Effect of an 8-week technical tethered swimming program on front kinematics in adolescent swimmers

Abstract

The aim of this study was to assess the effect of long-term use of technical tethered swimming on kinematics of front crawl swimming, and swimming performance. The experiment was attended by 19 girls Thirty-nine adolescent swimmers (19 girls and 20 boys, aged 13 years) participated in the study. The participants were randomly assigned to two groups. The experimental group received additional technical tethered swimming protocol. At this time, the control group performed standard technical training in free swimming. The study included an 8-week training cycle. Changes in swimming technique were assessed using a device equipped with a triaxial gyroscope and a triaxial accelerometer, with a particular focus on body roll and yaw rotationof the athletes. Significant changes were observed in yaw rotation (ω_maxY, ab_max) and body roll (A_maxR) variables. The study showed that a training protocol involving technical tethered swimming can positively affect front crawl kinematics in adolescent athletes when applied over the long term.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-24816-9.

Introduction

The concept of tethered swimming (TS) was pioneered by Magel¹. Tethered swimming has been recognized as one of the most specific methods for simulating ergometer swimming (Filho & Denadai, 2008), due to the similarity of movement in interaction with the environment², cycle mechanics³ and physiological aspects⁴. At the same time, contemporary perspectives in biomechanics emphasize the role of movement analysis not only in performance enhancement but also in injury prevention and rehabilitation, highlighting the need for applied approaches that translate laboratory insights into real training contexts⁵.

In addition, it is a method commonly used to measure thrust, both in relation to swimming performance correlation⁶, and to seasonal changes in propulsive force⁷, as well as a training tool to help develop physical abilities such as strength and power⁸. Although it is often emphasized in the scientific literature that the technique of tethered swimming is significantly different from that of free swimming, few comparative studies on the subject have been conducted to date⁹. These differences particularly relate to the mechanics of movement and the drag forces generated¹⁰.

Tethered swimming was introduced not only as a measurement method but also as a swimming training method⁷. There are two main methods used in competitive swimming training using tethered swimming. Fully tether swimming restricts forward displacement and is mainly used for near-static power traning¹¹. Semi-tethered swimming, incorporates inertial loading that permits controlled forward movement, thereby more closely replicating free swimming conditions¹². These take the form of resistance training, understood as that type of exercise that requires an impact against an opposing force usually generated by some type of training device. Such training aims to increase strength, power¹³.

Maglischo and Maglischo¹⁴ in their comparative study analysed the effects of tethered swimming and free swimming. The study included an analysis of swimmers performing a series of sprints with additional resistance to assess the effects of this method on the strength and performance parameters of swimmers. The researchers suggested that the technical changes observed during high-intensity tethered swimming may negatively affect technique. In another study, semi-tethered swimming was performed using a specially adapted Smith machine. This method is mainly used to perform high intensity tests or to assess swimmers’ skills¹⁵.

Olbrecht¹⁶ described two types of technical training: type I – training aimed at improving or teaching a new movement pattern; and type II – described as training swim drills aimed at automating the correct movement cycle at starting speed. Type I technical training should be planned at the beginning of the training session (after the warm-up), there should be a long break between exercises for a full rest, the working time should be short and the intensity low. Type II exercises, on the other hand, aim to ‘automate the movement’. The volume and distance of the repetitions should be progressively longer, the rests shorter and the intensity higher, and the exercise itself should reflect racing conditions as much as possible. The tethered training methods described in the literature do not fall under technical training, as they are not performed at low intensity, do not aim to improve or teach an appropriate movement pattern (type I) and do not reflect racing conditions (type II). However, fall within resistance training, as described in the literature¹⁷.

Practices based on exercises use the part method of teaching movement. They are used by coaches in various sports. They help correct technique and learn new skills. Still, there are doubts about whether the skills learned in such exercises effectively replicate key information and movement requirements in the target environment^18–20. An example of this would be single-arm crawl swimming, with the resting arm held close to the body. The purpose of this exercise is to allow athletes to focus on coordinating their breathing and improving their body position^21,22. However, there is no empirical evidence to demonstrate that these exercises affect the improvement of body alignment, breathing coordination or athletic performance. Instead, it is argued that the whole method may better facilitate learning for the repetitive and continuous movements that occur in swimming²³.

There seems to be a belief among coaches that tethered swimming has a positive effect on swimming technique. Compared to the other described uses of TS, the exercises performed for this purpose are characterized by low intensity and focus on stability of the arm and trunk during the active phase of the arms. For the purposes of this study, this type of use of tethering has been termed technical tethered swimming (TTS).

The observations of Maglischo and Maglischo¹⁴ were used as theoretical confirmation of the assumptions. It was observed that athletes tended to move their arms in a smaller arc during tethered swimming and took longer to perform the arm movement. Samson et al²⁴. observed less medial-lateral arm movement during tethered swimming compared to free swimming.

This raises the question of whether technical tethered swimming has real benefits as a training tool using a holistic method of teaching movement.

The aim of this study was to assess the effect of long-term use of technical tethered swimming on kinematics of front crawl kinematics and swimming performance.

Adolescence is considered a sensitive period for motor learning, during which technique-focused interventions can provide long-term benefits for athletic progress. Evidence from youth sport indicates that structured task-based exercises not only improve technical performance but also support broader cognitive abilities such as problem solving and creativity²⁵. In this context, incorporating technical rope swimming (TTS) into training programmes for adolescents may be particularly valuable, as it allows swimmers to refine body alignment, improve proprioceptive control, and reinforce effective movement patterns during a critical stage of skill development.

In both tethered and free-swimming conditions, the average forward velocity of the body is constant (equal to zero in the tethered case). As a result, the net impulse of force in the longitudinal axis is zero. In free swimming, the frontal drag is balanced by propulsive force, whereas in tethered swimming the frontal drag is minimal; however, the limb velocity relative to the water is greater, generating larger propulsive forces that are counterbalanced by the tether tension. If the swimmer expends the same amount of energy, the net impulse of force remains similar, which implies that the propulsive limb movement must be slower during the active stroke phase when tethered. Importantly, tethered swimming does not provide the stabilizing effect of water flow around the body; thus, any lateral force components or yaw moments lead to loss of stability. This requires precise propulsion during the arm stroke. These specific conditions suggest that technical tethered swimming should be considered primarily as a method of technical skill development rather than strength training. Despite its common use for strength and power development, the constraints imposed by tethered swimming enhance proprioceptive demands, improve trunk control, and facilitate conscious correction of technical errors.

Materials and methods

Participants

Nineteen girls (age 13.18 ± 0.66 years; body height 163.6 ± 5.2 cm; body weight 50.8 ± 4.42 kg; body fat 20.5% ± 2.0%) and 20 boys (age 13.33 ± 0.60 years; body height 167.8 ± 8.76 cm; body weight 52.46 ± 8.8 kg; body fat 14.1% ± 2.3%) participated in the study. All subjects were athletes of a local sports club. Inclusion criteria were: age between 12 and 14 years, possession of a valid swimming license, regular participation in competitions organized by the Polish Swimming Federation, and a minimum of 2 years of systematic swimming training. Exclusion criteria included musculoskeletal injuries in the 3 months preceding the study, medical contraindications for high-intensity training. Written informed consent was obtained from all participants and their legal guardians.

Participants were randomly allocated to the experimental group (EG) or control group (CG) using a computer-generated randomization procedure in Microsoft Excel. The randomization list was created prior to the start of the intervention.

The study was approved by the Senate Committee on Research Ethics of the Józef Piłsudski Academy of Physical Education in Warsaw (SKE 01–31/2023). All procedures involving human participants were performed in accordance with relevant guidelines and regulations, and in adherence to the Declaration of Helsinki. Informed consent was obtained from all participants and, due to the age of the subjects, from their parents or legal guardians. Participants and their legal guardians were fully informed about the nature, purpose, and course of the study, including any potential risks or benefits associated with participation.

Sample size and determination

The required sample size was estimated a priori using GPower software. For a repeated-measures ANOVA, assuming a medium effect size (Cohen’s f = 0.25), α = 0.05, and desired statistical power (1–β) = 0.80, the minimum sample size was calculated as 34 participants (17 per group). Our final sample of 39 participants exceeded this threshold, ensuring adequate statistical power to detect meaningful interaction effects.

Procedure

The study participants were randomly divided into two groups, in each maintaining gender parity. Randomization was performed using an Excel function. In the experimental group (EG), additional tethered technical training was introduced. At this time, the control group (CG) performed standard technical training in free swimming. The study included an 8-week training cycle.

Additional technical tether swimming was performed once a week and lasted 45 min. It included the following training procedure:

6 × 10 tethered swimming crawl cycles with 10ʹʹ rest,

1 × 150 front crawl snorkel swimming.

The athletes performed technical training in a tether with a snorkel to facilitate breathing. Free swimming was also performed with a snorkel to standardize training methods. The main objectives of performing this task were:

1. Focusing on the length of the stroke.

2. Deliberately slow execution of movements.

3. Stabilizing the trunk by reducing lateral arm movement.

4. Low-intensity work, maintained below the aerobic threshold

At the same time, the control group (CG) performed technical training for front crawl, following an identical schedule and technical guidelines as the experimental group (EG), and it concluded with a trending procedure:

45 min (x50 rest 10ʹʹ−20ʹʹ).

1. One-arm front crawl

2. Catch-up stroke drill

3. One-arm alternate catch-up

4. Front crawl

Both workouts were instructed and coordinated by two experienced swimming coaches (level I – certified by Polish National Swimming Association).

Before and after the split, the whole group performed the same swimming training together. The split was only introduced during the 45 min of training described above.

Training load during TTS was monitored using tether force measurement, heart rate (Polar Verity Sense), and the Children’s OMNI Scale of Perceived Exertion, with the target intensity set below 150 bpm. The purpose of the task was explained to the swimmers before each training bout, and during its execution the coaches focused exclusively on controlling intensity and duration. No additional technical feedback was provided, as two training sessions were conducted in parallel, and it was important to ensure that neither group received preferential attention.d.

Pre-test and post-test included:

1. Measurement of height and weight.

2. Measurement of the time taken to swim a distance of 50 m in front crawl at maximum speed while measuring acceleration using an accelerometer placed on the back of the pelvic girdle. A number of variables were measured in 3 major body movements:

• Translational motion – represented by averaged acceleration along the vertical axis.

• Body roll (rotational movements around the vertical axis of the body) – recorded by velocities around the vertical axis.

• Yaw rotation (rotational movements around the sagittal axis) recorded by angular velocities around the sagittal axis and accelerations along the transverse axis.

The analysis presents the most relevant variables, whose values are averaged.

The most relevant variables were selected for analysis, and the values presented were averaged.

From the recorded data, average numerical values of the swimming cycle were calculated, including arithmetic means and standard deviations for relative time (relative to cycle duration), all measured from the beginning of the cycle. These calculations were performed using STA1v0 software (Zbigniew Staniak, Institute of Sport – National Research Institute, Poland).

The following components were measured and analysed: av_max – vertical acceleration during propulsion, ω_maxR – angular velocity around the vertical axis in rolling movements, ω_maxY – angular velocity around the sagittal axis in yaw rotation, and A_maxR – maximum angle of pelvic tilt around the vertical axis during rotational movements. The data were analysed separately for movements involving the left upper limb (Left) and right upper limb (Right), as well as for the first lap (I25) and the second lap (II25).

The selected kinematic variables were chosen because they represent critical aspects of trunk rotation and stability in front crawl swimming. Roll angle and angular velocity have been widely investigated as indicators of body alignment and hydrodynamic efficiency^26,27. In contrast, yaw movements have received much less attention in swimming research, with only a few studies addressing this variable in recent years. In our view, yaw rotation is an unjustly neglected aspect of swimming kinematics, as it directly reflects trunk control and lateral deviations that may increase drag and impair efficiency. Including yaw among the analyzed variables therefore provides a broader and more comprehensive picture of swimmers’ body alignment.

The measurements took place in a 25 m pool. The athletes started from the water (without a starting dive). They swam the 25 m front crawl, did the standard freestyle turnaround and swam the second lap of the 50 m distance. The subjects were asked to complete the task in the shortest possible time. Before starting the swim, they performed a start warm-up consisting of a 1200 m swim as detailed by Skorulski et al.²⁸.

Research instruments

For each swimmer, the individual characteristics of the kinematics of average hip movement during the first and second halves of the pool were determined. A recorder (REJ62g by JD Jarosław Doliński, Poland) was used to track changes in speed and acceleration. The device contained a triaxial gyroscope and triaxial accelerometer (65 × 50 × 30 mm, 95 g). It was placed in a foam cover to minimise hydrodynamic drag and provide stability in the swimmer’s lower back, near the pelvic girdle. The centre of the recorder was aligned at the base of the sacrum. A special two-part belt was used to secure it: a non-elastic rope attached the recorder, while an elastic band was placed on the swimmer’s lower abdomen (Fig. 1).

The pelvic region was selected for sensor placement because it represents the central link in trunk alignment and force transfer during swimming. Previous research has shown that pelvic yaw and roll are sensitive indicators of trunk stability and hydrodynamic efficiency, with excessive rotation associated with increased drag and optimized rotation linked to improved swimming performance^26,29.

Measurements were taken at a sampling frequency of 200 Hz, with acceleration measured within a range of ± 2 g. The signal underwent analogue low-pass filtering at a cutoff frequency of 93 Hz. To measure angular velocity during rotation, a range of ± 500 deg·s⁻¹ was used.

The accuracy of acceleration measurements was verified statically against ground acceleration, with an absolute error of ± 0.2 m·s⁻². The precision of angular velocity readings was checked indirectly by measuring and calculating the recorder’s rotation angle within a 90-degree range around each axis. The absolute error for angle calculation was ± 1 degree, and for angular velocity, it was ± 0.6 deg·s⁻¹.

The validity of this device for swimming kinematics has been demonstrated previously: Staniak et al³⁰. confirmed its ability to distinguish breaststroke cycles in comparison with video-based reference analysis, while Staniak et al³¹. reported its sensitivity to phase duration and correlation with performance in butterfly, also validated against high-speed video. More recently, Skorulski et al²⁸. showed that the device can detect fatigue-induced changes in front crawl, confirming technical sensitivity.

The recorded waveforms were smoothed using a four-pole low-pass Butterworth filter with a cutoff frequency of 20 Hz. This frequency was selected to ensure that the calculated amplitude of motion speed would not decrease by more than 0.5% as a result of filtering, while keeping important acceleration waveform points for movement analysis clearly visible.

Body weight and composition were measured using a Tanita BC-545 N scale (Tanita Corporation, Tokyo, Japan).

A tether with a force meter (ZPS5-BTU1kN, Staniak, Poland) recorded the pulling force at 100 Hz and sent the data to a computer program for further analysis (MAX6v0M software, Poland).

During training, the test subjects were monitored using Polar Verity sense sensors, which allow real-time measurement of heart rate using Polar Team software. After a task, each athlete was asked to rate the task using the Children’s OMNI Scale of Perceived Exertion (OMNI) (Robertson et al., 2000).

Heart rate and fatigue levels were monitored by experienced swim coaches using the PolarTeam app and Variety Sense sensors. The aim of the training was to keep the heart rate below 150 h (bpm). The athletes had already been introduced to the OMNI scale and to working with heart rate sensors.

Test subjects performing tethered technical swimming were subjected to pulling force measurements during technical work to determine the maximum force with which they performed the technical task. A tether with a force meter (ZPS5-BTU1kN, Staniak, Poland) recorded the pulling force at 100 Hz and sent the data to a computer program for further analysis (MAX6v0M software, Poland).

Test subjects performing technical tethered swimming were subjected to a thrust measurement during technical work to determine the maximum force with which they performed the technical task.

Training volume and intensity

The training was designed and performed by the trainers in charge of the study training group. The structure of the training used is shown in Table 1.

Table 1.

Training volume completed by the control group, expressed as swimming distance (m), in the successive weeks of the experiment by designated training task.

Week	Total[m]	REC[m]	AEC1[m]	AEC2[m]	AEP[m]	ANP[m]	ANC[m]	SP[m]
1	22,170	6950	13,600	0	0	0	600	1020
2	33,070	19,920	12,850	0	0	0	300	0
3	29,950	17,500	11,950	0	0	0	500	0
4	27,450	11,100	11,600	3000	0	300	600	850
5	22,850	13,150	6500	0	2600	0	600	0
6	32,300	12,050	17,400	2000	0	0	0	850
7	21,050	18,500	2000	0	0	550	0	0
8	29,280	13,610	11,400	2000	0	500	720	1050

Table 1. REC – swim sets included warm up, cool down, kicking, pulling, and drill sets; AEC1 (aerobic capacity type 1) – swim set included swimming up to 30 min in one style, high volume, low intensity, and short rest. AEC2 (aerobic capacity type 2) – combination of short, intensive repetitions with AEC1; AEP (aerobic power) – included swimming sets with high intensity work, short rest, competitions of 200 m and more; ANP (anaerobic power) – swimming sets included extremely hard efforts with short rest, competitions up to 100 m. ANC (anaerobic capacity) – included short interval (25-50 m), long rest, very high to maximal intensity; SP (sprint sets) – very short sets ( <10 s) with maximal velocity, and very long rest.

Statistical analysis

Statistical analyses were carried out using STATISTICA software, version 13.1 (TIBCO Software Inc., 2017). The Kolmogorov-Smirnov test was used to verify the normality of data distribution, with a p-value greater than 0.20 indicating a normal distribution. All examined variables met the assumption of normality.

To assess differences in means, a repeated measures ANOVA (general linear model) was used. The analysis included three within-subject factors: MEASUREMENT (pretest, posttest), LAP (first, second), and SIDE (left, right), while GROUP (experimental, control) was treated as a between-subject factor. Post-hoc analyses were conducted using Fisher’s least significant difference (LSD) test. Statistical significance was set at the level of α = 0.05. Data were reported as means ± standard deviations and accompanied by 95% confidence intervals. Effect sizes were estimated using partial eta squared.

Results

Of all the measured velocity and acceleration variables, those that showed the most significant group × measurement or group × measurement × side interactions were selected and are described in the following section. Detailed descriptive statistics (means ± SD with 95% CI) are presented in Table 2, while the results of the repeated-measures ANOVA, including F values, p-values, and effect sizes (η²), are summarized in Table 3.

Table 2.

Mean ± SD values with 95% confidence intervals (CI) of the analyzed acceleration components during the first and second lengths of the pool, measured before and after the 8-week training cycle in the experimental group (technical tethered swimming) and the control group.

Variables	Side	Control group				Experimental group
		I25		II25		I25		II25
		PreTest	PostTest	PreTest	PostTest	PreTest	PostTest	PreTest	PostTest
ωmaxR	Left	217.0 ± 40.3 (CI: 198.14–235.86)	226.7 ± 47.8 (CI: 204.33–249.07)	233.6 ± 49.3 (CI: 210.53–256.67)	236.51 ± 43.2 (CI: 216.29–256.73)	241.6 ± 45.4 (CI: 219.72–263.48)	228.0 ± 49.0 (CI: 204.38–251.62)	253.3 ± 45.2 (CI: 231.51–275.09)	240.0 ± 46.6 (CI: 217.54–262.46)
	Right	237.6 ± 51.1 (CI: 213.68–261.52)	229.1 ± 47.8 (CI: 206.73–251.47)	241.9 ± 43.6 (CI: 221.49–262.31)	230.7 ± 43.0 (CI: 210.58–250.82)	244.8 ± 54.6 (CI: 218.48–271.12)	233.0 ± 56.5 (CI: 205.77–260.23)	246.9 ± 50.1 (CI: 222.75–271.05)	244.6 ± 44.4 (CI: 223.20–266.00)
t_ωmaxR	Left	0.27 ± 0.11 (CI: 0.22–0.32)	0.31 ± 0.12 (CI: 0.25–0.37)	0.29 ± 0.11 (CI: 0.24–0.34)	0.31 ± 0.14 (CI: 0.24–0.38)	0.24 ± 0.11 (CI: 0.19–0.29)	0.24 ± 0.12 (CI: 0.18–0.30)	0.27 ± 0.12 (CI: 0.21–0.33)	0.28 ± 0.13 (CI: 0.22–0.34)
	Right	0.23 ± 0.09 (CI: 0.19–0.27)	0.25 ± 0.11 (CI: 0.20–0.30)	0.24 ± 0.08 (CI: 0.20–0.28)	0.27 ± 0.11 (CI: 0.22–0.32)	0.26 ± 0.12 (CI: 0.20–0.32)	0.23 ± 0.11 (CI: 0.18–0.28)	0.27 ± 0.13 (CI: 0.21–0.33)	0.27 ± 0.12 (CI: 0.21–0.33)
ωmaxY	Left	108.2 ± 18.8 (CI: 99.40–117.00)	117.3 ± 22.6 (CI: 106.72–127.88)	112.9 ± 21.5 (CI: 102.84–122.96)	120.2 ± 21.1 (CI: 110.32–130.08)	106.9 ± 25.5 (CI: 94.61–119.19)	101.1 ± 25.0 (CI: 89.05–113.15)	110.9 ± 27.0 (CI: 97.89–123.91)	104.9 ± 24.2 (CI: 93.24–116.56)
	Right	109.6 ± 21.8 (CI: 99.40–119.80)	108.8 ± 18.7 (CI: 100.05–117.55)	111.3 ± 19.7 (CI: 102.08–120.52)	114.9 ± 16.1 (CI: 107.36–122.44)	111.8 ± 24.3 (CI: 100.09–123.51)	106.9 ± 21.5 (CI: 96.54–117.26)	112.7 ± 22.7 (CI: 101.76–123.64)	112.9 ± 20.9 (CI: 102.83–122.97)
t_ωmaxY	Left	0.28 ± 0.09 (CI: 0.24–0.32)	0.29 ± 0.12 (CI: 0.23–0.35)	0.33 ± 0.11 (CI: 0.28–0.38)	0.32 ± 0.12 (CI: 0.26–0.38)	0.30 ± 0.09 (CI: 0.26–0.34)	0.28 ± 0.08 (CI: 0.24–0.32)	0.32 ± 0.10 (CI: 0.27–0.37)	0.31 ± 0.09 (CI: 0.27–0.35)
	Right	0.27 ± 0.09 (CI: 0.23–0.31)	0.29 ± 0.09 (CI: 0.25–0.33)	0.30 ± 0.08 (CI: 0.26–0.34)	0.31 ± 0.10 (CI: 0.26–0.36)	0.31 ± 0.10 (CI: 0.26–0.36)	0.29 ± 0.10 (CI: 0.24–0.34)	0.31 ± 0.12 (CI: 0.25–0.37)	0.32 ± 0.11 (CI: 0.27–0.37)
abmax	Left	12.14 ± 3.44 (CI: 10.53–13.75)	12.38 ± 4.06 (CI: 10.48–14.28)	12.20 ± 3.14 (CI: 10.73–13.67)	12.91 ± 3.84 (CI: 11.11–14.71)	13.20 ± 2.83 (CI: 11.84–14.56)	12.67 ± 2.79 (CI: 11.33–14.01)	12.87 ± 3.1 (CI: 11.38–14.36)	12.03 ± 2.93 (CI: 10.62–13.44)
	Right	11.01 ± 1.63 (CI: 10.25–11.77)	11.67 ± 1.75 (CI: 10.85–12.49)	11.43 ± 2.00 (CI: 10.49–12.37)	12.51 ± 2.02 (CI: 11.56–13.46)	12.23 ± 2.66 (CI: 10.95–13.51)	11.11 ± 2.11 (CI: 10.09–12.13)	12.11 ± 2.22 (CI: 11.04–13.18)	11.43 ± 2.09 (CI: 10.42–12.44)
AmaxR	Left	28.62 ± 4.92 (CI: 26.32–30.92)	29.81 ± 4.77 (CI: 27.58–32.04)	33.10 ± 4.89 (CI: 30.81–35.39)	34.42 ± 4.23 (CI: 32.44–36.40)	28.83 ± 4.22 (CI: 26.80–30.86)	27.23 ± 5.57 (CI: 24.55–29.91)	32.91 ± 5.24 (CI: 30.38–35.44)	31.84 ± 4.68 (CI: 29.58–34.10)
	Right	27.61 ± 4.58 (CI: 25.47–29.75)	29.34 ± 4.00 (CI: 27.47–31.21)	32.30 ± 4.39 (CI: 30.25–34.35)	33.92 3.81	27.94 ± 4.20 (CI: 25.92–29.96)	27.02 ± 5.28 (CI: 24.48–29.56)	31.94 ± 5.40 (CI: 29.34–34.54)	31.26 ± 4.62 (CI: 29.03–33.49)
t_AmaxR	Left	0.55 ± 0.11 (CI: 0.50–0.60)	0.55 ± 0.14 (CI: 0.48–0.62)	0.61 ± 0.08 (CI: 0.57–0.65)	0.62 ± 0.12 (CI: 0.56–0.68)	0.53 ± 0.04 (CI: 0.51–0.55)	0.53 ± 0.06 (CI: 0.50–0.56)	0.60 ± 0.05 (CI: 0.58–0.62)	0.59 ± 0.08 (CI: 0.55–0.63)
	Right	0.50 ± 0.10 (CI: 0.45–0.55)	0.53 ± 0.11 (CI: 0.48–0.58)	0.56 ± 0.11 (CI: 0.51–0.61)	0.59 ± 0.11 (CI: 0.54–0.64)	0.53 ± 0.05 (CI: 0.51–0.55)	0.52 ± 0.06 (CI: 0.49–0.55)	0.58 ± 0.07 (CI: 0.55–0.61)	0.58 ± 0.07 (CI: 0.55–0.61)

Left – motion during active left upper limb, Right – motion during active right upper limb, I25 – the first part (25 m) of the analyzed effort, II25 – the second part (25 m) of the analyzed effort, ω_maxR – maximum angular velocity around the long axis in rotation movements, t_ω_maxR – time needed to achieve maximum angular velocity around the long axis (roll), ω_maxY – maximum angular velocity around the sagittal axis of the athletes in yaw movements (yaw rotation), t_ω_maxY – time needed to achieve maximum angular velocity around the sagittal axis of the athletes in yaw movements (yaw rotation), ab_max – acceleration along the transverse axis in yaw movments, A_maxR – maximum angle around the long axis in rotation movements, t_A_maxR – time needed to achieve maximum angle around the long axis in rotation movements.

Table 3.

Results of repeated-measures ANOVA for kinematic variables. The table reports F values, p-values, and effect sizes (η²) for the MEASUREMENT × GROUP and MEASUREMENT × GROUP × SIDE interactions.

Variables	INTERACTION
	MESURMENTS x GROUP	MESURMENTS x GROUP x SIDE
ωmaxR	F1.37 = 0.86, p = 0.358 η² = 0.15	F1.37 = 5.57, p = 0.024, η² = 0.13
t_ωmaxR	F1.37 = 41.62, p = 0.210, η² = 0.04	F1.37 = 0.91, p = 0.34, η² = 0.15
ωmaxY	F1.37 = 1.83, p = 0.180, η² = 0.05	F1.37 = 4.15, p = 0.048, η² = 0.10
t_ωmaxY	F1.37 = 1.11, p = 0.291, η² = 0.01	F1.37 = 1.11, p = 0.291, η² = 0.01
abmax	F1.37 = 7.89 p = 0.008, η² = 0.18	F1.37 = 0.22, p = 0.636, η² = 0.01
AmaxR	F1.37 = 5.48, p = 0.025, η² = 0.13	F1.37 = 0.78, p = 0.781, η² = 0.01
t_AmaxR	F1.37 = 0.26, p = 0.064, η² = 0.10	F1.37 = 0.25, p = 0.75, η² = 0.01

ω_maxR – maximum angular velocity around the long axis in rotation movements, t_ω_maxR – time needed to achieve maximum angular velocity around the long axis (roll), ω_maxY – maximum angular velocity around the sagittal axis of the athletes in yaw movements (yaw rotation), t_ω_maxY – time needed to achieve maximum angular velocity around the sagittal axis of the athletes in yaw movements (yaw rotation), ab_max – acceleration along the transverse axis in yaw movments, A_maxR – maximum angle around the long axis in rotation movements, t_A_maxR – time needed to achieve maximum angle around the long axis in rotation movements.

Yaw rotation

Significant changes were observed in the variables describing yaw rotation (yaw). Maximum acceleration in the direction of the transverse axis (ab_max) decreased in the experimental group while an increase in this variable was observed in the control group (interaction GROUP x MEASUREMENT: F_1.37 = 7.89 p = 0.008, η² = 0.18). At the same time, there were no significant differences between the groups in the time needed to achieve the abmax (Time_ab_max). The post-hoc test (GROUP × MEASUREMENT) for the ab_max variable showed a significant difference (p = 0.035) in the experimental group between measurements before and after the application of technical tethered swimming.

In addition, post-hoc analysis for the MEASUREMENT × GROUP × SIDE interaction for the variable ω_maxY showed a significant difference (p = 0.024) in the experimental group after the application of tethered technical swimming during the active phase of the left arm.

A significant difference (p = 0.024) was also observed between the experimental and control groups after the training intervention, in the value of ω_maxY during the active phase of the left arm.

The study showed an interaction (GROUP × MEASUREMENT × SIDE: F_1.37 = 4.15, p = 0.048, η² = 0.10) in maximum angular velocity around the vertical axis (ω_maxY). Post-hoc analysis for the MEASUREMENT × GROUP × SIDE interaction showed a significant difference (p = 0.024) in the value of ω_maxY during the active phase of the left arm in the experimental group after the application of tethered technical swimming. For the same interaction, a significant difference (p = 0.024) was observed between the experimental group and the control group after the training intervention, in the value of ω_maxY during the active phase of the left arm. No significant differences were observed between the CG and EG in ω_maxY during the active phase of the right arm and in the time required to reach ω_maxY (Time_ω_maxY).

Body roll

In the study, a significant MEASUREMENT × GROUP interaction (F_1.37 = 5.48, p = 0.025, η² = 0.13) indicated that the groups reacted differently to the intervention in the maximum pelvic angle about the vertical axis (A_maxR).

The study showed a significant interaction (MEASUREMENT × GROUP x SIDE: F_1.37 = 5.57, p = 0.024, η² = 0.13) in the maximum angular velocity around the vertical axis (ω_maxR).

The post-hoc test (MEASUREMENT × GROUP × SIDE) showed a difference (p = 0.008) in ωmaxR during the active phase of the left arm in the experimental group after the application of technical tethered swimming.

Training control

During the technical tethered swimming (TTS) series, the subjects in the experimental group achieved an average of 48.8 ± 12.45 N, scoring 1.5 ± 0.51 on the OMNI scale, while the control group scored technical training in free swimming at 1.36 ± 0.49 on the same scale. The Mann-Whitney U test showed no significant difference between the groups (p = 0.49).

Discussion

The results of the study showed no significant differences in the time for swimming 50 m front after an 8-week training cycle in either the experimental group (EG) or the control group (CG).

The lack of change in time results suggests that the time achieved over this distance alone is not directly dependent on the training used. Although technique-related parameters changed, they did not translate into a change in athletic performance. Although immediate performance gains were not observed, the observed kinematic changes are practically relevant for long-term technical development in adolescent swimmers. The lack of time results may be due to the complexity of the training process. The periodization of the training was not intended as a pretest and posttest. The timing of the training intervention was selected to allow for an uninterrupted 8-week training cycle. Breaks in training were determined by the school calendar. As a result, participants were not at optimal readiness for competition.

The absence of significant improvements in 50 m performance times may be partially explained by the high training loads undertaken during the intervention period. As illustrated in Table 1, swimmers completed weekly volumes ranging from ~ 22,000 to 33,000 m. Importantly, the post-test was conducted immediately after Week 8, when the total training volume (29,280 m) was higher than the mean weekly load across the intervention (~ 27,000 m). Testing under these conditions of elevated load and accumulated fatigue likely limited the transfer of improved kinematics into measurable sprint performance.

Yaw rotation

The use of accelerometers placed on the athletes’ backs provides a modern tool to accurately analyse yaw rotation³¹. The basis for the analysis of these movements is the angular velocity about the sagittal axis. An auxiliary variable in the description of yaw rotation is the acceleration about the transverse axis, which describes the dynamic changes in the transverse movement of the body³². stated that acceleration is only one element that influences yaw rotation analysis. They also demonstrated that elite swimmers had more controlled and less variable yaw rotation than less experienced swimmers, and that high yaw stability helped to maintain a faster swimming pace with fewer arm cycles.

To describe the dynamics of yaw rotation in detail, two additional variables were used: the time needed to reach maximum angular velocity, and the time needed to reach maximum acceleration in the transverse axis direction.

In the study, significant changes in variables describing yaw rotation were observed. Acceleration in the transverse axis direction (ab_max) decreased in the EG after technical tethered swimming by 4% in the first length during the active phase of the left arm (I25Left) and 10% during the active phase of the right arm (I25Right), and decreased by 7% in the second length during the active phase of the left arm (II25Left) and 6% during the active phase of the right arm (II25Right). A decrease of this value indicates improved trunk stabilization. Athletes swam more economically, by lowering the force needed to overcome water resistance in lateral movements.

At the same time, the study showed significant differences between the groups in the maximum angular velocity around the sagittal axis (ω_maxY) during the active phase of the left arm. In the EG, a decrease of 5.7% was observed in I25Left and II25Left. In contrast, an increase of 7.8% in I25Left and 6.0% in II25Left was observed in the CG. These changes indicate that technical tethered swimming (TTS) may have helped participants to improve control of body rotation, but the lack of significant difference in the active phase of the right arm suggests that these changes are asymmetrical and mainly involve movements during the active phase of the left arm, which may be related to the respiratory phase. As we know from research, athletes prefer a unilateral breathing phase, especially during maximal efforts (87% of those surveyed were right-handed)^33,34. A decrease in ab_max with a simultaneous decrease in ωmaxY may be indicative of a more stable body position in the water, which reduces energy loss due to excessive lateral movements, after application of TTS.

Body roll

Based on the findings of a computer simulation study, the authors suggested that body rotation about the vertical axis can have a significant effect on hand trajectory, promote the development of propulsive forces and therefore improve swimming performance²⁷. In the literature, angular velocity and maximum pelvic angle around the vertical axis are used to describe body roll^26,35. The description was extended with variables describing the time required to reach the maximum angular velocity around the vertical axis. This variable enriches the analysis with the dynamics of the swimmer’s movement.

The study revealed a significant difference between groups in the maximum angle of pelvic roll relative to the vertical axis (A_maxR). A_maxR in the EG decreased by 5.9% in I25Left, 3.4% in I25Right, 3.7% in II25Left and 2.2% in II25Right, while increasing by 4.0% in I23Left, 5.9% in I25Right, 3.8% in II25Left and 4.8% in II25Right in CG. A reduction in this value may indicate better stabilization, reduced water resistance and increased efficiency. We know from research that hip rotation is inversely correlated with swimming speed²⁹, which has a positive effect on improving sports performance. There was a significant increase in the value of ω_maxR in the CG of 4.29% in I25Left and 1.24% in II25Left, and a decrease in the value of this variable in the EG of 6.0% in I25Left and II25Left. The significant differences observed between the groups in movements during active left arm work suggest better trunk stabilization. The less effective stabilization is probably caused by the unilateral breathing phase in the athletes.

Practical significance of kinematic changes

Technical asymmetries in front crawl

The asymmetrical changes observed in this study, particularly during the active phases of the left arm, highlight the influence of unilateral breathing patterns on trunk stability. Most swimmers adopt right-side breathing^33,34, which may increase instability and accentuate asymmetry on the contralateral side. Our results are consistent with this notion, as greater improvements were observed during left-arm phases. While such asymmetries are common in front crawl, future research should apply established asymmetry assessment methodologies to quantify their magnitude and determine their long-term implications for technical efficiency and injury risk.

In our study, percentage reductions of 4–10% in yaw variables and 3–6% in body roll parameters were observed. Previous studies indicate that changes of this magnitude are biomechanically meaningful: Andersen et al²⁹. reported 3–7% differences in hip/shoulder roll across swimming speeds, while Barden and Barber³³ showed 5–10% alterations in hip roll depending on breathing laterality. Similarly, Hyodo et al³⁶. demonstrated that even ~ 5% changes in trunk twist correlated strongly with swimming velocity. Taken together, these findings suggest that the percentage changes observed in the present study are practically relevant, reflecting improvements in trunk stability and a potential reduction in hydrodynamic drag in adolescent swimmers.

Optimal range of body roll

Our results demonstrated a reduction in body roll amplitude (AmaxR) and angular velocity (ωmaxR), particularly during the left-arm phase, which we interpret as improved trunk stability in adolescent swimmers. Nevertheless, excessive or uncontrolled body roll has been associated with increased hydrodynamic drag and reduced efficiency²⁷. Recent research indicates that body roll should not be indiscriminately minimized but rather maintained within an optimal range. For instance, Gonjo et al.³⁷ reported that roll amplitude decreases with increasing swimming speed in elite swimmers, while He & Cheng²⁶ highlighted the importance of adequate trunk rotation for maintaining effective stroke mechanics. Taken together, these findings suggest that the reductions observed in our study reflect enhanced control and stabilization, and should be interpreted as adaptations toward greater efficiency rather than a universal reduction of body roll.

Limitation

Although the a priori power analysis indicated that a minimum of 34 participants (17 per group) would be sufficient to detect meaningful effects, and our final sample of 39 exceeded this threshold, the relatively modest sample size should still be considered a limitation. The number of participants may restrict the generalizability of the findings, particularly to other age categories or competitive levels. Moreover, recruiting substantially larger groups of adolescent swimmers of comparable age and training background is highly challenging in applied sports settings, where training squads rarely exceed 40 athletes and interventions must be coordinated with regular training and competition schedules.

Another limitation is that, although the accelerometer/gyroscope system used in this study has been previously validated in swimming kinematics research³¹, we did not conduct a separate reliability analysis (e.g., test–retest) on our sample. This may restrict the strength of methodological inferences, and we suggest that future studies should include such analyses to establish intraclass correlation coefficients (ICC) for kinematic variables and thereby strengthen methodological rigor.

The repeated-measures ANOVA design with multiple statistical comparisons, although supported by effect size and confidence interval reporting, may have increased the probability of Type I error. We did not perform any additional correction procedures (e.g., Bonferroni) across all dependent variables, as the outcomes represented conceptually distinct domains (time performance vs. kinematic parameters) and were not interpreted jointly. Each variable was considered separately, and the absence of significant effects (e.g., for performance time) was itself an important finding. While this approach avoids excessive Type II error, we acknowledge that it carries an increased risk of Type I error, which has been explicitly noted as a limitation.

Another limitation is that the intervention was scheduled in accordance with the swimmers’ regular training process rather than designed to maximize pre–post performance testing; while this may have limited direct performance improvements, it also represents a strength by reflecting the practical conditions of applied coaching.

The use of Fisher’s LSD for post-hoc testing, while common in exploratory intervention studies in adolescent swimmers, is less conservative than alternative approaches. This limitation should be considered when interpreting the present findings, and future research may benefit from applying more stringent procedures (e.g., Tukey’s HSD or Bonferroni correction) to increase statistical robustness.

Despite these limitations, the study provides novel insights into the use of technical tethered swimming in adolescent swimmers and offers practical implications for coaches.

Moreover, the post-test was conducted during a week of higher-than-average training load, which likely increased fatigue and may have masked performance transfer. Additionally, asymmetry was not formally quantified using established indices, which should be addressed in future studies.

Suggested future directions

In addition to external training load, which in our study was systematically quantified as weekly volume and intensity distribution (Table 1), future research could also incorporate internal load metrics such as session-RPE. Recent studies have also emphasized the usefulness of simple in-water indices for monitoring training adaptations³⁸.

Although HR monitoring is often used by coaches to control low-intensity zones, previous studies^39,40 suggest that HR may not reliably reflect the total training stress in swimming, particularly in adolescents. An integrated approach combining external and internal load measures may therefore provide further insight into the interaction between training stress and technical adaptations.

Future studies should employ multi-sensor or multi-dimensional analyses, which would enable a more comprehensive assessment of swimming kinematics, beyond the single pelvic-mounted accelerometer used here. We appreciate the reviewer’s observation. We agree that a single 45-minute TTS session per week represents a relatively small proportion of the swimmers’ total training volume. This dosage was chosen deliberately to integrate the intervention into regular training without overloading young athletes. Despite this, significant changes in kinematic variables were observed, suggesting that even a low-frequency protocol can have measurable effects. We have acknowledged this limitation in the revised Discussion and noted that future studies should explore higher training frequencies and different integration models to better determine the optimal TTS dosage.

Although prior validation studies support the use of this device, we acknowledge that external validity may vary depending on age, level, or swimming style. Future research should therefore include formal test–retest reliability, criterion validity against motion capture or high-speed video, and expanded testing across different conditions.

Conclusion

As we know from research²⁸, during long maximal efforts, numerous adverse changes in swimming technique occur in adolescent athletes. Based on the literature^27–35,41, it can be concluded that the changes observed after the application of the training protocol with technical tethered swimming (TTS) are positive. Nevertheless, they did not significantly improve the sports result in relation to the group performing technical training in free swimming. However, this may be related to the training loads applied before the end of the experiment according to the training periodization (Table 1).

The changes observed during the active phase of the left arm are likely to be the result of a correction in the technical errors acquired during the earlier training process that occur under the influence of asymmetric breathing pattern in athletes. It is probable that after using technical tethered swimming (TTS), the athletes made better use of the arm pull during the breathing phase by making smaller yaw rotation.

The changes observed during tethered technical training can improve swimming performance and reduce hydrodynamic drag²⁶, even if they are not immediately apparent in improved athletic performance. This suggests that TTS can be a valuable part of a training programme. The method helps permanently correct technical errors and improves swimming mechanics. It bases exercises on a whole-movement teaching practice. Training based on whole movement teaching practice tends to promote greater improvements in executive functions. This may result from the need for integrated planning and precise temporal coordination required to perform entire motor sequences, which are less emphasized in part-based practice⁴².

The performance of TTS exercises is characterized by low intensity, comparable to the intensity performed during technical exercises in free swimming. It is worth considering in the future whether extending the time of such training in the training microcycle can significantly improve athletic performance. It is also important to study how TTS affects swimming technique in the short term, for example when used as part of a swimming warm-up. In addition, the impact of this method on kinematics and athletic performance in other swimming styles would need to be assessed. Understanding these relationships can help to better match training methods to individual players’ needs and maximize their potential.

We did not apply a global Bonferroni-type correction across all dependent variables, as the outcomes represented conceptually distinct domains (time performance vs. kinematic parameters), and their interpretations were not conditional upon each other.

An important methodological consideration concerns the statistical approach. The use of repeated-measures ANOVA with multiple within-subject factors inherently increases the risk of Type I error due to multiple comparisons. To mitigate this limitation, we reported effect sizes and confidence intervals alongside p-values, which allow for a more robust interpretation of the findings. Furthermore, although the a priori power analysis indicated that our sample size (n = 39) was sufficient to detect medium effect sizes, we recognize that more complex interaction effects may remain relatively underpowered. This should be considered when generalizing the results, and future studies with larger cohorts and complementary statistical approaches (e.g., mixed-effects modeling or correction procedures for multiple testing) are warranted.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M.S. and J.G. conceptualized the study. M.S. designed the research protocol and conducted data collection. M.S., J.G., and S.K. performed data analysis. M.S. prepared the figures and wrote the main manuscript text. M.S., J.G., and S.K. reviewed and approved the final version of the manuscript. J.G. supervised the project in the role of academic advisor.

Funding

This research was funded by the Polish Ministry of Education and Science in the years 2023–2024 under the University Research Project no. 3 at Józef Piłsudski University of Physical Education in Warsaw, Poland: “Postural assessment and accelerometric characterisation of movement technique in selected sports disciplines.”

Data availability

The dataset supporting the findings of this study has been deposited in the RepOD repository under the DOI: 10.18150/E3FL3W. Data link: https://doi.org/10.18150/E3FL3W.

Declarations

Competing interests

The authors declare no competing interests.