126
EQUINE EXERCISE PHYSIOLOGY 6
Equine vet. J., Suppl. 34 (2002) 126-130
The cost of transport in an extended trot
S. J. WICKLER, D. F. HOYT*, E.A. COGGER and R. MCGUIRE
Equine Research Center and Departments of Animal and Veterinary Science and *Biological Sciences,
California State Polytechnic University, Pomona, California 91768, USA.
Keywords: horse; locomotion; energetics; gait transition; stride frequency
Summary
We hypothesised that trotters during an extended trot have
lower energetic costs of locomotion (CT) than horses not bred
for this behaviour. VO2 was measured as a function of speed
in 7 Arabian horses (3 trained to extend their trotting speeds)
and in 2 horses, of similar mass, bred to trot (Hackney). Both
oxygen consumption and CT increased with speed and there
was, contrary to our hypothesis, no difference between
breeds. In Arabians at 6.5 m/s, CT had increased 25% above
the CT at 5.0 m/s (normal transition speed). For Hackneys at
6.8 m/s, the CT was almost 35% higher. Stride frequencies
increased linearly in all horses up to 5.0 m/s. At the canter at
5.0 m/s, the frequency increased 9% to 111 strides/min, but
then increased minimally with speed. In the Hackneys and
the Arabians that extended the trot, stride frequencies were
approximately 102 and did not increase with speed. Stride
length (SL) increased linearly with speed in both trotting
and cantering horses, and cantering SL were lower than
trotting (at 5.0 m/s, SL for trotting = 3.04 m and for
cantering SL = 2.68 m). There were no differences between
breeds in stride frequency or stride length.
Extending the trot can have profound energetic
requirements that could limit athletic performance and may
lead to increased concussive impact on the limbs.
Introduction
The relationship between metabolism and speed during steadystate locomotion has been the focus of many studies using a variety
of animals and this relationship is generally reported as linear
(Taylor et al. 1970; Taylor 1994). However, in the pony and horse,
careful analysis within a single gait demonstrates that the rate of
oxygen consumption (VO2) is a curvilinear function of speed (Hoyt
and Taylor 1981; Wickler et al. 2000). The importance of the
distinction between linear and curvilinear relationships lies in the
resulting cost of transport, CT, which is the metabolic expenditure
to move a given distance (Taylor et al. 1970). If oxygen
consumption increases with speed in a linear fashion, then CT (VO2
divided by speed) is constant and is independent of speed. If the
slope of VO2 vs. speed is curvilinear, then the CT calculated
produces a U-shaped relationship between CT and speed with a
minimum, i.e. a speed that is most energetically economical.
When given the choice on open, level ground and up an
incline, trotting ponies (Hoyt and Taylor 1981) and horses
(Wickler et al. 2000) chose a relatively narrow range of speeds (a
‘preferred speed’) in spite of the fact that, on a treadmill, these
animals trotted over a much larger range of speeds. This preferred
speed coincided with the speed at which CT was at a minimum.
As speed increases above the preferred speed, CT increases. In
ponies, it was observed that they made a transition to the gallop
(this was probably a 3-beat gait, or canter; however, biomechanists
do not always make this distinction) to minimise energetic costs
(Hoyt and Taylor 1981). We hypothesised that the energetic
consequence of continuing to trot beyond normal trotting speeds
(we will refer to this, in the context of this paper, as an extended
trot) is an elevated CT, and we hypothesised that CT would increase
progressively at even higher speeds. But what about horses which
have been bred or trained to extend their trot? We hypothesised that
trotters would be more economical and have a reduced CT
compared to breeds not selected for this ability. We compared
Arabian horses that normally make a trot-gallop transition with a
breed that has been selected for extending the trot, the Hackney
horse. In addition, we trained 3 Arabian horses to extend their trot
and then measured the metabolic consequences.
Materials and methods
The Institutional Animal Care and Use Committee approved this
study. The study population (Table 1) consisted of purebred
Arabians (n = 7) and Hackney horses (n = 2).
Horses were familiarised to exercise on a treadmill (Sato I)1
and to wear a loose-fitting mask. Training occurred for a minimum
of 3 months prior to any data collection. Horses were exercised
either on the treadmill or outside, for a minimum of 5 days a week
during this 3 month period. A heart rate monitor (Polar Advantage
XL)2 was used to track daily workouts and assess fitness. All
animals were free of lameness and were considered moderately fit.
The highest speed at which the horses would consistently
maintain a trot for 1 min was determined in the Arabian horses.
Treadmill speeds were incremented by 0.25 m/s at 1 min
intervals. Horses were tested by increasing and decreasing the
speed on different days. The highest speed at which the horses
trotted for 1 min was averaged for each horse (a minimum of 9
trials using increasing speeds and 9 trials using decreasing
speeds). Following determination of the highest trotting speed,
3 of the 7 horses were trained using voice commands to extend
the gait while on the treadmill. This training process involved
only positive reinforcement.
Measurements of VO2 were made using an open-flow system.
Air was drawn through a loose fitting mask at rates of
3,000–6,000 l/min. Excurrent airflow was monitored by measuring
the pressure drop (2110 Smart Pressure Gauge)3 across a 15 cm
orifice plate (304SS)3. A sample of excurrent gas was withdrawn
and CO2 (Ascarite)4 and H2O (Dreirite)5 were removed, prior to
determination of oxygen concentration (Ametek S3A/II O2
analyzer)6. The rate of oxygen consumption was calculated using
equation 4b of Withers (1977). Values were calculated as ml O2/g.h.
127
5
0.25
4
0.20
Cost of transport, ml O2/kg m
VO2 ml/g.h
S. J. Wickler et al.
3
2
0.15
0.10
0.05
1
0.00
0
1
2
3
4
Speed (m/s)
5
6
1
7
2
3
4
Speed (m/s)
5
6
7
Fig 1: Oxygen consumption vs. speed is a curvilinear function in
Arabians (●
● ) and Hackneys (▲
▲ ). X = the 3 Arabian horses trained to
extend their gait.
Fig 2: Cost of transport, CT, expressed per body mass, is a curvilinear
function of speed. For explanation of symbols, see Figure 1.
Calibration of flow rates, Ve, was performed daily using a nitrogenbleed technique method (Saltin and Rose 1994). Nitrogen flow was
measured with a flow meter (Rotameter, Model No.
R-8M-R5-4F)7 and was calibrated with a dry gas meter (Model No.
DTM-325)8. All gas volumes were corrected to standard
temperature and pressure. Steady state oxygen consumption was
defined as the lowest 2 min average over a 12 min period. Data
from the O2 analyser and the differential pressure transducer were
collected at 1 Hz using an acquisition system9. Tread speed was
measured continuously using the voltage output from a motor
attached to the treadmill axle. This output was manually calibrated
weekly by timing a minimum of 10 tread revolutions with a digital
stopwatch while the horse was on the treadmill.
A preliminary warm-up consisted of 3 min at 1.6 m/s and
5 min at 3.5 m/s on the level. For the Arabian horses, one 12 min
bout of exercise was assigned randomly daily for trotting speeds
ranging between 2.0 and 4.5 m/s (speeds at which there was no
transition to a gallop during that 12 min period) and between 2.0
and 6.75 m/s for the Hackneys. Only one speed was run per day
and speeds were randomised.
Stride frequency determinations were made after a minimum
of 30 s at any speed and 30 strides were counted manually and
timed with a digital stopwatch. Two observers made a minimum
of 2 counts each and times were compared. If the values between
observers or observations differed by more than 5%, another
count was taken. For comparison with trotting values, stride
frequencies were also reported for the canter. Stride lengths were
calculated by dividing speed by stride frequency. Because stride
parameters are influenced by leg length (Hoyt et al. 2000), leg
lengths were measured (Table 1). There were no differences
within or between breeds.
To determine whether there was a difference between
Hackney and Arabian breeds for the relationship of VO2 and
speed, an ANCOVA was performed using speed as the covariate
(although with only n = 2 for the Hackneys, the power of this
analysis was limited). Data for VO2 vs. speed were then analysed
using stepwise regression analysis (Statview)10 to determine
whether a polynomial equation described the relationship
significantly better than a linear relationship. Stepwise regressions
with speed and speed-squared as independent variables were
performed with a probability to enter set at 0.05, while the
probability to remove was set at 0.10. Simple linear regressions
were used to analyse stride frequency as a function of speed, for
speeds (1) below and (2) above 5.0 m/s. Stride length was also
analysed by linear regression.
Results
TABLE 1: Physical data on the horse population. Leg lengths
were measured as in Hoyt et al. (2000).
Horse
Breed
Reign*
Red*
GT*
Thai
Larry
Royalty
Jobe
Nick
Don
Arabian
Arabian
Arabian
Arabian
Arabian
Arabian
Arabian
Hackney
Hackney
Sex Age (years) Weight (kg) Leg length (m)
M
G
M
G
G
M
G
G
G
4
7
3
15
15
5
11
10
9
413
466
415
500
420
496
408
450
445
1.34
1.28
1.31
NA
NA
1.36
NA
1.33
1.34
M = mare; G = gelding; *Arabian horses that were trained to extend
the trot; NA = measurements not available.
Before behavioural conditioning, the highest trotting speed for the
Arabians was approximately 5.0 m/s. For the 3 horses trained to
extend the trot, their normal transition speed prior to training
averaged 4.99 m/s (s.d. = 0.43 with 15 trials per horse).
Oxygen consumption as a function of speed was defined better
by a curvilinear function than a simple linear function (Fig 1).
This was true for the Arabians, with extended data excluded (P for
speed = 0.055; P for speed2 = <0.0001) and for the Hackneys
(P for speed = 0.135; P for speed2 = <0.0001). Although there was an
effect of breed on oxygen consumption, this was true because of
relatively few high data points from one Hackney at the higher
speeds. When these points were excluded, there was no breed effect
(P = 0.130 for speeds <6.1 m/s), so the trend line in Figure 1 is a
composite of all data: VO2 = 0.0964(speed2) - 0.267(speed) + 1.296,
128
The cost of transport in an extended trot
120
4.5
4.0
3.5
Stride length (m)
Stride frequency (strides/min)
110
100
90
3.0
2.5
2.0
80
1.5
2
3
4
5
Speed (m/s)
6
7
8
2
3
4
5
Speed (m/s)
6
7
8
Fig 3: Stride frequency increases linearly in trotting horses, but is
relatively independent of speed at the canter and in the extended trot in
both Arabians and Hackney horses. For explanation of symbols, see
Figure 3. + = values from dressage horses at collected, working, medium
and extended trot, as defined by Clayton (1994) (see Discussion).
Fig 4: Stride length (speed/stride frequency) increased for Arabian horses
trotting less than 5.0 m/s (●
●), in an extended trot greater than 5.0 m/s (X),
cantering (●) and in Hackneys trotting (▲
▲). + = values from dressage
horses at collected, working, medium and extended trot, as defined by
Clayton (1994) (see Discussion).
r = 0.961. Cost of transport, CT, plotted against speed was also
curvilinear and there was no difference between breeds
(P = 0.072). The curve represented in Figure 2 is also a composite
of all data and is: CT = 0.00685(speed2) - 0.0561(speed) + 0.238,
r = 0.644. The statistical power based on n = 2 is limited, but it was
clear that metabolic costs for the Hackney horses were not more
economical in an animal bred to trot.
For speeds between 2.0 and 5.0 m/s, stride frequency (SF)
increased (Fig 3) in a linear fashion for Arabians (SF = 8.65[speed]
± 58.8, r = 0.891) and for Hackneys (SF = 10.76[speed] ± 50.8, r =
0.970). When the Arabians made a gait transition to the canter,
stride frequencies increased 9% from 102 strides/min to 111
strides/min. Stride frequency at the canter increased with speed,
but changed less than 3% over the range of cantering speeds used
in this study (SF = 3.45[speed] + 93.6, r = 0.380, P = 0.020). In the
3 Arabians trained to extend their trot, stride frequencies did not
change with speed above 5.0 m/s (P = 0.18). Similarly, for
Hackneys, stride frequency did not change above a speed of 5.0
m/s (P = 0.22).
Stride lengths increased in a predictable fashion (Fig 4). The
slopes of the regressions for stride length vs. speed were not
different. However, the regression line for cantering was lower
than for trotting Arabians and Hackneys (Table 2).
Discussion
TABLE 2: Regression coefficients for the slope and intercept of
the line describing stride length (m/stride) vs. speed (m/s) with
the standard errors in parentheses. Values for Arabian (Arab)
horses trotting (<5.0 m/s), cantering, in an extended trot (‘x-trot’,
>5.0 m/s) and for trotting Hackney horses
Arab, trot
(n = 267)
Arab, canter
(n = 78)
Arab, x-trot
(n = 64)
Hackney
(n = 125)
Slope
0.471 (0.007) 0.466 (0.015) 0.525 (0.031) 0.468 (0.006)
Intercept 0.684 (0.024) 0.353 (0.078) 0.289 (0.171) 0.719 (0.028)
n = number of samples obtained from 7 Arabians trotting and
cantering, 3 Arabians extending their trot and 2 Hackney horses
There was no difference in behaviour or energetics between
Arabian and Hackney horses, in spite of the fact that the
Hackneys are gaited horses and can trot at speeds 60% higher
than Arabians. For the Arabian and Hackney horses, oxygen
consumption was a curvilinear function of speed (Fig 1). This
curvilinear relationship resulted in a cost of transport that had a
U-shaped relationship (Fig 2); a relationship that has been
emphasised previously in ponies (Hoyt and Taylor 1981) and
horses (Wickler et al. 2000, 2001). The speed that produced a
minimum cost (0.123 ml O2/kg m) was in the middle of the range
of trotting speeds. If horses are permitted to choose the speed at
which to trot, they will behaviourally choose this speed, a
preferred speed, that produces a minimum cost of transport (Hoyt
and Taylor 1981). This is true if the horses are weighted (Wickler
et al. 2001) or moving up an incline (Wickler et al. 2000). As the
speed increased above that which produced a minimum CT,
energetic costs grew progressively. At the highest speed
measured in Arabians, 6.5 m/s, the cost had increased by 32% to
0.162 ml O2/kg m and a 25% increase above the cost at the
normal transition speed. For one of the Hackney horses, the cost
was almost 60% higher than the minimum CT. Assuming this
same progressive increase in metabolic costs, estimates can be
made for horses trotting even faster. The top speed for one
Hackney horse was reported as 8.5 m/s (Edwards 1994), a speed
that would produce a cost of transport of 0.256 ml O2/kg m. That
represents a 107% increase over the minimum cost. Using this CT
and speed relationship, VO2 in the Hackney is estimated to be
130 ml O2/kg min, close to the VO2max of some Standardbreds
(Rose and Evans 1987). Higher VO2 levels (increased by 15%)
have been reported for Standardbreds (Tyler et al. 1996) but it is
unclear if these animals were trotting. In Standardbred races,
horses trot at speeds over 14.5 m/s for 1 mile events (United
States Trotting Association).
Stride frequency increases linearly with speed at the trot (Fig 4).
As frequency increases, the duration of stance phase of the stride
S. J. Wickler et al.
(time of contact, tc) decreases. During locomotion, all the force
necessary to support an animal’s bodyweight must be generated
during this period that the foot is in contact with the ground (tc) and
it follows that the rate of this force application is proportional to the
inverse of tc (1/tc). Kram and Taylor (1990) provided convincing
evidence that rate of force application (1/tc ) explains most of the
speed-related increases of metabolism with increasing speed.
Roberts et al. (1998) reported similar results in 6 species of bipeds.
As tc becomes shorter, muscles must contract more rapidly (increase
in velocity, v). This change in speed of contraction relative to the
maximal shortening velocities (Rome et al. 1990) (v/vmax) leads to
the requirement to recruit more fibres with higher contraction
velocities (Valberg 1986) and concomitant higher energetic costs.
Stride frequency at the trot-canter transition in the Arabians
averaged 102 strides/min (5.0 m/s). This agrees well with allometric
predictions based on equations published by Heglund and Taylor
(1988) using an average mass of 445 kg (= 101 strides/min), but the
transition speed in our horses (5 m/s) was lower than predicted by
their allometric equations (c.f. 5.75 m/s). When our horses made the
transition to a canter at 5.0 m/s, there was a marked increase in the
stride frequency (from 102 at the trot to 111 strides/min at the
canter). This increase is not noted in Heglund and Taylor (1988). In
their determination of the trot-gallop transition speed, they fitted
separate straight lines to the stride frequency vs. speed data for
trotting and galloping gaits. The trot-gallop transition speed was
then determined from the intersection of the 2 lines. This same
technique had been used earlier (Heglund et al. 1974) on a smaller
data set, but one that contained horses. Using the allometric
equations from that paper, the trot-gallop transition speed was
predicted to be 6.61 m/s and the stride frequency at the transition
was 114 strides/min. Interestingly, using separate regression
equations for trotting and cantering in our Arabians, the intersection
occurs at a stride speed of 6.69 m/s. Horses do not make the trotgallop transition at the speed where stride frequency of trotting
equals stride frequency of cantering.
If the horses extended their trot, stride frequency did not
increase in either the Hackneys or in the 3 Arabians taught to
extend their gait, at least at the range of speeds measured in this
study. Similar results were observed in Standardbred trotters at
similar speeds (Leach and Drevemo 1991). An increase in speed
without an increase in stride frequency dictates an increase in stride
length. One means of increasing stride length is to apply a greater
force to the ground (Weyand et al. 2000). In trotters, stride length
is the primary determinant of both speed and aerobic energy
expenditure, so that more successful trotters have longer strides at
the same speed, approximately 11.5 m/s (Roneus et al. 1995).
Extending the trot has not only metabolic consequences but,
extrapolating from the work of Farley and Taylor (1991), probably
leads to increases in ground reaction forces. In their ponies, peak
vertical forces increased linearly with speed while trotting. At a
‘trigger point’, where the ponies switched to a canter, forces
dropped. Extending the trot would increase forces transmitted to
the musculoskeletal system and might lead to osteoarthritis,
degenerative joint disease and long bone fractures (Radin et al.
1973, 1984; Simon et al. 1972; Pool 1990).
In the context of this paper, we defined an extended trot to be
when a horse continued to trot at speeds above the normal trotgallop transition. In dressage, the extended trot is defined as the
horse covering as much ground as possible. Their step lengthens
to the utmost as a result of great impulsion from the hindquarters,
but they do this while maintaining the same cadence (Anon 1999).
We have taken stride data (frequency and stride length) from
Clayton (1994) concerning trotting in dressage horses and have
plotted those in Figures 3 and 4 for comparative purposes. In the
dressage horses, stride frequencies were lower than in our horses
129
and changed little with speed at the 4 defined trots (collected,
working, medium and extended). Although stride frequency is
influenced by bodyweight (Heglund et al. 1974) and we would
expect the larger dressage horses to have lower stride frequencies,
the slope of stride frequency vs. speed in the dressage horses is
decidedly less than our horses. Stride length increases with speed,
but more so in dressage horses. These differences emphasise the
need for careful definitions in biomechanical analyses when
working with horses.
Acknowledgements
Debbie Mead provided the Hackney horses and, with the help of
Shannon Garcia, trained the Arabians to extend their gait. Holly
Greene, Equine Research Technician, provided technical support.
Support for this project came from a NIH grant #1S06GM53933
to DFH and SJW.
Manufacturers’ addresses
1Säto AB,
Lovisedalsvägen, Knivsta, Sweden.
CIC, Port Washington, New York, USA.
Instruments, Cleveland, Ohio, USA.
4Thomas Scientific, Swedesboro, New Jersey, USA.
5Hammond Dreirite Co., Xenia, Ohio, USA.
6Ametek, Pittsburgh, Pennslyvania, USA.
7Brooks Instrument, La Habra, California, USA.
8Measurement Control Systems, Santa Ana, California, USA.
9Sable Instruments, Las Vegas, Nevada, USA.
10Abacus Concepts, Berkeley, California, USA.
2Polar
3Meriam
References
Anon (1999) Rules for Dressage Events, 20th edn., Federation Equestre
Internationale, Switzerland. p 16.
Clayton, H. (1994) Comparison of the stride kinematics of the collected, working,
medium and extended trot in horses. Equine vet. J. 26, 230-234.
Edwards, E.H. (1994) The Encyclopedia of the Horse, Dorling Kindersely, New
York. p 379.
Farley, C.T. and Taylor, C.R. (1991) A mechanical trigger for the trot-gallop
transition in horses. Science 253, 306-308.
Heglund, N.C. and Taylor, C.R. (1988) Speed, stride frequency and energy cost per
stride: how do they change with body size and gait? J. expt. Biol. 138, 301-318.
Heglund, N.C., Taylor, C.R. and McMahon, T.A. (1974) Scaling stride frequency and
gait to animal size: Mice to horses. Science 186, 1112-1113.
Hoyt, D.F. and Taylor C.R. (1981) Gait and the energetics of locomotion in horses.
Nature 292, 239-240.
Hoyt, D.F., Wickler, S.J. and Cogger, E.A. (2000) Time of contact and step length:
The effect of limb length, running speed, load carrying and incline. J. expt. Biol.
203, 221-227.
Kram, R. and Taylor, C.R. (1990) Energetics of running: a new perspective. Nature
346, 265-267.
Leach, D.H. and Drevemo, S. (1991) Velocity-dependent changes in stride frequency
and length of trotters on a treadmill. In: Equine Exercise Physiology 3, Eds:
S.G.B. Persson, A. Lindholm and L.B. Jeffcott, ICEEP Publications, Davis,
California. pp 136-140.
Pool, R.R. (1990) Pathologic findings and pathogenesis of racetrack injuries. Vet.
Clin. N. Am.: Equine Pract. 6, 1-30.
Radin, E.L., Parker, H.G., Pugh, J.W., Steinberg, R.S., Paul I.L. and Rose, R.M.
(1973) Response of joints to impact loading. III. Relationship between trabecular
microfractures and cartilage degeneration. J. Biomech. 6, 51-57.
Radin, E.L., Martin, B.M., Burr, D.B., Caterson, B., Boyd, R.D. and Goodwin, C.
(1984) Effects of mechanical loading on the tissues of the rabbit knee. J. orthop.
Res. 2, 221-234.
Roberts, T.J., Kram, R., Weyand, P.G. and Taylor, C.R. (1998) Energetics of bipedal
running. I. Metabolic cost of generating force. J. expt. Biol. 201, 2745–2751.
Rome, L.C., Sosnicki, A.A. and Goble, D.O. (1990) Maximum velocity of shortening
of three fibre types from horse soleus muscle: Implications for scaling with body
size. J. Physiol. 431, 173-185.
Roneus, N., Essen-Gustavsson, B., Johnston, C., Drevemo, S. and Persson, S. (1995)
Lactate response to maximal exercise on the track: relation to muscle
characteristics and kinematic variables. Equine vet. J., Suppl. 18, 191-194.
130
Rose, R.J. and Evans, D.L. (1987) Cardiovascular and respiratory function in the
athletic horse. In: Equine Exercise Physiology 2, Eds: J.R. Gillespie and N.E.
Robinson, ICEEP Publications, Davis, California. pp 1-24.
Saltin, B. and Rose, R.J. (1994) The racing camel (Camelus dromedaries). Acta
Physiol. Scand. 150, Suppl. 67, 12-18.
The cost of transport in an extended trot
Changes in maximum oxygen uptake during prolonged training, overtraining,
and detraining in horses. J. appl. Physiol. 81, 2244-2249.
Valberg, S. (1986) Glycogen depletion patterns in the muscle of Standardbred trotters
after exercise of varying intensity and duration. Equine vet. J. 18, 479-484.
Simon, S.R., Radin, E.L., Paul, I.L. and Rose, R.M. (1972) The response of joints to
impact loading. II. In vivo behavior of subchondral bone. J. Biomech. 5, 267-272.
Weyand, P.G., Sternlight, D.B., Bellizzi, M.J. and Wright, S. (2000) Faster top
running speeds are achieved with greater ground forces not more rapid leg
movements. J. appl. Physiol. 89, 1991-1999.
Taylor, C.R. (1994) Relating mechanics and energetics during exercise. In:
Comparative Vertebrate Exercise Physiology: Unifying Physiological Principles.
Advances in Veterinary Science and Comparative Medicine, Ed: J.H. Jones,
Academic Press, San Diego. pp 181-215.
Wickler, S.J., Hoyt, D.F., Cogger, E.A. and Hirschbein, M. (2000) Preferred speed
and cost of transport: the effect of incline. J. expt. Biol. 203, 2195-2200.
Taylor, C.R., Schmidt-Nielsen, K. and Raab, J.L. (1970) Scaling of energetic cost of
running to body size in mammals. Am. J. Physiol. 219, 1104-1107.
Tyler, C.M., Golland, L.C., Evans, D.L., Hodgson, D.R. and Rose, R.J. (1996)
Wickler, S.J., Hoyt, D.F., Cogger, E.A. and Hall, K. (2001) The effect of load on
preferred speed and cost of transport. J. appl. Physiol. 90, 1548-1551.
Withers, P.C. (1997) Measurement of VO2, VCO2, and evaporative water loss with a
flow-through mask. J. appl. Physiol. 42, 120-123.