Share this article on:

Arm and leg power output in swimmers during simulated swimming


Medicine & Science in Sports & Exercise: July 2000 - Volume 32 - Issue 7 - p 1288-1292
Applied Sciences: Biodynamics

SWAINE, I. L. Arm and leg power output in swimmers during simulated swimming. Med. Sci. Sports Exerc., Vol. 32, No. 7, pp. 1288–1292, 2000.

Purpose Previously, it has not been possible to compare power output of the arms and legs during simulated swimming using dry-land ergometry. The purpose of this study was to determine arm-pulling and leg-kicking power using isokinetic dry-land ergometry.

Methods Twenty-two highly trained male swimmers of mean (± SD) age, 23 ± 3.6 yr; body mass, 78 ± 5.9 kg; and stature, 1.79 ± 0.04 m were recruited to the study. First, subjects performed 10 s of all-out exercise at each of five resistance settings, with 1 h rest in between, to determine the best maximal pull velocity (MPVopt). Second, they performed an all-out 30-s test at MPVopt, which was repeated the following day. These repeated 30-s tests were performed separately using simulated front-crawl arm-pulling and leg-kicking, on a computer-interfaced swim bench and purpose-built leg-kicking ergometer. Peak and mean power output (PPO; MPO) were determined from regression analysis of the power vs time relationship.

Results The mean (± SEM) PPO for arms and legs were 304 ± 22 W versus 435 ± 36 W. For MPO, the means were 225 ± 31 W vs 312 ± 26 W, respectively. These values were attained at mean MPVopt of 2.5 ± 0.2 m·s1 for arms and 2.3 ± 0.4 m·s1 for legs. The variation in PPO from repeated testing was 7.3% for arms and 8.3% for legs.

Conclusions These results show that the legs can sustain greater power output than the arms during simulated swimming. Also, the intra-subject variation in measurement of power output is small using these dry-land ergometers. These methods of assessment might be useful in explaining swimming performance and in monitoring changes that take place during training.

Physiology of Exercise, De Montfort University Bedford, Bedford, UNITED KINGDOM

Submitted for publication March 1997.

Accepted for publication November 1997.

Address for correspondence: Ian L. Swaine, De Montfort University Bedford, 37, Lansdowne Rd., Bedford, MK40 2BZ, United Kingdom; E-mail:

Sport-specific assessment methods for power output of the arms and legs for swimming are poorly developed compared with other sports (1,23). Separate arm and leg power output measures would be useful in evaluating training programs and in understanding the importance of power output for swimming performance. The difficulties in assessing power output of the arms and legs during swimming arise from the absence of suitable transducers that can detect the force being exerted by the limbs. These difficulties are mainly due to continuous displacement of the water. Assessment of the power output of swimmers has therefore developed into two main methodological categories: water-based and dry-land ergometry.

The water-based methods have involved measurement of power using pressure pads (MAD system) (6), estimations of power output from cinematography (14), or determination of propulsive power from tethering devices (8). However, the MAD system cannot measure power output from the arms and legs separately. The accuracy of estimations from cinematographic techniques have been questioned (12). Furthermore, propulsive power, as measured in tethered swimming, differs significantly with power output of the limbs (21).

Because of the inadequacy of water-based techniques, attempts have been made to assess power output in swimmers using dry-land devices. These have included arm-cranking (4) and gym-based equipment (13). However, the most widely used dry-land device in swimming research has been the swim bench (13,15,20). In addition to being useful for measuring peak arm power, the swim bench is also used to assess critical power (16) and cardiopulmonary responses to incremental exercise (9,18). Although previously swim benches have only permitted measurements of arm movements and the specificity of the swim bench to swimming appears to be poor (11), this type of ergometer appears to be more suitable for swimmers than other ergometers.

Recently, it has become possible to measure power output during exercise that is similar to the front crawl leg-kicking action (17). An advantage is that the swimmer adopts a prone position on this ergometer similar to swimming. Also, it is possible to measure power during the upward and downward action of leg-kicking simultaneously. Because this leg-kicking device is an adaptation of the swim bench, it allows kicking-force, -distance, and -duration data to be stored and handled in a similar way to that used for arm-pulling (18).

Using cycle ergometry, accurate methods for determination of power output have been developed (1). These methods have been used to determine power output in athletes from various sports (19). The developments in cycle ergometry have included assessment of variation in measurement due to repeated testing (1). Procedures have been developed to optimize the force-velocity characteristics of the exercising musculature (10,22). However, none of these developments have been used in assessment of swimmers. Therefore, the purpose of this study was to establish reliable values for power output during all-out simulated front crawl arm-pulling or leg-kicking in swimmers using isokinetic dry-land ergometry.

Back to Top | Article Outline



Twenty-two male swimmers of (mean ± SD) age, 18.8 ± 3.2 yr; stature, 1.76 ± 0.05 m; and body mass, 61.7 ± 5.9 kg gave their written informed consent and were recruited to this study. All procedures were performed in accordance with the Declaration of Helsinki. The swimmers were involved in training sessions of 1.5 h on at least five occasions per week for at least a 1-yr period preceding the study. None was involved in any dry-land training using the swim bench.

Back to Top | Article Outline

Experimental Design

There were two parts to the exercise testing. The first was designed to allow selection of the best resistance setting on the arm-pulling and leg-kicking ergometers. The second part involved assessment of peak power output (PPO) and mean power output (MPO) during repeated 30-s all-out exercise tests using either the arms or legs.

Back to Top | Article Outline

Swim bench.

The swim bench (H. and M. Engineering, Gwent, UK) is represented graphically in Figure 1 a and has been described in detail previously (18). It consists of paddles attached to two pull-ropes that induce rotation of the isokinetic resistance devices. When force is applied to the hand paddles, the pull-rope pays out at a velocity which ranges up to a maximum. This has been termed maximal pull velocity (MPV; 15). The resistance unit can provide seven MPV settings. Power output was determined by use of a transducer unit, through which each pulley-rope passed. These transducers recorded tensile force, the distance through which the force was applied, and the duration of the applied force for each arm-pull logged at 100 Hz. Mean power output was calculated by a microcomputer using pulling-force, -distance, and -duration as logged throughout each arm stroke.

Figure 1

Figure 1

The calibration procedure for swim benches has been given in detail previously (15,18) and involves suspension of weights from the pull-ropes. When using the swim bench, subjects were lying in a prone position, with a strap around the torso and swim bench. The purpose of this strap was to prevent slippage of the torso on the bench during exercise. The front-crawl arm stroke movement was simulated by alternately pulling with each arm. During all testing swimmers were instructed to perform arm-pulling in a way similar to free swimming. Subjects were encouraged to maintain maximal stroke length at all times.

Back to Top | Article Outline

Leg-kicking ergometer.

The design of the leg-kicking device that allows adaptation of the swim bench has been given in detail (17). This device, represented in Figure 1b, was constructed to be used with the existing swim bench’s resistance and transducer units. The leg-kicking ergometer consisted of two inverted stirrups that held the feet during simulated front crawl leg-kicking against resistance (University of Warwick, Warwick, UK). In addition, two lightweight alloy wheels of 30-cm diameter, which could freely rotate, were positioned 1.25 m apart along a vertical steel structure. The outer rims of the two wheels were connected by wire cable so that they could rotate by 100° either clockwise or anti-clockwise in unison. The inverted steel stirrups were attached to the wire cable on each side.

By using pulleys, it was possible to redirect the pull-ropes from the swim bench vertically upward along the length of the wire cable on each side of the leg-kicking ergometer. These pull-ropes were attached to the lower part of each stirrup. When force was exerted either on the upstroke with the right foot or on the downstroke of the left foot it registered on the right pulley-rope and vice versa. This allowed the swimmer to generate power in an upward or downward during simulated front-crawl leg-kicking. The wire cables allowed lateral movement during kicking (15 cm). Because the pull-ropes passed through the swim bench’s transducer unit, computing of power output was similar to that for arm-pulling.

Back to Top | Article Outline

Selection of the best maximal pull velocity setting.

To find the best MPV setting, five 10-s all-out exercise tests were performed. These resistance settings gave maximal pull velocities (MPV) of 1.75, 2.0, 2.25, 2.5, and 2.75 m·s1. These five tests were administered in a randomized order and were performed on subsequent days. The optimum value was determined by quadratic curve-fitting procedures (10). The MPVopt was identified as that MPV value at which the highest peak power output (PPO) occurred.

Back to Top | Article Outline

Reproducibility of power output.

All swimmers performed two 30-s all-out tests using either the swim bench or the leg-kicking ergometer, on subsequent days, 24 h apart. These tests were performed in a randomized order.

Back to Top | Article Outline

All-out 30 s exercise test.

All subjects performed an all-out 30-s exercise test on each of four separate occasions. After being attached to the swim bench or leg-kicking ergometer the swimmers were instructed to warm up by exercising at 25 W for 2 min. Subjects were verbally coached throughout the exercise. Time indications were given after 15, 20, and 25 s.

Back to Top | Article Outline

Statistical Analysis

The power versus time relationships were explored using Pearson’s product moment correlation and regression analysis. The repeated tests for arm-pulling and leg-kicking were compared using two-way analysis of variance (ANOVA) and post hoc repeated t-tests with Bonferroni correction. The variation in repeated measurement of arm and leg power was expressed as 95% confidence interval.

Back to Top | Article Outline


Selection of Best MPV Setting

The PPO from the MPV selection procedure (10-s all-out exercise tests) ranged from 172 ± 27 W to 366 ± 46 W for arm-pulling, and from 293 ± 34 W to 422 ± 37 W for leg-kicking. The individual relationships between PPO and MPV revealed a mean PPO of 358 ± 23 W for the arms and 415 ± 28 W for the legs. These PPO values occurred at a MPVopt of 2.55 m·s1 and 2.33 m·s1, respectively. The group mean PPO at each of the five MPVs for arm-pulling and leg-kicking is displayed in Figure 2.

Figure 2

Figure 2

Back to Top | Article Outline

Power Output from 30 s All-out Test

The relationships between power output and time during the 30-s tests were linear in all cases for arm-pulling and leg-kicking (P < 0.01). The mean PPO as calculated from the power versus time relationship was 304 ± 22 W for arms and 435 ± 36 W for legs. The arms produced 69.9% of the PPO values of the legs. An example of the relationship between power output and time for the arms and legs is given in Figure 3.

Figure 3

Figure 3

Back to Top | Article Outline

Repeated Testing

None of the mean PPO or MPO values were different between the repeated tests. For arm-pulling, the average PPO values were 291 ± 31 W versus 317 ± 32 W. For leg-kicking the means were 454 ± 45 W versus 416 ± 37 W. The relationship between values from the two tests for arms and legs are represented in Figure 4. The MPO values for arm-pulling were 247 ± 32 W versus 203 ± 24 W for first and second tests, respectively. The equivalent values for leg-kicking were 304 ± 24 W versus 320 ± 31 W. The average PPO and MPO values from repeated 30-s tests were significantly related for arm-pulling (PPO, r = 0.97;P = 0.01 vs MPO, r = 0.96;P = 0.01) and for leg-kicking (PPO, r = 0.96;P = 0.01 vs MPO, r = 0.96;P = 0.05). Variation in measurement of PPO from repeated tests was 7.3% for arm-pulling and 8.3% for leg-kicking. For MPO, repeat-test variation was 6.9% and 7.4% for arm-pulling and leg-kicking, respectively.

Figure 4

Figure 4

Back to Top | Article Outline


There are few studies of PPO during simulated front crawl arm-pulling and leg-kicking in swimmers using dry-land ergometry. The arm-pulling PPO values from the present study are in agreement with measurements made on a swim bench during a single arm pull (15) and are similar to those reported for arm-cranking exercise (4,5). The mean PPO value for leg-kicking in the present study is lower than previously reported for all-out cycling in swimmers (4,5). This difference in leg PPO can probably be explained by the greater mass of exercising musculature involved in cycling (5).

It might be considered surprising to find such high leg-kicking power output values for swimmers, because the leg-kicking action contributes so little to propulsion during swimming (2). These findings, in addition to the reports of aerobic power during dry-land leg-kicking (17), appear to contradict the established view that the power output and cardiopulmonary function of the swimmer is almost entirely located in the arms (21). Furthermore, most studies that have investigated power output of the arms and legs in swimmers have suggested that leg power is of little importance in swimming. The present results suggest that it might be necessary for swimmers to develop arm and leg power equally in dry-land training. Unfortunately, currently there are few dry-land training devices that allow simulation of the front crawl leg-kicking action against resistance.

The ratio of arm:leg power found in the present study (70%) was higher than previously found in swimmers using other types of ergometry (4). Hawley and Williams (4) reported an arm:leg power ratio of 45%. However, these investigators admitted that the ratio might have been depressed by the long-distance nature of their subjects’ training. This raises the question of the possible changes in leg-kicking power during training. This issue has not been investigated previously. Most studies of power output changes throughout training in swimmers have measured only arm power (15). The only other study reporting the arm:leg power ratio involved untrained children and found values between 60 and 70% using arm-cranking and cycling (7).

Of those studies that have used the swim bench, few have selected the MPV setting which maximizes individual performance on the swim bench. Sharp et al. (15) reported power output values for a series of single arm-pulls at a MPV ranging from 1.6 to 3.3 m·s1. These authors found that the MPVopt was 2.4 m·s1, which is similar to the MPV setting found to be optimal in the present study. However, Sharp et al. (15) did not use the same MPVopt setting in subsequent tests (15). Rather, a MPV was chosen that was close to mean maximal swimming speed over 100 yards. The importance of selecting an appropriate resistance setting before all-out exercise testing has been previously emphasized (1).

There are no published studies of the reproducibility of dry-land simulated front crawl arm-pulling or leg-kicking power in swimmers. Indeed, there are few studies of the reproducibility of measurement of dry-land arm power in general. Sharp et al. (15) found a reliability coefficient of 0.93 for repeated testing of single arm-pull power using a swim bench. The repeat test reliability coefficient for leg power on a cycle ergometer has been shown to be 0.96 (1,3). This is in agreement with that found in the present study. The small variation in repeated measurement of simulated arm-pulling and leg-kicking might provide a basis upon which changes in power output could be detected during training.

It is well established that the movements involved in swim bench ergometry do not replicate the movements involved in swimming (11). Nevertheless, it is clear that swimmers are more suited to this type of ergometry when compared to arm-cranking and cycling, not least because the exercising posture is similar to that of swimming. Furthermore, during swimming it is the resultant power due to propulsion that is quantified and not the power output of the limbs. It is known that these two measures of power output can differ according to the effectiveness of the swimming stroke technique (21). Because stroke technique can vary during training, it would complicate the interpretation of changes in power output if these changes were measured only during swimming itself.

The difficulties in assessment of leg power in swimmers have prevented determination of the freely chosen contribution of leg-power to the overall power output of the swimmer when using arms and legs. The local muscle metabolism involved in exercising arm or leg body segments simultaneously is poorly understood. Previous studies have investigated the contribution of the leg-kick to overall swimming speed (2). However, this has given little insight into the relative arm and leg metabolism at given exercise intensities. Rather, this would require the simultaneous use of the arm-pulling and leg-kicking ergometers described in the present study.

In summary, the results of this study demonstrate that power output can be assessed during simulated front crawl arm-pulling and leg-kicking. The intra-subject variation in measurement of power output appears to be small. The procedures used in the present study are highly reproducible. This study also introduces a procedure for selecting the MPV that allows maximal power output during subsequent testing. The ability to measure power output during arm-pulling and leg-kicking in swimmers could be of value in studies investigating adaptations to training in arm versus leg power. Furthermore, measurement of power output from the legs in swimmers might inform the swimming coach about the importance of leg power in front crawl swimming, despite its minor contribution to propulsion.

Thank you to the competitive swimmers of Bedford Modernians and Northampton Swimming Clubs for their participation in this study. Thank you to E. M. Winter, Ph.D. (for assistance in preparation of the manuscript) and to C. L. Zanker, M.Phil., and W. Riley (for artwork), all at De Montfort University Bedford, Bedford, UK.

Back to Top | Article Outline


1. Bar-Or, O. The Wingate anaerobic test: an update on methodology, reliability and validity. Sports Med. 4:381–394, 1987.
2. Bucher, W. Influence of the leg kick and the arm stroke on the total speed during the crawl stroke. In: Biomechanics in Swimming II, J. P. Clarys and L. Lewillie (Eds.). Baltimore: University Park Press, 1974, pp. 180–187.
3. Evans, J. A., and H. A. Quinney. Determination of resistance settings for anaerobic power testing. Can. J. Appl. Sports Sci. 6:53–56, 1981.
4. Hawley, J. A., and M. M. Williams. Relationship between upper body anaerobic power and freestyle swimming performance. Int. J. Sports Med. 12:1–5, 1991.
5. Hawley, J. A., M. M. Williams, M. M. Vickovic, and P. J. Handcock. Muscle power predicts freestyle swimming performance. Br. J. Sports Med. 26:151–155, 1992.
6. Hollander, A. P., G. De Groot, G. J. Van Ingen et al. Measurement of active drag forces during swimming. J. Sports Sci. 4:21–30, 1986.
7. Inbar, O., and O. Bar-Or. Relationship between anaerobic and aerobic arm and leg capacities to swimming performance of 8–12 year old children. In: Frontiers of Activity and Child Health, R. J. Shephard and R. Lavellee (Eds.). Quebec: Pelican, 1977, pp. 283–292.
8. Klika, J. R., and W. G. Thorland. Physiological determinants of sprint swimming performance in children and young adults. Pediatr. Exerc. Sci. 6:59–68, 1994.
9. Meerloo, A. I., M. L. Collis, and H. A. Wenger. Determination of VO2max in competent swimmers when using a continuous versus a discontinuous swim bench protocol. J. Hum. Mov. Stud. 13:249–254, 1987.
10. Nakamura, Y., Y. Mutoh, and M. Myashita. Determination of the peak power output during maximal brief pedalling bouts. J. Sports Sci. 3:181–187, 1985.
11. Olbrecht J., and J. P. Clarys. EMG of specific dry land training for the front crawl. In: Biomechanics and Medicine in Swimming, A. P. Hollander, P. A. Huijing, and G. De Groot (Eds.). Champaign, IL: Human Kinetics, 1983, pp. 136–141.
12. Payton, C. J., and R. M. Bartlett. Estimation of propulsive forces in swimming from three-dimensional kinematic data. J. Sports Sci. 13:447–454, 1995.
13. Rohrs, D. M., J. L. Mayhew, C. Arabas, and M. Shelton. The relationship between seven anaerobic tests and swim performance. J. Swim. Res. 6:15–19, 1990.
14. Schliehauf, R. E., L. Gray, and J. De Rose. Three-dimensional analysis of swimming propulsion in the sprint front crawl stroke, In:Biomechanics and Medicine in Swimming. A. P. Hollander, P. A. Huijing, and G. De Groot (Eds.). Champaign, IL: Human Kinetics, 1983, pp. 173–184.
15. Sharp, R. L., J. P. Troup, D. L. Costill. Relationship between power and sprint freestyle swimming. Med. Sci. Sports Exerc. 14:53–56, 1982.
16. Swaine, I. L. The relationship between 1500m swimming performance and critical power using an isokinetic swim bench. In:Biomechanics and Medicine in Swimming VII, J. P. Troup, et al. (Eds.). London: E and FN Spon, 1996, pp. 229–234.
17. Swaine, I. L. Cardiopulmonary responses to exercise in swimmers using a swim bench and a leg kicking ergometer. Int. J. Sports Med. 18:359–363, 1997.
18. Swaine, I. L, and C. L. Zanker. The reproducibility of cardiopulmonary responses to exercise using a swim bench. Int. J. Sports Med. 17:140–144, 1996.
19. Taunton, J. E., H. Maron, and J. G. Wilkinson. Anaerobic performance in middle and long distance runners. Can. J. Appl. Sports Sci. 6:109–113, 1981.
20. Takahashi, S., M. Bone, J. M. Cappaert, et al. Validation of a dryland swimming specific measurement of anaerobic power. In:Biomechanics and Medicine in Swimming VI, D. MacLaren, T. Reilly, and A. Lees. (Eds.). London: E and FN Spon, 1992, pp. 301–305.
21. Toussaint, H. M., and P. J. Beek. Biomechanics of competitive front crawl swimming. Sports Med. 13:8–24, 1992.
22. Vandewalle, H., G. Peres, J. Heller, and H. Monod. All out anaerobic capacity tests on cycle ergometers: a comparative study on men and women. Eur. J. Appl. Physiol. 54:222–229. 1985.
23. Winter, E. M. Cycle ergometry and maximal intensity exercise. Sports Med. 11:351–357. 1991.


© 2000 Lippincott Williams & Wilkins, Inc.