POLLOCK, COURTNEY L.1; JENKYN, THOMAS R.2,3,4; JONES, IAN C.2; IVANOVA, TANYA D.1; GARLAND, S. JAYNE1,5
The rowing stroke consists of drive and recovery phases. The drive phase starts with the catch position, when the blade of the oar enters the water and is immediately loaded as the rower applies pressure at the fixed foot stops initiating the drive. The finish position of the drive phase occurs when the blade is removed from the water. The recovery phase is the forward motion of the body and arms to prepare the oar for reentry into the water. The force generated by the legs and the acceleration of the body are the main components of the drive that determine the force developed at the handle (2).
During the rowing stroke, the trunk acts as a link in the kinetic chain both generating and transferring forces from the legs and arms to the oar. Baudouin and Hawkins (2) suggest that the ability of the trunk to transfer forces through to the arms and subsequently to the handle is imperative for the resulting force on the oar. There is limited information about the control of the pelvis and the trunk because it pertains to the transfer of forces and the stabilization of the spine during loading.
The spine is thought to be vulnerable during early drive as it is loaded in a flexed position and must change quickly from reaching out to a highly loaded extension movement with the transfer of large forces generated by the legs. Peak forces estimated at L4/L5 during a 2000-m race simulation on a rowing ergometer were found to be 2694 N of compressive force (4.6 times the rower's body weight) and 660 N of shear force (22). The cumulative effect of the rowing stroke also needs to be considered given the fact that the athlete will load the spine in this way approximately 230-260 times in a single 2000 m race, which may be repeated three to six times during training sessions. The most common injuries reported in rowing relate to low back injuries often associated with repetitive strain or cumulative microtrauma (10,26,27). Muscular dysfunction and motor control errors in maintaining spine stability have been suggested as possible factors contributing to low back disorders and chronic back pain (12,19,23).
Motor control studies of the trunk during static upright posture have demonstrated varying levels of coactivation of trunk flexor and extensor musculature with increased loads or challenges to postural stability (6,8,9,28). In these studies, a great deal of attention has been paid to the concept of coactivation of trunk muscles as a means to stabilize the spine and how levels of coactivation are mediated in dynamic conditions. Of the dynamic investigations, those most closely resembling the activity of rowing include lifting and pulling tasks (14-16,30,31). The level of coactivation of the trunk flexor and extensor muscles has been demonstrated to increase during asymmetrical versus symmetrical pull tasks (14) in subjects experiencing low back pain (LBP) (16,30) and during lifts with unstable loads (31).
There is a paucity of literature about motor control and spinal stabilization strategies during rowing and during sport in general. EMG recordings of the trunk muscles during rowing have been limited and have not included concurrent investigation of detailed kinematics of the spine and pelvis (3,25). The goals of this investigation are 1) to characterize the muscle activation patterns of the trunk and pelvis together with segmental kinematics of the pelvis, lumbar, and lower thoracic spine and 2) to investigate the concept of coactivation among the trunk flexor and the extensor muscles of the spine during the rowing stroke.
National team candidates (open class women) who train at the national training center for Rowing Canada Aviron were invited to participate in the study. Participation in the study was voluntary, and written informed consent was obtained from individuals. The study was approved by the University Ethics Review Board. Twelve of these individuals consented to participate. Nine subjects' data were included for analysis. Data from three of the twelve subjects were excluded due to technical difficulties during data collection. The mean age of the subjects was 25.8 ± 2.6 yr, height was 179.2 ± 2.1 cm, and weight was 75.8 ± 5.5 kg. Three subjects had past injuries related to the trunk involving rib stress fractures, which required 2 wk to 2 months off rowing, and three different subjects had a history of LBP, only one of these subjects took any time (1 wk) away from training due to the injury. All subjects reported being healthy, that is, without injury that was impacting training. Seven subjects trained with both sculling and sweep rowing, whereas two subjects trained only in sweep rowing.
Subjects performed a standardized 2000-m race simulation on a Concept 2 rowing ergometer (Concept II Inc., Morrisville, VT). The warm-up was self-selected on the ergometer, approximately 20 min in length. Subjects were encouraged to approach the 2000-m race simulation with maximal effort, as they normally would when submitting their scores to the national team. This session of data collection was in place of a regular workout in the athletes' training week. Two athletes performed personal bests during the testing.
Testing was performed in a biomechanics laboratory equipped with a motion capture system consisting of eight high-resolution digital cameras (Motion Analysis Corp., Santa Rosa, CA). Twenty-two autoreflective markers were attached to each subject at landmarks that define the three-dimensional position and orientation of the rigid segments and tracked by the real-time motion analysis system using Eagle Real Time Software (Motion Analysis Corp). The reflective markers were placed over the spinous process of C7, T4, T7, T10, L1, L3, S1, and the right scapula and bilaterally at the lateral midline of iliac crest, greater trochanter, knee, ankle, wrist, elbow, and acromion process (Table 1). Markers were attached with double-sided adhesive discs and hypoallergenic liquid skin adhesive (Graftobian, Madison, WI). Kinematic data were processed with a fourth-order Butterworth digital filter at 6 Hz and stored on a computer for offline analysis. Postprocessing of data was completed to ensure proper marker recognition. Instances of markers lost to camera view were brief and corrected with interpolation of points based on the existing trajectory of the marker. Error of the marker location with the motion capture system in the laboratory has been established previously as 0.5 mm (13).
EMG data were collected with TeleMyo 2400T telemetric surface real-time EMG system (Noraxon Inc., Scottsdale, AZ). Recording of the muscle activity was unilateral using disposable Ag-AgCl surface electrodes (Tyco Healthcare Group LP, Mansfield, MA); therefore, this study is not designed to investigate differences between sides but rather between flexors and extensors on the same side. Electrodes in bipolar configuration were placed over the belly of the muscle at a distance of 3 cm (center to center) and were secured using a hypoallergenic liquid skin adhesive (Graftobian). EMG data were sampled at 1200 Hz with a 12-bit analog to digital conversion and band-pass filtered at 16-500 Hz. All data were stored on computer for later analysis.
The placement of the EMG electrodes followed previously established protocols in bipolar configuration (4,15,17). Trunk flexors included rectus abdominis (RA), external oblique (EO), and transversus abdominis/internal oblique (TrA/IO) muscles considered together secondary to the likelihood of crosstalk and the similar function of these muscles (17,29). Extensors included trunk extensors, erector spinae lumbar (ES lumbar), erector spinae thoracic (ES thoracic), latissimus dorsi (LD), and hip extensors, biceps femoris (BF), and gluteus maximus (GM) muscles (Table 1).
EMG and kinematic data were collected for 30 s starting at 250 m and at 250-m increments of the race simulation. Only the first 250-m collection period was analyzed for the purpose of this investigation. There were usually 14-15 strokes in the 30-s collection period. This period in the 2000-m race simulation has been chosen because it represented steady state after the initial "push" and was likely to be before the onset of significant fatigue. The catch was defined as the most forward position of the wrist marker in the sagittal plane and the finish as the furthest position back. Each stroke was aligned to the catch, with a stroke represented as finish to finish positions of the wrist in the sagittal plane. This ensured that EMG activity and kinematics leading into the catch would be clear. Stroke time was converted to 100% of stroke to allow for comparison among strokes with the catch represented as 0% of the stroke. Five strokes that were free of movement artifact and most consistent in timing were averaged (Fig. 1).
Specifically, two areas of the drive will be discussed. The first is the period of peak force production, and the second is the phase of the stroke at which coactivation between the flexor and the extensor muscle groups was identified on the EMG. The period of peak force production was calculated based on the data of McGregor et al. (21) who reported the timing of peak force production among elite open class female rowers during maximal effort rowing to be 18.8% ± 1.8% postcatch, as defined by the onset of force production at the handle. To ensure capturing an accurate representation of EMG and kinematic data surrounding this point, we elected to use a range of 14-24% postcatch as the period of peak force production. Although we did not measure the force production during the current data collection due to lack of availability of equipment, subsequently we were able to measure the peak force on the handle. Two of the nine rowers were able to return to the laboratory, and both rowers displayed peak force at the handle within the range of 14-24% postcatch. The period of coactivation between the flexors and the extensors was identified from the onset of the earliest flexor muscle to the offset of the latest extensor muscle. Figure 2 depicts averaged rectified EMG from eight muscles of a single subject with the periods of peak force production and coactivation noted.
Kinematic data were analyzed with a custom written program (Matlab, Mathworks Inc., Natick, MA). Not all kinematic markers were used for the purpose of this investigation. The angles calculated in the custom program were within 0.5° of known angles created with the markers on a goniometer. Spinal segment angles were calculated and reported in the sagittal plane (see angle β, Fig. 3C). The spinal segment angles reported are representative of the motion of the upper vertebral segment marker relative to the lower vertebral segment marker (e.g., within the T7-T10 segment, the degree of movement is reflective of the motion of T7 relative to T10). The pelvis was determined as the plane created by three points: the markers at the right and left iliac crests and the sacral marker (Fig. 3B). Pelvic angles in the sagittal plane were representative of the angle between the pelvis and a horizontal vector at 0° as a constant base of reference (see angle γ, Fig. 3B). Angular velocities were calculated by differentiation of the position throughout the stroke. Peak angular velocities identified the percentage of stroke, during which each segment of the spine and pelvis was showing the greatest change in position over time.
EMG data were processed using Spike2 software (Cambridge Electronic Design, Cambridge, England, UK). EMG data for each muscle were filtered using a high-pass filter set at 30 Hz, rectified and averaged according to the catch. The onset and the duration of the EMG burst, EMG burst area, and peak amplitude of the burst were calculated. Determination of the onset and offset of the burst of muscle activity was performed using visual inspection due to previously identified challenges with the use of established algorithms to determine onset in trunk muscle activity (1,11). However, the data processing software displayed a line that was 2 SD above baseline to assist in the visual inspection. Onset of EMG activity was determined as the earliest maintained rise of EMG, greater than 2 SD beyond the baseline of activity, and offset as the return to baseline. EMG activity during each period of interest is presented as a percentage of the entire burst activity of each muscle during the drive to be reflective of the relative level of activation of each muscle.
The range of motion and the EMG parameters were compared between spinal segments and muscles using ANOVA and Tukey post hoc analysis of truly significant difference. Statistical significance was defined as P < 0.05. Results are expressed as mean ± SD.
The total duration of the 2000-m race simulation was 7.0 min ± 11 s. The mean stroke rate at 250 m was 33 ± 2 strokes per minute, power was 313 ± 24 W, and the mean split (a measure of speed) was 1.44 ± 0.03 min per 500 m. During the collection period starting at 250 m, the drive represented 44.3% ± 1.2% of the complete stroke with a mean stroke length of 149.13 ± 7.82 cm. Stroke length and timing were consistent within each subject during the collection period.
Initiation of the drive.
The catch represents the position initiating the drive and, therefore, the initiation of extension forces within the muscles of the trunk and pelvis. In the first 10% of the stroke (postcatch), all subjects demonstrated pelvic and lumbar segments moving from positions of flexion into extension. The point of peak angular velocity reveals when the segment position changes at the greatest rate. The mean peak extension angular velocities of the upper two thoracic segments (T4-7 and T7-10) occurred in the first 10% of stroke (Table 2). The early peak extension angular velocity may arise due to the thoracic spine resistance. This resistance may be opposing the flexion moment created by resistance at the handle at initiation of the drive. Once this early position of extension is reached within the upper thoracic spine, there is little positional change within the thoracic spine during the period of peak force production (Table 3).
The lower lumbar segment reaches peak extension angular velocity early in the period of peak force production, at approximately 14% postcatch (Table 2). This demonstrates that the activity of the extensors of the spine quickly extend this segment so it is not maintained at its position of maximal flexion during peak force production. As displayed in Figure 4, the lower lumbar segment extends quickly as force production would be developing, along with a steady increase in extension angular velocity of the pelvis. During the period of peak force production, the pelvis is the segment exhibiting an increasing angular velocity.
Sequencing of muscle activity, as determined by the onset of the EMG bursts, demonstrated that the extensors of the pelvis (GM and BF) and the spine (ES lumbar, ES thoracic, and LD) were initiated within the first 6% of the stroke postcatch (Table 4). Onsets of EMG bursts of all pelvic and spinal extensors were not significantly different from each other.
Period of peak force production.
The clear grouping of activation among the extensor muscles is maintained as indicated by timing of peak EMG amplitude (Table 4). The BF, GM, ES lumbar, and ES thoracic peak first postcatch (BF at 18.3% ± 2.4%, GM at 17.1% ± 4.7%, ES lumbar at 18.5% ± 3.5%, and ES thoracic at 18.5% ± 4.8%), with no significant difference among the timing of peak amplitudes. Collectively, the prime extensor muscles of the pelvis, lumbar, and mid-lower thoracic spine demonstrate peak activity during this phase, occurring within approximately a 1.5% range of the stroke cycle.
The LD reaches its peak amplitude (23.1% ± 6.4%) significantly later in the stroke compared with the preceding extensors. The peak amplitude of LD falls near the end of the period of peak force production of the drive; perhaps as it is differentiated from the remaining extensors by contributing also to glenohumeral joint extension during force production of the stroke. This is reflective of the overall sequence of the stroke (legs followed by trunk, followed by arms in completing the stroke).
During the period of peak force production, there were minimal amounts of coactivation among the flexors and the extensors of the trunk (Table 5). The amount of activity of each flexor and extensor muscle, as demonstrated by percentage of total burst area, during this period was significantly different (P = 0.0001). Extensor activity predominates during the period of peak force production with over 60% of the total burst area occurring during this period (Table 5). The flexor group demonstrates less than 6% of total burst area during the period of peak force production. It is unlikely that EMG activity displayed within the flexor group represents a spinal stabilizing coactivation strategy in response to the extensor activity generated during the period of peak force production within the stroke.
Among the extensor group, the hamstrings (BF) and the gluteal (GM) muscles demonstrate similar percentage of burst area during the period of peak force production consistent with being the main muscles of hip extension during the leg drive (Table 5). It is interesting to note that the percentage of burst area of the ES lumbar (62.9%) is similar to that of the hip extensors (60.7% and 66.5%) during the period of peak force production (Table 5). These muscles have also similar burst durations. In contrast, ES thoracic and LD demonstrate more prolonged bursts and therefore less percentage of total activity within this period specifically (Figs. 2 and 5). This suggests a differentiation of the erector spinae roles during the stroke.
During the period of peak force production, the pelvis and all segments of the spine are extending. Table 3 displays the range of motion of each segment during this percentage of the stroke. The pelvis is the main segment extending the trunk, by 9.5° ± 3.4°, with very little extension of the spinal segments. The coordination of activity of the extensor muscles appears to achieve effective stabilization of the spine with the pelvis, facilitating a somewhat rigid lever on which to transfer forces. The lower lumbar segment (L3-S1) demonstrates the most extension of the spinal segments (2.9° ± 1.0°, P = 0.0001). Although this is a small excursion, it demonstrates an increase in mobility of the lower lumbar segment during the period of peak force production.
Period of coactivation.
The period of coactivation is defined as the EMG activity between the onset of the first flexor muscle and the end of the last extensor muscle burst for each subject. This period of coactivation did not include the period of peak force production in any subject. The mean period of coactivation was from 28.2% ± 2.2% postcatch to 36.8% ± 3.2%. The mean duration of the period of coactivation was 7.9% ± 2.9% of the stroke. This represents mid to late drive as the end of the drive occurs at 44.3% ± 1.2% of the stroke.
During the period of coactivation within the stroke, flexor muscles demonstrate increasing EMG activity; however, the peak of flexor activity occurs outside of this period, approximately 40% postcatch (Figs. 2 and 5). The extensor group demonstrates decreasing activity, without any extensor muscle demonstrating peak activity in this period. This period represents the transition into the finish of the stroke and the slowing of trunk extension and the influence of the gravitational force. The flexor group, TrA/IO, RA, and EO, shows consistency with no significant difference found between timing of onset (Table 4). This consistency is also seen in the timing of peak amplitude of the flexor group (40.4-41.3%; Table 4). The flexor group demonstrates similar levels of activity as ES thoracic and LD (approximately 20% of total burst area; Table 5).
Pelvic peak extension angular velocity occurred at the initiation of the period of coactivation (27.9% ± 1.4% postcatch; Table 2). This coincides with onset of the flexor muscle group (at approximately 29% postcatch) directly following the peak extension angular velocity of the pelvis. Peak extension angular velocities of T10-L1 and L1-L3 segments occurred at 29.2% ± 9.3% and 32.7% ± 2.3% of stroke, respectively (Table 2). Although both lumbar segments and lower thoracic segment continue to extend during this period, the thoracic segments of T4-T7 and T7-T10 began to flex. This resulted in flexion being experienced superiorly through the trunk as the pelvis is displaying a slowing of extension angular velocity. The flexion of the upper thoracic segments may be influenced by multiple factors such as the arm pull of the handle toward the body, the goal of the subject to maintain a somewhat level horizontal linear trajectory of the head and handle throughout the drive.
The main findings of the current study include the following: 1) the pelvic and the spinal extensor muscles demonstrated similar timing of activation, as did the flexor muscles; 2) during the period of peak force production, there was minimal coactivation of the flexors and the extensors; 3) the motion of the spine was minimal during peak force production, although there was significantly more movement at L3-S1 than at any other segment; and 4) low levels of coactivation of the trunk flexors and extensors existed in late drive during the transition between predominately extension to predominately flexion activity.
Extensor muscle recruitment in early drive.
The group of elite female rowers investigated in the current study closely resembles the group of rowers investigated by McGregor et al. (21). Within that study, it was reported that peak force occurred at 18.8% ± 1.8% postcatch, in mid-drive of the stroke. However, McGregor et al. (21) defined catch as the onset of force production at the handle, whereas catch was defined as the most forward position of the wrist in the current study. This will introduce discrepancies in the timing between the two studies because McGregor et al. (21) noted that the point at which the handle ceased moving toward the ergometer (analogous to the furthest point forward of the wrist and catch as defined by our study) occurred 2.7-3.0% of the stroke cycle before the onset of force production at the handle. The current study is not trying to match precisely one point in our data set with that of McGregor et al. (21); instead, we created a window of 10% of the rowing stroke centered around McGregor et al.'s (21) point of peak force production (18.8%) to characterize the period of peak force production. Although force was not measured in the current study, the timing of the peak of EMG activity of BF, GM, ES lumbar, and ES thoracic was 17-20.6% postcatch, suggesting that peak force production of the stroke occurred in this period of the drive.
Our findings suggest that coordination of the extensors of the spine and the pelvis, after the catch, may be an effective strategy to support the spine as forces increase with the initiation of the drive. There was remarkable consistency of the onset of EMG activity among extensors of the spine and the pelvis. The peak extension angular velocity of the lower lumbar (L3-S1) segment and the upper two thoracic segments (T4-T7, T7-T10) occurs relatively early in the drive before peak loading. The early extension of these segments may serve to prevent further flexion of the spine before the occurrence of peak forces.
Kinematics associated with peak extensor EMG activity highlight the effectiveness of the motor control strategies of the subjects in maintaining the spine position relatively fixed. The trunk experiences forces inferiorly and superiorly created by the leg drive and the resistance of the handle through the pull of the arms. If the spine were to flex in response to the transfer of forces, the result would be decreased force transmission to the handle (2). Greater extension range of motion of a spinal segment could create a greater risk of shear forces introduced to the segment while experiencing high compression forces. The lower lumbar spine appears to be the segment at risk as compared with the upper lumbar spine. The larger movement at the lower lumbar spine may be a consequence of its proximity to the pelvis, which demonstrates high extension angular velocity.
Recruitment of trunk flexors in late drive and recovery.
The pattern of trunk flexor EMG activity demonstrates the main role of the abdominals in slowing extension of the trunk as the center of mass is brought further into extension, the layback. During the late drive, gravitational forces increase and resistance from the handle decreases once the pull of the arms is completed. The trunk EMG activity will shift increasingly toward flexion as it must stop extension and reverse direction, flexing toward the catch.
The onset of the flexors of the trunk is associated with coactivation of flexor and extensor activity. Of the erector spinae group, ES thoracic remains active into the coactivation period, more so than ES lumbar. ES lumbar activity is similar to BF and GM, which also demonstrate less EMG burst area during this period than ES thoracic. This may be reflective of differentiation of the two erector spinae muscles based on mechanical advantage of the longer, increasingly multisegmental thoracic portions of this group to demonstrate improved effectiveness as global stabilizers as well as extensors of the spine (7). Latissimus dorsi (LD) also demonstrates a longer duration of burst. The peak activity of LD occurs during the period of peak force and remains active as flexor activity is initiated. The activity of LD may be differentiated from the remainder of the extensors due to its involvement with the extension of the humerus during the pull of the handle to the trunk, signifying the end of the stroke.
Peak burst amplitude of TrA/IO, RA, and EO occurs at approximately 40% postcatch, with the end of the drive occurring at approximately 44% when force on the handle is minimal. The peak activity of TrA/IO, RA, and EO falls outside the range of which the trunk muscles exhibit coactivation demonstrating that the peak force production of the flexors occurs at the end of the drive phase when there is limited activity of the extensors. The flexor group is able to flex the trunk forward over the extended legs without having to overcome forces of extension that would be created if the extensor group was strongly active during this action.
Coactivation of trunk musculature.
During the period of peak force production, there were minimal amounts of coactivation between the trunk flexor and the extensor muscle groups. During this phase of the stroke, the extensors of the pelvis and the spine predominate, with peak activity of extensor bursts occurring with very low levels of flexor EMG. Currently there are limited data to develop empirical measures of spinal stability and trunk muscle coactivation within the literature. Therefore, it remains unclear whether low levels of abdominal muscle activity, as seen in this study, contribute to stabilization of the spine.
Lavender et al. (14) found similar levels of low trunk flexor and extensor coactivation during a symmetrical pull activity performed leaning back in a semisquat position. Of studies investigating coactivation of trunk flexors and extensors during trunk extension tasks, this investigation represents the task most closely resembling the activity of rowing. However, levels of coactivation displayed by subjects within our study are low. This may be reflective of the subjects and the activity investigated. The current investigation involved elite athletes performing a skilled, trained activity. A higher level of coactivation may be seen in novice rowers as compared with elite rowers as they may demonstrate less refined ability to maintain balance within these conditions (5,20). Higher levels of coactivation of the flexor muscles during peak force production would negatively impact the peak extension force created by the extensor muscle group and hence the power of the stroke. If the antagonist activity were heightened, the trunk extensor muscle activity would need to increase to maintain the overall power of the stroke. Increased coactivation of the flexors and the extensors would also result in increased compressive spinal forces and may therefore be injurious (9).
Coactivation has been shown to increase in tasks introducing asymmetrical or unstable loads (14,28,30). There may be more coactivation in sweep rowing than that seen on the ergometer because sweep rowing incorporates an element of spinal rotation and asymmetrical loading. It is also possible that the unstable base of the rowing shell in water versus the stable base of the ergometer may increase the level of trunk coactivation to maintain balance, a condition further challenged by rough water. Reeves et al. (24) demonstrated that increased coactivation of the trunk muscles while seated on an unstable surface degraded postural control, as seen by an increase in center of pressure velocity and excursion. Therefore, higher coactivation may be seen on the water or with novice rowers than with the ergometer in elite athletes in the current study.
Implications for rowing.
This investigation into the kinematics and the EMG of the trunk musculature revealed that muscle recruitment is quite precise in elite female rowers. This is associated with limited motion of the trunk in peak force production, thereby creating a stable lever to transfer the forces generated by the legs and a stable base on which the upper extremities are able to generate and transfer these forces to the handle. Timing of muscle activation has been demonstrated to be impacted by skill level or stage of motor learning (5), presence of pain within the segment of interest (30), kinematic position of the segment (18), direction of load (14), unanticipated external perturbations (28), and fatigue (3). Poor timing of muscle recruitment may result in technique that may compromise how the spine is loaded, as suggested by McGregor et al. (21) concerning novice rowers who demonstrated altered lumbopelvic rhythm. Further research is needed to explore the kinematics and the muscle recruitment strategies of the trunk under conditions of injury, fatigue, and different levels of rowing experience (novice rowers). Investigation with bilateral EMG recordings would also enable the identification of any motor control strategies unique to sweep or sculling rowing technique.
The authors wish to thank Dr. Trevor Birmingham, codirector of the Wolf Orthopaedic Biomechanics Laboratory, for his assistance with data collection and Angela Kedgley for her assistance with data analysis. Dr. Volker Nolte, head coach of the Western Men's Rowing team, provided helpful commentary on a draft of this article. This work was funded, in part, through a grant from the Natural Sciences and Engineering Research Council of Canada awarded to SJ Garland. The authors report no conflict of interest or endorsement by ACSM.
1. Allison GT. Trunk muscle onset detection technique for EMG signals with ECG artefact. J Electromyogr Kinesiol
2. Baudouin A, Hawkins D. A biomechanical review of factors affecting rowing performance. Br J Sports Med
3. Caldwell JS, McNair PJ, Williams M. The effects of repetitive motion on lumbar flexion and erector spinae muscle activity in rowers. Clin Biomech
4. Callaghan JP, McGill SM. Low back joint loading and kinematics during standing and unsupported sitting. Ergonomics
5. Chapman AR, Vicenzino B, Blanch P, Hodges PW. Patterns of leg muscle recruitment vary between novice and highly trained cyclists. J Electromyogr Kinesiol
6. Cholewicki J, Panjabi MM, Khachatryan A. Stabilizing function of trunk flexor-extensor muscles around a neutral spine posture. Spine
7. Crisco JJ, Panjabi MM. The intersegmental and multisegmental muscles of the lumbar spine. A biomechanical model comparing lateral stabilizing potential. Spine
8. Granata KP, Lee PE, Franklin TC. Co-contraction recruitment and spinal load during isometric trunk flexion and extension. Clin Biomech
9. Granata KP, Marras WS. Cost-benefit of muscle cocontraction in protecting against spinal instability. Spine
10. Hickey GJ, Fricker PA, McDonald WA. Injuries to elite rowers over a 10-yr period. Med Sci Sports Exerc
11. Hodges PW, Bui BH. A comparison of computer-based methods for the determination of onset of muscle contraction using electromyography. Electroencephalogr Clin Neurophysiol
12. Holt PJ, Bull AM, Cashman PM, McGregor AH. Kinematics of spinal motion during prolonged rowing. Int J Sports Med
13. Jenkyn TR, Birmingham T, Jones I, Wervey R, Irby S, Kaufman KR. A multi-clinic, multi-user evaluation of marker resolution and error in gait analysis laboratories. In: Proceedings of the Eighth Annual Meeting of the Gait and Clinical Movement Analysis Society
; Wilmington (DE); 2003; p. 223-24.
14. Lavender SA, Chen SH, Li YC, Andersson GB. Trunk muscle use during pulling tasks: effects of a lifting belt and footing conditions. Hum Factors
15. Leinonen VM, Kankaanpaa M, Airaksinen O, Hanninen O. Back and hip extensor activities during trunk flexion/extension: effects of low back pain and rehabilitation. Arch Phys Med Rehabil
16. Marras WS, Ferguson SA, Burr D, Davis KG, Gupta P. Functional impairment as a predictor of spine loading. Spine
17. Marshall P, Murphy B. The validity and reliability of surface EMG to assess the neuromuscular response of the abdominal muscles to rapid limb movement. J Electromyogr Kinesiol
18. McGill SM, Hughson RL, Parks K. Changes in lumbar lordosis modify the role of the extensor muscles. Clin Biomech
19. McGregor A, Anderton L, Gedroyc W. The assessment of intersegmental motion and pelvic tilt in elite oarsmen. Med Sci Sports Exerc
20. McGregor AH, Bull AM, Byng-Maddick R. A comparison of rowing technique at different stroke rates: a description of sequencing, force production and kinematics. Int J Sports Med
21. McGregor AH, Patankar ZS, Bull AM. Spinal kinematics in elite oarswomen during a routine physiological "step test." Med Sci Sports Exerc
22. Morris FL, Smith RM, Payne WR, Galloway MA, Wark JD. Compressive and shear force generated in the lumbar spine of female rowers. Int J Sports Med
23. Panjabi MM. Clinical spinal instability and low back pain. J Electromyogr Kinesiol
24. Reeves NP, Everding VQ, Cholewicki J, Morrisette DC. The effects of trunk stiffness on postural control during unstable seated balance. Exp Brain Res
25. Rodriguez RJ, Rogriguez RP, Cook SD, Sandborn PM. Electromyographic analysis of rowing stroke biomechanics. J Sports Med Phys Fitness
26. Soler T, Calderon C. The prevalence of spondylolysis in the Spanish elite athlete. Am J Sports Med
27. Teitz CC, O'Kane JW, Lind BK. Back pain in former intercollegiate rowers. A long-term follow-up study. Am J Sports Med
28. Thomas JS, Lavender SA, Corcos DM, Andersson GB. Trunk kinematics and trunk muscle activity during a rapidly applied load. J Electromyogr Kinesiol
29. Urquhart DM, Hodges PW, Story IH. Postural activity of the abdominal muscles varies between regions of these muscles and between body positions. Gait Posture
30. van Dieen JH, Cholewicki J, Radebold A. Trunk muscle recruitment patterns in patients with low back pain enhance the stability of the lumbar spine. Spine
31. van Dieen JH, Kingma I, van der Bug JCE. Evidence for a role of antagonistic cocontraction in controlling trunk stiffness during lifting. J Biomech