Archery and Mathematical Modelling 1
B.W. Kooi
1
Introduction
One way of studying ancient bows is to make replicas and use them for experiments. In
the present paper the emphasis is on a different approach, the use of mathematical models.
Such models permit theoretical experiments on computers to gain insight into the performance of different types of bow. The use of physical laws and measured quantities, such
as the specific mass of materials, in constitutive relations yields mathematical equations.
In many cases the complexity of the models obtained will not permit the derivation of the
solutions by paper and pencil operations. Computers can then be used to approximate the
solution. However, even this procedure will mostly necessitate simplifications. Sometimes
essential detailed information is missing. In other situations assumptions need to be made
to keep the model manageable. In that case the model has to be validated by the comparison of predicted results with actually measured quantities to justify the assumptions. For
that purpose, fortunately, replicas can be employed.
Mathematical models must accommodate all quantities which determine the action of
the bow. Such quantities are often called design parameters. Calculations are possible only
if all the parameters are known. Descriptions of bows in the literature are often incomplete,
so that comprehensive evaluation becomes impossible.
Theoretical experiments with models consists to a large extent of the research on the
influence of the design parameters on the performance of the bow. This presupposes
definition of good performance which fits the context of interest. Flight shooters are only
interested in a large initial velocity. For target archery, on the other hand, the bows has
to shoot smoothly and steadily.
In the 1930’s bows and arrows became the object of study by scientists and engineers;
see Hickman, Klopsteg and Nagler1, 2 . Their work influenced strongly the design and construction of the bow and arrow. Experiments were performed to determine the influence of
different parameters. Hickman made a very simple mathematical model for flatbows. Later
Schuster3 and Marlow4 also developed mathematical models to describe the mechanical action of a bow. Schuster dealt with the ballistic of the modern, so called working-recurve,
bow. Schuster’s model has the strange feature that bows appear to have 100% efficiency.
Marlow introduced an elastic string in the model in order to explain this discrepancy with
reality.
The description of our mathematical model is beyond the scope of this paper. The
reader is referred to papers5–7 . The developed model is much more advanced, so that more
detailed information is obtained. This gives a better understanding of the action of rather
general types of bow. Elsewhere8, 9 we have shown how this model can be adapted for the
description of the ballistics of a modern bow. The predicted efficiency is smaller than 100%
1
Journal of the Society of Archer-Antiquaries 34:21-29 (1991)
1
because in this model part of the available energy remains in the limbs and string and is
not transferred to the arrow.
This model was validated by a comparison of the measured initial velocity of an arrow
shot with a modern bow with the predicted value10 . As part of the Mary Rose project11
the measured weight of a replica was correlated with the predicted value. In both cases
the predictions were sufficiently good.
The aim of the present paper is to use the model for an evaluation of the performance of
bows used in the past and in our time. We try to uncover the function of the siyahs or ears
of the Asiatic composite bow and to find the reason for the differences in the performance
of the longbow and the Turkish bow in flight shooting.
2
Mathematical modelling
In essence the bow proper consists of two elastic limbs, often separated by a rigid middle
part, the grip. The bow is braced by fastening a string between both ends of the limbs.
After an arrow is set on the string the archer pulls the bow from braced situation into full
draw. This completes the static action in which potential energy is stored in the elastic
parts of the bow. After aiming, the arrow is loosed or released. The force in the string
accelerates the arrow and transfers part of the available energy as kinetic energy to the
arrow. Meanwhile the bow is held in its place and the archer feels a recoil force in the
bowhand. After the arrow has left the string the bow returns to the braced position because
of damping.
As stated before, a complete description of the mathematical model is beyond the
scope of this paper. An extensive discussion is presented elsewhere5–9 . A summary of all
important quantities in the model which determine the mechanical action of the bow is
listed below:
bow
string
arrow
length of the limbs
length of the grip
shape of the unstrung limbs
shape of cross-section of the limbs at all positions along the limbs
elastic properties of the materials of the limbs
specific mass of the materials of the limbs
shape and mass of the ears, if these are present
mass of the horns
fistmele
draw length
mass of the string
elastic properties of the string
mass of the arrow
These quantities, the design parameters, determine the weight of the bow. In practice the
bowyer tillers the bow such that it has finally the desired weight for a particular draw
2
length. The archer on the other hand sets the fistmele by the adjustment of the length of
the string.
For flight shooting the initial velocity of the arrow leaving the string is very important.
The larger this velocity the larger the maximum attainable distance. The actual distance
depends also on the elevation angle (nearly 45◦ ) and the drag of the arrow in the air. A
requirement for target shooting and hunting is that the bow shoots smoothly. It is difficult
to translate this feature into mathematics. High efficiency is a good criterion. However, a
heavy arrow always yields a high efficiency and, unfortunately so, a small initial velocity
and therefore a short distance. Hence, we have a combination of factors. The recoil-force,
i.e. the force, the archer feels in the bowhand after the release, also seems to be important.
The way this force changes in time can be calculated with the model, but it cannot be
summarized by a single number.
The bow should not exaggerate human error. To assess the sensitivity of the bow, its
performance is calculated repeatedly with slightly different values for the design parameters. If the performance depends strongly on a design parameter, the archer has to take
care that the value of this parameter is as constant as possible. To achieve this archers
need skill besides technique.
3
Validation of the mathematical model
Mathematical models may be beautiful by themselves and the way to solve them interesting, but they should mimic the mechanical action of the bow and arrow good enough if
they are used in the design of a bow or a sensitivity study.
We checked static action by comparing the measured weight of a replica of longbows
found on the recovered Mary Rose with calculated values. The Mary Rose was Henry
VIII ’s warship which sank in 1545 in The Solent, a mile outside Portsmouth. She was
recovered in 1982 with 139 yew longbows. Tests with these bows have demonstrated
that while it is possible to string and draw the bows to 30 inch, considerable degradation
within the cell structure of the wood has prevented a realistic assessment of the original
weight. A replica was made by Roy King, bowyer to the Mary Rose Trust. Prof. P.
Pratt, Imperial College of Science & Technology London, measured all parameters which
are required to calculate the mechanical performance of a bow. The weight of this replica
was also measured. It compared very well with the predicted value calculated with the
mathematical model (differences within 1%)11 . These results imply that if a good estimate
of the original modulus and density can be obtained, the original mechanical performance
of the longbows can be calculated from the dimensions of these recovered bows.
Data obtained with the test set-up described extensively elsewhere10 , permitted a comparison of predicted and measured arrow velocities. The dynamic action of bows could be
checked in this way. We used a modern bow made of maple in the core and glass fibres
embedded in strong synthetic resin at both sides of the core. All the essential parameters
listed above were measured. We measured the density and elastic modulus of both the
wood and the fibreglass and at a number of stations along the limbs the shape of the cross3
a)
b)
Figure 1: Non-recurve bows: Static deformation shapes (a) of the KL-bow and (b) of the ANbow.
sections. The results were used to determine the bending properties of the limbs. Finally
the elastic modulus and the mass of the string were measured.
The predicted weight was too high and therefore a knockdown factor was used for the
bending stiffness of the limbs, so that the calculated weight became equal to the measured
value. The predicted amount of energy stored in the bow by drawing it from the braced
situation to full draw, differed only slightly from the measured value. The measured
efficiency was a few percent below the calculated value. In the model internal and external
damping are neglected. This explains part of the discrepancy.
4
Classification of the bow
The classification of the bows we use is based on the geometrical shape and the elastic
properties of the limbs. The bows of which the upper half is depicted in Figure 1 are called
non-recurve bows. In the model the bow is assumed to be symmetric with respect to the
line of aim. So we need to deal with only one half of the bow. These bows have contact
with the string only at their tips. In the case of the static-recurve bow, see Figure 2, the
outermost parts of the recurved limbs (the ears) are stiff. In the braced situation the string
rests on string-bridges. These string-bridges are fitted to prevent the string from slipping
past the limbs. When such a bow is drawn, at some moment the string leaves the bridges
and has contact with the limbs only at the tips. In a working-recurve bow the limbs are
also curved in the ’opposite’ direction in the unstrung situation, see Figure 3. The parts
of a working-recurve bow near the tips, however, are elastic and bend during the final
part of the draw. When one draws such a bow, the length of contact between string and
limb decreases gradually until the point where the string leaves the limb coincides with the
4
a)
b)
Figure 2: Static-recurve bows: Static deformation shapes (a) of the PE-bow and (b) of the
TU-bow.
a)
b)
Figure 3: Working-recurve bows: Static deformation shapes (a) of the ER-bow and (b) of the
WR-bow.
5
Table 1: Quality coefficients for various bows. Note that the values for the working recurve
WR-bow are adapted for the masses of the arrow and string but the stiffness of its string
is about twice those of the other bows.
Bow
q
η
ν
KL-bow
AN-bow
PE-bow
TU-bow
ER-bow
WR-bow
0.407
0.395
0.432
0.491
0.810
0.434
0.765
0.716
0.668
0.619
0.417
0.770
2.01
1.92
1.94
1.99
2.08
2.09
tip. The string remains in that position during the final part of the draw. Elsewhere5 we
dealt with the statics (before arrow release) of these three types of bow. We studied the
dynamics (after arrow release) of the non-recurve bow6 , the dynamics of the static-recurve
bow7 and finally that of the working-recurve bow8 .
In the model the action of a bow and arrow combination is fixed by one point in a
high dimensional parameter space. Representations of different types of bow used in the
past and in our time form clusters in this parameter space. We study the performance of
different types of bow and start with a straight-end bow described by Klopsteg1 . This bow
is referred to as the KL-bow. The shape of the KL-bow for various draw lengths is shown
in Figure 1(a). The AN-bow represents another non-recurve bow, the Angular bow found
in Egypt and Assyria. The shape of the unstrung bow, shown in Figure 1(b), implies that
in the braced situation the limbs and the string form the characteristic triangular shape.
We consider two static-recurve bows, one from China, India and Persia, to be called the
PE-bow, and one which resembles a Turkish flight bow, to be called the TU-bow. The
shapes of these bows for various draw-lengths are shown in Figure 2. One of the workingrecurve bows, to be called the ER-bow, possesses an excessive recurve. It resembles a bow
described and shot by Hickman1 . The other working-recurve bow is the modern one which
was used for the validation of the model10 . This bow shown in Figure 3(b), is referred to
as the WR-bow.
Three quality coefficients for these types of bow are shown in Table 1. These coefficients
are defined for equal weight, draw length and mass of the limbs. Moreover the mass of the
arrows and strings were the same for all reported bows. This makes an honest comparison
possible. Unfortunately the stiffness of the string of the WR-bow is about twice that of
the other bows. The static quality coefficient q measures how much recoverable energy is
stored in the fully drawn bow. It is defined as the additional deformation energy stored
6
Figure 4: Static (F ) and dynamic (E) forcedraw curves for the static-recurve PE-bow.
in the elastic limbs and string by drawing the bow from the braced into the fully drawn
position divided by the weight timed the draw length. The efficiency η is the kinetic energy
transferred to the arrow divided by the just mentioned additional deformation energy. So,
it is the part of the available amount of energy which is transferred to the arrow as useful
energy. The third quality coefficient ν is proportional to the initial velocity. The constant
depends only on the weight, draw length and mass of the limbs.
The static quality coefficient is 1 when the draw-force is uniformly equal to the weight
for all draw lengths for a fictitious bow with no fistmele. Just as the efficiency, this
coefficient gives the actual value relative to a basic, characteristic value. The results show
that in practice q is slightly smaller that 0.5 except for the ER-bow with the extreme
recurve.
In what follows we will comment on a statement made by Hamilton12 :
”The function of the ”ears”, or siyahs, is well known today and no one can
question the superiority of the type of bow which still holds the world record
of shooting an arrow 972 yards (Klopsteg2 )”.
Hamilton continues:
”The siyah contributes in three ways to improve cast in the arrow”.
”(1) It provides leverage for the bowstring so the bow can be designed to
approach maximum weight earlier in the draw allowing more energy to be
stored for the cast”.
This statement is in agreement with our results. The static quality coefficient of the PEbow is larger then of the straight-end KL-bow.
In Figure 4 the static and dynamic force draw curve are shown for the PE-bow. The
line indicated with F shows a bend at the place where the string leaves the bridges. The
TU-bow stores even more energy in the fully drawn position, obviously because of the
recurve of the working part of the limbs. So the good static performance of flight bows
may result only partly from the use of the stiff ears.
7
”(2) Upon release, the bowstring imparts its energy to the arrow more uniformly
and at a higher and more sustained rate of thrust than in a ordinary bow
without siyahs.”
This statement is not supported by the results obtained with the model. Because of
the relatively heavy ears, there is a sudden decrease in the force in the string and, by
implication, in the acceleration force upon the arrow. The result of this is oscillatory
behaviour as shown in Figure 4. Consequentially the efficiency of static-recurve bows a
rather low. The amplitude of the oscillations depends largely on the modulus of elasticity
of the string and the mass of the arrow relative to the mass of the ears.
”(3) When the bow string reaches the bridges it is in effect shortened, increasing
the tension again on the bowstring and giving the arrow a final snap as it leaves
the bow.”
The dynamic force draw curve (E in Figure 4) shows that the acceleration of the arrow is
rather large when the string has contact with the bridges.
Notwithstanding this, the efficiency η of the PE-bow and certainly that of the TUbow, is rather low. This implies that the initial velocity ν is not as large as one would
expect on the basis of the static performance. This is caused by the relative heavy ears.
These considerations demonstrate why these bows can, inherently, not be better than long
straight-end bows. A large part of the available energy remains in the vibrating limbs and
string after arrow exit.
This holds even to a larger extent for the ER-bow. The string cannot slow down the
now light ends of the limbs during the final part of the acceleration of the arrow when the
bow is close to its braced situation again.
The modern WR-bow seems to be a good compromise between the non-recurve bow
and the static-recurve bow. The recurve yields a good static quality coefficient and the
light tips of the limbs give a reasonable efficiency.
5
Construction of the bow
But what made the Turkish flight bow a superb type of bow for flight shooting? Until now
we dealt with the mechanics of the bow but not with its construction. The efficiency is
greatly affected by the relative mass of the arrow relative to that of the limbs. For a fixed
mass of the arrow, the lighter the limbs the better the efficiency. This is the item where
technology becomes important. The minimum mass of the limbs for a fixed weight and
draw is determined largely by the appropriateness of the material to store energy.
In the past man used bows which differ not only in shape but also in the materials
applied. Simple bows made out of one piece of wood, straight and tapering towards the
ends have been used by primitives in Africa, South America and Melanesia. In the famous
English longbow the different properties of sapwood and heartwood were deliberately put
to use. Eskimoes used wood together with cords plaited of animal sinews and lashed to the
wooden core at various points. The Angular bow found in Egypt and Assyria are examples
8
Table 2: Mechanical properties and the energy per unit of mass referred to as δ bv for some
materials used in making bows13 .
material
working stress elastic modulus
kgf/cm2 × 102 kgf/cm2 × 105
specific mass
δ bv
3
−6
kg/cm × 10
kgf cm/kg
steel
sinew
horn
yew
maple
glassfibre
70.0
7.0
9.0
12.0
10.8
78.5
7800
1100
1200
600
700
1830
21.0
0.09
0.22
1.0
1.2
3.9
1300
25000
15000
11000
7000
43000
of composite bows. In these bows more than one material was used. In Asia the bow
consisted of wood, sinew and horn; ”Just as man is made of four component parts (bone,
flesh, arteries and blood) so is the bow made of four component parts. The wood in the
bow corresponds to the skeleton in man, the horn to the flesh, the sinew to the arteries,
and the glue to the blood.” These bows were used by the Mongolian races of Eastern Asia.
They reached their highest development in India, in Persia and in Turkey. In modern bows
maple and glass or carbon fibres, embedded in strong synthetic resin are used.
In Table 2 indications of the mechanical properties are given for some materials used
in making bows. From this table we conclude that it is possible to store much more energy
per unit of mass in the materials of the composite bow, sinew and horn, than in wood, the
material of the old simple bow. Moreover, in the composite bows not only better materials
were used, but they were also used in a more profitable manner. Sinew is very strong in
tension. It is therefore used on the back side. Horn withstands compression very good. It
is applied on the belly side of the limbs. Hence, a composite bow with the same mass as
a simple wooden bow can have a much larger weight. This explains the good performance
of the composite flight bow in flight shooting.
In Table 3 we give values for the weight, draw, mass and length, for a number of
bows described in the literature. The longbow is the replica of the Mary Rose bow. The
calculated weight of this bow called MRA1 was 102.4 lbs. If the same values are used for
the material properties of the Mary Rose bow called A812, the estimated weight becomes
108 lbs. W.F. Paterson14, 15 also investigated this bow. Dr. Clarke calculated that the draw
weight would be about 76 1/2 lb, depending on the modulus of elasticity of the yew15 . The
late Paterson in a letter to the author informs that Dr. Clarke: ”admits an error by the
factor of two in his calculations. His estimate should now read 153 lb”. Unfortunately no
value for the elastic modulus is mentioned14, 15 , but the final discrepancy is probably caused
9
Table 3: Parameters for a number of bows and an estimation of dbv the weight times draw
length divided by the mass of one limb. For comparison also the estimated values for δ bv
are given. When materials are used to their full extend, dbv divided by about 4 equals δ bv .
Ref.
Type
weight
kgf
draw
cm
mass
kg
length
cm
δ bv
kgf cm/kg
dbv
kgf cm/kg
1
flatbow
longbow
steelbow
Tartar
Turkish
modern
15.5
46.5
17.2
46.0
69.0
12.6
71.12
74.6
71.12
73.66
71.12
71.12
0.325
0.794
0.709
1.47
0.35
0.29
182.9
187.4
168.9
188.0
114.0
170.3
9000
9000
1300
20000
20000
30000
6800
8700
3500
4600
28000
6200
11
16
17
18
10
by a difference in this mechanical property. We decided to use a value of 0.75 105 kgf/cm2 .
Parenthetically, the spread in the modulus of elasticity of yew yields makes the predictions
of the weight (almost proportional to the modulus of elasticity of yew ) of the Mary Rose
bows uncertain.
The quantity denoted by dbv is proportional to the amount of energy stored in the bow
per unit of mass. It equals the weight times draw length divided by the mass of one limb.
When materials are used to their full extend dbv divided by about 4 should equal δ bv .
We saw that because of the stiff ears or a recurve of the working parts of the limbs,
much energy is stored in the static-recurve bow. In a recurved bow the amount of energy in
the braced position is already large. This implies that the limbs must be relative heavy in
order to store this extra and not usable energy, in addition to the recoverable energy. This
is the price paid for a larger static quality coefficient. On the other hand, sinew and horn
are relative tough and flexible materials, see Table 2. This explains why the use of these
materials fits well with the recurved shape of the unstrung bow. The values in Table 3
show that the Turkish bow is very strong but also light. This indicates why it permits one
to shoot a light arrow a long distance. A short bow is moreover easier in operation and is
suited for the use on horse back. In a letter19 to the author E. McEwen informs that: ”Pope
did not properly test his larger ‘Tartar’ (actually Manchu-Chinese) bow. He only drew it
36 inches and bows of this type and size are made to draw as much as 40 inches”. Pope
only mentions the weight for a draw length of 29 inches. If the weight of this bow with a
draw length of 101.6 cm is 70 kgf, we have dbv = 9700 kgf cm/kg. This value is still rather
low and this means that the materials of this bow are used only partly. This supports
McEwens19 view that: ”this bow was probably a ‘test’ bow used for exercise and for the
military examinations and not meant for actual shooting”. The values obtained for the
10
straight-end bows look very realistic. In the modern bow there is a surplus of material near
the riser section. This affects the efficiency only slightly. For, this part of the limb moves
hardly and therefore the involved kinetic energy is small. In this bow there is also a rather
large amount of unrecoverable energy in the braced position. This puts a constraint on the
amount of recurve. With respect to this, it is perhaps more important that the efficiency
of working-recurve bows decreases with increasing recurve.The mechanical properties of
the materials of these bows, however, are much better than those of the ancient composite
bows. Indeed, the modern bow holds now the longest flight shooting record.
Additional features were added to improve the performance especially for target shooting; relative immunity of the mechanical properties to temperature and humidity variations,
no tendency to follow the string, use of stabilizers, sculptured long centre-shot riser section,
bow sights and last but not least stronger materials for bowstrings. Finally an improved
arrow design adds to the steadiness of the equipment.
6
Conclusions
We conclude that these results indicate that the initial velocity is about the same for all
types of bow under similar conditions. So, within certain limits, the design parameters
which determine the mechanical action of a bow arrow combination appear to be less
important than is often claimed. We would endorse a view one could call holistic. It is not
always possible to isolate a single feature and state that it solely accounts for a good or bad
performance of the whole bow, as Hamilton12 did. Rausing20 studies the development of the
composite bow. According to him, the fact the static quality coefficient of the short staticrecurve bow to be larger than that of the short straight bow, disposes of the statement of
Pitt-Rivers, Balfour and Clark: ”the composite bow has no inherent superiority over the
wooden self-bow, so long as the latter was made from the most favourable kinds of timber
and expertly used”. The results obtained with the mathematical model suggest that, if the
word inherent has the meaning within the context we used it in Section 4, their statement
is true. A combination of many technical factors made the composite flight bow better for
flight shooting.
The quality coefficients of the modern bow are only slightly better than those of the
other types of bow. Materials used in modern working-recurve bows can store more deformation energy per unit of mass than the materials used in the past. Moreover the mechanical properties of these materials are more durable and much less sensitive to changing
weather conditions. This contributes most to the improvement of the modern bow.
We hope that we have shown that mathematical modelling can be a helpful tool in the
research on archery, not only for the design of new bow equipment but also for understanding the development of the bow in the past.
11
References
1
C. N. Hickman, F. Nagler, and P. E. Klopsteg. Archery: the technical side. National Field
Archery Association, Redlands (Ca), 1947.
2
P.E. Klopsteg. Turkish archery and the composite bow. Simon Archery Foundation, Manchester, 1987.
3
B. G. Schuster. Ballistics of the modern-working recurve bow and arrow. Am. J. Phys.,
37:364–373, 1969.
4
W. C. Marlow. Bow and arrow dynamics. Am. J. Phys., 49:320–333, 1981.
5
B. W. Kooi. On the Mechanics of the Bow and Arrow. PhD thesis, Rijksuniversiteit Groningen,
1983.
6
B. W. Kooi and J. A. Sparenberg. On the static deformation of a bow. Journal of Engineering
Mathematics, 14:27–45, 1980.
7
B. W. Kooi. On the mechanics of the bow and arrow. Journal of Engineering Mathematics,
15:119–145, 1981.
8
B. W. Kooi. The ’cut and try’ method in the design of the bow. In H. A. Eschenauer,
C. Mattheck, and N. Olhoff, editors, Engineering Optimization in Design Processes, volume 63
of Lecture Notes in Engineering, pages 283–292, Berlin, 1991. Springer-Verlag.
9
B. W. Kooi. On the mechanics of the modern working-recurve bow. Computational Mechanics,
8:291–304, 1991.
10
C. Tuijn and B. W. Kooi. The measurement of arrow velocities in the students’ laboratory.
European Journal of Physics, 13:127–134, 1992.
11
R. Hardy. Longbow. Mary Rose Trust, Portsmouth, 1986. pp. 4-6.
12
T. M. Hamilton. Native American Bows. Missouri Archaeological Society, Columbia (Ms),
1982.
13
J. E. Gordon. Structures or Why Things Don’t Fall Down. Pelican, Harmondsworth, 1978.
14
W. F. Paterson. ”Mary Rose” - a preliminary report. Journal of the Society of ArcherAntiquaries, 23:29–34, 1980.
15
W. F. Paterson. ”Mary Rose” - a second report. Journal of the Society of Archer-Antiquaries,
24:4–6, 1981.
16
R. P. Elmer. Target Archery. Hutchinson’s Library of sports and pastimes, London, 1952.
17
S. T. Pope. Bows and Arrows. University of California Press, Berkeley (CA), 1974.
18
Sir R. Payne-Gallwey. The Crossbow. Holland Press, London, 1976.
19
E. McEwen. Private communications, 1991.
12
20
G. Rausing. The Bow. Acta Archaeologica Lundensia. CWK Gleerups Förlag, Lund, Sweden,
1967.
13