Dynamics of the Shoulder and Elbow Joints of the Throwing Arm During a Baseball Pitch
Michael Feltner and Jesús Dapena
(Indiana University)
Fastball pitches of eight intercollegiate varsity baseball pitchers were filmed using the direct linear transformation (DLT) method of three-dimensional cinematography. Coordinate data were obtained, and the resultant joint forces and torques at the shoulder and elbow joints were calculated. Various kinematic parameters were also calculated to help describe the motions of the shoulder and elbow joints throughout the pitch. At the instant of stride foot contact, a horizontal adduction torque was present at the shoulder joint, and the shoulder was externally rotating. After the onset of the horizontal adduction torque, abduction and internal rotation torques were also present at the shoulder joint and a varus torque was present at the elbow joint. After the instant of maximum external rotation (30 ms prior to ball release), the upper arm started to internally rotate, but it was still in a position of external rotation at the instant of release. This paper discusses the roles of the torques in producing the observed motions of the throwing arm.
@@@Qualitative descriptions of the kinematics of the throwing arm during the baseball pitch (Figure 1) have appeared in most kinesiology textbooks. However, quantitative kinematic data have been obtained and reported by a relatively small number of investigators (Atwater, 1970, 1980; Elliott, Grove, Gibson, & Thurston, 1986; Feltner, 1984; Gainor, Piotrowski, Puhl, Allen, & Hagen, 1980; Pappas, Zawacki, & Sullivan, 1985; Sakaris, 1978; Tarbell, 1971). Studies of the kinetic parameters of the throwing arm have been even more limited (Feltner, 1984; Gainor et al., 1980; Horn, 1984).
@@@In biomechanical kinetic analysis, the multiple forces exerted by a body segment on an adjacent segment can be substituted by a resultant force exerted through the joint center ("joint force") and a resultant couple ("joint torque"). The joint torque reflects the net muscular activity at the joint, except near the limits of the range of motion, where ligaments, bones, and other passive structures may also contribute to it (Andrews, 1982). Thus, analysis of the forces and torques at the joints permits a better understanding of the causal factors responsible for the production of the observed motions.
@@@In the study by Gainor et al. (1980) the subjects did not attempt to maximize ball velocity at the instant of release. Consequently, the torque reported by these authors (internal/external rotation at the shoulder joint) should be expected to differ from that of a maximum effort pitch.
@@@Horn (1984) calculated the joint torques and forces at the shoulder, elbow, and wrist joints. The single subject used in this study also did not make a maximum effort during the pitch. According to Horn's computations, the velocity of the ball/hand segment reached a maximum value of only 21.6 m/s, and it subsequently dropped to 4.5 m/s before release. These values were considerably smaller than the average of the ball velocities at release (35.1 m/s - 39.0 m/s) reported by other investigators (Atwater, 1970; Elliott et al., 1986; Feltner, 1984; Sakaris, 1978; Selin, 1959; Tarbell, 1971). Other important shortcomings of Horn's investigation included the procedures used for coordinate data optimization, cubic spline smoothing, and nine-point chord average differentiation. These procedures contributed to large fluctuations in the computed joint torque and force data, thus casting doubts on the accuracy of the values obtained for these parameters.
@@@Feltner (1984) conducted a three-dimensional (3D) study of the torques and forces exerted at the shoulder joint during pitches made in actual games. Its main shortcoming was that the uniforms hampered the accurate identification of the positions of the shoulder and elbow joints, thus increasing the amount of "noise" in the raw coordinate data and raising doubts about the accuracy of the calculated torques and forces.
@@@The present investigation was essentially a modification of the previous study (Feltner, 1984). The subjects threw without shirts (to eliminate the problems caused by the concealment of body landmarks), and the scope of the project was expanded to include the elbow joint. The main goals of the present study were to obtain quantitative information on the torques at the shoulder and elbow joints of the throwing arm during full-effort baseball pitches, and to initiate the interpretation of the cause-effect relationships between the muscle actions and the motions of the arm.
Methodology
Filming Procedures
Eight male intercollegiate varsity baseball pitchers (M height 1.86 m; M weight 84 kg) were filmed during a team practice session while they threw from the bullpen pitching mound to a protective screen behind home plate . The subjects wore no shirts and, to aid in digitizing the elbow and wrist joints, a black band (approximately 2-4 cm wide) was painted around each of these joints using a waterbased paint. Two fastball pitches of each subject were filmed. The players were encouraged to make a maximum effort while executing the pitches, and informed consent was obtained from each subject prior to filming.
@@@The subjects were filmed using the DLT method of 3D cinematography (Abdel-Aziz & Karara, 1971; Walton, 1981). Two battery-powered LOCAM cameras were used to record the trials on 16mm Fujicolor 500 ASA film. The cameras viewed the subjects from the rear and from the throwing arm side, and they were set at nominal frame rates of 200 fps.
Computation of 3D Coordinates
A previous study (Feltner, 1984) showed little variability among the fastball pitches of any given player. This implied that any trial of a given subject should be expected to be representative of most fastball pitches thrown by that subject. Consequently , a single trial of each player was arbitrarily selected for analysis in the present study.
@@@A Vanguard projector head was used to project the film images onto the surface of a Houston Instrument COMPLOT digitizer connected to a CDC Cyber mainframe computer through a terminal. The suprasternale, the right hip, shoulder, elbow, wrist, and third knuckle , the left hip and shoulder, and the center of the ball were digitized in each projected film image (the suprasternale was defined as a point on the longitudinal axis of the trunk and at the level of the suprasternal notch).
@@@The correspondence between the frames of the two cameras for each trial was calculated following a procedure described by Dapena (1984). Due to the absence of mechanical synchronization of the two cameras, the instants of frame exposure of one film did not correspond exactly with the instants of frame exposure of the other film. To solve this problem, quintic spline functions (developed by L. Jennings and reported by Vaughan, 1980) were fitted, with no smoothing , to the film coordinate-time data obtained from each camera. Subsequently, the functions were used to compute interpolated values for times intermediate between the frames and which did correspond in the two cameras. For convenience, and to facilitate comparisons among trials, the interpolated values were calculated for instants ("output frames") separated by intervals of exactly 0.005 s, and the time t = 10.000 s was arbitrarily assigned to the instant of ball release. Analysis of the trials began at the instant when the ball left the glove, and ended at an instant approximately 0.05 s after ball release.
@@@A CDC Cyber computer used the digitized film data to calculate the coordinates of the ball and of the body landmarks for the time of each output frame in terms of reference frame R1, a right-handed orthogonal reference frame. R1 had its origin at the midpoint of the rear edge of the pitching rubber, and its axes were defined by vectors X1, Y1, and Z1(Figure 2). Z1 was vertical; X1 was horizontal and directed along the rear edge of the pitching rubber toward the third base side of the playing field; Y1 was perpendicular to the other two, and pointed toward home plate.
Data Smoothing and Kinematics
Quintic spline functions were used to smooth the time-dependent coordinates of each landmark. All subsequent calculations were performed with these smoothed landmark data.
@@@To aid in the calculation of the abduction/adduction , horizontal abduction/adduction, and internal/external rotation angles at the shoulder joint , a noninertial reference frame R2 , attached to the trunk, was defined (Figure 3). The values of its direction vectors (X2, Y2 , Z2) were calculated in terms of reference frame R1 for each output frame. X2 pointed from the suprasternale to the throwing shoulder; Y2 pointed anteriorly from the suprasternale, and was defined by the cross product of X2 and a vector pointing from the suprasternale to the mid-hip point; Z2 was defined as the cross product of X2 and Y2. It was also necessary to define two vectors, V1 and V2, that coincided, respectively , with the longitudinal axes of the upper arm and of the forearm (Figure 3).
@@@Joint Angles. The abduction/adduction angle at the shoulder joint i΅) was calculated as the angle formed by vectors V3 and X2, where V3 was the projection of V, on the plane determined by X2 and Z2 (Figure 4a). Standard analytic geometry procedures were used for this and all subsequent vector operations. The horizontal abduction/adduction angle at the shoulder joint (ΐ) was calculated as the angle formed by vectors V4 and X2, where V4 was the projection of V1 on the plane determined by X2 and Y2 (Figure 4b). The internal/external rotation angle at the shoulder joint (Α) was calculated as the angle formed by vectors V5 and V6, where V5 and V6 were the projections of vectors Z2 and V2, respectively, on the plane perpendicular to the longitudinal axis of the upper arm (Figure 4c) . The angle of flexion/extension at the elbow joint i¦) was calculated as the angle formed by vector V2 and the reverse of vector V1 (Figure 4d). Sketches defining the reference values and the signs for the four angles are shown in Figure 5.
@@@A special type of plot was used to facilitate the visualization of the motion of the upper arm relative to the trunk throughout the pitch (see Figure 11). For this plot, a theoretical hemisphere was defined for each output frame (Figure 6a). The sphere was centered at the shoulder joint, and the pole was aligned with vector X2. The point of intersection of vector V1 with the hemisphere defined the orientation of the longitudinal axis of the upper arm with respect to the right upper trunk, assuming a right-handed pitcher (see Figures 6a and b). The distal cross section of the upper arm was then represented by an ellipse centered at the intersection of the longitudinal axis with the sphere (Figure 6c). The plane determined by the longitudinal axes of the upper arm and of the forearm defined the lateral half of the arm (shaded in Figures 6d and e). In the polar representation, the projection of the longitudinal axis of the forearm onto the surface of the ellipse defined the lateral half of the ellipse (shaded in Figure 6f).
@@@Joint Angular Velocities. @Quintic spline functions (with no smoothing) were fitted to the time-dependent angular displacement values. The first derivative of the functions provided the angular velocity values.
Kinetics
Joint Forces and Torques. @The throwing arm was modeled as a four-link kinetic chain composed of the ball, hand, forearm, and upper arm (the ball was excluded from the kinetic chain after release) . The mass, as a percentage of total body mass, and the location of the center of mass (CM) of each body segment were taken from Dempster's cadaver data (1955). Moment-of-inertia values were taken from Whitsett (1963), and they were personalized for each subject using a procedure described by Dapena (1978).
@@@The ball was assumed to be subjected to two forces: weight, acting at its CM, and a force made by the hand on the ball, passing through the CM of the ball. The hand segment was assumed to be subjected to a proximal couple and three forces: weight, acting at its CM, a distal force (the reaction to the force made by the hand on the ball), and a force exerted by the forearm at the proximal joint. The forearm and upper arm segments were each assumed to be subjected to the force of their own weight, acting at the CM, plus a torque and a force at both the proximal and distal joints.
@@@The instantaneous CM location and the local angular momentum of each segment about its own CM were computed using a modification of the procedures detailed by Dapena (1978). The net force exerted on each segment was calculated from the second derivative of the location value of its CM. The net torque about the segment CM was computed as the first derivative of its local angular momentum. A procedure described by Andrews (1974, 1982) was then used to calculate the proximal joint force and torque exerted on each link of the kinetic chain, beginning with the ball.
@@@The joint force and torque values obtained were computed in terms of reference frame R1. To provide anatomically relevant meaning to these values, two noninertial reference frames, R3 and R4 , were defined at the shoulder and elbow joints , respectively .
@@@Reference Frame R3. @The direction vectors of this reference frame (X3 , Y3 , Z3) were expressed for the instant of each output frame in terms of reference frame R1 (Figure 7). The X3 vector pointed from the elbow joint to the shoulder joint; Y3 was the cross product of X3 and a vector pointing from the suprasternale to the mid-hip point; Z3 was the cross product of X3 and Y3.
@@@The net force and torque vectors at the shoulder joint were projected onto the three direction vectors of reference frame R3, for the instant of each output frame. The X3 torque represented internal/external rotation; the Y3 torque, abduction/adduction; the Z3 torque, horizontal abduction/adduction . The positive X3 force was directed proximally along the longitudinal axis of the upper arm, while the Y3 and Z3 forces were shear forces.
@@@Reference Frame R4. @The direction vectors of this reference frame (X4, Y4 , Z4) were expressed for the instant of each output frame in terms of reference frame R1 (Figure 8). Direction vector Y4 pointed from the wrist joint to the elbow joint; X4 was the cross product of Y4 with direction vector X3 of reference frame R3; Z4 was the cross product of X4 and Y4.
@@@For each output frame, the net force and torque vectors at the elbow joint were projected onto the three direction vectors of reference frame R4. The X4 torque represented flexion/extension; the Y4 torque, pronation/supination; the Z4 torque, valgus/varus rotation. The positive Y4 force was directed proximally along the longitudinal axis of the forearm, while the X4 and Z4 forces were shear forces.
Kinematic Data
All subjects showed very similar kinematic and kinetic patterns. The plots presented in the remaining figures correspond to the subject with the fastest ball velocity at the instant of release, but they are representative of the plots of all subjects. The magnitudes of the joint torques were negligible up to the instant of stride foot landing (Figure 11). Therefore, detailed analysis was limited to the period of the pitch between an instant approximately 0.075 s prior to stride foot contact and an instant approximately 0.050 s after ball release.
@@@Figure 9 shows the angular displacements and angular velocities at the shoulder joint. Figure 10 shows the elbow flexion/extension angle and angular velocity. The polar plot in Figure 11 shows the motion of the upper arm relative to the shoulder, from an instant near the time when the ball left the glove (t = 9.53 s) until t = 10.05 s, at 0.02 s intervals (see Figure 6). Figure 12 shows two views of the motion of the throwing arm from t = 9.96 s until t = l0.02 s.
@@@The times of the most important events of the pitch (stride foot contact, maximum external rotation, and ball release), and the values of the angular displacements at the shoulder and elbow joints at the instants of these events , are presented in Table 1.
@@@These results permitted a detailed description of the kinematics of the pitch. All data in the following description are the mean values for all subjects, unless stated otherwise. At an instant 0.075 s prior to stride foot contact, the upper arm was in a position of adduction (28), horizontal abduction (24), and internal rotation (80), with the elbow at an angle of 144. From this instant until the instant of stride foot contact, the upper arm was slightly abducted, horizontally adducted, and externally rotated, and the elbow joint was flexed. At stride foot contact (t = 9.817 s), the shoulder was still in a position of adduction (14), horizontal abduction (18), and internal rotation (44), with the elbow at an angle of 117. From this instant, the upper arm continued to abduct, horizontally adduct, and externally rotate until the instant of maximum external rotation. The elbow joint reached its maximum flexion (89}8 ) approximately halfway between the instants of stride foot contact and maximum external rotation, and then it began to extend.
@@@At the instant of maximum external rotation (t = 9.968 s), the upper arm was in an abducted (12), horizontally adducted (11), and externally rotated (80) position, and the elbow joint had been extended to 106. The upper arm was subsequently internally rotated and slightly adducted and horizontally abducted until the instant of ball release. The elbow joint was rapidly extended during this time period, and it reached its maximum angular velocity of extension (2200/s}400/s) shortly before release (t = 9.985s } 0.003s). The peak angular velocity of internal rotation (6100/s}1700/s) coincided approximately with the instant of ball release (t = 10.004 s } 0.005 s). At the instant of ball release, the angles of abduction and horizontal adduction of the upper arm were both positive, but very small (2 ). The upper arm, while rapidly internally rotating at the instant of release, was still in an externally rotated position at this instant (23of external rotation) . The elbow joint was near its maximum extension value at release, but still 20short of the fully extended position. The average speed of the baseball at release was 33.5 m/s.
@@@After ball release, the upper arm continued to undergo internal rotation, and again began to abduct and horizontally adduct. The neutral position of internal/external rotation (0) was reached just after the instant of ball release (t = 10.005 s }0.005 s). The elbow joint flexed again after release.
@@@Figure13 shows the relationship between the angles of internal/external rotation at the shoulder and of flexion/extension at the elbow . The plot shows that there was a combination of angular velocities of elbow flexion and shoulder external rotation until the instant of maximum elbow flexion. Then the elbow started to extend, while the upper arm continued rotating externally. After the instant of maximum external rotation, the upper arm internally rotated while the elbow joint kept extending. The plot clearly demonstrates that the motion of elbow extension preceded the motion of' internal rotation at the shoulder joint.
Kinetic Data
Figures14 and 15 show the torques and forces at the shoulder joint, and Figures16 and 17 show the torques and forces at the elbow joint.
@@@In all subjects, there was a horizontal adduction torque between the instants of stride foot contact and ball release (peak value, 110 Nm } 20 Nm) . The horizontal adduction torque preceded the abduction and internal rotation torques, which began at approximately t = 9.90 s (Figure 14). The peak abduction (70 Nm } 20 Nm) and internal rotation (90 Nm } 20 Nm) torques were reached just prior to the instant of maximum external rotation, at t = 9.952 s (} 0.013 s) and t = 9.954 s (} 0.006 s), respectively. The horizontal adduction torque continued briefly beyond the instant of release, but the abduction and internal rotation torques were markedly reduced by the time of release.
@@@The X3 shoulder joint force was always positive between the instants of stride foot contact and ball release (Figure 15). The magnitude of this force increased rapidly between the instants of maximum external rotation and ball release in all subjects , reaching its peak value (860 N } 120 N) near the instant of ball release (t = 10.005 s } 0.005 s). The Y3 force was generally negative between the instants of stride foot contact and release, indicating a forward force. The Z3 force was near zero until approximately t = 9.90 s, when it became positive. It remained positive until an instant between the times of maximum external rotation and ball release, when it reversed its direction and became negative. @@@
@@@@The pronation/supination torque at the elbow joint was very small (< 10 Nm) throughout the pitch (Figure 16). The flexion/extension torque was negligible until approximately halfway between the instants of stride foot contact and maximum external rotation. It then became an extension torque (peak value, 20 Nm } 10 Nm), and was generally present until near the instant of maximum external rotation. The time at which the maximum extension torque occurred was rather variable (t = 9.933 S } 0.024 s). After the instant of maximum external rotation, the flexion/extension torque approached zero or became a flexion torque. The valgus/varus torque was small until approximately t = 9.90s, when it became a varus torque and increased rapidly in magnitude. The peak varus torque value (100 Nm } 20 Nm) was reached at approximately the same time by all subjects (t = 9.953 s } 0.012 s). Its magnitude then decreased throughout the remainder of the pitch.
@@@All elbow joint forces were comparatively small (< 170 N) until approximately t = 9.90 s (Figure17). After this instant, the Y4 force became positive and increased rapidly in magnitude. Its peak value (830 N } 80 N) occurred at the instant of ball release in 6 of the 8 subjects, and within }0.005s of release in the remaining subjects. The X4 and Z4 forces were generally positive between the instants of stride foot contact and release. The X4 force reached its peak value (320 N } 60 N) prior to the instant of maximum external rotation (t = 9.952s } 0.006s), while the Z4 force was small throughout the pitch.
The average ball speed at release was slightly lower in the present investigation (33.5 m/s) than in some of the previous studies (35.1 m/s - 39.0 m/s in the analyses by Atwater, 1970; Elliott et al., 1986; Feltner, 1984; Sakaris, 1978; Selin, 1959; Tarbell, 197 1), but it was large enough to warrant the classification of the trials as full-effort pitches.
@@@The kinematic values obtained in the present study were generally consistent with the results of previous authors, although it was often difficult to make comparisons, due to limitations inherent in the two-dimensional approach followed by some researchers, to the absence of data smoothing prior to the computation of derivative values in most studies, and to differences in the definitions of angles.
@@@In two of the previous studies in which joint torques were computed (Horn, 1984; Gainor et al., 1980), the subjects did not attempt to maximize ball velocity, and the values of some of the kinematic parameters confirmed that the patterns of motion were quite different from those of normal full-effort pitches. Consequently, no comparisons were attempted between the joint torques reported in these papers and those of the present investigation.
@@@The joint torque patterns obtained in the present investigation were very similar to those obtained in the preliminary study carried out at our laboratory (Feltner, 1984), but the fluctuations in the plots were greatly reduced, probably because of improvements in landmark digitization due to the fact that the subjects of the present study did not wear shirts. To determine to what extent the remaining fluctuations were real or merely artifacts due to "noise" in the digitized data, one trial was redigitized by a separate experienced operator. The force and torque values obtained from the data of the two operators showed the same general pattems, but the magnitudes of the individual fluctuations, and their temporal locations, varied. This indicated that although the general patterns of the force and torque functions could be trusted, the instantaneous values of parameters that exhibited large fluctuations should be considered with caution. @@@The kinematic data and general patterns of the joint torques permitted a tentative interpretation of the roles of muscle groups in the generation of the motions of the throwing arm. The remainder of the discussion will explain this interpretation.
@@@Joint forces and torques of rather small magnitude are used during the early stages of the pitch to bring the upper arm into a position of approximately 15of adduction, 10 of horizontal abduction, and 0of internal/external rotation, with the elbow joint flexed at approximately 90, just after the instant of stride foot contact (see Figures 9a and 10a). From this position, the elbow is moved forward by a horizontal adduction torque at the shoulder (Figure 14). This torque agrees with the EMG activity of the pectoralis major reported by Jobe, Moynes, Tibone, and Perry (1984) for the period of time between stride foot contact and ball release. Somewhat later, the pitcher starts an abduction torque that lifts the elbow (Figure 14). At approximately the same time as these upper arm rotations take place, the trunk performs a counterclockwise rotation about its longitudinal axis and a lateral leaning motion toward the left (viewed from the rear). These motions of the trunk, produced by the actions of the leg and trunk muscles, hide much of the motion of the upper arm relative to the ground in the plots of anatomical abduction and horizontal adduction at the shoulder joint shown in Figure 9a. The amplitude of the motions of the upper arm relative to the ground are demonstrated more vividly by the stick figure sequence at the top of Figure 9 .
@@@While the upper arm is subjected to motions of horizontal adduction and of abduction, it also experiences a motion of external rotation that eventually takes it to an extremely externally rotated position (80) (see Figures 9 and 11). Surprisingly, this external rotation takes place against the action of an internal rotation torque exerted at the shoulder joint (Figure 14). This torque agrees with the EMG activities of the latissimus dorsi and pectoralis major reported by Jobe et al. (1984) for the period of time between stride foot contact and ball release.
@@@The causes for the external rotation of the upper arm could be explained by the following mechanism. The muscles in the anterior part of the shoulder joint exert a horizontal adduction torque (see Figure 14, and THA in Figure 18a). This leads to a counterclockwise angular acceleration of the arm in an overhead view, and consequently to a forward linear acceleration of the CM of the arm (G in Figure 18a) . The linear acceleration of G requires a forward force at the shoulder joint (negative FY3 value in Figure 1 5 , and vector FY3 in Figure 18a) , and this force exerts in turn a counterclockwise torque with respect to the CM of the flexed arm in a view along the longitudinal axis of the upper arm (Figure l8b). Through this indirect mechanism, the muscles that exert horizontal adduction torque also encourage external rotation. In normal baseball pitching, the external rotation effects of the horizontal adduction torque evidently outweigh the effects of the internal rotation torque exerted by the shoulder musculature (shown in Figure 14, and designated TlR in Figure 18b), and this leads to an increase in the angular velocity of external rotation immediately after stride foot contact (Figure 9b).
@@@As the arm externally rotates, the moment arm of the external rotation torque exerted by FY3 about G becomes smaller (Figure 18d). To maintain the emphasis on external rotation, the shoulder abduction musculature then begins to contract (Figure 14). Following a mechanism similar to that used to produce FY3, the torque exerted by these muscles (TA in Figure 18c) evokes an upward force at the shoulder joint (FZ3 in Figure 15 and in Figures 18 c and d). When the flexed arm is externally rotated past the 0 angle, the resultant of FY3 and FZ3 (FR in Figure 18d) is more effective for the production of external rotation than FY3 alone.
@@@This mechanism for the production of external rotation was already described in the preliminary report (Feltner, 1984) , and it agrees with the hypothesis proposed by Kreighbaum and Barthels ( 1985) that the external rotation is produced by the inertial lag of the forearm and hand as the more proximal segments rotate forward.
@@@While the arm externally rotates due to the combined effects of the abduction and horizontal adduction musculature, part of the internal rotation torque that resists against it is transmitted through the humerus to the elbow joint, where it is reflected in a varus torque exerted by the upper arm on the forearm (Figure 16). This large varus torque is probably exerted mainly through tensile forces on the ulnar collateral ligament and on the articular capsule of the medial side of the joint, and through compressive forces on the bones of the lateral side of the joint, although part of the torque could also be produced by the muscles originating on the medial epicondyle of the humerus as they perform their primary functions of wrist and finger flexion and radioulnar pronation . This phenomenon, termed medial tension overload or valgus extension overload, and the injuries associated with it, have been discussed extensively in the literature (Andrews & Wilson, 1985; Barnes & Tullos, 1978; DeHaven & Evarts, 1973; Indelicato et al., 1979; King. Brelsford, & Tullos, 1969; Wilson, Andrews, Blackburn, & McCluskey, 1983).
@@@Approximately halfway before maximum external rotation is reached, the elbow begins to extend (Figures 10 and 13), and within less than 0.10 s it reaches a very large speed of extension (2200/s } 400/s[see Figure 10b). Jobe et al. (1984) reported modest biceps EMG activity after the instant of stride foot contact, followed by triceps EMG activity. These data confirmed the EMG patterns of the control trial presented by Dobbins (reported in Roberts , 1971), and they agreed with the results of the present study, which showed negligible torque or sometimes a slight flexion torque at the elbow joint immediately after stride foot contact, followed by an extension torque (Figure 16) . However, the value of the extension torque is quite small (peak value, 20 Nm), and this strongly suggests that the extension of the elbow is not due primarily to the action of the triceps but to the resultant joint force exerted by the upper arm on the forearm at the elbow . For instance, a force pointing from the elbow joint toward the shoulder joint would lead to elbow extension.
@@@Such a force could be associated with a centripetal acceleration of the elbow joint as the upper arm performs its abduction and horizontal adduction rotations about the shoulder joint, or it could be produced by a linear acceleration of the trunk toward the left. In any case, the small magnitude of the extension torque at the elbow joint suggests that the acceleration of the ball may ultimately be due primarily to the actions of muscles other than the elbow extensors. This would be consistent with the findings of Dobbins (reported by Roberts, 1971). Dobbins used a differential nerve block to paralyze triceps activity. After six practice trials, the subject was able to throw the ball at over 80 % of the speed attained prior to the paralyzation of the triceps.
@@@Eventually the motion of external rotation at the shoulder stops (Figures 9 and 12). This could be due to increased internal rotation torque at the shoulder joint (due to active and passive torques exerted by the stretched subscapularis , latissimus dorsi, pectoralis major, and teres major muscles, and to passive torques exerted by the joint capsule near the limits of the joint range of motion) , and/or to a decrease in the external rotation torque exerted by FR about the CM of the arm as the elbow joint extends.
@@@After stopping the external rotation, the pitcher could conceivably keep the arm in its position of maximum external rotation and simply increase the speed of elbow extension to give the ball a large velocity at release. However, the actual pattern of motion is somewhat different from this (Figures 12 and 13): the arm undergoes a rapid motion of internal rotation immediately after reaching its position of maximum external rotation (Figure 9) , and the elbow stops short of full extension (Figure 10a) . The motion of internal rotation may be unavoidable, due to the stretch of the internal rotation musculature and the inability of the abduction and horizontal adduction torques to elicit much external rotation when the arm is nearly straight. Still, regardless of whether the motion of internal rotation is voluntary or involuntary, the combination of this motion with a slowing down of the elbow extension may protect the elbow against injury.
@@@If the elbow joint reached a maximum speed of extension just prior to the instant of full extension, this would lead to a large ball speed, but it would also risk injury to the posterior part of the elbow joint when the elbow locked straight immediately afterward (Figure 19a). The risk of injury would force the pitcher to limit the speed of the ball prior to release. The pattern actually used by the pitchers may be a good solution to this problem: by stopping the extension of the elbow before the attainment of full extension, and combining this with a rapid internal rotation at the shoulder joint (Figures 12 and 19b) , injury of the posterior part of the elbow joint can be avoided while permitting the hand to move forward beyond the position of the elbow without slowing down. The stopping of the elbow extension is probably achieved through the elbow flexor torque exerted shortly before ball release (Figure 16) This torque agrees with the biceps EMG activity found in the control trial studied by Dobbins (reported by Roberts, 1971).
@@ The above interpretation of the cause-effect mechanisms that link the joint forces and torques with the kinematics of the throwing arm during baseball pitching should be considered only a tentative explanation. For a more definitive interpretation, it will be necessary to examine the motions of the whole body of the pitcher and the interrelationships among joint forces and torques and linear and angular positions, velocities, and accelerations. This will probably require the development of a special model, perhaps similar to the one used by Putnam (1980) to study the cause-effect mechanisms of kicking motions.