The parallel squat is a complex lift involving the ankle, knee, and hip joints. Due to its multijoint nature, the squat is referred to as the "pillar of strength" exercise for the lower extremity (10). The squat is generally included in a weight training program to develop the quadriceps (rectus femoris, vastus lateralis, vastus medialis, and vastus intermedius), hamstrings (semimembranosus, semitendinosus, and biceps femoris), and triceps surae (gastrocnemius, soleus), although muscle groups such as the hip adductors and abductors and the erector spinae are also loaded (15). The parallel squat has also been suggested as an appropriate exercise for rehabilitation of the knee following anterior cruciate ligament injury because it imposes relatively less stress on the ligament than open chain exercises (23).
The squat is frequently performed by centering a weighted barbell across the posterior deltoids at the base of the trapezius. The exercise consists of alternating descent and ascent phases. During the descent phase, the lifter simultaneously flexes at the hip and knee joints and dorsiflexes the ankle joint to lower the body until the posterior surface of the thighs are parallel to the floor (13). In the ascent phase, the lifter extends the hip and knee joints and plantar flexes the ankle joint to return to the standing position.
Some controversy exists regarding isolation of specific muscle groups when performing the squat with stances of different widths. One popular bodybuilding text (17) suggests greater activity of the inner thigh muscles (vastus medialis and adductors) during the squat when a lifter abducts the thighs to assume a wider than shoulder width stance and laterally rotates the feet 45° away from the midline of the body. In addition, the authors state that the lateral anterior thigh muscle (vastus lateralis) is more active when a lifter adducts the thighs to assume a narrow stance. A final claim regarding leg position suggests overall thigh development occurs when using a shoulder width stance with the legs laterally rotated from the sagittal plane.
Of additional concern when resistance training is the muscle activity patterns in response to different loads. Traditional resistance, set, and repetition protocols vary among lifters. Strength trainers most often train with higher loads, fewer sets, and fewer repetitions whereas bodybuilders train with moderate loads, increased sets, and increased repetitions. According to McDonagh and Davies (12) an 80% of one repetition maximum (1RM) load is considered as a high load and one that is heavy enough to create an overload on muscle without causing extreme fatigue during completion of five repetitions. A 60% of 1RM load is considered to be a lower load because it less completely activates available muscle motor units.
An action potential is the electrical signal responsible for stimulating muscle activation. Surface electromyography (EMG) records the action potentials from multiple motor units within a monitored muscle, with the recorded signal reflecting both recruitment and rate coding of active units within the detection field of the electrode (9). EMG provides a technique to measure and compare muscle activity during phases of exercise performance (4) and between variations in resistance exercise technique (1,5,11). Although the parallel squat is the most frequently recommended lift for increasing overall power, muscular size, speed, and explosiveness in the legs, EMG has not been used extensively to evaluate patterns of leg muscle activity during performance of the parallel squat with varied leg stances and weight loads, and results of the limited published research studies tend to enhance the debate between scientists and practitioners. Tesch and Dudley (19,20) used magnetic resonance imaging (MRI) to evaluate muscle activity during the squat and reported no significant differences in lateral or medial thigh muscle utilization with different stances of the squat but reported greater thigh adductor activity with a narrow stance. However, MRI analysis provides only an indication of muscle utilization during a lift and does not provide a method to compare activity patterns during meaningful phases of an exercise (1). Signorile et al. (18) reported no difference in quadriceps muscle activity during squats performed with the toes pointed out, in or straight forward, although the width of foot placement did not vary extensively among the different conditions.
The purpose of this study was to compare activity in six muscles crossing the hip and/or knee joints when the parallel squat is performed with different stances and bar loads. The independent variables were load (60% and 75% of one repetition maximum), leg stance (narrow, shoulder width, and wide), and phase (descent, ascent). The dependent variables were the IEMG values for the rectus femoris, vastus medialis, vastus lateralis, adductor longus, biceps femoris, and gluteus maximus. IEMG was quantified as a single measure reflecting muscle activity, recognizing that the value will be affected by both changes in the number of motor units recruited and the firing frequency of recruited motor units. Different loads were used in the study primarily to provide a measure of external validity of the measurement technique, by providing a known criterion for determining physiological significance of any statistically significant differences in IEMG values. If the analysis technique was unable to identify statistically significant differences in IEMG values between training loads known to induce differential neuromuscular training effects, then use of the technique would be inappropriate to analyze the effects of different stances during lift performance.
Nine men volunteered to participate in the study. Each subject read, discussed and signed an informed consent in accordance with university policy. At the time of testing, each subject had been regularly engaged in a strength-training program, including the parallel squat, for at least 1 yr. Experienced lifters were used to facilitate muscle identification (by developed musculature and low subcutaneous fat reserves) and to ensure proper lifting technique. Descriptive statistics for the subjects include average age 22 yr (± 1 yr), height 183 cm (± 8 cm), mass 92 kg (± 14 kg) and 7 yr (± 2 yr) of lifting experience.
Data were collected in two phases. In phase 1, the 1RM in the shoulder width squat was estimated using the prediction method described by Lombardi (10). Following a warm-up, each subject completed sets of five repetitions with progressively higher loads until a set could not be completed. Rest was allowed between each set. The 1RM was predicted as the product of the coefficient for the weight used and the number of repetitions performed in the uncompleted final set. The predicted 1RM ranged from 118 to 250 kg. In phase 2, conducted within 1 wk, muscle activity and joint position data were collected during parallel squat performance at 75% (high load) and 60% (low load) of the subject's 1RM using the wide, shoulder width and narrow stances. The wide and narrow stances were defined as 140% and 75%, respectively, of the subject's shoulder width stance. To preserve the subjects natural squatting technique, subjects were allowed to maintain a preferred foot position in each stance. It was observed that subjects tended to keep the feet and knees aligned in a common plane as is recommended when performing the exercise.
A commercially available surface EMG system (Neuromuscular Research Center, Boston, MA) was used to record activity in six muscles: rectus femoris, vastus medialis, vastus lateralis, adductor longus, biceps femoris, and gluteus maximus. A single polyurethane housing containing two 1 mm × 10 mm silver bar surface electrodes spaced 10 mm apart (model MDI-X10) was placed longitudinally in the direction of the muscle fibers at the approximate motor point of each muscle (21). A ground electrode was secured to the styloid process of each subject during data collection. To reduce skin impedance at the site of the electrode attachment, the skin was shaved and washed with alcohol to remove excess skin debris. Flexible prewrap and tape were used to secure the electrode housing to the segment. Once applied, the electrodes remained secured to the skin during collection of EMG data. The relayed electrical signals from the muscles were differentially amplified 1000 to 3000 times with the isolated myoelectric preamplifier to provide recorded output in the range of ± 2.5 V.
To monitor flexion and extension of the knee joint, an "M" series twin axis goniometer (Penny and Giles Co., Santa Monica, CA) was positioned with its center of rotation on the lateral surface of the right knee. Endblocks were securely affixed to the subject without impeding range of motion. Alignment of the goniometer over the joint was visually verified between each trial.
Each subject was asked to refrain from any lower body lifting or strenuous activity for 48 h before data collection. Subjects completed five squat trials for each load and stance condition (total = 30 repetitions). Each subject performed all conditions in a randomized sequence to prevent order effects, with rest (at least 3 min) between each trial and condition to prevent fatigue. All lifts were performed in a power rack with a spotter. Use of a back belt was allowed if requested. To control the velocity of the descent and ascent phases of the squat, subjects performed each trial synchronized with a tape-recorded cadence of 3.2 s for the duration of the lift. The duration of the descent and ascent phases were 1.7 s and 1.5 s, respectively. EMG and goniometer data were collected for 4.5 s, beginning 0.25 s before descent commenced. Subjects were not allowed to bounce during the transition from descent to ascent.
Muscle activity and goniometer signals were simultaneously sampled (800 Hz) using a single analog to digital board installed in a personal computer. The sampling rate was limited by the number of channels recorded and the length of the sampling period. On completion of each trial, data were plotted on the VDT screen to assess quality, then stored on a hard drive for later processing.
The goniometer data were used to identify the descent and ascent phases of each trial. The descent phase was defined as the period between first knee joint flexion and maximum knee joint flexion. The ascent phase was defined as the period between maximum knee joint flexion and the return to a fixed, extended knee joint position. EMG values in microvolts (μV) were calculated from the A/D units for each muscle during each trial. Raw EMG signals for each trial were converted to linear envelopes by zeroing to the baseline, rectifying the signal and low-pass filtering with a cut-off set at 3 Hz (14,22). Integrated EMG values (in μV·s) were calculated for the descent and ascent phases of each trial by determining the area under the appropriate section of the linear envelope of each muscle during each trial. Reliability of the methodology was established for each muscle monitored by calculating the intraclass correlation coefficient (2) across the five trials collected for the ascent and descent phases of each condition for all subjects. The five-trial mean value of the IEMG for each muscle in each phase of the lift was calculated for each subject, and entered into a three-way repeated measures ANOVA (3 stance by 2 loads by 2 phases) for each muscle (α = 0.05). An estimate of the average error to be expected in each subject's IEMG value for each muscle was calculated as the standard error of measurement (SEM) using the equation SEM = sx (1 − rxx,)0.5, where sx is the group standard deviation and rxx, is the intraclass correlation coefficient, respectively, for the IEMG values of a specific muscle.
Intraclass correlation coefficients were calculated to quantify the reliability of the methodology used in the study, and these values are presented in Table 1. The coefficients ranged from a high of 0.9920 for the vastus medialis to a low of 0.9447 for the biceps femoris. These values indicate high internal consistency in the calculated IEMG values from trial to trial for each muscle. The standard error of measurement values for the IEMG values are also presented in Table 1, and range from a high of 2.02 μV·s for the vastus lateralis to a low of 0.51 μV·s for the adductor longus.
Three-way repeated measure ANOVA was used to compare the mean values of the IEMG values for each muscle across loads, stances, and phases of the parallel squat. No significant three-way interactions were identified for the IEMG values of any of the muscles monitored. The only two-way interaction involving stance was present among the mean values for the adductor longus and the gluteus maximus. Although significant load and phase main effects were identified for different muscles, no main effect for stance was identified for any of the muscles.
Quadriceps muscle. Descriptive statistics of the IEMG values for each component of the quadriceps are presented in Tables 2, 3, and 4. No significant three- or two-way interactions were identified by the repeated measure ANOVA. A significant load main effect was identified for the rectus femoris (F1,8 = 20.85, P = 0.002), the vastus medialis (F1,8 = 16.79, P = 0.003), and the vastus lateralis (F1,8 = 69.98, P < 0.0005). The effect of bar load on the activity pattern in all three components of the quadriceps is graphically summarized in Figure 1, which presents a five-trial ensemble average for the rectus femoris muscle of a single representative subject using a shoulder width stance with a low and high load. For all three components of the quadriceps, the general pattern of muscle activity was consistent between low and high loads, differing only in magnitude. On average across the stances and phases of all three components, the IEMG values for the high load were 20% greater than the IEMG values for the low load. The identification of statistically significant differences in IEMG values when comparing training loads known to induce physiologically different training effects justifies the use of this methodology to identify meaningful differences in muscle activity (1).
A significant phase main effect was identified only for the rectus femoris (F1,8 = 5.26, P = 0.05). On average across the low and high loads, the IEMG values for the ascent phase were 32% greater than those during the descent phase. Figure 1 graphically demonstrates the significant difference in muscle activity of the rectus femoris between the ascent and descent phases. For both the low load and the high load conditions, the muscle activity patterns showed increasing muscle activity with greater knee flexion during the descent phase. Following maximum muscle activity early in the ascent phase, muscle activity decreased as the subject extended the knees and hips to rise back to the starting position. Figure 2 demonstrates the lack of a significant phase main effect for the vastus lateralis, with the graph presenting a five-trial ensemble average for a single representative subject performing the squat with narrow and wide stances and a high bar load. The pattern was similar for the vastus medialis. In both muscles, activity increased through the initial 10% of the lift during the descent phase, and remained relatively constant until declining in the last 30% of the lift during the ascent phase.
Regardless of the stance width utilized, the mean IEMG values were not significantly different for any of the three components of the quadriceps muscle. The lack of a significant stance main effect for any component of the quadriceps muscle is evident from Figure 2 and in the stance means of Tables 2, 3, and 4. Similar muscle activity patterns were exhibited in all three stances used.
Adductor longus. Descriptive statistics of the IEMG values for the adductor longus are presented in Table 5. A significant stance by phase interaction was identified (F2,16 = 6.14, P = 0.01). A Tukey post hoc test indicated the IEMG values during the ascent phase were significantly greater than the IEMG values during the descent phase, but only when using the wide stance. The IEMG values for the ascent phase were approximately 50% greater than those of the descent phase with use of a wide stance compared to approximately 20% greater when using the narrow and shoulder width stances.
A load main effect was identified (F1,8 = 17.11, P = 0.003) for the IEMG values of the adductor longus. The IEMG values for the high load were, on average, 28% greater than those for the low load.
Gluteus maximus. Descriptive statistics of the IEMG values for the gluteus maximus are presented in Table 6. A significant load by stance interaction was present (F2,16 = 9.64, P = 0.002). The Tukey post hoc test indicated greater IEMG values during a squat with a wide stance compared to the narrow stance, but only with the high load.
A significant phase main effect (F1,8 = 27.49, P = 0.001) was identified for the gluteus maximus. On average, the IEMG values for the ascent phase were about 2.25 times higher than those for the descent phase.
Biceps femoris. Descriptive statistics of the IEMG values for the biceps femoris are presented in Table 7. A significant phase main effect was identified (F1,8 = 20.61, P = 0.002). The IEMG values for the ascent phase were, on average, more than 50% greater than the IEMG values during the descent phase.
Variability in EMG measures may arise from the procedures used when monitoring muscle activity. In this study, electrodes were secured to the skin over the six muscles monitored and remained in place throughout collection of the EMG data during the six conditions. As is evident in the high intraclass correlation coefficients, ranging from 0.9447 to 0.9920, this technique ensured high internal consistency during muscle monitoring. Calculating the five-trial mean value from internally consistent measures provides a reliable measure of each subject's true score (2) and allows for a high degree of confidence when interpreting the comparisons of IEMG measures across the six conditions of squat performance.
Each of the three components of the quadriceps muscles tested in this study, the rectus femoris, vastus medialis, and vastus lateralis, exhibited significantly greater IEMG values with a 75% 1RM load compared with a 60% 1RM load. On average across the stances and phases compared, IEMG values were 20% greater when lifting the high load compared with the low load. The differences in muscle activation between the loads used in this study uphold the findings of Berger (3), Adams et al. (1) and McCaw and Friday (11). Each study noted an increase in muscular activity as the load increased even though they studied different lifts than that used in the current study. The identification of greater activity in prime movers with greater load, a finding in accordance with the physiological principles of resistance training, validates the use of EMG as a technique to compare variations of strength training exercises.
There were no interactions involving stance, or main effects of stance, for any of the quadriceps muscles, contrary to the claim of many lay texts (7,16,17) concerning the relationship between stance width and muscle activation. Without explaining why this should occur, these texts claim selective muscle activation of the components of the quadriceps may be attained when a lifter alters stance width during lift performance. Specifically, the authors state that use of a wide stance with feet laterally rotated 45 degrees during squat performance results in increased activation of the vastus medialis, use of a narrow stance with feet pointed forward results in greater activation of the vastus lateralis, while use of a shoulder width stance with the feet laterally rotated results in greater overall quadriceps activation. This conjecture was not supported by the similar IEMG values measured for the quadriceps across all three stances. These results concur with the results of Tesch and Dudley (19,20) and Signorile et al. (18), who similarly reported no difference in quadriceps activity during variations of squat technique, and the results of Grabiner et al. (6), who compared isometric activity in the components of the quadriceps with varying levels of concomitant hip adductor isometric activity.
When active concentrically, the primary function of the quadriceps muscle is to extend the knee joint. Conversely, the quadriceps muscle controls flexion of the knee joint when active eccentrically. The results of this study suggest the three components of the quadriceps perform as a group regardless of concentric or eccentric activation during varied stance and load conditions. Because the vastus medialis and vastus lateralis are uniarticular muscles, crossing only the knee joint, varying stance width by altering thigh position in the frontal plane would not affect the length of the muscles, a factor that could influence recruitment patterns and magnitudes. The lack of a significant stance width effect on IEMG values for the rectus femoris suggests limited physiological implications of any change in the length of this biarticular muscle that may occur when hip position varies within the altered stance widths used in this study, although length changes of the rectus femoris were not quantified.
Selective activation within the quadriceps between the descent and ascent phases was observed only for the rectus femoris, with, on average, 32% greater activity observed during the ascent phase compared with the descent phase. Because greater peak knee extensor torque is required to overcome gravity when extending the knee joint during the ascent phase compared with controlling knee flexion during the descent phase (4), the increased rectus femoris activity could be interpreted as indicative of selective recruitment of the rectus femoris to provide a greater force output to contribute to the initial greater knee extensor torque when rising from the squat. However, this interpretation overlooks potential differences in recorded EMG related to differences in the efficiency of eccentric compared to concentric activity (1). Further analysis of the altered pattern of rectus femoris activity during the phases of the squat is warranted.
As with the quadriceps muscles, the significant load effect identified for the IEMG values of the adductor longus indicates the methodology utilized was able to discriminate between physiologically different muscle activity patterns. Lifting a 75% 1RM load results in, on average, 28% greater IEMG values compared with lifting a 60% 1RM load. Greater muscle activity with a higher load would be expected because altered motor units recruitment and firing patterns are an assumed contributor to the training effect during a systematic program of resistance exercise. Although it is plausible that use of a wide stance would require increased activation of all the thigh adductor muscles, including the adductor longus, to prevent excessive thigh abduction during descent and to cause thigh adduction during the ascent phase (7,16,17), this presumption is only supported in part by the results of this study. A stance by phase interaction was identified, with the mean IEMG values indicating greatest adductor longus activity during the ascent phase of the squat when a wide stance was used. The lack of consistently higher adductor longus activity with a wide stance suggests selective recruitment of this muscle when performing the squat with varied stance widths. In a squat, the thigh abducts and internally rotates as gravity acts to flex the hip and knee joints during the descent phase, with both actions occurring to a greater extent when a wide stance is utilized. During the ascent phase, the adductor longus acts with the other adductor muscles to pull the thigh closer to the midline of the body and to laterally rotate the segment back to a neutral position at full knee and hip extension. Greater activity in the adductor longus would be required with use of a wide stance because the range of motion of both adduction and external rotation is increased. Although Signorile et al. (18) claimed no benefits are to be gained by using altered foot placement during squat performance, the significant differences in IEMG values for the adductor longus suggest that the contribution of the medial thigh muscles is beneficially affected through altering stance width. The "burn" along the medial thigh reported by lifters following the squat performed with a wide stance probably reflects increased utilization of the adductor longus as opposed to selective recruitment of the vastus medialis. There is a need to quantify stress patterns on the ligaments of the knee joint to determine if the altered muscle patterns induced by use of an altered stance is offset by potentially deleterious increases in ligament stress.
The gluteus maximus is a prime mover for hip extension. The IEMG value during ascent was approximately twice as great as during descent. De Looze et al. (4) reported similar relative muscle activity levels between ascent and descent while the hip joint torque was similar between the two phases, except for slightly higher peak torque values during ascent. The greater IEMG values during the ascent phase compared with the IEMG values during the descent phase reflects the common finding of reduced muscle activity when a muscle is active eccentrically (4). Although increased activation of the gluteus maximus might be expected with an increased bar load, the observed load by stance interaction suggests that factors other than bar load alone influence recruitment of this muscle. The gluteus maximus exhibited significantly more activity when a lifter assumed a wide stance using a high load, but compared only to the narrow stance condition. An explanation for this finding could stem from the effects of stance width on length of the gluteus maximus. The gluteus maximus inserts distally on the gluteal tuberosity of the femur and the iliotibial tract. Thus, the lateral rotation and abduction of the thigh characterizing a wide stance squat could reduce the efficiency of the gluteus maximus by shortening the muscle length, placing it on a less optimal position on the lengthtension curve. To compensate for the reduced capability of force production, additional motor units would need to be recruited, and the recruited motor units would need to be activated with a higher frequency, reflected in an increased IEMG value. Future study should include quantification of any length changes of the gluteus maximus to explore this explanation.
As opposed to the uniarticular gluteus maximus, the hamstring muscles, including the biceps femoris, are biarticular, located posteriorly over both the hip and knee joints. Although not quantified in this study, the concurrent patterns of hip and knee flexion during descent, and hip and knee extension during ascent, may result in the biceps femoris maintaining a relatively constant length during the lift. Instead of being active concentrically during hip extension and eccentrically during hip flexion, the muscle may in fact be active in a quasi-isometric fashion through both the descent and ascent phases. Mean IEMG values indicate the biceps femoris was most active during the ascent phase, similar to the findings of De Looze et al. (4) and Isear et al. (8). The indication that significant hamstring activation was present during the squat is in contrast to what Tesch and Dudley (19,20) reported during an MRI investigation of the squat, which suggested the hamstring muscles are inactive during the squat. Although popular exercise manuals (7,15-17) do not strongly emphasize the role of the hamstring muscles during the squat, they do not suggest inactivity. The biceps femoris may activate to a greater degree during the ascent phase to contribute to the large hip extensor torque required to return to the upright position and to help stabilize the knee joint (4).
The current study was limited to an evaluation of muscle activity during squat performance using subjects with extensive experience with the lift. In acknowledgment of identified differences in the kinematics of squat performance between experienced and inexperienced lifters (13), future investigations should compare muscle recruitment and activity levels among lifters of different skill levels. It would also be beneficial to use subjects with previous injury to the anterior cruciate ligament to determine how observed differences in muscle recruitment is manifest during squat performance.
In conclusion, the results of this study do not substantiate claims in the lay literature on resistance exercise regarding selective recruitment of the components of the quadriceps muscle in response to varying stance width during performance of the parallel squat. In addition, variation in stance width alters the contribution of the biceps femoris and gluteus maximus to observed hip motion, but the changes in activation are not as great as those in response to differences in bar load.
1. Adams, G. R., M. R. Duvoisin, and G. A. Dudley. MRI and electromyography as indexes of muscle function. J. Appl. Physiol.
2. Baumgartner, T. A. Norm-referenced measurement: reliability. In: Measurement Concepts in Physical Education and Exercise Science,
M. J. Safrit and T. M. Wood (Eds.). Champaign, IL: Human Kinetics, 1989, pp. 45-72.
3. Berger, R. A. Effect of varied weight training programs on strength. Res. Q.
4. de Looze, M. P., H. M. Toussaint, J. H. van Dieen, and H. C. G. Kemper. Joint moments and muscle activity in the lower extremities and lower back in lifting and lowering tasks. J. Biomech.
5. Deluca, C. J. The use of surface electromyography in biomechanics. J. Appl. Biomech.
6. Grabiner, M. D., T. J. Koh, and L. Von Haefen. Effect of concomitant hip joint adduction and knee joint extension forces on quadriceps activation. Eur. J. Exp. Musculoskel. Res.
7. Holman, S. Home Gym Handbook: A Complete Guide to Training at Home.
Los Angeles: Homebody Productions, 1990, p. 82.
8. Isear, J. A., J. C. Erickson, and T. W. Worrell. EMG analysis of lower-extremity muscle recruitment patterns during an unloaded squat. Med. Sci. Sport Exerc.
9. Kranz, H., J. F. Cassell, and G. F. Inbar. Relation between electromyogram and force in fatigue. J. Appl. Physiol.
10. Lombardi, V. P. Beginning Weight Training.
Dubuque, IA: W. C. Brown, 1989, pp. 201-204.
11. McCaw, S. T., and J. J. Friday. A comparison of muscle activity between a free weight and machine bench press. J. Strength Cond. Res.
12. McDonagh, M. J., and C. T. Davies. Adaptive responses of mammalian skeletal muscle to exercise with high loads. Eur. J. Appl. Physiol.
13. McLaughlin, T. M., C. J. Dillman, and T. J. Lardner. A kinematic model of performance in the parallel squat by champion powerlifters. Med. Sci.
14. Olney, S. J., and D. A. Winter. Predictions of knee and ankle moments of force in walking from EMG and kinematic data. J. Biomech.
15. O'Shea, P. The parallel squat. Natl. Strength Condit. J.
16. Pearl, B., and G. T. Moran. Getting Stronger.
Bolinas, CA: Shelter, 1986.
17. Schwarzenegger, A., and B. Dobbins. The Arnold Schwarzenegger Encyclopedia of Modern Bodybuilding.
New York: Simon and Schuster, Inc., 1985, pp. 461-463.
18. Signorile, J. F., K. Kwiatkowski, J. F. Caruso, and B. Robertson. Effect of foot position on the electromyographic activity of the superficial quadriceps muscles during the parallel squat and knee extension. J. Strength. Cond. Res.
19. Tesch, P. A., and G. A. Dudley. MRI examines the squat. Musc. Fit.
November: 116, 1992.
20. Tesch, P. A., and G. A. Dudley. Muscle meets magnet. Stockholm: PA Tesch AB, 1993 pp. 51-52.
21. Warfel, J. H. The Extremities: Muscles and Motor Points,
6th Ed. Philadelphia: Lea & Febiger, 1985, pp. 74-76.
22. Winter, D. A. Electromyogram recording, processing, and normalization: procedures and considerations. J. Hum. Muscle Perform.
23. Yack, H. J., C. E. Collins, and T. J. Whieldon. Comparison of closed and open kinetic chain exercise in the anterior cruciate ligament-deficient knee. Am. J. Sports Med.