Race strategies to maintain the highest swimming speed (V in meters per second) are commonly analyzed by observing the changes in the stroke parameter values, that is, stroke length (SL in meters) and stroke rate (SR in hertz) (5,7,8,17,26), in which V = SL × SR. It is assumed that SR and SL reflect the effect of motor processes responsible for the generation of propulsive and resistive impulses and the way they are coordinated. An impulse (N·s) corresponds to the integral of a force with respect to time. Recently, Sanders (20) proposed to take into account these impulses in the analysis arguing that concentration only on SL and SR distracted analysts from considering factors that might explain differences in performance.
The swimmers can generate the same propulsive impulse by varying the amplitude of force and the duration of the application thereof such that the impulse remains the same. At constant speed, this propulsive impulse should balance the resistive impulse defined by the drag-time integral (28). The analysis of this ensemble could be useful to better understand the SR-SL values observed for a swimmer when swimming at a specific speed.
As long as the speed is on average constant, average propulsion and drag are in balance. Considering the impulse, this assertion thus assumes that the resistive impulse per stroke cycle (I−/stroke) equals the propulsive one (I+/stroke):
The resistive impulse per stroke cycle could be estimated by the following formula:
where I− corresponds to the resistive impulse, Fdrag to the cycle averaged drag force, and Tstroke to the stroke cycle duration (s).
Two sources of propulsion can be used, the arm stroke and the leg kick. Because the direct contribution to overall propulsion of the leg kick is considered to be small (estimated to 10% (32)), the present analysis will focus on the impulses generated by the arms.
The propulsive impulse the swimmer has to produce from the arms corresponds to
where I+r and I+l are propulsive impulses yielded by left and right arm, respectively, F (N) is the average force, pullr and pushr and pulll and pushl represent the durations (s) of the pull and push phases of the right and the left arm, respectively.
By combining equations 5 and 6, one obtains
To overcome Fdrag, the swimmer can act on three parameters, that is, the propulsive force magnitude, the duration over which the propulsive force is applied, and the rate of application of these impulses. However, the equation at this step conceals the possibility the swimmer has to overlap the application of the impulse of left and right arms. The index of coordination (IdC) characterizes the coordination of one arm action to the other by measuring the lag time (s) between the propulsive phases of each arm. IdC is expressed in percentage of the mean duration of the stroke (for the complete calculation methodology, see Chollet et al. (4)). When IdC equals 0, the stroke coordination is called "opposition." If so, the time at which one arm begins the pull phase coincides with the end of the push phase of the other arm. When there is a nonpropulsive gap between the propulsive phases of each arm, IdC is <0, and the stroke coordination is called "catch-up." When there is an overlap between the propulsive phases of the two arms, IdC >0, and the arms' coordination is said to be in superposition. Hence, the absolute time allotted to propulsion (Tprop) within a stroke cycle is also dependent on the IdC according to
IdC represents the average lag time between two consecutives arm propulsive phases relative to the stroke cycle duration. Because there are two lag times within a stroke cycle (one between the end of push of one arm and begin of pull of the other arm and vice versa), the IdC value has to be multiplied by 2 to obtain the amount of lag time within a stroke cycle.
Combining equations 5 and 6, one obtains
Thus, by varying the intrastroke time allotted to propulsion, the arm coordination (IdC) would change and represent an additional means to overcome the required drag force.
An important factor-limiting performance is the occurrence of fatigue. Fatigue is defined as an exercise-induced reduction in the maximal capacity to generate power output (10,30). During all out exercise, the power output generated by the swimmer decreases and the swimming speed decrease as recently demonstrated by Toussaint et al. (28) during a 100-m front crawl race. In other so-called time to exhaustion tests, the effect of fatigue is studied by imposing a fixed constant speed on the swimmer, and the swimmer is instructed to swim as long as possible, that is, until total exhaustion. Exhaustion has been defined as an inability to sustain exercise at the required intensity (30). In swimming, this corresponds to the inability to sustain the imposed pace. During such constant speed swims, a progressive increase in the SR value with a concomitant decrease in SL has been observed to keep the imposed pace (2,9,16,31). Such adaptations have been attributed to fatigue development, a tired swimmer being gradually unable to produce the force necessary to overcome resistance to forward movement. Further analysis revealed that the percentage of time devoted to propulsion within the stroke cycle is increased. This most probably leads to a more even impulse production per stroke cycle due to a reduced lag time between propulsion from the left and the right arm (2). Hence, the fatigue seems to affect the way the required speed (and so power output) is produced. Apparently, this influences the generation of the propulsive impulse too, but how (and why) this is so remains unclear.
Indeed, the cause of the increase of the relative time for propulsion within the stroke cycle as fatigue develops is difficult to understand. Data on the changes of the absolute duration of these phases are nonexistent because studies based on this model (4) always expressed the duration of these phases relative to the stroke cycle duration, that is, in percentage of cycle time (1,2,4,21-25). This raises the question whether the absolute time of the propulsive phase is lengthened or whether the nonpropulsive phases of the stroke are shortened, both yielding a similar relative result when the absolute stroke time is shortened (i.e., stroke frequency increases). Consideration of the absolute durations of the stroke phases would allow a better understanding of how the swimmer adapts to fatigue, and thus how the observed SR and SL maintain a pace reflects the adapted coordination to the progressive fatigue development. Moreover, because the change in the duration of the propulsive impulse with fatigue development remains unknown, it seems to be useful to estimate the time allotted by the swimmer to propulsion per lap (Tprop/distance). This can be estimated as follows:
where D/SL corresponds to the number of stroke cycles needed to cover the distance.
What is calculated is the total time allotted to propulsion (including overlap of left and right arm) per stroke cycle multiplied by the number of strokes per lap. This formula incorporates the distance covered during streamlining, which could lead to an underestimation of the number of stroke cycles actually needed to cover a certain distance. However, this overestimation seems minimal as Chollet and Pelayo (6) showed that the inclusion of the turn effect in the calculation of SL overestimate the true SL value by only 2.6% on an all-out 200-m front crawl stroke.
Several strategies are available for the swimmer to keep a determined speed during paced exercises. Because such exercises are currently used during training sessions, data on strategies adopted by the swimmers to manage their propulsive impulses as fatigue develops could provide a better understanding of the observed changes in stroke technique. The aim of this study is to provide a first explanation to the changes in the stroking parameters value during paced front crawl swimming. The study of changes in the duration of the different stroke phases and in the coordination of left and right arm propulsive action during front crawl swimming will be considered as convenient mean for this purpose.
MATERIALS AND METHODS
Ten well-trained swimmers (20.3 ± 1.7 yr; two females and eight males) volunteered for this study. Height, body mass, and arm span mean values were 180 ± 6 and 170 ± 7 cm, 72.8 ± 5.3 and 61.5 ± 5.3 kg, and 187 ± 5 and 172 ± 7 cm, respectively, for males and females. They trained for 12 ± 2 yr, at a frequency of 6 ± 2 training sessions per week during the study. Their performances in the 400-m front crawl stroke were 280.23 ± 13.73 and 322.73 ± 12.47 s, which correspond to a mean speed that represents 76.8% ± 3.7% and 73.3% ± 2.7% of the mean speed of the short course pool world record for men and women, respectively. They were informed of the procedure, the potential risks involved, and the benefits of the study. They gave a written consent to participate and then underwent a complete medical examination. The protocol has been reviewed and approved by the consultative committee for the protection of human subjects in biomedical research of the Nord-Pas-de-Calais (France) before the start of the study. During the testing period, the subjects maintained their training programs but adapted the intensity and the total volume of training consequently.
The tests were performed in a 25-m indoor swimming pool, using the front crawl stroke. The start of each trial took place in the water. Environmental conditions were standardized with water temperature fixed at 29°C. The subjects performed all the trials with at least 24 h of rest between each test to minimize fatigue effects. To minimize possible influence of the circadian rhythm, subjects performed their trials at the same time of day (3). They were also asked to maintain their normal diet throughout the study. For the day before each trial, subjects were asked to refrain from the ingestion of alcohol and caffeine.
All swimmers started the study with a 400-m maximal swim of which the mean velocity (V400) was calculated.
Subjects were then required to swim three times to exhaustion (TTE in seconds) at predetermined speeds, which corresponded to 95%, 100%, and 110% of the previously recorded V400 (TTE95, TTE100, and TTE110). Speeds were imposed by two operators walking on each side of the pool at the prescribed pace. Markers spaced by 5 m were laid out on the pool deck, and the corresponding split times were provided to the operators. The swimmer was asked to maintain pace by keeping their feet at the level of the operator walking on the deck. When the operator's feet caught the swimmer's head, the test was stopped.
The swimmers were filmed during all tests by two cameras; one above the surface of the water (JVC Mini DV GR-DVX 407EG operating at 25 Hz and contained in a box JVC WR-DV96) and one below (Sony Mini DV DCR-HC40E operating at 25 Hz and contained in a waterproof Sony HandyCam MPK-DVF6) carried by a trolley and pushed by an operator. This apparatus allows the analysis of each stroke cycle in a sagittal plane. Views of both cameras were postsynchronized with a visual signal that was visible on recordings for both cameras. DartTrainer software (Version 4.0.5 DartFish TeamPro) was used for the video analysis enabling the analysis of each stroke with an accuracy of 0.02 s.
All the TTE tests were realized in a randomized order to avoid a systematic (fatigue) effect on the results. The swimmers were asked to carry out their own warm-up and to repeat this warm-up before each TTE. Before the actual test, swimmers swam at the prescribed speed for a few lengths to get accustomed to the required pace. The subjects were encouraged during each TTE test to perform as well as possible.
As time to exhaustion varies from one swimmer to another, the duration of each TTE test was normalized to 100%. For the analysis of the changes in the variables, the TTE was split into six parts corresponding to 0%, 20%, 40%, 60%, 80%, and 100% of the TTE. For each part, the first stroke cycle after the push off and the last stroke of each length was not considered to avoid the influence of turns. At 0%, the first three stroke cycles have not been considered for analysis because swimmers did not instantly catch up with the right pace. To ensure that the studied technical changes occur during the imposed speed, the end of the time to exhaustion test has been determined in two steps; first, when the operator's feet caught the swimmer's head and the test was stopped. Second, video analysis was used to obtain the duration of the test. The last pool length included in the analysis was the one preceding two consecutive pool length swam at a speed lower than 97.5% of the required pace. This threshold of 2.5% has been chosen according to the methodology of Seifert et al. (21,22).
Stroking parameters analysis.
During the entire duration of each TTE test, the stroking parameters were measured using a PC software developed in our laboratory. Using that software, the clean swimming time per pool length (s) was measured. The clean swimming speed (V in meters per second) was then calculated. The stroke rate (SR in hertz) was measured cycle by cycle. Finally, the stroke length (SL in meters) was calculated as the ratio between V and SR. For each part, the corresponding values of SR, SL, and V were recorded.
The stroke cycle was broken down into four distinct phases as defined by Chollet et al. (4):
Phase A: entry and catch. This phase corresponded to the time from the hand's entry into the water to the beginning of its backward movement.
Phase B: pull. This phase ends at the hand's arrival in the vertical plane below the shoulder.
Phase C: push. This phase ends with the release of the hand from the water.
Phase D: recovery. This phase ends with a new entry of the hand into the water.
Phases B and C were assumed to be propulsive, and consequently, phases A and D were the nonpropulsive ones. Their durations were expressed in absolute time (A, B, C, and D in seconds). The mean duration of a complete arm movement, defined as the sum of the four distinct phases (A + B + C + D), was calculated. For each part of the TTE test (at 0%, 20%, 40%, 60%, 80%, and 100%), three consecutive arm stroke cycles were analyzed.
The beginning (start of phase B) and the end of the propulsive action (end of phase C) of each arm were determined. These data were used to calculate the index of coordination (IdC).
The video analysis allows also the quantification of the number of leg kicks realized in each stroke cycle. Three main modalities have been detected; two, four, and six leg beat kicks.
Means and SD (s) were used to represent the average and scatter of the studied variables. The normal Gaussian distribution of the data was verified by the Shapiro-Wilk's test. The compound symmetry, or sphericity, was checked using the Mauchley test (34). When the assumption of sphericity was not met, the significance of F-ratios was adjusted according to the Greenhouse-Geisser procedure when the epsilon correction factor was <0.75 or according to the Huyn-Feld procedure when the epsilon correction factor was >0.75 (29). A one-way ANOVA for repeated measures was used to compare the mean values for each variable at each point of analysis (0%, 20%, 40%, 60%, 80%, and 100% of each TTE).
To detect significant differences, post hoc tests were used. If the epsilon correction factor was <0.5, the Bonferroni post hoc test was used. If the epsilon correction factor was >0.5, the Tukey post hoc test was used (27).
Correlations between the variables that differed significantly have been calculated according to the following process. Variations of the variables between 0% and 100% of each TTE test have been calculated for each swimmer according to the following formula:
where Δ is the variation of the variable, and M1 and M2 are the values of the variable at 0% and 100% of the TTE test, respectively. The variation of the IdC values between both conditions has been calculated by subtracting the value at 0% from the value at 100% of the TTE test. This method is justified by the fact that huge Δ values can be obtained when the denominator is near 0 as is sometime the case for the IdC value. These Δ values have been correlated to detect if any variation of one variable during the TTE test was related to variations of other variables. When normality of the distribution was met, a Bravais-Pearson coefficient of correlation was used; otherwise, a Spearman coefficient of correlation has been used.
Finally, a chi-squared test has been used to detect any significant changes in the leg kick.
The threshold for significance was set at the 0.05 level of confidence. For the statistical analysis, STATISTICA 6.0 for PC (Statsoft®, Tulsa, OK) was used.
The durations of the TTE tests were 670 ± 117, 238 ± 43, and 68 ± 14 s for TTE95, TTE100, and TTE110, respectively.
Figure 1 represents the changes in the stroke parameters, the temporal structure of the stroke cycle, and the arm coordination in each TTE test.
A progressive increase of the SR was associated with a decrease in SL for each TTE test. For SR, the increases from the beginning to the end of the TTE test were 10.5% ± 8.3%, 8.7% ± 6.5%, and 2.7% ± 4.8% for TTE95, TTE100, and TTE110, respectively. The decreases from the beginning to the end of the TTE test for SL were −9.7% ± 7.5%, −7.7% ± 4.7%, and −4.0% ± 3.1% for TTE95, TTE100, and TTE110, respectively. The recovery phase (D) was significantly shortened (−12.6% ± 9.9%, −12.9% ± 5.7%, and −5.7% ± 2.7% from the beginning to the end for TTE95, TTE100, and TTE110, respectively), whereas the entry + glide phase (A) decreased significantly for TTE95 and TTE100 (−17.9% ± 14.1% and −15.0% ± 9.6%, respectively). For TTE110, phase A decreased nonsignificantly (−5.3% ± 10.5%).
The time allotted to the propulsive phases changed differently during the TTE depending on the required pace. No significant variations of B and C have been observed between the start and the end of TTE95 and TTE110 (P > 0.05). For TTE100, a significant decrease in the push phase (C) is observed from the beginning to the end of the test (−7.9% ± 5.7%; P < 0.05).
Finally, the index of coordination (IdC) and the time allotted to propulsion per pool length (Tprop) increased throughout each TTE test (P < 0.05). Variations for these parameters were 4.6 ± 1.9, 2.8 ± 2.3, and 2 ± 1.6 U (%) for IdC and 11.0% ± 6.6%, 4.6% ± 5.9%, and 7.5% ± 4.8% for Tprop for TTE95, TTE100, and TTE110, respectively.
The change in SR (ΔSR) was strongly negatively correlated to the variation of the duration of the glide + catch phase (A) for the each TTE test (Table 1). ΔSR was also significantly correlated with phase (D) except for TTE95 (P < 0.05). Given that changes in SR are compensated by variations in SL to keep V constant, it is no surprise that similar albeit slightly less strong correlations have been observed between ΔSL and all other variables previously considered. No significant relationship has have been found between variation of the stroking parameters and those of ΔIdC and ΔTprop.
The type of kick (two, four, or six beat kick) adopted at the start of each TTE represented 20%, 50%, and 30% for TTE95, 20%, 30%, and 50% for TTE100, and 10%, 30%, and 60% for TTE110. No significant change with respect to the end of the TTE in kicking was observed: 20%, 40%, and 40% for TTE95, 20%, 30%, and 50% for TTE100, and 10%, 20%, and 70% for TTE110.
Most of the studies that analyze the effect of the exercise duration on performance during all-out distance trials reported a decrease in speed (5,7,8,17,26), which can be attributed to a decrease in the power output generating capacity (28). In the latter study, a decrease of 24% in the power output during an all-out 100-m swim was observed. In the present study, a constant speed was imposed on the swimmer enforcing a constancy of the required propulsive impulse to overcome the corresponding drag impulse. By varying (within limits) the magnitude of the generated propulsion during the stroke cycle, the swimmer has some freedom of choosing the combination of SR and IdC. The swimmer has freedom to reduce the lag time between the left and the right arm propulsion with the option to overlap these impulses. The development of fatigue will progressively limit the capacity to vary the magnitude of the propulsive force within the stroke cycle with consequent limited liberty in apportioning the amount of impulse that can be applied per stroke. With the progress of all TTE tests, an increase of SR (with a concomitant decrease in SL) was observed as well as an increase in the IdC.
The observed SR-SL changes are in accordance with those described in the literature in which effects of fatigue on stroke technique are studied (9,11-13,16,31,33). Our results showed that speed is maintained by an increase in SR as reported previously. It is interesting that the absolute time the swimmer allots to generating propulsion during each (half) stroke cycle did not vary significantly. Given that the IdC increased implying that propulsive impulses from left and right hand were closer spaced in time or even overlapped (see Fig. 2), the total time propulsion is generated per distance unit increased. Because impulse is the product of time and force magnitude, the conclusion is that the changes in the coordination resulted in a reduction of the average magnitude of the propulsive force per stroke with the development of fatigue (while of course keeping the average propulsion equal).
In rowing, Peltonen and Rusko (18) observed a decrease in the average force produced, whereas the maximal force remains more or less constant throughout exercises of 6- and 12-min durations. The duration over which the force acted did not differ significantly. At the same time, the rowers maintained relatively constant power per stroke levels throughout the tests or even increased it at the end of the 6- and 12-min tests by compensating the average force decrease by a greater increase in the stroke rate (from 42 to 45 cycles per minute and from 37 to 44 cycles per minute for the 6- and 12-min test, respectively). This strategy seems similar to that observed in the present study.
Toussaint et al. (28) suggested that SR is a scaling factor to apportion the propulsive force to the reduced power output with consequent lower speed during an all-out 100-m front crawl stroke race. In essence, the reduced SR was related to the reduced capacity of propulsion production, so that the stroke technique remains more or less unaffected. In the present study, the change in SR plays a similar role in the sense that SR is increased to solve the problem of generating adequate propulsion while the development of fatigue leads to a reduced force generating capacity. Consequently, the observed increase of SR is mainly made through a significant reduction in time allotted to the nonpropulsive part of the arm stroke cycle. Indeed, significant and high coefficients of correlation between variations in SR and those for ΔA and ΔD are observed. However, the duration of the cycle in which propulsion was generated did not change except for TTE100 where the push phase (C) declined. Rather, swimmers preferred to decrease the glide + catch and recovery phases such that the gaps between propulsive actions decreased or the overlap increased (Fig. 2). One consequence is the increase of the IdC value. The subsequent better chain of the propulsive impulses allows an increase of the number of impulses per pool length, which lead to a greater time allotted for propulsion as illustrated by the increase of the Tprop even for TTE100 in which propulsive phase C declined. This behavior suggests a compensation for the decline of the force generating capacity by elongating the time the propulsive impulse is produced so that the propulsive force-integral over a pool length is constant. Some swimmers showed a positive IdC value at the end of the TTE test, which corresponds to a partial overlap of their propelling actions. Such a behavior leads to a summation of the propulsive force of the two arms during the overlap so that the balance between propulsive and resistive impulse could be maintained whereas the propulsive force per arm was insufficient to do so. These motor adaptations contribute to the maintenance of the speed investigated till exhaustion. The exhaustion is associated with the point where another increase in SR, and so a better chain of propulsive impulses and an even partial overlap of them, became insufficient to compensate for the decrease of the force generated per arm stroke. The averaged balance between propulsion and drag is thus broken and the speed declines, which leads to the end of the TTE test. The decline of the force generating capacity, suggested by our results, has already been observed by Monteil et al. (15) during an all-out 400-yd front crawl stroke.
It should be noted that an individual fixed drag-velocity relationship is assumed implying that the drag to overcome at the start of the TTE is assumed equal to that at the end provided the speed did not change. However, it could be possible that fatigue also affects the technique such that drag increases for example due to less streamlining as suggested by Craig et al. (7). If so, it raises the question what aspects of the changes in technique are effective in supporting the generation of propulsion and what are counterproductive by increasing drag.
It is interesting to observe that there is no correlation between ΔSR with ΔIdC and between ΔSR and ΔTprop. It reflects that some swimmers of our experimental sample showed larger variation in IdC for changes observed in SR than others. This result suggests that the IdC could reflect a separate aspect from SR to solve the problem of generating an adequate propulsive impulse with ongoing fatigue development. As previously evocated in rowing, the rowers have only the solution of increasing the stroke rate to maintain the required power output as two arms act in the same time. Our results suggest that overlap of the propulsive impulse is available for the swimmers as an additional strategy to the increase in SR, that is, change their chain of propulsive actions with the choice to make them overlap. In rowing, the arms act in concert as if there is only one engine, whereas swimmers have to manage two main engines (i.e., their two arms) by varying both the time that separate two consecutive impulses of the same arm (which refers mainly to an increase of SR) but also the lag time between consecutive propulsive actions of both arms (which mainly refers to the increase of the IdC). This freedom to overlap impulses of the left and the right arm stroke gives additional room for compensating a reduced force producing capacity of each arm. Such a suggestion does not mean that SR and IdC are completely separate means to solve the problem of the ongoing fatigue development, but it suggests that swimmers could favor one of the solutions or both as no strong correlation have been observed between IdC and SR. Flexibility in the motor organization to deal with the task constraint could be one explanation (25). However, such a hypothesis needs further study to better understand the underlying process, which could justify such strategies of motor adaptation (increase SR rather than IdC and vice versa) in swimming conditions with ongoing fatigue development.
The leg kicks are considered to indirectly contribute to propulsion as they stabilize and streamline the swimmer's body (32). For all tested speeds, no change in kicking has been observed, implying that the swimmers did not use their legs more when getting more tired. This strategy could be justified given that for a unit propulsion, the energy costs of the kick are much higher compared with that of the arm action (19). Yanai (35) observed a perfect match between the intracycle change of the external fluid force yielded by the leg beat and the intracycle change of the moment of inertia of the body. The constant number of leg kicks per stroke cycle observed in the present study could be also related to the necessity to conserve this moment of inertia balance thereby avoiding an increase in drag magnitude, which would finally fit with the assumption previously made by Watkins and Gordon (32).
The extend in which the studied parameters varied to the ongoing fatigue development differs depending on the imposed speed. Craig and Pendergast (8) showed that the higher the speed, the lower the usable number of SL-SR combinations. This relationship could explain why in the present study the magnitude of changes in SR and SL decreased from the lowest to the highest speed. A similar trend in the rate of change for the entry + glide, recovery phases, and IdC depending on imposed speed has been observed. Thus, the usable number of SL-SR combinations could reflect the possible changes allowed for the intrastroke parameters for the maintenance of a required speed.
This study highlighted the intracycle changes in motor organization with the aim to better understand the observed compensatory mechanism between stroke length and stroke rate to keep an imposed pace. The results show that with fatigue in constant swimming speed tests, the SR and the IdC gradually increased, increasing the duration over which the propulsive force acted per distance unit as the force capacity was reduced. Finally, our results suggest that SR and IdC are two more or less separate strategies to solve the problem of generating an adequate propulsive impulse with ongoing fatigue development.
The authors thank the swimmers for their kind participation and JC Dusse for his help to conduct the experiment.
1. Alberty M, Sidney M, Huot-Marchand F, Hespel JM, Pelayo P. Intracyclic velocity variations and arm coordination during exhaustive exercise in front crawl stroke. Int J Sports Med
2. Alberty M, Potdevin F, Dekerle J, Pelayo P, Gorce P, Sidney M. Swimming technique changes during time to exhaustion at freely chosen and controlled stroke rate. J Sports Sci
3. Atkinson G, Reilly T. Circadian variation in sports performance. Sport Med
4. Chollet D, Chalies S, Chatard JC. A new index of coordination for the crawl: description and usefulness. Int J Sports Med
5. Chollet D, Pelayo P, Delaplace C, Tourny C, Sidney M. Stroking characteristic variations in the 100-m freestyle for male swimmers of differing skill. Percept Mot Skills
6. Chollet D, Pelayo P. Effects of different methodologies in calculating stroke length in swimming. J Hum Mov Stud
7. Craig AB, Skehan PL, Pawelczyk JA, Boomer WL. Velocity, stroke rate, and distance per stroke during elite swimming competition. Med Sci Sports Exerc
8. Craig AB, Pendergast DR. Relationships of stroke rate, distance per stroke and velocity in competitive swimming. Med Sci Sports Exerc
9. Dekerle J, Nesi X, Lefevre T, et al. Stroking parameters in front crawl swimming and maximal lactate steady state speed. Int J Sports Med
10. Green H. Mechanisms of muscle fatigue in intense exercise. Journal of Sports Sciences
11. Keskinen KL, Komi PV. Interaction between aerobic/anaerobic loading and biomechanical performance in freestyle swimming. In: Ungerechts BE, Reischle K, editors. Biomechanics and Medicine in Swimming. Swimming Science V
. Champaign (IL): Human Kinetics; 1988a. p. 285-93.
12. Keskinen KL, Komi PV. The stroking characteristics in four different exercises in free style swimming. In: de Groot G, Hollander AP, Huijing PA, van Ingen Schenau GJ, editors. Biomechanics XI-B (International Series on Biomechanics)
Amsterdam: Free University Press;1988b. p. 839-43.
13. Keskinen KL, Komi PV. Stroking characteristics of front crawl swimming during exercise. J Appl Biomech
14. Maglischo EW. Swimming Fastest. Champaign (IL): Human Kinetics; 2003. p. 792.
15. Monteil KM, Rouard AH. Free swimming versus paddles swimming in front crawl. J Hum Mov Stud
16. Ogita F, Onodera T, Tabata I. Effect of hand paddles on anaerobic energy release during supramaximal swimming. Med Sci Sports Exerc
17. Pai YC, Hay JG, Wilson BD. Stroking techniques of elite swimmers. J Sports Sci
18. Peltonnen J, Rusko H. Interrelations between power, force production and energy metabolism in maximal leg work using a modified rowing ergometer. J Sports Sci
19. Pendergast D, Zamparo P, di Prampero PE, et al. Energy balance of human locomotion in water. Eur J Appl Physiol
20. Sanders R. New analysis procedures for giving feedback to swimming coaching and swimmers. In: Gianikellis KE, Mason BR, Toussaint HM, Arellano R, Sanders RH, edtors. Scientific Proceedings-Applied Program, XXth International Symposium on Biomechanics in Sports-Swimming
. Caceres (Spain): University of Extramedura; 2002. p. 1-14.
21. Seifert L, Boulesteix L, Chollet D. Effect of gender on the adaptation of arm coordination in front crawl. Int J Sports Med
22. Seifert L, Chollet D, Bardy BG. Effect of swimming velocity on arm coordination in the front crawl: a dynamic analysis. J Sports Sci
23. Seifert L, Chollet D, Allard P. Arm coordination symmetry and breathing effect in front crawl. Hum Mov Sci
24. Seifert L, Boulesteix L, Carter M, Chollet D. The spatial-temporal and coordinative structures in elite male 100-m front crawl swimmers. Int J Sports Med
25. Seifert L, Chollet D, Rouard A. Swimming constraints and arm coordination. Hum Mov Sci
26. Sidney M, Delhaye B, Baillon M, Pelayo P. Stroke frequency evolution during 100-m and 200-m events front crawl swimming. In: Keskinen KL, Komi PV, Hollander AP, editors. Biomechanics and Medicine in Swimming
, Swimming Science VIII
. Jyväskylä (Finland): Gummerus Printing; 1999. p. 71-6.
27. Stevens J. Applied Multivariate Statistics for the Social Sciences
. 3rd Edition. Mahwah (NJ): Erlbaum; 1996. p. 712.
28. Toussaint HM, Carol A, Kranenborg H, Truijens MJ. Effect of fatigue on stroking characteristics in arm-only 100-m front-crawl race. Med Sci Sports Exerc
29. Vincent JW. Statistics in Kinesiology
. 2nd ed. Champaign (IL): Human Kinetics; 1999. p 194.
30. Vollestad N. Measurement of human muscle fatigue. J Neurosci Met
31. Wakayoshi K, D'Acquisto J, Cappaert JM, Troup JP. Relationship between metabolic parameters and stroking technique characteristics in front crawl. In: Troup JP, Hollander AP, Strass D, Trappe SW, Cappaert JM, Trappe TA, editors. Biomechanics and Medicine in Swimming
, Swimming Science VII
. London: E&FN Spon; 1996. p. 152-8.
32. Watkins J, Gordon AT. The effects of leg action on performance in the sprint front crawl stroke. In: Hollander P, Huijing P, Groot G, editors. Swimming Science IV
. Champaign (IL): Human Kinetics; 1983. p. 310-5.
33. Weiss M, Reischle K, Bouws N, Simon G, Weicker H. Relationship of blood lactate accumulation to stroke rate and distance per stroke in top female swimmers. In: Ungerechts BE, Reischle K, editors. Swimming Science V
. Champaign (IL): Human Kinetics; 1988. p. 295-303.
34. Winter EM, Eston RG, Lamb KL. Statistical analyses in the physiology of exercise and kinanthropometry. J Sports Sci
35. Yanai T. Stroke frequency in front crawl: its mechanical link to the fluid forces required in non-propulsive directions. J Biomech