Normal Gait

A major contribution to the mechanical analysis of walking, also from the California group, was made by Bresler and Frankel (1950). They performed free-body calculations for the hip, knee and ankle joints, allowing for ground reaction forces, the effects of gravity on the limb segments and the inertial forces. The analytical techniques developed by these workers formed the basis of many current methods of modelling and analysis.

An important paper describing the possible mechanisms which the body uses to minimise energy consumption in walking was published by Saunders et al. (1953). Further important work on energy consumption and, in particular, the energy transfers between the body segments in walking was published by Cavagna and Margaria (1966). In the 1960s, research also began to concentrate on the variability of walking, the development of gait in children and the deterioration of gait in old age. Patricia Murray published a series of papers on these subjects, including a detailed review (Murray, 1967).

Once the motions of the body segments and the actions of the different muscles had been examined and documented, attention moved to the forces generated across the joints. Although limited calculations of this type had been made previously, the study by Paul (1965) was the first detailed analysis of hip joint forces during walking. A subsequent paper by Paul also included an analysis of the forces in the knee (Paul, 1966). Since then, there have been many mathematical studies of force generation and transmission across the hip, knee and ankle.

The 1970s and 1980s saw great improvements in methods of measurement. The development of more convenient kinematic systems, based on electronics rather than photography, meant that results could be produced in minutes rather than days. Reliable force platforms with high-frequency response became available, as well as convenient and reliable EMG systems. The availability of high-quality three-dimensional data on the kinetics and kinematics of walking, and the ease of access to powerful computers, made it possible to develop increasingly sophisticated mathematical models. Gait laboratories now routinely measure joint moments and powers for the hip, knee and ankle, and estimates can also be made of muscle, ligament and joint contact forces.

Recent decades have seen the emergence of increasingly powerful systems, higher-speed cameras, greater portability, smaller markers, a wide variety of marker sets and larger data collection volumes.

Fig. 2.2 shows the timings of initial contact and toe off for both feet during a little more than one gait cycle. Right initial contact occurs whilst the left foot is still on the ground and there is a period of double support (also known as double limb stance) between initial contact on the right and toe off on the left. During the swing phase on the left side, only the right foot is on the ground, giving a period of right single support (or single limb stance), which ends with initial contact by the left foot. There is then another period of double support, until toe off on the right side. Left single support corresponds to the right swing phase, and the cycle ends with the next initial contact on the right.

In each double support phase, one foot is forwards, having just landed on the ground, and the other one is backwards, being just about to leave the ground. When it is necessary to distinguish between the two legs in the double support phase, the leg in front is usually known as the leading leg and the leg behind as the trailing leg. The leading leg is in loading response, sometimes referred to as braking double support, initial double support or weight acceptance. The trailing leg is in pre-swing, also known as second, terminal or thrusting double support, or as weight release.

In each gait cycle, there are thus two periods of double support and two periods of single support. The stance phase usually lasts about 60% of the cycle, the swing phase about 40% and each period of double support about 10%. However, this varies with the speed of walking, with the swing phase becoming proportionately longer and the stance phase and double support phases shorter as the speed increases (Murray, 1967); see Blanc et al. (1999) for a detailed study of gait cycle timing. The final disappearance of the double support phase marks the transition from walking to running. Between successive steps in running there is a flight phase, also known as the float, double-float or nonsupport phase, when neither foot is on the ground. For more detail, this book now includes a chapter considering gait analysis of running (Chapter 8, Gait Analysis in Running and the Management of Common Injuries).

The terms used to describe the placement of the feet on the ground are shown in Fig. 2.3. The stride length is the distance between two successive placements of the same foot. It consists of two step lengths, left and right, each the distance by which the named foot moves forwards in front of the other. In pathological gait, it is common for the two step lengths to be different. If the left foot is moved forwards to take a step and the right one is brought up beside it rather than in front of it, the right step length will be zero. It is even possible for the step length on one side to be negative; for example, if the left foot never catches up with the right foot, the distance between the left and right feet will be negative. However, the stride length starting with the left heelstrike must always be the same as the stride length starting with the right heelstrike, unless the subject is walking around a curve where the inside leg will have a shorter stride length than the outside leg. This definition of a stride, consisting of one step by each foot, breaks down in some pathological gaits in which one foot makes a series of hopping movements whilst the other is in the air (Wall et al., 1987). There is no satisfactory nomenclature to deal with this situation.

The walking base (also known as the stride width or base of support) is the side-to-side distance between the line of the two feet, usually measured at the midpoint of the back of the heel but sometimes below the centre of the ankle joint. The preferred unit for stride length and step length is the metre and for the walking base, millimetres. The pattern of walking known as tandem gait involves walking with the heel of one foot placed directly in front of the toes of the other; that is, with a walking base close to zero. Although this pattern is not typically seen, even as a pathological gait it requires good balance and coordination.

It is obvious that you need to walk more carefully on ice than on asphalt. Whether or not the foot slips during walking depends on two things: the coefficient of friction between the foot and the ground, and the relationship between the vertical force and the forces parallel to the walking surface (front to back and side to side). The ratio of the horizontal to the vertical force is known as the utilised coefficient of friction, and slippage will occur if this exceeds the actual coefficient of friction between the foot and the ground. In normal walking, a coefficient of friction of 0.35 to 0.40 is generally sufficient to prevent slippage; the most hazardous part of the gait cycle for slippage is initial contact. The literature on foot-to-ground friction and slippage is fairly extensive; examples include Cham and Redfern (2002) and Burnfield et al. (2005).

The cadence is the number of steps taken in a given time period, the usual unit being steps per minute. In most other types of scientific measurement, complete cycles are counted, but as there are two steps in a single gait cycle, the cadence is a measure of half-cycles. The normal ranges for both cadence and cycle time in both genders at different ages are shown in Table 2.1, where we consider the effect of age in more detail.

The speed of walking is the distance covered by the whole body in a given time, and should be measured in metres per second. Many authors use the term velocity in place of speed, but this is an incorrect usage of the term, unless the direction of walking is also stated, since velocity is a vector. The instantaneous speed varies from one instant to another during the walking cycle, but the average speed is the product of the cadence and the stride length, providing appropriate units are used. The cadence, in steps per minute, corresponds to half-strides per 60 seconds or full strides per 120 seconds. The speed can thus be calculated from cadence and stride length using the formula:

speed (m/s)=stride length (m)×cadence (steps/min)/120

If cycle time is used in place of cadence, the calculation becomes much more straightforward:

speed (m/s)=stride  length (m)/cycle time (s)

The walking speed thus depends on the two step lengths, which in turn depend to a large extent on the duration of the swing phase on each side. The step length is the amount by which the foot can be moved forwards during the swing phase so that a short swing phase on one side will generally reduce the step length on that side. In pathological gait, the step length is often shortened, but it behaves in a way which is counterintuitive. When pathology affects one foot more than the other, an individual will usually try to spend a shorter time on the ‘bad’ foot and a correspondingly longer time on the ‘good’ one. Shortening the stance phase on the bad foot means bringing the good foot to the ground sooner, thereby shortening both the duration of the swing phase and the step length on that side. Thus a short step length on one side generally means problems with single support on the other side.

When making comparisons among individuals, particularly children, it is useful to allow for differences in size. This is done by dividing a measurement by some aspect of body size, such as height (stature) or leg length, a procedure generally known as normalisation. It is thus fairly common to see walking speed expressed in statures per second or to see measures such as step factor, which is step length divided by leg length (Sutherland, 1997).

Since walking speed depends on both cadence and stride length, it follows that speed may be changed by altering only one of these variables—for instance, by increasing the cadence whilst keeping the stride length constant. In practice, however, people normally change their walking speed by adjusting both cadence and stride length. Sekiya and Nagasaki (1998) defined the walk ratio as step length (m) divided by step rate (steps/min) and found that it was fairly constant in both males and females over a range of walking speeds from very slow to very fast. Macellari et al. (1999) conducted a detailed study of the relationships among gender, body size, walking speed, gait timing and foot placement.

The upper body moves forwards throughout the gait cycle. Its speed varies a little, being fastest during the double support phases and slowest in the middle of the stance and swing phases. The trunk twists about a vertical axis, with the shoulder girdle rotating in the opposite direction to the pelvis. The arms swing out of phase with the legs so that the left leg and the left side of the pelvis move forwards at the same time as the right arm and the right side of the shoulder girdle. Lamoth et al. (2002) made a detailed study of the relative motion between the pelvis and the trunk at different walking speeds. Murray (1967) found average total excursions of 7 degrees for the shoulder girdle and 12 degrees for the pelvis in adult males walking at free speed. The fluidity and efficiency of walking depend to some extent on the motions of the trunk and arms, but these movements are commonly ignored in clinical gait analysis and have been relatively neglected in gait research. The whole trunk rises and falls twice during the cycle, through a total range of about 46 mm (Perry, 1992), being lowest during double support and highest in the middle of the stance and swing phases. An approximation to this vertical motion can be seen in the position of the hip joint in Fig. 2.4. The trunk also moves from side to side, once in each cycle, with the trunk being over each leg during its stance phase, as might be expected from the need for support. The total range of side-to-side movement is also about 46 mm (Perry, 1992). The pelvis twists about a vertical axis and tips slightly, both backwards and forwards (with an associated change in lumbar lordosis) and from side to side. The spinal muscles are selectively activated so that the head moves less than the pelvis, which is important for providing a stable platform for vision (Prince et al., 1994).

The hip flexes and extends once during the cycle (Fig. 2.5). The limit of flexion is reached around the middle of the swing phase, and the hip is then kept flexed until initial contact. The peak extension is reached before the end of the stance phase, after which the hip begins to flex again.

Initial Contact (Fig. 2.11)

Loading Response (Fig. 2.12)

Opposite Toe Off (Fig. 2.13)

Mid-stance (Fig. 2.14)

Heel Rise (Fig. 2.15)

Opposite Initial Contact (Fig. 2.16)

Toe Off (Fig. 2.17)

Feet Adjacent (Fig. 2.18)

Tibia Vertical (Fig. 2.19)

Terminal Foot Contact (Fig. 2.11)

The gait cycle ends at the next initial contact of the same foot (in this case, the right foot). Because it is confusing to refer to the end of the cycle as initial contact, it is sometimes known as terminal foot contact.

Winter (1980) coined the term support moment to describe the sum of the sagittal plane moments about the hip, knee and ankle joints:

MS=MH+MK+MA    Winter (1980)

where MS, MH, MK and MA are the support, hip, knee and ankle moments, respectively.

Winter noted that the support moment was far less variable than its individual components, suggesting that a decreased moment about one joint could be compensated for by an increased moment about one or both of the other joints. However, it was difficult to interpret this in biomechanical terms, since the sign convention was based on flexion and extension rather than on clockwise and anticlockwise moments, which caused the direction of the knee moment to be opposite to that of the hip and ankle moments. Hof (2000) published a justification for the support moment and suggested that it is responsible for preventing collapse of the knee. Based on his analysis, he suggested the following revised formula for its calculation:

MS=½ MH+MK+½ MA  Hof (2000)

Fig. 2.24 illustrates the support moment calculated from the sagittal plane hip, knee and ankle internal moments from the normal subject used for illustration throughout this chapter, using the formula suggested by Hof (2000). Anderson and Pandy (2003) suggested that a better alternative to the support moment would be the sum of the vertical components of force from individual muscles during walking.

Inman et al. (1981) and Rose and Gamble (1994) quoted an equation, based on several studies, for the relationship between walking speed and energy consumption per unit time. Energy consumption included both the basal metabolism and the overheads. They showed, not surprisingly, that energy consumption per unit time is less for slow walking than for fast walking. Translating their equation into SI units, it becomes:

Ew=2.23+1.26 v2

where E_w is the energy consumption in watts per kilogram of body mass and v is the speed in m/s.

As an example of the application of this equation, a 70-kg person walking at 1.4 m/s, which is a typical speed for adults, would consume energy at a rate of 330 W. The term v² in the equation shows that energy consumption increases as the square of the walking speed.

Energy Consumption per Unit Distance (Oxygen Cost)

The energy consumption per metre walked, also known as energy cost, has a less straightforward relationship with walking speed, as both very slow and very fast walking speeds use more energy per metre than intermediate walking speeds. The equation describing this relationship, again converted into SI units, is:

Em=2.23/v+1.26 v

where E_m is the energy consumption in joules per metre per kilogram of body mass and v is the speed in m/s. The energy cost of walking is higher in children and decreases steadily with age up to adulthood.

Minimum energy usage is predicted by this equation at a speed of 1.33 m/s. A 70-kg person walking at this speed would use 235 J/m, or 235 kJ/km. A typical candy bar contains around 1000 kJ and would thus supply enough energy to walk 4.26 km, or more than 2.5 miles!

The preceding equations merely give average values for adults, which may be modified by age, gender, walking surface, footwear and so on. Pathological gait is frequently associated with an energy consumption which is considerably above average values, due to some combination of abnormal movements, muscle spasticity and co-contraction of antagonistic muscles. To provide a baseline for studies of pathological gait, Waters et al. (1988) made a detailed study of the energy consumption of 260 normal children and adults of both genders, walking at a variety of speeds.

Winter et al. (1976) studied the energy levels of the limb segments and of the quaintly named HAT (head, arms and trunk). These authors criticised some earlier studies that had included the kinetic energy of linear motion but had neglected the kinetic energy due to rotation, which is responsible for about 10% of the total energy of the shank. Winter et al. studied only the sagittal plane, regarding energy exchanges in the other planes as negligible. They confirmed the exchange between potential and kinetic energy, described earlier, and estimated that roughly half of the energy of the HAT segment was conserved in this way. The thigh conserved about one-third of its energy by exchanges of this sort and the shank virtually none. They also noted that the changes in total body energy were less than the changes in energy of the individual segments, indicating a transfer of energy from one segment to another. In one subject, during a single gait cycle, the energy changes were shank 16 J, thigh 6 J and HAT 10 J, for a total of 32 J. However, the total body energy change was only 22 J, indicating a saving of 10 J by intersegment transfers. Siegel et al. (2004) performed a detailed analysis on the relationship between lower limb joint moments and mechanical energy during gait.

If the knee is kept straight, a movement of the hip from a flexed position to an extended one, such as occurs in the stance phase of gait, will result in the centre of mass of the body moving forwards, but also in its rising and then falling again. The amount of forward movement and the amount of rising and falling both depend on the total angle through which the hip joint moves from flexion to extension (Fig. 2.25A). Since the forward movement is equal to the stride length, it follows that the greater the stride length, the greater will be the angles of flexion and extension of the hip and the more the centre of mass will move vertically between its highest and lowest positions. The first determinant of gait is the way in which the pelvis twists about a vertical axis during the gait cycle, bringing each hip joint forwards as that hip flexes and backwards as it extends. This means that for a given stride length, the hip joint itself moves forwards through a smaller distance than the foot so that less flexion and extension of the hip is required. A proportion of the stride length thus comes from the forward-and-backward movement of the hip joint. The reduction in the range of hip flexion and extension leads to a reduction in the vertical movement of the hip (Fig. 2.25B).

As described in the preceding section, flexion and extension of the hip are accompanied by a rise and fall in the height of the hip joint. If the pelvis were to remain level, the trunk would follow this up-and-down movement. However, the second determinant of gait is the way the pelvis tips about an anteroposterior axis, raising first one side and then the other so that when the hip of the stance phase leg is at its highest point, the pelvis slopes downwards, resulting in the hip of the swing phase leg being lower than that of the stance phase leg. Since the height of the trunk does not depend on the height of either hip joint alone but on the average of the two of them, this pelvic obliquity reduces the total vertical excursion of the trunk (Fig. 2.26). However, it can only be achieved if the swing phase leg can be shortened sufficiently to clear the ground (normally by both flexing the knee and dorsiflexing the ankle), when the height of its hip joint is reduced.

The third, fourth and fifth determinants of gait (Fig. 2.27) are concerned with adjusting the effective length of the leg, by lengthening it at the beginning and end of the stance phase and shortening it in the middle, to keep the hip height as constant as possible. The third determinant is the stance phase flexion of the knee. As the femur passes from flexion of the hip into extension, if the leg remained straight, the hip joint would rise and then fall, as described earlier. However, flexion of the knee shortens the leg in the middle of this movement, reducing the height of the apex of the curve.

Complementary to the way in which the apex of the curve is lowered by shortening the leg in the middle of the movement from hip flexion to extension, the beginning of the curve is elevated by lengthening the leg at the start of the stance phase, or initial contact. This is achieved by the fourth determinant of gait, or the ankle mechanism. Because the heel sticks out behind the ankle joint, it effectively lengthens the leg during the loading response (Fig. 2.27).

In the same way that the heel lengthens the leg at the start of the stance phase, the forefoot lengthens it at the end of the stance phase, in the fifth determinant, or the terminal rocker (Fig. 2.27). From the time of heel rise, the effective length of the lower leg increases as the ankle moves from dorsiflexion into plantarflexion.

The first five determinants of gait are all concerned with reducing the vertical excursions of the centre of gravity. The sixth is concerned with side-to-side movement. If the feet were as far apart as the hips, the body would need to tip from side to side to maintain balance during walking (Fig. 2.28A). By keeping the walking base narrow, little lateral movement is needed to preserve balance(Fig. 2.28B). The reduction in lateral acceleration and deceleration leads to a reduction in the use of muscular energy. The main adaptation which allows the walking base to be narrow is a slight valgus angulation of the knee, which permits the tibia to be vertical whilst the femur inclines inwards, from a slightly adducted hip.

It should be obvious that although the six determinants of gait have been described separately, they are integrated during each gait cycle. The combined effect is a much smoother trajectory for the centre of gravity and (according to the original description) a much lower energy expenditure. According to Perry (1992), the determinants of gait reduce the vertical excursions of the trunk by about 50% and the horizontal excursions by about 40%.

Although a number of studies described the development of gait in children, Sutherland et al. (1988) is one of the most detailed. Following are the main ways in which the gait of small children differs from that of adults:

These differences in gait mature at different rates. The characteristics numbered (3), (4) and (5) in the preceding list have changed to the adult pattern by the age of 2, and (1) and (6) by the age of 4. The cycle time, stride length and speed continue to change with growth, reaching normal adult values around the age of 15.

Fig. 2.30, which is based on data from Sutherland et al. (1988), shows the average sagittal plane motion at the hip, knee and ankle joints in 49 children between 11 and 13 months of age. It should be compared with Fig. 2.5, which shows the same parameters for a normal adult female. Sutherland et al. only gave the timing of initial contact and toe off on the two sides and used a different definition of hip angle; the data in Fig. 2.30 have been adjusted into extension by 15 degrees to make them comparable with the other figures in this book.

The pattern of hip flexion and extension differs from that in adults in that the degree of extension is reduced and the hip does not remain flexed for so long at the end of the swing phase. The knee never fully extends, but this is seen at all ages in Sutherland’s data and may reflect the method of measurement. There is some stance phase knee flexion in infants, but it is both smaller in magnitude and earlier than in adults. The flexion of the knee in the swing phase is also somewhat reduced at the age of 1, and most adults have more swing phase flexion than is seen in Fig. 2.5.

Initial contact in small children is by the whole foot, with heelstrike being replaced by foot flat. The ankle is plantarflexed at initial contact and remains so into the early stance phase, in contrast to the adult pattern, in which the ankle is approximately neutral at initial contact but moves rapidly into plantarflexion. The pattern of dorsiflexion followed by plantarflexion through the remainder of the stance phase is essentially the same at all ages.

Since children are smaller than adults, it is not surprising that they walk with a shorter stride length and at a slower speed. Sutherland et al. (1988) showed that stride length is closely related to height and that the ratio of stride length to stature is similar to that found in adults. The change in stride length with age mirrors the change in height, showing a rapid increase up to age 4 and a slower increase thereafter. Todd et al. (1989) detailed the relationships between the height of children and their general gait parameters. Small children walk with a short cycle time (rapid cadence), the mean at the age of 1 being about 0.70 s (171 steps/min). Cycle time increases with age but is still around 0.85 s (141 steps/min) at age 7, which is well below the typical adult values of 1.06 s (113 steps/min) for males and 1.02 s (118 steps/min) for females. The shorter cycle time partly compensates for the short stride length and the speed ranges from 0.64 m/s at age 1 to 1.14 m/s at age 7, compared with the typical adult values of 1.46 m/s for males and 1.30 m/s for females. Sutherland did not report on the gait of children beyond the age of 7 and did not distinguish between the results from male and female children. Table 2.1 gives the normal ranges for the general gait parameters in children, derived in part from Sutherland’s data. However, values based on age alone may be misleading; stride length depends on height and walking speed, both of which may be lower in children with a disability than in normal children of the same age.

As can be seen in Fig. 2.30, the swing phase occupies a smaller proportion of the gait cycle in very small children than in adults, thus minimising the time spent in the less stable condition of single-legged stance. The relative duration of the swing phase increases with age, reaching the adult proportion around the age of 4 years. There is symmetry between the two sides at all ages. Sutherland et al. (1988) related the width of the walking base to the width of the body at the top of the pelvis, using the somewhat confusing ‘pelvic-span/ankle-spread’ ratio. Changing the measurement units for the sake of clarity, the walking base is about 70% of the pelvic width at the age of 1 year, falling to about 45% by the age of 3 1/2 years, at which level it remains until the age of 7. An average value for adults is not readily available, but it is probably less than 30%.

At the very youngest ages, the EMG patterns showed that there is a tendency to activate most muscles for a higher proportion of the gait cycle than in adults. With the exception of the triceps surae, adult patterns are established for most muscles by the age of 2. Sutherland et al. (1988) found that children could be divided into two groups depending on whether the triceps surae was activated in a prolonged (infant) pattern or the normal (adult) pattern. More than 60% of the children younger than 2 years of age showed the infant pattern; the proportion dropped to less than 30% by the age of 7. The authors speculated that this might relate to delayed myelination of the sensory branches of the peripheral nerves.

An excellent review of the main changes in gait occurring during childhood was given by Sutherland (1997). The gait of children younger than 2 was examined in detail by Grimshaw et al. (1998). The joint moments and powers of this same group were studied by Hallemans et al. (2005).

Cunha (1988) discussed the gait of the elderly and pointed out that many pathological gait disorders are incorrectly thought to be part of the normal ageing process. Identification of an underlying cause, which may be treatable, could result in improved quality of life for the patient and a reduced risk of falls and fractures. Cunha classified the causes of the gait disorders of old age as follows: neurological, psychological, orthopaedic, endocrinological, general, drugs, senile gait and associated conditions. He described the features of the gait in many conditions affecting the elderly and suggested a plan for the investigation and management of these patients.

A number of investigations have been made of the changes in gait which occur with advancing age, especially by Murray et al. (1969), who studied the gait of males up to the age of 87. The description which follows is confined to the effects of age on free-speed walking, although Murray et al. also examined fast walking. A companion paper (Murray et al., 1970) studied the gait of females up to age 70. It did not provide as much information on the effects of age, but generally confirmed the observations made on males.

The gait of elderly people is subject to two influences: the effects of age itself and the effects of pathological conditions, such as osteoarthritis and parkinsonism, which become more common with advancing age. Provided patients with pathological conditions are carefully excluded, the gait of elderly people appears to be simply a ‘slowed down’ version of the gait of younger adults. Murray et al. (1969) were careful to point out that ‘the walking performance of older men did not resemble a pathological gait’.

Typically, the onset of age-related changes in gait takes place in the decade from 60 to 70 years of age. There is a decreased stride length, a variable but generally increased cycle time (decreased cadence) and an increase in the walking base. Many other changes can also be observed, such as a relative increase in the duration of the stance phase as a percentage of the gait cycle, but most of them are secondary to the changes in stride length, cycle time and walking base. The speed (stride length divided by cycle time) is almost always reduced in elderly people. Table 2.1 gives normal ranges for the general gait parameters up to the age of 80.

Some of the differences between the gait of the young and the elderly are apparent in Fig. 2.31, which is taken from Murray et al. (1969). These authors suggested that the purpose of gait changes in the elderly is to improve the security of walking. Both decreasing the stride length and increasing the walking base make it easier to maintain balance whilst walking. Increasing the cycle time (reducing the cadence) leads to a reduction in the percentage of the gait cycle for which there is only single-limb support, since the increase in cycle length is largely achieved by lengthening the stance phase and hence the double support time.

Changes in the angular excursions of the joints in the elderly include a reduction in the total range of hip flexion and extension, a reduction in swing phase knee flexion and a reduction in ankle plantarflexion during the push off phase. However, all of these depend on both cycle time and stride length and are probably within normal limits if these factors are taken into account. Nigg et al. (1994) confirmed these observations in a detailed study on three-dimensional joint ranges of motion in walking, in male and female subjects from 20 to 79 years of age. The vertical movement of the head is reduced and its lateral movement increased, probably secondary to the changes in stride length and walking base, respectively.

The trajectory of the toe over the ground is modified in old age, giving an improved ground clearance during the first half of the swing phase. This is probably another mechanism for improving security. The heel rises less during pre-swing and the foot attitude is closer to the horizontal at initial contact, both of these changes being related to the reduction in stride length. There is also an increase in the angle of toe out in elderly people and changes in the posture and movements of the arms, the elbows being more flexed and the shoulders more extended. The reasons for these differences are not known.

The dividing line between normal and abnormal may be difficult to define in elderly people. A condition known as idiopathic gait disorder of the elderly has been described, which is essentially an exaggeration of the gait changes which normally occur with age and is characterised by a cautious attitude to walking, with a prolonged cycle time (low cadence), a short stride length and an increased step-to-step variability. For a comprehensive review of the changes in gait with advancing age, see Prince et al. (1997).