Relationship between Body Center of Mass Velocity and Lower Limb Joint Angles during Advance Lunge in Skilled Male University Fencers

: We investigated the inﬂuence of advance lunging in fencing from the perspective of velocity and lower limb joint angles to identify how the joint angles contribute to the peak velocity in a lunge with advance (LWA). Fourteen skilled athletes (age: 19.6 ± 0.9 years, height: 171.2 cm ± 5.2 cm, weight: 63.7 kg ± 5.3 kg, and fencing experience: 9.7 ± 3.1 years) participated by performing two types of attacking movements, and data were collected with a 3D movement analysis system. A correlation between the peak velocity of the body center of mass (CoM) in an advance lunge and several joint angle variables (rear hip peak ﬂexion angle (r = 0.63), rear ankle peak dorsiﬂexion angle (r = − 0.66), rear ankle range of motion (r = − 0.59), and front hip peak extension angle (r = 0.54)) was revealed. In addition, the joint angle variables that signiﬁcantly predicted peak CoM velocity during an LWA were the rear knee peak ﬂexion angle ( β = 0.542), rear knee peak extension angle ( β = − 0.537), and front knee peak extension angle ( β = − 0.460). Our ﬁndings suggest that the rear leg hip joint, rear leg ankle joint, and front leg hip joint may control the acceleration generated by an LWA. Furthermore, more ﬂexion of the rear leg knee joint in the early phase of the lunge and greater extension of the rear and front leg knee joints at the end of the lunge phase may help increase peak velocity.


Introduction
Fencing is a sport in which victory or defeat is decided by scoring points when two fencers face each other, and most of the points are scored by 'touch' attacks [1,2]. The most basic 'touch' attacking movement is the lunge [3], which is characterized by extending the arm holding the sword, pushing the back leg backwards, and kicking the front leg forward to land [4]. Various aspects of lunge performance have been investigated [5], and one of the factors that substantially impacts the score is 'speed'. Studies that have focused on 'speed' during the lunge have investigated the time from the start of the movement until the target is hit (movement time during the lunge) [6][7][8][9][10][11][12], and some have investigated peak velocity during the lunge [4,[7][8][9][10][12][13][14]. Of these two indices, peak velocity during the lunge is considered the more optimal indicator because the time of movement during the lunge affects the distance set for the lunge in each fencer [14]. Furthermore, the joint angle of the lower limb is often used as an indicator of a factor affecting the peak velocity of the lunge [4,14], and quantification of the lunge movement in the attack phase of fencing from the aspect of peak velocity and joint angle has been attempted.
Bottoms et al. [4] discovered that the rear leg knee joint range of motion (ROM) in the sagittal plane during the lunge and the peak flexion angle (PFA) of the rear and front leg hip joints were predictors of the lower limb joint angles that contributed most to sword velocity. This finding indicates that a low posture with knee joint flexion is important for a more effective performance. Guan et al. [14] also reported that the ROM of the rear leg knee joint during the lunge increases the horizontal peak velocity. However, in these studies [4,14], participants started the lunge movement from a stationary posture (en garde posture) and the movement was analyzed until the touch. The issue remains that the characteristics of actual matches are not fully reflected in the analysis. Specifically, in actual matches, when facing an opponent, an advance brings the opponent closer (the target distance) and leads to a rush for a touch [3]. Generally, the lunge and advance are added to the series of attacking movements (lunge with advance (LWA)) in the attacking phase, implying that the analysis of attacking movements in fencing remains insufficient, and LWA movements require further investigation. However, to the best of our knowledge, none of the previous studies we reviewed met these criteria.
To expand the scope of previous research to include sports other than fencing, we analyzed taekwondo, a combat sport similar to fencing, which includes a preliminary movement similar to advance lunging in fencing. In this study, a back kick and a back kick with a jump were compared to investigate the effect of the jump on the ground reaction force (GRF) during the kick's execution [15]. Reportedly, adding a preliminary jump to the kick action, i.e., an increase in velocity, increases the flexion and extension angles of the lower limb joint angles (the knee joint in this study) and increases the GRF value. In fencing, an additional preliminary movement (an increase in velocity) during the same attack phase is expected to increase the velocity in the latter half of the movement and substantially affect the lower limb joint angles. However, as mentioned above, there is no such previous research on fencing, and the relationship between the two remains unclear. Thus, it is expected that clarification of the characteristics of an LWA, which is often used in actual competitions, in terms of lower limb joint angles and velocities will lead to the accumulation of more game-specific knowledge. This knowledge can be incorporated into evidence-based coaching and the quantitative evaluation of athletes' performances during competitions.
This study aims to clarify the influence of the advance (preliminary movement) on the lunge (attacking movement) in fencing from the viewpoint of velocity and lower limb joint angles. Furthermore, we aimed to identify the joint angle factors that contribute to the peak velocity of the LWA. In an LWA, as the peak velocity of the advance increases, the lower limb joint angles are also expected to change significantly. Furthermore, similar to previous studies on the lunge without advance (LWOA) [4,14], we hypothesized that the ROM of the rear knee joint, PFA of the rear hip joint, and PFA of the front hip joint would affect the peak velocity of the LWA.

Participants
Most previous studies included highly skilled male fencers as participants [6,14]. Since this study is positioned as an extension of those studies, the participants of this study were 14 male fencers (age: 19.6 ± 0.9 years, height: 171.2 cm ± 5.2 cm, weight: 63.7 kg ± 5.3 kg, and fencing experience: 9.7 ± 3.1 years, hereafter expressed as the mean ± standard deviation) who specialize in foil. The sample size was estimated using G*Power (G*Power 3.1.9.6; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany), and an effect size of 0.8, α level of 0.05, and power (1-β) of 0.80 were obtained, which indicated that 12 participants were needed. All participants were right-handed and were members of the university fencing team, two of whom were national team members. The purpose and method of this study were fully explained in writing and orally in advance, and the participants signed an informed consent form after fully understanding the content of the written explanation. The participants confirmed that they did not sustain any musculoskeletal injuries in the past 6 months prior to the test. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Graduate School of System Design and Management, Keio University (approval number: SDM-2022-E001).

Study Design
In this observational study, 14 fencers performed trials of two types of attacking movements, an LWA and LWOA, and the correlation between the center of body mass (CoM) peak velocity during the advanced phase of the LWA and the amount of change in the lower limb joint angle variables (PFA, peak extension angle (PEA), and ROM of the hip and knee joints, as well as peak dorsiflexion angle (PDA), peak plantar flexion angle (PPA), and ROM of the ankle joint) during the lunge phase of the LWA and LWOA was analyzed to determine the effect of the advance on the lower limb joint angle during the lunge. Furthermore, we identified the lower limb joint angle variables during the lunge that contribute to the peak velocity of the CoM during the LWOA and LWA and clarified the kinematic characteristics of the LWA.

Trial Conditions
The experimental setup is shown in Figure 1. Lunge distance was defined as the horizontal distance from the rear leg toe to the target in the 'en garde' (start) position, as previously described, and was defined as 1.5 times the length of each participant's height [7,16]. The advance distance was set as the length between each participant's left and right acromion. The distance between the participant and the target was 256.8 ± 7.8 cm for the lunge distance and 46.3 ± 2.0 cm for the advance distance. For the target, a 30 cm × 30 cm square lamé jacket cloth used for fleuret events was attached to the chest and adjusted so that the target surface was level with the tip of the participant's lungethrusting sword.
Biomechanics 2023, 4, FOR PEER REVIEW 3 in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Graduate School of System Design and Management, Keio University (approval number: SDM-2022-E001).

Study Design
In this observational study, 14 fencers performed trials of two types of attacking movements, an LWA and LWOA, and the correlation between the center of body mass (CoM) peak velocity during the advanced phase of the LWA and the amount of change in the lower limb joint angle variables (PFA, peak extension angle (PEA), and ROM of the hip and knee joints, as well as peak dorsiflexion angle (PDA), peak plantar flexion angle (PPA), and ROM of the ankle joint) during the lunge phase of the LWA and LWOA was analyzed to determine the effect of the advance on the lower limb joint angle during the lunge. Furthermore, we identified the lower limb joint angle variables during the lunge that contribute to the peak velocity of the CoM during the LWOA and LWA and clarified the kinematic characteristics of the LWA.

Trial Conditions
The experimental setup is shown in Figure 1. Lunge distance was defined as the horizontal distance from the rear leg toe to the target in the 'en garde' (start) position, as previously described, and was defined as 1.5 times the length of each participant's height [7,16]. The advance distance was set as the length between each participant's left and right acromion. The distance between the participant and the target was 256.8 ± 7.8 cm for the lunge distance and 46.3 ± 2.0 cm for the advance distance. For the target, a 30 cm × 30 cm square lamé jacket cloth used for fleuret events was attached to the chest and adjusted so that the target surface was level with the tip of the participant's lunge-thrusting sword.

Procedure
Referring to the protocol of Gutiérrez-Dávila et al. [8], subjects performed stretching, running, fencing footwork, etc., for about 15 min on their own as a warm-up before the experiment, and then practiced LWOAs and LWAs several times each to get used to the experimental conditions. The participants then took the starting position and assumed the

Procedure
Referring to the protocol of Gutiérrez-Dávila et al. [8], subjects performed stretching, running, fencing footwork, etc., for about 15 min on their own as a warm-up before the experiment, and then practiced LWOAs and LWAs several times each to get used to the experimental conditions. The participants then took the starting position and assumed the 'en garde' posture in a stationary position, from which they started the movements at any time and performed LWOAs and LWAs until achieving three successful trials each (a minimum of six trials per athlete) [17]. To avoid the influence of fatigue on performance, we set a rest period of 30 s between trials.
In this experiment, a participant was considered to have failed if the tip of the sword was out of the range of the target, if the participant stopped moving during the lunge, or if the participant's body balance was considerably off. The participants performed an average of 3.1 LWOAs and 3.6 LWAs, including failed attempts. Each participant was instructed to perform the fastest possible movement from the start to the touch point in each trial without stopping the movement.
A three-dimensional (3D) motion analysis system was used for data collection. Reflective markers (14.0 mm in diameter) attached to each part of the participant's body were photographed during each test, and the analysis software measured the 3D coordinate values. The participant was surrounded by eight high-resolution cameras (Qualisys Standard Motion Cameras Miqus M3, Goteborg, Sweden), and the sampling frequency was set at 500 Hz. The data captured by the cameras were stored in the data station of the 3D motion analysis system using a dedicated cable. In this study, a stationary coordinate system was set up with the Y-axis for the direction of the attacking motion, the X-axis for the direction orthogonal to the Y-axis, and the Z-axis for the vertical direction.
The participants wore motion capture suits and fencing masks on their heads and held a sword in their dominant hand. Forty-four tracking markers (14.0 mm in diameter) were placed on the bilateral sides of the head, shoulder, trunk, arm, thigh, knee, shank, ankle, and foot, and seven points were placed on the fluted sword (fifty-one points in total). The reflective markers used in this study were placed on the crown of their head, forehead, right and left sides of the head, acromion processes, second cervical spinous process, twelfth thoracic spinous process, upper arms, lower end of the ribs, anterior superior iliac spines, posterior superior iliac spines, lateral epicondyle of the humerus, medial epicondyle of the humerus, ulnar styloid processes, radial styloid processes, third metacarpal bones, xiphoid process of the sternum, sacrum, lateral epicondyle of the femur, medial epicondyle of the femur, lateral malleolus, thighs, tibial tuberosity, head of the second metatarsus, head of the fifth metatarsus, and heel bones (44 points), with 7 points also located on the foil (top of the blade, near the top, middle of the blade, near the guard, and guard (guard center, guard right, and guard left)), for a total of 51 points. In this study, the fencers used a fleuret No. 5 sword (BF Allstar, Germany; blade length: 90 cm) and a mask (Allstar, Germany) that conformed to international standards and were standardized. Each participant provided their own fencing shoes. Before the trials, the fencers were asked to stand in an anatomical reference position, and data from static trials were collected.

Analysis
The video data were digitized, and the markers on the body were tracked and labeled using Qualisys Track Manager and exported as C3D files. Marker coordinates reconstructed in a 3D space were quantified using Visual 3D and filtered with a Butterworth fourth-order zero-lag filter at a frequency of 8 Hz. The positions of the CoM and joint centers were calculated from the filtered marker coordinates. The CoM was calculated as a weighted average of the CoM positions of 15 segments based on the Visual 3D skeletal model [18], and the horizontal displacement and peak velocity during the lunge were obtained. The hip joint center was calculated from a regression equation that provides an estimate of the distance from the origin of the pelvis [19,20]. The knee joint center was assumed to be on a plane defined by the hip joint center, femoral and knee markers, and midway between the femoral condyles. The ankle joint center was assumed to be on the plane defined by the knee joint center, shank markers, and ankle markers and to be one-half the distance between the ankles.
To ensure accurate data analysis, the advance and lunge phases were defined according to previous studies [14,21,22]. The participant landed on both feet and started the movement from the stationary state of 'en garde'. The toe-off of the front leg was defined as the beginning of the advance phase, and the landing of the front leg, attraction of the rear leg, and toe-off of the front leg again were the end of the advance phase and the beginning of the lunge phase. After the front leg swung forward in coordination with the rear leg, the heel contact of the front leg was defined as the end of the lunge phase ( Figure 2). For each participant, the trial with the fastest peak velocity of the CoM among the three trials successfully performed for each LWOA and LWA was adopted as the representative value. Time normalization was performed by setting the start-to-end time as 100%, and the mean and standard deviation of the trials were calculated for each 1% time. Since many previous studies have analyzed joint angles of hip and knee joint flexion (+)/extension (−) and ankle dorsiflexion (+)/plantar flexion (−) in the sagittal plane [4,14], the relative angles of the two adjacent segments (pelvis and femur for hip, femur and tibia for knee, and tibia and foot for ankle) were also calculated for the present study.
the rear leg, the heel contact of the front leg was defined as the end of the lunge phase ( Figure 2). For each participant, the trial with the fastest peak velocity of the CoM among the three trials successfully performed for each LWOA and LWA was adopted as the representative value. Time normalization was performed by setting the start-to-end time as 100%, and the mean and standard deviation of the trials were calculated for each 1% time. Since many previous studies have analyzed joint angles of hip and knee joint flexion (+)/extension (−) and ankle dorsiflexion (+)/plantar flexion (−) in the sagittal plane [4,14], the relative angles of the two adjacent segments (pelvis and femur for hip, femur and tibia for knee, and tibia and foot for ankle) were also calculated for the present study.
The kinematic variables calculated are as follows: • Velocity parameter: peak velocity of the CoM during the advance phase and peak velocity of the CoM during the lunge phase;

Statistical Analyses
The data were acceptable for all procedures, and no missing values were identified. The normal distribution of all variables was confirmed by the Shapiro-Wilk test, and in order to compare LWOA and LWA items, parametric data were subjected to the paired ttest, and nonparametric data to the Wilcoxon signed-rank test. The significance level was set as <5% for each item, and Cohen's d [23] and r were obtained to confirm the effect size. The kinematic variables calculated are as follows:

•
Velocity parameter: peak velocity of the CoM during the advance phase and peak velocity of the CoM during the lunge phase; • Lower limb joint angle parameters: PFA, PEA, PDA, PPA, and ROM of front and rear leg during the lunge phase; • Amount of change in the lower extremity joint angle variables: the LWA minus the LWOA for each lower extremity joint angle variable in the lunge phase.

Statistical Analyses
The data were acceptable for all procedures, and no missing values were identified. The normal distribution of all variables was confirmed by the Shapiro-Wilk test, and in order to compare LWOA and LWA items, parametric data were subjected to the paired t-test, and nonparametric data to the Wilcoxon signed-rank test. The significance level was set as <5% for each item, and Cohen's d [23] and r were obtained to confirm the effect size. The peak velocity of the CoM during the advance phase in the LWA and the mean change in lower limb joint angles (PFA, PEA, PDA, PPA, and ROM) during the lunge phase in the LWOA and LWA were quantified using Pearson's correlation coefficient (r). The correlation coefficients were interpreted based on previous studies (0-0.09, trivial; 0.1-0.29, small; 0.3-0.49, moderate; 0.5-0.69, large; 0.7-0.89, very large; 0.9-0.99, nearly perfect; and 1, perfect) [24]. Furthermore, using the peak velocity of the CoM during the lunge phase as the dependent variable and the lower limb joint angles (PFA, PEA, PDA, PPA, and ROM) as predictor variables, the joint angle factors contributing to the peak velocity of the LWOA and LWA were determined using the stepwise method of multiple regression analysis. Significance was set at the α level of p < 0.05. IBM SPSS Statistics 28 (IBM Corporation) software was used for all statistical analyses.

Results
The peak velocity of the CoM was 1.88 ± 0.16 m/s for the LWOA, whereas it was 2.41 ± 0.20 m/s for the LWA, which was a 28% increase on average (Figure 3). The PFA, PEA, PDA, PPA, and ROM of the lower limb joints (hip, knee, and ankle joints) in the LWOA and LWA showed a significant difference (p < 0.05) in the rear hip PFA, rear hip PEA, rear hip ROM, rear knee PEA, rear knee ROM, rear ankle PDA, rear ankle PPA, front hip PFA, front hip PEA, front knee PFA, front knee ROM, front ankle PDA, front ankle PPA, and front ankle ROM (Table 1). Of these, the front hip PFA, front knee PFA, front knee ROM, front ankle PPA, front ankle ROM, rear hip PFA, rear hip PEA, rear hip ROM, rear knee PEA, rear knee ROM, and rear ankle PPA were significantly greater in the LWA (p < 0.05). In contrast, the front hip PEA, front ankle PDA, and rear ankle PDA were significantly greater in the LWOA (p < 0.05). and 1, perfect) [24]. Furthermore, using the peak velocity of the CoM during the lunge phase as the dependent variable and the lower limb joint angles (PFA, PEA, PDA, PPA, and ROM) as predictor variables, the joint angle factors contributing to the peak velocity of the LWOA and LWA were determined using the stepwise method of multiple regression analysis. Significance was set at the α level of p < 0.05. IBM SPSS Statistics 28 (IBM Corporation) software was used for all statistical analyses.

Results
The peak velocity of the CoM was 1.88 ± 0.16 m/s for the LWOA, whereas it was 2.41 ± 0.20 m/s for the LWA, which was a 28% increase on average (Figure 3). The PFA, PEA, PDA, PPA, and ROM of the lower limb joints (hip, knee, and ankle joints) in the LWOA and LWA showed a significant difference (p < 0.05) in the rear hip PFA, rear hip PEA, rear hip ROM, rear knee PEA, rear knee ROM, rear ankle PDA, rear ankle PPA, front hip PFA, front hip PEA, front knee PFA, front knee ROM, front ankle PDA, front ankle PPA, and front ankle ROM (Table 1). Of these, the front hip PFA, front knee PFA, front knee ROM, front ankle PPA, front ankle ROM, rear hip PFA, rear hip PEA, rear hip ROM, rear knee PEA, rear knee ROM, and rear ankle PPA were significantly greater in the LWA (p < 0.05). In contrast, the front hip PEA, front ankle PDA, and rear ankle PDA were significantly greater in the LWOA (p < 0.05).    Figure 4 show the correlation between the peak velocity of the CoM in the advance phase of the LWA and the amount of change in the lower limb joint angle variables in the LWOA and LWA. The results showed that the rear hip PFA, rear ankle PDA, rear ankle ROM, and front hip PEA were significantly (p < 0.05) related to the peak velocity of the CoM in the advance phase (r = 0.54, 0.63, −0.59, −0.66). Among these, a positive correlation (r = 0.54, 0.63) was confirmed for the rear hip PFA and front hip PEA, and a negative correlation (r = −0.59, −0.66) was found for the rear ankle PDA and rear ankle ROM.     Table 3 shows the results of a multiple regression analysis for predicting the peak velocity of the CoM from the lower limb joint (hip, knee, and ankle) angle variables (PFA, PEA, PDA, PPA, and ROM) of the front and rear legs in the LWOA and LWA. In the LWOA, the rear hip ROM (β = 0.592) was significantly predictive. For the LWA, the rear knee PFA (β = 0.542), rear knee PEA (β = −0.537), and front knee PEA (β = −0.460) were significantly predictive. Regarding the multicollinearity of the explanatory variables, the VIF value was less than 10, indicating that multicollinearity was not observed.   Table 3 shows the results of a multiple regression analysis for predicting the peak velocity of the CoM from the lower limb joint (hip, knee, and ankle) angle variables (PFA, PEA, PDA, PPA, and ROM) of the front and rear legs in the LWOA and LWA. In the LWOA, the rear hip ROM (β = 0.592) was significantly predictive. For the LWA, the rear knee PFA (β = 0.542), rear knee PEA (β = −0.537), and front knee PEA (β = −0.460) were significantly predictive. Regarding the multicollinearity of the explanatory variables, the VIF value was less than 10, indicating that multicollinearity was not observed.

Discussion
In this study, we clarified the influence of the advance on the lunge in fencing from the perspective of velocity and lower limb joint angles and identified the joint angles contributing to the peak velocity of an LWA. The results show that the higher peak velocity of the CoM in the advance phase significantly changed several joint angle variables in the lunge phase, supporting our hypothesis. However, our hypothesis was only partially supported as we identified joint angles contributing to the peak velocity as a predictor. The following subsections discuss the results obtained in this study.

Effects of Advance Movements on Lower Limb Joint Angles during the Lunge
A positive correlation (r = 0.63) between the peak velocity of the CoM during the advance phase of the LWA and the amount of change in the rear hip PFA during the lunge phase was confirmed. This result suggests that the greater the peak velocity of the CoM during the advance phase, the greater the flexion of the rear hip joint at the end of the advance phase and at the beginning of the lunge phase (Figure 3e), which may have controlled the acceleration that occurred during the advance phase.
In contrast, the rear ankle PDA and the rear ankle ROM indicate that the greater the peak velocity of the CoM in the advance phase, the smaller the ankle dorsiflexion in the lunge phase, and the ankle joint is found to be in a plantar flexed position (Figure 3g). These results suggest that after the rear leg ankle joints are extended in the advance phase, the greater flexion again in the lunge phase decreases the velocity so that the ankle joints are moved closer to the plantar flexion position, such as in a toe-up, to increase the lunge velocity efficiently (Figure 3g).
For the front leg, the amount of change in the front hip PEA during the lunge phase of the LWA was correlated with the peak velocity of the CoM during the advance phase (r = 0.54). This means that as the peak velocity of the CoM in the advance phase increases, the front hip extension angle becomes smaller (the flexion angle becomes larger) (Figure 3b). Therefore, it can be inferred that the front leg hip, as well as the rear leg hip, flexes more at the beginning of the lunge phase, and thus may serve as a control function for the acceleration generated by the advance.

Lower Limb Joint Angle Variables Contributing to Peak Velocity of LWA
The results of the multiple regression analysis of the lower limb joint angles contributing to the peak velocity of the CoM in the lunge showed that the rear hip ROM was a significant variable in the LWOA. However, the coefficient of determination was small (adjusted R 2 = 0.297). Bottoms et al. [4] reported that the rear hip PFA and rear knee ROM were significantly correlated with sword velocity in the lunge, and the results of this study were partially similar; however, the results for the knee joint were not consistent with those of Bottoms et al. [4]. One possible reason for this is that the target distance in this study was shorter than that in previous studies. As the target distance increases, the knee and hip joints become involved, implying that velocity and distance may greatly affect the joint angle during the lunge movement.
In contrast, for the LWA, the front knee PEA, rear knee PFA, and rear knee PEA were the lower limb joint angle variables that significantly affected the peak velocity of the CoM during the lunge phase (adjusted R 2 = 0.665). Guilhem et al. [22] reported that the rear leg's gluteus medius, vastus lateralis, and gastrocnemius showed increased muscle activity during the lunge phase of the LWA and that the maximal muscle strength power of the rear knee extensors was closely related to the horizontal peak velocity of the lunge. This finding implies that the gastrocnemius of the rear leg is involved in the rear knee PFA, and the vastus lateralis of the rear leg is involved in the rear knee PEA. The study by Guilhem et al. [22] supported these results. For the front leg, the front knee PEA during the lunge phase of the LWA significantly affected the peak velocity of the CoM, suggesting that sufficient extension of the front knee joint at the end of the lunge phase generated a forward propulsive force and contributed to the peak velocity of the CoM. Similar results have been reported for the LWOA in previous studies, which supports the results of this study [6,14].

Implications for Coaching of Fencers
The findings of this study will help fencers and coaches better understand the kinematics of the LWA. Further, these results will contribute to developing new tactics and training plans for fencers and coaches for competitions by helping them understand the characteristics of the LWA and evaluating their movements during competitions. For example, when predicting the opponent's attacking movements during a match, if the advance is fast, the hip and knee joints will flex more during the transition to the lunge phase, which may give the opponent a chance to counterattack [25]. To increase the peak velocity during the lunge phase, it may be effective for coaches to suggest increasing the total range of motion of the rear leg hip joint during the lunge in an LWOA. In an LWA, it may be important to assume a low position with the rear leg knee joint more flexed at the beginning of the lunge phase and to extend the rear leg knee and front leg knee joint to a greater extent at the end of the lunge phase. Furthermore, training strategies should emphasize the importance of rear hip flexors and extensors in the LWOA and rear knee flexors and extensors and front knee extensors in the LWA.

Limitations and Future Work
This study had some limitations. First, the trials were conducted without protective gear. In actual competition situations, protective gear is worn, but in this study, anatomical markers had to be affixed. Second, in order to strictly control the relevant variables, the experiment was conducted in a constrained environment with a non-moving opponent and a target that was more limited than the actual effective surface. It is conceivable that these different conditions could alter the movement patterns of fencers. In future studies, we would like to conduct investigations in an environment similar to a match situation, in which opponents are expected to advance and retreat, in order to better understand the relationship between the defender and the fencer.

Conclusions
In conclusion, to elucidate the kinematic characteristics of an LWA, we compared the LWOA and LWA and clarified the influence of the peak velocity of the CoM in the advance phase of the LWA on the lower limb joint angle variables in the lunge phase. Furthermore, we identified and examined the lower limb joint angle variables that contribute to the peak velocity of the CoM of the LWA. We discovered significant differences in the peak velocity of the CoM and several lower limb joint angles during the lunge phase in the LWA compared to that in the LWOA. The higher the speed of the advance, the more deeply the front hip joint and the rear leg hip joint are flexed at the beginning of the lunge phase, and the less flexion of the rear leg ankle joint in the middle phase of the lunge, which may control the acceleration generated by the advance. Furthermore, the joint motions of the knee joints of the rear and front leg are closely related to the peak velocity of the CoM during the lunge in the LWA. Therefore, the peak velocity of the CoM during the lunge phase is increased by increasing the flexion of the rear leg knee joint in the early phase and extending the rear leg and front leg knee joints considerably in the late lunge phase.  Informed Consent Statement: Informed consent was obtained from all participants involved in the study.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author. The data are not publicly available due to privacy concerns.