Content uploaded by Augusto Barbosa
Author content
All content in this area was uploaded by Augusto Barbosa on Feb 18, 2019
Content may be subject to copyright.
Original Article
The road to 21 seconds: A case report
of a 2016 Olympic swimming sprinter
Augusto Carvalho Barbosa
1,2
, Pedro Frederico Valada
˜o
1,3
,
Carolina Franco Wilke
1,4,5
, Felipe de Souza Martins
1
,
Dellano Ce
´zar Pinto Silva
1
, Scott Alexander Volkers
1
,
Cla
´udio Olı
´vio Vilela Lima
1
, Jose
´Ricardo Claudino Ribeiro
1
,
Nata
´lia Franco Bittencourt
1
and Renato Barroso
6
Abstract
This study aimed to describe training characteristics as well as physical, technical and morphological changes of an elite
Olympic swimming sprinter throughout his road to 21 s in the 50 m freestyle. Over a 2.5-year period, the following
assessments were obtained: external training load, competitive performance, instantaneous swimming speed, tethered
force, dry-land maximal dynamic strength in bench press, pull-up and back squat and body composition. From 2014 to
2016, the athlete dropped 3.3% of his initial best time by reducing total swimming time (i.e. the total time minus 15-m
start time – from 17.07 s to 16.21 s) and improving the stroke length (from 1.83m to 2.00m). Dry-land strength (bench
press: 27.3%, pull-up: 9.1% and back squat: 37.5%) and tethered force (impulse: 30.5%) increased. Competitive perform-
ance was associated to average (r ¼0.82, p ¼0.001) and peak speeds (r ¼0.71; p ¼0.009) and to lean body mass
(r ¼0.55; p ¼0.03), which increased in the first year and remained stable thereafter. External training load presented a
polarized pattern in all training seasons. This swimmer reached the sub-22 s mark by reducing total swimming time,
which was effected by a longer stroke length. He also considerably improved his dry-land strength and tethered force
levels likely due to a combination of neural and morphological adaptations.
Keywords
Biomechanics, performance, sport, strength, tethered swimming, training
Introduction
The 50 m freestyle is the fastest event in competitive
swimming and was introduced into the Olympic
Games in Seoul 1988, when Matt Biondi set the new
World Record with 22.14 s. Two years later, Tom Jager
became the first man to swim this event under 22 s
(21.98 s) and until December of 2017 a total of only
76 swimmers reached this mark. The ability to swim
the 50 m freestyle under 22s amongst men has become
imperative to succeed in international events; therefore,
effort has been put into better describing the determinant
factors of a successful performance in this race.
Recently, several studies verified the influence of iso-
lated factors on sprint swimming performance, such as
race analysis,
1,2
swim kinematics,
3
tethered force,
4
dry-
land strength and power,
5–7
training load distribution,
8
body composition
9,10
and inter-arm coordination.
11
To
some extent, these findings have provided a stronger
basis for planning sprinters’ training programs.
However, the competitive level of the swimmers tested
Reviewer: In
˜igo Mujika
Ryan Atkison
Tiago Barbosa
1
Sport Sciences Department, Minas Te
ˆnis Clube, Belo Horizonte, Brazil
2
Meazure Sport Sciences, Sao Paulo, Brazil
3
Neuromuscular Research Center, Faculty of Sport and Health Sciences,
University of Jyva
¨skyla
¨, Jyva
¨skyla
¨, Finland
4
Exercise Physiology Laboratory, Federal University of Minas Gerais, Belo
Horizonte, Brazil
5
Sport and Exercise Discipline Group, Faculty of Health, University of
Technology Sydney, Sydney, Australia
6
Department of Sports Science, School of Physical Education, State
University of Campinas, Campinas, Brazil
Corresponding author:
Augusto Carvalho Barbosa, Sport Sciences Department, Minas Te
ˆnis
Clube, Rua da Bahia, 2244, Lourdes, Belo Horizonte, MG 30.160-012,
Brazil.
Email: augusto.barbosa@meazure.pro
International Journal of Sports Science
& Coaching
0(0) 1–13
!The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1747954119828885
journals.sagepub.com/home/spo
in some of these studies may hinder the application of
these experimental findings for elite athletes.
12
Studies comprising elite swimmers typically present
cross-sectional data
13–15
and/or do not focus on the
integration of the different aspects of an athlete’s
preparation. Differently, Slominski and Nowacka
16
described in detail the external training load planned
and implemented within a four-year cycle that led one
swimmer to European, World and Olympic medals.
Similarly, Hellard et al.
8
quantified the external training
load of the last 11 weeks leading to the main competi-
tions of 20 seasons (from 1991 to 2012) in 138 elite
swimmers, modelled its relationship to swimming per-
formance and provided useful insights for planning
sprint, middle-distance and distance training programs.
Indeed, the description and the interpretation of
long-term data of elite athletes can help to understand
key features of swimming performance. However, to
date, there is no study concerning elite sprinters’ train-
ing load and its effects on sprint performance-related
variables, such as swim kinematics, tethered force, dry-
land strength and body composition. Considering the
small number of swimmers who have been able to swim
the 50 m freestyle under 22 s and the lack of docu-
mented technical and physical modifications along
this way to 21 s, the aim of this study was to describe
training characteristics and physical, technical and
morphological changes of an elite Olympic sprinter
throughout his road to 21 s in 50 m freestyle.
Methods
Participant
The male swimmer in this investigation (age: 25 years,
height: 1.80 m, arm span: 1.80 m) has been ranked in
the top 50 in the World since 2014 and achieved his best
ranking in 2016 (14th) with his first sub-22 s mark
(21.82 s, 880 FINA points). He joined our training pro-
gram in 2014, when his best time was 22.57 s (795 FINA
Points) obtained in 2011. He had been involved in sys-
tematic training for 10 years and did not present any
injuries during the studied period. He provided verbal
and written consent to participate in this study and
procedures received approval of Campinas State
University’s Ethics Committee.
Study design
This case study had a retrospective longitudinal obser-
vational design and was carried out from January 2014
to September 2016. The following assessments were
obtained to monitor the athlete’s training responses
(Figure 1): external training load, competitive perform-
ance, instantaneous swimming speed, tethered force,
dry-land maximal dynamic strength and body compos-
ition. Due to logistical restrictions in 2016, the second
tethered force measure was obtained in February 2017.
External training load
The training load prescribed by the coach was analysed
and the in-water training volume was divided into three
intended intensity zones
17
: zone 1 – low lactate zone:
intensity below the first ventilatory threshold, typically
blood lactate below <2 mmolL
1
; zone 2 – lactate
accommodation zone: from the first to the second ven-
tilatory threshold, typically blood lactate between 2 and
4 mmolL
1
; and zone 3 – lactate accumulation zone:
above the second ventilatory threshold, where blood
lactate production exceeds maximum clearance rates,
typically blood lactate above 4 mmolL
1
. The training
loads executed during the week of the main competition
were not computed. Dry-land strength training load
complied with American College of Sports Medicine
(ACSM) guidelines
18
(Table 2).
Competitive performance
The competitive performance was evaluated by elec-
tronic timing in 50 m freestyle long course official com-
petitions. The first 2014 National Championship was
used as baseline (athlete’s first major competition
after joining our squad) whereas the first, the second
and the third 2015 National Championships, the first
2016 National Championship and the 2016 Olympic
Games (heats) were used to follow up his competitive
performance (Figure 1).
Figure 1. Timeline of experimental procedures.
C: competitive performance; ISS: instantaneous swimming speed; TF: tethered force; 1RM ¼maximal strength; BC: body composition.
2International Journal of Sports Science & Coaching 0(0)
Due to organizational limitations of the competi-
tions, only one camera was used to video record the
races. It was placed at the 25 m mark on a tripod and
an operator rotated it to follow the swimmer’s displace-
ment. The acquisition frequencies were 30 (in 2014, in
the first and second 2015 National Championships, and
in the 2016 Olympic Games – video provided by the
Brazilian biomechanist in Rio 2016) and 60 Hz (in the
third 2015 National Championship and in the first 2016
National Championship – cameras were upgraded).
Digital lines were superimposed onto the videos at 15,
25 and 35 m on Kinovea (v.0.8.24, Paris, France) using
gold-standard references positioned on both sides of
the pool.
Start time was considered from the start signal (i.e.
the light emitted by the official start system that was
visible by the camera) to the instant in which the swim-
mer crossed the digital line at the 15-m mark.
1
Split
times were attained from 0–25 m and 25–50 m, respect-
ively, whereas clean swimming was measured between
15 m and 35 m. The centre of the swimmer’s head was
the reference for the assessments at 15, 25 and 35 m.
Stroke rate was obtained from the time to complete
eight cycles, whereas stroke length was calculated as
the ratio between swimming speed and stroke rate,
1
both between 15 and 35 m. Stroke count corresponded
to the total of strokes performed during the race. Total
swimming time was obtained by subtracting the start
time from the final time.
The perspective error was accounted in a pilot study
which compared this one-camera approach to another
with three fixed cameras, positioned at 10 m (camera
1¼15 m’s view), 25 m (camera 2 ¼25 m’s view) and
40 m (camera 3 ¼35 m’s view). Seven swimmers per-
formed one 50 m maximal swim, so the times at the 15,
25 and 35-m marks could be obtained. Comparison
between both methods indicated low typical errors of
measurement (15 m ¼0.02 s, 25 m ¼0.01 s and
35 m ¼0.01 s), low coefficient of variations (CVs;
15 m ¼0.3%, 25 m ¼0.1% and 35 m ¼0.2%) and very
high intra-class correlation coefficients (ICCs;
15 m ¼0.95, confidence interval (CI) 95% ¼0.70–0.99,
F¼39.796, p ¼0.0002; 25 m ¼0.99, CI 95% ¼0.93–
1.00, F ¼191.414, p <0.0001; 35 m ¼0.96, CI
95% ¼0.77–0.99, F ¼54.407, p ¼0.0001).
In a second experiment, we also estimated the intra-
examiner reliability of the one-camera approach, which
was considered excellent for start (0–15 m: typical error
of measurement ¼0.01 s, CV ¼0.2%, ICC ¼0.94, CI
95% ¼0.71–0.99, F ¼34.480, p ¼0.0001), split
(0–25 m: typical error of measurement ¼0.01 s,
CV ¼0.1%, ICC: 0.99, CI 95% ¼0.94–1.00,
F¼174.323, p <0.0001 and 25–50 m: typical error of
measurement ¼0.01 s, CV ¼0.1%, ICC ¼1.00, CI
95% ¼0.98–1.00, F ¼496.935, p <0.0001) and clean
swimming times (15–35 m: typical error of measure-
ment ¼0.03 s, CV ¼0.3%; ICC ¼0.95, CI
95% ¼0.76–0.99, F ¼43.970, p <0.0001).
Instantaneous swimming speed
Instantaneous speed was measured (Figure 1) with a
speedometer
19
(CEFISE, Nova Odessa, Brazil)
attached to the swimmer’s hips during one push-off
20–25 m maximal sprint with breath held and self-
selected stroke rate. An analogic underwater camera
attached to a monopod recorded the trial from the
15-m mark (the operator rotated it to follow the swim-
mer’s displacement). A custom-designed software
received and synchronized both filtered speed (50–
240 Hz – system’s sampling capacity was improved
over time) and video data (30 Hz). Video and speed
data were synchronized by interpolation.
The first two strokes after the break-out were dis-
carded to attenuate push off effects. Main points
(Figure 2) were obtained in six consecutive complete
cycles (i.e. the interval between two successive lowest
points in the velocity–time curve) for the assessment of
peak speed (the highest speed value between two con-
secutive minimum speed values, CV ¼0.42%), minimum
speed (the minimum speed value found immediately after
hand’s entry in the water – CV ¼0.14%) and average
speed (the average of all speed values within a cycle –
CV ¼0.03%). The average of six cycles of each variable
was retained for analysis. Results from the tests that
occurred within three weeks of any competition were
used for correlation analysis.
Tethered force
A fully tethered swimming system (200 Hz; CEFISE,
Nova Odessa, Brazil) evaluated the tethered force
on June 2014 and February 2017 (Figure 1). The test con-
sisted of a 10-s maximal swimming with breath held and
self-selected stroke rate. All testing procedures complied
with previous investigation.
20
Peak force (CV ¼3.4%),
average force (CV ¼8.4%), impulse (CV ¼0.2%), rate
of force development (RFD, CV ¼2.6%) and stroke dur-
ation (CV ¼8.9%) were assessed during cycles (i.e. the
interval between two successive lowest points in the
force-time curve), as used previously.
20
Main points
used for analysis are shown in Figure 3. Variables from
the trial with the highest impulse were retained for
analysis.
Dry-land maximal dynamic strength
Absolute maximal dynamic strength was assessed with
1RM for pull-up, bench press and back squat in the
same testing session, interspaced by 10 min. All these
Barbosa et al. 3
exercises were mentioned in previous investigations, as
reported by a recent review about resistance training in
swimming.
21
Relative maximal dynamic strength was
obtained by dividing the total load lifted by the swim-
mer’s body mass. After a standard warm-up performed
prior to each exercise (eight repetitions with 60%
1RM þfour repetitions with 80% 1RM with 3 min of
rest), the athlete performed up to five attempts with
3 min of rest. The load was progressively increased
until a failed attempt terminated the test. Tests were
performed once a year: (1) July 2014, (2) July 2015
and (3) September 2016 (Figure 1). In 2016, the staff
decided to test the athlete in September (i.e. after the
Olympic Games) because he spent a relative long time
competing in Europe in June, when the training sched-
ule and loads had to be adjusted. So, July was strategic
for improving his strength and conditioning towards
the Olympic Games.
Body composition
Body composition was measured (Figure 1) by the same
researcher in the mornings. Body mass was obtained
using an electronic scale (Toledo, Model 2096-PP,
Sao Paulo, Brazil). Seven-site skinfolds (chest, abdom-
inal, thigh, triceps, sub-scapular, supra iliac and mid
axillary) were measured with a plicometer (Lange,
Cambridge Scientific Instruments, Cambridge, MD) fol-
lowing ACSM’s
22
guidelines. Test–retest ICC of all
skinfolds exceeded 0.90. Afterwards, body density
23
and fat percentage
24
were estimated. Lean body mass
(LBM) was determined by subtracting absolute body
Figure 2. Typical speed–time curve. 1: peak speed points; 2: minimum speed points; 3: average speed range; L: entry of left hand in
the water; R: entry of right hand in the water.
Figure 3. Example of two consecutive cycles of front-crawl force-time curve and the main points used for analysis.
Fpeak: peak force; ImpF: Impulse; Fmin: minimum force.
4International Journal of Sports Science & Coaching 0(0)
fat from total body mass. The swimmer received nutri-
tional guidance throughout the study period and the
whole process was regularly monitored and adjusted
by a certified nutritionist. However, dietary intake
was not registered.
Statistical analysis
Absolute data and percent changes were used to present
time effects. Shapiro-Wilk was used to test data nor-
mality. The relationship between competitive perform-
ance and body composition variables was assessed
through Pearson coefficient correlation, interpreted as:
0 to 0.30 ¼small, 0.31 to 0.49 ¼moderate, 0.50 to
0.69 ¼large, 0.70 to 0.89 ¼very large and 0.90 to
1.00 ¼nearly perfect.
25
Regression analyses were per-
formed between competitive performance and peak,
minimum and average swimming speed. Significance
level was set at p <0.05. Analyses were conducted
using SPSS for Windows (Version 16.0).
Results
External training load
The years were divided into 2 (2016) or 3 preparations
(2014 and 2015), according to Brazilian’s main compe-
titions. The preparations ranged from 10 to 17 weeks
(13.9 2.0 weeks). The regular weekly training routine
comprised 8 to 11 sessions, which consisted of 6 to 8
in-water (40–120 min) and 2 to 3 dry-land sessions
(30–120 min). The external training load of 715 in-water
training sessions out of 818 performed by the swimmer
(i.e. 87.4%; 289 in 2014, 232 in 2015 and 194 in 2016)
was quantified (Table 1 and Figure 4). Due to technical
problems, some training sessions did not get registered,
especially in the second preparation of 2015 (Table 1
and Figure 4). In general, the average session volume
was 3272 773 m, in which Z1, Z2 and Z3 corres-
ponded to 89.7 2.9%, 0.6 1.0% and 9.7 2.8%,
respectively. Weekly volumes in Z1, Z2 and Z3 are
shown in Figure 4.
A typical dry-land cycle lasted 15 weeks (Table 2)
and was divided into three mesocycles plus two weeks
of taper. All sessions were supervised by a strength and
conditioning professional to ensure appropriate tech-
nique and loads.
Competitive performance
The swimmer raced the 50 m freestyle 21 times during
the study (Figure 5), obtained four personal bests
(1.4%, 0.5%, 0.3% and 1.2%, respectively)
and dropped 3.3% of his initial best time, from
22.57 s to 21.82 s. After approximately 2.5 years of
training, he first swam under 22 s in Olympic trials
(April 2016, 21.82 s), and again in the heats of the
Olympic Games (August 2016, 21.96 s). Results from
race analysis in 2014, 2015 and 2016 are shown in
Table 3.
Figure 4. Weekly volume in Z1, Z2 and Z3 in (a) 2014, (b) 2015 and (c) 2016.
C: main competition; *Number of unquantified sessions.
Barbosa et al. 5
Table 2. Description of the dry-land strength training.
Mesocycle Objective Training load
I
(2–4 weeks)
Morphological adaptation 3 sets x 10-8RM with 60 s rest and slow velocity (1s concentric / 3s eccentric)
Exercises: 5 per session comprising chest, shoulder, legs, back (2x). Example:
dumbbell bench press, shoulder press, back squat, low pulley row, pull up.
II
(4–6 weeks)
a) Maximal strength
and power
a) 3-4 sets x 6RM + 4(first 2 weeks); 3-4 sets x 3(1RM*) + 4-10 Movement velocity
on the first part was slow due to the high resistance. Athlete was instructed to
try to move the weight as fast as possible though. Rest: 120-180 s.
yPower exercises: low resistance and high velocity, executed at the end of each set.
*Assisted repetitions: the strength coach provided help or resistance to keep
movement velocity constant and slow (duration: ~5s per repetition), ensuring
maximal voluntary muscle actions in each repetition (i.e. considering the force-
length curve of each exercise). Exercises: 5 per session comprising chest,
shoulder, legs and back (2x). Examples: bar bench press + push up, leg press + box
jump, hang clean, pull up + pulley pull down, medicine ball throws (e.g. shoulder
extension, shoulder horizontal adduction, shoulder abduction).
b) Power (strength
endurance)
b) 3 circuits, each with 5 exercises, 3 sets x 6-10, all with high concentric velocity,
Rest: 30-60 s. All exercises in the circuit were done in sequence; thus the rest
was only after all repetitions of each exercise was performed. Rest between
circuits was 3 minutes. Circuit example: pulley pullover (shoulder extension) +
pull up + burpees + medicine ball (shoulder extension) throw + pulley pull down
(shoulder extension + scapula adduction/downward rotation).
III
(3 weeks)
Maintenance Usually one session of each mesocycle per week. Session choice was based on the
athlete’s body composition, fatigue level, in-water performance and psycholo-
gical factors. When mesocycle 1: modified to 3 sets of 6RM. When mesocycle 2:
aorb.
IV
(2 weeks)
Taper First week: 2 short sessions (~30 minutes) 72 hours apart. 4 exercises in each
session. 2 sets x 6-12 (low resistance, high velocity). Rest: 120 s rest. Exercise
examples: unilateral or bilateral box jump, medicine ball throws (e.g. shoulder
extension, shoulder horizontal adduction, shoulder abduction), push-ups, pulley
pull down, pull-ups.
Note: In all mesocycles there were 2 strength sessions (2 sessions/week, ~90-120 minutes, interspaced by 72 hours) and a third session focusing on
trunk strength and preventative/postural exercises (1 session with 30-60 minutes, each with 2-3 sets of 6-10 repetitions and 60-90s of rest, 6 exercises,
e.g., shoulder internal and external rotation, trunk extension and flexion - isometric and dynamic, hip flexion, scapula posterior tilt and planks).
Table 1. Weekly volume and training intensity distribution of each preparation analyzed.
Year
Preparation Sessions Volume Z1 Z2 Z3
(duration) P/Q (m/wk) m/wk % m/wk % m/wk %
2014 1 (10 wks) 77/77 24,704 5463 22,109 5022 89.5 3.1% 95 96 0.4 0.4% 2500 964 10.1 2.9%
2 (15 wks) 114/113 24,339 4917 22,185 4499 91.2 1.6% 80 77 0.4 0.5% 2073 652 8.4 1.7%
3 (13 wks) 100/99 25,344 5281 23,393 4951 92.2 1.8% 31 63 0.2 0.3% 1921 564 7.6 1.6%
2015 1 (13 wks) 96/91 22,298 7167 20,283 6958 90.6 3.8% 188 282 1.1 1.9% 1826 736 8.3 2.9%
2 (17 wks) 126/46 27,608 2015 24,875 2325 90.0 2.0% 300 224 1.1 0.8% 2433 415 8.9 1.8%
3 (15 wks) 109/95 24,032 4454 21,092 4237 87.6 2.8% 46 166 0.2 0.7% 2894 698 12.2 2.6%
2016 1 (14 wks) 100/100 22,489 5966 19,893 5434 88.4 2.5% 143 122 0.7 0.6% 2453 821 10.9 2.5%
2 (14 wks) 96/94 21,317 6890 18,795 6176 88.0 2.2% 129 138 0.9 1.2% 2393 951 11.1 2.5%
P: prescribed sessions; Q: quantified sessions.
6International Journal of Sports Science & Coaching 0(0)
Instantaneous swimming speed
Data from all 18 assessments performed over the study
period are shown in Table 4. The swimmer used regular
trunks in the first 13 assessments and a competition suit
in the last 5. This may have influenced his speed and is
assumed as a limitation. Twelve speed tests were per-
formed close to competitions and were used for
Figure 5. Competitive performance in 50 m freestyle.
IT: initial best time; PB: personal best; LOC: local event; NAT: national event; TRIAL: Brazilian Olympic Trial; INT: international event;
OG: Olympic Games. Arrows indicate the main competitions.
Table 3. Comparison between race analysis in 2014, 2015 and 2016 and percent changes from April 2014 to April 2016.
APR
2014
APR
2015
AUG
2015
DEC
2015
APR
2016
SEP
2016
% 2014
vs. 2016
Official time (s) 22.71 22.35 22.14 22.08 21.82 21.96 3.9%
Start time 0–15 m (s) 5.64 5.81 5.60 5.65 5.61 5.64 0.5%
Split times (s)
0–25 m 10.37 10.51 10.16 10.27 10.10 10.14 2.6%
25–50 m 12.34 11.84 11.98 11.81 11.72 11.82 5.0%
Total swimming time (s) 17.07 16.54 16.54 16.43 16.21 16.32 5.0%
Clean swimming time (s) 9.64 9.40 9.24 9.31 9.09 9.17 5.7%
Stroke rate (c/min) 67.9 64.3 65.7 65.9 66.1 66.6 2.6%
Stroke length (m) 1.83 1.99 1.98 1.95 2.00 1.97 þ8.9%
Stroke count (n) 42 38 39 40 39 39 7.1%
Table 4. Matching competitive performance (CP), average (ASS), minimum (MSS) and peak (PSS) swimming speeds in all testing
sessions.
Year
2014 2015 2016
Month May Jun Jul Oct Jan Apr Jul Sep Oct Feb Feb
a
Mar Apr May Jun Jul Jul Aug
a
CP (s) 22.74 – – 22.94 – – 23.10 22.86 22.65 22.28 – 22.60 21.82 22.24 22.40 22.53 – 21.96
ASS (m/s) 1.90 1.94 1.85 1.87 1.90 1.88 1.91 1.87 1.89 1.92 1.92 1.93 2.01 1.93 1.98 1.97 2.04 2.03
MSS (m/s) 1.61 1.70 1.60 1.50 1.53 1.56 1.65 1.56 1.63 1.64 1.67 1.56 1.66 1.53 1.60 1.54 1.64 1.60
PSS (m/s) 2.15 2.21 2.11 2.19 2.20 2.13 2.23 2.22 2.21 2.27 2.25 2.29 2.31 2.27 2.32 2.28 2.38 2.32
a
Swimmer stopped before completing six strokes, so only four were considered.
Barbosa et al. 7
regression analysis (Figure 6). Very large correlations
were found between competitive performance and
speedometer average (r ¼0.82, CI 95%: 0.95/
0.46; p ¼0.001) and peak speeds (r ¼0.71; CI
95%: 0.91/0.23; p ¼0.009), whereas no relationship
was detected for minimum speed (r ¼0.28; CI 95%:
0.73/0.35; p ¼0.39).
Tethered force
Results concerning tethered force are shown in Table 5.
All tethered swimming test variables considerably
increased from 2014 to 2017, indicating that the train-
ing was effective for improving specific strength.
Dry-land maximal dynamic strength
Absolute maximal strength increased in all exercises
(Figure 7(a)). From 2014 to 2015, the percent changes
were 2.8% for pull-up, 6.4% for bench press and 8.3%
for back squat. Improvements from 2015 to 2016 were
6.2%, 19.7% and 26.9%, for pull-up, bench-press and
back squat, respectively, greater than those from 2014
to 2015. The same pattern was observed for relative
strength (Figure 7(b)) which increased 0.5% for pull-
up, 4.0% for bench press and 5.9% for back squat from
2014 to 2015, and 5.1%, 18.4% and 25.6% from 2015
to 2016, respectively.
Body composition
Body composition data are shown in Figure 8. The CV
for body mass, % fat, LBM and fat mass (considering
all measurements throughout the study period) were
1.1%, 12.8%, 1.5% and 13.2%, respectively. When
the swimmer reached the sub-22 s mark (i.e. April and
August 2016), his body mass, % fat, LBM, fat mass
values were 77.0 and 77.3 kg, 8.4 and 7.1%, 70.5 and
71.8 kg and 6.5 and 5.5 kg, respectively. Fifteen meas-
ures matched competitions dates (Figure 8) and were
used for correlation analysis. LBM correlated mod-
erately to competitive performance (r ¼0.55;
CI 95% ¼0.83/0.06; p ¼0.03), whereas no relation-
ship was detected for the other body composition
variables.
Figure 6. Regression analyses between competitive performance and (a) average, (b) minimum and (c) peak speeds.
Table 5. Tethered force variables.
Jun 2014 Feb 2017 D%
Peak force (N) 211.1 245.9 þ16.5%
Average force (N) 133.7 149.2 þ11.6%
RFD (N/s) 611.9 711.0 þ16.2%
Duration (ms) 463 552 þ19.2%
Impulse (Ns) 62.4 81.5 þ30.6%
RFD: rate of force development.
8International Journal of Sports Science & Coaching 0(0)
Discussion
Elite athletes’ data are seldom published for several
reasons including the time-consuming task of paper
writing and submission, the gap between researchers
and practitioners and/or the team’s policy of maintain-
ing data secret for competitive reasons. This is the first
study to describe training characteristics and long-term
changes in physical and technical variables of an elite
50 m freestyle swimmer, and our main findings were as
follows: (1) the external training load presented a polar-
ized distribution in all training cycles; (2) competitive
performance improved mainly by reducing total
swimming time; (3) clean swimming time (15–35 m)
improved due to a longer stroke length; (4) dry-land
strength and tethered force considerably increased;
(5) competitive performance correlated with average
and peak speed during testing and (6) LBM increased
in the first year, remained stable thereafter and was
inversely correlated to competitive performance.
The sprinter’s training was mostly performed in Z1
(89.7%), followed by Z3 (9.7%), and very little in Z2
(0.6%), which is similar to the polarized intensity
distribution model. The polarized model consists of
significant proportions of training at both high (15%–
20%) and low intensities (75%–80%), and only a small
proportion of threshold training (5%–10%).
26,27
This
pattern has been utilized by coaches from different
sports as a way to benefit from high intensity training
effects without using high volumes, which could lead to
overreaching or overtraining.
28
Although it is unclear
whether there is an optimal training load distribution
Figure 8. Body mass (a), % fat (b), lean body mass (c) and fat mass (d). White bars indicate assessments in both body composition
and competitive performance. Arrow points correspond to athlete’s personal best times.
Figure 7. Progression of maximal absolute and relative strength in all three exercises.
Barbosa et al. 9
for swimming sprinters, we observed that this pattern
succeeded in improving athlete’s performance herein.
Nevertheless, the current ‘‘90-0.5-9.5’’ distribution
across the three zones is considerably different from
those verified earlier for high-trained endurance
athletes (e.g. ‘‘75-5-20’’
17
and ‘‘80-0-20’’
27
) and for 100-
and 200-m elite swimmers (‘‘77-12-11’’
29
), indicating that
distribution across the three zones should be adjusted
according to athletes’ competitive requirements.
It is also interesting that the 9.5% of the volume
performed in the high-intensity zone is 10% lower
than that reported previously.
17
Considering that
during the 50-m race a large amount of energy is rap-
idly required (anaerobic contribution estimated at 96%:
ATP-PCr ¼38% and glycolytic ¼58%),
30
an increase
in zone 3 volume could be expected to further develop
the metabolic and neuromuscular mechanisms involved
in this task. However, the short duration of the race
also prevents the occurrence of high level of acidosis,
30
so in order to succeed the swimmer should be able to
generate and maintain the highest speed level as pos-
sible. Such characteristic is different from endurance
sports
17,27
in which a relatively high intensity (not max-
imum) should be maintained during long distances and
thus could explain the difference between the ‘‘90-0.5-
9.5’’ and the typical ‘‘75-5-20’’ polarized patterns.
Therefore, we suggest that the current balance between
zone 1 and zone 3 may allow swimmers to reach com-
petitive speed more frequently in training without get-
ting chronically fatigued throughout the season and
improve long-term adaptations by enhancing training
specificity.
The athlete obtained four personal best perform-
ances and dropped 3.3% of his initial best time
(22.57 s) during the study period. Despite reaching the
sub-22 s mark in Rio 2016, he was slower compared to
the Olympic trials (21.96 s vs. 21.82 s). Although both
cycles presented similar training intensity distributions,
weekly volume was lower in the preparation immedi-
ately before the Olympics (22 km vs. 21 km), espe-
cially in Z3 (60 m per week). The reduction of 60 m
per week over 14 weeks equals 840 m, which represent
1/3 of the volume performed in a week. This volume
also corresponds to 2.5% of the total volume swam in
Z3. Although we cannot be sure about this volume’s
effect, we believe it may be relevant and may have influ-
enced his performance as this intensity is more specific
for 50 m race. Accordingly, Hellard et al.
8
showed that
an increase in high intensity training load is important
for 50-m sprinters’ performance in the last 10 weeks
prior the main competition. In addition, it also likely
that psychosocial pressure of competing at home
during the Olympics Games may have impacted his
performance, especially in the semi-finals, when there
may have been a greater public expectation.
Race analysis revealed that performance enhance-
ment occurred mainly due to a progressive reduction
in total swimming time over time (17.07 s in 2014 vs.
16.21 s in 2016), which reached 5.0% in 2016.
Swimming speed is the combination of stroke rate
and stroke length, and despite a 2.6% decrease in
stroke rate, stroke length increased 8.9% at the end.
These results are in accordance with previous findings
that faster swimmers achieve greater distances per
stroke.
31
Additionally, such increase in stroke length
was likely related to his increased ability to produce
force in the tethered swim test. The greater impulse
observed in 2017 was a consequence of increased peak
force, RFD and stroke duration. We acknowledge that
the second tethered swimming test was considerably far
from the athlete’s best competitive performance and
that these variables may change over time, but we con-
sider conceivable that increased tethered force contrib-
uted to performance enhancement as it has been shown
to be sensitive to identify training-induced adaptations
in swimming.
32
Tethered swimming performance depends on the
force applied to the water, which is influenced by
both technique and the neuromuscular ability to pro-
duce strength, assessed herein through 1RM test.
Increases in maximal strength were detected for all
exercises and likely had an impact on both tethered
force and stroke length. Interestingly, changes in max-
imal strength of back squat and bench press were larger
than the improvements of tethered force variables,
whereas the pull-up maximal strength modification
was slightly lower. Force transference is still a challen-
ging topic in sports, even more in swimming due to the
great influence of technique on force application. It is
becoming clear though that factors such as body pos-
ition, type of muscle action and pattern of neural acti-
vation utilized in training affect transference to specific
motor tasks,
33
such as arm stroke and leg kicking.
Although it is not possible to determine the contribu-
tion of strength gains in each exercise to the increase in
tethered force variables, it is conceivable higher trans-
ference from pull-up due to its kinesiologic similarity to
the arm stroke. Accordingly, Perez-Olea et al.
7
verified
strong correlations between 50 m freestyle performance
and different mechanical variables of the pull-up in
both 1RM and maximum number of repetition tests.
Regarding strength gains, they possibly resulted
from a combination of morphological and neural adap-
tations. Although a direct measure of muscle morph-
ology was not available (e.g. cross-sectional area and/or
pennation angle), hypertrophy may be inferred by the
2 kg increase in LBM from 2014 to 2015. Changes in
muscle mass may not be directly related to strength
gains,
34
but may allow greater strength production
after a period of training to induce neural changes.
35
10 International Journal of Sports Science & Coaching 0(0)
Accordingly, all maximal strength levels kept increasing
from 2015 to 2016 despite no relevant changes in LBM.
Interestingly, the strength gains were greater from
2015 to 2016 than from 2014 to 2015. Although dry-
land training was designed to improve strength and
power in all cycles, it progressed carefully at the begin-
ning to avoid injuries. Over time, training intensity was
further increased and led to this greater strength
improvement from 2015 to 2016.
LBM moderately correlated with competitive per-
formance, suggesting that its increase may lead to a
lower time in competition. This seems reasonable
since increased LBM is attained mainly by augmenting
muscle mass, which is responsible for producing
strength and power. However, this result should be
carefully interpreted as LBM increased in the first
year and remained relatively stable thereafter, whereas
competitive performance continued improving.
Additionally, there was a considerable variation of
competitive performance (from 23.2 s to 21.8 s) for
similar LBM values (70.5 kg).
Body mass varied slightly over time. The difference
between the highest and the lowest values is 5% and
the CV was 1.1%, indicating that it remained stable
during both preparatory and competitive periods, and
that the swimmer trained and competed in very similar
conditions. Of note, his top five competitive perform-
ances were obtained with %fat around 8% (8.4%,
7.1%, 6.8%, 7.1% and 8.7%). Keeping body fat
within certain limits is important as it enlarges body
surface and may ultimately increase drag and reduce
swimming speed.
36
Moreover, fat mass always
increased after main competitions possibly due to a
week off from training for recovery to the next
preparation.
Data also revealed that speedometer average and
peak speeds can predict competitive performance.
Peak speed is attained when propulsion is greater
than drag at the maximum level within an arm-stroke.
Therefore, to swim faster, one should find the best body
position throughout the stroke cycle (i.e. drag reduc-
tion) and increase the ability to produce power (i.e.
neuromuscular capacity and technique). Our results
are reasonable as this swimmer improved performance
mainly by reducing total swimming time (0.86 s) with
a smaller change in start time (0.03 s). The athlete
reached the sub-22 s mark when peak speed reached
2.30 m/s, suggesting that swimmers should attempt
to approximate or be better than this ‘‘threshold.’’
Conversely, minimum speed is reached at the end of
entry phase (i.e. the beginning of hand’s backward
movement)
11
and lower values are likely associated to
non-favourable body and/or arm positions that may
increase drag and/or propulsive discontinuity.
Although minimum speed may differentiate regular
and elite sprinters, it seems to not have major influence
on the current swimmer’s average speed or competitive
performance. Future studies on elite-level swimmers are
encouraged to investigate whether such rationale is
individual-specific or can be applied to this population.
As much as these speed parameters may provide
world-class references, the data presented herein were
attained under rested conditions and may better repre-
sent the first half of the race, whereas the 50 m perform-
ance also depends on the second 25-m split. For
instance, the current swimmer had a greater time
improvement in the second half of the race (5.0%)
compared to the first one (2.6%). Therefore, the
swimmer should be able to achieve such minimum
and/or maximum values references and also keep
them in great levels under a more fatigued condition.
Finally, this study has inherent limitations of case
study designs. As only one swimmer was analysed, con-
clusions may vary according to individual’s strong and
weak physical and technical characteristics. Although
his competitive performance improved mainly due to
reduced total swimming time, other swimmers may
improve more by improving their starts, as it corres-
ponds to 30% of the total race distance. We also
acknowledge that more assessments of maximal
strength, tethered force and race analysis would pro-
vide a better understanding of the mechanisms under-
lying his road to 21 s. Additionally, internal training
load was not analysed and could have provided more
information about physiological effects of training.
37
Nevertheless, our results can be useful and revealing
since average data are not always capable of explaining
elite ‘‘outlier’’ performances.
Practical applications
This study has important practical applications as it
provides long-term training, testing and competitive
data of a world-class 50 m freestyle swimmer. Data
reported herein may be used as reference for setting
training characteristics as well as physical and technical
goals for teams and/or individuals. Additionally, this
study not only highlights the usefulness of simple, inex-
pensive and practical assessments such as anthropom-
etry and maximal strength for monitoring elite athletes’
progression but also points to the important role of
technology (e.g. speedometer, tethered swimming and
video analysis) on providing access to more complex
variables required within high-performance sports
context.
In line with previous studies on swimming biomech-
anics, our data also highlight the importance of increas-
ing stroke length to improve swimming performance,
which was likely achieved through an increased ability
to produce dry-land and in-water force levels.
Barbosa et al. 11
Our results also suggest that reaching adequate levels of
LBM is important, especially within competitive peri-
ods, as this variable may influence both drag and pro-
pulsive force. Hence, such multifactorial and complex
nature of 50 m swimming performance should be taken
into account by the coaching staff when planning indi-
vidual athlete’s training and testing programs.
Conclusions
This swimmer reached the sub-22 s mark mainly by
reducing total swimming time, which was effected by
a longer stroke length. He also considerably improved
his dry-land strength and in-water tethered force levels,
likely due to a combination of neural and morpho-
logical adaptations.
Acknowledgements
We would like to thank the participant for his unwavering
dedication and authorization the disclosure of such valuable
data, to Minas Tenis Clube for the support given to the tech-
nical staff in the preparation of this study, to Leonardo Leis
for his work on quantifying the external training load and to
Paulo Cezar da Silva Marinho for his cooperation with the
athlete’s race video in the Olympic Games.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
ORCID iD
Augusto Carvalho Barbosa http://orcid.org/0000-0003-
3406-8524
References
1. Morais JE, Marinho DA, Arellano R, et al. Start and turn
performances of elite sprinters at the 2016 European
Championships in swimming. Sports Biomech 2018; 1–15.
Epub ahead of print 26 March 2018. DOI: 10.1080/
14763141.2018.1435713.
2. Veiga S, Roig A and Gomez-Ruano MA. Do faster swim-
mers spend longer underwater than slower swimmers at
World Championships? Eur J Sports Sci 2016; 16:
919–926.
3. Craig AB Jr, Skehan PL, Pawelczyk JA, et al. Velocity,
stroke rate and distance per stroke during elite swimming
competition. Med Sci Sports Exerc 1985; 17: 625–634.
4. Morouco P, Keskinen KL, Vilas-Boas JP, et al.
Relationship between tethered forces and the four swim-
ming techniques performance. J Appl Biomech 2011; 27:
161–169.
5. Loturco I, Barbosa AC, Nocentini RK, et al. A correl-
ational analysis of tethered swimming, swim sprint per-
formance and dry-land power assessments. Int J Sports
Med 2016; 37: 211–218.
6. Morouco P, Neiva H, Gonzalez-Badillo JJ, et al.
Associations between dry land strength and power meas-
urements with swimming performance in elite athletes: a
pilot study. J Hum Kinet 2011; 29A: 105–112.
7. Perez-Olea JI, Valenzuela PL, Aponte C, et al.
Relationship between dryland strength and swimming
performance: pull-up mechanics as a predictor of swim-
ming speed. J Strength Cond Res 2018; 32: 1637–1642.
8. Hellard P, Scordia C, Avalos M, et al. Modelling of opti-
mal training load patterns during the 11 weeks preceding
major competition in elite swimmers. Appl Physiol Nutr
Metab 2017; 42: 1106–1117.
9. Strzala M, Stanula A, Krezalek P, et al. Influence of
morphology and strength on front crawl swimming
speed in junior and youth age group swimmers.
J Strength Cond Res, Epub ahead of print 8 July 2017.
DOI: 10.1519/JSC.0000000000002084.
10. Lowensteyn I, Signorile JF and Glitz K. The effect of
varying body composition on swimming performance.
J Strength Cond Res 1994; 8: 149–154.
11. Chollet D, Chalies S and Chatard JC. A new index of
coordination for the crawl: description and usefulness. Int
J Sports Med 2000; 21: 54–59.
12. Pyne DB, Lee H and Swanwick KM. Monitoring the
lactate threshold in world-ranked swimmers. Med Sci
Sports Exerc 2001; 33: 291–297.
13. Olstad BH, Zinner C, Vaz JR, et al. Muscle activation in
world-champion, world-class, and national breaststroke
swimmers. Int J Sports Physiol Perform 2017; 12:
538–547.
14. Toussaint HM and Truijens M. Power requirements for
swimming a world-record 50-m front crawl. Int J Sports
Physiol Perform 2006; 1: 61–64.
15. Seifert L, Vantorre J and Chollet D. Biomechanical ana-
lysis of the breaststroke start. Int J Sports Med 2007; 28:
970–976.
16. Slominski P and Nowacka A. The structure and volume
of training loads in the four-year training cycle of an elite
olympic athlete. Pol J Sport Tourism 2017; 24: 162–169.
17. Seiler KS and Kjerland GO. Quantifying training inten-
sity distribution in elite endurance athletes: is there
evidence for an ‘‘optimal’’ distribution? Scand J Med
Sci Sports 2006; 16: 49–56.
18. Garber CE, Blissmer B, Deschenes MR, et al. Quantity
and quality of exercise for developing and maintaining
cardiorespiratory, musculoskeletal, and neuromotor fit-
ness in apparently healthy adults: guidance for prescrib-
ing exercise. Med Sci Sports Exerc 2011; 43: 1334–1359.
19. Barbosa TM, Morouco PG, Jesus S, et al. The interaction
between intra-cyclic variation of the velocity and mean
swimming velocity in young competitive swimmers. Int J
Sports Med 2013; 34: 123–130.
20. Barbosa AC, Barroso R and Andries O Jr. Post-
activation potentiation in propulsive force after specific
swimming strength training. Int J Sports Med 2016; 37:
313–317.
12 International Journal of Sports Science & Coaching 0(0)
21. Crowley E, Harrison AJ and Lyons M. The impact of
resistance training on swimming performance: a system-
atic review. Sports Med 2017; 47: 2285–2307.
22. American College of Sports Medicine. ACSM’s guidelines
for exercise testing and prescription, 8th ed. Philadelphia:
Lippincott Williams & Wilkins, 2009, p.366.
23. Jackson AS and Pollock ML. Prediction accuracy of
body density, lean body weight, and total body volume
equations. Med Sci Sports 1977; 9: 197–201.
24. Siri WE. Body composition from fluid spaces and dens-
ity. In: Brozek J and Henschel A (eds) Techniques for
measuring body composition. Washington: National
Academy of Science, 1961, pp.223–244.
25. Hopkins WG, Marshall SW, Batterham AM, et al.
Progressive statistics for studies in sports medicine and
exercise science. Med Sci Sports Exerc 2009; 41: 3–13.
26. Stoggl T and Sperlich B. Polarized training has greater
impact on key endurance variables than threshold, high
intensity, or high volume training. Front Physiol 2014; 5: 33.
27. Neal CM, Hunter AM, Brennan L, et al. Six weeks of a
polarized training-intensity distribution leads to greater
physiological and performance adaptations than a thresh-
old model in trained cyclists. J Appl Physiol (1985) 2013;
114: 461–471.
28. Hydren JR and Cohen BS. Current scientific evidence
for a polarized cardiovascular endurance training
model. J Strength Cond Res 2015; 29: 3523–3530.
29. Mujika I, Chatard JC, Busso T, et al. Effects of training
on performance in competitive swimming. Can J Appl
Physiol 1995; 20: 395–406.
30. Rodrigues FA and Mader A. Energy system in swim-
ming. In: Seifert L, Chollet D and Mujika I (eds) World
book of swimming: from science to performance.
Hauppauge: Nova Science Publischers, 2010, pp.225–240.
31. Craig AB Jr, Skehan PL, Pawelczyk JA, et al. Velocity,
stroke rate and distance per stroke during elite swimming
competition. Med Sci Sports Exerc 1985; 17: 625–634.
32. Papoti M, Martins LE, Cunha SA, et al. Effects of taper
on swimming force and swimmer performance after an
experimental ten-week training program. J Strength Cond
Res 2007; 21: 538–542.
33. Cormie P, McGuigan MR and Newton RU. Developing
maximal neuromuscular power: part 2 – training consid-
erations for improving maximal power production.
Sports Med 2011; 41: 125–146.
34. Buckner SL, Dankel SJ, Mattocks KT, et al. The problem
of muscle hypertrophy: revisited. Muscle Nerve 2016; 54:
1012–1014.
35. Hakkinen K, Pakarinen A, Alen M, et al. Neuromuscular
and hormonal adaptations in athletes to strength training
in 2 years. J Appl Physiol (1985) 1988; 65: 2406–2412.
36. Toussaint H and Truijens M. Biomechanical aspects of
peak performance in human swimming. Anim Biol 2005;
55: 17–40.
37. Bourdon PC, Cardinale M, Murray A, et al. Monitoring
athlete training loads: consensus statement. Int J Sports
Physiol Perform 2017; 12: S2161–S2170.
Barbosa et al. 13
