Journal Logo


Shoulder and Hip Roll Changes during 200-m Front Crawl Swimming


Author Information
Medicine & Science in Sports & Exercise: December 2008 - Volume 40 - Issue 12 - p 2129-2136
doi: 10.1249/MSS.0b013e31818160bc
  • Free


Body roll, defined as the rolling action of the trunk around its longitudinal axis, seems to have important functions in front crawl swimming and to be linked to swimming performance. For example, Yanai (24) stated that body roll facilitates the breathing action, whereas Payton et al. (15) suggested that body roll influences the hand displacement relative to the water, thereby contributing to hand velocity. In view of the possibility that body roll may play an important role in improving swimming performance, many investigators have tried to calculate body roll and to determine its effect on front crawl kinematics.

Some investigators have measured body roll for the whole trunk, by attaching a wooden fin on swimmers' backs and calculating its deviation from the vertical axis using one camera and two-dimensional analysis techniques (6,11,15-17). Despite some interesting findings reported in these studies, the assumption that the whole trunk rolls as a rigid segment might not be tenable because evidence from other studies suggests that the shoulders and hips might roll to different extents (5). Thus, body roll needs to be examined with methods that do not rely on that assumption.

The analyses by Cappaert et al. (5) using video data obtained from swimmers during competition and those from Yanai's studies (23,24) on university swimmers seem to have been the only studies conducted in three dimensions (3D) and to quantify separately the shoulder roll (SR) and hip roll (HR) of swimmers. It was reported that swimmers rolled their shoulders considerably more than the hips (5,24) and that swimmers decreased SR by 9° when swimming velocity (V) increased from 1.3 to 1.6 m·s−1 (23). However, the significance of the aforementioned differences and relationships was not assessed statistically.

Although the above-mentioned 3D studies provided some useful data on SR and HR, more research could be beneficial with respect to improving the understanding and expanding the knowledge in this area. Moreover, the use of sophisticated research equipment and designs could facilitate the reduction of possible errors associated with not considering the influence of breathing actions, extrapolation beyond the calibration volume, reduced digitizing reliability due to limited number of cameras, adjustments made for combining above and below water data, and image distortion and refraction. In addition, there are some areas that remain to be investigated. For example, the changes that occur in SR and HR throughout the course of a race remain unknown. Also, there are no data on the relationship between the timings of SR and HR peaks on each side during a stroke cycle (SC). Further, qualitative analysis has indicated that bilateral asymmetries in body roll are common among swimmers and might be related to swimming V (3). However, SR and HR have been calculated for one side only, without identification of the side used, and thus, there remains a lack of information regarding bilateral asymmetries in the magnitude of SR and HR and their association with V.

The purpose of this study was to determine accurately the magnitude and changes in SR and HR throughout a 200-m maximum front crawl swim and whether SR and HR were associated with swimming V. The timings of SR and HR peaks on each side, as well as the bilateral asymmetries in SR and HR and their relationship with swimming V were also investigated. It was hypothesized that SR would be different to HR and that SR and HR would increase with a decrease in V during the test. Finally, given that intersubject variability in kinematic parameters has been identified in many swimming studies and that the calculation of group values only might mask any individual differences, individual profiles and intersubject variability were also explored in the present study.



Ten male front crawl swimmers (age = 17.1 ± 0.9 yr, height = 181.0 ± 5.4 cm, body mass = 72.4 ± 5.7 kg, and personal best performance in the 200-m front crawl = 121.5 ± 4.9 s) of national and international level participated in this study. All swimmers were free from injuries and illness, and the 200-m front crawl was one of their specialist events. The test procedures were approved by the University of Edinburgh ethics committee, and written informed consent forms were obtained from all subjects before their participation in the test.


To minimize any overtraining effects on test performance, swimmers and coaches were instructed to avoid any stressful training the days before the test day. On the testing day, after a personalized warm up, each swimmer performed a 200-m maximum front crawl swim using his exact competition pacing and strategy. A push start was used to eliminate the influence of the dive on the kinematics of the SC analyzed for the first length. To eliminate any effects of breathing on SR and HR (17), swimmers were instructed to avoid breathing while swimming through the 6.5-m calibrated space. Nevertheless, in view of the possibility that the breathing preference could be associated with roll asymmetries even when swimmers do not breathe, the swimmers were requested to inform the investigators of their breathing preference. The right side was the preferred breathing side for all the swimmers tested.

To ensure that test performance would be at a similar level to competition performance, considering the effect of the push start on the final time, a test was considered acceptable if a swimmer's time for the 200-m was less than 105% of his personal best performance of the season. All swimmers satisfied these criteria on their first attempt.

Camera and calibration setup.

All tests were conducted in a 25-m indoor pool. The space of interest was calibrated with a 6.75-m3 calibration frame with orthogonal axes (4.5 m × 1.5 m × 1.5 m, for the direction of swim (X), the vertical (Y), and the lateral direction (Z), respectively), which was positioned in the midsection of the pool with half the frame above and half below the water. Six stationary and synchronized JVC KY32 CCD cameras (four below and two above the water) recorded a space 6.5 m long, extending 1 m beyond each side of the frame for the swimming direction. All cameras were a priori genlocked, and the time codes were displayed on each frame of all video recordings to facilitate subsequent processing. The calibration setup and the accuracy and reliability procedures have been described in detail by Psycharakis et al. (18). The small errors in the calculations suggested very good reconstruction accuracy and reliability and negligible image distortion and refraction. Moreover, it has been shown that small calibrated volumes increase the possibility of larger errors and inaccuracies caused by extrapolations beyond these volumes (7). Thus, the large volume of the calibrated space in the present study minimized the possibility of extrapolation beyond that space, increasing further the accuracy of the measurements. Figure 1 illustrates the camera and frame positions in the pool throughout the recordings as well as the field of view recorded by each one of the six cameras.

Camera and calibration frame setup and cameras' field of view for data collection. (Below water: cameras 1 to 4; Above water: cameras 5 and 6).

Data processing.

One SC was recorded for each 50-m length, and therefore, each variable of interest was calculated for four SC (SC1, SC2, SC3, and SC4) during the 200-m test. Nineteen body landmarks (vertex; shoulder, elbow, wrist, hip, knee, ankle, and metaphalangeal joints; the end of the middle fingers and the big toes) were digitized for each field (50 fields per second) using the Ariel Performance Analysis System (APAS). The 3D reconstruction was performed using the Direct Linear Transformation method (1) incorporated in APAS. The accuracy of locating submerged markers was maximized by having four cameras. This meant that for the vast majority of the digitized frames, each marker was clearly visible by at least two different cameras, minimizing the incidence of "guessed points" being used in the DLT calculation.

The above- and below-water sequences were digitized and transformed separately. The different sequences were then combined into a single file. A Fourier transform and inverse transform were used to filter and smooth the raw displacement data by retaining harmonics up to 6 Hz in the inverse transform.

Data analysis.

The trunk vector was defined by connecting a line from the midpoint of the shoulder to the midpoint of the hip joints. The normal to the shoulder/trunk plane was defined as the cross product of the trunk unit vector and the unit vector in the direction of the line connecting the shoulder joints. The SR angle (in degrees) was calculated as the angle between the vertical and the projection of the normal onto the Y-Z plane. The HR angle (in degrees) was calculated in the same manner as SR, except that the normal to the hip/trunk plane rather than the normal to the shoulder/trunk plane was projected onto the Y-Z plane. The unit vector representing the line of the hips was in the direction of the line joining the hip axes. The magnitude of SR and HR, being the sum of the roll to the right and left sides, was calculated for each SC that was analyzed. Maximum SR and HR to each side and their timing as a percentage of the SC were calculated to identify any in-phase or out-of-phase relationships between SR and HR. Timing of achieving the neutral (0°) SR and HR positions was also determined. The average swimming V (m·s−1) for each swimmer was calculated by taking the mean of instantaneous V values for one complete SC. Also, given that Yanai (23) reported that changes in body roll might be associated with changes in stroke frequency (SF), SF (Hz) in the present study was calculated for each SC to explore any association with changes in SR and HR.

Bilateral asymmetries in roll were calculated as the absolute value of the peak roll difference between the left and right sides. Moreover, it has been reported that technique asymmetries in swimming, such as asymmetric arm coordination and arm pulling patterns, could be related to handedness and/or breathing side preference of swimmers (21). In view of the possibility that asymmetries in rolling patterns might also be related to handedness, roll dominance in the present study was examined on the basis of handedness. In line with other studies in this area (e.g., (21)) arm dominance was determined with the use of a laterality questionnaire and based on the methods of Annett (2) and Oldfield (14). Given that all swimmers who participated in the test had right arm dominance, the roll values for the right and left sides were calculated to identify any side dominance and the extent to which such dominance was associated with handedness. Rolling to the right was defined as the rolling of the shoulders/hips during the phase of the stroke that the right shoulder/hip was higher than the left shoulder/hip, and vice versa.

Digitizing reliability.

One complete SC of one swimmer was digitized 10 times for all six cameras. For each variable of interest, the SD and coefficient of variation (CV) across all digitizations were calculated as an indication of reliability. These calculations indicated good reliability. The SD and CV values for each variable were as follows: 0.002 m·s−1 and 0.09% for average V, 2.41° and 1.88% for SR, 1.85° and 3.92% for HR, 0.47% of SC and 1.03% for the timing of maximum SR on each side, and 0.62% of SC and 1.87% for the timing of maximum HR on each side.

Statistical analysis.

To identify significance of changes in variables across the stages of the swim, a two-way repeated-measures ANOVA was performed for each variable, with SC as the fixed factor and subjects as the random factor. In addition, post hoc tests were conducted to identify the significance of the findings for all six SC pair combinations. To eliminate the possibility of type I errors in these post hoc tests, a Bonferroni adjustment to reduce the alpha level was applied as described by Vincent (22). For all the repeated-measures ANOVA, the assumption of sphericity was tested. Given that this assumption was not violated, no further data adjustments were required.

To assess the nature and strength of correlations between SR/HR and V for each of the race stages, the Pearson's product moment correlation coefficient (r) was calculated. This meant that there were 10 scores (one for each participant) for each pair of variables for each race stage (SC1 to SC4 and mean 200-m scores). A paired samples t-test was performed to identify significant differences between SR and HR, as well as between the peak roll values for each side for both shoulders and hips. Finally, SR and HR asymmetries were also quantified with the use of the symmetry index (SI), as described by Robinson et al. (19), according to the following equation:

where RL is the roll value recorded for the left side and RR is the roll value recorded for the right side. As suggested, a value between −10% and 10% for the SI implies symmetry. Left- and right-side asymmetries are indicated when SI < −10% and when SI > 10%, respectively. For all statistical calculations, significance was accepted at P < 0.05. All statistical analyses were conducted with the use of the Statistical Package for Social Sciences software (version 14.0; SPSS Inc., Chicago, IL).


Differences in the magnitude of SR and HR.

Swimmers were rolling their shoulders significantly more than their hips throughout the test (P < 0.001), with the magnitude of HR being less than half the magnitude of SR. Figure 2 shows a roll-time profile that reflects the mean values for the shoulders and hips of the swimmers during SC1. The profiles are clearly very different in terms of the magnitude of SR and HR.

Mean values for the shoulder (continuous line) and hip roll (dashed line) pattern of the group of swimmers during the first stroke cycle. (Positive roll values represent roll to the right and negative roll values roll to the left side).

Changes in swimming V, SF, SR, and HR throughout the test.

Table 1 shows the values and the repeated-measures ANOVA tests for V, SF, SR, and HR. Swimmers generally decreased V during the test; V in SC1 was significantly higher than the other three SC (P < 0.001) and V in SC2 was significantly higher than in SC3 (P = 0.004) and SC4 (P = 0.009). SF was significantly higher in SC1 than in the other three SC (P ≤ 0.001). The SD values were particularly high for HR, with the significant between-swimmer differences (P < 0.001) indicating that the magnitude of roll varied among swimmers. Unlike SR, there was a trend to increase HR during the test, with the values for SC4 being significantly higher than in SC1 (P = 0.001).

Velocity, shoulder roll, hip roll, and stroke frequency values and repeated-measures ANOVA throughout the 200-m freestyle test.

Correlations between swimming V and SR/HR.

SR had a negative and significant correlation with swimming V for all SC and the mean 200-m values (−0.663 ≤ r ≤ −0.634, 0.037 ≤ P ≤ 0.049), with the exception of SC4, indicating that, in general, faster swimmers were rolling their shoulders less than slower swimmers. The correlations between HR and swimming V did not reach significance.

Changes and differences in the timings of SR and HR.

Table 2 shows the timings of SR and HR peak values for the left and right sides and at the two points of neutral (0°) roll as well as the differences in these timings. Despite individual differences observed between swimmers (P < 0.001), there was no indication of out-of-phase relation between SR and HR for any side or at the points of 0° roll. The repeated-measures ANOVA showed no changes in the timing differences during the test and no correlations between V and the timing differences in SR and HR.

Magnitude of shoulder roll (SR) and hip roll (HR) peaks for the left and right sides and timings of SR and HR peaks for the left and right sides and at the two points of neutral (0°) roll throughout the 200-m freestyle test.

Bilateral asymmetries in SR and HR.

Bilateral asymmetries of considerable magnitude were found for both SR and HR. These asymmetries as well as the SI for both shoulders and hips are shown in Table 3. Swimmers rolled their shoulders significantly more to the left than to the right side throughout the test (0.000 ≤ P ≤ 0.022), for which seven swimmers had SI lower than −10% for SR. No significant side bias was identified for HR on the mean population. Nevertheless, SI < −10% for two swimmers (indicating left asymmetry), SI > 10% for four swimmers (indicating right asymmetry), and −10% ≤ SI ≤ 10% for four swimmers (indicating symmetry) suggested several swimmer profiles (Table 4). The repeated-measures ANOVA showed no statistically significant changes in the magnitude of bilateral asymmetries throughout the test for either SR or HR. Swimming V was not significantly correlated with the magnitude of asymmetries in both SR and HR.

Shoulder and hip roll asymmetries and symmetry index (SI) throughout the 200-m freestyle test.
Shoulder and hip roll asymmetry profiles for the 200-m freestyle test.

Swimmer profiles according to SR and HR asymmetries and timing differences.

Table 4 presents the swimmer profiles according to the bilateral roll asymmetries, whereas details of the roll asymmetries and SR − HR timings for each swimmer are shown in Table 5. All swimmers were right-handed, and their preferred breathing side was the right. There was a large variability in the roll profiles, and on the basis of the data presented in Table 4, four main profiles can be identified: (a) SR and HR symmetry (two swimmers), (b) SR or HR asymmetry only (three swimmers), (c) SR and HR asymmetry on the same side (two swimmers), and (d) SR and HR asymmetry on opposite sides (three swimmers). Although no consistent pattern was found for the group, many differences were observed among swimmers (between-swimmer differences significant for P < 0.001), with the shoulders and hips being out of phase in some cases but with no particular dominant pattern identified. As mentioned previously, neither the roll asymmetries nor the timing differences were associated with velocity.

Mean ± SD 200-m values for velocity, shoulder roll (SR), and hip roll (HR) symmetry index (SI) and differences in timing of peak SR and HR on the left and rights sides for all swimmers.


The present study indicated that the magnitudes of SR and HR are significantly different, and therefore, body roll must be examined separately for shoulders and hips for the purposes of describing accurately the rolling characteristics of the trunk and exploring further their influence on swimming performance. Moreover, considering the identified intersubject variability, the study of body roll should include the analysis of roll profiles of individual swimmers.

Swimmers increased HR as swimming V decreased during the test. There was also a decrease in SF that could be possibly associated with longer SC times, which would allow extra time for swimmers to roll their hips more. Indeed, swimmers spent significantly less time in SC1 than in the other three SC (0.000 ≤ P ≤ 0.003). The increase in HR is in agreement with the findings of Yanai (23) who reported that swimmers increased body roll when swimming slower. However, contrary to Yanai's data, no changes were found in SR during the present study. A possible explanation is the differences in the level of participants in the two studies (and, subsequently, the range of velocities tested), as Yanai tested collegiate swimmers, whereas national- and international-level swimmers were tested in the present study. Despite that the effect of skill on body roll has not been clearly established yet, evidence from previous studies suggests that skill might affect the magnitude and patterns of SR and HR (5). Moreover, despite that a decrease in body roll for a higher V was reported in the former study, this decrease was not assessed statistically for shoulders and hips. Cappaert et al. (5) suggested that possibly SR and HR changes in swimming are linked to resistive forces. Given that active drag is affected by frontal surface area, large differences in the magnitude of SR and HR would increase the frontal surface area and create large resistive forces. Because the increase in HR during the test resulted in smaller differences between the magnitudes of SR and HR, it seems that as the test progressed, the swimmers experienced lower resistive forces at the positions of peak roll. Finally, the fact that swimmers changed the magnitude of HR but not SR during the test underlined that separate calculation of SR and HR is essential in front crawl swimming.

Faster swimmers had less SR than slower swimmers, but no similar pattern was identified for HR. Although possibly the magnitude of SR is constrained by the duration of the SC, swimming V was not correlated with SF or SC time for any SC during the test. Given that faster swimmers had lower SR than but not significantly different HR to slower swimmers, the differences between SR and HR would be lower for the faster swimmers. Thus, it may be hypothesized that the amount of "twist" in the upper body somehow influences performance, perhaps by affecting frontal surface area and, therefore, resistive forces, something that requires further investigation in future studies. Yanai (24) stated that it seems paradoxical that swimmers are often recommended to increase body roll to enhance their swimming performance, because a complex mechanical association with propulsion (which will be hard to accomplish without reducing V) would be required for a swimmer to increase body roll at a given speed. Yanai added that such mechanical propulsion would include arms and legs producing forces in nonpropulsive directions, thus reducing the efficiency of propulsive forces at the swimming direction.

The bilateral asymmetries found for both SR and HR emphasized the necessity to calculate these kinematic variables for both sides. Several swimmer profiles were identified on the basis of roll asymmetry. Despite the individual asymmetries in HR, no consistent side bias was identified. However, the data indicated left-side dominance in SR. Bilateral asymmetries were evident in both faster and slower swimmers, and no relationship was identified between swimming V and the magnitude of asymmetries. Seifert et al. (21) also reported asymmetries for swimmers of all levels, showing that arm coordination asymmetry (with propulsive discontinuity on one side and propulsive superposition on the other) varied greatly among swimmers and that technique asymmetry was associated with the handedness and unilateral breathing pattern of swimmers. It has also been reported that swimmers might apply larger propulsive forces when pulling with the dominant arm (10,12). Given that the swimmers in the present study were right-handed and that maximum SR to the left side occurred during the underwater phase of the right arm, possibly the left-side dominance in SR could be related to differences in the magnitude, duration, timing, or direction of propulsive forces between the underwater phases of the dominant and nondominant arm. Finally, after a review of dominance and symmetry in gait studies, Sadeghi et al. (20) suggested that the dominant limb might be used mainly for propulsion and the nondominant for control and support. Similarly, the observed asymmetries in the roll values for the group of swimmers in the present study could also be related to differences in the main functional roles of the upper limbs.

The analysis of the timings of SR and HR did not reveal a dominant pattern among swimmers. Nevertheless, despite the mean group values not being consistent and not significantly different across the SC of the 200 m, the patterns used by swimmers were highly individual, with differences in timings of maximum SR and HR. Moreover, the timing of maximum SR relative to maximum HR varied between swimmers, as well as between the left and right sides of individual swimmers. Nevertheless, swimming performance was not linked to these differences, providing no evidence that the hips peaking at the same time as the shoulders or leading/trailing the shoulders can lead to an improvement in swimming performance. This indicates that swimmers optimize their coordination in ways that suit their individual characteristics and supports modern views from the motor control literature that optimal patterns of coordination are specific to each individual (e.g., (13)).

There was no clear indication with respect to how each of the different roll profiles could be linked to performance or if some profiles are better than others. This is an area that can be explored further in future studies, possibly by combining the roll analysis with an analysis of intracycle velocity and acceleration patterns, as well as with other factors that might be related to roll profile differences such as anthropometric characteristics. Nevertheless, the inter- and intraindividual variability observed in the roll profiles and the timings of SR and HR could be explained from a functional perspective. Glazier et al. (9) stated that, as the resistive drag forces are constantly fluctuating, an effective front crawl swimming technique must be sufficiently flexible and adaptable to enable emerging patterns of coordination to be modified according to this and other constraints impinging on the swimmer. These authors suggested that when the magnitude of body roll changes the patterns of movement produced by the swimmers' arms, the orientation of the hands during the underwater phase needs to be modified to increase the propulsive forces generated by the hands. In addition, the technique adaptations made by individual swimmers also depend on other factors related to resistive forces, such as swimming V and body posture in different phases of the SC. Thus, it seems that the asymmetric roll profiles of swimmers in the present study might be related to other technique asymmetries and to differences between the right and left underwater phase of the stroke in the resistive forces and the magnitude, duration, timing, or orientation of propulsive forces. Therefore, the differences in roll profiles and timing of roll peaks among swimmers might be attributed to attempts of individual swimmers to adjust their technique to compensate for constraints such as constantly fluctuating resistive forces and, hence, to maximize effectiveness. The above could also explain the intraindividual variability observed during the test. Bartlett et al. (4) and Davids et al. (8) stated that even elite athletes are unable to reproduce invariant movement patterns and that such movement and coordination variability could indeed be functional, to reduce injury risk (as overuse injuries might be related to low variability) and facilitate changes in coordination patterns due to ever-changing constraints imposed on them by various factors. Further research on roll differences, with consideration on coordination patterns and the effects of constraints such as resistive forces, would facilitate the confirmation of the above assumptions and improve the understanding on the associations between different roll profiles and swimming performance.

In this study, only nonbreathing SC were analyzed. However, possibly asymmetries in kinematic characteristics of breathing SC could still influence the technique of nonbreathing SC, especially for swimmers using unilateral breathing patterns. For example, swimmers' SR toward the breathing side may be greater during a breathing than during a nonbreathing SC. Thus, given that all the swimmers' preferred breathing side was the right, possibly swimmers rolled their shoulders more to the left while trying to adjust from the higher roll of a breathing SC. Seifert et al. (21) also reported that swimmers with unilateral breathing patterns displayed asymmetric arm coordination, whereas two swimmers with bilateral breathing patterns were symmetrical. The investigators stated that by consistently turning the head to the same side to inhale, swimmers stabilize this automatism and develop a breathing laterality that could lead to asymmetries in other variables, such as stroke times and propulsive actions of the arms. Given that the influence of breathing actions and breathing laterality on SR/HR values and symmetry has not been examined in front crawl swimming, in future studies, it would be of interest to explore changes in SR and HR between breathing and nonbreathing SC and the effects of unilateral and bilateral breathing on roll asymmetries. Furthermore, such an investigation could be combined with an analysis of the association of the roll maxima with propulsive and resistive phases of the underwater part of the SC, to enable the identification of any links between roll and propulsion/resistance during swimming.


The present study indicated that separate calculation is required for SR and HR to explore their influence on front crawl swimming, because swimmers rolled their shoulders significantly more than their hips. Faster swimmers were generally rolling their shoulders less than slower swimmers, whereas all swimmers decreased HR during the 200-m test. The group of right-handed swimmers in this study displayed left-side dominance in SR, implying that factors related to handedness might affect the SR symmetry in swimming.

Dr. Stelios Psycharakis thanks the Greek State's Scholarship Foundation for its financial support throughout the period of this study.


1. Abdel-Aziz YI, Karara HM. Direct linear transformation from comparator coordinates into object space coordinates in close range photogrammetry. In: American Society of Photogrammetry Symposium on Close Range Photogrammetry; 1971 Jan 26-29: Falls Church (VA). American Society of Photogrammetry; 1971. p. 1-18.
2. Annett M. A classification of hand preference by association analysis. Br J Psychol. 1970;61(3):303-21.
3. Arellano R, Lopez-Contreras G, Sanchez-Molina JA. Qualitative evaluation of technique in international Spanish junior and pre-junior swimmers: an analysis of error frequencies. In: Chatard JC, editor. Biomechanics and Medicine in Swimming IX. St Etienne (France): University of St Etienne Publications; 2003. p.87-92.
4. Bartlett R, Wheat J, Robins M. Is movement variability important for sports biomechanists? Sports Biomech. 2007;6(2):224-43.
5. Cappaert JM, Pease DL, Troup JP. Three-dimensional analysis of the men's 100-m freestyle during the 1992 Olympic games. J Appl Biomech. 1995;11(1):103-12.
6. Castro F, Minghelli F, Floss J, Guimaraes A. Body roll angles in front crawl swimming at different velocities. In: Chatard JC, editor. Biomechanics and Medicine in Swimming IX. St Etienne (France): University of St Etienne Publications; 2003. p. 111-4.
7. Challis JH. A multiphase calibration procedure for the direct linear transformation. J Appl Biomech. 1995;11(3):351-8.
8. Davids K, Glazier P, Araujo D, Bartlett R. Movement systems as dynamical systems, the functional role of variability and its implications for sports medicine. Sports Med. 2003;33(4):245-60.
9. Glazier PS, Wheat JS, Pease DL, Bartlett RM. Dynamic systems theory and the functional role of movement variability. In: Davids K, Bennett S, Newell K, editors. Movement System Variability. Champaign (IL): Human Kinetics; 2006. p. 49-72.
10. Keskinen OP, Keskinen KL. Velocity profiles of competitive swimmers and triathlonists during an all-out 100-m swim. In: XII FINA World Congress on Swimming Medicine; 1997 Apr 12-17: Goteborg (Sweden). FINA; 1997. p. 351-6.
11. Liu Q, Hay JG, Andrews JG. Body roll and handpath in freestyle swimming: an experimental study. J Appl Biomech. 1993;9(3):238-53.
12. Maglischo EW, Maglischo CW, Santos TR. Patterns of forward velocity in the four competitive swimming strokes. In: Proceedings of the VIIth International Symposium of the Society of Biomechanics in Sports; 1989 Jul 2-7: Footscray (Australia). Footscray Institute of Technology; 1989. p. 139-49.
13. Newell KM. Coordination, control and skill. In: Goodman D, Franks I, Wilberg RB, editors. Differing Perspectives in Motor Learning, Memory, and Control. Amsterdam (The Netherlands): North-Holland; 1985. p. 295-317.
14. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9(1):97-113.
15. Payton CJ, Baltzopoulos V, Bartlett RM. Contributions of rotations of the trunk and upper extremity to hand velocity during front crawl swimming. J Appl Biomech. 2002;18(3):243-56.
16. Payton CJ, Bartlett RM, Baltzopoulos V. The contribution of body roll to hand speed in front crawl swimming-an experimental study. In: Keskinen KL, Komi PV, Hollander AP, editors. Biomechanics and Medicine in Swimming VIII. Jyvaskyla (Finland): University of Jyvaskyla; 1999. p. 65-70.
17. Payton CJ, Bartlett RM, Baltzopoulos V, Coombs R. Upper extremity kinematics and body roll during preferred-side breathing and breath-holding front crawl swimming. J Sports Sci. 1999;17(9):689-96.
18. Psycharakis SG, Sanders R, Mill F. A calibration frame for 3D swimming analysis. In: Proceedings of XVII International Symposium on Biomechanics in Sports; 2005 Aug 22-27: Beijing (China). The China Institute of Sports Sciences; 2005. p. 901-5.
19. Robinson RO, Herzog W, Nigg BM. Use of force platform variables to quantify the effects of chiropractic manipulation on gait symmetry. J Manipulative Physiol Ther. 1987;10(4):172-6.
20. Sadeghi H, Allard P, Prince F, Labelle H. Symmetry and limb dominance in able-bodied gait: a review. Gait Posture. 2000;12(1):34-45.
21. Seifert L, Chollet D, Allard P. Arm coordination symmetry and breathing effect in front crawl. Hum Mov Sci. 2005;24(2):234-56.
22. Vincent WJ. Statistics in Kinesiology. 3rd ed. Leeds (UK): Human Kinetics Europe Ltd; 2005. p. 344.
23. Yanai T. Stroke frequency in front crawl: its mechanical link to the fluid forces required in non-propulsive directions. J Biomech. 2003;36(1):53-62.
24. Yanai T. What causes the body to roll in front-crawl swimming? J Appl Biomech. 2001;17(1):28-42.


©2008The American College of Sports Medicine