body roll and handpath in freestyle · pdf file · 2007-08-29body roll and handpath...

JOURNAL OF APPLIED BIOMECHANICS, 1993,9,238-253 O 1993 by Human Kinetics Publishers, lnc.

Body Roll and Handpath in Freestyle Swimming: An Experimental Study

Qi Liu, James 6. Hay, and James G. Andrews

The purpose of this study was to determine the influence of (a) body roll, and (b) the motion of the arm relative to the trunk, on the medial-lateral component of the path followed by the hand during the pull phase in freestyle swimming. Ten male swimmers swam three trials of freestyle at a long- distance workout pace. Three-dimensional (3D) underwater videography was used to record the body roll angle-time history and the path followed by the hand during the pull phase. A mathematical model was used to character- ize the motion of a swimmer's right upper limb in accord with 3D data from the videotape images, and to determine what the path of the hand would have been as a result of body roll alone. The contribution of body roll to the actual handpath was found to be nearly equal to the contribution of medial-lateral motions of the hand relative to the trunk.

The path followed by the hand of a freestyle swimmer during the pull phase might be attributed to motion of the arm relative to the trunk, to rolling of the trunk around the longitudinal axis, or to a combination of these two. Although these possibilities have been explored in a computer simulation study (Hay, Liu, & Andrews, 1993), there appears to have been no attempt to quantify the influence of body roll and motion of the hand relative to the trunk on the handpath in actual freestyle swimming. The purpose of this study was to determine the influence of body roll: and the influence of the motion of the arm relative to the trunk, on the medial-lateral component of the path followed by the ' .nd during the pull phase in freestyle swimming.

Methods

Subjects

Ten male swimmers from a university swimming team were used as subjects. The shoulder width, length of right arm, length of right forearm, and length of

Qi Liu is with Nike Sport Research Laboratory, Nike Inc., Beaverton, OR 97005- 6453. James G. Hay is with the Department of Exercise Science and James G. Andrews is with the Departments of Mechanical Engineering and Exercise Science, University of Iowa, Iowa City, IA 52242-1 111.

Freestyle Swimming: Experimental Study 239

right hand were measured for each subject, with the subject standing. The shoulder width was defined as the distance between the acromion processes of the right and left scapulae; the arm length as the distance between the acromion process of the scapula and the midpoint of the olecranon process of the ulna; the forearm length as the distance between the midpoint of the olecranon process of the ulna and the styloid process of the ulna; and the hand length as the distance between the styloid process of the ulna and the middle fingertip measured parallel with the long axis of the forearm and hand. Distances were measured with a metric tape. These descriptive data are shown in Table 1.

Data Collection

Two half-periscope systems (Hay & Gerot, 1991) were used to record each subject's performance. System 1, which was used to record each performance in side and bottom views (Figure I), consisted of a video camera (Panasonic AG450), a medium-size mirror on the side of the pool, a large mirror on the bottom of the pool, and a surface-wave deflector. System 2, which recorded each performanck in front view (Figure 2), had similar components to System 1 except a small mirror, attached to the top of the wave deflector, replaced the large mirror. The data were collected with both cameras operating with a sampling frequency of 60 Hz and a shutter speed of 11250 s. A timing system, consisting of two groups of three light-emitting diodes (LEDs) and a control circuit, was used to synchronize data from the two cameras. One group of LEDs was placed in the optical field of each video camera.

Before the first trial started, a plumb line with a 22-N (5-lb) weight was held approximately in the center of the optical field of the two cameras and its position recorded in side and front views. This line was used as a vertical reference line in the subsequent analyses. A 200 cm x 5 cm x 5 cm piece of wood was then positioned underwater, with its long axis horizontal and perpendicular to the end of the pool, and directly above the midline of the large bottom mirror.

Table 1

Descriptive Data on the Subjects (cm)

Shoulder Arm Forearm Hand Subject width length length length

Liu, Hay, and Andrews

Figure 1 - System 1, used to record below-water technique of sub-jects in side and bottom views. A = tripod, B = videocamera, C = side mirror and supporting frame, D = surface wave deflector, and E = bottom mirror.

Figure 2 - System 2, used to record above- and below-water techniques of subjects in front view. A-D, as in Figure 1; E = additional small mirror attached to edge of surface wave deflector.

The position of this piece of wood was recorded in bottom view. Later, when the videotape of the bottom view was analyzed, the 200-cm length of the piece of wood was used as a forward horizontal reference. Finally, a control object with known dimensions (Figure 3) was placed below the water surface and its position recorded in front, side, and bottom views for projection angle calibration. The mean differences between the known projection angles for the control object and the projection angles measured from videotape images were found to be 2" for the side and front views and 3" for the bottom view. Later, when the tapes

Freestyle Swimming: Experimental Study

Figure 3 - Control object used to validate measures of projection angles. A = fixed board, B = movable board, C = hinge, D = variable angle friction support bracket, E = supporting rods, and F = baseboard fastened to sidewall of pool.

were analyzed, the corresponding mean difference was added to all projection angles measured in the side, front, and bottom views.

Before the subject entered the water, a balsa wood fin mounted on a curved aluminum base was strapped to his back. The position of the fin was recorded in front view for determination of body roll angle-time histories.

Each subject swam three trials without a break. In each trial, the subject swam 15 m at a long-distance workout pace. The subject was required to swim straight toward the front-view camera and directly above the midline of the bottom-view mirror and to perform a complete right-arm pull phase while passing over the bottom-view mirror. (Note: For all of the subjects the right side was the preferred breathing side.)

Videotape Analysis

The videotapes were analyzed with the aid of a Peak 2D Motion Measurement System (Peak Performance Technologies, Inc., Denver, CO). A right-handed, Cartesian base reference frame B:Qxyz was established for the purpose of determining the body roll angle, the joint locations, and the arm segment projection angles (Figure 4). The origin Q of B was fixed to the body at the midpoint of the line connecting both shoulders. The x-axis was horizontal and directed perpendicular to the side of the pool, the y-axis was horizontal in the direction of swimming, and the z-axis was directed vertically upward. The body roll angle was defined as the projection angle 8 of the shoulder axis SQ to the horizontal x-axis (Figure 4 t a n angle equal to the projection angle defined by the forward edge of the fin and the vertical-as recorded in front view. Two points were


F Figure 4 - The base reference frame B:Qxyz had its origin Q a t the midpoint of the shoulders, its x-axis horizontal and directed toward the side of the pool on the subject's right, its y-axis in the direction of swimming, and its z-axis directed vertically upward. The body roll angle 0 was the projection angle defined by the shoulder axis SQ and the horizontal, a s seen in front view, an angle equal to the angle between the midline of the forward edge of the fin and the upward vertical.

digitized on the midline of the forward edge of the fin. The Peak 2D software was then used to calculate the angle between the fin and the vertical.

The estimated locations of the right shoulder S, elbow E, wrist W, and middle fingertip F were digitized in each of the three (front, side, and bottom) views. Peak 2D software was then used to calculate front-view projection angles between the vertical and the projection of SQ, the projection of ES, the projection of WE, and the projection of FW. The same procedure was also used with the side- and bottom-view tapes. Cubic spline functions were fit to the projection angle data obtained with each camera, and the fitted data were synchronized by use of a linear interpolation procedure.

Through use of the known distance between S and Q (the half shoulder width = 0.5 Ds), and the projection angles for the line segment SQ in the base reference frame B, the B coordinates locating the right shoulder S relative to Q (rS,,-J were found with a projection angle reconstruction method (see the appendix). With a similar procedure, the B coordinates locating the right elbow E relative to S (rWS), the right wrist W relative to E (rWE), and the right middle fingertip F relative to W (rFW) were determined. The location of F relative to Q was then found with the equation


Mathematical Model

The 3D mathematical model described by Hay, Liu, and Andrews (1993), with its four associated coordinate systems, was used to determine the handpath that would have been recorded by each subject if there had been no medial-lateral motion of the hand relative to the trunk during the pull phase. The trunk-fixed reference frame T:QxTyTzT was used to describe the medial-lateral motion of the fingertip of the swimmer's right hand relative to the trunk during the pull phase of the stroke. For the sake of brevity in what follows, the term handpath will be used to describe the locus, in the translating base reference frame B:Qxyz, of the middle fingertip F of the swimmer's right hand during the pull phase of the stroke. In a similar manner, the expression relative motion of the hand will be used to describe the medial-lateral motion of the middle fingertip of the swimmer's right hand in the rotating trunk-fixed reference frame T:QxTyTzT, during the pull phase of the stroke.

Three simplifying assumptions are implicit in the use of this model:

The right shoulder's flexion-extension axis XT remains in the translating vertical x,z plane of the base reference frame, B. The right shoulder S, elbow E, wrist W, and fingertip F remain in the rotating x,,yA plane of the arm-fixed reference frame, A. The longitudinal axes of the right forearm (EW) and right hand (WF) remain collinear (i.e., there is no right wrist motion).

The handpath histories in the translating reference frame B were found with the expression

where xF = (0.5Ds-a)cosO - b sin@ sine, yF = bcos@, zF = -(O.5DS-a)sinO - b sin@ cose, 8 is the body roll angle, (I is the shoulder extension angle, D, is the constant shoulder width of the subject (measured before the performance), a is the magnitude of the negative x, coordinate of F, and b is the magnitude of the y, coordinate of F (Figure 5). (Note: a is constant when there is no relative motion of the hand.)

The shoulder extension angle @ was determined from the B coordinates of S and E with the expression

where r , ,~ = xEi + y j + zEk and rs,, = xsi + y j + zsk. The constant a was taken as the x, coordinate of F at the beginning of the pull phase. The variable length b is given by the expression b2 = R2 - a2, where R, the length of the straight line between S and F, was found with the expression

Thus the simplified motion of the model (i.e., no relative motion of the hand) forced the right fingertip F to remain in the xA,yA plane at a fixed distance (a = constant) from the rotating YA,ZA plane, and at the varying distance R from S.


Figure 5 - Arm-fixed reference frame A:SxAyAzA shown in an overhead view.

Data Analysis

Body Roll Angle Versus Time. For the purpose of comparison among subjects, time was expressed as a percentage of the duration of the pull phase (normalized time). Cubic spline functions were fitted to the body roll angle versus time data so that the body roll angle at selected percentages of the time for the pull phase could be determined. The mean and standard deviation of the body roll angle at these selected percentages were computed for the 30 trials analyzed.

Maximal Body Roll Angle. The maximal body roll angles for each subject were identified, and the mean and standard deviation of the maximal body roll angles for the entire sample were calculated.

Influence of Body Roll on Actual Handpath. A method was developed to determine the average influence of the body roll on the actual handpath during the pull phase. This method was based upon a representation of the medial-lateral motions of the fingertip of the right hand in the translating base reference frame B, as seen in top view (Figure 6).

If a swimmer performed the pull phase without either body roll or relative motion of the hand, his hand would follow a straight line path from the point of entry to the point of exit, shown by Curve C, Figure 6. If he performed it with some body roll but without relative motion of the hand, the handpath would have the general form of Curve B, Figure 6. And if he performed it with both body roll and relative motion of the hand, as is generally the case in practice, the handpath would have the general form of Curve A, Figure 6.

The contribution of medio-lateral relative motion of the hand to the actual path recorded (Curve A) was determined according to the following argument. If the subject had performed the pull phase as a simple shoulder extension without either body roll or relative motion of the hand, his fingertip would have been at point N on Curve C at a certain instant. Alternatively, if he had rolled his body as he actually did in this trial, but without relative motion of the hand, his fingertip would have been at point R on Curve B at that instant. However, because the


-100-1

Figure 6 - Paths followed by the fingertip of the right hand, as viewed in the translating horizontal x,y plane of B:Qxyz, under varying conditions.

fingertip was actually at point P on Curve A at this instant, the subject must have moved his fingertip laterally relative to his rotating trunk when he rolled his body. Furthermore, for the sum of the medio-lateral motions of the hand due to body roll and to relative motion of the hand to match the actually observed location of the hand, the lateral distance traveled by the fingertip relative to the trunk up to that instant must have been equal to the distance between points R and P. This distance was therefore added to the x-coordinate of point N to yield the x-coordinate of point M.

The systematic repetition of the corresponding calculations throughout the time interval from entry to exit of the hand yielded Curve D, which represents the path the fingertip would have followed due to the relative motions of the hand alone-that is, in the absence of body roll. The medial-lateral location of any point on Curve A (e.g., point P) relative to Curve C is thus the algebraic sum of the medial-lateral deviation of the corresponding point on Curve B (point R) from Curve C, and the corresponding medial-lateral deviation of the point on Curve D (point M) from Curve C. The relationship among these quantities can be expressed as

N P = N R + R P = N R + N M .

The influence of body roll on handpath was determined by measuring the closed area (A,) between Curve B and Curve C, and the influence of upper limb relative motion on handpath was determined by measuring the closed area (A,) between Curve D and Curve C (Figure 7). The percent contribution P of body roll to medial-lateral deviations of the actual path followed by the fingertip was then computed with the following expression:


Y (c(i"j Curve c

Figure 7 - The influence of body roll on handpath, as viewed in the translating horizontal x,y plane of B:Oxyz, was obtained by expressing the enclosed area A, as a percentage of the sum of the enclosed areas A, and A,.

The areas A, and A2 were found with an electronic graphics calculator (Numonics Corp., Lansdale, PA).

Results and Discussion

Body Roll

The maximal body roll angle ranged from 51.5" to 66.0°, with a mean of 60.8" (Table 2). This mean value for the maximal body roll to the nonbreathing side exceeded previous estimates, recommendations, and measured values.

Counsilman (1977) observed that "the body rolls 35 to 45 degrees on each side as the swimmer makes a complete arm cycle. The swimmer rolls more on the side on which he breathes than on the nonbreathing side" (p. 161). Colwin (1991) concurred, in part: "The total amount of roll varies from 70 degrees to 90 degrees or 35 degrees to 45 degrees on each side of the body's long axis" (P. 33).


Table 2

Maximal Angle of Body Roll

Subject Maximal angle of body roll (")

Maglischo (1982) stated, "Although it is possible to roll too much, most swimmers roll too little. Crawl stroke swimmers should roll at least 45 degrees to each side (from a prone position). Most swimmers roll more than 45 degrees toward their breathing side" (p. 94). Maglischo then directed the reader's attention to photo sequences in which "the proper amount of roll is shown." These photo sequences show maximum body roll angles that in one instance (Maglischo's Figure 2.6B) appear to approach 80" and in the other (Maglischo's Figure 2.9H), taken from a difficult oblique angle, appear to be in the 70-80" range. In both cases, these maximum angles appear to be to the swimmer's breathing side. None of the previously cited sources provided quantitative data in support of their observations or recommendations.

The only known quantitative data were reported by Beekman (1986) and Levinson (1987). Beekman reported mean values for the maximum body roll angle of 47.8" and 59.7" to the nonbreathing and breathing sides, respectively. These values support the observations of Counsilman and Maglischo that the body rolls more to the breathing than to the nonbreathing side. The 47.8" value for the nonbreathing side is, however, considerably less than the mean of 60.8" found in the present study. Levinson (1987) reported values for an Olympic sprint freestyle champion of 45" (to one side) and >50° (to the other, presumably breathing, side). The first of these is similar to the mean value reported by Beekman for the nonbreathing side and again is considerably less than the corresponding mean value of the present study.

The difference between the mean value reported by Beekman and that obtained in the present study is unlikely to have been due to differences in the methods used to obtain the measures of body roll. The method used-recording in front view the excursions of a dorsal fin strapped to the subject's back-was the same in both studies. There were, however, some major differences between

248 Liu, Hay, and Andrew5

the two studies in the subjects and the speeds at which they were asked to swim. The subjects of the Beekman study included two female high school swimmers, four male college swimmers, four female college swimmers, and one female graduate student who had competed in swimming as an undergraduate; they were asked to perfom a set of six repetitions over 25 yd (22.9 m) at the sprinting pace they would use when swimming such a set in practice. The subjects of the present study were 10 male college swimmers; they were asked to swim a set of three 15-m repetitions at long-distance workout pace. In short, the subjects of one study were mostly females and the subjects of the other were exclusively male; the former swam at sprint speed and the latter at a long-distance pace. Although it is not clear how these differences might cause a difference in the mean body roll angle recorded, it does seem likely that they are somehow responsible for this difference.

Means and standard deviations of the body roll angle are shown as a function of the duration of the pull phase in Figure 8. The relatively small standard deviations (3.9-6.4") recorded for the first 80% of the pull phase are indicative of a fairly high degree of consistency among subjects in the execution of this part of the body roll. The larger values (6.7-8.9") recorded for the final 20% of the pull phase are indicative of an increasing variability in the angular speed of the rolling action. Although there is nothing in the data of the present study to either support or reject the notion, this increasing variability might conceivably be associated with the acceleration of the hand during the latter half of the pull phase, an acceleration that has been shown to be characteristic of skilled performance in freestyle swimming (Counsilman, 1981). Thus, it might be that the more skilled subjects developed a higher velocity of the hand, and an associated higher angular speed of body roll, during the final part of the pull phase than did the less skilled subjects.

1 1 1 ' 1 ' 1 ~ 1 " ' l " ' 0 20 40 60 80 1

Time (%) -20-

Figure 8 - Means and standard deviations computed from the body roll angle data for the 10 subjects at consecutive 5% increments of the duration of the pull phase.


If the maximum angle of body roll were reached midway through the duration of the pull phase, and if the curve representing the mean values as a function of the duration of the pull phase were symmetrical, the angular speed of the rolling motion in one direction (as a function of time) would have been the same as that in the reverse direction. This, however, was not the case. Instead, the maximum angle of body roll was reached earlier, at 45% of the duration of the pull phase (Figure 8). This implies that, on average, the angular speed of the rolling motion was a little greater during the first part of the pull phase than it was during the second (or return) part.

Handpath

The paths followed by the fingertip of the right hand are shown in an overhead view for each of the 10 subjects in Figure 9. Critical points in these paths are presented in Table 3. These data show that in 8 of the 10 cases the fingertip of the right hand was brought to or across the midline (x = 0) of the subject's body, and that in the remaining 2 cases it was brought very close to the midline (x = 2 cm and x = 5 cm). These same data, coupled with the data on the shoulder widths of the subjects (Table I), showed that in 2 cases (Subjects 7 and 9) the right fingertip was brought beyond, and indeed far beyond, the left side of the subject's body. In the case of Subject 7, the fingertip extended 25.8 cm [49 - (46.4/2)] beyond the left side of the body, and in the case of Subject 9, it extended 24.35 cm [49 - (49.3/2)] beyond the left side of the body.

The point at which the maximum medial deviation of the right fingertip was recorded was forward of the line of the subject's shoulders (y = 0) in 6 cases, level with this line in 1 case, and behind it in the remaining 3 cases. The mean value for the 10 cases was 10 cm forward of the line of the shoulders. This value was distorted by the presence of two extreme values: the 60 and 65 cm recorded at entry for Subjects 2 and 9, respectively. With these two subjects excluded, the mean value reduced to a more representative -4 cm.

Body Roll, Relative Motion, and Handpath

A comparison of the contributions of body roll and relative motion of the hand to the actual handpath in the x,y plane of B:Qxyz (top view) is presented in Table 4. The contribution of body roll to the actual handpath as seen in this plane ranged from 41.1% to 73.6%, with a mean value of 52.1%. This mean value indicated that, on average, body roll and relative motion of the arm at the shoulder, elbow, or both contributed about equally to determining the path followed by the hand.

Visual inspection of Curves A and D for each subject (see Figure 6) revealed that, for the majority of the subjects, Curve D (the curve showing the medial- lateral motion of the fingertip due to the relative motion of the hand) was to the right of Curve C (the straight line joining the points of entry and exit) throughout the pull phase, and that Curve D moved first away from, and then back toward, the midline of the trunk, represented by the y-axis of the graph. This meant that the relative motion was exactly the reverse of what is commonly supposed. Instead of the subjects moving the arm first toward, and then away from, the

Figure 9 - Path followed by the right fingertip of each subject (Sl-S10) as seen in the x,y plane, from entry (at top) to exit (at bottom). The x-axis corresponds with the line of the shoulders and the y-axis with the midline of the trunk.


Table 3

Coordinates of the Right Fingertip at Critical Instants in the Pull Phase (cm)

Instant (coordinate)

Subject Entry (x) Exit (x) Most medial (x) position Most medial (y) position

Note. Positive values of x are to the right of the long axis of the body (x = O), and positive values of y are forward of the line of the shoulders (y = 0).

Table 4

Contribution of Body Roll to Actual Handpath

Total area, A, + A, Body roll area, A, P Subject (digitizer units) (digitizer units) Vo)

midline of the body as the pull progressed, the majority of the subjects moved it away from, and then toward, the midline. A few of the subjects preceded this sequence of relative motions by a brief and short-range motion of the hand toward the midline immediately after entry.

252 Liu, Hay, and Andrew

Conclusion

The findings of this study indicate that the medio-lateral motions of the hand observed in male, collegiate freestyle swimmers performing at a long-distance workout pace do not owe their existence solely to motions of the hand relative to the trunk, as might be expected. Instead, they are due about equally to the rolling action of the swimmer's trunk about its long axis and to the medial-lateral motions of the hand relative to the trunk. Furthermore, these relative motions do not occur in a medial-then-lateral sequence but instead in a lateral-then-medial sequence. In other words, the swimmer sweeps the hand away from the rotating trunk in the first part of the pull and toward the rotating trunk in the second part. These findings contradict what appear to be widely held beliefs concerning the relative motion of the hand in freestyle swimming. They are thus deserving of close scrutiny by teachers, coaches, and scientists concerned with the techniques employed in swimming.

References

Beekman, K.M. (1986). Characteristics of the front crawl techniques of swimmers with shoulder impingement syndrome. Unpublished master's thesis, University of Iowa, Iowa City.

Colwin, C.M. (1991). Swimming into the 21st century. Champaign, IL: Leisure Press. Counsilman, J.E. (1977). Competitive swimming manual for coaches and swimmers.

Bloomington, IN: Counsilman. Counsilman, J.E. (1981). The importance of hand speed and acceleration in swimming the

crawl stroke. In J.M. Cooper and B. Haven (Eds.), Proceedings of the Biomechanics Symposium, Indiana University, October 26-28, 1980 (pp. 226-238). The Indiana State Board of Health.

Hay, J.G., & Gerot, J.T. (1991). Periscope system for recording the underwater motions of a swimmer. International Journal of Sport Biomechanics, 7(4), 392-399.

Hay, J.G., Liu, Q., & Andrews, J.G. (1993). Body roll and handpath in freestyle swimming: A computer simulation study. Journal of Applied Biomechanics, 9, 227-237.

Levinson, D.A. (1987). Internal stroke motions and the effective coaching of stroke mechanics. Journal of Swimming Research, 3(2), 21-28.

Maglischo, E.W. (1982). Swimming fast. Palo Alto, CA: Mayfield.

Appendix

Derivation of Projection Angle Reconstruction Method

The B coordinates of points F, W, E, and S were obtained from measured distances between these anatomical landmarks for each subject and from the measured projection angles (q,, q,, . . ., q,,) that orient the line segments FW, WE, ES, and SQ in B. To illustrate the procedure used, consider the problem of determining the B coordinates of the right shoulder S (x,, y,, 2,) given the shoulder width Ds, the front view (body roll) projection angle q, = 0 of SQ, the side view projection angle q, of SQ, and the bottom view projection angle q8 of SQ (Figure lo), where Q has B coordinates XQ = 0 = yQ = .q.


Figure 10 - Projection angles for the shoulder axis, SQ.

The following four relationships can be deduced from Figure 10:

Any combination of two of the three projection angles (q,, qx, q,) of the line SQ can be used to determine the B coordinates of S. For example, if the side and bottom view projection angles qu and q, are used, substitution of Equations A3 and A4 into A1 yields

from which xs can be determined. With xs now known, ys can be found from Equation A4, and then zs can be determined from Equation A3.

Acknowledgments

We are grateful to Dr. Warren G. Darling of the University of Iowa for his insight and suggestions; to the members of the men's swimming team of the University of Iowa, who served as the subjects of the study; and to John Porter, Bing Yu, John Chow, and Gang Huang, who assisted with the data collection.

body roll and handpath in freestyle · pdf file · 2007-08-29body roll and handpath...

Documents