Professional Documents
Culture Documents
1.1 INTRODUCTION
rest to a velocity of about 230 ft/s and a spin rate between 2000 and 3000
rpm. The ball deforms extensively during contact with the club, compressing
to about 80% of its diameter (Fig. 1.2), but regains its original shape within a
couple of diameters along its trajectory.
o ......
~-+-_ _-+--+-_ _ -+--+-_+--I-+--+-_+--I-+--+-_~-+--+-_+--
no 300
30
The second phase is the actual flight of the golf ball, which is about 250
to 300 yards long and about 35 yards high at its maximum point.
Aerodynamics plays a crucial role in this phase, and it will dictate where and
when the ball lands. The third and final phase of the motion begins when the
ball impacts the ground. It may bounce a number of times before rolling
some distance and finally coming to rest. Aerodynamics does not play a
significant role in this phase.
The study of golf ball aerodynamics may have a history as long as the
game itself (Thomson, 1910). The most significant event in golf ball
aerodynamics, however, took place soon after the introduction of the gutta-
percha ball in 1848, invented by Rev Adam of St. Andrews (before that time,
golf balls were made of leather covers stuffed with feathers, called the
"feathery" golfball). Gutta percha is a natural rubber, and the ball was made
by filling a mold with latex, which was then pressed and cured at high
temperature. Golfers discovered, probably by accident, that the ball flew
further and better when scored or marked. Chase (1981) credits this
discovery to a professor at St. Andrew's University, but it clearly became
common knowledge very rapidly, and it initiated a wave of innovation in
cover design (Fig. 1.3).
Chap. 1. Aerodynamics of the Golf Ball 5
Figure 1.2. Defonnation of a golf ball on impact with a driver (photo courtesy of Callaway
Golf Company).
Figure 1.3. Some early gutta percha balls showing different surface treatments. From left to
right: "Smooth gutty," circa 1850's; "Knife cut gutty," circa 1860 to 1870's; "Ocobo molded
gutty," circa 1890's; "Finch bramble gutty" circa 1890-1900. Images courtesy of Leo M.
Kelly, Jr.: http://www.oldgolf.com.
By the tum of the century, the solid gutta-percha ball had given way to
the wound ball, where rubber thread was wound under high tension around a
small rubber core, and the wound core was encased in a gutta-percha cover.
This was the Haskell ball, designed and patented by Coburn Haskell in 1898.
Balata (a thennoset plastic) then replaced gutta percha for the cover, since it
had a better adhesion to the rubber windings, and a better color and
toughness. It is the first rubber-cored ball. The modem golf ball is generally
of a "two-piece" design, first introduced by Spalding through the Executive
ball in 1972, where the rubber windings were replaced by a solid
polybutadiene core and the cover was made of blends of synthetic polymers
that gave greater consistency and better durability (Sullivan and Melvin,
6 Alexander Smits and Steven Ogg
Figure J.4. Callaway CTU 30 ball, featuring a solid polybutadiene core and a cast thermoset
urethane cover. Image courtesy of Callaway Golf Company.
Once the ball leaves the club, its motion is governed by the laws of
aerodynamics and the force of gravity. The trajectory of the golf ball is
independent of its construction, and the only important characteristics are its
geometrical shape, size, mass, and its moment of inertia. That is, the shape
and distribution of the dimples on a golf ball, together with its size and
inertial characteristics, completely determine its aerodynamic performance.
It is for this reason that there is currently such intense competition among
golf ball manufacturers to produce innovative dimple shapes and patterns.
Very small changes in dimple design can have important consequences for
the ball trajectory, particularly its "carry" distance (the distance between tee
Chap. 1. Aerodynamics of the Golf Ball 7
and impact with the ground), the maximum trajectory height, and the angle
of incidence at the point where it hits the ground. These quantities are
important to a golfer, and there is continuous innovation required by the
manufacturers to satisfy the demand from golfers for a better, longer, more
accurate ball.
The desire for increased distance comes squarely up against the Rules of
Golf, as published jointly by the United States Golf Association (USGA) and
the Royal and Ancient Golf Club of St. Andrews. In particular, the Overall
Distance Standard requires that the ball, when struck by a mechanical golfer
using a driver under specified conditions, not travel more than a total
distance (carry and roll combined) of 296.8 yards, which includes a +6%
"testing tolerance" that could be reduced to 4% as testing methodology
improves. The golf ball manufacturers, therefore, have invested heavily in
developing sophisticated tools and techniques for measuring and predicting
the performance of golf balls, both to design better balls, and, most
importantly, not to infringe the Rules.
Once the ball is launched, the only forces acting on the ball are those due
to gravity and the air acting on the ball. The aerodynamic force has a lift
component (due to the spin on the ball) and a drag component (due to the
friction between the air and the surface of the ball).
In principle, the forces acting on the ball can be found by solving the
equations governing the flow of fluids, known as the Navier-Stokes
equations, together with the appropriate boundary conditions. For most
practical flows, this represents a very challenging problem, and it is almost
always necessary to resort to a computer. However, for the flight of a golf
ball under realistic conditions, it is extremely difficult to obtain computer-
generated solutions, even when using the most powerful computers available
today. We resort, by necessity, to experimental work. Experiments are also
challenging, and it can be labor-intensive to obtain accurate and complete
results. Fortunately, we can use a powerful tool called dimensional analysis
to significantly reduce the amount of work involved. Dimensional analysis is
a common tool in aerodynamics, as well as in other fields where the
complexities of the physical problem do not allow analytical or
computational solutions. In particular, it allows the organization and
presentation of experimental data in a very efficient and elegant manner (see,
for example, Kline, 1965).
8 Alexander Smits and Steven Ogg
F = f(V,D,oo,p,ll,k) (1.1)
I
F
7:PV A
2
=
g
(PVD ooR
Il
!)
, V 'D (1.2)
(1.3)
and
CD =g2(Re, W,f;) (1.4)
where CL and CD are the lift and drag coefficients, respectively, defined by
Chap. 1. Aerodynamics of the Golf Ball 9
and (1.5)
The other non-dimensional groups are the Reynolds number Re, the spin rate
parameter W , and the relative roughness, E, are given by
Re=--,
pVD w = roR , E
k
=-. (1.6)
I.l V D
To understand the role played by the dimples and the spin rate, it is useful
to consider first the case of a non-spinning ball. Here, there is no lift, only
drag (we shall see why later), and Eq. 1.4 reduces to
(1.7)
That is, the drag coefficient is only a function of Reynolds number and
relative roughness. Figure 1.6 shows this behavior, obtained as the result of
many experiments on smooth and rough spheres, including a representative
golf ball (the range of Reynolds numbers experienced by a golf ball during a
typical driver shot is approximately 50,000 to 200,000, corresponding to
balls traveling at 60' to 240 ftls in air at 70or' and atmospheric pressure).
There are different curves depending on the value of the relative roughness,
but all curves show a similar behavior: at low Reynolds numbers the drag
coefficient is approximately constant at a value of about 0.5, while at higher
Reynolds numbers the drag coefficient suddenly reduces to a much lower
value of about 0.1 or 0.25, before slowly increasing.
The value of the Reynolds number where the drag coefficient takes a
sudden dip is called the "critical" Reynolds number. For a golf ball its value
lies somewhere between 40,000 and 60,000, whereas for a smooth ball it is
between 350,000 and 400,000 (Fig. 1.5). For a Reynolds number above the
critical value, a golf ball has about half the drag of a smooth ball. That is,
for velocities greater than about 70 ftls, a golf ball experiences less than half
the air resistance experienced by a smooth ball of the same size.
10 Alexander Smits and Steven Ogg
5
.6 kiD x 10
900 1250 500 150 rough spheres
\\ \\r--r --~-4k~- .-
.5 '~1-""\ (Achenbach,1974)
.4
CD
.3
·/·V· " smooth sphere
.J'Y'-'''-<''-~- I (Achenbach, 1972)
,.--
I
./
.2
v
/
. I
I
, .,"
_----
"'-- .,"
.1 \
0
2 2 4
Re
Figure 1.5. Drag coefficients as a function of Reynolds number for spheres with different
degrees of roughness. Adapted from Bearman and Harvey (1976). With permission of the
Royal Aeronautical Society
v_ 8\0
~
n
- -
---- 1-----1
Figure 1.6. Flow over the ball at Reynolds numbers below the critical value. A laminar
boundary layer is found on the upstream face of the ball, and the flow separates near the point
of maximum cross-section. A large wake region is formed.
Although the boundary layer is now turbulent, the flow just outside the
boundary layer is still very much the same as that seen at lower Reynolds
numbers: that is, the freestream velocity reaches a maximum value near the
crest, and then begins the decrease. Again, a pressure difference is
generated that leads to a force on the fluid in the boundary layer that acts in
the upstream direction. However, a turbulent boundary layer responds quite
differently to a laminar boundary layer when acted upon by this "adverse"
pressure gradient, primarily because of the turbulent mixing that occurs
inside the boundary layer. First, the turbulent mixing tends to even out the
velocity gradients in the boundary layer, so that the velocity profile is
"fuller" (see Fig. 1.7), and the flow enters the region of adverse pressure
gradient a with relatively high momentum near the wall. A turbulent
boundary layer, therefore, will be able to resist the decelerating force due to
pressure differences more effectively than a laminar one. Second, even as
the boundary layer enters the adverse pressure gradient region, turbulent
mixing continues to take place, helping to replenish the momentum loss near
the wall by bringing higher momentum fluid from the outer part of the
boundary layer closer to the wall. As a result of these two phenomena, both
related to turbulent mixing, a turbulent boundary layer is able penetrate
further into the adverse pressure gradient before separating. The width of
the wake will be reduced, and the total momentum deficit in the wake will
also be reduced, leading to a lower drag force.
We see that the change from laminar to turbulent flow in the boundary
layer on the front face of the ball has a profound effect on the drag force.
The transition to turbulence reduces the size of the wake, which results in a
considerably lower total drag force. Although the friction drag in a turbulent
Chap. 1. Aerodynamics of the Golf Ball 13
.- --- - -
-- I--
b
-:;;I -
J I
~ /
-:;:JII'" ~
Figure I. 7. Flow over the ball at Reynolds numbers above the critical value. The boundary
layer becomes turbulent on the upstream face of the ball, and the flow separates well
downstream of the point of maximum cross-section. A smaller wake region is fonned than in
the case with fully laminar flow.
same size and weight of a regular golf ball, will only drive it about 140
yards. Roughness, and dimple design, are critical factors in influencing golf
ball performance.
However, rougher is not necessarily better. Each individual roughness
element, that is, each dimple, makes a contribution to the overall drag, and
when the roughness becomes too large, the gains achieved by manipulating
the critical Reynolds number are offset somewhat by the drag due to the
roughness elements. This effect is seen in the slow rise of the drag
coefficient at Reynolds numbers beyond the critical value (see Fig. 1.5).
Therefore, the design of the dimple pattern, and the choice of dimple sizes
and shapes, is a balance between achieving the lowest critical Reynolds
number and the lowest form drag contribution at the same time.
One of the most significant differences among golf balls relates to their
performance near the end of the trajectory. An effective dimple pattern will
promote transition to turbulence over the entire Reynolds number range
experienced in flight, so that the flight envelope is always above the critical
Reynolds number. If the dimples are not effective at low Reynolds numbers,
transition to laminar flow may occur, and the drag on the ball will suddenly
rise near the end of its trajectory. It will literally seem to fall out of the sky,
reducing the total carry. In addition, because its velocity at impact will be
reduced, and its angle of impact with the ground is increased, the roll
distance will also be reduced.
Therefore, the dimple pattern should be designed to promote turbulence
at the lowest possible Reynolds number, for the range of spin rates
encountered in flight, without creating an undesirable amount of extra drag
due to roughness. For example, it may be possible to promote turbulence at
very low Reynolds numbers using highly exaggerated dimple shapes, but it
is likely that the total drag at higher Reynolds numbers would not be optimal
because of the relatively high skin friction drag due to the added roughness
(even if the skin friction drag is small compared to the form drag, it may still
make an important contribution to the total drag).
Non-dimensionalizing, we have
SRD=h.(W,Re,M), (1.9)
where the non-dimensional spin rate decay is given by SRD =0) {R/V)2 ,
r/
and M = R is the non-dimensional radius of gyration.
Typical spin rate decay measurements are given in Fig. 1.8. By using the
non-dimensional representation shown, the results for this particular ball
collapse onto a single curve for all Reynolds numbers, which is close to a
straight line, and suggests that at constant velocity the spin rate decreases
exponentially.
16 Alexander Smits and Steven Ogg
~-------------------- ....
o
.
o
o 0
.
o
.-
'0
0ij A
o e s A
<-N
'"
N
0
O~-tOA
~
'3 )( 0 0 .(?t"lf1~Jb·~A Reynolds No.
0= 96453.3
o
~. 0=124045.4
A = 154843.5
ci + = 186052.0
x = 214114.6
0=247927.5
~+---...,...---.,..----r--~,.....--.,...---t
I 0.00 0.06 0.10 0.16 0.20 0.26 0.30
wR/V
Figure 1.8. Spin rate decay for different Reynolds numbers. Reprinted from Smits and Smith
(1994), with permission ofE & FN Spon.
Consider the free body diagrams shown in Fig. 1.9a,b. Figure 1.9b
depicts the forces acting on the ball as it ascends. The lift force is defined to
act normal to the velocity vector. The drag force is defined to act in a
direction opposite the velocity vector. Note that there is a component of the
lift vector, Vh, that acts in a horizontal direction opposing the travel of the
ball. There is a component lift force opposing the gravitational force, Lv,
but there is also a horizontal component, Lh, that is resisting the down-range
travel of the ball. Also note that the drag force has a component that acts in
Chap. 1. Aerodynamics of the Golf Ball 17
a vertical direction opposing the vertical component of lift. The drag force is
not only causing the ball to decelerate, but is pulling the ball down! Given
that with the creation of additional lift comes increased drag, it becomes
clear that for ascending flight minimization of lift is critical.
This picture becomes quite different after the golf ball has passed the
apex and is in descending flight (Fig. 1.9a). Here the lift force is not only
opposing the force of gravity, via component Lv, but as indicated by the
horizontal component, Lh, is pulling the ball forward! The lift force is
therefore contributing to increased carry distance and, by reducing the
incoming angle to the ground, increased roll distance. But what about the
increase in drag associated with maximizing lift at this low speed condition?
While the drag force certainly has a negative decelerating impact on the ball,
observe that it now has a vertical component, Dv, that is opposing the force
of gravity and hence helping to keep the golf ball in the air.
~-+-l~ ...... Dh
Velocity
mg mg
Figure 1.9. Free body diagrams showing the forces acting on a golf ball in flight: (a)
Descending ball; (b) Ascending ball.
With reference to Fig. 1.9, for a ball in flight without crosswind, hook or
slice, we have:
where m is the mass of the ball, FD is the total drag force, FL is the total lift
force, and a. is the angle the ball trajectory makes with the horizontal
direction, x, and g is the acceleration due to gravity in the vertical direction
y. The notation xdenotes the second derivative ofx with respect to time t.
If we define a parameter K = pA/2m, we have:
18 Alexander Smits and Steven Ogg
Since .e + y = V 2
, and x= V cosa and y = V sina , we obtain
These equations can be solved numerically to find the ball trajectory and its
point of impact once the initial conditions are specified. The trajectories
shown in Fig. 1.1 were obtained in this manner. However, the equations
cannot be solved without accurate measurements of the lift and drag
characteristics of golf balls since the variation of the lift and drag
coefficients as functions of Reynolds number and spin rate is a necessary
input to the trajectory calculation.
detennined. The spin rate was varied up to 8000 rpm (a spin rate parameter
Wof 0.089), but only one tunnel speed V of 105 ftJs was used
(corresponding to a Reynolds number of about 94,000), whereas typical
driver shots have a maximum velocity of approximately 235 ftJs. Davies
also presented some trajectory calculations using the measured lift and drag
infonnation, but, as he stated: "Neither the manner in which L and D vary
with translational speed nor the time rate of change of rotational speed are
known."
The next major contribution was made by Beannan and Harvey (1976),
who suspended a large model of a golf ball in a wind tunnel using fine wire
supports. The ball could be spun in either direction using a motor mounted
inside the ball. The interference due to the support wires was detennined by
reference to previous results for spinning smooth spheres, and it was found
that when the ratio of the support wire diameter to the ball diameter was less
than about 0.005 the interference effects were negligible. Results were
obtained for Reynolds numbers between 1.26xl05 and 2.38xI0 5, and spin
rate parameters of 0.02 to OJ, and they demonstrated a marked difference
between hexagonal dimples and conventional dimples by trajectory
calculations. It should be pointed out that in these calculations the time rate
of change of rotational speed, that is, the spin-rate decay parameter SRD
was assumed to be zero.
More recently, Aoyama (1990) adapted Davies' method for finding lift
and drag from drop tests by recording the trajectory of the ball as it fell using
short exposure video techniques. This approach can give more accurate
results for lift and drag since there is no need for simplifying assumptions
regarding the trajectory shape, the equations of motion, or the relative
velocity differences between the ball and the airstream. Lift and drag
coefficients were presented for ball velocities of 100 to 250 ftJs, and spin
rates from 1000 to 3500 rpm. They showed that the lift and drag coefficients
for the balls using the 384 icosahedron dimple pattern were consistently
lower than the 336 Atti patterned balls (an octahedron pattern, split into
eight concentric straight-line rows, named after the main producer of molds
for golf balls).
Smits and Smith (1994) used wind tunnel testing to develop an
aerodynamic model of the golf ball. The balls were mounted in a drag
balance using thin metal spindles for support. The spindles allowed the ball
to rotate freely, and one side could be connected to a motor to spin the ball
to high rotation rates. When the desired rotation rate was reached, the
spindle was released from the driving motor and the motor was withdrawn
from the tunnel. As the spin on the ball decreased, lift and drag
measurements were taken simultaneously. The measurements were
corrected for the interference effects of the spindles, one of the problems in
using a drag balance. Spin rate decay measurements over the entire
Reynolds number range were taken in a separate series of tests. The results
20 Alexander Smits and Steven Ogg
were obtained using only one brand of golf ball, the Slazenger two-piece ball
with a Surlyn cover. This brand was chosen because previous tests had
indicated the balls are highly symmetrical, thereby avoiding many of the
mechanical resonance problems experienced at high spin rates with some
other brands.
Typical results for the lift coefficient as a function of spin rate are given
in Fig. 1.10. The data at spin rates less than about 0.04 is not reliable and it
should be discounted. Clearly, the Reynolds number dependence is weak, at
least for Reynolds numbers greater than approximately 90,000. Below this
value of 90,000, however, there is a decrease in lift coefficient, followed by
a sudden increase at a Reynolds number of about 60,000.
... r-..,----,-......,....--,-----,,--~-,-......,...--,---,
! I i j
-.--~_.-- .•~.--~ ••-.-J--.-~-L-.-.l-- . U ...-
..
.40
.11 l-ii'\----+--t
• i-·r-r--1---1--~--r--·t·--1
• • 1 i 1 j J i
o • .M " • • ~ • • • M
Spin Rate. wR/V Spin Rate. wR/V
Figure 1.10. Lift coefficient as a function of spin rate parameter for different Reynolds
numbers. Reprinted from Smits and Smith (1994), with permission ofE & FN Spon.
Typical results for the drag coefficient as a function of spin rate are
shown in Fig. 1.11. Three observations can be made. First, all the results
indicate a strong dependence on spin rate. Second, the low Reynolds
number results «50,000) show a minimum in the drag coefficient at a
certain non-dimensional spin rate, which corresponds to a particular rate of
rotation (about 43 revolutions/s). Third, for Reynolds numbers greater than
about 50,000, the drag coefficient increases with Reynolds number. Smits
and Smith (1994) noted that when the Reynolds number exceeds
approximately 200,000, there appeared to be a relatively sudden decrease in
drag coefficient, which they ascribed to compressibility effects. These
trends are more evident in Fig. 1.12, where the data of Fig. 1.11 are re-
plotted as a function of Reynolds number. This figure also demonstrates the
drag rise below a Reynolds number of about 60,000, the "critical" Reynolds
number. The evidence from Fig. 1.11 indicates that the critical Reynolds
number is a function of spin rate.
Chap.t. Aerodynamics of the Golf Ball 21
Figure 1.11. Drag coefficient as a function of spin rate parameter for different Reynolds
numbers. Reprinted from Smits and Smith (1994), with permission ofE & FN Spon.
..
..---,...- - " " ; - - - ' - - - " - - - ' - ;- - ,
__ -~._.._..-
... ._--.---.l·_··_··__ ·~····_·_·······t-_·_···_·-L
31 ·_··_········i···_.. -_····;·····_····..···1··..··--.1--_·_-"'!-"'---'"
,.. _.._ _-.;..._-_.__ ~_ __ ; ----4-.----.~._._ .
.;-~_I]ii~t~
o : ::::~t;t~t~d;;-~1=::
: =~=-.J====1=~~:~:I=~~~~==E=~~
! l i
-
~ j
.1& ··········.··!·---·-·--··1-··---··--1·-·----r·---·r-·-..-.
: :
I'" _ _
' :
...
Raynoldo No. Reynold. No.
Figure J. J2. Drag coefficient as a function of Reynolds numbers for different values of the
spin rate parameter. Reprinted from Smits and Smith (1994), with permission of E & FN
Spon.
Indoor testing seems to be the best approach for obtaining high quality
data. The ftrst successful Indoor Testing Range (ITR) was designed and
constructed at the USGA Research and Test Center in Far Hills, New Jersey
(Zagarola et al. 1994). The test range uses a calibrated launching machine to
provide precisely known initial velocity, launch angle and spin rate. The
velocity of the ball is then measured at three down range stations, along with
the vertical and horizontal position, all with a high degree of precision. This
facility can be used to obtain fundamental aerodynamic data and
22 Alexander Smits and Steven Ogg
"r------------------------,
."
0.'
o_~
,r -
Sj:lIn-rprr:
-----~~:
." -------200'"
01 / ------'00(0
O,O~
.... _'00"
,O+--_ --_--_--_-_--~--_-_----J
11000)) 13(00)
'''''''
Reynolds Number
11000(· ,"".
Figure 1.13. Lift coefficient as a function of Reynolds numbers and spin. Courtesy of
Callaway Golf Company.
O."'r---::-------------------------,
0,)13
o.:~
012
i O~
C 03
(.l
~."
Q
4000
-----_3000
0.2'::
0.' ""------~===========:::::~:~
O"I--_~--_--~--_-_--~--_-_--J
5((lOO 10000 l'OOOO IlOO(() 1:,(1,») 17{1(IQ) 'Ol'lOO ZU!OCO
Reynolds Number
Figure 1.14. Drag coefficients as a function of Reynolds numbers at 3000 rpm. From Ogg
Patent 6,290,615 (1999).
1.6 REFERENCES
Achenbach, E., 1972, Experiments on the flow past spheres at very high Reynolds numbers.
J. Fluid Meehan. 54: 565.
Aoyama, S., 1990, A modem method for the measurement of aerodynamic lift and drag on
golf balls. Science and Golf, A. J. Cochran, ed., E. & F. N. Spon, London, pp. 199-204.
Bearman, P. W., and Harvey, 1. K. 1976, Golfball aerodynamics, Aeronautical Quart. 27:
112-122.
Chikaraishi, T., Alaki, Y., Maehara, K., Shimosaka, H., and Fukazawa, F., 1990, A new
method on measurement of trajectories of a golf ball., Science and Golf, A. 1. Cochran,
ed., E. & F. N. Spon, London, pp. 193-198.
Davies, J. M., 1949, The aerodynamics of golf balls, J. Applied Physics. 20: 821-828.
Kline, S. 1., 1965, Similitude and Approximation Theory, McGraw-Hill, New York.
Maccoll, J. ,1928, Aerodynamics ofa spinning sphere. J. Royal Aeronautical Soc. 32: 777.
Martin, J., 1968, The Curious History ofthe GolfBa//, Horizon Press, New York.
Mehta, R., 1985, Aerodynamics of sports balls, Ann. Rev.Fluid Meehan. 17: 151-189.
Moin, P. and Kim, 1., 1997, Tackling turbulence with supercomputers, Sci. Amer. 276: 62-76.
Smits, A. 1., and Smith, D. R., 1994, A new aerodynamic model of a golf ball in flight, Proc.
Second World Scientific Congress ofGolf, E. & F. N. Spon, London, pp. 340-347.
Smits, A. 1., and Zagarola, M.V., 2000, Applications of Dense Gases to Model Testing for
Aeronautical and Hydrodynamic Applications, Proceedings International Workshop on
Dense Gas Dynamics, Institute for Advanced Physics, Boulder, CO.
Sullivan, M. 1. and Melvin, T., 1994, The relationship between golf ball construction and
pefonnance, Science and GolfII, A. J. Cochran and M. R. Farrally, eds., E. & F. N.
Spon, London, pp. 334-339.
Chap. 1. Aerodynamics of the Golf Ball 27
Thomas, F., 2002, The aerodynamics of golf, The Falconer, No. 22, pp. 2-5.
Thomson, J. J., 1910, The dynamics ofa golf ball, Nature, 85: 2151-2157.
Zagarola, M. V., Lieberman, B., and Smits, A. J., 1994, An indoor testing range to measure
the aerodynamic performance of golf balls, Proc. Second World Scientific Congress of
Golf, E. & F. N. Spon, London, pp. 348-354.