NUMBER 51 FALL 2001

Summaries of articles in this issue; mast . . . . . . . . . . . . . . . 2

Contributions to Human Power . . . . . . . . . . . . . . . . . . . . . . 2
Erratum
Correction to article by Vernon Forbes in HP 50. . . . . . . . . 2
Articles
Determination of the crank-arm length to maximize
power production in recumbent cycle ergometry
Danny Too and Chris Williams . . . . . . . . . . . . . . . . . . . . . 3
Bicycle pitchover characteristics
Frederick Matteson. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
CdA and Crr measurement
John C. Snyder Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Technical notes
Efficiency of bicycle chain drives
Claire L. Walton & John C. Walton . . . . . . . . . . . . . . . . . 14
Further comments on the Spicer article
on drive-train efficiency, John S. Allen. . . . . . . . . . . . . . . 15
Bicycle stability after front-tire deflation
Dave Wilson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Book reviews
The Athenian Trireme, by J.S. Morrison and others,
reviewed by Theo Schmidt. . . . . . . . . . . . . . . . . . . . . . . . . 18
The Dancing Chain by Frank Berto
reviewed by Dave Wilson. . . . . . . . . . . . . . . . . . . . . . . . . . 19
Bicycle Design by Mike Burrows
reviewed by Dave Wilson. . . . . . . . . . . . . . . . . . . . . . . . . . 19
Editorials
Velomobiles, Joachim Fuchs . . . . . . . . . . . . . . . . . . . . . . . . 20
Open Road: the end of a dream, Dave Wilson . . . . . . . . . . 22
Tiresome, Dave Wilson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Number 51
Fall 2001 $5.50
HUMAN
POWER
TECHNI CAL JOURNAL OF THE I HPVA
H U MA N P O WE R
H U MA N P O WE R
is the technical journal of the
International Human Powered Vehicle
Association
Human Power 51, Fall 2001
Editor
David Gordon Wilson
21 Winthrop Street
Winchester, MA 01890-2851 USA
[email protected]
Associate editors
Toshio Kataoka, Japan
1-7-2-818 Hiranomiya-Machi
Hirano-ku, Osaka-shi, Japan 547-0046
[email protected]
Theodor Schmidt, Europe
Ortbhlweg 44
CH-3612 Steffisburg, Switzerland
[email protected]
Philip Thiel, Watercraft
4720 - 7th Avenue, NE
Seattle, WA 98105 USA
Production
JS Design & JW Stephens
IHPVA
Paul MacCready, Honorary president
Chris Broome, USA, Chair
Ben Wichers Schreuer, The Netherlands,
Vice-chair,
Open, Secretary/treasurer
Publisher
IHPVA
PO Box 1307
San Luis Obispo, CA 93406-1307 USA
Phone: +805-545-9003; [email protected]
Human Power (ISSN 0898-6908) is pub-
lished irregularly for the International
Human Powered Vehicle Association,
an organization dedicated to promoting
improvement, innovation and creativity in
the use of human power generally, and
especially in the design and development
of human-powered vehicles.
Material in Human Power is copyrighted
by the IHPVA. Unless copyrighted also by
the author(s), complete articles or repre-
sentative excerpts may be published else-
where if full credit is given prominently
to the author(s) and the IHPVA. Individual
subscriptions are available to non-IHPVA
or HPVA members, as are individual
issues.
No. 51 Fall 2001 $5.50/IHPVA members, $4.50
IN THIS ISSUE
Optimum crank-arm length for
recumbents
Danny Too and Chris Williams tested
nineteen subjects using the recumbent-
seating position found in earlier studies
to permit maximum power output to
be developed. Each person pedaled at
maximum effort using, in turn, five differ-
ent crank lengths. (One subject produced
over 1.1 kW). The recommendations for
the best lengths of cranks for different
races are bound to be followed closely.
Bicycle pitchovers
Fred Matteson is concerned for the
safety of bicyclists, particularly when
braking on steep descents. His analysis
has produced a graph on which each rider
can enter her/his body and bike character-
istics, and thereby learn on which hills
her/his level of braking can be critical.
CdA and Crr measurements
John Snyder has developed two
methods of measuring ones coefficients
of aerodynamic drag and of rolling drag.
The first uses two hills of different slope
but similar surface, and the terminal
coasting velocities and other easily
measured data give the coefficients. The
second method involves one hill, and
two coasting runs down the hill, in one
case with a drogue chute. John gives all
instructions.
Technical notes
Chain-drive efficiency. Claire Walton
and John Walton have analyzed the Spicer
data from the last issue of HP, and have
shown graphically how increased chain
tension increases transmission efficiency.
Improvements in chain-loss
measurements? In another technical note
on the Spicer data, John Allen suggests
a feedback system for the driver
torquemeter from the driven torque, so
avoiding the inexactness of measuring the
difference between two similar quantities.
Bicycle stability after front-tire
deflation. Your editor reports on studies
by Andy Oury and others on tires
that produce instabilities when they
go flat. The prime recommendation is
that standards of tire-to-rim fits be
promulgated.
Book reviews
The Athenian Trireme, a book on a
human-powered warship of several-hun-
dred years BC, is reviewed enthusiastical-
ly by Theo Schmidt.
The Dancing Chain, by Frank Berto,
Ron Shepherd and Raymond Henry, is
a wonderful compendium of derailleur
gears from the earliest to the latest times,
favorably reviewed by your editor.
Bicycle Design by Mike Burrows is
another placed in the must read
category by your editor.
Editorials
Joachim Fuchs contributes a guest
editorial on velomobiles.
Your editor writes a sad farewell
and thank you to Open Road, publisher
of Encycleopedia and Bike Culture
Quarterly, among other notable
productions. I also write somewhat
angrily, again, about tires.
Dave Wilson
CONTRIBUTIONS TO HUMAN POWER
The editor and associate editors (you may choose with whom to correspond) welcome
contributions to Human Power. They should be of long-term technical interest (notices
and reports of meetings, results of races and record attempts and articles in the style of
Building my HPV should be sent to HPV News). Contributions should be understandable
by any English-speaker in any part of the world: units should be in S.I. (with local units
optional), and the use of local expressions such as two-by-fours should be either avoided
or explained. Ask the editor for the contributors guide (available in paper, e-mail and
pdf formats). Many contributions are sent out for review by specialists. Alas! We cannot
pay for contributions. They are, however, extremely valuable for the growth of the human-
power movement. Contributions include papers, articles, reviews and letters. We welcome
all types of contributions, from IHPVA members and nonmembers.
ERRATUM
In HP 50, p. 11, figure 3, the labels for the two lower lines were inadvertently
reversed. The lowest line is that for the Ritchey OCR rim, and the middle line is for
the Bontrager. Apologies to Vernon Forbes and to these manufacturers.
ABSTRACT
The purpose of this study was to
determine the crank-arm length that
would maximize peak, mean and min-
imum power outputs in a recumbent
cycling position. Nineteen male volun-
teers were each tested with five pedal-
crank-arm lengths (110, 145, 180, 230
and 265 mm) according to a ran-
domized sequence on a free-weight
Monark cycle ergometer. The 30-second
Wingate Anaerobic Cycling test was
performed in a recumbent position (75
seat-tube angle, backrest perpendicular
to the ground) against a resistance of
85 g/kg of the subjects body mass
(5.0 J/crank rev/kg BM). Curve estima-
tion with regression analysis revealed
that the crank-arm lengths to maximize
peak power, mean power and minimum
power are 124 mm, 175 mm and
215 mm, respectively.
INTRODUCTION
It is well documented that recum-
bent human-powered vehicles with
aerodynamic fairings, having a smaller
drag coefficient and cross-sectional
area, are faster than the standard
racing bicycle (Kyle, 1982). However,
with the current speed record of
117.06 km/hr 72.74 mph), established
in 2000 by a single rider (Sam Witting
-ham) on a Varna recumbent bicycle
Mephisto, designed and built by
Georgi Georgiev, it becomes question-
able whether a more aerodynamically
effective human-powered vehicle can
be designed. If future speed records
are to be attained, it is necessary to
focus not only on the aerodynamics,
but also to examine the variables that
affect power production in recumbent
cycling and the interactions that would
maximize it. Investigations in this
area of recumbent cycling and power
production have included an exami-
nation of changes in seat-tube angle
(Too, 1991) and trunk/backrest angle
(Too, 1994).
Too (1991), examining a systematic
change in seat-tube angle (0, 25, 50,
75 and 100), reported the largest
peak power and mean power to be
found with the 75 seat-tube angle
and a parabolic curve (quadratic trend)
best describing the change in peak
power and mean power with changing
seat-tube angles. Seat-tube angle was
defined by the angle formed between
the seat tube and a vertical line
(perpendicular to the ground) passing
through the crank spindle. Using a
75 seat-tube angle, Too (1994) inves-
tigated the effect of three trunk/seat-
backrest angles (60, 90 and 120) on
power production. A parabolic trend
in peak power and mean power was
found with changes in trunk/seat-back-
rest angle, with the largest peak power
and mean power reported using the 90
trunk angle.
Based on muscle force-length and
force-velocity power relationships,
changes in crank-arm length will
affect joint angles, muscle length,
force, torque and power production in
cycling. Since the literature involving
traditional upright cycling positions
have reported an effect on power
output with changes in crank-arm
length (Hull & Gonzalez, 1988; Inbar,
Dotan, Trousil & Dvir, 1983; Too &
Landwer, 2000), it can be assumed that
power production will also be affected
in a recumbent cycling position with
different crank-arm lengths. Therefore
the purpose of this study was to deter-
mine the trend in power production
with changes in crank-arm length,
and the crank-arm length that would
maximize peak power, mean power
and minimum power in a recumbent
cycling position.
METHOD
Nineteen healthy volunteer male par-
ticipants (mean age = 24.8 4.4 yr.,
weight = 81.76 11.84 kg, height = 1.80
0.08 m) subjects were tested with
a free-weight Monark cycle ergometer
(Model 814E) at five pedal-crank-arm
lengths (110, 145, 180, 230 and 265 mm),
as defined by the distance between the
center of the crank spindle and pedal
spindle. (The normal crank-arm length
for a Monark cycle ergometer is 170
mm). To accomplish this, two adjust-
able crank arms allowing for manipula-
tions from 0 to 300 mm were used (Too
& Landwer, 2000). All subjects were
Human Power Number 51, Fall 2001 3
Determination of the crank-arm length to maximize
power production in recumbent-cycle ergometry
Danny Too and Chris Williams
Figure 1. Recumbent position with a 75 degree seat-tube angle
2 Number 51, Fall 2001 Human Power
and minimal pedaling rates occur with
the 110-mm crank-arm length. From
regression equations, the predicted
crank-arm lengths to maximize peak
power, mean power and minimum
power are 124 mm, 175 mm and
215 mm, respectively.
DISCUSSION
Since no literature could be found
examining the effect of changes in
crank-arm length on cycling per-
formance in a recumbent position,
comparisons will be made with the lit-
erature available for an upright posi-
tion. The parabolic curve observed in
peak power and mean power with
increasing crank-arm length is consis-
tent with the trend for an upright posi-
tion reported by: (1) Inbar et al. (1983)
for five crank-arm lengths (125, 150,
175, 200 and 225 mm); and (2) Too
and Landwer (2000) for five crank-arm
lengths (110, 145, 180, 230 and 265 mm).
From best-fitting parabolic curves,
Inbar et al. (1983) described the peak
power and mean power to occur at
a crank-arm length of 166 mm and
164 mm, respectively; whereas Too and
Landwer (2000) predicted peak power
and mean power to be maximized with
crank-arm lengths of 164 and 200 mm,
respectively. This is quite in contrast
with the predicted crank-arm lengths
(124 and 175 mm) to maximize peak
power and mean power, respectively,
for a recumbent position.
The largest peak power (762.7 W)
and mean power (615.9 W) values
reported by Inbar et al. (1983), and
those reported by Too and Landwer
(2000; largest peak power and mean
power values to be 968 W and 718 W,
respectively) are less than the largest
peak power (1144 W) and mean power
(845 W) values recorded for the recum-
bent position in this investigation. In
fact, except for the 265 mm crank-arm
length condition, peak power values
(and all mean power values) in the
recumbent position were greater than
the largest peak and mean power
values reported by Inbar et al (1983)
and by Too and Landwer (2000)
for an upright position. The smaller
peak power and mean power values
reported by Inbar et al. (1983) may
be attributed to a smaller load used
(75 g/kg body mass) and/or to the
different stature of the subjects tested
(approximately 10.5 kg smaller, 73 mm
shorter than the subjects of this inves-
tigation). However, the smaller peak
and mean power values reported by
Too and Landwer (2000) are probably
attributed to differences in lower-limb
joint angles (between an upright and
recumbent position) and/or to a
smaller force production potential in
an upright position (since there is no
seat-backrest to push against).
Based on the predicted crank-arm
lengths to maximize the different
power variables, and the trend of
peak power, mean power and mini-
mum power with changes in crank-arm
length, it would appear that an interac-
tion exists between crank-arm length
and power production, with the opti-
mal crank-arm length to maximize
power dependent on load and pedaling
rate. Since power is a function of both
force and velocity, the optimal crank-
arm length to maximize peak power
would be one where the maximum
pedaling rate is produced and main-
tained with the largest load that can
be applied. Although manipulation of
load was not examined in this inves-
tigation, changes in crank-arm length
would alter the torque on the crank
arm (when the same force is applied)
and would be analogous to a change
in load. Based on the force-velocity
relationship, a longer crank-arm length
resulting in a lower load experienced
by the lower limbs will result in a
greater linear velocity at the pedal
(when compared to the same pedaling
rate with a shorter crank-arm length).
This was confirmed when the maximal
pedaling rates determined for the dif-
ferent crank-arm lengths of this investi-
gation were converted to maximal lin-
ear pedal velocity. The maximal linear
pedal velocity was found to increase
(although the maximal pedaling rate
decreased) with increasing crank-arm
lengths from 110 to 265 mm. Similarly,
an increase in crank-arm length from
110 to 265 mm also resulted in
an increase in minimum linear pedal
velocity (as determined from the mini-
mum pedaling rates) and is also con-
sistent with that expected from force-
velocity relationships. Since parabolic
curves in power were observed with
increasing crank-arm lengths, and the
largest values for peak, mean and mini-
mum power were found with three dif-
ferent crank-arm lengths, this would
indicate that the optimal crank-arm
length to maximize power is dependent
on the type of power examined.
In this investigation, the optimal
crank-arm lengths predicted to maxi-
mize peak, mean and minimum power
with a load of 85 g/kg BM, were
124 mm, 175 mm and 215 mm,
respectively. The interaction between
crank-arm length, pedaling rates and
load (as evidenced by parabolic curves
for power), would suggest that the
optimal crank-arm length for peak,
mean and minimum power would
change with different loads. Based on
the force-velocity-power relationship,
increased loads to maximize power,
resulting in a decreased pedal rate
would favor longer crank-arm lengths.
Changes in crank-arm length will
affect not only the force-velocity-power
relationship, but also the muscle force-
length relationship. From the force-
length curve, a muscle can produce
its largest force at resting length, with
a decrement in force at increasing or
decreasing lengths. Systematic incre-
ments in crank-arm length (from 110 to
265 mm) for an upright cycling position
have been reported to result in signifi-
cant decrements in minimum hip and
knee angle, and significant increments
in hip and knee range of motion (Too
and Landwer, 2000). Whether it is more
advantageous to use a long crank arm
or a short crank arm is unknown
because there is a complex interaction
among changes in joint angles, muscle
length and muscle-moment-arm length
to produce force and torque with
changes in crank-arm length. This com-
plexity is further increased when multi-
joint muscles that cross the hip and
knee, or knee and ankle are involved
and interact with force-velocity-power
relationships. Additional research into
the interaction of crank-arm length,
pedaling rate and load on power pro-
duction is needed before the limits of
performance in human-powered vehi-
cles can be reached.
Human Power Number 51, Fall 2001 5
tested in each of the five pedal-crank-
arm-length conditions, with the order of
testing randomly assigned. There was
a minimum of 24 hours of recovery
between test sessions. For each condi-
tion, pedal toe clips were worn, and
the subject was strapped to the seating
apparatus at the hip and trunk.
The recumbent cycling position used
for all test sessions, was defined by a
75angle formed between the bicycle
seat tube and a vertical line passing
through the crank spindle (see figure 1;
Too, 1991). To obtain this seating
position, a variable seating apparatus,
allowing for manipulations in seat-tube
angle, backrest angle and seat-to-pedal
distance was used and interfaced to a
Monark cycle ergometer (Model 814E).
The seat backrest was kept perpendicu-
lar to the ground and the seat-to-pedal
distance adjusted to 100% of the total
leg length of each subject, as measured
from the right femur to the ground
(Too, 1991). The test protocol involved
a computerized 30-second Wingate
Anaerobic Cycling Test. To initiate the
test, the subject pedaled the cycle
ergometer with no load. Once the
ergometers inertial resistance had been
overcome, the appropriate load
(85 g/kg of the subjects body mass)
was instantaneously applied using cali-
bration weights, and the subject ped-
aled as hard and as fast as possible for
30 seconds. A Sports Medicine Industry
(SMI) opto-sensor (Model 2000) with a
sampling rate of 50 Hz, interfaced with
a Zenith 386 micro-computer, in con-
junction with 16 reflective markers on
the ergometer flywheel, was used to
monitor and record flywheel revolu-
tions during the test. Peak power was
calculated from the highest average fly-
wheel speed during any consecutive
five seconds, mean power was deter-
mined from the mean flywheel speed
for the entire 30-second test, and mini-
mum power was calculated from the
lowest mean flywheel speed during any
consecutive five seconds (which was
always the last five seconds). The dif-
ferent power variables were calculated
using the following equation:
Peak power (watts) = [load (N)]
[distance covered by flywheel with
one revolution (1.615 meters per
revolution) average number of
recorded flywheel revolutions for
five seconds (rpm)] / [1 min/60 sec].
Additionally, maximum and mini-
mum pedaling rates were calculated
from flywheel speed recorded for peak
power and minimum power, respective-
ly. The equation used in this calculation
was:
Pedaling rate (rpm) = average
flywheel rpm for five seconds / 3.7
flywheel revolution per pedal-crank
revolution (Gledhill and Jamnik, 1995).
This would be equivalent to a
52/14 gear ratio. Curve estimation with
regression analysis was used to deter-
mine: (1) the trend in peak power,
mean power and minimum power
with changes in crank-arm length;
and (2) the crank-arm length that
would maximize peak power, mean
power and minimum power during a
30-second test.
RESULTS
With changes in crank-arm lengths,
the mean SD values of peak power,
mean power, minimum power, maxi-
mum and minimum pedaling rates are
presented in table 1.
Based on regression analysis the
change in peak power, mean power
and minimum power with increasing
crank-arm length, appears to be best
described by a parabolic curve, repre-
sented by the equation: y = -x2 + x +
C (where y represents power and x rep-
resents crank-arm length) as shown in
figure 2. The specific regression equa-
tions for the various measures of power
were as follows:
Peak power (quadratic trend,
p = 0.006): y = 0.011x
2
+ 2.8x + 972
(SE = 11)
Mean power (quadratic trend,
p = 0.011): y = 0.011x
2
+ 3.8x + 513
(SE = 5)
Minimum power (quadratic trend,
p = 0.002): y = 0.007x
2
+ 2.8x 325
(SE = 2).
From table 1, several observations can
be made: (1) regardless of crank-arm
length, peak power is greater than
mean power, and mean power is great-
er than minimum power; (2) peak
power is greatest with the 145-mm
crank-arm length and least with the
265-mm crank-arm length; (3) mean
power is greatest with the 145- and
180-mm crank-arm lengths and least
with the 265-mm crank-arm length;
(4) minimum power is greatest with
the 230-mm and least with the 110-mm
crank-arm length; and (5) maximal
4 Number 51, Fall 2001 Human Power
Crank-arm length (mm)
110 145 180 230 265
PP (W) M 1139 1144 1097 1025 916
SD 206 214 223 193 167
MP (W) M 802 845 845 819 762
SD 177 192 168 166 134
MINP (W) M 555 598 619 627 612
SD 141 156 145 148 117
MAXPED (rpm) M 174.1 171.7 167.5 153.1 135.2
SD 11.7 10.7 12.4 11.1 11.3
MINPED (rpm) M 82.0 88.4 91.9 92.7 91.4
SD 12.7 12.2 12.0 12.3 9.1
PP = peak power; MP = mean power; MINP = minimum power
MAXPED = maximum pedaling rate;
MINPED = minimum pedaling rate
Table 1. Peak power, mean power, minimum power, maximum and
minimum pedaling rate with changes in crank arm length
600
800
1000
1200
110 145 180 230 265
Crank-arm length (mm)
Peak
Mean
Minimum
0
Figure 2. Predicted peak power, mean power and minimum power with
increasing crank-arm length.
P
o
w
e
r

(
W
)
in the USA may not be present. Signs,
striping, guard rails and signals may
be absent. Dangers may appear with-
out warning and such hazards as very
steep local inclines, improper banking,
holes or damage may suddenly appear.
Very narrow roads are common in
mountainous areas, sometimes wide
enough for only one vehicle. These
dangers may well result in the cyclist
having to use sudden and strong brak-
ing. Braking on steep descents can eas-
ily lead to a pitchover accident and this
will be discussed as the main subject
of this article.
ANALYSIS OF PITCHOVER
Because the rider on a conventional
bicycle sits high and the wheelbase
is fairly short, braking tends to cause
the bicycle and rider to pitch over.
On level ground rather severe braking
is required to result in a pitchover.
However, as the bicycle inclines
forward, such as in going down a
hill, the weight vector inclines forward
such that more weight is carried by
the front wheel than on level ground.
Application of brakes produces a
pitching moment as on level ground.
Much less braking than on level ground
can result in a pitchover. Analysis
can tell us how much. The subject
of pitchover has been covered by
DeLong (1978, 208209) [and Sharp
(1977 [1896], 216220) Ed.]. Herein an
equation will be derived showing the
braking required as a fraction of the
total weight of bicycle and rider for a
range of slopes of the road. The sketch
(figure 1) shows a bicycle and rider on
a road with a slope of . The height of
the combined bicycle-and-rider center-
of-gravity on level ground is h. The
SUMMARY
Pitchover is explained and a graph
developed showing boundaries versus
slopes. Situations and road characteris-
tics are discussed.
INTRODUCTION
Pitchovers, wherein the bicycle and
rider rotate forward about the front
wheel, have been a problem since the
early days of cycling when the high-
wheel, direct-drive bicycle, commonly
referred to as the Ordinary and later,
derisively, as the Penny Farthing, was
used. The position of the rider, high and
forward with respect to the front axle,
made these cycles likely to pitch for-
ward particularly in descents with brak-
ing. It was this danger that led to the
development of the chain-driven safety
bicycle still in use today. Todays bicy-
cles, with the rider well back between
the wheels, are far safer, but pitchovers
can and do still occur.
This article concerns the matter of
hills. That steep hills are an expecta-
tion of touring cyclists is evident by
the installation of triple chainwheels
and wide-range gearing on touring
machines. Before the time of the auto-
mobile steep roads were common in
this country. Horses could climb steep
hills, but cars had limited climbing
ability. The author recalls seeing Ford
Model Ts stop at the bottom of a hill,
turn around and back up because they
could climb a steeper hill in reverse
gear. Such a practice was not reason-
able and the trend has been towards
less-steep public roads in the United
States. The process of building safe,
high-speed roads has consisted of
straightening and leveling, often at
great expense and difficulty. Abroad,
and in particular in lesser-developed
lands or where there are fewer
automobiles, even main roads may
be unsealed, crooked and containing
steep slopes. Safety features common
SUMMARY
The predicted crank-arm lengths to
maximize peak power, mean power
and minimum power in a recumbent
cycling position, using a resistance
load of 85 g/kg body mass, were
124 mm, 175 mm and 215 mm, respec-
tively. This would suggest that for
human-powered vehicle competitions
of short duration, where maximal peak
power is necessary, a shorter crank-
arm length is recommended. For com-
petitions of longer duration where
fatigue is a factor and the largest mean
power and minimum power become
important, it is suggested that longer
crank-arm lengths be used.
REFERENCES
Gledhill, N. and Jamnik, R. 1995.
Determining power outputs for cycle
ergometers with different sized fly-
wheels. Medicine and Science in
Sports and Exercise 27:134135.
Hull, M. L. and Gonzalez, H. 1988.
Bivariate optimization of pedalling
rate and crank arm length in cycling.
Jl. of Biomechanics 21:839849.
Inbar, O., Dotan, R., Trousil, T. and Dvir,
Z. 1983. The effect of bicycle crank-
length variation upon power perfor-
mance. Ergonomics 26:11391146.
Kyle, C.R. 1982. Bicycling 23:5966.
Too, D. 1991. The effect of hip
position/configuration on anaerobic
power and capacity in cycling.
International Jl. of Sports
Biomechanics 7:359370.
Too, D. 1994. The effect of body ori-
entation on power production in
cycling. The Research Quarterly for
Exercise and Sport 65:308315.
Too, D. 2000. The effect of pedal crank
arm length on joint angle and power
production in upright cycle ergome-
try. Jl. of Sports Sciences 18:153161.
ABOUT THE AUTHOR
Danny Too is an associate professor
in the Department of Physical Educa-
tion and Sport at the State University
of New York at Brockport and has
been involved in human-powered-vehi-
cle research since 1985.
Danny Too
State University of NY at Brockport
350 New Campus Drive
Brockport, NY 14420 USA
E-mail: <[email protected]>
Bicycle pitchover characteristics
by Frederick H. Matteson
distance of the center-of-gravity from
the front-wheel contact point, P, on
level ground is l. The total weight of
bicycle and rider is W. The braking
force is B. The inclination of the
bicycle causes the line of action of the
weight vector to go forward an amount
l on the ground. The pitchover will
be initiated when the inertial moment
from the braking about P is equal to
the moment of the weight. Noting that,
l = h tan Eq. 1
MP = 0 = Bh W(l l) cos Eq. 2
Substituting for l and simplifying,
B/W = l/h cos sin Eq. 3
This expression yields the brake force
as a fraction of the total weight where
pitchover will take place for various
ratios of l/h and angles, . It may
be seen that on level ground ( =
0, cos = 1.0, and sin = 0) that
B/W = l/h. Equation 3 is plotted for a
range of values of l/h versus the slope,
, on figure 2.
If one is to descend a hill at a
constant speed and not accelerate,
the resistance to motion must equal
the accelerating component of gravity.
Assuming that that resistance is the
force supplied by the brakes, B,
B = W tan Eq. 4
The braking force for this steady
state increases as the slope increases.
On a steep hill this required braking
force could exceed that which would
result in a pitchover. Under these
conditions the rider is in serious
trouble. The boundary is shown on
figure 2 as a dashed line. To show
where actual bicycles and riders would
lie on the figure, the authors two
touring bicycles, with him on them,
were chosen. One is a lightweight
touring bicycle of conventional design
and the other is a Bike Friday
New World Tourist model. These two
bicycles differ greatly in appearance,
but fit and perform similarly. Each was
fitted with fenders (mudguards), rear
pannier rack, frame pump and empty
water bottle. The values of l and h
were determined. Because both
bicycles are custom built, the values
are not considered applicable to other
peoples bicycles. The curves of B/W
versus are shown. The differences
between the two curves is surprisingly
small. The center-of-gravity of the Bike
Friday was slightly lower than the
conventional bicycle, making it a bit
more stable.
DISCUSSION
Figure 2 shows, as an example, that
if one is on a bicycle and the value Figure 1. Sketch of forces and moment arms on bicycle and rider
Figure 2. Pitchover points versus descent angle
DEFINITIONS
F slope
h height of the combined bicycle-
and-rider center-of-gravity on
level ground
P front-wheel contact point
W total weight of bicycle and
rider
B braking force
l horizontal distance from bicy-
cle-and-rider center of gravity
on level ground to contact
point, P
l = h tan
6 Number 51, Fall 2001 Human Power Human Power Number 51, Fall 2001 7
of l/h is 0.5 (curves descending to the
right) and the slope is ten degrees
(vertical lines) that their intersection
occurs at a value of B/W (horizontal
lines) of approximately 0.32. That
means that a braking force of about
a third of the weight of the bicycle
plus rider would be enough to cause a
pitchover. This level of braking can be
easily attained. But because the point
lies well above the curve for braking
for steady state, less braking applied
than that for pitchover would allow
deceleration. If the slope were to
increase to about 14 degrees, the
pitchover-braking and steady-state-
braking curves coincide and there is
no margin for slowing. If one follows
the l/h curves down to the x-axis, those
angles where tan = h/l, result in a
pitchover with no braking. The impor-
tance of figure 2 to the cyclist is not in
such calculations, but rather in permit-
ting an understanding of the nature of
the problem. In practice one probably
does not know the steepness of the hill
precisely. The values of l/h vary with
the position of the rider; the rider can
increase l/h by lowering his body or
sliding back on the saddle. Further,
the rider can only roughly judge the
amount of braking force. Lastly, the
likely situation for a pitchover will
often not be in a steady descent, but in
some sort of emergency condition.
It is obvious from the analysis that
the slope of the roadway is a most
critical variable. In the USA criteria
and standards exist governing slopes
of roads. The American Association of
State Highway and Transportation Offi-
cials (Merritt, 1983, Table 16-6) has set
forth limiting standards for slopes. On
Interstate highways and primary roads
the limits are five to seven percent
(34 degrees). On secondary roads
and in mountainous terrain limits of
10 percent are suggested (6 degrees).
These standards are suggested stan-
dards only and each state, in fact,
establishes its own standards. In
reality roads may vary widely and
often exceed the standards in the USA.
In other countries the slopes may be
greater.
In the vicinity of Hollister, California,
the author measured slopes. On local
roads and highways the slopes mea-
sured did not exceed eight degrees (14
percent). However in residential sub-
divisions, where children live, roads
used by cyclists often approached or
reached ten degrees (17.6 percent). The
county [San Benito] allows up to 15-per-
cent slopes for long stretches or 16 per-
cent for up to 122 meters (400 feet).
In mountainous areas of the world
the cyclist is likely to encounter sharp
turns often referred to as hairpin turns
or switchbacks. These turns pose spe-
cial dangers. Good practice dictates
that the turn not be too sharp, i.e., that
an ample radius of the inner road edge
should exist. In the case of a descend-
ing turn, to keep the cross section
of the roadway level requires that the
inner portion of the road be cut away
along radial lines resulting in a helical
shape. If the road extends to the center
of the curve then that portion of
the road at the center descends ver-
tically. Such turns exist. These turns
often restrict visibility severely. As one
rounds a turn one may find oneself fac-
ing a bus or truck inching up the slope,
blocking the road and requiring sudden
braking and possibly making the cyclist
hug the inner side of the road. An acci-
dent may not be avoidable. It is the
local slope of the surface that is
critical and the braking used, even
momentarily, which will result in the
pitchover. Speed is important too
because, although it does not in itself
determine pitchover, the moving bicy-
cle pitching over can act as a catapult
on the rider and make the accident
more serious.
What can be done to avoid pitching
over? If the rider has time to react he
might lower himself or move aft on the
bicycle. It would seem helpful to lower
or move aft the baggage. Panniers
can be carried low. They tend to get
dirty from splashing and sometimes hit
curbs, but otherwise seem to be satis-
factory. Moving the load aft is difficult
and may adversely affect handling.
However, if, as often happens in
touring, the baggage is removed, say,
to climb a mountain for sightseeing,
the load is not available for the
dangerous descent.
Although they may have other short-
comings in hills, tandems do have supe-
rior longitudinal stability. Recumbent
bicycles come in a large variety
of forms, but the lower center of grav-
ity should generally alleviate the pitcho-
ver problem. The long-wheelbase style
should be very stable. Sometimes plan-
ning can be beneficial. If one is making
a round trip on a road with a steep
hill in one direction, the direction of
travel should be safer climbing rather
than descending that hill. It is one
purpose of this article that understand-
ing the danger and the mechanism
of pitchover should make for a safer
cyclist. Another hope is that road
designers and builders will understand
that some designs and practices may
not be particularly hazardous to motor
vehicles and yet be very hazardous to
bicycles. In many cases hazards can be
rectified, reduced or avoided.
REFERENCES
DeLong, F. 1978. DeLongs guide to
bicycles and bicycling. Radnor, PA:
Chilton.
Merritt, F. S. 1983. Standard handbook
for civil engineers, 3rd ed. NY:
McGraw-Hill.
Sharp, A. 1977. Bicycles and tricycles.
Cambridge, MA: The MIT Press.
(Originally published in London by
Longmans, Green, 1896.)
THE AUTHOR
Frederick H. Matteson, a retired
engineer, has been riding bicycles for
70 years and touring in the USA and
abroad over a span of almost 60 years.
Frederick H. Matteson
1874 Cushman Street
Hollister, CA 95023-5525 USA
Phone: 831-637-6598
8 Number 51, Fall 2001 Human Power
ABSTRACT
This paper provides the conceptual basis, and examples, of
ways to estimate an HPVs rolling and aerodynamic resistance
by utilizing the slope-intercept form of a linear equation as
applied to coastdowns on down-grades.
INTRODUCTION
This testing protocol was developed to evaluate objectively
two of the qualities affecting the performance of a bicycle.
The procedures detailed do not seek to model fully all of
the physical factors involved. Instead, they serve as a means
to make quantitative observations during actual riding condi-
tions. Of significance, the methods rely upon measurable
steady states.
Requirements for constant-slope hills and for wind-free
conditions impose a limit to when and where the protocol
may be employed. The reader is encouraged to study works
by Chester Kyle and others listed in the suggested reading
list, to explore indoor and similar methods used for evaluat-
ing aerodynamic drag and rolling resistance.
PREMISE
There are six losses that a cyclist must overcome while
pedaling: gravity (if she or he is climbing), air drag (if
adverse), rolling resistance, linear and rotational inertia
changes (if accelerating), and drive-train inefficiencies [1]. At
a constant velocity, inertia changes are null. While coasting,
drive-train losses are non-existent [2]. Terminal velocity of a
downhill-coasting bicycle refers to the phenomenon whereby
the down-road component of the force of gravity achieves
equilibrium with the two remaining primary loads: aerody-
namic drag and rolling resistance.
mass gravity [sine(angle) Crr]
Cd =
air density area terminal velocity
2
0.5
The above formula, describing the coefficient of drag (Cd)
of a land vehicle when coasting at terminal speed, squares
velocity and contains a sine function. Therefore it does not
appear at first inspection to follow the pattern of a linear
equation. The equation may be treated as though it were
linear when only Cd and the coefficient of rolling resistance
(Crr) serve as its variables and its other terms are regarded
as constants.
Employing the following redistribution and substitutions
facilitates setting the equation into linear slope-intercept
form:
mass gravity
CdA =[sine(angle)Crr],
air density terminal velocity
2
0.5
and putting
Y CdA,
X Crr,
mass gravity
m , and
air density terminal velocity
2
0.5
b m sine(angle), then the equation can be given as
Y = (1)mX + b.
By establishing values for mass, gravity, air density, termi-
nal velocity, and hill angle occurring in two dissimilar settings
(prime and double prime), i.e., each producing unique values
for linear slope (m) and the y-axis intercept (b), it becomes
plausible to solve for Crr (X) using:
b b
Crr = .
m + m
Once Crr is known, the corresponding CdA (Y) value may be
determined with either:
CdA = (1)mX + b, or
CdA = (1)mX + b .
The opportunity now exists to identify experimental situ-
ations whereby two separate coasting events might be
conducted, measured, and compared.
TWO-HILL COMPARISON
Four constants appear within the formula (gravity, mass,
air density, and angle) that, when modified, result in a
corresponding change in coasting velocity. For the moment,
the most consequential of those four to alter by a known
amount is angle. It is hypothesized that a comparison of
terminal velocities as achieved on two unlike hills, with
similar road textures, permits quantifying a bicycles unique
CdA and Crr solution set pair.
In practice, comparing the terminal velocities achieved on
two different hills, as well as the single-hill test that appears
later, should be recognized as a method to approximate the
values describing performance.
Input values and dimensions [3]
Reflect for a moment on the constants. Each needs to
be known with some assurance as it applies to a specific
coasting event. Knowing if calm wind conditions exist is also
essential, as the formulation tacitly implies air and ground
velocity equal one another. Fortunately, convivial means exist
to quantify each.
Mass
The systems total mass can be found quickly by standing
on a scale while holding the bicycle and all equipment that
will be carried while riding. Ideally this measurement should
be taken just prior to coasting.
Gravity
Acceleration due to gravity varies only slightly at different
points on the earths surface. It can generally be considered
to be a value of 9.81 m/s
2
.
Wind
A length of about 300 mm of single-ply tissue paper when
held hanging straight down reveals whether or not calm wind
conditions prevail. This deceptively uncomplicated tool dem-
onstrates sensitivity to subtle movements of air. For the pres-
ent need, a wind-speed-measuring device need only indicate
zero wind velocity (see table 1).
Air density
The density of air, often referred to by the Greek letter ,
varies considerably with altitude, temperature, and humidity.
Human Power Number 51, Fall 2001 9
CdA and Crr measurement
by John C. Snyder, Jr.
straw
string
weight
protractor
Conditions, hill #2 (double prime)
combined rider & vehicle mass: 100kg
grade: 8%
temperature: 23C
air pressure: 1016 hPa
terminal velocity: 16 m/s
surface: smooth asphalt
wind: still air
angle = arc tangent(grade/100)
= arc tangent(0.08)
= 4.57 degrees
air density = P/RT
1016 hPa
=
2.87 (23C + 273.15)
= 1.20 kg/m
3
mass gravity
m=
air density terminal velocity
2
0.5
100kg 9.81 m/s
2
=
1.20 kg/m
3
16
2
m/s 0.5
= 6.41 m
2
b = m sine(angle)
= 6.41 m
2
sine 4.57 degrees
= 10.64 m
2
0.08
= 0.51 m
2

Determination of CdA and Crr
Determination of the coefficients
of rolling resistance and aerodynamic
drag times area now occurs in the fol-
lowing manner:
b b
Crr =
m + m
0.53m
2
0.51m
2
=
6.41m
2
+ 10.64m
2
= 0.0047, and
CdA = m(Crr) + b
= 10.63m
2
0.0047 + 0.53m
2
= 0.48m
2
SINGLE-HILL COMPARISONS
The previous treatment creates a
foundation for additional testing
methods. Physically modifying a
bicycles CdA by a known amount
during one of two coastdowns will
cause the terminal velocity to change
in a predictable fashion (Kyle, 1984,
2240 [5]). This observation suggests
comparative coastings to determine
CdA and Crr might be conducted on a
single hill in conjunction with a drogue
device (see page 13).
Auxiliary drag
If
Y vehicle CdA,
Y drogue a known modifying value of
CdA, and if
(Y vehicle + Y drogue) = CdA total, then
(Y vehicle + Y drogue) = (1)mX + b,
and
Y vehicle = (1)mX + b Y drogue.
These relationships permit the com-
parison of a coasting vehicles unmod-
ified (prime) and modified (double
prime) configurations to give Crr:
(1)mX + b =(1)mX + b Ydrogue
b b Y drogue
X = = Crr .
m + m
Though there exist various ways to
alter a bicycles aerodynamic proper-
ties, the deployment of a small para-
chute [6], or the attachment of a rigid
plate off to the side of the vehicle [7],
represent ideas that have been success-
fully adopted in the past for increasing
total aerodynamic drag in a controlled
manner. It is assumed by this mathe-
matical model that any auxiliary source
of drag will be configured such that it
will not significantly interact with the
normal performance characteristic of
the tested vehicle. A drogue device also
implies a more elegant handling.
When
M mass
g gravitational constant
G grade [8]
air density
v velocity,
the following equation depicts the equi-
librium between aerodynamic drag and
rolling resistance, and grades effect
when at the steady state,
(0.5)()(v
2
)(CdA)+Mg(Crr)=(1)Mg(G)
This equation may be rearranged into:
(0.5)()(v
2
)(CdA) = (1)Mg(G + Crr).
If a vehicles CdA changes by a known
amount (CdA) then the following also
applies:
(0.5)()(v
2
)(CdA +CdA)=(1)Mg(G+Crr)
If two coastdowns to terminal veloc-
ity are conducted, only one of which
has been modified, and both occur dur-
ing identical weather conditions on the
same hill, permitting cancellation of
values that have not changed, then:
(v
2
)(CdA) = (v
2
)(CdA + CdA),
which, after having been solved for
CdA, reveals the following description:
CdA drogue
CdA vehicle =
(v/v)
2
1
By solving this equation, a bicycles
CdA may be assessed even if mass,
grade, and air density are unknown val-
ues, but are consistent from one coast-
ing event to the next.
Example B
The following depicts measuring the
CdA of a bicycle coasting to terminal
velocity twice on a single hill. The
first coasting occurs without changes
to the system. The second coasting is
performed while a drogue device is
deployed.
Conditions
CdA drogue = 0.44 m
2
terminal velocity, unmodified (v)=15 m/s
terminal velocity, modified (v)=9 m/s
Determination of CdA
CdA drogue
CdA =
(v/v)
2
1
0.44m
2
=
(15 m/s / 9 m/s)
2
1
= 0.25 m
2
LIMITATIONS AND STRENGTHS
The accuracy of determining the
hills angles, air density, mass, and
velocity values, ultimately control the
quality of the solution. Though simple
in design the suggested instruments are
pragmatic and easily obtained. It is con-
ceivable that by exercising due care any
error brought about by input data could
be made negligible.
There are several conceptual con-
cerns. First is the models presumption
that a precise single CdA and Crr solu-
tion set pair exists at all. It is unlikely
that Crr can be identical on distinctly
different roadways. Cd is not a constant
throughout a range of velocities [9,10].
The assumption that a drogue could be
deployed such that it does not interact
in any manner with the vehicles normal
performance characteristic is indefensi-
ble. However, by utilizing comparative
velocities, and other conditions which
are close in value to one another the
significance of these inherent errors
will be lessened.
There exist logistical concerns. Any
However, a reasonable value may be
obtained by applying air temperature
and pressure to the ideal-gas law:
= P/(RT)
where:
air density (kg/m
3
)
P pressure (Pa)
R constant (J/kg K)
T temperature (K)
After accounting for modern weather
services custom of using hPa and
Celsius, the equation becomes
hPa
(kg/m
3
)=
2.87(degreesC+273.15)
Hill angle
Obtaining the angle of the roadway
can be forthright as well. One credible
method is to acquire a large protractor
with a hole at its vertex through which
to affix a length of thread or thin string.
The other end of the thread will be tied
to a small weight. A straight hollow cyl-
inder, such as a drinking straw, taped
or glued to the base of the protractor
serves to form a sighting guide (fig. 1).
While looking through the sight one
moves the device until the inverted pro-
tractors base becomes parallel with the
slope of the road. Then by pinching the
thread tight to the protractor, the hills
angle in degrees may be read directly.
Alternatively, one may use a com-
mercially available inclinometer such
as found on some magnetic pocket
compasses, or a tripod-mounted transit
designed specifically for the purpose.
Occasionally, reliable survey data
will be available. If a hills slope is
expressed as a grade percent, convert
to degrees of angle with the following:
degrees = arc tangent(grade/100)
Some computer spreadsheet
programs perform trigonometric func-
tions exclusively with radians, which
can be converted with the following
relationships:
1 degree = /180 radians,
1 radian = 180/ degrees.
Terminal velocity
Transient conditions influencing
coasting hold the potential of exerting
profound, even if subtle, accumulating
effects. Any coasting event should
be recognized as the sum result of
many unidentifiable and a few identifi-
able controlling factors. As examples:
edging past a tiny unseen pebble, trav-
eling through a small dip in the road
bed, or even making slight unavoidable
movements in steering, will induce
a momentary velocity change, thus
modifying the total time and distance
traveled.
When observing an equilibrium con-
dition, such as represented by the
essentially unchanging value of ter-
minal velocity, minor
variations in a road
bed and other
transient phenomena
exhibit little or no
influence.
The rate of velocity
increase becomes
minuscule as a down-
hill-coasting bicycle
approaches terminal
speed. The digital
output of a cycle
computer typically
rounds units of kilometers or miles to
one or two decimal places. For those
reasons, a practical reading of the
steady state will appear conveniently
prior to the actual occurrence of
terminal velocity. When coasting down
a constant slope, one needs only to
monitor a cycle computer to determine
when velocity no longer increases or
decreases.
The maximum-velocity function
available on some digital cycle
computers will display the terminal
velocity value, but only if terminal
velocity has not been exceeded due
either to pedaling or variation of the
roadbed. As with any data-collection
procedure, obtaining as many samples
as practical is desirable.
These unit conversions may prove
helpful:
mph 0.4469 = m/s
km/h 0.2778 = m/s
EXAMPLE A
The following depicts a bicycle that
has coasted to its equilibrium condi-
tion on two separate hills [4].
Conditions, hill #1 (prime)
combined rider & vehicle mass: 100kg
grade: 5%
temperature: 26C
air pressure: 1014 hPa
terminal velocity: 12.5 m/s
surface: smooth asphalt
wind: still air
angle = arc tangent(grade/100)
= arc tangent(0.05)
= 2.86 degrees
air density = P/RT
1014 hPa
=
2.87 (26C + 273.15)
= 1.18 kg/m
3
mass gravity
m =
air density terminal velocity
2
0.5
100kg 9.81 m/s
2
=
1.18 kg/m
3
12.5
2
m/s 0.5
= 10.64 m
2
b = m sine(angle)
= 10.64 m
2
sine 2.86 degrees
= 10.64 m
2
0.05
= 0.53 m
2

Human Power Number 51, Fall 2001 11
J
o
h
n
C
.
S
n
y
d
e
r
Figure 1. Inclinometer
Force Strength km/h mph Observation
0 Calm 01 Smoke rises vertically; tissue hangs vertically.
1 Light air 15 13 Smoke drifts; tissue moves slighty.
2 Slight breeze 611 47 Leaves rustle; tissue becomes horizontal
Table 1. A portion of the Beaufort wind scale
10 Number 51, Fall 2001 Human Power
Construction of a drogue parachute suitable for HPV
experimentation is an easy project. Using scissors, cut the
two side creases and the bottom fold of a large, plastic
lawn-and-leaf bag (2.4 micrometer thickness) to obtain two
equally-sized panels of material. Set one aside; the other
will serve as a rectangular parachute canopy. To each of the
four corners, which have been knotted, tie lengths of thin
nylon cord or twine. The other ends of the cords then can
be attached, via a fishing-line swivel, to a handle. A circular
hole cut in the center of the rectangle helps stabilizes
the parachute when its open and filled with moving air.
The coefficient of aerodynamic drag of any drogue device
is defined by:
drag
CdA drogue
air density air velocity2 0.5
In order to determine the product of coefficient of drag
(Cd) and area (A), it is not necessary to know the individual
values of either Cd or A as they exist separately.
A dimensional analysis reveals
N
CdA drogue =
(kg/m
3
) (m/s)
2
(no unit)

kg m/s
2
=
kg/m
3
(m/s)
2

= (m
2
).
Thus, a drogue devices CdA may be learned by finding
representative values for: air velocity, air density and the
force due to air resistance. If connected to a land vehicle
traveling through still air, a drogue devices air velocity
equals the vehicles ground speed. Estimating air density
occurs by applying the prevailing air temperature and air
pressure to the ideal-gas law.
The force is found with a little more effort. A spring-scale
may be adopted as a hand hold when attached to a small
parachutes shroud lines. While riding aboard a tandem
bicycle or an automobile, a passenger holds the open para-
chute by the scales handle into a non-turbulent region of
air flowing past the vehicle. The resulting drag causes an
SI-unit instrument to display a reading in newtons (N). If
the type of spring-scale available registers only kilograms,
multiplying that unit number by 9.81 m/s
2
(g) will yield the
drag force (N).
EXAMPLE
wind conditions: calm
relative air velocity: 8.5 m/s
air drag: 20.5 N
air temperature: 26 C
air pressure: 1014 hPa
air density = P/RT
1014 hPa
=
2.87 (26C + 273.15)
= 1.18(kg/m
3
)
drag
CdA drogue =
air density air velocity
2
0.5
20.5 N
=
1.18 kg/m
3
(8.50 m/s)
2
0.5
= 0.48(m
2
)
When deploying a drogue parachute to estimate either
the parachute's or a bicycles CdA it is strongly advised
first to deploy the canopy at low initial air/ground speeds,
then slowly and cautiously increase the towing vehicle's
velocity to a higher, though still modest, constant rate. The
parachute described herein must not be used to slow an
over-speeding vehicle.
J
o
h
n
C
. S
n
y
d
e
r
Drogue parachute
Finding
the CdA
of a
para-
chute
testing done while other traffic is pres-
ent will be affected. The air must
be still, a rare condition most fre-
quently occurring shortly after sunrise.
Constant slopes of sufficient length,
with reasonable constancy and accessi-
bility are essential. And, above all safety
must be in the forefront of an inves-
tigators thoughts at all times while
operating a bicycle. These hurdles are
often fully surmountable. With recog-
nized limitations, testing as presented
can provide an accessible way to esti-
mate meaningful frontal CdA and com-
posite Crr values for an individual
vehicle. Most important, these esti-
mates may be based on observation
of steady-state conditions as occurring
during actual road conditions.
NOTES AND REFERENCES
1. Witt, F. R. and Wilson, D. G.
1982. Bicycling science, 2nd ed.,
Cambridge, MA: MIT Press.
2. Crr has been modeled here as includ-
ing all conditions which impede the
rolling motion of a bicycle at a con-
stant velocity. Thus, hub bearing loss-
es are accounted for in the resulting
Crr values as well as road texture,
effects of tire design, tire air pres-
sure, operating temperature, axle nut
torque, et.al. It is recognized that
bearing losses might be accounted
for separately with additional han-
dling and measurements.
3. Personal correspondence with Israel
Urieli , Associate Professor, Dept.
of Mechanical Engineering, Ohio
University, Athens, Ohio, Fall 1999.
(Reference applies to text section
devoted to input values.)
4. Although m and b are shown as
area, their unit serves only to facili-
tate dimensional analysis.
5. Kyle, C. R. 1984. Improving the
racing bicycle system. In Second
International Human Powered
Vehicle Symposium: Proceedings.
Indianapolis, IN: IHPVA.
6. Most commonly associated with land-
ing aircraft, hang gliders, and racing
automobiles, deceleration parachutes
are not new to land HPVs. A
notable example appeared during
the 98 Winter X-Games in Crested
Butte, Colorado. During mountain-
bike races on a steep, snow-covered
ski run, several riders deployed large
parachutes worn as belt packs as a
marginally successful attempt to end
descents having maximum speeds
over 110 km/h (70 mph).
7. Compton, Tom. Coast-down tests,
Hard-core-bicycle-science discussion
thread (moderated e-mail list).
05 Aug 1997.
http://www.cyclery.com/lists/
hardcore-bicycle-science
8. Grade, instead of angle, has been
used here to simplify the appearance
of the equations. The value ultimately
cancels out.
9. Wilson, D. G. 1997. Wind-tunnel tests,
in review of Tour, das Radmagazin.
Human Power 12:4 (spring):78.
10. Tamai, Goro. 1999. The leading
edge: Aerodynamic design of
ultra-streamlined land vehicles.
Cambridge, MA: Bentley.
ADDITIONAL READING
AND INTERNET RESOURCES
11. International Human Powered
Vehicle Association (IHPVA). 1982.
The First International Human
Powered Vehicle Symposium:
Proceedings, Allan Abbott, ed. Seal
Beach, CA: IHPVA.
12. Cameron, A. 1999. Measuring
drive-train efficiency, Human Power
46:57.
13. Compton, Tom. Analytic Cycling
(interactive website)
http://www.analyticcycling.com/
QCHome_Page.html
14. Flanagan, M. J. 1996. Considerations
for data quality and uncertainty
in testing of bicycle aerodynamics,
Cycling Science, Fall.
15. HPV [internet] mailing list threads:
Measure wind resistance started
Thur., 05 Jan. 1996, Coast down
by computer started Mon., 05 May
1996; and Special Aero Question
started Fri., 10 May 1996, http://
www.ihpva.org.
16. Cycling software, Powercalc.
Jones, N. (shareware).
http://www.xsystems.co.uk/
machinehead/powercal.html
17. Martin, J., Milliken, D., Cobb, J.
McFadden, K. and Coggan, A. 1998.
Validation of a mathematical model
for road cycling power. Jl. of Applied
Biomechanics.14:276291.
18. Palmer, Chad. 1999. Understanding
air density, USA Today, 6 January.
http://www.usatoday.com/weather/
wdensity.htm
19. Papadopoulos, J. 1999. Simple
approximations for the effects of tire
resistance, wind, weight, and slope,
Human Power 48:1013.
20. Parks College Parachute Research,
student area (contains a method for
estimating a small parachutes CdA).
http://www.pcprg.com/student.htm.
21. Pivit, R. 1990. Measuring
aerodynamic drag, Radfahren,
Feb.:4749. (English language trans-
lation at: http://www.damonrinard.com/
aero/measuring.htm)
22. Sanders, Jol. 1999. A primer on
bicycle mechanics, with a spread-
sheet for power calculations (mono-
graph).
23. Snyder, J. C. and Tetz,
J. Tools: Spreadsheets. See http://
www.ihpva.org
24. Starkjohann, Christian. Drag
measurement on HPVs, http://
hal.kph.tuwien.ac.at/~cs/drag/
index.htm
25. The bicycle tachograph,
http://hal.kph.tuwien.ac.at/~cs/tacho/
index.htm
26. Tamai, G. 1998. Aerodynamic drag
components, Human Power 13:2
(spring):1517.
ACKNOWLEDGMENT
The author wishes to express his
deepest gratitude to the following indi-
viduals for their encouragement, input,
and keen insights: Carl Etnier, Jol
Sanders, Theo Schmidt, Jean Seay, John
Tetz, Israel Urieli, Dave Wilson and
Susan Snyder.
THE AUTHOR
John Snyder is a cyclist from Great
Falls, Montana.
John Clark Snyder, Jr.
12 Golden Valley Loop
Great Falls, MT 59404-6114 USA
Phone: 406-727-4132;
<[email protected]>
Human Power Number 51, Fall 2001 13 12 Number 51, Fall 2001 Human Power
Human Power Number 51, Fall 2001 15
FURTHER COMMENTS
ON THE SPICER ARTICLE
ON DRIVE-TRAIN EFFICIENCY
John S. Allen
The Spicer article on bicycle-drive-
train efficiency is interesting, and the
research appears to me to have been
well-conducted. The conclusion that
much of the power loss was not con-
verted to heat (that is, went into vibra-
tion instead) is interesting, as are the
conclusions that loss is not much great-
er with an unlubricated chain or with
chainline offset (though I think the loss-
es with offset would be greater with
older types of chain with flat side plates
which do not engage and disengage
smoothly). The conclusion that larger
sprockets increase efficiency is expect-
ed from theory.
Spicer used two torque-measuring
devices, one attached to the crank and
the other to the rear wheel. At the high
efficiencies typical of chain drives,
this approach to measurement is some-
what prone to error, because the
measurement of interestthe differ-
ence between the actual efficiency and
100%is a small difference between
two large quantities.
One way around this problem is to
use a single measuring device to mea-
sure a torque difference. Implementa-
tion of this approach is simple with a
unity drive ratio: the torque from the
motor at the input of the drive-train
may be applied directly to the brake
REFERENCE
Spicer, J.B., J.K. Christopher, M.J.
Richardson, J. Ehrlich, and J.R.
Bernstein. 2000. On the efficiency of
bicycle chain drives. Human Power
50:39.
AUTHORS
Claire Walton is a senior at Franklin
High School in El Paso, Texas. She is
interested in pursuing a career in math,
physics, and music.
John Walton is an engineering profes-
sor at the University of Texas at El Paso
who rides recumbent cycles.
John Walton <[email protected]>
Associate Professor
Department of Civil Engineering
University of Texas at El Paso
El Paso, TX 79968 USA
efficiency from the higher chain
tension is more important than the
loss of efficiency from having the
smaller sprocket. Clearly more experi-
mental data using different chainring
and sprocket combinations will be
required to answer questions on chain
efficiency definitively.
NOMENCLATURE
r
w
= wheel radius, m
= rotation rate, rev/s
F
w
= propulsive force, N
r
s
= effective sprocket radius, m
T
c
= chain tension, N

s
= chain power loss factor
or efficiency
The paper by James B. Spicer et
al., (2000) presents very useful and
relevant information for the further
understanding of HPV transmissions.
Its conclusions concerning the effects
of lubrication, rotation rate, and
tension on efficiency are highly valu-
able. We believe this contribution will
be viewed as even more significant
when the data are presented from a
slightly different perspective.
The testing apparatus was set up to
maintain the rpm of the front chainring
and the power applied to it at constant
levels. A single 52-tooth chainring was
tested. This leads to the observation
that the largest rear wheel sprocket
is most efficient. Though correct, this
result is not necessarily widely appli-
cable. When applied power, crank rpm,
and chainring size are held constant,
the velocity of the vehicle and the
force applied to the rear wheel must
vary. Since the same work and chain-
ring rpm are producing different veloc-
ities, a different force must be reacting
against the wheel. A physical analogy
for the columns in table 1 and the
results in figure 2 of the paper would
be a situation where the 52:11 gearing
represents downhill, 52:15 is level
ground, and 52:21 is uphill. In this
situation the 52:21 going uphill has
greatest chain efficiency. This is a valid
conclusion, but not the primary ques-
tion in HPV design and operation. Simi-
larly, in figure 3 the chain tension
is kept constant. At constant tension
larger sprockets are more efficient, but
they would also be delivering more
torque.
We are more interested in the case
of constant power and constant rpm
of the rear sprocket (i.e., constant
velocity of a vehicle on the road at
constant power supply to the wheel).
What sprocket will be most efficient
for a vehicle at constant velocity?
The experimental results suggest a
trade-off. At constant vehicle velocity
(and other conditions) a smaller
sprocket will have a greater chain
tension than a larger sprocket. The
higher tension in the smaller sprocket
will tend to counteract the inherent
lower efficiency of the smaller
sprocket (when sprockets are
compared at constant tension). Which
effect is more important for the situa-
tion of a vehicle traveling at constant
velocity?
In order to address this question we
take the measured data (specifically
the linear fits from figure 3 in Spicer
et al., (2000)) and present the results
in a revised format. The experimental
results are not changed: they are
merely presented differently using
simple algebraic manipulations.
The power supplied to the wheel, Pw
and by the chain, Pc are given by:
Pw = 2rwFw Eq. 1
Pc = 2rsTc Eq. 2
The difference in power is from the
loss of efficiency in the chain.
Pw = sPc Eq. 3
Pw
s = Eq. 4
Pc
Substituting the equations for power
and simplifying.
2rwFw = 2rsTcs Eq. 5
Fwrw
Tc = Eq. 6

c
r
s
In order to eliminate variables we take
a ratio of the chain tension from the
use of two different sprockets (11
and 21-tooth) while keeping power
supplied to the rear wheel constant:
T
21

11
r
11
11
11
= = Eq. 7
T
11

21
r
21
21
21
A similar equation is used for the
15-tooth sprocket. The linear fits of
chain tension versus efficiency for
the different sprockets in figure 3 of
Spicer et al., (2000) provide relation-
ships between efficiency and tension
for each sprocket. We assume that effi-
ciency is independent of chain speed
(as reported by Spicer et al., 2000) and
that most of the chain loss is asso-
ciated with the rear sprocket. Figure
3 of the paper indicates that at equal
tension, larger sprockets are more effi-
cient. It follows logically that most
of the loss in chain efficiency occurs
in association with the rear sprockets
which are all much smaller than the
52-tooth chainring.
Solving the system of equations pro-
vides an estimate of the chain efficien-
cy for the case of constant vehicle
velocity and power applied to the rear
wheel. The alternative data presenta-
tion is shown here in figure 1 with effi-
ciency for each sprocket size given as
a function of the chain tension in the
11 tooth sprocket. For example, when
the tension in the 11-tooth sprocket
is 1/0.006 = 167 Newtons, the efficien-
cies are 92% for the 11-tooth sprocket,
90.5% for the 15-tooth sprocket, and
88.5% for the 21-tooth sprocket
assuming the same vehicle velocity and
power to the rear wheel for all three
sprockets. The corresponding tensions
and efficiencies for the 15 and 21 tooth
sprockets are calculated using equa-
tion 7. The lines are truncated approxi-
mately at the limits of the experimental
data. Over most of the experimental
range the smaller sprockets give great-
er chain efficiency. The lines appear
to converge at high tensions, with all
three sprockets giving high efficiency.
The surprising and counterintuitive
result is that the smaller sprockets
have greater estimated chain effi-
ciency at constant vehicle velocity
and applied power than the larger
sprockets. Therefore, the increased
TECHNICAL NOTES
Efficiency of bicycle chain drives:
results at constant velocity and supplied power
by Claire L. Walton and John C. Walton
0.002 0.004 0.006 0.008 0.01 0.012 0.014
82.5
85.0
87.5
90.0
92.5
95.0
97.5
100.0
E
f
f
i
c
i
e
n
c
y
,

%
11
21
15
Reciprocal tension in 11 tooth sprocket, N
-1
Figure 1. Efficiency of the 11, 15, and 21-tooth sprockets at constant
vehicle velocity and power to the rear wheel.
Table 1. Drive efficiencies for different chain configu-
rations
50 RPM 60 RPM 70 RPM 60 RPM 60 RPM
100 W 100 W 100W 150 W 175W
5211 92.5 91.1 88.7 94.6 95.5
5215 94.7 92.3 90.4 96.2 97.5
5221 95.2 93.8 92.0 97.4 98.2
80
85
90
95
100
0.002 0.004 0.006 0.008 0.01 0.012 0.014
52-11
52-15
52-21
D
r
iv
e
tr
a
in
e
ffic
ie
n
c
y
(
%
)
Reciprocal tension (N
-1
)
Figure 3 (left) and Table 1 (above) provided by James
B. Spicer for his article in Human Power 50, "On
the efficiency of bicycle chain drives" are repro-
duced here for the convenience of readers of the
technical notes submitted by Claire C. Walton,
John Walton, and John Allen.
14 Number 51, Fall 2001 Human Power
Human Power Number 51, Fall 2001 17
cessful in keeping the matter of product
safety and testing entirely to them-
selves!
Chris Juden is now on a committee
of the European standards organization
CEN, which enforces its standards on
member countries. ISO standards are
only, it seems, recommended.
John LRaY Stephens wrote You
should get some tire-and-rim-industry
experts involved with this [question of
standards for tire fits]. Unfortunately, I
have never heard of any such persons.
Tires just seem to float down out of
heaven (or rise up from hell?). After
considerable efforts to reach tire manu-
facturers I was told that Vredestein, the
Netherlands manufacturer of the tire
on the German recumbent on which I
had my most-recent episode of instabil-
ity, was conducting a study on run-flat
behavior. However, when eventually I
received a courteous response from Mr.
U. K. Banerjee, it turned out that he was
investigating puncture prevention.
I bought a product called Snake-
charmer from Bikewise International,
Mammoth Lakes, CA. It was a length of
dense solid trapezoidal-shaped rubber,
intended to be fed into the rimwell
under the tube to prevent pinch flats
and presumably to give some run-flat
capability. It was produced only for
large ATB tires, at least at the time of
my purchase,
and I could not
test it. It would
add a consid-
erable amount
of mass to a
wheel and, I
would think,
stress the tube,
which would
have to wrap
around it.
In September
1998 I added
the problem
statement on
flat-tire stability to my list of undergrad-
uate-thesis topics at MIT. Andy Oury,
then a senior, responded enthusiastical-
ly, carried out several valuable experi-
ments, and has allowed me to report
some of his results here. We drew up
a too-ambitious program in which we
recumbents, and Joshua Putnam, who
considered the problem serious enough
to institute the practice of letting the air
completely out of the front tire when
trying out a new bike. Bill Volk wrote,
I too find the situation to be unaccept-
able. I run heavy, inefficient thorn tubes
because of my fear that a blowout at
high speed would be a disaster. Why
cant we have rims that retain the tires
even at no inflation? And perhaps a
rubber strip that is placed around the
rim, under the tube, that supports the
bike on loss of air pressure. I had
Performance semi-slick 26" tires that fit
so snugly that you could safely ride no-
inflation. That should be the standard.
Presumably because of a tight-fitting
tire, Ed Deaton of Fools Crow Cycles,
faced with difficult choices, rode five
miles (8 km) on a flat front tire: he
had IRC Roadlites with Sun M14 rims.
Similarly Andy Milstein of Princeton
had no trouble riding with a flat
front tire. It was a Tioga Comp Pool,
measured by Mark B. of Wheel Life
Cycles to be 46-mm wide, on a Sun
CR-18 20 x 1.75" rim of about 27-mm
width. (That was significant because
one of my early suspicions, and a
concern of Larry Black, was that a
wide tire on a narrow rim might have
a greater tendency to flop alternately
left and right.)
Bill Volk mentioned that Sutherlands
Handbook for Bicycle Mechanics had a
good section on fits between different
brands of rims and tires, but my edition
did not have this section, and I
could not get an answer to my letter
to Sutherland asking about standards
of fit. John Allen, prominent bicycle
expert and author, sent me a copy of his
Japanese Industrial Standards D 9421,
Rims for Bicycles, giving a tolerance
of +/- 3 mm for rim circumference, and
of standard K 6302 Rubber pneumatic
tires for bicycles, which, he pointed
out, gave neither tolerances nor dimen-
sions of tire beads. (Later, Andy Oury,
see below, found that the International
Standards Organization ISO 5775/1
Standards for bicycle tires and rims
also had tolerances for rim diameters
but not, as far as he could determine,
for tire beads. This was confirmed by
Chris Juden, below.)
My instinct tells me that the old inch
sizes had some specified or customary
standards because my old 27x1-1/4"
and other inch tires were all at-least
good fits on the rims. Now, it seems
from our experience and that of many
people who wrote to me, it is entirely
by chance that one gets a tire that is a
tight fit on a rim and that will therefore
provide a substantial degree of safety
in the event of a front-tire blowout.
However, Doug Milliken, a long-time
consultant to Alex Moulton, wrote that
Moulton controls both the rim diameter
and the bead size of his 17" tires.
I wrote to Andrew M. Fischer, a
Boston-area attorney who specializes in
helping bicyclists with liability claims,
but he had had no experience of this
problem.
Chris Juden, technical officer of the
Cyclists Touring Club (UK), and a
resource on every aspect of bicycle per-
formance, wrote: There are ISO stan-
dards for tyres and rims: ISO 5775 parts
1 and 2. The only trouble is: they
were written by tyre manufacturers for
their own convenience so -2 places
rather tight tolerances upon rim-bead-
seat diameter (plus or minus 0.48 mm)
whilst -1 says nothing at all about the
corresponding tyre-bead dimension!
Having once been involved in rim
manufacture, I can tell you we used
to have some interesting arguments
with Raleigh around the fact that a
lightweight alloy rim inevitably shrinks
some 0.46 mm in diameter when you
put properly tensioned spokes in it!
Since this standard doesnt say if its
talking about pre- or post-build dimen-
sions, we had to restrict ourselves to
only the top half of that measly tol-
erance or else Raleigh quality control
would measure our bare rims or built
wheels respectively, depending upon
whether their latest shipment from
Michelin were a tight or loose fit!
On the BSI committee we found the
tyre people always played their cards
very close to the chest and would never
be drawn when invited to submit appro-
priate limits and fits for their products
or even the criteria for a simple blow
off test. With many an Ah yes, but and
its not so simple as that, the cycle-tyre
industry has thus been remarkably suc-
in complication) by replacing the chain
drive with a gear drive or with knife
edges positioned by a jackscrew to
achieve the desired torque ratios. I
would be most interested to hear from
someone who attempts any of these
approaches.
Spicers conclusions suggest some
additional tests which he did not con-
duct. His infrared photographs show
that much heat was generated in the
derailleur pulleys. Most derailleur pul-
leys have primitive plastic sleeve bear-
ings, yet no test was done of bearing
lubrication, or of ball-bearing pulleys.
Spicer never tested for how much fric-
tion is reduced by eliminating one or
both derailleur pulleys. And would larg-
er pulleys, with their smaller chord
factor, increase efficiency by reducing
vibration as well as bearing friction?
John S. Allen
<[email protected]>
http://www.bikexprt.com/
BICYCLE STABILITY AFTER
FRONT-TIRE DEFLATION
Dave Wilson (reporting partly for Andy
Oury)
Drawings by author, 2000.11.13
THE PROBLEM
On three occasions I have had front-
tire blowouts, or at least rapid loss of
pressure, that have resulted in my hav-
ing been thrown off my bicycle with
some violence. One was when riding a
Moulton road bike as a bus was about
to pass; one was on an Avatar LWB
recumbent; and one on a CLWB recum-
bent, when I narrowly avoided being hit
by a large truck. A friend told me about
someone who was, in fact, killed after
his front tire burst, causing him to be
propelled into the path of a car. A photo
intended to be humorous in Bicycling
[Magazine] showed two men on a tan-
dem competing in the Davis Double
Century just after their front tire
deflated (almost certainly after braking
sharply at a corner during a mountain
descent, thus overheating the rim). The
caption stated that as they hit the
ground their bones breaking sounded
like a famous breakfast cereal (snap!
crackle! pop!).
The reporting from dead bicyclists is
zero, and the reporting of and examina-
tion of bicycle accidents is so perfunc-
tory that it is highly probable that a
considerable number of deaths and
serious inju-
ries are the
result of
instability
following
front-tire
deflation.
Therefore
this has to
be regarded
as a serious
problem.
OUR STUDY OF THE PROBLEM
In the summer of 1998 I wrote about
flat-tire instability to a list [Internet
discussion list] then called HBS, for
Hardcore bicycle science, moderated
by Jim Papadopoulos (the name is now
shortened to Bicycle science and it
is moderated by Sheldon Brown). No
one reported previous studies of this
problem apart from one described by
Doug Milliken, who wrote a letter
Flat-tire directional performance to
Human Power in spring 1991 (9:1, 17).
He tested a motor-cycle fitted with pro-
prietary run-flat tires on the rear wheel.
The tires had a flap of rubber on the
outside of the tire that fitted tightly
over the rim and acted as a bead-reten-
tion system. One with a small flap did
not in fact hold the bead when the
tire was flat, and the bead fell into
the well in the rim. The tire flopped
around, causing the motor-bike to go
unstable, even though the tire was on
the rear wheel. The second tire with a
wider flap held the beads in place.
With this tire, Milliken found that
he could run the bike at high speed
(80 km/h) and could perform various
maneuvers without problem. He
thought that good run-flat bicycle tires
would probably be tubeless.
I wrote also to the HPV list, and sev-
eral writers on this and on the HBS
list contributed valuable experiences
and suggestions. Some reported similar
occurrences to mine, including Dave
Larrington of the British Human Power
Club, who had had instant crashes
from front-tire flats on regular bikes
(upwrong, in his words) and on
at the output, where it cancels except
for the difference due to power loss.
At 100% efficiency, the torque of the
motor and brake cancel, and so do the
measurement errors. Assuming that a
reasonably accurate measurement of
the input torque can be taken, this
approach promises a high degree of
accuracy for a high-efficiency system.
Implementing this approach is some-
what more difficult in the case of a
bicycle chain drive, with its step-up
ratio. The torque-combining system
must have the same ratio. Suppose, for
example, that the bicycle's chain drive
has a 52/15 drive ratio. Then we could,
for example, use another chain drive
with the same 52/15 ratio to combine
the torques at the motor and brake.
What objections might be made to
this approach? A first objection might
be that inefficiency of the torque-com-
bining drive system would corrupt the
measurement. But on second thought,
it need not. The torque-combining drive
system is not in motion, and so it
absorbs only power which has already
been lost through vibration of the pri-
mary drive system. And that vibration
converts the sliding friction of the sta-
tionary torque-combining chain drive
into viscous friction (as also happens,
for example, with the pivots of phono-
graph tone-arms when subjected to the
vibrations transmitted from the stylus
in the record groove).
There is another, real and serious
problem, however, and it also occurs
because the torque-combining chain is
not in motion. Chord factors average
out in a chain drive that is in motion,
but not in one which is stationary. The
chord factor of the 15-tooth sprocket
of our example is 1/cos 12, or 1.022,
and the chord factor of the 52-tooth
chain-wheel is 1/cos 3.46 or 1.0018.
The resulting range over which mea-
surement may vary is the product of the
two chord factors. An error range of
over 2% is hardly desirable, given that
the goal of the suggested measurement
technique is inherent, high accuracy.
The chord-factor problem may be
greatly reduced by doubling or tripling
the number of teeth on the torque-
combining chain-wheel and sprocket,
or may be eliminated (at some cost
Figure 2. "Snake-
charmer" run-flat insert
(note rim without bead
seats)
16 Number 51, Fall 2001 Human Power
Figure 1. Tire bead off seats:
tread flops to side
to be the more common arrangement
depicted by artists. The pros and
cons of the different arrangements are
described as well as the reasoning
which led to the reconstruction plan of
the Olympias, which was then launched
in 1987. However before this, models
and even a full-scale floating section
were built in order to study the rowing
geometry. The ship was built with more
or less traditional materials and meth-
ods, but not exclusively, as the major
interest lay in operational research and
not historical ship-building. It took the
combined resources of two trusts (one
British, one American) to build the
Olympias, and also of the Hellenic navy,
which owns and operates it.
Part II is completely new in this sec-
ond edition and describes the results
of numerous sea trials in the years
19871992. The Olympias nearly lives
up to its expectations, but its 5- to
5.5-knot cruising speed is slightly slow-
er than hoped for. Peak speeds over
8 knots were recorded. The total effi-
ciency of the oar system was estimated
at about 1/3. Rowing the Olympias is
hard, difficult, and sometimes unpleas-
ant work, but apparently there is usu-
ally no shortage of volunteers, whether
civilian or navy. Only the top level of
rowers can see their oars, whereas
the bottom-level rowers can see
Human Power Number 51, Fall 2001 19
nothing in their poorly venti-
lated and smelly workspace
(where they are also being
dripped on by the sweat from
the upper rowers). The logis-
tics of operation are remark-
able: for example, a days
journey requires that almost
two tons of water be on
board for consumption by the
rowers. Olympias also per-
forms well under its two
square sails. Combined sail-
ing and rowing allows slightly
higher speeds with a bit less
effort.
The report concludes with
a discussion of the results
and suggestions for slight
improvements. However, it
appears that the Athenian
Trireme project was just
about optimal, and the same
can be said for the book.
Theo Schmidt
THE DANCING CHAIN by Frank Berto,
Ron Shepherd and Raymond Henry.
ISBN 1-892495-21-X, available from Van
der Plas Publications, San Francisco,
CA or, signed, from Frank Berto
<[email protected]> for US$58.00
including shipping in the US, more for
other destinations.
Reviewed by Dave Wilson
This is a large, beautiful, hard-cover,
profusely illustrated and comprehen-
sive book on derailleur gears. (It also
has intriguing paragraphs on hub gears,
bicycle companies, bicycle magazines,
the people who invented, developed,
sold and rode the gears and bikes, and
much more.) I have been interested in
bicycle gears for many decades, but I
found myself continually saying "Wow!"
to myself as I learned about aspects of
gears that I had long wanted to know. I
used to look forward to every article of
Frank Berto's on his exhaustive tests of
new derailleurs in Bicycling and other
magazines, and I have great respect
for him and his two co-authors,
Raymond Henry from France and
Ron Shepherd from Australia. Walter
Ulreich of Germany and Tony Hadland
of Britain also contributed to and
checked parts of the book. Van der Plas
Publications have put the same effort
REVIEWS
THE ATHENIAN TRIREME, 2nd ed.,
by J. S. Morrison, J. E. Coates and N.B.
Rankov, Cambridge University Press.
ISBN 0-521-56456-5 (pbk.)
Reviewed by Theo Schmidt
The Athenian Trireme is a compre-
hensive report on the reconstruction
and testing of a Greek trireme with 170
rowers, probably the largest and most
successful such project in recent times,
even if the object of interest dates back
to several hundred years BC. These
Greek warships were numerous and
very successful in their day, but none
survive, as they were unsinkable and
were thus never preserved in bottom
mud as were some other ships. It may
also be characteristic of the time that
a great deal was written about the bat-
tles in which these ships were used, but
almost nothing about the vessels them-
selves. Well known is their technique of
ramming and holing other ships with a
protruding and strengthened bow.
Part I of the book, which was orig-
inally published in 1986, starts with
painstaking detective work assembling
and correlating the meagre data avail-
able such as not-to-scale pictures on
vases, coins and reliefs. It is not even
clear whether the name trireme (Greek:
trieres) actually means three vertically
displaced levels of single-oared rowers,
whose feasibility the project aims to
prove, or single levels of oars manned
by several rowers each, which seems
into producing this book as it does for
the series of proceedings of the confer-
ences on bicycle history.
Every reader of Human Power
should, if you can afford it, buy this
book. It gives you all the history, the
fundamentals, the reasons for contin-
ual changes in design, the pitfalls to
avoid, and so on that you will ever
need. It may seem expensive, but it
has been produced at what is likely
to be a considerable loss (even if the
authors receive no compensation for
their years of dedicated work). It has
been published by Frank Berto because
no publisher would undertake so large
a publication at so little possibility of
sales sufficient to cover expenditures.
Accordingly, there are only a few avail-
able. Rush to get your copy! And give
thanks to Frank Berto and his collabo-
rators for their major contributions to
the human-power movement.
BICYCLE DESIGN by Mike Burrows.
Open Road Publishers, UK,
ISBN 1-898457-07-7; US edition by
AlpenBooks Press, ISBN 0-9669795-2-4
Reviewed by Dave Wilson
Mike Burrows is the best-known and
probably the foremost bicycle designer
in the world today. He is also one of
the top designers and builders of HPVs.
He has therefore done a great deal to
bring together two of the very different
branches of bicycles and bicycling and
to endow HPVs with respect from the
regular biking community. He is also
an everyday bicyclist and a racer in
his Windcheetah Speedy tricycle or his
wanted to look not only at bead reten-
tion but also at the effect of the ratio
of tire width to rim width (ATB tires
in particular are usually bulbous, hav-
ing a pear-shaped cross-section on what
seems like a small rim) and of tire-
sidewall stiffness. Andy Oury worked
on what the correspondents just quoted
thought was the most important factor,
bead retention.
THE EXPERIMENTS
We first thought that we could do a
highly controlled experiment by having
my troublesome bicycle wheel and tire,
held in a frame, running on the surface
of an inverted portable belt sander.
However, the tire did not display the
extraordinary alternating flops, left and
right, that had thrown me off my bike,
and that had prevented me even from
pushing the bike subsequently. Oury
found that, for the flopping behavior
to occur, he had to rig up a bike to
run along a simulated roadway with a
similar number of degrees of freedom
as has a bicycle when it is being ridden
(or pushed).
The simplest way of producing bead
retention on the shoulders of the wheel
rim after deflation seemed indeed for
them to be a tight fit. I have had tires
that could be stretched over the wheel
rims only with great difficulty. When
these were inflated, the tire beads
remained in the rim well until the
tube inflation pressure reached around
80% of normal full pressure. They then
snapped over the rim shoulders with
a satisfying crack. My experience fol-
lows that of Doug Milliken and Bill
Volk: I have never found tire instability
with tires that were a tight fit on the
rims, and which, therefore, did not flop
loosely around in the rim when they
became deflated. I confess that I can-
not remember if I have had a front-tire
blowout with a good-fitting tire. I would
certainly remember something like the
instability that made staying on the
three bikes mentioned above absolutely
impossible.
The tires that caused me the prob-
lems were exceedingly loose. This char-
acteristic made puncture repair almost
a pleasure, because one could get the
tires on and off without tire levers.
They were so loose, in fact, that cen-
tering them during subsequent inflation
became difficult: it was easy to produce
an eccentric rolling surface, even to
the extent of having the tube pop out
between tire and rim where the tire was
high. Oury built up the rim shoulders
using standard masking tape, and he
put on fifteen layers before the tires
were retained and the flat-tire flopping
was inhibited. His experiments there-
fore did a great deal to confirm the
premise: that a slack fit between tire
bead and wheel rim is the prime cause
of flat-tire instability and that a tight
fit will therefore provide a substantial
degree of safety in the event of a front-
tire blowout.
RECOMMENDATIONS
The International Standards Organi-
zation should form a committee of tire
and rim manufacturers to agree on
standards of rim diameter and shape
and of tire-bead diameters so that
a tight fit could be relied upon in
all circumstances. Then manufacturers
should agree to observe these stan-
dards.
Dave Wilson
<[email protected]>
*Oury, Andrew P. (1999). Run-flat perfor-
mance of bicycle tires and rims Thesis,
B.S.M.E., M.I.T., Cambridge MA.
Right: Cover of the latest paperback ver-
sion of this report on the reconstruction
and testing of a Greek trireme. The first
edition was published in 1986 (ISBN
0521564190) and as a textbook in 1990
(ISBN 0521311004). Hardcover edition of
the 2nd ed., ISBN 0521564190
From the book jacket: For this second edition, the text
has been recast and a number of substantive changes
have been made in the light of the sea trials and
new research. In addition, there is an entirely new chap-
ter which describes the trials of Olympias in detail,
reports the performance figures obtained, and outlines
the changes which the authors would wish to incorporate
into any second reconstruction. There are nineteen new
illustrations, including eleven photographs of Olympias at
sea demonstrating features of the design which could be
represented only by drawings in the first edition.
Figure 3. Tire beads retained: symmetric
tread
18 Number 51, Fall 2001 Human Power
GUEST EDITORIAL
YOUR NEXT VEHICLE: A VELOMOBILE?
Joachim Fuchs
Velomobiles are fully faired recum-
bent vehicles for everyday use. Many
people consider that they have the
potential to play an increasing role
within different types of human-pow-
ered vehicles. In addition, they could
give a positive contribution on our
future traffic. Or, more precisely: can
fully faired everyday recumbents
replace cars and normal bicycles? This
article gives a view over the recent
developments in Europe.
First of all, velomobiles are human-
powered vehicles that differ from
normal bicycles in function and
appearance. There are many types,
produced as prototypes and in small-
scale manufacturing. Velomobiles are
fully faired recumbent cycles that
are constructed for everyday use and
provide full rain protection. The fairing
also gives better protection from
accidents for the driver.
An important question is: Why should
I use a velomobile, and what are the
advantages compared to a bicycle on
the one hand, and a car on the other
hand?
An obvious example of the difference
between the rider of a velomobile and
that of a normal bicycle is that users
of velomobiles wear almost the same
clothes in summer and winter. This is
one main argument for velomobiles:
there is no need for a look outside in
the morning. No shapeless rain suits
hinder ones pedaling. In addition, both
women and men can ride in business
suits if they wish.
This implies that there must be good
ventilation, an important factor in velo-
mobile design. Because of the absence
of direct wind, adjustable air flaps are
integrated in the fairing. The air stream
within the fairing is moderate com-
pared with that on an unfaired bicycle.
Therefore, the rider learns to moder-
ate his or her own power. My own expe-
rience shows that one sweats less in a
well-ventilated velomobile even in sum-
mer. In contrast, on a normal bicycle
one is getting blown dry by the wind
and sweating starts intensively after
riding. This is unpleasant when riding
to work regularly. Properly mounted
air flaps within a fairing can avoid this
effectively.
When riding uphill, passive ventila-
tion does reach its limit, on a regular
the chapter on the future of bicycles.
Ive realized that just about everything
that I hope to see in future bikes, like
cantilever wheels that one can change
rapidly when one has a flat or when one
wants to put on a studded tire (as on the
day of writing), and all-enclosed trans-
missions, and disk brakes, have been
worked on by Mike Burrows. The man
is a master and his book is a must
read.
Dave Wilson
deep-section aero wheels, which can
have very stiff rims, and the need for
more-forgiving rims on all-terrain bikes,
and his frank statements on what he
doesnt like, are all high-value and high-
octane. Likewise his comments on sus-
pension, braking and monoblades are
pure common-sense that isnt as com-
mon as we would like.
As it happens, Im in what I hope are
the closing stages of writing the third
edition of Bicycling Science (with Jim
Papadopoulos), and I am working on
bicycle and in a velomobile, because
the speed of the vehicle is not enough
to produce a sufficient air flow. The
question is often asked: Is it possible
to ride uphill in a velomobile? Velo-
mobiles are around 15 per cent heavier
than upright bicycles if the rider is
included in both cases. The speed loss
uphill can therefore never be larger
than this 15 per cent. On small
or moderate gradients uphill, the
lower air resistance of the velomobile
compensates for this disadvantage.
Velomobiles normally have a smaller
effective frontal area (which governs
the air resistance). This is the reason
for the higher speeds that can be
reached with some velomobiles.
Higher speeds are attractive especially
for riders that are used to
physical exercise and have fun
riding with their own power.
Those people who like riding
at 1.53 m/s (510 mph) will
not feel a difference. With a
little more power input, riders
who are not very sportive
become astonished when they
can ride at 13 m/s, 30 mph,
for some time. This is indeed
possible with sportive velo-
mobiles.
There are, naturally, many
different kinds of velomobiles.
Most velomobiles are tricy-
cles. They are stable, anyone
can ride them immediately,
and they have good luggage capacity.
Two-wheelers are ridden by sportive
people because they can normally go
faster and can lean in turns. Examples
of such velomobiles in Europe are
Aeolos and Desira. In everyday use,
the handling is very important. Getting
in and out should not be hindered by
the fairing. This is the precondition
for switching from a bicycle to a velo-
mobile: it should be practicable for
short distances (buying bread rolls on
Sunday morning).
A velomobile that exhibits its advan-
tages only in rainy weather would not
find many users, because the addi-
tional place to park such a velomobile
is a problem at least in urban cities in
Europe. To some extent, one can say
that velomobiles combine the advan-
tages of cars and bicycles. As bicycles,
velomobiles can be used on roads
on which cars are banned. Often,
everyday distances can be covered
through a nice landscape whereas
car drivers have to use boring main
roads. Besides that, velomobiles are
economic even though they are expen-
sive when purchased. This is espe-
cially the case if velomobiles are often
used and if they replace other vehicles.
Compact velomobiles can be taken
along in a train with a bicycle compart-
ment, at least in some countries in
Europe. Some designs can be taken
apart to make them easier to stow,
which is necessary with most tricycles.
People who like to ride with other
(non-velomobile) riders should take
20 Number 51, Fall 2001 Human Power
A young velomobile enthusiast during a test...
Children have a lot of fun sitting in velomo-
biles even they cannot see through the wind-
shield!
Ratcatcher short-wheelbase recumbent
bike. Several years ago he was hired
by Giant of Taiwan, one of the largest
bicycle manufacturers in the world. His
influence is therefore already major and
is likely to increase.
Mikes book on bicycle design has
been eagerly awaited. When Open Road
Publishers failed recently we were con-
cerned that we would not get to see
it, but we are fortunate, at least on the
American continent, that AlpenBooks
has picked it up. It is a sturdily bound
paperback of 160 pages, on bright-white
stock, with some color centerfolds of
"Mikes favourite bikes. All the photo-
graphs are clear and good, B/W and
color, as are the diagrams. There are
also several excellent cartoons by Jo
Burt and Geoff Apps.
The book starts with a gracious fore-
word by Richard Ballantine, paying trib-
ute to Burrows many characteristics,
including some that he has recently
learned: diplomacy and gentle advo-
cacy, which increase his effectiveness
as something close to a revolutionary.
Then we plunge into what I can only
describe as pure Burrows: fifteen chap-
ters of Mikes strong views on every-
thing from ergonomics to monoblades
and cantilever wheels. They are well
written (or well edited by Tony
Hadland) and expressed with nice mod-
esty as well as pride in his many innova-
tions, which he often credits to others.
For instance, the monoblades and canti-
lever wheels he saw on an 1889 cross-
frame Invincible in a museum. (He
is also kind enough to state that he
wants his book to fit in the gap between
Richards Bicycle Book and Whitt and
Wilsons Bicycling Science. He suc-
ceeds superbly! He wanted no algebra
or equations, and he managed that.) He
apologizes that his book is written from
a British perspective. It is, but he gives
credit to non-Britons. The Giant com-
pany also comes out well.
Many of the topics that occupy much
discussion space on HPV and bicycling
mailing lists would be enlightened by
Burrows trenchant observations and
opinions. In discussing frame design he
draws a distinction between torsional
stiffness and vertical compliance that
makes a lot of sense. His guidance on
Human Power Number 51, Fall 2001 21
The variety of velomobiles indicates that there is still potential for further developments.
Different velomobiles present at meeting in Oktober
2000 in Germany. From front to rear: Leitra, Aeolos
and Cab-bike. In contrast to the other vehicles, Aeolos
(a development of the author) is a two-wheeler. Further
informations can be found at http://www.velomobile.de
into account that one can chat while
pedaling. This can also be useful
for communicating with other traffic
users, mostly car drivers. A properly
constructed velomobile can be pushed
along sidewalks and shopping malls.
(You will have a built-in shopping
cart!) Most enthusiasts first think of
rain protection when the talk is about
velomobiles. On the one hand, there
are sly unfaired riders who calculate
that it rains only a small amount of
time. On the other hand, in practice,
people who get wet once are more
likely to use the car next time.
To match the demands of practi-
cability, constructors have to design
their products with considerable skill.
Protection from cold wind in winter
time is as important as from rain.
On a normal bicycle, it is often hard
to choose the right clothes. After
commuting some time, one begins to
sweat under the warm clothes. Velo-
mobiles avoid this problem, because
one can adjust the air flow and do not
have to change clothes. And what is
the feeling when getting in a velomo-
bile? First of all, velomobiles are
quite narrow. This is a necessary prop-
erty, because velomobiles should be
built light in weight and compact
to consume only little space when
parking. The feeling in a velomobile
is individually quite variable. Some
elderly people feel ill at ease in the
fairing even if they dont touch it.
Others report that they feel safe and
secure in the fairing because of the
protection effect. The well-being is
further influenced by other factors.
The sight through the windscreen for
example should not be affected by
reflections. A velomobile should have
a low noise level inside the fairing.
All this contributes to the feeling that
is specific for velomobiles. Most test
riders get along quickly with the new
vehicle. The main advantage of an
ideal velomobile is that it is the
proper means of transportation in
most everyday situations. Thus,
partially faired vehicles with the head
outside the fairing have the disad-
vantage that the head might have to
face a very strong wind. Nevertheless,
some bicycle riders choose that kind of
recumbent vehicle, believing that they
have more advantages than upright
bicycles. Velomobiles moreover offer
a built-in rain protection, advanced
aerodynamics and a protection from
cold windand all that in one vehicle.
There are several velomobiles
commercially available. The first velo-
mobile to attain widespread use was
the Leitra. The Leitra follows the
classic concept with a lightweight steel
frame and a glass-fiber fairing fixed to
the frame. This offers the advantage
that there is less noise than in a
monocoque vehicle. Later velomo-
biles often use a construction that is
easier to realize. One example of that
kind is the Cab-bike. Such velomobiles
dont have a frame in its own sense.
Instead, the fairing forms a closed shell
with all components mounted to it.
Vehicles of that kind have fewer parts
and are cheaper to produce. But in
the case of damage, it is necessary
to repair it skillfully to ensure the
shell recovers its rigidity. There are
further aspects to take into account,
for example eye-level height, which
is important in urban traffic. Besides
the commercially available velomo-
biles there are some vehicles that
were either produced for personal use
or have at least the potential for a
commercial product. The inventors of
prototypes add to the diversity of
velomobiles. To give some examples:
Veleric, Hajen, Jouta, Desira, Pedicar,
Muscar. Each vehicle was constructed
for different purposes; the Desira even
exists in several versions.
Although velomobiles offer lots of
advantages, one should remember that
velomobiles serve a niche market. The
price of more than approx. US$5500.00
is far higher than that of most upright
bicycles. The breakthrough would be if
everyone could find a velomobile that
fits the demands of daily commuting.
Can you see yourself in a velomobile
soon?
REFERENCES
A lot of information can be found on
the internet:
General information: www.ihpva.org
www.velomobile.de This is an internet
platform for velomobiles that was
just started (please have a look on
it now and then to get current infor-
mation). The velomobiles mentioned
in this text are introduced or linked
there.
Several publications in the Proceedings
of the European Seminars On
Velomobile Design I- IV. Further
information at www.futurebike.ch
Printed papers:
Curneal, Steve, 1990. Omega: The evo-
lution of a recumbent. HPV News
Jan/Feb:1011.
Stuart, Bob. 1994. Coroplast HPV body
construction. HPV News Dec:1719.
Dovydenas, Vytas. 1990. Velomobile.
Berlin: Verlag Technik GmbH [ISBN
3-341-00790-3; originally published
in Lithuania and translated into
German; out of print]
Joachim Fuchs <[email protected]>
EDITORIALS
THE END OF A DREAM
My principal activity other than work
on Human Power seems to be trying
to finish (with Jim Papadopoulos) the
third edition of Bicycling Science. I
have been working on the chapter
about what we can expect and what
we might want in our future bicycles.
The easy way out was to refer readers
to Encycleopedia and Bike Culture
Quarterly and other publications of
Open Road. But the shocking news of
the bankruptcy and the laying off of all
employees of Open Road has just hit us.
It has seemed to be a miracle that the
company could do what it did: to pro-
duce (since 1993) a series of superbly
produced texts and magazines and vid-
eos on alternatives in cycling. Every
issue of everything it did was not only
a resource for the cycling enthusiast:
every photo was so beautifully done
and reproduced that each item of out-
put became a coffee-table book. Im
using that as a term of admiration, not
disparagement. Each publication could
be left on a table at a doctors office
and would be guaranteed to be looked
at with wonder by a wide range of peo-
ple. Thus it spread acceptance and even
respect for the more adventurous, and
the quirkier, human-powered vehicles.
We enthusiasts would seize each issue
of each series and be inspired by the
quality of the publication and by the
ingenuity of the subjects. We marveled
that this could be done without adver-
tisements of macho trucks and SUVs on
every other page. Open Road had been
going from strength to strength for the
better part of a decade, publishing in
English and German, with agents in
four countries, organizing Bike Culture
Weeks in its home territory in and near
York, UK, and, lately, publishing two
superb books.
I had thought that Open Road must
have an angel funder, in the way that
the early IHPVA had infusions for prizes
from Du Pont especially, but it seems
that there was none. The speed of the
collapse of Open Road, and the large
amount of debt at the end, are sadden-
ing and sobering. We in the new, reor-
ganized IHPVA are far less ambitious,
even timid by comparison, and yet we
are hanging by a slim financial thread.
The temptation to draw some parallels
is irresistible: if we want superb pub-
lications like those of Open Road
to continue, and the less-ambitious
but irreplaceable magazines such as
Recumbent Cyclist News (RCN),
Recumbent UK, and all the other publi-
cations of our national and local HPV
associations (in which we hope we
may include Human Power), we must
support them with subscriptions and
with the recruitment of others to join.
People like Jim McGurn and his asso-
ciates at Open Road performed the
miracles they did in the spirit of mis-
sionaries with a vision, at low or zero
pay. Many selfless people also invested
in a dream, and have lost all their
money. We give heartfelt thanks and
appreciation to all involved for their all-
too-short period of brilliance, one that
shone on us all. We hear that some
of the former staff have plans for new
publications to try to carry on the tradi-
tion, and we wish them god speed.
Dave Wilson
TIRESOME
Pneumatic tires were patented twice,
in 1845 by Thomson and in 1888 by
another Scot, Dunlop. (Patent proce-
dures can still be as capricious.) When
I was a child, motor-tire failures were
to be expected in regular driving.
Nowadays a flat on almost any motor
vehicle is, or was, very rare. Racers go
around the turns of Indianapolis and
the twists of European Grand Prix cir-
cuits at over three-hundred km/h, at
very high tire temperatures, with amaz-
ing reliability considering the condi-
tions. At one time the favorite tires at
Indy were Firestone. So how did it
come about that Firestone tires were
implicated in many failures in Ford
Explorer vehicles at far-lower speeds
and temperatures? How could Ford
design a vehicle that would roll over
after an event as expectable as a tire
failure? And how could Ford make a
vehicle (on which its profit margin is
allegedly very high) that, when it rolled,
had no inbuilt roll cage, so that it
crushed passengers still in the vehicle?
Tires have also been blamed for
the crash of the supersonic Concorde.
The investigators have tentatively
concluded that, during a take-off run,
one tire or pair of tires picked up a
piece of metal that was on the runway,
and either spun it off like a projectile
into a fuel tank, or spun off pieces of
tire that perforated the fuel tank(s).
This seemed to be a horrible piece of
bad luck, until reports were aired that
tires on Concordes had failed in similar
fashion more than once previously. So
had engineers or managers just wished
the problem wasn't going to recur?
An approximately similar number of
people died as allegedly did from the
Firestone-Ford tire problems.
In this issue of Human Power we
report on a problem with bicycle tires:
a flip-flop behavior that can throw rid-
ers suddenly off their machines when
a front tire deflates. It appears to be
caused simply by poor fits of tires on
rims. If this is so, it could be solved
quickly by industry agreement on the
dimensions of rims and tires, spurred
possibly by government specifications.
We dont know how many lives have
been lost from this unnecessary series
of failures. Bicycle accidents are not
taken seriously enough to be investi-
gated in depth. There has been no out-
cry. Your editors letters to the U.S.
Consumer Product Safety Commission
and to several industry organizations
have remained unanswered.
Remedies for bicyclists have the
same status as so-called orphan
drugs. These drugs are not developed
for fatal but relatively rare diseases
because drug companies see insuf-
ficient profit. Is the bicycle-tire-rim
case a situation where industry is
not being sued enough? The much-
maligned product-liability lawyers can
correct serious deficiencies in industry
responses, or lack of responses, to
shoddy practice.
Dave Wilson
Human Power Number 51, Fall 2001 23
HUMAN POWER PUBLISHING
RECORD, 19952000
Human Power 11:4 (Fall/Winter
199495)
Human Power 12:1 (Spring 1995)
Human Power 12:2 (Fall 1995)
Human Power 12:3 (Winter/Summer
1996)
Human Power 12:4 (Spring 1997)
Human Power 13:1 (Fall 1997)
Human Power 13:2 (Spring 1998)
Volume 13:3 (Summer/Fall 1998)
In 1998 we moved to a simpler
numbering system for Human Power,
since we are not able to publish on a
regular, pre-defined schedule.
Adding up all the issues we could
find back to issue 1:1, we numbered
the next issue #46. After a long-time
member noticed that we had left out
a number, we re-numbered that winter
199899 issue #47.
Thus, for 1999 and 2000, we pub-
lished the following:
Human Power 48 (Summer 1999)
Human Power 49 (Winter 19992000)
Human Power 50 (Spring 2000)
Human Power 51 (this issue)
We expect to publish at least two
issues of Human Power in 2001 and
have already begun work on Human
Power 52.
22 Number 51, Fall 2001 Human Power
International Human
Powered Vehicle
Association
IHPVA
PO Box 1307
San Luis Obispo, CA 93406 USA
http://www.ihpva.org