JOURNALOF

Vol. 39, No.

APPLIED PHYSIOLOGY
1, July 1985. Printed in U.S.A.
Force platforms as ergometers
GIOVANNI A. CAVAGNA
Istituto di Fisiologia Umana, Uniuersitci di Milan0 and
Centro di Studio per la Fisiologia de1 Lavoro Muscolare del CNR, Milano, Italy
CAVAGNA, GIOVANNI A. Force platforms as ergometers. J. Appl.
Physiol. 39( 1) : 174-l 79. 1975.-Walking and running on the
level involves external mechanical work, even when speed aver-
aged over a complete stride remains constant. This work must be
performed by the muscles to accelerate and/or raise the center of
mass of the body during parts of the stride, replacing energy which
is lost as the body slows and/or falls during other parts of the
stride. External work can be measured with fair approximation
by means of a force plate, which records the horizontal and ver-
tical components of the resultant force applied by the body to the
ground over a complete stride. The horizontal force and the verti-
cal force minus the body weight are integrated electronically to
determine the instantaneous velocity in each plane. These veloci-
ties are squared and multiplied by one-half the mass to yield the
instantaneous kinetic energy. The change in potential energy is
calculated by integrating vertical velocity as a function of time to
yield vertical displacement and multiplying this by body weight.
The total mechanical energy as a function of time is obtained by
adding the instantaneous kinetic and potential energies. The
positive external mechanical work is obtained by adding the in-
crements in total mechanical energy over an integral number of
strides.
biomechanics; locomotion;
external and internal work
walking . running;
>
mechanical work;
lowered, requiring external positive work during the increase in
potential energy and the speed of progression oscillates from below
to above the average value requiring external positive work during
the increase in kinetic energy of the centre of mass of the body.
The muscles are active and exerting a force during the phases
of the step of decreasing mechanical energy to retard and control
the movement; in these phases they are being stretched and the
muscular force is doing negative work. This braking action re-
quires the expenditure of chemical energy by the muscles and is
equivalent to applying the brakes on an automobile. Thus the
chemical energy expended in applying the brakes and the me-
chanical energy which is not stored in and recovered from elastic
elements appears as heat in the muscle and is lost. This mechani-
cal energy is then replaced by contraction of the muscles as they
do positive work to raise and reaccelerate the center of mass of
the body.
If complete storage and recovery of energy in elastic elements
(contracted muscles and tendons) were possible, then the decrease
in potential and/or kinetic energy during one phase of the step
could be utilized in another phase of the step, and no additional
energy from the muscles would be needed except that necessary
to maintain tension (a slack muscle is unable to store any appre-
ciable amount of elastic energy). Although there is some recovery
of energy in elastic elements (1, 2, 5, 13), there is presently no
way to measure it directly.
MECHANICAL WORK IN WALKING AND RUNNING. Muscles transform
chemical energy into mechanical work during exercise. The rate
of energy utilization is routinely determined by measuring oxygen
consumption and lactic acid production. However, the rate at
which muscles perform mechanical work is seldom measured, and
then for specialized types of exercise, such as pedaling a bicycle
ergometer. The mechanical work performed during such common
exercises as walking and running on the level has only been meas-
ured a few times (3-5, 9, 10, 11) and then with an enormous
effort. This paper describes a convenient means of measuring the
external mechanical work performed in walking, running, and
jumping from the forces applied by the body to the ground.
It is frequently argued that walking and running at a constant
speed on the level involves only a very small amount of external
mechanical work to overcome air resistance. This argument is
based on the fact that the mechanical energy possessed by the
body (both potent .ial a nd kinetic) is the same at the beginning
and the end of each step.
On the other hand, it is commonly concluded that external
mechanical work is done when a change in the total mechanical
energy of the body is observed after one or more steps, for example,
when a hill is climbed (increasing potential energy) or a sprinter
accelerates (increasing kinetic energy). These situations differ
from level walking and running at a constant speed only in the
duration of time between performance of positive and negative
work. For instance within a step cycle of level running at a
constant speed, the center of gravity of the body is raised and
FORCEPLATFORMASANERGOMETERTOMEASUREMECHANICALWORR
As long ago as 1885 Marey and Demeny (12) used a force plat-
form to measure the vertical component of the force exerted by the
feet against the ground during standing vertical jumps. In the
1930s Fenn (11) utilized a similar type of platform to measure
forward and backward components of force applied by the body
to the ground during running. From these force measurements he
calculated the mechanical work necessary to account for the
velocity changes of the center of mass of a running man during
each stride. In 1939 Elftman (8) used a force platform to measure
the force exerted by one foot against the ground during walking.
Although he did not measure the actual velocity and the displace-
ments of the center of gravity, he showed that this was possible
cc
. . . by proper evaluation of integration constants.
Cavagna and colleagues (3, 5) utilized force platforms to meas-
ure the external mechanical work done in level running and walk-
ing at different speeds. The procedure was laborious since these
authors were limited by I) the small dimensions of the platforms
(35 x 35 cm; in walking only the phase of the step in which a single
foot was on the ground could be studied, and in running it was
difficult to step on the platform without altering the length of the
stride); 2) the elaboration of the force-time tracing from the plat-
form, made entirely by graphical computation (this was neces-
sarily discontinuous, and the shape irregularities of the tracing
and the oscillations due to the natural frequency of the platform
at the highest speeds decreased the accuracy); 3) the method was
SO slow that only a small number of determinations of mechanical
work could be obtained.
174
FORCE PLATFORMS AS ERGOMETERS
175
These difficulties have been overcome I) by using a large plat-
form (4 x 0.5 m) sensitive to the forward and the vertical compo-
nents of the force exerted by the foot and 2) by electronically in-
tegrating the force measurements to yield a direct readout of
velocity.
CALCULATIONS INVOLVED IN UTILIZING MEASUREMENT
OF FORCE TO DETERMINE EXTERNAL WORK
The external work required to change the kinetic and/or po-
tential energy of the center of mass of the body is defined as
W ezt = F ,s = /F/./s/COSQ = Fd, + Ffsf + F,sl
= w, + Wf + WI
(1)
where s is the displacement of the center of mass of the body, F is
the resultant of all external forces exerted against the body by the
ground and the air, and Q is the angle between the two vectors F
and s; F. , s,, , Ff , etc. are projections of F and s in the vertical,
forward, and lateral directions; W, , Wf , and WI are the work
done, respectively, by the vertical, forward, and lateral compo-
nents of the force. During level walking and running sf is obviously
much greater than s0 and s1 , and F, is much greater than Ff and
FI due to the acceleration of gravity. As a result of the greater
displacement and force, respectively, Wf and W, are much greater
than WI (= 100 times in walking (4)). Thus WI can be neglected.
The work involved in rotating the body about its center of mass
(7) was also neglected.
To determine W,,t one can then start by measuring forces in
the vertical (Fo) and forward (Ff) directions over a complete
stride.
The force exerted by the feet on the platform in the vertical
direction is
Fv = weight (P) + frictional forces + m (I~
and in the forward (or lateral) direction
(2)
Ff = frictional forces + m .af
(2)
where m is the mass of the body, a is the acceleration of the center
of gravity of the body and the frictional forces are u) air resistance
and 6) the force which opposes the displacement of the center of
gravity within the body during a nonelastic deformation of the
body itself. The frictional forces oppose muscular force during the
acceleration of the body whereas they cooperate with it in de-
celerating the body. For the purpose of the present calculations
the forces of friction even in man sprinting at high speed (2) can
be neglected as discussed later. Therefore Eq. 2 and 2 can be
simplified to
F,, - P = m,a,
(3)
Ff
= m,af
(3')
F, - P can be measured directly during exercise by zeroing the
platform while the subject stands on it quietly immediately before
the experiment. Thus it is possible to calculate the acceleration of
the center of gravity from direct measurements of the force exerted
in the vertical and forward directions.
It is much easier to work from instantaneous velocity than from
instantaneous force in calculating the mechanical work. This
procedure abolishes the interference caused by vibrations of the
plate and provides a much cleaner starting point. In addition the
velocity tracings are much simpler and their interpretation much
easier. Since the velocity is the integral of acceleration it can be
obtained directly by electronically integrating the signal measured
with the force plate (Fig. 1)
J(Fo - P)dt = mV, + const
(4)
jFfdt = mV$ + const
(4')
VF (m/set) 0.5r
const.
t
-051
FF (kg) 50
r
0
-50
t
Vv (m/set) 1
const
-1
F,(kg) 0
100
200
WALK
0.5
RUN
const
-05 E
50
-50 1
0 i
2
const
-2 t
d \
0 1 2 set
FIG. 1. Tracings recorded when a subject was walking at 5.5 km/h
(left) and running at 11.9 km/h over the force platform. Ff and F,
are, respectively, the horizontal and vertical components of the result-
ant force impressed by the feet on the platform; whereas F, oscillates
around the zero (positive values = acceleration forward of subject,
i.e., backward push on platform), F,, oscillates around a force value
(59 kg) equal to the subjects body weight, P (corresponding to 1 g =
9.8 m/s2 on the right-hand scale). Simultaneously the horizontal force
and the vertical force minus the body weight (F, - P) are integrated
electronically to determine the velocity tracings Vf and V,; these give
the forward and vertical components of the velocity of the center of
gravity of the whole body plus an integration constant to be deter-
mined later according to the procedure indicated in Fig. 2. Electronic
integrators are operated by subject crossing photocells at the platform
level (2.7 m apart when walking and 3.5 m apart when running, Fig.
5); photocells are placed in such a way that the subject crosses them
when he does not contact the ground before or after the platform.
The integration constants must be known in order to calculate
the absolute velocity. The value of the integration constants in
Eq. 4 and 4 is zero when the subject begins to move on the plat-
form from the resting condition (V = 0). In this case the deter-
mination of the instantaneous velocity is easy and both absolute
velocity and external mechanical work can be recorded as a
function of time simultaneously with the performance of the
exercise (1, 2, 6, 13). However, when the subject arrives on the
platform with a velocity above zero, the values of the constants
in Eq. 4 and 4 are unknown and the absolute velocity cannot be
measured simultaneously with the performance of exercise. The
absolute forward velocity is obtained by I) measuring the average
forward velocity of the subject moving across the platform by
means of two photocells: these also turn on and off the integrators
while the subject is over the platform without contacting the
ground before or after it (Fig. 24); 2) measuring the area under
the recorded velocity tracing during the time interval while the
integrators are on (either by planimetry or by utilizing a compu-
ter); and 3) dividing the area by the time required to travel the
distance between the photocells to position the average velocity
forward in the tracing (Fig. 2B). Instantaneous forward velocity
can then be calculated. In this procedure it is assumed that the
average speed of the trunk between photocells is equal to the
average speed of the center of gravity: this is reasonable since the
displacements of the center of gravity within the body and the
tilting of the trunk (5, 11) are small in comparison with the dis-
tance between the sights.
A similar procedure is followed to determine the absolute value
of the vertical component of the velocity (Fig. 2, C and D); in
this case however the area A below the vertical velocity tracing is
measured only for an interval of time, nr, corresponding to an
integral number of steps. The ratio A/w gives an average vertical
velocity equal to zero on the assumption that during an integral
number of cycles, the upward displacement of the center of gravity
176 G. A. CAVAGNA
Al Forward velocity tracing :
0 0.5 set 1
B) Integrate during time integrators
are on and divide by thetime they
are on in order to locate VF on tracing:
>
4i
I-12
0 0.5 set 1
From the absolute values of VF,V, and Sv
the kinetic energies EKF :$mV,f, E,, :+rnV:
the potential energy: EP = weight S,
and the total energy:
E
TOT
= EKF+ E,,+ Ep
are easily obtained ( Fig. 3 >
Cl Vertical velocity tracing :
f time integrators on
ON OFF
0 0.5 set 1
0) Integrate during time of one or more
complete strides (oblique hatching ) and
divide by this time to Locate v,,which is zero
0.5 set 1
El Vertical displacement downwards ( vertical
hatching ) and upwards (horizontal hatching )
IS obtained by integration of Vv tracing:
0
is equal to the downward displacement (i.e., that in level walking
and running the height of the center of gravity on the average is
constant). This assumption may not be true at the end of a single
step, but it is certainly valid over a number of steps. Once the
zero vertical velocity is determined then the instantaneous vertical
velocity is known.
From the instantaneous velocity in the forward and vertical
directions the instantaneous kinetic energy Ekf = >s(rnVj2) and
Ekv =
>s(mV,2) can be calculated; +AEk is the positive work
required to accelerate the center of gravity of the body.
The instantaneous potential energy Ep = Pas, can be obtained
by integrating the vertical velocity to determine the vertical dis-
placement of the center of mass (Fig. 2E)
SKdt = s, + const
(4
Geometrically the upward displacement is represented by the
upper area enclosed by the instantaneous velocity tracing, V, ,
and the zero (Fig. 20). The positive work against gravity is given
by the increments of the potential energy: +AE, . The total
mechanical energy is then obtained by adding the instantaneous
potential energy Ep and the instantaneous kinetic energies EkV
and EIcj
E tot = E, + Ekv + Ekf (6)
It is clear that potential energy can be converted into kinetic
energy and vice versa if the increase in one is out of phase with
the increase in the other, and indeed this happens during walking
(4). On the other hand, during running, the increase and decrease
in potential and kinetic energy are almost completely in phase,
and there is little interconversion between kinetic and potential
energy (5).
Finally the positive external mechanical work is obtained by
adding the increments in mechanical energy, +AEt,t , over an
integral number of strides.
In Fig. 3 the upward curve indicates how the kinetic energy
Ekf of the center of mass of the body of a 78-kg subject oscillates
during two steps of running at 15.5 km/h. This kinetic energy
0.5 set 1
FIG. 2. Procedure followed to deter-
mine the absolute value of the velocity
in the forward (A and B) and the ver-
tical (C and D) directions from the
velocity + constant tracings, recorded
during the exercise, (Fig. 1). From the
absolute value of velocity given in B and
D the corresponding kinetic energy can
be calculated ; vertical velocity tracing
given in D is integrated further to yield
the vertical oscillations of the center of
gravity sV (E) and then the potential
energy changes (Fig. 3).
curve was calculated from the forward velocity curve given in
Fig. 2B. The E, curve below indicates the oscillation of the poten-
tial energy of the body and was calculated from the vertical dis-
placement curve given in Fig. 2E. The curve just above it is the
sum of the potential and the kinetic energy Ekv (calculated from
curve in Fig. 20). In the interval in which the sum Ep + Ekv is
constant the subject is off the ground and the kinetic energy EkV
is transformed into potential energy and vice versa. At the top
and bottom points the two curves coincide since at these points
the vertical velocity and then EkV is nil. Whereas Ekf is given in
absolute units (the average kinetic energy of the subject being
about 171 Cal), the potential energy Ep is not: in each trial the
starting point for the computation of Ep was taken as zero. The
two curves below indicate the total mechanical energy Etot =
Ekf + Ep -/- EkV and, for comparison, the sum Ep -/- Ekf as usually
measured in the past (Fig. 5 in (5)). Etot may differ from Ep + Ekf
not only for the slope of the curve, giving the rate of work produc-
tion, but also for the amplitude of the oscillation of the curve
which represents the external mechanical work done; the differ-
ence, negligible in running, becomes appreciable in walking. The
positive external mechanical work done is taken as the increments
of the curve Etot during the interval of time, of one or more com-
plete strides, in which the V, tracing was integrated to locate the
average vertical velocity (Fig. 20). The same is done for the curve
Ekf to determine the work necessary to sustain the kinetic energy
changes Wf , and for the curve Ep + EkV to determine the work
against gravity, WV . All the calculations reported in Fig. 2 as well
as the curves in Fig. 3 were made directly by a computer.
As mentioned above if the subject starts moving on the plat-
form the constant of integration in Eq. 4 and 4 is zero and it is
possible to integrate the vertical velocity directly to give displace-
ment (and potential energy). Simultaneously other operational
amplifiers can be used to square velocity to obtain a direct readout
of kinetic energy and to sum instantaneous potential and kinetic
energy to give total instantaneous energy. Thus mechanical work
can be recorded simultaneously with the exercise (1, 2, 6, 13). The
limit of this method is that any drift of the first integrator (velocity)
is squared by the second (displacement).
FORCE PLATFORMS AS ERGOMETERS 177
,E
KF
EP+ EKV
/
0. 0.50 1.00 SEC
FIG. 3. Mechanical energy of the centre of mass of a 78 kg subject
running at 15.5 kn/h. E kf is the kinetic energy calculated from the
forward velocity curve given in Fig. 2B. EP is the potential energy
calculated from the sU curve in Fig. 2E. Ekv is the kinetic energy calcu-
lated from the curve in Fig. 211. Etot = Ekj j-- EkV -j-- Ep. All the calcu-
lations reported in Fig. 2 as well as the curves above were made
directly by a computer.
ASSUMPTIONS INVOLVED IN CALCULATIONS
TO DETERMINE EXTERNAL WORK FROM FORCE
The kinetic energy increase (+AE,& calculated disregarding
the forces of friction, is greater than the total positive work actually
done by the subject to accelerate himself (+AE,& to overcome
air resistance and to deform the body ( VVlosses), i.e.
+AEk > + AEk + &,,,,, = positive work actually
done by muscles
(7)
During deceleration, when the forces of friction and muscular force
have the same direction (-F = - m a + forces of friction), the
apparent kinetic energy change ( - AEh), calculated as described
above, will be smaller than the negative work done, i.e.
-AEk < -AI& + Wlosses = negative work actually
done by muscles
00
Inequalities 7 and 8 can be derived mathematically as follows.
During the push (positive work phase), the area below the force-
time tracing recorded by the platform (JFdt) equals the momen-
tum actually gained by the body (Jm a dt) plus the time integral
of the force, FfT, necessary to overcome the frictions against the
air and within the body (JFf,dt)
$Fdt = J m a dt + J Ff, dt
i.e., assuming the initial velocity equal to zero
(9)
??lV = m v + m vfr
m
where m vfr represents an additional momentum which the body
would have acquired because of the push in absence of friction.
When the kinetic energy is calculated from the velocity v it can
be seen that
>d(rn v2> > >a(rn d2) + +$ (m vfr2>
which is inequality 7.
During negative work
-JFdt = -J m a dt + J Ffr dt
w
i.e., assuming that the deceleration is sufficient to reduce the
velocity v to zero
-mu = -mu + mvf,
W)
from which it appears that -m v, the momentum actually lost,
is the sum of the momentum lost because of a) the braking action
of the muscles against the platform, -m v, and 6) the force of
friction, - m vf T . Expressing the work as a kinetic energy change,
inequality 8 is obtained
->a(rn v2) < - %(m vr2) + $5 h vfr2) (14)
The signs - indicate that the initial kinetic energy, which is a
positive quantity, is subtracted from the final kinetic energy
(equal to zero) in order to get the kinetic energy change.
Fortunately the error done by assuming AE, = AEk + Wlosses
is not great. This was checked by measuring the work actually
done (AEk + VVl osses) by a sprinter to accelerate forward and by
comparing it with the same work measured as AE, (2). The posi-
tive and the negative work actually done by the muscles at each
step was determined by multiplying the average horizontal force,
IQ , exerted on the platform during the push or during the brake,
by the actual displacement forward of the trunk Sf during these
intervals determined by means of photocells. Since the displace-
ments of the center of gravity within the trunk (11) were small in
comparison with the distance between sights (3 m), the displace-
ment of the trunk Sf could not differ appreciably from the dis-
placement of the center of gravity. The mechanical work done at
each step
Ws tep
= Ff Sf P)
is positive during acceleration and negative during deceleration;
the positive and the negative work were calculated from Eq. 15
for 19-20 steps taken by two subjects after the start. The total
positive work is given by
+wot = z + Wstep
and the total negative work
- wtot = x - Wstep
(16)
178
The algebraic sum
+ Wtot - Wtot = Wtot (17)
was, according to inequalities 7 and 8, smaller than the work
measured as AEk . The difference however was only 8% (2). I f
some skidding took place part of this difference would represent
the work done against friction between the foot and the ground.
Taking into account that in sprint running one attains the highest
speed, i.e., the greatest air resistance, and that the forces deform-
ing the body are the greatest met in locomotion, it is therefore
possible to conclude that the error done with the procedure de-
scribed above is tolerable.
The vertical displacements of the body usually take place at a
speed appreciably smaller than that attained in the forward
direction; in addition the cross-sectional area of the body per-
pendicular to the direction of the movement is much smaller in
the vertical than in the forward direction. Both these factors make
air resistance to a vertical movement much smaller than that to a
forward movement and therefore probably negligible. In addi-
tion the vertical component of the push acting along the vertebral
column is probably less effective in deforming the body than an
equal force acting in forward direction. On the other hand the
work lost in the deformation of the body due to the forward com-
ponent of the push was found to be smaller than that due to air
resistance in sprint running (2). These considerations induce one
to think that V, N V, and therefore Ep N E, In addition ex-
periments in progress indicate that data of work done against
gravity in running, obtained with the present method, are in good
VERTICAL FORWARD
4
I
SIDE
9 VIEW :
7 &
1 I
i 500 mm I
TOP
VIEW
4
69
FORWARD FORCE
,$
P 15OVOC * b 150 voc 4
FIG. 4. Schema of vertical and forward units forming each of the
eight plates constituting the whole platform. Forward unit is placed
on top of the vertical one and fixed to it by the indicated bolts. Strain
gauges indicated by D undergo a deformation opposite to that of
strain gauges Z when a force is applied to the unit. All the strain
gauges of the eight units sensitive to the force in a given direction
(vertical or forward) are wired together in a Wheatstone bridge of 96
resistors as indicated at the bottom.
G. A. CAVAGNA
agreement with those determined at the same speed by Fenn (11)
with motion picture analysis.
APPARATUS INVOLVED IN MEASURING EXTERNAL
MECHANICAL WORK DURING WALKING AND RUNNING
The force platform used in these studies consists of a single row
of eight (0.5 rn) plates. Each plate has separate vertical and for-
ward units and utilizes twelve spring assemblies (6 in the vertical
and 6 in the forward) consisting of four springs and two strain
gauges (Fig. 4). Half of the strain gauges in the assemblies are
compressed and half are stretched when a force is applied in one
direction. Each half is wired in series with those of the other plat-
forms forming two opposite arms of a Wheatstone bridge as indi-
cated in Fig. 4. The platform is inserted, with its surface at the
level of the floor, 30 m from the beginning of a corridor 50 m long.
Its natural frequency is 42 Hz in the forward and 30 Hz in the
vertical direction. The maximal difference found between the
electrical response of the eight platforms, when acting with a given
force on the central portion of each one of them, is: 4y0 (with full
scale 5 kg) or 4.7 7 (full scale 80 kg) in the vertical direction; and
8-9 $& (full scale 10 kg) or 7-lO$$, (full scale 50 kg) in the hori-
zontal direction. The maximal difference found between the 32
corners of the 8 platforms was 11.3-14%. These differences can
be reduced by shunting (with appropriate resistors) the strain
gauges corresponding to the most sensitive points. The average
response of the horizontal units to a force directed forward (de-
celeration) was usually found slightly greater than that to the
same force directed backward (acceleration); the difference how-
ever was less than 5%. The platform was tested from 1 to 250 kg
in the vertical and from 0.5 to 100 kg in the forward direction
and found to give a linear response within an average error of 5
and 7 %, respectively.
The output from each Wheatstone bridge (forward and vertical,
Fig. 4) after amplification by a Philips PR 7510 preamplifier
ERTlCAL I - 1 FORWARD
FIG. 5. Diagram of the setup used to record the tracings given in
Fig. 1: when the subject crosses the first photocell the integrators are
operated and the velocity tracings begin (Vv and Vf in Fig. 1). Ve-
locity tracings together with a ZOO-Hz signal are also recorded on tape
for later treatment by a computer according to the procedure indicated
in Fig. 2. For further details see text.
FORCE PLATFORMS AS ERGOMETERS 179
goes a) directly to a Hewlett-Packard amplifier (350-1000B) and
recorder (77 19) to yield force, 6) through a DC offset, to an inte-
grator (which is turned on and off by the photocells) and then to
the amplifier and recorder to yield velocity. The output of this
amplifier is also recorded on tape (Hewlett-Packard 3960A tape
recorder) for later treatment of the signal by a computer. The
photocells also turn on and off a ZOO-Hz signal which is also re-
corded on the tape (Fig. 5). This provides the time base needed
for calculating the integration constants as described earlier. In
addition it is used to trigger the analog-to-digital converter. Before
integrating the vertical component of the force (F V
=P+
is necessary to subtract the body weight as mentioned
m a,> it
earlier.
This is done by setting the platforms output exactly to zero using
the DC offset voltage while the subject stands absolutely still on
the platform before the beginning of the exercise. This operation,
and also the setting of the forward output exactly to zero with
another DC offset voltage, must be done with great care immedi-
ately before each trial otherwise an appreciable drift of the output
of the integrators takes place. In our setup the base line from the
platform to the integrators has noise, mainly due to vibrations of
the ground, attaining about & 100 g, and its drift during
of a trial is usually less than 50 g.
the time
i.e., the external work. In exercise, however, also internal work is
done, and this cannot be measured by means of the force platform.
The main source of internal work (and the only one directly
measurable) is the kinetic energy changes of the limbs calculated
from their velocity relative to the center of gravity. This can be
measured by motion picture analysis as described for example by
Fenn (10). The increments of the curve indicating the total kinetic
energy of the limbs represent the internal work done. This curve
should not be added to the curve of the mechanical energy level,
E tot, given in Fig. 3 (as sometimes reported in the literature),
because by this summation one implicitly assumes the transfer of
some of the work done by the internal forces to an increase in the
energy level of the center of mass, which is not possible by defini-
tion The absolute value of the increments of the curve Etot (giving
the external positive work) must therefore be added to the abso-
lute value of the increments of the curve of the kinetic energy of
the limbs (giving the internal positive work) to obtain the total
positive work done.
The author thanks Prof. C. R. Taylor of Harvard University for
his much helpful and constructive advice during the completion of
this manuscript. The program to obtain the records of Fig. 3 was pre-
pared by Dr. Carlo Cavagna of the Politecnico of Milan. The plat-
EXTERNAL, INTERNAL, AND TOTAL WORK
form was projected by the Viterra firm, Wallisellen (Zurich, Switzer-
The methods described above allow us to determine only the
land).
work necessary to sustain the displacements of the center of gravity, Received for publication 25 October 1974.
REFERENCES
1. CAVAGNA, G. A., L. KOMAREK, G. CITTERIO, AND R. MARGARIA.
Power output of the previously stretched muscle. Med. Sport 6:
159-167, 1971.
2. CAVAGNA, G. A., L. KOMAREK, AND S. MAZZOLENI. The me-
chanics of sprint running. J. Physiol., London 2 17 : 709-72 1, 197 1.
3. CAVAGNA, G. A., AND R. MARGARIA. Mechanics of walking. J.
Ap~l. Physiol. 21 : 271-278, 1966.
4. CAVAGNA, G. A., F. P. SAIBENE, AND R. MARGARIA. External
work in walking. J. A/@. Physiol. 18 : 1-9, 1963.
5. CAVAGNA, G. A., F. P. SAIBENE, AND R. MARGARIA. Mechanical
work in running. J. AppZ. Physiol. 19 : 249-256, 1964.
6. CAVAGNA, G. A., A. ZAMBONI, T. FARAGGIANA, AND R. MARCARIA.
Jumping on the moon: power output at different gravity values.
Aerospace Med. 43 : 408-4 14, 1972.
7. ELFTMAN, H. The rotation of the body in walking. Arbeitsphysiol-
ogie 10: 477-484, 1939.
8. ELFTMAN, H. The force exerted by the ground in walking. Ar-
beitsphysiologie 10 : 485-49 1, 1939.
9. ELFTMAN, H. The work done by muscles in running. Am. J.
Physiol. 129 : 672-684, 1940.
10. FENN, W. 0. Frictional and kinetic factors in the work of sprint
running. Am. J. Physiol. 92 : 583-611, 1930.
11. FENN, W. 0. Work against gravity and work due to velocity
changes in running. Am. J. Physiol. 93 : 433-462, 1930.
12. MAREY, J., AND G. DEMENY. Locomotion humaine, mecanisme
du saut. Compt. Rend. Acad. Sci. 101 : 489-494, 1885.
13. THYS, H., T. FARAGGIANA, AND R. MARGARIA. Utilization of
muscle elasticity in exercise. J. ApfZ. Physiol. 32 : 491-494, 1972.