APPLIED SCIENCES

Biodynamics

Effects of power and flexibility training on

vertical jump technique
JOSEPH P. HUNTER and ROBERT N. MARSHALL
Department of Sport and Exercise Science, The University of Auckland, Auckland, NEW ZEALAND

ABSTRACT
HUNTER, J. P., and R. N. MARSHALL. Effects of power and flexibility training on vertical jump technique. Med. Sci. Sports Exerc.,
Vol. 34, No. 3, pp. 478 – 486, 2002. Purpose: The purpose of this study was to assess the effects of power and flexibility training on
countermovement and drop jump techniques. Method: All jumps were executed with the goal of attaining maximum height and no
restrictions were placed on the magnitude of countermovement or ground contact time. Subjects underwent initial testing followed by
random allocation to one of four groups: power training to increase vertical jump height (P), stretching to increase flexibility (S), a
combination of power and stretch training (PS), and a control group (C). Training lasted for 10 wk, followed by retesting. Jump height
was calculated in addition to the following technique variables: eccentric lower-limb stiffness produced during the countermovement
phase, magnitude of countermovement, and in the case of the drop jumps, ground contact time. Results: Groups PS, P, and S all
increased countermovement jump (CMJ) height, but only groups PS and P increased drop jump height (DJ30, DJ60, and DJ90 for drop
jumps performed from 30-, 60-, and 90-cm drop heights). The technique changes associated with power training were increases in
magnitude of countermovement (CMJ, DJ30, DJ60, and DJ90) and increases in ground contact time (DJ30 and DJ60). In addition, the
eccentric lower-limb stiffness produced during the countermovement phase of the jumps increased for CMJ and decreased for DJ30,
DJ60, and DJ90. Stretching appeared to have no significant effect on CMJ or drop jump technique. Conclusion: The results of this
study show that when the training goal is maximum jump height alone, it is likely that drop jump technique will change in the direction
of a lower eccentric leg stiffness, greater depth of countermovement, and a longer ground contact time, whereas for a countermovement
jump eccentric leg stiffness and the depth of countermovement will both increase. It is proposed that these technique changes are a result
of attempting to optimize a complex combination of factors involved in jumping (e.g., utilization of elastic energy, Golgi tendon organ
inhibition, and contractile component contribution). Key Words: JUMP TECHNIQUE CONTINUUM, BOUNCE DROP JUMP,
COUNTERMOVEMENT DROP JUMP, ECCENTRIC LOWER-LIMB STIFFNESS, STRETCHING

A
drop jump involves stepping from a small height,
landing, and then explosively jumping (usually, but
not always) vertically. Various restrictions can be
set when performing such jumps, ultimately affecting jump
technique and the height obtained. An example of such a
restriction is to instruct the athlete to minimize ground
contact time while still attempting to jump as high as pos-
sible. To aid in achieving this, feedback of the ratio of “jump
height to ground contact time” can be given after each jump
(11). The jump technique used with such restrictions has
been referred to as a “bounce drop jump” (i.e., a quick
rebound style of jump) of which the antithesis is a “coun-
termovement drop jump” (i.e., using a relatively larger
countermovement) (2).
When restricted to using a bounce drop jump or counter-
movement drop jump technique, the highest jump heights
0195-9131/02/3403-0478/$3.00/0
MEDICINE & SCIENCE IN SPORTS & EXERCISE®
Copyright © 2002 by the American College of Sports Medicine
Submitted for publication January 2000.
Accepted for publication July 2001.
478
are achieved with the latter (11). However, when a group of
athletes perform unrestricted drop jumps and are instructed
to jump for maximum height, the technique will range from
that of a bounce drop jump to a countermovement drop jump
(2). That is, there exists a “jump technique continuum” in
which technique ranges from use of a small amplitude
countermovement, production of high lower-limb stiffness
during the eccentric phase of the jump, and a short ground
contact time to the converse at the other end of the
continuum.
The existence of drop jump technique variation and the
effects of instructions have been documented (1,2,11). How-
ever, the effects of various training methods on jump tech-
nique (with regard to the jump technique continuum) remain
undocumented. For example, lower-limb stiffness used dur-
ing the eccentric phase of a stretch-shortening cycle jump is
likely to have a large influence on both jump technique and
the height jumped. In the interest of making more informed
training program decisions, it would be valuable to know
the changes in stiffness associated with various training
programs. The suggested optimal level of stiffness for a
stretch-shortening cycle task has ranged from stiff (7) to
compliant (10). This issue remains unresolved, and it is
possible the authors were referring to different “sources” of
stiffness (see the Discussion for more on this). However, if
there is an optimal stiffness to use during the eccentric phase
of a jump, then it is possible, when unrestricted, subjects
will adapt to using such a technique. Therefore, the purpose
of this study was to assess the effects of power training and
stretching on unrestricted jump performance, in particular
the effects of training on the technique used with regard to
the jump technique continuum.
METHODS
Approach to the problem and experimental de-
sign. This study included four randomly allocated groups:
a power training group (P), which trained to increase ver-
tical jump performance; a stretching group, which trained to
increase flexibility (S); a combined power and stretching
group (PS); and a control group (C). The study was ap-
proved by The University of Auckland Human Subjects
Ethics Committee, and all subjects attended a test familiar-
ization session at which written informed consent was ob-
tained. A pre-training testing session was followed by 10 wk
of training and then a post-training testing session. The tests
included stretch tolerance of the hamstrings and quadriceps,
and four variations of vertical jumps.
The jumps included drop jumps from 30-, 60-, and 90-cm
drop heights (DJ30, DJ60, and DJ90) and also a counter-
movement jump (CMJ). These jumps were chosen due to
the wide range of eccentric loading experienced by the
performer. The performance goal for all jumps was to obtain
maximum height. With the intention of allowing the sub-
jects to optimize their own jump technique, no restrictions
were placed on the time spent in contact with the ground and
the magnitude of the countermovement used. All jumps
were performed with hands placed on hips. In addition to
jump height, the technique variables measured (with regard
to the jump technique continuum) included eccentric lower-
limb stiffness, magnitude of countermovement, and, in the
case of drop jumps, ground contact time.
Training took place over a 10-wk period. Power training
consisted of two supervised sessions per week. Stretch (flex-
ibility) training consisted of one supervised session and
three unsupervised sessions per week. Subjects who per-
formed the stretch training kept a log of date and duration of
the unsupervised sessions. Control subjects were requested
not to start any specific lower-body power or stretch training
for the duration of the 10 wk. Throughout the duration of the
study, all subjects were requested to maintain their sporting
involvement at the volume and intensity as it was for the
weeks before the study. Therefore, the sole form of inter-
vention was the addition of a training stimulus of higher
intensity than that usually experienced.
Subjects. All subjects were male, from a variety of
sporting backgrounds (but primarily basketball and volley-
ball), and generally had little consistent resistance training
experience. None of them had been involved in any struc-
TRAINING EFFECTS ON VERTICAL JUMP TECHNIQUE
tured plyometric program or stretching program (designed
to increase flexibility) in the past year. The mean age,
height, and body mass (⫾ SD) of all the subjects were 24 ⫾
4 yr, 1.82 ⫾ 0.06 m, and 82.93 ⫾ 11.71 kg, respectively.
From an initial 60 subjects that were randomly assigned
to each group (i.e., 15 per group), one subject discontinued
from each of groups PS and C, and four subjects discontin-
ued from each of groups P and S. This left a total of 50
subjects who completed the study requirements (PS ⫽ 14, P
⫽ 11, S ⫽ 11, and C ⫽ 14). Part way through the study, one
subject in group PS suffered discomfort in one knee when
performing drop jumps. In the interest of safety, his program
was slightly modified and drop jump data were not obtained
in the final testing session. Nonetheless, the subject did
perform CMJ and all other test requirements.
Training procedures. Group P performed a power
training program, group S performed a stretching (flexibil-
ity) program, and group PS performed a program combining
power training and stretching. The “double volume” of
training that the PS group undertook was considered irrel-
evant due to the fact that the two programs aimed to improve
different aspects related to performance. All subjects kept a
logbook of their required tasks.
The power training program involved twice a week train-
ing performing a combination of resistance training exer-
cises and plyometrics (see Table 1). The resistance training
sessions comprised of a deadlift/squat hybrid exercise and
weighted countermovement jumps with dumbbells held in
the hands. The plyometrics included drop jumps and coun-
termovement jumps.
The deadlift/squat hybrid exercise was performed with a
bar that allowed subjects, standing with arms hanging by
their sides, to squat down and grasp two fixed parallel
handles slightly wider than shoulder width. Although sim-
ilar to a conventional barbell deadlift in concept, the actual
movement pattern is more like that of a squat due to the
unobstructed path of the bar about the knees. The bar was
positioned on blocks to allow the subject to start from a knee
angle of approximately 90°. The advantage of such an
exercise was the limited amount of equipment required to
perform the exercise in safety.
The weighted and unweighted countermovement jumps
and drop jumps were all performed without the use of the
arms (i.e., either hands on hips or holding dumbbells) and
with the only goal being to jump for maximum height. All
jump exercises were performed with maximum effort right
from the beginning of the program.
The lower-body stretching program was performed 4 d
per week. One of the sessions was supervised whereas the
other three were unsupervised. The program included a
variety of common static stretches for the hamstrings, quad-
riceps, hip extensors, hip adductors and abductors, and
plantarflexors. Each stretch was held at the point of only
mild discomfort with an emphasis on relaxing the entire
body. As the discomfort subsided, the subjects were in-
structed to slowly and very gently take the stretch further. If
required, subjects were permitted to take small breaks (a few
seconds’ duration in which the stretching tension was
Medicine & Science in Sports & Exercise姞 479
TABLE 1. 10-wk power training program.
Weighted CMJ (kg ⴝ mass Deadlift/squat (% ⴝ perceived
Unweighted placed on each dumbbell percentage of maximum effort for
Week Session CMJ handle) Drop Jumps (cm ⴝ drop height) the last repetition of each set)
1 A 2 ⫻ 10 @ DB handles 3 ⫻ 10 @ 60%
2 B 2 ⫻ 10 @ DB handles 3 ⫻ 10 @ 70%
C 2 ⫻ 10 @ 2.5 kg 2 ⫻ 10 @ 30 cm 2 ⫻ 10 @ 80%
3 D 2 ⫻ 10 @ 2.5 kg 2 ⫻ 10 @ 30 cm 2 ⫻ 10 @ 85%
E 2 ⫻ 10 @ 2.5 kg 2 ⫻ 10 @ 30 cm 2 ⫻ 10 @ 90%
4 F 1⫻8 2 ⫻ 8 @ 3.75 kg 2 ⫻ 8 @ 45 cm 2 ⫻ 8 @ 95%
G 1⫻8 2 ⫻ 8 @ 3.75 kg 2 ⫻ 8 @ 45 cm 2 ⫻ 8 @ 80%
5 H 1⫻8 2 ⫻ 8 @ 5 kg 1 ⫻ 8 @ 45 cm; 1 ⫻ 8 @ 60 cm 2 ⫻ 8 @ 100%
I 1⫻8 2 ⫻ 8 @ 5 kg 1 ⫻ 8 @ 45 cm; 1 ⫻ 8 @ 60 cm 2 ⫻ 8 @ 80%
6 J 3⫻3 2 ⫻ 8 @ 5 kg 2 ⫻ 8 @ 100%
K 3⫻3 1 ⫻ 8 @ 45 cm; 4 ⫻ 8 @ 60 cm
7 L 4⫻6 3 ⫻ 6 @ 6.25 kg 3 ⫻ 6 @ 100%
M 4⫻3 1 ⫻ 6 @ 45 cm; 1 ⫻ 6 @ 60 cm; 3 ⫻ 6 @ 75 cm
8 N 4⫻6 3 ⫻ 6 @ 6.25 kg 3 ⫻ 6 @ 100%
O 4⫻3 1 ⫻ 6 @ 45 cm; 1 ⫻ 6 @ 60 cm; 3 ⫻ 6 @ 75 cm
9 P 4⫻6 3 ⫻ 6 @ 7.5 kg 3 ⫻ 6 @ 100%
Q 4⫻3 1 ⫻ 6 @ 60 cm; 1 ⫻ 6 @ 75 cm; 3 ⫻ 6 @ 90 cm
10 R 4⫻6 3 ⫻ 6 @ 7.5 kg 3 ⫻ 6 @ 100%
S 4⫻3 1 ⫻ 6 @ 60 cm; 1 ⫻ 6 @ 75 cm; 3 ⫻ 6 @ 90 cm
Exercises are prescribed as “sets ⫻ reps @ load,” where load can refer to the mass used, drop height, or percentage of maximum effort.
Warm-up/preparation sets are also performed for each exercise but are not listed in the above program.
Rest durations between reps and sets respectively are: 20 s and 2 min for unweighted CMJ; 20 s and 3 min for weighted CMJ and drop jumps. The deadlift/squat exercise
is performed for sets of continuous repetitions with 4 min between sets.
Once at 100% of perceived maximal effort in the deadlift/squat exercise the weight used is increased whenever possible.
Apart from the deadlift/squat, all other exercises are performed with maximal effort on every single repetition.
DB, dumbbell; CMJ, countermovement jumps.

released) during each stretch repetition. From week 4 on- mation recorded in these logbooks and insight gained from
ward, some proprioceptive neuromuscular facilitation regular discussion with each subject suggested that in any
(PNF) stretching was included only in the supervised ses- particular week at least three-quarters of the set program
sions (using intermittent bouts of 10-s submaximal contrac- was completed. If a subject did not achieve these standards
tions followed by a relaxed stretch). See Table 2 for infor- on a regular basis, then their involvement in the study was
mation regarding sets and repetitions of the stretching discontinued. There existed no notable differences in adher-
program. ence to stretching programs between groups. From logbooks
At the initial testing session the control subjects com- and contact kept with the control subjects, it was inferred
pleted a form to indicate their physical activities during the that these subjects did adhere to their requirements of main-
past 3 wk. During the 10 wk between their initial test and taining involvement in their regular activities.
final test, each control subject was required to fill out a daily Testing procedures. Each subject was tested before
physical activity log. This provided a means of checking for and after the 10 wk of training. On each occasion, subjects
any significant change in physical activity over the 10 wk. were requested to refrain from any physically exhausting
Adherence to programs. All power training sessions exercise 48 h before the testing session and to have done no
were supervised, and if a subject could not attend a session, previous exercise on the day of the test. After obtaining
a replacement appointment was made. Only one from every measures of body mass and height, there were two main
four stretching sessions was supervised; however, each sub- sections to the testing session: a) stretch tolerance test for
ject kept a logbook of the unsupervised sessions. Subjects in hamstrings and quadriceps and b) four variations of a ver-
the stretching groups were encouraged to adhere to the tical jump.
programmed unsupervised sessions but also to record a true Measures of stretch tolerance of the hamstrings and quad-
reflection of the stretching that was performed. The infor- riceps were preceded by a warm-up on a cycle ergometer
(Monark 818, Varberg, Sweden) for 5 min at 120 W. One
TABLE 2. 10-wk stretching program. experimenter performed all of the flexibility measures on all
Duration of Repetitions Estimated Duration subjects. The procedures used to measure stretch tolerance
Each Stretch for Each of Stretching were designed specifically for this study; however, some of
Week (s) Stretch Session (min)
the methods were similar to those suggested by Hubley-
1 and 2 20 3 18
3 and 4 30 3 27
Kozey (6). The subject’s body (particularly his pelvis) was
5 and 6 40 3 36 held still by strapping the subject to a specially designed
7 and 8 50 3 45 table. A Leighton Flexometer (Leighton Flexometer, Spo-
9 and 10 60 3 54
kane, WA) was used to measure joint angle while moving
At any one stretching session, two common static stretch exercises are performed for
each of the following muscle groups: hamstrings, quadriceps, gluteals and hip abduc- the joint through its range of motion. All stretches were
tors, hip adductors, ankle plantarflexors. passive; that is, an external force (provided manually by the
Four stretching sessions are programmed in each week.
From week 4 onward, some proprioceptive neuromuscular facilitation (PNF) experimenter) moved the joint through its range of motion.
stretching is included once per week. This was performed slowly yet progressively over three
480 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org
periods of 30 s while gaining feedback from the subject
regarding the tolerance of the stretch. The subjects were
informed that they were in control of how far the stretch was
performed and at the end of the last 30-s period they should
aim to reach a point at which they can tolerate the stretch
being held for at least 10 s yet feel that they would not be
injured. For each trial, the maximum amount of flexion (hip
or knee) was measured and used as the approximate starting
point for the following trial. The maximum range of motion
measured on the last trial was used as the measure of stretch
tolerance. Great care was taken to ensure continual feedback
from the subjects to safeguard against injuries.
To test the stretch tolerance of the hamstrings, the subject
laid supine on a table with one strap across the anterior
superior iliac spines and another across the upper thigh of
one leg. A flexometer was attached to the lateral side of the
“free” leg, just above the knee. The flexometer was set to
zero while the knee was extended and the leg flat against the
table. The subject was requested to keep his leg completely
straight while the experimenter raised it in the subject’s
sagittal plane. Both left and right legs were tested, and the
average of the two used for analysis.
To test the stretch tolerance of the quadriceps, the subject
laid prone on a table with knee joints protruding just over
the end of the table. Two straps fastened the subject tightly
to the table: one strap across the upper thighs immediately
below the buttocks and one strap across the 5th lumbar
vertebrae. A flexometer was strapped to one ankle and faced
laterally. While the subject laid with his chest against the
table, the leg to be tested was fully extended, completely
relaxed, and the flexometer was set at zero. For the first two
trials, the subject laid flat with the chest against the table as
the experimenter applied a force to flex the subject’s knee.
Before the third trial, the subject rose up on his elbows in a
“sphinx position” with his upper arms vertical and forearms
flat against the table. This was to aid holding the subject’s
pelvis flat and prevent any hip flexion. This was only done
on the last trial as some subjects found it too uncomfortable
to remain in this position for longer periods of time.
All jumps were performed on a force plate (Bertec 6090;
Bertec Corporation, Columbus, OH) used to collect vertical
ground reaction force data. The force plate amplifier (Bertec
AM6-3) was interfaced via an A/D MacLab System (ADI
MacLab Systems, Dunedin, New Zealand) to a Macintosh
computer (Power PC 8100,100AV) running Chart version
3.4/s. A two-point calibration of vertical force application
was performed with known weights immediately before
each test. The sampling rate was 400 Hz.
After a jump-specific warm-up, subjects performed four
variations of a vertical jump (all with hands placed firmly on
hips) in the following order: CMJ, DJ30, DJ60, and DJ90.
Three maximal effort trials of each jump variation were
attempted, preceded by one or two submaximal trials to
become familiar with the “new” jump. It was considered
more appropriate not to randomize the order of the jumps
but rather to allow the subjects opportunity to optimize their
technique by performing each jump variation in consecutive
order. Approximately 1–2 min were taken between individ-
TRAINING EFFECTS ON VERTICAL JUMP TECHNIQUE
ual jumps. The only outcome goal was maximal height; that
is, there were no restrictions placed on the magnitude of
countermovement or ground contact time (in the case of
drop jumps).
Calculations. Jump heights for all jumps were calcu-
lated, and the greatest height for each of CMJ, DJ30, DJ60,
and DJ90 were used. For CMJ, it was assumed the vertical
velocity of the body mass center, before performing the
jump, was zero. Therefore, the vertical velocity of the body
mass center at take-off was calculated using the impulse-
momentum relationship. Jump height (h) was then calcu-
lated using h ⫽ v2/(⫺2a) where v is the velocity at take-off
and a is acceleration due to gravity.
Jump heights of the drop jumps (DJ30, DJ60, and DJ90)
could not be determined from the vertical impulse method as
the velocity of the initial landing from the drop height box
was unknown. An alternative method of calculating jump
height from flight time was used. Jump height (h) can be
estimated from flight time using: h ⫽ (a·t2)/2 where t is
equivalent to 1/2 flight time and a is acceleration due to
gravity. The assumption made with this calculation is that
the height of the body mass center is identical for both the
instants of take-off and landing. To help achieve this, sub-
jects jumped with hands on hips; however, it was clear that
for some subjects this method overestimated jump height.
As a result, a corrected flight time method was used to
calculate drop jump height.
From an assumption that the expected error (of the flight
time calculation method) for the drop jumps would be
similar to that of CMJ, the expected error was derived by
calculating all three CMJ heights with both the flight time
method and the impulse method, subtracting the former
from the latter and then taking the mean of the three differ-
ences. This expected error was then subtracted from the
flight time calculated heights of the best drop jumps.
Stiffness of the combined lower limbs produced during
the eccentric muscle action phase (i.e., eccentric lower-limb
stiffness) was calculated for the highest jumps of each of
CMJ, DJ30, DJ60, and DJ90. Stiffness was calculated from
the ratio of change in vertical ground reaction force to the
simultaneous change in height of the subject’s body mass
center. This was obtained via double integration of the
vertical acceleration data measured from the force platform
(3,5). In this process, knowledge of the velocity at some
point in time is required. For the CMJ, this was obtained
from where the subject stood still before initiating the jump
(i.e., velocity equalled zero). For DJ30, DJ60, and DJ90, the
velocity point used was that of the instant of jump take-off
(v) calculated using: v ⫽ (⫺2a·h)0.5 where a is acceleration
due to gravity and h is the height jumped. Also, to calculate
absolute position data, the position of center of mass at some
point in time was required. For CMJ, the height of the
subject’s body mass center when standing erect was defined
as zero; for all drop jumps the height of the subject’s center
of mass at initial contact (after dropping from the drop
height) was defined as zero. Note that each individual ath-
lete was capable of performing jumps with a wide variation
in eccentric lower-limb stiffness. Therefore, it should be
Medicine & Science in Sports & Exercise姞 481

FIGURE 1—Vertical ground reaction force as a function of position of
body mass center for the duration of the eccentric phase of (A) a
countermovement jump and (B) three different subjects all performing
drop jumps from a 60-cm height. Lowering of the subject’s body mass
center was considered positive, thereby resulting in positive measures
of lower-limb stiffness. For CMJ, standing upright and the end of the
eccentric phase are indicated in the graph, as is the linear portion over
which lower-limb stiffness was calculated. For the drop jumps, the
initial contact for each subject occurs at point (0,0), and the end of the
eccentric phase is indicated by the “free end” of each tracing; the slope
of the straight lines represents lower-limb stiffness.
kept in mind that this measure of eccentric lower-limb
stiffness is that which each subjects “chose” to produce for
each particular jump variation so as to obtain maximum
jump height.
Figure 1 shows representative graphs for vertical ground
reaction force as a function of vertical position of the body
mass center. Note that lowering of the center of mass has
been defined as positive. Figure 1A is that of a CMJ. The
start of the CMJ eccentric phase was defined as the point
where acceleration reached a minimum; the end was when
the subject’s body mass center reached its lowest position.
Typically, the eccentric phase contained a relatively linear
portion when excluding the first and last 5 cm of vertical
displacement. It was considered that the slope of this “mid-
dle” portion would best describe the predominant level of
eccentric lower-limb stiffness for the CMJ.
Figure 1B shows vertical ground reaction force as a
function of vertical position of the body mass center for
482 Official Journal of the American College of Sports Medicine
TABLE 3. Contrast coefficients used for statistical analysis.
Contrast Coefficients
PS P S C Interpretation
0.5 0.5 ⫺0.5 ⫺0.5 CPG vs CNPG
1 ⫺1 0 0 PS vs P
0 0 1 ⫺1 S vs C
The interpretation shows which two groups (or which two combined groups) were
contrasted. PS, power & stretch group; P, power group; S, stretch group; C, control
group; CPG, combined power groups (i.e., PS and P); CNPG, combined nonpower
groups (i.e., S and C).
three different subjects all performing drop jumps from a
60-cm drop height. Unlike the CMJ, there was typically no
distinct linear portion for the eccentric phase of the jumps.
Eccentric lower-limb stiffness for the drop jumps was cal-
culated as the change in vertical ground reaction force over
the change in position of the body mass center for the
complete eccentric muscle action phase during ground con-
tact. This approach permitted the detection of differences in
“average” stiffness produced as indicated by the three sub-
jects in Figure 1 (with the slope of the straight lines repre-
senting lower-limb stiffness).
From the body mass center position data, derived via
double integration of the vertical acceleration data, the ver-
tical displacement of the body mass center during the ec-
centric phase of each highest jump was calculated. The
starting reference point for CMJ was the position of center
of mass while standing erect and for DJ was the position of
center of mass at initial ground contact. For both CMJ and
DJ, the end of the eccentric phase was defined as the lowest
position of the body mass center.
Ground contact time was measured for the highest jumps
of each of DJ30, DJ60, and DJ90. This variable was mea-
sured from the instant of landing to the instant of take-off as
detected by the vertical ground reaction force data.
Statistical methods. The statistical methods of this
study were chosen so as to allow sample size to be kept at
a practical level and to maintain a reasonable level of
statistical power (80% chance of detecting an effect size of
0.35 in vertical jump height with P ⬍ 0.10). This involved
using a limited amount of preplanned contrasts with no
adjustments made for multiple tests. It was determined that
14 subjects in each of the four groups would be sufficient.
Subject numbers did drop slightly below this figure for some
of the groups, which would have slightly decreased the
statistical power in detecting small effect sizes for contrasts
between two groups. However, many of the contrasts did
involve combining groups (see explanation in next few
paragraphs), which ensured ample subjects for these
contrasts.
The differences between pre- and post-training measures
for each variable were analyzed with a one-way ANOVA
and with a limited number of orthogonal preplanned con-
trasts (Table 3). These preplanned contrasts were designed
to answer three specific questions for each of the variables.
They were: a) what effect did the power training have,
regardless of whether stretching was performed or not; b)
what effect did the addition of the stretching program to a
power training program have; and c) what effect did the
http://www.acsm-msse.org
TABLE 4. Stretch tolerance of hamstrings and quadriceps.
Combined
Stretch
Groups
Hamstrings (°) Pre 69 ⫾ 10
Post 98 ⫾ 14
§*** Change 29 ⫾ 10
Quadriceps (°) Pre 155 ⫾ 14
Post 165 ⫾ 8
§*** Change 10 ⫾ 10
Values are mean ⫾ SD. Evidence of differences in “change” is indicated via the symbol:
§ denotes CSG increased stretch tolerance more than CNSG; *** P ⬍ 0.01.
stretching program have in contrast to a control group? Note
that the contrast related to the first question combined both
groups which performed power training (groups PS and P)
and both groups which did not (groups S and C). These are
later denoted in the text as CPG (combined power groups)
and CNPG (combined nonpower groups), respectively.
The exception regarding use of these contrast coefficients
was one other preplanned contrast between groups PS and S
combined (CSG ⫽ combined stretching groups) and groups
P and C combined (CNSG ⫽ combined nonstretching
groups) for stretch tolerance of the hamstrings and quadri-
ceps. This was designed to answer the question: Was the
stretching program effective (regardless of whether power
training was performed)?
Alpha was set at 0.10, however, the level of evidence of
an effect (8,9) is indicated in the tables with either * P ⬍
0.10, ** P ⬍ 0.05, or *** P ⬍ 0.01. Any contrasts that
“approached significance” are also indicated in the tables.
RESULTS
Analyses on changes in stretch tolerance were made be-
tween all subjects who performed stretching (CSG) and all
those who did not (CNSG). For the hamstrings stretch
tolerance test, there were 25 subjects in each group. Due to
some subjects reaching the maximum level of knee flexion
for their initial test (and thereby not included in analysis),
the quadriceps stretch tolerance analysis included only 22
subjects in the CSG and 20 subjects in the CNSG. Stretch
tolerance results are shown in Table 4.
Fifty subjects were included in all the CMJ test analyses,
whereas only 49 were included in the drop jump analyses
(due to knee discomfort experienced by one subject during
drop jumps). Jump height results are shown in Table 5.
For DJ30, DJ60, and DJ90 eccentric lower-limb stiffness,
two or three outliers existed (i.e., subjects who landed with
markedly greater stiffness than the others). Therefore, anal-
yses were performed with and without these outliers re-
moved to see whether the same conclusions were revealed.
Results for when the outliers were removed are shown in
Table 6. When outliers were not removed (and the data were
transformed to satisfy the assumptions of ANOVA), the
statistical inferences were relatively similar except in the
case of DJ90 where there was no evidence of a difference.
TRAINING EFFECTS ON VERTICAL JUMP TECHNIQUE
Results for vertical displacement of the body mass center
Combined are shown in Table 7. Ground contact time was determined
Nonstretch for DJ30, DJ60, and DJ90 only and are shown in Table 8.
Groups
77 ⫾ 23
82 ⫾ 22
6⫾7
DISCUSSION
159 ⫾ 11 Often research is conducted with the goal of determining
158 ⫾ 12 if one form of training is superior to another. However, in
0⫾5 many cases, training effects other than performance out-
come (e.g., effects on technique) remain unknown. This
current study assessed the effects of power training and
flexibility training on jump technique, in addition to jump
performance. Insight gained regarding the effects of the
training methods on technique could possibly contribute to
a more informed choice of training methods by coaches and
athletes. From the results of this study, it is likely that if the
training goal for drop jumps is maximum jump height alone,
technique will change in the direction of a lower eccentric
leg stiffness, greater depth of countermovement, and possi-
bly a longer ground contact time, whereas for a counter-
movement jump eccentric leg stiffness and the depth of
countermovement may both increase. Depending on the
desired outcome, these changes may be either detrimental,
advantageous, or irrelevant to the athlete’s sport. The deci-
sion remains with the coach and athlete.
The results from this study also suggest that stretching has
little effect on the jump technique variables measured but
may offer some performance outcome benefits, at least in
the case of CMJ. Other researchers have also reported im-
proved stretch-shortening cycle performance with the addi-
tion of stretching to a strength program (in subjects previ-
ously unfamiliar with regular flexibility training; 10). At this
point in time, these results suggest that stretching should not
be ignored as a possible source of improvement in stretch-
shortening cycle tasks, at least for those athletes who are
inflexible. However, further research is required.
The level of eccentric lower-limb stiffness produced dur-
ing the best (highest) jump for each jump variation was used
for analysis. Although each subject was capable of produc-
ing a variety of stiffness levels (i.e., acute regulation of
stiffness), the level produced during the highest jump was
considered that which allowed optimal technique so as to
maximize jump height. Subsequently, because of the direct
effect of eccentric lower-limb stiffness on jump technique,
it was included (with magnitude of countermovement and
ground contact time) under the term “technique variables.”
Farley et al. (4) measured lower-limb stiffness produced
by subjects when performing two-legged hopping in place at
a preferred frequency. These authors described the behavior
of the lower limbs as being “spring-like.” That is, the
vertical ground reaction force was dependent on the dis-
placement of the body mass center. In another similar study
(5), they explained their force displacement graphs of two-
legged hopping as being “. . .approximately linear at high
levels of force and displacement.” This was not the case for
the drop jumps in the current study, although CMJ were
mostly linear (see Fig. 1 for an example). As a result of this,
the reader must be aware, especially in the case of the drop
Medicine & Science in Sports & Exercise姞 483
TABLE 5. Jump heights; the maximum jump height of each jump variation was used for analysis.
Power & Stretch Power Stretch Control
CMJ (cm) Pre 38.5 ⫾ 5.3 36.6 ⫾ 4.8 35.1 ⫾ 5.4 37.0 ⫾ 5.7
Post 43.4 ⫾ 6.2 39.5 ⫾ 6.0 36.4 ⫾ 5.9 36.7 ⫾ 5.7
¥** ‡** §*** Change 4.9 ⫾ 2.2 2.9 ⫾ 2.6 1.3 ⫾ 1.5 ⫺0.3 ⫾ 1.2

DJ30 (cm) Pre 38.7 ⫾ 5.2 34.6 ⫾ 4.7 34.6 ⫾ 4.3 35.2 ⫾ 6.1
Post 42.2 ⫾ 4.9 38.3 ⫾ 5.1 35.7 ⫾ 5.0 35.8 ⫾ 6.3
§*** Change 3.5 ⫾ 2.5 3.7 ⫾ 2.8 1.2 ⫾ 2.7 0.6 ⫾ 1.6

DJ60 (cm) Pre 38.5 ⫾ 4.7 34.5 ⫾ 4.9 35.0 ⫾ 4.7 35.0 ⫾ 6.3
Post 42.0 ⫾ 5.0 38.0 ⫾ 4.8 34.9 ⫾ 4.7 35.9 ⫾ 6.3
§*** Change 3.5 ⫾ 3.5 3.5 ⫾ 3.7 ⫺0.1 ⫾ 2.7 0.8 ⫾ 1.8

DJ90 (cm) Pre 38.5 ⫾ 5.0 34.4 ⫾ 4.8 33.6 ⫾ 5.1 34.5 ⫾ 6.4
Post 41.9 ⫾ 5.5 36.7 ⫾ 4.2 34.0 ⫾ 5.1 34.9 ⫾ 5.8
§*** Change 3.4 ⫾ 3.3 2.3 ⫾ 2.0 0.3 ⫾ 3.8 0.3 ⫾ 2.1
Values are mean ⫾ SD. Evidence of differences in “change” are indicated via the symbols: ¥ denotes PS increased more than P; ‡ denotes S increased more than C; § denotes CPG
increased more than CNPG; ** P ⬍ 0.05; *** P ⬍ 0.01.

jumps, that the measures of lower-limb stiffness are calcu-
lated simply from force and position differences over two
points in time. Nonetheless, even for the drop jumps, these
measures of lower-limb stiffness were considered to suffi-
ciently describe the differences in jumping technique used
by the subjects. To emphasize that stiffness was calculated
over the eccentric phase of the jump the term “eccentric
lower-limb stiffness” has been used in this paper.
Interestingly, an inconsistency existed in the effects of
training on the level of eccentric lower-limb stiffness pro-
duced during CMJ and the drop jumps. Power training,
regardless of whether stretching was performed or not, was
shown to increase the level of eccentric lower-limb stiffness
produced during CMJ but to decrease that of DJ30, DJ60,
and DJ90. No evidence existed to suggest that stretching had
any effect on the level of eccentric lower-limb stiffness
used. CMJ and drop jump eccentric lower-limb stiffness
were derived from the vertical force-displacement graphs
with slightly different methodologies. However, it is un-
likely that such slight variation in the methodologies would
explain the extreme differences of the effect of the training
on the level of eccentric lower-limb stiffness produced (i.e.,
an increase for CMJ versus decreases for the drop jumps).
Before discussing the possible reasons of the effect of
power training on the level of eccentric lower-limb stiffness
produced, it might be helpful to consider the changes in the
other jump-technique variables. Evidence suggested that
power training, regardless of whether stretching was per-
formed or not, resulted in a greater vertical displacement of
the body mass center. Furthermore, there was some evi-
dence that ground contact time increased for the drop jumps.
Just as with eccentric lower-limb stiffness, there was gen-
erally no evidence to suggest that stretching had any effect
on vertical displacement of the body mass center and ground
contact time.
When only considering the drop jumps, it appears that the
power training enabled the subjects to decrease the level of
eccentric lower-limb stiffness used and to absorb the early
ground reaction forces over a greater distance and time. It is
possible the subjects, at the initial testing session, performed
the drop jumps with suboptimal technique. However, with
habitual jumpers chosen for the study, the familiarization
process used, and the best jump of each jump variation used
for analysis, this inference seems unlikely. A more likely
explanation would be that the training adaptations allowed
the new technique to evolve. Just as in testing, during
training there were no restrictions placed on the depth of
countermovement used and ground contact time. Without
these restrictions, the subjects were free to modify their
technique so as to optimally use their current neuromuscular
capabilities.

TABLE 6. Eccentric lower limb stiffness data with outliers removed.

Power & Stretch Power Stretch Control
CMJ (kN䡠m⫺1) Pre 4.7 ⫾ 2.8 3.7 ⫾ 2.7 4.1 ⫾ 2.1 5.6 ⫾ 4.1
Post 5.5 ⫾ 1.9 5.5 ⫾ 1.1 4.4 ⫾ 2.5 4.7 ⫾ 2.9
‡** Change 0.8 ⫾ 2.2 1.8 ⫾ 2.2 0.3 ⫾ 2.5 ⫺0.9 ⫾ 2.1

DJ30 (kN䡠m⫺1) Pre 7.1 ⫾ 2.3 6.8 ⫾ 1.8 5.7 ⫾ 2.9 6.4 ⫾ 2.9
Post 5.7 ⫾ 1.8 5.1 ⫾ 0.8 6.0 ⫾ 3.7 6.1 ⫾ 2.2
§*** Change ⫺1.4 ⫾ 1.8 ⫺1.7 ⫾ 2.3 0.3 ⫾ 1.7 ⫺0.4 ⫾ 1.4

DJ60 (kN䡠m⫺1) Pre 6.9 ⫾ 3.1 6.7 ⫾ 2.9 4.7 ⫾ 1.3 5.0 ⫾ 1.7
Post 5.0 ⫾ 1.0 4.4 ⫾ 0.4 4.1 ⫾ 0.8 5.2 ⫾ 1.6
§*** Change ⫺1.9 ⫾ 3.0 ⫺2.3 ⫾ 3.0 ⫺0.7 ⫾ 0.9 0.2 ⫾ 1.0

DJ90 (kN䡠m⫺1) Pre 5.4 ⫾ 1.5 5.2 ⫾ 1.5 4.3 ⫾ 1.4 4.5 ⫾ 1.4
Post 4.6 ⫾ 0.9 4.1 ⫾ 0.4 3.9 ⫾ 1.0 4.3 ⫾ 0.8
§* Change ⫺0.8 ⫾ 1.1 ⫺1.1 ⫾ 1.5 ⫺0.4 ⫾ 0.7 ⫺0.2 ⫾ 1.0
Values are mean ⫾ SD. Evidence of differences in “change” are indicated via the symbols: ‡ denotes CPG increased more than CNPG; § denotes CPG decreased more than CNPG;
* P ⬍ 0.10; ** P ⬍ 0.05; *** P ⬍ 0.01.

484 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

TABLE 7. Vertical displacement of the body mass center to the lowest point of the eccentric phase.
Power & Stretch Power Stretch Control
CMJ (cm) Pre 35.8 ⫾ 8.3 40.4 ⫾ 7.8 36.8 ⫾ 6.2 37.5 ⫾ 7.1
Post 43.9 ⫾ 6.1 45.1 ⫾ 6.6 39.1 ⫾ 8.3 37.8 ⫾ 6.7
§*** Change 8.1 ⫾ 4.7 4.7 ⫾ 8.5 2.3 ⫾ 6.2 0.3 ⫾ 6.3

DJ30 (cm) Pre 30.4 ⫾ 10.0 34.1 ⫾ 7.9 37.1 ⫾ 9.1 36.9 ⫾ 11.0
Post 41.8 ⫾ 9.0 41.8 ⫾ 5.8 36.9 ⫾ 12.1 36.3 ⫾ 9.8
§*** Change 11.3 ⫾ 10.6 7.7 ⫾ 9.6 ⫺0.3 ⫾ 5.6 ⫺0.5 ⫾ 5.5

DJ60 (cm) Pre 37.1 ⫾ 10.0 38.5 ⫾ 10.0 41.2 ⫾ 11.9 42.6 ⫾ 12.3
Post 48.1 ⫾ 7.6 47.4 ⫾ 6.8 45.7 ⫾ 8.8 40.5 ⫾ 8.3
‡* §*** Change 11.0 ⫾ 8.2 8.9 ⫾ 9.5 4.5 ⫾ 6.8 ⫺2.1 ⫾ 11.3

DJ90 (cm) Pre 41.6 ⫾ 11.5 44.0 ⫾ 9.7 46.5 ⫾ 12.4 47.5 ⫾ 12.8
Post 51.5 ⫾ 7.5 50.3 ⫾ 6.4 50.4 ⫾ 10.6 49.2 ⫾ 8.8
§** Change 9.9 ⫾ 7.6 6.4 ⫾ 9.0 4.0 ⫾ 5.6 1.7 ⫾ 10.4
Values are mean ⫾ SD. These positive measures indicate to what extent the subjects lowered their body mass center; i.e., down is defined as positive. Evidence for differences are indicated
via the symbols: ‡ denotes S increased their displacement more than C; § denotes CPG increased their displacement more than CNPG; * P ⬍ 0.10; ** P ⬍ 0.05; *** P ⬍ 0.01.

In contrast to the drop jumps, there was strong evidence
to suggest that the level of eccentric lower-limb stiffness
produced during CMJ increased for all those subjects who
performed power training (regardless of performing stretch-
ing or not). Despite this, there was strong evidence that
vertical displacement of the body mass center increased.
That is, these subjects used a greater magnitude of
countermovement.
The question remains, “Why did the subjects appear to
benefit from the addition of a stretching program for CMJ
but not for any of the drop jumps?” A possible explanation
might be the changes in jump technique that were adopted
over the 10 wk of training. Komi (7) has suggested that
greater lower-limb stiffness during the eccentric phase of a
stretch-shortening cycle task might be an advantage by
allowing greater storage and release of elastic energy. Note
that for all those subjects who performed power training, the
level of eccentric lower-limb stiffness produced during CMJ
increased. CMJ was the only jump variation in which this
occurred and was the only jump variation in which group PS
appeared to benefit (with a greater jump height improve-
ment) over group P, likewise for group S over group C. It is
possible that the greater eccentric lower-limb stiffness (sup-
posedly via acute regulation of the contractile components
of the musculature involved) did allow greater utilization of
elastic energy, particularly in the subjects of group PS.
Wilson et al. (10) have reported benefits (to a rebound bench
press exercise) gained from stretching program. The authors
presented evidence of a stretching program causing a de-
crease in series elastic component stiffness and proposed
that it allowed greater storage of elastic energy, thereby
resulting in greater stretch-shortening cycle performance.
The subjects in group S of the current study may have
benefited in a such a way. Group S did not increase their
level of eccentric lower-limb stiffness produced during
CMJ; rather, it remained approximately the same. Greater
compliance of the series elastic component (due to increased
flexibility; 10) with the same eccentric lower-limb stiffness
would have still allowed more of the eccentric force to be
stored in the series elastic component, supposedly resulting
in benefits due to recoil of elastic energy. However, this is
speculation and requires further research.
In the cases of the drop jumps, the level of eccentric
lower-limb stiffness produced decreased, and ground con-
tact time and magnitude of the countermovement used in-
creased. Furthermore, there was no apparent advantage to
performing the stretching program. This agrees with the
above hypothesis in the following way. Greater storage (and
then utilization) of elastic energy is only possible through
increased eccentric lower-limb stiffness, despite any effects
that stretching might have had. That is, all subjects in groups
PS and P decreased the level of eccentric lower-limb stiff-
ness produced (possibly to help absorb the high impact
forces and prevent Golgi tendon inhibition) and in doing so
decreased their opportunity to store elastic energy. If this
was the case, then the stretching program would possibly
not be of the same advantage as when training for CMJ (or
any other stretch-shortening cycle task in which eccentric

TABLE 8. Ground contact time.

Power & Stretch Power Stretch Control
DJ30 (ms) Pre 440 ⫾ 126 486 ⫾ 111 557 ⫾ 109 535 ⫾ 149
Post 576 ⫾ 101 606 ⫾ 39 581 ⫾ 177 546 ⫾ 117
§*** Change 136 ⫾ 134 120 ⫾ 133 24 ⫾ 142 11 ⫾ 63

DJ60 (ms) Pre 478 ⫾ 134 511 ⫾ 139 538 ⫾ 136 553 ⫾ 163
Post 577 ⫾ 87 617 ⫾ 29 605 ⫾ 60 571 ⫾ 101
§* Change 99 ⫾ 117 106 ⫾ 138 67 ⫾ 125 18 ⫾ 116

DJ90 (ms) Pre 489 ⫾ 124 550 ⫾ 110 555 ⫾ 123 588 ⫾ 159
Post 593 ⫾ 88 628 ⫾ 30 623 ⫾ 80 609 ⫾ 85
# Change 105 ⫾ 96 78 ⫾ 103 69 ⫾ 71 21 ⫾ 123
Values are mean ⫾ SD. Evidence of differences in “change” are indicated via the symbols: § denotes CPG increased more than CNPG; * P ⬍ 0.10; *** P ⬍ 0.01; # denotes CPG
increased more than CNPG with P ⫽ 0.12.

TRAINING EFFECTS ON VERTICAL JUMP TECHNIQUE Medicine & Science in Sports & Exercise姞 485
lower-limb stiffness was at least maintained over the 10 wk
of training).
Another possible explanation of the decreased drop jump
eccentric lower-limb stiffness (and increased countermove-
ment and ground contact time) for CPG is that, with train-
ing, strength was developed in the deeper countermovement
position, allowing for higher vertical ground reaction forces
to be produced over a longer time period (i.e., greater
impulse). It is possible that the benefits to be gained from
such technique were greater than the benefits that would be
gained from maintaining or increasing lower-limb stiffness
in an attempt to utilize more elastic energy.
In attempting to explain the training-induced changes in
jump technique, it must be considered that the maximization
of one single factor of performance (e.g., utilization of
elastic energy) is probably not optimal. It is possible that the
technique used in a jump that does maximize utilization of
elastic energy might result in suboptimal concentric perfor-
mance by the contractile component. In a drop jump, say,
maximization of elastic energy use might occur with very
high levels of eccentric lower-limb stiffness; however, this
means the subject would only be taking a relatively small
countermovement and, therefore, have less distance and
time over which to produce impulse in the concentric phase.
More benefits are likely to be gained from using a technique
that allows the optimal combination of elastic utilization,
impulse produced by the contractile component, and other
important factors.
In conclusion, it was found that the variables used to
measure jump technique (with regard to the “jump tech-
nique continuum”) changed with 10 wk of power training
REFERENCES
1. BOBBERT, M. F., R. A. HUIJING, and G. J. VAN INGEN SCHENAU. Drop
jumping: the influence of jumping technique on the biomechanics
of jumping. Med. Sci. Sports Exerc. 19:332–338, 1987.
2. BOBBERT, M. F., M. MACKAY, D. SCHINKELSHOEK, P. A. HUIJING,
and G. J. VAN INGEN SCHENAU. Biomechanical analysis of drop and
countermovement jumps. Eur. J. Appl. Physiol. 54:566 –573,
1986.
3. CAVAGNA, G. A. Force platforms as ergometers. J. Appl. Physiol.
39:174 –179, 1975.
4. FARLEY, C. T., R. BLICKHAN, J. SAITO, and C. R. TAYLOR. Hopping
frequency in humans: a test of how springs set stride frequency in
bouncing gaits. J. Appl. Physiol. 71:2127–2132, 1991.
5. FERRIS, D. P., and C. T. FARLEY. Interaction of leg stiffness and
surface stiffness during human hopping. J. Appl. Physiol. 82:15–
22, 1997.
6. HUBLEY-KOZEY, C. L. Testing flexibility. In: Physiological Testing
of the High Performance Athlete, 2nd Ed., J. D. Duncan, H. A.
486 Official Journal of the American College of Sports Medicine
designed to increase vertical jump height in which maximal
height was the only goal. Power training was attributed with
the following changes in drop jump technique: decrease in
eccentric lower-limb stiffness, and increases in magnitude
of countermovement and ground contact time. Stretching
(flexibility training) appeared to offer no added benefits to
drop jump height and had no significant effect on drop jump
technique. Power training was associated with the following
changes in CMJ technique: increases in eccentric lower-
limb stiffness and the magnitude of countermovement.
There was evidence that stretching (flexibility training) did
aid in increasing CMJ height, and possible explanations
were discussed. However, stretching appeared to have no
significant effect on CMJ technique. Coaches and athletes
should consider these technique changes when deciding on
the instructions (i.e., outcome goals) associated with jump
training. Finally, it is proposed that the technique changes
that occurred were a result of attempting to optimize a
complex combination of factors (and their changes as a
result of the training programs) involved in jumping. Further
research is required to investigate: a) the reasons behind
such changes in technique, b) the possible benefits a stretch-
ing (flexibility) program may have to stretch-shortening
cycle tasks, and c) the effects on technique of a jump
training program with instructions to “minimize ground
contact time while still trying to jump as high as possible.”
This research was funded by grant no. 9154 3498044 from Sports
Science New Zealand.
Address for correspondence: Joseph P. Hunter, Department of
Sport and Exercise Science, The University of Auckland, Private Bag
92019, Auckland, New Zealand; E-mail: joe.hunter@xtra.co.nz.
Wenger, and H. J. Green (Eds.). Champaign, IL: Human Kinetics,
1982, pp. 309 –360.
7. KOMI, P. V. Stretch-shortening cycle. In: Strength and Power in
Sport, P. V. Komi (Ed.). London: Blackwell Science, 1992, pp.
169 –179.
8. THOMAS, J. R., and J. K. NELSON. Research Methods in Physical
Activity, 3rd Ed. Champaign, IL: Human Kinetics, 1996, pp.
107–108.
9. WILD, C. J., and G. A. F. SEBER. Chance Encounters: A First
Course in Data Analysis and Inference. New York: Wiley, 1999,
pp. 389 –391.
10. WILSON, G. J., B. C. ELLIOT, and G. A. WOOD. Stretch shortened
cycle performance enhancement through flexibility training. Med.
Sci. Sports Exerc. 24:116 –123, 1992.
11. YOUNG, W. B., J. F. PRYOR, and G. J. WILSON. Effects of instruc-
tions on characteristics of countermovement and drop jump per-
formance. J. Strength Cond. 9:232–236, 1995.
http://www.acsm-msse.org