A feedback-controlled treadmill (treadmill-on-demand) and the spontaneous speed of walking and running in humans.

A novel apparatus, composed by a controllable treadmill, a computer, and an ultrasonic range finder, is here proposed to help investigation of many aspects of spontaneous locomotion. The acceleration or deceleration of the subject, detected by the sensor and processed by the computer, is used to accelerate or decelerate the treadmill in real time. The system has been used to assess, in eight subjects, the self-selected speed of walking and running, the maximum "reasonable" speed of walking, and the minimum reasonable speed of running at different gradients (from level up to +25%). This evidenced the speed range at which humans neither walk nor run, from 7.2 +/- 0.6 to 8.4 +/- 1.1 km/h for level locomotion, slightly narrowing at steeper slopes. These data confirm previous results, obtained indirectly from stride frequency recordings. The self-selected speed of walking decreases with increasing gradient (from 5.0 +/- 0.8 km/h at 0% to 3.0 +/- 0.9 km/h at +25%) and seems to be approximately 30% higher than the speed that minimizes the metabolic energy cost of walking, obtained from the literature, at all the investigated gradients. The advantages, limitations, and potential applications of the newly proposed methodology in physiology, biomechanics, and pathology of locomotion are discussed in this paper.     

Classification of gait patterns in the time-frequency domain

This paper describes the classification of gait patterns among descending stairs, ascending stairs and level walking activities using accelerometers arranged in antero-posterior and vertical direction on the shoulder of a garment. Gait patterns in continuous accelerometer records were classified in two steps. In the first step, direct spatial correlation of discrete dyadic wavelet coefficients was applied to separate the segments of gait patterns in the continuous accelerometer record. Compared to the reference system, averaged absolute error 0.387 s for ascending stairs and 0.404 s for descending stairs were achieved. The overall sensitivity and specificity of ascending stairs were 98.79% and 99.52%, and those of descending stairs were 97.35% and 99.62%. In the second step, powers of wavelet coefficients of 2 s time duration from separated segments of vertical and antero-posterior acceleration signals were used as features in classification. Our results proved a reliable technique of measuring gait patterns during physical activity.     

Asymmetries in postural control and locomotion in chimpanzees (Pan troglodytes)

Posture and locomotion are two of the most primitive and basic motor manifestations of an organism's behavior. Although the restrictions they impose on other motor functions are evident, few studies have considered the possibility of asymmetries in these behaviors in human and nonhuman primates, and how they might impact other asymmetries at higher functional levels. The aim of the current study was to explore in a group of 10 chimpanzees at the Madrid Zoo-Aquarium the degree of asymmetry in four behaviors related to locomotion (walking, ascending, descending, and brachiating) and four behaviors associated with posture (sitting, lying, hanging, and changing postures). Few subjects showed individual preferences, but significant trends in the group for some of the behaviors were found, including right-hand use when initiating quadruped walking, and left-hand use when descending and hanging. Some significant correlations also emerged: a negative one between walking and descending, and a positive one between walking and brachiating and between sitting and changing postures. No correlations were found between locomotor and postural modes. Although we cannot make generalizations on the population level at this time, these findings highlight the importance of considering postural and locomotion factors when studying motor asymmetries in primates.     

Development and validation of a new method to measure walking speed in free-living environments using the actibelt® platform

Walking speed is a fundamental indicator for human well-being. In a clinical setting, walking speed is typically measured by means of walking tests using different protocols. However, walking speed obtained in this way is unlikely to be representative of the conditions in a free-living environment. Recently, mobile accelerometry has opened up the possibility to extract walking speed from long-time observations in free-living individuals, but the validity of these measurements needs to be determined. In this investigation, we have developed algorithms for walking speed prediction based on 3D accelerometry data (actibelt?) and created a framework using a standardized data set with gold standard annotations to facilitate the validation and comparison of these algorithms. For this purpose 17 healthy subjects operated a newly developed mobile gold standard while walking/running on an indoor track. Subsequently, the validity of 12 candidate algorithms for walking speed prediction ranging from well-known simple approaches like combining step length with frequency to more sophisticated algorithms such as linear and non-linear models was assessed using statistical measures. As a result, a novel algorithm employing support vector regression was found to perform best with a concordance correlation coefficient of 0.93 (95%CI 0.92-0.94) and a coverage probability CP1 of 0.46 (95%CI 0.12-0.70) for a deviation of 0.1 m/s (CP2 0.78, CP3 0.94) when compared to the mobile gold standard while walking indoors. A smaller outdoor experiment confirmed those results with even better coverage probability. We conclude that walking speed thus obtained has the potential to help establish walking speed in free-living environments as a patient-oriented outcome measure.     

Measurement of Human Daily Physical Activity

Objectives: To validate a new device, Intelligent Device for Energy Expenditure and Activity (IDEEA), for the measurement of duration, frequency, and intensity of various types of human physical activity (PA). Research Methods and Procedures: The ability of IDEEA to identify and quantify 32 types of PA, including the most common daily exercise and nonexercise PA, was tested in 76 subjects: Subjects included males (N = 33) and females (N = 43) ranging in age from 13 to 72 years with a mean body mass index (BMI) of 24.7 kg/m² (range: 18.4 to 41.0) [43 females: 13 to 72 years old and BMI 18.4 to ～41.0 kg/m² (mean = 24.7 kg/m² ); 33 males: 15 to ～72 years old and BMI 21.0 to ～38.4 kg/m² (mean = 25.9 kg/m²)]. Postures, limb movements, and jumping were tested using a timed protocol of specific activities. Walking and running were tested using a 60‐meter track, on which subjects walked and ran at 6 self‐selected speeds. Stair climbing and descending were tested by timing subjects who climbed and descended a flight of stairs at two different speeds. Results: Correct identification rates averaged 98.9% for posture and limb movement type and 98.5% for gait type. Pooled correlation between predicted and actual speeds of walking and running was high (r = 0.986, p ≤ 0.0001). Discussion: IDEEA accurately measured duration, frequency, type, and intensity of a variety of daily PAs.     

Projections of excitatory and inhibitory motor neurones to the circular and longitudinal muscle of the guinea pig colon

The aim of this study was to identify myenteric pathways to the circular and longitudinal muscle of the guinea pig proximal colon. To identify excitatory and inhibitory muscle motoneurones, we applied the neuronal retrograde tracer DiI onto the circular or longitudinal muscle layer and performed additional immunohistochemistry for nitric oxide synthase (NOS) and choline acetyltransferase (ChAT). On average 166 +/- 81 circular muscle motoneurones (CMMN) and 100 +/- 74 longitudinal muscle motoneurones (LMMN) were labelled by DiI tracing. Myenteric pathways innervating the muscle were either ascending (DiI-labelled neurones with oral projections) or descending (DiI-labelled neurones with anal projections). The circular muscle was preferentially innervated by ascending pathways (66.0 +/- 9.1%). Most ascending CMMN were ChAT-positive (87.2 +/- 8.5%), whereas descending CMMN were mainly NOS-positive (82.3 +/- 14.6%). Most ascending (62.2 +/- 11.1%) and descending (82.0 +/- 12.5%) CMMN had circumferential projection preferences (circumferential projections were longer than projections along the longitudinal gut axis). In contrast to the polarised projections to the circular muscle, the longitudinal muscle was equally innervated by ascending (46.2 +/- 15.1%) and descending (53.9 +/- 15.1%) neurones. Ascending and descending pathways to the longitudinal muscle consisted predominantly of ChAT-positive neurones (98.1 +/- 1.9% and 68.0 +/- 8.5%, respectively), and both pathways had prominent longitudinal projection preferences. Only 25.5% of the descending LMMN were NOS-positive. In conclusion, the circular muscle in the proximal colon is innervated by descending inhibitory (NOS-positive neurones) and ascending excitatory (ChAT-positive neurones) pathways. In contrast, the longitudinal muscle is primarily innervated by ascending and descending excitatory motoneurones, and only a small proportion of the descending pathway consisted of inhibitory motoneurones.     

Development and Validation of a New Method to Measure Walking Speed in Free-Living Environments Using the Actibelt? Platform

Walking speed is a fundamental indicator for human well-being. In a clinical setting, walking speed is typically measured by means of walking tests using different protocols. However, walking speed obtained in this way is unlikely to be representative of the conditions in a free-living environment. Recently, mobile accelerometry has opened up the possibility to extract walking speed from long-time observations in free-living individuals, but the validity of these measurements needs to be determined. In this investigation, we have developed algorithms for walking speed prediction based on 3D accelerometry data (actibelt®) and created a framework using a standardized data set with gold standard annotations to facilitate the validation and comparison of these algorithms. For this purpose 17 healthy subjects operated a newly developed mobile gold standard while walking/running on an indoor track. Subsequently, the validity of 12 candidate algorithms for walking speed prediction ranging from well-known simple approaches like combining step length with frequency to more sophisticated algorithms such as linear and non-linear models was assessed using statistical measures. As a result, a novel algorithm employing support vector regression was found to perform best with a concordance correlation coefficient of 0.93 (95%CI 0.92–0.94) and a coverage probability CP1 of 0.46 (95%CI 0.12–0.70) for a deviation of 0.1 m/s (CP2 0.78, CP3 0.94) when compared to the mobile gold standard while walking indoors. A smaller outdoor experiment confirmed those results with even better coverage probability. We conclude that walking speed thus obtained has the potential to help establish walking speed in free-living environments as a patient-oriented outcome measure.     

Load along the tibial shaft during activities of daily living

External load at the tibia during activities of daily living provides baseline measures for the improvement of the design of the bone–implant interface for relevant internal and external prostheses. A motion analysis system was used together with an established protocol with skin markers to estimate three-dimensional forces and moments acting on ten equidistant points along the tibial shaft. Twenty young and able-bodied volunteers were analysed while performing three repetitions of the following tasks: level walking at three different speeds, in a straight-line and with sudden changes of direction to the right and to the left, stair ascending and descending, squatting, rising from a chair and sitting down. Moment and force patterns were normalised to the percentage of body weight per height and body weight, respectively, and then averaged over all subjects for each point, about the three tibial anatomical axes, and for each task. Load patterns were found to be consistent over subjects, but different among the anatomical axes, tasks and points. Generally, moments were higher in the medio/lateral axis and influenced by walking speed. In all five walking tasks and in ascending stairs with alternating feet, the more proximal the point was the smaller the mean moment was. For the remaining tasks the opposite trend was observed. The overall largest value was observed in the medio/lateral direction at the ankle centre in level walking at high speed (9.1% body weight * height on average), nearly three times larger than that of the anterior/posterior axis (2.9) during level walking with a sidestep turn. The present results should be of value also for in-vitro mechanical tests and finite element models.     

How do low horizontal forces produce disproportionately high torques in human locomotion?

Although horizontal ground forces are only approximately 15% of vertical forces, they account for 47% and 33% of the metabolic cost in walking and running. To explain these disproportionately high metabolic costs, we hypothesized that low horizontal ground forces generate relatively high torques on body segments during locomotion and this is mediated by long moment arms. We compared external force moment arms and discreet torques applied to the body segments by horizontal and vertical forces during walking and running. Sixteen subjects (21.9+/-1.9 years) walked at 1.5m/s and ten subjects (23.2+/-2.0 years) ran at 3.83 m/s. Segmental torques in the sagittal plane were partitioned into components due to horizontal and vertical forces and quantified by their angular impulses. The mean (+/-S.E.) ratios of horizontal to vertical ground forces (GF ratio) and angular impulses (AI ratio) in walking were 0.131 (+/-0.003, 95% confidence interval (CI) 0.124-0.137) and 0.530 (+/-0.018, CI 0.497-0.569). Results were similar in running. In both gaits the AI ratios were significantly greater than the GF ratios because the respective CI's did not overlap. The horizontal forces produced 53% and 41% as much angular impulse on the body segments, as did the vertical forces in walking and running despite being only 13% as large. In the two movements the moment arms for the horizontal forces averaged across foot, leg, thigh, and trunk body segments were 3.8 fold larger than those for the vertical forces. The data supported the hypothesis and suggest that the relatively low horizontal vs. vertical forces accounted for a disproportionately higher percentage of the angular impulses placed on the body segments and this effect was due to relatively long moment arms for horizontal forces. These results partially explain the relatively large metabolic cost of generating relatively low horizontal forces.     

Pygmy locomotion

The hypothesis that Pygmies may differ from Caucasians in some aspects of the mechanics of locomotion was tested. A total of 13 Pygmies and 7 Caucasians were asked to walk and run on a treadmill at 4–12 km · h^–1. Simultaneous metabolic measurements and three-dimensional motion analysis were performed allowing the energy expenditure and the mechanical external and internal work to be calculated. In Pygmies the metabolic energy cost was higher during walking at all speeds (P < 0.05), but tended to be lower during running (NS). The stride frequency and the internal mechanical work were higher for Pygmies at all walking (P < 0.05) and running (NS) speeds although the external mechanical work was similar. The total mechanical work for Pygmies was higher during walking (P < 0.05), but not during running and the efficiency of locomotion was similar in all subjects and speeds. The higher cost of walking in Pygmies is consistent with the allometric prediction for smaller subjects. The major determinants of the higher cost of walking was the difference in stride frequency (+9.45, SD 0.44% for Pygmies), which affected the mechanical internal work. This explains the observed higher total mechanical work of walking in Pygmies, even when the external component was the same. Most of the differences between Pygmies and Caucasians, observed during walking, tended to disappear when the speed was normalized as the Fronde number. However, this was not the case for running. Thus, whereas the tested hypothesis must be rejected for walking, the data from running, do indeed suggest that Pygmies may differ in some aspects of the mechanics of locomotion.     

Tuataras and salamanders show that walking and running mechanics are ancient features of tetrapod locomotion

The lumbering locomotor behaviours of tuataras and salamanders are the best examples of quadrupedal locomotion of early terrestrial vertebrates. We show they use the same walking (out-of-phase) and running (in-phase) patterns of external mechanical energy fluctuations of the centre-of-mass known in fast moving (cursorial) animals. Thus, walking and running centre-of-mass mechanics have been a feature of tetrapods since quadrupedal locomotion emerged over 400 million years ago. When walking, these sprawling animals save external mechanical energy with the same pendular effectiveness observed in cursorial animals. However, unlike cursorial animals (that change footfall patterns and mechanics with speed), tuataras and salamanders use only diagonal couplet gaits and indifferently change from walking to running mechanics with no significant change in total mechanical energy. Thus, the change from walking to running is not related to speed and the advantage of walking versus running is unclear. Furthermore, lumbering mechanics in primitive tetrapods is reflected in having total mechanical energy driven by potential energy (rather than kinetic energy as in cursorial animals) and relative centre-of-mass displacements an order of magnitude greater than cursorial animals. Thus, large vertical displacements associated with lumbering locomotion in primitive tetrapods may preclude their ability to increase speed.     

Maximal oxygen uptake during treadmill walking and running at various speeds.

Three groups of male subjects, average fitness (AF, N = 12), high fitness (HF, N = 7) and highly fit competitive race walkers (CRW, N = 3) performed maximal treadmill tests walking at 3.5 and 4.5 mph and running at 4.5, 5.5, 7.0, and 8.5 mph. In addition, the HF group performed a running test at 10.0 mph and the CRW group performed a walking test at 5.5 mph. All maximal oxygen uptake (VO2 max) tests with the exception of the 3.5 mph walking test (modified Balke test) were discontinuous in nature. VO2 max obtained from walking tests was similar regardless of speed within each group. Walking VO2 max was significantly lower than running VO2 max which was found to be similar over a speed range of 4.5 to 8.5 mph in the AF group. Running at 4.5 mph (HF group) and 4.5 and 5.5 mph (CRW group) resulted in lower VO2 max levels than running at speeds greater than or equal to 7.0 mph. Associated physiological variables (heart rate, ventilation, and respiratory exchange ratio) did not demonstrate a discernable pattern with reference to mode of locomotion (walking versus running) or speed. It was concluded that VO2 max elicited during walking is independent of speed and less than VO2 max obtained during running. Running VO2 max was interrelated with speed of running and state of training.     

Effects of order of presentation of exercise intensities and of sauna baths on perceived exertion during treadmill running

Thirteen male subjects performed a running test on the treadmill consisting of four standard exercise intensities [65%, 75%, 85%, 95% maximal O2 uptake (VO2 max)] presented in ascending, descending or random order. At the end of each exercise intensity, O2 consumption, heart rate (fc), venous blood lactate concentration [( Ia]b) and perceived exertion were assessed. This last variable was determined according to the Borg nonlinear CR-20 scale. The same variables were also determined during exercise at a standard intensity (65% or 95% VO2 max) performed before and after a Finnish sauna bath. Ratings of perceived exertion showed a good test-retest reliability (r = 0.77); they were the same when the exercise intensity was expressed in relative (%VO2 max) or absolute (speed) terms, and were independent of the order of presentation of the exercise. The latter had no effect on fc either but it did, however, influence [Ia]b, which was significantly higher in the descending, as compared to the ascending or random modes of presentation. The sauna bath increased fc at a given exercise intensity, but left perceived exertion and [Ia]b unchanged. It was concluded that at least under the present experimental conditions, fc and venous [Ia]b do not play a major role as determinants of perceived exertion.     

The energy cost of walking or running on sand

Oxygen uptake (VO2) at steady state, heart rate and perceived exertion were determined on nine subjects (six men and three women) while walking (3-7 km.h-1) or running (7-14 km.h-1) on sand or on a firm surface. The women performed the walking tests only. The energy cost of locomotion per unit of distance (C) was then calculated from the ratio of VO2 to speed and expressed in J.kg-1.m-1 assuming an energy equivalent of 20.9 J.ml O2-1. At the highest speeds C was adjusted for the measured lactate contribution (which ranged from approximately 2% to approximately 11% of the total). It was found that, when walking on sand, C increased linearly with speed from 3.1 J.kg-1.m-1 at 3 km.h-1 to 5.5 J.kg-1.m-1 at 7 km.h-1, whereas on a firm surface C attained a minimum of 2.3 J.kg-1.m-1 at 4.5 km.h-1 being greater at lower or higher speeds. On average, when walking at speeds greater than 3 km.h-1, C was about 1.8 times greater on sand than on compact terrain. When running on sand C was approximately independent of the speed, amounting to 5.3 J.kg-1.m-1, i.e. about 1.2 times greater than on compact terrain. These findings could be attributed to a reduced recovery of potential and kinetic energy at each stride when walking on sand (approximately 45% to be compared to approximately 65% on a firm surface) and to a reduced recovery of elastic energy when running on sand.     

Neural Basis of Stimulus-Angle-Dependent Motor Control of Wind-Elicited Walking Behavior in the Cricket Gryllus bimaculatus

Crickets exhibit oriented walking behavior in response to air-current stimuli. Because crickets move in the opposite direction from the stimulus source, this behavior is considered to represent ‘escape behavior’ from an approaching predator. However, details of the stimulus-angle-dependent control of locomotion during the immediate phase, and the neural basis underlying the directional motor control of this behavior remain unclear. In this study, we used a spherical-treadmill system to measure locomotory parameters including trajectory, turn angle and velocity during the immediate phase of responses to air-puff stimuli applied from various angles. Both walking direction and turn angle were correlated with stimulus angle, but their relationships followed different rules. A shorter stimulus also induced directionally-controlled walking, but reduced the yaw rotation in stimulus-angle-dependent turning. These results suggest that neural control of the turn angle requires different sensory information than that required for oriented walking. Hemi-severance of the ventral nerve cords containing descending axons from the cephalic to the prothoracic ganglion abolished stimulus-angle-dependent control, indicating that this control required descending signals from the brain. Furthermore, we selectively ablated identified ascending giant interneurons (GIs) in vivo to examine their functional roles in wind-elicited walking. Ablation of GI8-1 diminished control of the turn angle and decreased walking distance in the initial response. Meanwhile, GI9-1b ablation had no discernible effect on stimulus-angle-dependent control or walking distance, but delayed the reaction time. These results suggest that the ascending signals conveyed by GI8-1 are required for turn-angle control and maintenance of walking behavior, and that GI9-1b is responsible for rapid initiation of walking. It is possible that individual types of GIs separately supply the sensory signals required to control wind-elicited walking.     

Limited Transfer of Newly Acquired Movement Patterns across Walking and Running in Humans

The two major modes of locomotion in humans, walking and running, may be regarded as a function of different speed (walking as slower and running as faster). Recent results using motor learning tasks in humans, as well as more direct evidence from animal models, advocate for independence in the neural control mechanisms underlying different locomotion tasks. In the current study, we investigated the possible independence of the neural mechanisms underlying human walking and running. Subjects were tested on a split-belt treadmill and adapted to walking or running on an asymmetrically driven treadmill surface. Despite the acquisition of asymmetrical movement patterns in the respective modes, the emergence of asymmetrical movement patterns in the subsequent trials was evident only within the same modes (walking after learning to walk and running after learning to run) and only partial in the opposite modes (walking after learning to run and running after learning to walk) (thus transferred only limitedly across the modes). Further, the storage of the acquired movement pattern in each mode was maintained independently of the opposite mode. Combined, these results provide indirect evidence for independence in the neural control mechanisms underlying the two locomotive modes.     

Isometric muscle length-tension curves do not predict angle-torque curves of human wrist in continuous active movements

In this study we tested the hypothesis that during steady contractions of human wrist extensors or flexors, the torque-angle relationship during movements imposed about the wrist is predicted by the classical isometric muscle length-tension curve, with ascending, descending and ascending limbs. Angle-torque relationships were measured during steady muscle activation (10% of maximal voluntary contraction: MVC), elicited either by electrical stimulation or voluntary regulation of the electromyogram (EMG). Flexion-extension movements of constant speed (+/-10 degrees /s) were imposed on the subjects' hands with a servo actuator, either through the full physiological range of motion +/-50 degrees, or through +/-10 degrees. During extensor contractions, angle-torque curves in +/-50 degrees movements had ascending, descending and ascending limbs, as in isometric contractions. However, in +/-10 degrees movements, torque always increased with increasing muscle length and decreased with decreasing length, even over angles corresponding to the descending limb of isometric curves. For flexor activation, angle-torque curves had similar properties, though descending limbs were less obvious or absent. During imposed movements, hysteresis was observed in the angle-torque curves. This was attributed to non-linearities of the active muscles. Hysteresis reached a maximum at intermediate wrist angles and declined at maximal muscle length, contradicting the recent hypothesis that sarcomere non-uniformity is responsible for the hysteresis. We conclude that the classical isometric length-tension curve, with its prominent descending limb, does not predict angle-torque curves of human wrist muscles in continuous movements. A more appropriate model is one in which stiffness about the wrist is always positive and hysteresis is a significant factor.     

In vitro fatigue failure of cemented acetabular replacements: a hip simulator study

Although hip simulators for in vitro wear testing of prosthetic materials used in total hip arthroplasty (THA) have been available for a number of years, similar equipment has yet to appear for endurance testing of fixation in cemented THA, despite considerable evidence of late aseptic loosening as one of the most significant failure mechanisms in this type of replacements. An in vitro study of fatigue behavior in cemented acetabular replacements has been carried out, utilizing a newly developed hip simulator. The machine was designed to simulate the direction and the magnitude of the hip contact force under typical physiological loading conditions, including normal walking and stair climbing, as reported by Bergmann et al. (2001, Hip 98, Freie Universitaet, Berlin). A 3D finite element analysis has been carried out to validate the function of the hip simulator and to evaluate the effects of boundary conditions and geometry of the specimen on the stress distribution in the cement mantle. Bovine pelvic bones were implanted with a Charnley cup, using standard manual cementing techniques. Experiments were carried out under normal walking and descending stairs loading conditions with selected load levels from a body weight of 75-125 kg. Periodically, the samples were removed from the test rigs to allow CT scanning for the purpose of monitoring damage development in the cement fixation. The hip simulator was found to be satisfactory in reproducing the hip contact force during normal walking and stair climbing, as reported by Bergmann et al. Finite element analysis shows that the stress distributions in the cement mantle and at the bone-cement interface are largely unaffected by the geometry and the boundary conditions of the model. Three samples were tested up to 17 x 10(6) cycles and sectioned post-testing for microscopic studies. Debonding at the bone-cement interface of various degrees in the posterior-superior quadrant was revealed in these samples, and the location of the failures corresponds to the highest stressed region from the finite-element analysis. Preliminary experimental results from a newly developed hip simulator seem to suggest that debonding at the bone-cement interface is the main failure mechanism in cemented acetabular replacements, and descending stairs seem to be more detrimental than normal walking or ascending stairs with regard to fatigue integrity of cement fixation.     

The physiological cost of carrying light and heavy loads

Nine subjects walked on a treadmill with load weights equal to 10% and 40% of body weight carried on the back. Although the speed of the treadmill was selected so that the measured oxygen consumption (VO2) was the same for both load conditions, the heavier load placed an extra strain on the cardiopulmonary system and was perceived by all subjects as harder work than the lighter load. When the subjects worked at their own pace, walking on a level road or climbing stairs with load weights equal to 10% and 40% of body weight, they compensated for the heavier load by decreasing walking speed or climbing rate. Although the energy costs calculated from walking speed, body and load weight for self-paced walking and the external work of stair climbing were the same for both load conditions, the heavier load was again perceived as harder work. These findings are discussed as they relate to the definition of acceptable load weights.