You are on page 1of 14

Eur J Appl Physiol (2009) 106:1–14

DOI 10.1007/s00421-009-1008-7

INVITED REVIEW

Efficiency in cycling: a review


Gertjan Ettema Æ Håvard Wuttudal Lorås

Accepted: 2 February 2009 / Published online: 20 February 2009


 Springer-Verlag 2009

Abstract We focus on the effect of cadence and work rate been studied extensively, not the least cadence. Research
on energy expenditure and efficiency in cycling, and present has also been directed towards other factors, including
arguments to support the contention that gross efficiency can those related to task variations (e.g., crank arm character-
be considered to be the most relevant expression of effi- istics, load, ring shape, body position), environmental
ciency. A linear relationship between work rate and energy conditions (e.g., uphill cycling), and also subject charac-
expenditure appears to be a rather consistent outcome among teristics (e.g. patient groups, athletic level) in relation to
the various studies considered in this review, irrespective of energy cost.
subject performance level. This relationship is an example of It is common practice to express the energy expenditure
the Fenn effect, described more than 80 years ago for muscle as efficiency (i.e., the ratio of work generated to the total
contraction. About 91% of all variance in energy expendi- metabolic energy cost). This is often done in order to
ture can be explained by work rate, with only about 10% compare various studies that have been done at different
being explained by cadence. Gross efficiency is strongly work rates, as a higher work rate typically requires a higher
dependent on work rate, mainly because of the diminishing energy consumption. Another reason is to get a better
effect of the (zero work-rate) base-line energy expenditure insight into how external work rate affects the physiolog-
with increasing work rate. The finding that elite athletes have ical stress. Moreover, the efficiency measure may also
a higher gross efficiency than lower-level performers may provide more insight into the mechanisms behind the
largely be explained by this phenomenon. However, no firm effects of various factors on energy consumption, and
conclusions can be drawn about the energetically optimal thereby into the function of the metabolic processes
cadence for cycling because of the multiple factors associ- involved in work production. One of the major challenges
ated with cadence that affect energy expenditure. in studying physical activity is finding an accurate mea-
surement of the (external) work done (e.g., van Ingen
Keywords Efficiency  Cycling  Energy expenditure  Schenau and Cavanagh 1990). Cycling, especially on a
Cadence  Work rate cycle ergometer, is one of the few exceptions. Thus, it is
not surprising that many studies on the relationship
between efficiency and energy consumption have been
Introduction conducted within the context of cycling.
Efficiency in cycling has been studied for almost a
The study of energy consumption in cycling has a long century (Benedict and Cathcart 1913), and di Prampero
history. Many factors affecting energy consumption have (2000) recently provided an overview of this issue. The
discussion that will be developed in this review is based on
an old debate. It deals in essence with the definition of
G. Ettema (&)  H. W. Lorås efficiency and what may be termed ‘‘baseline subtraction’’
Human Movement Science Programme, Faculty of Social
(i.e., the subtraction from the measured oxygen uptake of
Sciences and Technology Management, Norwegian University
of Science and Technology, 7941 Trondheim, Norway that associated with the baseline condition—rest, unloaded
e-mail: gertjan.ettema@svt.ntnu.no pedalling etc.). Baseline subtractions were used in early

123
2 Eur J Appl Physiol (2009) 106:1–14

experiments (e.g., Benedict and Cathcart 1913; Dickinson regard the human body (i.e., its locomotor muscles) as a
1929; Garry and Wishart 1931, 1934), but criticised much system that is not isolated. Indeed, energy from its envi-
later (e.g., Cavanagh and Kram 1985; Stainbsy et al. 1980; ronment can flow into the system (metabolites and negative
van Ingen Schenau and Cavanagh 1990). Even so, various work) and out (in the form of work and heat). The ther-
definitions of efficiency and baseline subtractions are used modynamic potential of a system, enthalpy (H) is the sum of
currently. The essence of the debate on the validity of base- the internal energy of a system and the product of pressure
line subtractions relates indirectly to the fundamental laws and volume. When considering energy changes in muscle
of thermodynamics and mechanics. These laws and their contraction, this can be simplified to that enthalpy change is
direct consequences for treatment and interpretation of data equal to internal energy change because muscle volume is,
are not always followed strictly, or are not explicitly practically speaking, constant. Enthalpy consists of two
clarified in the literature. This lack of clarity can lead to forms, free energy (G) and entropy (S). The change in free
misunderstandings. The principal aim of this paper is energy drives reactions: when doing work, one spontaneous
therefore to present a review of the literature on the rela- reaction occurs (ATP hydrolysis) and liberates energy (DG
tionships among cycling efficiency, cadence and work rate. is negative) that drives another reaction (work production,
We focus on gross efficiency because, in our opinion, this DG is positive). However, not all free energy that is liber-
is the most relevant efficiency measure for such consider- ated in the driving reaction is utilised in the driven reaction.
ations, as we explain later. As a result, some of our The difference is transferred as heat, leading to an increase
conclusions may differ from those expressed in the in S. Efficiency of an energy converting system is the ratio
literature. of free energy outflow over the total free energy inflow.
We start with a brief overview of a number of principles In muscle contraction, the free energy outflow is work.
that originated from thermodynamics and classic mechan- This efficiency ratio is also found by taking the energy
ics. In doing so, we clarify our standpoint on the use of rate (power) for both nominator and denominator in this
various definitions of efficiency. We then discuss a number equation:
of detailed issues that relate directly to the calculation and work power
interpretation of efficiency, especially in cycling. Ee then e¼ ¼
energy cost energy rate
critically review the literature on efficiency in cycling and
its dependence on the two main variables, power and The energy cost is usually expressed as metabolic cost.
cadence. Finally, the ‘‘trainability’’ of cycling efficiency is When discussing the ways in which efficiency can be
briefly discussed. It is beyond the scope of this paper to defined, it is important to have a clear system definition and
include comprehensive discussion of factors such as mus- be aware of the fact that efficiency is a parameter
cle fibre type (Horowitz et al. 1994; Coyle 2005), bicycle describing a quality of the energy flow that runs through
models (Minetti et al. 2001), crank systems (Zamparo et al. that particular system. In a previous paper (Ettema 2001),
2002), chain rings (Cullen et al. 1992; Hull et al. 1992), we briefly touched on this issue. Here, it is taken as an
crank inertial load in uphill cycling (Hansen et al. 2002), explicit departure point: the definition of what the energy
foot position (Van Sickle and Hull 2007), and environ- conversion system entails fully determines what is to be
mental conditions (Ferguson et al. 2002; Green et al. 2000). considered as energy inflow (cost) and what as energy
We hope that this analysis contributes to setting a outflow (work and heat). Thus, this system definition
framework by which studies can be better compared, determines the definition of efficiency. In our opinion,
and efficiency can be interpreted and discussed more unfortunately, an explicit and formal definition of the
appropriately. energy conversion system is often lacking, which may lead
to confusion and misunderstanding for the reader.
The complication with locomotion of the (human) body
Mechanisms of energy conversion and efficiency is that the energy transforming system and the physical
definitions, as applied to cycling body may physically be overlapping, but are not neces-
sarily identical entities. In locomotion, one can define the
Thermodynamics and efficiency definition energy converting system as the total physiological human
body, and the object that work is done against as the
The first law of thermodynamics, i.e., the law of conser- physical human body (mass). The energy inflow is then
vation of energy, states that the total energy of a system and contained in the food intake. This implies that the energy
its surroundings is constant. Thus, in a system isolated from required for swallowing and digestion is to be considered
its surroundings, the total amount of energy is constant and in the calculations of efficiency. One could go as far as
no energy can be produced or lost, only transformed. By arguing that all the energy consumption required to main-
using the terms energy loss and production, we implicitly tain homeostasis should be considered as energy inflow, as

123
Eur J Appl Physiol (2009) 106:1–14 3

without this maintenance cost (i.e., the cost merely to stay essentially isolated from the process of doing work (the last
alive), it would be impossible to do work at all. This issue not commonly being discussed in the literature). In
maintenance entails far more than only the process of other words, the line of thought is that maintenance is, of
digestion and is an ongoing process with, in principle, a life course, necessary to support a system in the first place, but
long energy flow, leading to efficiency values equal to zero the energy flow otherwise is not related to that of doing
for any activity. This line of thought is rarely followed, work in whatever form: one assumes that two independent
however. Yet, it exemplifies problems that may arise when energy flows run in ‘parallel’, while not affecting or
interpreting the term efficiency and it shows how important relying on each other. One energy flow is that of the
it is to define the system explicitly and accurately. If we are musculoskeletal system as a work generator (assuming that
more interested in how muscles generate work, and not in constitutes the defined energy transformation system), and
the energy cost of food intake and other metabolic pro- the other flow is that of system maintenance. It may be
cesses in the body, we may define the system as the incorrect to assume that the energy cost of base-line
musculoskeletal system of the body. In this case, the processes is unaffected by the work rate during exercise (see
energy inflow is defined as the energy from glycogen and e.g., Cavanagh and Kram 1985; Stainbsy et al. 1980; van
other substrates, but it does not include the abovemen- Ingen Schenau and Cavanagh 1990). In fact, there is ample
tioned processes, nor the costs of supporting cardiac output evidence that this is not the case, especially at higher work
and ventilation, for example. rates; various processes are affected, e.g., gastrointestinal,
splanchnic metabolism, general metabolic processes due to
Net efficiency temperature increase via the Q10 effect, and ventilation (e.g.
Stainbsy et al. 1980). On the other hand, gross efficiency
Energy consumption is often calculated from oxygen increases with work rate (see below) because of the
uptake and lactate production, the latter especially in case diminishing effect of offset (base-line) metabolism (at rest
of exercise performed above the lactate threshold. Often, or zero work rate) as work rate increases. In other words,
one is interested in the musculoskeletal system as the when studying the effect of work rate, the unqualified
energy conversion system. In such a case, it is common use of gross efficiency seems rather meaningless when
practice to subtract all energy costs that are not related attempting to enhance our understanding of the energy flux
directly to work production (e.g., associated with resting process; gross efficiency will, of necessity, increase with
metabolism) from the estimation based on oxygen uptake work rate. Refuting the use of net efficiency as a true
and lactate production. Various methods and efficiency expression of efficiency does not mean that we disagree
definitions have been proposed to handle this challenge, but about the existence of various energy consuming body
none are without problems. The basic principle is that one functions that have no, or little, direct bearing on
estimates the energy consumption of all processes that are mechanical work production. Rather, we disagree with
not part of the energy flow through the defined system of the philosophy that these functions have no impact on work
interest (e.g., muscle; contractile element). This is referred production and that work production does not rely at all on
to as base-line subtraction (Gaesser and Brooks 1975; these processes.
Stainbsy et al. 1980). The main aim of base-line subtraction
is to establish a measure that refers to the efficiency of Internal work and work efficiency
muscle contraction in vivo. While this procedure may be
relatively simple in engineering, it is not so in biology Internal work is often defined as the work necessary to
because of the many interactions and interdependences move the body segments relative to the body’s centre of
between physiological systems. Perhaps the most funda- mass, i.e., changing the relative kinetic and potential
mental of base-line subtractions is that of the resting energy of these segments. Often, this is redefined as posi-
metabolic energy rate (e.g., Gaesser and Brooks 1975). The tive work done to accelerate the limbs relative to the centre
reasoning is that the resting metabolic rate is needed to of mass of the body (e.g., Cavagna et al. 2008). The
maintain overall system homeostasis, irrespective of the negative component, i.e., reduction of body-segment
work being performed, and thus is not associated with movement energy relative to the centre of mass, is often
doing this work. ‘‘removed’’ from the calculations (by taking the absolute
values of their energy changes, e.g., Thys et al. 1996;
power
enet ¼ : Willems et al. 1995). This approach implies the presump-
metabolic rate  rest metabolic rate tion that this component is an energy loss (i.e., always
By using such subtraction, one implicitly assumes that converted into heat, never into external work). There is no
the processes related to resting metabolism are independent, basis for this presupposition. For cyclic movements such as
constant (Stainbsy et al. 1980) and, more importantly, walking, running, and cycling it is obvious that, over an

123
4 Eur J Appl Physiol (2009) 106:1–14

entire movement cycle, the net change of this relative When summating internal work and external work as the
movement energy equals zero. In other words, the total total work done by a system (e.g., Minetti et al. 2001;
internal work in cyclic movements equals zero. Neptune Winter 1979), two questionable steps are taken. Firstly, the
and Herzog (1999) found that more negative work is done system definition is unclear, as an ‘internal’ energy con-
on the crank with increasing cadence, particularly above version is, in a way, regarded an outflow. For example,
90 rpm. Thus, it is likely that more energy is lost by Winter (1979) defined mechanical efficiency as the sum of
decreasing mechanical effectiveness. It should be noted internal and external work divided by the metabolic cost.
that Neptune and Herzog (1999) did not link this increase However, he did not define the energy converting system
to internal work, but rather to muscle activation dynamics, that this efficiency is a measure for. According to logical
i.e. neuromuscular coordination aspects. The definition of semantics, the term ‘internal’ cannot appear in the
work efficiency is based on the assumption that the meta- numerator in an efficiency definition. Secondly and more
bolic cost for unloaded cycling (or any other activity) is not importantly, one runs the risk of counting twice a part of
utilized for doing external work. The rationale is well the work done, overestimating the calculated efficiency.
explained by Whipp and Wasserman (1969) and illustrated For example, applying this method, Widrick et al. (1992)
in their Fig. 1: report mechanical efficiencies of above 40%, i.e. suffi-
power ciently high to suggest a flaw. Furthermore, their data (see
ework ¼ : Fig. 1 in Widrick et al. 1992) indicate a negative intercept
metabolic rate  metabolic rateunloaded
The metabolic rate at unloaded cycling is the sum of the for the work rate—energy expenditure relationship. This
resting metabolic rate and the metabolic rate required for implies a negative muscular efficiency, if it is assumed that
doing internal work. We do not agree with the use of work the sum of internal and external work is a valid measure for
efficiency as a measure for muscular efficiency, nor do we muscular work. The use of the misleading notion that
support the notion that summation of internal and external internal work is not related to doing external work has been
work is a valid estimate for muscular work. The reasons are similarly criticized (e.g., Kautz et al. 1994; Kautz and
explained below. Neptune 2002). Figure 1 shows a conceptual diagram
explaining how, in principle, the various energy deposits
and work transitions are linked to each other. The changes
Energy influx
in the energy levels of the ‘‘internal energy depots’’ (elastic
energy and body-segment movement energy) are internal
or external work transitions, but not both at the same time.
The main message here is that often we do not have
information on one of the depots (elastic energy), but not
muscle contraction on the direct transition of muscle work to external work.
pos internal work
This lack of information makes it impossible to judge,
muscle contraction

internal a priori, a reduction in body-segment movement energy as


external work

pos. internal
work
movement a loss. We do not argue here that the measurements on
energy
internal work, or better body-segment movement energy, is
pointless, but we do argue against the interpretation of this
neg. internal

internal
work

neg. internal measure as an ‘‘internal loss of energy’’, and against the


elastic
work
energy summation of internal and external work as a measure for
muscular work. For the same reason, we argue that the use
of work efficiency as a measure for muscular efficiency is
based on a flawed assumption.
A practical and at first sight elegant solution for mea-
suring the true metabolic cost of losses by doing internal
work is the measurement of the metabolic cost during
External work unloaded movements (e.g., Dickinson 1929; Whipp and
Wasserman 1969; Gaesser and Brooks 1975; Hagberg
et al. 1981; Hintzy-Cloutier et al. 2003; Nickleberry and
Brooks 1996). If one creates a condition in which no
Fig. 1 Conceptual model for energy flow during exercise. The terms external work can be done, all work done by the muscular
‘positive’ (pos) and ‘negative’ (neg) indicate direction of energy flow
system is related to the body-segment movement energy,
only. One of the main issues is that internal work, as described in the
literature, cannot be assumed to be loss of energy. Further, every and will be dissipated into heat. This is in essence a pure
energy conversion (arrow) implies energy loss (as heat) base-line subtraction. The problem that arises is the same

123
Eur J Appl Physiol (2009) 106:1–14 5

as discussed previously, namely that this approach pre- Whipp and Rossiter 2005). Hintzy-Cloutier et al. (2003)
sumes two independent energy flows. Furthermore, by found that the extrapolation method results in lower val-
measuring the energy cost of the unloaded movement, not ues than true unloaded cycling, and discuss some reasons
only is all internal work dissipated into heat, but it must be for this.
dissipated into heat because the circumstances do not allow
external work production. There is no reason to believe Delta efficiency
that in the case where external work can be done, the same
internal work is not (partly) converted to external work. It is only a small step from work efficiency to delta-effi-
Moreover, it appears that by moving the lower extremities ciency (e.g., Bijker et al. 2001, 2002 ; Chavarren and
passively, i.e. by external forces, metabolic rate increases Calbet 1999; Coyle et al. 1992; Garry and Wishart 1931;
significantly (Nobrega et al. 1994). This indicates that Marsh et al. 2000; Mora-Rodriguez and Aguado-Jimenez
other processes than simply the active limb movements 2006; Nickleberry and Brooks 1996; Sidossis et al. 1992)
also affect metabolic rate. Unpublished data from our defined as:
laboratory indicate that muscle activity in unloaded cycling
Dpower
is extremely low and can hardly account for the total eD ¼ ;
increase of metabolic rate that is usually observed in Dmetabolic rate
unloaded cycling (e.g., about 200–450 W energy con- where Dpower and Dmetabolic rate stand for increment of
sumption at 0 external work rate, at 60–120 rpm, Hagberg power and metabolic rate with increasing work rate. The
et al. 1981; Sidossis et al. 1992). Two other processes that advantage of such a measure is that knowledge about
at first sight are obvious candidates are the enhanced resting metabolic rate is not required, and the measure is
heart—and ventilation rate. However, there is ample evi- likely to be less affected by potential changes in the base-
dence that these processes require comparatively little line energy cost caused by work rate. However, the same
additional energy consumption (e.g., McGregor and fundamental problem remains, namely, that implicitly one
Becklake 1961; Aaron et al. 1992; Kitamura et al. 1972) assumes that the energy flow for the Dpower production is
and thus can hardly explain this phenomenon. Further- independent of the energy flow for the first amount of
more, doing more ineffective work due to coordination power produced. This is the same as stating that when
challenges (Neptune and Herzog 1999) likely enhances increasing work rate one turns on an ‘extra’ engine
metabolic rate in passive cycling. This was substantiated (muscle, motor units) that runs independently from the
by Bell et al. (2003), who found considerable muscle other engine(s) that produces the initial amount of power
activity and a coinciding increase in metabolic rate during production. Delta efficiency is not, by definition, an
pure passive cycling compared to other modes of passive integral measure for the entire energy conversion process.
leg movements. In that study, subjects reported it was It will be independent of the protocol (work rate incre-
difficult to relax completely in passive cycling. Thus, it ments) only if the metabolic cost–work rate relationship is
seems difficult to experimentally determine the energy cost linear (e.g., Bijker et al. 2001, 2002; Chavarren and
of true internal work (i.e., work that never appears as Calbet 1999; see also Fig. 2). This implies that all ‘extra’
external) in loaded cycling. engines that are turned on with increasing work rate will
Another method to determine the costs of body seg- show the same efficiency. In that case, the efficiency of
ments’ movement energy changes is by extrapolating the one engine is equal to the efficiency of all engines when
relationship between external work rate and energy cost to they are considered together as one unit. In other words,
a zero work rate (e.g., Hintzy-Cloutier et al. 2003). Such when using delta efficiency as a measure for muscular
an approach requires that the several work rates that are efficiency, the a priori assumption is made that efficiency
used entail the same body segments’ movement energy. is independent of work rate. Note that a linear relationship
In cycling, this requirement is fulfilled by using the same does not imply that delta efficiency provides a valid
cadence. Furthermore, it is expected that the energy measure, and neither does the finding that the efficiency
cost—work rate relationship is linear, which has been value obtained are realistic (i.e., between say 20 and
substantiated empirically in many studies (e.g., Bijker 25%). These findings merely hold this option open.
et al. 2001, 2002; Chavarren and Calbet 1999; Hintzy- Nevertheless, theoretically impossible efficiency values
Cloutier et al. 2003; Widrick et al. 1992; see also the indicate a flaw in the calculations. For example, Bijker
literature results gathered in this review, Fig. 2), although et al. (2001, 2002) report a delta efficiency for running of
it should be noted that for sustained work rates that around 50%, which seems to be an unlikely, if not
exceed the lactate threshold this relationship may become impossible, true efficiency. Note that we do not claim that
nonlinear because of the influence of the so-called ‘‘slow the use of delta efficiency is meaningless; rather we claim
component’’ of oxygen uptake (e.g., Poole et al. 1994; that it is not a measure for efficiency. Instantaneous

123
6 Eur J Appl Physiol (2009) 106:1–14

30 30
a b

Gross efficiency (%)

Gross efficiency (%)


25 25

**
20
* * 20
*
** Luthanen et al., (1987)

15 * 15

10 * 10

5 5
30 45 60 75 90 105 120 0 100 200 300 400 500
Cadence (rpm) External power (W)
2500
30
c d
2000

Gross efficiency (%)


Metabolic rate (W)

25

1500
20

Samozino et al. (2006)


1000 Moseley and Jeukendrup (2001) 15
Luhtanen et al. (1987)

Chavarren & Calbet (1999)


500 10
Moseley et al. (2004)

Sallet et al. (2006)

0 5
0 100 200 300 400 500 0 100 200 300 400 500
External power (W) External power (W)

2500 30
e f
2000
Gross efficiency (%)
Metabolic rate (W)

25

1500 20

1000 15
120

500 10
60

0 5
0 100 200 300 400 500 0 100 200 300 400 500
External power (W) External power (W)

efficiency is the same as delta efficiency for an infinitely offset (i.e., base-line metabolism) with increasing work
small Dpower, and in fact a more accurate measure than rate. Gaesser and Brooks (1975) considered this a cal-
delta efficiency. However, the same problems apply to culation artifact and therefore rejected the use of gross
instantaneous efficiency as for delta efficiency. efficiency, a standpoint we dispute; it is rather a mat-
ter of interpretation of what a true measure is, and
The use of gross efficiency that depends on the definition of the energy converting
system.
The curved work rate–gross efficiency relationship is a In a final remark on efficiency definitions and interpre-
consequence of the decreasing relative contribution of the tation, we wish to note once more that the problem

123
Eur J Appl Physiol (2009) 106:1–14 7

b Fig. 2 Overview of literature data explored in this review and used in Efficiency in cycling
the quantification of efficiency. The different symbols indicate
different studies. Figures a–d show the mean values at a particular
Here, we are concerned with the literature on efficiency in
cadence or external power for each study. a Gross efficiency against
cadence. *: Dickinson (1929), net efficiency for comparison. b Same cycling, particularly regarding the influence of work rate
data against external power. A low efficiency (Luhtanen et al. 1987) is and cadence. This paper does not aim to discuss perfor-
indicated. Two values at high power are boxed (Lucia et al. 2002 mance enhancement in competitive cycling by optimisation
(open circle); Coyle 2005 (filled circle)] and discussed in more detail
for energy expenditure in cycling. Nevertheless, it appears
in the text. c Metabolic rate against external power, based on same
data as in a and b. Boxed values are same as boxed in b. Data by that one tends to freely choose a pedalling rate that is
McDaniel et al. (2002) are shown in grey squares for comparison. A somewhat above the energetically optimal one (e.g., Foss
few studies, with somewhat different findings, are indicated in the and Hallén 2005). Some authors argue that this hampers
legend and discussed in the text. d Same data as in b, but depicting a
performance (e.g., Foss and Hallén 2005; Hansen and
possible error of measurement of 5%. Thick curve is the average
curve, based on the regression line in b. Thin curves indicate ranges if Ohnstad 2008). Others use this finding to argue that the
both metabolic rate and external power have deviation (error) of 5%, human body apparently does not ‘care’ about minimising
but in opposite directions. A thick vertical error bar indicates the energy expenditure (e.g., Redfield and Hull 1986), or
same range if only one of the measures has a 5% deviation; the thin
consider other optimisation criteria, such as muscle acti-
horizontal arrows indicate the efficiency difference following from
this error. Filled marker represents the highest power in the study by vation (e.g., Neptune and Hull 1999). While we do not take
Luhtanen et al. (1987). Its deviation from power measures in other a stand on this issue in this paper, the findings summarised
studies is discussed in the text. e, f Same diagrams as c and d, here may be of relevance for its ultimate resolution.
respectively, but now showing all reported values for different
cadences at one particular power. Thin curves are identical to figures
c and d, and thick curves are those based on all values. Inset in e Cadence and work rate
shows data from Chavarren and Calbet (1999), indicating effect of
cadence (60–120 rpm). Studies presented in the figure: Cannon et al. The two most obvious variables that may affect efficiency
(2007), Chavarren and Calbet (1999), Coyle et al. (1992), Coyle
are cadence and work rate. Cadence is often thought of a
(2005), Delextrat et al. (2003), Foss and Hallén (2004), Gaesser and
Brooks (1975), Hansen et al. (2002), Hintzy et al. (2005), Hopker rather simple and straightforward ‘gear between force
et al. (2007), Horowitz et al. (1994), Lucia et al. (2002, 2004), (torque) and velocity (angular velocity) of muscle con-
Luhtanen et al. (1987), Moseley and Jeukendrup (2001), Moseley traction. Thus, the energetically optimal cadence is likely
et al. (2004), Mora-Rodriguez and Aguado-Jimenez (2006); Mourot
to be found at a muscle contraction speed that is close to
et al. (2004); Nickleberry and Brooks (1996); Sallet et al. (2006);
Samozino et al. (2006); Sidossis et al. (1992); Unpublished data— maximal power and efficiency in isolated muscle (i.e.,
Elite; Unpublished—Trained; Widrick et al. (1992); Zameziati et al. around 0.3 of maximal force and contraction velocity; e.g.,
(2006) see Barclay et al. 1993). Kohler and Boutellier (2005)
argued that the freely chosen cadence may not follow the
principle of minimising energy cost because of the dis-
presented here is not new. For example, Cavanagh and crepancy between velocities giving maximal power and
Kram (1985), and before them Stainbsy et al. (1980), efficiency. However, their analysis does not account for a
presented the problem of base-line subtractions and inter- number of processes that are affected by cadence as well.
pretation of such efficiencies in depth. Cavanagh and Kram For example, because of activation-relaxation dynamics,
(1985) refer to net-efficiency, delta-efficiency as concep- relatively more time is consumed at higher cadences by the
tually flawed, and van Ingen Schenau (1998) argued activation and relaxation process. Furthermore, inertial
strongly against the subtraction of internal work. It seems, effects by rotating lower limb masses lead to a change from
however, that the use of these methods in the scientific muscle performance to performance on the pedals (e.g.,
literature is continues unabated. In the remainder of this Ettema et al. 2009; Kautz and Hull 1993; Kautz and
review on efficiency in cycling, we will take gross effi- Neptune 2002; Lorås et al. 2009). Ettema et al. (2009)
ciency as the departure point, and discuss delta efficiency demonstrated that details in cycling technique change with
with the critical note that we do not consider it to be a valid cadence. In other words, the concept of treating choice of
measure for efficiency. Stainbsy et al. (1980) also indicated cadence as a mere ‘gearing’ between force and velocity of
that gross efficiency does not resemble muscular efficiency muscle contraction may be attractive, but it probably does
(independently of how the muscle is defined as an energy not fully hold.
conversion system). In other words, for theoretical and To summarize the extensive literature on efficiency in
practical reasons, an attempt to determine muscular effi- cycling and the effect of work rate and cadence, we plotted
ciency in whole body movements seems a fruitless the results of a large (but certainly not complete) set of
exercise. Accordingly, in the remainder of this paper, we studies that report gross efficiency in cycling. The studies
will focus on gross efficiency in cycling, reflecting effi- include untrained up to elite and professional cyclists
ciency of the entire human body in action. (including world top), different exercise protocols, and

123
8 Eur J Appl Physiol (2009) 106:1–14

500
a cadence. The inter-study variation may easily be thought to

External power (W)


400 be caused by methodological differences. However, when
300 plotting the same pool of data against external power, a
200
different picture is shown. A very consistent relationship
100
between work rate and efficiency is found. This relation-
ship is even more clearly demonstrated by plotting the
0
0 20 40 60 80 100 120 140 metabolic rate against work rate (Fig. 2c). A linear rela-
Cadence (rpm) tionship is found, which is not unexpected but merely
reflecting what various studies have reported explicitly
2200
b (e.g., Anton-Kuchly et al. 1984; Bijker et al. 2001, 2002;
2000
Chavarren and Calbet 1999; Coast and Welch 1985;
1800
Francescato et al. 1995; Gaesser and Brooks 1975; Hintzy-
Metabolic rate (W)

1600
Cloutier et al. 2003; McDaniel et al. 2002; Moseley et al.
1400
2004; Widrick et al. 1992). As stated before, the curved
1200
work rate–gross efficiency relationship is a consequence of
1000
the offset (y-intercept) of the work rate–metabolic rate
800
relationship. Note, that this offset does not, per sé, indicate
600
any fixed baseline energy cost that, physiologically, is
400 independent of work rate. The rather surprising aspect of
200 the result is the high consistency between the various
0 studies regarding the work rate–metabolic rate relationship,
0
25 where it seems to be lacking as a function of cadence.
Ca 50
den 75 Although one should be cautious with the interpretation of
ce 400 500
(rpm 100 300
) 100 200 correlations here, that between metabolic rate and external
125 0 wer (W)
External po power amounts to 0.97 (n = 93, p \ 0.0001; 26 studies, 29
Fig. 3 a Cadence plotted against work rate for all data considered in
conditions/subject groups, meaning that 94% of the varia-
this overview (see Fig. 2). b Energy expenditure plotted against work tion among all (mean) energy expenditure values for all
rate and cadence (same data as in Fig. 3a and Fig. 2e and f. Coloured these situations is explained by absolute work rate. This
mesh is the two-dimensional linear regression (red marker indicates outcome is only slightly more ambivalent when separate
extrapolated intercept at zero load and zero cadence). Markers
distinguish data below (filled) and above the mesh (open). The
data for all different cadences at the same power output
regression is described by: Metabolic rate = 39.7 (±29.3) ? 2.84 were entered (in 9 studies), as shown in Fig. 2e. Also when
(±0.34) 9 rpm ? 3.73 (±0.08) 9 External Power converting the data to work rate-efficiency curves, only
small differences with the original calculations occur
various procedures to calculate metabolic rate and external (Fig. 2f), with the correlation being reduced to 0.95
power. However, all studies used respiratory exchange (r2 = 0.91). In other words, factors other than work rate,
ratio values to convert oxygen uptake rate to metabolic including cadence, explain less then 10% of the variation in
rate. Where the anaerobic contribution was considerable, energy expenditure. Adding cadence as a dependent factor,
either the relevant data points were not considered in the the explained variance is increased to 94% (cadence
original reports or lactate levels were taken into consider- explains about 10% on its own). These findings, both
ation and converted to metabolic rate, according to, for correlation values as well as the absolute cost-work rate
example, di Prampero (1981). To avoid over-representation relationship, agree well with McDaniel et al. (2002)
of studies that examined a matrix of cadence and work rate, (redrawn in grey in Fig. 2c, but not included in the anal-
we averaged their results according to cadence and work ysis), who looked at cadence, work rate and movement
rate before entering them into the figures (Figs. 2a–d). speed (by altering crank length). In their study, 95% of all
Thus, per study, one single data entry for each work rate variation in metabolic cost, including all experimental
and each cadence was used. All separate combinations of conditions, was explained by work rate. In the present data
cadence and power are shown in Fig. 2e, f. pool, cadence and power are correlated to some extent
Figure 2a shows the data according to cadence. Even (r = 0.171, p \ 0.019; Fig. 3a), which complicates the
though most studies report a clear negative effect of interpretation somewhat as these two factors share some of
cadence on gross efficiency, the overall picture shows a their variance. Still, both factors seem reasonably evenly
minimal effect. The inter-study variation is much larger spread over all data considered in this overview (Fig. 3a).
than any visible trend, and some studies show the opposite Therefore, it is unlikely that this correlation between work
(positive) effect or an inverted u-shape with an optimal rate and cadence has a strong effect on the findings.

123
Eur J Appl Physiol (2009) 106:1–14 9

Table 1 Average values of the reciprocal slope of work rate–meta- not only on aerobic, but also on anaerobic contributions if
bolic rate relationship (RSep-mr) calculated from (and reported by) a relevant. Still, the estimates of the anaerobic contribution are
number of studies on cycling
bound to be less accurate than the aerobic counterpart. The
Source RSep-mr curvilinear increase of metabolic rate with work rate is likely
related to the slow component of O2 uptake that emerges at
Chavarren and Calbet (1999) 22.2
intensities above lactate threshold (e.g., Poole et al. 1994;
Gaesser and Brooks (1975) 26.2
Whipp and Rossiter 2005). That is, for constant work-rate
Hansen et al. (2002) 24.4
exercise, VO2 shows a further slow increase (after a delay of
Luhtanen et al. (1987) 17.8
2–3 min). Thus, some of the differences between studies
Moseley and Jeukendrup (2001) 25.5
may be due to the emergence of the VO2-slow component
Moseley et al. (2004) 21.5
especially at high work rate (e.g., Luhtanen et al. 1987;
Nickleberry and Brooks (1996) 25.0
Moseley and Jeukendrup 2001; Moseley et al. 2004).
Samozino et al. (2006) 21.6
Clearly, both work rate and exercise duration are of impor-
Sidossis et al. (1992) 22.1
tance when comparing efficiency results. The studies
Widrick et al. (1992) 25.4 discussed in this paper tend to have relatively short time
Zameziati et al. (2006) 27.6 periods of measurement at constant work rates (2–3 min.).
Zamparo et al. (2002) 23.4 Thus, the impact of the VO2-slow component is likely not
Unpublished—elitea 22.7 more than moderate.
Unpublished—trainedb 27.4 Coyle and coworkers (e.g., Coyle 2005; Sidossis et al.
Mean 23.8 1992) report that gross efficiency is independent of work
Standard deviation 2.6 rate. This seems in contradiction with the current overview
a
Cyclists from the national Norwegian team (time trial). Measure- of the literature that includes their publications. However,
ments done in same lab as (b) considering Fig. 2, it becomes clear that the impact of work
b
Data are from the same study as Ettema et al. (2009) and Lorås rate on gross efficiency diminishes strongly from about
et al. (2009), but not reported in these publications 150 W. And, as mentioned above, a negative trend can be
discerned in some studies, but these they are explained (at
least partially) by relatively high work rates close to the
Interestingly, the intercept of the two-dimensional regres- individual’s maximum.
sion at zero work rate and zero cadence (Fig. 3b), which In summary, absolute external power determines not
would be the theoretical value for energy expenditure while only metabolic rate, but also gross efficiency in a more
sitting still on a bicycle, reaches a value of 40 W (not consistent manner than cadence does.
statistically significant from zero). This value is too low,
but still physically possible, despite the rather large Reciprocal slope of work rate–metabolic rate
extrapolation range from the experimental data. Overall, it relationship (delta efficiency)
seems that the very original findings by Fenn (1924) on
isolated muscle also apply to the entire human body in As we dispute the idea that delta efficiency reflects true
cycling in a very consistent manner. efficiency, we will refer to it here as the reciprocal slope of
In the literature data in Fig. 2c, some deviations appear: in work rate–metabolic rate relationship (RSep-mr). The data
two cases the offset in metabolic rate is somewhat higher in Fig. 2c show a RSep-mr of 25.5% (26.1% when using all
(Chavarren and Calbet 1999; Samozino et al. 2006), and in a data separately, Fig. 2e). These values are similar to Bijker
few others the metabolic rate increases exponentially at high et al. 2001, 2002. Coyle (2005) reports delta efficiency to
work rates (Luhtanen et al. 1987; Moseley and Jeukendrup be very similar to gross efficiency in cycling (around 21–
2001; Moseley et al. 2004). The higher offset cannot be 23%) for one of the world top cyclists. This would imply
explained, but the exponential increase may be because the that the corrected baseline approaches zero or is negligible
subjects exercised above lactate threshold and approached with regard to total metabolic rate, which may in fact be the
their maximal work capacity. In such an instance, one may case as these efficiencies were recorded at extremely high
expect that an increase in work rate requires a dispro- work rates. Sidossis et al. (1992) find that delta efficiency
portional amount of metabolic input (e.g., because of deviates from gross efficiency mainly at high cadence
deterioration of coordination). In the case of Luhtanen et al. (100 rpm). This is explained by that the base-line meta-
(1987), where the highest three work rates were at and above bolic rate (i.e., at zero work rate) depends on cadence. The
anaerobic threshold, this leads to a negative relationship literature data in the present paper indicate that most, if not
between gross efficiency and work rate (Fig. 2b). Note that all, studies have very similar RSep-mr values (Fig. 2c),
for all data presented in Fig. 2, the metabolic rate was based the overall value being close to 26%. Table 1 shows the

123
10 Eur J Appl Physiol (2009) 106:1–14

RSep-mr values of a number of studies that reported meta- one assumes that resting metabolic rate and the cost of limb
bolic rate at various work rates. Most of these values are movements is (physically) independent of work rate, delta
not taken from the original papers, but calculated from the efficiency is a measure of the increasing cost directly
data collected for this review. Thus, for all studies the same linked to increasing muscle work. Marsh et al. (2000)
algorithm was used. Some studies show substantially lower basically substantiated that the impact of increasing work
reciprocal slopes than the average, but none much higher. rate is independent of movement speed. Chavarren and
This is explained by the different weighting in the calcu- Calbet (1999), however, report a significant positive effect
lations of these studies: the studies with the higher slopes of cadence on delta efficiency. Their work rate–metabolic
have more data points. Thus, it seems more appropriate to rate data are redrawn in Fig. 2e, inset. Although the trend
conclude that the RSep-mr as found in the literature averages of a changing slope is evident, it is small considering the
around 23–24% rather than 26%. range of cadence (eD = 21.5% at 60 rpm to near 24% at
120 rpm). On the other hand, Chavarren and Calbet (1999)
Is there an energetically optimal cadence? report a stronger negative effect of cadence on gross effi-
ciency, indicating that we are not only dealing with
When considering overall effectiveness for the entire processes associated with work rate, but also other pro-
human body (i.e. total energy cost in relation to external cesses (e.g., force-velocity properties, activation-relaxation
power), most studies that looked at a rather wide range of dynamics, energy flow associated with internal work,
cadences and are represented in the analysis in this review ventilation and circulation).
report that the lowest pedalling rate is most effective In early work, Dickinson (1929), but also Garry and
(Chavarren and Calbet 1999; Gaesser and Brooks 1975; Wishart (1931, 1934), studied the relationship between
Lucia et al. 2004; Nickleberry and Brooks,1996; Samozino cadence and efficiency, basing their analysis on Hill’s
et al. 2006; Sidossis et al. 1992; Widrick et al. 1992). Two force-velocity equation. (Note that they expressed speed in
studies claim the highest rate to be most effective, and 2 terms of time of contraction.) Dickinson (1929) established
others an optimal cadence (Foss and Hallén 2005; own efficiency for maximal efforts (i.e., well over the lactate
unpublished data). Other studies (e.g., Coast et al. 1986) threshold, and with no steady state) by including oxygen
also find an optimal cadence with regard to efficiency. In uptake during recovery. She subtracted resting metabolism
most of these studies, the optimal cadence lies between 60 (i.e., calculated net efficiency). We calculated gross effi-
and 80 rpm. These contradictory results may, in part, be ciencies from these original data, leading to values around
explained by the interaction between work rate and 4–9%. These extremely low values occur because of the
cadence. In short, when an optimal cadence is found, it long period of oxygen uptake measurement (30 min) in
increases with work rate (Böning et al. 1984; Coast and relation to the exercise period (1–10 min), once more
Welch 1985; Francescato et al. 1995; Foss and Hallén clearly addressing the practical and theoretical challenges
2004; Gaesser and Brooks 1975; Seabury et al. 1977), and around true efficiency of work production. Some of
the impact of cadence on efficiency seems most remarked Dickinson’s original data are re-plotted in Fig. 2a against
at lower work rates (Chavarren and Calbet 1999; Samozino cadence. The optimal frequency lies around 35 rpm (all
et al. 2006; Widrick et al. 1992). Also di Prampero (2000), data to the left that are omitted were lower). This is rather
in his review, concluded that efficiency in cycling is striking as these measurements were made at a high work
affected by cadence and the optimum by work rate. rate. On the other hand, the data are not out of the ordinary
Extrapolations from cadence studies to the force- compared to the more recent studies accounted for in Fig. 2.
velocity relationship of muscle should, however, be made Furthermore, Dickinson calculated net efficiency, assuming
with caution. Indeed, Hill (1934) warned against this in his a constant and work independent of resting metabolism.
critical comments on Garry and Wishart (1931, 1934). Not An important consequence of the analysis of cadence and
only muscle shortening velocity, but also activation- work rate effects on efficiency is that power differences
relaxation dynamics are strongly affected by cadence. may be a confounder in experimental studies on cadence. In
McDaniel et al. (2002) attempted to distinguish between experimental testing, relatively small differences in external
these two factors by manipulating crank length. They found power that are related to cadence may occur. These dif-
that pedal speed (marker for muscle shortening speed) and ferences can have a relatively large impact on the cadence-
power (work rate) were the main determinants for meta- efficiency relationship. Furthermore, relatively small errors
bolic rate, not cadence (marker for activation-relaxation in power measurements affect the efficiency value consid-
dynamics). Marsh et al. (2000) found no effect of cadence erably (see below, Fig. 2d). An error of 5% at 285 W
on delta efficiency, where values were around 23–26%. (14.25 W) gives an efficiency error of 1% (Fig. 2d). In other
This does not contradict findings about optimal cadence words, it is important that possible systematic errors in
with regard to energetic cost. For the sake of argument, if power linked to cadence are eliminated.

123
Eur J Appl Physiol (2009) 106:1–14 11

In summary, because of the fundamental challenges of measurements with regard to studies on cycling efficiency.
discriminating the various mechanisms of energy expen- Figure 2d shows the ranges of efficiency calculations that
diture and losses that relate differently to cadence (e.g., arise from 5% error in both metabolic rate and external
expenditure not directly associated to doing work, force- power going in opposite directions. The vertical bar shows
velocity characteristics of muscle), as well as accuracy the range near 300 W if only one of these measures has that
issues, a true cadence-efficiency relationship has still not same error. Only one data point falls clearly outside the
been established. Overall (Fig. 3b), there seems to be a range of 5% error (filled circle). This is the result from
small negative effect of increasing cadence on efficiency. Luhtanen et al. (1987) at the highest work rate, which was,
as mentioned earlier, well above the lactate threshold and
Do elite athletes have high efficiencies and does thus bound to result in a lower efficiency. Thus, the dif-
training improve efficiency? ference between studies may be partly explained by
differences in (systematic) errors. This merely strengthens
Lucia et al. (2002) reported rather high gross efficiency the notion that cycling is an extremely consistent exercise
values for some top cyclists. The average for the group model with regard to the relationship between metabolic
amounted to 24.5% (with a peak individual value at rate and external power. Thus, the situation presented in
28.1%). Jeukendrup et al. (2003) argued that these results Fig. 2c may constitute a very solid framework for the
were extremely high from a theoretical point-of-view and interpretation of past, present and future studies.
must have been affected by errors in the measurements (see
also below, next section). They furthermore concluded that
if these data were correct, ‘‘some interesting physiological Conclusion
adaptations may exist…’’. Coyle (2005) reported an
increase in efficiency over a period of 7 years of training We conclude this review by putting what was discussed
and competing in one of the most outstanding cyclists of earlier into a simple theoretical framework. In the tradition
modern times from about 21–23%. Coyle proposed that of Fenn (1924), one can factorise the metabolic costs as
biochemical adaptations may have caused this improve- found in cycling (Fig. 2): E = I ? kW, where E is meta-
ment (i.e., a greater contribution from aerobically-efficient bolic costs, I is the constant intercept (maintenance), k a
type I fibres). When considering these data and their constant (reciprocal delta efficiency), and W the external
placement within the data derived from the literature work done. When using delta efficiency as a measure for
(Fig. 2b, c; data enclosed in a grey square; only overall muscular efficiency, one assumes that the intercept of the
average is shown for both studies), these values do not work rate-metabolic rate relationship is not associated with
seem extraordinary, although Lucia et al. (2002) appear to muscular contraction and all energy increase is linked to
show a slightly high efficiency value. This is supported by the work accomplished. This is, of course, a tempting
values from Sallet et al. (2006) on elite and professional thought, but the physiological basis for it can be chal-
riders who score even higher efficiencies at powers above lenged; the only matter that is clearly established is a very
400 W (data most to the left in Fig. 2b, c). The main reason consistent linear work rate-metabolic rate relationship. The
why gross efficiency is relatively high is likely because of equation is likely better rewritten as E = Ic ? k(Iw ? W),
the high work rate. Also the improvement in efficiency with Iw = qW. In other words, total metabolic rate is built
reported by Coyle (2005) may be explained by an increased up from a constant rate (Ic), a work related rate (kW), and a
power at which these values were determined. Neverthe- component of energy consumption that is not directly
less, the studies by Sallet et al. (2006) and Lucia et al. associated with the work conversion process (muscle
(2002) show metabolic rates below the regression line in contraction) but changes linearly with it (Iw). These pro-
Fig. 2c, which may indicate either measurement error or, cesses may be, for example, ventilation and circulation, but
indeed, some physiological changes that enhance efficiency also digestive processes. Energy loss associated with rela-
above the increase that is directly linked to that for the tive movements of segments (note once more, not the
work rate. It is interesting to note that the same group entire internal work) would logically be accounted for by
(Lucia et al. 2004) report a lower efficiency is reported the Ic component: losses associated with internal work do
(23.4 vs. 24.5%) at a slightly lower power (366 vs. 385 W). not depend on external work rate, but more likely on
cadence (i.e., the amount of kinetic energy changes of the
How accurate are efficiency measurements? moving limbs). Indeed, both Sidossis et al. (1992) and
Chavarren and Calbet (1999) clearly show that with
Irrespective of definitions and concepts, a framework for the increasing cadence only Ic increases in a more or less linear
accuracy of efficiency measurements can be established. It fashion (see Fig. 2e, inset). It is tempting, but likely
seems reasonable to allow for a 5% error in biological incorrect, to conclude that this increase is solely due to

123
12 Eur J Appl Physiol (2009) 106:1–14

increase of internal work losses. Whatever the case, the Cavanagh PR, Kram R (1985) The efficiency of human move-
equation can be rewritten as follows: ment—a statement of the problem. Med Sci Sports Exerc 17:
304–308
E ¼ Ic þ kðq þ 1ÞW: Chavarren J, Calbet JA (1999) Cycling efficiency and pedalling
frequency in road cyclists. Eur J Appl Physiol Occup Physiol
While the efficiency of the pure work production system (a 80:555–563. doi:10.1007/s004210050634
precise definition of this system is not presented here) is Coast JR, Welch HG (1985) Linear increase in optimal pedal rate
k-1, the measured efficiency is {k(q ? 1)}-1. Somewhat with increased power output in cycle ergometry. Eur J Appl
Physiol Occup Physiol 53:339–342. doi:10.1007/BF00422850
speculatively, one can argue that if all non-associated Coast JR, Cox RH, Welch HG (1986) Optimal pedalling rate in
energy costs (Ic and kqW) could be accounted for accu- prolonged bouts of cycle ergometry. Med Sci Sports Exerc
rately in measurements, one should obtain a muscular 18:225–230. doi:10.1249/00005768-198604000-00013
efficiency of close to 30%. As delta efficiency reported in Coyle EF (2005) Improved muscular efficiency displayed as Tour de
France champion matures. J Appl Physiol 98:2191–2196. doi:
the literature is about 26% or less, one can conclude that q 10.1152/japplphysiol.00216.2005
must be about 0.1–0.15. This is, of course, based on the Coyle EF, Sidossis LS, Horowitz JF, Beltz JD (1992) Cycling
assumption that muscular efficiency in vivo and in isolated efficiency is related to the percentage of type I muscle fibers.
muscle are similar. However, we cannot take this as a Med Sci Sports Exerc 24:782–788
Cullen LK, Andrew K, Lair KR et al (1992) Efficiency of trained
departure point if we want to gain knowledge about muscle cyclists using circular and noncircular chainrings. Int J Sports
efficiency in vivo through experimentation. As long as we Med 13:264–269. doi:10.1055/s-2007-1021264
do not have better knowledge about the value for q, or Delextrat A, Tricot V, Bernard T et al (2003) Drafting during
about the validity of the entire equation for that matter, the swimming improves efficiency during subsequent cycling.
Med Sci Sports Exerc 35:1612–1619. doi:10.1249/01.MSS.
use of delta efficiency (or any other efficiency) in the 0000084422.49491.2C
search for muscular efficiency is fruitless. As many new di Prampero PE (1981) Energetics of muscular exercise. Rev Physiol
methods are being developed that can monitor energy Biochem Pharmacol. 89:143–222 (Cited in: di Prampero PE
consumption locally in the body under in vivo conditions, et al. (1993) Energetics of best performances in middle-distance
running. J Appl Physiol 74:2318–2324)
the future holds a number of challenges that may be di Prampero PE (2000) Cycling on Earth, in space, on the Moon. Eur J
realised. Appl Physiol 82:345–360. doi:10.1007/s004210000220
Dickinson S (1929) The efficiency of bicycle-pedalling, as affected by
speed and load. J Physiol 67:242–255
Ettema GJ (2001) Muscle efficiency: the controversial role of
elasticity and mechanical energy conversion in stretch-shorten-
References ing cycles. Eur J Appl Physiol 85:457–465. doi:10.1007/
s004210100464
Aaron EA, Seow KC, Johnson BD, Dempsey JA (1992) Oxygen cost Ettema G, Loras H, Leirdal S (2009) The effects of cycling cadence
of exercise hyperpnea: implications for performance. J Appl on the phases of joint power, crank power, force and force
Physiol 72:1818–1825 effectiveness. J Electromyogr Kinesiol 19:e94–e101 (online
Anton-Kuchly B, Roger P, Varene P (1984) Determinants of only). doi:10.1016/j.jelekin.2007.11.009
increased energy cost of submaximal exercise in obese subjects. Fenn WO (1924) The relation between the work performed and
J Appl Physiol 56:18–23 the energy liberated in muscular contraction. J Physiol 58:373–
Barclay CJ, Constable JK, Gibbs CL (1993) Energetics of fast- and 395
slow-twitch muscles of the mouse. J Physiol 472:61–80 Ferguson RA, Ball D, Sargeant AJ (2002) Effect of muscle
Bell HJ, Ramsaroop DM, Duffin J (2003) The respiratory effects of temperature on rate of oxygen uptake during exercise in humans
two modes of passive exercise. Eur J Appl Physiol 88:544–552. at different contraction frequencies. J Exp Biol 205:981–987
doi:10.1007/s00421-002-0771-5 Foss O, Hallén J (2004) The most economical cadence increases with
Benedict FG, Cathcart EP (1913) Muscular work. Publications no. increasing workload. Eur J Appl Physiol 92:443–451. doi:
187, Carnegie Institute of Washington (as referred to in 10.1007/s00421-004-1175-5
Dickinson (1929)) Foss O, Hallén J (2005) Cadence and performance in elite cyclists.
Bijker KE, de Groot G, Hollander AP (2001) Delta efficiencies of Eur J Appl Physiol 93:453–462. doi:10.1007/s00421-004-1226-y
running and cycling. Med Sci Sports Exerc 33:1546–1551. doi: Francescato MP, Girardis M, di Prampero PE (1995) Oxygen cost of
10.1097/00005768-200109000-00019 internal work during cycling. Eur J Appl Physiol Occup Physiol
Bijker KE, de Groot G, Hollander AP (2002) Differences in leg 72:51–57. doi:10.1007/BF00964114
muscle activity during running and cycling in humans. Eur J Gaesser GA, Brooks GA (1975) Muscular efficiency during steady-
Appl Physiol 87:556–561. doi:10.1007/s00421-002-0663-8 rate exercise: effects of speed and work rate. J Appl Physiol
Böning D, Gonen Y, Maassen N (1984) Relationship between work 38:1132–1139
load, pedal frequency, and physical fitness. Int J Sports Med Garry RC, Wishart GM (1931) On the existence of a most efficient
5:92–97. doi:10.1055/s-2008-1025887 speed in bicycle pedalling, and the problem of determining
Cannon DT, Kolkhorst FW, Cipriani DJ (2007) Effect of pedaling human muscular efficiency. J Physiol 72:425–437
technique on muscle activity and cycling efficiency. Eur J Appl Garry RC, Wishart GM (1934) The efficiency of bicycle pedalling in
Physiol 99:659–664. doi:10.1007/s00421-006-0391-6 the trained subject. J Physiol 82:200–206
Cavagna GA, Legramandi MA, Peyre-Tartaruga LA (2008) Old men Green HJ, Roy B, Grant S et al (2000) Increases in submaximal
running: mechanical work and elastic bounce. Proc Biol Sci cycling efficiency mediated by altitude acclimatization. J Appl
275:411–418. doi:10.1098/rspb.2007.1288 Physiol 89:1189–1197

123
Eur J Appl Physiol (2009) 106:1–14 13

Hagberg JM, Mullin JP, Giese MD, Spitznagel E (1981) Effect of cycling. Med Sci Sports Exerc 32:1630–1634. doi:10.1097/
pedaling rate on submaximal exercise responses of competitive 00005768-200009000-00017
cyclists. J Appl Physiol 51:447–451 McDaniel J, Durstine JL, Hand GA, Martin JC (2002) Determinants
Hansen EA, Ohnstad AE (2008) Evidence for freely chosen pedalling of metabolic cost during submaximal cycling. J Appl Physiol
rate during submaximal cycling to be a robust innate voluntary 93:823–828
motor rhythm. Exp Brain Res 186:365–373. doi:10.1007/ McGregor M, Becklake MR (1961) The relationship of oxygen cost
s00221-007-1240-5 of breathing to respiratory mechanical work and respiratory
Hansen EA, Jorgensen LV, Jensen K et al (2002) Crank inertial load force. J Clin Invest 40:971–980. doi:10.1172/JCI104336
affects freely chosen pedal rate during cycling. J Biomech Minetti AE, Pinkerton J, Zamparo P (2001) From bipedalism to
35:277–285. doi:10.1016/S0021-9290(01)00182-8 bicyclism: evolution in energetics and biomechanics of historic
Hill AV (1934) The efficiency of bicycle pedalling. J Physiol 82:207– bicycles. Proc Biol Sci 268:1351–1360. doi:10.1098/rspb.2001.
210 1662
Hintzy F, Mourot L, Perrey S, Tordi N (2005) Effect of endurance Mora-Rodriguez R, Aguado-Jimenez R (2006) Performance at high
training on different mechanical efficiency indices during pedaling cadences in well-trained cyclists. Med Sci Sports Exerc
submaximal cycling in subjects unaccustomed to cycling. Can 38:953–957. doi:10.1249/01.mss.0000218139.46166.ec
J Appl Physiol 30:520–528 Moseley L, Jeukendrup AE (2001) The reliability of cycling
Hintzy-Cloutier F, Zameziati K, Belli A (2003) Influence of the base- efficiency. Med Sci Sports Exerc 33:621–627. doi:10.1097/
line determination on work efficiency during submaximal 00005768-200104000-00017
cycling. J Sports Med Phys Fitness 43:51–56 Moseley L, Achten J, Martin JC, Jeukendrup AE (2004) No
Hopker JG, Coleman DA, Wiles JD (2007) Differences in efficiency differences in cycling efficiency between world-class and
between trained and recreational cyclists. Appl Physiol Nutr recreational cyclists. Int J Sports Med 25:374–379. doi:10.1055/
Metab 32:1036–1042. doi:10.1139/H07-070 s-2004-815848
Horowitz JF, Sidossis LS, Coyle EF (1994) High efficiency of type I Mourot L, Hintzy F, Messonier L, Zameziati K, Belli A (2004) Supra-
muscle fibers improves performance. Int J Sports Med 15:152– maximal cycling efficiency assessed in humans by using a new
157. doi:10.1055/s-2007-1021038 protocol. Eur J Appl Physiol 93:325–332. doi:10.1007/s00421-
Hull ML, Williams M, Williams K, Kautz S (1992) Physiological 004-1179-1
response to cycling with both circular and noncircular chain- Neptune RR, Herzog W (1999) The association between negative
rings. Med Sci Sports Exerc 24:1114–1122. doi:10.1249/ muscle work and pedaling rate. J Biomech 32:1021–1026. doi:
00005768-199210000-00008 10.1016/S0021-9290(99)00100-1
Jeukendrup A, Martin D, Gore CJ (2003) Are world-class cyclists Neptune RR, Hull ML (1999) A theoretical analysis of preferred
really more efficient? Med Sci Sports Exerc 35:1238–1239. doi: pedaling rate selection in endurance cycling. J Biomech 32:409–
10.1249/01.MSS.0000074558.64862.3B 415. doi:10.1016/S0021-9290(98)00182-1
Kautz SA, Hull ML (1993) A theoretical basis for interpreting the Nickleberry BL Jr, Brooks GA (1996) No effect of cycling experience
force applied to the pedal in cycling. J Biomech 26:155–165. on leg cycle ergometer efficiency. Med Sci Sports Exerc
doi:10.1016/0021-9290(93)90046-H 28:1396–1401. doi:10.1097/00005768-199611000-00008
Kautz SA, Neptune RR (2002) Biomechanical determinants of Nobrega AC, Williamson JW, Friedman DB, Araujo CG, Mitchell JH
pedaling energetics: internal and external work are not indepen- (1994) Cardiovascular responses to active and passive cycling
dent. Exerc Sport Sci Rev 30:159–165. doi:10.1097/00003677- movements. Med Sci Sports Exerc 26:709–714. doi:10.1249/
200210000-00004 00005768-199406000-00009
Kautz SA, Hull ML, Neptune RR (1994) A comparison of muscular Poole DC, Barstow TJ, Gaesser GA, Willis WT, Whipp BJ (1994)
mechanical energy expenditure and internal work in cycling. J VO2 slow component: physiological and functional significance.
Biomech 27:1459–1467. doi:10.1016/0021-9290(94)90195-3 Med Sci Sports Exerc 26:1354–1358
Kitamura K, Jorgensen CR, Gobel F, Taylor HL, Wang Y (1972) Redfield R, Hull ML (1986) On the relation between joint moments
Hemodynamic correlates of myocardial oxygen consumption and pedalling rates at constant power in bicycling. J Biomech
during upright exercise. J Appl Physiol 32:516–522 19:317–329. doi:10.1016/0021-9290(86)90008-4
Kohler G, Boutellier U (2005) The generalized force-velocity Sallet P, Mathieu R, Fenech G, Baverel G (2006) Physiological
relationship explains why the preferred pedaling rate of cyclists differences of elite and professional road cyclists related to
exceeds the most efficient one. Eur J Appl Physiol 94:188–195. competition level and rider specialization. J Sports Med Phys
doi:10.1007/s00421-004-1283-2 Fitness 46:361–365
Lorås H, Leirdal S, Ettema G (2009) Force effectiveness during Samozino P, Horvais N, Hintzy F (2006) Interactions between
cycling at different pedalling rates. J Appl Biomech 25:85–92 cadence and power output effects on mechanical efficiency
Lucia A, Hoyos J, Perez M et al (2002) Inverse relationship between during sub maximal cycling exercises. Eur J Appl Physiol
VO2 max and economy/efficiency in world-class cyclists. 97:133–139. doi:10.1007/s00421-006-0172-2
Med Sci Sports Exerc 34:2079–2084. doi:10.1097/00005768- Seabury JJ, Adams WC, Ramey MR (1977) Influence of pedalling
200203000-00021 rate and power output on energy expenditure during
Lucia A, San Juan AF, Montilla M et al (2004) In professional road bicycle ergometry. Ergonomics 20:491–498. doi:10.1080/
cyclists, low pedaling cadences are less efficient. Med Sci 00140137708931658
Sports Exerc 36:1048–1054. doi:10.1249/01.MSS.0000128249. Sidossis LS, Horowitz JF, Coyle EF (1992) Load and velocity of
10305.8A contraction influence gross and delta mechanical efficiency. Int J
Luhtanen P, Rahkila P, Rusko H, Viitasalo JT (1987) Mechanical Sports Med 13:407–411. doi:10.1055/s-2007-1021289
work and efficiency in ergometer bicycling at aerobic and Stainbsy WN, Gladden LB, Barclay JK, Wilson BA (1980) Exercise
anaerobic thresholds. Acta Physiol Scand 131:331–337. doi: efficiency: validity of baseline subtractions. J Appl Physiol
10.1111/j.1748-1716.1987.tb08247.x 48:518–522
Marsh AP, Martin PE, Foley KO (2000) Effect of cadence, cycling Thys H, Willems PA, Saels P (1996) Energy cost, mechanical work
experience, and aerobic power on delta efficiency during and muscular efficiency in swing-through gait with elbow

123
14 Eur J Appl Physiol (2009) 106:1–14

crutches. J Biomech 29:1473–1482. doi:10.1016/0021-9290(96) Whipp BJ, Wasserman K (1969) Efficiency of muscular work. J Appl
84543-X Physiol 26:644–648
van Ingen Schenau GJ (1998) Positive work and its efficiency are at Widrick JJ, Freedson PS, Hamill J (1992) Effect of internal work on
their dead-end: comments on a recent discussion. J Biomech the calculation of optimal pedaling rates. Med Sci Sports Exerc
31:195–197 24:376–382. doi:10.1249/00005768-199203000-00014
van Ingen Schenau GJ, Cavanagh PR (1990) Power equations in Willems PA, Cavagna GA, Heglund NC (1995) External, internal and
endurance sports. J Biomech 23:865–881. doi:10.1016/0021- total work in human locomotion. J Exp Biol 198:379–393
9290(90)90352-4 Winter DA (1979) A new definition of mechanical work done in
Van Sickle J Jr, Hull ML (2007) Is economy of competitive human movement. J Appl Physiol 46:79–83
cyclists affected by the anterior–posterior foot position on the Zameziati K, Mornieux G, Rouffet D, Belli A (2006) Relationship
pedal? J Biomech 40:1262–1267. doi:10.1016/j.jbiomech.2006. between the increase of effectiveness indexes and the increase of
05.026 muscular efficiency with cycling power. Eur J Appl Physiol
Whipp BJ, Rossiter HB (2005) The kinetics of oxygen uptake: 96:274–281. doi:10.1007/s00421-005-0077-5
physiological inferences from the parameters. In: Jones AM, Zamparo P, Minetti AE, di Prampero PE (2002) Mechanical
Poole DC (eds) Oxygen uptake kinetics in health, disease. efficiency of cycling with a new developed pedal-crank. J
Routledge Publs, London, pp 64–94 Biomech 35:1387–1398. doi:10.1016/S0021-9290(02)00071-4

123

You might also like