Experimental Approach to the Problem
A randomized crossover and counterbalanced design were used to compare the effect of bent over barbell row as antagonist CA against agonist bench press and push-up CA on ball throwing distance. Before launching 3 of the main experiment trials, subjects participated in 10 weeks of resistance training program to be familiarized with testing protocols, improve chosen exercises’ techniques, and increase strength. Moreover, during the last week of familiarisation, subjects performed 3 medicine ball throws (as used during the experiment) after each training visit to familiarize themselves with the throwing technique.
There were three experimental sessions 48 hours apart, where CA was included in randomized order for individuals (Figure 1). After standardized warm-up subjects completed 3 experimental trials involving a baseline ball throwing with a 2kg medicine ball followed by a CA. The push-up CA was performed in 2 sets of 4 repetitions with 3 minutes of rest between sets. Plyometric and isometric push-up exercises used as CA significantly improved shot put performance among female throwers , to adapt this exercise to the strength of the subjects it has been decided to perform push-ups as described below. The bent-over barbell row (ROW) and bench press (BP) was performed in 2 sets of 4 repetitions with respectively 60% and 80% one-repetition maximum (1RM) with 3 minutes rest between sets. This intensity was chosen because both stronger and weaker individuals respond to CA better when stimuli are higher. Retesting of ball throwing was measured 3 times in total respectively after 3, 6, and 9 minutes. The load of 80% 1RM was chosen because it is enough to awake PAPE reaction with decreasing possible fatigue harmful effects on results . Recovery time was based on prior studies which show that gender and strength level influence rest time [6,11,26].
Power calculations indicated determined the minimum sample size of 12 participants would be required to detect an effect size of 0.4, collected from the average effect size reported in the Wilson metanalysis  (repeated measures, within-between interactions ANOVA power = 0.8, alpha = 0.05, correlation among rep measures = 0.8, number of groups = 3, number of measurements = 2; G*Power 188.8.131.52). Thirteen (n = 13) professional Czech softball players from National Team and 1st division Clubs volunteered in study (mean ± SD: age 22.2 ± 3.1 years; height 169 ± 4.5 cm; body mass (BM) 68.3 ± 11.3 kg; bench press 1RM 40.5 ± 6.8 kg; BP 1RM/BM 0.6 ± 0.1; bent over barbell row 1RM 39.5 ± 6.7 kg; ROW 1RM/BM 0.7 ± 0.1; softball experience 7.3 ± 2.4 years). Subjects were recruited on the basis that they were healthy, injury-free, and engaged in a resistance-training program for the last 10 weeks. They were able to perform bent-over barbell row and bench press with proper technique as assessed by certified strength and conditioning coach.
Experimental trials were separated by 48 hours from each other (Figure 1). Moreover, subjects were instructed to avoid upper body workouts during the time of measurements. In the beginning, participants started with a standardized warm-up protocol consisting of 5 minutes of running on a treadmill with a constant speed of 6 km/h followed by dynamic stretching with an emphasis on stretching the chest musculature. Two minutes after warm-up, subjects performed ball throw as pre-measurement, and 3 minutes after pre-measurements, participants in a randomized order completed CA of either bent over barbell row, bench press, or push-up. After PAP activation subjects rested for 3 minutes before starting POST measurements.
Conditioning exercises and one-repetition maximum measurements
1 RM measurements
1RM of bench press and bent over barbell row was measured at the end of 10 weeks of resistance training familiarisation. Before measuring 1RM, all participants underwent a standardized warm-up consisting of 5 minutes of running on a treadmill with a constant speed of 6 km/h followed by dynamic stretching with an emphasis on stretching the chest musculature. All subjects started the exercise with 8 repetitions at 50% 1RM measured during the preparatory 10 weeks resistance training period. Then subjects performed 4 repetitions at 70% 1RM and 3 repetitions at 80% 1RM, respectively. After the final warm-up, subjects began lifting 1RM with maintaining proper technique and a full range of motion with the weight starting from 5kg added to the previous 1RM. If an attempt was successful, 2.5kg were added with rest intervals between each attempt for 3 minutes until proper 1RM was reached. Bent over barbell row 1RM was measured using the same warm-up and measurement protocol as during bench press.
Bench press technique
During a bench press, subjects were instructed to lay prone on the bench with the leg resting on the floor where the knee was positioned at a 90º angle. The grip was pronated with hands spaced in the distance between each other of 1.5 widths of the shoulder. Subjects were required to control the bar’s descent until the chest was touched approximately 3cm above to xiphoid process and, without pause, push it. The cadence of the move was 2 seconds down and voluntary tempo up in the concentric phase controlled by the coach.
Bent over barbell row technique
The bent-over barbell row position required the head to lean on a bench with high adjusted to each person to maintain 90 degrees of flexion in the hip with knees slightly bent (Figure 2). Participants were instructed to pull the bar at the height of the bottom of the sternum. The grip position was the width of the shoulders. The cadence of the move was 1 second up and 2 seconds down.
Push up technique
The push-up (PU) exercise was performed with wide hands positioned on the floor. It was completed from a plank position with the body remaining straight from the head to the heels. The starting position was with the placement of hands on the shoulder line with fingers pointing forward. When the view from the side, hands fell directly below the shoulders. From this position, subjects were instructed to lower the body with elbows directed to the sides until their chest almost touched the floor. After reaching that point subject changed direction without pausing. The cadence of the move was maintained 2 seconds down and voluntary tempo in the concentric phase
Medicine ball chest throw performance
The medicine ball throwing test used during this measurement is recommended by Harasin et al.  to measure maximum throwing performance. The test was performed in a sitting position where both legs were placed on the ground with 90º flexion in the knee. During each throw head and shoulders had to be in touch with the wall, and the trunk had to be in touch with a chair. The 2kg medicine ball (circumference 65cm) was used for throwing as participants practiced this weight before. Each participant was instructed to throw the ball with maximal effort and possibly furthest. Throws started from the chest position with elbows abducted from the trunk (Figure.2). Measurement tape was placed on the ground to assess throw distance by an assistant with a distance meter. Each time ball had been thrown 3 times in the row without rest, the ball was passed to the participant immediately after a throw by an assistant. In front of the throwing position at the distance of 6m was a standing assistant holding the radar (The Stalker ATS II, Version 184.108.40.206, Applied Concepts, Dallas, TX, USA) evaluating throwing speed (m/s). The best of 3 throws from each protocol was used to make statistical comparisons.
All statistic has been performed using STATISTICA software (TIBCO, PaloAlto, CA, USA), at alfa level 0.05. The normality was calculated by the Shapiro-Wilk test. One-way ANOVA was used to check the differences between initial throwing distance and maximum speed before the activation exercises. Two-way ANOVA for the repeated measure was used to calculate the throwing distance and maximum speed differences between pre and post-measurements (repeated factor) and between the activation exercise type (repeated pre-post measure x exercise), followed by paired sample T-test as a post hoc test. The effect of the activation exercise was calculated by Cohen’s d effect size considering 0.2, 0,5, and 0,8 as small, medium, and large effect sizes, respectively.