Journal of Human Evolution 44 (2003) 563579

Energy transformation during erect and bent-hip, bent-knee walking by humans with implications for the evolution of bipedalism
W.J. Wang *, R.H. Crompton, Y. Li, M.M. Gunther
Department of Human Anatomy and Cell Biology, The University of Liverpool, PO Box 147, Liverpool L69 3BX, UK Received 1 November 2002; accepted 3 March 2003

Abstract We have previously reported that predictive dynamic modeling suggests that the bent-hip, bent-knee gait, which some attribute to Australopithecus afarensis AL-288-1, would have been much more expensive in mechanical terms for this hominid than an upright gait. Normal walking by modern adult humans owes much of its eciency to conservation of energy by transformation between its potential and kinetic states. These ndings suggest the question if, and to what extent, energy transformation exists in bent-hip, bent-knee gait. This study calculates energy transformation in humans walking upright, at three dierent speeds, and walking bent-hip, bent-knee. Kinematic data were gathered from video sequences and kinetic (ground reaction force) data from synchronous forceplate measurement. Applying Newtonian mechanics to our experimental data, the uctuations of kinetic and potential energy in the body centre of mass were obtained and the eects of energy transformation evaluated and compared. In erect walking the uctuations of two forms of energy are indeed largely out-of-phase, so that energy transformation occurs and total energy is conserved. In bent-hip, bent-knee walking, however, the uctuations of the kinetic and potential energy are much more in-phase, so that energy transformation occurs to a much lesser extent. Among all modes of walking the highest energy recovery is obtained in subjectively comfortable walking, the next highest in subjectively fast or slow walking, and the least lowest in bent-hip, bent-knee walking. The results imply that if bent-hip, bent-knee gait was indeed habitually practiced by early bipedal hominids, a very substantial (and in our view as yet unidentied) selective advantage would have had to accrue, to oset the selective disadvantages of bent-hip, bent-knee gait in terms of energy transformation.  2003 Elsevier Science Ltd. All rights reserved.
Keywords: Energy exchange; Phase-shift; Bent-hip, bent-knee; Erect walking; Evolution of bipedalism

Introduction
* Corresponding author. Tel.: +44-151-794-6867; fax: +44-151-794-5517 E-mail addresses: wangwj@liv.ac.uk (W.J. Wang), rhcromp@liv.ac.uk (R.H. Crompton), yu.li@bristol.ac.uk (Y. Li), michaelg@liv.ac.uk (M.M. Gunther).

It has been proposed (Cavagna et al., 1975), and experimental studies conrm (Cavagna et al., 1976, 1977, 1983, 2000), that the energy conservation characteristic of human walking is the result of

0047-2484/03/$ - see front matter  2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0047-2484(03)00045-9

564

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

out-of-phase uctuations in kinetic and potential energy of the body centre of mass (CM). In general, as the forward velocity of the CM decreasesfrom heel-strike to mid-stancethe height of the CM increases, as the body passes over the stance leg: the potential energy of the CM, therefore, increases over the time when the kinetic energy is decreasing. Contrarily, the kinetic energy of the CM will increase from mid-stance to toe-o, as the forward velocity of the CM increases, while the potential energy decreases, as the height of the CM falls. It was further proposed (Alexander and Jayes, 1980; Alexander, 1992) that energy transformation in humans is dependent on a so-called sti gait (i.e. one where the hip and knee joint tend to be kept in relatively extended postures). The gait is associated with a characteristically double-humped curve for vertical ground reaction forces (GRFs). According to this hypothesis, if the knee is allowed to remain in substantially exed postures, the vertical ground reaction force curves will show a single hump, and energy transformation should be reduced or absent. A biomechanical link between the form of vertical ground reaction force curves and the kinematics of the hip and knee joint has subsequently been experimentally conrmed (Li et al., 1996). Energy transformation has been investigated in chimpanzees trained to walk bipedally, in an upright posture (Kimura, 1996). However, a link between bent-hip, bent-knee (BHBK, or compliant) gait and low rates of energy transformation in the CM has not yet been demonstrated. While human running is characterized by higher muscle forces and GRFs than human walking (Winter, 1990), normal human running may be expected to benet from compliance (cycle time being short enough to permit return of energy by elastic recoil), and moderately exed knee postures should therefore be tolerable. The mechanisms of bipedal walking and running are, thus, very dierent. The evolution of bipedal walking is generally regarded as the Rubicon of hominization. While our closest relatives, the African apes, do exhibit voluntary bipedalism, it is a relatively rare event, and typically characterized by exed postures of the hip and knee joints (Jenkins, 1972). The earliest relatively complete skeletal evidence for the acqui-

sition of bipedalityand hence that for which we can reasonably expect to be able to determine its mode (Wade, 2002)remains the 3.18 million year old skeleton of Australopithecus afarensis AL-288-1 Lucy (Johanson et al., 1982; Kimbel et al., 1994; Leakey et al., 1995; Sarmientos [e.g. 1998] suggestion that this hominid was a quadruped is almost universally rejected.) The nature or mode of bipedalism in A. afarensis, however, remains disputed, since individual features of the skeleton suggest adaptations for both bipedality and for arboreal climbing (see, e.g. Susman et al., 1984) It has been proposed (e.g. Stern and Susman, 1983; Hunt, 1994), therefore, that bipedalism in A. afarensis may have been facultative rather than habitual, and their gait more like the occasional bent-hip, bent-knee (BHBK) or compliant bipedalism characteristic of other (untrained) living African apes than the erect walking of modern humans. However, some are unconvinced by the evidence for arboreality, and regard A. afarensis as a committed upright biped (see, e.g. Latimer et al., 1987, and also Ward, 2002). The assessment that Lucys bipedalism was compliant is problematic, since it suggests that, (to the extent which early human ancestors walked rather than ran, see above) their bipedalism would have been of a form that might be expected to be mechanically (and presumably physiologically) inecient, lacking the kinematic requirements for energy transformation. (We shall report an experimental physiological evaluation of BHBK gait in humans elsewhere). Since it is extremely dicult to measure the metabolic costs of dierent gaits for untrained non-human primates, no unequivocal evidence exists that erect bipedalism oers direct advantages over BHBK gait, despite an extensive literature (see, e.g., Taylor and Rowntree, 1973; Rodman and McHenry, 1980; Carrier, 1984; Leonard and Robertson, 1995; Steudel, 1996). Inverse dynamic modelling studies, based on limb proportions, suggest that A. afarensis could have been a mechanically eective upright biped (Kramer, 1999), but would have incurred greatly increased mechanical costs in BHBK walking (Crompton et al., 1998). Others have suggested that as a compliant gait reduces peak vertical

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

565

GRFs during walking, the (peak) loads imposed on the sacroiliac and other joints by bipedalism would have been reduced, favouring compliant/ BHBK walking as a transitional gait during the acquisition of bipedality (Schmitt et al., 1996). It has also been suggested that exed joint postures may be benecial to changes of direction and acceleration (Preuschoft and Witte, 1991). Sellers and collaborators (2003) have recently demonstrated that forwards dynamic modelling (where motion is driven by tension generators, representing muscles, and taking into account some of their physiological properties, rather than by sets of kinematics) can predict experimentally derived metabolic costs of human upright walking within 15%. This modelling approach is currently being applied to BHBK gaits, where costs may be validated by comparison to the experimental assessments of metabolic costs of Carey (1998). In this paper, we address only the eects of BHBK gaits on transformation of mechanical energies. Using particle mechanics, we set out to determine: 1) whether and to what extent BHBK bipedal walking in humans can benet from energy transformation, comparing the rates of transformation with those in erect walking by humans in self-assessed slow, comfortable and fast speeds; and 2) whether the characteristic changes in the pattern of GRFs in upright and BHBK walking are accompanied respectively by relatively out-ofphase and relatively in-phase uctuations in the kinetic and potential energy of the body centre of mass.

surface, and was used to record ground reaction forces (GRFs) to computer disk via an AD converter, using DIA/DAGO software (GfS, Aachen). To obtain general 3D kinematic data, as well as particular information on the double support phase and the velocity of the CM, two genlocked standard CCD PAL video cameras, giving a 50 Hz sampling rate, were set parallel and at 90( respectively to the long axis of the walkway. Recordings were made split screen via a special eects generator, and were synchronised with the force records using LEDs in the eld of one camera, triggered by any of the four force transducers. All subjects walked barefoot and wore a tight-tting swimsuit. Kinematics were analyzed using our own, specially written software (Wang, 1999). The subjects started walking well before, and nished walking well after, the force platform, so that the forward velocity of the body centre of mass (CM) was kept as constant as possible, while permitting as natural a gait as possible. A total of 80 trials for each mode of walking were retained for analysis. After deleting some recording failures (such as where only half of the foot landed on the force platform, or where step length was determined to have been adjusted by the subject to permit foot contact with the force platform) 70 trials for each gait remained available for the calculation of energies. Calculation of energy There are various ways of calculating uctuations in the energies of the body centre of mass (see, e.g. Zarrugh, 1981; Williams and Cavanagh, 1983; Winter, 1983 and 1990; Williams et al., 1995). Using Newtonian mechanics and employing a force platform, Cavagna and colleagues integrated GRFs to calculate the kinetic and potential energy in the CM, and then estimated the recovery of work done (Cavagna, 1975; Cavagna et al., 1976). This method is less than ideal. Firstly, it may give slightly low estimates of the total work done (Donelan et al., 2002). Secondly, as indicated by authors including Winter (1979) and Williams et al. (1995), the energy changes of the CM do not fully represent the energy changes of the whole body: symmetrical, reciprocal movements of the limbs, which are typical for walking (erect or

Materials and methods Subjects The subjects were 8 adult men and women, aged between 20 and 40 years and 1.61.85 m in height. Each subject walked along a 25 m plywood walkway in four (subjectively determined) modes: slow, comfortable and fast erect walking, and BHBK (compliant) walking. 10 trials were recorded for each subject for each mode. A Kistler 9281B force platform (surface dimensions: 0.4 0.6 m) was set into the walkway, level with its

566

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

compliant) do not result in changes in the position of the CM. The dierences in energy changes of the CM between dierent gaits, therefore, reect only part of the energetics of walking. Nevertheless, the method remains a simple and straightforward approach. As this study does not concern itself with a complete estimate of segment energies, we employ a similar technique to that of Cavagna and colleagues (1975 and 1976). However, our approach is a slightly modied version. Unlike Cavagna and colleagues, who calculated energy exchange from absolute values of work done, we utilize the value of uctuation in energy, which should enable us to take the eects of both energy output and absorption into consideration. In this study, therefore, work done is estimated by calculating the uctuations in potential and kinetic energies and the sum of both. The uctuations of kinetic and potential energies are dened as the work done in maintaining motion of the body CM, and the uctuation of the sum of the kinetic and potential energy as the work produced by the body. To permit use of a single forceplate, avoiding the problems of ensuring contact with two plates, and thus permitting a more natural gait, we assumed that the subjects walked symmetrically. GRFs for one side were mirrored to the other side and oset by 50% of a stride cycle. By integrating force platform data, we readily obtain curves of kinetic energy and potential energy, and can, then, investigate their dynamic trends during walking. The method is described in detail below. The energies of the CM can be obtained by calculation from GRFs. If the whole body is considered as a particle, Newtons Second Law can be written as: Fx,y max,y Fz mg maz (1) (2)

reaction forces for both feet, taking the double support phase into consideration. Thus, the acceleration of the CM, a, can be obtained. By integrating a once, we obtain velocity (Eq. 34) and twice, displacement (Eq. 5): 1 vx,y(t) vx,y(t0) m
t t

Fx,y(t)dt
t0

(3)

1 vz(t) vz(t0) m

(Fz(t) mg)dt
t0 t

(4)

sx,y,z(t) sx,y,z(t0)
t0

vx,y,z(t)dt

(5)

where v is velocity and s displacement. From the denition of mechanical energy, the translational kinetic energy and the gravitational potential energy of the CM of a subject are: KE 1 2 mv 2 c (6) (7)

PE mgzc

where PE is the potential energy; KE the kinetic energy; vc the velocity of the CM in horizontal and vertical directions; and zc the displacement of the CM in the vertical direction. Fluctuation of energies To analyse the work done by the whole system (the body) to maintain motion, the uctuation of the energies, E, over the total stride (i.e. stance and swing phases together) is computed as follows: E max(E) min(E) (8)

where F is the GRF acting on a subject, measured by a force platform; m the body mass; a the acceleration of the subjects CM; and x, y and z respectively represent the transverse axis, the anterior-posterior axis, and the vertical axis. F in this case is total ground force, i.e., the sum of the

E represents the work done by the body to maintain whole body movement, and can be determined for kinetic energy, potential energy or their sum. The larger E, the more work done. KE and PE indicate the energy for maintaining

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

567

motion and (KE + PE) signies the energy output of the body, or in other words, the work done by the body tissues (see Appendix A). Energy-transfer value To compare the eect of energy transformation in dierent modes of walking, we have dened a coecient, , the energy-transfer value, as: KE1 PE (PE1KE) (9)

where PE is the maximum change in the potential energy; KE the maximum change in the kinetic energy; and (PE + KE) the maximum change in the sum of the two energies. Although the Eq. (11) has a dierent form from that of Cavagna et al., (1976), it demonstrably reects the eectiveness of energy transformation between the two forms of energies (see Appendix B). In order to test whether or not there are signicant dierences between the experimental gaits and the calculated parameters, a statistical analysis was carried out using ANOVA (Bowker and Lieberman 1959).

This dimensionless coecient reects the ratio of energy transformation and is thus independent of the mode of walking. The larger , the more the energy exchanged. Energy-velocity value To compare the eect of per unit energy on the velocity of the CM, another coecient, , is created. This may be termed the energy-velocity value, and is dened as below: VCM E (10)

Results Range of joint angles The recorded kinematic data show that during BHBK walking, the range of hip angles (angle between the long axes of the thigh and trunk) of the subjects averagely ranged from a mean 23 to a mean 69(, the knee angle (dened as the angle between the long axes of the thigh and the lower leg) from 35 to 92( and the ankle angle (dened as the angle between the foot and lower leg) from 2.86 to 34(. In upright walking, in comparison, hip angles ranged from 12 to 34(, knee angles from 0 to 60( and ankle angles from 12 to 20( (Fig. 1.ac). In the literature (Winter, 1991, pp.29), the joint angles in erect walking are about 10 to 20( for the ankle, 0 to 60( for the knee and 18 to 23( for the hip. The slight dierences between the literature and our data may result from two causes: 1) the marker positions dened on subjects for kinematic measurement may be dierent; and/or 2) the velocities of the subjects may be slightly dierent: those for fast, normal, and slow walking in Winter (1991, pp. 12) are about 1.6824, 1.325, and 0.998 (m/s), while the velocities for our subjects are 1.93, 1.47, and 1.04 (m/s) respectively (see Table 1). Since the four gaits in our study were determined by the subjects themselves and are internally consistent, the slight dierences in joint angles between Winters (1991) and our data may safely be ignored (see Fig. 1.ac).

where VCM is the average velocity of the CM and E is the range of uctuation of the energy of the whole body. In this case, VCM is dened as the distance covered in a complete stride divided by the strides duration, and is calculated using the recorded kinematic data, and E is the uctuation of the sum of kinetic and potential energies, (KE + PE). therefore expresses the per unit eect of energy use on the velocity of the CM. The larger , the higher the velocity achieved for a given expenditure of energy. Recovery of energy In order to evaluate the eectiveness of energy transformation, we have dened a new coecient of energy recovery: Recoveryn ( PE1 KE)K (PE1KE) ( PE1 KE) (11)

568

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

Fig. 1. Two gaits: erect walking and bent-hip, bent-knee walking. a) Digitised stick-gure illustration of typical kinematics of comfortable erect walking in the sagittal view. b) Digitised stick-gure illustration of typical kinematics of bent-hip, bent-knee walking in the sagittal view. c) Averaged ranges of joint angle for the subjects in dierent modes of walking. Right: erect walking. Left: BHBK walking.

Energy uctuation From the computed results, it was found that in erect walking, the uctuations of the two energies are largely out-of-phase or nearly so (i.e. one increases while the other decreases). However, in BHBK walking, the uctuations are reasonably

in-phase or nearly so (i.e. both increase or decrease roughly at the same time) (see, especially, Figs. 25). The averages of the computed energies (KE, PE and KE + PE) in dierent modes of walking, and their standard deviations, are also given in Figs. 25. In Figs. 2.b5.b, the curves for kinetic energy (KE) and potential energy (PE) have been moved

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579 Table 1 Comparison of energy uctuations in dierent modes of walking SF Mean BW FW SW CW 0.3900 0.4500 0.3950 0.4250 VCM(m/s) Mean STD 1.2261 1.9329 1.0388 1.4703 0.23 0.14 0.11 0.13 KE(J/kg) Mean STD 0.2801 0.6475 0.3130 0.4850 0.10 0.15 0.05 0.08 PE(J/kg) Mean STD 0.2548 0.4175 0.2650 0.3075 0.12 0.13 0.09 0.10 (KE + PE)(J/kg) Mean STD 0.3883 0.5204 0.2944 0.3529 0.16 0.13 0.10 0.09

569

1. BW, bent-hip, bent-knee walking; FW, fast erect walking; SW, slow erect walking; CW, comfortable erect walking (n = 70 trials each mode). 2. SF, swing factor, the proportion of the swing time to the cycle duration. 3. VCM, velocity of the body centre of mass. 4. KE, range of change of kinetic energy; PE, range of change in potential energy; (KE + PE), range of change in the sum of kinetic and potential energies.

to the coordinate system of the sum of the kinetic and potential energies (KE + PE) in order to facilitate comparison of energy changes. The curves start at left heel strike (LHS), then right toe o (RTO), right heel strike (RHS), left toe o (LTO) and nish at the next left heel strike (LHS). Because the period of double support diers between the modes of walking, the timing of these parameters diers. RHS however is always at 50% of the cycle, as the right and left side of the body were assumed to move symmetrically. In comfortable walking, RTO occurs at a mean of 7.5% and LTO at a mean of 57.5% (S.D. 0.027) of a gait cycle. In fast walking, RTO occurs at a mean of 5% and LTO at a mean of 55% (S.D. 0.038) of the cycle. In slow walking RTO occurs at a mean of 10% and LTO at a mean 60% (S.D. 0.033) of the cycle, and in BHBK walking RTO occurs at a mean of 11% and LTO at a mean 61% (S.D. 0.031) of the cycle (see Figs 2.b5.b, and see also the swing factor in Table 1). It may readily be seen from Figs 2.b, 3.b and 4.b that in upright walking the kinetic and potential energies uctuate alternately: KE increases as PE decreases and viceversa. This permits energy transformation or energy exchange. In erect walking, the sum of kinetic and potential energy (PE + KE) oscillates within a relatively smaller range than does KE and PE. In BHBK walking (Fig. 5.b), however, the range of (KE + PE) is not obviously smaller than that of KE or of PE.

As expected, kinetic energy increases with increase of the velocity of CM (see Fig. 7), while potential energy does not change with increasing velocity (see Fig.6). Calculated results of energies for all trials in the dierent modes of walking are shown in Figs. 68, and the averages listed in Table 1. Direct comparisons of KE and PE in the dierent modes of walking should not be taken too far, since the KE and PE of any subject are proportional to the velocity of the CM (VCM). Thus, as VCM in BHBK walking is relatively small, so too are KE and PE (see Figs 6 and 7). However, we may note that the uctuation of the sum of the two energies (i.e. (KE + PE)) is relatively larger-compared to KE and PE in BHBK walking than it is in other modes (Fig. 68). Further, though the VCM in BHBK walking is lower than that in comfortable walking, the uctuation of (KE + PE) is at least as large (see also Table 1). Finally, we can also see (Table 1) that because energy exchange is so small, BHBK walking is higher in (KE + PE) for a lower velocity. The mean of for comfortable walking is the greatest, at 2.479; while the values for fast and slow walking are almost exactly the same: 2.2014 and 2.2002, but the for BHBK walking is the smallest, at 1.5241 (see Fig. 9 and Table 2), meaning that comfortable walking achieves the largest energy transformation and BHBK walking the

570

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

Fig. 2. Energy uctuation in comfortable walking. a) Top: mean range of uctuation in PE (potential energy, solid line) and its standard deviations (dashed lines) in J/kg. Middle: mean range of uctuation in KE (kinetic energy, solid line) and its standard deviations (dashed lines) in J/kg. Bottom: mean range of uctuation in KE + PE (sum of kinetic and potential energies, solid line) and its standard deviations (dashed lines). Trails = 70. b) Mean uctuation of the centre of mass in KE + PE (the sum of kinetic and potential energies, * line) in KE ( lines) and in PE (B line) in J/kg, trails = 70. LHS: left heel strike, RTO: right toe o, RHS: right heel strike, and LTO: left toe o. Comfortable walking obtains the highest values in the eectiveness of energy transformation among all modes of walking.

Fig. 3. Energy uctuation in fast walking. a) All symbols are the same as those in Fig. 2a, b) All symbols are the same as those in Fig. 2b. Fast walking obtains higher values in the eectiveness of energy transformation than that in BHBK walking.

least. All trials are displayed in Fig. 9, and the average values for dierent modes of walking are listed in Table 2. The highest energy-velocity value ( ) is obtained, once again, from comfortable walking

(mean 4.5449) and the lowest from BHBK walking (mean 3.7638) (see Table 2). Values of for all trials are displayed in Fig. 10, and the average values of for dierent modes of walking are listed in Table 2. The values of energy recovery show a similar trend to other parameters: the values decrease in turn from comfortable walking (55%) to fast walking (51%), to slow walking (49%), and nally to BHBK walking (27%) (see Table 2). The results of the ANOVA for inuences of gaits on calculated parameters are given in Table 3.

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

571

Fig. 4. Energy uctuation in slow walking. a) All symbols are the same as those in Fig. 2a, b) All symbols are the same as those in Fig. 2b. Slow walking obtains higher values in the eectiveness of energy transformation than that in BHBK walking.

Fig. 5. Energy uctuation in bent-hip, bent-knee walking. a) All symbols are the same as those in Fig. 2a, b) All symbols are the same as those in Fig. 2b. Bent-hip, bent-knee walking obtains the lowest values in the eectiveness in energy transformation among all modes of walking.

In brief, almost all variables show signicant dierences between the groups (see Table 3).

Discussion Possible reasons Two inuences on energy transformation in walking may readily be identied. The rst derives from the shape of the fore-aft and vertical ground reaction force curves from which our gures for

transformation are derived. The averages of these GRFs are given in Fig. 11. In comfortable and fast walking, the valley in the vertical (Z) force curves is well marked. This indicates that vertical acceleration decreases as the displacement (height) of the CM increases, which will have the eect of converting kinetic energy into gravitational energy. The valley is smaller in slow walking and nearly absent in BHBK walking, so that less transformation can occur. However, this factor cannot explain why, while the valley is deepest in

572

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

Fig. 6. The ranges of uctuation in PE in dierent modes of walking for all trials. PE: uctuations in potential energy. B = bent-hip, bent-knee walking, = slow walking, + = fast walking * = comfortable walking.

fast walking, energy transformation in this gait is less than that in comfortable walking, and in fact similar to that in slow walking. The second, and in view of the above, probably more important factor is the phase relationship between the uctuations in the kinetic and potential energies. In Fig. 11, we can see that in normal walking, the fore-aft (Y) force changes sign, from negative to positive, at about mid-stance. This implies that the velocity of the CM, and, hence, the KE will be lowest at mid-stance. On the other hand, the simultaneous valley in the vertical force (Z) indicates that the vertical displacement (height) of the CM, and, hence, the PE is greatest at mid stance. In particular, Fig. 2.b shows that in comfortable walking, KE reaches its minimum and PE its maximum at about 30% and again at about 80% of the gait cycle. In BHBK walking, as in erect walking (see Fig. 5.b), the fore-aft force (Y) changes sign at about mid-stance, so that the velocity of CM, and thus the KE, is the lowest at the mid-stance, but the minimum vertical (Z) force, and, hence, maximum PE, occurs after midstance. The shift of phase, bringing KE and PE

into phase or nearly so, prevents substantial energy transformation (see Figs 11 and 5.b). Figs. 3.b and 4.b show that in fast and slow walking, the uctuations, and, hence, the energy transformation, are intermediate between these extremes, although considerably closer to the pattern in comfortable walking than to that in BHBK walking. Comparison with other studies The above suggests that the mode of gait may indeed be a major factor in determining energy recovery. From the calculated results, comfortable walking obtains the highest recovery, 55%, fast and slow walking about 50%, and BHBK walking the lowest, at 27%. Even though recovery in this study has a somewhat dierent meaning from that in Cavagna et al. (1976, 1977) and some other studies, the values fall within the range given by Cavagna and colleagues for humans. However, a smaller dierence was found between comfortable and fast walking than the nearly-twofold dierence suggested by Cavagna and colleagues. In his study of chimpanzees trained in upright bipedalism,

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

573

Fig. 7. The ranges of uctuation in KE in dierent modes of walking for all trials. KE: uctuations in kinetic energy. B = bent-hip, bent-knee walking, = slow walking, + = fast walking and * = comfortable walking.

Kimura (1996) found that energy recovery falls o much more sharply with increasing speed: it is highest, 40%, at 0.5 m/s, but zero by around 1.6 m/s. From our results, recovery in human BHBK walking is nearly half that in human erect, comfortable, and slow walking, although BHBK walking achieved similar speeds. McMahon and colleagues (1985, 1987) compared Groucho running (in essence, BHBK running) to normal running, and found that in this gait, both stiness of the legs and GRFs were reduced, but the rates of O2 consumption increased by as much as 50%. The results suggest that even in running compliance may not always result in return of elastic energy. Carey (1998) measured the metabolic costs of BHBK walking for adults at dierent speeds, and found that costs are much greater in BHBK walking than in erect walking (Carey and Crompton, 1997). Finally, Schmitt et al. (1996) tested compliant and normal gaits using a force platform and found that peak (but not mean) vertical force is reduced by 1025% of body weight during compliant gait, a conclusion with which our results agree. Thus, kinetic and

physiological experiments, both on BHBK walking and on running suggest that although peak GRFs may well be reduced, metabolic costs are greatly increased in gaits where hip and knee exion is substantially larger than the values seen in erect walking. We do not here address the mechanism whereby higher mechanical costs and low rates of transformation in BHBK gaits may be related to higher metabolic costs: but these parameters are more than likely to be functionally related. Implications for the evolution of bipedalism What does this study imply for the evolution of bipedalism? Biomechanical factors (Rose, 1991) may, all other things being equal, be expected to select for changes in morphology that will reduce energetic costs or increase performance in any species most ecologically or reproductively crucial locomotor behaviour. While natural selection sometimes appears to produce solutions other than the most energetically optimal one (e.g., the probably inecient knuckle-walking gait of African apes [Richmond and Strait 2000]), it

574

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

Fig. 8. The ranges of uctuation in the sum of (KE + PE) in dierent modes of walking. (KE + PE): uctuations in the sum of two energies. B = bent-hip, bent-knee walking, = slow walking, + = fast walking and * = comfortable walking.

should equally be borne in mind that such solutions may be functional consequences of strong selection in favour of other (more expensive, more stress-inducing or more ecologically crucial) behaviours, an issue we shall address elsewhere. Maintenance of extended hip and knee joint posture is demonstrably possible in trained African apes and in untrained orang-utans, despite a nonhuman-like joint conformation, and a non-humanlike lumbar spine. Recalling the prediction that BHBK walking in Australopithecus afarensis would have high mechanical joint power requirements (Crompton, et al., 1998), our present nding that energy transformation is much lower in bentknee, bent-hip walking than in erect walking, suggests that for BHBK gait to have practiced by early bipeds, very large osetting selective advantages would have had to accrue to BHBK (but not erect bipedalism), for the former to be adopted as habitual gait. In our view, neither (putatively) reduced peak loads on the sacroiliac joint (Schmitt et al., 1996) nor advantages for change in direction or speed (Preuschoft and Witte, 1991) are entirely convincing as such osetting selective advantages. Such selective advantages remain to be identied.

As Rose (1991) pointed out, many ecological, social and morphological factors will have inuenced the evolution of bipedalism. Therefore, biomechanical factors would be expected to operate to nd a best-compromise solution for performance/eciency in several dierent aspects of locomotor adaptation (sensu lato) (and see also Alexander, 1996 and 1991; Wang et al., 2003), but energetic eectiveness in walking would certainly be expected to be among the most important factors in determining the compromise. Summary This study has investigated whether there are dierences in energy transformation between the erect and bent-knee-bent-hip walking. A group of human adults were required to walk using various modes of walking. Force platform and kinematic data were collected, and energy uctuations in the body centre of mass calculated. The results show that in erect walking, the energy uctuations are substantially out-of-phase, so that the kinetic and potential energies can frequently be exchanged with each other; however, in bent-knee, bent-hip

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

575

Fig. 9. Values of energy-transfer coecient, . The average for comfortable walking is the highest and that for bent-hip, bent-hip walking the lowest. B = bent-hip, bent-knee walking, = slow walking, + = fast walking, and * = comfortable walking.

Table 2 Comparison of the eects of energy transformation in dierent modes of walking (n = 70 trials each mode) VCM BW 1.2261 FW 1.9329 SW 1.0388 CW 1.4700 0.23 0.14 0.11 0.13 3.7638 4.0075 3.9041 4.5449 1.70 1.23 1.24 1.49 1.5241 2.2014 2.2002 2.4790 0.58 0.80 0.87 0.99 Recoveryn % 27 51 49 55

hip walking. The results imply that if bent-knee, bent-hip gait was indeed habitually practiced by early hominids, an as yet unidentied selective advantage of BHBK over erect bipedalism would have had to exist, sucient to oset its demonstrably large energetic disadvantage.

Note: 1. BW, FW, SW and CW: see Table 1. 2. VCM, velocity of the centre of mass. 3. , energy-velocity coecient, see Eq. (10) in the Methods. 4. , energy-transfer value, see Eq. (9) in the Methods. 5. Recoveryn, energy recovery value, see Eq. (11) in the Methods.

Acknowledgements We are grateful to Profs. R. McN. Alexander, H. Preuschoft and M. D. Rose for their helpful comments. We would like to thank the associate editor and three referees for the constructive comments during peer-review of the manuscript. The authors thank Drs. A. Conant and R. Payne for the assistance with early draughts. This research is funded by grants from the Biotechnology and Biological Sciences Research Council, the Natural Environment Research Council, and The Leverhulme Trust, UK.

walking, the energy uctuations are relatively in-phase or nearly so, so that little or no energy transformation between kinetic and potential energies would be possible. Regarding energy recovery, among the four modes of walking investigated, the highest energy recovery occurs in comfortable walking, the next highest in fast or slow walking, and the lowest in bent-knee, bent-

576

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

Fig. 10. Values of energy-velocity coecient, . The average for comfortable walking is the highest, and for bent-hip, bent-knee walking the lowest. Other symbols are the same as those in Fig. 9.

Table 3 Analysis of variance (ANOVA) of the data from row data on variables F VCM PE KE (PE + KE) Recoveryn s2/s2 1 2 406.46 192.18 30.96 41.77 3.94 16.80 19.90 k 0.077 0.049 0.054 0.060 0.696 0.401 0.096 P <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05

t1

t1

t1

W
t0 t1

Fxdx
t0

Fydy
t0 t1

(Fz mg)dz
t1

maxdx
t0 t1

maydy
t0 t1 t0

(maz mg)dz
t1

dvx dx m dt mvxdvx

dvy dy m dt
t1

dvz dz m dt
t1

t1

mgdz
t0

(see general statistics texts). Note: 1. F = 2. F0.05;(3,276) = 2.65. 3. k = k*(s2/n)0.5 used to decide the distance between groups 2 (see Bowker and Lieberman, 1959, p.298).

t0 t1

t0 t1

t0

mvydvy
t0 t0

mvzdvz
t0

mgdz

t0

Appendix A By the denition of work, there is an equation: dW ds (1)

1 1 2 2 m(vx(t1) vx(t0)) m(v2(t1) v2(t0)) y y 2 2 1 2 2 m(vz (t1) vz (t0)) mg(z(t1) z(t0)) 2 KE(t1) KE(t0) PE(t1) PE(t0)

(2)

where F is all forces and s displacement. Thus the total work done is

Therefore, the work done can be expressed by the changes in energy. When the work done is calculated over a stride cycle, the total work should,

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579

577

Fig. 11. Characteristics of ground reaction forces. Vertical axis: force in units of body weight (N/BW). Solid line: mean; dashed line: standard deviation. Left: Fz, ground reaction forces in the vertical direction; right: Fy, ground reaction forces in the forward direction. From top to bottom: SW: slow erect walking; FW: fast erect walking; CW: comfortable (normal) walking; and BHBK: bent-hip, bent-knee walking.

theoretically, be zero because the velocities in three directions and the displacement in the direction z will repeat the values occurring in the previous cycle. Thus, we may consider the maximum change in energy uctuations to represent the work done in walking (see Eq. 8). Appendix B Cavagna et al. (1975, 1976, 1983 and 2000) have expressed energy recovery in terms of work: Recovery Wf Wv Wext Wf Wv (1)

As we express the work done by the change in energies (see Appendix A), the energy recovery can be calculated using Eq. (11): Recoveryn ( PE1 KE)K (PE1KE) ( PE1 KE) (11)

Equation (11) has a clear physical meaning: it estimates the eectiveness of energy transformation between the kinetic and potential energies.

References
Alexander, R.McN., 1991. Characteristics and advantages of human bipedalism. In: Rayner, J.M.V., Wooton, R.J. (Eds.), Biomechanics and Evolution. Cambridge University Press, Cambridge, pp. 255265. Alexander, R.McN., 1992. A model of bipedal locomotion on compliant legs. Phil. Trans. Roy. Soc. Lond. (B), 189198.

where Wf is the positive work in the forward direction; Wv the positive work in the vertical direction; and Wext the positive work in the vertical and forward directions.

578

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579 Latimer, B., Ohman, J.C., Lovejoy, C.O., 1987. Talocrural joint in African hominoids: implications for Australopithecus afarensis. Am. J. Phys. Anthropol. 74, 155176. Leonard, W.R., Robertson, M.I., 1995. Energetic Eciency of Human Bipedality. Am. J. Phys. Anthropol. 97, 335338. Li, Y., Crompton, R.H., Gunther, M.M., Alexander, R.McN., Wang, W.J., 1996. Characteristics of ground reaction forces in normal and chimpanzee-like bipedal walking by humans. Folia Primatol. 66, 137159. McMahon, T.A., 1985. The role of compliance in mammalian running gaits. J. Exp. Biol. 115, 263282. McMahon, T.A., Valiant, G., Frederick, E.C., 1987. Groucho running. J. Appl. Physiol. 62, 23262337. Preuschoft, H., Witte, H., 1991. Biomechanical reasons for the evolution of hominid body shape. In: Coppens, Y., Senut, B. (Eds.), Origine(s) de la bipedie chez les Hominides. Paris, Editions du C.R.N.S, pp. 5978. Richmond, B.G., Strait, D.S., 2000. Evidence that humans evolved from a knuckle-walking ancestor. Nature. 404, 382385. Rodman, P.S., McHenry, H., 1980. Bioenergetics and the origin of human bipedalism. Am. J. Phys. Anthropol. 52, 103106. Rose, M.D., 1991. The process of bipedalisation in hominids. In: Coppens, Y., Senut, B. (Eds.), Origine(s) de la bipedie chez les hominides. Paris, Editions du CRNS, pp. 3748. Sarmiento, E.E., 1998. Generalized quadrupeds, committed bipeds and the shift to open habitats: An evolutionary model of hominid divergence. Am. Mus. Nov. 3250, 178. Schmitt, D., Stern, J.R., Larson, S.G., 1996. Compliant gait in humans: implications for substrate reaction forces during Australopithecine bipedalism. Am. J. Phys. Anthropol. Suppl. 22, 209. Sellers, W.I., Dennis, L.A., Crompton, R.H. 2003. Predicting the metabolic energy costs of bipedalism using evolutionary robotics. J. Exp. Biol. 206, 11271136. Steudel, K., 1996. Limb morphology, bipedal gait, and the energetics of hominid locomotion. Am. J. Phys. Anthropol. 99, 345355. Stern, J.T., Susman, R.L., 1983. The locomotor anatomy of Australopithecus afarensis. Am. J. Phys. Anthropol. 60, 279317. Susman, R.L., Stern, J.T., Jungers, W.L., 1984. Arboreality and bipedality in the Hadar hominids. Folia Primatol. 43, 113156. Taylor, C.R., Rowntree, V.J., 1973. Running on two or on four legs: Which consumes more energy? Science 179, 186187. Wang, W.J. 1999. The mechanics of bipedalism in relation to load-carrying: biomechanical optima in hominid evolution. PhD thesis, The University of Liverpool. Wang, W.J., Crompton, R.H., Li, Y., Gunthor, M.M., 2003. Optimum ratio of upper to lower limb lengths in handcarrying of a load under the assumption of frequency coordination. J. Biomech. 36, 249252. Ward, C.V., 2002. Interpreting the posture and locomotion of Australopithecus afarensis: where do we stand? Yrbk Phys. Anthropol. 45, 185215.

Alexander, R.McN., 1996. Optima for Animals. Princeton University Press, Princeton, NJ. Alexander, R.McN., Jayes, A.S., 1980. Fourier analysis of forces exerted in walking and running. J. Biomech. 13, 383390. Bowker, A.H., Lieberman, G.J., 1959. Engineering statistics. Prentice-Hall, Englewood Clis (N.J.). Carrier, D.R., 1984. The energetic paradox of human running and hominid evolution. Curr. Anthropol. 25, 483495. Carey, T.S. 1998. Energetics of early hominid bipedalism. PhD thesis, The University of Liverpool. Carey, T.S., Crompton, R.H., 1997. The bipedal locomotion of early hominids. J. Morphol. 232(3), 240. Cavagna, G.A., 1975. Force platforms as ergometers. J. App Physiol. 39, 174179. Cavagna, G.A, Thys, H., Zamboni, A., 1976. The sources of external work in level walking and running. J. Physiol. 262, 639657. Cavagna, G.A., Heglund, N.C., Taylor, C.R., 1977. Mechanical work in terrestrial locomotion: two basic mechanisms for minimizing energy expenditure. Am. J. Physiol. 233, 243261. Cavagna, G.A., Franzetti, P., Fuchimoto, T., 1983. The mechanics of walking in children. J. Physiol. 343, 323339. Cavagna, G.A., Williams, P.A., Heglund, N.C., 2000. The role of gravity in human walking: pendular energy exchange, external work and optimal speed. J. Physiol. 528(3), 657668. Crompton, R.H., Li, Y., Wang, W.J., Gunther, M.M., Savage, R., 1998. The mechanical eectiveness of erect and bentknee, bent-hip bipedal walking in Australopithecus afarensis. J. Hum. Evol. 35, 5574. Donelan, J.M., Kram, R., Kuo, A.D., 2002. Simultaneous positive and negative external mechanical work in human walking. J. Biomech. 35, 117124. Hunt, K.I., 1994. The evolution of human bipedality: ecology and functional anatomy. J. Hum. Evol. 26, 183202. Jenkins, F.A. Jr., 1972. Chimpanzee bipedalism: cineradiographic analysis and implications for the evolution of gait. Science 178, 877879. Johanson, D.C., Lovejoy, C.O., Kimbel, W.H., White, T.D., Ward, S.C., Bush, M.E., Latimer, B.M., Coppens, Y., 1982. Morphology of the Pliocene Partial hominid skeleton (AL 288-1) from the Hadar formation, Ethiopia. Am. J. Phys. Anthropol. 57, 403452. Kimbel, W.H., Johanson, D.C., Rak, Y., 1994. The rst skull and other new discoveries of Australopithecus afarensis at Hadar, Ethiopia. Nature 368, 449451. Kimura, T., 1996. Centre of gravity of the body during the ontogeny of chimpanzee bipedal walking. Folia Primatol. 66, 126136. Kramer, P.A., 1999. Modelling the locomotor energetics of extinct hominids. J. Exp. Biol. 202, 28072818. Leakey, M.G., Fiebel, C.S., McDougall, I., Walker, A.C., 1995. New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature 376, 565571.

W.J. Wang et al. / Journal of Human Evolution 44 (2003) 563579 Williams, P.A., Cavagna, G.A., Heglund, N.C., 1995. External, internal and total works in human locomotion. J. Exp. Biol. 198, 379393. Williams, K.R., Cavanagh, P.R., 1983. A model for calculation of mechanical power during distance running. J. Biomech. 16, 115128. Winter, D.A., 1979. A new denition of mechanical work done in human movement. J. Appl. Physiol. 46, 7983. Winter, D.A., 1983. Energy generation and absorption at the ankle and knee during fast, natural, and slow cadences. Clin. Orthop. Rel. Res. 175, 147154.

579

Winter, D.A., 1990. Biomechanics and Motor Control of Human Movement (2nd edition). Wiley-Interscience, New York. Winter, D.A., 1991. The Biomechanics and Motor Control of Human Gait: Normal, Elderly and Pathological. University of Waterloo Press, Waterloo, Ontario. Zarrugh, M.Y., 1981. Power requirements and mechanical eciency of treadmill walking. J. Biomech. 14, 157165.