Treatment – Negotiating Impact ................ 102 4.3.3 The Repeatability of the Tennis Serve .......................................................... 104

4.4 CONCLUSION.................................................................................................... 108

viii

CHAPTER 5: BIOMECHANICAL COMPARISON OF THE HIGH PERFORMANCE FLAT AND KICK TENNIS SERVES.......................................................................... 109

5.1 INTRODUCTION................................................................................................ 109 5.2 METHODOLOGY ................................................................................................ 111

5.2.1 Subject Preparation and Performance........................................................... 111 5.2.2 Data Treatment and Statistical Analysis ........................................................ 112

5.3 RESULTS .......................................................................................................... 113 5.3.1 Effect of Serve Type on Absolute and Planar Racquet Velocity ........................ 113 5.3.2 Body Kinematics that Characterise Serve Performance ................................... 114 5.3.3 Shoulder Joint Kinetics that Characterise the FS and KS ................................. 118 5.3.4 Pre-impact Shoulder Joint Loading as a Predictor of Serve Velocity ................. 122

5.4 DISCUSSION..................................................................................................... 123 5.4.1 Effect of Serve Type on the 3D Profile of Racquet Velocity ............................. 123 5.4.2 Variation in Body Kinematics in the FS and KS............................................... 125 5.4.3 Relationship between Serve Type and Shoulder Joint Kinetics: Implications for

Injury and Performance .............................................................................. 130 5.4.3.1 Cocking............................................................................................... 130 5.4.3.2 Forwardswing...................................................................................... 131 5.4.3.3 Follow-through .................................................................................... 133

5.4.4 Contribution of Pre-impact Shoulder Joint Kinetics to Serve Velocity................ 134 5.5 CONCLUSION.................................................................................................... 135

CHAPTER 6: RELATIONSHIP BETWEEN LOWER LIMB COORDINATION AND SHOULDER JOINT KINETICS IN THE TENNIS SERVE........................................... 136



6.3 RESULTS .......................................................................................................... 140 6.3.1 Serve Technique as a Function of Lower Limb Kinematics .............................. 140 6.3.2 Effect of Variable Foot Placement and Lower Limb Drive on 3D Racquet

Velocity ..................................................................................................... 144 6.3.3 Body Kinematics that Characterise Serve Performance ................................... 145 6.3.4 Shoulder Joint Kinetics that Characterise the Performance of the FU, FB and

ARM Serves................................................................................................ 147 6.3.5 Pre-impact Shoulder Joint Loading Predicted by Lower Limb Joint Kinematics .. 153

6.4 DISCUSSION..................................................................................................... 154 6.4.1 Lower Limb Kinematics that predict Serve Technique..................................... 154 6.4.2 Relationship between Variable Foot Placement and Lower Limb Drive on 3D

Racquet Velocity......................................................................................... 156 6.4.3 Variation in Body Kinematics in the FU, FB and ARM Serves ........................... 157 6.4.4 Effect of Variable Lower-Extremity Joint Kinematics on Shoulder Joint

Loading: Implications for Injury and Performance. ........................................ 159 6.4.4.1 Cocking............................................................................................... 160 6.4.4.2 Forwardswing...................................................................................... 161 6.4.4.3 Follow-through .................................................................................... 163

6.4.5 Lower Limb Kinematics that Predict the Average Rate of Pre-impact Maximum Compressive Force Loading ......................................................................... 164

6.5 CONCLUSION.................................................................................................... 165

CHAPTER 7: SHOULDER JOINT KINETICS OF THE ELITE WHEELCHAIR TENNIS SERVE – A CASE STUDY........................................................................................ 167



7.3 RESULTS .......................................................................................................... 170

ix

7.3.1 Effect of Wheelchair Serve Type on 3D Racquet Velocity................................ 170 7.3.2 Upper-Extremity Kinematics that Describe the Wheelchair FS and KS .............. 172 7.3.3 Shoulder Joint Kinetics that Characterise the Wheelchair FS and KS ................ 175

7.4 DISCUSSION..................................................................................................... 178 7.4.1 Differential 3D Racquet Velocity in the WFS and WKS, and in Comparison with

the Velocity Developed During the Able-Bodied Serve .................................... 179 7.4.2 Variation Between the Upper-Extremity Joint Kinematics of the WFS and WKS,

and in contrast to the Able-Bodied Serves..................................................... 181 7.4.3 Effect of Serve Type on Shoulder Joint Loading in Wheelchair Players:

Implications for Injury and Performance ....................................................... 184 7.4.3.1 Cocking............................................................................................... 185 7.4.3.2 Forwardswing...................................................................................... 185 7.4.3.3 Follow-through .................................................................................... 186

7.5 CONCLUSION.................................................................................................... 189

CHAPTER 8: SUMMARY AND CONCLUSIONS........................................................ 190 8.1 SUMMARY ........................................................................................................ 190 8.2 CONCLUSIONS.................................................................................................. 193 8.3 RECOMMENDATIONS FOR FUTURE RESEARCH.................................................... 195

8.3.1 Subject Preparation and Data Analysis.......................................................... 195 8.3.2 Continued Investigation of the Relationship between Shoulder Joint Loading

and Tennis Player Performance.................................................................... 195 8.3.3 Quantification of Models of Optimal Serve Performance ................................. 196

REFERENCES......................................................................................................... 197

x

LIST OF TABLES

Table 2.1. Distribution of injuries per body part. ........................................................ 23

Table 2.2. Sites of upper-extremity injury.................................................................. 23

Table 2.3. Pathology of tennis injury......................................................................... 23

Table 3.1. Upper- and lower-body retroreflective marker naming convention and locations. ................................................................................................ 72

Table 3.2. Sequence and direction of rotations comprising joint coordinate systems. .... 80

Table 4.1. A summary of the mean most appropriate MSEs for selected kinematic variables in the three FS and three KS, as performed by two professional players.................................................................................................... 91

Table 4.2. A summary of the mean most appropriate MSEs for selected kinetic variables in the three FS and three KS, as performed by two professional players.................................................................................................... 91

Table 4.3. Coefficients of Multiple Correlations (CMC) of selected kinematic and kinetic variables during the forwardswing of the FS and KS.................................... 99

Table 4.4. Coefficients of Multiple Correlations of selected shoulder kinetic variables post-impact. ............................................................................................ 99

Table 5.1. Comparison of linear racquet kinematics across FS and KS (* p<0.01). ...... 113

Table 5.2. Upper and lower body kinematics that characterise the FS and KS (* p<0.01), and that are reported to relate to shoulder joint loading in the serve. ................................................................................................... 115

Table 5.3. Comparison of shoulder joint kinetics considered to represent shoulder joint load across FS and KS (n = 12; * p<0.01). ....................................... 119

Table 5.4. Results of step-wise regression on pre-impact loading variables and absolute racquet velocity in the FS (* p<0.05). ........................................ 123

Table 5.5. Results of step-wise regression on pre-impact loading variables and absolute racquet velocity in the KS (* p<0.05). ........................................ 123

Table 6.1. Variables included in the stepwise discriminant analysis procedure (* p <0.001). ........................................................................................ 141

Table 6.2. Canonical discriminant functions extracted from the analysis..................... 141

Table 6.3. Classification of serve type based on discriminant functions. ..................... 141

Table 6.4. Excluded from the stepwise discriminant analysis, descriptive statistics of peak front knee joint flexion for the FU, FB and ARM serves. ..................... 142

Table 6.5. Comparison of linear racquet kinematics across FU, FB and ARM serves (* p<0.01). ........................................................................................... 145

Table 6.6. Upper and lower body kinematics that characterise and may relate to shoulder joint loading in the FU, FB and ARM serve (* p<0.01).................. 147

xi

Table 6.7. Shoulder joint kinetics for the FU, FB and ARM serves (* p<0.01). ............ 148

Table 6.8. Results of step-wise regression analyses on selected lower limb kinematics and the average rate of maximum compressive force loading in the swing phase of the FB and ARM serves (* p<0.05). ........................................... 154

Table 7.1. Comparison of mean linear racquet kinematics between the WFS (n=3) and WKS (n=3) as performed by two subjects (S1 and S2), as well as in contrast to the mean able-bodied FS and KS (n=12). ................................ 170

Table 7.2. Mean shoulder joint kinematics, related to shoulder joint loading, that characterise the WFS and WKS, and as compared with the able-bodied FS and KS. ................................................................................................. 173

Table 7.3. Mean 3D motion of shoulder alignment during the WFS and WKS, as well as in comparison with the able-bodied FS and KS. .................................... 175

Table 7.4. Mean shoulder joint kinetics that punctuate WFS and WKS performance, and as compared with the able-bodied FS and KS..................................... 177

Table 7.5. Mean shoulder joint kinetics of the wheelchair and able-bodied serves expressed relative to maximum pre-impact absolute racquet velocity. ........ 178

Table A.1. Mean (± SD) shoulder joint kinematic and kinetic data modelled with the original and modified upper arm triads for the FS and the KS...................... xxii

xii

LIST OF FIGURES

Figure 1.1. Two-dimensional (2D) functional representation of the knee joint flexion-extension angle. ...................................................................................... 12

Figure 1.2. 2D functional representation of separation angle....................................... 13

Figure 1.3. 2D functional representation of lateral flexion separation angle. ................. 13

Figure 1.4. 2D functional representation of shoulder alignment forward flexion angle. .. 14

Figure 1.5. 2D functional representation of shoulder alignment lateral flexion angle...... 14

Figure 1.6. 2D functional representation of shoulder alignment rotation angle.............. 15

Figure 2.1. Articulations that comprise the shoulder girdle (from Whiting and Zernicke (1998, p178). .......................................................................................... 28

Figure 2.2. Architecture of rotator cuff muscle group from anterior (left) and posterior (right) aspects (reprinted with permission; Pluim, 2001).............................. 29

Figure 3.1. Frontal (left) and sagittal (right) illustration of the magnitude of shoulder joint plane of elevation (PoE), elevation (E) and internal-external rotation (IRER) describing four different attitudes of the upper arm. A and B – PoE: -190º, E: 90º, IRER: -85º; C and D - PoE: -140º, E: 150º, IRER: -70º; E and F – PoE: -195º, E: 30º, IRER: -45º; G and H – PoE: -120º, E: 90º, IRER: -50º . ........................................................................................... 58

Figure 3.2. Illustration of the a) mathematical effect of gimbal-lock as upper arm abduction reaches 90º, and b) how Bodybuilder uses its ‘internal record of rotations’ to avoid gimbal-lock................................................................... 61

Figure 3.3. Shoulder joint internal (+) and external (-) rotation, during the swing phase of FSs (n=10) using the Euler Z-X-Y convention. ........................................ 62

Figure 3.4a. Mean shoulder joint internal (+) and external (-) rotation during the swing phase of a FS (n=10), using the Euler Z-X-Y convention (error bars represent sample standard deviation). ....................................................... 62

Figure 3.4b. Mean shoulder joint flexion (+) and extension (-) during the swing phase of a FS (n=10), using the Euler Z-X-Y convention (error bars represent sample standard deviation)....................................................................... 63

Figure 3.5a. Shoulder joint longitudinal rotation, during the swing phase of a FS (n=10), using the ISB convention.............................................................. 64

Figure 3.5b. Mean shoulder joint longitudinal rotation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation)................................................................................................ 64

Figure 3.6. Shoulder joint plane of elevation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation). . 65

Figure 3.7. Shoulder joint elevation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation)................. 65

xiii

Figure 3.8. Marker positions of able-bodied players during static trials (anterior view, left; posterior view; right; refer to Table 3.1 for marker key). ...................... 69

Figure 3.9. Design of the semi-malleable ‘T-bar’ cluster.............................................. 69

Figure 3.10. Design of the aluminium ‘T-bar’ cluster................................................... 69

Figure 3.11. Marker positions of wheelchair players during the dynamic trials (anterior view; refer to Table 3.1 for marker key)..................................................... 73

Figure 3.12. Diagrammatic representation of the spherical pointer used to locate humeral and femoral epicondyles (Op represents the origin of the three independent coordinate systems calculated using markers m1-m5). ............. 74

Figure 3.13. Placement (medial to the axillary folds) of anterior and posterior shoulder markers required for estimation of the GH joint centre. ............................... 75

Figure 3.14. Customised foot rig used to define the foot segment. .............................. 79

Figure 3.15. Schematic of the testing environment (left) and camera positions (right). . 85

Figure 4.1. Example of the residual analysis performed to ascertain the most appropriate MSE for horizontal racquet velocity. ......................................... 92

Figure 4.2. Example of the residual analysis performed to ascertain the most appropriate MSE for the right shoulder internal-external rotation moment..... 92

Figure 4.3. Effect of different MSEs on the upper arm - thorax elevation angle during the forwardswing of a professional player’s FS............................................ 93

Figure 4.4. Effect of different MSEs on the shoulder joint internal rotation moment generated during the forwardswing of a professional player’s FS.................. 94

Figure 4.5. Effect of data treatment and smoothing procedures on the internal-external rotation angle of the racquet arm’s shoulder joint (as calculated by the Z-X-Y Euler decomposition) during the swing phase of a professional player’s FS............................................................................................... 96

Figure 4.6. Effect of data treatment and smoothing procedures on the internal-external rotation moment of the racquet arm’s shoulder joint during the swing phase of a professional player’s FS...................................................................... 97

Figure 4.7. Comparison of the shoulder joint internal rotation moment during the forwardswing of a professional player’s highest quality FS, mean three highest quality FSs, and mean five highest quality FSs. ............................. 100

Figure 5.1. Comparison of mean 3D linear racquet velocities during the forwardswing of the FS and KS. ................................................................................... 114

Figure 5.2. Comparison of the mean internal rotation of the upper arm during the forwardswing of the FS and KS. .............................................................. 116

Figure 5.3. Contrast in the 3D alignment of the shoulders at impact in the FS (left) and KS (right)........................................................................................ 117

Figure 5.4. Comparison of the mean front knee joint extension during the lead leg drive phase of the FS and KS. ................................................................. 118

xiv

Figure 5.5. Angular external rotation velocity (within the thorax) and internal rotation moment about the long axis of the upper arm during the forwardswing of a FS and a KS. ....................................................................................... 120

Figure 5.6. Shoulder joint internal-external rotation power term during the forwardswing of a FS and a KS................................................................ 120

Figure 5.7. Angular internal (+) and external (-) rotation velocity as well as internal (-) and external (+) moment about the long axis of the upper arm during the follow-through of a FS and a KS. ............................................................. 121

Figure 5.8. Shoulder joint internal-external rotation power term during the follow- through of a FS and a KS........................................................................ 122

Figure 6.1. Mean extension of the front knee during the lead leg drive phase of the FU, FB and ARM serves. ......................................................................... 143

Figure 6.2. Mean extension of the rear knee during the rear leg drive phase of the FU, FB and ARM serves. ......................................................................... 143

Figure 6.3. Mean angular velocity of rear knee extension during the rear leg drive phase of the FU, FB and ARM serves. ...................................................... 144

Figure 6.4. Comparison of mean absolute racquet velocity during the forwardswing of the FU, FB and ARM serves..................................................................... 145

Figure 6.5. Mean internal rotation of the upper arm during the forwardswing of the FU, FB and ARM serve. ........................................................................... 146

Figure 6.6. Representative external (+) and internal (-) shoulder joint rotation moments during the cocking of the FU, FB and ARM serves as performed by a high performance player.................................................................. 149

Figure 6.7. Representative external rotation (expressed in the thorax) angular velocity and internal rotation moment about the long axis of the upper arm during the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player......................................................................... 150

Figure 6.8. Representative internal rotation moments about the long axis of the upper arm during the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player.................................................. 150

Figure 6.9. Representative shoulder joint internal-external rotation power term during the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player......................................................................... 151

Figure 6.10. Representative angular internal (+) and external (-) rotation velocities as well as internal (-) and external (+) moments about the long axis of the upper arm during the follow-through of a FU, FB and ARM serves as performed by a high performance player.................................................. 152

Figure 6.11. Representative shoulder joint internal-external rotation power terms during the follow-through of a FU, FB and ARM serves as performed by a high performance player......................................................................... 152

Figure 7.1. Comparison of mean absolute horizontal racquet velocity during the forwardswing of the WFS and WKS, as performed by both subjects. .......... 171

xv

Figure 7.2. Comparison of mean absolute lateral racquet velocity during the forwardswing of the WFS and WKS, as performed by both subjects. .......... 171

Figure 7.3. Mean internal rotation of the upper arm during the forwardswing of the WFS and able-bodied FS......................................................................... 173

Figure 7.4. Mean internal rotation of the upper arm during the forwardswing of the WKS and able-bodied KS. ....................................................................... 174

Figure A.1. Position of the original and modified upper arm triads. ...............................xx

Figure A.2. The effect of divergent upper arm triad placement on shoulder joint internal (+) and external (-) rotation angular velocities (as expressed in the thorax) during the forwardswing of the FS and KS. ................................... xxiii

xvi

CHAPTER 1: THE PROBLEM

1.1 INTRODUCTION

Professional tennis players travel extensively year round, with tournaments on the

professional men’s and women’s calendars numbering 325 and 614 respectively in 2003

(International Tennis Federation (ITF), 2003). The Tours’ players and health care

providers have voiced their concern at the growing incidence of injury (Clarey, 2005). In

the male game, 66 players cited injury as reason for their withdrawal from tournaments

in 1994 (Association of Tennis Professionals (ATP), 2003). Eight years on, and that

figure had more than doubled with injuries preventing players from participating in

tournaments on 133 different occasions (ATP, 2003).

In most athletic endeavours, high-speed performance is a desired outcome. Whether it is

crossing the finish line first or hitting the ball the ‘hardest’, high forces are developed to

rotate segments in a fashion commensurate with this outcome. These segment rotations

need to be coordinated; with associated tissues generating and absorbing considerable

load. Whiting and Zernicke (1998) identify load as a product of the magnitude, location,

direction, duration, frequency, variability and rate of force application. In sporting

parlance, this load can be considered internal: a product of the muscle activity required

for skill execution, or external, as applied by an external force such as the ground to the

feet or gravity to a segment. In tennis stroke production, loading can therefore be

operationally defined as a combination of mean and peak joint kinetics. In this way, joint

loading is thought to correlate positively with velocity generation (Elliott et al., 2003), so

when movement speed is at a premium, loading is likely to be high.

The muscle activity that provides for force and torque development can be concentric,

isometric or eccentric. In sport, concentric muscle activity is typically associated with

force generation, while isometric contractions contribute predominantly to joint stability.

Eccentric work is more often linked to muscle fibre pre-stretch or stretch under tension,

during the deceleration of fast-moving segments. Contracting muscles in this way

spawns more tension than isometric or concentric contractions (Lieber et al., 1991), and

has been linked to exercise-induced muscle damage (Evans et al., 1985) and

1

compromised muscle activation post-exercise (McCully and Faulkner, 1985; Lieber and

Friden, 1988). Most researchers also agree that high intensity eccentric muscle

contraction is a mechanism linked to tissue injury (Evans et al., 1985; Friden and Lieber,

1992). It then follows that this form of loading represents a serious threat to athletes

that need to repeatedly accelerate and decelerate body segments at high speed.

In tennis, high loads are generated repetitively and high levels of eccentric muscle

activity are common to selected strokes. Repetitive tissue stress that overwhelms a

player’s ability for tissue repair (Herring and Nilson, 1987), is therefore a real injury

concern. Peak loading and the rate of loading also have implications for both injury and

performance. It is therefore of little surprise that most tennis conditioning programs are

tailored to harness and manage load in terms of volume, magnitude and rate

(Verstegen, 2003).

Loading of the upper extremity in dynamic motion is the subject of extensive research in

some athletic populations. Several efforts have been made to quantify the kinetics of

throwing and baseball pitching; two skills that involve a similar upper arm motion to the

tennis serve. Indeed, shoulder and elbow injuries associated with baseball pitching are

well documented (Pappas and Zawacki, 1991), and thought to be related to the

repetitive, high forces generated as player’s position, accelerate and decelerate

segments of the upper limb (Atwater, 1979; McLeod and Andrews, 1986; Fleisig et al.,

1995). For example, during the pitch’s cocking phase, Fleisig et al. (1995) reported

maximum shoulder internal rotation torques of 67±11Nm and anterior shear forces in

excess of 450N. When coupled with the high maximum compressive (1090±110N) and

posterior shear (400±90N) forces experienced at the shoulder during arm deceleration

and follow through (Fleisig et al., 1995), the shoulder’s susceptibility to injury becomes

appreciable. Elbow varus torques as high as 65Nm have been recorded during arm

cocking and are a proposed mechanism of injury at this joint (Fleisig et al., 1996).

An epidemiological overview of injury in professional tennis sees shoulder injuries rank

among the game’s most prevalent (ATP, 2003; Reque, personal communication, 2005).

Like baseball pitchers throwing a fast ball, tennis players wishing to hit their serves at

2

high speed need to generate large shoulder joint torques. Over the period 1994-2002,

75 male professionals were forced to withdraw from tournaments as a result of injuries

sustained to their shoulders (ATP, 2003). Trainers employed by the men’s professional

tour go further to suggest that as many as 25% of singles players complain of shoulder

pain or discomfort each week (Reque, personal communication, 2005). Sports coaching

and rehabilitation literature has postulated a series of factors that may predispose

players to shoulder injury (Kibler, 1995; Ellenbecker and Roetert, 2003). Improper

mechanics, high speed stroke performance, deficits in strength, overuse, inflexibility and

muscle imbalance are commonly cited.

Recently, greater loading conditions at the shoulder joint during the serve were

hypothesised to increase the potential for injury at this joint (Elliott et al., 2003).

Suggestions also abound that shoulder joint loading is strongly associated with the type

of service technique used (Elliott et al., 2003). However, researchers are yet to quantify

the effects of serve type or the use of selected service techniques on the loading profile

at the shoulder. Such information would be of clear benefit from injury prevention and

player development standpoints.

The complexity of a movement skill, where it lies along the continuum of closed to open

motor performance, as well as the technical aptitude of the performer, influence the

(athlete’s) reproduction and (coach’s) analysis of skilled performance (Knudson and

Morrison, 2002). Elite performers are able to reproduce movement skills with greater

precision than novice performers (Arutyunyan et al., 1968). In studying the repeatability

of the kinematic and kinetic characteristics of normal adult gait, Kadaba et al. (1989)

concluded that it was reasonable to base clinical decisions on a single gait evaluation.

Joint angle motion and kinetic patterning in selected planes were found to be quite

repeatable when subjects walked at their normal or preferred speed (Kadaba et al.,

1989). Research evaluating the intra-subject variability of upper extremity angular

kinematics in the tennis forehand demonstrated that college level players reproduced

consistent wrist and elbow positions (coefficient of variation, CV < 5.9%) but with highly

variable wrist and elbow joint angular velocities (CV= 90.6%) and accelerations (CV=

129.5%) (Knudson, 1990). Other researchers have also questioned the validity of

3

assuming one superior performance to be representative of an athlete’s overall

technique (Bates et al., 1992; Mullineaux et al., 2001). In tennis, the kinetic and

kinematic repeatability of the serve has not been explored. In light of the noted

methodological concerns, quantification of the stroke’s mechanical consistency, when

performed by high performance players, appears necessary before presenting any

implications for performance.

Similarly, valid biomechanical analysis of the service motion necessitates the selection of

appropriate data treatment techniques. More specifically, determination of an optimal

filtering or smoothing procedure is needed to best represent the higher-order kinematics

that describe the service motion, particularly near racquet-ball impact (Vint and Hinrichs,

1996; Knudson and Bahamonde, 2001). While various approaches have been used to

assist the mechanical analysis of tennis groundstrokes (Knudson and Bahamonde, 2001;

Reid and Elliott, 2002; Knudson, 2005), their appropriateness in describing the three-

dimensional (3D) kinematics and kinetics of the tennis serve requires scrutiny.

1.2 STATEMENT OF THE PROBLEM

Tennis stroke production is characterised by the development of high forces and torques

at the shoulder joint. Injuries to the shoulder joint, or its associated structures, are

among the most prevalent sustained by professional tennis players. Several authors

associate the loads generated and absorbed by the tissues of the shoulder during the

serve to the risk of shoulder injury (Chandler et al., 1992; McCann and Bigliani, 1994;

Kibler, 1995). Certain serve techniques have also been suggested to load the shoulder to

variable extents (Elliott et al., 2003). So, elucidation of the effects of serve technique

and type on loading at the shoulder joint has clear application for the development of

specific coaching, injury prevention and rehabilitation strategies.

Therefore, the aims of this thesis are to:

• determine the repeatability of kinematic and kinetic data in the high performance

tennis serve, so that the number of trials required for representative data can be

established (study 1, Chapter 4);

4

• establish an optimal smoothing routine, inclusive of the most appropriate level of

smoothing (mean squared error, MSE) and treatment of racquet-ball impact, to

best represent serve kinematics and kinetics (study 1, Chapter 4);

• investigate the effect of high performance serve type (flat serve, FS, versus kick

serve, KS) on loading of the shoulder joint (study 2, Chapter 5);

• examine the relationship between variation in lower limb involvement in the high

performance FS (foot-up (FU), foot-back (FU) and no leg drive (ARM)) and

shoulder joint load (study 3, Chapter 6);

• explore the effect of serving with no leg drive (i.e. from a wheelchair) on the

shoulder joint kinetics of two high performance wheelchair players (study 4;

Chapter 7).

1.3 JUSTIFICATION OF THE STUDY

The shoulder is a key joint in tennis stroke production and is commonly implicated in

tennis injury. The joint has been described as a ‘funnel’ that facilitates the generation,

summation, transfer and regulation of forces from the legs to the hand (Kibler, 1995).

Tennis strokes, and in particular the serve, require that high forces be generated

through large ranges of motion at the shoulder. Fleisig et al. (2003) has reported that

professional players move through 90° of upper arm internal rotation during the service

forwardswing (i.e. from upper arm maximum external rotation (MER) to impact) and in

doing so record upper arm rotational velocities as high as 3000°.s-1. The development of

these forces and subsequent rotational velocities during a match or throughout a playing

career coupled with inadequate muscle strength, poor motor control or the use of a

specific technique, may increase a player’s susceptibility to tissue injury (McCann and

Bigliani, 1994), either instantaneously or accumulatively.

On the professional tour, a regulation singles match will see players hit between 50-150

FS and KS. When multiplied by 60 (target number of competitive singles matches per

year) and serving during doubles matchplay and practice is added, one begins to

appreciate the need for good shoulder joint function, as well as sound serve technique

(Elliott, 2001). While there are several mechanical characteristics common to the world’s

best servers, variation in others can be readily observed. The arrangement of the feet

5

and the involvement of the lower limbs are two such characteristics. Variations in serve

technique are thought to load the shoulder joint differently and therefore have some

implications for injury (Elliott et al., 2003). Kinetic analyses of the serve that establish

relationships between serve technique and type and loading at the shoulder would

therefore be invaluable from both performance and injury prevention perspectives.

1.4 HYPOTHESES

The hypotheses for study 1 are:

1. Quintic splines using a MSE similar to that used to represent the true signal of

other high-speed sports motions (i.e. cricket bowling MSE 23) will be suitable for

the analysis of the tennis serve;

2. Smoothing through impact in the high performance serve does not compromise

the representativeness or accuracy of related shoulder kinematic and kinetic pre-

impact data;

3. Three serves are required to gain reliable kinematic and kinetic data on the high

performance FS and KS.


1. Flat serves develop higher peak pre-impact horizontal and vertical racquet

velocities, while KSs generate higher maximum pre-impact lateral racquet

velocities;

2. The FSs are characterised by larger peak, shoulder joint anterior forces and

average rates (to peak) of anterior force loading during the cocking phase than

the KSs;

3. Kick serves are characterised by larger mean shoulder joint compressive forces

during the forwardswing and follow-through than in the FS;

4. Higher average rates of shoulder joint compressive force loading are more

common to the KS than the FS during the swing phase;

5. Higher peak, shoulder joint pre-impact internal rotation moments and post-

impact external rotation moments are experienced in the FS as compared with

the KS;

6

6. Different pre-impact shoulder joint kinetics predict the development of racquet

velocity in the FS and KS.


1. The magnitude of lead and rear knee extension can predict serve technique;

2. Serving with a leg drive (i.e. use of the FU or FB technique) as compared with

minimal leg drive (i.e. the ARM serve) produces higher absolute racquet velocities

during the forwardswing;

3. The ARM serve increases the upper arm peak external rotation moment in

cocking, while serving with a leg drive (i.e. use of the FU or FB technique)

produces larger peak upper arm internal rotation moments during the

forwardswing;

4. Players using the FU service technique generate higher mean pre-impact and

post-impact shoulder joint compressive forces than FB and ARM servers;

5. Higher average rates of maximum compressive force loading during the swing

phase characterise the FU serve as compared with the FB and ARM serve;

6. FB serves are characterised by larger peak, shoulder joint anterior forces and

average rates of peak anterior force loading during the cocking phase than FU

and ARM serves;

7. ARM serves increase the upper arm peak external rotation moment in cocking;

8. More pronounced transverse plane trunk rotation punctuates the forwardswing of

the ARM serve, while larger amounts of lateral flexion are involved in the

forwardswings of the FU and FB techniques;

9. Different lower limb kinematics predict shoulder joint loading between serves.


1. Wheelchair flat serves (WFSs) develop higher peak pre-impact horizontal racquet

velocities, while wheelchair kick serves (WKSs) generate higher maximum pre-

impact lateral racquet velocities;

2. Serving with no leg drive (i.e. from a wheelchair) produces reduced peak

absolute and horizontal pre-impact racquet velocities as compared with serving

with a leg drive (i.e. the able-bodied serves);

7

3. The magnitude of MER of the upper arm is independent of wheelchair serve type

but lower when serving with no leg drive (i.e. the wheelchair serves) as

compared with serving with leg drive or with minimal leg drive (i.e. the able-

bodied serves);

4. As compared with the able-bodied serve, more pronounced transverse plane

trunk rotation characterises the forwardswing of wheelchair serves;

5. Although independent of wheelchair serve type, higher peak upper arm external

rotation moments are generated during the cocking phase of the WFSs and WKSs

than in the able-bodied FS and KS;

6. Higher peak shoulder joint internal rotation moments are experienced in the

forwardswing of the WFS as compared with the WKS;

7. Relative to absolute racquet velocity, higher peak shoulder joint internal rotation

moments, mean compressive forces and average rates of compressive force

loading are produced in the forwardswings of the wheelchair serves as compared

with the able-bodied serves;

8. During the follow-through, peak shoulder joint external rotation moments and

mean compressive forces are generated independent of wheelchair serve type;

9. Relative to absolute racquet velocity, higher post-impact peak shoulder joint

external rotation moments and mean compressive forces are experienced by

players who employ no leg drive (i.e. the wheelchair serves) as compared with

some leg drive (i.e. able-bodied serves).

1.5 LIMITATIONS

1.5.1 Variance in the maturation and anthropometric strength measures of the sample

may complicate the interpretation of results. No attempt was made to

standardise these measures.

1.5.2 The assumption that the sample is representative of high performance tennis

players.

1.5.3 The sample size reflected the number of West Australian male tennis players,

who were considered to have high performance serves.

1.5.4 The lack of similar research in tennis ensured the derivation of some definitions

were introspective.

8

1.5.5 Error introduced as a result of skin movement under the markers was not

quantified, however, the use of cluster marker sets reduced this error.

1.5.6 The assumption that differences in shoulder joint loading can be evaluated by

variance in serve technique.

1.5.7 The testing environment approximated a tennis court.

1.5.8 The FSs and WFSs were required to pass under a rope extended 0.3m above net

height. Kick serves and WKSs were required to pass over a rope extended 0.5m

above the net. All serves were required to land in 1x1 metre target area

bordering the ‘T’ of the first service box.

1.5.9 The inertial properties for the upper and lower limbs reported by De Leva (1996)

were used.

1.5.10 The deduction that high performance players were able to reliably and repeatedly

vary their serve technique.

1.6 DELIMITATIONS

The study was delimited to:

1.6.1 Twelve players, whom were identified by three professional coaches to possess

high performance service techniques.

1.6.2 A marker set developed by the UWA School of Human Movement and Exercise

Science was used (modified version of Lloyd et al., 2000).

1.6.3 All subjects used a Wilson Pro Staff tour 95 (mass: 0.380kg, length: 0.690m).

1.6.4 A tennis court was recreated in the laboratory and surrounding space to improve

ecological validity.

1.7 DEFINITION OF TERMS

Tennis is a sport with many of its own terms. Definitions of those and standardisation

other terms specific to this course of studies are:

• Foot-up: the player’s feet (as represented by calcaneal markers) are separated by

≤ 0.20m at maximum front knee flexion.

• Foot-back: the player’s feet are separated > 0.20m at maximum front knee

flexion.

9

• Leg drive: the magnitude and rate of angular change about the joints of both the

front and rear legs during a player’s propulsive upward and forward movement to

racquet-ball impact in the serve.

• 3D coordinate system: x was in the direction of the serve, y was defined along

the baseline and z was defined as the vertical axis.

• Flat serve: a serve hit with maximal effort – and ‘no’ spin – over a rope extended

0.3m above net height, to a 1x1 metre target area bordering the ‘T’ of the first

service box.

• Kick serve: a serve hit with maximal effort and ‘kick’, over a rope extended 0.5m

above net height, to a 1x1 metre target area bordering the ‘T’ of the first service

box.

• Arm serve: players hit maximal effort flat serves – over a rope extended 0.3m

above net height, to a 1x1 metre target area bordering the ‘T’ of the first service

box – but minimised the extent to which they actively flexed and extended their

ankle, knee and hip joints (i.e. minimised their leg drive).

• Kick: a spin that sees the ball, upon contacting the ground, bounce high and

obliquely away (i.e. to the left of a player returning a right-hander’s serve).

• First service box: otherwise known as the ‘deuce’ side, it is the service box to the

left of the server (when standing in the right half of the court), and to which the

server hits when point scores are even (i.e. 15-15).

• Second service box: otherwise known as the ‘ad’ side, it is the service box to the

right of the server (when standing in the left half of the court), and to which the

server hits when point scores are odd (i.e. 30-15).

• Y-X-Y: an Euler angle decomposition that establishes a plane of elevation, the

amount of elevation, and finally axial rotation, which when interpreted in a globe

system facilitates the description of 3D shoulder joint motion.

• Z-X-Y: the Euler angle decomposition defining the order of rotations of one

moving coordinate system about another moving coordinate system as flexion-

extension, abduction-adduction and internal-external rotation.

• Serve type: pertains to the use of a FS, KS, WFS or WKS.

• Serve technique: refers to technical or mechanical variations typically common to

both FS and KS performance (i.e. the use of the FU, FB or ARM techniques).

10

The following events corresponding to meaningful temporal or kinematic characteristics

of the serve were identified as appropriate in each service trial:

• Racket glide (RG): marked by the initial shifting of the racquet backward and

often accompanied by an analogous transfer of weight.

• Ball toss (BT): moment at which the tossing arm released the tennis ball.

• Maximum lead knee joint flexion (MKF): corresponding to the point in time at

which lead or front knee was maximally flexed.

• Back foot up (BFU): coinciding with the positioning of the back or rear foot

alongside the front foot in FU or ARM serve.

• Racquet high point (RHP): coinciding with the racquet tip reaching a maximum

positive magnitude in the Y direction, prior to MER of the upper arm.

• Maximum external rotation: represented as the time at which the upper (i.e.

racquet) arm reaches a position of MER prior to ball impact.

• Front foot off (FFO): coinciding with the front foot leaving the ground

independent of the rear foot.

• Back foot off (BFO): represented as the rear foot leaving the ground independent

of the front foot.

• Feet off (FO): marked by both feet leaving the ground simultaneously.

• Impact (IMP): defined as 1/250th of a second prior to racquet and ball impact.

• Post-collision (PC): 1/250th of a second post racquet and ball impact.

• Finish (FIN): moment at which the front foot impacted the ground as the body

retreated from being projected upward during the drive to impact (or 0.12s ‘post-

collision’ when the front foot remained in contact with the ground, as when

players hit ARM serves). This interval corresponded to the average time that

elapsed from ‘post-collision’ to ‘finish’ when players served a FS. The same

interval was applied to define ‘finish’ in the wheelchair tennis serve.

The following phases (i.e. between events) were used throughout data analysis:

• Backswing: from RG to MKF (able-bodied serve) or from RG to RHP (wheelchair

serve).

• Rear leg drive: from BFU or MKF to MER, BFO or FO depending on the subject

and/or serve performed.

11

• Lead leg drive: from MKF to MER or FO depending on the subject and/or serve

performed.

• Swing: from MKF to IMP.

• Cocking: from MKF to MER in the able-bodied serve, and from RHP to MER in the

wheelchair serve.

• Forwardswing: from MER to IMP.

• Follow-through: from PC to FIN.

References made to pre-impact kinematics and/or kinetics will relate to all pre-impact

motion. Terms such as deceleration and post-impact will be used to discuss

characteristics of the follow-through.

Definitions of 3D shoulder joint angles will be provided in Chapter 3. The following

absolute and relative angles were also examined in studies 1-4, and thus warrant

explanation:

• Knee joint flexion-extension: relative excluded angle about z axes of the lower

leg and thigh, where flexion is positive (Figure 1.1).

Figure 1.1. Two-dimensional (2D) functional representation of the knee joint flexion-extension angle.

12

• Separation angle: relative included angle between the y axes of the pelvis and

shoulders, where rotation of the right shoulder beyond the pelvis (i.e. left

shoulder is leading) is negative (Figure 1.2).

Figure 1.2. 2D functional representation of separation angle.

• Lateral flexion separation angle: relative included angle between the x axes of

the pelvis and shoulders, where flexion of the trunk to the right is positive (Figure

1.3).

Figure 1.3. 2D functional representation of lateral flexion separation angle.

13

• Shoulder alignment forward flexion angle: absolute angle between the z axes of

the shoulder alignment and global coordinate system, where forward flexion of

the shoulder alignment is positive (Figure 1.4).

Figure 1.4. 2D functional representation of shoulder alignment forward flexion angle.

• Shoulder alignment lateral flexion angle: absolute angle between the x axes of

the shoulder alignment and global coordinate system, where lateral flexion of the

shoulder alignment to right (i.e. right shoulder down) is positive (Figure 1.5).

Figure 1.5. 2D functional representation of shoulder alignment lateral flexion angle.

14

• Shoulder alignment rotation: absolute angle between the y axes of the shoulder

alignment and global coordinate system, where rotation of the shoulder

alignment to the left is positive (Figure 1.6).

Figure 1.6. 2D functional representation of shoulder alignment rotation angle.

15

CHAPTER 2: LITERATURE REVIEW

2.1 INTRODUCTION

The aetiology of musculoskeletal injuries generally has a kinetic mechanical cause

(Whiting and Zernicke, 1998). Applied to the world of professional tennis, this means

that the kinetic loads that characterise stroke production and movement are likely

associated with injuries sustained by the game’s combatants. As the magnitude of

loading has been purported to increase with movement velocity (Elliott et al., 2003), elite

tennis players for whom high movement speed is a priority, may be subject to elevated

injury risk.

In sport, loading of a tissue can occur accumulatively or instantaneously. Accumulative

loading is a feature of tennis play and has been linked to the overuse injuries sustained

by players (Kibler et al., 1992; Kibler et al., 1996). Overuse relates to repetitive loading

of a tissue, which ultimately exceeds that tissue’s ability to adapt. Tissues can also be

damaged instantaneously, where load at any one point in dynamic motion stresses them

injuriously. Faulty or inefficient stroke or movement technique that intensifies joint

loading may perpetuate a player’s potential for injury (Kibler and Safran, 2000).

The serve, along with the forehand and backhand, form the nucleus of tennis stroke

production. Advancements in technology and physical preparation have seen high-

velocity groundstrokes feature more prominently in the repertoires of modern day

players, yet the serve remains the most explosive and important stroke in the game. It

follows that joint loading is considered to be most pronounced in the serve, and that its

repetitive performance, along with other similar overhead ballistic motion, is commonly

implicated in upper extremity joint injury. Indeed injuries or pain at the shoulder rank

among the most common upper extremity complaints and plague approximately 25% of

male professional players during any tournament week (Reque, personal communication,

2005). One of the foci of player development is thus to foster service actions that

provide for high racquet speeds and variation in ball placement, while minimising the

potential for injury.

16

Several mechanical characteristics are common to good servers, yet variation exists

among the service techniques used. Elliott and Wood (1983), for example, were the first

to investigate the differing ground reaction force profiles of players who serve with their

feet positioned together (FU) or apart (FB). Subsequent studies have examined the

effect of these feet arrangements on performance measures such as service velocity,

and across playing standard (Bartlett et al., 1994; Bahamonde and Knudson, 2001;

Girard et al., 2005). Service motions can also be observed to employ more or less leg

drive, and a full or abbreviated backswing (Elliott et al., 2003). With wheelchair tennis

becoming recently and increasingly popular, the mechanical variation of the wheelchair

tennis serve is also worthy of consideration, particularly as it is a serve which derives

minimal if any benefit from players’ lower limbs.

Numerous studies have described the serve from kinematic perspectives (Elliott and

Wood, 1983; Elliott et al., 1986; Elliott et al., 1995). Kinetic interpretations of the stroke

however, are few in number and fewer still, if upper extremity, and more particularly

shoulder joint loading is the area of concern. Specific research efforts to substantiate

how serve technique and type affect the development of racquet speed but more

specifically shoulder joint loading is keenly anticipated. Correspondingly, kinetic

comparisons of different serve motions may assist coaches develop more efficacious

performance models.

The fact that biomechanical analyses of the tennis serve have generally assumed one

superior serve performance to be representative of a player’s overall service motion is of

interest. The incongruities of describing movement in this way are well documented

(Bates et al., 1992; Mullineaux et al., 2001), and necessitate that the repeatability of

sports skills is quantified before representative and valid conclusions can be drawn. In

light of this, the kinematic consistency of the forehand drive executed by collegiate level

players has been evaluated (Knudson, 1990; Knudson and Blackwell, 2005). Players

were shown to reproduce consistent wrist and elbow positions at impact (coefficient of

variation, CV < 5.9%) but with highly variable higher-order wrist and elbow joint angular

kinematics (CV > 90%) during the forwardswing (Knudson, 1990). As many as ten trials

were reported as necessary to reliably describe racquet kinematics near impact

17

(Knudson, 2005). Collectively, these studies highlight the need to corroborate tennis

stroke repeatability, even when performed by highly skilled players. It then follows that

substantiation of the kinetic and kinematic repeatability of the high performance tennis

serve should to be pursued, prior to extrapolating results to player development.

With this backdrop in mind, the following review will examine the role of loading in

musculoskeletal injury and serve performance, before evaluating the repeatability

literature with reference to the serve. Further emphasis shall be placed on establishing

potential links between serve technique and shoulder joint loading.

2.2 LOADING AND MUSCLE INJURY

Mechanisms of injury are many and varied. From a sports medicine perspective however,

Leadbetter (1994) has described seven basic mechanisms of injury: 1) dynamic

overload, 2) overuse, 3) structural vulnerability, 4) inflexibility, 5) muscle imbalance, 6)

rapid growth and 7) contact or impact. Some of these mechanisms relate directly to

mechanical loading, with dynamic overload and overuse perhaps the two most

commonly implicated in the sport of tennis.

In a sports context, loading can be internal, as generated by normal muscle use, or

external, as applied by an external force. In tennis, internal loading is then the product

of the tissue response that provides for stroke production and court movement. The

interactions of the feet with the ground and the hand with the racquet are examples of

the game’s external mechanical loads.

Normal physical activity sees the body’s tissues continuously experience loads with no

obvious injury. These “typical” loads are reported to be within a physiological range,

whereafter the likelihood of injury increases. Put simply, when loads exceed the

physiological range, tissues experience loads that can be injurious. Tissue injury can

therefore occur instantaneously, when a single loading episode exceeds a tissue’s

maximum tolerance, or accumulatively, courtesy of repeated force application with

insufficient recovery time. Although acute and chronic tissue injuries, as they are

otherwise and respectively known, are typically distinguishable, they can also be related.

18

For example, chronic injury stems from repetitive episodes of microtrauma overwhelming

the body’s ability for tissue repair (Herring and Nilson, 1987), yet the chronic-to-acute

injury cycle of repetitive loading (overuse) weakening tissue to increase the likelihood of

acute injury is similarly common (Whiting and Zernicke, 1998).

Elite tennis players hit strokes at high speed up to 450 times per match. This repetitive

high-speed stroking, coupled with the added leverage provided by the racquet to

magnify associated loads, is purported to contribute to upper extremity injury in tennis.

Lower limb injuries are also frequent. Indeed the frequency and pattern of tennis injury

may vary with the age, level of play and conditioning, and training status of the

population (Cassell and McGrath, 1999). These variables aside, researchers do however

agree that muscle sprains are the most frequent acute tennis injury, while most chronic

tissue injuries, such as some muscle strains, largely relate to overuse (Hutchinson et al.,

1995; Kibler et al., 1988; Winge et al., 1989).

Load magnitude and repetition are noted to contribute to musculoskeletal injury

(Stauber, 2004). More particularly, the high forces associated with eccentric contractions

are often held to account (Lieber and Friden, 1993). Eccentric contractions, where

muscles lengthen under tension, have been noted to produce more load than when

muscles shorten under tension or contract concentrically (Lieber, 1992). Several

researchers have demonstrated that muscle injury or strain occur during powerful

eccentric muscle contractions (Garrett, 1990; Noonan and Garrett, 1999). In tennis,

these contractions occur frequently and can be most often observed as players activate

selected musculature to decelerate the rotating trunk, upper limb and racquet following

the racquet-ball impact of all strokes. When serving, the muscles, collectively known as

the rotator cuff – responsible for slowing the upper arm as it internally rotates and

horizontally abducts post-impact - are regularly implicated in shoulder injury. An effect,

exacerbated by another variable noted to intensify the prospect of muscle strain or

injury: the high frequency of these eccentric contractions during tennis play (Stauber,

2004).

19

A muscle’s length and fibre type are also associated with its susceptibility to strain, but

to lesser extents. That is, Lieber and Friden (1993) suggested that muscle damage may

not be a function of force during eccentric contractions but rather the magnitude of the

muscle fibre (active) strain. However, by the researchers’ own admission, these results

may not be generalisable to muscles more highly pennated than the investigated tibialis

anterior, whose muscle fibres assume a virtually longitudinal architecture (Lieber and

Friden, 2000). Similarly, the gradual change the researchers elicited in muscle length

did not reflect the dynamic, intermittent muscle contractions that characterise the high-

speed performance of sports skills.

In the workplace, Stauber (2004) recommended reducing the velocity of movements to

attenuate the prospect of muscle strain; an unrealistic, nevermind undesirable

proposition in high performance tennis. Likewise, other than by playing less, or more

aggressively, are tennis players likely to reduce the number of loading episodes they

encounter. So, developing an efficient serve motion, such that loading conditions are

attenuated but high racquet speeds produced becomes critical for the high-performance

player. Similarly important is the need to satisfactorily condition the musculature of the

shoulder, which first generates and then absorbs the ‘brunt’ of the upper extremity’s

high joint loads.

2.3 EPIDEMIOLOGY OF INJURIES IN TENNIS PLAYERS

The transformation of the former amateur game into a multi-million dollar professional

industry has seen many aspiring, as well as seasoned professional players, place

themselves under significant performance and training pressures. Coupled with the

dynamicity of the modern game, the manifestation of these pressures sees some players

demand increasing amounts from their bodies and ultimately subject themselves to

greater injury risk. The recent emergence of the professional wheelchair tennis tours,

and to a lesser extent world championships for veteran players, has also made the game

increasingly attractive to these competitors, who likely have varying levels of physical

conditioning. Although these intrinsic risk factors (age, physical conditioning …) are

unlikely the sole mechanisms of tennis injury, there is little doubt that they have

contributed to the increasing incidence of injury in certain tennis playing populations.

20

Extrinsic or environmental aspects such as the weather, the type and condition of

playing surfaces and tennis equipment are also considered to elevate injury risk (Nigg

and Segesser, 1988; Mohtadi and Poole, 1996; Llana et al., 1998).

2.3.1 Profile of Tennis Injury

Good population-based studies that accurately depict the incidence of injury in tennis

players are few in number (Kibler, 1994; Mohtadi and Poole, 1996). Legal and ethical

issues are reported to limit the information available at the professional level (Martin,

personal communication, 2004), while the retrospective nature of several studies make

the reliability of recall regarding injury cause, site and recovery, tenuous. Exposure to

injury (i.e. injury rate/hours of play) is similarly, poorly documented (Cassell and

McGrath, 1999). Table 2.1 summarises the distribution of upper body, lower body and

core injuries among different groups of players, while Table 2.2 highlights the most

common sites of upper extremity injury.

2.3.1.1 Injury distribution

The assessment of injury patterns in tennis is complicated by disparate research

populations, methodologies and data presentation. Variation in definitions of injury type

and severity further cloud the interpretation of results, while the lack of information on

exposure to tennis injury (i.e. injury rate per 100 hours played) confounds the valid

comparison of reported injury rates. Consequently, Table 2.1 only presents the injury

data published on elite junior and professional players. This information is also specific to

the intended course of studies, where the sample will comprise of high performance

players.

2.3.1.1.1 Junior tennis In elite junior tennis, the lower limbs (50-59%) appear to be the most commonly injured

area of the junior player’s body. Among the 1440 junior male players evaluated by

Hutchinson et al. (1995), the incidence of lower extremity injury was significantly greater

and virtually double that of upper extremity injury (Table 2.1). The thigh, hamstrings

and ankle were the most common anatomical sites of lower limb injury, while players

were also regularly troubled by shoulder and back complaints.

21

22

In a retrospective review of the medical and physiotherapy records of 45 elite, junior

male and female players, Reece et al. (1986) found injury to the lower extremity, and in

particular the ankle, similarly prevalent. The study of Kibler et al. (1988) reinforced this

bias for lower limb injury in junior tennis, but did report the shoulder as the most

commonly injured body site (Table 2.2). The finding that shoulder pain plagues elite

junior players is in agreement with Lehman’s (1988a) revelation that 24% of the 270

junior players he examined had experienced shoulder complaints in the 12 months

preceding their examination.

A link between the injuries players sustain and the predominant surface upon which they

play has been hypothesised (Lehman, 1988b; Bastholt, 2000). So, with the majority of

the players in these three studies from Australia or the United States – two nations

where player development largely occurs on hard courts – comparable injury profiles

may be expected. Intuitively, it may also be assumed that as hard court surfaces

introduce higher peak impact forces (or ground reaction forces) and are characterised by

higher co-efficients of friction than clay courts, the risk of lower extremity injury from

accumulative or instantaneous strain may be increased (Bastholt, 2000; Miller and Cross,

2003).

Population Age Sample Upper Extremity Trunk and/or Spine Lower Extremity Reference

Elite Australian juniors 16-20 45 20 21 59 Reece et al. (1986)

Elite US and International juniors 11-14 97 35 11 54 Kibler et al. (1988)

Elite US juniors 14-17 1440 26 24 50 Hutchinson et al. (1995)

Elite Danish players 14-48 89 45.7 11.0 39.0 Winge et al. (1989)

Competitive English players >10 131 35 20 45 Chard and Lachman (1987)

Top-ranked male professionals* 18-34 1151 29 25 46 ATP (2003)

(* Based on injuries presented to account for tournament withdrawals)

Table 2.1. Distribution of injuries per body part.

Upper Extremity Population Age Sample

Shoulder Arm and/ or elbow Wrist / hand Reference

Elite US and International juniors 11-14 97 21 6 7 Kibler et al. (1988)

Elite US juniors 14-17 1440 33 30 37 Hutchinson et al. (1995)

Elite Danish players 14-48 89 38 38 24 Winge et al. (1989)

Top-ranked male professionals 18-34 1151 38 18 44 ATP (2003)

Table 2.2. Sites of upper-extremity injury.

Study Sample Overuse Strains Sprains Fractures Other

Kibler et al. (1988) 97 elite junior players - 25% 65% 12% 7%

Winge et al. (1989) 61 elite male & 28 elite female players 67% 14% 17% 2% 5%

Hutchinson et al. (1995) 1440 elite junior boys - 13% 59% - 28%

Table 2.3. Pathology of tennis injury.

2.3.1.1.2 Elite adult and professional tennis Legal and ethical issues are noted to restrict the release of information that details the

injuries sustained by professional tennis players (Martin, personal communication, 2004).

However, the recent presentation of injury data at an ATP meeting in December 2003

does go some way to helping the game’s coaches, trainers and medical professionals

better appreciate the epidemiological landscape of professional tennis. Of important note

though, is that the listed injuries and complaints were responsible for the withdrawal of

competitors from ATP tournaments held between 1992 and 2002. This is significant as

ATP players, coaches and trainers acknowledge that while complaints may be legitimate,

there can also be other underlying reasons for tournament withdrawal (Morris, personal

communication, 2006). Nonetheless, indications are that injuries to the lower extremity

are also prevalent in the professional game. That is, with complaints distributed

relatively equally between the hip, thigh, knee, ankle and foot, almost 50% of all

recorded injuries were to the lower extremity, while upper extremity (29%) and trunk

(25%) injuries were responsible for a similar number of tournament withdrawals (Table

2.1).

These results contrast with the findings of Winge et al. (1989), whom provide one of the

few other insights into the epidemiology of injury in elite adult players. In documenting

the incidence of injury among 89 high performance Danish players, some of whom were

professionally ranked, the researchers found injuries to the upper extremity (46%) to

outweigh those sustained to the lower extremity (39%) and trunk/spine (11%) (Winge

et al., 1989). The conflicting patterns of injury reported by the ATP (2003) and Winge et

al. (1989) suggest, contrary to earlier reports (Cassell and McGrath, 1999), that injury

blueprints between elite adult and elite junior players may be similar. Indeed, it could be

argued that higher incidence of upper extremity injury recorded by Winge et al. (1989)

may be related to marked variation in specific intrinsic (age of the sample: 14 to 48

years old) and extrinsic (racquet type: wood, graphite and aluminium) risk factors.

Common to both of these data sets however was the high incidence of shoulder pain

(Table 2.2). That is, injuries to the shoulder accounted for 17% of all injuries and were

the single most frequent complaint recorded by Winge et al. (1989). Shoulder

24

pathologies also comprised 38% of all upper-extremity injuries sustained by professional

players over a 12 year period (ATP, 2003). Indeed, Juan Reque, a physiotherapist on the

men’s tour, suggests that up to one in every four players complain of shoulder pain

during any tournament week, and that every 2-3 years approximately 1/3 of top 100

players miss tournaments due to shoulder problems (Reque, personal communication,

2005). Reque’s assertion is supported by Priest’s (1988) findings that more than 50% of

his sample of 84 high performance players suffered symptoms of shoulder pain at some

point in their playing careers. Shoulder pain or discomfort has been suggested similarly

prevalent in other adult and even wheelchair tennis playing populations (Lehman,

1988a; Pluim and Bullock, 2003).

2.3.2 Types of Injuries

The types of musculoskeletal injuries sustained by athletes provide obvious insights into

the mechanisms of injury, and are to some extent, sport-specific. As aforementioned,

most musculoskeletal injury can also be linked to the magnitude or rate of tissue

loading. With this in mind, the complex stroke and movement demands of tennis give

rise to varied injury histologies. Again, there is limited scientific literature cataloguing the

types of injuries that elite players sustain, and the interpretation of which, is often

complicated by variation in injury classifications. Nevertheless, just as the most common

injury sites are comparable between elite playing populations, Table 2.3 seems to show

similar homogeneity in injury type.

Ligamentous sprains accounted for 59% of the injuries sustained by elite junior

population investigated by Hutchinson and colleagues (1995). Of interest is that almost

nine in every ten sprains occurred at the knee or ankle, highlighting potential differences

in the mechanism of injury between the upper and lower extremities. Muscle strain

comprised 12% of injuries and led the researchers to identify overuse as a primary

factor in selected muscle injury (Hutchinson et al., 1995). Reece et al. (1986) and Kibler

et al. (1988) present divergent aetiologies whereby the earlier study differentiated

between intrinsic and overuse injury, while the latter classified 63% of all injuries as

overuse and 37% as acute. In the only study to have documented types of injuries in an

elite adult tennis playing population, Winge et al. (1989) attributed 67% of all tennis

25

injury to overuse and 14% to muscle strains. Indeed, overuse has been most regularly

implicated in the shoulder joint injuries sustained by tennis players (Kibler et al., 1988;

Winge et al., 1989).

To summarise, it would appear that acute injuries present more commonly in the lower

limbs, while overuse pathologies are frequently linked to injuries of the trunk and upper

extremity.

2.4 SHOULDER INJURIES IN TENNIS PLAYERS

Tennis coaching and rehabilitation literature has postulated a series of factors that may

predispose the tennis player to shoulder injury. Improper mechanics, overuse and

muscle inflexibility, imbalance and / or weakness are commonly cited (Kibler, 1995).

Nevertheless quantification of these proposed aetiologies has not been forthcoming.

Tennis strokes, and in particular the serve, require that high forces be generated

through large ranges of motion at the shoulder. Fleisig et al. (2003) has reported that

professional players move through 90° of upper arm internal rotation during the serve’s

forwardswing and in doing so, record upper arm rotational velocities as high as

3000°.s-1. Kibler (1995) further apportioned 21% of the total force generated during the

stroke to the musculature of the shoulder joint. Both researchers postulated, rather than

quantified, implications for injury and performance. This contrasts with Elliott et al.

(2003) who substantiated a link between serve technique and loading at the shoulder

joint such that the use of specific techniques may increase the potential for injury at the

joint.

This program of research will extend the knowledge concerning the relationship between

serving technique and type, and loads at the shoulder joint so that specific coaching,

injury prevention and rehabilitation strategies may be developed. To best understand

this relationship however, the coach and/or trainer need to develop an appreciation of

the shoulder joint’s architecture, and thus the structures that harbor the most load

during the serve’s execution. An understanding of variables known to magnify loading

26

conditions, and therefore fuel the prospect of players sustaining shoulder injury, is also

necessary.

2.4.1 Anatomy of the Shoulder

Four separate articulations comprise the shoulder girdle. They are the sternoclavicular

joint, the acromioclavicular (AC) joint, the glenohumeral (GH) joint and the

scapulothoracic articulation. Successful performance of the overhead tennis serve motion

requires movement from all four articulations.

The SC joint is a freely moveable synovial joint that links the upper extremity to the

torso. Its function resembles that of a ball-and-socket joint, allowing motion in nearly all

planes, including rotation. The ability to thrust the arm and shoulder forward and

upward, as in the tennis serve, requires sound SC joint function. However, joint injury is

typically sustained through the substantial direct or indirect application of force to its

supporting ligaments, and is not common among tennis players (see 2.3.2.1.5).

The synovial joint between the lateral aspect of the clavicle and the medial surface of

the scapula’s acromion process is the AC joint. A capsule and wedge shaped meniscus

hold the acromion to the clavicle (Bosworth, 1955). The AC joint, the distal end of the

clavicle, the coracoacromial ligament and the anterior acromion comprise the roof of the

shoulder or its subacromial arch. The space between the subacromial arch and the

rotator cuff musculature houses the subacromial bursa, a structure prone to

inflammation or impingement during the repetitive upper arm flexion and internal

rotation movement that characterises the tennis serve (Lehman, 1988a).

The GH joint is a synovial ball and socket joint whose anatomical architecture and

ligamentous support affords inherently poor stability but considerable mobility. A soft-

tissue sheath of fibrocartilage, the glenoid labrum, encases the glenoid rim to enhance

the socket’s articulation and depth. The glenoid’s shape also permits a small amount of

bony contact between it and the humeral head. The capsule of the shoulder attaches to

both the humerus and scapula, and its associated laxity allows a wide range of motion.

Thickened areas of the capsule comprise the superior, middle and inferior GH ligaments.

27

These ligaments are the primary static stabilisers of the GH joint, and are responsible for

anterior stability in different shoulder positions (Townley, 1986; O’Brien et al., 1990;

Terry et al., 1991). For example, when the humerus is externally rotated and in 90° of

abduction, as observed in the tennis serve, the primary static restraint to anterior

instability becomes the inferior GH ligament (Blevins, 1997).

Figure 2.1. Articulations that comprise the shoulder girdle (from Whiting and Zernicke (1998, p178).

The “rotator cuff” muscles are the foremost dynamic stabilisers of the GH joint,

supplementing the static stability provided by the GH ligaments (Figure 2.2). The rotator

cuff, comprised of subscapularis, supraspinatus, infraspinatus, and teres minor, is also

responsible for humeral head depression and active external rotation of the humerus

(Blevins, 1997), both of which are fundamental for successful and prolonged tennis

performance.

28

Figure 2.2. Architecture of rotator cuff muscle group from anterior (left) and posterior (right) aspects (reprinted with permission; Pluim, 2001).

During tennis stroke production, these muscles collectively help to centre the humeral

head in the glenoid fossa. Ryu et al. (1988) demonstrated that while hitting

groundstrokes and serves, all muscles of the cuff were significantly and specifically

active. This research group, not unlike others, compared the rotator cuff muscle

patterning of the serve to that of overhead throwing. The cuff’s role in centering the

humeral head allows the “driver” muscles of the shoulder, that possess longer lever arms

(i.e. latissimus dorsi, and pectoralis major), to move the humerus in relation to the

scapula. For example, in the serve, selected shoulder muscles - flexor biceps brachii,

supraspinatus, infraspinatus, serratus anterior, and anterior deltoid – become moderately

to highly active as players’ upper arms approach maximal external rotation (van Gheluwe

and Hebbelinck, 1986; Ryu et al., 1988; Morris et al., 1989). Then with the cuff optimally

aligning the humeral head, the triceps brachii, subscapularis, infraspinatus, serratus

anterior, anterior deltoid, pectoralis major, and latissimus dorsi contract vigorously to

facilitate the acceleration of the upper arm and racquet to the ball. Here, the high

activity of the upper arm’s internal rotators is consistent with the finding that internal

rotation of the upper arm is a significant contributor to the generation of racquet velocity

in the serve (Elliott et al., 1995).

Optimal shoulder joint function requires sound scapular control. As outlined by Kibler

(1998), the scapula must provide a stable base for GH articulation, rotate as needed,

and facilitate the transfer of forces from the trunk to the arm. The scapular muscles’ role

29

in allowing the scapula to move in congruence with the rotating upper arm during

overhead activities is thus all important. For example, the first 30-50ْ of GH abduction

should see the scapula move laterally (Kibler, 1998), before rotating about a fixed axis

through an arc of approximately 65ْ as the shoulder reaches full extension (Poppen and

Wlaker, 1976). This motion is believed to account for the 2:1 ratio between the GH

abduction and scapulothoracic rotation that is observed in most throwing athletes

(Kennedy, 1993). Kibler (1998) suggested that such homogeny provides for the

maintenance of a strong shoulder joint angle, reduces the prospect of increased rotator

cuff compression and attenuates the stress placed on the GH joint’s capsule and

ligaments.

2.4.2 Proposed Mechanisms of Shoulder Injury

The aetiology of shoulder pain in tennis players remains a topic of considerable debate

and investigation. Nevertheless, its most common genesis relates to injury of the rotator

cuff. As aforementioned, the cuff’s four muscles stabilise the GH joint in dynamic motion

and selectively contract to externally rotate the humerus. Injury to the cuff is considered

multifactorial but has been popularly linked to overuse. Or, in other words, while it is

agreed that overuse pathologies are the main cause of presenting cuff injuries, the

process of degeneration is disputed between authors.

Lehman (1988a) and McCann and Bigliani (1994) proposed that most shoulder pain,

especially in young players, is caused by primary subacromial impingement, as the

tendons of the rotator cuff and long head of biceps are placed in vulnerable positions

from repetitive movements at/above 90º shoulder abduction. Nirschl (1992) preferred to

contend that in the majority of cases, tendinitis is the primary condition and is triggered

by tensile overload of the rotator cuff as it works maximally to stabilise the humeral head

in the glenoid in overhead motion. Subsequent degeneration leads to active instability

and may subject the cuff to secondary impingement. According to Ellenbecker (1995)

the extreme ranges of motion that characterise upper arm external rotation in the serve

may also lead to anterior capsule attenuation, producing laxity patterns similar to those

reported for professional baseball players and thus further stress the dynamic stabilisers.

30

Other reported conditions causing shoulder pain include cervical radiculopathy, calcific

tendinitis, and degenerative joint disease of the AC joint (McCann and Bigliani, 1994).

2.4.2.1 Mechanisms of injury to the rotator cuff

Rotator cuff injury in tennis players and other athletic populations can be classified

according to the mechanism of injury to the cuff. Blevins (1997), in his review of the

subject, identified cuff injury through macrotrauma but more likely repetitive

microtrauma, primary tensile overload, or impingement and tensile overload secondary

to instability. Improper throwing mechanics and muscle fatigue can also increase stress

on the cuff, resulting in a vicious cycle of cuff pathology. That is, as loads on the cuff

increase, compromising its ability to centre the humeral head in the glenoid, increased

anterior and superior translation occurs (Blevins, 1997). Ultimately mechanical

impingement, collagen tensile failure and further cuff dysfunction may result. The upper

and posterior portion of the cuff (supraspinatus, infraspinatus, and teres minor) are

suggested to run the highest risk of failure as they harbour the most load near the end

of the backswing and during the follow-through phase (Pluim, 2001). In the section to

follow, mechanisms of rotator cuff injury will be summarised.

2.4.2.1.1 Primary impingement Cuff impingement may occur primarily but typically develops secondary to tendon

overload or GH instability. So, rare in athletes under the age of 40, this type of

impingement may result from the acromion’s bony anatomy limiting the space available

for the rotator cuff between the acromion’s undersurface and the humeral head (Blevins,

1997). The literature describes three types of acromions and an increased incidence of

rotator cuff pathology has been noted in individuals with hooked acromions (Bigliani et

al., 1986). Primary subacromial impingement, as it is otherwise known, may lead to

inflammation, fibrosis and tearing of the cuff (Neer, 1972).

2.4.2.1.2 Primary tensile overload (rotator cuff failure) In overhand sports, the rotator cuff can be stressed beyond its ability to adapt resulting

in tissue failure, inflammation and degeneration (Fleisig et al., 1995). Some reports

attribute this stress to a sudden increase in activity intensity and/or duration (Blevins,

31

1997), while Jobe and Bradley (1988) have stated that injuries sustained to the rotator

cuff in training are commonly due to overuse. It is during the deceleration phases of

motions like the serve, where the cuff eccentrically contracts to resist distraction,

horizontal adduction and internal rotation of the humerus and is believed to be at

greatest risk of tensile failure (Jobe et al., 1984). So, service techniques that produce

greater shoulder joint loads – particularly as the upper arm decelerates – likely

predispose players to this pathology.

Indeed, Fleisig et al. (1996) attributed many of the rotator cuff tears that occur in

throwers, especially between the posterior midsupraspinatus and the midinfraspinatus

(Andrews and Angelo, 1988), to tensile failure of the cuff. Subsequent rotator cuff

dysfunction (i.e. of infraspinatus and teres minor) has been linked to increased GH

translations and secondary impingement (Blevins, 1997; Fleisig et al., 1995). More

specifically, secondary impingement sees the humerus migrate superiorly so that the

greater tuberosity of the humerus, the rotator cuff muscles or the biceps muscles

impinge against the inferior surface of the acromion or coracoacromial ligament (Fleisig

et al., 1995).

2.4.2.1.3 Instability Instability describes abnormal laxity of the GH joint as a result of traumatic dislocation or

subluxation. It may also arise through the repetitive stretching of the joint’s stabilising

structures (Blevins, 1997). In tennis, this occurs during the serve. That is, as the

humerus is abducted through 90º and maximally externally rotated, the anterior capsule

and GH ligaments resist anterior translation of the humeral head. In the process, they

are stretched, and over time anterior GH joint instability may develop. While players with

genetically mobile joints may suffer multidirectional GH joint instability, anterior

instability is more common among tennis players (Pluim and Safran, 2004). Those

players with tight, posterior capsules and/or weak rotator cuff and scapula stabilising

musculature are believed to be at greatest risk of suffering this pathology (Pluim and

Safran, 2004).

32

Unfortunately for tennis players, failure of the GH joint’s passive subsystem (i.e. anterior

capsule and GH ligaments) to limit the anterior translation of the humeral head may also

lead to cuff injury through direct mechanical compression (primary impingment) or

through increased cuff loading (secondary impingement) (Jobe and Kvitne, 1989; Meister

and Andrews, 1993). The prospect of GH instability may therefore increase among

players who employ serving techniques characterised by augmented shoulder joint

loading near MER of the upper arm.

2.4.2.1.4 Secondary impingement As its name may suggest, this type of impingement occurs secondary to the

aforementioned conditions. That is, when the upper arm is overhead, instability and/or

cuff dysfunction (stemming from primary tensile overload) can see the humeral head

translate superiorly so the supraspinatus or the biceps muscles impinge under the

coraco-acromial arch (Blevins, 1997; Fleisig et al., 1995).

More recently, athroscopic observation has given rise to internal impingement as a sub-

classification of secondary impingement. Internal impingement occurs when the arm is

maximally externally rotated and results from the pinching of posterior supraspinatus

and anterior infraspinatus between the humerus and the posterior-superior glenoid rim.

Anterior GH joint instability, limited upward rotation of the scapula and hyperangulation

(extension of the abducted, externally rotated upper arm beyond the plane of the

scapula) are suggested to increase the likelihood of throwers and tennis players suffering

this condition (Blevins, 1997). Consequently, service techniques that exhibit

hyperangulation of the humerus and probably greater horizontal abduction torques

during the backswing may be pathogenic.

2.4.2.1.5 Macrotrauma Cuff injury may also occur as the result of a direct blow to the shoulder (Blevins et al.,

1996). However, this pathology is likely limited to participants in contact sports, and its

incidence in tennis is negligible.

33

2.4.3 Variables that may contribute to Shoulder Injury

2.4.3.1 Technique-related factors

Common to the acceleration phase of most throwing activities is the maintenance of the

throwing shoulder at approximately 90º (relative to the trunk-thorax plane). Matsuo et

al. (1999) demonstrated that this angle correlated positively with reduced shoulder and

elbow joint loading, while still allowing for high speed ball release. A seemingly strong

and desirable position for the shoulder to function, Fleisig et al. (2003) confirmed a

similar alignment of the upper arm and trunk in the high performance serve. Overhead

athletes unable to approximate this trunk-upper arm aspect may experience

compromised scapula mechanics or place structures of the GH joint under greater stress.

Subsequent pain may in turn inhibit muscle recruitment and scapula tracking.

With different techniques noted to produce variable kinetic patterning in the serve (Elliott

et al., 2003), it is probable that the activation of certain muscles and by extension their

fatiguability and potential for injury, may be affected. To this end, Burnett et al. (1995)

investigated the effect of a 12-over spell (i.e. 12 mulitples of 6 bowls/deliveries) on fast

bowling cricket technique, an overhand skill not dissimilar to the tennis serve. While no

significant kinematic differences were found throughout the 12-over effort, there was

some evidence of kinematic change depending on the fast bowling action employed. This

finding was enough to prompt Portus et al. (2000) to examine the relationship between

bowling technique and shoulder counter-rotation throughout an 8-over spell. Varying

amounts of shoulder counter-rotation have been observed in different fast bowling

actions, and increased counter-rotation has been linked to low back injury in this athlete

population. Indeed the researchers were able to demonstrate that shoulder counter-

rotation (between overs 2 and 8) increased with the number of bowls delivered by

bowlers using front-on techniques (Portus et al., 2000). Whether or not this type of

bowling action magnifies tissue load and/or accelerates muscle fatigue to progressively

increase shoulder counter-rotation was not investigated. In tennis, similar relationships

may exist between serve technique, repetition, and shoulder injury. Certainly altered

throwing mechanics secondary to muscle fatigue or soreness about the shoulder is

thought to increase the demands on the cuff (Blevins, 1997). Also, while fatigue has

34

been shown to have no effect on the length at which muscles are injured, fatigued

muscles do absorb less energy before reaching the degree of stretch that causes injury

(Mair et al., 1996). In practice, this may mean that the use of certain serving techniques,

especially those that may be less mechanically efficient, could perpetuate greater injury

risk.

2.4.3.2 Inflexibility

Research has failed to substantiate a clear cause-and-effect relationship between

inflexibility and injury (Warren and Jones, 1987), yet inflexibility has been shown to be a

risk factor in shoulder injury (Ellenbecker, 1995). For example, deficits in GH internal

rotation and total rotation - adaptations typical in high performance tennis players - are

reported to make these athletes more susceptible to shoulder pathologies (Kibler et al.,

1996). Further, a correlation between poor dominant arm internal rotation and the

presence of shoulder pain has been substantiated (Vad et al., 2003), while Kibler et al.

(1992) have also inferred inflexibility to contribute to humeral stress fractures by

affecting normal mechanical function.

As intimated above, tennis players lose range of motion in dominant shoulder internal

rotation and gain in external rotation over the course of their careers (Kibler et al., 1996;

Schmidt-Wiethoff et al., 2004). Compared with their non-dominant shoulders, the 39

elite male (M) and female (F) players analysed by Kibler et al. (1996) displayed reduced

dominant arm internal rotation (M: -26.3º, F: -30.0º), more external rotation (M:

+12.3º, F: +5.8º), and less total rotation (M: -13.4º, F: -24.6º). These findings are

reinforced by Schmidt-Wiethoff et al. (2004) who showed 27 male professional players

had significantly less internal rotation (43.8º v 60.8º) and significantly greater range of

external rotation (89.1º v 81.2º) in their dominant arms than in their non-dominant

arms. Of further interest is that while Kibler et al. (1996) showed losses of internal

rotation to increase with years of tournament play and age, the data of Ellenbecker

(1992) and Schmidt-Wiethoff et al. (2004) indicate that this kinematic imbalance is likely

present in all elite junior and professional players.

35

That the loss of internal rotation seemingly overshadows the gain in external rotation to

result in reduced total shoulder range of motion has been a consistent finding of the

aforementioned studies. Kibler et al. (1998) reported that lax anterior capsules are

common to tennis players and may allow increased external rotation. However, Schmidt-

Wiethoff et al. (2004) hypothesised that losses in total range can be attributed to

contracted posterior capsules that result from the repetitive eccentric loading conditions

and subsequent, recurring microtraumatisation sustained during tennis play. In turn,

Harryman et al. (1990) suggested that these structural changes may propagate anterior

and superior migration of the humeral head in the abducted and flexed humerus. Tight

posterior structures may also exacerbate anterior shear forces on the shoulder joint

during overhead motion, while excessive external rotation has been implicated in rotator

cuff injury (Davidson et al., 1995).

2.4.3.3 Muscle imbalance

Muscular strength imbalances, reported to occur naturally or as an adaptation to the

demands of training (Parker et al., 1983), have been suggested to increase the risk of

tissue injury (Cook et al., 1987; Kibler et al., 1988). Certainly in tennis, selective

strengthening of the muscles surrounding the shoulder joint is a commonly accepted by-

product of training and competitive play (Ellenbecker and Tiley, 2001). Put simply, the

muscles that insert on the anterior and medial aspects of the humerus are those that

contribute most significantly to the generation of racquet velocity in both the serve and

forehand. These two shots comprise the cornerstone of most players’ technical

repertoires and the musculature responsible for accelerating the hitting limb and

racquet, is presented with constant training stimuli to which it must adapt. On the other

hand, the muscles which attach to the scapula or the humerus posteriorly such as

infraspinatus and teres minor may contribute to the development of racquet speed in the

backhand but are more regularly involved in upper arm and racquet deceleration in the

serve and forehand. Consequently, the internal and external rotators achieve

disproportionate strength and power gains, creating a functional muscle imbalance,

which has been suggested a causative factor in overuse injuries to the serving shoulder

(Chandler et al., 1992). Stress fractures of the humerus, albeit not an overly common

tennis injury, could also arise from such muscular strength imbalances (Kibler et al.,

36

1992). Kibler et al. (1992) infer that absolute or relative force may overload weak

muscles, resulting in those forces being transferred to the skeletal system to potentially

precipitate a stress reaction in the humerus.

Chandler et al. (1992) confirmed increased internal rotation strength in the dominant

arms (peak torque, M: 20.6-29.9Nm; M: 16.0-23.8Nm for non-dominant arms) of college

tennis players, while simultaneously reporting no differences in power and total work in

dominant and non-dominant arm external rotation. The work of Ellenbecker and Roetert

(2003) corroborated this strength differential among elite, young male and female

players, while comparable bilateral upper arm rotational strength of female collegiate

players has been attributed to variation in playing experience and skill (Kraemer et al.,

1995). Similarly disparate external:internal rotation strength ratios between dominant

and non-dominant arms have been reported among baseball pitchers (Alderink and

Kluck, 1986; Cook et al., 1987) and other sports athletes (Ivey et al., 1984). The

distinction between the primary roles of the internal and external rotators of the

shoulder coupled with differences in load repetition, their architecture and cross-

sectional area help account for this localised strength imbalance. Nevertheless, increases

in dominant shoulder internal rotation without accompanying improvements in external

rotation strength have been proposed to overload the shoulder joint, predisposing tennis

players to injury (Chandler et al., 1992). Or, in more practical terms, the inability of the

external rotators to negotiate the repetitive and high forces generated by the internal

rotators during the serve is likely to place tennis players at greater shoulder injury risk.

To this end, kinematic analyses of the serve and forehand strokes highlight the

important contribution of upper arm internal rotation in the development of racquet

velocity (Elliott et al., 1995; Takahashi et al., 1996). This suggests that the muscles

responsible for this action should be trained. However, with selective strengthening of

the internal rotators already a consequence of tennis play, their further development

must be considered holistically. Recommendations are that the concentric strength ratio

between external:internal rotation approximate 66-75%, or that concentric external

rotator strength be at least 2/3 that of the internal rotators (Alderink and Kluck, 1986;

Cook et al., 1987; Codine et al., 1997; Ellenbecker and Roetert, 2003). While deviation

37

from which may not systematically nor necessarily lead to tendiomuscular pathologies of

the shoulder (Codine et al., 1997), there is some evidence linking altered ratios with

pathologic conditions of the shoulder (i.e. impingement and instability; Warner et al.,

1990).

There are however, a number of factors that need to be considered when interpreting

these shoulder joint internal:external rotation strength ratios. Firstly, they largely

pertain to concentric muscle action. Where the internal rotator musculature is indeed

concentrically active during the acceleration phase of the forehand and serve, such

evaluations are not representative of, and therefore fail to specifically assess, the

dynamic eccentric actions of the external rotators during upper arm and racquet

deceleration. Noting that this concentric external-to-concentric internal ratio fails to

functionally correspond to the predominant activity of overhand athletes, Noffal (2003)

compared an eccentric external-to-concentric internal ratio between throwers and non-

throwers. In finding ratios from 1.17-1.60, Noffal (2003) showed the eccentric strength

of the external rotators to be greater than the concentric strength of the internal rotator

musculature. Interestingly however, these ratios were lower amongst the throwers,

implying that gains in internal rotation strength – typical of such populations – are not

observed in external rotation, and are likely related to the musculature’s aforementioned

dichotomous roles. Secondly, isokinetic testing provides for the generation of limb

velocities that are appreciably (90%) lower than those reported in high performance

serving, potentially reducing the functional relevance of reported results.

2.4.3.4 Improper scapula mechanics

Sufficient upward rotation of the scapula is believed critical for the injury-free

performance of overhead sports skills (Kibler, 1998). A deficit in this motion has been

implicated in shoulder injury (Endo et al., 2001; Ludewig and Cook, 2002), and more

specifically in subacromial impingement (Kibler, 1998; Michener et al., 2003). That is,

accompanying insufficient upward scapular rotation is decreased acromial elevation,

which perpetuates the impingement of the subacromial structures. Of additional note is

that shoulder muscle fatigue has been demonstrated to impair the maintenance or

achievement of this scapular motion (Birkelo et al., 2003; Tsai et al., 2003). With this in

38

mind, serving techniques that introduce the largest forces at the shoulder joint,

presumably advancing the onset of shoulder muscle fatigue, may heighten a player’s

chances of suffering subacromial impingement.

Although shoulder muscles are commonly injured from microtrauma-induced muscle

strain leading to weakness and force couple imbalance, they can also be inhibited by

painful conditions around the joint (Kibler, 1998). Muscle inhibition is common in GH

abnormalities whether from instability, labral lesions or arthrosis (Moseley et al., 1992;

Kuhn et al., 1995). It is seen both as a decreased ability for the muscles to exert torque

and stabilise the scapula and as a disorganisation of the normal muscle firing patterns of

the muscles around the shoulder (Glousman et al., 1988; Warner et al., 1992; Kuhn et

al., 1995).

Injury to the long thoracic nerve or the spinal accessory nerve can alter the muscle

function of the serratus anterior and trapezius muscles respectively. These types of

nerve injury occur in less than 5% of shoulder muscle function abnormalities, but can

severely compromise the motor control and stabilisation strategies of the region. That is,

the serratus anterior and lower trapezius form a crucial part of the force couple

responsible for elevating the acromion and are reportedly the muscles first and most

affected by inhibition-based muscle dysfunction (Glousman et al., 1988; Pink and Perry,

1996; Kibler, 1998). Kibler (1991) and Warner et al. (1992) further implicate poor

acromial elevation and resultant impingement in the early stages of many shoulder

problems, including cuff tendonitis and GH instability.

If inhibition does lead to the scapula becoming deficient in motion or position, force

transfer from the lower extremity to the upper extremity can be severely impaired

(Kibler, 1998). Kibler (1995) further highlights the important conciliatory role of the

scapula by suggesting that a 20% decrease in kinetic energy delivered from the hip and

trunk to the arm would demand an 80% increase in mass or a 34% increase in

rotational velocity at the shoulder to deliver the same amount of resultant force to the

hand.

39

2.5 THE RELATIONSHIPS BETWEEN SERVING MECHANICS, SHOULDER

LOADING AND PERFORMANCE

The tennis serve is the most important stroke in tennis. A mechanically sound serve that

provides for velocity generation and variation in ball placement while minimising the

potential for injury, is an integral part of player development (Elliott, 2001). Viewing the

serves of top professionals clearly illustrates that there is no single technique used, yet

certain serves or mechanical characteristics that are part thereof may augment injury

risk (Elliott et al., 2003). These characteristics, as a function of the serve, their potential

influence on shoulder joint kinematics and kinetics, and therefore injury, will be

discussed below.

2.5.1 Serve Type

The players’ tactical intention, shaped by factors such as individual strengths and

weaknesses, the court surface and match score, is the principal determinant of the type

of serve used (Crespo and Miley, 1998). Nevertheless, skilled tennis players tend to use

their first serves to assert immediate authority on the point, and thus engage flat or

power serves. This type of serving is characterised by high, horizontal ball velocities.

Second serves on the other hand, see players face the prospect of losing points outright

should the serve fail to land in the designated service box. A premium is therefore placed

on accuracy rather than speed, and skilled players largely use a spin (slice or topspin)

serve as their second delivery (Trabert and Hook, 1984; Douglas, 1992; Groppel, 1992;

Chow et al., 2003a). The game’s elite introduce the ball in similar fashions yet often hit

their second serves more aggressively, applying a more oblique (vertical and lateral)

rotation to the ball causing it to ‘kick’ or bounce forward and to the left of the opponent

(Miller and Cross, 2003).

To achieve the different tactical objectives of the first and second serves, players

manipulate the placement of their ball toss. That is, Chow et al. (2003a) reported right-

handed male professional players to impact flat serves in front (≈0.6m) and marginally

to the left (≈0.2m) of their front foot, while racquet-ball contact of second serves

occurred closer to the baseline and further to the left (≈0.6m). In reporting impact

location with respect to centre of mass, Vorobiev et al. (1993) confirmed that right-

40

handed players contacted the ball further to the left for second serves (0.36m) than first

serves (0.12m).

The theoretical notion, and belief among many coaches, that speed is exchanged for

accuracy between the first and second serves would appear consistent with other speed-

accuracy trade-offs in human performance (Fitts, 1954). Furthermore, with Elliott et al.

(2003) demonstrating upper extremity joint loading to increase along with service speed,

it may be assumed that first (or flat) serves would stress shoulder and elbow joints more

than second (or kick) serves. However, Chow et al. (2003a) found no significant

difference between the resultant, pre-impact racquet velocities of the first and second

serves of four professional players. The directional components of racquet velocity did

differ with serve type, but this finding clouds the above assumption nonetheless.

Corroboration of which serve, FS or KS, loads upper extremity joints, and in particular

the shoulder joint, to a greater extent warrants investigation.

2.5.2 Foot Arrangement

In the serve, the stance that players assume can be described as either FU or FB (Elliott

and Wood, 1983). While found to have no effect on serving velocity, these lower limb

arrangements do interact with the court differently such that resultant ground reaction

forces (GRF’s) and the players’ ‘drive’ to impact vary (Elliott and Wood, 1983;

Bahamonde and Knudson, 2001). The FU technique sees players bring their back foot

forward to their front, prior to pushing upward and forward to the ball. Other players will

leave their back foot behind, near their original position (FB technique) before driving

obliquely upward to racquet-ball impact.

The FU technique has been reported to produce higher vertical GRF’s (1.9-2.1

bodyweight (BW) vs FB: 1.3-1.5BW), while higher horizontal GRF’s characterise the FB

technique (Elliott and Wood, 1983; Bahamonde and Knudson, 2001). Elliott and Wood

(1983) also revealed the FU technique helped players attain significantly higher hitting

positions compared with the FB serve at impact (2.65m vs 2.54m above the court). The

researchers considered this a by-product of the FU technique’s greater vertical impulse.

The FB stance, on the other hand, is believed beneficial for serve and volleyers as it

41

creates comparatively less horizontal braking force to slow the forward momentum of

the body (Bahamonde and Knudson, 2001). No attempt has been made to investigate

potential links between service stances and variable loading profiles of the upper

extremity.

2.5.3 Leg Drive

Not until the 2003 study of Elliott et al. had there been an attempt to quantify the link

between lower limb interaction and upper extremity coordination during the serve.

Empirically, the significance of the relationship was suspected, yet lower limb analyses of

the serve had traditionally preferred to examine the connection between GRF and foot

placement (Elliott and Wood, 1983; Bahamonde and Knudson, 2001).

On the back of the early work of Elliott and Wood (1983), most coaching texts propound

the concept of ‘leg drive’. Although some tennis coaches refer to leg drive and high

amounts of knee joint flexion almost interchangeably, it encompasses more

comprehensive joint and muscle action and can be considered the effective, sequentially

coordinated extension of the ankle, knee and hip joints. Successful execution of this

triple joint extension is purported to increase impact height, facilitate the transfer of

angular momentum between the lower body and trunk, and increase MER of the upper

arm to lengthen the racquet’s forwardswing to the ball and place the internal rotator

musculature of the upper arm on stretch (Elliott, 2001). It is considered the first link in

the coordinated body movement or ‘kinetic chain’ and thus essential for high-speed

serving (Elliott and Wood, 1983; Kibler, 1998).

In spite of its multi-joint nature, angular change about the knee joint is, from a coach’s

perspective, the most observable indicator of the efficiency of a player’s leg drive. For

this reason, Elliott et al. (2003) used variation in knee joint flexion to appraise leg drive,

and its effect on upper extremity joint kinetics. More specifically, they divided a sample

of professional players competing at the 2000 Olympics into those with an effective leg

drive (flexed at their front knee joints by >10° at MER of the upper arm) and those with

a less effective drive (flexed by <10°) to investigate the hypothesis that the amount of

42

front knee flexion at MER was linked to shoulder joint torque at MER and therefore the

potential for injury.

The players with a more effective leg drive or larger knee flexion at MER recorded lower

internal rotation torques at MER (43.7 Nm) than the less effective group (57.8 Nm).

These torques were similar to those detailed by Bahamonde (1989) to characterise the

serves of college level tennis players, but less than the 94 Nm reported by Noffal and

Elliott (1998). Elliott et al. (2003) suggested that the lower internal rotation torques of

the players with more effective leg drives may reflect the use of a facilitatory inertial

transfer from the trunk to the upper limb to externally rotate the upper arm. It was

paradoxically inferred that the players whose knees were more extended at MER more

actively used their upper arms’ external rotators to achieve MER. No external rotation

moments were reported to corroborate these extrapolations however. Bahamonde

(2000) nevertheless reported large negative shoulder joint internal-external rotation joint

power (-220 ± 72W) near MER of the typical collegiate serve, indicating that high level

players place large eccentric loads on their internal rotator musculature to thus benefit

from the storage of elastic energy during subsequent humeral internal rotation.

A more effective leg drive was also linked to lower peak internal rotation torques (≈55 vs

65 Nm) during the forwardswing to impact. Indications are thus that less load was

placed on the shoulder joint as the associated muscles concentrically contracted to

rotate the upper arm to impact. The players with less knee flexion experienced a similar

torque to the ≈ 70 Nm reported for professional pitchers (Fleisig et al., 1999).

In investigating the lower limb activity that characterised the serves of players of

different playing standards, Girard et al. (2005) found elite players to generate higher

vertical GRF’s than both intermediate and beginner players. This fact coupled with the

elite players’ more explosive transition from maximum knee flexion to take-off (i.e. feet

leaving the force platform) led the researchers to ascertain that they served with more

vigorous knee extension, or leg drive, than their lower-level counterparts. The magnitude

of leg muscle activation however, was similar across all players, while some differences

in the timing of leg muscle activation were suggested to hint at more precocious

43

activation strategies among the elite players. Interestingly, the researchers also

evaluated the vertical hopping ability of all subjects and found that maximum leg power

and leg stiffness were similar, irrespective of playing ability. As subsequently suggested,

the similarity of these lower limb neuromuscular characteristics may indicate that the

observed kinetic and subtle electromyographic (EMG) differences were primarily due to

the coordinative abilities that characterised the varying levels of subject expertise. With

all subjects assuming a stance somewhere between the FU and FB techniques, the

researchers were not able to evaluate potential relationships between the serve’s stance

and the EMG and kinetic characteristics of the lower limb.

So, in spite of these recent efforts to elucidate the contribution of the leg drive to the

serve, it remains clear that more comprehensive evaluation of lower limb joint action,

and not just that of the knee joint, and its influence on serve performance is needed.

Already, decreased knee joint flexion at MER has been linked to increased shoulder joint

loading but its effect may vary temporally, in relation to the action of the ankle and hip

joints, or depending on serve type.

2.5.4 Follow-through

Scientific studies of the tennis serve have paid scant attention to the mechanics of the

follow-through. Almost exclusively they have described pre-impact serve kinematics and

kinetics; preferring to extrapolate their findings to the follow-through, and on occasion,

to injury. Consequently, for insights into the upper extremity joint loads players harbor

during the serve’s follow-through, we must rely upon those reported to characterise the

baseball pitch following ball release (Jobe et al., 1984; DiGiovine, 1992; Fleisig et al.,

1995, 1996).

As aforementioned eccentric muscle contractions produce more tension and are more

commonly linked to muscle injury than their concentric counterparts. It should therefore

come as little surprise that many of the shoulder injuries sustained by overhead athletes

are associated with the high load their shoulder musculature, and in particular the

rotator cuff, tolerates in decelerating the throwing or hitting limb/apparatus during the

44

follow-through. Indeed, in baseball pitching Fleisig et al. (1996) consider both the

shoulder and elbow joint to be susceptible to injury during arm deceleration.

Rotation of the trunk about its three axes and upper arm internal rotation continue

during the early phase of the follow-through (Elliott, 2001). These actions are needed to

minimise the eccentric load placed on the shoulder muscles responsible for slowing the

rotating upper arm and racquet. Indeed, while Kellis and Baltzopoulos (1995) implicate

the elastic elements of this musculature in the development of the necessary upper arm

external rotation torque, it is nonetheless here where the rotator cuff eccentrically

contracts to resist distraction, horizontal adduction and internal rotation of the humerus

and is believed to be at greatest risk of tensile failure, and by extension muscle strain or

tear (Jobe et al., 1984; Fleisig et al., 1996). Similarly, if the humerus is repeatedly

allowed to go through more than 60º of internal rotation during the follow-through

(Kibler, 1995), stressing the anterior and posterior bands of the GH ligament (O’Brien et

al., 1990), the prospect of players encountering GH instability increases.

After baseball pitch release, the posterior shoulder muscles (i.e. teres minor,

infraspinatus and posterior deltoid) have also been shown to produce posterior and

compressive forces and a horizontal adduction torque (DiGiovine, 1992; Fleisig et al.,

1995). In the absence of this kinetic patterning, superior translation of the abducted,

horizontally adducted and internally rotated humerus would be expected, and if also a

feature of the tennis serve, the likelihood of players experiencing primary tensile

overload and secondary impingement problems would increase (Fleisig et al., 1996).

Meister (2000) further implies that rotator cuff weakness, fatigue or improper mechanics

may impair a player’s ability to generate these kinetics, and thus predispose the player

to shoulder injury.

So with the loads encountered at the shoulder during the follow-through empirically

implicated in its injury, corroboration of as much would appear essential. Certainly, from

performance and injury prevention perspectives, information detailing potential

relationships between the follow-through’s kinetics, and serve type or technique may be

most useful.

45

2.5.5 Wheelchair Tennis Serve

Like able-bodied tennis players, wheelchair players generate high racquet velocities

during the serve. However unlike able-bodied players, they generate these high racquet

velocities without the aid of their legs. Wheelchair players are thus devoid of the lower

limb drive believed to facilitate able-bodied players in externally rotating their upper

arms and achieving optimal racquet displacement during the service backswing. By

extension wheelchair players (particularly the majority that do not counter-rotate their

chairs to help create an inertial humeral external rotation lag) may need to rely on

greater concentric contraction of their external rotator musculature to attain comparable

levels of upper arm external rotation. From a loading perspective, certain parallels are

likely therefore to exist between the upper extremity joint kinetics of wheelchair players

and able-bodied players, whom possess less effective ‘leg drives’. Alternatively however,

it could be that wheelchair players consciously enter into less upper arm external

rotation and are thus forced to rely on the concentric work of their internal rotators -

more explicitly and through a comparatively reduced ROM - to build racquet speed.

Expectations would thus be that the upper arm internal rotators of wheelchair tennis

players are preferentially strengthened or that their external:internal rotation strength

ratio is altered. The findings of Bernard et al. (2004), in which wheelchair athletes

trended toward developing higher upper arm internal rotation torques than selected

tennis players and other non-athletes but displayed relative weakness in upper arm

external rotation depending on their neurological level of lesion, offer some support to

both hypotheses.

In tennis, the game’s governing body, the ITF, consider any player to have a medically

diagnosed permanent physical disability resulting in a substantial loss of function in one

or both lower extremities as eligible to compete on the wheelchair tennis tour. Where a

large number of amputees and sufferers of perthes disease compete, players with spinal

cord injuries represent the largest playing population. As disability type impacts

performance differently (Goosey-Tolfrey and Moss, 2005), an understanding of the types

of spinal cord injury as well as their influence on physical performance is necessary prior

to examining potential links between service technique and upper extremity joint

kinetics.

46

The effects of spinal cord injury depend on the neurological level of the injury and its

classification as complete or incomplete. A complete injury sees players exhibit no

function (i.e. sensation or voluntary movement) below the level of the injury; with both

sides of the body being equally affected. Conversely, an incomplete injury may affect

one side of the body more than the other and provides some functioning (i.e. voluntary

limb movement, movement sensation) below its primary level. The level of spinal cord

injury also provides an insight into what parts of the player’s body might be affected by

paralysis and loss of function. However, in general terms the higher level of spinal cord

injury, the more affected the player.

It therefore follows that most elite wheelchair tennis athletes that have sustained spinal

cord injury at or below the thoracic level possess varying levels of physical function. For

example, some wheelchair tennis players can strengthen their trunk and abdominal

muscles to benefit their movement and stroke play, while other players cannot. Although

seemingly trivial, such a difference is likely to have ramifications on performance,

especially from a kinetic standpoint (Goosey-Tolfrey and Moss, 2005). That is, the notion

of the kinematic or kinetic chain is commonly referred to in sports where velocity

generation is important. Equally the failing of one of the chain’s “links” to contribute

optimally to the development and transfer of momentum has been suggested to

compromise skilled, high-speed performance, and may indeed result in a compensatory

response from one of the chain’s other links. So already localised by virtue of the skill’s

lack of lower limb involvement, the inability of some players to fully activate their trunk

musculature further abbreviates the wheelchair serve’s kinematic chain as compared

with that of able-bodied players.

Indeed, the effects of an abbreviated kinematic chain on throwing performance have

already been investigated. For example, Toyoshima et al. (1974) constrained technique

to evaluate the contributions of hip, trunk and shoulder rotations to ball velocity, and

found that when lower body and trunk involvement was restricted and no forward stride

allowed, peak ball velocity dropped to 63.5% of normal overhead throw speed. Similarly,

a model of sequential muscle activation elaborated by Alexander (1991) showed

systematic restriction of full trunk motion to reduce typical maximum ball throw

47

velocities by approximately 50%. The work of Whiting et al. (1985) inferred a

comparable reduction in the speed of water polo throws as compared with land-based

throws aided by the generation of GRF’s. Significantly, McMaster et al. (1991) also

associated the lack of a base of support in the water polo throw to heightened shoulder

joint forces. Intuitively, it follows that a similar compensatory kinetic outcome may be

observed in the wheelchair tennis serve.

In the non-athletic wheelchair population, the incidence of shoulder pain and

impingement is high (Pentland and Twomey, 1994; Curtis et al., 1999). These individuals

require the continual use of their upper extremities, not just for mobility but for

transfers, weight-relief tasks and reaching activities. The relatively high mechanical

shoulder joint loads associated with these everyday tasks, coupled with their high

frequency, has been implicated in the development of these shoulder complaints (Bayley

et al., 1987; Reyes et al., 1995; Perry et al., 1996; Harvey and Crosbie, 2000; van

Drongelen et al., 2005). Unsurprisingly, injuries to the upper extremity, and more

particularly the shoulder, have also largely dominated all wheelchair athlete

epidemiology (Ferrara and Davis, 1990). Indeed, as with able-bodied players, wheelchair

tennis players commonly suffer from shoulder overuse injuries (Ferrara and Davis, 1990;

Burnham et al., 1993). Paraplegics with shoulder problems have been shown to exhibit

weakness in shoulder joint adduction and internal and external rotation, as compared

with those without shoulder problems (Burnham et al., 1993). However, as alluded to

above, wheelchair athletes displayed greater strength in upper arm internal rotation than

able-bodied tennis players and non-athletes, so these pathologies appear more likely

related to exaggerated agonist/antagonist peak torque ratios (1.7:1, Bernard et al.,

2004). Pluim and Bullock (2003) agreed and suggested that shoulder muscle imbalance,

with comparative weakness of the humeral head depressors may be a causative and

perpetuating factor in rotator cuff injury in wheelchair athletes. Indications are also that

players with high-level spinal cord injury may show greater prevalence and intensity of

shoulder pain than players with lower-level spinal cord injury (Curtis et al., 1999).

As aforementioned, players that use their lower limbs less effectively when serving have

been shown to load their shoulders more than players who derive greater benefit from

48

their lower limbs’ interactions with the ground (Elliott et al., 2003). Extrapolating this

relationship to the wheelchair game would tend to suggest that its combatants may be

expected to generate higher amounts of pre-impact, shoulder joint force and torque

relative to serve velocity. Similarly, devoid of optimal trunk rotation, the wheelchair

tennis serve, as compared with the able-bodied serve, may further amplify the role of

the shoulder in generating velocity, and presumably harbouring load. While no research

has attended the upper-extremity joint kinetics of the able-bodied serve post-impact, it

would appear reasonable to also assume that wheelchair players may experience

comparatively larger shoulder joint loads after racquet-ball impact. That is, in the

absence of the facilitatory post-impact, lower body and trunk rotation to indirectly help

decelerate the continued progression of upper arm and racquet, wheelchair players may

be required to more forcibly eccentrically contract the shoulder external rotators.

Obviously, these assertions require quantification but may lead to the revision of current,

technical and strength and conditioning instruction of wheelchair tennis athletes.

2.6 REPEATABILITY OF KINETIC AND KINEMATIC DATA IN MOTION

To most accurately describe the kinetics and kinematics of the tennis serve or for that

matter any sports skill, one must first ascertain the repeatability of its performance. The

biomechanical assessment of the serve involves the measurement of many aspects of

performance, whose reliability, or repeatability, may vary widely and be affected by

factors such as biological or physiological variation of the subject, limitations with the

equipment and subject motivation (Hunter et al., 2004).

Some variation in kinetic and kinematic patterning across multiple attempts at the same

task is accepted as motor control fact (Latash and Turvey, 1996). Yet it remains common

for sports biomechanics research to describe athletic performance based on a single

examination of a sports skill (Knudson, 1990). Indeed, investigations evaluating the

mechanics of tennis stroke production have typically assumed one superior performance

to be representative of a player’s overall stroke technique and thus analysed the highest

velocity stroke or the stroke subjectively assessed to be of the highest quality (Reid and

Elliott, 2002). With load shown to positively relate to stroke velocity (Elliott et al., 2003),

the inference is then that analysis of the highest velocity trial would highlight the skill’s

49

peak loading conditions. Doubts nonetheless remain as to how representative these

loads are of the skill’s repeated performance, and thus those most commonly implicated

in tissue injuries.

The execution of a sports skill with a high degree of repeatability should increase along

with the competency of the performer. A link between the legitimacy of using one trial as

a representative measure of a performer’s technique and ability level is also likely.

Bootsma and Wieringen (1990) reported that analysis of a single trial can be justified

with expert sportspersons whose within-trial consistency (movement repeatability) is

high. In studying the repeatability of the kinematic and kinetic characteristics of normal

adult gait, Kadaba et al. (1989) also concluded that it was reasonable to base clinical

decisions on a single gait evaluation. Another body of research exists however,

highlighting associated methodological concerns (Salo et al., 1997; Mullineaux et al.,

2001). For example, Salo et al. (1997) showed that depending on the variable analysed,

between 1 and 78 trials were needed to reliably represent the average movement

pattern of sprint hurdles, as performed by seven British national-level hurdlers. Bates et

al. (1992), in analysing the effect of trial size on statistical power, mounts a similar

argument for evaluating more than one trial.

In studying the upper extremity angular kinematics of the tennis forehand, Knudson

(1990) showed that collegiate players attained similar wrist and elbow positions

(coefficient of variation, CV < 5.9%) despite inconsistent wrist and elbow joint angular

velocities (CV= 90.6%) and accelerations (CV= 129.5%). Bootsma and Van Wieringen

(1990) demonstrated similar variation in the direction of bat travel during the

forwardswing of the elite table tennis forehand drive, but with very consistent bat-ball

alignments at impact. Expert marksmen have also been observed to make compensatory

upper-arm movements to achieve low variability in the position of the pistol barrel when

shooting (Arutyunyan et al., 1968). This illustrates that skilled performers can obtain

similar outcomes despite variation in coordination strategies.

So, in summary, while researchers concede that some variables may never be reliable

enough to monitor small changes in performance (Hunter et al., 2004), they

50

nevertheless implore sports scientists to pursue improvements in reliability by using the

average score of multiple trials (Hunter et al., 2004; Mullineuax et al., 2001). In some

instances however, like where subject fatigue may compromise results or where data

collection and processing become inordinately long, there will likely be an upper limit to

the number trials examined (Hunter et al., 2004). In tennis, normative data describing

the key mechanical characteristics of selected strokes is widespread, yet the number of

strokes upon which these data should be based has rarely been documented. An

oversight made all the more compelling by the fact that intra-subject variability has been

observed in tennis stroke production (Knudson, 1990), highlighting the anomalies of

conclusions drawn from single trials across several performers. Certainly, a recent effort

by Knudson and Blackwell (2005) to substantiate the variability of impact kinematics in

the tennis forehand has reinvigorated interest in the area. Averaging data from ten

forehands was required to obtain reliable racquet kinematics near impact; but this

number likely varies with stroke and subject playing level. Nonetheless, it reinforces the

need to verify the repeatability of the high performance FS and KS and therefore

determine the minimum number of executions, from which accurate, representative

observations can be made.

2.6.1 Selection of Data Treatment Procedures

Where repeatable data are needed to ensure the meaningfulness of subsequent analysis,

equally important is the selection of appropriate methods of data treatment. This

involves the derivation of an apposite filtering or smoothing technique, inclusive of an

optimal cut off frequency (COF) or MSE, to accurately represent higher-order kinematics

throughout the entire movement range (Giakis et al., 1998; Challis, 1999). In tennis

stroke production, the speed at which segments are rotated and balls hit exacerbates

the difficulty of identifying these best modes of treatment (Knudson and Bahamonde,

2001; Knudson, 2005).

Most previous examinations of tennis stroke kinematics have filtered their raw data using

routines with established COFs or MSEs (Bahamonde and Knudson, 2001; Reid and

Elliott, 2002; Knudson, 2005). Although an investigative norm, it is unlikely that these

routines are specific to the movement frequency that characterises tennis stroke

51

production (Yu et al., 1999). Challis (1999) raised further concerns related to the failure

of these routines to satisfactorily account for variations in the quality of sampled

displacement data. Alternative procedures such as residual analyses (Winter, 1990) are

often recommended as preferable for the determination of filtering routines that are

data-set specific. Indeed, the substantiation of a best level of filtering of the raw data of

the tennis serve is central to the validity of the serve’s kinematic and kinetic

representation.

Recently, the work of Knudson and colleagues (2001; 2005) has also demonstrated the

difficulties associated with attaining representative kinematic data near impact in the

forehand groundstroke. Earlier biomechanical analyses of tennis strokes were

subsequently shown to have treated displacement data inappropriately (Ariel and

Braden, 1979; Groppel and Nirschl, 1986; Elliott et al., 1989; Sprigings et al., 1994;

Elliott and Christmass, 1995; Takahashi et al., 1996), leading to some misrepresentation

of the actual movement performed, particularly around impact (Woltring, 1985; Giakis et

al., 1998; Knudson and Bahamonde, 2001). Consideration of only pre-impact data has

since become common practice (Reid and Elliott, 2002; Elliott et al., 2003; Fleisig et al.,

2003), yet Knudson and Bahamonde (2001) also demonstrated that this approach can

produce spurious results. That is, in investigating the effectiveness of different

extrapolation and smoothing techniques in negotiating the influence of racquet-ball

impact on the representativeness of forehand kinematics, these researchers showed that

extrapolated, then smoothed data were more representative than when no extrapolation

method was used. In other words, end-point error (i.e. misrepresentative kinematic

data) proliferated when only pre-impact data was smoothed. Whether similar data

consideration and treatment confounds the kinematics and kinetics of the tennis serve

needs to be determined.

2.7 CONCLUSION

In tennis, there is no single service technique used. For the most part, they are thought

to reflect the player’s morphology or flair. However, doubts have recently been raised as

to the efficacy of mechanical variation in serve technique, particularly from an injury

perspective. High performance players can be injured relatively frequently, and shoulder

52

injuries are among their most common complaints (Kibler, 1995). The shoulder joint

loads developed during the serve, which increase along with serve speed (Elliott et al.,

2003), are commonly implicated in the shoulder injuries sustained by this playing

population. Quantification of kinetic differences that may be related to the variable foot

arrangements, lower limb interactions and backswings observed in the serve would thus

seem of considerable importance from both performance and injury prevention

standpoints. Verification of shoulder joint loading during the serve’s follow-through

assumes similar significance.

53

CHAPTER 3: GENERAL METHODOLOGY

3.1 OVERVIEW OF 3D INTERPRETATION OF THE SHOULDER JOINT

Non-invasive measurement of high-velocity 3D upper extremity joint motion, as in the

tennis serve, is best achieved through stereophotogrammic means (Chiari et al., 2005).

Accordingly, the Vicon 612 (Oxford Metrics, Oxford, UK) system at the School of Human

Movement and Exercise Science was utilised to track retroreflective markers placed on

tennis players perfoming the tennis serve. Twelve cameras, operating at 250 Hertz (Hz),

minimised the prospect of marker occlusion and optimised marker reconstruction during

dynamic trials.

Accurate 3D representation of segments requires at least three non-collinear markers, or

points. Additional markers are required to define points of anatomical relevance and to

define joints. In the upper body, while estimation of wrist and elbow movement is

relatively simple as both joints can be represented by two degrees of freedom,

biomechanical modelling of the shoulder is complicated by its three degrees of freedom

as well as the high rotational velocities and large ranges of motion that punctuate its

involvement in functional tasks. Further confounding the description of the shoulder joint

are divergent marker placements, coordinate axis definitions and angle decompositions

(Rab et al., 2002).

For the humerus, the lateral and medial humeral epicondyles are referenced universally,

yet determination of the third point of anatomical relevance, typically the estimated

centre of GH joint rotation, is not standardised. While this poor consensus has

contributed to inconsistent modelling methods, small errors in defining the GH joint’s

centre are believed to influence angular data only minimally due the relative (greater)

length of the humerus (Anglin and Wyss, 2000). Even Wu et al. (2005) in their position

paper for the International Society of Biomechanics (ISB) prefer not to standardise the

definition of the centre of GH rotation, leaving it to the discretion of individual

researchers. In this course of studies, an approximation of the GH joint centre was

attained as described in Chapter 3.2.5.1.

54

Significantly, as the position of the humerus is achieved through simultaneous rotations

of the acromioclavicular, sternoclavicular and GH joints as well as scapulothoracic

motion, consideration must also be given to the roles of the clavicle and scapula (van der

Helm and Pronk, 1995). Theorists and clinicians agree that substantiation of clavicular

and scapular positions and rotations during dynamic GH joint motion would provide a

more complete understanding of the shoulder complex in functional activities. However,

accurate assessment of the clavicle’s dynamics is confounded by the requirement of

collinear marker placements and marker-associated skin movement (Wu et al., 2005).

Tracking of scapular motion necessitates skeletal pins, time-consuming palpation or

complex imaging techniques, and is often impractical in applied research settings (Rab et

al., 2002). Consequently, devoid of non-invasive instrumentation to monitor clavicle or

scapula position, researchers have tended to describe the humerus relative to the

thorax, thus creating a virtual thoraco-humeral joint (van der Helm and Pronk, 1995;

Anglin and Wyss, 2000). In this thesis, analyses of the shoulder joint kinetics associated

with the tennis serve has described upper arm position, and therefore GH joint motion,

in like kind (Chapter 3.2.7.1).

Clinically, all shoulder positions are referenced to the anatomical posture, considered

joint neutral or zero, as well as in three orthogonal planes (sagittal, frontal, or

transverse) or around a longitudinal axis of rotation (Doorenbosch et al., 2003). Motion

in any direction is therefore calculated with reference to the zero position in a

standardised sequence of rotations. Most commonly, the sequence of rotations described

by Grood and Suntay (1983), also known as Euler Z-X-Y decompostion, are used. This

sequence calculates flexion first, followed by abduction and finally axial rotation. While

satisfactory for describing shoulder joint movements in clinical settings, its validity in

biomechanically describing shoulder positions during dynamic functional tasks has been

questioned. In part, the genesis of this concern relates to when rotation occurs outside

one of the orthogonal planes (i.e. non-planar), whereby descriptions of shoulder position

become increasingly ambiguous. Failure to follow these conventions can also result in

differential descriptions of the same end point position of the arm (Doorenbosch et al.,

2003).

55

Furthermore, decomposition in three sequential angles contributes to the gimbal-lock

problem known to plague movement interpretation at the shoulder. Gimbal-lock, where

angles become ill-defined as axes coincide, occurs when the order of rotations are about

different axes and the second rotation in the sequence (i.e. abduction in Euler Z-X-Y

decomposition) approximates ±90º (Anglin and Wyss, 2000). So, in applying this

decomposition to describe the upper extremity joint motion that characterises the tennis

serve, gimbal-lock would occur as the humerus abducts to 90º, perpendicular to the

trunk. Conversely, with reference to the lower extremity, 90º of abduction or adduction,

for example at the knee, is anatomically impossible and the Euler Z-X-Y sequence is

considerably less likely to create undetermined positions.

Gimbal-lock is a feature of all Euler decomposition formats, yet recent evidence points

toward the selection of appropriate measurement procedures being task-dependent. In

this way, research has advanced techniques to describe 3D joint rotations with global or

spherical coordinate systems (Cheng, 2000; Doorenbosch et al., 2003). In these

systems, the three rotations of the upper arm relative to the trunk are described as

latitudes and longitiudes, with gimbal-lock restricted to the anatomical position of 0º,

and 180º (i.e. arm pointing upward). Accurate, unambiguous definitions of rotations are

also central to these systems, and the establishment of a plane of elevation, the amount

of elevation, and finally the axial rotation of the humerus are recommended (Figure 3.1-

3.4; Anglin and Wyss, 2000). To this end, the Standardisation and Technology

Committee of the ISB has proposed that the Euler rotation sequence Y-X-Y be used to

best compute motion of the humerus relative to the thorax (Wu et al., 2005; see Chapter

3.2.7.1).

Although clinically difficult to interpret, indications are that these global or spherical

systems coupled with Y-X-Y decompositions can more accurately record shoulder joint

kinematics (Cheng, 2000; Doorenbosch et al., 2003), especially when large amounts of

upper arm abduction are present, as in the tennis serve. Nonetheless, no study has

reported the efficacy of the Euler Z-X-Y and ISB recommended decompositions in

describing the shoulder joint motion of dynamic overhead actions similar to the serve.

Consequently, the documented preference for the ISB decomposition, while valuable as

56

a guide, remains largely unsubstantiated. For this reason, a short analysis was

undertaken to contrast the modelled angular data of the shoulder using both

decompositions (Figures 3.2 - 3.7). Visual inspection of selected pre- and post-impact

kinematic outputs confirmed the preferential use of the ISB decomposition in

consistently describing motion of the shoulder, particularly about its longitudinal axis (i.e.

internal and external rotation). In light of these observations and industry standard, the

ISB decomposition was used in the UWA model to describe shoulder joint motion, while

the Euler Z-X-Y sequence continued to be referenced in describing the movement about

all other joints.

57

A B

C D

F E

G

H

Figure 3.1. Frontal (left) and sagittal (right) illustration of the magnitude of shoulder joint plane of elevation (PoE), elevation (E) and internal-external rotation (IRER) describing four different attitudes of the upper arm. A and B – PoE: -190º, E: 90º, IRER: -85º; C and D - PoE: -140º, E:

150º, IRER: -70º; E and F – PoE: -195º, E: 30º, IRER: -45º; G and H – PoE: -120º, E: 90º, IRER: -50º 1.

1 Retrospective analysis and discussion of the magnitude of shoulder joint internal-external rotation – as related to the UWA marker set – is undertaken in Chapter 5.4.2 and Appendix A.

58

3.1.1 Analysis of Upper Arm Position with Euler Z-X-Y and ISB

Decompositions

To best appreciate the confounding influence of gimbal-lock, as well as the measures

previously undertaken to negotiate its effect, basic 3D rigid body dynamics need to be

further considered.

As abovementioned, coordinate systems consisting of three orthogonal axis define

segments in 3D motion, The subsequent elaboration of matrices allows the

determination of angles between parent and child coordinate systems. A rotation matrix

is a 3*3 matrix consisting of the cosines between the axes of the parent and child

coordinate systems, between which the rotation is defined. The nine variables of the 3*3

matrix contain three independent variables (or unknowns), which can be solved through

Euler angle decompositions (van der Helm, 1997). The euler angles can then be

interpreted as rotations of the child coordinate system around axes of the parent

coordinate system.

There are 12 Euler angle decompositions, Z-X-Y X-Y-Z Z-Y-X X-Z-Y Y-X-Z Y-Z-X Y-X-

Y Y-Z-Y Z-X-Z Z-Y-Z X-Y-X X-Z-X. Each angle decomposition produces different

angular kinetmatics such that standardised decompositions need to be applied in motion

analysis. Significantly, as alluded to in Chapter 3.1, gimbal-lock occurs in all Euler angle

decompositions, yet its form is affected by the decomposition or order of rotations

selected. That is, in rotation sequences involving repeated rotation about an axis (i.e. Y-

X-Y), it presents when the second rotation is at or near 0º or 180º. Alternatively, if

rotations are about different axes (i.e. Z-X-Y), angles become ill-defined as the second

rotated axis approximates ±90º.

In matrix mathematics the term Beta (β) is used to describe the calculation to determine

the first of the above-mentioned solveable independent variables. Beta, or more

specifically the cosine of β, is then used to help determine alpha (α) and gamma (γ). In

the Z-X-Y angle decomposition, which characterised the original UWA upper body model,

β represents abduction and adduction such that when the upper arm abducts/adducts to

90º, α (i.e. flexion-extension) and γ (i.e. humeral rotation) become indeterminate.

59

Further confounding any subsequent analyses are the mathematical procedures that

researchers or programmers employ in an effort to negotiate this irregularity.

To this end, the method used by Vicon (Oxford Metrics) Bodybuilder software is to

constrain β in the calculation of α and γ so that the cosine of β is never indeterminate.

Vicon (2002, p.102) describes that “in dynamic trials, if a segment rotates continuously,

BodyLanguage prevents discontinuities by keeping an internal record of rotations in a

form which does not suffer gimbal-lock”. Figure 3.2 contrasts Bodybuilder’s

determination of 3D shoulder angular displacement data with the same mathematically

computed angles (i.e. using the Z-X-Y decomposition devoid of Bodybuilder intervention)

during the forwardswing of a FS. The effects of gimbal-lock are clearly evidenced as the

shoulder abduction angle reaches 90º, while the ambiguity with which the Bodybuilder

intervention accesses its ‘internal records to prevent discontinuities’ in shoulder joint

flexion-extension and axial rotation is demonstrable. While this evidence alone may be

sufficient to support defining the order of rotations according to the ISB convention,

further verification of this preference is provided through the following comparisons

(Figures 3.3 – 3.7).

60

-200

-100

0

100

200

300

400

0 20 40 60 80

Swing (Temporally normalised)

Ang

le (D

egre

es)

100

Bodybuilder Flexion-ExtensionBodybuilder Abduction-AdductionBodybuilder Internal-External RotationMathematical Flexion-ExtensionMathematical Internal-External RotationMathematical Abduction-Adduction

Figure 3.2. Illustration of the a) mathematical effect of gimbal-lock as upper arm abduction reaches 90º, and b) how Bodybuilder uses its ‘internal record of rotations’ to avoid gimbal-lock.

Figures 3.3a and 3.3b highlight the variation in longitudinal shoulder joint rotation during

the swing phase of the FS, as computed by the Euler Z-X-Y and ISB conventions

respectively. Indeed the effects of the anti-flip code/intervention used by Bodybuilder to

calculate internal-external rotation as the upper arm abducts to 90º during the serve, can

again be observed (in Subjects 1, 2, 5, 6 and 10). It is also likely that the

aforementioned ambiguity of this ‘BodyLanguage’ contributed to the disparate angular

displacement data describing the same joint action portrayed in Figure 3.3. While some

kinematic variation can be expected between subjects, interpretation of Figure 3.3

alongside Figures 3.4a and 3.4b, which describe the group means and standard

deviations of shoulder joint longitudinal rotation and flexion-extension respectively,

further indicates that this convention inadequately describes 3D shoulder joint motion

during the serve.

61

-500

-400

-300

-200

-100

0

100

200

300

0 20 40 60 80 100


Ang

le (D

egre

es)

Subject 1

Subject 2

Subject 3

Subject 4

Subject 5

Subject 6

Subject 7

Subject 8

Subject 9

Subject 10

Figure 3.3. Shoulder joint internal (+) and external (-) rotation, during the swing phase of FSs (n=10) using the Euler Z-X-Y convention.

-250

-200

-150

-100

-50

0

50

100

150

0 20 40 60 80


Ang

le (D

egre

es) 100

Figure 3.4a. Mean shoulder joint internal (+) and external (-) rotation during the swing phase of a FS (n=10), using the Euler Z-X-Y convention (error bars represent sample standard deviation).

62

-150

-100

-50

0

50

100

150

200

250

0 20 40 60 80


Ang

le (D

egre

es)

100

Figure 3.4b. Mean shoulder joint flexion (+) and extension (-) during the swing phase of a FS (n=10), using the Euler Z-X-Y convention (error bars represent sample standard deviation).

Figures 3.5a and 3.5b, on the other hand, show a more consistent and representative

description of longitudinal shoulder joint rotation during subjects’ FS with the ISB Y-X-Y

convention. Devoid of the gimbal-lock problems that plague the Euler Z-X-Y

decomposition at 90º of shoulder joint abduction, this convention accurately describes

movement in the plane of elevation (Figure 3.6) as well as upper arm elevation (Figure

3.7). Consequently, the Y-X-Y convention was selected as the preferred means through

which to calculate 3D shoulder joint position in this thesis.

63

-500

-400

-300

-200

-100

0

100

200

300

0 20 40 60 80 100


Ang

le (D

egre

es)

Subject 1

Subject 2

Subject 3

Subject 4

Subject 5

Subject 6

Subject 7

Subject 8

Subject 9

Subject 10

Figure 3.5a. Shoulder joint longitudinal rotation, during the swing phase of a FS (n=10), using the ISB convention.

-140

-120

-100

-80

-60

-40

-20

00 20 40 60 80


Ang

le (D

egre

es)

100

Figure 3.5b. Mean shoulder joint longitudinal rotation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation).

64

-250

-200

-150

-100

-50

00 20 40 60 80


Ang

le (D

egre

es)

100

Figure 3.6. Shoulder joint plane of elevation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation).

0

20

40

60

80

100

120

140

0 20 40 60 80Swing (Temporally normalised)

Ang

le (D

egre

es)

100

Figure 3.7. Shoulder joint elevation during the swing phase of a FS (n=10), using the ISB convention (error bars represent sample standard deviation).

65

3.2 METHODS OF DATA COLLECTION

Data collection procedures that were consistent across all studies are detailed below.

The UWA marker set along with its subsequent use in coordinate system definitions and

the UWA model (a synthesis of previous models developed by the Biomechanics Group

at UWA) is also introduced. The protocols for completion of the dynamic service trials as

well as any procedures that were study-specific are provided in the respective chapters.

3.2.1 Subject Information

Up to twelve state to national representative calibre male tennis players (22.7±4.2

years), identified by three professional coaches as having high performance FSs and KSs,

were selected to participate in this course of studies. All players were competitors in the

West Australian state-pennant competition, while 50% currently or previously held a

professional ranking. Two, top internationally ranked male wheelchair players (24-29

years) also agreed to participate.

Participation was contingent on all subjects reading the Subject Information Sheet and

signing the Consent Form (Appendix C). Ethics were submitted, and approved by the

Ethics in Human Research Committee of the University of Western Australia (Appendix

B).

3.2.2 Overview of the UWA Model

The development of the UWA model was based on the ‘calibrated anatomical systems

technique’ (CAST; Cappozzo, 1984), which involves cluster marker sets and pointer

anatomical landmark calibration procedures to reduce the two largest sources of error in

3D motion analysis: skin movement artefact and incorrectly located critical anatomical

landmarks. By establishing bone embedded technical (TCS) and anatomical coordinate

systems (ACS) for each rigid body segment, the UWA model allows full body kinematics

to be be calculated and standard inverse dynamics to be used to determine joint

kinetics.

Definition of a biomechanical model of the skeletal system and determination of the

instantaneous orientation and position of orthogonal sets of axes considered embedded

66

in the rigid bony segments that comprise the model is required to reconstruct human

motion (Cappozzo et al., 1996). Typically, these orthogonal sets of axes are referred to

as TCSs. A minimum of three reconstructed markers per bony segment are required to

estimate the orientation matrices and position vectors of the TCSs relative to the

laboratory (global) coordinate system. These markers can be arbitrarily placed on the

skin surface, alleviating the need for markers on anatomically-significant and palpable

bony prominences, which may otherwise introduce skin movement artefact and become

obtrusive, during dynamic trials. However, the meaningfulness of the resultant

movement data is limited from morpoholgical and functional standpoints (Della Croce et

al., 2003). To provide for more edifying and repeatable data, the kinematic and kinetic

scalar information must be referenced to ACSs (Della Croce et al., 2003). Consequently,

anatomical landmarks must be defined, with their position vectors determined relative to

the relevant TCS. Where at least three anatomical landmarks exist per bony segment,

ACSs can be constructed and associated position vectors and orientation axes calculated

(Cappozzo, 1984; Della Croce et al., 1997). With knowledge of the position vectors and

orientation axes of the ACSs of two adjacent segments, joint kinematics and kinetics can

be estimated (Grood and Suntay, 1983; Woltring 1994).

In the UWA model, bilateral lower limb TCS were established for feet, thighs and lower

legs, while additional TCS were constructed for both humerus, forearms and hands of

the upper extremity. Technical coordinate systems were also formulated for the pelvis,

thorax, torso, head and racquet. Static calibration trials provided for the determination of

key anatomical landmarks in the bilateral lower leg, thigh, humerus and forearm TCSs

such that those landmarks could be reconstructed relative to the relevant TCS during

service trials. More accurate identification of elbow and knee epicondyles was facilitated

through the use of a pointer method in accordance with the CAST (Cappozzo, 1984).

Segmental ACSs were then derived during service trials to enable joint kinematics and

kinetics to be determined.

3.2.3 UWA Marker Set

To implement CAST, a compatible full-body marker set was designed. Subjects were

fitted with 62 individual retro-reflective markers, all 16mm in diameter, using non-

67

allergenic double-sided adhesive tape (Figure 3.8). As aforementioned, three markers

were required to reconstruct bone-representative TCSs. Constrained cluster methods,

characterised by three markers affixed to semi-malleable pieces of ‘T-bar’ shaped plastic

(Figure 3.9) or to rigid, moulded aluminium ‘T-Bars’ (Figure 3.10), facilitated this process

and minimised error associated with skin and soft tissue movement (Manal et al., 2000).

An additional five markers were attached to the racquet to define a TCS (Figure 3.8),

while three markers were also applied to the ball to assist with identification of racquet-

ball impact, and thus the events IMP and PC. Elaboration of all marker locations for both

static calibration and service trials is provided in Table 3.1 below.

During the service trials, 50 markers, inclusive of eight ‘T-bar’ clusters were affixed to

the subjects. The remaining 12 markers were used in the static calibration to identify the

position of key anatomical landmarks for the defintion of segmental ACSs. Markers

affixed to the medial and lateral malleoli of the ankles, anterior and posterior aspects of

the shoulder, and styloid processes of the ulna and radius were removed following the

static calibration. A spherical pointer was used to identify the left and right epicondlyes

of bilateral humerus and femur, outputing virtual markers at these locations as per the

method outlined in Chapter 3.2.4.

68

Figure 3.8. Marker positions of able-bodied players during static trials (anterior view, left; posterior view; right; refer to Table 3.1 for marker key).

Figure 3.9. Design of the semi-malleable ‘T-bar’ cluster.

Figure 3.10. Design of the aluminium ‘T-bar’ cluster.

69

Dynamic markers used in service trials to determine relative segment motion

Segment Marker Location

Upper body

Thorax C7 7th cervical vertebrae

T10 10th thoracic vertebrae

CLAV Clavicular notch

STRN Xyphoid process of the sternum

Head LFHD Left front head

LBHD Left back head

RFHD Right front head

RBHD Right back head

Right Shoulder RACR Right acromion process

Left Shoulder LACR Left acromion process

Right Upper Arm RUA1 Right superior upper arm cluster

RUA2 Right middle upper arm cluster

RUA3 Right inferior upper arm cluster

Left Upper Arm LUA1 Left superior upper arm cluster

LUA2 Left middle upper arm cluster

LUA3 Left inferior upper arm cluster

Right Forearm RFA1 Right superior forearm cluster

RFA2 Right middle forearm cluster

RFA3 Right inferior forearm cluster

Left Forearm LFA1 Left superior forearm cluster

LFA2 Left middle forearm cluster

LFA3 Left inferior forearm cluster

Right Hand RCAR Right carpal

RHNR Right dorsal radial knuckle

RHNU Right dorsal ulnar knuckle

Left Hand LCAR Left carpal

LHNR Left dorsal radial knuckle

LHNU Left dorsal ulnar knuckle

70

Racquet RRHD Right mid racquet head

LRHD Left mid racquet head

THT1 Right aspect of racquet throat

THT2 Left aspect of racquet throat

RTIP Racquet tip

Ball BAL1 Right ball

BAL2 Mid ball

BAL3 Left ball

Lower Body

Pelvis RPSIS Right posterior superior iliac spine

LPSIS Left posterior superior iliac spine

RASI Right anterior superior iliac spine

LASI Left anterior superior iliac spine

Left thigh LTH1 Left superior thigh cluster

LTH2 Left middle thigh cluster

LTH3 Left inferior thigh cluster

Right thigh RTH1 Right superior thigh cluster

RTH2 Right middle thigh cluster

RTH3 Right inferior thigh cluster

Left lower leg LTB1 Left superior tibia cluster

LTB2 Left middle tibia cluster

LTB3 Left inferior tibia cluster

Right lower leg RTB1 Right superior tibia cluster

RTB2 Right middle tibia cluster

RTB3 Right inferior tibia cluster

Left foot LCAL Left calcaneous

Left metatarsal 1st head (approximate) LMT1

Left metatarsal 5th head (approximate) LMT5

Right foot RCAL Right calcaneous

Right metatarsal 1st head (approximate) RMT1

Right metatarsal 5th head (approximate) RMT5

71

Static markers defining key antomical landmarks at the shoulder, wrist and ankle joints

Joint Marker Location

Shoulder RASH Right Anterior Shoulder

RPSH Right Posterior Shoulder

LASH Left Anterior Shoulder

LPSH Left Posterior Shoulder

Wrist RWRR Right Radial Wrist

RWRU Right Ulnar Wrist

LWRR Left Radial Wrist

LWRU Left Ulnar Wrist

Ankle RLMAL Right Lateral Mallelous

RMMAL Right Medial Mallelous

LLMAL Left Lateral Mallelous

LMMAL Left Medial Mallelous

Table 3.1. Upper- and lower-body retroreflective marker naming convention and locations.

3.2.3.1 Marker set used for wheelchair players

In the final study, performed with wheelchair players, markers were affixed to the same

upper body locations as for the able-bodied players (Figure 3.11). However, due to the

players’ seated position, the marker on the 10th thoracic vertebrae could not be placed

without compromising player comfort and potentially serve mechanics. No lower body

markers were used.

72

Figure 3.11. Marker positions of wheelchair players during the dynamic trials (anterior view; refer to Table 3.1 for marker key).

3.2.4 Pointer Method

A pointer wand with 16mm retroreflective markers was used to more accurately define

the bilateral position of the medial and lateral epicondyles of the elbows and knees. Each

joint required two static calibration trials. At the knee, with the subject standing upright,

the first trial involved the investigator positioning the tip of the pointer directly on the

most lateral point of the lateral femoral epicondyle. Accordingly, the second trial saw the

pointer located on the most medial aspect of the medial femoral epicondyle and the

positions of both femoral epicondyles were stored with respect to the TCS of the femoral

triad cluster. The same procedure was repeated at both elbows such that virtual humeral

epicondlyes were created and stored with respect to the TCS of the humeral triad

cluster. To this end, successful reconstruction of the five pointer markers, as well as, the

three markers from the relevant thigh or upper arm cluster was needed for epicondyle

virtual marker output. The UWA model code required pointer markers to be labelled as

m1 to m5, with markers m1 to m4 positioned on a spherical plane perpendicular to the

pointer shaft and the vector defined from m5 to the pointer origin (Figure 3.12). The

average position of markers m1-m4 was calculated as the pointer origin.

73

This pointer origin was utilised as the origin of three pointer coordinate systems, with

different orientations, which were determined using markers m1-m5. Three virtual ‘end

of pointer positions’ were calculated from each coordinate system so that, when

averaged, a final pointer tip position could be derived. This method allowed for

epicondylar locations to be expressed relative to the TCS of the related upper arm or

thigh cluster. These reconstructed positions relative to the appropriate cluster could then

be determined during the service trials.

Figure 3.12. Diagrammatic representation of the spherical pointer used to locate humeral and femoral epicondyles (Op represents the origin of the three independent coordinate systems

calculated using markers m1-m5).

3.2.5 Joint Centre Definitions

The following joint centres are required for ACS definitions. As they cannot be directly

palpated, their positions are estimated from repeatable external anatomical landmarks.

3.2.5.1 Shoulder

Estimation of the GH joint’s centre of rotation (SJC) was reported as the mean 3D

position of markers identifying the acromion process and situated on the anterior and

posterior aspects of the shoulder during a static trial (Figure 3.13). The position of the

GH joint centre was then referenced to the T-bar upper arm cluster (with its length

aligned with the long axis of the humerus), removing the need for the anterior and

posterior shoulder markers to remain affixed to the participant during service trials.

Continued tracking of the acromion marker was similarly redundant in the calculation of

74

the GH joint centre during dynamic trials. Nevertheless, the appendage of these markers

was necessary to compute variables such as shoulder alignment.

Figure 3.13. Placement (medial to the axillary folds) of anterior and posterior shoulder markers required for estimation of the GH joint centre.

3.2.5.2 Elbow

As aforementioned, pointer trials identified the medial and lateral epicondyles of the

elbows such that the midpoints of these virtual markers represented the elbow joint

centres (EJCs). During the dynamic trials, the position of these virtual elbow markers

was expressed relative to the T-bar clusters on the respective humerus.

3.2.5.3 Wrist

The wrist joint centres (WJCs) were defined by bisection of the markers placed on the

styloid processes of the ulna and radius. The position of each WJC was stored with

respect to the TCS of the forearm triad cluster placed directly superior to the WJC. The

ulna and radius styloid markers were removed for dynamic trials as they interferred with

normal service motion.

3.2.5.4 Hip

Hip joint centres (HJC) were determined relative to the pelvis ACS and estimated using

the Orthotrack (Motion Analysis Corp., Santa Rosa, CA) predictive technique regression

75

equation (Shea et al., 1997). This predictive equation is based upon displacements along

the anterior-posterior, medial-lateral and superior-inferior axes of the pelvis ACS. The

magnitudes of these displacements are calculated as fixed percentages of the distance

between the anterior superior iliac spines (ASIS) of the pelvis and are translated -21% in

the anterior-posterior direction, ± 32% in the medial-lateral direction and 34% in the

superior-inferior direction.

3.2.5.5 Knee

As explained in Chapter 3.2.2, four static calibration trials, using a spherical pointer to

locate the lateral and medial femoral epicondyles, were used to create virtual femoral

epicondlyes. The point midway between the medial (ME) and lateral (LE) epicondyle

represented the knee joint centre (KJC).

3.2.5.6 Ankle

The bisection of the markers placed on the medial malleoli of the tibiaes and lateral

malleoli of the fibulaes were calculated as the ankle joint centres (AJCs), and were

stored with respect to the lower leg TCSs during the dynamic trials.

3.2.6 Segment Coordinate Definitions

As previously outlined, TCSs for the upper arm, forearm, hand, thigh, lower leg, and foot

were determined from marker clusters on each of these segments. The use of the

aformentioned anatomical landmarks thus permitted ACSs and joint coordinate systems

to be defined relative to the TCSs throughout service trials. Anatomical coordinate

systems for the pelvis, thighs and lower legs were defined according to the procedure

put forward by Kadaba et al. (1989), while new ACSs were established for the upper

extremity, trunk, racquet and feet. To follow, the ACSs used to determine full body

kinematics and upper extermity kinetics from the UWA model are highlighted.

3.2.6.1 Head

With the head reconstructed to provide a point of somatic reference (i.e. no kinematic or

kinetic data was derived), only a TCS was defined. Markers were aligned along the

lateral aspects of the head, in the anterior and posterior borders, such that the midpoint

76

of all four markers represented the head’s origin (HO). A positive x axis ran from the

midpoint of the two posterior head markers to the HO, while a positive z axis followed

the midpoint between the two markers of the head’s left aspect to the midpoint of the

two markers on the head’s right aspect. Orthogonal to the x-z plane, and pointing

upward was a positive y axis.

3.2.6.2 Thorax and torso

Markers representing the clavicle (clavicular notch), sternum (xyphoid process), and C7

and T10 vertebraes were used to delineate the ACS of the trunk, both superiorly (i.e.

thorax) and inferiorly (i.e. torso). The midpoint between C7 and the clavicle defined the

thorax origin, THO, while the vector from C7 to the clavicle represented the throrax’s

positive x axis. A positive y axis ran along a line from the bisection of T10 and the

sternum to THO, while a positive z axis was to the orthogonal right of the x-y plane.

The torso’s origin, TOO, was defined as the midpoint between the markers positioned at

T10 and the sternum. The vector from T10 to sternum represented the torso’s positive x

axis, while a positive y axis ran from TOO to THO, and a positive z axis to the orthogonal

right of the x-y plane.

3.2.6.3 Upper arm

A positive z axis was calculated from the lateral to medial epicondyle of the the right arm

and medial to lateral epicondyle of the left arm. A positive y axis ran up the long axis of

the humerus, from the EJC to the SJC, while a positive x axis was anterior and ortogonal

to the z-y plane.

3.2.6.4 Forearm

The wrist joint centres represented the origins of both forearm segments. For left and

right forearms, positive z axes ran from the radial to ulna styloid processes and vice

versa respectively. Positive y axes were defined from WJCs to EJCs, with positive x axes

being anterior and orthogonal to the z-y planes.

77

3.2.6.5 Hand

The point midway between the markers placed on the radial and ulnar knuckles defined

the origin of the TCS of the hand. Positive z axes were defined from ulna to radial

knuckle and from radial to ulna knuckle for the left and right forearms respectively. The

line from origin to WJC represented the positive y axes, while positive x axes were

anterior and orthogonal to the z-y plane.

3.2.6.6 Pelvis

The ACS of the pelvis was defined according to an origin located midway between the

ASISs. With a positive z-axis along the line of left ASIS to right ASIS, the positive x-axis

ran from the sacrum marker to the origin. The y-axis was anterior and orthogonal to the

x-z plane.

3.2.6.7 Femur

The KJC represented the origin of the thigh ACS. The line passing through the KJC to the

HJC (positive being superior) defined the y-axis, while an orthogonal z-axis was parallel

to the plane defined by the ME and LE (positive pointing from left to right). A positive x-

axis was anterior and orthogonal to the y-z plane.

3.2.6.8 Lower leg

Determination of ACSs for the lower legs was achieved by using an origin at the AJCs,

with a positive y-axis calculated from the AJC to the KJC and a z-axis defined from the

MMAL to LMAL. As with the knee, a positive x-axis was anterior and orthogonal to the y-

z plane.

3.2.6.9 Foot

A customised foot rig was used to define the orientation of the foot segment, and thus

overcome the previously described large errors associated with palpating and placing

markers on anatomical landmarks of the feet (Della Croce et al., 1997). Four retro-

reflective markers attached to the customised rig defined the rig TCS (Figure 3.14), and

foot measurements were made with respect to this coordinate system. Subjects stood on

the customised rig in a comfortable stance with their heels against a small metal plate at

78

the rear of the rig. The long axis (x) of the foot segment was assumed parallel to the x–z

(horizontal) plane of the rig’s TCS, while it was further assumed that this axis was

rotated around the y-axis of the rig. Defined as the line bisecting the calcaneus and the

midpoint between the 2nd and 3rd metatarsal heads, the resultant foot abduction-

adduction was measured using a goniometer secured to the alignment rig. A subsequent

assumption that the foot was rotated in inversion/eversion about its x-axis permitted the

derivation of a rear-foot angle relative to the x-z plane of the rig and perpendicular to

the foot’s x-axis. This rear-foot inversion/eversion angle was measured using an

inclinometer (Dasco Pro Inc., Rochford IL). These rotational sequences complied with

the ISB standard (Wu and Cavanagh, 1995) and were used to define the foot ACS,

whereby two virtual markers were created and expressed relative to the foot TCS.

Together with the calcaneus marker, the two virtual markers formulated the ACS of the

foot. The calcaneus was the origin of the ACS of the foot.

Figure 3.14. Customised foot rig used to define the foot segment.

3.2.6.10 Racquet

The bisection of the markers attached to the left and right most lateral aspects of the

racquet head represented the racquet’s origin (Figure 3.8). For the right-handed player,

a positive y axis was defined from the racquet tip to the origin, a positive z axis from the

left to right lateral aspects and a positive x axis orthogonal and anterior to the y-z plane.

79

3.2.7 Definition of Joint Coordinate Systems

Having defined bone-representative ACSs, a third coordinate system – a joint coordinate

system, JCS – needed to be determined for the functional interpretation of 3D data.

Fundamentally, JCSs outline the order of rotations required to represent one coordinate

system with respect to another.

All joints, with the exception of the shoulder, corresponded to the standard Euler Z-X-Y

convention. The sequence that was adhered to throughout all motion was

flexion/extension, adduction/abduction and internal/external rotation of the moving

segment coordinate system with respect to the fixed segment coordinate system (Table

3.2). As aforementioned, the shoulder, or ‘thoracohumeral joint’, was described by a

globe system based on a more recent ISB convention (Wu et al., 2005).

Joint

Axis Pelvis Hip Knee Ankle Elbow Wrist

+ Anterior tilt Flexion Flexion Dorsiflexion Flexion Flexion Z

- Posterior tilt Extension Extension Plantarflexion Extension Extension

+ Upward pelvic

obliquity Adduction Adduction Adduction Adduction

Ulna

deviation X

- Downward pelvic

obliquity Abduction Abduction Abduction Abduction

Radial

deviation

+ Forward rotation Internal

rotation

Internal

rotation Inversion Pronation n/a

Y

- Backward

rotation

External

rotation

External

rotation Eversion Supination n/a

Table 3.2. Sequence and direction of rotations comprising joint coordinate systems.

3.2.7.1 Shoulder

The JCS of the shoulder, or theoretically thoraco-humeral, joint was defined in

accordance with the aforementioned ISB recommendations (i.e. with a Y-X-Y order of

rotation). The first y axis was fixed to the thorax and coincident with the y axis of the

ACS of the thorax (superior being positive). The second y axis represented axial rotation

80

of the humerus, as defined positively from EJC to SJC (i.e. the y axis of the humeral

ACS), while the positive x axis was fixed to the humerus and coincident with the positive

x axis of the upper arm.

3.2.8 Calculation and Interpretation of 3D Joint Kinetics

To best appreciate the 3D shoulder joint kinetics reported in subsequent Chapters,

elaboration of the computation processes used by both the UWA model and Oxford

Metrics Vicon software (i.e. Bodybuilder) is required.

3.2.8.1 Definitions of force, moment and power

The net force acting on a rigid body (i.e. segment) is the sum of the different forces

acting on it at any given time.

Moments describe the ability of forces to make a rigid body rotate, and are thus central

to the kinetic analysis of dynamic sports skills in which high-speed segment rotations are

key (i.e. the tennis serve). The product of the net force acting on a body and the

perpendicular distance separating its line of action (i.e. segment centre of mass) from an

anatomically significant reference point (i.e. joint centre), net moments can also be

divided into three components. Shoulder joint moments are calculated about the

shoulder joint centre and expressed in the coordinate system of the joint’s distal or child

segment (i.e. upper arm).

Joint power is the dot product of the moment vector and the angular velocity vector.

While the reported shoulder joint moments are calculated at the COM and expressed in

the coordinate system of the upper arm, the derivation of joint power requires that

moments be expressed in the coordinate systems of the proximal or parent (thorax)

segments. The shoulder joint moment vector is thus expressed in the ACS of the thorax.

In accordance with this convention, the angular velocity of the child (upper arm) is

calculated with respect to the parent (thorax) ACS (Craig, 2005). Although standard

practice in past 3D biomechanical analyses, particularly of lower limb joint motion

(Bellchamber and van den Bogert, 2000; Ferber et al., 2003), commensurate expression

81

of 3D shoulder joint angular velocities and power confounds the functional interpretation

of shoulder joint kinetics, as discussed in Chapter 3.2.8.3.

Power = [Mx,My,Mz] . [ωx,ωy,ωz]

= Mx.ωx + My.ωy + Mz.ωz

Mathematically, it is thus a scalar quantity, and has been used extensively in clinical

biomechanics to describe the rate of change of mechanical energy at a joint. Recently

however, debate has proliferated as to whether each power term in the dot product

equation would provide more meaningful information if left separate to determine

changes in kinetic energy for motion along its respective reference axis (Buczek, 2006).

While still a source of some contention, there is growing support for the use of this

approach in understanding joint motion. That is, with the calculation of individual

reference axis joint power terms considered to be functionally and increasingly relevant

to dynamic motion (Buczek, 2006; van den Bogert, 2006), this approach was followed in

the analyses that thesis has undertaken into the shoulder joint kinetics in the tennis

serve.

3.2.8.2 Bodybuilder computation of 3D joint kinetics

Bodybuilder uses a ‘reaction’ function to group together a force and a moment with the

moment reference point – around which the force is deemed to act – as a single

convenient unit. Calculations hard-coded into the software then allow for the output to

be split into three components that correspond to resultant points of application, forces

and moments.

3.2.8.3 Functional interpretation of force, moment and power

At the shoulder, forces are thus expressed in the distal coordinate system of the

shoulder joint, or in other words, the upper arm (child) segment. The TCS and axes of

the upper arm are defined as outlined above. Distractive (-) and compressive (+) forces

act along the long (y) axis of the upper arm, while the range of upper arm internal-

external rotation (and abduction) common to the serve ensures that forces acting along

both the x and z axes are interpreted at discrete points to accurately determine anterior

82

and posterior shoulder joint force. For example, anterior and posterior forces would act

in the direction of the x axis of the upper arm if a player assumed the anatomical

position. However, near positions of maximum upper arm external rotation during the

serve, the forces acting in the direction of the z axis of the upper arm would better

represent anterior and posterior shoulder joint force. Although not reported in this

thesis, the magnitude of forearm pronation in the serve similarly confounds the

interpretation of elbow joint forces.

Internal (-) and external (+) rotation moments are computed about the longitudinal (y)

axis of the upper arm. As with the interpretation of shoulder joint forces, the range of

upper arm internal-external rotation (and abduction) necessitates that moments

calculated about the x or z axis be considered selectively (i.e. in accordance with upper

arm attitude) to best analyse abduction (+) / adduction (-) and/or flexion (-) / extension

(+) moments at the shoulder.

As alluded to above, the moment and angular velocity vector transformations necessary

for the derivation of joint power are also required for the calculation of individual joint

power terms, introducing some challenges with respect to the functional interpretation of

3D shoulder joint angular velocities and power. For example, when the upper arm

abducts and flexes overhead, the y axes of the thorax and upper arm are orientated

differentially (i.e. the positive y axes point in opposite directions) such that an upper arm

internal rotation angular velocity is indicative of counter-clockwise rotation in the

coordinate system of the child (i.e. upper arm) but clockwise rotation in the coordinate

system of the parent (i.e. thorax). Unfortunately, during the execution of the high

performance tennis serve, players’ upper arms are rarely flexed and abducted directly

overhead. That is, with shoulder joint abduction angles known to approximate 110º

(Fleisig et al., 2003) at impact, the y axes of the thorax and upper arm are not

coincident. Indeed, in this scenario, an upper arm internal rotation angular velocity in

the coordinate system of the child would be represented as part upper arm external

rotation and part upper arm extension in the coordinate system of the parent.

Subsequent interpretation of shoulder joint power through the modelled internal-external

rotation moments, which are expressed in the coordinate system of the child (upper

83

arm), alongside the modelled internal-external rotation angular velocities, which are

expressed in the coordinate system of the parent (thorax), therefore becomes notionally

difficult.

A mode of expression that was considered retrospectively was to calculate and interpret

all joint kinetics and angular velocities in a shoulder joint coordinate system. While axes

would not be orthogonal and thus limit the absolute validity of kinetic expression, data

may be more meaningful from a functional standpoint. Preliminary work was successful

in determining kinetics and angular velocities in the JCS of the elbow; however similar

vector transformation into the Y-X-Y decomposition used to describe shoulder joint

kinematics was beyond the scope of this thesis. So, in spite of its interpretive challenges,

all joint moments are expressed in the coordinate system of the child segment, while

joint angular velocities are expressed in the coordinate system of the parent segment. At

the shoulder, the product of an internal rotation moment of the upper arm in the child’s

coordinate system and an external rotation angular velocity of the upper arm expressed

in the coordinate system of the parent (i.e. thorax) would then represent a positive joint

power term. Significantly, in acknowledging the limitations of this approach, the ensuing

chapters will exhibit caution in discussing joint power in detail. The comparative nature

of the research questions in this thesis will nevertheless help to ensure that the validity

of any contrasts made with respect to shoulder joint angular velocities.

3.2.9 Data Collection

Participants were guided through a standardised eight minute physical warm-up that

included a variety of tennis-specific movement patterns, dynamic stretches and/or

simulated service swings. They were then provided time to familiarise themselves with

the testing environment and hit an appropriate number of practice serves such that they

were ready to commence data analysis, serving with maximal effort.

Players were instructed to hit specific serves (FS, KS, FU, FB, ARM, WKS, WFS), with

maximal effort, to a 1x1 metre target area bordering the ‘T’ of the first service box (see

Figure 3.15). Up to five successful executions were required for each serve type or

84

technique. For serves to be considered successful, they had to satisfy the criterion

provided within the specific serve definitions detailed in Chapter 1.

Figure 3.15. Schematic of the testing environment (left) and camera positions (right).

All players used the same 0.690m, 0.380kg Wilson 6.0 Pro Staff racquet, whose inertial

parameters were calculated using procedures outlined by Brody et al. (2002; Appendix

E). Although the Wilson 6.0 was not the preferred racquet of every participant, all

players expressed familiarity with the racquet (i.e. had practiced with it) and indicated

that they felt comfortable with its weight, balance and stiffness. Pertinent subject

anthropometry was documented (Appendix D), while the segmental masses and

moments of inertia required for kinetic analysis were provided from data reported by de

Leva (1996) and Clauser et al. (1969).

To facilitate interpretation of results, certain data collection and analysis procedures will

be reinforced in the ensuing Chapters, while other protocols (i.e. number of subjects,

serves performed and statistical analysis) that are specific to the individual studies that

comprise this thesis will also be outlined where relevant.

3.2.10 Data Reduction

Two-dimensional data captured by each of the 12 cameras (operating at 250Hz) was

reconstructed into 3D trajectories in Vicon Workstation software (Oxford Metrics, UK).

Each trial of interest was visually inspected to eliminate random marker movement not

congruent with the motion being performed. Incorrect marker identification by the Vicon

(Oxford Metrics, UK) tracking system and random reflections momentarily registering as

85

‘phantom markers’ can manifest, and were removed to minimise measurement error. A

quintic spline with an optimal MSE (25; as determined using residual analysis in Chapter

4) was applied to the data such that the UWA upper and lower body models could then

be applied for the calculation of all relevant kinematic and kinetic data. Kinematic and

kinetic data were output in the ASCI text file format. Customized Matlab software

allowed for select continuous variables to be normalised (i.e. to 100 data points) for pre-

determined phases that facilitated direct comparison of the temporal sequencing

between players and serves.

86

CHAPTER 4: QUANTIFICATION OF DATA TREATMENT AND ANALYSIS

TECHNIQUES OF THE TENNIS SERVE

4.1 INTRODUCTION

Quantitative kinematic and kinetic analyses are gaining recognition as valuable tools in

the objective assessment of sports skills. In tennis, their widespread acceptance has

been hindered by the time lag associated with and the appropriateness of subsequent

feedback. The questionable reliability of these measures has also inhibited their seamless

integration into many elite tennis programs.

For higher-order kinematics of sporting movements to be represented accurately,

appropriate filtering or smoothing technique is all-important (Giakis et al., 1998). So too,

is the selection of a degree of filtering or cut off frequency that will optimally remove

noise associated with the movement’s kinematic representation (Challis, 1999). The

high-speed segment rotations and racquet-ball impacts in tennis exacerbate these

challenges (Knudson and Bahamonde, 2001; Knudson, 2005).

For the most part, biomechanical studies of tennis stroke production have assumed that

one superior performance characterises a player’s movement coordination for that

stroke. Similar procedures have been followed in the motion analyses of other dynamic,

sporting skills. However, the anomalies underlying this assumption are well documented

(Bates et al., 1992; Mullineaux et al., 2001) and should lead to quantification of

movement repeatability in sporting contexts becoming a methodological norm.

Kadaba et al. (1989) investigated the repeatability of gait in healthy subjects and

suggested that significant clinical decisions may be based on a single trial. More common

are preferences for studies to consider at least three trials in an effort to provide for

more representative and valid data (Mullineaux et al., 2001). In calculating reliable

racquet kinematics near impact of the advanced forehand, Knudson and Blackwell

(2005) and Knudson (2005) recently opted to average data from 5 and 10 trials

respectively. Previously, Knudson (1990) had used the same stroke and similar playing

population to demonstrate that upper arm displacement data at impact, across trials,

87

was quite consistent. Interestingly, this was achieved with significant variability in the

angular velocities and accelerations of the rotating upper limb during the forwardswing.

This inconsistent patterning of higher-order kinematic variables, and presumably the

resulting joint kinetics, illustrates the need to ascertain the repeatability of all sporting

movements that involve high-speed segment rotations.

The serve, widely regarded the most important stroke in tennis, is also the game’s only

closed skill; devoid of the variable incoming ball speeds, heights and spins that can

affect the game’s other strokes. Substantiating the repeatability of the high performance

FS and SS, and therefore the minimum number of times a coach should observe them

for critical evaluation will assist these professionals in their daily work and

simultaneously provide for more concise future research.

The hypotheses for this study are:

1. Quintic splines using a MSE similar to that used to represent the true signal of other

high-speed sports motions (i.e. cricket bowling MSE 23) will be suitable for the analysis

of the tennis serve;

2. Smoothing through impact in the high performance serve does not compromise the

representativeness or accuracy of related kinematic and kinetic data;



4.2 METHODOLOGY OF DATA TREATMENT

The following section will be presented in three parts. First, the subjects used for data

collection will be described. The methods used to ascertain the favoured data treatment

techniques to describe the tennis serve will then be detailed. An outline of the

procedures used to establish the repeatability of the serve will follow.

4.2.1 Subject Preparation and Performance

Following an appropriate warm-up, two male, full-time professional tennis players (mean

age: 20 years; mass: 85 kilograms) were instructed to hit, with maximal effort, five

successful FS and five successful KS to a 1x1 metre target area bordering the ‘T’ of the

88

first service box. For serves to be considered successful, they had to satisfy the FS and

KS criteria defined in Chapter 1. The players along with two high performance coaches

agreed to the quality ranking of these serves, so that they could be analysed in

sequence. In turn, to answer this study’s hypotheses, the following serves were

analysed:

− Consistent with the preference of Mullineaux and colleagues (2001), data from both

subjects’ three highest quality FSs and KSs were used to establish the most

appropriate MSE.

− The highest quality FSs of the two professionally ranked players were treated using

five different data treatment techniques near impact.

− Serve repeatability was ascertained by analysis of the five successful FSs and KSs as

well as the three highest quality FSs and KSs, as performed by both subjects.

Both players used the same 0.690m, 0.380kg Wilson 6.0 Pro Staff racquet described in

Chapter 3. For kinetic analysis, segmental masses and moments of inertia were provided

from data reported by de Leva (1996) and Clauser et al. (1969). The 3D serve motions

were recorded using a 12-camera 250 Hz, Vicon motion analysis system (Oxford Metrics

Inc.), employing the customised UWA full-body marker set outlined in Chapter 3. Static

calibration trials were performed to determine key anatomical landmarks in relevant

TCSs such that those landmarks could be reconstructed during the service trials, as

detailed in Chapter 3.2.2. The raw data were then filtered (Woltring filter/quintic spline)

at different (see Chapter 4.2.2) or best frequencies, before being modelled with UWA’s

customised lower and upper body models.

4.2.2 Selection of the MSE for Data Smoothing

Obtaining a best approximation of the true signal, or in other words, the movement

performed, requires optimal filtering. Of the various approaches that exist, the residual

analysis proposed by Winter (1990) is the most widely implemented; yet concerns

relating to its accuracy and methodological validity are commonly expressed (Yu et al.,

1999). While research has put forward alternative procedures, it has also conceded that

these protocols may be data-set specific (Yu et al., 1999) and derivative specific (van

89

den Bogert & de Koning, 1996). A residual analysis, which involves raw data being

filtered at different frequencies or MSE values, and the determination of residuals

between the filtered and raw data at those MSEs (Winter, 1990), was therefore used to

establish the most appropriate level of filtering for the high performance tennis serve.

The procedure, adapted from Winter (1990), was applied to specific variables considered

important to shoulder joint loading in the serve (Elliott et al., 2003).

The raw data from the three highest quality FSs and KSs of both professional players

were filtered (Woltring filter/quintic spline) at a comprehensive range of frequencies (i.e.

mean square error, MSE: 10, 15, 20, 25, 30, 35, 50, 65, 80 and 120 Hz), prior to being

modelled as detailed above. The subsequent residual procedure that provided for

authentication of the most suitable MSE for each loading variable is highlighted in

Figures 4.1 and 4.2. That is, the point at which the curve deviated from linearity was

considered to occur when R2 < 0.95. A trendline was then extended to the Y axis, where

a line - parallel to the X axis - was serially projected from this intersection back to the

curve. Finally, a perpendicular line was dropped to the X axis to estimate the most

appropriate MSE for each variable of interest (Winter, 1990).

Average MSEs for each kinematic and kinetic variable as well as gross averages for all

kinematic and kinetic variables were then calculated. Visual inspection of the

differentially filtered data helped to confirm the selection of a most appropriate MSE

(Figures 4.1 and 4.2). This method of visual inspection was recently favoured by Hunter

et al. (2004) and Portus (2006) in the assessment of the reliability of the sprinting

mechanics and cricket bowling respectively.

4.2.2.1 Results

Tables 4.1 and 4.2 highlight the most suitable (mean) MSEs for the different kinematic

and kinetic variables reported to affect shoulder joint kinetics in the serve. Derived from

residual analyses, the gross mean for all of the kinematic MSEs listed in Table 4.1 was

28.2 ± 3.2, while the mean MSE for kinetic variables (Table 4.2) was 21.6 ± 1.2. A

resultant average MSE of 24.9 was derived from these gross kinematic and kinetic

means.

90

Kinematic variable FS average (n=6) KS average (n=6) All serve average

(n=12)

Vertical trajectory of right hip

joint centre 29 29 29

Forward trajectory of racquet

tip 29 27 28

Separation angle 27 23 25

Shoulder joint internal-

external rotation angle 34 32 33

Upper arm - thorax elevation

angle 29 23 26

Table 4.1. A summary of the mean most appropriate MSEs for selected kinematic variables in the three FS and three KS, as performed by two professional players.

Kinetic variable FS average (n=6) KS average (n=6) All serve average

(n=12)

Anterior-posterior shoulder

joint force 20 22 21

Shoulder joint internal-

external rotation moment 23 21 22

Table 4.2. A summary of the mean most appropriate MSEs for selected kinetic variables in the three FS and three KS, as performed by two professional players.

Figures 4.1 and 4.2 provide examples of the residual analyses undertaken to determine

the most appropriate MSEs for the horizontal racquet velocity and shoulder joint internal-

external rotation moment generated during the FS of one subject. Mean square errors of

20-25 can be interpreted such that signal noise would be expected to increase with the

selection of less stringent MSEs.

91

R2 = 0.95

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

90.00

100.00

110.00

120.00

Woltring MSE

Inve

rse

Res

idua

l

Figure 4.1. Example of the residual analysis performed to ascertain the most appropriate MSE for horizontal racquet velocity.

R2 = 0.95

0.015

0.017

0.019

0.021

0.023

0.025

0.027

0.029

0 10 20 30 40 50 60 70 80 90 100

110

120

Woltring MSE

Inve

rse

Res

idua

l

Figure 4.2. Example of the residual analysis performed to ascertain the most appropriate MSE for the right shoulder internal-external rotation moment.

92

Figure 4.3 sees the upper arm - thorax elevation angle used to demonstrate the effect of

different levels of filtering on the raw data. Plotted throughout the forwardswing phase

of one subject’s FS, the outputs filtered at an MSE of 25 are shown to accurately

represent the true signal, and therefore the upper arm and trunk movement performed.

Visual inspection of the shoulder joint internal rotation moment developed by a

professional player during the FS forwardswing (Figure 4.4) confirms that raw data

filtered at an MSE of 25 appears to authentically describe the shoulder joint kinetics that

characterise elite serve performance.

-90

-80

-70

-60

-50

-40

-301 11 21 31 41 51 61

Frame Number

Ang

le (D

egre

es)

RAW

MSE 25

MSE 10

MSE 50

Figure 4.3. Effect of different MSEs on the upper arm - thorax elevation angle during the forwardswing of a professional player’s FS.

93

-100

-80

-60

-40

-20

0

20

40

1 6 11 16 21 26 31 36

Frame Number

Mom

ent (

Nm

)

RAW

MSE 10

MSE 25

MSE 50

Figure 4.4. Effect of different MSEs on the shoulder joint internal rotation moment generated during the forwardswing of a professional player’s FS.

4.2.3 Assessment and Selection of Data Treatment – Negotiating Impact

In investigating the effect of different filtering procedures on forehand kinematics when

impact and post-impact data were removed, Knudson and Bahamonde (2001) and

Knudson (2005) demonstrated that extrapolated, then smoothed data were more

accurate than when no extrapolation method was used. However, Vint and Hinrichs

(1996) were able to show that use of the quintic spline without extrapolation was

superior to that of other filtering techniques coupled with linear extrapolation. The

quintic spline has been vindicated for data treatment in several tennis studies

(Bahamonde, 2000). Nevertheless, with a view to attaining the most representative

kinematic and kinetic data near impact, the effects of five different data treatment

conditions were compared. Impact is reported to last 0.004s (Knudson and Bahamonde,

2001). Conservative visual inspection of all data suggested that any effect on segmental

and racquet rotations dissipated 0.020s post racquet and ball collision for both subjects.

Deletion and interpolation of data for periods of 0.024s (or 6 frames) around impact

eliminated any misrepresentative movement signal associated with the impact event,

and was consistent with the time intervals used by Knudson and Bahamonde (2001) to

94

evaluate the efficacy of five-point linear extrapolation and polynomial extrapolation in

modelling forehand impacts. The data describing the FSs of the two professional players

were treated as follows:

1. Data were smoothed through impact with a Woltring filter (MSE = 25, as

determined by the aforementioned residual analyses);

2. Trials were cropped and digitised to the frame preceding impact (-0.004s). The

data were then smoothed using a Woltring filter (MSE = 25);

3. Data from three frames either side of impact were cropped and deleted. All

trajectories were then interpolated by Vicon’s cubic spline “fill gaps” function,

before being smoothed as in condition 1;

4. Data from one frame pre-impact and five frames post-impact were cropped and

deleted. All trajectories were then interpolated by Vicon’s cubic spline “fill gaps”

function, before being smoothed as in condition 1;

5. Data from five frames pre-impact and one frame post-impact were cropped and

deleted. All trajectories were then interpolated by Vicon’s cubic spline “fill gaps”

function, before being smoothed as in condition 1.

Plotted, modelled trajectories as well as kinematic and kinetic outputs deemed important

to shoulder joint loading in the serve were visually compared between conditions, from

MKF to IMP. As research has largely associated endpoint error with cropped and

smoothed pre-impact data, only data pertaining to this phase were analysed.

Furthermore, with comparable 3D racquet velocities reported to characterise first and

second serves (Chow et al., 2003a), only FSs were analysed. The treatment procedure

whose data were considered to best represent the actual serve movement, minimising

the effects of impact-induced oversmoothing or end-point error, was subsequently

selected.

4.2.3.1 Results

Figures 4.5 and 4.6 contrast the effect of different data treatment and smoothing

procedures on selected shoulder joint kinematics during the swing phase of one of the

professional player’s series of FSs. These differential effects were representative of those

95

observed in all modelled data. More specifically, Figure 4.5 presents data describing the

internal-external rotation angle about the racquet arm’s shoulder joint, while Figure 4.6

depicts the internal-external rotation moment about the same joint. Although exclusive

consideration of pre-impact data would appear to introduce some endpoint error,

indications are that smoothing through impact may not compromise the true

representation of racquet and segment rotation. Indeed, whether data were just

smoothed through impact or differentially cropped either side of impact and then

smoothed, the effect on the representation of the shoulder joint internal-external angle

was minimal. More variation existed in the modelled shoulder joint internal-external

moment, especially when data were cropped to the last frame pre-impact, whereby the

acting moment changed from internal rotation to external rotation as the upper arm and

racquet approached impact.

-45

-40

-35

-30

-25

-20

-15

-100 20 40 60 80

Swing (Temporally Normalised)

Ang

le (d

egre

es)

100

5 pre 1 post1 pre 5 post6 framesPre impactThru imp

Figure 4.5. Effect of data treatment and smoothing procedures on the internal-external rotation angle of the racquet arm’s shoulder joint (as calculated by the Z-X-Y Euler decomposition) during

the swing phase of a professional player’s FS.

96

-40

-30

-20

-10

0

10

20

0 20 40 60 80

Forwardswing (Temporally Normalised)

Mom

ent (

Nm

)

100

5 pre 1 post1 pre 5 post6 framesPre impThru imp

Figure 4.6. Effect of data treatment and smoothing procedures on the internal-external rotation moment of the racquet arm’s shoulder joint during the swing phase of a professional player’s FS.

4.2.4 Determining the Repeatability of the Tennis Serve

Research indicates that healthy subjects perform specific locomotive tasks repeatably

both within and between test days (Kadaba et al., 1989). This occurs in spite of an

individual’s inherent physiological and mechanical variability and factors such as test day

residual/induced fatigue and hydration status, which have been suggested to further

magnify this variance, especially where high velocity movement is desired (Portus et al.,

2000). The tennis serve is indeed executed at a high velocity and quantification of the

repeatability of the high performance serve motion is most warranted. To date,

kinematic analyses of the forehand by Knudson (1990; 2005) remain the only efforts to

validate intra-subject tennis stroke variability.

The repeatability of locomotor tasks is commonly evaluated through coefficients of

multiple correlations (CMCs). This method was chosen to determine the repeatability of

the high performance FS and KS, with an r ≥ 0.90 indicating that continuous pre- and

post-impact kinematics and kinetics (time normalised to 101 points) were of acceptable

similarity between trials. Data were collected as outlined in Chapter 4.2.2, and then

97

98

cropped, interpolated and filtered at the most appropriate MSE (25; as in condition 4) as

determined by the aforementioned data treatment analyses.

4.2.4.1 Results

The CMCs for specific mechanical variables of both players across the best three and all

five trials are presented in Tables 4.3 and 4.4. With CMCs > 0.90 indicative of strong,

positive correlations (Yu et al., 1997), the results in Table 4.3 suggest that the tennis

serve, when executed by high performance players, is generally a highly repeatable skill.

Select displacement data and most kinetic variables of interest all recorded CMCs > 0.90,

irrespective of the serve hit or number of trials (three or five) analysed. A point

illustrated by the fact that the lowest CMC recorded for variables such as knee extension

and anterior-posterior shear force at the shoulder was 0.93. Table 4.4 highlights some

variation in the patterning of post-impact shoulder joint kinetics across players. That is,

while relatively consistent for Player 1’s FS and KS (Figure 4.7), the shoulder joint

internal rotation moment (CMC ≈ 0.61-0.87) and anterior force (CMC ≈ 0.79-0.98)

generated by Player 2 are more irregular.

Kinematics Kinetics

Knee Extension (º)

Angular velocity of

lateral trunk flexion

(rad.s-1)

Internal rotation

shoulder moment

(Nm)

Anterior shoulder

joint force (N)

Elbow valgus force

(N)

Serve type FS KS FS KS FS KS FS KS FS KS

Subject 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2

Five 0.93 0.98 0.94 0.99 0.94 0.97 0.71 0.78 0.97 0.93 0.91 0.86 0.99 0.98 0.93 0.97 0.99 0.94 0.90 0.98 CMC Sample

(n) Three 0.94 0.97 0.94 0.99 0.96 0.99 0.73 0.75 0.98 0.98 0.98 0.79 0.98 0.98 0.96 0.96 0.99 0.93 0.96 0.99

Table 4.3. Coefficients of Multiple Correlations (CMC) of selected kinematic and kinetic variables during the forwardswing of the FS and KS.

Internal-external rotation

shoulder moment (Nm)

Anterior shoulder joint force

(N)

Serve type FS KS FS KS

Subject 1 2 1 2 1 2 1 2

Five 0.95 0.87 0.89 0.61 0.96 0.79 0.98 0.98 CMC

Sample (n) Three 0.95 0.86 0.88 n/a 0.96 0.89 0.97 0.97

Table 4.4. Coefficients of Multiple Correlations (CMC) of selected shoulder kinetic variables post-impact.

Figure 4.7 contrasts the shoulder joint internal rotation moments generated during the

forwardswings of Players 1’s highest quality FS, mean three highest quality FSs, and

mean five highest quality FSs. Irrespective of the number of serves analysed, minimal

variation characterises the magnitude and shape of the pre-impact shoulder joint internal

rotation moment outputs. This provides pictorial evidence of the high CMCs (0.93-0.98)

recorded for pre-impact shoulder joint internal rotation moments in the FS, and

indicates that one successful FS execution may in fact provide some

representative mechanical data, while more conservative approaches would use

mean kinematic and kinetic data from three to five successful executions.

-35

-30

101

-25

-20

-15

-10

-5

01 21 41 61 81

Mom

ent (

Nm

)

Forwardswing (Temporally normalised)

Highest quality FS

Mean 3 highest quality FSs

Mean 5 highest quality FSs

Figure 4.7. Comparison of the shoulder joint internal rotation moment during the forwardswing of a professional player’s highest quality FS, mean three highest quality FSs, and mean five highest

quality FSs.

4.3 DISCUSSION

Valid and reliable mechanical descriptions of sports skills necessitate that an optimal

amount of filtering be applied to remove signal noise associated with their kinematic

representation. In like kind, to most accurately describe the mechanics of skills involving

100

impacts, such as the tennis serve, data must be smoothed to minimise the effects of

impact or end-point error if only pre-impact data are considered. Furthermore,

comprehensive and accurate skill analysis or enhancement may require that multiple

executions need to be observed or analysed.

4.3.1 MSE for Best Representation of Tennis Serve Motion

Research hypothesis:

1. Quintic splines using a MSE similar to that used to represent the true signal of other

high-speed sports motions (i.e. cricket bowling MSE 23) will be suitable for the analysis

of the tennis serve.

The kinematic and kinetic analysis of human movement are based on the calculations of

derivatives from sampled displacement data. Signal noise that contaminates this

displacement data can confound these calculations and misrepresent the movement

performed. The selection of an appropriate level of filtering to remove unwanted

distortions is therefore considered essential to most human movement research.

Most papers to have kinematically described tennis stroke production have filtered their

raw data using routines with set MSEs (Bahamonde and Knudson, 2001; Reid and Elliott,

2002; Knudson, 2005). While largely considered a reliable means of data treatment,

these routines do not account for variations in the quality or type of the data (Giakis and

Baltzopoulos, 1997; Challis, 1999). So, prior to assessing the repeatability of the high

performance FS and KS, an appropriate or best level of filtering should first be

substantiated.

The most appropriate MSEs for selected kinematic and kinetic variables considered

important to shoulder joint loading in the serve were reported in Tables 4.1 and 4.2. The

mean MSE for the selected kinematic variables was 28.2 ± 3.2, while the mean kinetic

MSE was 21.6 ± 1.2. As these MSE values were reasonably similar, their mid-point (i.e.

28.2+21.6/2 = 24.9 or ≈ 25) was accepted as the most appropriate MSE for use with

the Woltring filter in analysis of the high performance FS and KS. Although van den

Bogert and de Koning (1996) recommended that raw data be filtered at different

101

frequencies depending on the outputs (i.e. kinematic or kinetic) of interest, the

homogeneity of the above kinematic and kinetic MSEs was considered to justify this

approach. Indeed the resultant MSE (25) compares favourably to the MSE of 23 that

Portus (2006) visually inspected as optimal for upper-extremity and torso kinematics in

the elite overarm cricket bowl. It is important to note however, that while residual

analyses performed on other high performance tennis strokes may be expected to reveal

similar MSEs, variation in the ability levels of the subjects as well as analysis systems

and data collection procedures would likely produce different results.

Hypothesis determination:

1. Sampled displacement data of the tennis serve can be smoothed with quintic

splines using MSEs near 25 to accurately calculate serve kinematics and kinetics.

4.3.2 Assessment and Selection of Data Treatment – Negotiating Impact


2. Smoothing through impact in the high performance serve does not compromise the

representativeness or accuracy of shoulder kinematic and kinetic pre-impact data.

The complications associated with reporting accurate kinematic and kinetic data near

impacts, like those that occur when striking a tennis ball, are well documented (Knudson

and Bahamonde, 2001; Knudson, 2005). Some investigations into the mechanics of

tennis stroke production have failed to account for the effect of smoothing the large

changes in segment displacement near impact, likely oversmoothing the data as a result

(Ariel and Braden, 1979; Groppel and Nirschl, 1986; Elliott et al., 1989; Sprigings et al.,

1994; Elliott and Christmass, 1995; Takahashi et al., 1996). Ramifications are such that

the smoothed points throughout the data set, which are assumed to represent the actual

movement (Knudson and Bahamonde, 2001), are affected (Woltring, 1985; Giakis et al.,

1998). Efforts to account for this anomaly by only considering pre-impact data are

confounded by the difficulty in identifying the precise timing of impact, as well as

endpoint error caused by data smoothing (Knudson and Bahamonde, 2001). Subsequent

research evaluating the merits of different extrapolation and smoothing techniques in

providing for representative kinematic data near the end of, and throughout data sets,

102

has therefore proliferated (Vint and Hinrichs, 1996; Giakis et al., 1998; Knudson and

Bahamonde, 2001). Indeed Knudson and Bahamonde (2001) used data describing the

angular displacement and velocity of the wrist joint near impact in the tennis forehand to

analyse the effect of endpoint conditions in tennis.

To obtain accurate displacement and velocity data at impact, Knudson and Bahamonde

(2001) demonstrated that the quintic spline coupled with five-point linear extrapolation

or polynomial extrapolation (i.e. to estimate positions at impact and five frames beyond)

was more effective than using the spline to simply smooth through impact. More

recently, a comparable case study on an intermediate-level player executing a forehand,

reported commensurate results with smoothing through impact not only shown to

significantly affect the timing and amplitude of peak racquet velocity but also account for

49% of its variance (Knudson, 2005). While this approach was likely responsible for the

observed decreases in racquet speed in earlier investigative efforts (Elliott et al., 1989;

Takahashi et al., 1996), the efficacy of smoothing through the impact of the high

performance tennis serve may show less distortion when treated similarly. Interpretation

of Figures 4.5 and 4.6 appear to support this assertion with minimal differences noted

between the kinematic and kinetic outputs irrespective of the way data were treated

prior to their smoothing through impact. Interestingly, consideration of only pre-impact

data seems to be significantly affected by end-point error, compromising the true

representation of segment and racquet coordination near and through impact.

With the effect of smoothing through impact seemingly negligible in the kinematic and

kinetic analysis of the tennis serve in this study, indications are that data extrapolation –

as endorsed by Knudson and Bahamonde (2001) – may not be necessary. Rather,

analysis of data that has been interpolated and then smoothed through impact should

allay doubts as to the authenticity of pre- and post-impact segmental positions.

Treatment of data in this fashion also allows for the evaluation of post-impact upper

extremity joint kinematics and kinetics. More specifically, comparison of the modelled,

differentially cropped and smoothed data indicated that all conditions which involved

smoothing through impact appeared to accurately depict the analysed movement. So,

where any of these conditions could be appropriate for data reduction, the technique

103

that included as much raw pre-impact data as possible (i.e. up to 1 frame pre-impact)

was preferred. This approach was considered the most prudent as it maximised the

amount of true movement signal analysed, while also being sufficiently conservative to

guard against any movement artefact suspected among frames immediately post-impact.

In analysing Figures 4.5 and 4.6 alongside the findings of Knudson and Bahamonde

(2001) and Knudson (2005), it’s similarly important to understand that the serve, unlike

groundstrokes or volleys, is a closed skill, that when optimally executed is largely devoid

of variable racquet-ball impact locations, especially with high-performance players.

Theoretically these more consistent, centrally-located racquet-ball impacts should

minimise any movement distortion of racquet or body; more likely to be observed

following off-centre impacts. Also, with the serve’s tossed ball possessing virtually no

horizontal momentum (Chow et al., 2003a), the collision between racquet and ball would

interfere minimally with the racquet’s forward progression, and therefore the serve’s

entire movement. In the execution of groundstrokes however, the incoming ball’s

momentum must be overcome, intensifying the changes in the pre- and post-impact

acceleration profile of the racquet to thereby augment the potential for resultant

oversmoothing.


2. Smoothing through interpolated impact data in the high performance serve does

not compromise the validity of the related data analysis.

4.3.3 The Repeatability of the Tennis Serve




The execution of a sports skill with a high degree of positional repeatability should

increase along with the competency of the performer. A link between the legitimacy of

using one trial as a representative measure of a performer’s technique and ability level is

also likely. Indeed, Bootsma and Wieringen (1990) reported that analysis of a single trial

104

can be justified with expert sportspersons whose within-trial consistency (movement

repeatability) is high. In studying the repeatability of the kinematic and kinetic

characteristics of normal adult gait, Kadaba et al. (1989) also concluded that it was

reasonable to base clinical decisions on a single gait evaluation. However, another body

of research exists to highlight the methodological concerns over such study designs (Salo

et al., 1997; Mullineaux et al., 2001; Knudson and Blackwell, 2005). For example, Salo et

al. (1997) in establishing the number of trials required to reliably represent the average

movement pattern of sprint hurdles, showed that depending on the variable analysed,

between 1 and 78 trials were needed. Bates et al. (1992), in analysing the effect of trial

size on statistical power, mounts a similar argument.

While normative data on various stroke descriptors abound, information pertaining to the

number of tennis strokes that should be recorded to accurately represent specific

kinematic and kinetic variables is limited. Rather, investigative forays into the mechanics

of stroke production have preferred to analyse the highest velocity stroke or the stroke

subjectively assessed to be of the highest quality (Elliott et al., 1989; Reid and Elliott,

2002). The underlying assumption is that one superior performance is representative of

that player’s overall stroke technique. Certainly, from a loading perspective, the highest

velocity stroke would theoretically present the highest loading conditions and thus be of

the most research interest. Interestingly however, intra-subject variability has been

shown to feature in tennis stroke production, bringing into question this methodological

norm, particularly if the research questions involve derivative data (i.e. velocity,

acceleration, force). In examining the upper extremity angular kinematics of the tennis

forehand, Knudson (1990) demonstrated that college level players reproduced consistent

wrist and elbow positions (coefficient of variation, CV < 5.9%) but with highly variable

wrist and elbow joint angular velocities (CV = 90.6%) and accelerations (CV = 129.5%).

With no mention made of mean racquet and/or post-impact ball velocity this inconsistent

patterning of higher order kinematic variables may be expected. Alternatively, even in an

experimental setting with incoming ball velocity controlled, variation in ball impact height

and a player’s stance may vary from one forehand to the next, thereby affecting

segment coordination. These confounds were less evident in a more recent evaluation of

105

the variability of impact kinematics in the tennis forehand hit by advanced players

(Knudson, 2005). Most racquet kinematic variables were very consistent (mean CV <

6.3%) across five trials, yet no effort was made to substantiate the minimum number of

trials required to obtain reliable stroke data. Furthermore, as playing standard likely

influences stroke repeatability it may be that more trials need to be considered when

analysing the strokes of advanced players (ITN 2-4; i.e. Knudson and colleagues, 1990,

2005) as compared with those of the professional players (ITN 1) used in this

investigation.

Consistent with the findings of Knudson (1990), Bootsma and Van Wieringen (1990)

demonstrated variation in the direction of bat travel during the forwardswing of the elite

table tennis forehand drive, but with very consistent bat-ball alignments at impact.

Expert marksmen have also been observed to make compensatory upper-arm

movements to achieve low variability in the position of the pistol barrel when shooting

(Arutyunyan et al., 1968). This illustrates that skilled performers can obtain similar

outcomes, despite variation in coordination strategies, which is analogous with the

concept of synergies in motor control.

Coefficients of multiple correlations equal to or in excess of 0.90 are considered to

indicate highly repeatable movement performance (Yu et al., 1997). Pre-impact knee

extension (CMC ≥ 0.93), shoulder joint anterior-posterior shear force (≥ 0.93) and elbow

valgus-varus force (≥ 0.90) averaged over the forwardswings of the three and five FS

and SS can therefore be regarded as highly repeatable. The angular velocity of lateral

flexion (≥ 0.94) and the shoulder joint’s internal rotation moment (≥ 0.93) also satisfied

this criterion during the FS. Greater inconsistency however was observed in these

variables when one or both players executed the KS (CMC ≈ 0.71-0.86), which may

reflect inherent variation in the KS motion itself. Indeed with no control exhibited over

the location of the ball toss, it may be that the more laterally displaced KS toss displays

greater variability, resulting in more capricious coordination of the trunk and upper limb.

Certainly, some high performance players have been observed to impart more or less

“kick” across their KS, with the variable lateral displacement of their tosses largely

considered to influence the amount of kick they generate (Goetzke, personal

106

communication, 2005). The CMC’s calculated on post-impact shoulder joint kinetics

revealed varying degrees of repeatability between players, and may be linked to

differences in how the same player dissipates energy and readies him or herself for

subsequent point play.

Conceptually the more closed nature of the tennis serve may be thought to lend itself to

greater movement repeatability than previously analysed skills like the tennis forehand.

As aforementioned however, skilled tennis players have been shown to reproduce

consistent body and racquet positions with variable speeds and accelerations of segment

rotation (Knudson, 1990). Indications are that players coordinate their bodies and

racquets similarly when serving. Certainly, an increased sample of players may have

provided further insight as to whether or not this variance was representative of the

serve of the wider elite tennis playing population. Nonetheless, as alluded to previously,

inherent variation likely exists not only between and within variables but also subjects

(Bates et al., 1992; Salo et al., 1997).

A reduction in the number of trials analysed (i.e. from five to three) however, did not

compromise the CMC’s of the kinematic and kinetic variables linked to increased shoulder

joint loading in the serve. In other words, increasing the sample of serves beyond three

appears to do little to enhance the degree to which mean data becomes more

representative of the performed movement. Consequently, it may be reasonable to

assume that an accurate representation of a player’s typical serve motion can be

obtained from three trials. This would also satisfy the recommendation of Mullineaux et

al. (2001) that studies of human movement consider at least this number of trials in an

effort to provide for more representative and valid data.


3. Reliable kinematic and kinetic data can be gleaned through three high


107

4.4 CONCLUSION

There is little doubt that tennis stroke analysis can benefit from quantifiable and

systematic methods of observation (Knudson and Morrison, 2002). The three-fold

purpose of this study sees us firstly recommend that a tennis serve’s sampled

displacement data be smoothed with a quintic spline at a MSE near 25 for accurate

kinematic and kinetic representation. Unlike in the forehand, this smoothing protocol

may be applied through filled impact data in the high performance serve without

compromising the validity of pre- and post-impact data analysis. While selecting a

protocol that permits a maximum amount of raw pre-impact data may be preferred,

other options that involve the smoothing of combinations of deleted and interpolated

data (0.004-0.02s) either side of impact also appear appropriate for data reduction.

Thirdly, we encourage researchers and coaches to consider critically observing at least

three FS or KS, if technique improvement or enhanced mechanical understanding of the

serve are desired.

108

CHAPTER 5: BIOMECHANICAL COMPARISON OF THE HIGH PERFORMANCE

FLAT AND KICK TENNIS SERVES

5.1 INTRODUCTION

On the professional tour, singles players hit between 50-150 first and second serves per

match. When multiplied by 60 – a common target number of singles matches per year –

and serving during doubles matchplay and practice is added, one begins to appreciate

the need for sound service technique. While there are several mechanical characteristics

common to the world’s best servers, variation among individual techniques as well as

between serve types can be readily observed. At the elite level for example, a player’s

tactical objective will likely differ between first and second serves, resulting in varied

coordinative demands. A fact illustrated by ball impact locations and racquet velocity

profiles changing with serve type (Chow et al., 2003b), and further differences reported

in trunk muscle activation during the flat, slice and topspin serves (Chow et al., 2003a).

From a tactical standpoint, high performance players use the first serve to win or

dominate the point. More often than not, secure in the knowledge of a second “back-up”

delivery; players will coordinate body segments so that the FS is hit with near maximum

horizontal velocity. The second serve, on the other hand, sees players wanting to get the

serve “in”, but at the same time, execute with sufficient precision and speed so as to

make returning uncomfortable for the receiver. High performance male players

simultaneously achieve these goals by imparting a large amount of spin to the second

serve. This spin can be slice but is more commonly topspin, as this type of ball rotation

benefits from the Magnus effect to increase a serve’s margin of error for a given post-

impact ball speed (Brody et al., 2002). When pre-impact racquet trajectories are such

that an oblique and forward rotation is applied to the ball, serves are observed to “kick”.

In other words, the ball’s trajectory is similar to that of a topspin serve, but upon

contacting the ground rather than just bounce high (i.e. as compared to a FS) it also

“kicks” away (i.e. to the left of a player returning a right-hander’s serve). The KS

therefore is the preferred second serve of most professional male players. Anecdotal

reports from elite coaches imply the FS and KS to differentially load the body and

racquet arm (Goetzke, personal communication, 2005), yet kinetic variation between the

high performance FS and KS has not been investigated.

109

The shoulder is a key joint in the serve. High upper arm rotational velocities (i.e. internal

rotation ≈3000º.s-1) developed through large ranges of motion (≈270° circumduction)

are believed to contribute ≈20% of the total force generated during the stroke (Kibler,

1995). Unsurprisingly, injury to the tennis player’s shoulder is often allied to the serve

(Elliott et al., 2003). Further, variation in serve technique has been shown to load the

shoulder joint differently and therefore have some implications for injury. Superficially,

positive associations between serve velocity and shoulder joint loading may implicate the

FS, with its higher horizontal racquet velocities, in shoulder injury (Elliott et al., 2003).

However, Chow et al. (2003a) revealed no difference between first and second serve

pre-impact racquet speeds, but significant variation in the components of racquet

velocity. Where higher vertical and lateral racquet velocities were observed to

characterise the second serve, the opposite was true for horizontal racquet velocity.

Players may therefore experience comparable gross shoulder joint loading in the FS and

KS but with differential 3D joint kinetics. To elaborate, players may generate higher

anterio-posterior joint force in the FS but magnified compressive force in the KS.

Kinetic analyses to establish these theorised relationships between serve type and

loading at the shoulder have not been pursued. The derivation of this information, along

with matchplay statistics such as a player’s first and second serve percentage over the

course of a tournament or career, may provide an insight into that player’s susceptibility

to shoulder injury. Indeed, disparate FS and KS kinetics may lead some players,

especially those with a history of shoulder problems, to reconsider their serving strategy.

Hypothetically, armed with the knowledge that KSs produce higher compressive forces

throughout the forwardswing to impact, symptomatic players could adjust first or second

serve strategy and thus the frequency with which these serves are executed. Similarly,

substantiation of the serve’s shoulder joint kinetics may assist practitioners in the

diagnosis of injury to the joint, while also providing coaches with information that may

lead to the re-assessment of improvement strategies.

110




velocities;


average rates (to peak) of anterior force loading during the cocking phase than

the KSs;

3. Kick serves are characterised by larger mean shoulder joint compressive forces

during the forwardswing and follow-through than in the FS;

4. Higher average rates of shoulder joint compressive force loading are more

common to the KS than the FS during the swing phase;


impact external rotation moments are experienced in the FS as compared with

the KS;

6. Different pre-impact shoulder joint kinetics predict the development of racquet

velocity in the FS and KS.

5.2 METHODOLOGY


Subsequent to an appropriate warm-up, 12 high-performance male players hit maximal

effort FSs and KSs to a 1x1 metre target area bordering the ‘T’ of the first service box.

As reported in Chapter 4, reliable kinematics and kinetics can be derived from normative

data of at least three serves. Consequently, all players were required to execute three

successful FSs and three successful KSs. For these serves to be considered successful,

they had to satisfy the respective FS and KS definitions provided in Chapter 1. An

average of six FSs and three KSs were hit by each player; with never more than four FS

and two KS serves unsuccessfully landing in the target area. These rates of success

compare favourably to the FS (55-60%) and KS (95-98%) percentages typical of

professional male matchplay (ATP, 2006). Mean (continuous and discrete) subject data

were then interpreted in the data analysis procedures outlined below.

111

5.2.2 Data Treatment and Statistical Analysis

The same 0.690m, 0.380kg Wilson 6.0 Pro Staff racquet, whose inertial parameters were

calculated using procedures outlined by Brody et al. (2002), was used by all players. The

data of De Leva (1996) and Clauser et al. (1969) provided the necessary segmental

masses and moments of inertia. Players were fitted with a customised UWA full-body

marker set (see Chapter 3) and a 12-camera 250 Hz, Vicon motion analysis system

(Oxford Metrics Inc.) recorded the 3D marker trajectories, and thus reconstructed the

service motions. Data were treated as recommended in Chapter 4, and the filtered

outputs were modelled with UWA’s customised lower and upper body models in

preparation for analysis.

As detailed in Chapter 1, events corresponding to meaningful temporal or kinematic

characteristics of the serve were identified as appropriate in each service trial.

Elaboration of phases between events (see Chapter 1) also facilitated the analysis of

peak loading conditions as well as associated kinematic variables. Customised Matlab

software allowed for the temporal normalisation of phases to assist with comparison

across serve types. Sixteen paired T-tests were used to ascertain statistically significant

differences between the kinematic variables considered to relate to shoulder joint

loading in the FS and KS. A further seven paired comparisons were performed to

determine any statistically significant variation in shoulder joint kinetics with serve type.

The kinetic variables analysed were those identified in the literature as being

representative of shoulder joint ‘load’ or empirically linked to shoulder injury, and thus

those presented in the research hypotheses. Due to the exploratory nature of this

research a stringent partial Bonferroni correction (p<0.01) was applied to delineate

statistical significance and reduce the Type 1 error rate. Further examination of the

relationships between the variables considered to represent pre-impact shoulder joint

load and service kinematics was undertaken through regression analyses.

112

5.3 RESULTS

5.3.1 Effect of Serve Type on Absolute and Planar Racquet Velocity

As depicted in Table 5.1, significant differences existed between the pre-impact racquet

velocity profiles of the FS and KS. That is, where higher absolute (43.22 ± 3.1 m.s-1),

horizontal (40.55 ± 3.3 m.s-1) and vertical (30.03 ± 3.2 m.s-1) racquet velocities

characterised the forwardswings of the FS, players generated higher lateral racquet

velocities at impact for the KS (-10.19 ± 2.3 m.s-1; FS: -1.42 ± 5.5 m.s-1). For the right-

handed player, movement of the racquet left to right along the global Y-axis is negative.

Figure 5.1 illustrates these differential mean racquet velocity profiles during the

forwardswing of both serves.

FS KS Linear racquet

kinematics

Phase /

EVENT Mean (SD) T stat p

Maximum absolute

velocity (m.s-1)

Forward-

swing 43.22 (3.1) 40.28 (2.9) 4.410 0.001 *

Maximum horizontal

velocity (m.s-1)

Forward-

swing 40.58 (3.4) 35.04 (2.9) 8.502 0.000 *

Maximum vertical

velocity (m.s-1)

Forward-

swing 30.03 (3.2) 27.85 (2.9) 3.949 0.002 *

Lateral velocity (m.s-1) IMP -1.42 (5.5) -10.19 (2.3) 6.326 0.000 *

Table 5.1. Comparison of linear racquet kinematics across FS and KS (* p<0.01).

113

-20

100


-10

0

10

20

30

40

50

0 20 40 60 80

Velo

city

(m.s

-1)

FS Horizontal

KS Horizontal

FS Lateral

FS Vertical

KS Lateral

KS Vertical

Figure 5.1. Comparison of mean 3D linear racquet velocities during the forwardswing of the FS and KS.

5.3.2 Body Kinematics that Characterise Serve Performance

Selected angular displacement and velocity data describing upper and lower body motion

of the FS and KS is presented in Table 5.2. Upper arm MER approximated 115º±15º for

both serves and was antecedent to the upper arm moving through ≈40º of longitudinal

rotation (see Figure 5.2) and externally rotating (in the thorax) at high speeds (mean:

10.4 - 10.8 rad.s-1) during the forwardswing. The upper arm plane of elevation angle

trended marginally higher in the KS (-161.45º ± 10.2º; FS: -158.93º ± 8.5º), while the

elevation of the upper arm with respect to the thorax at impact was independent of

serve type (FS: 108.85º ± 14.1º; KS: 107.72º ± 19.7º).

114

FS KS Kinematic characteristic

EVENT /

Phase Mean (SD) T stat p

Maximum external rotation of the

racquet shoulder (º) MER

-115.86

(18.3)

-119.04

(18.3) 0.693 0.502

Upper arm plane of elevation

angle (º) MER

-158.93

(8.5)

-161.45

(10.2) 2.407 0.035

Peak shoulder joint longitudinal

rotation angular velocity (rad.s-1)

Forward-

swing

-10.80

(4.7)

-10.60

(3.1) -0.594 0.564

Upper arm – thorax elevation

angle (º) IMP

108.85

(14.1)

107.72

(19.7) 0.335 0.744

Lateral flexion separation angle

(º) MKF

31.52

(7.3)

31.55

(7.5) -0.121 0.906

Shoulder alignment lateral flexion

(º) IMP

-41.68

(7.8)

-33.43

(10.2) -3.903 0.002 *

Shoulder alignment rotation (º) IMP

-41.55

(18.5)

-64.36

(14.3) 7.152 0.000 *

Shoulder alignment forward

flexion (º) IMP

56.40

(15.1)

67.23

(9.4) -3.865 0.003 *

Maximum front knee joint flexion

(º) MKF

73.41

(19.3)

74.61

(17.1) -1.324 0.212

Peak front knee joint extension

angular velocity (rad.s-1)

Lead leg

drive

7.57

(2.0)

8.22

(1.8) 2.795 0.017

Maximum rear hip vertical velocity

(m.s-1)

Rear leg

drive

2.06

(0.3)

2.25

(0.3) -3.927 0.002 *

Table 5.2. Upper and lower body kinematics that characterise the FS and KS (* p<0.01), and that are reported to relate to shoulder joint loading in the serve.

115

-140

100

-120

-100

-80

-60

-40

-20

00 20 40 60 80

Forwardswing (Temporally normalised)A

ngle

(deg

rees

)

FS

KS

Figure 5.2. Comparison of the mean internal rotation of the upper arm during the forwardswing of the FS and KS.

Expression of the pelvis coordinate system in that of shoulder alignment allowed for the

computation of the angle between their respective X-axes, or in other words, a lateral

flexion separation angle. As compared to the pelvis, the shoulders were similarly laterally

flexed to the right at MKF in both the FS (31.52º ± 7.3º) and KS (31.55º ± 5.5º). At

impact however, the 3D alignment of the shoulders (with respect to the global

coordinate system) varied significantly between serves. That is, where the shoulders

were more rotated (i.e. front-on; FS: -41.55º ± 18.5º; KS: -64.36º ± 14.3º) and tilted to

the left (FS: -41.68º ± 7.8º; KS: -33.43º ± 10.2º) in the FS, they were flexed further

forward in the KS (67.23º ± 9.4º; KS: 56.40º ± 15.1º) (see Figure 5.3).

116

Figure 5.3. Contrast in the 3D alignment of the shoulders at impact in the FS (left) and KS (right).

Maximum front knee joint flexion (≈74º±18º) was consistent for both serves (Figure

5.4), however peak velocity of knee joint extension trended higher in the KS. The

difference observed in the peak vertical velocity of the rear hip between serves (FS: 2.06

± 0.3 m.s-1; KS: 2.25 ± 0.3 m.s-1) also hints toward some differential higher order lower

limb kinematics characterising the FS and KS.

117

00 20 40 60 80 100

Lead leg drive (Temporally Normalised)

10

20

30

40

50

60

70

80

Ang

le (D

egre

es)

FSKS

Figure 5.4. Comparison of the mean front knee joint extension during the lead leg drive phase of the FS and KS.

5.3.3 Shoulder Joint Kinetics that Characterise the FS and KS

No significant differences were recorded in the shoulder joint kinetics of the FS and KS

(Table 5.3). During the cocking phase of both serves, players developed homogeneous

maximum anterior forces (FS: -167.29 ± 46.7N; KS: -160.35 ± 52.5N) at comparable

rates (FS: -208.30 ± 56.2N.s-1; KS: -189.92 ± 55.9N.s-1). Peak internal rotation moments

approximating -23Nm were also generated during the forwardswing irrespective of serve

type.

Similar pre-impact compressive force profiles punctuated the performance of both

serves. More specifically, average rates of maximum compressive force loading were

near to 325N.s-1 and mean compressive forces approximated 220N regardless of serve

performed. Further non-significant differences marked the post-impact shoulder joint

kinetics of the FS and KS. However, there was some suggestion of mean compressive

118

forces (FS: 87.11 ± 39.6N; KS: 75.66 ± 32.5N) and peak external rotation moments (FS:

18.76 ± 10.0N; KS: 14.73 ± 6.6N) trending higher during the follow-through of the FS.

FS KS Shoulder joint forces and

moments

EVENT /

Phase Mean (SD) T stat P

Maximum anterior force (N) Cocking -167.29

(46.7)

-160.35

(52.5) -0.555 0.590

Average rate of maximum

anterior force loading (N.s-1) Cocking

-208.30

(56.2)

-189.92

(55.9) -1.435 0.179

Peak internal rotation moment

(Nm)

Forward-

swing

-22.66

(7.6)

-23.48

(5.4) 0.423 0.680


compressive force loading

(N.s-1)

Swing 333.82

(61.3)

321.18

(83.8) 1.021 0.329

Mean compressive force (N) Forward-

swing

228.63

(52.4)

210.69

(54.2) 1.571 0.144

Mean compressive force (N) Follow-

through

87.11

(39.6)

75.66

(32.5) 2.644 0.023

Peak external rotation moment

(Nm)

Follow-

through

18.76

(10.0)

14.73

(6.6) 1.840 0.093

Table 5.3. Comparison of shoulder joint kinetics considered to represent shoulder joint load across FS and KS (n = 12; * p<0.01).

Figure 5.5 presents an example of the moment and angular velocity of upper arm

internal-external rotation during the forwardswing of a FS and KS performed by the

same subject. The joint power term at the shoulder joint throughout the majority of this

phase is positive (Figure 5.6), and thus consistent with what is to be expected from the

prevailing internal rotation moment and external rotation upper arm angular velocity

(expressed in the thorax). These profiles are representative of the upper arm

longitudinal rotation that characterised the service forwardswing of all the players

studied.

119

-30

1

-25

-25

-20

-15

-10

-5

01 21 41 61 81 10

Mom

ent (

Nm

)

-20

-15

-10

-5

0


Angular velocity (rad.s

-1)

FS Shoulder joint internal rotation moment

FS Shoulder joint external rotation angular velocity

KS Shoulder joint internal rotation moment

KS Shoulder joint external rotation angular velocity

Figure 5.5. Angular external rotation velocity (within the thorax) and internal rotation moment about the long axis of the upper arm during the forwardswing of a FS and a KS.

-200

0 20 40 60 80 100


0

200

400

600

800

1000

1200

1400

Pow

er (W

)

FS

KS

Figure 5.6. Shoulder joint internal-external rotation power term during the forwardswing of a FS and a KS.

120

The moment and angular velocity of upper arm internal-external rotation during the

follow-through of a subject’s FS and KS are presented in Figure 5.7. The product of the

external rotation moment and internal rotation upper arm angular velocity (expressed in

the thorax) produced the negative shoulder joint power term (Figure 5.8) generated

throughout this period. These profiles are representative of the upper arm longitudinal

rotation that characterised the service follow-through of all the players studied.

-30

1

Follow-through (Temporally normalised)-20

-20

-10

0

10

20

30

1 21 41 61 81 10

Mom

ent (

Nm

)

-15

-10

-5

0

5

10

15

20

Angular velocity (rad.s

-1)

FS Shoulder joint internal rotation moment

FS Shoulder joint external rotation angular velocity

KS Shoulder joint internal rotation moment

KS Shoulder joint external rotation angular velocity

Figure 5.7. Angular internal (+) and external (-) rotation velocity as well as internal (-) and external (+) moment about the long axis of the upper arm during the follow-through of a FS and

a KS.

121

-500

-400

100

Follow-through (Temporally normalised)

-300

-200

-100

0

100

200

300

400

500

600

0 20 40 60 80

Pow

er (W

)

FS

KS

Figure 5.8. Shoulder joint internal-external rotation power term during the follow-through of a FS and a KS.

5.3.4 Pre-impact Shoulder Joint Loading as a Predictor of Serve Velocity

While the reported pre-impact joint loading variables were similar between serves, to

determine whether they differentially contributed to FS and KS absolute pre-impact

racquet velocity (the goal of both serves), multiple regression analyses were performed.

The reported loading variables were entered as predictor variables into a stepwise

regression with absolute racquet velocity serving as the criterion variable.

Average rate of anterior force loading in the cocking phase was shown to be the most

significant predictor (F = 6.010, p = 0.037) of maximum absolute FS velocity (Table

5.4), while mean pre-impact compressive force further strengthened the prediction

equation. The model realised the following prediction equation: Maximum absolute FS

velocity = 29.442 + (-0.420*average rate of anterior force loading + 0.022*mean pre-

impact compressive force), which explained 81.5% of the variance in peak FS absolute

racquet velocity.

122

Predictor variable Regression co-

efficient Beta weight

Multiple

correlation R value

Average rate of anterior

force loading * -0.420 -0.733 -0.733 0.691

Mean pre-impact

compressive force * 0.022 0.365 0.365 0.815

Constant 29.442

Table 5.4. Results of step-wise regression on pre-impact loading variables and absolute racquet velocity in the FS (* p<0.05).

The average rate of compressive force loading (F = 31.145, p = 0.000) during the swing

phase accounted for ≈75% of the variance in peak KS absolute racquet velocity (Table

5.5). The resulting prediction equation was: Maximum absolute KS velocity = 30.472 +

0.031*average rate of pre-impact compressive force loading.

Predictor variable Regression co-

efficient

Beta

weight

Multiple

correlation R value

Average rate of compressive

force loading * 0.031 0.870 0.870 0.757

Constant 30.472

Table 5.5. Results of step-wise regression on pre-impact loading variables and absolute racquet velocity in the KS (* p<0.05).

5.4 DISCUSSION

Evaluation of tennis serve kinematics has principally focused on the FS, while all kinetic

analyses have exclusively considered this type of serve. From a tactical standpoint, few

players approach the FS and KS in the same way, yet previous research of tennis serves

has elucidated few mechanical differences (Chow et al., 2003a; 2003b).

5.4.1 Effect of Serve Type on the 3D Profile of Racquet Velocity




velocities.

123

As hypothesised, significant differences existed in the pre-impact racquet velocity profiles

of the FS and KS. The significantly higher peak horizontal and vertical racquet velocities

generated during the forwardswing of the FS and the higher peak lateral velocities at

impact in the KS were also in agreement with the differential velocity profile reported by

Chow et al. (2003a). From a practical standpoint, these differences are related to the

divergent ball toss locations of the FS and the KS. For example, the displacement of the

ball toss significantly further forward in the FS (Chow et al., 1999, 2003a) likely

facilitates the development of high horizontal racquet velocities, whereas the KS, with its

exaggerated lateral ball toss position, would appear to predispose players to generating

higher lateral racquet velocities. Tangentially, the higher peak vertical racquet velocities

developed during the forwardswing of the FS (30.03 ± 3.2 m.s-1; KS: 27.85 ± 2.9 m.s-1)

may have assisted players in attaining higher subsequent hitting positions (relative to

standing height (ST); FS: ≈1.56xST; KS: ≈1.52xST) at impact.

Of further interest is that where Chow et al. (2003a) reported professional male players

to generate similar pre-impact absolute racquet velocities in both first and second serves

(≈38m.s-1), the players in this study developed significantly higher absolute 3D velocities

in the FS (43.22 ± 3.1 m.s-1) than in the KS (40.28 ± 2.9 m.s-1). These differences could

be attributed to the more elite playing sample of the Chow et al. (2003a) study, which

analysed the serves of four top 100 ranked professional male players. However, the

variance may be better explained by methodological incongruence. For example, in the

current investigation all players were instructed to hit maximal effort FS and KS to

location. In contrast, although Chow et al. (2003a) controlled for ball placement, the first

and second serves were likely executed hit with varying tactical intent (i.e. slice or

topspin), thereby confounding the comparison of absolute 3D velocities between specific

types of serves.

Professional players have been reported to generate relatively homogeneous pre-impact

horizontal (≈33 m.s-1) and absolute (≈38 m.s-1) racquet velocities for both first and

second serves (Chow et al., 2003a). Correspondingly, documentation of post-impact ball

velocities in excess of 50 m.s-1 or 180 km.hr-1 is common (Elliott and Wood, 1983; Elliott

124

et al., 2003). The magnitudes of the recorded peak, pre-impact horizontal (FS: ≈40m.s-1;

KS: ≈35m.s-1) and absolute (FS: ≈43m.s-1; KS: ≈40m.s-1) racquet velocities in this study

were thus comparable to the serves of other elite players. The peak lateral racquet

velocities of the FS (-1.42 ± 5.5 m.s-1) and KS (-10.19 ± 2.3 m.s-1) are appreciably

higher (from left to right) than those which Chow et al. (2003a) found to characterise

the typical first (3.47 ± 3.7 m.s-1) and second (-1.83 ± 2.4 m.s-1) serves of their sample.

Similar variation marks the peak vertical racquet velocities recorded between the two

studies (≈28m.s-1 vs ≈17m.s-1). Again, the likelihood is that this is a product of disparate

study protocols as compared to any differences in playing populations. For example,

where this study reported peak vertical racquet velocities during the forwardswing to

impact, Chow et al. (2003a) determined all measures of pre-impact racquet velocity from

the two video fields preceding racquet-ball impact.


1. Higher peak horizontal and vertical racquet velocities are generated in the

forwardswing of the FS, while higher peak pre-impact lateral racquet

velocities are developed in the KS.

5.4.2 Variation in Body Kinematics in the FS and KS

The tilted alignment of the shoulders and pelvis prior to cocking coincides with MKF, and

is considered by many coaches as key to high speed serving. Indeed, afforded the label

of “power” or “trophy” position, this platform is believed to trigger resultant knee

extension and trunk rotation, which in turn is associated with higher impact heights

(Elliott and Wood, 2003), increased racquet displacement (Elliott, 2001), and assisted

upper arm external rotation. Importantly, it also favours the development of angular

momentum about the anterior-posterior axis of the thorax, which Bahamonde (2000)

revealed as a characteristic of higher speed servers.

The maximum flexion angles of the front knee were similar for the FS (73º ± 19º) and

KS (74º ± 17º), and consistent with the magnitude of front knee flexion advocated in

the coaching literature (Elliott and Alderson, 2003). While this characteristic is readily

125

observable, it should not form the sole basis upon which coaches evaluate how actively

players engage their lower limbs in the serve. More complete analyses are required; a

fact reinforced by comparing the peak velocity of knee joint extension and the peak

vertical velocity of the rear hip in the FS and KS. That is, despite both serves being

characterised by ≈75º of maximum front knee flexion, the peak vertical velocity of the

rear hip in the KS (2.25 ± 0.3 m.s-1) was significantly higher than in the FS (2.06 ± 0.3

m.s-1). Although failing to reach statistical significance (p = 0.017), the peak knee

extension angular velocity trended similarly higher, albeit modestly in comparison to the

more variable 800±400º of peak knee extension reported by Fleisig et al. (2003), in the

KS (8.22 ± 1.8 rad.s-1; FS: 7.57 ± 2.0 rad.s-1). Together, these characteristics appear to

suggest that a more precocious lower limb drive punctuates the KS.

Conceptually, Elliott et al. (1986) and Elliott (1988) have proposed that a more forceful

lower limb drive produces an off-centre force relative to the hitting shoulder such that

the racquet arm and racquet are forced down and away from the body during the

cocking phase of the serve. If KS are indeed characterised by more dynamic leg drives,

intuitively it could be surmised that the magnitude of upper arm external rotation would

also be more pronounced as compared to the FS. However, MER of the upper arm

approximated 115º ± 15º for both serves so variance in lower limb drive may be more

related to the serve’s shoulder joint kinetics. The homogeneity in the peak upper arm

external rotation velocities (expressed in the thorax) during the forwardswings of the FS

(10.8 ± 4.7 rad.s-1) and KS (10.6 ± 3.1 rad.s-1) appears to support this affirmation. At

MER, a marginally higher mean upper arm plane of elevation angle characterised the KS

(-161º ± 10º) as compared with the FS (-159º ± 9º). Significantly, only one subject

recorded a mean plane of elevation angle in excess of 180º at MER such that the upper

arms of all other subjects largely remained in front of their shoulder alignments. This

mean upper arm plane of elevation position is comparable to the previously reported 7º

± 9º of upper arm horizontal adduction at same reference point of the serve (Fleisig et

al., 2003). It would therefore appear that hyperangulation (i.e. extension of the

abducted, externally rotated arm beyond the plane of the scapula) contributing to

secondary impingement may not be a cause of concern among players that align their

bodies in this fashion when executing a FS and/or a KS.

126

The abovementioned magnitudes of mean FS and KS maximum upper arm external

rotation (115º ± 15º) are sizeably less than the 170º-185º of upper arm external

rotation previously reported to characterise the serve (Fleisig et al., 2003) and baseball

pitch (Dillman et al., 1993; Fleisig et al., 1999; Escamilla et al., 2001; Matsuo et al.,

2002). Similarly, the mean peak upper arm longitudinal rotation velocities recorded in

this study amount to approximately 25% of the 2000º-3000º.s-1 detailed in past

kinematic investigations of the tennis serve (Elliott et al., 1995; Kibler, 1995; Fleisig et

al., 2003). This anomaly would appear to be related to methodological differences

between studies and potential systematic error in the current data collection protocol.

That is, earlier efforts to evaluate the kinematics and/or kinetics of the tennis serve

derived 3D coordinate data from 2D digitised images utilising the direct linear

transformation method. The subsequent modelling techniques were restricted to inter-

segment vector comparisons projected onto planes, rather than the elaboration of ACS

matrices from the position of TCSs as performed in this study. For example, Elliott et al.

(2003) calculated axial rotation of the humerus from a vector defining the longitudinal

axis of the forearm relative to anterior direction of the shoulder in the transverse plane

of the upper arm. Fleisig and colleagues (1996; 1996b; 2003) in their kinetic study of the

tennis serve and throws of other sports employed the same analysis technique. While

this indirect measure of computing internal-external humeral rotation is touted as

accurate in all but extended elbow joint positions (Gordon and Dapena, 2006), the reality

is that its validity in representing actual 3D humeral joint motion is questionable. Indeed,

by inferring longitudinal rotation of the shoulder joint through changes in the sagittal

plane position of the forearm, these measures effectively discount the role of the trunk

in contributing to humeral motion. Furthermore, such approaches are commonly affected

by down plane error where any incorrect identification of the flexion-extension axes of

the shoulder but more likely elbow joint centre produces cross talk. The inaccuracy of

subsequent kinematic or kinetic descriptions of humeral rotation are magnified even

further when decompositions calculate it as the third rotation in sequence.

In an effort to negotiate the limitations involved with this technique, the customised

UWA marker set attempted to correctly replicate 3D humeral motion via the TCS of the

127

humerus. As documented in Chapter 3, the challenge of accurately tracking marker

trajectories in optical 3D motion analysis is significant (Cappozzo et al., 1996). Similarly,

the effects of soft tissue artefact have been shown to confound the representation of 3D

lower-extremity joint motion in walking (Reinschmidt, 1997a) and running (Reinschmidt,

1997b). To this end, the design and more pertinently placement of the humeral triad

alongside the bulky soft tissue of the upper arm, may have underestimated the

magnitude of axial humeral rotation in this study. Indeed, Gordon and Dapena (2006)

highlighted similar error to confound the calculation of axial rotation of the upper arm in

their recent critique of the contribution of joint rotations to serve speed. In application,

this likely manifested in reduced shoulder joint longitudinal rotation kinematics and

kinetics as when compared with those calculated through inter-segment vector

comparisons. Consequently, comparison of the current upper arm internal-external

rotation data to previously reported shoulder joint longitudinal rotation kinematics (and

kinetics) is confounded by the dichotomous data collection techniques. Significantly, the

mathematical derivation of shoulder joint position as described by the plane of elevation

and elevation angles should be unaffected. Furthermore, the comparative nature of this

study ensures that comparisons within subjects and between techniques are also valid,

and thus can be discussed in light of the hypothesised differences.

Noteworthy here is that while voluminous research has been conducted to demarcate

the most representative TCSs of lower limb bone motion (i.e. under the rigid body

assumption), it is clear that further work is needed to elaborate similarly representative

TCSs for the upper extremity. The positions of the anatomically relevant landmarks

required for the ACS definition are held in the TCS during dynamic motion. While

considerable work has been devoted to best defining anatomical landmarks, and thus

ACSs, comparatively less attention has been afforded to ascertaining the position of the

TCS to optimally represent bone motion. One possible solution is to align the upper arm

triad more distally along the humerus (i.e. closer to the elbow joint centre). A

retrospective case study was performed to determine whether this alignment reduced

skin movement and soft tissue artefact to provide for improved coupling between the

TCSs and the underlying bone motion as compared with the triad’s current orientation.

Results are presented and discussed in Appendix A.

128

The lateral flexion separation angle of ≈31º at MKF for both the FS and KS is consistent

with that expected of the “power” position described above. Noteworthy is that the need

to maintain balance and the large range of motion of the trunk dictates that the

magnitude of lateral pelvis flexion will be less than that which is possible at the

shoulders. Nonetheless, the greater lateral flexion of shoulder alignment - and by

extension the trunk - to a player’s racquet side would, when complemented by back leg

drive, assist players to produce more angular momentum about the abduction-adduction

axis (i.e. x-axis) of the trunk and better transfer angular momentum to the upper limb

(Elliott, 1988; Bahamonde, 2000). To this end, although Bahamonde (2000) illustrated

angular momentum generated about flexion-extension axis as the largest source of

angular momentum in the serve, players that laterally flexed their trunks to produce

angular momentum in a clockwise direction during the forwardswing to impact served

with greater ball speeds. Given the comparable laterally flexed alignments that describe

this sample’s FS and KS at MKF, it appears likely that high-performance players align

their trunks to generate angular momentum about the trunk’s anterior-posterior axis

independent of serve type. While certainly plausible, the 3D alignment of the shoulders

at impact varies significantly between FS and KS and suggests that the fashion in which

the trunk rotates during the forwardswing does differ between serves.

That is, at impact the alignment of the shoulders was significantly more rotated, tilted to

the left and extended in the FS than in the KS. Indications are thus that the

forwardswing of the FS may be characterised by larger amounts of lateral trunk flexion

and transverse plane trunk rotation, while more forward flexion is involved in the

execution of the KS. The fact that Chow et al. (2003a) found increased abdominal

muscle activity in the upward swing of the topspin serve as compared with that of flat or

slice serve may support this latter assertion. Nevertheless, in spite of these different

resultant orientations of shoulder alignment at impact, the angle of upper arm elevation

with respect to the thorax approximated 110º for both serves. This finding is consistent

with the 101º ± 11º of shoulder abduction that Fleisig et al. (2003) reported to

characterise the upper extremity motion of professional players at serve impact, and is

compatible with Atwater’s (1979) contention that ≈90º of shoulder abduction is typical

129

of all overhand sports skills. Significantly, it is also similar to the angle of 100º ± 10º

that Matsuo et al. (1999) detailed as producing maximum pitching ball velocity and

minimum shoulder loading in baseball pitching.

5.4.3 Relationship between Serve Type and Shoulder Joint Kinetics:

Implications for Injury and Performance

None of the seven variables considered to represent shoulder joint load differed across

serve type. In other words, comparable pre- and post-impact shoulder joint loads

punctuated the execution of both the FS and KS. These findings are in stark contrast to

the hypothesised differences but somewhat consistent with the comparable shoulder

joint kinematics across techniques. Furthermore, as selected racquet and body

kinematics were shown to vary with serve type, it is likely that kinetic analyses of other

joints would unearth technique-related differences.

Research hypotheses:


average rates of peak anterior force loading during the cocking phase than

the KSs;

3. Kick serves are characterised by larger mean shoulder joint compressive

forces during the forwardswing and follow-through than in the FS;

4. Higher average rates of compressive force loading are more common to the

KS than the FS during the swing phase;


impact external rotation moments are experienced in the FS as compared

with the KS.

5.4.3.1 Cocking

Peak anterior forces of ≈165N were recorded during the cocking phase of both the FS

and KS. Similarly, the average rate of peak anterior force loading was homogeneous

between serves, with mean rates of 208.30 ± 56.2N.s-1 and 189.92 ± 55.9N.s-1

punctuating the FS and KS respectively. This failed to support the hypothesis that

players would experience higher peak shoulder joint anterior forces and average rates of

130

peak anterior force loading when executing FS as compared with KS. These comparable

anterior force loading profiles may then suggest that the GH joint’s anterior capsule and

ligaments are similarly stressed near MER independent of serve type. These passive

structures play an important role in limiting anterior translation of the humeral head to

mitigate the prospect of GH instability.

Previous kinetic analyses of the tennis serve have not reported peak anterior forces prior

to or at MER; preferring to report maximum values during the forwardswing.

Consequently, the larger peak anterior forces reported by Noffal and Elliott (445N; 1998)

and Elliott et al. (males: 291.7 ± 119.8N; females: 185.1 ± 60.9N, 2003) may be typical

of the forwardswing to impact.

5.4.3.2 Forwardswing

Common to tennis pedagogy is the concept that dynamic lower limb activity coupled with

timely 3D trunk rotation ‘drives the shoulder up, the racquet down and the upper arm

into external rotation’ to propagate a stretch shorten cycle response in the related

anterior shoulder joint (i.e. internal rotation) musculature. Theoretically, the internal

rotator muscles would contract eccentrically to assist the anatomical constraints to

external rotation. Simultaneous storage of elastic energy would in turn facilitate the

concentric contraction of the upper arm’s internal rotators during the forwardswing to

impact (Wilson et al., 1989; Bahamonde, 1997; Walshe et al., 1998; Elliott et al., 1999).

In documenting the average individual shoulder joint power terms generated during the

serves of five male collegiate players, Bahamonde (1997) provided some evidence of

negative longitudinal rotation power (-220±72W) during the late backswing to confirm

this pattern of eccentric internal rotator muscle activity. However, interpretation of

Figures 5.5 and 5.6 fails to categorically support this notion; albeit there is some

suggestion of opposing longitudinal rotation joint moment and angular velocity, and thus

a negative joint power term, in the early forwardswing. As would be expected and

elaborated on below, the latter stages of the forwardswing see the internal rotation

moment and external rotation angular velocity (expressed in the thorax) of the upper

arm reach near peak values, and high positive power – similar to that which was

reported by Bahamonde (1154±1179W; 1997) – is produced.

131

As intimated above, internal rotation of the upper arm is considered key to the

development of high racquet velocities in the serve. Indeed, Elliott et al. (1995) have

demonstrated that this longitudinal rotation of the upper arm contributes upward of 40%

of the racquet’s horizontal velocity at impact. Of subsequent and recent investigative

interest has been the magnitude of the internal rotation moments that generate this

rotation during the serve’s forwardswing. For example, Elliott et al. (2003) observed

peak shoulder internal rotation torques of 71.2 ± 15.1Nm and 47.8 ± 16.3Nm for male

and female professional players respectively, while Bahamonde (1989) reported lower

torques (33Nm) to drive upper arm longitudinal rotation just prior to impact.

Comparatively smaller peak internal rotation moments were generated during the FS (-

22.66 ± 7.6Nm) and KS (-23.48 ± 5.4Nm) forwardswings of players in this sample. As

aforementioned, these lower values likely relate to divergent data collection and

modelling techniques. However, with high horizontal racquet velocities more central to

the foremost tactical goal of the FS as compared to the KS, it was hypothesised that

higher peak internal rotation moments would punctuate the forwardswing of the FS. As

evidenced by non-significant t-test result (0.423, p=0.680), the hypothesis was not

supported and players would appear to develop similar peak pre-impact internal rotation

moments independent of serve type.

Throughout the upper arm’s extension and internal rotation to impact, portions of the

rotator cuff (along with associated connective tissue, the joint capsule and biceps) need

to provide the compressive force necessary to centre the humeral head in the glenoid

fossa (Blevins, 1997). Ultimately, failure to do so would result in superior migration of

the humeral head and the supraspinatus or biceps muscles impinging under the coraco-

acromial arch (secondary impingement, Fleisig et al., 1995; Blevins, 1997). Throughout

the forwardswing to impact, the mean compressive force applied to the upper arm

approximated ≈220N in both the FS and KS; with no discrimination between serve type

possible. Similarly, poor distinction was made between the FS and the KS according to

their average rates of compressive force loading (FS: 333.8 ± 61.3 N.s-1; KS: 321.2 ±

83.8N.s-1). So, as players seem to generate pre-impact compressive forces of similar

magnitudes and at similar rates during the performance of the FS and KS, secondary

132

impingement brought on by high compressive force loading conditions may be no more

likely in the FS than it is during the KS.

5.4.3.3 Follow-through

The rotator cuff is reported to eccentrically contract to resist distraction, horizontal

adduction and internal rotation of humerus during the follow-through phase of high-

speed overhand sports skills like the tennis serve (Jobe et al., 1984; Fleisig et al., 1996).

With Kibler (1995) suggesting that the humerus may internally rotate by as much as 60º

post-impact, the inference is then that an external rotation moment must be applied to

slow continued upper arm internal rotation during the follow-through phase of the serve.

This interaction is illustrated in Figure 5.7, where the continued external rotation of the

subject’s upper arm (in the thorax) appears to be resisted by an external rotation

moment during early follow-through. Correspondingly, the resultant shoulder joint

internal-external rotation power term was negative (Figure 5.8). Visual inspection of

Figures 5.7 and 5.8 suggest that the inferred eccentric load placed on the shoulder

muscles responsible for slowing the rotating upper arm and racquet is similar between

serves. Determination of the specific muscles, or as suggested by Kellis and Baltzopoulos

(1995), the elastic properties that produce this external rotation moment would

however, require simultaneous 3D kinetic and EMG analysis. Nonetheless, by extension it

may be suggested that the rotator cuff is at no greater risk of tensile failure, muscle

strain or tear through repeated deceleration of FS or KS racquet and upper-extremity

motion.

Contrary to the hypothesised post-impact shoulder joint kinetic differences between the

FS and KS, and in spite of some evidence of a trend implicating the FS in marginally

higher post-impact shoulder joint loading conditions, the results of the paired

comparisons suggest that no distinctive loading profile characterises the follow-through

of either serve. That is, deceleration of the above-mentioned continued internal rotation

of the upper arm is facilitated by similar peak FS (18.8 ± 10.0Nm) and KS (14.7 ±

6.6Nm) post-impact external rotation moments. Likewise, as in the forwardswing,

indications are that selected shoulder muscles need to produce ≈80N of mean post-

impact compressive force to resist humeral distraction in both the FS and KS. As

133

theorists have typically preferred to postulate rather than quantify post-impact shoulder

joint kinetics, comparisons of the reported magnitudes of post-impact force and torque

to the literature are not possible.

Hypotheses determination:

2. Similar peak, shoulder joint anterior forces and average rates of peak anterior

force loading are experienced in the cocking phase of both the FS and KS.

3. Similar mean shoulder joint compressive forces are generated during the

forwardswing and follow-through of both the FS and the KS.

4. Comparable average rates of compressive force loading punctuate the swing

phase of both the KS and the FS.


impact external rotation moments are not experienced in the FS as compared

to the KS.

5.4.4 Contribution of Pre-impact Shoulder Joint Kinetics to Serve Velocity


6. Different pre-impact shoulder joint kinetics predict the development of

racquet velocity in the FS and KS.

Considerable variation in peak absolute racquet velocity was accounted for by different

shoulder joint kinetics in the FS and KS. That is, where 81% of this variation was

explained by a combination of pre-impact average rate of peak anterior force loading

and mean compressive force during the FS, average rate of maximum compressive force

loading in the swing phase was the most significant predictor (75%) of peak absolute

racquet velocity in the KS.

So, while Chapter 5.3.3 illustrated that these three kinetic variables load the shoulder to

similar extents independent of serve type, their contribution to racquet velocity appears

to differ. More broadly, it is possible that while comparable shoulder joint kinetics

punctuate the performance of both the FS and KS, they may exert varying influence on

the coordination of more distal segments and ultimately racquet velocity. Furthermore,

134

with more than hint of variable lower limb drive and shoulder alignment between serves,

this predictive difference in racquet velocity may also point to more subtle differences in

each serve’s ‘kinetic’ chain.


6. The pre-impact shoulder joint kinetics that predict the development of

racquet velocity vary with serve type.

5.5 CONCLUSION

Players generate significantly higher pre-impact horizontal, vertical and absolute racquet

velocities in the FS as compared with the KS. Conversely, higher lateral velocities are

developed during the forwardswing of the KS. The shoulder joint kinetics that contribute

to these differential velocity profiles do not however vary depending on the type of serve

performed. Nonetheless, it should be noted that individual players whom experience

higher loading conditions in a FS or KS may indeed be more susceptible to shoulder joint

pathologies through repetitive, long-term performance of that serve.

135

CHAPTER 6: RELATIONSHIP BETWEEN LOWER LIMB COORDINATION AND

SHOULDER JOINT KINETICS IN THE TENNIS SERVE

6.1 INTRODUCTION

In the tennis serve, dynamic lower limb motion is considered a precursor to high speed

trunk and upper extremity segment rotation, and thus the origin of the stroke’s ‘kinetic

chain’. By extension, this precocious leg action is also reported as key to achieving the

high service speeds that are pivotal to success in the modern professional game.

However, unlike in other overhand sports skills, such as elite baseball pitching, where a

certain uniformity punctuates the pitcher’s leg action, considerable coordinative lower

limb variation can be observed in the high performance tennis serve.

To this end, Elliott and Wood (1983) were the first to investigate the effect of stance in

the serve. In so doing, they described players as using the FU or FB technique, and

along with Bahamonde and Knudson (2001), revealed different GRF’s to characterise

each technique. That is, in developing comparable peak serving speeds, players using

the FU technique were generally shown to generate larger vertical GRF’s, whereas

players who adopted the FB arrangement developed higher horizontal propulsive forces

(Elliott and Wood, 1983; Bahamonde and Knudson, 2001). In turn, the FU stance

produced higher impact positions, while the FB arrangement – devoid of as much

horizontal braking force – was suggested to facilitate a player’s subsequent forward

progression to the net. Whether there is a corresponding kinetic difference in upper

extremity joint motion, and more specifically shoulder joint motion, is not known.

Certainly if one technique was shown to preferentially load the shoulder joint, there

would be obvious implications from a player development standpoint.

The velocity profile of the leg or lower limb drive, considered a by-product of the

effective, sequentially coordinated extension of the ankle, knee and hip joints, helps

develop the abovementioned GRF’s, and is thus assumed critical to high performance

serving. From a coach’s perspective, angular change about the front knee joint is widely

used to assess the proficiency of a player’s leg drive. However, with Girard et al. (2005)

indicating that this proficiency increases along with playing skill, more comprehensive

136

kinematic and/or kinetic analysis is required to detect increasingly subtle mechanical

differences among an elite playing population. It was in part, for the above practical

reason that Elliott et al. (2003) used front knee joint flexion at MER of the upper arm to

represent and evaluate the effect of leg drive on upper extremity joint loading.

The study’s results suggested that players with more effective leg drives (>10° front

knee joint flexion at MER) recorded lower internal rotation moments at MER (43.7 Nm)

than those players reported to use their legs less effectively (<10° front knee joint

flexion, 57.8 Nm). Better leg drives were also linked to lower peak internal rotation

torques (≈55 vs 65 Nm) during the forwardswing to impact. With peak racquet velocities

developed independent of leg drive, the inference was that players experience reduced

pre-impact shoulder joint loading when using more effective leg drives. Unfortunately

however, the classification used by Elliott et al. (2003) likely misrepresents leg drive and

thus may be of only limited functional value. For example, as players’ feet tend to leave

the ground antecedent to MER of the upper arm, describing the angular position of the

front knee appears capricious. Indeed, while it could be argued that greater knee flexion

at MER is indicative of earlier, larger peak knee flexion angles, it’s equally plausible that

the reverse is true. That is, players may have recorded <10° front knee joint flexion at

MER having gone through greater front knee extension or as a result of higher mean

front knee extension angular velocities. So, the very players reported to possess less

effective leg drives may have actually used their legs more dynamically.

Of further note is that research has only considered motion about the knee, and not the

ankle or hip, in their appraisals of leg drive. Albeit with respect to forward propulsion,

biomechanical analyses of walking and running have revealed impulse generated at the

ankle and hip joints to primarily contribute to these respective forms of locomotion

(Novacheck, 1995). It would therefore appear plausible that ankle and/or hip joint

kinematics play a similarly important role in a tennis player’s lower limb drive during the

serve.

In summary, although players and coaches may be familiar with information describing

the different GRF’s produced by the FU and FB techniques, there is a comparative lack of

137

evidence examining whether stance affects upper extremity, and more particularly

shoulder joint kinetics. Similarly, in spite of recent efforts to elucidate the contribution of

the leg drive to the serve, it remains clear that more comprehensive evaluation of lower

limb joint action and its influence on serve performance is needed.


1. The magnitude of lead and rear knee extension can predict serve technique;

2. Serving with a leg drive (i.e. use of the FU or FB technique) as compared with

minimal leg drive (i.e. the ARM serve) produces higher absolute racquet

velocities during the forwardswing;

3. The ARM serve increases the upper arm peak external rotation moment in



forwardswing;

4. Players using the FU service technique generate higher mean pre-impact and

post-impact shoulder joint compressive forces than FB and ARM servers;

5. Higher average rates of maximum compressive force loading during the

swing phase characterise the FU serve as compared to the FB and ARM

serve;



FU and ARM serves;

7. ARM serves increase the upper arm peak external rotation moment in

cocking;

8. More pronounced transverse plane trunk rotation punctuates the

forwardswing of the ARM serve, while larger amounts of lateral flexion are

involved in the forwardswings of the FU and FB techniques;


6.2 METHODOLOGY

The procedures detailed in Chapter 5 for the kinetic comparison of FS and KS were

followed; any deviation from this methodology is outlined below.

138


Twelve players hit three successful maximal effort, FU and FB FSs to a 1x1 metre target

area bordering the ‘T’ of the first service box. All but two players (n=10) also hit three

successful, flat ARM serves to the same location. This disparity in sample size was due to

two players being unable to hit ARM serves successfully without compromising their

typical serve technique.

The FU stance was the preferred foot arrangement of seven players, while the remainder

of the sample employed FB stances in competition. All subjects were nevertheless

comfortable serving with both stances, indicating that they had either previously

experimented with or used their non-preferred stance in tournament play. Players

assumed their preferred stances in hitting the ARM serves.

Players executed an average of six FU, FB and ARM serves; with never more than four

unsuccessful FU, FB or ARM serves performed by any player. Mean kinematic and kinetic

data were interpreted for each subject’s three successful FU, FB and ARM FSs using the

data analysis methods detailed in Chapter 6.2.2.


Data were treated as in Chapters 4 and 5.1.2, while all service trials were characterised

by the identifiable events and phases defined in Chapter 1. Matlab software was

employed to temporally normalise phases to facilitate comparison between serves.

A discriminant analysis was undertaken to determine the lower limb kinematics that

distinguished serve technique, and by extension, leg drive. Eleven one-way repeated

anovas ascertained statistically significant differences in the kinematic variables

considered to relate to shoulder joint loading in FU, FB and ARM serves. A further nine

anovas were conducted to resolve different shoulder joint load between serves. The

kinetic variables considered to represent load, and thus examined, were:

− upper arm peak external rotation moment in cocking.

139

− peak anterior force in cocking;

− average rate of peak anterior force loading in cocking,

− average rate of maximum compressive force loading during the swing phase,

− mean compressive force and peak internal-external rotation moments during the

forwardswing and follow-through.

On account of the high number of comparisons, a partial Bonferroni correction (p<0.01),

was used to detect statistically significant differences between service techniques. In the

event of a significant F ratio, post hoc Tukey analyses were conducted to best ascertain

variation between the three service techniques at a p<0.05 level.

6.3 RESULTS

6.3.1 Serve Technique as a Function of Lower Limb Kinematics

Discriminant analysis was conducted to determine which lower limb kinematics best

predicted serve technique (i.e. FU, FB and ARM serve). The commonly referenced

maximum front knee flexion angle was included in a stepwise analysis procedure along

with the ranges and peak velocities of hip, knee and ankle extension during the

respective drive phases of both rear and front legs. The one-way comparisons indicated

significant differences on all but one predictor variable (range of rear ankle plantar-

flexion, p=0.136). The lower limb kinematics shown to best discriminate between serve

technique were range of rear knee extension (F=35.582, p<0.001), range of front knee

extension (F=80.908, p<0.001) and peak angular velocity of rear knee extension

(F=25.351, p<0.001; Table 6.1). Two discriminant functions were extracted (F1: Wilks’

Lambda = 0.076, Chi-square = 77.231, p<0.001; F2: Wilks’ Lambda = 0.615, Chi-square

= 14.571, p≤0.001), with their corresponding high eigenvalues (F1: 7.074; F2: 0.625)

and canonical correlations (F1: 0.936; F2: 0.620) – especially of function 1 – indicative of

strong discriminatory power (Table 6.2; Betz, 1987). Indeed, 94.1% of all serves were

correctly classified (Table 6.3), while subsequent cross-validation to avoid any

confounding circular terms of reference still saw 91.2% of all serves classified correctly.

Predictor variable EVENT /

Phase FU FB ARM

Wilks’

Lambda F

140

Mean (SD)

Range of front knee joint

extension (º)

Lead leg

drive

54.11

(11.7)

65.48

(12.6)

33.92

(12.6) 0.161 80.908 *

Range of rear knee joint

extension (º)

Rear leg

drive

59.44

(6.6)

44.83

(8.3)

17.25

(8.5) 0.099 35.582 *

Peak rear knee joint extension


Rear leg

drive

-9.33

(1.2)

-7.24

(0.9)

-3.50

(1.3) 0.076 25.351 *

Table 6.1. Variables included in the stepwise discriminant analysis procedure (* p <0.001).

Eigenvalues Wilks' Lambda

Function Eigenvalue

Canonical

Correlation

Wilks'

Lambda Chi-square Sig.

1 7.074 .936 .076 77.231 .000

2 .625 .620 .615 14.571 .001

Table 6.2. Canonical discriminant functions extracted from the analysis.

Predicted Group Membership %Serve type

1.00 2.00 3.00 Total *#

Foot Up 91.7 8.3 - 100.0

Foot Back - 100.0 - 100.0

Arm - 10.0 90.0 100.0

* 94.1% of original

grouped cases correctly

classified.

# 91.2% of cross-

validated grouped cases

correctly classified.

Table 6.3. Classification of serve type based on discriminant functions.

The descriptive statistics of the three variables included in the discriminant analysis

procedure are also detailed in Table 6.1. With the individual Wilks’ Lambda values close

to 0, most of the variance in the range of rear knee extension, range of front knee

extension and peak angular velocity of rear knee extension is attributable to between

group differences. As compared to the ARM serve, the more dynamic involvement of the

lower limb during the FU and FB technique is clearly evidenced. Further variation also

marked the range of front (FU: 54.11±11.7º; FB: 65.48±12.6º) and rear (FU:

59.44±6.6º; FB: 44.83±8.3º) knee extension as well as the peak angular velocity of rear

knee extension (FU: -9.33±1.2rad.s-1; FB: -7.24±0.9rad.s-1) between the FU and FB

serves. While these three variables best predict serve technique, tennis coaching

141

literature – as intimated above – commonly infers peak front knee joint flexion as

representative of the effectiveness of a player’s leg drive. Consequently, to allow for

more comprehensive comparison to past literature and insights into current coaching

practice, the descriptive statistics for peak front knee joint flexion, computed as part of

the discriminant analysis, are presented in Table 6.4. Again, differences between

techniques point to reduced lower limb involvement in the ARM serve (53.01±17.1º),

and some stance-dependent variation (FB: 85.67±14.5º; FU: 69.92±14.9º) in peak front

knee flexion.

FU FB ARM Predictor variable

EVENT /

Phase Mean (SD)

Maximum front knee joint

flexion (º) MKF 69.92 (14.9) 85.67 (14.5) 53.01 (17.1)

Table 6.4. Excluded from the stepwise discriminant analysis, descriptive statistics of peak front knee joint flexion for the FU, FB and ARM serves.

Figures 6.1-6.3 contrast the lower limb kinematic variables that discriminate between

serve technique. The higher mean range of front knee extension common to the FB

serve is evident in Figure 6.1, while Figures 6.2 and 6.3 confirm the more pronounced

extension of the rear knee joint during the FU serve. The significantly reduced lower limb

drive that characterised the ARM serve is apparent throughout.

142

00 10 20 30 40 50 60 70 80 90 100

Lead Leg Drive (Temporally Normalised)

10

20

30

40

50

60

70

80

90A

ngle

(Deg

rees

)

Foot Up (n =12)Foot Back (n =12)Arm (n=10)

Figure 6.1. Mean extension of the front knee during the lead leg drive phase of the FU, FB and

ARM serves.

00 10 20 30 40 50 60 70 80 90 100

Rear Leg Drive (Temporally Normalised)

10

20

30

40

50

60

70

80

90

Ang

le (D

egre

es)


Figure 6.2. Mean extension of the rear knee during the rear leg drive phase of the FU, FB and ARM serves.

143

-10

-9

100

Rear Leg Drive (Temporally Normalised)

-8

-7

-6

-5

-4

-3

-2

-1

0

1

0 20 40 60 80

Ang

ular

vel

ocity

(rad

.s-1

)

Foot Up (n =12)Foot Back n =12)Arm (n=10)

Figure 6.3. Mean angular velocity of rear knee extension during the rear leg drive phase of the FU, FB and ARM serves.

6.3.2 Effect of Variable Foot Placement and Lower Limb Drive on 3D Racquet

Velocity

A difference was recorded in the maximum pre-impact absolute (F = 5.983, p = 0.006)

racquet velocity between serves, with post hoc tests indicating that the forwardswings of

the FU (43.60 ± 3.0m.s-1, p<0.05) and FB (42.64 ± 3.1m.s-1, p<0.05) techniques were

characterised by higher racquet speeds than those generated in the ARM (39.38 ±

3.4m.s-1) serve (Table 6.5). Pictorial confirmation of this differential development of

mean pre-impact absolute racquet velocity is provided through Figure 6.4. The difference

in the peak pre-impact horizontal racquet velocities (F = 5.274, p = 0.011) generated in

the FU (41.04 ± 2.9m.s-1), FB (39.82 ± 3.2m.s-1) and ARM (36.89 ± 2.9m.s-1) serves

also bordered on significance, while the peak pre-impact vertical racquet velocity trended

higher when players served with a lower limb drive (FS: 30.41 ± 3.4m.s-1; FB: 29.84 ±

3.1m.s-1; ARM: 27.35 ± 1.9m.s-1). Considerably less variation was evident in peak post-

impact horizontal racquet deceleration regardless of foot arrangement or lower limb

drive (F=1.329, p=0.279; FU: 29.38 ± 14.2 m.s-2; FB: 29.63 ± 7.4m.s-2; ARM: 23.18 ±

7.3m.s-2).

144

FU FB ARM Linear racquet

kinematics Phase

Mean (SD) F p

Maximum absolute

velocity (m.s-1)

Forward-

swing

43.60

(3.0)

42.64

(3.1)

39.38

(3.4) * 5.983 0.006

Maximum horizontal

velocity (m.s-1)

Forward-

swing

41.04

(2.9)

39.82

(3.2)

36.89

(2.9) 5.274 0.011

Maximum vertical

velocity (m.s-1)

Forward-

swing

30.41

(3.4)

29.84

(3.1)

27.35

(1.9) 3.331 0.049

Peak horizontal

deceleration (m.s-2)

Follow-

through

29.38

(14.2)

29.63

(7.4)

23.18

(7.3) 1.329 0.279

Table 6.5. Comparison of linear racquet kinematics across FU, FB and ARM serves (* p<0.01).

0

5

10

15

20

25

30

35

40

45

0 10 20 30 40 50 60 70 80 90 100Forwardswing (Temporally Normalised)

Velo

city

(m.s

-1)


Figure 6.4. Comparison of mean absolute racquet velocity during the forwardswing of the FU, FB and ARM serves.

6.3.3 Body Kinematics that Characterise Serve Performance

Angular displacement and velocity data describing upper and lower body motion in the

FU, FB and ARM serves are presented in Table 6.6. Regardless of serve performed, the

upper arm rotated about its longitudinal axis to a similar position of MER (F=0.171,

145

p=0.843; 115º ± 20º), before moving through a comparable amount (≈40º) of internal

rotation to impact (see Figure 6.5). Non-significant differences were also observed in the

subsequent peak pre-impact shoulder joint external rotation (expressed relative to the

thorax) angular velocity in the FU (11.15±4.0 rad.s-1), FB (10.20±3.3 rad.s-1) and ARM

(8.28±2.3 rad.s-1) serves. Of similar homogeny, and thus independent of coordinative

lower limb variation, was the angle of the elevation between the upper arm and thorax

at impact (F=0.148, p=0.863; FU: 107.70º ± 14.6º, FB: 108.61º ± 15.2º, ARM: 110.93

º ± 12.4º).

-140

-120

-100

-80

-60

-40

-20

00 10 20 30 40 50 60 70 80 90 100


Ang

le (D

egre

es)


Figure 6.5. Mean internal rotation of the uppeARM serve.

during

e forwardswing were also uniform, irrespective of stance and lower limb drive.

r arm during the forwardswing of the FU, FB and

Consistent to the FU, FB and ARM serves (F=0.195, p=0.824) was a lateral flexion

separation angle of ≈31º ± 5º at MKF. The ensuing ranges through which subjects

laterally flexed (≈17º ± 6º) and rotated (≈22º ± 12º) their shoulder alignments

th

146

FU FB ARM Kinematic

characteristic

EVENT /

Phase Mean (SD) F p

Maximum external

rotation of the racquet MER (23.5) (17.6) (15.3)

0.171 0.843

shoulder (º)

-118.77 -116.53 -113.93

Peak shoulder joint

longitudinal rotation


Forward-

swing

-11.15

(4.0)

-10.20

(3.3) (2.3) 2.065 0.144

-8.28

Upper arm – thorax IMP 0.148 0.863

elevation angle (º)

107.70

(14.6)

108.61

(15.2)

110.93

(12.4)

Lateral flexion separation MKF

(6.4) (7.0) (4.1) 0.195 0.824

angle (º)

30.45 31.14 32.08

Range of shoulder

alignment lateral flexion Forward-

0.789 0.463

(º) swing

16.61

(5.7)

15.78

(6.7)

19.06

(6.5)

Range of shoulder

a

Forward-0.213 0.810

22.42 21.17 24.68

lignment rotation (º) swing (13.8) (12.5) (11.3)

Table 6.6. Upper and l elate to shoulder joint loading in the FU, FB and ARM serve (* p<0.01).

int Kinetics that Characterise the Performance of the FU, FB

homogenous during this same phase (≈2 ± 1Nm: F=0.923,

=0.408; Figure 6.6).

ower body kinematics that characterise and may r

6.3.4 Shoulder Jo

and ARM Serves

Table 6.7 summarises the descriptive results of the kinetics considered to load the

shoulder joint in the FU, FB and ARM serves. Peak anterior force (FU: -164.52 ± 39.5N;

FB: -166.15 ± 46.0N; ARM: -151.42 ± 28.2Ns-1: F=0.451, p=0.641) and the average

rate at which this force was developed (FU: -202.63 ± 53.0N.s-1; FB: -208.57 ± 52.0N.s-

1; ARM: -169.58 ± 27.1N.s-1: F=2.159, p=0.133) during the cocking phase were

comparable in all three service techniques. Peak external rotation moments were

negligible and similarly

p

No significant difference was ascertainable between the peak internal rotation moments

generated during the forwardswings of the FU (-22.78 ± 5.9Nm), FB (-23.40 ± 7.4Nm)

and ARM (-19.20 ± 3.8Nm) serves. Some variation however, appeared to punctuate the

147

profile of pre-impact compressive force at the shoulder, with a near significant difference

(F = 4.995, p = 0.013) recorded in the average rate of maximum compressive force

loading during the swing phase of the FU (344.29 ± 53.0N.s-1), FB (329.50 ± 61.9N.s-1)

and ARM (272.26 ± 50.1N.s-1) serves. Indeed, the adoption of a less conservative p

value – perhaps justifiable given the study’s originality – would have seen significance

achieved. To this end, the mean pre-impact compressive force also trended higher in the

FU (231.08 ± 47.4N) and FB (220.72 ± 45.6N) serve as compared with the ARM (184.65

29.6N) technique.

44, p=0.275) and peak external rotation moments of

18±8Nm (F=1.237, p=0.304).

FU ARM

±

Regardless of stance and leg drive, the follow-through was characterised by ≈75±30N of

mean compressive force (F=1.3

≈

FB Shoulder joint forces and EVENT /

Mean (SD) F p

moments Phase

Maximum anterior force (N) Cocking -164.52 -166.15 -151.42

0.451 0.641 (39.5) (46.0) (28.2)


anterior force loading (N.s-1) Cocking

-202.63

(53.0)

-208.57

(52.0)

-169.58

(27.1) 2.159 0.133

Peak internal rotation 0.923 0.408

moment (Nm) Cocking

2.23

(1.5)

2.07

(1.6)

1.41

(1.3)

Peak internal rotation Forward-

swing

-22.78 -23.40 -19.20 1.516 0.235

moment (Nm) (5.9) (7.4) (3.8)


compressive force loading -1

Swing 4.995 0.013

(N.s )

344.29

(53.0)

329.50

(61.9)

272.26

(50.1)

Mean compressive force (N) Forward-

3.533 0.041 swing

231.08

(47.4)

220.72

(45.6)

184.65

(29.6)

Mean compressive force (N) Follow-

through 1.344 0.275

87.87

(41.4)

77.49

(33.7)

62.81

(30.3)

Peak external rotation

moment (N1.237 0.304

Follow- 21.76 19.00 14.95

m) through (11.4) (11.00) (6.8)

Table 6.7. Shoulder joint kinetics for the FU, FB and ARM serves (* p<0.01).

148

Figure 6.6 presents a representative example of the upper arm longitudinal rotation

moments inherent to the cocking phase of FU, FB and ARM serves. Figures 6.7 and 6.8

respectively depict the moments and angular velocities of upper arm long-axis rotation

during the forwardswing of the same three serves. As in the FS and KS (see Chapter 5),

throughout the majority of the forwardswing, joint power at the shoulder joint is positive

(Figure 6.9), and therefore consistent with a prevailing upper arm internal rotation

moment and external rotation angular velocity (expressed in the thorax). These profiles

are representative of the upper arm longitudinal rotation that characterised the FU, FB

and ARM service of all players.

-14

Cocking (Temporally Normalised)

-12

-10

-8

-6

-4

-2

0

2

0 10 20 30 40 50 60 70 80 90 100

Mom

ent (

Nm

)

FU Moment

ARM Moment

FB Moment

Figure 6.6. Representative external (+) and internal (-) shoulder joint rotation moments during the cocking of the FU, FB and ARM serves as performed by a high performance player.

149

-25

100

-20

-15

-10

-5

00 20 40 60 80

Forwardswing (Temporally normalised)A

ngul

ar v

eloc

ity (r

ad.s

-1)

FU Ang VelARM Ang VelFB Ang Vel

Figure 6.7. Representative external rotation (expressed in the thorax) angular velocity and

internal rotation moment about the long axis of the upper arm during the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player.

-35

100

-30

-25

-20

-15

-10

-5

00 20 40 60 80


Mom

ent (

Nm

)

FU MomentARM MomentFB Moment

Figure 6.8. Representative internal rotation moments about the long axis of the upper arm during

the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player.

150

-200

0 10 20 30 40 50 60 70 80 90 100


0

200

400

600

800

1000

1200

1400

1600

1800Po

wer

(W)

FU

FB

ARM

Figure 6.9. Representative shoulder joint internal-external rotation power term during the forwardswing phase of the FU, FB and ARM serves as performed by a high performance player.

Figure 6.10 depicts the moment and angular velocity of upper arm internal-external

rotation during the follow-through of a subject’s FU, FB and ARM serve. The product of

the internal rotation upper arm angular velocity (expressed in the thorax) and the

external rotation moment produced the negative shoulder joint longitudinal rotation

power term (Figure 6.11). These profiles are representative of the upper arm

longitudinal rotation that characterised the service follow-through of all players.

151

-30

1

Follow-through (Temporally Normalised)-30

-20

-10

0

10

20

30

1 21 41 61 81 10

Ang

ular

Vel

ocity

(rad

.s-1

)

-20

-10

0

10

20

30

Mom

ent (

Nm

)

FU Ang VelFU MomentARM Ang VelARM MomentFB Ang VelFB Moment

Figure 6.10. Representative angular internal (+) and external (-) rotation velocities as well as internal (-) and external (+) moments about the long axis of the upper arm during the follow-

through of a FU, FB and ARM serves as performed by a high performance player.

-600

100

Follow-through (Temporally Normalised)

-400

-200

0

200

400

600

800

1000

0 20 40 60 80

Pow

er (W

)

FU

FB

ARM

Figure 6.11. Representative shoulder joint internal-external rotation power terms during the follow-through of a FU, FB and ARM serves as performed by a high performance player.

152

6.3.5 Pre-impact Shoulder Joint Loading Predicted by Lower Limb Joint

Kinematics

With the difference in the average rate of maximum compressive force loading in the

swing phase approaching significance, follow-up regression analyses were performed to

ascertain whether variance in this variable could be explained by differential lower limb

serve kinematics. The three lower limb kinematic variables shown to best discriminate

between serve technique in Chapter 6.3.1 were entered as predictor variables into a

stepwise regression, with average rate of pre-impact compressive force loading acting as

the criterion variable. A separate regression analysis was conducted for each serve.

None of these variables was shown to significantly predict the average rate of maximum

compressive force loading in the swing phase of the FU serve, and thus no prediction

model was elaborated. Conversely, the range of front and rear knee extension were

shown to significantly predict this loading variable during the FB (F = 11.764, p = 0.006)

and ARM serves (F=6.608, p=0.033) respectively (Table 6.8). The resultant FB model

realised the following prediction equation: Average rate of maximum compressive force

loading in the swing phase = 566.915 + (-3.626*range of front knee extension), which

explained 54.1% of the variance in the average rate of maximum compressive force

loading during the FB serve.

The equation derived from the ARM model: Average rate of maximum compressive force

loading in the swing phase = 204.29 + (3.94*range of rear knee extension) accounted

for a similar amount of this variable’s variance (45.2%) during the ARM serve.

153

Serve

technique

Predictor

variable

Regression

co-efficient

Beta

weight

Multiple

correlation R value

Range of front knee

extension * -3.626 -0.735 -0.735 0.541

FB

Constant 566.915

Range of rear knee

extension * 3.940 0.673 0.452 0.384

ARM

Constant 204.290

Table 6.8. Results of step-wise regression analyses on selected lower limb kinematics and the average rate of maximum compressive force loading in the swing phase of the FB and ARM

serves (* p<0.05).

6.4 DISCUSSION

Analyses of serving mechanics have demonstrated variation in the GRF’s produced by

the FU and FB techniques. The relationship between the effectiveness of a player’s leg

drive and upper extremity joint kinetics has also been a source of recent investigative

interest. However, there has been no previous attempt to systematically corroborate the

effects of variable stances and lower limb kinematics on shoulder joint loading.

6.4.1 Lower Limb Kinematics that predict Serve Technique


1. The magnitude of lead and rear knee extension can predict serve technique.

Discriminant analysis of 13 kinematic variables describing the lower limb joint motion of

the serve revealed range of rear knee extension, range of front knee extension and peak

angular velocity of rear knee extension to discriminate between the FU, FB and ARM

serve with ≈95% accuracy. In practice, extrapolation of this finding may indicate that

observation of these three variables should provide coaches with sufficient kinematic

information from which to assess a player’s leg drive.

As compared to the FU and FB techniques, the ARM serve’s significantly reduced

magnitudes and rates of angular change in knee extension confirmed that subjects

successfully minimised their leg drive. In turn, the fact that players using the FB stance

154

went through greater front knee joint extension (65.48±12.6º; FU: 54.11±11.7º) is

likely related to this variable’s strong positive correlation (0.910) with peak front knee

joint flexion, and a probable by-product of this stance’s wider base of support permitting

greater squat depth. Conversely, larger ranges of rear knee joint extension were

observed in the FU technique (59.44±6.6º; FB: 44.83±8.3º), and when interpreted

alongside similarly augmented peak angular velocities of rear knee joint extension (FU:

-9.33±1.2 rad.s-1; FB: -7.24±0.9 rad.s-1), may account for some of the higher vertical

GRF’s reported to characterise serves performed with a FU stance (Elliott and Wood,

1983; Bahamonde and Knudson, 2001).

Of significant note is that the abovementioned positive correlation between maximum

front knee joint flexion, and its subsequent extension, suggests that coaches are justified

in attending maximum flexion of the front knee when assessing leg drive. To this end,

the FU and FB techniques were characterised by maximum front knee joint flexion

angles that were similar to those reported by Bartlett et al. (≈111º; 1994) and within the

boundaries of acceptability recommended by Elliott and Alderson (2003); those of the

ARM serve (≈55º), naturally enough, were not. Importantly, it would be remiss of

coaches to only interpret this variable; particularly as it was shown to correlate poorly

with both the range (0.300) and peak angular velocity (0.093) of rear knee extension.

Of similar applied interest is that, as inferred in Chapter 6.1, the length of a baseball

pitcher’s stride forward is known to approximate 90% of the pitcher’s height. As part of

this stride forwards, Dillman et al. (1993) reported that the stride foot lands lateral to

the back heel and in a position of adduction. Young et al. (1996) suggested that failure

to ‘set-up’ accordingly can result in incorrect femur position, premature hip and/or trunk

rotation, disrupted lumbo-pelvic rhythm and compromised spinal muscle activation such

that the role of the shoulder musculature in velocity generation is amplified (Young et

al., 1996). It’s conceivable that similar ideal ‘set-ups’ epitomise serve performance,

whereby the development of GRF’s, shoulder joint loads and racquet velocity is optimal.

Indeed, where Bartlett et al. (1994) originally raised the prospect of a most

advantageous maximum front knee flexion angle, optimal amounts of foot separation

and/or lower limb involvement likely proliferate.

155

To summate, short of measuring lower limb joint kinetics, evaluation of the range of

front and rear knee joint extension as well as the peak angular velocity of rear knee joint

extension best discriminates between serve technique. Collectively, these variables can

also be interpreted to provide a more representative measure of leg drive, than the

angle of maximum front knee joint flexion alone.


1. The magnitude of lead and rear knee extension in combination with peak

angular velocity of rear knee extension can predict serve technique.

6.4.2 Relationship between Variable Foot Placement and Lower Limb Drive on

3D Racquet Velocity


2. Serving with a leg drive (FU or FB technique) as compared to minimal leg

drive (ARM serve) produces higher absolute racquet velocities during the

forwardswing.

As expected, the FU (43.60 ± 3.0m.s-1) and FB (42.64 ± 3.1m.s-1) technique generated

significantly higher pre-impact peak absolute racquet velocities than the ARM serve

(39.38 ± 3.4m.s-1). When using a leg drive, the players in this sample developed similar

absolute racquet velocities as those of previously analysed professional players

(≈38m.s-1; Chow et al., 2003a). Peak pre-impact horizontal (≈40m.s-1) and vertical

(≈30m.s-1) racquet velocities also trended higher when facilitated by a leg drive; with

their magnitudes again comparing favourably to those (horizontal ≈33m.s-1; vertical

≈17m.s-1) previously reported in the literature (Chow et al., 2003a; Elliott et al., 2003).

It would thus appear that racquet speed is generated independent of stance (i.e. in

agreement with Elliott and Wood, 1983), but significantly affected by differential leg

drive. This finding lends some support to the well documented but largely anecdotal

contribution of lower limb drive to the development of racquet velocity.

156

The study of Elliott et al. (2003) ranks as the most recent of a small number of

investigations to have compared service speed between players with variable lower limb

kinematics. In some contrast to the current study, these researchers found no significant

difference between post-impact ball velocities (162km.hr-1), irrespective of the

proficiency of a player’s leg drive. As previously discussed, this finding may be

confounded by their potentially misrepresentative classification of leg drive. A similar

methodological flaw appears to emasculate the link between the FB serve and the

production of higher post-impact ball speeds proposed by Bartlett et al. (1994). That is,

as higher-ranked players were reported to use the FB technique and lower-ranked

players tended to employ the FU technique, it is unclear whether the difference in ball

velocity was due to stance or level of play.

Chow et al. (2003a) demonstrated variation between the horizontal, vertical and lateral

racquet velocities that punctuated the follow-through phase of first and second serves.

However, with peak post-impact horizontal deceleration of the racquet comparable in the

FU, FB and ARM serve, indications are that lower extremity joint motion has a minimal

effect on racquet deceleration during the follow-through.


2. Serving with a leg drive is necessary to produce maximum pre-impact

absolute racquet velocities.

6.4.3 Variation in Body Kinematics in the FU, FB and ARM Serves


3. More pronounced transverse plane trunk rotation characterises the

forwardswing of the ARM serve, while larger amounts of lateral flexion are

involved in the forwardswings of the FU and FB techniques.

Shoulder and pelvis tilt prior to cocking is a feature of FS technique, irrespective of foot

arrangement or lower limb drive. As documented in Chapter 5, this tilted alignment

facilitates the development of angular momentum through lateral trunk flexion during

the forwardswing, which as demonstrated by Bahamonde (2000), is key to high velocity

157

serving. Furthermore, this homogeneity in lateral flexion separation angle at MKF

indicates that high performance players endeavour to position their shoulders relative to

their pelvis so as to always derive benefit from subsequent lateral trunk flexion.

With substantiated differences in the leg drive that characterised the FU, FB and ARM

techniques, variation in the magnitude of upper arm MER could be expected. That is,

given the theorised key role of lower limb drive in forcing the racquet arm and racquet

down and away from the body (Elliott et al., 1986; Elliott, 1988), a reduction in upper

arm MER was anticipated in service techniques involving less dynamic leg actions.

Interestingly however, with MER of the upper arm ranging from 115º-120º in all serve

techniques, it appears that leg drive plays a lesser role in increasing racquet

displacement (Elliott, 2001) and assisting upper arm external rotation than hypothesised.

The comparable peak upper arm external rotation velocities (expressed in the thorax)

during the forwardswings of the FU (-11.15 ± 4.0 rad.s-1), FB (-10.20 ± 3.3 rad.s-1) and

ARM (-8.28 ± 2.3 rad.s-1) serves further suggest that the value of leg drive may be more

related to its interaction with and effect on other, preceding links in the ‘kinetic chain’,

such as the trunk. Paradoxically, the issue of misrepresentative marker placement,

discussed in detail in Chapters 5 and 8 cannot be discounted #. More considered

exploration of the relationships between lower limb and trunk motion should nonetheless

be advanced as a source of future research interest.

Despite the abovementioned sychronicity in shoulder and pelvis lateral alignments at

MKF, it was proposed that more shoulder alignment rotation and less shoulder alignment

lateral flexion would feature in the forwardswing of the ARM serve. This posited,

preferential rotation was based on the notion that reduced leg drive would mitigate

lateral flexion in the ARM serve and thus predispose players to rotate increasingly about

the long axis of their trunks. However, contrary to expectations, comparable amounts

(F=0.789, p=0.463) of left lateral shoulder alignment flexion punctuated the

forwardswings of all serves (≈17º ± 7º). As alluded to above, indications are thus that

high-performance players laterally flex to generate angular momentum about their

trunk’s ab-adduction axis, regardless of service technique. In like kind, players were also

158

shown to rotate their shoulder alignments independent of lower limb coordination

throughout the forwardswing (FU: 22.42º ± 13.8º; FB: 21.17º ± 12.5; ARM: 24.68º ±

11.3, F=0.213, p=0.810). Indeed, if anything, a trend suggested players laterally flexed

and rotated their shoulder alignments, and by extension their trunks, to greater extents

in the ARM serve. In a practical context, this may be indicative of a compensatory

increase in 3D trunk rotation among players who serve with a reduced lower limb drive.

As in the FS and KS (Chapter 5), the angle of upper arm elevation with respect to the

thorax approximated 110º at racquet-ball impact in all three service techniques.

Consistent with the ≈100º of shoulder joint abduction previously observed to

characterise the professional tennis serve (Fleisig et al., 2003), this alignment of the

upper arm and thorax also borders the 100º ± 10º shown by Matsuo et al. (1999) to

produce maximum ball velocity and minimum shoulder joint loading in baseball pitching.


3. Similar amounts of transverse plane trunk rotation and lateral flexion

characterise the forwardswings of the ARM, FU and FB serves.

6.4.4 Effect of Variable Lower-Extremity Joint Kinematics on Shoulder Joint

Loading: Implications for Injury and Performance.




FU and ARM serves.

5. ARM serves increase the upper arm peak external rotation moment in



forwardswing;

6. FB serves produce average rates of maximum compressive force loading

during the swing phase that are higher than in the ARM serve but lower

compared to the FU serve;

# Retrospective analysis and discussion of the magnitude of shoulder joint internal-external rotation – as related to the UWA marker set – is undertaken in Chapter 5.4.2 and Appendix A.

159

7. Players using the FB service technique generate lower mean pre-impact and

post-impact shoulder joint compressive forces than when using the FU serve,

but higher than with the ARM serve;

8. Similar peak external rotation moments are produced in the follow-through,

irrespective of stance and leg drive.

Not until the 2003 study of Elliott et al. had there been an attempt to quantify the link

between lower limb coordination and upper extremity joint loading in the serve.

Empirically, the significance of a relationship had been suspected, yet analyses of the

serve’s lower limb mechanics had traditionally attended the association between GRF

and stance (Elliott and Wood, 1983; Bahamonde and Knudson, 2001). The current

study, against expectations, has highlighted that variation in stance and leg drive

appears to have little effect on the shoulder joint kinetics that characterise the FU, FB

and ARM serves. However, as in Chapter 5’s comparison of the FS and KS, selected

racquet and body kinematics differ between serves, raising the prospect of technique-

related kinetic differences propagating at other joints.

6.4.4.1 Cocking

Courtesy of the FB technique’s substantiated more horizontal lower limb propulsion

(Elliott and Wood, 1983; Bahamonde and Knudson, 2000), it was proposed that players

would be subject to heightened shoulder joint anterior loads during the cocking phase of

this serve. However, as players tolerated similar peak anterior forces (≈ -160 ± 35N)

and developed those forces at comparable rates (-150N.s-1 to -200N.s-1) during the FU,

FB and ARM serves, this contention was not supported. With previous investigations

preferring to document maximum anterior forces during the forwardswing of the serve,

comparative comment is limited. Nonetheless, these data are consistent with the peak

anterior force and average rate of peak anterior force loading reported to characterise

the cocking phase of the FS (-167N; 208N.s-1) and KS (-160N; 190N.s-1) in Chapter 5. In

light of these comparable anterior shoulder joint loads, the passive structures (i.e. GH

joint’s anterior capsule and ligaments) that limit forward translation of the humeral head,

and thus help to prevent GH instability, would appear similarly stressed near MER

independent of stance and leg drive.

160

In demonstrating players with more effective leg drives to record lower internal rotation

moments at MER (43.7 Nm) than players with less effective drives (57.8 Nm), Elliott et

al. (2003) suggested that the more effective group benefitted from a superior trunk-

shoulder inertial transfer to facilitate external rotation of the upper arm. It was further

inferred that players whose knees were more extended at MER (i.e. representative of

less effective leg drives) more actively engaged their upper arm external rotator

musculature to achieve MER. Contrastingly and in deference to this latter assertion,

players in the current study were shown to generate similar, negligible peak external

rotation moments from MKF to MER (≈ -3 ± 1Nm), irrespective of lower limb drive and

stance. Indeed, as evidenced by Figure 6.6, internal rotation moments were produced

throughout the majority of the cocking phase. Similar dominant internal rotation

moments have been shown to decelerate external rotation of the upper arm during the

baseball pitch (Fleisig et al., 1995).

This contrast in shoulder joint kinetic patterning between the two studies may relate to

the former investigation’s ambiguous definition of leg drive. That is, as aforementioned,

front knee flexion angle at MER likely misrepresents the ebullience of a player’s leg drive,

such that a more valid delineation in Elliott et al. (2003) may have also produced

comparable internal rotation moments at MER, irrespective of leg action. To this end, the

results of the current study suggest that players can benefit from trunk-shoulder inertial

transfer as long as some lower limb drive (and trunk rotation) is present. Future

examination of the relationships between lower limb and trunk mechanics, as well as the

ensuing interaction of the trunk and upper extremity could elaborate performance

models to substantiate just how much leg drive and trunk rotation is required to derive

such a benefit.


With the action of the legs long considered to influence the longitudinal rotation of the

upper arm, and, in turn serving speed, examination of the kinetics responsible for

humeral rotation during the forwardswings of serves with contrasting lower limb drives

was expected to unveil technique-related differences. Indeed, some evidence already

161

exists to implicate ineffective leg actions in the production of higher peak internal

rotation moments during this phase (65Nm, versus 55Nm following more effective

drives; Elliott et al., 2003). The comparable peak internal rotation moments that typified

the forwardswings of the FU (-22.78 ± 5.9Nm), FB (-23.40 ± 7.4Nm) and ARM (-19.20

± 3.8Nm) serves in this study however, challenge the posited amplification of

longitudinal rotation moment accompanying reduced leg drive. Furthermore, similar

internal rotation moments and upper arm external rotation angular velocities (expressed

in the thorax) combined to produce increasing positive longitudinal rotation power terms

throughout the service forwardswing (Figures 6.7 - 6.9), irrespective of stance and lower

limb drive.

As elaborated on in Chapter 5, comparison of the peak pre-impact internal rotation

moments (≈20 Nm) of the FU, FB and ARM serves with those previously reported as

common to the service forwardswing (30-100 Nm; Noffal and Elliott, 1998; Bahamonde,

1999; Elliott et al., 2003) or baseball pitch (≈ 70 Nm; Fleisig et al., 1999) is likely

confounded by disparate data collection and modelling procedures. Here, it is

nonetheless worth reflecting on the observed difference in absolute racquet velocity; for

in spite of the demonstrated importance of upper arm internal rotation to the serve

(Elliott et al., 1995), it would appear that other segmental rotations, or a combination

thereof, better explain variation in the development of high performance players’ service

speeds. Certainly some of Bahamonde’s work has shown dichotomous trunk angular

momentum to partially account for differences in the service velocities of college level

players (Bahamonde, 2000). Rather than specific, independent segmental contributions

determining racquet velocity however, it remains probable the sequential coordination of

the entire ‘kinetic chain’, or at a minimum, its foremost components, is needed for the

development of high 3D racquet velocities.

For the upper arm to safely and continually rotate at high speeds overhead, Blevins

(1997) emphasised the key role of compressive forces in centering the humeral head in

the glenoid. Augmentation of humeral distraction, and thus the need to generate higher

pre-impact mean compressive forces and average rates of maximum compressive force

loading were hypothesised by-products of the FU technique’s more pronounced vertical

162

lower limb drive (Elliott and Wood, 1983; Bahamonde and Knudson, 2000). Short of

offering this contention categorical support, a trend did suggest that the application and

magnitude of compressive forces was magnified when players served with a FU stance.

This same trend offered further support to the hypothesis that the ARM serve, with its

minimal leg drive, would produce lower average rates of maximum compressive force

loading (272.26 ± 50.1N.s-1) and mean compressive forces (184.65 ± 29.6N) as

compared to when serving with a leg drive (i.e. FU: 344.29 ± 53.0N.s-1, 231.08 ±

47.4N; FB technique: 329.50 ± 61.9N.s-1, 220.72 ± 45.6N). Together, these results hint

at the profile of pre-impact compressive force loading being influenced by the leg drive

and stance that players employ when serving. Equally, with the ARM serve characterised

by lower absolute racquet velocities than the FU and FB techniques, it seems likely that

velocity generation is to some extent load-related. Further extrapolation implies

secondary impingement resulting from high compressive force loading conditions, as

more likely when serving with a leg drive, and increasingly probable when that leg drive

is produced from a FU stance.


As depicted in Figure 6.10, continued internal rotation of a player’s upper arm is resisted

by an external rotation moment during the early service follow-through, regardless of

preceding lower limb coordination. Consistent with the notion of eccentric rotator cuff

muscle activity precluding distraction, horizontal adduction and internal rotation of

humerus during the follow-through (Jobe et al., 1984; Fleisig et al., 1996), resultant

shoulder joint longitudinal rotation power is negative (Figure 6.11). Moreover,

interpretation of these figures indicates that the pattern of eccentric external rotator

muscle loading is independent of lower limb drive. Conceptually, it could thus be

assumed that subtle variations in trunk motion affect shoulder joint kinetics to a greater

extent than variable lower limb mechanics.

Of similar incredulity was that no differences were recorded in the investigated discrete

post-impact shoulder joint kinetics of the FU, FB and ARM serves. That is, slowing of the

above-mentioned continued internal rotation of the upper arm was assisted by

comparable peak external rotation moments (≈18±8Nm) irrespective of stance and leg

163

drive. Variable lower limb coordination also had little bearing on the mean compressive

forces (≈75±30N) generated to resist post-impact humeral distraction. Interestingly,

where there was some suggestion of technique-related differences in pre-impact mean

compressive force, this variation was not as evident post-impact. No clear explanation is

apparent, yet the less repeatable nature of players’ follow-throughs, as highlighted in

Chapter 4, may be partly responsible. As alluded to in Chapter 5, this course of studies is

the first attempt to analyse 3D shoulder joint kinetics during the follow-through phase of

the serve such that comparisons to previous tennis literature are not possible.


4. No difference exists in the peak, shoulder joint anterior forces and average

rates of peak anterior force loading generated during the cocking phase of

FB, FU and ARM serves;

5. Peak upper arm external rotation moments in cocking and peak upper arm

internal rotation moments during the forwardswing are produced independent

of stance and leg drive;

6. Mean pre-impact and post-impact shoulder joint compressive forces are

comparable irrespective of stance and leg drive;

7. FB serves produce similar average rates of compressive force loading during

the swing phase as in the ARM and FU serve;

8. Similar peak external rotation moments are produced in the follow-through,

irrespective of stance and leg drive.

6.4.5 Lower Limb Kinematics that Predict the Average Rate of Pre-impact

Maximum Compressive Force Loading



In light of the observed differences in lower limb kinematics as well as the trend for

varied average rates of pre-impact maximum compressive force loading in the FU, FB

164

and ARM serves, it may be reasonable to assume some underlying relationship between

stance/leg drive and this indicator of shoulder joint load. Certainly the results of the

regression analyses performed for the FB and ARM serves lend some support to this

assertion. That is, the range of front knee extension was shown to share a negative

relationship with, and explain ≈54% of the variance in the average rate of maximum

compressive force loading during the swing phase of the FB technique. Likewise, the

predictive equation elaborated for the ARM serve accounts for ≈45% this variable’s

variance when serving with minimal leg drive; and infers that the average rate of

maximum compressive force loading increases along with rear knee joint extension in

this serve.

In contrast to the FB and ARM serves, variables other than the range of front and rear

knee extension, and peak angular velocity of rear knee extension appear to explain the

variance in the average rate of maximum compressive force loading in the swing phase

of the FU technique. This comes as some surprise as the rationale for hypothesising

augmented compressive force loading in the FU serve related to the greater vertical

lower limb propulsion (and by extension presence of leg drive), previously reported to

punctuate serving with a FU stance (Elliott and Wood, 1983; Bahamonde and Knudson,

2000). These findings nevertheless reveal different relationships between lower limb

kinematics and shoulder joint kinetics in the FS depending on the stance adopted and

leg drive used.


9. The range of front and rear knee joint extension account for significant

amounts of the variance in the average rates of pre-impact compressive force

loading in the FB and ARM serve respectively. Variables other than those

considered to represent leg drive explain this variable’s variance in the FU

serve.

6.5 CONCLUSION

Coaches should be able to glean sufficient information from a server’s range of front and

rear knee joint extension as well as his/her peak angular velocity of rear knee joint

extension to ascertain the stance and quality of leg drive used. Indeed, when facilitated

165

by a leg drive, high-performance players can generate similar absolute pre-impact

racquet velocities using either a FU or FB service stance. Devoid of a leg drive however,

players are less capable of developing high absolute racquet velocities, regardless of

service stance.

Interestingly, comparable shoulder joint kinetics evolved from the differential lower limb

mechanics that characterised the FU, FB and ARM techniques. However, a trend in the

average rates of maximum compressive force loading and the mean compressive forces

applied to the shoulder joint throughout the swing to impact did hint at some technique-

related differences. Indeed, the noted dichotomy in the absolute racquet velocities

between serves could point to serve speed increasing along with pre-impact shoulder

joint load. More specifically, the application and magnitude of compressive forces needed

to resist pre-impact humeral distraction appear most pronounced in the FU serve and

least so in the ARM serve. To this end, there is also some evidence to suggest that the

lower limb kinematics that relate to the generation of these compressive force profiles

differ with variation in stance and leg drive.

166

CHAPTER 7: SHOULDER JOINT KINETICS OF THE ELITE WHEELCHAIR TENNIS

SERVE – A CASE STUDY

7.1 INTRODUCTION

As in professional able-bodied tennis, the development of high racquet speeds is a goal

of the elite wheelchair game. Interestingly, wheelchair players generate these high

racquet speeds without the assistance of the lower limb drive that features so

prominently in able-bodied stroke production. A coordinative difference most pronounced

in the serve, where this propulsion of the lower extremities is believed to increase

racquet displacement and optimise subsequent segmental action (Elliott et al., 1986;

Bartlett et al., 1994; Girard et al., 2005). In turn, shoulder joint loading is reportedly

reduced (Elliott et al., 2003) and the development of racquet speed facilitated. So, given

the conceptual importance of lower limb drive to the able-bodied serve, is it conceivable

that wheelchair players experience comparatively greater shoulder joint loading when

serving?

The ITF consider any player to have a medically diagnosed permanent physical disability

resulting in a substantial loss of function in one or both lower extremities as eligible to

compete on the wheelchair tennis tour. Where some players may be amputees, others

will suffer from perthes disease, and another group of competitors will have sustained

spinal cord injuries. Among this latter group of players, generally the higher the level of

spinal cord injury, the more affected the player. More complete injuries (or breaks of the

spinal cord) also exacerbate impairment of bilateral neuromuscular control and function

below the level of the injury. Most elite wheelchair tennis athletes that have sustained

spinal cord injuries at or below the thoracic level therefore possess varying levels of

physical function (Bullock, personal communication, 2006). For example, some

wheelchair tennis players can strengthen their trunk and abdominal muscles to benefit

their movement and stroke play while other players cannot. Obviously this inability to

optimally engage such an important “link” in the movement chain has ramifications for

serve performance; likely affecting shoulder joint kinetics and more significantly, the

prospect of injury (Goosey-Tolfrey and Moss, 2005).

167

The shoulder is a key joint in wheelchair locomotion and most commonly implicated in

injury among virtually all wheelchair populations (Ferrara and Davis, 1990). Given the

joint’s prominent role in tennis stroke production it is of little surprise that these

pathologies also permeate this tennis playing populace. Several aetiologies have been

proposed (Burnham et al., 1993; Pluim and Bullock, 2003; Bernard et al., 2004), so

quantification of the shoulder joint kinetics that characterise the wheelchair serve would

further facilitate diagnostics, while also assisting coaches evaluate the efficacy of their

current technical instruction and exercise prescription.

The hypotheses for this case study are:

1. WFSs develop higher peak pre-impact horizontal racquet velocities, while

WKSs generate higher maximum pre-impact lateral racquet velocities;


absolute and horizontal pre-impact racquet velocities as compared with

serving with a leg drive (i.e. the able-bodied serve);

3. The magnitude of MER of the upper arm is independent of wheelchair serve

type but lower when serving with no leg drive (i.e. the wheelchair serve) as

compared with serving with leg drive or with minimal leg drive (i.e. the able-

bodied serve);


trunk rotation characterises the forwardswing of wheelchair serves;

5. Although independent of wheelchair serve type, higher peak upper arm

external rotation moments are generated during the cocking phase of the

WFSs and WKSs than in the able-bodied FS and KS;


forwardswing of the WFS as compared with the WKS;

7. Relative to absolute racquet velocity, higher peak shoulder joint internal

rotation moments, mean compressive forces and average rates of

compressive force loading are produced in the forwardswings of the

wheelchair serves as compared with the able-bodied serves;

168


mean compressive forces are generated independent of wheelchair serve

type;



players who employ no leg drive (i.e. the wheelchair serves) as compared

with some leg drive (i.e. able-bodied serves).

7.2 METHODOLOGY

The procedures detailed in Chapter 5 for the kinetic comparison of the able-bodied FS

and KS were followed. All methodological differences or addendums are outlined below.


Two, top-30 professionally ranked wheelchair players were fitted with customised UWA

upper-body marker sets prior to hitting three successful maximal effort, WFS and WKS to

a 1x1 metre target area bordering the ‘T’ of the first service box.

Subject 1, a quarter-finalist at the Australian Wheelchair Open in 2005, suffered from an

incomplete injury at the T12 level but a complete break at the level of L1. Subject 2, on

the other hand, suffered from an incomplete T10 spinal cord injury, and thus exhibited

more trunk and lower limb function.

Players hit an average of six WFSs and six WKSs; with never more than four

unsuccessful WFSs or WKSs executed by either player. Mean kinematic and kinetic data

for each subject’s three successful WFS and WKS were interpreted using the data

analysis procedures highlighted below.


Data were treated according to the procedures detailed in Chapters 4 and 5.1.2, while

the events and phases defined in Chapter 1 were identified for all service trials. Phases

were again temporally normalised with Matlab software to facilitate data analysis.

169

With the small sample size prohibiting most statistical methods, comparative descriptive

analyses of the two players’ mean WFS and WKS kinematics and kinetics were

undertaken. Comparison of these data to those reported in Chapter 5 to characterise the

high-performance able-bodied FS and KS was also pursued.

7.3 RESULTS

7.3.1 Effect of Wheelchair Serve Type on 3D Racquet Velocity

Comparable mean maximum absolute racquet velocities (≈31.95 ± 1.0 m.s-1) were

generated during the forwardswings of both players’ WFS and WKS (Table 7.1).

However, higher pre-impact horizontal racquet velocities characterised the WFS

(≈29m.s-1; WKS: ≈26m.s-1), while the lateral racquet velocities that players developed at

impact in the WKS (S1: -4.10 ± 1.3 m.s-1; S2: -14.84 ± 1.4 m.s-1) were effectively

double those generated at WFS impact (S1: 0.91 ± 0.5 m.s-1; S2: -7.77 ± 4.0 m.s-1).

Pictorial confirmation of these differential profiles of horizontal and lateral racquet

velocity is provided through Figures 7.1 and 7.2 respectively.

Subject 1 developed lower absolute, horizontal and lateral racquet velocities than

Subject 2. Observable differences also mark the racquet velocities produced in the

wheelchair serves as compared with the able-bodied serves (Table 7.1).

WFS WKS Able-bodied

S1 S2 S1 S2 FS KS Linear racquet

kinematics

Phase /

EVENT Mean (SD)

Maximum absolute

velocity (m.s-1)

Forward-

swing

30.68

(0.4)

33.20

(0.7)

30.76

(0.4)

32.56

(0.7)

43.22

(3.1)

40.28

(2.9)

Maximum horizontal

velocity (m.s-1)

Forward-

swing

28.01

(1.0)

30.53

(0.7)

26.10

(0.4)

26.15

(0.3)

40.58

(3.4)

35.04

(2.9)

Lateral velocity (m.s-1) IMP 0.91

(0.5)

-7.77

(4.0)

-4.10

(1.3)

-14.84

(1.4)

-1.42

(5.5)

-10.19

(2.3)

Table 7.1. Comparison of mean linear racquet kinematics between the WFS (n=3) and WKS (n=3) as performed by two subjects (S1 and S2), as well as in contrast to the mean able-bodied

FS and KS (n=12).

170

-15Forwardswing (Temporally Normalised)

-10

-5

0

5

10

15

20

25

30

35

0 10 20 30 40 50 60 70 80 90 100

Velo

city

(m.s

-1)

WFS Subject 1

WKS Subject 1

WFS Subject 2

WKS Subject 2

Figure 7.1. Comparison of mean absolute horizontal racquet velocity during the forwardswing of the WFS and WKS, as performed by both subjects.


-15

-10

-5

0

5

10

0 10 20 30 40 50 60 70 80 90 100

Velo

city

(m.s

-1)

WFS Subject 1

WKS Subject 1

WFS Subject 2

WKS Subject 2

Figure 7.2. Comparison of mean absolute lateral racquet velocity during the forwardswing of the WFS and WKS, as performed by both subjects.

171

7.3.2 Upper-Extremity Kinematics that Describe the Wheelchair FS and KS

Kinematic data describing 3D shoulder joint motion in the WFS and WKS are presented

in Table 7.2. Wheelchair players were observed to rotate their upper arms to similar

positions of MER (93º±2º) irrespective of wheelchair serve type, but through ≈20º less

external rotation than in the able-bodied FS and KS (Figures 7.3 - 7.4). At MER, the two

wheelchair players recorded plane of elevation angles from 145º to 165º, indicating that

neither player rotated his humerus beyond the alignment of his shoulders.

The elevation of the upper arm with respect to the thorax at impact was consistent

between wheelchair players and independent of wheelchair serve type (≈120º ± 4.0º).

During the forwardswing to impact, slightly higher peak upper arm external rotation

velocities (expressed in the thorax) were developed by Subject 2, while a similar trend

characterised the performance of the WKS (3.5 ± 0.8 rad.s-1) as compared with the WFS

(5.5 ± 1.0 rad.s-1). Notable differences also punctuated the wheelchair players’ pre-

impact shoulder alignment and by extension trunk motion (Table 7.3). That is, while

both players laterally flexed their shoulder alignments through similar ranges (≈15º)

irrespective of serve performed, more pronounced shoulder alignment forward flexion

and rotation was observed in the serves of Subject 2 (forward flexion: ≈26º; rotation:

≈50º) than those of Subject 1 (forward flexion: ≈11º; rotation: ≈8º). Interestingly, even

greater dichotomy was evident in the 3D orientation of both subjects’ shoulder

alignments at impact.

172

WFS WKS Able-bodied

S1 S2 S1 S2 FS KS Kinematic

characteristic

EVENT /

Phase Mean (SD)

Maximum external rotation

of the racquet shoulder (º) MER

-93.63

(0.8)

-91.40

(0.3)

-96.80

(1.2)

-94.32

(2.0)

-115.86

(18.3)

-119.04

(18.3)

Upper arm plane of

elevation angle (º) MER

-142.13

(5.2)

-150.86

(5.1)

-144.94

(2.3)

-165.82

(0.7)

-158.93

(8.5)

-161.45

(10.2)

Peak shoulder joint

longitudinal rotation


Forward-

swing

-2.57

(1.7)

-4.63

(0.1)

-4.83

(1.2)

-6.07

(1.2)

-10.80

(4.7)

-10.60

(3.1)

Upper arm – thorax

elevation angle (º) IMP

122.77

(5.0)

119.32

(5.5)

117.78

(1.1)

121.22

(5.2)

108.85

(14.1)

107.72

(19.7)

Table 7.2. Mean shoulder joint kinematics, related to shoulder joint loading, that characterise the WFS and WKS, and as compared with the able-bodied FS and KS.


-100

-80

-60

-40

-20

00 10 20 30 40 50 60 70 80 90 100

Ang

le (D

egre

es)

WFS Subject 1

WFS Subject 2

Able-bodied FS

Figure 7.3. Mean internal rotation of the upper arm during the forwardswing of the WFS and

able-bodied FS.

173


-120

-100

-80

-60

-40

-20

00 10 20 30 40 50 60 70 80 90 100

Ang

le (D

egre

es)

WKS Subject 1

WKS Subject 2

Able-bodied KS

Figure 7.4. Mean internal rotation of the upper arm during the forwardswing of the WKS and

able-bodied KS.

In comparison to able-bodied players, wheelchair players appear to employ less forward

trunk flexion (wheelchair: 10º-30º; able-bodied: ≈50º) but similar amounts of lateral

trunk flexion (≈15º) during the service forwardswing. The contrasting ranges of shoulder

alignment rotation suggest that the serves of individual wheelchair players involve

greater variation in long-axis trunk rotation (5º-60º) than the serves of able-bodied

players (≈20º).

174

WFS WKS Able-bodied

S1 S2 S1 S2 FS KS Kinematic characteristic EVENT /

Phase Mean (SD)

Range of shoulder alignment

lateral flexion (º)

Forward-

swing

14.13

(1.1)

13.55

(3.1)

16.52

(3.1)

20.74

(9.1) 16.41 (6.13)

9.59 (4.1)


forward flexion (º)

Forward-

swing

11.43

(1.0)

21.90

(3.8)

11.27

(7.2)

31.82

(5.6) 45.78 (10.8)

51.91 (8.2)


rotation (º)

Forward-

swing

11.13

(3.7)

39.25

(6.8)

5.94

(0.6)

60.67

(1.7) 22.45 (13.7)

17.66 (9.3)

Shoulder alignment lateral

flexion (º)

IMP -26.47

(5.3)

-42.45

(5.8)

-20.45

(2.7)

-43.00

(3.5)

-41.68

(7.8)

-33.43

(10.2)

Shoulder alignment forward

flexion (º)

IMP 23.52

(2.0)

47.67

(9.4)

19.73

(5.0)

54.19

(9.4)

56.40

(15.1)

67.23

(9.4)

Shoulder alignment rotation

(º)

IMP -156.45

(6.5)

-110.35

(10.5)

-169.81

(0.7)

-104.85

(12.3)

-41.55

(18.5)

-64.36

(14.3)

Table 7.3. Mean 3D motion of shoulder alignment during the WFS and WKS, as well as in comparison with the able-bodied FS and KS.

7.3.3 Shoulder Joint Kinetics that Characterise the Wheelchair FS and KS

The descriptive statistics of the kinetic variables considered to load the shoulder joint in

the WFS and WKS are presented in Table 7.4. For comparative purposes, the same

mean data pertaining to the performance of the able-bodied FS and KS are also

included.

As in the able-bodied serves, homogeneous and negligible peak external rotation

moments (≈1.0 ± 1 Nm) assisted external rotation of the upper arm during the cocking

phase of the WFS and WKS. Subsequent peak upper arm internal rotation moments

(≈15Nm) were also similar in the forwardswings of both the WFS and WKS. However,

some inter-subject variation appeared to exist, whereby Subject 2 (WFS: -14.95 ±

0.8Nm; WKS: -18.20 ± 0.2Nm) generated slightly higher peak pre-impact internal

rotation moments than Subject 1 (WFS: -13.46 ± 0.5Nm; WKS: -14.07 ± 0.8Nm). A

comparable pattern epitomised the peak external rotation moments of the follow-

through, where these moments were analogous between WFS and WKS, but slightly

175

more pronounced in the deliveries of Subject 2. All peak external rotation moments

assisting humeral deceleration in the WFS and WKS were nevertheless lower than in the

able-bodied FS and KS.

The peak compressive forces and the rates at which they were developed prior to impact

were similar for the WFS and WKS. However, distinguishable inter-subject variation was

again noted. Specifically, the serves of Subject 2 were characterised by pre-impact peak

compressive forces (≈130 ± 7N) and average rates of peak pre-impact compressive

force loading (≈270 ± 20N.s-1) that were appreciably (≥100%) higher than those

generated during Subject 1’s serves. The respective mean compressive and distractive

forces that prevailed during the service follow-throughs of Subject 1 and Subject 2

provide further evidence of some mechanical variation marking the service techniques of

these two wheelchair players.

176

WFS WKS Able-bodied

S1 S2 S1 S2 FS KS Shoulder joint kinetics EVENT /

Phase Mean (SD)


moment (Nm) Cocking

0.56

(0.3)

2.18

(0.4)

0.36

(0.1)

1.21

(0.9)

2.17

(1.5)

1.93

(1.3)

Peak internal rotation

moment (Nm)

Forward-

swing

-13.46

(0.5)

-14.95

(0.8)

-14.07

(0.8)

-18.2

(0.2)

-22.66

(7.6)

-23.48

(5.4)



(N.s-1)

Swing 79.19

(18.2)

265.48

(33.0)

85.13

(10.3)

273.67

(25.2)

333.82

(61.3)

321.18

(83.8)

Mean compressive force

(N)

Forward-

swing

61.74

(13.6)

130.00

(6.5)

48.76

(9.0)

130.84

(7.9)

228.63

(52.4)

210.69

(54.2)

Mean compressive (+)

and distractive forces (-)

(N)

Follow-

through

-20.75

(11.0)

49.93

(9.9)

-36.69

(1.3)

39.10

(7.2)

87.11

(39.6)

75.66

(32.5)


moment (Nm)

Follow-

through

5.98

(2.3)

11.87

(7.5)

6.35

(0.8)

11.01

(2.7)

18.76

(10.0)

14.73

(6.6)

Table 7.4. Mean shoulder joint kinetics that punctuate WFS and WKS performance, and as compared with the able-bodied FS and KS.

In absolute terms, wheelchair players appear to harbour less load at the shoulder joint

during the serve than able-bodied players (Table 7.4). However, when shoulder joint

kinetics are expressed relative to maximum pre-impact absolute racquet velocities:

Peak external rotation moment in the WKS i.e.,

Maximum pre-impact absolute racquet velocity in the WKS X 100 = … % ,

certain wheelchair players may be subject to loads more commensurate to those

generated during the able-bodied serve. This is particularly true for shoulder joint

internal-external rotation moments, where the able-bodied FS and KS as well as the WFS

and WKS of Subject 2 were shown to be characterised by similar relative internal

rotation moment loads (≈40-60%). The reduced kinetics of Subject 1’s serves as

compared with those of Subject 2 nevertheless suggests that other wheelchair players

177

likely encounter lower absolute and relative shoulder joint loading conditions than able-

bodied players.

WFS WKS Able-bodied

S1 S2 S1 S2 FS KS Shoulder joint kinetics EVENT /

Phase % of maximum pre-impact absolute racquet velocity


moment (Nm) Cocking 1.8 6.6 1.2 3.7 5.0 4.8

Peak internal rotation

moment (Nm)

Forward-

swing 43.9 45.0 45.7 55.9 52.4 58.3



(N.s-1)

Swing 258.1 799.6 276.8 840.5 772.4 797.4

Mean compressive force

(N)

Forward-

swing 201.2 391.6 158.5 401.8 529.0 523.1

Mean compressive (+)

and distractive forces (-)

(N)

Follow-

through -67.6 150.4 -119.3 120.1 201.6 187.8


moment (Nm)

Follow-

through 19.5 35.8 20.6 33.8 43.4 36.6

Table 7.5. Mean shoulder joint kinetics of the wheelchair and able-bodied serves expressed relative to maximum pre-impact absolute racquet velocity.

7.4 DISCUSSION

Current technical instruction of the wheelchair tennis serve is largely intuitive, guided to

some extent by the substantiated biomechanical information describing the able-bodied

serve. The link between shoulder pain and serve performance among wheelchair players

has similar origins. Delineation of the shoulder joint kinetics that contribute to the

development of racquet velocity in the WFS and WKS, and that are associated with

shoulder joint injury, is thus important.

178

7.4.1 Differential 3D Racquet Velocity in the WFS and WKS, and in

Comparison with the Velocity Developed During the Able-Bodied Serve


1. WFSs develop higher peak pre-impact horizontal racquet velocities, while WKSs

generate higher maximum pre-impact lateral racquet velocities;


absolute and horizontal pre-impact racquet velocities as compared to when

serving with a leg drive (i.e. the able-bodied serve).

Consistent with expectations, the WFSs were characterised by higher peak pre-impact

horizontal racquet velocities (≈29m.s-1) than the WKSs (≈26m.s-1). As further

hypothesised, higher peak lateral velocities were developed during the forwardswings of

the WKSs (≈-9m.s-1) as compared with the WFSs (≈-4m.s-1). These results agree with

the dichotomous pre-impact racquet velocity profiles previously reported to characterise

the performance of different able-bodied serves (see Chapter 5; Chow et al., 2003a).

Moreover, they also indicate that similar service strategies epitomise high-performance

able-bodied and wheelchair tennis play. That is, where a premium is placed on the

development of higher horizontal ball velocities in the WFS and FS, players’ second

deliveries are generally geared toward accuracy and spin such that more oblique pre-

impact racquet trajectories are necessary.

Interestingly, it appears that wheelchair players generate comparable peak pre-impact

absolute racquet velocities in accommodating these different tactical objectives (WFS:

32.04 ± 0.7 m.s-1; WKS: 31.41 ± 0.6 m.s-1). This finding contrasts with the higher

absolute racquet velocities generated by high-performance able-bodied players in the FS

(43.22 ± 3.1 m.s-1) as compared with the KS (40.28 ± 2.9 m.s-1), but concurs with the

results of an earlier investigation by Chow et al. (2003a). In the absence of an obvious

explanation, it may be that this distinction relates to the typical right-handed able-bodied

player’s preference for hitting KSs to the ‘second’ rather than ‘first’ service box. This

preference is anecdotally noted to produce more effective KSs to the ‘second’ side, and

may in turn, correspond to similar peak absolute racquet velocities punctuating the

forwardswings of high-performance able-bodied players’ FSs and KSs, when hit to this

179

side. However, it could be argued that an analogous preference pervades the elite

wheelchair game.

With both wheelchair players lacking the dynamic leg actions of able-bodied players and

possessing variable but restricted trunk motion, the comparatively lower peak pre-impact

absolute and horizontal racquet velocities of the WFS and WKS were anticipated. In fact,

as inferred by Brody (1987), due to wheelchair players’ low hitting heights, its largely

impractical (i.e. in terms of serve percentage), if not impossible, for them to generate

the same horizontal racquet velocities as their able-bodied counterparts (i.e. 33 m.s-1;

Chow et al., 2003a) and still have their serves land successfully (and consistently) in the

court. Nonetheless, the ≈33% reduction in the maximum pre-impact absolute and

horizontal racquet velocities developed in the wheelchair serve compares favorably with

the 30-50% decreases observed in throwing velocity when throwers experience

constrained hip and trunk motion (Toyoshima et al., 1974; Alexander, 1991).

Noteworthy is also that contrastive pre-impact racquet velocities were developed by the

individual wheelchair players. That is, Subject 2 generated higher 3D racquet velocities,

especially laterally, in both the WFS and WKS. Certainly as both players suffered from

injuries of similar vertebral level but different severity, some variation in racquet

kinematics could be expected. To elaborate, where Subject 1 sustained a complete break

at the T12 level, Subject 2 suffered an incomplete spinal cord injury – albeit at a

marginally higher thoracic level – such that he exhibited greater control over his lower

limbs, and could in fact walk. Subject 1, on the other hand, possessed no lower limb

function. Although strapped to the chair, Subject 2 suggested that his comparatively

superior leg use enabled him to gain some ‘push’ against the chair to ‘drive upward’

when serving, and more importantly provided a more stable platform for subsequent

high-speed segment coordination (Weekes, personal communication, 2006).

Furthermore, the incomplete nature of Subject 2’s injury would typically imply that this

subject possessed superior trunk strength as compared with Subject 1. Although

quantification of as much was beyond the scope of this study, it is probable that these

neuro-physiological factors account for some of the variation observed in the racquet

velocities developed by these elite wheelchair players.

180


1. Wheelchair players develop higher peak pre-impact horizontal racquet velocities

in their WFSs, but greater maximum pre-impact lateral racquet velocities in their

WKSs.

2. As compared to when serving with a leg drive (i.e. the able-bodied serve),

serving with no leg drive (i.e. from a wheelchair) produces lower peak absolute

and horizontal pre-impact racquet velocities.

7.4.2 Variation Between the Upper-Extremity Joint Kinematics of the WFS and

WKS, and in contrast to the Able-Bodied Serves


3. The magnitude of MER of the upper arm is independent of wheelchair serve type

but lower when serving with no leg drive (i.e. the wheelchair serve) as compared

with serving with leg drive or with minimal leg drive (i.e. the able-bodied serve).


trunk rotation characterises the forwardswing of wheelchair serves.

As illustrated in Chapters 5 and 6, able-bodied players rotate their upper arms ≈115º to

a position of MER when serving. Several authors have previously linked the magnitude of

humeral MER to the forcefulness of the able-bodied player’s leg drive (Elliott et al., 1986;

Bartlett et al., 1994). While the results of Chapter 6 suggest this assertion to be overly

simplistic, the hypothesis that wheelchair players would achieve less externally rotated

positions of the upper arm than able-bodied players was supported. That is, without the

utility of a leg drive, the upper arms of the two wheelchair players reached ≈95º of MER

in both the WFS and WKS. Significantly, as one of the limitations of shoulder joint

modelling is that this computed external rotation angle is actually a combination of trunk

hyperextension, scapulothoracic motion and true GH rotation, the disparity between the

upper arm MER achieved in the wheelchair and able-bodied serves may also reflect some

variation in trunk hyperextension and scapulothoracic motion; the extent to which is

currently unquantifiable using external markers.

181

The absolute peak angular velocities of humeral external rotation (expressed in the

thorax) during the forwardswing of the wheelchair serves were also lower (≈40-80%)

than in the able-bodied FS and KS #. Intuitively, a link can be made between the able-

bodied serves’ comparatively larger magnitudes of upper arm MER and their augmented

peak external rotation (expressed in the thorax) angular velocities. That is, greater

external rotation of the upper arm in cocking could correspond to able-bodied players

benefitting from increasingly vigorous eccentric stretch, and thus recoil, of associated

musculature to thereby assist internal rotation of the upper arm during the forwardswing

to impact.

At MER, marginally higher upper arm plane of elevation angles were observed in the

serves of Subject 2 (WFS: -151º ± 5º; WKS: -165º ± 1º) as compared with Subject 1

(WFS: -142º ± 5º; WKS: -145º ± 2º). Importantly, both subjects recorded mean plane

of elevation angles <180º so that their upper arms remained in front of their shoulder

alignments. Similar humeral displacements characterised the high-performance able-

bodied FS and KS (see Chapter 5) as well as the serves of professional ATP players

(Fleisig et al., 2003). Consequently, the threat of hyperangulation (i.e. extension of the

abducted, externally rotated arm beyond the plane of the scapula), which can contribute

to secondary impingement problems, appears similarly negligible among wheelchair

players executing WFSs and/or WKSs.

However, at impact, wheelchair players assume upper arm-thorax elevation angles that

are ≈15º higher than those observed during the high-performance and professional

able-bodied tennis serves (Chapter 5; Fleisig et al., 2003). Although previous research

could be interpreted to implicate these more obtuse upper arm-thorax elevation angles

in injurious shoulder joint kinetics, it’s likely that the angle which permits maximum

wheelchair serve speed with minimal shoulder joint loading is different to the 100º ± 10º

reported by Matsuo et al. (1999) in baseball pitching. To this end, more elevated upper

arms may be related to dichotomous 3D trunk rotation in the wheelchair tennis serve, or

a by-product of wheelchair players’ reduced hitting heights.

# Retrospective analysis and discussion of the magnitude of shoulder joint internal-external rotation – as related to the

182

Clear variation pervaded the 3D rotation of shoulder alignment, and by extension the

trunk, in the WFS and WKS of the two wheelchair players. During the forwardswing to

impact, Subject 2 forward flexed (≈25º) and rotated (≈50º) his shoulder alignment

more than Subject 1 (F ≈ 11º; R ≈ 8º) irrespective of serve type. Further, where both

players left laterally flexed their trunks by ≈17º during this same phase, Subject 2

impacted the ball in more left laterally flexed positions (≈25º) than Subject 1 (≈42º). At

impact, the shoulder alignments of Subject 2 were also flexed further forward but less

rotated than those of Subject 1. As aforementioned, the contrasting injury pathologies of

the two players obviously influence the extent to which each player is able to actively

engage the musculature of his trunk. Indeed by Subject 2’s own admission, his greater

lower limb and trunk mobility and strength allows him to more vigorously forward flex

and rotate his trunk, without compromising his serve accuracy and/or chair balance.

Simply put, the two wheelchair players explained Subject 2’s service motion to more

closely resemble that of an able-bodied player. The more pronounced shoulder

alignment motion of Subject 2’s WKS as compared with his WFS appears consistent with

the heightened abdominal muscle activity that Chow et al. (2003a) found to feature in

topspin serves as compared with flat or slice serves, and would seem to lend additional

weight to this assertion.

Some trunk-related coordinative difference was also evident between the wheelchair and

able-bodied serves. While the nature of shoulder alignment rotation during the

forwardswing of the WFS and WKS was distinctly individual, indications are that certain

wheelchair players – likely those with incomplete injuries – rotate about the long-axes of

their trunks more than able-bodied players when serving. This finding offers some

support to the hypothesised augmentation of transverse plane trunk rotation in the

wheelchair serve. On the contrary, able-bodied players appear capable of involving

larger amounts of forward trunk flexion during their serves than wheelchair players.

Bahamonde (2000) elucidated the important contribution of angular momentum

developed through truncal flexion to serve speed, so this variance in pre-impact shoulder

alignment forward flexion could partly explain the observed differences in the racquet

velocities generated by wheelchair and able-bodied players. As all players went through

UWA marker set – is undertaken in Chapter 5.4.2 and Appendix A.

183

similar ranges of pre-impact left lateral flexion and impacted at the ball at comparably

oblique alignments, the above-mentioned more obtuse upper arm-thorax elevation

angles of wheelchair players at impact are likely related to more elevated positions of the

upper arm rather than the differential alignment of the thorax.


3. Maximum external rotation of the upper arm occurs independent of wheelchair

serve type but is less pronounced in the wheelchair serve as compared with the

able-bodied serve.

4. The magnitude of transverse plane trunk rotation during the forwardswing of

wheelchair serve is distinctly individual and can be more or less pronounced than

in the able-bodied serve.

7.4.3 Effect of Serve Type on Shoulder Joint Loading in Wheelchair Players:

Implications for Injury and Performance


5. Although independent of wheelchair serve type, higher peak upper arm external

rotation moments are generated during the cocking phase of the WFS and WKS

than in the able-bodied FS and KS.


forwardswing of the WFS as compared with the WKS.

7. Relative to absolute racquet velocity, higher peak shoulder joint internal rotation

moments, mean compressive forces and average rates of compressive force

loading are produced in the forwardswings of the wheelchair serves as compared

to the able-bodied serves.


mean compressive forces are generated independent of wheelchair serve type.



players who employ no leg drive (i.e. the wheelchair serves) as compared with

some leg drive (i.e. able-bodied serves).

184

7.4.3.1 Cocking

As discussed in Chapter 6.4.4.1, previous research has inferred a negative association

between leg drive and the role of a player’s external rotator musculature in achieving

MER of the upper arm during the serve (Elliott et al., 2003). Data from the current study

shows that wheelchair players, whom possess no leg drive, generate negligible peak

external rotation moments during the cocking phase of both the WFS (0.5 – 2.2Nm) and

WKS (0.4 – 1.2Nm). Similarly insignificant peak moments were reported to assist

external rotation of the upper arm in the able-bodied FS and KS (≈ 3 ± 1Nm). Scant

support was therefore offered to the above contention and the hypothesised increase in

the peak upper arm external rotation moment in the wheelchair serve as compared with

the able-bodied serve. It would then appear that even wheelchair players produce MER

of the upper arm as a consequence of the inertial lag of the forearm, hand and racquet.


Contrary to expectations, peak pre-impact internal rotation moments appear to be

developed independent of wheelchair serve type. Indeed, if anything, there was some

suggestion of larger peak internal rotation moments driving the forward progression of

the arm and racquet in the WKS. While definitive corroboration would need to be sought

through analyses of larger sample sizes, it is possible that internal rotation of the upper

arm plays a more pronounced role in the development of racquet speed in the WKS than

it does in the WFS.

Several researchers have demonstrated the ill-effects of abbreviated or restricted

segment coordination on throwing performance (Toyoshima et al., 1974; Whiting et al.,

1985; Alexander, 1991; McMaster et al., 1991). For example, the inability to rotate hips

and/or generate GRFs has been shown to negatively affect throwing velocity, while

McMaster et al. (1991) associated the lack of a base of support in the water polo throw

to heightened shoulder joint forces. In turn, it was assumed that as wheelchair players

gain virtually no lower limb drive and exhibit varied trunk function, a relative increase in

shoulder joint kinetics would be observed. In other words, the shoulder would need to

harbour proportionately higher joint loads in the wheelchair serve as compared with the

185

able-bodied serve. More specifically, larger relative internal rotation moments would help

offset any reduced contribution of the trunk to velocity generation, while higher relative

mean compressive forces and rates of maximum compressive force loading would help

counter any intensified tendency for humeral distraction that could occur subsequent to

the desire to increase impact height.

Surprisingly however, there was little evidence of any compensatory kinetic response.

When expressed relative to maximum pre-impact absolute racquet velocity, similar peak

internal rotation moments (≈55%) prevailed regardless of serve type or disability, while

mean pre-impact compressive forces were actually lower in the wheelchair serves (S1:

160-200%, S2: 390-400%; able-bodied FS and KS: ≈525%). As compared to the able-

bodied serves (≈780%), the relative average rates of pre-impact maximum compressive

force loading were similar for the WFS and WKS of Subject 2 (≈820%) but lower for the

WFS and WKS of Subject 1 (≈265%). These observed differences in the pre-impact

compressive force profile of the two wheelchair players may be related to slight variation

in their respective impact heights. That is, where Subject 2 impacted all serves at 1.8 x

his sitting height, Subject 1 made racquet-ball contact in less extended positions (1.71 x

his sitting height). In application, these contrastive sitting-impact heights suggest that

Subject 2 possessed a more ‘up and out’ service action (Elliott et al., 1986), which may

have necessitated the production of higher compressive forces to maintain the humeral

head centred in the gleniod.

To summate, as compared to the high-performance able-bodied serve, indications are

that the elite wheelchair serve subjects its exponents to lower absolute and similar

relative pre-impact shoulder joint loading conditions. Collectively, these results lend

weight to assertion in Chapter 6 that racquet velocity may be load-dependent.


As expected, no distinctive loading profile characterised the follow-through of either

wheelchair serve. Deceleration of the internally rotating upper arm was facilitated by

similar peak WFS (6-11Nm) and WKS (6-11Nm) post-impact external rotation moments.

186

Comparable mean compressive forces were also generated following ball impact in both

subjects’ WFSs and WKSs.

Of note however, was that some variation marked the mean forces applied along the

humerus throughout the follow-throughs of both players’ serves. That is, where Subject

2 generated mean compressive forces (WFS: 50 ± 9N; WKS: 39 ± 7N) to help resist

post-impact humeral distraction, Subject 1 applied mean distraction forces (WFS: -21 ±

11N; WKS: -36 ± 1N) to maintain the upper arm in equilibrium. This dichotomous kinetic

patterning relates to stylistic differences in the players’ follow-throughs. As is familiar to

the able-bodied serve, Subject 2 followed through ‘across his body’. Subject 1, on the

other hand, finished ‘away from the chair’. Both types of follow-throughs are commonly

used by elite wheelchair players, and ITF Wheelchair Development Officer, Mark Bullock

(2006, personal communication), suggests that while players with more complete spinal

cord injuries may find it difficult to follow-through ‘across their bodies’, certain coaches

prefer players to finish ‘away’ irrespective of disability. The fact that Subject 1 suffered

from a more complete injury than Subject 2, and indicated that he had difficulty in

maintaining the necessary balance to comfortably follow-through ‘across his body’

supports Bullock’s affirmation.

Indeed, by following through ‘across his body’, Subject 2 continued to rotate his trunk

(about all 3 axes) while also allowing his upper arm to move down (i.e. decreasing upper

arm-thorax elevation) and toward the midline of his thorax (i.e. decreasing shoulder joint

plane of elevation). So, in paralleling the ‘finish’ of the able-bodied serve and other

overhand sports skills, Subject 2 produced mean compressive forces like those previously

reported to assist upper arm deceleration in the serve and baseball pitch. In contrast,

the ‘away’ finish saw Subject 1 undergo limited 3D trunk rotation and maintain more

obtuse shoulder joint elevation and plane of elevation angles such that mean distraction

forces were needed to support the mass of the humerus. As injury to the rotator cuff is

often associated with its role in generating the compressive forces needed to resist

humeral distraction during the deceleration phase of overhand sports skills (Jobe et al.,

1984; Fleisig et al., 1996), interpretation of these data would appear to suggest that

Subject 1’s post-impact kinetics are less likely to elicit this type of cuff injury. However,

187

upon closer retrorespective inspection, Subject 1 was revealed to apply higher absolute

shoulder joint posterior forces (WFS: ≈60N; WKS: ≈70N) to help resist superior

translation of the humeral head than Subject 2 (WFS: ≈35N; WKS: ≈55N). Collectively,

these results point to both follow-throughs subjecting wheelchair players’ shoulders to

high but different loads that may be equally injurious.

Finally, similar to that which was hypothesised pre-impact, expectations were that the

follow-throughs of the wheelchair serves would be characterised by higher relative peak

shoulder joint external rotation moments and mean compressive forces than the able-

bodied deliveries. However, as illustrated in Table 7.4, even in controlling for maximum

pre-impact absolute racquet velocities, relative post-impact peak external rotation

moments (190-100%) and mean compressive forces (≈40%) in the able-bodied serves

remained appreciably higher than the corresponding measures of load in the wheelchair

serves (70-120%; ≈30%). So, despite lacking the same lower body and trunk rotation,

wheelchair players seem no more likely to forcibly contract their shoulder joint

musculature or develop high shoulder joint loading conditions to help decelerate the

upper arm and racquet, than able-bodied players.


5. Peak upper arm external rotation moments are generated independent of

wheelchair serve type and player disability during the cocking phase of the serve.

6. Similar peak shoulder joint internal rotation moments were developed during the

forwardswing of both the WFS and WKS.

7. Relative to absolute racquet velocity, comparable peak shoulder joint internal

rotation moments, mean compressive forces and average rates of compressive

force loading were produced in the forwardswings of the wheelchair and able-

bodied serves.


mean compressive forces were generated independent of wheelchair serve type.

9. Relative to absolute racquet velocity, post-impact peak shoulder joint external

rotation moments and mean compressive forces were no higher in the wheelchair

serve than the able-bodied serve.

188

7.5 CONCLUSION

Wheelchair players generate similar pre-impact absolute racquet velocities in the WFS

and WKS. However, where higher pre-impact horizontal racquet velocities are produced

in the WFS, higher lateral velocities are developed during the forwardswings of the WKS.

The shoulder joint kinetics that contribute to these differential velocity profiles appear to

vary between individual players (and level and severity of spinal cord injury) but remain

consistent across wheelchair serve type. In comparison to the high-performance able-

bodied serve, indications are that elite wheelchair players are subject to similar or

reduced relative pre- and post-impact shoulder joint loading conditions when hitting both

types of wheelchair serve. Indications are thus that the generation shoulder joint load

positively relates to the development of racquet velocity. Furthermore, subsequent

suggestions that higher shoulder joint loads punctuate wheelchair as compared with

able-bodied serve performance such that this playing population is predisposed to

elevated shoulder joint injury risk are unfounded.

189

CHAPTER 8: SUMMARY AND CONCLUSIONS

In this chapter, a summary of results from this program of research is provided. This is

followed by a synthesis of its practical applications, and finally, recommendations for

future study.

8.1 SUMMARY

1. Different data smoothing protocols were compared to identify the most appropriate

technique for treating 3D high performance FS and KS data. Concurrent

quantification of the movement repeatability of these serves was pursued.

a. Two male, full-time professional tennis players (mean age: 20 years; mass: 85

kilograms) hit, with maximal effort, five FS and five KS to a 1x1 metre target

area bordering the ‘T’ of the first service box. First serves were to be hit “flat”,

while KS were to be hit with maximum “kick”. For serves to be considered

successful, the ball had to clear the net and bounce in the target area.

b. Accurate kinematic and kinetic representation of the service motion can be

achieved through smoothing sampled displacement data with a quintic spline

using a MSE near 25.

c. Antecedent to smoothing through impact, interpolation of data describing the

serve from one frame pre-impact to five frames post-impact is effective in

minimising error associated with end point conditions.

d. FSs and KSs are highly repeatable skills when performed by high performance

players. Reliable kinematic and kinetic data can be gleaned through three

successful executions of each serve.

2. Substantiation of the shoulder joint kinetics produced during the FS and KS was

undertaken; allowing for the elucidation of hypothesised relationships between

serve type (FS and KS) and shoulder joint loading.

190

a. Twelve high performance male tennis players hit, with maximal effort, three

successful FSs and three successful KSs to a 1x1 metre target area bordering

the ‘T’ of the first service box. First serves were to be hit “flat”, while KSs were

executed with maximum “kick”.

b. Differences in the tactical use of the FS and KS are reflected in the higher peak

horizontal racquet velocities of the FS, and the higher peak lateral racquet

velocities of the KS.

c. Shoulder joint loading as represented by: maximum pre-impact anterior force

and average rate of maximum pre-impact anterior force loading in cocking,

average rate of maximum compressive force loading during the swing phase,

mean compressive force and peak internal-external rotation moments during

the forwardswing and follow-through, and peak abduction moment during the

follow-through, is similar between serves. The GH joint would therefore appear

subject to instantaneous and accumulative stress independent of serve type.

d. Although loading was homogenous across serve type, different shoulder joint

kinetics explained the variance in the absolute racquet velocity generated in the

FS (average rate of anterior force loading in the cocking, and mean pre-impact

compressive force) and KS (average rate of maximum compressive force

loading during the swing phase).

3. The relationship between varied lower limb coordination, as epitomised by

dichotomy in stance and leg drive, and shoulder joint kinetics in the FS was

analysed.

a. Twelve high performance male tennis players hit three successful maximal

effort, flat FU and FB serves to a 1x1 metre target area bordering the ‘T’ of the

first service box. All but two players also hit three successful, flat ARM serves to

the same location.

b. Differences in the stance and leg drive that players use when serving are

encapsulated by variation in the range of lead and rear knee extension as well

as peak angular velocity of rear knee extension.

191

c. In the high performance FS, absolute racquet velocity is generated independent

of stance but significantly and positively affected by the presence of a leg drive.

d. Shoulder joint kinetics are largely developed independent of lower limb

coordinative differences, however, the application and magnitude of

compressive forces needed to resist pre-impact humeral distraction trended

higher in FU serve and lower in the ARM serve.

e. The range of front knee joint extension significantly predicts the average rate

of maximum pre-impact compressive force loading in the FB serve, whereby

greater front knee joint extension corresponds to reduced compressive rates of

loading. Conversely, the range of rear knee joint extension significantly predicts

this same loading variable in the ARM serve, whereby greater rear knee joint

extension corresponds to heightened average rates of maximum pre-impact

compressive force loading. Variables other than the lower limb kinematics

shown to discriminate between serve techniques explain the variance in the

average rate of maximum compressive force loading in the swing phase of the

FU serve.

4. A case study comparison of the shoulder joint kinetics that characterised the elite

WFSs and WKSs was undertaken. Subsequent contrasts with the able-bodied FS

and KS provided an insight into the shoulder joint loads that wheelchair and able-

bodied players harbour during the performance of their serves.

a. Two male top 30-ranked international wheelchair players hit, with maximal

effort, three successful WFSs and three successful WKSs to a 1x1 metre target

area bordering the ‘T’ of the first service box. Wheelchair first serves were to be

hit “flat”, while WKSs were executed with maximum “kick”.

b. Without the benefit of a propulsive leg action, wheelchair players develop lower

peak absolute and horizontal pre-impact racquet velocities than able-bodied

players.

c. As in the able-bodied serve, the divergent tactical use of the WFSs and WKSs is

achieved through higher respective horizontal and lateral racquet velocities.

192

d. Wheelchair players maximally externally rotate their upper arms independent of

serve type, but less so than in the able-bodied FS and KS.

e. The magnitude of trunk rotation during the WFS and WKS appears to be heavily

influenced by individual player’s level and severity of spinal cord injury.

f. Shoulder joint kinetics are mostly developed independent of wheelchair serve

type, but vary between individual players (and level and severity of spinal cord

injury).

g. During the serve, elite wheelchair players appear subject to similar or reduced

relative pre- and post-impact shoulder joint loading conditions than able-bodied

players.

8.2 CONCLUSIONS

The stakes are high on the professional tennis tour. Whether as a ‘tour

journeyman/woman’ or a player preparing for a tilt at their first Grand Slam title, the

same goal remains: to win. Career-defining tournaments are forever around the corner,

yet, equally omnipresent are career-ending injuries.

Injuries to the GH joint, or its associated structures, are among the most prevalent and

debilitating sustained by professional tennis players. Popular literature has associated the

loads generated and absorbed by the tissues of the shoulder during stroke production,

and more particularly the serve, to GH joint injury (Chandler et al., 1992; McCann and

Bigliani, 1994; Kibler, 1995). Indeed, variations in serve technique have been suggested

to load the shoulder joint differently and therefore have some implications for injury

(Elliott et al., 2003).

Decisions to overhaul a player’s service technique are not trivial however, and ill-

informed analyses can halt a player’s ascension up the ranks just as much as a shoulder

injury may. Evaluation of serve technique should therefore be based on at least three

observations of the same, successful serve. In like kind, interventions to improve a

player’s service technique should have some evidence base, and not be solely founded

on anecdotal lore. To this end, the following summative insights into the relationships

193

between different service types and techniques and shoulder joint loading may assist

this process.

Similar shoulder joint kinetics aid the development of pre-impact 3D racquet velocity in

the FS and KS. Despite higher peak horizontal, vertical and absolute racquet velocities

epitomising FS performance and higher lateral velocities characterising the KS, the

homogeneously high loads tolerated at the shoulder point to the repetitive, long-term

performance of either serve as relevant in associated shoulder joint injury pathologies.

Coordinative lower limb variation in the serve, as summarised by players’ ranges of front

and rear knee joint extension and their peak angular velocities of rear knee joint

extension, also affects the development of serve speed. More explicitly, when facilitated

by a leg drive, high-performance players can generate similar absolute pre-impact

racquet velocities from either a FU or FB service stance. Less dynamic use of their legs

however, results in players generating appreciably lower pre-impact absolute racquet

velocities. In spite of the marked variation in lower limb kinematics between FU, FB and

ARM serves, comparable shoulder joint kinetics were inherent to the different service

techniques. Nevertheless there was some suggestion of the application and magnitude

of compressive forces trending higher with a FU stance and a more pronounced leg

drive. Interestingly, with higher absolute racquet velocities developed during the FU and

FB serves, this trend also inferred shoulder joint loading to relate positively to racquet

velocity. Of further note was that these divergent absolute racquet velocities were

generated from comparable shoulder joint kinetics but variable lower limb kinematics;

raising the prospect that other links in the ‘kinetic chain’, such as the trunk, may be

more affected by dichotomous leg actions.

In contrast to the able-bodied serve, similar peak pre-impact absolute racquet velocities

pervade the WFS and WKS. As in the able-bodied game however, divergent serve tactics

necessitate that higher peak pre-impact horizontal and lateral racquet velocities

punctuate the WFS and WKS respectively. The shoulder joint kinetics that contribute to

these differential velocity profiles are consistent across wheelchair serve type, but

specific to individual players, likely varying with level and severity of spinal cord injury.

194

Relative to racquet velocity, both high-performance able-bodied and elite wheelchair

players experience similar pre- and post-impact shoulder joint loads, lending support to

proposed positive relationships between racquet velocity and shoulder joint load, while

also intimating that both playing populations appear subject to analogous shoulder joint

injury risk.

8.3 RECOMMENDATIONS FOR FUTURE RESEARCH

This course of studies has highlighted related areas of potential research interest.

Methodological recommendations can also be born out of the technical note detailed in

Appendix A.

8.3.1 Subject Preparation and Data Analysis

As aforementioned, determination of specific surface marker locations that best

represent 3D humeral motion is key to the validity of future 3D upper-extremity

kinematic and kinetic analyses of the tennis serve. Correspondingly, prospective

evaluations of shoulder joint function during the serve should elaborate a shoulder joint

coordinate system consistent with the Y-X-Y decomposition. This would ensure the

derivation of shoulder joint moments and angular velocities in the same functionally-

relevant coordinate system to thereby provide for the more meaningful and constructive

interpretation of shoulder joint power. Concurrent EMG analysis of selected shoulder

musculature would stand to corroborate any inferences drawn with respect to muscle

activation.

8.3.2 Continued Investigation of the Relationship between Shoulder Joint

Loading and Tennis Player Performance

This is the first program of research to investigate 3D shoulder joint kinetics of the high

performance tennis serve through stereophotogrammetry. Subsequent analyses of the

effects of serve technique and type, as performed by high performance adult male

players, on shoulder joint loading were also pursued. Further research, employing alike

protocols and directed toward the quantification of similar relationships among high

performance female players and/or junior (i.e. 12-16 years old) players would certainly

add to the tennis coaching and sports medicine fraternities’ understanding of how serve

195

mechanics may relate to shoulder joint injury. Indeed, use of the current technology to

examine the mechanics of the game’s other power strokes, such as the forehand and

backhand, could provide similarly beneficial insights.

8.3.3 Quantification of Models of Optimal Serve Performance

Emanating from this thesis is the need to further explore the relationship between lower

limb and trunk motion as well as the ensuing interaction of the trunk and upper

extremity. As aforementioned, with racquet velocity seemingly generated independent of

shoulder joint kinetics, it appears likely that serve type and technique may differentially

load other ‘joints’ in the kinetic chain.

Statistical techniques, like structural equation modelling, should be used to best

determine these hypothesised segmental interactions and further contrast the predictive

value of segment contributions to shoulder joint load and/or racquet velocity in different

serves. The formation of theoretical ‘player height to stance width’ ratios or best

morphological fits for leg drive may also evolve.

196

REFERENCES

Alderink, G. J., & Kluck, D. J. (1986). Isokinetic shoulder strength of high school and college aged pitchers. J Orthop Sports Phys Ther, 7, 163-172. Alexander, R. M. (1991). Optimum timing of muscle activation for simple models of

throwing. J Theor Biol, 150, 349-372. Andrews, J. R., & Angelo, R. L. (1988). Shoulder arthroscopy for the throwing athlete.

Tech Othrop, 3, 75-81. Anglin, C., & Wyss, U. P. (2000). Review of arm motion analyses. Proc Inst Mech Eng

[H], 214(5), 541-555. Ariel, G. B., & Braden, V. K. (1979). Biomechanical analysis of ballistic vs tracking

movement in tennis skills. Paper presented at the A National Symposium on the Racket Sports, Champaign, IL.

Arutyunyan, G. H., Gurfinkel, V. S., & Mirskii, M. L. (1968). Investigation of aiming at a target. Biophysics, 13, 536-538.

ATP. (2003). Overview of injuries and tournament withdrawals on the ATP tour (1994-2002). Unpublished document. Florida: Association of Tennis Professionals.

ATP. (2006). www.atptennis.com/en/players/matchfacts, March 15, 2006. Florida: Association of Tennis Professionals.

Atwater, A. E. (1979). Biomechanics of overarm throwing movements and of throwing injuries. Exerc Sport Sci Rev, 7, 43-85.

Bahamonde, R. E. (1989). Kinetic analysis of the serving arm during the performance of the tennis serve. Paper presented at the XIIth International Society of Biomechanics [abstract #99], Los Angeles.

Bahamonde, R. (1997). Joint power production during flat and slice tennis serves. Paper presented at the 15th International Symposium on Biomechanics in Sports, Texas.

Bahamonde, R. E. (2000). Changes in angular momentum during the tennis serve. J Sports Sci, 18(8), 579-592.

Bahamonde, R. E., & Knudson, D. (2001). Ground reaction forces of two types of stances and tennis serves. Med Sci Sports Exerc, 33(5 Supplement 1), S102.

Bartlett, R. M., Piller, J., & Miller, S. (1994). A three dimensional analysis of the tennis serves of National (British) and County standard male players. In Reilly, T., Hughes, M., & Lees, A. (Eds.), Science and racket sports. London: E&FN Spon.

Bastholt, P. (2000). Professional tennis (ATP tour) and number of medical treatments in relation to type of surface. Medicine and Science in Tennis, 5(2), 9.

Bates, B. T., Dufek, J. S., & Davis, H. P. (1992). The effect of trial size on statistical power. Med Sci Sports Exerc, 24, 1059-1068.

Bayley, J. C., Cochran, T. P., & Sledge, C. B. (1987). The weight-bearing shoulder. The impingement syndrome in paraplegics. J Bone Joint Surg Am, 69, 676-678.

Bellchamber, T. L., & van den Bogert, A. J. (2000). Contributions of proximal and distal moments to axial tibial rotation during walking and running. J Biomech, 33, 1397-1403.

Bernard, P. L., Codine, P., & Minier, J. (2004). Isokinetic shoulder rotator muscles in wheelchair athletes. Spinal Cord, 42(4), 222-229.

Bigliani, L. V., Morrison, S. D., & April, E. W. (1986). The morphology of the acromion and its relationship of rotator cuff tears [abstract]. Othrop Trans, 10, 228.

197

Birkelo, J. R., Padua, D. A., Guskiewicz, K. M., & Karas, S. G. (2003). Prolonged overhead throwing alters scapular kinematics and scapular muscle strength. J Athl Train, 38, S10-S11.

Blevins, F. T., Hayes, W. M., & Warren, R. F. (1996). Rotator cuff injury in contact athletes. Am J Sports Med, 24(3), 263-267.

Blevins, F. T. (1997). Rotator cuff pathology in athletes. Sports Med, 24(3), 205-220. Bootsma, R. J., & van Wieringen, P. C. W. (1990). Timing an attacking forehand drive in

table tennis. J Exp Psych, 16, 21-29. Bosworth, D. M. (1955). The role of the orbicular ligament in tennis elbow. J Bone Joint

Surg Am, 37-A(3), 527-533. Brody, H. (1987). Tennis science for tennis players. Pennsylvania: University of

Pennsylvania Press. Brody, H., Cross R., & Lindsey, C. (2002). The physics and technology of tennis.

California: Racquet Tech Publishing. Buczek, F. L. (2006). Reply to: Calculation of scalar joint power or individual joint power

terms. BIOMCH-L. Bullock, M. (2006). The effect of player disability on wheelchair serve technique.

Personal communication, May 2006. Burnett, A. F., Elliott, B. C., & Marshall, R. N. (1995). The effect of a 12-over spell on

fast bowling technique in cricket. J Sports Sci, 13(4), 329-341. Burnham, R. S., May, L., Nelson, E., Steadward, R., & Reid, D. C. (1993). Shoulder pain

in wheelchair athletes. The role of muscle imbalance. Am J Sports Med, 21(2), 238-242.

Cappozzo, A. (1984). Compressive loads in the lumbar vertebral column during normal level walking. J Orthop Res, 1(3), 292-301.

Cappozzo, A., Catani, F., Leardini, A., Benedetti, M. G., & Croce, U. D. (1996). Position and orientation in space of bones during movement: Experimental artefacts. Clin Biomech (Bristol, Avon), 11(2), 90-100.

Cassell, E., McGrath, A. (1999). Lobbing injury out of tennis: A review of literature. Monash University, Melbourne.

Challis, J. H. (1999). A procedure for the automatic determination of filter cutoff frequency for the processing of biomechanical data. J Appl Biomech, 15, 303-317.

Chandler, T. J., Kibler, W. B., Stracener, E. C., Ziegler, A. K., & Pace, B. (1992). Shoulder strength, power, and endurance in college tennis players. Am J Sports Med, 20(4), 455-458.

Chard, M. D., & Lachman, S. M. (1987). Racquet sports - patterns of injury presenting to a sports injury clinic. Brit J Sport Med, 21(4), 150-153.

Cheng, P. L. (2000). A spherical rotation coordinate system for the description of three-dimensional joint rotations. Ann Biomed Eng, 28(11), 1381-1392.

Chiari, L., Croce, U. D., Leardini, A., & Cappozzo, A. (2005). Human movement analysis using stereophotogrammetry, Part 2: Instrumental errors. Gait Posture, 21, 197-211.

Chow, J. W., Carlton, L. G., Lim, Y. T., Chae, W. S., Shim, J. H., Kuenster, A. F., Kokubun, K. (2003a). Comparing the pre- and post-impact ball and racquet kinematics of elite tennis players' first and second serves: A preliminary study. J Sports Sci, 21(7), 529-537.

198

Chow, J. W., Shim, J. H., & Lim, Y. T. (2003b). Lower trunk muscle activity during the tennis serve. J Sci Med Sport, 6(4), 512-518.

Clarey, C. (2005). As game evolves, so do injuries. International Herald Tribune, Tuesday, February 15, 2005, p. 24.

Clauser, C. E., McConville, J. T., & Young, J. W. (1969). Weight, volume and centre of mass segments of the human body. In AMRL Technical Report 69-10. Dayton, OH: Wright-Patterson Air Force Base.

Codine, P., Bernard, P. L., Pocholle, M., Benaim, C., & Brun, V. (1997). Influence of sports discipline on shoulder rotator cuff balance. Med Sci Sports Exerc, 29(11), 1400-1405.

Cook, E. E., Gray, V. L., Savinar-Nogue, E., et al. (1987). Shoulder antagonistic strength ratios: A comparison between college-level baseball pitchers and non-pitchers. J Orthop Sports Phys Ther, 8, 451-461.

Craig, J. J. (2005). Introduction to Robotics: Mechanics and Control (3rd Ed). New Jersey: Prentice Hall.

Crespo, M., & Miley, D. (1998). Advanced coaching manual. London: ITF Ltd. Curtis, K. A., Drysdale, G. A., Lanza, R. D., Kolber, M., Vitolo, R. S., & West, R. (1999).

Shoulder pain in wheelchair users with tetraplegia and paraplegia. Arch Phys Med Rehabil, 80, 453-457.

Davidson, P. A., Elattrache, N. S., Jobe, C. M. et al. (1995). Rotator cuff and posterior-superior glenoid labrum injury associated with increased glenohumeral motion: A new site of impingement. J Shoulder Elbow Surg, 4(5), 384-390.

de Leva, P. (1996). Adjustments to zatsiorsky-seluyanov's segment inertia parameters. J Biomech, 29(9), 1223-1230.

Della Croce U., C. A., Kerrigan C., & Lucchetti L. (1997). Bone position and orientation errors: Pelvis and lower limb anatomical landmark identification reliability. Gait Posture, 5(2), 156-157.

Della Croce, U., Camomilla, V., Leardini, A., & Cappozzo, A. (2003). Femoral anatomical frame: Assessment of various definitions. Med Eng Phys, 25(5), 425-431.

DiGiovine, N. M. (1992). An electromyographic analysis of the upper extremity in pitching. J Shoulder Elbow Surg, 1, 15-25.

Dillman, C. J., Fleisig, G. S., & Andrews, J. R. (1993). Biomechanics of pitching with emphasis upon shoulder kinematics. J Orthop Sports Phys Ther, 18(2), 402-408.

Doorenbosch, C. A., Harlaar, J., & Veeger, D. H. (2003). The globe system: An unambiguous description of shoulder positions in daily life movements. J Rehabil Res Dev, 40(2), 147-155.

Douglas, P. (1992). The handbooks of tennis. London: Pelhams Books. Ellenbecker, T. (1992). Shoulder internal and external rotation strength and range of

motion of highly skilled junior tennis players. Isok Ex Sci, 2(2), 65-72. Ellenbecker, T. S. (1995). Rehabilitation of shoulder and elbow injuries in tennis players.

Clin Sports Med, 14(1), 87-110. Ellenbecker, T., Tiley, C. (2001). In Roetert, P. & Groppel., J. (Eds.), World-class tennis

technique. Champaign: Human Kinetics. Ellenbecker, T., & Roetert, E. P. (2003). Age specific isokinetic glenohumeral internal and

external rotation strength in elite junior tennis players. J Sci Med Sport, 6(1), 63-70.

Elliott, B. C., & Wood, G. A. (1983). The biomechanics of the foot-up and foot-back tennis service techniques. Aust J Sport Sci, 3(2), 3-6.

199

Elliott, B., Marsh, T., & Blanksby, B. (1986). A three-dimensional cinematographic analysis of the tennis serve. Int J Sport Biomech, 2, 260-271.

Elliott, B. C. (1988). Biomechanics of the serve in tennis. A biomedical perspective. Sports Med, 6(5), 285-294.

Elliott, B., Marsh, T., & Overheu, P. (1989). A biomechanical comparison of the multi-segment and single unit topspin forehand drives in tennis. Int J Sport Biomech, 5, 350-364.

Elliott, B., Marshall, R., & Noffal, G. (1995). Contribution of upper limb segment rotations during the power serve in tennis. J Appl Biomech, 11, 433-442.

Elliott, B., & Christmass, M. (1995). A comparison of the high and low backspin backhand drives in tennis using different grips. J Sports Sci, 13(2), 141-151.

Elliott, B., Baxter, K., & Besier, T. (1999). Internal rotation of the upper-arm segment during a stretch-shorten cycle movement. J Appl Biomech, 15, 381-395.

Elliott, B. (2001). The serve. ITF Coaching & Sport Science Review, 24, 3-5. Elliott, B., & Alderson, J. (2003). Biomechanical performance models. In Elliott, B., Reid,

M., & Crespo, M. (Eds.), ITF Biomechanics of Advanced Tennis (pp. 155-175.). London: ITF Ltd.

Elliott, B., Fleisig, G., Nicholls, R., & Escamilla, R. (2003). Technique effects on upper limb loading in the tennis serve. J Sci Med Sport, 6(1), 76-87.

Endo, K., Ikata, T., Katoh, S., & Takeda, Y. (2001). Radiographic assessment of scapular rotational tilt in chronic shoulder impingement syndrome. J Orthop Sci, 6, 3-10.

Escamilla, R. F., Fleisig, G. S., Zheng, N., Barrentine, S. W., & Andrews, J. R. (2001). Kinematic comparisons of 1996 olympic baseball pitchers. J Sports Sci, 19(9), 665-676.

Evans, W. J., Meredith, C. N., Cannon, J. G., Dinarello, C. A., Frontera, W. R., Hughes, V. A., Jones, B. H., & Knuttgen, H. G. (1985). Metabolic changes following eccentric exercise in trained and untrained men. J Appl Physiol, 61, 1864-1868.

Ferber, R., McClay Davis, I., & Williams III, D. S. (2003). Gender differences in lower extremity mechanics during running. Clin Biomech, 18(4), 350-357.

Ferrara, M. S., & Davis, R. W. (1990). Injuries to elite wheelchair athletes. Paraplegia, 28(5), 335-341.

Fitts, P. M. (1954). The information capacity of the human motor system in controlling the amplitude of movement. J Exp Psych, 47, 381-391.

Fleisig, G. S., Andrews, J. R., Dillman, C. J., & Escamilla, R. F. (1995). Kinetics of baseball pitching with implications about injury mechanisms. Am J Sports Med, 23(2), 233-239.

Fleisig, G. S., Barrentine, S. W., Escamilla, R. F., & Andrews, J. R. (1996). Biomechanics of overhand throwing with implications for injuries. Sports Med, 21(6), 421-437.

Fleisig, G., Escamilla, R. F., Andrews, J. R., Matsuo, T., Satterwhite, Y., & Barrentine, S. W. (1996b). Kinematic and kinetic comparison between baseball pitching and football passing. J Appl Biomech, 12, 207-224.

Fleisig, G. S., Barrentine, S. W., Zheng, N., Escamilla, R. F., & Andrews, J. R. (1999). Kinematic and kinetic comparison of baseball pitching among various levels of development. J Biomech, 32(12), 1371-1375.

Fleisig, G., Nicholls, R., Elliott, B., & Escamilla, R. (2003). Kinematics used by world class tennis players to produce high-velocity serves. Sports Biomech, 2(1), 51-64.

Friden, J., & Lieber, R. L. (1992). Structural and mechanical basis of exercise-induced muscle injury. Med Sci Sports Exerc, 24(5), 521-530.

200

Garrett, W. E., Jr. (1990). Muscle strain injuries: Clinical and basic aspects. Med Sci Sports Exerc, 22(4), 436-443.

Giakis, G., & Baltzopoulos, V. (1997). Optimal digital filtering requires a different cut-off frequency strategy for the determination of the higher derivatives. J Biomech, 30, 851-855.

Giakis, G., Baltzopoulos, V., & Bartlett, R. M. (1998). Improved extrapolation techniques in recursive digital filtering: A comparison of least squares and prediction. J Biomech, 31, 87-91.

Girard, O., Micallef, J. P., & Millet, G. P. (2005). Lower-limb activity during the power serve in tennis: Effects of performance level. Med Sci Sports Exerc, 37(6), 1021-1029.

Glousman, R., Jobe, F., Tibone, J., Moynes, D., Antonelli, D., & Perry, J. (1988). Dynamic electromyographic analysis of the throwing shoulder with glenohumeral instability. J Bone Joint Surg Am, 70(2), 220-226.

Goetzke, R. (2005). Implications for injury due to technical differences in first and second serve technique. Personal communication, April 2005.

Goosey-Tolfrey, V. L., & Moss, A. D. (2005). Wheelchair velocity of tennis players during propulsion with and without the use of racquets. Adapt Phys Act Q, 22, 291-301.

Gordon, B. J., & Dapena, J. (2006). Contributions of joint rotations to racquet speed in the tennis serve. J Sport Sci, 24(1), 31-49.

Grood, E. S., & Suntay, W. J. (1983). A joint coordinate system for the clinical description of three-dimensional motions: Application to the knee. J Biomech Eng, 105(2), 136-144.

Groppel, J. L. (1992). High tech tennis. Champaign: Leisure Press. Groppel, J. L., & Nirschl, R. P. (1986). A mechanical and electromyographical analysis of

the effects of various joint counterforce braces on the tennis player. Am J Sports Med, 14(3), 195-200.

Harryman, D. T., Sidles, J. A., Clark, J. M., et al. (1990). Translation of the humeral head on the glenoid with passive glenohumeral motion. J Bone Joint Surg Am, 72(A), 1334-1343.

Harvey, L. A., & Crosbie, J. (2000). Biomechanical analysis of a weight-relief maneuver in C5 and C6 quadriplegia. Arch Phys Med Rehabil, 81, 500-505.

Herring, S. A., & Nilson, K. L. (1987). Introduction to overuse injuries. Clin Sports Med, 6(2), 225-239.

Hunter, J. P., Marshall, R. N., & McNair, P. (2004). Reliability of biomechanical variables of sprint running. Med Sci Sports Exerc, 36(5), 850-861.

Hutchinson, M. R., Laprade, R. F., Burnett, Q. M., Moss, R., & Terpstra, J. (1995). Injury surveillance at the usta boys' tennis championships: A 6-yr study. Med Sci Sports Exerc, 27(6), 826-830.

ITF. (2003). Professional tennis events calendar. London: ITF Ltd. Ivey, F. M., Calhoun, J. H., Rusche, K., & Bierschenk, J. (1984). Isokinetic testing of

shoulder strength: Normal values. Arch Phys Med Rehabil, 66(6), 384-386. Jobe, F. W., Moynes, D. R., Tibone, J. W., & Perry, J. (1984). An emg analysis of the

shoulder in pitching: A secondary report. Am J Sports Med, 12, 218-220. Jobe, F. W., & Bradley, J. P. (1988). Rotator cuff injuries in baseball. Prevention and

rehabilitation. Sports Med, 6(6), 378-387.

201

Jobe, F. W., & Kvitne, R. S. (1989). Shoulder pain in the overhand or throwing athlete. The relationship of anterior instability and rotator cuff impingement. Orthop Rev, 18, 963-975.

Kadaba, M. P., Ramakrishnan, H. K., Wootten, M. E., Gainey, J., Gorton, G., & Cochran, G. V. (1989). Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait. J Orthop Res, 7(6), 849-860.

Kellis, E., & Baltzopoulos, V. (1995). Isokinetic eccentric exercise. Sports Med, 19(3), 202-222.

Kennedy, K. (1993). Rehabilitation of the unstable shoulder. Oper Tech Sports Med, 1, 311-324.

Kibler, W. B., McQueen, C., & Uhl, T. (1988). Fitness evaluations and fitness findings in competitive junior tennis players. Clin Sports Med, 7(2), 403-416.

Kibler, W. B. (1991). The role of the scapula in the overhead throwing motion. Contemp Orthop, 22, 525-532.

Kibler, W. B., Chandler, T. J., & Stracener, E. S. (1992). Musculoskeletal adaptations and injuries due to overtraining. Exerc Sport Sci Rev, 20, 99-126.

Kibler, W. B. (1994). Clinical biomechanics of the elbow in tennis: Implications for evaluation and diagnosis. Med Sci Sports Exerc, 26(10), 1203-1206.

Kibler, W. B. (1995). Biomechanical analysis of the shoulder during tennis activities. Clin Sports Med, 14(1), 79-85.

Kibler, W. B., Chandler, T. J., Livingston, B. P., & Roetert, E. P. (1996). Shoulder range of motion in elite tennis players. Effect of age and years of tournament play. Am J Sports Med, 24(3), 279-285.

Kibler, W. B. (1998). The role of the scapula in athletic shoulder function. Am J Sports Med, 26(2), 325-337.

Kibler, W. B., & Safran, M. R. (2000). Musculoskeletal injuries in the young tennis player. Clin Sports Med, 19(4), 781-792.

Knudson, D. (1990). Intrasubject variability of upper extremity angular kinematics in the tennis forehand drive. Int J Sport Biomech, 6, 415-421.

Knudson, D., & Bahamonde, R. (2001). Effect of endpoint conditions on position and velocity near impact in tennis. J Sports Sci, 19(11), 839-844.

Knudson, D. V., & Morrison, C. S. (2002). Qualitative analysis of human movement (2nd ed). Champaign: Human Kinetics.

Knudson, D. (2005). Effect of tennis ball mass and smoothing on peak racket velocity. In Q. Wang (Ed.) Proceedings of XXIII International Symposium on Biomechanics in Sports (pp. 757-760). Beijing: China Institute of Sport Science.

Knudson, D. V., & Blackwell, J. R. (2005). Variability of impact kinematics and margin for error in the tennis forehand of advanced players. Sports Engineering, 8, 75-80.

Kraemer, W. J., Triplett, T., Fry, A. C., Koziris, L. P., Bauer, J. E., Lynch, J. M., McConnell, T., Newton, R. U., Gordon, S. E., Nelson, R. C., & Knuttgen, H. G. (1995). An in-depth sports medicine profile of women college tennis players. J Sport Rehabilitation, 4, 79-98.

Kuhn, J. E., Plancher, K. D., & Hawkins, R. J. (1995). Scapular winging. J Am Acad Orthop Surg, 3, 319-325.

Latash, M. L., & Turvey, M. T. (1996). Dexterity and its development. Mahwah, New Jersey: Lawrence Erlbaum Associates Publishers.

202

Leadbetter, W. B. (1994). Soft tissue athletic injury. In F. H. Fu, Stone, D. A. (Eds.), Sports injuries: Mechanisms, prevention & treatment. (pp. 733-780.). Baltimore: Williams & Wilkins.

Leardini, A., Chiari, L., Della Croce, U., & Cappozzo, A. (2005). Human movement analysis using stereophotogrammetry. Part 3. Soft tissue artifact assessment and compensation. Gait Posture, 21(2), 212-225.

Lehman, R. C. (1988a). Shoulder pain in the competitive tennis player. Clin Sports Med, 7(2), 309-327.

Lehman, R. C. (1988b). Surface and equipment variables in tennis injuries. Clin Sports Med, 7(2), 229-232.

Lieber, R. L., & Friden, J. (1988). Selective damage of fast glycolytic muscle fibres with eccentric contraction of the rabbit tibialis anterior. Acta Physiol Scand, 133(4), 587-588.

Lieber, R. L., Woodburn, T. M., & Friden, J. (1991). Muscle damage induced by eccentric contractions of 25% strain. J Appl Physiol, 70(6), 2498-2507.

Lieber, R. L. (1992). Skeletal muscle structure and function. Baltimore: Williams & Wilkins.

Lieber, R. L., & Friden, J. (1993). Muscle damage is not a function of muscle force but active muscle strain. J Appl Physiol, 74(2), 520-526.

Lieber, R. L., & Friden, J. (2000). Functional and clinical significance of skeletal muscle architecture. Muscle Nerve, 23(11), 1647-1666.

Llana, S., Brizuela, G., Alcantara, E., Dura, J. V., & Garcia, A. C. (1998). Epidemiological aspects of tennis and their relation with footwear and tennis courts. Paper presented at the ISBS.

Lloyd, D. G., Alderson, J., & Elliott, B. C. (2000). An upper limb kinematic model for the examination of cricket bowling: A case study of Mutiah Muralitharan. J Sports Sci, 18(12), 975-982.

Ludewig, P. M., & Cook, T. M. (2002). Translations of the humerus in persons with shoulder impingement syndromes. J Orthop Sports Phys Ther, 32, 248-259.

Mair, S. D., Seaber, A. V., Glisson, R. R., & Garret W. E. (1996). The role of fatigue in susceptibility of acute muscle strain injury. Am J Sports Med, 24(2), 137-143.

Manal, K., McClay, I., Stanhope, S., Richards, J., & Galinat, B. (2000). Comparison of surface mounted markers and attachment methods in estimating tibial rotations during walking: An in vivo study. Gait Posture, 11(1), 38-45.

Martin, K. (2004). Legal and ethical constraints to the dissemination of injury data from professional women's tennis. Personal communication, February 2004.

Matsuo, T., Matsumoto, T., Takada, Y., & Mochizuki, Y. (1999). Influence of different shoulder abduction angles during baseball pitching on throwing performance and joint kinetics. Paper presented at the XVII International Symposium on Biomechanics in Sports, Perth.

Matsuo, T., Matsumoto, T., Mochizuki, Y., Takada, Y., & Saito, K. (2002). Optimal shoulder abduction angles during baseball pitching from maximal wrist velocity and minimal kinetics viewpoints. J Appl Biomech, 18, 306-320.

McCann, P. D., & Bigliani, L. U. (1994). Shoulder pain in tennis players. Sports Med, 17(1), 53-64.

McLeod, W. D., & Andrews, J. R. (1986). Mechanisms of shoulder injuries. Phys Ther, 66(12), 1901-1904.

203

McMaster, W. C., Long, S. C., & Caiozzo, V. (1991). Isokinetic torque imbalances in the rotator cuff of the elite water polo player. Am J Sports Med, 19(1), 72-75.

Meister, K., & Andrews, J. R. (1993). Classification and treatment of rotator cuff injuries in the overhand athlete. J Orthop Sports Phys Ther, 18, 413-421.

Meister, K. (2000). Injuries to the shoulder in the throwing athlete, Part I. Am J Sports Med, 28(2), 265-275.

Michener, L. A., McClure, P. W., & Karduna, A. R. (2003). Anatomical and biomechanical mechanisms of subacromial impingement syndrome. Clin Biomech (Bristol, Avon), 18, 369-379.

Miller, S., & Cross, R. (2003). Equipment and advanced performance. In Elliott, B., Reid, M., Crespo, M. (Eds.), ITF Biomechanics of advanced tennis (pp. 177-200). London: ITF Ltd.

Mohtadi, N., & Poole, A. (1996). Racquet sports. In Caine, C. G., Caine, D. J., & Lindner, K. J. (Ed.), Epidemiology of sports injuries. Champaign: Human Kinetics.

Morris, C. (2006). The real reasons behind tournament withdrawal. Personal communication, February 2006.

Morris, M., Jobe, F. W., Perry, J., Pink, M., & Healy, B. S. (1989). Electromyographic analysis of elbow function in tennis players. Am J Sports Med, 17(2), 241-247.

Moseley, J. B., Jobe, F. W., Pink, M. M. et al. (1992). Emg analysis of the scapular muscles during a shoulder rehabilitation program. Am J Sports Med, 20, 128-134.

Mullineaux, D. R., Bartlett, R. M., & Bennett, S. (2001). Research design and statistics in biomechanics and motor control. J Sports Sci, 19(10), 739-760.

Neer, C. S. (1972). Anterior acromioplasty for the chronic impingement syndrome in the shoulder: A preliminary report. J Bone Joint Surg Am, 54(1), 41-50.

Nigg, B. M., & Segesser, B. (1988). The influence of playing surfaces on the load on the locomotor system and on football and tennis injuries. Sports Med, 5(6), 375-385.

Nirschl, R. P. (1992). Elbow tendinosis/tennis elbow. Clin Sports Med, 11(4), 851-870. Noffal, G., & Elliott, B. (1998). Three-dimensional kinetics of the shoulder and elbow

joints in the high performance tennis serve: Implications for injury. Paper presented at the 4th International Conference of Sports Medicine and Science in Tennis, Coral Gables, Florida.

Noffal, G. J. (2003). Isokinetic eccentric-to-concentric strength ratios of the shoulder rotator muscles in throwers and non-throwers. Am J Sports Med, 31(4), 537-541.

Noonan, T. J., & Garrett, W. E., Jr. (1999). Muscle strain injury: Diagnosis and treatment. J Am Acad Orthop Surg, 7(4), 262-269.

Novacheck, T. F. (1995). Walking, running, and sprinting: a three-dimensional analysis of kinematics and kinetics. Instr Course Lect, 44, 497-506.

O'Brien, S. J., Neues, M. C., Arnoczky, S. J. et al. (1990). The anatomy and histology of the inferior glenohumeral ligament of the shoulder. Am J Sports Med, 18, 449-456.

Pappas, A. M., & Zawacki, R. M. (1991). Baseball: Too much on a young pitcher's shoulders? Physician Sports Med, 19, 107-117.

Parker, M. G., Ruhling, R. O., Holt, D., Bauman, E., & Drayna, M. (1983). Descriptive analysis of quadriceps and hamstring muscle torque in high school football players. J Orthop Sports Phys Ther, 5, 2-6.

Pentland, W. E., & Twomey, L.T. (1994). Upper limb function in persons with long term paraplegia and implications for independence. Part I. Paraplegia, 32, 211-218.

204

Perry, J., Gronley, J. K., Newsam, C. J., Reyes, M. L., & Mulroy, S. J. (1996). Electromyographic analysis of the shoulder muscles during depression transfers in subjects with low-level paraplegia. Arch Phys Med Rehabil, 77, 350-355.

Pink, M. M., & Perry, J. (1996). Biomechanics. In F. W. Jobe (Ed.), Operative techniques in upper extremity sports injuries. (pp. 109-123). St. Louis: Mosby.

Pluim, B. (2001). Tendon injury of the shoulder. In Crespo, M., Pluim, B., & Reid, M. (Eds.), Tennis medicine for tennis coaches (pp. 64-68). London: ITF Ltd.

Pluim, B., & Bullock, M. (2003). Wheelchair tennis and physical conditioning. In Reid, M., Quinn, A., & Crespo, M. (Ed.), ITF strength and conditioning for tennis (pp. 203-209). London: ITF Ltd.

Pluim, B., & Safran, M. (2004). From breakpoint to advantage. Vista: Racquet Tech Publishing.

Poppen, N. K., & Walker, P. S. (1976). Normal and abnormal motion of the shoulder. J Bone Joint Surg Am, 58(2), 195-201.

Portus, M. R., Sinclair, P. J., Burke, S. T., Moore, D. J., & Farhart, P. J. (2000). Cricket fast bowling performance and technique and the influence of selected physical factors during an 8-over spell. J Sports Sci, 18(12), 999-1011.

Portus, M. R. (2006). Technique effects on low back pain in cricket bowlers. Unpublished Doctoral Dissertation, University of Western Australia, Perth.

Priest, J. D. (1988). The shoulder of the tennis player. Clin Sports Med, 7(2), 387-402. Rab, G., Petuskey, K., & Bagley, A. (2002). A method for determination of upper

extremity kinematics. Gait Posture, 15(2), 113-119. Reece, L. A., Fricker, P. A., & Maguire, L. M. (1986). Injuries to elite young tennis

players at the australian institute of sport. Aust J Sci Med Sport, 18, 11-15. Reid, M., & Elliott, B. (2002). The one- and two-handed backhands in tennis. Sports

Biomech, 1(1), 47-68. Reinschmidt, C., van den Bogert, A. J., Lundberg, A., Nigg, B. M., Murphy, N., Stacoff,

A., & Stano, A. (1997a). Tibiofemoral and tibiocalcaneal motion during walking: External vs skeletal muscles. Gait Posture, 6(2), 98-109.

Reinschmidt, C., van den Bogert, A. J., Lundberg, A., Nigg, B. M., & Murphy, N. (1997b). Effect of skin movement on the analysis of skeletal knee joint motion during running. J Biomech, 30(7), 729-732.

Reque, J. (2005). Epidemiology of shoulder injuries in professional men's tennis. Personal communication, June 2005.

Reyes, M. L., Gronley, J. K., Newsam, C. J., Mulroy, S. J., & Perry, J. (1995). Electromyographic analysis of shoulder muscles of men with low-level paraplegia during a weight relief raise. Arch Phys Med Rehabil, 76, 433-439.

Ryu, K. N., McCormick, J., Jobe, F.W., Moynes, D. R., & Antonelli, D. J. (1988). An electromyographic analysis of shoulder function in tennis players. Am J Sports Med., 16, 481-485.

Salo, A., Grimshaw, P. N., & Viitasalo, J. T. (1997). Reliability of variables in the kinematic analysis of sprint hurdles. Med Sci Sports Exerc, 29(3), 383-389.

Schmidt-Wiethoff, R., Rapp, W., Mauch, F., Schneider, T., & Appell, H. J. (2004). Shoulder rotation characteristics in professional tennis players. Int J Sports Med, 25(2), 154-158.

Shea, K. M., Lenhoff, M. W., Otis, J. C., & Backus, S. I. (1997). Validation of a method for location of the hip joint center. Gait Posture, 5(2), 157-158.

205

Sprigings, E., Marshall, R., Elliott, B., & Jennings, L. (1994). A three-dimensional kinematic method for determining the effectiveness of arm segment rotations in producing racquet-head speed. J Biomech, 27(3), 245-254.

Stauber, W. T. (2004). Factors involved in strain-induced injury in skeletal muscles and outcomes of prolonged exposures. J Electromyogr Kinesiol, 14(1), 61-70.

Takahashi, K., Elliott, B., & Noffal, G. (1996). The role of upper limb segment rotations in the development of spin in the tennis forehand. Aust J Sci Med Sport, 28(4), 106-113.

Terry, G. C., Hammon, D., France, P., & Norwood, L. A. (1991). The stabilizing function of passive shoulder restraints. Am J Sports Med, 19(1), 26-34.

Townley, C. O. (1986). The capsular mechanism in recurrent dislocation of the shoulder. J Bone Joint Surg, 68, 887-891.

Toyoshima, S., Hoshikawa, T., Miyashita, M., & Aguri, T. (1974). Contribution of body parts to throwing performance. In Biomechanics V (pp. 169-174.). Baltimore: University Park Press.

Trabert, T., & Hook, J. (1984). The serve: Key to winning tennis. New York: Dodd, Mead & Co.

Tsai, N. T., McClure, P. W., & Karduna, A. R. (2003). Effects of muscle fatigue on 3-dimensional scapular kinematics. Arch Phys Med Rehabil, 84, 1000-1005.

Vad, V. B., Gebeh, A., Dines, D., Altchek, D., & Norris, B. (2003). Hip and shoulder internal rotation range of motion deficits in professional tennis players. J Sci Med Sport, 6(1), 71-75.

van den Bogert, A. J., & de Koning, J. J. (1996). On optimal filtering for inverse dynamics analysis. Proceedings of the IXth Biennial Conference of the Canadian Society for Biomechanics, Vancouver, British Columbia.

van den Bogert, A. J. (2006). Reply to: Calculation of scalar joint power or individual joint power terms. BIOMCH-L.

van der Helm, F. C., & Pronk, G. M. (1995). Three-dimensional recording and description of motions of the shoulder mechanism. J Biomech Eng, 117(1), 27-40.

van der Helm, F. C. (1997). A standardised protocol for motion recordings of the shoulder. Paper presented at the First Conference of the International Shoulder Group, Maastricht.

van Drongelen, S., Van der Woude, L. H., Janssen, T. W., Angenot, E. L., Chadwick, E. K., & Veeger, D. H. (2005). Mechanical load on the upper extremity during wheelchair activities. Arch Phys Med Rehabil, 86(6), 1214-1220.

van Gheluwe, B., & Hebbelinck, M. (1986). Muscle actions and ground reaction forces in tennis. Int J Sport Biomech, 2, 88-99.

Verstegen, M. (2003). Developing strength. In Reid, M., Quinn, A., & Crespo, M. (Eds.), ITF strength and conditioning for tennis (pp. 113-135). London: ITF Ltd.

Vicon Motion Systems. (2002). Bodybuilder for biomechanics. Oxford Metrics Ltd: Oxford.

Vint, P. F., & Hinrichs, R. N. (1996). Endpoint error in smoothing and differentiating raw kinematic data: An evaluation of four popular methods. J Biomech, 29(12), 1637-1642.

Vorobiev, A., Ariel, G., & Dent, D. (1993). Biomechanical similarities and differences of A.Agassi's first and second serves. Paper presented at the 11th Symposium of the International Society of Biomechanics in Sports, Amherst, MA.

206

Walshe, A., Wilson, G., & Ettema, A. (1998). Stretch-shorten cycle compared with isometric preload: Contributions to enhanced performance. J Appl Physiol, 84, 97-106.

Warner, J. J., Micheli, L. J., Arslanian, L. E., Kennedy, J., & Kennedy, R. (1990). Patterns of flexibility, laxity, and strength in normal shoulders and shoulders with instability and impingement. Am J Sports Med, 18(4), 366-375.

Warner, J. J., Micheli, L. J., Arslanian, L. E., Kennedy, J., & Kennedy, R. (1992). Scapulothoracic motion in normal shoulders and shoulders with glenohumeral instability and impingement syndrome. A study using moire topographic analysis. Clin Orthop Relat Res, 285, 191-199.

Warren, B. L., & Jones, C. J. (1987). Predicting plantar fasciitis in runners. Med Sci Sports Exerc, 19, 71-73.

Weekes, B. (2006). Subjective impressions of personal wheelchair serve performance. Personal communication, May 2006.

Whiting, W. C., Puffer, J. C., Finerman, G. A., Gregor, R. J., & Maletis, G. B. (1985). Three-dimensional cinematographic analysis of water polo throwing in elite performers. Am J Sports Med, 13(2), 95-98.

Whiting, W., & Zernicke, R. (1998). Biomechanics of musculoskeletal injury. Champaign: Human Kinetics.

Wilson, G., Elliott, B., & Wood, G. (1989). The effect on performance of imposing a delay during a stretch-shorten cycle movement. Med Sci Sports Exerc, 23, 364-370.

Winge, S., Jorgensen, U., & Lassen Nielsen, A. (1989). Epidemiology of injuries in danish championship tennis. Int J Sports Med, 10(5), 368-371.

Winter, D. A. (1990). Biomechanics and motor control of human movement. New York: John Wiley & Sons.

Woltring, H. J. (1985). On optimal smoothing and derivative estimation from noisy displacement data in biomechanics. Hum Mov Sci, 4, 229-245.

Woltring, H. J. (1994). 3-D attitude representation of human joints: A standardization proposal. J Biomech, 27(12), 1399-1414.

Wu, G., & Cavanagh, P. R. (1995). ISB recommendations for standardization in the reporting of kinematic data. J Biomech, 28(10), 1257-1261.

Wu, G., van der Helm, F. C., Veeger, H. E., Makhsous, M., Van Roy, P., Anglin, C., Nagels, J., Karduna, A. R., McQuade, K., Wang, X., Werner, F. W., & Buchholz, B. (2005). ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion--Part II: Shoulder, elbow, wrist and hand. J Biomech, 38(5), 981-992.

Young, J. L., Herring, S. A., Press, J. M., & Casazza, B. A. (1996). The influence of the spine on the shoulder in the throwing athlete. J Back Musc Rehab, 7, 5-17.

Yu, B., Keinbacher, T., Growney, E. S., Johnson, M. E., & An, K. (1997). Reproducibility of the kinematics and kinetics of the lower extremity duing normal stair-climbing. J Orthop Res, 15, 348-352.

Yu, B., Gabriel, D., Noble, L., & An, K. (1999). Estimate of optimum cutoff frequency for the butterworth low-pass digital filter. J Appl Biomech, 15, 318-329.

207

xvii

LIST OF APPENDICES

xviii

APPENDIX A

TECHNICAL NOTE

xix

A.1 Introduction

The original nature of this thesis dictated that several developmental modelling issues

ran in parallel. For example, as detailed in Chapter 4, determination of optimal data

treatment techniques was required. More specifically, the lack of comparable 3D

analyses of the serve necessitated the derivation of appropriate smoothing procedures,

while further demanding the substantiation of the repeatability of the high-performance

serve. Early data analysis also revealed gimbal lock, and the resultant poor

representation of 3D shoulder joint position courtesy of angle calculations based on a

Z-X-Y sequence of Euler angle rotations. A Y-X-Y decomposition was thus utilised to

negotiate this problem, and provide for improved validity in interpreting 3D shoulder

joint motion during the serve.

Finally, as alluded to in Chapters 3-7, data analysis uncovered a source of systematic

error related to the position of the upper arm triad. As discussed in Chapter 3, the

representation of segment motion necessitates the definition of two coordinated

systems. The upper arm or humeral triad directly defines the axes of the TCS. The

position of the anatomically significant landmarks, previously defined in the ‘static trials’,

are recalled in dynamic motion trials with respect to the position of the TCS. As such,

any misapplication of the humeral TCS will result in erroneous kinematics and inverse

dynamics calculations.

In lay terms, the placement of these triads along the long axis of the upper arm and

around the bicep appears to underestimate the actual longitudinal rotation of the

humerus. Similar surface marker placement recently confounded the interpretation of

shoulder joint internal-external rotation data in serves performed by college level players

(Gordon and Dapena, 2006). Indeed, Leardini et al. (2005) elaborated on the problem of

soft tissue artefact (i.e. skin and muscle deformation and displacement, STA) in the

estimation of skeletal system kinematics; referring to it as the most critical source of

error in human movement analysis. To this end, evaluation of the effect of STA in lower

limb joint motion during walking (Reinschmidt et al., 1997a) and running (Reinschmidt et

al., 1997b) demonstrated spurious representations of underlying bone motion through

surface markers. Significantly, at the thigh – where muscle bulk was highest and STA

xx

most pronounced – a ≈10º difference was observed between actual skeletal motion and

the marker-estimated skeletal motion. Relative to this course of studies, this Type 1

error does not confound the legitimacy of the comparisons made between serve types or

techniques. Unfortunately though – with data collected, analysis underway and access to

the same sample unlikely – comprehensive remedial action was not possible.

Nonetheless, in an effort to estimate this error, a single case study was performed with

the coupling of the original upper arm triad (OT) compared with a modified, second triad

(MT), located proximal to the humeral epicondyles but distal of the triceps muscle belly

(Figure A.1).

Figure A.1. Position of the original and modified upper arm triads.

A.2 Data Collection, Treatment and Analysis

Consistent with the methods outlined in Chapter 5 and following an appropriate warm-

up, the subject hit three successful, maximal effort FSs and KSs. Data from the three FSs

and KSs were treated as detailed in Chapter 4. For each trial, the two humeral triads

were labelled and modelled independently. Consequently, two sets of mean kinematic

and kinetic data, pertaining to the differentially represented upper arms, were produced

to describe the shoulder joint motion of the same serves.

xxi

With the validity of the kinematics and kinetics associated with the OT’s representation

of longitudinal shoulder joint rotation of principal concern, predominantly those data

were analysed. Indeed, comparison of the related mean kinematic and kinetic data of

the OT to that of the more distally placed MT, allowed for the representativeness of both

to be qualitatively assessed.

A.2 Results

Table A.1 presents the descriptive statistics of selected shoulder joint kinematics and

kinetics, modelled with the OTs and MTs, for the FS and the KS. While the homogeny of

upper arm position, as described by the plane of elevation and elevation angles at

impact, is evidenced regardless of triad placement, axial rotation of the humerus is

clearly affected. That is, both the magnitude of upper arm MER and the range through

which the humerus internally rotates during the follow-through are ≈40% higher when

calculated using the MT. This difference is even more pronounced (≈150%) when

comparing the peak pre-impact external rotation (expressed in the thorax) angular

velocities between the MT (FS: -24.58±4.5 rad.s-1; KS: -18.99±1.0 rad.s-1) and OT (FS: -

9.65±4.5 rad.s-1; KS: -9.08±1.1 rad.s-1). Indeed, Figure A.2 contrasts the differentially

represented angular velocities of longitudinal upper arm rotation during the

forwardswing of the FS and KS.

xxii

Variable EVENT / Phase

Serve type Original Modified

FS -84.82 (11.5)

-126.31 (3.6) Maximum upper arm external rotation (º) MER

KS -86.59 (11.5)

-120.48 (1.5)

FS -7.98 (4.7) -22.23 (4.4) Peak internal rotation angular velocity (rad.s-1)

Forward-swing KS -9.17 (0.4) -22.96 (1.7)

FS -9.65 (4.5) -24.58 (4.5) Mean internal rotation moment (Nm) Forward-swing KS -9.08 (1.1) -18.99 (1.0)

FS -27.38 (4.0) -33.12 (3.8) Peak internal rotation moment (Nm) Forward-swing KS -30.23 (2.8) -26.98 (3.8)

FS -171.85 (2.3)

-165.42 (5.2)

Plane of elevation (º) IMP KS -188.21

(2.5) -183.88 (2.4)

FS 93.43 (4.6) 100.10 (4.7) Elevation (º) IMP KS 98.78 (3.8) 105.48 (1.8) FS 67.98 (4.2) 85.00 (12.8) Range of longitudinal rotation (º)

Follow-through KS 71.30 (2.1) 104.96 (11.7)

FS 4.86 (1.0) 9.35 (1.2) Mean external rotation moment (Nm) Follow-through KS 3.30 (0.7) 7.29 (0.7)

Table A.1. Mean (± SD) shoulder joint kinematic and kinetic data modelled with the original and modified upper arm triads for the FS and the KS.


-20

-15

-10

-5

0

5

10

15

Ang

ular

vel

ocity

(rad

.s-1

)

FS (Original)KS (Original)FS (Modified)KS (Modified)

xxiii

Figure A.2. The effect of divergent upper arm triad placement on shoulder joint internal (+) and external (-) rotation angular velocities (as expressed in the thorax) during the forwardswing of

the FS and KS.

Unsurprisingly, these differences were compounded in the related kinetic data. That is,

marked variation was observed in the triads’ mean pre-impact internal rotation moments

(MT: ≈-22Nm; OT: ≈-9Nm) and mean post-impact external rotation moments (MT:

≈8Nm; OT: ≈4Nm). However, peak internal rotation moments during the forwardswing

were comparable irrespective of triad (≈-30Nm). Collectively, this moment data suggests

that the coupling between the OT and humeral rotation was not directly related. The

bicep – close to which the OT was placed – was visibly observed to rotate with the

humerus through mid-range (i.e. where peak internal rotation moments are produced)

but not at end range (i.e. near maximum external or internal rotation), thereby

confirming this conception. Similarly miscellaneous coupling of surface based markers

and humeral rotation during the serve has been inferred by Gordon and Dapena (2006).

A.3 Practical Application

This rudimentary analysis provides some insight into how divergent upper arm marker

placement can confound the representation of 3D humeral motion. Similar to that which

has been observed in lower extremity motion analyses (Reinschmidt et al., 1997a,

1997b), evaluation of shoulder joint kinematics and kinetics is affected by varying

amounts of STA, depending on marker application.

In relation to this course of studies, evidence implicates the OTs, which were used to

define the TCSs of the humerus, in erroneously estimating the upper arm’s long-axis

rotation during the tennis serve. This finding is consistent with the data of Gordon and

Dapena (2006), which have previously implicated surface markers attached to fleshy

locations of the upper arm in inaccurate estimations of humeral rotation. Significantly, as

the reported upper arm elevation and plane of elevation angles were consistent

irrespective of triad used, it may be that the spurious effects of differential triad

placement are restricted to data describing humeral rotation. To this end, the OT

appears to underestimate humeral rotation by ≈50%; with the contrast in the associated

higher order kinematics and kinetics even more pronounced (≈100%). Unfortunately, as

this variation was not entirely linear and likely differed between subjects, the elaboration

xxiv

of a ‘correction factor’ by which the current studies’ related shoulder joint kinematics and

kinetics could be multiplied was precluded.

Notwithstanding these results, of further interest is that although the kinematics and

kinetics calculated with the MT were 50-150% higher than the same data from the OT,

they remain appreciably lower than those previously reported to describe humeral

rotation during the tennis serve through the aforementioned inter-segment vector

comparison approach (Elliott et al., 2003; Fleisig et al., 2003). The most probable

explanation of this disparity continues to relate to the divergent modelling methods

these investigations employed to determine 3D shoulder joint motion. Here, of relevant

and important note is that while researchers have advanced several methods to help

compensate for STA, Leardini et al. (2005) have been equally quick to acknowledge that

reliable estimation of 3D motion using skin markers is yet to be satisfactorily achieved.

For example, Elliott et al. (2003) calculated axial rotation of the humerus from a vector

defining the longitudinal axis of the forearm relative to anterior direction of the shoulder

in the transverse plane of the upper arm. Fleisig and colleagues (1996; 1996b; 2003) in

their kinetic study of the tennis serve and throws of other sports employed the same

analysis technique. While this indirect measure of computing internal-external humeral

rotation is touted as accurate in all but extended elbow joint positions, the reality is that

its validity in representing actual 3D humeral joint motion is questionable. Indeed, by

inferring longitudinal rotation of the shoulder joint through changes in the sagittal plane

position of the forearm, these measures effectively discount the role of the trunk and

scapula in contributing to humeral motion.

Nevertheless, indications are that future 3D biomechanical examinations of shoulder joint

motion in the tennis serve can attempt to reduce the spurious effects of STA by defining

the TCS of the humerus through markers placed distal to the biceps and/or triceps

muscle belly. Regrettably, determination of more specific surface marker locations to

optimally represent underlying 3D humeral motion was beyond the scope of this case

study, but should precede the abovementioned prospective analyses.

xxv

APPENDIX B

UNIVERSITY OF WESTERN AUSTRALIA

APPLICATION FORM TO UNDERTAKE RESEARCH

INVOLVING HUMAN SUBJECTS

xxvi

Human Research Ethics Committee Research Services

35 Stirling Highway, Crawley, WA 6009 Telephone +61 8 9380 3703 Facsimile +61 8 9380 1075 Email [email protected] http://www.research.uwa.edu.au/hethics.html

APPLICATION TO UNDERTAKE RESEARCH INVOLVING HUMAN

SUBJECTS

(RESPONSES MUST BE TYPED) NB: Please answer all questions fully in terms which can be readily understood by an informed layperson. This form is designed to ensure compliance with the National Statement on Ethical Conduct in Research Involving Humans. http://www.health.gov.au/nhmrc/publications/humans/contents.htm 1. TITLE OF PROJECT (In lay terms):

Loading and velocity generation in the high performance tennis serve

2. CHIEF INVESTIGATOR:

(Must be a member of Staff of The University of Western Australia.)

Name

Machar Reid

Position:

Doctorate of Philosophy Student

School:

Department of Human Movement and Exercise Science

http://www.health.gov.au/nhmrc/publications/humans/contents.htm

xxvii

Contact Address:

c/Department of Human Movement and Exercise Science Parkway Entrance 3 Crawley 6009 WA

Telephone

(BusinessHours):

+61 401 077 441

Specify Adjunct or Clinical position if held

Email Address:

[email protected]

If this is a student project please include the name, degree course and departmental address of the student. Telephone and email address should be provided if possible. NB: If this is a PhD project, students must indicate this and provide student number.

Student no.: 0041128

3.

EXPECTED DURATION OF PROJECT:

Please note that the research or recruitment of participants must not commence until a date after final approval has been obtained from the HREC.

From Date of Initial Recruitment: July 2004

Date of Expected Completion: January 2005

4. FUNDING: Is this protocol the subject of a grant application? Yes [ ] No [ X ]

xxviii

If 'Yes', what is the Agency?

Provide details of any affiliation or financial interest in funding source and/or commercialisation of research results

5. OTHER ETHICAL APPROVALS: Has the protocol previously been submitted to the Human Research Ethics Committee? Yes [ ] No [ X ] Has the protocol been submitted to another Institutional Ethics Committee? Yes [ ] No [ X ] If ‘Yes’ to which Committee/s has it been submitted?

What was the outcome of the submission?

Has the protocol been submitted to the Confidentiality of Health Information Committee (CHIC)? ie: Health Department or Agency Yes [ ] No [ X ] Does the Project require a “Covenant of Confidentiality”*? Yes [ ] No [ X ] (*ie: accessing private practitioner’s medical records – see Human Ethics Office web page: Code of Practice for the use of Name–Identified Data and Covenant of Confidentiality) 6 PRIVACY LEGISLATION: Does this research project involve access to data held by a Commonwealth Department or agency? Yes [ ] No [ X ]

xxix

If your proposed research project involves access to data held by a Commonwealth Department or agency, you will have to comply with the privacy principles established under Commonwealth Privacy Legislation. Information and further documentation relating to these issues must be obtained from the Secretary to the Committee (Tel: 9380-3703).

Is the data to be collected, used or disclosed from an organisation Yes [ ] No [ X ] in the private sector? Does the data include information that identifies the individual(s) Yes [ ] No [ X ] concerned? 7 AIMS OF THE PROJECT: Please give a concise and simple description of the aims of the project. This must be in lay terms.

Identify characteristics of tennis serve technique that load the shoulder to greater or lesser extents

8 PARTICIPANT GROUP: (a) Who will be the participants? Please include size of sample(s) and variables such as

age, sex and state of health. Please state clearly whether children, mentally ill individuals or persons in dependent relationships such as teacher/student, doctor/patient, staff etc. will be recruited.

Participants will number 18 and will be aged 25±7 years. They will be high

performance tennis players and coaches from the Perth metropolitan area.

(b) From where and how will participants (including controls if applicable) be recruited?

How will the initial contact be made with the participants?

Participants will be known to the researcher and initial contact will be made

by email or a telephone conversation.

xxx

(c) Does recruitment involve the circulation/publication of an advertisement, circular, letter , email list, bulletin etc?

Yes [ ] No [ X ] If ‘Yes’, please provide copies and details of publication

(d) Does the research specifically target Aboriginal people or is the sample

likely to include a significant percentage of Aboriginals? Yes [ ] No [ X ]

If 'Yes', please refer to the NHMRC publication Values and Ethics: Guidelines for Ethical Conduct in Aboriginal and Torres Strait Islander Health Research. Please provide a statement demonstrating how the study recognises the values outlined in the above publication and which Aboriginal groups or organisations have been consulted.

9 DETAILS OF PROCEDURES: (a) Please describe briefly the project methodology. Describe all procedures to which

participants will be subjected, highlighting any which may have adverse consequences.

Participants will have 30 retroflective markers placed on anatomical

landmarks and will be asked to hit 20 serves to designated area within the service court. Filming and 3D reconstruction of serve technique will be achieved with the Vicon, 12 camera, analysis system. There will be no adverse consequences for participants.

(b) Will any chemical compounds, drugs or biological agents be

administered? Yes [ ] No [ X ] If 'Yes', describe names, dosages, routes of administration, frequency of

administration, and any known or suspected adverse effects. All suspected adverse events should be listed on the Information Sheet/Consent Form.

(c) Does the research involve use of unmarketed drugs? Yes [ ] No [ X ] If 'Yes', Clinical Trial Notification (CTN) or Clinical Trial Exemption (CTX)

approval must be obtained before the project may proceed.

http://www.health.gov.au/nhmrc/publications/synopses/e52syn.htm

http://www.health.gov.au/nhmrc/publications/synopses/e52syn.htm

xxxi

Investigational brochure enclosed. Yes [ ] No [ ] CTN approval has been requested. Yes [ ] No [ ] CTX approval has been requested. Yes [ ] No [ ] (d) Will blood or other tissue samples be taken? Yes [ ] No [ X ] If 'Yes', please state site, frequency and volume of any blood or other tissue

sampling.

If 'Yes', list all personnel who will be involved in this procedure.

(e) Will there be any invasive procedures other than blood or

tissue sampling? Yes [ ] No [ X ]

If 'Yes', please provide details of these procedures.

(f) Will participants be exposed to ionising or non-ionising radiation? Yes [ ] No [

X ]

(i) If 'Yes', please provide details including the quantitative assessment of the absorbed dose, supported either by dosimetric calculation or other information.

xxxii

(ii) If 'Yes', has the radiation Protection Office been

asked for approval? Yes [ ] No [ X ] If 'Yes', please attach copy of approval notification.

10 NHMRC NATIONAL STATEMENT ON ETHICAL CONDUCT IN RESEARCH

INVOLVING HUMANS (1999) http://www.health.gov.au/nhmrc/publications/humans/contents.htm

(a) Please indicate whether the protocol conforms to the National Statement on Ethical Conduct in Research Involving Humans. Yes [ X ] No [ ]

(b) Please indicate whether the protocol conforms to the National Statement on Ethical

Conduct in Research Involving Humans with regard to the following areas of research:

(i) Research involving children, young people, persons with intellectual

or mental impairment, persons highly dependent on medical care or persons in dependent or unequal relationships Yes [ ] No [ ] N/A [ X ]

(ii) Research involving collectivities Yes [ ] No [ ] N/A [ X ]

(iii) Research involving Aboriginal and Torres

Strait Islander Peoples Yes [ ] No [ ] N/A [ X ]

(iv) Research involving ionising radiation Yes [ ] No [ ] N/A [ X ]

(v) Research involving assisted reproductive

technology Yes [ ] No [ ] N/A [ X ]

(vi) Clinical trials Yes [ ] No [ ] N/A [ X ]

(vii) Innovative therapy or intervention Yes [ ] No [ ] N/A [ X ]

http://www.health.gov.au/nhmrc/publications/humans/contents.htm

xxxiii

(viii) Epidemiological research Yes [ ] No [ ] N/A [ X ]

(ix) Use of human tissue samples Yes [ ] No [ ] N/A [ X ] (x) Human genetic research Yes [ ] No [ ] N/A [ X ] (xi) Research involving deception of participants,

concealment or covert observation Yes [ ] No [ ] N/A [ X ] (c) Please address the ethical considerations of the proposed research to satisfy the

Committee that the research protocol gives adequate consideration to participants’ welfare, rights, beliefs, perceptions, customs and cultural heritage both individual and collective (Refer Item 2.22 of the National Statement on Ethical Conduct in Research Involving Humans)

11 ETHICAL ISSUES Please indicate which of the following ethical issues are involved in this research. (a) Does data collection require access to confidential

data without the prior consent of participants? Yes [ ] No [ X ] (b) Will visual recordings be made, eg: photo, video, etc? Yes [ X ] No [ ] (c) Will audio recordings be made, eg: tape or digital, etc? Yes [ ] No [ X ] (d) Will participants be asked to commit any act which might

diminish self-respect or cause them to experience shame, embarrassment or regret ? Yes [ ] No [ X ]

(e) Will any procedure be used which may have an unpleasant

or harmful side effect? Yes [ ] No [ X ]

xxxiv

(f) Does the research use any stimuli, tasks, or procedures,

which may be experienced by participants as stressful, noxious, or unpleasant? Yes [ ] No [ X ]

(g) Will the research use no-treatment or placebo control

conditions? Yes [ ] No [ X ] (h) Will any samples of body fluid or body tissue be

required specifically for the research, which would not be required in the case of the ordinary treatment? Yes [ ] No [ X ]

(i) Does the research involve a fertilised human ovum? Yes [ ] No [ X ] (j) Does the project use embryos beyond a period of

fourteen days after fertilisation? Yes [ ] No [ X ] (k) Does the project involve the implantation of embryos,

which have been the subjects of prior experimentation? Yes [ ] No [ X ] (l) Are there in your opinion any other ethical issues

involved in the research? Yes [ ] No [ X ] If the answer to any of the above questions is 'Yes' please amplify below. Details

required of secure storage of recordings, preferably within Departmental facilities.

All video footage taken of the participants will be secured in the Department of Human Movement and Exercise Science’s storage facilities and will not for any reason other than internal data analysis.

12. INFORMATION SHEET AND INFORMED CONSENT FORM: Normally, each participant is given an information sheet and is required to sign a

consent form. Do you undertake to obtain written consent for each participant? Yes [ X ] No [ ]

xxxv

(a) If 'Yes', please attach a copy of the Information Sheet and the Consent Form to be

given to and signed by all participants and/or their responsible signatory. These forms should be on departmental letterhead. The Information Sheet should describe all the procedures proposed in clear, simple terms. It should list any potential short- or long-term side effects and any hazards. The required standard paragraph must be included at the bottom of all Consent Forms, or Information Sheets where appropriate. (As the majority of concerns raised by the Human Research Ethics Committee are raised in connection with the Information Sheet and Consent Form, it is strongly recommended that you consult the Guidelines for Preparation of Information Sheet and Consent Form, available on the Human Ethics Office web page.)

(b) If 'No', please justify this departure from normal procedure.

13. POTENTIAL BENEFITS AND RISKS: (a) What are the possible benefits of this research?

(i) To the participant: Identification of specific characteristics of the participant’s serve that may be

more likely to predispose the participant to shoulder injury. Any subsequent modification of serve technique would then reflect a want to reduce this injury risk.

(ii) To humanity generally; The course of studies will quantify the kinetics of different tennis serve

techniques with a view to providing coaches and players information to minimise upper extremity joint loading while maximising serve speed. Shoulder injury is one of the most common in tennis playing populations and it is widely recognised to be closely associated with the large forces and torques developed about the shoulder during the service motion.

(b) What in your view are the possible hazards of this research to the participants?

xxxvi

As the research will be performed safely and comprehensively, I do not believe there are any hazards for the participant.

14. REMUNERATION: Is any financial remuneration or other reward being offered to participants in the study? Yes [ ] No [ X ] If 'Yes', please state how much will be offered and for what purpose, e.g. to cover

travelling expenses, time spent etc. Volunteers may be recompensed for inconvenience and time spent, but any such payment or compensation should not be so large as to be an inducement to participate.

15. EXTERNAL AUDITS: Will individual results of this study be subject to an audit by

any agency external to the University? Yes [ ] No [ X ]

If 'Yes', who will be conducting the audit?

xxxvii

CHECK LIST To assist with processing this application, I have: 1. answered all the questions fully; Yes [ X ] No [ ] 2. signed this application form and obtained the Head

of Department's signature; Yes [ X ] No [ ] 3. enclosed the original application form, with a copy

of the Information Sheet/Consent Form and any related documentation (questionnaires, letters, etc.) attached; Yes [ X ] No [ ]

4. enclosed eleven, complete, collated copies of the

application form including Information Sheet/Consent Form and related documents as per 3 above. (making 11 copies in all including the original). Yes [ X ] No [ ]

5. and enclosed two copies of the full grant/research

proposal. Yes [ X ] No [ ] Chief Investigator: • I certify that I am the Chief Investigator named on the front page of this application

form. • I undertake to conduct this project in accordance with all the applicable legal

requirements and ethical responsibilities associated with its carrying out. I also undertake to ensure that all persons under my supervision involved in this project will also conduct the research in accordance with all such applicable legal requirements and ethical responsibilities.

• I certify that adequate indemnity insurance has been obtained to cover the personnel

working on this project. • I have read the Code of Practice for the use of Name-identified Data. I declare that I

and all researchers participating in this project will abide by the terms of this Code. • I make this application on the basis that it and the information it contains are

confidential and that the Human Research Ethics Committee of The University of Western Australia will keep all information concerning this application and the matters it deals with in strict confidence.

xxxviii

Name (Please print): Signed: Date:

Head of School:

I am aware of the content of this application and approve the conduct of the project within this school.

Name: (Please print): Signed: Date:

Last updated 1 July 2002

xxxix

APPENDIX C

DOCUMENTS OF INFORMED CONSENT AND INFORMATION

PACKS ISSUED TO ALL SUBJECTS

xl

Loading and velocity generation in the high performance tennis serve.

[email protected] Tel: +61 401 077 441

www.hmes.uwa.edu.au

School of Human Movement & Exercise Science35 Stirling Highway

Crawley WA 6009 Australia

CONSENT FORM

Investigator Responsibilities - Participants Rights

1) As a subject you are free to withdraw your consent to participate at any time. 2) The researcher will answer any questions you may have in regard to the study at any time. Questions concerning the study can be directed to: Machar Reid, B.App.Sc.(Hons) ph 0401077441 School of Human Movement and Exercise Science University of Western Australia I, (print your name) _________________________________ have read the information contained within this consent form and any questions I have asked have been answered to my satisfaction. I agree to participate in this project, realising I can withdraw at any time without being subject to any penalty or discriminatory treatment. I agree that the purpose of this research and potential risks or discomforts involved with the testing procedures have been sufficiently explained to me. I also agree that research data gathered for this study may be published providing my name and confidential details are not used. I have read the aforementioned criteria, been provided with written explanations of the procedures and understand my rights as a participant. ______________________ _________________ Signature of participant Date ______________________ _________________ Signature of investigator Date The Human Research Ethics Committee at the University of Western Australia requires that all participants are informed that, if they have any complaint regarding the manner, in which a research project is conducted, it may be given to the researcher or, alternatively to the Secretary, Human Research Ethics Committee, Registrar’s Office, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009 (telephone number 6488-3703). All study participants will be provided with a copy of the Information Sheet and Consent Form for their personal records.

xli


The University of Western Australia 35 Stirling Highway, Crawley WA 6009 + 61 401 077 441

[email protected]


— Subject Information Sheet —

Purpose

To investigate the effect of tennis serve technique to shoulder joint loading.

Procedures

Participants will be required to present to testing sessions in a rested state. Data

collection will be performed in the Sports Biomechanics Laboratory at the University of

Western Australia. Participants will wear appropriate match-play attire and footwear.

Participants will be required to complete one testing session, during which their service

techniques will be recorded using a three-dimensional (3D) opto-reflective motion

analysis system (Vicon 612, Oxford Metrics, Oxford UK). Subjects will be fitted with 62

individual retro-reflective markers, all 16mm in diametre, using non-allergenic double-

sided adhesive tape such that the Vicon motion analysis system can reconstruct 3D serve

motions.

Subsequent to an appropriate warm-up (jog, run throughs, arm swings and practice

serves), participants will be asked to hit, with maximal effort and their preferred

techniques, three flat and three kick serves to a 1x1 metre target area bordering the ‘T’

of the first service box. They will then be asked to hit a further three successful FS using

two different actions with which they will be familiar: foot-up or foot-back, and with

minimal lower limb involvement. For serves to be considered successful, the ball must

clear the net and bounce in the target area. Participants will use the same 0.675m,

0.385 kg Wilson 6.0 Pro Staff racquet.

xlii

Risks

The protocol requires players to use a standardised racquet and make minor temporary

adjustments to their preferred service technique. However, all participants will be

familiar with the racquet, which is representative of that used by the high-performance

playing population. They will also possess sufficient expertise to seamlessly make the

technical adjustments, all of which will have been trialled at some point in their playing

careers. Furthermore, all care will be taken to ensure risks, hazards or health problems

will be avoided.

The use of non-allergenic double-sided tape alleviates any concerns with respect to the

manifestation of allergic reactions on behalf of participants.

Benefits

Shoulder injuries are among the most common and debilitating injuries suffered by

tennis playing populations and are widely recognised as being closely associated with

the large forces and torques developed about the shoulder during the service motion.

So, by quantifying the upper extremity joint kinetics of different tennis serve techniques,

coaches and players will be provided information to minimise shoulder joint loading

while maximising serve speed.

Also, identification of specific characteristics of a participant’s serve as more likely to

predispose that individual to shoulder injury would facilitate subsequent technique

modification so as to reduce this injury risk.

Subject Rights

Participation in this research is voluntary and you are free to withdraw from the study at

any time without prejudice. You can withdraw for any reason and you do not need to

justify your decision.

If you do withdraw we may wish to retain the data that we have recorded from you but

only if you agree, otherwise your records will be destroyed. Your participation in this

xliii

study does not prejudice any right to compensation that you may have under statute of

common law.

If you have any questions concerning the research at any time please feel free to ask

the researcher who has contacted you about your concerns. Further information

regarding this study may be obtained from Machar Reid (+61 401 077 441).

xliv

APPENDIX D

EXAMPLE OF SUBJECT ANTHROPOMETRIC MEASUREMENTS

xlv


The University of Western Australia 35 Stirling Highway, Crawley WA 6009 + 61 401 077 441

[email protected]


— Subject Anthropometry Sheet —

Name: Age: Height (cm): Weight (kg): Foot length (cm): Foot Alignment Right foot Left foot Inversion (º) Eversion (º) Abduction (º)

Adduction (º)

xlvi

APPENDIX E

RACQUET DIMENSIONS AND INERTIAL CHARACTERISTICS

xlvii

Racquet: Wilson 6.0 Pro Staff Tour

Length: 690mm

Mass: 385g

Calculations have been made in accordance with the procedures recommended by Brody

et al. (2002).

Racquet mass

Balance point

(Giacomini)

P (cm, from COM to 2nd top

string)

Y (cm, from COM to axis of

handle)

Time for 1 oscillation

(sec) Swingweight (SW, from 2nd top string)

380g 33.1cm 29.4cm 22.94cm 1.348 SW = 375.500 Recoilweight (RW, moment of inertia about COM in ‘x’) = SW - MY2 RW = 172.700

Racquet mass

Balance point

(Giacomini)

P (cm, from centre mains

to bar)

Y (cm, from COM to axis of

handle)

Time for 1 oscillation

(sec) Polar moment (PM)

380g 33.1cm 17.7cm As above 0.897 PM = 15.304

Then, using perpendicular axis theorem, moment of inertia of third axis, z = RW + PM = 188.004 kg.cm2

Moments of inertia expressed as percentages of racquet length (as required in the UWA Model):

Moments of inertia about racquet COM (kg.m2)

X 0.250 Y 0.022 Z 0.272

By - the UWA Profiles and Research Repository · Machar Reid This Thesis is Presented for the...

Documents

Transcript of By - the UWA Profiles and Research Repository · Machar Reid This Thesis is Presented for the...