- Email: [email protected]
Segment interactions within the swing leg during unloaded and loaded running
SEGMENT INTERACTIONS WITHIN THE SWING LEG DURING UNLOADED AND LOADED RUNNING PHILIP E. MARTIN*? and PETER
R. CAVANAGH:
and Sport Research Institute. Arizona State University. Tempe, AZ 85287, U.S.A. and Center for Locomotion Studies. Penn. State University , University Park. PA 16802. U.S.A.
tExercise
:The
Abstract-In this study, designed to determine the effect of lower extremity inertia manipulation on joint kinetics and segment energetics during the swing phase, I5 male distance runners were filmed as they performed treadmill running (3.35 m s-l) under five load conditions: no added load and loads oCO.25 kg and 0.50 kg added to each thigh or each foot. Results of this study demonstrated that the energetics of the lower extremity movements during the swing phase of the running cycle were dominated by mechanical energy transfers between adjacent segments attributed to the joint reaction forces. which acted to redistribute mechanical energy within the system. These contributions were considerably greater than those of the net joint moments. which primarily retlccted muscular generation and dissipation of mechanical energy. Lower extremity loading caused little change in the movement pattern of the swing leg. However, increases in the joint reaction forces and net moments and in the amount of work done and the energy transfer attributed to the reaction forces and moments were observed, but were limited to thejoints proximal to the location of the added load. These results were consistent with the increased aerobic demand associated with increases in lower extremity inertia that have been reported elsewhere and also have implications for the manner
in which the nellromuscular system controls the motion of the legs during running.
INTWODUCTION
Despite the apparent simplicity of human gait and the ease with which most individuals are able to initiate, sustain, and modulate walking and running patterns, there is little doubt that the control of individual body segments during these activities represents a highly complex neuromuscular ellbrt. Segment motions are orchestrated not only by grAvitational and muscle forces acting upon the segments, but also by forces from interactions between adjacent segments. The swing phase of the gait cycle has been the focus of numerous analyses (e.g.. Chapman and Caldwell. 1983; Elliott, 1977; Mena er al.. 1981; Mochon and McMahon. 1980; Phillips et ol.. 1983) and provides an interesting motion for the consideration of interactions between segments and the coordination of their movements. While numerous biomechanical features of the swing phase have been studied previously, there has been little consideration the inertial
properties
influence lower extremity
of how manipulation
of the segments
of
of the leg
dynamics during this period.
researchers have considered the physiolgical consequencesof carrying added load on the lower ex&emities and have consistently shown higher aerobic demands under loaded conditions (e.g., Catlin and Dressendorfer, 1979; Claremont and Hall. 1988; Martin, 1985; Myers and Steudel, 1985). It may be expected that increasesin aerobic demand due to increased lower extremity inertia would be accompanied by biomechanical changes, particularly in joint kinetics and segment energetics within the In contrast
Receiwd l
to this, numerous
in jnol jorm I 5 A ugusr 1989. To whom correspondence should be addressed.
lower extremity. This, however, has not been evaluated previously. Consequently, the purpose of our study was to determine the etTectof lower extremity inertia manipulation on these kinetic features during the swing phase. It was believed that results of this study could be important in understanding factors that influence the aerobic demand or economy of running, specifically those related to the increased demands under loaded conditions. PROCF.DURW Fifteen healthy male long-distance
runners partici-
pated in three test sessionswhich involved running on
a motor driven treadmill. Mean subject age was 29.3 yr (range =20-47) and mean body mass was 72.0 kg (range 3: 54.7-90.3). The iirst session was used to habituate the subjects to the treadmill and to introduce them to the loading procedures prior to data collection. Each subject performed three IO-min bouts of treadmill running separated by IO-min rest intervals at the experimental speed of 3.35 ms-‘. Those subjects who had no previous experience on a treadmill comple!ed an additional 30 min of treadmill training prior to data collection. Each subject then completed two identical experimental sessions in which his response to the loading conditions was measured. Five load conditions were used such that the eflects of both load magnitude and position could be examined. These included: (a) an unloaded condition, (b) 0.25 kg added to each thigh, (c) 0.25 kg added to each foot, (d) 0.50 kg added to each thigh, and (e)O.SOkg added to each foot. The loads were added to the leg in the form of small packets of lead shot. Running shoes with elastic pockets sewn onto the medial and lateral aspects of
530
P. E. MARTINand P. R. CAVANAGH
each shoe and elastic bicycling shorts with pockets sewn onto the medial and lateral sides of the distal aspect of each leg were provided for the subjects and served to stabilize the loads on the segments. Primarily because of differences in anthropometric characteristics between subjects, it was not possible to position the loads on the thighs at the same relative location (i.e., same percentage of thigh length) for all subjects. Since the goal of the load manipulation was simply to produce distal and proximal loading of the leg, the inability to precisely control the positioning of the thigh loads was not considered a major limitation. The location of the thigh loads was quantified for each subject as part of the motion analysis process. The mean location for these loads was approximately 68% of thigh length from the hip joint center (range = 59-80%). An identical testing protocol was used on both experimental days, which were separated by a minimum of I day and a maximum of 3 days of rest. On both test days, each subject first completed a 3-min warm-up run on the treadmill. This was then followed by alternating periods of I2 min of rest and 8 mitt of running at 3.35 m s- ’ until the five randomly ordered load conditions were completed. The 8-min running periods were used in an effort to obtain steady-state conditions of the runners both physiologically and mechanically prior to actual data collection. Standard two-dimensional high-speed cinematographical procedures were used to record a sagittal plane view of the running motions of the subjects under each of the five load conditions. A single Locam camera operating at a nominal ftXIlle rate of 100 frames s- ’ was used to tilm the subjects near the end of each I(-min trial. Camera speed was calibrated using an elcvtronic timing unit positioned in the tield of view of the camera. For each subject, the swing phase for one stride under each load condition was analyzed from each day of testing. Data from the two test strides were pooled to decrease the effect of potentially spurious anomalous results from a single test day on group trends. Coordinate data representing the spatial locations of toe, heel, ankle, knee, hip, and inferior aspect of the ear on the left side of the body were generated from film analysis and served to define the segment endpoints of the foot, shank, thigh, and trunk. Even though the analysis focused on the lower extremity, quantification of the angular position of the trunk was necessaryto fully describe the interaction of the thigh with the trunk. In addition. it was assumed that with limited motion of the head, the inferior aspect of the ear would be a more suitable superior endpoint of the trunk than the shoulder, which can move considerably due to shoulder and trunk rotation. Coordinate data were scaled and digitally filtered using a fourth-order zero-lag Butterworth filter. Selection of cut-otf frequencies was based on an analysis using a residual procedure described by Wells and Winter (1980). Cutoffvalues ranged from 3 Hz for the horizontal coordi-
nate data of the ear to 8 Hz for the coordinate data of the toe and heel. So that data from this study could be compared with those of overground analyses, the horizontal coordinate data were also adjusted for mean treadmill belt displacement in the measurement interval. Values for a series of mechanical power, work. and energy variables were then quantified from the coordinate data using a mechanical power model introduced by Elftman (1939a. b) and later clarified and extended by Quanbury et al. (1975) and Robertson and Winter (1980). These variables included the instantaneous power attributed to the joint reaction forces (PF) and moments (PM) at the ankle, knee, and hip and their associated mechanical work contributions to the foot, shank, and thigh. These were computed at each joint using the following equations: PF,, = F, * <, PM,, = Ml,
*ik
(1) (2)
where j, F, V, rcI and w represent the joint of interest. joint reaction force, joint center velocity, net joint moment, and angular velocity of segment k (either distal or proximal to joint j) at instant i. The role of the joint reaction force has been described as a simple mechanism of energy transfer since the rate at which mechanical energy is lost by one segment is equivalent to the rate at which energy is gained by its neighbor (Alcshinsky, 1986; Robertson and Winter, 1980: Winter. 1979). Unlike the simple transfer function of the joint reaction force, the contributions of the net joint moments were further subdivided into mechanical energy generation (i.e., concentric muscle actions reflecting positive work), dissipation (i.e.. eccentric muscle actions reflecting negative work), and transfer components by comparison of the instantaneous moment powers for adjacent segments (Quanbury et al.. 1975; Robertson and Winter, 1980; Chapman and Caldwell. 1983). From these calculations, the mechanical power approach provides a means of considering the sources of the mechanical work done on the segments and the relative contributions of the musculature and neighboring segments to the work done. Necessary kinematic data describing lower extremity and trunk segment motions were derived from the filtered coordinate data using finite difference equations. Standard link segment mechanics using an inverse dynamics approach were then used to generate values of the ankle, knee, and hip joint reaction forces and moments (Bresler and Frankel, 1950; Winter, 1979). Segment inertial characteristics needed for these calculations were estimated using data and procedures described by Clauser et al. (1969) and Hanavan (1964). To quantify the net energy changes of the foot, shank, and thigh, the contributions of the reaction forces and moments at both the proximal and distal ends of the segments were utilized. Since these data were initially calculated as power values, they were numerically integrated with respect to time using
Segment interactions during running
trapezoidal integration (Cheney and Kincaid. 1980) to derive energy data associated with the generation, dissipation. and transfer functions. Changes in these energy measures were then quantified in order to determine the mechanical work attributed to each of these functions. Once the various mechanical power and energy calculations were completed. these data were normalized with respect to time so that group mean responsescould be computed. All data presented in the figures that characterize the results of the analysis reflect mean data calculated from two cycles for each of the 15 subjects. Data for the mechanical work attributed to thejoint reaction forces and moments under the live load conditions were also analyzed statistically using analysis of variance with repeated measures (Games et al., 1979) to more objectively assess the influence of segment loading. Necessary post hoc analyses were conducted using the Tukey WSD test.
0
R
2501
so nhmtx)
100
80
,
0 207
-150
THGH
I -02
-
1
I 42
‘““l----l 241
-127 I
SHANK I
-193
I
I
-2OOJ
RFSULTS
condition
A necessary prerequisite to evaluating changes in joint kinetics and segment energctics as a consequence of lower extremity loading is a somewhat dctailcd consideration of thcsc characteristics when no load was added to the leg. Figures l-5 summarize power and energy patterns for the lower extremity for the unloaded condition. As rcflectcd in Fig. I. all three lower extremity segments experienced increases in mechanical energy level during the early portion of the swing phase (thigh, 19 J; shank, 44 J; foot, 41 J). With the exception of a brief period during midswing when a trivial increase in thigh energy (2 J) occurred, positive work on the thigh was completed in the first 20% of the swing. The mechanical energy increases for the shank and foot. however, occurred over greater proportions of the swing phase (40% and 68%). As a consequence,the average rates at which positive work was done on the shank (241 W) and thigh (207 W) were comparable (Fig. 1B). The foot, on the other hand, displayed a more gradual increase in mechanical energy and thus a lower net average power (138 W). In the remaining portion of the swing phase, all three segments experienced decreases in energy level (thigh. 32 J. shank. 51 J; foot, 45 J). Temporally, the initiation of negative work was reflected first in the thigh and was followed sequentially by energy losses for the shank and foot. Consequently, the average negative power level of the foot was the highest of the three segments (-312 W vs - 127 and - 193 W for the thigh and shank) even though its energy loss was comparable to that of the shank. From Figs 2-4. it is apparent that energy transfers between adjacent segments attributed to the joint reaction forces dominated the energetics of the lower extremity segments. The energy components for the
40
20
0
Unloaded
531
~~~~~
0
40
20
&I
TIME
80
100
(X)
40 30 320 G 5
10
Ed 0 -10
204 0
.
1
20
-
1
JO
.
,
60
.
,
80
.
(
100
nua tx) Fig. I. The net instantaneous power (A), the average power for phases in which energy was either being gained or lost (B). and the resulting energy changes(C) for the foot. shank, and thigh are shown with respect to time expressed as a pereentage of mean swing time (436ms). The presentation of average power in(B) tends to simplify the somewhat complicated patterns of instantaneous power seen in(A). Also included in (A) are diagrams depicting the position of the leg for one subject at various stages of the swing phase.
hip, knee, and ankle joint reaction forces displayed accumulated energy transfers between segments of 368 J during the swing phase, whereas the combined positive and negative work attributed to the net joint moments was only 134 J. Analysis of the net moments also revealed that I I2 J of the 134 J were linked to generation and dissipation functions, primarily about the hip and knee, while the remaining 22 J were energy
532
P. E. MARTIN and P. R. CAvAhAGn
A600
HPJRF
-e +
1
HPM
-I
0
.
1
20
.
1
KNEEJRF KNEEM ANKLE JRF
-
40
I
64
.
I
.
1
60
100
60
100
nME (~1
75 SO 2
25
5 6 3
-25
0 w
E
0
-50 43
-75
.1001 0
-50 20
40
60
60
100
TIME (X)
0
20
40
60
nME (Xi
Fig. 2. lnstantancous power (A) and energy changes (B) for the hip and knee reaction forces and net moments as they allictcd the cncrgctics of the thigh during the swing phase. In the early portion of swing. the hip reaction force and net moment pcrformcd positive work on the thigh while both the knee reaction lorcc and moment acted to dccrcasc scgmcnt. energy. Near the end of the swing. the hip and ktxr reaction lorccs ruverscd their roles while the hip and knee moments continued their gcncration and dissipation roles. rcspcctivcly.
Fig. 3. lnstantancous power (A) and energy changes (B) for the kna and ankle reaction forrxs and net moments as they atTcctcdthe cncrgctics of the shank during the swing phase. The kntzz reaction force made the largest contribution to the cncrgctics ol’thc shank, transferring energy into the shank in the early swing and out of the shank near the end of swing The ankle reaction force showed the opposite trend. ftnt transferring energy out of the shank (early)and then into the shank (late). While the ankle moment made little contribution to shank cncrgctics, the knrv moment served to dissipate energy from the shank throughout the swing.
transfers via the moments. Figure 5 summarizes schematically the energetic llux for the entire leg during the early and latter portions of the swing phase and serves to emphasize the relative contributions of distal and proximal energy transfers attributed to the joint reaction forces and the generation and dissipation of energy attributed to the net joint moments.
on the feet, the net moment at the knee and the mechanical power attributed to the net moment increased relative to the unloaded condition. The average per cent increasesin peak maximum and minimum values when 0.50 kg was added to the foot were 17.7% for the knee moment and IS. 1% for the knee moment power. Similar increasesin quantified kinetic variables also occurred at the ankle and hip when the feet were loaded. When load was added to the thighs, however, only the magnitudes of the kinetic parameters of the hip were affected and in a much smaller and lessconsistent manner. In contrast to these kinetic changes, modifications in lower extremity kinematics were considerably smaller. For example, the difference in average peak shank angular velocity between the load conditions represented in Fig. 6 was just 3.0%. Table 1, which summarizes the total energy transfers due to the hip, knee, and ankle reaction forces and the mechanical work attributed to the net moments during the swing phase for the five load conditions, illustrates further the kinetic modifications that resulted from inertial changes of the lower extremity.
Segment loading influences
When load was added to the lower extremities. systematic changes in joint kinetics and associated mechanical work functions were observed. As expected, these changes were more prominent for the more distally positioned loads and as load magnitude increased. In addition, modifications in joint kinetics were generally limited to the joints proximal to the added load. To provide an example of the effect of loading on joint kinetics and kinematics, Fig. 6 displays the net moment about the knee joint, shank angular velocity, and instantaneous power of the knee moment for two load conditions (no added load and 0.50 kg added to the foot). When load was positioned
Segment interactions during running
EARLY
.lo! 0
.,-,-,.,., 20 40
60
80
100
ns4Etsb) Fig. 4. lnstantancous power (A) and energy changes(B) for the ankle reaction force and net moment as they alTcctcdthe encrgctics of the foot during the swing phase. Energy transfer bctwccn the shank and foot attributed to the ankle reaction force accounted for nearly all of the positive and negative work done on the foot during the swing.
When analyzed statistically, there were no significant changes in the energy transfer and work values due to the addition of load to the thigh, even though there was a consistent trend for energy transfer attributed to the hip reaction force to increase as the load was increased. All variables showed significant increases, however, when load was added to the feet. Energy transfer and work measures for the reaction forces and moments relative to the unloaded condition increased an average of 9. I % for the 0.25 kg load condition and 19.7% when 0.50 kg was carried on the foot.
Reaction
force
and moment
contributions
Results showed that the majority of the energy changes within the lower extremity segments were attributed to intersegment energy transfers associated with the joint reaction forces and that there was a clear pattern of energy transfer distally in the early portion of the swing phase and proximally later in the swing (see Fig 5). Rather than acting as a source of energy generation or a sink for dissipation, the joint reaction forces appear to redistribute the mechanical energy generated from other sources (Aleshinsky, 1986; Rob-
533
LATE
Fig. 5. Schematic summary ol the contributions of the hip. kntz, and ankle joint reaction forccs (bold arrows) and net moments (small arrows) to segment energy fluctuations during early and late portions of the swing phase. All data are expressed in joules. These data show that intersegment interactions reprcscntcd by the joint reaction forces dominated the cncrgy flux of the leg during the swing phase. These intcrscgmenl interactions resuhed in the transfer of mozhanical energy distally during the early portion of the swing and proximally in the later stages of the swing. Dcspitc their smaller magnitude. significant energy gcncration and dissipation contributions were also made by the hip and knee musculature. rcspcctivcly.
crtson and Winter. 1980; Winter, 1979; Winter and Robertson, 1978). Through eccentric and concentric actions, as retlected by the net moments, the musculature surrounding the joints served more modest roles in generating and dissipating energy during the course of the swing phase. When considered in total, these results are consistent with previous research suggesting that segment motions of the swing leg are generated and controlled proximally, particularly by musculature about the hip and by the interaction of the thigh with the trunk (Chapman and Caldwell. 1983; Elliott, 1977; Phillips et al.. 1983). That is, a sufficient amount of mechanical energy was input into the lower extremity via the mechanisms of the hip joint reaction force (68 J) and net moment (36 J) in the early portion of the swing to account for the increases in mechanical energy of all three segments. With respect to muscular contributions, represented mechanically by the net moments about the joints and the resulting mechanical powers attributed to them, only the hip musculature played a significant role in generating mechanical energy to the lower extremity. Musculature about the knee, as represented by the net moment, served primarily to dissipate mechanical energy, while the ankle moment made little contribution. The mechanical energy input into the system via these mechanisms about the hip was subsequently redistributed first distally via the knee and ankle reaction forces and then proximally in
534
P. c
MARTIN
and P. R. CAVANAGH
-*MO’
W-
O.SOKGFUJT --m
(4 Fuixm
-20-
(-1 EXTENSOR
-40-r
, 0
I
-
20
,
.
40
,
.
60
,
.
00
( 100
nME (K)
-
OSOKGFDDT UNLOADED
E O. 0: -loog B
mo400-
’
400-r 0
0.SOKGFOOT --IJNBED -
I 20
-
I 40
.
I
so
.
1 80
.
1 100
n04E(~1 Fig. 6. Comparison of the ncl knee moment(A). shank angular velocity(B). and knee moment power(C) during the swing phase for unloaded running and running with 0.50 kg add4 to each foot. Results demonstrated consistentmodilications in joint kinetic parameters proximal to the location where load was added but little change in segment kinematics.
the later stages of the swing phase as the energy levels of the segments declined. Response to inertial
property
manipulation
In considering the efiect of lower extremity loading on the kinematics of the swing leg, it was anticipated that the increased inertia due to the loading would produce modifications in the lower extremity movement patterns. The results of this analysis and results reported previously (Martin, 1985). however, dem-
onstrated that only modest adjustments occurred in the kinematics of the gait pattern when the lower extremity was loaded. For example, no statistically significant changes in single leg support time and Right time were observed, while the difference in step length between the unloaded condition and that condition in which 0.50 kg was added to each foot was 1.4 cm. Similarly, changes in swing time, peak vertical position of the ankle joint above the ground, peak velocity of the ankle, and the vertical displacement of the hip joint were small (+9 ms, +0.8 cm, -0.23 m s- ‘, and +0.6cm, respectively) when load was added to the feet (Martin. 1985). From a motor control perspective, the resultsof our analysis suggest that the neuromuscular system responded to the modifications in lower extremity inertial characteristics by increasing muscular output while maintaining swing phase kinematics. The observed increases in net joint moments and reaction forces in the absence ofsubstantial changes in kinematics are consistent with results reported by Lestienne (1979) for a simple elbow tlexion-extension task under different velocity and inertial load conditions. He found that the duration of agonist electromyographic activity and the timing of the onset of antagonist activity were related to movement velocity but were independent of inertial load. Under similar velocity conditions, results showed that the level of excitation of the musculature increased with increases in the inertial load so as to maintain the timing of agonist and antagonist activity. Even though our mechanical results are in apparent agreement with Lestiennc’s results, our interpretation of increased muscular output must be viewed with caution in the absence of electromyographic support since the swing phase of running is obviously a considerably more complex movement than the single joint task studied by Lestienne. The apparent kinematic invariance within the range of inertial manipulation used in this study may also be compatible with a mathematical theory of movement coordination which suggests that many voluntary movements are produced so as to maximize the smoothness of the motion (Flash and Hogan, 1985; Hogan, 1984; Hogan et al.. 1987). According to this theory, the movement for which smoothness is maximized is one in which the mean squared magnitude of the time derivative of acceleration (jerk) is minimized. While not tested directly in the present study, the results raise the possibility that the neuromuscular system may be attempting to maintain smoothness of the motion of the swing leg despite modifications in the inertia of the leg. The observed increases in the net moments and their associated work contributions as higher and more distally positioned loads were applied to the leg are suggestive of increased contributions by lower extremity musculature and therefore are consistent with the increased aerobic demand that has been reported consistently for conditions in which lower
Segment interactions during running
535
Table 1. Mean values ( f S.D.) for energy transfer measures for the joint reaction forces (JRF ) and for the work attributed to the hip and knee moments (.Vf) under the five load conditions. Load conditions
Baseline
0.25 kg. thigh
0.50 kg. thigh
0.25 kg, root
0.50 kg. foot
Hip JRF (energy transfer)
IS’.3 (24.6)
153.3 (23.6)
158.4 (28.7)
158.9. (24.7)
166.8. (27.6)
Knee IRF (energy transfer)
133.1 (15.6)
131.8 (15.0)
130.8 (16.2)
139.5.
(15.5)
150.0. (18.8)
Ankle JRF (energy transfer)
82.7 (8.0)
82.9 (8.6)
82.1 (9.4)
93.6, (8.6)
104.9. (10.0)
Hip bf (work on the thigh)
52.6 (8.9)
51.2 (8.2)
52. I (8.9)
56.9. (9.1)
63.1. (11.4)
Knee &f (work on the thigh)
21.1 (4.3)
20.6 (4.2)
20.7 (4.1)
23.9. (4.4)
27.?* (5.4)
Knee ,M (work on the shank)
44.6 (5.4)
44.2 (5.3)
44.3 (6.4)
47.7’ (5.5)
51.7’ (6.7)
Ankle &f (work on the shank)
(Z,
7.0 (1.4)
7.0 (1.5)
7.8’ (1.4)
8.5* (1.7)
Ankle hf (work on the foot)
8.6 (1.8)
8.6 (2.0)
8.6 (2.1)
9.5’ (2.0)
IO.50 (2.4)
Variable
The energytransferand work valuesrepresentthe sum of the ahsolutc values for positive and negative energy changes. All values arc expreaed in joules. l Significantly different from the baseline condition.
extremity inertia has been increased (e.g., Catlin and Dressendorfer, 1979; Claremont and Hall. 1988; Martin, 1985; Myers and Steudel. 1985). An interesting implication of these results pertains to the possible effect of individual ditTercncesin anthropometric characteristics of the body, particularly the inertial characteristics of the lower extremity segments, on the economy of walking and running. There is little doubt that the explanation for individual differences in the economy of motion is a complicated one that involves a complex interaction of a variety of structural, biomechanical, and physiological factors (Cavanagh and Kram, 1985; Frederick, 1985). Nevertheless, ifall other factors were the same, results from the present study support the notion that a runner with a proportionately smaller amount of total body mass concentrated in the lower extremities would require lesseffort to generate the movements of the legsdue to the lower inertia of the segments (Cavanagh and Kram, 1985; Myers and Steudel. 1985). While it would be expected that a proportionately higher trunk mass would require an increased muscular effort, research has shown that the cost of supporting, accelerating. and decelerating mass located centrally on the body is considerably lower than that for mass carried on the extremities (Inman et al., 1981; Myers and Steudel. 1985: Ralston and Lukin. 1969). Despite the economy and motor control implications raised in the preceding discussion, there are a number of issues that the present study was not designed to address. Specifically, electromyographic data would provide a clearer indication of the response of the neuromuscular system to the modilic-
ations in lower extremity inertia and the contributions of individual muscics to the swing phase. Of particular interest are the contributions of the biarticular muscles spanning the joints of the lower extremity during unloaded and loaded running because of suggested implications for aerobic demand. From computations of net joint moment power during running, Elftman (1940) speculated that the biarticular architecture within the lower extremity reduces the muscular work required during running in comparison to a hypothetical situation in which only monoarticular muscles contribute. He further indicated that this reduction in mechanical work is reflected in a saving of energy expenditure. Finally, it would be of interest to evaluate the applicability of the kinematic-based maximum smoothness principle (Flash and ffogan, 1985; Hogan, 1984; Hogan et al., 1987) to the motions of the gait cycle and to consider the compatibility of this principle with the concept that movement patterns are developed so as to minimize aerobic demand (Cavanagh and Kram, 1985). The lack of available information on these and other issues related to extremity loading suggests the need for further research on this topic. REFERENCFS
Aleshinsky. S. Y. (1986) An energy ‘sources’ and ‘fractions’ approach to the mechanical energy expenditure problem-Ii. Movement of the multi-link chain model. J. Biomechunio 19. 295-300. Bresler. B. and Frankel. J. P. (1950) The forces and moments in the leg during level walking. Frans. ASME 72, 27-36. Catlin. M. J. and Dressendorfer. R. H. (1979) ElTect of shoe
536
P. E.‘~*.RTIN and P. R. CAYANAGH
weight on the energy cost of running. Med. Sci. Sports Exert. 11.80. Cavanagh. P. R. and Kram. R. (1985) Mechanical and muscular factors affecting the efficiency of human movement. Med. Sci. Sports Exert. 17. 32633 1. Chapman. A. E. and Caldwell, G. E. (1983) Factors determining changes in lower limb energy during swing in treadmill running. J. &mechanics 16. 69-77. Cheney. W. and Kincaid. D. (1980) Numerical Mathematics and Computing. pp. 6670. Brooks/Cole, Monterey, CA. Claremont. A. D. and Hall. S. J. (1988) Effects of extremity loading upon energy expenditure and running mechanics. Med. Sci. S~rls Exrrc. 20. 167-l 71. Clauser. C. El. McConville. j. ‘I: and Young. J. W. (1969) IC’piyht, Volume and Center or Mass of Segments of fhe Human Body (AMRL-TR-69-70). Wright Patterson Air Force Base. Ohio. Elftman. H. (1939a) Forces and energy changes in the leg during walking Am. 1. Physiol. 125. 339-356. Elftman. H. (1939b) The function of muscles in locomotion. Am. J. Physiol. 12S. 357-366. Elftman. H. (194O)The work done by muscles in running. Am. 1. Physiol. 129. 672-684. Elliott. B. C. (1977) The co-ordination of force summation in the lower limb during the recovery phase of treadmill running. J. Human Movemenr Srud. 3. 82-87. Flash. T. and Hogan. N. (1985) The coordination of arm movements: an ~experimentally confirmed mathematical model. J. Neurnsci. 5. 1688-t 703. Frederick. E. C. (1985) Synthesis. cnpcrimcntation. and the biomcchanics of economical movcmcnt. Med. Sri. Sports Exrrc. 17.44-47. Games. P. A., Gray. G. S., Herron. W. L., Pcntz. A. and Sinatra, R. (1979) Anulysis cg Variance with Repeated Meusures. The Pennsylvania State University Computation Center, University Park. PA. Hanavan. E. (1964) A Murhemuficul Model 01 rhr llumun Eody (AMRL-TR-64-102). Wright Patterson Air Force Base, Ohio. Hogan. N. (1984) An organizing principk for a class of
voluntary movements. J. Neurosci. 4 2745-2754. Hogan. N, Bizi. E.. Mussa-lnvaldii F. A. and Flash. T. (1987)Controlling multijoint motor behavior. Exert. Sport Sci. Reu. lS, 153-190. Inman. V. T.. Ralston. H. J. and Todd. F. (1981) Human Walkiny. p. 71. Williams and Wilkins. Baltimore, MD. Lestiennc. F. (1979) Effects of inertial load and velocity on the braking process of voluntary limb movements. Exp. Brain Res. 3s. 407418. Martin, P. E. (1985) Mechanical and physiological responses to lower extremity loading during running. Med. Sci. Sports Exert. 17, 427-433. Mena D.. Mansour. J. M. and Simon. S. R. I198 I1 Analvsis and synthesis of human swing leg motion during gait and its clinical applications. 1. Biomrchanics 14.823-832. Mochon. S. and McMahon. T. A. (1980) Ballistic walking. J. Biomechanics 13.49-57. Myers, M. J. and Steudel, K. (1985) Effect oflimb mass and its distribution on the energetic cost of running. . J. exp. Biol. 116.363-373. Phillips, S. J.. Roberts, E. M. and Huang, T. C. (1983) Quantification of intersegmental reactions during rapid swing motion. J. Biomechanics 16.41 l-417. Quanbury. A. 0.. Winter, D. A. and Reimer. G. D. (1975) Instantaneous power and power flow in body segments during walking. J. Human Mouemenf Srud. 1. 59-67. Ralston. H. J. and Lukin. L. (1969) Energy levels of human body segments during walking. Ergon&ics 12, 3946. Robertson. D. G. E. and Winter. D. A. 11980) Mechanical energy generation, absorbtion, and transfer gmongst segments during walking. J. Biomechanics 13. 845-854. Wells, R. P. and Winter. D. A. (1980) Assessment of signal and noise in the kinematics of normal. pathological, and sporting gaits. Humun Lucomorion I, pp. 92-93. London, Ontario, Canada. Winter. D. A. (1979) Biomeehunics of ffuman Movement, pp. 84-107. John Wiley. New York. Winter, D. A. and Robertson. D. G. E. (1978) Joint torque and energy patterns in normal gait. Biol. Cykn. 29, 137-142.
R. CAVANAGH:
and Sport Research Institute. Arizona State University. Tempe, AZ 85287, U.S.A. and Center for Locomotion Studies. Penn. State University , University Park. PA 16802. U.S.A.
tExercise
:The
Abstract-In this study, designed to determine the effect of lower extremity inertia manipulation on joint kinetics and segment energetics during the swing phase, I5 male distance runners were filmed as they performed treadmill running (3.35 m s-l) under five load conditions: no added load and loads oCO.25 kg and 0.50 kg added to each thigh or each foot. Results of this study demonstrated that the energetics of the lower extremity movements during the swing phase of the running cycle were dominated by mechanical energy transfers between adjacent segments attributed to the joint reaction forces. which acted to redistribute mechanical energy within the system. These contributions were considerably greater than those of the net joint moments. which primarily retlccted muscular generation and dissipation of mechanical energy. Lower extremity loading caused little change in the movement pattern of the swing leg. However, increases in the joint reaction forces and net moments and in the amount of work done and the energy transfer attributed to the reaction forces and moments were observed, but were limited to thejoints proximal to the location of the added load. These results were consistent with the increased aerobic demand associated with increases in lower extremity inertia that have been reported elsewhere and also have implications for the manner
in which the nellromuscular system controls the motion of the legs during running.
INTWODUCTION
Despite the apparent simplicity of human gait and the ease with which most individuals are able to initiate, sustain, and modulate walking and running patterns, there is little doubt that the control of individual body segments during these activities represents a highly complex neuromuscular ellbrt. Segment motions are orchestrated not only by grAvitational and muscle forces acting upon the segments, but also by forces from interactions between adjacent segments. The swing phase of the gait cycle has been the focus of numerous analyses (e.g.. Chapman and Caldwell. 1983; Elliott, 1977; Mena er al.. 1981; Mochon and McMahon. 1980; Phillips et ol.. 1983) and provides an interesting motion for the consideration of interactions between segments and the coordination of their movements. While numerous biomechanical features of the swing phase have been studied previously, there has been little consideration the inertial
properties
influence lower extremity
of how manipulation
of the segments
of
of the leg
dynamics during this period.
researchers have considered the physiolgical consequencesof carrying added load on the lower ex&emities and have consistently shown higher aerobic demands under loaded conditions (e.g., Catlin and Dressendorfer, 1979; Claremont and Hall. 1988; Martin, 1985; Myers and Steudel, 1985). It may be expected that increasesin aerobic demand due to increased lower extremity inertia would be accompanied by biomechanical changes, particularly in joint kinetics and segment energetics within the In contrast
Receiwd l
to this, numerous
in jnol jorm I 5 A ugusr 1989. To whom correspondence should be addressed.
lower extremity. This, however, has not been evaluated previously. Consequently, the purpose of our study was to determine the etTectof lower extremity inertia manipulation on these kinetic features during the swing phase. It was believed that results of this study could be important in understanding factors that influence the aerobic demand or economy of running, specifically those related to the increased demands under loaded conditions. PROCF.DURW Fifteen healthy male long-distance
runners partici-
pated in three test sessionswhich involved running on
a motor driven treadmill. Mean subject age was 29.3 yr (range =20-47) and mean body mass was 72.0 kg (range 3: 54.7-90.3). The iirst session was used to habituate the subjects to the treadmill and to introduce them to the loading procedures prior to data collection. Each subject performed three IO-min bouts of treadmill running separated by IO-min rest intervals at the experimental speed of 3.35 ms-‘. Those subjects who had no previous experience on a treadmill comple!ed an additional 30 min of treadmill training prior to data collection. Each subject then completed two identical experimental sessions in which his response to the loading conditions was measured. Five load conditions were used such that the eflects of both load magnitude and position could be examined. These included: (a) an unloaded condition, (b) 0.25 kg added to each thigh, (c) 0.25 kg added to each foot, (d) 0.50 kg added to each thigh, and (e)O.SOkg added to each foot. The loads were added to the leg in the form of small packets of lead shot. Running shoes with elastic pockets sewn onto the medial and lateral aspects of
530
P. E. MARTINand P. R. CAVANAGH
each shoe and elastic bicycling shorts with pockets sewn onto the medial and lateral sides of the distal aspect of each leg were provided for the subjects and served to stabilize the loads on the segments. Primarily because of differences in anthropometric characteristics between subjects, it was not possible to position the loads on the thighs at the same relative location (i.e., same percentage of thigh length) for all subjects. Since the goal of the load manipulation was simply to produce distal and proximal loading of the leg, the inability to precisely control the positioning of the thigh loads was not considered a major limitation. The location of the thigh loads was quantified for each subject as part of the motion analysis process. The mean location for these loads was approximately 68% of thigh length from the hip joint center (range = 59-80%). An identical testing protocol was used on both experimental days, which were separated by a minimum of I day and a maximum of 3 days of rest. On both test days, each subject first completed a 3-min warm-up run on the treadmill. This was then followed by alternating periods of I2 min of rest and 8 mitt of running at 3.35 m s- ’ until the five randomly ordered load conditions were completed. The 8-min running periods were used in an effort to obtain steady-state conditions of the runners both physiologically and mechanically prior to actual data collection. Standard two-dimensional high-speed cinematographical procedures were used to record a sagittal plane view of the running motions of the subjects under each of the five load conditions. A single Locam camera operating at a nominal ftXIlle rate of 100 frames s- ’ was used to tilm the subjects near the end of each I(-min trial. Camera speed was calibrated using an elcvtronic timing unit positioned in the tield of view of the camera. For each subject, the swing phase for one stride under each load condition was analyzed from each day of testing. Data from the two test strides were pooled to decrease the effect of potentially spurious anomalous results from a single test day on group trends. Coordinate data representing the spatial locations of toe, heel, ankle, knee, hip, and inferior aspect of the ear on the left side of the body were generated from film analysis and served to define the segment endpoints of the foot, shank, thigh, and trunk. Even though the analysis focused on the lower extremity, quantification of the angular position of the trunk was necessaryto fully describe the interaction of the thigh with the trunk. In addition. it was assumed that with limited motion of the head, the inferior aspect of the ear would be a more suitable superior endpoint of the trunk than the shoulder, which can move considerably due to shoulder and trunk rotation. Coordinate data were scaled and digitally filtered using a fourth-order zero-lag Butterworth filter. Selection of cut-otf frequencies was based on an analysis using a residual procedure described by Wells and Winter (1980). Cutoffvalues ranged from 3 Hz for the horizontal coordi-
nate data of the ear to 8 Hz for the coordinate data of the toe and heel. So that data from this study could be compared with those of overground analyses, the horizontal coordinate data were also adjusted for mean treadmill belt displacement in the measurement interval. Values for a series of mechanical power, work. and energy variables were then quantified from the coordinate data using a mechanical power model introduced by Elftman (1939a. b) and later clarified and extended by Quanbury et al. (1975) and Robertson and Winter (1980). These variables included the instantaneous power attributed to the joint reaction forces (PF) and moments (PM) at the ankle, knee, and hip and their associated mechanical work contributions to the foot, shank, and thigh. These were computed at each joint using the following equations: PF,, = F, * <, PM,, = Ml,
*ik
(1) (2)
where j, F, V, rcI and w represent the joint of interest. joint reaction force, joint center velocity, net joint moment, and angular velocity of segment k (either distal or proximal to joint j) at instant i. The role of the joint reaction force has been described as a simple mechanism of energy transfer since the rate at which mechanical energy is lost by one segment is equivalent to the rate at which energy is gained by its neighbor (Alcshinsky, 1986; Robertson and Winter, 1980: Winter. 1979). Unlike the simple transfer function of the joint reaction force, the contributions of the net joint moments were further subdivided into mechanical energy generation (i.e., concentric muscle actions reflecting positive work), dissipation (i.e.. eccentric muscle actions reflecting negative work), and transfer components by comparison of the instantaneous moment powers for adjacent segments (Quanbury et al.. 1975; Robertson and Winter, 1980; Chapman and Caldwell. 1983). From these calculations, the mechanical power approach provides a means of considering the sources of the mechanical work done on the segments and the relative contributions of the musculature and neighboring segments to the work done. Necessary kinematic data describing lower extremity and trunk segment motions were derived from the filtered coordinate data using finite difference equations. Standard link segment mechanics using an inverse dynamics approach were then used to generate values of the ankle, knee, and hip joint reaction forces and moments (Bresler and Frankel, 1950; Winter, 1979). Segment inertial characteristics needed for these calculations were estimated using data and procedures described by Clauser et al. (1969) and Hanavan (1964). To quantify the net energy changes of the foot, shank, and thigh, the contributions of the reaction forces and moments at both the proximal and distal ends of the segments were utilized. Since these data were initially calculated as power values, they were numerically integrated with respect to time using
Segment interactions during running
trapezoidal integration (Cheney and Kincaid. 1980) to derive energy data associated with the generation, dissipation. and transfer functions. Changes in these energy measures were then quantified in order to determine the mechanical work attributed to each of these functions. Once the various mechanical power and energy calculations were completed. these data were normalized with respect to time so that group mean responsescould be computed. All data presented in the figures that characterize the results of the analysis reflect mean data calculated from two cycles for each of the 15 subjects. Data for the mechanical work attributed to thejoint reaction forces and moments under the live load conditions were also analyzed statistically using analysis of variance with repeated measures (Games et al., 1979) to more objectively assess the influence of segment loading. Necessary post hoc analyses were conducted using the Tukey WSD test.
0
R
2501
so nhmtx)
100
80
,
0 207
-150
THGH
I -02
-
1
I 42
‘““l----l 241
-127 I
SHANK I
-193
I
I
-2OOJ
RFSULTS
condition
A necessary prerequisite to evaluating changes in joint kinetics and segment energctics as a consequence of lower extremity loading is a somewhat dctailcd consideration of thcsc characteristics when no load was added to the leg. Figures l-5 summarize power and energy patterns for the lower extremity for the unloaded condition. As rcflectcd in Fig. I. all three lower extremity segments experienced increases in mechanical energy level during the early portion of the swing phase (thigh, 19 J; shank, 44 J; foot, 41 J). With the exception of a brief period during midswing when a trivial increase in thigh energy (2 J) occurred, positive work on the thigh was completed in the first 20% of the swing. The mechanical energy increases for the shank and foot. however, occurred over greater proportions of the swing phase (40% and 68%). As a consequence,the average rates at which positive work was done on the shank (241 W) and thigh (207 W) were comparable (Fig. 1B). The foot, on the other hand, displayed a more gradual increase in mechanical energy and thus a lower net average power (138 W). In the remaining portion of the swing phase, all three segments experienced decreases in energy level (thigh. 32 J. shank. 51 J; foot, 45 J). Temporally, the initiation of negative work was reflected first in the thigh and was followed sequentially by energy losses for the shank and foot. Consequently, the average negative power level of the foot was the highest of the three segments (-312 W vs - 127 and - 193 W for the thigh and shank) even though its energy loss was comparable to that of the shank. From Figs 2-4. it is apparent that energy transfers between adjacent segments attributed to the joint reaction forces dominated the energetics of the lower extremity segments. The energy components for the
40
20
0
Unloaded
531
~~~~~
0
40
20
&I
TIME
80
100
(X)
40 30 320 G 5
10
Ed 0 -10
204 0
.
1
20
-
1
JO
.
,
60
.
,
80
.
(
100
nua tx) Fig. I. The net instantaneous power (A), the average power for phases in which energy was either being gained or lost (B). and the resulting energy changes(C) for the foot. shank, and thigh are shown with respect to time expressed as a pereentage of mean swing time (436ms). The presentation of average power in(B) tends to simplify the somewhat complicated patterns of instantaneous power seen in(A). Also included in (A) are diagrams depicting the position of the leg for one subject at various stages of the swing phase.
hip, knee, and ankle joint reaction forces displayed accumulated energy transfers between segments of 368 J during the swing phase, whereas the combined positive and negative work attributed to the net joint moments was only 134 J. Analysis of the net moments also revealed that I I2 J of the 134 J were linked to generation and dissipation functions, primarily about the hip and knee, while the remaining 22 J were energy
532
P. E. MARTIN and P. R. CAvAhAGn
A600
HPJRF
-e +
1
HPM
-I
0
.
1
20
.
1
KNEEJRF KNEEM ANKLE JRF
-
40
I
64
.
I
.
1
60
100
60
100
nME (~1
75 SO 2
25
5 6 3
-25
0 w
E
0
-50 43
-75
.1001 0
-50 20
40
60
60
100
TIME (X)
0
20
40
60
nME (Xi
Fig. 2. lnstantancous power (A) and energy changes (B) for the hip and knee reaction forces and net moments as they allictcd the cncrgctics of the thigh during the swing phase. In the early portion of swing. the hip reaction force and net moment pcrformcd positive work on the thigh while both the knee reaction lorcc and moment acted to dccrcasc scgmcnt. energy. Near the end of the swing. the hip and ktxr reaction lorccs ruverscd their roles while the hip and knee moments continued their gcncration and dissipation roles. rcspcctivcly.
Fig. 3. lnstantancous power (A) and energy changes (B) for the kna and ankle reaction forrxs and net moments as they atTcctcdthe cncrgctics of the shank during the swing phase. The kntzz reaction force made the largest contribution to the cncrgctics ol’thc shank, transferring energy into the shank in the early swing and out of the shank near the end of swing The ankle reaction force showed the opposite trend. ftnt transferring energy out of the shank (early)and then into the shank (late). While the ankle moment made little contribution to shank cncrgctics, the knrv moment served to dissipate energy from the shank throughout the swing.
transfers via the moments. Figure 5 summarizes schematically the energetic llux for the entire leg during the early and latter portions of the swing phase and serves to emphasize the relative contributions of distal and proximal energy transfers attributed to the joint reaction forces and the generation and dissipation of energy attributed to the net joint moments.
on the feet, the net moment at the knee and the mechanical power attributed to the net moment increased relative to the unloaded condition. The average per cent increasesin peak maximum and minimum values when 0.50 kg was added to the foot were 17.7% for the knee moment and IS. 1% for the knee moment power. Similar increasesin quantified kinetic variables also occurred at the ankle and hip when the feet were loaded. When load was added to the thighs, however, only the magnitudes of the kinetic parameters of the hip were affected and in a much smaller and lessconsistent manner. In contrast to these kinetic changes, modifications in lower extremity kinematics were considerably smaller. For example, the difference in average peak shank angular velocity between the load conditions represented in Fig. 6 was just 3.0%. Table 1, which summarizes the total energy transfers due to the hip, knee, and ankle reaction forces and the mechanical work attributed to the net moments during the swing phase for the five load conditions, illustrates further the kinetic modifications that resulted from inertial changes of the lower extremity.
Segment loading influences
When load was added to the lower extremities. systematic changes in joint kinetics and associated mechanical work functions were observed. As expected, these changes were more prominent for the more distally positioned loads and as load magnitude increased. In addition, modifications in joint kinetics were generally limited to the joints proximal to the added load. To provide an example of the effect of loading on joint kinetics and kinematics, Fig. 6 displays the net moment about the knee joint, shank angular velocity, and instantaneous power of the knee moment for two load conditions (no added load and 0.50 kg added to the foot). When load was positioned
Segment interactions during running
EARLY
.lo! 0
.,-,-,.,., 20 40
60
80
100
ns4Etsb) Fig. 4. lnstantancous power (A) and energy changes(B) for the ankle reaction force and net moment as they alTcctcdthe encrgctics of the foot during the swing phase. Energy transfer bctwccn the shank and foot attributed to the ankle reaction force accounted for nearly all of the positive and negative work done on the foot during the swing.
When analyzed statistically, there were no significant changes in the energy transfer and work values due to the addition of load to the thigh, even though there was a consistent trend for energy transfer attributed to the hip reaction force to increase as the load was increased. All variables showed significant increases, however, when load was added to the feet. Energy transfer and work measures for the reaction forces and moments relative to the unloaded condition increased an average of 9. I % for the 0.25 kg load condition and 19.7% when 0.50 kg was carried on the foot.
Reaction
force
and moment
contributions
Results showed that the majority of the energy changes within the lower extremity segments were attributed to intersegment energy transfers associated with the joint reaction forces and that there was a clear pattern of energy transfer distally in the early portion of the swing phase and proximally later in the swing (see Fig 5). Rather than acting as a source of energy generation or a sink for dissipation, the joint reaction forces appear to redistribute the mechanical energy generated from other sources (Aleshinsky, 1986; Rob-
533
LATE
Fig. 5. Schematic summary ol the contributions of the hip. kntz, and ankle joint reaction forccs (bold arrows) and net moments (small arrows) to segment energy fluctuations during early and late portions of the swing phase. All data are expressed in joules. These data show that intersegment interactions reprcscntcd by the joint reaction forces dominated the cncrgy flux of the leg during the swing phase. These intcrscgmenl interactions resuhed in the transfer of mozhanical energy distally during the early portion of the swing and proximally in the later stages of the swing. Dcspitc their smaller magnitude. significant energy gcncration and dissipation contributions were also made by the hip and knee musculature. rcspcctivcly.
crtson and Winter. 1980; Winter, 1979; Winter and Robertson, 1978). Through eccentric and concentric actions, as retlected by the net moments, the musculature surrounding the joints served more modest roles in generating and dissipating energy during the course of the swing phase. When considered in total, these results are consistent with previous research suggesting that segment motions of the swing leg are generated and controlled proximally, particularly by musculature about the hip and by the interaction of the thigh with the trunk (Chapman and Caldwell. 1983; Elliott, 1977; Phillips et al.. 1983). That is, a sufficient amount of mechanical energy was input into the lower extremity via the mechanisms of the hip joint reaction force (68 J) and net moment (36 J) in the early portion of the swing to account for the increases in mechanical energy of all three segments. With respect to muscular contributions, represented mechanically by the net moments about the joints and the resulting mechanical powers attributed to them, only the hip musculature played a significant role in generating mechanical energy to the lower extremity. Musculature about the knee, as represented by the net moment, served primarily to dissipate mechanical energy, while the ankle moment made little contribution. The mechanical energy input into the system via these mechanisms about the hip was subsequently redistributed first distally via the knee and ankle reaction forces and then proximally in
534
P. c
MARTIN
and P. R. CAVANAGH
-*MO’
W-
O.SOKGFUJT --m
(4 Fuixm
-20-
(-1 EXTENSOR
-40-r
, 0
I
-
20
,
.
40
,
.
60
,
.
00
( 100
nME (K)
-
OSOKGFDDT UNLOADED
E O. 0: -loog B
mo400-
’
400-r 0
0.SOKGFOOT --IJNBED -
I 20
-
I 40
.
I
so
.
1 80
.
1 100
n04E(~1 Fig. 6. Comparison of the ncl knee moment(A). shank angular velocity(B). and knee moment power(C) during the swing phase for unloaded running and running with 0.50 kg add4 to each foot. Results demonstrated consistentmodilications in joint kinetic parameters proximal to the location where load was added but little change in segment kinematics.
the later stages of the swing phase as the energy levels of the segments declined. Response to inertial
property
manipulation
In considering the efiect of lower extremity loading on the kinematics of the swing leg, it was anticipated that the increased inertia due to the loading would produce modifications in the lower extremity movement patterns. The results of this analysis and results reported previously (Martin, 1985). however, dem-
onstrated that only modest adjustments occurred in the kinematics of the gait pattern when the lower extremity was loaded. For example, no statistically significant changes in single leg support time and Right time were observed, while the difference in step length between the unloaded condition and that condition in which 0.50 kg was added to each foot was 1.4 cm. Similarly, changes in swing time, peak vertical position of the ankle joint above the ground, peak velocity of the ankle, and the vertical displacement of the hip joint were small (+9 ms, +0.8 cm, -0.23 m s- ‘, and +0.6cm, respectively) when load was added to the feet (Martin. 1985). From a motor control perspective, the resultsof our analysis suggest that the neuromuscular system responded to the modifications in lower extremity inertial characteristics by increasing muscular output while maintaining swing phase kinematics. The observed increases in net joint moments and reaction forces in the absence ofsubstantial changes in kinematics are consistent with results reported by Lestienne (1979) for a simple elbow tlexion-extension task under different velocity and inertial load conditions. He found that the duration of agonist electromyographic activity and the timing of the onset of antagonist activity were related to movement velocity but were independent of inertial load. Under similar velocity conditions, results showed that the level of excitation of the musculature increased with increases in the inertial load so as to maintain the timing of agonist and antagonist activity. Even though our mechanical results are in apparent agreement with Lestiennc’s results, our interpretation of increased muscular output must be viewed with caution in the absence of electromyographic support since the swing phase of running is obviously a considerably more complex movement than the single joint task studied by Lestienne. The apparent kinematic invariance within the range of inertial manipulation used in this study may also be compatible with a mathematical theory of movement coordination which suggests that many voluntary movements are produced so as to maximize the smoothness of the motion (Flash and Hogan, 1985; Hogan, 1984; Hogan et al.. 1987). According to this theory, the movement for which smoothness is maximized is one in which the mean squared magnitude of the time derivative of acceleration (jerk) is minimized. While not tested directly in the present study, the results raise the possibility that the neuromuscular system may be attempting to maintain smoothness of the motion of the swing leg despite modifications in the inertia of the leg. The observed increases in the net moments and their associated work contributions as higher and more distally positioned loads were applied to the leg are suggestive of increased contributions by lower extremity musculature and therefore are consistent with the increased aerobic demand that has been reported consistently for conditions in which lower
Segment interactions during running
535
Table 1. Mean values ( f S.D.) for energy transfer measures for the joint reaction forces (JRF ) and for the work attributed to the hip and knee moments (.Vf) under the five load conditions. Load conditions
Baseline
0.25 kg. thigh
0.50 kg. thigh
0.25 kg, root
0.50 kg. foot
Hip JRF (energy transfer)
IS’.3 (24.6)
153.3 (23.6)
158.4 (28.7)
158.9. (24.7)
166.8. (27.6)
Knee IRF (energy transfer)
133.1 (15.6)
131.8 (15.0)
130.8 (16.2)
139.5.
(15.5)
150.0. (18.8)
Ankle JRF (energy transfer)
82.7 (8.0)
82.9 (8.6)
82.1 (9.4)
93.6, (8.6)
104.9. (10.0)
Hip bf (work on the thigh)
52.6 (8.9)
51.2 (8.2)
52. I (8.9)
56.9. (9.1)
63.1. (11.4)
Knee &f (work on the thigh)
21.1 (4.3)
20.6 (4.2)
20.7 (4.1)
23.9. (4.4)
27.?* (5.4)
Knee ,M (work on the shank)
44.6 (5.4)
44.2 (5.3)
44.3 (6.4)
47.7’ (5.5)
51.7’ (6.7)
Ankle &f (work on the shank)
(Z,
7.0 (1.4)
7.0 (1.5)
7.8’ (1.4)
8.5* (1.7)
Ankle hf (work on the foot)
8.6 (1.8)
8.6 (2.0)
8.6 (2.1)
9.5’ (2.0)
IO.50 (2.4)
Variable
The energytransferand work valuesrepresentthe sum of the ahsolutc values for positive and negative energy changes. All values arc expreaed in joules. l Significantly different from the baseline condition.
extremity inertia has been increased (e.g., Catlin and Dressendorfer, 1979; Claremont and Hall. 1988; Martin, 1985; Myers and Steudel. 1985). An interesting implication of these results pertains to the possible effect of individual ditTercncesin anthropometric characteristics of the body, particularly the inertial characteristics of the lower extremity segments, on the economy of walking and running. There is little doubt that the explanation for individual differences in the economy of motion is a complicated one that involves a complex interaction of a variety of structural, biomechanical, and physiological factors (Cavanagh and Kram, 1985; Frederick, 1985). Nevertheless, ifall other factors were the same, results from the present study support the notion that a runner with a proportionately smaller amount of total body mass concentrated in the lower extremities would require lesseffort to generate the movements of the legsdue to the lower inertia of the segments (Cavanagh and Kram, 1985; Myers and Steudel. 1985). While it would be expected that a proportionately higher trunk mass would require an increased muscular effort, research has shown that the cost of supporting, accelerating. and decelerating mass located centrally on the body is considerably lower than that for mass carried on the extremities (Inman et al., 1981; Myers and Steudel. 1985: Ralston and Lukin. 1969). Despite the economy and motor control implications raised in the preceding discussion, there are a number of issues that the present study was not designed to address. Specifically, electromyographic data would provide a clearer indication of the response of the neuromuscular system to the modilic-
ations in lower extremity inertia and the contributions of individual muscics to the swing phase. Of particular interest are the contributions of the biarticular muscles spanning the joints of the lower extremity during unloaded and loaded running because of suggested implications for aerobic demand. From computations of net joint moment power during running, Elftman (1940) speculated that the biarticular architecture within the lower extremity reduces the muscular work required during running in comparison to a hypothetical situation in which only monoarticular muscles contribute. He further indicated that this reduction in mechanical work is reflected in a saving of energy expenditure. Finally, it would be of interest to evaluate the applicability of the kinematic-based maximum smoothness principle (Flash and ffogan, 1985; Hogan, 1984; Hogan et al., 1987) to the motions of the gait cycle and to consider the compatibility of this principle with the concept that movement patterns are developed so as to minimize aerobic demand (Cavanagh and Kram, 1985). The lack of available information on these and other issues related to extremity loading suggests the need for further research on this topic. REFERENCFS
Aleshinsky. S. Y. (1986) An energy ‘sources’ and ‘fractions’ approach to the mechanical energy expenditure problem-Ii. Movement of the multi-link chain model. J. Biomechunio 19. 295-300. Bresler. B. and Frankel. J. P. (1950) The forces and moments in the leg during level walking. Frans. ASME 72, 27-36. Catlin. M. J. and Dressendorfer. R. H. (1979) ElTect of shoe
536
P. E.‘~*.RTIN and P. R. CAYANAGH
weight on the energy cost of running. Med. Sci. Sports Exert. 11.80. Cavanagh. P. R. and Kram. R. (1985) Mechanical and muscular factors affecting the efficiency of human movement. Med. Sci. Sports Exert. 17. 32633 1. Chapman. A. E. and Caldwell, G. E. (1983) Factors determining changes in lower limb energy during swing in treadmill running. J. &mechanics 16. 69-77. Cheney. W. and Kincaid. D. (1980) Numerical Mathematics and Computing. pp. 6670. Brooks/Cole, Monterey, CA. Claremont. A. D. and Hall. S. J. (1988) Effects of extremity loading upon energy expenditure and running mechanics. Med. Sci. S~rls Exrrc. 20. 167-l 71. Clauser. C. El. McConville. j. ‘I: and Young. J. W. (1969) IC’piyht, Volume and Center or Mass of Segments of fhe Human Body (AMRL-TR-69-70). Wright Patterson Air Force Base. Ohio. Elftman. H. (1939a) Forces and energy changes in the leg during walking Am. 1. Physiol. 125. 339-356. Elftman. H. (1939b) The function of muscles in locomotion. Am. J. Physiol. 12S. 357-366. Elftman. H. (194O)The work done by muscles in running. Am. 1. Physiol. 129. 672-684. Elliott. B. C. (1977) The co-ordination of force summation in the lower limb during the recovery phase of treadmill running. J. Human Movemenr Srud. 3. 82-87. Flash. T. and Hogan. N. (1985) The coordination of arm movements: an ~experimentally confirmed mathematical model. J. Neurnsci. 5. 1688-t 703. Frederick. E. C. (1985) Synthesis. cnpcrimcntation. and the biomcchanics of economical movcmcnt. Med. Sri. Sports Exrrc. 17.44-47. Games. P. A., Gray. G. S., Herron. W. L., Pcntz. A. and Sinatra, R. (1979) Anulysis cg Variance with Repeated Meusures. The Pennsylvania State University Computation Center, University Park. PA. Hanavan. E. (1964) A Murhemuficul Model 01 rhr llumun Eody (AMRL-TR-64-102). Wright Patterson Air Force Base, Ohio. Hogan. N. (1984) An organizing principk for a class of
voluntary movements. J. Neurosci. 4 2745-2754. Hogan. N, Bizi. E.. Mussa-lnvaldii F. A. and Flash. T. (1987)Controlling multijoint motor behavior. Exert. Sport Sci. Reu. lS, 153-190. Inman. V. T.. Ralston. H. J. and Todd. F. (1981) Human Walkiny. p. 71. Williams and Wilkins. Baltimore, MD. Lestiennc. F. (1979) Effects of inertial load and velocity on the braking process of voluntary limb movements. Exp. Brain Res. 3s. 407418. Martin, P. E. (1985) Mechanical and physiological responses to lower extremity loading during running. Med. Sci. Sports Exert. 17, 427-433. Mena D.. Mansour. J. M. and Simon. S. R. I198 I1 Analvsis and synthesis of human swing leg motion during gait and its clinical applications. 1. Biomrchanics 14.823-832. Mochon. S. and McMahon. T. A. (1980) Ballistic walking. J. Biomechanics 13.49-57. Myers, M. J. and Steudel, K. (1985) Effect oflimb mass and its distribution on the energetic cost of running. . J. exp. Biol. 116.363-373. Phillips, S. J.. Roberts, E. M. and Huang, T. C. (1983) Quantification of intersegmental reactions during rapid swing motion. J. Biomechanics 16.41 l-417. Quanbury. A. 0.. Winter, D. A. and Reimer. G. D. (1975) Instantaneous power and power flow in body segments during walking. J. Human Mouemenf Srud. 1. 59-67. Ralston. H. J. and Lukin. L. (1969) Energy levels of human body segments during walking. Ergon&ics 12, 3946. Robertson. D. G. E. and Winter. D. A. 11980) Mechanical energy generation, absorbtion, and transfer gmongst segments during walking. J. Biomechanics 13. 845-854. Wells, R. P. and Winter. D. A. (1980) Assessment of signal and noise in the kinematics of normal. pathological, and sporting gaits. Humun Lucomorion I, pp. 92-93. London, Ontario, Canada. Winter. D. A. (1979) Biomeehunics of ffuman Movement, pp. 84-107. John Wiley. New York. Winter, D. A. and Robertson. D. G. E. (1978) Joint torque and energy patterns in normal gait. Biol. Cykn. 29, 137-142.