Sie sind auf Seite 1von 13

On the Use of Conformal Mapping in Shaping Wing Profiles

Author(s): R. S. Burington
Source: The American Mathematical Monthly, Vol. 47, No. 6 (Jun. - Jul., 1940), pp. 362-373
Published by: Taylor & Francis, Ltd. on behalf of the Mathematical Association of America
Stable URL: https://www.jstor.org/stable/2303633
Accessed: 06-03-2019 01:32 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Mathematical Association of America, Taylor & Francis, Ltd. are collaborating with
JSTOR to digitize, preserve and extend access to The American Mathematical Monthly

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
362 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

ON THE USE OF CONFORMAL MAPPING IN SHAPING


WING PROFILES*

R. S. BURINGTON, Case School of Applied Science

1. Introduction. The use of conformal mapping in fluid mechanics dates back


to the early work of Gauss, Riemann, Weierstrass, C. Neumann, H. A. Schwarz,
and Hilbert. The earliest comprehensive application of conformal mapping to
aerodynamics may be attributed to Lord Rayleigh (1877) [1]. In the early
of the present century Kutta (1902) [2] and Joukowski (1906) [3] published a
series of fundamental papers on airfoil theory which marked the beginning of a
new era in fluid mechanics. Within a very short time many improvements of
the theories of Kutta and Joukowski were advanced, the recent works of von
Kairmain, Tietjens, Prandtl, Mises, Trefftz, Hiihndorf, Glauert and many others
being of importance [4, 5]. As the limitations of these theoretical methods be-
came more fully appreciated, interest in the mathematical shaping of airfoils
waned. However, the recent improvement in the efficiency of airplanes and the
immense cost of designing and constructing the modern airliner has renewed
interest in the theoretical approach to aerodynamics and in particular to the
question of shaping wing profiles mathematically.
It is the purpose of the present paper to give a short elementary exposition
of the use of conformal mapping in ideal two-dimensional airfoil theories.

2. Basic principles of wing theory [4, 5, 6]. The basic principle underlying
the construction of all flying machines is the property of a body (such as an in-
clined plane), when moving horizontally through the air, of experiencing a force
R which may be decomposed into two components, one called the drag D in
the direction of the airflow and the other called the lift L in a direction perpen-
dicular to the flow. (Fig. 1.) The angle between the direction of the flow and the

FIG. 1

force R exerted on the body is a function of the geometrical shape and position
of the body with respect to its motion. The lift is essential for carrying the air-
plane but the drag is largely undesired and must be compensated for by the
thrust of the propeller.
The drag is largely due to the friction of the air along the airfoil and to the

* Extracted from an illustrated lecture given to the Ohio Section of the Mathematical Associa-
tion of America, April 8, 1939.

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 363

appearance of the lift. The lift is accompanied by the creation of a definite


flow pattern in the neighborhood and in the wake of the airfoil and demands a
continuous supply of energy. Study of the drag due to friction requires a con-
sideration of the viscosity of the fluid, while the study of the drag due to lift
may be made through the application of fluid mechanics to ideal fluids.
An ordinary type of airfoil can be described as having the form of a flat or
slightly cambered body usually symmetrical with respect to a medium plane.
The section made by the medium plane is called the profile of the airfoil.
The force experienced by the airfoil is due mainly to the difference in the
pressures acting upon the lower and upper surfaces. For steady motion of a
fluid of density p along a body at rest, the pressure p at any point of absolute
velocity U can be calculated from Bernoulli's equation, which in its simplest
form is

p + 2pU2 = Po + 2pv2,

where po is the pressure and V is the absolute velocity at "infinity."


For certain purposes the airfoil is taken to extend to infinity on both sides
in a direction normal to the airfoil. This assumption simplifies the problem since
the effects of wing tips need not be considered, reducing the flow to the two-
dimensional case.
The fundamental problem of wing theory is that of determining the flow
around a geometrically determined wing or wing system. The distribution of
velocities and pressures must be calculated, assuming that a body having the
shape of the wing is moving uniformly and with constant velocity through the
fluid. For the investigation of the stability of airplanes and for various other
problems it is necessary to know the moment of the force experienced by the air-
foil. A general account of the mathematical theory must furnish formulas for
calculating the lifting force and the energy to be expended in producing the flow
system.

3. Two-dimensional ideal fluid mechanics. To avoid certain serious difficul-


ties and for reasons of simplicity an ideal fluid is often assumed in the study of
fluid mechanics. An ideal fluid is one in which the viscosity is zero and the den-
sity is constant; the fluid is assumed to be continuous, homogeneous, perfectly
mobile and free of shearing movement; and such changes in pressure as may oc-
cur are assumed to be such as not to sensibly effect the density.
We consider an ideal fluid of unit depth flowing over the c-plane in such a
manner that the vector velocity U of the liquid is everywhere horizontal and
independent of the depth of the point at which it is measured. (Fig. 2.) Let C be
an arbitrary curve in the A-plane passing through the points A (a, b) and B(c, d),
and let S be the cylindrical surface through (j with elements perpendicular to
the c-plane. Let V, denote the component of U along C at the point P(S, q),
and let the S and q components of U at (S, 7) be u and v, respectively. The vol-
ume of fluid flowing per unit of time (flux H) across the arc AB of C through S is

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
364 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

(1) H - vdS + ud-.

Now H is independent of C if and only if [81

d(-) da"
(2) ()

In case (2) holds, the integral


r (zy)
(3) v6t 7- dS + udq
a, b)

defines a function i(S, ) known as the stream function. If (2) holds, the fluid is
incompressible.

L yB(c id)
V'

FIG. 2

The circulation along C from A to B is defined to be

(4) X V'(1s = X xdf + vd-q.

In case (4) is independent of C,


an ad
(5)

and conversely, and the motion of the fluid is then said to be irrotational. When
(5) holds the integral
r (xY)

(6) +()r)J udE + vd?7


(a,b)

defines a function 44t, ), known as the velocity potential.

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 365

The complex function

(7) F + i J

is known as the generalized potential function. Since aoq/d9= / =u, d0/O'q=


-Ot/'/Oi=v, F is an analytic function of the complex variable =+irq. A pla
flow which may be characterized by such a function F is called a potential flow.
Both $ and i are solutions of Laplace's equation, that is, 72= 0, 72, =t0.
The velocity components may be obtained from dF/dV, since

dF ad a4,
(8) d= - + i- =X- iv.

The curves of the family =c1, a constant, are called stream lines, and the
curves of the family ?$ = c2, a constant, are known as potential lines. These fami-
lies are mutually orthogonal.
The simple types of generalized potentials are well known. For a fluid flow-
ing with a uniform velocity Ue+ia, U and a being real, the potential is
FRz=R+iV/R= Ue-ia; for a simple source of strength m (real) at Do,
tial is F8 = (m/2r) log (D - to); for a vortex of circulation F (real) and center at
to, the potential outside a circular barrier of radius a and center to is Fv=
-(iP/2wx) log (D-Po)/a; for a doublet whose axis is the line of angle a with the
i-axis and of moment M at 0o, the potential is FD -Me i/( - 0).
If F1, , Fn are n generalized potentials then any linear combination of
these is also a generalized potential, hence by superposition of simple potential
flows, flow fields of various degrees of complexity can be constructed.

0w t~
go

FIG. 3

4. Flow about a circular barrier. A case of particular interest arises from a


superposition of a doublet and a vortex, of common center t0, upon a rectilinear
flow. A potential F,= +4'iQ for such a flow is
Me ia jJ (-
(9) FC= FR-FD-FV = Ue-ia(-) + +-log 1
v-P0 2wr a

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
366 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

We require that the flow across the boundary of the circle v-0o=aeil be zero.
This means that for v on this circle, ik = O. From (9), we then find that M= Ua2.
The velocity function for this flow is

dF Ua2eia ir 1
(10) - = Ue-ia ? = U-iV, .

The velocity at infinity in this case is Ueia. Figure 3 illustrates this case. Note
the barrier circle and barrier lines emanating from the stagnation points P and
p.
From Bernoulli's theorem and the relation

(11) R = pt + ipx, = -J W?d,

the total resultant force R due to pressure on the circular boundary =re il can
be calculated. When a =O, we find

(12) pN= 0, p1= F(2 +-2)


If a = r, p, = pP U. The circle of radius a may represe
body of circular cross-section placed in the stream. We conclude that a circular
cylinder of uniform cross-section and infinite length when placed in a field of f
composed of a translation, a vortex motion, and a doublet, will experience per unit
length a total force (lift) of measure pP U at right angles to the stream.
This result is known as the Kutta-Joukowski theorem.
It has been shown that this theorem remains true when the circular cylinder
is replaced by a cylinder whose cross-section is uniform and simply connected.
These results are not quite in accord with the facts for actual fluids, largely
due to the assumption of zero viscosity. If the circulation F were zero, the cylin-
der would experience a turning moment but would have no lift.

5. Potential functions for an arbitrary wing shape. A number of methods


have been used for determining the potential functions for a given wing shape.
For the two-dimnensional case these methods may be roughly divided as follows:
(1) superposition of elementary potentials, (2) solution of Laplace's equation,
(3) conformal mapping, (4) other methods.
The first method has been illustrated in ?4 for the flow past a circular cylin-
der and consists in a search for a superposition of elementary potentials whose
resulting streamline pattern imitates the flow past a wing shape.
The second method is usually very difficult and depends upon finding sb and /
as solutions of Laplace's equation, meeting the required boundary conditions.
The third method involves the use of conformal mapping [7, 8, 13] and is
discussed in the following sections.

6. Principle of application of mapping to wing theory [5, 4]. Independent

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 367

of the earlier work of Lord Rayleigh, Kutta (1902) calculated the stream lines
around an airfoil using conformal mapping. This method has proved to be of
great value, though by the nature of the mathematics used, the method is largely
restricted to two-dimensional cases.
The theory of the two-dimensional flow of an incompressible ideal fluid
around wing sections is based on the following principles.
We suppose that the potential function for a plane flow about a profile G of
boundary G in the A-plane (Fig. 4) is known and is

(13) F(v) = 4(t, 7) + ii/(j, n),


where on G, 41 = , since the flow across G must be zero.

AA

FIG. 4

The procedure consists in mapping by means of a suitable analytic function


the profile Q (and the corresponding flow picture) into a profile CP of boundary P
resembling airfoils used in aeronautic practice.
We subject the points in the c-plane to a transformation

(14) t=-t(z) , (z = x+iy),


which is analytic almost eve
of G there corresponds a definite point z of the new profile boundary P (and
that this correspondence is biunique); (b) dt/dz is finite and not zero over 1,
the whole of the A-plane outside of G, and so that I will be represented by the
domain S of the z-plane outside of the airfoil ?; (c) t(z) is analytic at infinity,
and that as z- > oo, - cc, and dt/dz->K, a real, finite, non-zero constant;
(d) everywhere outside of G, (14) has a single-valued inverse z=f(v).
The new potential function F(z) for the flow about cP may be obtained from
the potential F(D) and transformation (14), and is given by

(15) F(z) = F(t(z)) = 1(x, y) + i4(x, y).

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
368 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

Since for each v on G, V(t, )=0, and F(P) is real, it follows from (15) and
(a) that T'(x, y) = 0, for all points on P, so that the flow across P is zero.
The velocity function for the flow in the c-plane is dF/d=u-iv=w?, and
for the new flow in the z-plane is

dF dF dD dt
(16) = =Wr = lz ivz Wz.
dz dA dz dz

The restriction (c) is made in order that the velocity wz of the fluid at in-
finity in the z-plane be a real constant K times the fluid velocity wr at infinity in
the c-plane.
Since F(z) = F(z) with v =t(z), it follows that for each z and corresponding ,
q(t, q) =4(x, y) and 41(S, q) =TF(x, y). Hence, to each equipotential line q=c
in the c-plane (ci a real constant) there corresponds in the z-plane the equipoten-
tial line 1? = c1; and to each stream line 41 = c2 in the c-plane (C2 a real constant)
there corresponds in the z-plane the stream line T = C2.
The restriction in (b) that dt/dz shall be analytic, finite and non-zero over I
is made in order that the flow pattern in the region z formed by the families
0 =c1, 4 = C2 shall be mapped conformally (with preservation of angles) int
flow pattern of the z-plane formed by the families I = cl and T = C2.
Since b= c1, and 1'= C2 are orthogonal, so are 1 = c1 and T= C2.
The points in the z-plane where dF/dz = 0 are called the critical points of the
z-plane; at such points the velocity wz vanishes. From (16), we see that if dt/dz
is finite and non-zero at a critical point A' in the z-plane, then dF/d= 0, and
the point A in the c-plane corresponding to A' is a critical point of flow in the
c-plane.
If, however, C is a critical point in the c-plane and at C, dz/dt is zero, then
the right-hand side of (16) is indeterminate. Let C' be the point in the z-plane
corresponding to C. We require that the velocity dF/dz of flow at C' be finite
and non-zero. (The Joukowski condition.) Evidently, C' is not a critical point
in the z-plane.
Suppose that T and T', and A and A' are corresponding points (Fig. 4) and
that T, A', and A are critical points (and the only ones on G and P). T and T'
have the same value of velocity potential q. A short calculation shows that the
circulation I along G from T counter-clockwise through A to T is equal to the
circulation I' along P from T' counter-clockwise through A' to T', and that in
general IfO. Furthermore, if one calculates I (or I') around G (or P) starting
from some point other than T (or T'), one finds a sudden increase in the value
of the velocity potential of the same amount I when crossing T (or T').
We have shown how to obtain from a given potential F(v) and associated
profile G the potential F(z) for the flow of an ideal fluid around any profil
which is generated from G by a conformal transformation.
7. Application of the method. The usual starting potential is that considered
in (9), for the flow around a circular cylinder G of infinite span. Thus

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 369

Ua2eia ir (- -
(17) F(P) = Ue-ia( - ) + + log
- 0 27r a

At present no method is known for obtaining the exact value of the circula-
tion P. The only case in which we are able to calculate P in fair accord with ex-
perience, is that of a profile with a "sharp trailing edge," with moderate thick-
ness, slight curvature, and small angle of attack. In this case we obtain a definite
value of P-independent of viscous phenomena in the boundary layer of fluid
about the body-by using Joukowski's assumption of a "smooth flow" at the trail-
ing edge. According to this assumption, the circulation has in every case just the
amount necessary to secure a finite velocity of the fluid at the trailing edge.
This restriction to profiles with a sharp trailing edge implies that z=f(z) is
not conformal at the point T of G which corresponds to the trailing edge T' of P.
Let T be T and T' be ZT-
If at T the angle ir in the c-plane transforms into the angle (27r-r) at T' in
the z-plane, then at T, z =ff(z) has the development
00

(18) Z - ZT ( tT) (21T)IT + E a,( - T)%(21rT)IT


i=2

Since -r <7r,

dz-
(19) d =i O

By (16), in order that wz be finite at the trailing edge T, dF/djrT must vanish.
This is called the Joukowski condition for circulation.

FIG. 5

Let vT=0o+ae-ig. (In Figure 5, the angle is taken to be positive if T lies


on the lower half of G.) From (10) and the Joukowski condition, we have

dF- irF
(20) I - Ue-2- Uei(a?2) + -- ei = 0,
dn T 2cra

hence

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
370 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

27raU
(21) r - [e= i(a) - +] = 4-rUa sin (a + ,B).

Expressionsfor lift and moment. Suppose that for l


can be developed in the form

k1 k2 k3
(22) z = c + v +-_+-+- + **
~.~ 2 ~ -3

which meets the requirements (c) that z has no singularities at cc, and that at
= oo, dz/d =1. Select the center (to, -go) of G so that c=O0. From (11) the
sulting force acting on the airfoil can be calculated; it is equal to pI U, and acts
perpendicular to Ue+ia. Thus from (12) the lift is L=47rpU2a sin (a +i). Here
(a+f) is the effective angle of attack. In Figure 5, the direction of no lift-the
first axis of the profile-is shown.
The moment of the forces acting on 4P with respect to 0o is

(23) Mc = (z - o) w dz],

C= 27rpU2k2 sin 2(y - a).

The line through ?o with inclination Py to the a-axis is known as the second axis of
the profile. Now

k2

M = MF- L - cos (2y + -a),


a

where MF=27rpU2k2 sin 2(03+y). If MF>0, the airfoil T is unstable; if MF<0,


'P is stable. For neutral equilibrium, L and MC must vanish for the same angle
of attack.
The system of straight lines representing the lines of action of the lift for
different angles of attack envelope a (metacentric) parabola. The first axis is the
directrix of this parabola. Let the focus be F. Then F lies at a distance k2/a
from Po on a line CF which makes an angle 2(3+,y) with the first axis and (i3+,y
with the second axis.

8. The Joukowski family of airfoils [3, 4]. Case r = 0. Airfoil with vanishi
tail angle. Consider

Z - ZT ~
(24) z+ST \t+ T/
Z + ZT + vT

Let T be the point vT=b, where b is a real number. Evidently ZT=2 T if


dz/d= 1 at oo. From (24), we find

b2
(25) z = + '

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 371

which (with a properly defined inverse) satisfies the requirements listed in ?6.
The non-conformal (critical) points L and T of transformation (25) are ? = ? b.
Using transformation (25), we find that every circular arc through T and L
in the c-plane is represented in the z-plane by an airfoil of circular arc through L'
and T', where L' and T' are the points z = + 2b. The circle of radius b and center
at P = 0 is represented in the z-plane by the line segment (called the chord) join-
ing z= -2b and z=2b.
By using different circles passing through L and T in the D-plane as genera-
tors, we can obtain information concerning the flow about indefinitely thin air-
foils of circular shape with different curvatures.
Airfoils of finite thickness and rounded nose can be investigated by using as
generators circles passing through T and having L as an interior point. (Fig. 6.)
Let the center of the generators be

(26) t0 = aE cosf,, ,0 +=- ae sinf3


2

where e is a constant, keeping constant the direction of the first axis. The radius
of the generating circle is a(1 + e).

G~~~~
FIG. 6

Airfoils deduced from (25) in this manner are called the airfoils of the
Joukowskifamily. The skeleton L'T' of a Joukowski airfoil section is the image
of the circle whose radius is a and whose center is (0, h/2).

9. Kdrman-Trefftz family of airfoils [9]. There are two essential restriction


in the airfoil profiles constructed by the Joukowski process: (1) they all have
zero tail angles at the trailing edge, and (2) their thickness is relatively small.
These restrictions are of great importance due to strength and stability require-
ments of practical airfoils. Some of these restrictions can be removed by gen-
eralizing (25).

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
372 USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES [June-July,

The simplest way to obtain a given tail angle r is to use

9'7~~~~~ Z - ZT / T -
(27) = X =2- T/7,
Z + ZT \ + ~T/

a transformation which satisfies the general requirements stated in ?6. Let


CT=b. Then ZT=XvT. Any circle through tT and - T transforms by (27) into an
airfoil called a skeleton enclosed by circular arcs with tail angle r in the z-plane.
Furthermore, circles through tT and having -AT as an interior point transform
into profiles having round noses and sharp trailing edges of the desired tail angle.
Airfoils derived in this manner are known as the Kdrman-Trefftz family.
(Fig. 7.)

/~~

FIG. 7

The Karman-Trefftz airfoils give aerodynamic characteristics only slightly


changed from those of Joukowski airfoils. However, an improvement in regard
to velocity and pressure distribution along the airfoil is obtained. All these air-
foils are inherently unstable. In spite of this, transformation (27) has proved to
be very useful in computing aerodynamic characteristics of given airfoils.

10. Other families of airfoils. By using transformation z =f(t) having fo


or more critical points inside the generating cylinder G, Mises [10] deri
family of stable profiles having sharp trailing edges. Direct methods for finding
the aerodynamic characteristics of airfoils of a given shape have been developed
by Karman, Trefftz, . . ., MIuller, Hohndorf, Theodorsen, and others. The prob-
lem is identical with that of finding the conformal transformation between the
given airfoil and a circle.
In a recent paper Piercy, Piper, and Preston [12] have discussed a new
family of wing profiles which they claim has certain advantages over wing
shapes previously discussed by Joukowski and others. The method consists es-
sentially in selecting one branch of a hyperbola as a generating curve in the
z1-plane and applying a transformation z1z2 = ei which produces an inversion
and reflection with respect to a point C in the line segment CA, and which results

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms
1940] USE OF CONFORMAL MAPPING IN SHAPING WING PROFILES 373

in airfoils closely resembling those of successful modern wings. By applying a


sequence of conformal transformations, the hyperbola is transformed into a
circle permitting the application of the method to the potential function (17)
for a flow past a circle.
References

1. Rayleigh, Lord, On the irregular flight of a tennis ball, Messenger of Mathematics, vol. 7,
1877, p. 14.
2. Kutta, W., Lift forces in flowing fluids, Ill. Aero. Mitt., 1902.
3. Joukowski, N., On the profiles of airfoils, Z. Flugtech. Motorluftschiffahrt, vol. 1, 1910,
p. 281; vol. 3, 1912, p. 81.
4. Joukowski, N., Aerodymamique, Gauthier-Villars, Paris, 1916.
5. Aerodynamics, vols. 1-6, Durand, Editor-in-chief, 1934. This work includes an excellent
bibliography.
6. Tietjens, Applied Hydro and Aero-mechanics, McGraw-Hill, 1934.
Tietjens, Fundamentals of Hydro and Aero-dynamics, McGraw-Hill, 1934.
7. Curtiss, D. R., Analytic Functions of a Complex Variable, Open Court Publishing Co.,
1926.
8. Burington, R. S., and Torrance, C. C., Higher Mathematics, McGraw-Hill, 1939, pp. 268-
269; pp. 616-620.
9. von Karmfan, Th., and Trefftz, E., Potential flow round airfoil profiles, Z. Flugtech. Motor-
luftschiffahrt, vol. 9, 1918, p. 111.
10. von Mises, R., The theory of lift of profiles, Z. Flugtech. Motorluftschiffahrt, vol. 8, 1917,
p. 157; vol. 11, 1920, pp. 68 and 87; Z. Angew. Math. Mech., vol. 2, 1922, p. 71.
11. Aerodynamics, vol. 3, Durand. See section by C. Witoszynski and M. J. Thompson.
12. Piercy, N. A. V., Piper, R. W., Preston, J. H., A new family of wing profiles, Phil. Mag.,
ser. 7, vol. 24, no. 161, Sept. 1937, pp. 425-44.
13. Caratheodory, C., Conformal Representation, Cambridge Tract in Mathematics and
Physics, London, 1932.

This content downloaded from 192.38.67.116 on Wed, 06 Mar 2019 01:32:14 UTC
All use subject to https://about.jstor.org/terms

Das könnte Ihnen auch gefallen