Number 26

Produced, edited, and copyrighted by
Professor Brent S. Rushall, San Diego State University


Brent S. Rushall, Ph.D., R.Psy.
San Diego State University

Eric J. Sprigings, Ph.D
University of Saskatchewan

Jane Cappaert, M.S.
US Swimming

Harry A. King, Ph.D.
San Diego State University


Five stroking sequences from four male events were filmed during winning performances at either the 1991 World Championships or 1992 Olympic Games. Features analyzed were: (a) angle of the body relative to the horizontal plane, (b) angle of the central axis of the head relative to the horizontal plane, (c) extent of movement of the legs and positions of the feet, and (d) rolling motions of the hips and shoulders. It was found that the body and head in all Ss oscillated vertically to varying degrees, foot and leg positions were often in non-propulsive positions, and the hips and shoulders seldom moved in unison. The effects of these factors on propelling economy and streamline were discussed. Cautious suggestions for improvements were made. It was suggested that even these champions were not as efficient as they could be.


The characteristics of resistances in swimming have been discussed in detail (7, 8, 9). A re-awakened emphasis has been placed on the possible value of reducing movement resistance to facilitate easier progress through water. In the concept of propulsive economy, effective propulsive force is divided by the resistance to forward movement created by passive factors (e.g., frictional resistance from skin, swimming attire, hair, and minimum form resistance) and active factors (e.g., unnecessary movements or positions in the water). Without changing the amount of power applied, the speed of progression should increase if resistances are reduced.

Theorizing about the body (shoulders, torso, and hips), that is one aspect of the crawl stroke, in isolation from all other aspects is dangerous. The human body is ill-suited to progress efficiently through water as many of the actions required for propulsion present problems in terms of resistance. Swimming science has yet to determine the most beneficial characteristics for streamlining. In crawl stroke swimming, athletes have quite different shapes but perform in the relatively constant medium of competitive water. This leads to the possibility of considerable individual differences between swimmers in propulsive and resistive dynamics as their fixed properties are accommodated in an "optimizing" process. The choice of just how and when propulsive forces should be applied, and how and when efforts are made to reduce resistive forces, must take into consideration a swimmer's body shape and physiological capacities, race length, and competitive stroke. In very short races, where the generation of energy/power is of great importance, streamlining may not be crucial. On the other hand, in distance races, energy conservation and force application are both vital so the relationship between propulsive forces and streamlining should be important.

Swimming texts have not been consistent in the description of the body position. A streamlined, horizontal body position which creates the smallest form resistance and the least energy for its maintenance have been proposed (3, p. 20). In contrast, the East Germans advocated "The shoulder line should be raised above the water surface, the pelvis should be deep enough below the surface to permit convenient leg actions beneath the surface. In this position, the longitudinal axis of the body runs at a slight angle to the water surface to give the body a good hydrodynamic position" (4, p. 64). It is generally conceded that the body position, as described by its longitudinal axis, should be stable. Vertical movements should be avoided since they consume energy and increase resistance (3, p. 22).

The recommended head position has been less exact. Some coaches advocate the face should look forward, others that it look down. It has been shown that as the angle of the head increases, in relation to horizontal, so does resistance (3, p. 21). Thus, the lower the head angle, the greater would be the streamline. It is commonly accepted that the head should rotate on its own longitudinal axis when breathing and maintain the same axis when not breathing. The axis of the head should be stable throughout the total stroke (3, p. 23; 4, p. 64).

Despite detailed descriptions of body and head dynamics during the crawl stroke, carefully controlled observations of these kinds of kinematic features, particularly as used by elite swimmers in competitions, have been scarce.

This study analyzed kinematic data from champion male crawl-stroke swimmers in several events, at different stages in those events, and always during exceptional performances. The observation and measurement of athletes so their resultant "patterns" can be generalized, is a basic requirement for sport science. It is not reliable to infer "benefits" from observing one athlete or one set of data. Replication provides an assessment of reliability. For a long time, swimming coaches have attempted to discover what one champion does and then pattern the activities of others on that discovery. This approach still persists (1, 5) despite its inherent dangers. Studies are available that show the responses of elite athletes to training are very different (6) as are techniques of champions (2). Exact modeling from a single individual is erroneous, counter-productive, and violates the principle of individuality.


Different stages of four events were filmed during winning performances at either the 1991 World Championships (Perth, Australia) or the 1992 Olympic Games (Barcelona, Spain). The sequences (Ss) and locations were:

  1. Keiren Perkins at 40 m into his gold medal 1500 m performance at Barcelona.
  2. Keiren Perkins at 1440 m into his gold medal 1500 m performance at Barcelona.
  3. Evgenyi Sadovyi at 140 m into his gold medal 200 m performance at Barcelona.
  4. Alexandre Popov at 40 m into his gold medal 100 m performance at Barcelona.
  5. Joerg Hoffman in the mid-stage of his gold medal 1500 m performance at Perth.

Ss were filmed (60 fps) above and below the water. One set of two high-resolution video cameras was placed under the water on the pool bottom in lane 4 and focused upward and toward the center of the lane. A similar set of cameras was placed above the water on either side of lane 4 and focused downward toward the center of the lane. Cameras were focused at the 40-m mark of the pool.

One representative stroke cycle (a cycle being defined between successive LA entries) was selected from film taken during the five performances noted above. Positions of several body markers were noted in every second frame of the selected cycle. Locations of these markers were recorded digitally, and results combined to effectively make a composite 3-D "digital" recording of the changing locations of the markers. In this way it was possible to describe kinematic features of the stroke cycle. The following action features were noted.

  1. The angle of the body relative to the horizontal plane. The line of the body was determined by joining the center of the line between the hips to the center of the line joining the points of the shoulders.
  2. The angle of the central axis of the head relative to the horizontal plane.
  3. The extent of movement of the legs and positions of the feet.
  4. The rolling motions of the hips and shoulders.

From these records it was possible to develop pictures which would aid analyses of interest. Stick-figure representations (pictures) of the lateral profile of a swimmer's position at four stages during the chosen stroke cycles were made. The four stages were:

  1. When the arm was fully immersed at the left arm entry (LA entry).
  2. The point at which the upper LA was closest to vertical during the pull, but before the right arm entered.
  3. When the arm was fully immersed at the right arm entry (RA entry).
  4. The point at which the upper RA was closest to vertical during the pull, but before the LA entered.

These stages are pictured in Figure 1.

Figure 1

Figure 1. Five sequences of a full crawl stroke cycle for male World and Olympic Champions. "B" and "H" indicate the body and head angles at each stroke stage. K. Perkins (KP1) refers to Keiren Perkins' action at the 40 m mark of his 1500 m race while K. Perkins (KP2) refers to his action at the 1440 m mark of the same race.


Table 1 summarizes the five analyzed sequences for body angles and movements, head angles and movements, leg and foot positions and movements, and body rotations. Case-by-case appraisals of these features are contained below.



Body Angles and Movements
KP1  Oscillates between 3 and 8 degrees; related to LA and RA downward press
KP2  Oscillates between 6 and 12 degrees; related to LA entry and downward press
ES   Oscillates between 3 and 15 degrees; related to LA and RA downward press
AP   Oscillates between 17 and 20 degrees; consistently high
JH   Oscillates between 8 and 20 degrees; high angle related to LA entry; low
     angle related to RA entry

Head Angles and Movements
KP1  Oscillates between 18 and 56 degrees; related to RA press and pull 
KP2  Oscillates between 14 and 48 degrees; related to LA and RA entry and press
ES   Oscillates between 8 and 29 degrees; related to body angle changes
AP   Oscillates between 32 and 56 degrees; exaggerated during LA pull; greatest
     angle of all Ss
JH   Oscillates between 8 and 20 degrees; highest angle at LA entry; lowest
     angle at RA entry; associated with arm stroke phase

Leg and Foot Positions and Movements
KP1  Stay close to being in body shadow; at least one foot drags in three of the
     pictures; foot drags at entries
KP2  One action outside of body shadow; feet drag down in two pictures; one foot
     drags with RA entry
ES   To a minor degree outside of body shadow; at least one foot drags in each
     picture; one foot drags at each arm entry
AP   Big kick outside of body shadow; one foot drags at each entry
JH   At least one foot always drags

Body Rotations
KP1  Hips do not roll to the same degree as shoulders
KP2  Hips aligned with shoulders
ES   Hips stable or twisted in comparison to shoulder roll
AP   Body twisted during LA pull; hips and shoulders roll together during RA
JH   Uneven shoulder rotation; hips remain relatively stable

K. Perkins (1) (KP1)

It is assumed that this sequence represents a non-fatigued state because it occurs at 40 m into the 1500 m race.

  1. The body angle was consistently at 8 degrees in three of the four positions (#1, #3, and #4). It dropped to 3 degrees during the LA pull. The body angle increased again with the RA entry. The position and range of angles was the lowest of any S. The vertical shoulder movements seemed to be related to the downward pressing stage of each arm action.
  2. KP1's head angle varied considerably. At the LA entry it was lowest, but then rose to be highest during the RA pull. The two highest head angles were associated with the RA entry and initial pull (#3 and #4). The head movement was different for the LA and RA pulls. The changing angle of the head appeared to be independent of body angle.
  3. The legs trailed in the "shadow" of the frontal area of the body in pictures #1 and #4. In picture #2, the left leg was marginally outside of the shadow and in #3, the lower part of the hyperextended leg was exposed. Those positions would have increased form resistance. In pictures #1 and #2, both feet were angled such that no propulsive force was possible. At both arm entries, the foot of the opposite leg was angled to cause a non-propulsive braking action.
  4. In picture #1, KP1's shoulders had rolled but the hips rotated to a much lesser degree. However, in picture #2 the hips were aligned with the shoulders, but in #3 and #4, the shoulder action once again exceeded that of the hips. The shoulders rotated through a greater angle than did the hips.

K. Perkins (2) (KP2)

The sequencing of movements at this stage of fatigue of KP's 1500 m race is different to that of KP1. The changes exhibit some very different actions and angles.

  1. The body angle was greatest at, and associated with, the LA entry. it was lowest in pictures #2 and #3, when the RA was extended forward and entering. It rose again as the LA prepared to enter. The angles were greater than in the earlier nonfatigued sequence (KP1), and also exhibited a different movement pattern. The vertical shoulder movement associated with the RA press in KP1, was not exhibited in this sequence.
  2. KP2's head oscillated vertically at lower angles and through a decreased range when compared to KP1. At both arm entries, the head angle was lowest. It increased during each pull. The head movement appeared to be related to a rhythmic occurrence in the stroking pattern.
  3. In picture #2, as the RA neared the water and the LA pulled, the head was lifted, and the left leg dropped down, exposing the thigh from the body shadow thus, increasing form resistance. As occurred with KP1, the position of a foot in pictures #3 and #4 was one that could not create propulsion. In #3 it would increase drag resistance noticeably.
  4. The body rolled more than in the earlier KP1 sequence. The shoulders and hips rotated in unison through approximately the same range of angles.

E. Sadovyi (ES)

  1. The body angle of ES reciprocated vertically. At the LA entry (#1), the body angle was greatest as the arm pressed forward and/or straight down, but once it commenced its primarily backward push (#2), the body angle flattened considerably. At the RA entry (#3) the body rose again in concert with a forward and/or straight down press (#4).
  2. The head angle increased as the body angle increased (#2, #3, and #4), and decreased as the body angle decreased (#1 and #2). Vertical head and body movements appeared to be related.
  3. In pictures #3 and #4, a lower leg dropped down increasing form resistance. These exposures coincided with the highest head angles. In all pictures, including both the LA and RA entry, at lest one foot was in a nonpropulsive position.
  4. Picture #1 showed that the trunk was twisted. For the remainder of the stroke, the shoulders rotated through a greater range than did the hips. Generally, the hips remained flat.

A. Popov (AP)

In contrast to the three previous Ss, AP was a sprinter.

  1. AP's body angle was consistently high. It displayed the least variation of the five Ss.
  2. AP's head position was also the highest of the five Ss. It displayed the highest maximum and minimum angles. The greatest angle was associated with the LA pull, which seemed to be quite deep.
  3. In comparison to all Ss, AP's kick covered the greatest movement range. In three pictures (#1, #2, and #4), at least one of AP's legs was below hip level. A foot was positioned so that no propulsive force could be created at both entries.
  4. In pictures #1 and #2 (LA entry and pull), the body was twisted. At RA entry and during its pull, the hips faced the same side as the shoulders but were not at the same angle.

J. Hoffman (JH)

  1. The body angles of JH were more than the other two distance Ss (KP1, KP2, and ES). It was highest when the LA initiated a downward press. Vertical oscillations appeared to be related to the arm cycles.
  2. The head angle was greatest in picture #1 when the LA commenced its downward press, the RA exited the water, and the body angle was greatest. It was least at the RA entry, when the body angle was low. This suggests an association between body angle, head angle, and arm entry.
  3. In each picture, at least one foot was at an angle that could not produce propulsion. In picture #1, both feet were dragging coinciding with the highest body and head angles.
  4. The hips remained relatively flat throughout the sequence. The shoulder roll was uneven. The left shoulder remained flat at the LA entry and during its pull, but the right shoulder rolled considerable during the RA pull.


Several assumptions about crawl stroke swimming techniques underlie this analysis. Most have been justified by Rushall, Holt, et al. (8) and Rushall, Sprigings, et al. (9).

  1. The position of the body affects the peripheral actions of the limbs. It would be wrong to consider what the limbs do without relating them to the body's posture and movement.
  2. Actions which move large amounts of water (wave resistance), increase the frontal and eddy resistance of the system (form resistance), and increase the surface area of the swimmer (frictional resistance), are usually detrimental to propulsion.
  3. Movement economy can be assisted by reducing passive and active drag resistances as long as no propulsive movements are compromised.
  4. Observations taken during competitions are representative of the stroking form at that stage of the event.

Two factors complicate discussion of the observed phenomena: (a) the crawl stroke is so complex, that many of the observed actions could cause or result from other segment movements, and (b) verified dynamics of resistance reduction have not been demonstrated within a total stroke, although several have been exhibited in isolation. The interpretations offered below should be considered speculative in nature even though a great deal of caution was taken in their formulation.

Body Angles and Movements

All Ss were observed to have an inclined but varying body angle in their active posture. The angle decreased during the LA pull and the RA recovery for all Ss. When the RA pulled and the LA recovered the angle was stable in KP1, but increased in the others.

The greater the body angle, the greater would be the form resistance. As well, when the body is angled, it is normal for the legs and feet to drop (see AP). Changing the angle would produce movement induced wave resistance. Each alteration would require a counter-balancing reaction in some other part of the anatomy. In these observations, it was usually one or more feet.

On the other hand, it could also be argued that an angled body could contribute to beneficial lift (4). AP exhibited the greatest body angle, a position that accentuates, rather than diminishes, form resistance. Because of the poor hydrodynamics of the frontal shape and the relatively slow speed of progression of a swimmer, it is unlikely that benefits from lift would outweigh the detrimental costs of the increased drag. Further, if an angle was helpful, it should be maintained all the time throughout the stroking cycle. No swimmer exhibited a stable or nearly stable angle and so the lift explanation appears less credible than the "stable, flat body position" theory.

The lowest body angles were exhibited by KP1 and the highest by AP. The range of movements appear to be associated with particular actions in the stroke, that is, they could be caused. Four of the five Ss had the highest body angle at LA entry and close to the lowest angle at RA entry.

It should be possible to stabilize the body angle. It is proposed that the flatter the body angle the better, since body angle is an integral part of streamlining and the body is a nonpropulsive segment. A horizontal body alignment would minimize form resistance, and one that was stable would also reduce wave resistance. No S displayed this kinematic feature. It may be impossible to effect and may best serve as a technique goal to be striven for, with the realization that it will never be attained.

The surprising feature of this analysis was that all Ss oscillated vertically in their body alignment, a movement that could be deemed unnecessary. A vertical movement would not contribute in any way to propulsion but would move large amounts of water through both ascending and descending phases diverting a swimmer's energy to water movement rather than propulsion. When the body angle is highest, form (frontal) resistance would be increased because the hips would drop slightly to accommodate the greater angle. A swimmer's ultimate potential would be degraded by the amount of energy consumed and resistance created in this seemingly unnecessary action.

Since vertical movements of the body do not contribute to propulsion but do consume and dissipate energy, it is worthwhile to consider what might cause them. Four Ss exhibited the greatest body angle when the LA pressed down. That could be interpreted as meaning the shoulders rise as a reaction to the LA downward press. It would be desirable not to have oscillations. Some possible modifications of technique to prevent this undesirable feature are: (a) have some other part of the anatomy counteract the LA downward press; or (b) change the LA action after entry to be more propulsive by generating backward rather than vertical force. The latter effect could be achieved by bending the arm at the elbow upon entry rather than using a straight arm press.

Two possible counteracting movements are as follows:

  1. The other arm to that which is pressing. If one arm enters the water at the same time that the other exits the water, the vertically opposed actions could serve to cancel each other's forces. If that occurred, the body could remain stable.
  2. A kick to create a force that is equivalent to the downward pressure. The problem with this alternative is that the duration of a kick is much shorter than that of the arm press and so would not effectively cancel the full press. The duration of the downward press could be minimized by attempting to create propulsive forces by bending the elbow as soon as the hand enters the water.

No swimmer exhibited a static body angle as advocated by both theories of alignment. It is difficult to rationalize vertical movements of nonpropulsive body segments as contributing to forward propulsion. The oscillations could be considered technique deficiencies in all Ss although KP1 came closest to the theoretical ideal.

Head Angles and Movements

All Ss exhibited vertical head movements which were related more to phases in the stroking pattern than body positions. In three Ss (KP1, KP2, ES) the highest head angle was related to the RA pull, as opposed to the body angle which was related to the LA entry. In JH the head lift and body lift were both associated with the LA press. There is no obvious pattern of relationship between body and head movements across all Ss.

In three Ss, the highest head angle was associated with a foot position that could not be propulsive (KP2 #4; ES #4; JH #1). AP's thigh dropped to increase form resistance when the head angle was highest. It is possible that the drag created by the foot/limb counter-balanced the rotational force created by the head lift. If that is the case, then a head movement is detrimental to streamlining and propulsive economy.

The difference between the body and head angles is a measure of neck extension. In KP, the degree of extension is greater in the nonfatigued than fatigued state. It is hard to justify any benefit from such an action since the range of motion is reduced in fatigue, possibly to save energy.

It is hard to suggest any benefit to be gained from vertical head movement for it does not contribute to propulsion. The head is a very dense part of the body. Its movement would have to be offset by some other significant downward pressure. It appears that might be accomplished by the early part of arm actions in some Ss, significant leg-foot movements in others, or combinations of both. Such movement reactions would compromise the potential to minimize resistance.

For effective streamlining, it would be better to rotate the head only on the longitudinal axis even when breathing so that no vertical displacement occurs. Such an advocacy has been part of swimming coaching lore for many years but none of the Ss in this study displayed such a technique feature. Particular attention to suppressing vertical head actions in elite swimmers seems to be warranted.

Leg and Foot Positions and Movements

For perfect theoretical streamlining, the legs and feet should be maintained within the "shadow" of the cross-sectional area of the body. The challenge with leg movements is to keep them efficient, that is, creating more propulsion than resistance. To do this, both feet and leg sections have to be at least horizontal, or inclined upward and moving backward. The preparation to kick actually moves the back of the lower leg and foot in a slightly forward direction which increases both form and wave resistance.

If a leg section or foot emerges below the shadow, it is angled down and usually moving slightly forward, increasing both form and wave resistance. The emergence of the thigh, often in preparation for kicking, would have a secondary problematical effect. The surface would move largely downward and slightly forward. That would create a counter-productive "braking" force. The same could be said for foot and lower leg angles that fall below horizontal. Since a leg is a relatively large section of the body, any unnecessary movements would increase wave resistance excessively. The larger the movements, the greater would be the amount of water moved, and consequently, the greater the amount of energy lost.

All swimmers dropped some portion of the leg below the shadow. The thigh dropped down prior to completing a kick (KP1 #2; KP2 #2; ES #4; AP #2; JH #2), and the lower leg was inclined down at the end of a kick (KP2 #3; ES #3 and #4; AP #1 and #3; JH #4).

It is possible that braking movements are necessary to offset some other action and contribute to maintaining streamline and progress direction. For example, dragging the feet and legs tends to lift the hips and are usually performed to correct some exaggerated force (e.g., head lift or downward arm press).

These actions may cause or be caused by other actions in the complex crawl stroke movement. If they are necessary, then their resistance should be counter-balanced by greater force production somewhere else.

AP exhibited the largest leg action and "braking" effect. It is not possible to conclude that this inefficiency is really a bad aspect for a sprint swimmer. Although it may reduce streamline efficiency, it may allow an even greater offsetting propulsive force to be created by some aspects of arm actions. For example, a big kick may allow a bigger arm sweep that would facilitate achieving a greater terminal velocity. Since sprinting is not energy limited, that is a distinct possibility. It is supported by the fact that the world's top sprinters have very noticeable kicking actions. However, it would be interesting to see what would result if AP reduced the sweep of the kick and thus, achieved greater streamlining, and then evaluate its effect on forward progress, force production, and resistance.

Coaches frequently promote "kicking hard" as an aspect of swimming faster. It is accepted that harder kicking is necessary to at least partially counter-balance increased pulling forces. It may be necessary to kick very hard in maximum sprinting but unless care is taken, "over-kicking" may detract significantly from propulsion. Thus, it is necessary to analyze the efficiency of kicking actions to see that they do not go outside the range of necessary and contributory benefits.

Just when a hard kick turns from being beneficial to detrimental is not easy to tell. However, the larger the kick the greater will be the form and wave resistance incurred, and the greater will be the braking action. In all pictures, AP exhibits leg positions that could not produce propulsive forces.

The four distance swimmers (KP1, KP2, ES, JH) generally have their legs in, or closer to, the body shadow. Since endurance events promote swimming efficiency, a trailing high leg position should contribute to streamlining and not enhance resistance. At least for distance swimmers, the position of the legs (usually signaled by the position of the feet) should be monitored and corrected if necessary. It is reasonable to expect distance swimmers' heels to kick to the surface with the toes pointed. When not kicking, as in a two-beat kick, the legs-trail position should be high so that the legs are kept completely in the body shadow.

In all swimmers, at least one foot dragged causing a braking effect (KP1 #1 and #3; KP2 #3; ES #1 and #3; AP #1 and #3; JH #1 and #3). Usually, such actions are reactions to other movements. In at least one stroke in all Ss, the dragging foot occurred in concert with upward head and body movements as well as downward arm presses.

In JH and AP, the drag force component of the kick (the major force) generally would be close to vertical or forward rather than backward. These Ss attempt to swim forward with arm actions but brake against forward propulsion with the kick. It is possible that such a kicking action is a reaction to elevated body and head angles, particularly in AP. The downward/braking force of the kick would tend to lift the hips and legs and reduce the angled body position. In AP, it is also possible that the hips have to be low to facilitate a bigger kick (4) which in turn allows a greater arm force.

In these champions the kick is not always propulsive, at various stages appearing to increase resistance. When the head was high and/or the body angle was greatest, at least one foot attained a nonpropulsive position. That suggests the potential of the kick to assist in propulsion is diminished the more the body and head depart from horizontal. In stroke analysis for determining possible improvements, this action is one that should be considered for minimization because it does not assist propulsion. To achieve that, it may be necessary to correct other parts of the stroke, possibly by aligning the head and body horizontally, and reducing or eradicating their vertical movements as well as reducing vertical force components at the arm entries.

Body Rotations

It is not as easy to define the parameters of body rotation during the crawl stroke as it is to describe longitudinal properties. Part of the problem is that the flexibility and strength of a swimmer determines how much rotation is necessary to place the arm in the most advantageous position for generating propulsive force. However, it is proposed that certain aspects of shoulder and hip rotation must be demonstrated in order to optimize the body's musculature for enduring force application and to minimize detrimental resistance.

Figure 2 illustrates the effect of hip and shoulder rotation on form resistance. When the hips remain stable and flat and the shoulders rotate to an appreciable angle, the frontal area increases. However, if the hips rotate in a similar manner to the shoulders, the resistance area is decreased from the previous conceptualization. Streamlining would be greatly assisted by a combined and similar hip and shoulder rotation.

Figure 2

Figure 2. The shoulder and hip rotations of a hard-kicking swimmer. "H" is the angle of the hips, "S" is the angle of the shoulders, and "C" is the potential angle of the hips if they were rotated in concert with the shoulders. "A" and "B" are symbols for portions of the cross-sectional area of the body. If the hips were not rolled, then the cross-sectional area would be "A" + "B." If they were rolled to match the shoulders, the cross-sectional area would be reduced to "A." Similar shoulder and hip roll angles are advantageous for streamlining.

Figure 3 illustrates two frontal perspectives of KP in the Barcelona 1500 m swim. The first depiction is at the 40 m mark of the race (KP1) when reasonably hard kicking and fast speed occurred. The hips (19 degrees) have not "rolled" as much as the shoulders (49 degrees). The right side of the frontal form has "opened-up" to increase this aspect of resistance. However, at the 1440 m mark of the event (KP2), when kicking and swimming speed were lessened and the swimmer was fatigued, the hips (41 degrees) and shoulders (42 degrees) are in harmony and the reduction in cross-sectional area is obvious.

Figure 3

Figure 3. Shoulder and hip rotations for KP1 and KP2. As the race progressed, hip rotation increased, improving streamline because of reduced form resistance. In both instances, the shoulder roll is greater than 40 degrees.

The KP illustration needs to be interpreted with caution. The two pictures are intended to be as similar as possible but are not at identical stages of the stroke. This is because the timing of arm recoveries and pulls, and consequently kicking, changed as the race progressed. However, the degree of rotation of the shoulders is greater than 40 degrees in both cases and the upper arms in the pulls and recoveries greater than 45 degrees. It is hypothesized that as the race progressed, streamlining, from a form resistance perspective, improved primarily due to increased hip rotation. However, this interpretation must be made against the observation that KP frequently "surges" in races, such actions being accompanied by more intense (i.e., six-beat) kicking. The first 40 m illustration contains a six-beat kicking action while the 1440 m illustration contains a two-beat action.

These two "differences" in the analysis suggest that when "harder" kicking is performed, and in particular four-beat kicking, the hips do not rotate as much as with a two-beat kick. Thus, it is possible that "hard" kicking may increase the potential of the arms to generate extra force but, by restricting hip rotation, may also increase the cross-sectional area of the swimmer's body and thus, increase form resistance. When that is added to the increased form and wave resistance that results from bigger leg movements (see Figure 1), the scope of kicking movements may have to be a feature of swimmer streamlining that has to be attended to more than has been done in the past. If this is not considered, the gains from harder kicking may be offset by losses in streamline. However, given the data available, it is not possible to state exact ranges of tolerance for this factor.

Shoulder-hip roll. Figure 1 provides some perspective of the similarities of shoulder and hip rolls in each S. The greater the shoulder and hip rotations, the longer are their respective lines in the illustrations. KP2 has a greater similarity in shoulder and hip roll for the entire stroking sequence than in any S.

ES and AP have hip movements that do not follow shoulder movements. Some pictures have the hips rotating one way while the shoulders go the other (e.g., ES #1; AP #1, #2, and #3). At those times, form resistance is exaggerated. AP displays the largest foot movement range of all Ss. With such pronounced kicking, the hips remain horizontal or have a twist (#1) similar to that of ES. JH also has shoulder rotation that is not matched by hip roll. The asymmetry of the shoulder rotation does not allow much to be concluded.

All swimmers rolled their shoulders considerably. Knowing that KP rolled at least 40 degrees, when the angle of the shoulder lines are compared to the other Ss in Figure 1, or the body exposure area is considered, it can be seen that all Ss rolled the shoulders as much as, if not more than KP. Flat shoulders positions are not evidenced. The concept of planing over the surface of the water with a flat chest and shoulders is not supported by these observations. Rather, it seems that a roll in the vicinity of 45 degrees to either side would be preferable.

When swimmers kick hard there is a strong tendency for the hips to remain flat even though the shoulders rotate (see ES and JH). That is how kicks are practiced on a kicking board. If flat kicking actions were altered and combined to match the shoulder roll, the same amount of propulsion and/or counter-balancing would result but streamlining would be enhanced. It would seem to be a reasonable and improved skill practice activity to kick with the hips rotating symmetrically and rhythmically, probably without using a restrictive kicking board.

It is proposed that swimmers be taught to rotate the hips as much as the shoulders. Propulsive forces will not depreciate even though the kick may be performed to the "side." It is the horizontal propulsive force of the kick that is important and that will occur irrespective of the rotational position of the kicking action.

Shoulder to upper arm angle. One cause of shoulder problems in swimmers is dominant internal rotator activation in the crawl stroke (10). Figure 4 illustrates angular relationships between the upper arm and the shoulder line. In the left example, the internal rotator muscles (anterior deltoid, pectoralis major, latissimus dorsi) are stronger and activated more than the external rotators (infraspinatus, teres minor, supraspinatus). The excessive force subluxes the greater head of the humerus so that it contacts the glenoid labrum of the scapula. With excessive repetitions, the external rotators weaken and the glenoid labrum eventually becomes irritated and/or damaged. If the angle of the shoulders and that of the upper portion of the pulling arm are not aligned, shoulder problems could develop. The angular position at the shoulder: (a) places more reliance on the internal rotator muscles to produce force and (b) reduces the effectiveness and function of the external rotators. As training progresses, the internal rotators strengthen and the external rotators weaken, exacerbating the problem further.

Figure 4

Figure 4. Two alignments of the upper arm and shoulders in the crawl stroke. In the left figure the shaded portion of the circled shoulder joint indicates that the internal rotator (IR) muscles are activated more than the external rotators (ER) resulting in a minor inward movement of the greater head of the humerus to irritate the glenoid labrum. In the right figure, a position similar to that displayed by KP2, the alignment of the humerus and the shoulder rotation allows both IR and ER muscle groups to function in "balance" so that the shoulder joint works correctly, produces no irritation, and more muscles are used to produce propulsive force.

With regard to propulsion, an angular change at the shoulder joint that favors one group of muscles over the other means that fewer muscles are used to generate and sustain force. Those fewer muscles will fatigue easier than if more muscles were used to generate the same amount of force. Thus, not only is a swimmer threatened by injury with an angled shoulder pull, the mechanical and endurance properties of the resulting action also will be degraded.

A more desirable relationship between the upper arm and shoulder is displayed by KP2. The right example in Figure 4 illustrates a KP2 arm-shoulder alignment. Both internal and external rotator muscles work to adduct the arm in the central part of the pull. That adduction allows the head of the humerus to nestle correctly in the glenoid fossa. Mechanically, the swimmer is advantaged by being able to use both internal and external rotator muscles to generate propulsive force. When more muscles can be used to generate force, the load is distributed across a greater number of muscles, consequently allowing each muscle to do less work and thus, maintain a greater endurance capacity in a swimming performance. That is one of the technique strengths of KP.

The desirable technique feature of the upper arm, when combined with shoulder rotation, is to have the two aligned. Usually, the recovering arm should also be aligned in the manner displayed in Figure 3. This could be an important technique criterion for effective crawl stroke swimming.


The nature of perfect streamlining is purely theoretical. Much of what Counsilman (3, pp. 20-25) proposed is still pertinent today. One could assert the following as being of "obvious" advantage: (a) reduce resistance, particularly in non-propulsive body segments; and (b) remove unnecessary movements. The greatest caution when offering opinions has to be that observed movements that create resistance may well be reactions to other movements which are beneficial, the result being that the "gains" outweigh the costs.

It could be argued that one should strive to have the least cross-sectional area of the body as the maximum frontal aspect. To achieve that, the head would have to be down inside that aspect with the hips and legs trailing in its "shadow" and exhibiting similar rotation to the shoulders. That position would minimize form resistance and would place the total head-body-legs posture on the water surface at the level supported by natural buoyancy and where resistance is least. Characteristics of good streamlining in this position would be as follows:

A flat, perfectly aligned position floating buoyantly, would result in minimal passive drag. When actions are added, active drag occurs. When exaggerated unproductive actions (e.g., vertical oscillations of the body and head) are exhibited, active drag is increased and propulsive economy reduced. No Ss in this study displayed these characteristics. this suggests all Ss might have room for significant technique improvements.

KP1 could be considered to be the most streamlined of all Ss. The attainment and maintenance of a very low body angle with legs trailing in the shadow of the body cross-section suggests a very low level of form resistance. However, the lowest body angle of 3o was only maintained for a brief period. All other swimmers reflected less streamlining, that is they proceeded through the water at varying angles that increased wave and form resistances.

It would seem to be prudent to experiment with teaching swimmers to strive for perfect streamlining. For crawl stroke streamlining, the back of the head, upper back, buttocks, and heels at the top of the kick would all have to be on the surface and possibly, partly exposed. That would require a straight, rather than hyperextended neck, and a naturally curved lumbar region rather than one that is exaggerated as a result of tilting the body angle and raising the head to look forward. Few swimmers demonstrate such a position. However, in some parts of a race, coaches actually instruct athletes to perform some of those postural characteristics. For example, in order to have a swimmer "go faster" as the finish is approached, swimmers commonly are told to "bury your head and do not breathe." The result of that is a straightening of the neck, the hips rising in reaction, and overall form resistance being reduced. One has to ask, if it is advantageous to alter the swimming posture in that manner to increase velocity when approaching a finish, why would it not be advantageous to do that for the total race? If it was done for the total event, then the swimmer would enjoy the benefits of the fast-finishing action for the full duration.

However, the counter theoretical argument to that of the previous paragraph has to be recognized (4). A small positive body angle could generate some vertical lift which might be beneficial in lowering overall frictional drag since more of the body would move through air than water. This argument may not be true, but since total resistance is an optimization matter involving several forms of drag, the true answer is not known. Until definitive answers are provided, the maximum streamlining approach is cautiously advised.

Meticulous streamlining is an area of swimming technique that should be thoroughly tested for it has a very large effect upon movement economy. It is possible that no swimmer has yet been observed to maximize streamlining. The relationship between perfect streamlining and ability to produce propulsive forces has not been measured. Attempts to improve streamline should be measured against performance changes to determine their worth. It is an avenue for producing performance improvement that does not require extra physical effort.


The analysis of body kinematics of champion crawl stroke swimmers is striking by the extent of elements that could be improved rather than by desirable characteristics that should be copied. It is proposed that given the performance standard of the sample of swimmers used, that there is still much improvement possible in crawl stroke events. It is likely that significant improvements in world records will continue, particularly if aspects of streamlining are attended to by the world's best swimmers.

These authors wish to reiterate the tentative nature of this analysis. Although bold interpretations and suggestions have been made, they are tentative. Essentially, they are hypothetical explanations that could serve as research topics but may also be cautiously "tested" in real-world settings. Until objectively verified conclusions are generated, the content and expression of this study must be guarded. Despite that caveat, an attempt has been made to extend these observations so that they have some practical, albeit tentative, implications.

Theoretically, what has been suggested could result in benefits. However, when contemplating changes in technique for older, more experienced swimmers, coaches should ask themselves whether major changes will be beneficial to the swimmers in the long-run, particularly if they have to spend very extended periods becoming comfortable with a different technique. The value of this analysis has been more of what not to do rather than substantiation of what to do. Possible problems have been identified and suggestions for their improvements have been offered.

All Ss appear to be able to improve streamlining. The magnitude of those possibilities is greater in some than in others. Beneficial changes could be derived from any or all of the following.

  1. Keeping the body horizontal and not raising the shoulders in concert with arm entries. The body should only be rotated on a horizontal, longitudinal axis.
  2. Keeping the head down, looking at the bottom and then to the side of the pool when breathing, so that rotation on a horizontal, longitudinal axis is achieved. It is detrimental to streamlining to look forward and/or to raise the head at any time, particularly when breathing.
  3. Keeping the legs and kick within the body shadow so that form resistance is minimized. The only time the legs should emerge from the position of maximum streamline is when detrimental effects are more than offset by arm-propulsion gains.
  4. Reducing downward pressure after the arms enter the water and achieving a horizontal pull component earlier. Bending at the elbow, immediately after entry, would assist in faster directional force change. If this could be achieved, there is every likelihood that streamlining will be better maintained.
  5. Rotating the hips and shoulders together to the same degree in a streamlined position. This will reduce form resistance.
  6. Allowing the kick to rotate on a longitudinal axis in concert with hip rotation. A concentration on vertical kicking increases form resistance whereas rotation has the potential to decrease it without any loss in forward velocity.
  7. Aligning the upper arm of the pull and recovery with the line of the shoulders to increase the musculature used to generate force as well as maintain the integrity of the shoulder joint. An emphasis on, and tolerance of, internal rotator muscle group activity will eventually cause problems. This proposal will introduce a greater role for the external rotator muscles during the pull.
  8. Maintaining streamlined feet when kicking so that drag is not created. However, it must be realized that a "braking" action may be a corrective movement for some other error.


  1. Busch, F. How they train: Chad Carvin. Swimming World and Junior Swimmer, 35(9), 33, 1994.
  2. Cappaert, J., and B. S. Rushall. Biomechanical analyses of champion swimmers. Sports Science Associates, Spring Valley, California, 1994.
  3. Counsilman, J. E. The science of swimming. Prentice-Hall, Englewood Cliffs, New Jersey, 1968.
  4. Lewin, G. Swimming. Sportverlag, Berlin, Germany, 1979.
  5. McAllister, B., R. Quick, and G. Troy. How they train . . . an overview of the training programs of: Janet Evans, Kristine Quance, Jenny Thompson, Lea Loveless, Trina Jackson. American Swimming, 4-11, August/September, 1994.
  6. Montpetit, R., A. Duvallet, J. P. Serveth, and G. Cazorla. Stability of VO2max during a 3-month intensive training period in elite swimmers. Paper presented at the Annual Meeting of the Canadian Association of Sport Sciences, Halifax, Nova Scotia, Canada, 1981.
  7. Rushall, B. S., L. E. Holt, E. J. Sprigings, and J. Cappaert. A re-evaluation of forces in swimming. Journal of Swimming Research, 10(fall), 6-30, 1994.
  8. Rushall, B. S., E. J. Sprigings, L. E. Holt, and P. R. Francis. Forces in swimming--current status. NSWIMMING Coaching Science Bulletin, 2(4), 1-25, 1994.
  9. Sheehan, D. P., and D. M. Laughrin. Device for quantitative measurements of hydrodynamic drag on swimmers. Journal of Swimming Research, 8, 30-33, 1992.
  10. Troup, J. P. (Ed.). (1992). International Center for Aquatic Research Annual: Studies by the International Center for Aquatic Research 1991-92. US Swimming Press, Colorado Springs, Colorado: US Swimming Press, 1992.

Return to Table of Contents for Swimming Science Bulletin.