Aerodynamics As The Basis of Aviation How Well Did It Do

Journal of Aeronautical History Paper No.
2018/01
Aerodynamics as the Basis of Aviation: How Well Did It Do?

J. A. D. Ackroyd
Timperley, Cheshire
In memory of the Royal Aircraft Establishment, its staff and contributions to aerodynamics
Abstract
The paper describes the role of aerodynamics in the enhancement of aeroplane performance.
To illustrate this, drag and lift-to-drag data are reviewed, covering the first half-century of
powered, controllable flight. The survey begins with the Wright Flyer and the biplanes
subsequently developed. This is followed by the monoplane’s ascendancy, the new ideas in
aerodynamics here leading to significant drag reduction and increased speed. For this phase,
data provided by the Royal Aircraft Establishment, which appear to be not widely known, are
discussed in some detail, together with the Establishment’s development of drag assessment
methods. The survey then turns to the emerging jet age, ending with the early British swept-
wing aircraft, forerunners of the Swift and Hunter. The influence of Reynolds number emerges
in the survey but the transonic drag rise due to compressibility is not covered. It is hoped
that this survey of aerodynamic drag reduction will be of interest to students new to the
subject and also to those wishing to learn more of the development of aeronautical science.
1. Introduction
This paper is, in one sense, an appendix to two earlier papers published in this Journal. The
(1)
first described the evolution of our understanding of the aeroplane’s lift and drag forces.
This, in part, served as the basis of the author’s Royal Aeronautical Society Cody Lecture in
(2)
November 2015 given to the Society’s Farnborough Branch. The second paper described
the evolution of the Spitfire’s aerodynamic design and sprang from the author’s lecture on
that subject given to the Society’s Spitfire Seminar at Hamilton Place in September 2016.
Both lectures included material not covered in the two papers, consisting of basic
aerodynamic data for a variety of aircraft. Such data, particularly those amassed by the Royal
Aircraft Establishment (RAE), Farnborough, around the period of the Second World War,
might be of interest to a wider audience.
The Cody Lecture had the title ‘Aerodynamics as the Basis of Aviation’ and this provides the
first part of this paper’s title. To a modern audience this is clearly a no-brainer; a grasp of
aerodynamics is nowadays accepted without question as an essential part of any successful
aeroplane design and, indeed, has often been the lead technology. However, the lecture’s title
was taken from a lecture presented by Sebastian Finsterwalder (1862 – 1951) at Lausanne in
(3)
1909 and published in the following year. One suspects that Finsterwalder had adapted the
title of an earlier book, widely read by aspiring aeronauts, written by the hang-gliding pioneer
(4)
Otto Lilienthal (1848 – 1896). That book had the title Bird Flight as the Basis of Aviation
and appeared in 1889. At that time and until the early years of the twentieth century,
aerodynamics had been an entirely experimental activity in which various wing shapes were
tested, shapes often loosely based on those in the natural world. Yet no one was able to say
1
Journal of Aeronautical History Paper No. 2018/01
why some wings performed better than others or indeed, more fundamentally, on what
(3)
rational basis did any of them function. In contrast, Finsterwalder was suggesting that an
emulation of bird flight no longer pointed the way forward; in future, guidance would be
provided by this new science of aerodynamics. This, he believed, had at last acquired a
rational basis resting on sound physical principles. In particular, he was drawing attention to
the new circulation theory of lift developed by his Munich colleague, Martin Wilhelm Kutta
(1867 - 1944), and the similar ideas of Frederick William Lanchester (1868 - 1946) in Britain.
Indeed, on that basis the essential elements of modern wing theory emerged during the next
decade, more details of which can be found in Reference 1.
Given the challenge implicit in Finsterwalder’s lecture title, one is entitled to ask: how well
did this new science do? That question provides the second part of this paper’s title. In
answer, the paper reviews certain core aircraft performance data over the years to illustrate
the improvements brought about by the application of this new branch of the physical
sciences. Two indicators of aerodynamic quality are surveyed, namely the drag coefficient at
zero lift CD0 and the maximum lift-to-drag ratio (L/D)max. The latter, however, is shown in
Appendix 1 to be intimately related to CD0 and wing aspect ratio. A further important
aerodynamic factor is the maximum lift coefficient CLmax, but this is not dealt with here. The
main interest is in drag and what we might call the ‘history of CD0 reduction’.
In this context the paper can also be seen as a supplement to Loftin’s book Quest for
(5)
performance. The evolution of modern aircraft . Written by the former Chief Aeronautical
Engineer at NASA’s Langley Research Center, this provides a masterly survey of the
technical development of the aeroplane which also emphasises the importance of the above
(5)
two aerodynamic quantities. However, whilst Loftin draws his early examples from
British and German aircraft of the First World War, later examples are predominantly those
aircraft produced in the United States. Here, British examples drawn from the RAE’s archive
(5)
supplement Loftin’s data. Moreover, in dealing with piston-engine aircraft, Loftin bases
his calculations for CD0 entirely on engine power. But with the advent of ducted radiators
and rearward ejecting exhausts in the mid-1930s, engine power became augmented slightly by
(5)
the additional thrusts produced by these devices. Thus Loftin slightly underestimates the
values of CD0 for the aircraft of what we might call the ‘Spitfire generation’. Here thrust
augmentation effects are included in the calculations leading to the data listed.
The next section of the paper describes the basic features of aerodynamic drag. There are two
contributions, namely the drag induced by the trailing vortices of an aeroplane’s lifting
system and the drag due to the aeroplane’s shape at its zero lift condition. The survey of
basic aerodynamic data begins in Section 3 and deals with the period from the Wright Flyer
of 1903 to the early 1930s, a period initially dominated by the biplane. Section 4 reviews the
lecture given to this Society in 1929 by Bennett Melvill Jones. Entitled ‘The Streamline
Aeroplane’, his lecture was, in essence, a plea for British designers to clean up their
aerodynamic act. Aptly timed, it came within the short period during which designers were
beginning to turn away from the well-understood biplane configuration of fabric-covered
structural members, initially of wood, in favour of the greater efficiency offered by the
monoplane of stressed-skin metal construction. Moreover, Jones was able, in effect, to
2
provide a target value for CD0 for his ‘Streamline Aeroplane’ to which designers might aim.
Section 5 surveys the RAE’s aerodynamic data for the piston-engine monoplanes developed
around the period of the Second World War. The final Section covers the beginning of the jet
age. British units are used throughout, in keeping with the original aircraft performance data
used.
2. Aerodynamic Drag and Its Evaluation
Before beginning this survey of aircraft drag reduction over the years, it is necessary to define
drag, understand its origins and be specific as to the measure of it. In particular, we must be
clear about the nature of those two aerodynamic quantities mentioned above, the drag at zero
lift CD0 and (L/D)max.
(6)
Using his newly-established principles of dynamics, Isaac Newton (1643 – 1727) showed
that the air resistance experienced by a body is proportional to the product of air density ρ,
the square of the flow velocity V, and an area S which is characteristic of the body.
Subsequently, a vast quantity of experimental evidence confirmed this fundamental
proportionality. Thus the resistance component along an aeroplane’s direction of flight, the
drag force D, is given by
2
D = ½ ρV S CD. (1)
The factor of ½ was introduced in the late 1920s because the instrument used to obtain
2
airspeed on aircraft and in wind tunnels, the Pitot-static tube, measures ½ρV directly. As to
the choice of the characteristic area S, the convention in aeronautics is that this is the total
wing planform area. The quantity CD, the ‘proportionality factor’ implicit in Newton’s
relationship, is called the drag coefficient. Equation (1) shows that CD is simply a number
2
having no dimensions since both D and ρV S have the dimension of force, lbs in the British
system or newtons in Système International (SI) units. The value of this non-dimensional
number is not a universal constant but, broadly speaking, depends on the shape of the aircraft
and its attitude to the airstream. Thus it provides a measure of how drag-prone is the aircraft;
for drag reduction, the lower its value the better.
To give some idea of the range of CD values encountered, a flat plate held normal to an
airstream has a CD value a little over unity, the drag coefficient being based on its planform
area S. This is also its frontal area, the area ‘seen’ by the approaching flow. A circular
cylinder, for which S is taken to be the ‘seen’ frontal area, the product of diameter and span,
has a CD value only slightly less than this. A well-streamlined aerofoil, when reared up
through 90° so as to present its under-surface head-on to the airstream in a very un-
streamlined posture, also has a high CD value, based on its planform area S, which is similar
to those of the plate and the cylinder. In contrast, that aerofoil set at zero lift and small
incidence has a CD value which is roughly one hundredth of these values. Such an aerofoil
has a maximum thickness of, say, one tenth of its chord. Thus its CD value based on its
frontal area, the product of maximum thickness and span, is ten times higher than that based
on its planform area, yet ten times less than that of the cylinder. The practical significance of
3
these convoluted CD examples, however, is revealed once they are translated into actual drag
forces. It follows from the above that a wing having this aerofoil section set at low incidence
and zero lift has the same drag at the same airspeed as a circular cylinder, or a wire, of the
same span but of a diameter one tenth the maximum thickness of the aerofoil. This provides
an indication of the scale of the drag experienced by the early biplanes and those few
monoplanes of the early years of powered flight, festooned as they were with the struts and
bracing wires needed for structural security.
As explained in Reference 1, the airflow close to a surface is slowed by friction, and the role
of the thin, highly viscous boundary layer in this is crucial to the explanation of the drags
described above. For the aerofoil at low incidence, it is its streamlined shape which controls
the boundary layer’s behaviour, persuading it to remain attached so as to separate only at the
sharp trailing edge. Consequently the drag is created solely by the attached boundary layer,
is relatively small and is mainly due to viscous skin friction. In contrast, the situation for the
normal plate and the cylinder is quite different. In the case of the plate the boundary layer
separates at its sharp edges; for the cylinder separation occurs near its maximum thickness.
In both cases, separation results in wide wakes possessing low pressures, partial vacuums
which suck the bodies backwards rather more than they are pushed rearwards by the flow’s
higher pressure impinging on their forward surfaces. It is this large pressure imbalance
which is the dominant contributor to their high drags, not skin friction.
Little of this was understood until the early years of the twentieth century. The boat builders
of antiquity must have been aware that streamlined shapes are advantageous but this does not
appear to have prompted wider use of the idea. There were, however, moments of insight as
the years progressed. The ‘Father of Aeronautics’, George Cayley (1773 – 1857), for
(7)
example, produced a streamlined shape in 1809, based on his measurements of a trout . He
(8)
explained his thinking in the last of his triple papers published in 1810, pointing out that
the shape of the rear of a body is as important as the front in reducing resistance. Without a
tapered rear, he explained, “a partial vacuity” would be created, its suction thereby increasing
the resistance. This, it must be added, went against current opinion, which held that it was
the shape of the front of a body, not its rear, which determines resistance.
That all this might be in some way related to the air’s viscosity seems not to have occurred to
the nineteenth century’s budding aeronautical community. Indeed, as late as 1891 Samuel
Pierpont Langley (1834 – 1906), sometimes credited with coining the term ‘aerodynamics’,
(9)
asserted that the value of the air’s viscosity coefficient was far too small to be of any
influence. An entirely contrary view emerged in the lecture delivered by Ludwig Prandtl
(10)
(1875 – 1953) to the mathematical congress held at Heidelberg in 1904 . In this he
pointed out that it was the conjunction of the air’s extremely small viscosity coefficient and
the enormous velocity changes within the thin boundary layer surrounding the surfaces of a
body which creates significant skin friction drag. For what we now call laminar flow, in
which the motion is smooth and devoid of random irregularities, he showed that for a flat
–½
plate at zero incidence the drag coefficient CD is proportional to Re . Here Re is the
Reynolds number, a further dimensionless quantity which expresses the ratio between forces
due to inertia and viscosity, and is given by
4
Re = ρ V l / μ, (2)
where l is the plate length and μ the viscosity coefficient. Using the flow equations for the
(10)
laminar boundary layer developed by Prandtl , his student Paul Richard Heinrich Blasius
(11)
(1883 – 1970) obtained
-1/2
CD = 2.654 Re . (3)
Experiment has confirmed this basic result although more accurate calculation gives the
numerical coefficient as 2.656. At the extremely high Reynolds numbers experienced in
flight, however, the laminar boundary layer is found to be unstable and trips into turbulence.
For drag estimation with such boundary layers, reliance has had to be placed more on
experimental evidence. As we shall see in Section 4, the CD relation for turbulent flow over a
flat plate takes a form similar to equation (3) although both the coefficient and the power to
which Re must be raised are rather different.
(10)
Prandtl’s lecture of 1904 also explained the circumstance in which boundary-layer
separation occurs. In passing around a body, the flow exterior to the boundary layer initially
accelerates up to the body’s maximum thickness but thereafter decelerates toward the rear. If
this deceleration is too severe, the boundary-layer flow, already retarded by viscous action,
may simply stop. Subsequently the boundary-layer flow peels away from the surface,
exhibiting the phenomenon of boundary-layer separation which is the cause of the wide low
pressure wake and excess drag. This, then, provides the reason for streamlining; the long
tapered tail produces gentle deceleration so that separation is avoided. As to the effect of
sharp edges such as those on the normal plate described above, the flow initially accelerates
rapidly in attempting to round the edge but then promptly decelerates so severely that
separation at the edge is inevitable. This also occurs at the sharp trailing edge of an aerofoil
but, as Reference 1 explains, it is by this deliberate enforcement of boundary-layer separation
there that this fundamental action of viscosity is used to generate the lifting system of a wing.
As to the lift L, the force component perpendicular to the flight direction, this mimics
equation (1) by taking the form
2
L = ½ ρV S CL. (4)
Although we are mainly interested in the drag coefficient CD, nonetheless the lift, and its
non-dimensional coefficient CL, enter the drag story. As explained in Reference 1, due to
their wings’ lift forces being generated continuously by the circulatory or bound vortices
about them, all aeroplanes in flight inevitably create a system of trailing vortices in their wakes.
These vortices produce a contribution to an aeroplane’s drag called induced drag. By the
close of the First World War the sound theoretical basis of this wing behaviour had emerged
(12, 13)
from Prandtl and his Göttingen associates . Hermann Glauert of the RAE (Figure 1),
who explained these ideas to a Britain largely innocent of such transformative advances,
(14)
showed that the induced drag coefficient CDi is given by
2
CDi = k CL /(πA). (5)
Here A is the wing aspect ratio defined as the wing span divided by the average chord
2
(the fore and aft dimension of the wing planform), or alternately as (wing span) / (wing
planform area S). The factor k has a minimum value of unity, which is obtained for a
5
monoplane wing alone experiencing an

elliptic distribution of lift loading. The
latter is most easily achieved by the use
of an elliptic planform wing shape;
other shapes produce k values slightly
higher than unity. However, further
contributions to k arise from the lifting
systems of the complete aircraft, notably
the tailplane and, in some cases, the
fuselage itself. Thus arriving at a k
value for a complete aircraft is often
problematic and relies on wind-tunnel
and flight-test data. As related in
Reference 2, it is thought that
Supermarine, for example, used the
value k = 1.15 drawn from experience
with the Spitfire. Biplanes and
multiplanes produce further
complexities in k’s estimation since the
trailing vortices of the individual wings
mutually interact in creating the total
(14)
induced drag. Glauert , for example,
Figure 1. Hermann Glauert FRS (1892-1934)*
calculated that k = 1.58 for biplane
Source: Royal Aeronautical Society
wings of rectangular planform and equal
(National Aerospace Library)
spans, aspect ratio 6, and a gap-to-span
ratio of 1/6. Here roughly 90% of the
increase in k above unity comes from the biplane correction, the remaining slight correction
being due to the fact that the rectangular wings’ planforms do not produce ideal elliptic
loading.
The coefficient for the total drag experienced by an aeroplane, CD, is then
CD = CD0 + CDi. (6)
Here CD0 is the drag coefficient at the zero lift condition, the coefficient representing the drag
caused by the airflow behaviour around the aircraft at this condition for which, there being no
trailing vortices, the induced drag is also zero. The value of CD0 partly depends on the effects
___________________________________________________________________________
* Born in Yorkshire, Hermann Glauert won a mathematics scholarship to Trinity College
Cambridge and in 1916 joined the Aerodynamics Department of the Royal Aircraft
Factory, Farnborough (from 1918 the RAE). His visit to Göttingen after the close of the
First World War provided him with the work there on boundary layers and wing theory. In
his subsequent career he introduced improved mathematical methods to wing theory,
extending this to a wide variety of aerodynamic problems. He became Head of the
Aerodynamics Department in 1931 but was killed in a tragic accident in 1934. He was the
first of a number of RAE staff to become Fellows of the Royal Society.
6
of the boundary layer such as surface skin friction and, if present, boundary-layer separation,
but also includes drag contributions caused by excrescences. The boundary-layer drag
contribution is often referred to as profile drag whereas that due to excrescences and such is
called parasite drag. Unlike induced drag, or CDi, which can be predicted with reasonable
accuracy, as above, the prediction of CD0 is more problematic, as we shall see as the drag
story unfolds.
An estimate of CD0 can be obtained from wind-tunnel tests, simply by measuring the aircraft
model’s drag at zero lift. However, because the model is smaller than the full-scale aircraft
and the testing speed usually much lower than that achieved in full-scale flight, the problem
of the ‘scale’ or Reynolds number effect (see, for example, References 1 and 2), can result in
inaccurate values for the full-scale value of CD0. Values of higher accuracy are obtained by
flight testing at full scale. In this case knowledge of the engine power or thrust required to
achieve the measured airspeed at a given altitude is used to calculate CD. Also from the
airspeed and the additional knowledge of the aeroplane’s weight the value of CL at this flight-
test condition can be obtained from equation (4). Equation (5) then provides an estimate for
CDi so that, from equation (6), the value of CD0 is obtained. More details of this calculation
procedure are given in Appendix 2.
At this stage it is useful to see how CL, CD, and CL/CD (= L/D) for an aeroplane vary
throughout the useful incidence range. Figure 2 illustrates these variations, although the
graphs are not drawn to a common vertical scale but merely illustrate the trends. Thus CL
increases almost linearly with incidence until the graph curls over to a maximum. At this
point wing stall occurs around CLmax, the stall being caused by gross boundary-layer
separation due to the wing upper surface’s flow deceleration becoming too severe.
It is important to understand that, at whatever point an aeroplane reaches in its flight

envelope, the aeroplane must be at some point on the CL ~ incidence graph. For example, at
high speed the lift L is ‘generated’
2
mainly by the ½ρV term in equation
(4) and the aeroplane flies at a low
value of CL/low incidence.
Conversely, at low speed a high
CL/high incidence is required.
According to equation (5), CDi varies

2
as CL so the CD graph shows a
quadratic curve rapidly rising upward
from its initial CD0 value as incidence
increases. For drag analysis, flight-
test results are often obtained at the
higher end of the speed spectrum so
that, as mentioned earlier, CL is low,
typically around 0.1. In this case CDi Figure 2 Variations of CL, CD and L/D with
is small, usually around 5% of CD. wing incidence α
7
Consequently, for the calculation of CD0 from flight-test data, an accurate value for k in the
estimation of CDi is not essential. Indeed, in Section 5 it will be found that, in British practice
during the 1940s, for simplicity the value of k selected was unity, the value for a wing alone
experiencing the ideal elliptic loading situation.
2
Because CDi varies as CL the graph for L/D shows a maximum value, not at CLmax but at
some intermediate CL value. In Appendix 1 it is shown that a high value of (L/D)max
produces the beneficial effects of small gliding angles and, for powered flight, enhanced
range. The ratio is thus an indicator of aerodynamic efficiency and it is therefore useful to
predict its value. The analysis for this is included in Appendix 1, the result being that
½
(L/D)max = ½(πA/(k CD0)) . (7)
Clearly, in order to produce a high value of (L/D)max it is necessary to have CD0 as small as
possible whilst the aspect ratio A should be as large as structural limits and mission
requirements allow. Because of the form of equation (7), the estimation of (L/D)max from
that equation requires a reasonably accurate value of k to be selected, unlike the case above in
the estimation of CD0.
Appendix 1 also includes a discussion of Figure 2’s other features, particularly those related
to the early wing theories proposed by Prandtl and his Göttingen colleagues and by Lanchester
in Britain.
During the first forty or so years of powered flight, propulsion was provided almost
exclusively by piston-engine/propeller combinations. The engine powers required for this
phase of development are discussed in Appendix 2, in which it is shown that engine power is
related to the cube of aircraft speed. As an illustration of what might be called the ‘tyranny
3
of V ’, it is shown that an aeroplane with the high drag characteristics and wing area of the
Wright Flyer would have required an engine power of about 10,000 hp to achieve a
maximum speed of around 300 mph. That such a speed was achieved with powers around a
tenth of that value is a tribute not only to engine designers but also to the successful
application of the new ideas in aerodynamics. The latter are shown in Appendix 2 to have
reduced the drag of the Flyer by a factor of ten. The achievement is all the more remarkable
in that it was accomplished in little over thirty years after the Flyer. That said, the basic
guidelines for this advance – the boundary-layer and wing theories - were in place ten years
earlier or, to re-iterate an earlier point, a decade after Finsterwalder’s lecture.
Appendix 2 also includes the effects of thrust augmentation made possible by the
introduction of ducted radiators and rearward-ejecting exhausts. In addition, the analysis
underlying the British practice in drag assessment is explained, in which all drag
contributions measured either in wind tunnels or in flight tests are scaled down to drag data at
a speed of 100 ft/s at sea level. This made it easier to compare the drag characteristics of one
aircraft type with another.
8
3. From the Wright Flyer to Aircraft of the Early 1930s
It would be appropriate to begin this survey of aircraft drag data with those for the first
aeroplane to achieve powered, controllable flight, the Wright Flyer flown at Kitty Hawk in
December 1903. However, little can be deduced from the short flights achieved. Luckily, as
part of the centenary celebrations of the Wrights’ achievements, wind tunnel results for a
scale model of its development, the Flyer 3 of 1905, are now available. Although the main
(15)
interest of the paper by Padfield and Lawrence centres on the Flyer 3’s unusual flight
dynamics, its basic aerodynamic data were obtained from wind-tunnel tests of a one-eighth
(16)
scale model. These data are given in Lawrence’s doctoral thesis and it is felt that data for
(16)
the 1903 Flyer would not differ significantly from these. Lawrence obtains CD0 = 0.1
whereas (L/D)max is around 5.6 and these results head Table 1 below. Using Glauert’s value of
k = 1.58 for an equal span biplane, wings of aspect ratio 6 and gap-to-span ratio 1/6 (= 0.167)
[the Flyer 3’s aspect ratio is 6.47 and gap-to-span ratio 0.15], equation (7) yields (L/D)max = 5.7,
(16) (5)
a result close to that given by Lawrence . Table 1’s other data are taken from Loftin , to
which have been added the year of first flight.
Table 1’s data cover the initial, largely biplane phase of aeroplane development followed by
the early years of the aerodynamically cleaner monoplane. In the former category, the
examples include a few monoplanes, also often heavily braced, but all possessing such
additional drag-producing features as open cockpits and fixed undercarriages.
Table 1 Zero lift drag, aspect ratio and maximum lift/drag, aircraft 1903 to 1935
CD0 Aspect ratio A Lift/drag (L/D)max
Wright Flyer 3 (1905) 0.10 6.28 5.6
B. E. 2c (1914) 0.037 4.47 8.2
Fokker E III (1915) 0.077 5.70 6.4
Airco DH-2 (1915) 0.043 3.88 7.0
Airco DH-4 (1916) 0.042 4.97 8.1
Albatros D-III (1916) 0.047 4.65 7.5
Sopwith F.1 Camel (1916) 0.038 4.11 7.7
Fokker Dr. 1 Triplane (1917) 0.032 4.04 8.0
Junkers D-I (1917) 0.061 5.46 7.0
Fokker D-VIII (1918) 0.055 6.58 8.1
Handley Page 0/400 (1918) 0.043 7.31 9.7
Fokker F-2 (1920) 0.047 7.10 9.4
Handley Page W8f (1924) 0.055 4.67 7.1
Ford 5-AT (1926) 0.047 7.26 9.5
Ryan NYP (1927) 0.038 6.63 10.1
Northrop Alpha (1930) 0.027 5.93 11.3
Lockheed Vega 5C (1931) 0.028 7.65 11.4
Lockheed Orion 9D (1931) 0.021 7.01 14.1
Boeing 247D (1933) 0.021 6.55 13.5
Douglas DC-3 (1935) 0.025 9.14 14.7
9
The Royal Aircraft Factory’s B. E. 2c is a particularly interesting case since it emerged so

(5)
early and, in terms of CD0 and (L/D)max, it appears to perform well. Yet as Loftin points
out, it was underpowered with a slow top speed and a poor rate of climb. One virtue
appeared to be its high stability, yet this resulted in sluggish manoeuvrability in defence
which earned for it the term ‘cold meat’ from German fighter pilots. Readers are encouraged
(5)
to study the well-judged assessments Loftin provides for many of these First World War
aircraft, particularly his reasons for why the few monoplanes (Fokker E III, Junkers D-1,
Fokker D-VIII) did little better than their biplane contemporaries. Suffice it to say that many
factors determined the usefulness or otherwise of aeroplanes deployed in that conflict, and
low drag was probably not the foremost consideration.
Economic considerations in commercial aviation became more of a driving force for

aerodynamic improvement during the late 1920s, particularly in the United States. Further
impetus for drag reduction arose when national prestige was at stake; one way or another,
money could be found to support such endeavours as the Schneider Trophy contests and the
MacRobertson Air Race from London to Melbourne in 1934.
A significant decrease in CD0 can be detected in Table 1 with the advent of the more
streamlined monoplane, examples being the Northrop Alpha and the two Lockheed aircraft,
the Orion being notable for its retractable undercarriage. Here the monoplane also began to
benefit from the adoption of solid-skin construction, the skin now assisting in carrying the
main structural loads. Table 1’s last two types, the Boeing and Douglas aircraft, provide
examples in which the further step of all-metal construction had been taken. Together with
this and the adoption of retractable undercarriages came the use of enclosed cockpits and
means to reduce engine drag: cowlings for radial engines and ducted radiators for liquid-
cooled engines. Further aerodynamic refinements such as wing-root fillets and better surface
finishes produced yet more drag improvements. The resulting reductions in CD0, together
with the use of higher wing aspect ratios for commercial aircraft, resulted in improvements in
(L/D)max, as Table 1 shows.
(17)
Anderson uses such data to illustrate these phases of development by constructing a
diagram showing a series of three downward steps in CD0 values as the years progressed. In
this diagram the first, largely biplane phase (1910 – 1925) has an average CD0 value around
0.045, but with notable scatter as Table 1’s data indicate. For the second phase (1927 – 1947)
in which the piston-engine monoplane takes the stage, the average CD0 value drops to about
0.027, but again with significant scatter. More data for this phase, drawn from the large RAE
(17)
archive, are given in Section 5 below. According to Anderson the final phase, that of the
jet-propelled aeroplane, produced an average CD0 of 0.015 and British data for the early years
of this phase are given in Section 6.
4. B. M. Jones and the ‘Streamline Aeroplane’ (1929)

(18)
The importance of the lecture given by Jones (Figure 3) to the Royal Aeronautical Society
in 1929 has been mentioned in Section 1. At this stage of his notable career, by now a
10
leading member of Britain’s Aeronautical

Research Committee (ARC), he was well-
placed to press for aerodynamic
improvements in British aircraft.
(18)
Jones began his lecture by reminding
his audience of “the effortless flight of the
sea birds and the correlated phenomenon
of the beauty and grace of their forms. We
all possess a more or less clear ideal of
what an aeroplane should look like; a kind
of albatross…” He went on to complain
that “progress towards this ideal, so far as
the general purpose craft is concerned is,
we must admit, painfully slow. It has
seemed to me that a contributory factor to
the slowness of this evolution has been the
lack of any generally understood and
easily visualised estimate of what could be
achieved were the difficulties in the way Figure 3 Bennett Melvill Jones FRS (1887 – 1975).
of realising the ideal form overcome. He joined the Aerodynamics Department of the
There is a natural tendency to decide on National Physical Laboratory in 1911, moved to the
one day that the gain – say 20 per cent. on Royal Aircraft Factory in 1914 and in 1916 to the
the total drag, or 7 per cent. on the speed – Air Armament Experimental Station, Orford Ness.
to be had by spending endless trouble on A Cambridge graduate, he returned there in 1919 to
become Francis Mond Professor of Aeronautical
improving the undercarriage design, is not
Engineering until retirement in 1952.
worth the trouble; on the next day to come
Image: Aircrew Remembered.
to a similar conclusion about the drag of http://aircrewremembered.com/
the engine cooling apparatus; on the next
day about the wires, struts and minor excrescences; and on the next about the pilot’s view;
omitting to notice that if all the improvements were made at once the total gain would not be
some insignificant percentage of the whole, but might reduce power consumption to a small
fraction of its original value and so extend the range and usefulness of the aeroplane into
realms which would otherwise be unattainable.”
(18)
Jones then explained that there are two distinct types of drag, as outlined in Section 2,
namely induced drag and the drag due to the aeroplane’s shape. The former, he pointed out,
can be calculated with reasonable precision using, in effect, equation (5) above. He
demonstrated that, at the high end of the speed spectrum, this induced drag is small, typically
5% of the total. Thus it is with the residual drag contribution that the main problem lies. He
then proposed an aeroplane shaped so that excrescences and such are entirely absent, a shape
so streamlined, so aerodynamically clean as to be devoid of parasite drag, that the only
remaining drag contribution is that which cannot be eradicated, namely the profile drag due
to the boundary layer. He rightly took the view that at the high Reynolds numbers of full-
scale flight the boundary layer would largely be turbulent. To this should be added the point
that all parts of an aircraft’s surface infected by a propeller’s turbulent wash will inevitably
11
be turbulent. For such conditions he replaced equation (3) for laminar flow with one based
(19)
on experiment provided by Prandtl which, in the notation used here, is
-0.15
CD = 0.076 Re . (8)
With this relation for turbulent skin friction on flat plates, he produced, in effect, an estimate
for CD0 for his ideal ‘streamline aeroplane’. The details of his calculations are given in
Appendix 3, where it is seen that his target value, though not stated in such terms, is
CD0 = 0.0128. (9)
Here bold italics are used to indicate that this CD0 value is for the clean aeroplane, a notation
which will be used later in Section 5.
(18)
On the above basis, in effect, and by the use of equation (5) for CDi, Jones constructed the
fairly narrow band of theoretical curves shown toward the base of Figure 4. The horizontal
ordinate here is the aircraft speed in miles per hour. Probably because he wished to present
his data in a form which would appeal to his audience at that time, he chose as his vertical
ordinate the quantity ‘Brake Horsepower per 1000 lbs weight’. Consequently, his theoretical
curves had to be calculated for a range of wing loadings (weight/S) and also span loadings
2
(weight/(span) ). Included in the figure are the data points for current aircraft, many of
which are British biplanes, and these provide a truly graphic illustration of how far in excess
of Jones’s ideal these were. The nearest to his theoretical curves is the monoplane Ryan NYP
‘Spirit of St. Louis’ (see Table 1) in which Charles Lindbergh had flown the Atlantic in 1927.
Figure 4 Power versus Speed; Jones, Reference 18
12
As was often the case during this period, the discussion following Jones’s lecture, carefully
(18)
recorded in its publication , was almost as interesting as the lecture itself. The audience’s
response was somewhat mixed; whilst some clearly got the message, others were rather
dubious. Charles Clement Walker (1877 – 1968) of de Havilland Aircraft, for example,
reported that “an ordinary commercial aeroplane did attain from 60 to 67 per cent of the
streamline speed”. He went on to list a number of excrescences and such which were, he felt,
unavoidable on aircraft: “the undercarriage, all the bracing, the cooling of the engine, the
cockpit”. He added that “in the case of commercial aircraft, which must have all sorts of
other excrescences, it was difficult to foresee how far one could go towards eliminating the
resistance. Such a machine had a radial engine sometimes at the front of the body, the cabin
had to be ventilated by means of structures like ships’ ventilators, sometimes there was a
starting engine mounted outside the fuselage, and it did not matter very much what one did in
the way of fairing after that. It would be interesting to know what Professor Jones thought
about the amount that could be saved”. The transcript of the discussion records audience
laughter at this point. The latter perhaps suggests that Walker’s contribution was seen by
some as apt criticism. Yet within the next few years a number of his listed drag creators were
either eliminated or reduced in their effect. One such item was raised by George Herbert
Dowty (1901 – 1975) who, whilst recognising the undercarriage as a significant drag creator,
took the view that “the retractable undercarriage is not desirable because of the complicated
retracting mechanism and the extra weight involved”. He went on to report that he was
working on a fixed undercarriage of reduced drag. Though not germane to the current topic
of drag reduction, it is worth noting that around this time similar reservations on the grounds
of complexity and weight were expressed in the discussion following a lecture to this Society
(20)
on variable pitch propellers .
(18) 3
In his response to Walker, Jones remarked that “Since power varies approximately as V ,
the realised speed of 70 per cent of the streamline speed corresponds to an expenditure of
power of about three times the streamline power for a given speed”. To Dowty he merely
replied that he was “interested to hear of his efforts to tackle the problem”. As to matters
raised by other audience members, much of their discussion turned into a debate on transition
to turbulence, providing an interesting snap shot of the limited state of aerodynamic knowledge
on this topic in Britain at that time. At one point David Randall Pye (1886 – 1960) asked if
data for aircraft competing for the Schneider Trophy could be added to Figure 4. To this
Jones replied that he had not included these aircraft since they were not comparable with the
other aeroplanes in his figure. He added that “it would be interesting to see them worked out
accurately by someone at the Air Ministry who has the facts at his fingers’ ends”. Such data
as are available to this author for the Supermarine S4 to S6B, together with the Spitfire’s
forerunner, the Type 224, are listed in Appendix 4. These are used to provide rather tentative
values for CD0.
At this point it is interesting to ask: was Jones’s ideal, put here as CD0 = 0.0128, achieved in
the subsequent drive to produce aerodynamically cleaner aeroplanes? The answer for the
piston-engine era is that it was not. Taking the case of the Spitfire, one of the cleaner British
military aircraft, we will find a CD0 value of 0.020 in the survey of Section 5 below. There it
will be seen that its boundary-layer drag is about 58% of the total, indicating that if all other
parasite drag contributions had been eradicated its CD0 value would have been 0.012, close to
13
Jones’s ideal. But as we shall see, for practical reasons total parasite drag eradication was not
possible. For the jet aeroplane era the results turn out to be rather better, as will become clear
in Section 6.
Finally, to return to Jones’s opening comments lauding the aerodynamic virtues of the
(21)
albatross, we might ask: how well does that bird do? In 2005 Sachs reviewed the various
estimates of albatross performance made between 1932 and 1982. Estimates for CD0 ranged
between 0.020 and 0.048 and, for his calculations, he selected 0.033. Aspect ratio estimates
varied between 15 and 20 so that (L/D)max values ranged from 18 to 24.6. For the latter ratio,
(21)
Sachs selected a value of 20. It appears then that with regard to CD0 the piston-engine
fighters of the Spitfire generation did really rather well in comparison.
5. Piston-Engine Monoplanes (1930s and 1940s)

5.1 Royal Aircraft Establishment figures of merit
With the arrival of the new generation of largely monoplane military aircraft in the mid-
1930s, the RAE’s Aerodynamic Staff felt the need to set down simple figures of merit when
(22)
assessing aeroplane performance . The emphasis was on what they termed residual drag,
the drag remaining once induced drag has been deducted. Residual drag, it will be recalled,
has two contributions, the profile drag created by the boundary layer’s action, and the parasite
drag created by protrusions, excrescences, leaks from access panels and bomb doors and, in
general, anything which creates departures from a well-streamlined clean form. The
(22)
Aerodynamic Staff’s report issued in January 1937 introduced four figures of merit, two
of which turned out to be CD0 and (L/D)max, the latter being seen as an indicator of transport
efficiency (see Appendix 1). However, a third figure of merit appeared, this being the
cleanness efficiency, a term which rapidly became renamed cleanness ratio, CR. This is
another measure of the residual drag, and was defined as the ratio of the residual drag of the
clean aeroplane, in other words the profile drag entirely due to the boundary layer, to the
(18)
actual residual drag. As Jones had done (see Appendix 3), the profile drag due to the
boundary layer was estimated using experimentally obtained flat plate skin friction formulae
assuming a turbulent boundary layer throughout. In the notation adopted in Section 4, it
follows that
CR = CD0 / CD0. (10)
Here, however, it must be understood that CD0, the drag coefficient entirely due to the
(18)
boundary layer on the clean aircraft, does not have the value assigned to it by Jones , the
result given by equation (9). That value was an average for the current aircraft surveyed by
Jones in 1929. Now the evaluation of CD0 is tailored to each individual aircraft. Using the
value for the skin friction coefficient appropriate to the flight Reynolds number as described
in Appendix 3, that coefficient is then applied to the estimated total wetted area of that
aircraft and finally CD0 is calculated knowing the ratio of the total wetted area to the wing
planform area S.
14
The final figure of merit introduced in Reference 22 compared the residual drag of an
equivalent flying wing, there being no other lifting surfaces, and the residual drag of the
actual aeroplane. However, this criterion was dropped in the next report on the matter issued
(23)
in October 1937 .
(22) (23)
Whilst the first report provided performance data for fifteen recent aircraft, the second
added data for a further six: Hawker Hurricane, Supermarine Spitfire, Fairey Battle, Bristol
Blenheim, Armstrong Whitworth Whitley and Handley Page Harrow. Of the first report’s
aircraft, the Avro 652 became the Anson. The less-well-known Burnelli UB-14A was an
American lifting-fuselage airliner, the Hendy Heck built by Parnall Aircraft a four-seat cabin
monoplane, the Martin B-10 the U S Army Air Corps’ first all-metal monoplane bomber, the
Caudron 460 a French racing monoplane and the Miles M.6 Hawcon was a research
monoplane built for the RAE on which four wing thicknesses could be investigated.
(22, 23)
The aircraft included in the two reports are brought together in Table 2 which lists their
data of more immediate interest here such as CD0 and (L/D)max. Also included are cleanness
ratio (CR) and propeller efficiency η, the latter being listed because those of its values above
0.80 seem rather high when compared with later evaluations. As to cleanness ratio, the
Heinkel, Caudron, Comet and Falcon lead the field with the Spitfire the best of the newer six
military aircraft. The Heinkel He. 70 with its extraordinarily low value of CD0 is the Kestrel
engine version bought by Rolls-Royce. Since this loomed large in RAE thinking at this time,
the subsequent extensive investigations into its performance are discussed separately in
Appendix 5. As will emerge presently, the Hurricane listed in Table 2 is the prototype. It is
likely that the Spitfire listed is also the prototype since its engine is stated to be the Rolls-
Royce PV 12.
5.2 Drag measurement by flight testing
Flight testing of the British aircraft was conducted either at the RAE or the Aeroplane and
Armament Experimental Establishment (A&AEE), Martlesham Heath near Ipswich, or, in the
case of the two flying boats, the Marine Aircraft Experimental Establishment, Felixstowe.
Except for the Heinkel assessed by Rolls-Royce, data for foreign aircraft were obtained from
reliable foreign sources.
As equation (7) indicates, aircraft having low values of CD0 and moderate to high aspect
ratios benefit in terms of (L/D)max. However, the latter’s values listed in Table 2 are
probably a little high. In evaluating k the wing alone was considered, taking account of
planform shape, taper ratio and the biplane factor as appropriate. Thus for the complete
aircraft it is likely that k would be slightly higher and this, by equation (7), would reduce
slightly the values of (L/D)max.
(23)
As indicated in Table 2, the second report states that the engine powers for the Hurricane,
Spitfire and Battle are suspected as being rather high. This, in conjunction with the perhaps
too high values of η, suggests that the propeller thrusts for these three aircraft are too high, in
15
which case one would expect CD0 to be rather lower. On the other hand, as yet the thrust
effects of ducted radiators have not been included in the assessments, effects which would
raise CD0 values. Perhaps in these cases the engine power over-estimation and the
augmentation effects have, by chance, cancelled each other out since the later, more accurate
CD0 values for the Hurricane and Spitfire are essentially those given in Table 2.
Table 2 Figures of merit from flight test measurements

Propeller Zero lift drag Cleanness
(L/D)max
efficiency η CD0 ratio CR
Airspeed Courier (1933) 0.755 0.0256 0.390 15.5
Avro 652 (Cheetah V) (1935) 0.755 0.0266 0.413 14.4
Avro 652 (Cheetah VI) (1935) 0.78 0.0262 0.424 14.5
Burnelli UB 14A (1934) 0.74 0.0252 0.297 14.9
DH 88 Comet (1934) 0.83 0.0197 0.531 19.0
Gloster Gauntlet (1933) 0.82 0.0314 0.329 10.0
Gloster Gladiator (1934) 0.825 0.0260 0.349 10.5
Heinkel He 70 (1932) 0.82 0.0150 0.540 18.5
Hendy Heck (1934) 0.765 0.0260 0.394 13.5
Martin B. 10 (1932) 0.77 0.0286 0.321 14.3
Caudron 460 (1934) 0.850 0.0215 0.564 15.3
Miles Falcon (1934) 0.79 0.0228 0.520 15.6
Miles M.6 Hawcon (1935) 0.80 0.0240 0.435 14.5
Short R 24/31 (1933) 0.72 0.0540 0.212 10.1
Short Singapore II (1930) 0.76 0.0514 0.341 8.61
Hawker Hurricane (1935)* 0.855 0.025 0.350 14.0
Supermarine Spitfire (1936)* 0.855 0.020 0.430 15.0
Fairey Battle (1936)* 0.79 0.023 0.380 14.5
Bristol Blenheim (1935) 0.825 0.026 0.350 14.0
Arm. Whit. Whitley (1936) 0.810 0.025 0.360 13.0
Handley Page Harrow (1936) 0.810 0.032 0.350 13.0
* Engine Power Estimates may be 5 – 10% too high

Source: RAE Staff, References 22, 23
5.3 Flight test and wind tunnel comparisons

(23)
The second report includes a graph (Figure 5) showing a comparison of data obtained
from flight tests and the Compressed Air Tunnel (CAT) at the National Physical Laboratory
(NPL). The CAT was capable of reaching significantly higher Reynolds number (Re) values
than more conventional wind tunnels. In Figure 5 the horizontal ordinate is Re based on
16
wing chord, whereas the vertical ordinate is the coefficient Cf. Cf is the drag coefficient at
zero lift, CD0, but based on the aircraft’s wetted area rather than its wing planform area S. In
(18)
Jones’s notation (see Appendix 3), E being the wetted area, Cf = S CD0 / E. Figure 5’s
lowest continuous curve is the result for turbulent boundary layers on flat plates similar to
C.A.T. TESTS.
FLIGHT TESTS.
COMPARISON OF C.A.T. AND FLIGHT DRAG MEASUREMENTS

Figure 5 Modified drag coefficient Cf against Re. Source, Reference 23
(18)
that used by Jones . The other curves are the CAT results for models of the Hurricane,
Battle and the BMW-powered version of the Heinkel He. 70. The data points for the aircraft
indicated are obtained from flight tests. These are at significantly higher Re values than those
achievable even in the CAT and the figure thus illustrates the difficulty in drag prediction at
full scale using wind-tunnel data. The wind-tunnel curve for the Heinkel might extrapolate
quite well to the data point for the full-scale aircraft, but the same cannot be said for the
Hurricane and Battle curves. As described in Section 3 of Reference 2, it was this difficulty
(24)
which Collar addressed in 1940 in his performance comparison of the Hurricane and the
Spitfire. His essential point was that it is crucial to compare like with like when relating CAT
data with full-scale results. Once various items of parasite drag had been deleted from his
drag assessments of the full-scale aircraft, so that the model and full-scale aircraft were
similar, the CAT results extrapolated quite well to the full-scale results.
To return to the cleanness ratio, CR, it is interesting to use its values together with those for
CD0 in Table 2 to obtain estimates for CD0. Using equation (10) and taking the cases of the
best of the monoplanes and biplanes, the Heinkel and the Gladiator respectively, the
following results are obtained:
17
Heinkel He. 70 CD0 = 0.008 Gladiator CD0 = 0.009.
(18)
As indicated in Appendix 3, these are close to the value Jones would have obtained (CD0 =
0.0086) had he not allowed a significant increase to boost his flat plate turbulent boundary-
layer drag calculations. The above two results suggest that the RAE Aerodynamic Staff were
allowing no such increases but were sticking rigorously to the turbulent boundary-layer curve
of Figure 5 in their estimation of boundary-layer drag. Three points are worth raising here.
Firstly, how accurate was this turbulent boundary-layer drag assessment method based on flat
plate data? The answer, shortly to be given, was that it underestimated this drag contribution.
Secondly, as Figure 5 illustrates, CD0 estimation based on maximum power and speed data
provides a snap-shot value at only one Re value. But from take-off to maximum speed an
aircraft’s speed can increase by a factor of five or six; the aircraft thus traverses a corresponding
range of Re values over which the boundary-layer drag contribution to CD0 will vary. Thus
some variation in CD0, albeit slight, will occur over an aircraft’s speed range due to this Re
effect. Finally, despite possible inaccuracies in boundary-layer drag estimation, the CR values
for many of the aircraft listed in Table 2 are very low. Thus it was perceived that there was a
major problem with regard to parasite drag and its prediction.
5.4 Sources of parasitic drag
In August 1938 Morgan (Figure 6) used the

assessments provided by the two RAE
(22, 23)
reports in an attempt to pin down the
(25)
sources of parasite drag . For this he
estimated the drag contributions from
various aircraft parts for five sample
aircraft. Four of these, the Handley Page
Harrow, Armstrong Whitworth Whitley,
Gloster Gladiator and the Hawker
Hurricane, have similar CR values, while
the fifth, the Heinkel He. 70 with its larger
CR value chosen for comparison, served as
his yardstick for cleanness. This exercise
(25)
was prompted, as Morgan puts it, “to
Figure 6 Morien Bedford Morgan FRS
find, if possible, the reason for the (1912 – 1978).
unexpectedly low cleanness efficiency of Source: Royal Aeronautical Society
the Hurricane prototype”, thus confirming (National Aerospace Library)
that it is this aircraft which appears in Table
2. The total drag in each case is the residual drag which remains after the induced drag has
been deducted. His results are listed in Table 3, his drag categories, he admits, being rather
arbitrarily chosen.
(25)
Morgan drew on 45 data sources (31 RAE and ARC reports, 12 NACA reports and 2
others) to estimate the various drag subtotal contributions listed in the first part of Table 3. In
additional data tables not reproduced here, he provided more detailed drag estimates for
18
individual items contributing to these subtotals. The more detailed individual drag data for
(24)
the Hurricane will be used later so as to compare his results with those of Collar . For the
(25)
estimation of skin-friction drag listed in the second part of Table 3, Morgan again used
flat plate turbulent boundary-layer data in conjunction with his estimates of surface wetted
area. This procedure he applied to the wings (and wing engines), body (and nose engine), tail
unit and undercarriage, struts and wires as appropriate.
As was usual in Britain by then, Table 3 lists the various drag contributions reduced to the
drag in lbs at a speed of 100 ft/s at sea level. Morgan’s detailed calculations of the various
skin-friction contributions to profile drag result in slight changes to the data listed in Table 2.
The Heinkel, in particular, emerges with an even higher cleanness ratio.
Table 3 Drag Contributions in lbs at 100 ft/s
Armstrong Hawker
Estimated Drag Heinkel Handley Gloster
Whitworth Hurricane
lbs at 100 ft/s He 70 Page Harrow Gladiator
Whitley Prototype
Engine Installation 6.8 50.0 57.0 10.0 15.4
Wings 27.8 104.0 121.0 23.5 32.8
Body & Cabin 17.2 80.0 62.5 11.3 14.5
Tail Unit 7.2 25.0 27.5 5.0 5.6
Excrescences 1.0 20.0 11.6 0.3 4.0
Undercarriage 4.0 63.9 9.7 0 17.5
Struts & Wires 0 4.0 1.4 0 6.9
Interference 1.0 12.0 9.0 2.0 4.0
Remainder 7.3 51.1 76.3 22.4 8.5
Total Residual Drag 72.3 410.0 376.0 74.5 109.2
Turbulent Skin Friction (Flat Plate)

Wings + wing engines 24.4 71.5 83.3 16.2 23.2
Body + nose engine 11.1 42.0 29.0 5.94 6.5
Tail unit 6.0 20.1 22.0 3.98 4.5
Undercar, struts, wires 0.2 10.1 1.1 0 4.0
Total skin friction 41.7 143.7 135.4 26.12 38.2
Cleanness Ratio 0.576 0.35 0.36 0.35 0.35
Source: Morgan, Reference 25
In his assessments of the two aircraft having fixed undercarriages, the Harrow and the
(25)
Gladiator, Morgan noted that these features account for about 16% of the residual drag.
However, he was particularly concerned with the item listed as ‘Remainder’ in Table 3. As a
percentage of the total residual drag, this is particularly high for the Hurricane (30%), whereas
for the Gladiator it is the least of the five at about 8%. He put this down to features the effects
19
of which were as yet not known but he suspected that surface irregularities together with leaks
around access panels and, for the bombers, bomb doors were probable culprits. He
recommended that the Hurricane be tested in the RAE’s 24 ft Open Jet Tunnel so that more
accurate data might be obtained. This was done and the detailed results proved most useful
(24)
in Collar’s investigation two years later.
5.5 Thrust from exhausts and radiators and other improvements
A number of improvements in the assessment procedures appeared around this time. In 1935
(26)
Meredith had introduced the idea of the ducted radiator for liquid-cooled engines and by
(27)
1937 Capon had provided an analysis from which the additional thrust produced by
radiator heat regeneration could be calculated. In March 1940 Hartshorn, Diprose and
(28)
Patterson provided an assessment of the thrust available from rearward-directed exhaust
momentum. Thus the thrust augmentation produced by these two means could be added to
the propeller thrust (ηP/V) provided by the engine. As the total thrust is equal to the total drag
in steady level flight at constant speed, a more accurate drag assessment could then be
obtained from flight-test thrust results.
Thus far the method of boundary-layer drag estimation had relied on that introduced by
(18)
Jones , being based on flat plate turbulent boundary layer data. This was seen to be
inaccurate as a predictor of boundary-layer drag since it took no account of velocity
variations over an aeroplane’s surfaces; in the flat plate case, such variations are absent.
Moreover, it was realised that the boundary layer creates a form of drag additional to skin
friction. This is caused by the slower moving boundary layer’s gradual thickening as it
develops over a surface. Thus the boundary layer ‘shoulders aside’, or outwardly displaces
slightly, the flow exterior to it. The exterior flow ‘sees’ a body shape slightly thicker than the
actual body, and this ‘seen’ body extends into the wake. The consequence is that the pressure
at the trailing edge does not return to the value it would have reached had the boundary layer
been absent and this creates the fore-and-aft pressure imbalance called boundary-layer
pressure drag. Again this effect is absent in the case of the flat plate held at zero incidence,
there being no streamwise force due to pressure on such a surface. Both of these deficiencies
in the flat plate method were rectified in the turbulent boundary-layer analysis developed by
(29)
Squire and Young (Figures 7 and 8) in 1937, more details of which are given in Reference 2.
The method gave good agreement with experiments on aerofoil sections and thus provided
(30)
more reliable drag estimates for wings. By 1939 Young had extended the method to
bodies of revolution which could be applied to fuselages.
(29)
An important feature of the Squire and Young analysis is that it shows a clear relationship
between the additional boundary-layer pressure drag and a wing’s thickness/chord ratio, t/c.
(31)
Squire emphasised this feature in the discussion following Relf’s lecture to the Royal
(32)
Aeronautical Society in 1938. Young , in reviewing Reference 29’s method, pointed out
that a good approximation to its results is
(Boundary-layer pressure drag)/(Total boundary-layer drag) ≈ t/c. (11)
20
The total boundary-layer drag is the sum of the skin-friction and pressure drags. Drawing on
(32)
Reference 30’s results, Young added that for bodies of revolution, maximum diameter d
and length l, the effect is rather less:
(Boundary-layer pressure drag)/(Total boundary-layer drag) ≈ 0.4d/ l. (12)
These results, in conjunction with the CAT investigation into drag for aerofoils having higher
(18)
t/c ratios mentioned in Section 4, corroborated Jones’s belief that aerofoil drag increases
with increasing thickness/chord ratio.
Figure 7 Herbert Brian Squire FRS Figure 8 Alec David Young FRS (1913 – 2005)
(1909 – 1961). Copyright: Godfrey Source: Royal Aeronautical Society
Argent Studio (National Aerospace Library)
Joining the Aerodynamics Department of He joined the Aerodynamics Department of the
the RAE in 1934, he moved to the RAE in 1936, moving to the College of
Aerodynamics Division of the National Aeronautics, Cranfield, in 1946. In 1950 he
Physical Laboratory in 1949. He became became Professor of Aerodynamics there and, in
Zaharoff Professor of Aviation at Imperial 1954, Professor of Aeronautical Engineering at
College, London, in 1952. Queen Mary College, University of London, until
retirement in 1978.
5.6 Drag of the Spitfire and Hurricane
Two early beneficiaries of these advances in thrust and drag estimation were Hufton (Figure
(33)
9) and Collar (Figure 10). Hufton’s report of April 1940 compares the drag of a standard
Spitfire with that of the High Speed Spitfire intended for an attempt at the Landplane World
(24)
Speed Record. Collar’s comparison of the Spitfire and Hurricane emerged in June 1940
21
and its findings are described in detail in Section 3 of Reference 2. For ease of reference,
(24) (25)
Collar’s data are repeated in Table 4A for comparison with Morgan’s Hurricane data
(33)
on the one hand and Hufton’s Spitfire data on the other.
(25)
As mentioned earlier, Morgan included more detailed drag data for individual airframe
parts which contributed to his subtotal categories listed in Table 3. These more detailed data
have been re-assigned here in an attempt to make them fit the subtotal categories listed in
(24) (33)
Table 4A, the latter categories being those chosen by Collar and Hufton . Even so,
Figure 10 Arthur Roderick Collar FRS (1908 –

1986) Source: Royal Aeronautical Society
Figure 9 Philip Arthur Hufton (1911 – (National Aerospace Library)[Royal Society]
1974) Source: Farnborough Air Sciences He joined the Aerodynamics Department of the
National Physical Laboratory in 1929, transferred to
Trust
the Structural and Mechanical Engineering
He joined the Aerodynamic Flight of the Department of the RAE in 1940, there to work on
RAE in 1934 and rose to be Deputy Director aeroelasticity. In 1946 he became Sir George White
(Aircraft). Professor of Aeronautics at Bristol University until
retirement in 1973
there are anomalies. For example, Morgan’s item listed as Profile drag (wings, fuselage and
tail) may have included Cooling Drag and Windscreen. As to Induced drag, this is unknown
(25)
since Morgan’s data are for residual drag only. The various contributions to Thrust are
(23)
also unknown, not having been listed in the original RAE report . However, it is known
that the Hurricane prototype did not possess ejector exhausts so that a value of zero is given
in that case. Similarly, the prototype had a retractable tailwheel so here again a zero value
can be assigned. As to gun holes, the prototype lacked these but did possess an aerial post, so
(25) (24)
this might explain the difference between Morgan’s assessment and that by Collar for
22
what was presumably an early production Hurricane. As to the high drag not accounted for
(25) (24)
by Morgan , about half of this could be what Collar attributes to leaks.
(23) (25)
It will be recalled that the data on which Morgan based his Hurricane assessment were
suspected of having rather too high values for both engine powers and propeller efficiencies,
but neglected to include radiator thrust. These, it is felt, may have cancelled each other out,
the result being that Morgan’s CD0 value is very close to that obtained from Collar’s more
(24)
detailed assessment as shown in Table 4B. Morgan’s CR and CD0 values quoted there,
based on flat plate drag data, are those listed in Table 3. However, the corresponding values
(24)
calculated from Collar’s Hurricane assessment are significantly higher, based as they are
(29) (30)
on the more accurate methods of Squire and Young and Young . Broadly speaking,
then, the aircraft assessed earlier were a good deal aerodynamically cleaner than the RAE had
hitherto supposed.
Table 4A Thrust/Drag Contributions in lbs at 100 ft/s

High speed
Hurricane Hurricane Spitfre Spitfire
Spitfire 1938
Morgan (25) Collar (24) Collar (24) Hufton (33)
Hufton (33)
Thrust
Engine/propeller thrust ? 73.0 51.5 52.15 44.95
Exhaust Thrust 0 9.0* 7.5* 6.35 6.75
Heat Regeneration Thrust ? 1.7 1.6
Total Thrust ? 82.0 59.0 60.2 53.3
Drag
Induced drag (including washout) ? 4.0 3.0 3.0 1.2
Residual Drag
Roughness (including rivets, joints,
--- 3.3 2.0 2.0 ---
paint)
Cooling drag (including oil cooler) --- 8.0 7.0 7.0 6.0
Air intake 1.5 1.0 1.0 1.0 1.0
Tail wheel or skid 0 2.0 2.0 2.0 1.0
Tailplane protection --- --- 0.3 --- ---
Gun holes, aerial post 0.3 0.8 0.8 0.7
Windscreen --- 1.5 1.2
Leaks (including control gaps) --- 11.0 5.0 1.2 0.9
Wing-body interference 2.0 9.9 4.5 1.5 1.5
Not Accounted For 22.4 --- --- 8.4 12.9
Total Drag ? 82.0 59.0 60.2 53.3
* Exhaust & Heat Regeneration Combined
23
(24) (33)
Neither Collar nor Hufton takes the further step of calculating CD0 but this has been
done here and the results are given in Table 4B. Included are the values for CR and CD0
together with aspect ratio, A. For want of precise values for the factor k in equation (7), that
½
equation is used merely to calculate k (L/D)max and thereby provide an indication of the
maximum lift to drag ratio.
Table 4B Derived Results
High speed
Hurricane Hurricane Spitfre Spitfire
Spitfire 1938
Morgan (25) Collar (24) Collar (24) Hufton (33)
Hufton (33)
Wing Span/ft 40 40 37 37 33.5
2
Wing Planform Area S/ft 264 264 242 242 232
Heat Regeneration Thrust ? 1.7 1.6
Aspect Ratio A 6.06 6.06 5.66 5.66 4.84
CR 0.35 0.52 0.58 0.56 0.54
CD0 0.009 0.013 0.012 0.011 0.010
CD0 0.025 0.025 0.020 0.020 0.019
½
k (L/D)max 13.8 13.8 14.9 14.9 14.1
(24) (33)
Turning now to the Collar and Hufton data in Table 4A for the standard Spitfire, these
(24)
are largely in agreement. As Collar explains, a number of the minor items (Air intake,
Tailwheel, Gun Holes) have been obtained from RAE tests on a Hurricane in the 24 ft Open
Jet Tunnel and such features are sufficiently similar to those on the Spitfire to justify
(33)
applying their results to that aircraft; presumable Hufton took the same view. The main
(24)
differences between the two assessments are that Collar has higher Leak and Interference
values but no Drag unaccounted for, whereas a value for the latter appears in Hufton’s
(33) (33)
assessment . However, Hufton allows that this item might be due to leaks but also to
such other factors as propeller-body interference, compressibility loss, errors in performance
measurements, thrust estimation and the boundary-layer transition point having been assumed
(24)
to lie at wing leading edges and fuselage noses. The latter is also assumed by Collar on
the grounds that turbulence in the propeller wash will infect the fuselage and the inboard
parts of the wing whilst gun holes will have a similar effect on much of the wings’ outboard
(24)
parts. A further slight difference between the two reports is that Collar combines exhaust
(33)
and radiator thrusts whereas Hufton lists them individually. Indeed, his is even more
detailed in that he provides corrections to engine power due to the back pressure of the ejector
exhausts. These corrections have been obtained from simulated altitude tests at 18,500 ft on
a Merlin II conducted by the RAE’s Engine Department. That report, dated May 1939,
shows that the power reduces from 1057 hp to 1033.5 hp. Martlesham Heath’s report of
January 1939 for flight tests with Spitfire K 9787 at the above altitude gives the maximum
(34) (24)
speed as 362.5 mph. This aircraft, K9787, is the first production Spitfire I . Collar is
less specific, stating a power of 1020 hp at 18,000 ft, at which altitude the maximum speed is
(33)
365 mph, the Hurricane’s speed being 40 mph lower. For propeller efficiency, Hufton
provides a detailed assessment of losses for the twin-bladed wooden Vickers Jablo unit which
(24)
results in a value for η of 0.77. In this respect, although Collar has consulted Hufton’s
(33)
report , he opts for a η value of 0.8 but does not state the type of propeller used. As
24
mentioned earlier, these values for η call into question the accuracy of those higher values
(22, 23)
listed in Table 2 taken from earlier reports .
Nonetheless, Collar’s and Hufton’s CD0 values listed in Table 4B for the standard Spitfire are
virtually identical and have very similar CD0 values. As to cleanness, the CR values are
(24)
therefore closely similar and both slightly better than that for the Hurricane. Collar points
out that the Hurricane’s higher Profile drag, 40.5 lbs compared with the Spitfire’s 32.2 lbs, is
not only due to its larger size and therefore greater wetted area but also to its wing’s higher
t/c ratio.
(24)
The mysterious item labelled Tailplane protection in Collar’s assessment has recently
(35)
been investigated by Brinkworth . He concludes that this was probably a small guard
fitted to the top of the fin, not the tailplane. Its purpose was to prevent the cable of an anti-
spin parachute, if deployed in an emergency, slipping between the fin and the rudder and thus
jamming the rudder.
(33)
Hufton’s assessment of the High Speed Spitfire is again meticulous, particularly with
regard to thrust estimation. The aircraft has a much more powerful engine than the standard
Spitfire and the RAE Engine Department has supplied him with power estimates at an
altitude of 3,000 ft for three speeds, namely 300 mph, 350 mph and 400 mph. At the highest
speed, and after corrections for exhaust back pressure, the power is estimated to be 1941 hp.
The propeller is a Watts four-bladed fixed-pitch unit of smaller diameter than that used on the
(33)
standard Spitfire which is therefore less prone to compressibility losses. For this Hufton
calculates a η value of 0.78. His drag estimates are calculated for all three speeds but his
results for the 400 mph case only are listed in Table 4A for brevity’s sake. At the lower
speeds his results for Profile and Induced drags differ slightly. In the former case this is due
to the boundary-layer drag obtained using References 29 and 30 varying with Reynolds
number, the result being that the profile drag, when reduced to the 100 ft/s standard,
decreases very slightly with increasing speed. As to Induced drag, this also reduces as speed
increases because the aircraft flies at lower CL values. Although the data of Table 4A show
that a significant drag reduction has been achieved when compared to the standard Spitfire’s
results, 53.3 lbs compared with 60.2 lbs for Total drag, Table 4B shows that the reduction in
CD0 is small. This is partly due to the fact that, whilst drag is reduced, so too is the planform
area S.
At first sight it appears odd, perhaps even erroneous, that Hufton’s Engine/propeller thrusts
(Table 4A) turn out to be higher for the standard Spitfire than for the High Speed Spitfire
when the latter clearly has a far more powerful engine. This is resolved, however, when
account is taken of Appendix 2’s procedure used to reduce flight data at high speed and
altitude to sea-level conditions at a speed of 100 ft/s. Equation (A2.10) is the engine-thrust
3
reduction relation and the crucial term there is P/(σ V ). The power P increases from 1033.5
hp (standard Spitfire) to 1941 hp (High Speed Spitfire) whereas the increase in V from 362.5
mph to 400 mph, even when cubed, is far less of a compensatory increase. A far larger
effect, however, is produced by the density ratio σ = ρ/ρ0, ρ0 being the sea level air density.
For the standard Spitfire at 18,500 ft σ = 0.56 whereas for the High Speed Spitfire at 3,000 ft
25
3
σ = 0.92. It is largely due to σ’s influence appearing in the denominator of P/(σ V ) which
compensates for the lower P of the standard Spitfire and thus results in its higher
Engine/propeller thrust recorded in Table 4A.
(33)
Before leaving the Collar (24) and Hufton assessments, it is interesting to compare their
(5)
results with those which would have been obtained using Loftin’s approach in which thrust
augmentation effects are ignored. Using Collar’s Hurricane data in Table 4A as an example,
according to Loftin’s method the thrust would be solely 73 lbs and this would be taken to be
the total drag. Deducting the Induced drag of 4 lbs leaves a Residual drag of 69 lbs, not
Collar’s value of 82 lbs, and this would yield the lower CD0 value of 0.021. This exercise
(5)
applied to Collar’s Spitfire data gives the lower value of 0.017. Thus the Loftin approach
for cases in which thrust augmentation is significant underestimates CD0 by about 15%.
By 1940 the RAE’s drag assessment method had reached a reasonably high level of accuracy.
(33)
Notably in Hufton’s case , this had resulted in a data layout similar to an accountant’s
company balance sheet. On the Credit side, engine thrusts corrected for back pressure and
propeller efficiency could be added to thrusts provided by the exhausts and ducted radiators
so as to provide reasonably accurate values for total thrust. That total Credit must then be
balanced against the Debit total produced by the various forms of drag. Prediction of the
(29) (30)
latter had improved significantly with the use of the Squire and Young and Young
profile drag analyses and the increasing amounts of more accurate wind tunnel data for
various contributors to parasite drag.
5.7 Drag of bombers

(36)
All this was put to use by Bottle in his RAE report of November 1940 which assessed
bomber performance. The aircraft assessed are the Avro Manchester (both 80 ft and 90 ft
span versions), Handley Page Halifax, Short Stirling, Douglas DB 7 (Boston in RAF service),
Glen Martin 167 (Maryland in RAF service), Heinkel He. 111 and Junkers Ju. 88. For
(36)
comparison, Bottle also includes recent data for the Vickers Wellington 1A, Handley
Page Hampden and Armstrong Whitworth Whitley. The newer British aircraft, the two
Manchesters, Halifax and Stirling, are all prototypes and therefore probably not
(36)
representative of subsequent Service aircraft. For example, Bottle states that the 90 ft
span Manchester has the turrets faired. All four aircraft had been tested by the A&AEE in
1940 after its move at the war’s outbreak from Martlesham Heath near the East Coast to the
safer location of Boscombe Down near Salisbury. The A&AEE’s tests of the Glen Martin
167 were not yet complete. Flight-testing of the other British aircraft, and probably the
Douglas DB 7, had been conducted at Martlesham Heath. There were no British performance
(36)
tests available for the two German aircraft. The Heinkel, Bottle believes, was a standard
aircraft, presumably a captured example although it is unclear if it was in a flyable condition.
(36)
As to the Junkers, Bottle comments dryly that “no complete machine was available for
inspection”. His further purpose with the two German aircraft and the Glen Martin was not
only to provide drag assessments but also to use their data to estimate maximum speed, the
latter being unknown.
26
(36)
Bottle’s thrust and drag data for the four newer British aircraft are given in Table 5A.
Table 5B shows the data for the two American and the two German aircraft. In both tables
(36)
only the total Engine installation data are given whereas Bottle lists individual drag items
such as those due to air intakes, exhausts and coolers both internal and external. The drag
item labelled Roughness is due to camouflage paint. For the British aircraft, the propellers
are de Havilland constant speed units, the metal blades being of thin section, and for these
(36)
units Bottle takes η to be 0.82 whereas for the others η = 0.8 is assumed.
Table 5A Thrust/Drag Contributions in lbs at 100 ft/s

Manchester Manchester
Halifax Stirling
(span 80 ft) (span 90 ft)
1939 1939
1939 1940
Thrust
Engine/propeller thrust 353 396 450 537
Exhaust Thrust 21 22.5 20 18.5
Heat Regeneration Thrust 6.5 7 8 0
Total Thrust 380.5 425.5 478 555.5
Drag
Induced drag 65 75 38 83
Residual Drag
Wings: Profile drag 96 106 127 131.5
Roughness 14 16 24 26
Controls 7 7 7 9
L. E. slats --- --- 15 ---
Fuselage: Profile drag 36 36 37.5 46
Roughness 2.5 4.5 3 3
Cabin 2 2 2
Rear turret 19 17 17 19
Front turret 3 3 3 3
Tail: Profile drag 26 26 30.5 26.5
Extra fin 2 2 --- ---
Roughness 7 7 6 5.5
Controls 8 8 7 11
Wing-fuselage interference 5 5 5 5
Engine installation 59 59 100
Miscellaneous: Wireless mast(s) 2 2 3 1.5
Pitot 0.5 0.5 1 0.5
Small protuberances 8 8 5 ---
Leaks 20 20 18
Not Accounted For -1.5 21.5 29 57
Total Drag 380.5 425.5 478 555.5
Total Residual Drag 315.5 350.5 440 472.5
Source: Bottle, Reference 36
27
(36)
Table 5C lists CD0 and cleanness ratio CR for all of the aircraft considered in Bottle’s report .
Here the CR values are those obtained using the boundary-layer profile drag analyses of
(36)
References 29 and 30. In his report, Bottle adds values based on the older flat plate
analysis though these are not included here. The bomber CD0 values are higher than those for
the fighters given in Table 4B due to the larger amounts of parasite drag created by engine
nacelles, gun turrets and such. Consequently, the CR values are smaller. Using the values
½
for CD0 and aspect ratios, A, in conjunction with equation (7), values for k (L/D)max have
been calculated and are listed in Table 5C. The two American aircraft, Boston and Maryland,
benefit here due to lower CD0 values and higher aspect ratios.
Table 5B Thrust/Drag Contributions in lbs at 100 ft/s

Douglas DB 7 Martin 167 Heinkel He. Junkers Ju.
(1939) (1939) 111H (1939) 88A (1938)
Thrust
Engine/propeller thrust 141 156 268.5 251.5
Exhaust Thrust 0 0 0 2.5
Heat Regeneration Thrust --- --- 3.5 ---
Total Thrust 141 156 272 254
Drag
Induced drag 7 9.5 31.5 46
Residual drag
Wings: Profile drag 43 46.4 79.5 48
Roughness 3 3.7 8.5 4
Controls 3 2 4 7
Fuselage: Profile drag 14 12.6 22.5 16
Roughness 1 0.9 2 1
Cabin 4 4 2 6
Upper turret/rear cockpit --- 3 15 ---
Lower gun position --- 4 9 6
Tail: Profile drag 14 13.7 20 14
Roughness 1 1.1 2.5 1
Controls 3.5 3 6 4
Wing-fuselage interference 2 4 2 5
Engine installation 36 36 51 48
Miscellaneous: Wireless mast(s) 0.5 1.2 1.5 1.5
Pitot 0.5 0.5 1 0.5
Small protuberances 1 4 3 6
External bombs --- --- --- 32
Leaks 8 6.4 12.5 8
Not Accounted For -0.5 0 0 0
Total Drag 141 156 272 254
Total Residual Drag 134 146.5 240.5 208
28
Table 5C Zero lift drag coefficients and cleanness ratios

Cleanness Aspect ½
CD0 k (L/D)max
ratio CR ratio A
Avro Manchester (80 ft) 0.0255 0.51 6.15 13.8
Avro Manchester (90 ft) 0.0265 0.49 7.28 14.7
Handley-Page Halifax 0.0293 0.45 7.76 14.4
Short Stirling 0.0266 0.44 6.70 14.1
Vickers Wellington 1A (1936) 0.0291 0.45 8.84 15.4
Handley-Page Hampden (1936) 0.0271 0.48 6.95 14.2
Arm. Whit. Whitley 0.0259 0.46 6.21 13.7
Heinkel He 111 0.0220 0.51 5.90 14.5
Junkers Ju 88 0.0320 0.38 6.51 12.6
Douglas DB 7 (Boston) 0.0243 0.53 8.09 16.2
Martin 167 (Maryland) 0.0232 0.50 11.40 19.6
(36)
It will be recalled that Bottle had no flight-test performance data for the Martin 167
Maryland and the two German aircraft but one purpose of his investigation was to estimate
their maximum speeds. However, it is not entirely clear how he achieved this. It may be that
he used his assessment procedure in reverse, so to speak. Using the analyses of References
29 and 30 and the known geometries of the aircraft, together with reasonable estimates for
their flight Reynolds numbers, he would be able to estimate profile drag. He would also have
sufficient data to make reasonable estimates of the parasite drag contributions listed in Table
5B. The resulting estimate for the total drag must then be equal to the total thrust. The latter
is then corrected for thrust augmentation effects and propeller efficiency (η he assumes to be
0.8). The engine thrust can then be calculated and from the engine’s maximum power, which
is known, an estimate for the maximum speed follows. His maximum speed estimates are
listed in Table 5D together with published performance data for the three aircraft. However,
there is a difficulty with regard to the latter in that their stated altitudes do not match those
(36)
chosen by Bottle . The published data chosen here are those which this author has to hand
(37)
and provide the closest match to Bottle’s altitudes. For the Martin 167, data for the
(36)
RAF’s Maryland I are given in Table 5D. For the Heinkel, Bottle states that the engines
are Jumo 211D units. This suggests that the Heinkel variant is the 111H with the glass-nosed
(38)
cockpit and it is its maximum speed which is listed in Table 5D. As to the Junkers, this
(38)
has Jumo 211B engines suggesting the 88A variant, data for which are also given in the
table. Comparison of these results suggests that Bottle’s estimates are reasonably accurate.
The biggest difference is that for the Junkers, but it must be pointed out that for this aircraft
(36)
Bottle includes in the data of Table 5B the drag of six 250 kg bombs carried externally, a
drag addition which contributes about 13% to the total.
Table 5D Estimated maximum speeds

Bottle (36) Cooper and Thetford (37) Green (38)
Martin 167 Maryland 297 mph @ 10,000 ft 298 mph @ 7,500 ft
Heinkel He 111H 254 mph @ 13,000 ft 255 mph @ 13,120 ft
Junkers Ju. 88A 266 mph @ 15,000 ft 280 mph @ 18,050 ft
29
5.8 Collected drag and thrust data
As the war years progressed and more new aircraft entered British service, performance data
obtained at Boscombe Down and at the RAE itself were passed to the RAE’s Aerodynamics
Department. When analysed there, these yielded a considerable quantity of basic
aerodynamic data. These the RAE began to issue in reports covering both aerodynamic and
structural data for the aircraft assessed. This was a joint effort by the Aerodynamics
Department and the Structural and Mechanical Engineering Department (SME). The SME’s
interest was in assessing various structural efficiency factors, an exercise which deserves a
survey in its own right. Here, however, we are interested only in the Aerodynamics
Department’s assessment of drag, particularly with regard to the estimation of CD0 using the
(33) (36)
methods used by Hufton and Bottle described above.
(39)
The first issue of the report , although undated, probably appeared in 1942. An addendum
of January 1943 corrected data for the Halifax. In November 1943 Issue 2 of the report
appeared, stating an intention to provide updates at intervals of six months to one year. Issue
3 appeared in August 1945, an addendum was added in October of that year giving data for
three British jet aircraft and this was followed by a further addendum in April 1946
correcting and adding further data.
As an indication of the detail provided by Reference 39, Table 6 lists the drag data for five
sample aircraft. Here, and in all cases listed in that reference, for simplicity the induced drag
has been estimated using equation (5) with k = 1.
With regard to the Mustang III’s low CD0 value compared with that for the Spitfire IX, both
aircraft have similar totals for Profile Drag, 29.4 lbs (Mustang) and 29.9 lbs (Spitfire),
whereas Power Plant Drag and Guns are significantly less for the Mustang. Since Power
Plant Drag includes cooling drag, it may be, as suggested in Reference 2, that the Mustang
benefits here from the greater area ratio of its radiator duct which reduces the drag of its
radiator matrix. As to Guns, the Mustang is stated to have had its 0.5 in machine gun ports
sealed whereas this was not the case for the cannon-armed Spitfire. Under the Miscellaneous
heading, the Mustang again benefits from its fewer excrescences. The Spitfire’s high value
for Drag Not Accounted For may be due, in part, to overestimation of Propeller Thrust. In
(39)
this respect, Issue 3 (August 1945) of the report stresses that the accuracy with which CD0
is obtained from flight tests depends entirely on the accuracy of the estimation of engine
power. It feels that a thrust or torque meter would be a great asset in improving the accuracy
of CD0 determination, an indication that there was some doubt as to engine thrust accuracy.
(5)
It will be recalled that Loftin’s method ignores thrust augmentation and for the P-51D he
gives a CD0 value of 0.0163. Applying that method to the Mustang III data of Table 6, the
residual drag of 48.8 lbs is replaced by 41.1 lbs. The latter yields CD0 = 0.0148, an
underestimation similar to that quoted for this method’s application earlier to Collar’s Spitfire
(24)
data . However, the Mustang III and the P-51D are not strictly comparable. The Mustang
III’s nearest American equivalent is the P-51B/C variant, not the P-51D which introduced the
lower rear fuselage decking and bubble cockpit hood providing better all-round vision.
30
The maximum speed tests on both the Spitfire IX and the Mustang III had initially been
conducted at altitudes around 26,000 ft, conditions which yielded higher values for CD0.
Reference 39 states the suspicion that such high-speed tests were encountering
compressibility effects even in straight and level flight. Therefore the tests were repeated at
altitudes close to sea level at which maximum speeds were lower and the sound speed higher,
the resulting CD0 values being noticeably reduced. The above suspicion is confirmed by the
author’s calculations indicating that the high altitude tests were conducted at Mach numbers
(40)
around 0.65. Mair’s report of 1950 (see Figure 5 of Reference 2) on high subsonic flight
tests at the RAE shows that compressibility effects causing higher values for CD0 were indeed
encountered at these conditions. Therefore only the lower altitude test data for such cases are
quoted in Tables 6 and 7.
The Sunderland V is the only aircraft listed in Table 6 which is powered by radial engines. In
Reference 39 it was recognised that the interior ducting of such engine cowlings would produce
heat regeneration thrust in similar fashion to that of a liquid-cooled engine’s ducted radiator.
(27)
In estimating this effect for aircraft with radial engines Capon’s analysis for ducted
radiators was again used but thrust results were multiplied by an efficiency factor of 0.8.
The data for the multi-engine aircraft of Table 6, Mosquito, Lancaster and Sunderland,
illustrate the consequences for Power Plant Drag inherent to such aircraft. For the two four-
engine aircraft there are additional drags due to gun turrets. Such unavoidable consequences
of their mission requirements lead inevitably to higher CD0 values and lower cleanness ratios.
(39)
All of the CD0 and other data from the various issues of the report are brought together in
Table 7. However, in Issue 2 of November 1943 the cleanness ratio (CR) results were
dropped from the report’s data tables. The author has calculated these quantities from the
detailed data in this and subsequent tables, the results being given in italics. Included in
½
Table 7 are the author’s calculations for k (L/D)max using equation (7), the results again
being given in italics.
The data in Table 7 are grouped in the categories adopted in Reference 39: single engine
fighter, single engine fighter (naval), twin engine fighter, naval torpedo & light bomber, twin
engine bomber, four engine heavy bomber, flying boat.
31
Table 6 Thrust/Drag Contributions in lbs at 100 ft/s

Spitfire IX Mustang III Mosquito Lancaster Sunderland
Propeller Thrust 65.0 43.0 NF XV
138.2 III
492.0 V
610.0
Exhaust Thrust 7.5 6.6 22.7 69.0 58.4
Heat Regeneration Thrust 1.4 1.1 4.4 7.0 4.4
Total Thrust 73.9 50.7 165.3 568.0 627.8
Induced Drag 1.4 1.9 9.8 79.2 84.0
Profile Drag Wings 19.0 17.5 39.6 116.9 147.0
Profile Drag Body 6.8 7.0 12.3 31.4 106.3
Profile Drag Tail 4.1 4.9 14.1 29.4 37.8
Power Plant Drag 15.7 13.4 32.6 100.0 85.6
Guns & Turrets 5.0 0 2.1 64.5 43.5
Radio 0.5 0.6 1.0 13.2 30.5
Roughness 1.2 1.2 11.3 19.2 31.4
Misc. Incl. Control Gaps 9.9 6.0 15.1 66.7 67.3
Drag Accounted For 62.2 50.6 128.1 441.3 549.4
Drag Not Accounted For 10.3 -1.8 27.4 47.5 39.4
Cleanness ratio (CR) 0.412 0.603 0.424 0.364 0.494
CD0 0.0252 0.0175 0.0273 0.0316 0.0293
Source: RAE Staff, Reference 39
Table 7 Drag and other data for a variety of types

CD0 Cleanness ratio Aspect ratio A k½ (L/D)max
CR
Single Engine Fighter
Hurricane I 0.0249 0.534 6.2 14.0
Hurricane II 0.0253 0.516 6.2 13.9
Hurricane IV 0.0307 0.413 6.2 12.6
Spitfire Vb 0.0213 0.518 5.6 14.4
Spitfire Vc 0.0218 0.508 5.6 14.2
Spitfire IX 0.0252 0.412 5.6 13.2
Spitfire XXI (Prototype) 0.0182 0.521 5.6 15.5
Typhoon I (Prototype) 0.0216 0.564 6.2 15.0
Typhoon I 0.0238 0.540 6.2 14.3
Tempest II 0.0185 0.565 5.6 15.4
Tempest V (Prototype) 0.0199 0.563 5.6 14.9
Mustang I 0.0179 0.624 5.9 16.1
Mustang III 0.0175 0.603 5.9 16.3
Mustang X 0.0227 0.519 5.9 14.3
Fw 190A3 0.0279 0.444 6.0 13.0
Single Engine Fighter (Naval)
Firefly III 0.0308 0.418 6.0 12.4
32
Firebrand 0.0228 0.587 6.7 15.2

Seafire IIc 0.0254 0.441 5.6 13.2
Seafire III 0.0235 0.460 5.6 13.7
Hellcat 0.0267 0.507 5.5 12.7
Twin Engine Fighter
Whirlwind (Prototype) 0.0315 0.408 8.1 14.2
Welkin I 0.0297 0.449 10.6 16.7
Mosquito IIF 0.0226 0.488 6.1 14.6
Mosquito NF X 0.0258 0.527 6.1 13.6
Mosquito NF XV 0.0273 0.424 6.1 13.2
Beaufighter I 0.0272 0.440 6.6 13.8
Beaufighter IIF 0.0266 0.450 6.6 14.0
Naval Torpedo & Light Bomber
Barracuda (Prototype) 0.0321 0.484 6.1 12.2
Barracuda II 0.0312 0.465 6.1 12.4
Beaufighter TF X 0.0354 0.349 6.7 12.2
Firebrand TF III 0.0281 0.436 6.8 13.8
Avenger I 0.0210 0.583 6.0 15.0
Twin Engine Bomber
Buckingham (Prototype) 0.0232 0.516 7.3 15.7
Mosquito XVI B 0.0294 0.395 6.1 12.8
Mosquito IX B 0.0252 0.459 6.1 13.8
Boston III A 0.0296 0.429 8.1 14.7
Mitchell II 0.0258 0.476 7.4 15.0
Ju 88C6 0.0315 0.378 7.4 13.6
Four Engine Heavy Bomber
Halifax I 0.0400 0.344 7.8 12.4
Halifax II 0.0365 0.357 7.8 13.0
Halifax III 0.0344 0.399 7.8 13.3
Lancaster I 0.0302 0.396 8.0 14.4
Lancaster II 0.0335 0.369 8.0 13.7
Lancaster III 0.0316 0.364 8.0 14.1
Lancaster VI 0.0333 0.344 8.0 13.7
Lincoln I 0.0324 0.364 10.1 15.6
Stirling 0.0314 0.374 6.7 13.0
Fortress II A 0.0253 0.467 7.6 15.4
Liberator III A 0.0407 0.355 11.7 15.0
Flying Boat
Lerwick 0.0346 0.469 7.5 13.4
Sunderland I 0.0304 0.470 7.5 13.9
Sunderland II 0.0314 0.456 7.5 13.7
Sunderland III 0.0307 0.466 7.5 13.9
Sunderland V 0.0293 0.494 7.5 14.2
Mariner 0.0299 0.518 9.9 16.1
33
The first issue of Reference 39 remarks that the drag of a production aircraft is usually higher
than its prototype. As an example, it points to the Typhoon data of Table 7. The report adds
that the drag of production aeroplanes alters during the life of the type, generally increasing
with time due to changes in operational equipment, engine developments and so on. This can
be seen in Table 7’s Merlin-powered Lancaster I, III and VI sequence (the Lancaster II was
Hercules-powered) as the aircraft acquired more parasite drag creating features such as the
blister housing H2S radar. Surprisingly, however, the reverse is the case with the Merlin-
powered Halifax I and II; the Halifax III was Hercules-powered. The effect of change in
armament is shown in Table 7’s high CD0 value for the Beaufighter TF X compared with that
for the fighter version, Beaufighter IIF. On test the Beaufighter TF X carried an 18 in Mk. XII
torpedo and the aircraft had been fitted with dive brakes which, when retracted, remained
proud of the wings.
6. The Early Jet Age

6.1 Straight wing jet aircraft
The addendum to Issue 3 of Reference 39, dated October 1945, provides test-flight data for
the Gloster E.28/39 and also for an interim version of the Gloster Meteor I and the De
Havilland Vampire. The Gloster E.28/39 is the first British jet aircraft and its evolution is
(41)
described in exemplary detail by Brinkworth . In April 1946 the data for all three aircraft
were incorporated in the final addendum’s complete data sheet. For the three aircraft the
thrust and drag estimates taken from that data sheet are given in Table 8. The engines are
stated to be Rolls-Royce W2B/23/101 (E.28/39), Rolls-Royce W2B/23-C (Meteor) and De
Havilland Goblin I (Vampire). For the E.28/39 a 4% loss of thrust is assumed due to the
length of the jet pipe.
As noted at the end of Table 8, a correction applies to the CD0 value for the Vampire. The
(39)
copy of the report seen by the author carries the following hand-written explanation:
“This drag figure includes some compressibility drag. RDT.1 and De Havilland both deduce
at low Mach numbers a profile drag of about 44 lb at 100 ft/s [CD0 = 0.0142].” “Profile drag”
here refers to the item labelled Total Residual Drag in Table 8.
Table 8 shows that the adoption of the jet engine resulted in significant drag improvements.
There are now fewer parasite drag creating features, items associated with cooling for piston
engines having been removed. Such improvements are seen to result in higher values for CR.
However, of the two fighters, the Meteor benefits the least due to its twin-engine layout.
Drag results at low and high altitudes supplied by English Electric for the Canberra B.1
(42)
(1949) were analysed by the Aerodynamic Staff in December 1950. It must be stressed
that the flight test data were obtained at a Mach number of 0.74. Nonetheless, the low CD0
values given in Table 9 at this fairly high subsonic condition are remarkable. For lower
speeds clear of compressibility effects they can be expected to be even lower and it would be
interesting to know what they are. The results have been included here since they illustrate
how much lower CD0 could become as the jet age progressed. For the Martin B-57B, the
(5)
American version of the Canberra with extended wings, Loftin gives CD0 = 0.0119 at an
34
(42)
altitude of 2,500 ft and M = 0.79. The purpose of the RAE report is to use the data in an
attempt to estimate boundary-layer transition position on the wings; in this respect the
assessment is rather inconclusive and further investigation is recommended.
Table 8 Thrust & Drag Contributions in lbs at 100 ft/s

Gloster E.28/39
Meteor I (1943) Vampire (1943)
(1941)
Engine Thrust 31.6 74.9 48.0
Induced Drag 0.8 2.1 1.7
Profile Drag Wings 9.8 23.6 17.7
Profile Drag Body 7.6 9.9 10.2
Profile Drag Tail 4.6 9.0 6.0
Power Plant Drag 0 13.0 2.0
Guns 0 1.2 1.6
Radio 0.3 0.3 0.3
Roughness 1.0 1.7 0.9
Misc. Incl. Control Gaps 5.9 8.9 5.4
Drag Accounted For 29.2 67.6 44.1
Drag Not Accounted For 1.6 5.2 2.2
Total Residual Drag 30.8 72.8 46.3
Cleanness ratio (CR) 0.714 0.584 0.732
CD0 0.0174 0.0163 0.0150*
* Handwritten Correction 0.0142
Table 9 Drag of the English Electric Canberra B.1 at M = 0.74

6 6
Reynolds number Re = 22.5×10 Re = 60×10
Altitude (ft) 40,000 10,000
CD0 0.0119 0.0107
6.2 Swept wing jets
By 1950 Britain had adopted swept wings for its high speed aircraft. In December of that
(43)
year, an RAE report by Courtney compared the low-speed drag characteristics of two
such British aircraft with those for the North American F-86A Sabre. The two British aircraft
are the Supermarine E.41/46 (Supermarine 510) and the Hawker E.38/46 (Hawker P.1052).
The Supermarine aircraft was a development of the straight-winged Attacker, to which swept
wing and tail surfaces were fitted. It retained the Attacker’s ‘elephant-ear’ intakes at the
forward sides of the fuselage, its rear jet pipe and tailwheel undercarriage. The Hawker
aircraft was a swept-winged development of the straight-winged Sea Hawk but retained that
35
aircraft’s unswept tail surfaces together with its intakes at thickened wing roots, twin jet pipes
at the trailing edge roots and nose-wheel undercarriage.
(39)
In the past, drag assessment based on flight-test results had usually used engine
power/thrust data at one known speed and altitude to determine the total drag coefficient CD.
An estimate for the induced drag coefficient CDi had then been deducted to obtain the
residual drag coefficient, the zero-lift coefficient CD0. For the three swept-wing aircraft of
(43)
Courtney’s report a new method had been used. In this the aircraft had been flown at a
number of CL values for which the corresponding values of the total drag coefficient, CD,
(43)
could be deduced. As to the latter, Courtney states that the RAE’s Aero Flight Section
estimated that, given the jet pipe pressure and temperature and an accurate test-bed
calibration of the engine, its thrust should be correct to within 2%. Thus a reasonably
accurate value for the total drag, and hence CD, can be determined from the flight-test thrust
2
results. According to equations (5) and (6), a graph of CD ~ CL should yield a straight line
variation which, on extrapolation to CL = 0, yields a value for CD0. A further benefit of this
approach is that the gradient of this graph yields a value for the induced drag factor k of
equation (5). This method therefore not only avoids the need to estimate induced drag but
also gives a value for k.
(43)
As Courtney points out, there is, however, a difficulty with this approach. Flight at
different CL values entails flying at different speeds and therefore at different Reynolds
numbers, Re. Profile drag and the other drag contributions are sensitive to Re, as illustrated
(42) (33)
by the Canberra results above. As mentioned in Section 5, Hufton had also encountered
this problem of sensitivity to Reynolds number (Re) in estimating the drag of the High Speed
2
Spitfire at three different speeds. Consequently, the resulting CD ~ CL graph is not the
(43)
desired straight line but one which is slightly curved. As Courtney explains, a straight
line can be chosen through the data points of the graph, in which case the value of CD0
obtained will correspond to some mean value of Re, and the value of k will also include the
effect of Re variation. Here it is worth adding that in wind tunnel testing, in contrast, this
difficulty can be avoided by using the same tunnel speed for all the tests at different CL
values (see below). The problem here, however, is the familiar one of wind-tunnel testing:
the consequent fixed Reynolds number is much less than that at full scale.
(43)
Commenting on the three aircraft’s drag coefficients derived using this method, Courtney
states that experience has shown that the uncertainty in determining CD0 from flight tests
usually exceeds the uncertainty of 2% within which the thrust can in theory be determined.
In view of the uncertainty regarding Re effects and the choice of the best curve through the
scattered experimental points, he suggests that an inaccuracy of as much as 5% may well
(43)
arise. His results for CD0 are given in Table 10A. From these Courtney obtains the
values for D0,100, the drag in lbs at a speed of 100 ft/s at sea level conditions.
36
Table 10A Drag of swept wing aircraft

CD0 D0,100 (lbs at 100 ft/s)
Supermarine E.41/46 (1948) 0.0164 53.5
Hawker E.38/46 (1948) 0.014 43.0
North American F-86A (1947) 0.013 44.5
Source: Courtney, Reference 43
(43)
The account sheet drawn up by Courtney for estimates of the various drag contributions to
match the total D0,100 is shown in Table 10B. These contributions have been obtained using
(39)
the analyses developed earlier . The item labelled ‘Chute housing’ for the E.41/46 refers
to the housing for an anti-spin parachute, a feature absent on the other aircraft.
Table 10B Residual Drag, lbs at 100 ft/s

Supermarine Hawker E.38/46 North American
E.41/46 Sabre F- 86A
Wing
Profile 18.85 15.5 18.45
Roughness 0 0 0
Control Gaps 0.95 0.8 0.90
Tail
Profile 5.85 5.0 6.0
Roughness 0.05 0.05 0.05
Control Gaps 1.45 1.25 1.5
Body
Profile 12.65 10.45 12.1
Roughness 0 0 0
Cabin 2.0 2.0 2.0
Miscellaneous
Intakes 4.0 2.0 1.0
Exit 0.5 0.5 0.5
Interference 2.0 1.5 1.5
Pitot 0.5 0.5 0.5
‘Chute Housing’ 0.5 --- ---
Wheel Well 1.0 ---
Total Accounted For 50.3 39.55 44.5
Total (Table 10A) 53.5 43.0 44.6
Unaccounted For 3.2 3.45 0.1
Unaccounted For % 6.0 8.0 0.2
Source: Courtney, Reference 43
(43)
Courtney judges that, so far as the comparison between flight test and estimated drags is
concerned, the unaccounted-for terms, of between zero and 8% of the total, indicate
satisfactory agreement. On past aircraft, mainly piston-engine, the unaccounted-for term is
37
on the average about 9% of the total residual drag, and he suggests that a similar percentage
may be found on jet aircraft. This he considers to show fairly good agreement between
estimated and flight-test figures having regard to the order of accuracy that can reasonably be
expected in estimating or measuring drag.
(43)
Courtney provides interesting comments on the finish of the two British aircraft. He feels
that the finish on the Supermarine is poor, the wing surfaces being, as he puts it, “somewhat
rough and there were several poor panel junctions. Some rivets were not properly filled and
others were not flush”. However, he excuses this on the grounds that it was designed and
built in a hurry, not so much to achieve high-speed flight but rather to acquire experience of
swept-wing flight as quickly as possible. As to the Hawker, no examination could be made at
the time since it had recently crashed. However, from memory he judged it to have had a
better finish, a judgement supported by his examination of the earlier Sea Hawk and the later
Hawker P. 1081 with swept tail surfaces and straight-through jet pipe. As to the F-86A, he
remarks that no example was yet available for inspection in Britain. His judgement is
therefore based on published accounts from which he merely notes that these show wing
leading-edge slats extending over most of the wing span, indicating a rather limited
assessment.
(43)
Commenting on the various intakes used on the three aircraft, Courtney judges that the
fuselage side intakes of the Supermarine are generally less efficient on the whole than the
body-nose or wing-root intakes used on the other two.
(43)
Courtney concludes with a lengthy and detailed appendix surveying the current state of
drag prediction. His opening paragraph is worth quoting in full.
“Drag estimation is by no means an exact science. It is based partly on theoretical
calculations for ideal shapes of wings and bodies, partly on such wind tunnel and flight
tests as are available on actual shapes, excrescences, leaks, roughness etc., and partly
on past experience in comparing estimated total drags with those deduced from flight
performance measurements. There is plenty of scope for error in the assumptions that
have to be made, eg. on the amount of laminar flow existing in flight, or on the general
standard of cleanliness – roughness of paint, amount of leakage, interference and so
on.”
(43)
On wing profile drag Courtney notes that this is estimated from the theoretically derived
data provided by the Squire and Young (29) method, which was conveniently summarised in
the Royal Aeronautical Society’s Data Sheets. Here it must be added that, during the war
years, the Society had established a Technical Department, its aim being to bring together
technical information from all available sources. Subsequently this became a separate entity,
(43)
the Engineering Science Data Unit (ESDU). Courtney points out that the Squire and
(29)
Young method was devised to deal with the earlier RAF and NACA aerofoil sections and
probably needed updating to cope with the high-speed sections now being used.
(43)
Commenting on fuselage profile drag, Courtney notes that assessments are based on
(30)
Young’s method which assumes an axially-symmetric body whereas few fuselages are of
38
circular cross-section; some errors can therefore be expected in this connection. As to the
remaining drag contributors, he remarks that the estimation of their drags “is largely a matter
of intelligent guesswork assisted by a woefully small amount of data derived from tunnel and
flight tests.” He adds that “External fittings and excrescences are allowed for by finding a
tunnel test on something resembling the item in question and applying a similar drag
coefficient”. The implications for further penetration into ‘the jet age’ are clear: more
advanced theoretical methods combined with extensive wind tunnel and flight tests will be
crucial to further progress.
It seems appropriate to end this survey with that most sleekly elegant of early British swept-
wing aircraft, the Hawker Hunter. Results for a 1/5 scale model of it tested in the RAE’s 11.5
(44)
ft tunnel are contained in a report, dated June 1950, by Kirby and Holford . The tunnel
6
speed was 120 ft/s and the test Reynolds number 1.6×10 based on the wing mean chord.
The model was fitted with a duct from the intakes to the tailpipe which was throttled to give
something like the airflow required for the engine. Although the report’s main concern
centres on stability issues, test results provide CD data over a range of CL values.
(45)
Brinkworth has analysed these data and reports a CD0 value of 0.0139. In this he has used
2
the procedure described above in which a graph of CD against CL is plotted. The graph’s
line here is gratifyingly straight, there being no Re variation involved in this case.
Extrapolation to CL = 0 then yields the above estimate for CD0. The graph’s gradient is 0.122
so that, the aspect ratio A being 3.24, the value for k is 1.24. It would be interesting to compare
the CD0 estimate with the value obtained from flight tests of the full-scale aircraft for which
the Reynolds number would be more than twenty times higher than that of the tunnel tests.
7. Concluding Remarks
The survey has attempted to set down the main steps needed to both understand the origins of
drag and reduce its adverse effects on aircraft performance. Having provided a large number
of examples, it is hoped that this survey might prove useful in the ongoing debate concerning
performance comparisons of famous aircraft.
(18)
It is felt that, for the British, it was Jones in 1929 who provided the much-needed impetus
for improvement. Lest it be thought that this came rather late in the day, it should be
remembered that the necessary wherewithal for this, the new boundary-layer and wing
theories, had only begun to penetrate Britain’s scientific and technical consciousness a mere
six or seven years earlier.
Jones’s plan for improvement was to pare away parasite drag to such an extent that the only
drag remaining was that which could not be eradicated, the drag due to the boundary layer.
However, later application of his method, based on flat plate turbulent boundary layer data,
led to underestimation of this drag contribution and improvement in this respect did not occur
(29, 30)
until RAE staff provided more accurate methods in the late 1930s. Meanwhile,
(25)
attempts such as that by Morgan at the RAE to pin down the various parasite drag
contributions, hampered as this was by boundary-layer drag underestimation on the one hand
39
and rather inaccurate parasite drag prediction on the other, nonetheless showed the way
forward. In this respect, a further problem concerned the accuracy with which total drag was
predicted from flight-test data. Total drag being equal to total engine thrust in steady level
flight, it was necessary to improve the accuracy with which propeller efficiency was
calculated and to take correct account of the thrust augmentation effects provided by ejector
exhausts and ducted radiators. Again, the RAE made considerable progress in these areas.
Around 1940 all of these advances were shown to advantage in the drag assessments made by
(24) (33)
Collar and, more particularly, by Hufton at the RAE for relatively well-streamlined
(36)
fighter aircraft. And at that time Bottle showed that this method worked almost equally
well when applied to necessarily more parasite drag prone bombers. These advances in drag
(43)
assessment then successfully carried over to the early jet age, as Courtney demonstrated
in 1950. During much of this period, whilst the RAE was engaged in developing more
specialist advances in aerodynamics, its staff were also occupied with the bread-and-butter
(39)
activity of compiling an extensive data base detailing the core aerodynamic features of a
wide range of aircraft. This has proved extremely useful here in tracing the ‘history of CD0’.
To reach this far in this ‘history’ has taxed the stamina of this old CD0 geek, and doubtless the
reader’s patience. Nonetheless, it would be interesting to bring the story closer to the present
day, not least because this would produce a far greater tribute to the RAE than that of this
paper. It was one thing for the RAE to deal with the aerodynamic exigencies of the wartime
period and its immediate aftermath of the early jet age but expertise of a greater order of
magnitude was needed to devise the aerodynamic advances for Concorde and the Airbus
wing design. To cover these major challenges faced by staff at both Farnborough and the
newly-founded Bedford site, however, would be no mean feat since it would depend heavily
on tracing such data as remain from wilfully abandoned government laboratories and now-
defunct manufacturers. Perhaps another writer might take up such a considerable challenge?
Acknowledgements
The author gratefully acknowledges the generous assistance of Brian Riddle, Librarian of the
Royal Aeronautical Society, and Anthony Pilmer of the National Aerospace Library,
Farnborough. Gratitude must also be expressed to the late Harry Fraser-Mitchell FRAeS and
Frank Armstrong FREng FRAeS for alerting the author to the RAE CD0 database’s existence,
to Geoffrey Butler FRAeS (Farnborough Air Sciences Trust (FAST)) for providing copies of
such documents, to Professor Brian Brinkworth FREng, FRAeS and Michael Marsden for
further data, to Professor Gareth Padfield FREng, FRAeS for data relating to the Wright
Flyers, to Dr-Ing Johannes Traugott (Technischen Universität München) for data on the
albatross, and to Dr Christopher Mitchell FRAeS and Evolve Design Consultants for
assistance with some of the graphics.
40
References
1. ACKROYD, J. A. D. Babinsky’s Demonstration: The Theory of Flight and Its Historical
Background. J. Aero. Hist., 2015, 5, pp 1-42. https://www.aerosociety.com/publications/jah-
babinsky-s-demonstration-the-theory-of-flight-and-its-historical-background/
2. ACKROYD, J. A. D. The aerodynamics of the Spitfire. J. Aero. Hist., 2016, 6, pp 59-86.
https://www.aerosociety.com/publications/jah-the-aerodynamics-of-the-spitfire/
3. FINSTERWALDER, S. Die Aerodynamik als Grundlage der Luftschiffahrt Vortrag.
Zeitschrift für Flugtechnik und Motorluftschiffahrt, 1910, 1, pp 6-10, 30-31.
4. LILIENTHAL. O. Der Vogelflug als Grundlage der Fliegekunst. Gaertners, Berlin, 1889.
[Birdflight as the Basis of Aviation. Trans. Isenthal, A. W., Longmans Green & Co., London,
1911.]
5. LOFTIN, L. Quest for performance. The evolution of modern aircraft, NASA SP 468,
NASA, Washington DC, 1985. https://history.nasa.gov/SP-468/contents.htm
6. NEWTON, I. Philosophiae Naturalis Principia Mathematica. London, 1687. [Mathematical
Principles of Natural Philosophy. Trans. Third Latin Edition by MOTTE, A, London, 1729.]
7. GIBBS-SMITH, C. H. Sir George Cayley’s Aeronautics 1796-1855. HMSO, London, 1962.
8. CAYLEY, G. On aerial navigation, A Journal of Natural Philosophy, Chemistry and the Arts,
1809, 24, pp 164-174; 1810, 25, pp 81-87, 161-169.
9. LANGLEY, S. P. Experiments in Aerodynamics. Smithsonian Institution, Washington DC,
1891.
10. PRANDTL, L. Über Flüssigkeitsbewegung bei sehr kleiner Reibung. Verhandlungen des
dritten internationalen Mathematiker-Kongresses in Heidelberg, 8-13 August 1904. pp 489-
491, Teubner, Leipzig, 1905.
11. BLASIUS, P. R. H. Grenzschichten in Flüssigkeiten mit kleiner Reibung. Zeit. für
Mathematik und Physik, 1908, 56, pp 1-37.
12. PRANDTL, L. Tragflügeltheorie, I Mitteilungen. Nach der kgl Gesellschaft der Wiss zu
Göttingen, Math-Phys Klasse, 1918, pp 451-477.
13. PRANDTL, L. Tragflügetheorie, II Mitteilungen. Nach der kgl Gesellschaft der Wiss zu
Göttingen, Math-Phys Klasse, 1919, pp 107-137.
14. GLAUERT, H. Aerofoil and Airscrew Theory. Cambridge University Press, 1926.
15. PADFIELD, G. D. and LAWRENCE, B. The birth of the practical aeroplane: An appraisal
of the Wright brothers’ achievements in 1905. Aeronaut J, 2005, 109, pp 421- 437.
16. LAWRENCE, B. The Flying Qualities of the Wright Flyers, PhD thesis, September 2004,
Dept of Engineering, University of Liverpool, UK. A paper based on the thesis is available at
http://pcwww.liv.ac.uk/eweb/fst/publications/Wright1905.pdf
17. ANDERSON, J. D. The Airplane. A History of Its Technology. AIAA. Reston VA, 2002.
18. JONES, B. M. The streamline aeroplane. JRAeS, 1929, 33, pp 357-385.
41
19. PRANDTL, L. Über den Reibungswiderstand strömender Luft. Ergebnisse der

Aerodynamischen Versuchsanstalt zu Göttingen, 1927, 3, pp 3-5.
20. HELE-SHAW, H. S. and BEACHAM, T. E. The variable pitch airscrew. JRAeS, 1928, 32,
pp 525–554.
21. SACHS, G. Minimum shear wind strength required for dynamic soaring of albatrosses. Ibis,
2005, 147, pp 1-10.
22. AERODYNAMIC STAFF, RAE. Some aerodynamic figures of merit. RAE Report No. BA
1275 (Rewritten), January 1937.
23. AERODYNAMIC STAFF, RAE. Notes on aerodynamic criteria for some recent service
aeroplanes. RAE Report No. BA 1430, October 1937.
24. COLLAR, A. R. The performance of the ‘Hurricane’ and ‘Spitfire’ aeroplanes. ARC, Ae
1674, June 1940.
25. MORGAN, M. B. Note on drag analysis of various types of aeroplane. RAE Report No. BA
1499, August 1938.
26. MEREDITH, F. W. Note on the cooling of aircraft engines with special reference to ethylene
glycol radiators enclosed in ducts. ARC, R & M No. 1683, 1935.
http://naca.central.cranfield.ac.uk/reports/arc/rm/1683.pdf
27. CAPON, R. S. The cowling of cooling systems. ARC, R & M No. 1702, 1936.
28. HARTSHORN, A. S., DIPROSE, K. V. & PATTERSON, G. N. On the thrust obtainable
from exhaust momentum. RAE Report No. BA 1578. March 1940.
29. SQUIRE, H. B. and YOUNG, A. D. The calculation of the profile drag of aerofoils. ARC,
R & M No. 1838, 1937. http://naca.central.cranfield.ac.uk/reports/arc/rm/1838.pdf
30. YOUNG, A. D. The calculation of the total and skin friction drags of bodies of revolution at
zero incidence. ARC, R & M No. 1874, 1939.
http://naca.central.cranfield.ac.uk/reports/arc/rm/1874.pdf
31. RELF, E. F. Recent research on the improvement of the aerodynamic characteristics of
aircraft. JRAeS, 1938, 42, pp 512-535.
32. YOUNG, A. D. Boundary Layers. BSP Professional Books, Oxford, 1989.
33. HUFTON, P. A. An analysis of the performance of the high speed Spitfire and a standard
Spitfire. RAE B. A. Departmental Note – Performance No. 12, April 1940.
34. MORGAN, E. B. and SHACKLADY, E. Spitfire: The History. Key Publishing, Stamford,
Lincs, 1987.
35. BRINKWORTH, B. J. Spitfire ‘Tailplane Protection’ and spinning trials. J. Aero. Hist.,
2017, 7, pp 16-24. https://www.aerosociety.com/publications/jah-spitfire-tailplane-
protection-and-spinning-trials/
36. BOTTLE, D. W. Analysis of the drag of a number of bombers at top speed. RAE B. A.
Departmental Note – Performance No. 24, November 1940.
37. COOPER, H. J. and THETFORD, O. G. Aircraft of the Fighting Powers, Vol. 1.
Harborough, Leicester, 1941.
42
38. GREEN, W. Warplanes of the Third Reich. MacDonald, London, 1970.

39. RAE STAFF. Structural weight and drag analyses for British and American aircraft. RAE
Tech. Note SME 96, Aero 1092. Issue 1 c1942 (Addendum January 1943), Issue 2 November
1943, Issue 3 August 1945 (Addenda October 1945, April 1946).
40. MAIR, W. A. (Ed). Research on high speed aerodynamics at the Royal Aircraft
Establishment from 1942 to 1945. ARC, R & M No. 2222, 1950.
41. BINKWORTH, B. J. On the aerodynamics of the Gloster E28/39 - a historical perspective.
Aeronaut. J., 2008, 112, (1131), pp 307–326.
42. AERODYNAMIC STAFF, RAE. Drag analysis of the Canberra and brief comparison with
flight tests. RAE Technical Memorandum No. Aero. 142, December 1950.
43. COURTNEY, A. L. Profile drag analysis of Supermarine E. 41/46, Hawker E. 38/46 and
North American F. 86A, with notes on method of drag estimation. RAE Tech. Note Aero.
2074, December 1950.
44. KIRBY, D. A. and HOLFORD, J. F. Low speed tunnel tests on a 1/5th scale model of a
single-jet fighter with a 40° sweptback wing (Hawker F.3/48). RAE Report Aero 2382, June
1950.
45. BRINKWORTH, B. J. Letter to the author. August 2017.
46. CAYLEY, G. Sir George Cayley’s Governable Parachutes, Mechanics’ Magazine, 1852, 57,
pp 241-244.
47. ANDERSON, J. D. Introduction to Flight, Its Engineering and History. McGraw-Hill, New
York, 1978.
48. KUTTA, M. W. Auftriebskräfte in strömenden Flüssigkeiten. Illus. Aeronautische
Mitteilungen, 1902, 6, pp 133-135.
49. ZHUKOVSKII, N. E. Über die Konturen der Tragflächen der Drachenflieger. Zeitschrift für
Flugtechnik und Motorluftschiffahrt, 1910, 1, pp 281-284.
50. ACKROYD, J. A. D. Lanchester – The Man (The 31st Lanchester Lecture). Aeronaut. J.,
1992, 96, (954), pp 119-140.
51. LANCHESTER, F. W. Aerodynamics. Constable & Co Ltd, London, 1907.
52. BAZZOCCHI, E. Technical aspects of the Schneider Trophy and the world speed record for
seaplanes. Aeronaut. J., 1972, 76, (734), pp 65-82.
53. ANDREWS, C. F. and MORGAN, E. B. Supermarine Aircraft since 1914. Putnam, London,
1981.
54. JAMES, D. N. Schneider Trophy Aircraft 1913 – 1931. Putnam, London, 1981.
55. ANDREWS, C. F. and COX, W. G. The Supermarine S4 – S6B. Aircraft in Profile, Vol. 2,
Profile Publications Ltd., Windsor, Berks. 1965.
56. HEINKEL, E. Schnellpostflugzeug He 70. Zeitschrift für Flugtechnik und
Motorluftschiffahrt, 1933, 24, (24), pp 669-676. [The high-speed Heinkel He 70 mail
airplane. NACA Tech. Memo. No. 746, 1934.]
43
57. JONES, R. and SMYTH, E. Experiments on a Heinkel He. 70 aeroplane in the Compressed
Air Tunnel. ARC, R & M No. 1709, 1936.
58. HUFTON, P. A. and SMITH, A. G. Short performance tests on Heinkel He 70 with Rolls
Royce Kestrel XVI (V. P.). RAE Report No. BA 1364. January 1937.
59. HUFTON, P. A. Preliminary note on the cleanness efficiency of the Heinkel He 70. RAE
Report No. BA 1401. May 1937.
60. SHAW, R. A. and DIPROSE, B. A. Tests on the Heinkel He 70 in the 24 ft. tunnel. RAE
Report No. BA 1445. December 1937.
61. CAMBRIDGE UNIVERSITY AERONAUTICAL LABORATORY. The measurement of
profile drag by the pitot-traverse method. ARC, R & M No. 1688, 1936.
62. HUFTON, P. A. An analysis of some performance tests on the Heinkel He 70 with Rolls
Royce Kestrel XVI. RAE Report No. BA 1474. May 1938.
44
Appendix 1
The Importance of L/D Ratio and Its Maximum Value
In his Notebook entry for 1st December 1804 Cayley states (7) that his first glider, the world’s
first aeroplane, descended steadily at a speed of 15 ft/s in still air conditions and at an angle
of 18° below the horizontal. The situation for any such glider is shown in Figure 11, the
gliding angle here being γ. The glider’s weight is W and, using the triangle of forces shown,
it is seen that the weight component along the glide path and in the direction of steady motion
is Wsin γ. The weight component perpendicular to this motion is Wcos γ. Thus Wsin γ
provides the propulsive thrust for this motion so the glider’s drag force, D, which precisely
opposes this is given by
D = Wsin γ.
The lift force, L, by definition, is perpendicular to the direction of motion so that
L = Wcos γ.
Consequently, the lift-to-drag ratio, L/D, is
L/D = cos γ/sin γ = 1/tan γ. (A1.1)
Figure 11 Gliding flight in still air conditions
Clearly it is advantageous to achieve the smallest possible gliding angle and, by equation
(A1.1), this requires the L/D ratio to be as high as possible. Cayley’s gliding angle, γ = 18°,
indicates an L/D ratio around 3. For his ‘Governable Parachute’ of 1852 Cayley (46), in
contrast, suggested that this glider would travel a horizontal distance 5 to 6 times its release
height, implying an L/D ratio between 5 and 6.
A further indicator of the importance of the L/D ratio is provided by a theoretical analysis to
determine the distance flown by a propeller-driven aeroplane until its fuel is completely
exhausted. The analysis is rather lengthy and can be found in a number of standard texts on
aeronautics, for example Reference 47. The situation envisaged in the analysis is rather
simplistic in the sense that no climb-out and descent-to-landing are included and the
aeroplane is assumed to fly throughout at a fixed value of the L/D ratio. The result for the
distance flown, known as the Breguet range relation, reveals that this distance is directly
45
proportional to the L/D ratio. Consequently improvement of that ratio provides a

proportionate improvement in range.
In order to determine an aeroplane’s maximum L/D ratio it is advantageous to put the

problem a slightly different way: under what circumstances is the D/L ratio a minimum?
This makes the mathematical analysis rather more straightforward. According to equations
(5) and (6),
2
CD = CD0 + BCL , where B = k/(πA).
2 2
Since Drag, D = ½ρV SCD and Lift, L = ½ρV SCL, then
2
D/L = CD /CL = CD0/CL + BCL / CL = CD0/CL + BCL.
The only variable on the right hand side of this relation is CL. Using differential calculus, the
rate of change of D/L with respect to variation in CL is
2
d(D/L)/dCL = - CD0/CL + B.
2
This is a maximum, a minimum or a point of inflexion when d(D/L)/dCL = 0, ie CL = CD0/B
2
or when CD0 = BCL , the induced drag coefficient then being equal to CD0. The calculation
2
of d2(D/L)/dCL at this value of CL reveals that the case of interest is a minimum. Therefore,
½ ½ ½ ½ ½
(D/L)min = CD0/(CD0/B) + B(CD0/B) = (BCD0) + (BCD0) = 2(BCD0) .
½
Thus, (L/D)max = ½(πA/(kCD0)) . (A1.2)
This maximum value is seen in Figure 2 of the main text. One further feature of that figure
worth discussing is the slope of the CL ~ incidence graph, the incidence angle being α.
Prandtl (12) showed that this slope depends on wing aspect ratio, A, the result for the elliptic
(14)
loading case (k = 1) given by Glauert being
dCL/dα = 2 π A/ (2 + A). (A1.3)
As the aspect ratio A tends to infinity, the slope thus correctly increases to the limit of 2π
(48) (49)
which is the result given by the aerofoil theories of Kutta in 1902 and Zhukovskii in
1910. This variation of the slope with aspect ratio springing from Prandtl’s work is shown in
Figure 12 taken from Reference 50. Included in the figure is the graph of the results obtained
(51)
by Lanchester in 1907. As Reference 50 shows, Lanchester’s result is
dCL/dα = 2 κ (ε + 1),
(51)
where κ and ε are empirical variables of dubious provenance for which Lanchester is
nonetheless able to estimate their values over a range of aspect ratios. Although Lanchester’s
results shown in Figure 12 underestimate Prandtl’s correct variation, they nonetheless show a
(51)
similar trend. Lanchester also predicted the variation of induced drag, giving the result
2
CDi = CL / [4κ (1 + ε) / (1 – ε)].
Prandtl’s equivalent result for the elliptic loading case k = 1 is (equation (5))
2
CDi = CL /(πA).
46
The two results differ only in the form of

their denominators and Figure 13, taken
from Reference 50, shows their variations
with aspect ratio. Lanchester’s result for
induced drag is thus seen to be close to that
of Prandtl.
As is well-known, Lanchester’s work of

(51)
1907 in aerodynamics was largely
ignored in Britain, despite his membership
of the Advisory Committee for Aeronautics
since its foundation in 1909. Thus his
rather rudimentary concept of the boundary
layer, which he called the inert layer, and
his semi-empirical circulation theory for
wing lift had no impact on aeronautical
research in Britain. Moreover, the British
aeronautical community remained entirely Figure 12 Variation of lift curve slope
innocent of the major and more rigorous with aspect ratio
Source: Ackroyd, Reference 50
advances abroad: Prandtl’s boundary-layer
(10)
theory , the circulation theories of
(48) (49)
Kutta and Zhukovskii for aerofoils
and their extension by Prandtl (12, 13) to
wings. Thus British work in
aerodynamics remained largely an
experimental activity. For example, the
various aerofoil section shapes devised,
often by cut-and-try methods, were
usually tested on wings of aspect ratio six,
the results being displayed graphically in
forms similar to Figure 2. Thus there was
no realisation that the slopes of the CL ~ α
graphs were, by equation (A1.3), close to Figure 13 Variation of induced drag factor with
three-quarters of 2π and that this could be aspect ratio Source: Ackroyd, Reference 50
improved by use of higher aspect ratios.
Similarly there was no understanding that the rise of CD with α was due to induced drag and
that this too could be improved by choice of higher aspect ratio. The situation changed
rapidly, however, once Glauert had visited Göttingen after the close of the First World War.
Thereafter, as described in Reference 1, British research in aerodynamics advanced rapidly
(18)
during the 1920s. What then became required, as evidenced by Jones’s plea of 1929, was
that designers should put it into common practice. As the 1930s advanced some, at least,
realised that the world had begun to drift into uncertain times. The doctrine at that time, it
will be recalled, was that the bomber would always get through and the fear was that this
might translate into an all-too-terrifying reality. Thus the 300 mph more heavily-armed
fighter with a high rate of climb would be crucial in countering this.
47
Appendix 2
The Aeroplane’s Power & Lift Requirements
For an aeroplane in steady level flight at constant speed the total propulsive thrust, T, must
precisely oppose the drag, D. The power required to sustain this motion is TV. For an
aeroplane solely dependent on an engine-propeller combination for its propulsion, the engine
power being P and propeller efficiency η, the power provided by the propeller is ηP. Thus,
3
ηP = TV = DV = ½ ρV S CD (A2.1)
3
or V = 2ηP/(ρ S CD). (A2.2)
(5)
Essentially this is the result stated by Loftin which he uses as the basis for his assessment
of aeroplane performance.
3
To see how this V relationship affects the speed performance of an aeroplane, consider the
case of the first powered aeroplane to achieve sustained, controllable flight, the Wright Flyer
1 of 1903. Its extensively braced biplane assembly produced a very high value of CD. This
and its low value of P resulted in a maximum value for V of around 30 mph. With no other
changes being made, suppose that P could be boosted by a factor of 100. In that case,
3
according to equation (A2.2), V would be equally enhanced and V itself increased by the
cube root of 100, which is roughly 4.6. Thus the speed would become a little short of 140
mph, a worthwhile speed enhancement but at the expense of an enormous power increase.
To put the latter to better use, suppose that by careful streamlining CD could be reduced to
one fifth of its previous value and sufficient lift could be obtained by a single wing, a
monoplane, of an area one half that of the Flyer. This combination of aerodynamic
improvements, 1/5 and ½, or a reduction to 1/10 of the product CD S, gives a further boost of
3
10 to our original power-produced boost of 100. This yields a total boost to V of not the
earlier 100 but 1,000. The latter’s cube root is 10 so the original top speed of 30 mph is
multiplied by this factor to yield a speed of around 300 mph. In contrast, if no aerodynamic
improvements had been employed then the reduction of 1/10 in the product CD S would have
to have been replaced by a power increase factor of 1,000. Given that the Flyer 1’s power
was around 10 hp, an aeroplane, however configured but possessing the Flyer’s high CD S
value, would have required an engine power of about 10,000 hp in order to achieve a speed of
around 300 mph.
But how might the above improvement in top speed affect the aeroplane’s performance at
low speed, in particular the pilot’s ability to handle the aircraft at that condition? According
to equation (4), the relation for the lift force, L, is
2
L = ½ ρV S CL, (A2.3)
CL being the lift coefficient which depends predominantly on a wing’s incidence angle (see
Figure 2).
In steady level flight at constant speed, the lift must precisely oppose the aeroplane’s weight,
W:
L = W.
48
By equation (A2.3),
2
V = 2w/(ρ CL), (A2.4)
where w = W/S
is the wing loading. To return to the examples chosen for the power ~ speed comparison
above, again suppose that a wing area half that of the Wright Flyer 1 is selected.
Furthermore, suppose that, with its much more powerful but heavier engine and a metal
structure of much greater weight, this monoplane’s total weight has increased, say, by a
factor of eight. Consequently, wing loading w has increased by a factor of sixteen. If there
are no changes to air density and the maximum value of CL available, according to equation
2
(A2.4), V must also increase by this same factor. The lowest value of V would then increase
by a factor equal to the square root of sixteen, which is four. Whereas the Flyer 1 might have
sustained itself in flight at a little over 20 mph, this much heavier monoplane’s lowest speed
would be of the order of 80 mph. In order to make this monoplane more manageable for the
pilot at the low speeds of take-off or landing, an increase in CL through the use of trailing
edge flaps/leading edge slats might bring this speed down to a more acceptable 60 mph value.
2
The above exercises are far from hypothetical. The Flyer 1 of 1903 has a wing area of 510 ft
and weighs 745 lbs. It is powered by an engine of 12 hp which, taking the propeller
efficiency η to be about 0.67, yields a power at the propeller around 8 hp. These data can be
compared with those of the early marks of the Spitfire, for which engine power is a little over
1000 hp. The propeller efficiency η is 0.80 so that power at the propeller is slightly above
2
800 hp, roughly one hundred times that of the Flyer. The Spitfire’s wing area is 242 ft , a
little less than half that of the Flyer, whereas its weight is 5820 lbs. Thus its wing loading, w,
is slightly more than sixteen times higher than that of the Flyer 1. As to the crucial factor of
drag, modern assessments of the Flyer suggest that at its highest speed it flew at a CD value
around 0.1. In contrast, having benefitted from a high degree of streamlining, the Spitfire and
its contemporaries such as the Hurricane and Bf 109 achieved CD values around 0.02,
roughly one fifth that of the Flyer. In fact, due to its CD value being the lowest of those
above, together with thrust assistance from its engine exhaust and radiator hot air ejection
(see below), the Spitfire achieved a maximum speed around 360 mph, significantly higher
3
than our 300 mph rough estimate. However, as a further illustration of the P ~ V relation
(A2.2), it is worth noting that during its impressive development the Spitfire’s engine power
roughly doubled. On that basis its maximum speed would be expected to increase by a factor
of roughly the cube root of two, which is about 1.26. This suggests a maximum speed around
450 mph, close to that actually achieved by the final Spitfire marks.
As noted above, the Spitfire and its contemporaries designed in the mid-1930s benefitted
from additional thrust, Tj, provided by rearward ejecting engine exhausts and heat
regeneration in their ducted radiators. The power enhancement provided by these devices is
then Tj V. Consequently equation (A2.1) now takes the form
3
ηP + TjV = DV = ½ ρV S CD. (A2.5)
The important point here is that a calculation of CD based solely on the power at the propeller
(5)
term ηP, the approach taken by Loftin , will result in an underestimation of CD.
49
In the late 1920s and throughout the period of the Second World War it became common
practice in Britain to scale drag results from wind tunnel and flight tests to the drag
experienced at sea level and at a speed of 100 ft/s. This made it easier to compare drag
results of one aircraft with another and with results obtained in the RAE’s 24 ft Open Jet
Tunnel. The analysis required to achieve this scaling has been given in Appendix 1 of
Reference 2 but is repeated here for ease of reference. It follows from equation (A2.5) that
3 2
ηP /(V ) + Tj/(V ) = ½ ρ S CD.
Writing σ = ρ/ρ0, ρ0 being the sea-level air density, then
3 2
ηP /(σV ) + Tj/(σV )= ½ ρ0 S CD.
However,
2
½ ρ0 S CD (100) = D100 (A2.6)
is the drag at sea level at a speed of 100 ft/s. Consequently,
4 3
10 ηP/ (σ V ) + Tj,100 = D100 (A2.7)
4 2
Tj,100 = 10 Tj/(σ V ), (A2.8)
the latter being the exhaust/radiator thrust at a speed of 100 ft/s. The term
4 3
10 ηP/ (σ V ) = Te,100 (A2.9)
is then the engine/propeller thrust at 100 ft/s.
In flight-test work with aeroplanes propelled by a piston engine the values of η, P and Tj, if
appropriate, must be known with certainty at the altitude at which the speed, V, has been
measured. With that information, CD can be calculated from the above relations. However,
for an aeroplane propelled solely by a jet engine it is the thrust Tj which is known. In this
case the basic relationship (A2.1) becomes
2
Tj = D = ½ ρV S CD. (A2.10)
When scaled to sea level conditions and a speed of 100 ft/s, the expression for Tj,100 is then
that given by equation (A2.8).
50
Appendix 3
The Value of CD0 for Jones’s ‘Streamline Aeroplane’
(18)
In order to estimate the boundary-layer drag on aeroplanes, Jones bases his calculations
on known skin-friction relations for flat plates experiencing laminar or turbulent boundary-
layer flows. Since these flows are due entirely to the action of the air’s viscosity, the latter’s
influence enters results in terms of the Reynolds number
Re = ρ V l /μ,
where l is the streamwise length of the body and μ is the viscosity coefficient. The skin
(18)
friction relations are shown in Figure 14 taken from Jones’s paper in which he plots the
skin friction coefficient kF against Reynolds number Re. This coefficient he defines as
2
kF = (skin friction drag)/(ρV E), (A3.1)
E being the total wetted surface area of the plate which is twice S, its planform area. This,
together with the lack of the factor of ½ in the definition of kF, gives kF = CD/4. For laminar
(11)
boundary layers he quotes the result given by Blasius (equation (3)) and confirmed by
experiment,
–½
kF = 0.66 Re . (A3.2)
(19)
For turbulent boundary layers he turns to Prandtl’s result drawn from a survey of
experimental data (equation (8)):
– 0.15
kF = 0.019 Re . (A3.3)
Figure 14 Variation of drag for plates and aerofoils with Reynolds Number
Source: Jones, Reference 18
51
(18)
Jones points out that, in practice, departure from the lower, laminar curve of Figure 14
can occur when turbulent transition begins, the curve looping upward to join the upper
turbulent curve. Although the process of transition was not understood at the time, it was
5 5
know that this often, but not always, tended to occur in the Re range 10 and 5×10 ; hence
the shaded area in the figure indicating the possible transition region. Included in Figure 14
are results for a variety of current aerofoil sections, most of which lie on or slightly above the
turbulent curve. He argues that, due to the high Reynolds numbers achieved in full-scale
flight, the skin friction drag on aeroplane wings will lie close to the turbulent curve. He then
turns to bodies of revolution, citing a number of experimental results obtained from a series
of airship shapes. The wetted area for these shapes he takes to be three-quarters the area of
their circumscribing cylinders. The results, when displayed in this form in Figure 15, are
strikingly similar to those for flat plates in Figure 14.
Figure 15 Variation of drag for axisymmetric bodies with Reynolds Number

Source: Jones, Reference 18
(18)
For the estimation of aeroplane drag due to its boundary layer, Jones explains that, having
selected a Reynolds number representative of full-scale flight, the corresponding value for kF
can be obtained from equation (A3.3). He points out that, for full-scale aeroplanes at full
6 7
speed, the Re values based on wing chord as the length scale vary between 4×10 and 10 ,
the corresponding values of kF ranging from 0.0020 to 0.0017. In contrast, representative Re
7 7
values based on fuselage length lie between 2×10 and 5×10 , the corresponding kF values
(18)
varying from 0.00155 to 0.00135. “So as to leave a little in hand”, as Jones puts it, he
selects a kF value of 0.0020 for his streamline aeroplane. He is then left with the problem of
selecting a value for E, the aeroplane’s total wetted area. Having surveyed a number of
52
current aircraft, he finds that the ratio E/S (S being the wing planform area) varies between
3.0 and 3.5. Taking E/S = 3.2 as an average value and using equation (A3.1), his result for
the streamline aeroplane’s boundary-layer drag at zero lift, D0, is
2 2 2
D0 = 0.0020 ρ V E = 0.0020 × 3.2 ρ V S = 0.0040 × 3.2 (1/2 ρ V S).
Although Jones does not take this step, when recast in terms of CD0 his result yields
2
CD0 = D0 / (½ ρ V S) = 0.0128.
7
Had he chosen the kF value of 0.0035 for the highest Reynolds number of 5×10 , his value for
CD0 would have been 0.0086. This is close to the values used later at the RAE as described
in Section 5.
(18)
Jones raises two further points of interest here. The first is that in Figure 14 the drags of
aerofoils of small thickness-chord ratio tend to lie slightly below the turbulent curve whereas
thicker aerofoils, particularly those marked by black triangles (▲) with thickness-chord ratios
between 12.5% and 20%, lie significantly above the curve. Subsequent wind-tunnel studies
in Britain in the mid-1930s seemed initially to contradict this. As described in Section 3 of
Reference 2, this led to some confusion when it came to the choice of wing thickness-chord
ratio for certain fighter aircraft. Luckily the matter was resolved within a couple of years and
Jones’s prognostication was proved correct.
The second point raised in Jones’s lecture concerns some of the experimental data shown in
(18)
Figure 15. Jones argued, correctly as it turned out, that the point at which transition to
turbulence occurred seemed to depend on the turbulence level of the wind tunnels used.
Those with lower turbulence levels appeared to produce later onsets of turbulence than those
(18)
having high turbulence levels. Jones then drew attention to the two experimental curves
6
shown as elongated S-shaped lines at Reynolds numbers slightly above 10 . These results, he
explained, had been obtained in the naval tank of the National Physical Laboratory. In this
case the airship models had been pushed steadily through water at rest, the results suggesting
that when the turbulence level is extremely low a delay in the onset of transition appeared
possible. This offered the possibility of extending the laminar boundary-layer region and
thereby reducing the drag. Again Jones was correct in this assessment and he pursued the
matter in more detail later in his career.
53
Appendix 4
Supermarine Schneider Trophy Seaplanes & Type 224
Mitchell’s sleekly streamlined Supermarine S4 with its unbraced monoplane wing was seen
as a game-changer in the Schneider Trophy contests. Its design was all the more astonishing
since it came from a designer previously wedded to heavily-braced biplane layouts. Its crash
at Chesapeake Bay in October 1925, which the pilot luckily survived, was suspected as being
due to the wing’s aeroelastic instability. Consequently, Mitchell’s subsequent Schneider
Trophy contenders featured wire-braced monoplane wings. A further change involved the
means by which engine cooling was achieved. According to Bazzocchi (52), Mitchell
calculated that 38% of the S4’s drag came from the aircraft’s underwing Lamblin radiators,
the probable cause of the very high CD0 value given below. Subsequent Supermarine
contenders used double skins on the floats, the coolant being pumped through the gap
between the skins and thereby eliminating cooling drag.
Taking up Jones’s suggestion (18), estimates are made here for the CD0 values of all of the
Supermarine Schneider Trophy contenders and that for the Supermarine Type 224, the
(53)
forerunner of the Spitfire. The floatplane data are taken from Andrews and Morgan , data
(54)
with which James largely agrees. These data had appeared in an earlier article by
Andrews and Cox (55) in which it is stated that the dimensional and speed data are “From
official and company records. Any variations from figures sometimes published are due to
differential loadings for test flight series”. Engine data are stated to be “From Napier and
Rolls-Royce records”. However, the Andrews and Morgan (53) data for the S6B list two
maximum speeds, 390 mph and 407.5 mph, but only one power value of 2350 hp. In
contrast, James (54) notes that the higher speed achieved by the S6B for the World Speed
Record of 407.5 mph was produced using a higher-powered ‘Sprint’ version of the Rolls-
Royce R engine of “some 2600 hp”. Bazzocchi (52) is more specific, giving the value of 2650
hp and it is this which is used here.
One assumption made in the calculation of CD using equation (A2.2) is that the performance
data for engine power and maximum speed are for straight and level flight at close to sea
level conditions. Sea level air density is taken from the International Standard Atmosphere
(ISA) tables (see, for example, Reference 37). As to propeller efficiency η, the values
assumed are those provided by Bazzocchi (52) for the S5 and S6B, the value for the latter
being applied to the S6. For the S4, η is taken to be 0.8, a value which is also used for the
Type 224. In the calculation of CD0 the values of CDi emerge as being very small, ranging
from 5 × 10 - 4 to 8 × 10- 4.
The CD0 results are shown in Table 11A, those for the S6 and S6B being gratifyingly
consistent for aircraft which are, aerodynamically, almost identical. The result for the S4 is
exceptionally high, 0.048, and, as noted above, probably due to its radiator system. In
(5)
contrast, Loftin produces a much lower value of 0.027 but this is based on a reduced
engine power, 450 hp, which appears erroneous. Bazzocchi (52) has given numerical results
for CD0 for the S4, S5 and S6 and these are listed in Table 11B, together with his assumed
engine powers. His result for the S4 is close to that calculated here and must surely have
been based on the speed listed in Table 11A, this having the official status of a World
Seaplane Record. The difference in CD0, around 2%, might be due to a different value
selected for η (unstated) and his choice of an air density value other than that of the ISA. His
54
result for the S5 is more problematic since, with a lower power, he obtains a higher CD0
value. The cause of this might, in part, be due to his choice of a maximum speed different to
that given in Table 11A. In this respect, it should be noted that CD is very sensitive to the
value of the speed selected. The P ~ V3 form of equation (A2.2) indicates that slight
differences in speed are magnified roughly by a factor of three in CD calculations. To
illustrate this, the calculation procedure has been reversed for Bazzocchi’s S5 data. His
power and CD0 data in Table 11B are used to estimate the maximum speed, listed there as 306
mph. This drop of roughly 4% in speed compared with that listed in Table 11A results in
roughly a 10% change in CD0. A similar problem appears in the case of Bazzocchi’s S6 data,
for which the speed given in Table 11A is again a World Speed Record which one would
expect Bazzocchi (52) to use. If, however, he did not, the result of the reverse calculation
yields the speed of 345 mph listed in Table 11B, a drop of around 3.6% which leads to the
10% change in CD0.
Despite the questionable accuracy of the CD0 results in the two Tables, their values are all
high compared with later monoplanes but are the inevitable consequence of the ‘fixed
undercarriage’ layout adopted. The floats necessary to their mission requirements, however
well-streamlined, inevitably create significant parasite drag.
Table 11A Characteristics of Supermarine Schneider Trophy and Type 224 Aircraft
S4 S5 S6 S6B S6B* Type 224
(1925) (1927) (1929) (1931) (1931) (1934)
Power (hp) 680 900 1,900 2,350 2,650 600
Span (ft) 30.58 26.75 30 30 30 45.83
Wing Area S (ft2) 139 115 145 145 145 295
Weight (lbs) 3,191 3,242 5,771 6,086 6,086 4,743
Prop. Efficiency η 0.8 0.814 0.84 0.84 0.84 0.8
Speed (mph) 226.75** 319.27 357.7** 390 407.5** 228
Aspect Ratio, A 6.73 6.22 6.21 6.21 6.21 7.12
CD0 0.048 0.028 0.033 0.033 0.033 0.030
* World Speed Record S6B with ‘Sprint’ R engine of higher power.

** World Speed Record
Table 11B Bazzocchi’s Results, Reference 52

S4 S5 S6 S6B
Power (hp) 680 875 1,900 2,650
CD0 0.049 0.031 0.037 ?
Speed (mph) 226.75** 306 345 407.5**
(52)
The Type 224 data included in Table 11A have been obtained from Andrews and Morgan
(34)
and Morgan and Shacklady which suggest that the aircraft achieved a maximum speed of
228 mph at an altitude of 15,000 ft when powered by a Goshawk engine of 600 hp. The
value of CD0 based on these data, the air density at altitude being taken from the ISA table, is
55
included in Table 11A. It is seen to be high, a result of such drag-enhancing features as its
large wing of high thickness-chord ratio, open cockpit and fixed, trousered undercarriage. As
described in Reference 2, it was the Supermarine team’s dissatisfaction with this aircraft’s
high drag which led to the complete redesign resulting in the Spitfire.
The CD0 calculations given here are all rather tentative. All that can be claimed for them is
that, having taken the initial data at face value and obeyed the stated assumptions, the listed
CD0 results follow. Readers in possession of more accurate data for any of these aircraft are
encouraged to communicate them via, perhaps, the Correspondence column of this Journal.
Appendix 5
The Heinkel He. 70 & The RAE’s Assessment
The Heinkel He. 70 first flew in December 1932 and its high speed performance rapidly
(56)
attracted considerable interest. In 1933 Ernst Heinkel’s report on the aircraft provided
performance data obtained from flight tests by the Deutsche Versuchsanstalt für Luftfahrt
(DVL) together with much detail on its design, structure and aerodynamic features. In the
following year the NACA produced an English translation of Heinkel’s report (see Reference 56).
(56)
To illustrate the aircraft’s superior aerodynamic quality, Heinkel used a version of the P ~ V3
relation (A2.2) which introduced what he called the ‘high-speed index’. This is η/CD and from
equation (A2.2) it follows that at sea-level conditions
3
η/CD = ½ ρ0 V S/P. (A5.1)
For a given aircraft the values of wing planform area, S, and engine power, P, are known and
its maximum speed, V, has been obtained from flight tests. At sea level, the air’s density is
ρ0. Consequently all the terms on the right hand side of equation (A5.1) are known and that
relation yields a value for the combination η/CD but not its individual constituents’ values
which remain unknown. Nonetheless, the aim is to produce the highest possible value for
this index since this implies not merely a high value for propeller efficiency η but, more
(56)
importantly, a low value for the drag coefficient CD. Heinkel provides values of η/CD for
a number of recent American aircraft which have led the field in speed performance and
(56)
those for the Northrop Alpha and the Lockheed Orion are listed in Table 12. Heinkel
points with approval to the retractable undercarriage on the Orion together with the cowling
for its radial engine. He admits that, until recently, German attempts to compete with the
Orion have been unsuccessful, citing the Junkers Ju. 60 (Table 12) as an example. However,
he stresses that this situation has changed markedly with the advent of the He. 70, as Table 12
shows.
56
Table 12 Drag of a number of aircraft

‘High-speed Index’ η/CD
Northrop Alpha (1930) 25.3
Lockheed Orion (1931) 35.6
Junkers Ju. 60 (1932) 26.1
Heinkel He. 70 (1932) 52.8
Source: Heinkel, Reference 56
The values of η/CD listed in Table 12 are consistent with the data for CD0 given in Tables 1
and 2 if account is taken of the fact that, due to induced drag, CD is slightly greater than the
Table 12 CD0 values and the propeller efficiencies are taken to lie in the region 0.75 – 0.80.
Performance data for various versions of the Heinkel He. 70 will be given later.
(56)
As to the choice of aerodynamic shape, Heinkel states that no wind tunnel tests were made
to aid in this. He favours an elliptic planform wing of aspect ratio 6 but makes no mention of
the fact that this choice minimizes induced drag. However, as indicated in Section 2, induced
(56)
drag is small at high speed and minimizing it is of minor benefit. Heinkel favours this
shape since it provides a broad chord inboard, offering ample room for the main
undercarriage’s outward retraction. As related in Reference 2, Mitchell had a similar reason
(56)
for his choice of this shape in the Spitfire’s design. As to thickness-chord ratio, Heinkel
states that this is 17.5% at the wing root but tapers considerably toward the tip. The emphasis
(56)
in the design, Heinkel stresses, is in the reduction of parasite drag. Thus the
undercarriage, including the tail skid, is fully retractable. So too is the radiator for the
twelve-cylinder BMW engine using ethylene glycol as coolant (see Condition A in Figure
16). Use of the latter also allows the radiator matrix to be smaller both in size and weight,
thus producing less drag when the radiator is
deployed.
(56)
A further drag-reducing feature, Heinkel notes, lies
in the geometry of the wing root attachment to the
circular-section fuselage. For this low wing design,
the wing root abuts the fuselage square to its surface.
This creates anhedral for the wing in the immediate
vicinity of its root but a short distance outboard this
turns to slight dihedral so as to provide lateral stability.
(56)
Heinkel states that the maximum speed was
increased slightly after wing root fillets were added.
He adds that items contributing parasite drag are
designed to minimize this effect; door knobs and foot-
steps are inset and the windows are flush-mounted.
Finally, the wings, fuselage and control surfaces are
shell-plated and flush-riveted.
Figure 16 Heinkel He. 70
radiators, Source: Hufton,
Reference 59
57
Such was the British interest in the He. 70 that Vickers constructed a 1/18th scale model of it
which was tested in the company’s wind tunnel in order to measure its drag coefficient.
Subsequently the model was tested at the RAE and then transferred to the National Physical
Laboratory for testing at higher Reynolds numbers in the Compressed Air Tunnel (CAT).
The CAT results, together with those obtained by Vickers and the RAE, appeared in an ARC
(57)
report by Jones and Smyth dated February 1936. It was estimated that at its maximum
speed the He. 70 flew at a CL value of 0.116 and the CAT tests were therefore carried out at
that value. Both the Vickers and the RAE tests had been conducted at the slightly higher CL
value of 0.136 and their CD results lay slightly above those from the CAT, in part due to the
effect of higher induced drag. However, once induced drag had been deducted, Jones and
(57) 6
Smyth conclude that at the CAT’s highest test Reynolds number, Re, of 5.6×10 ,
obtained with the tunnel running at 24 atmospheres pressure, the value of CD0 is 0.0151. This
extrapolates well to the full-scale aircraft’s similar CD0 value deduced from DVL data given
(56) 7
by Heinkel at the even higher Re value of 1.8×10 . As to the model’s CD0 variation with
Re, this is plotted in terms of the related coefficient Cf and it is this variation which is
included in Section 5’s Figure 5 taken from Reference 23. As mentioned in Section 5, Cf = S
CD0/ E, E being the aircraft’s total wetted area and S the wing planform area. The Vickers
2
estimate for the He. 70’s total wetted area E is 1230 ft and S is 393 ft2 so that for both the
model and the full-scale aircraft E/S = 3.13. This happens to be close to the average value
(18)
chosen by Jones , E/S = 3.2 (see Appendix 3), in calculating his theoretical performance
curves shown in Figure 4.
Rolls-Royce took delivery of its Kestrel-powered He. 70, G-ADZF, in March 1936 shortly
after the first flight of the Spitfire. As tested by the DVL, the early version of the He. 70 had
a retractable radiator which dropped baldly into the airstream with no surrounding cowl to
create further drag (Figure 16’s Condition A). G-ADZF had a forward radiator enclosed
within a fixed cowl designed by Rolls-Royce in which the radiator itself did not fully retract
(Figure 16’s Condition C). The airflow was straight-through, the cowl providing no entry
(26)
flow divergence as advocated by Meredith to reduce the radiator matrix’s drag. G-ADZF
also had a fixed tail skid, unlike the earlier He. 70.
After flight testing at Rolls-Royce’s Hucknall site a few miles north of Nottingham, the
aircraft was transferred to the RAE for assessment, the cowled radiator now moved back to a
(58)
ventral position (Figure 16’s Condition B). A short RAE report by Hufton and Smith ,
dated January 1937, outlines the test procedures adopted and gives preliminary results.
Initially the position error of the aircraft’s Pitot-static head was determined for high speeds by
a series of flights over a speed course seven miles long at a height of 2,000 ft using timing
from the aircraft; at low speeds a suspended static head was used. The conclusion was that,
with the radiator retracted as far as possible, the maximum speed was 258 mph at 14,500 ft
altitude, but that this was reduced by 10 mph with the radiator fully deployed within the cowl.
(58)
The rest of the report is concerned with rate of climb measurements. A later report, it
was stated, would analyse the speed results.
(59)
This analysis is contained in Hufton’s report of May 1937 in which he compares CD0 and
other data obtained from the maximum speed results of the DVL, Hucknall and RAE tests. In
58
particular, these data were used to determine the cleanness ratios for the three radiator
configurations examined. However, cleanness ratios are here still based on flat plate
turbulent boundary-layer data and are therefore of rather dubious accuracy. The results are
listed in Table 13. The values for the propeller efficiency η seem rather high compared with
(33)
those used later by Hufton . The values for gross wing area and wetted area differ slightly
since in analysing the DVL tests the values were the best that could be obtained from rather
crude drawings. At the RAE the dimensions of the actual aircraft were measured; the
(59)
difference, Hufton explains, is partly a genuine one due to differences in the wing root
fillets and tailplanes. Table 13’s CD0 and CR values for the Hucknall tests are those listed in
Table 2 of Section 5 drawn from Reference 22.
Table 13 Heinkel He. 70 characteristics

DVL RAE Hucknall
Engine BMW Kestrel XVI Kestrel XVI
Radiator Plain retractable Cowled retractable Forward cowled
Gross Wing Area, S, ft2 393 390 390
Wetted Area, E, ft2 1237 1305
Maximum Speed, mph 235 219 239
Altitude, ft 0 1900 3000
Power, hp 660 599 662
Propeller Efficiency, η 0.800 0.814 0.822
CL 0.134 0.164 0.132
CD 0.0154 0.01825 0.0159
CDi 0.00095 0.00141 0.000915
CD0 0.0144 0.01685
Cleanness Ratio, CR 0.559 0.479 0.544
Total Drag @ 100 ft/s,
72 84.5 74
lbs
Tail Skid Estimate, lbs 0 4 4
CD0 (No Tail Skid) 0.0144 0.01605 0.0142
CR (No Tail Skid) 0.559 0.518 0.575
Source: Hufton, Reference 59
The last four lines of Table 13

indicate Hufton’s interest in
estimating the drag improvement
if the tail skid is made retractable.
Morgan (25), in assessing the He.
70, comments that “the tail skid
was of a rather clumsy design” and
assigns to it 4 lbs in drag, as does
(59)
Hufton .
The next step in the RAE’s

assessment of the He. 70 entailed
testing it in the 24 ft Open Jet
Figure 17 Heinkel He. 70 in the RAE 24 ft Open Jet Tunnel.
Source Farnborough Air Sciences Trust.
59
Tunnel, a facility which had begun operation in 1935 (Figure 17). The results are given in
(60)
the report by Shaw and Diprose dated December 1937. Since the He. 70’s wing span is
slightly above 48 ft, some 12 ft of the outboard part of each wing projected beyond the
tunnel’s jet. However, a correction based on the known CD0 for the aerofoil section was
(60)
applied in an attempt to deal with this difficulty. The report takes the view that this “did
not detract much from the value of the tests since the investigation was mostly concerned
with the wing root, fuselage and engine details where it was felt that there was most chance
(60)
of effecting improvements”. The report proposes that similar tests be made on other high
(24)
speed aircraft when they become available. The later Hurricane tests mentioned by Collar
indicate that this advice was followed.
(60)
The investigation included drag measurements made with the ventral radiator, with the
nose radiator, after leaks had been sealed and after excrescences had been removed. Pitot-
static traverses were carried out in the fuselage boundary layer aft of the cabin hood and aft
of the wing root. Also wool tufts were used to investigate the flow around the entries to the
cowled radiators.
For the tests the tail skid was removed, as was the propeller which was replaced by a spinner.
In the investigation of leaks, sealing was effected with linen tape and adhesive. Shaw and
(60)
Diprose note that the He. 70 is comparatively free of the seams and removable panels
which are common leak sources on most military aircraft. So although leaks on this aircraft
amounted to only about 8% of the total drag, they felt that the exercise clearly pointed to the
importance of attempting to reduce leakage on more leak-prone military aircraft. In the
investigation of parasite drag, excrescences such as exhaust pipes and air intakes were
removed whereas small pipes and knobs which could not be removed were faired.
The drag reductions produced for the case in which the cowled radiator was sited at the nose
are listed in Table 14. The initial total drag, 88.7 lbs, is significantly higher than that listed in
(59)
Table 13, 74 lbs, which was deduced by Hufton from flight test data. This Shaw and
(60)
Diprose largely attribute to the difference in Reynolds number Re between the two cases,
the Re value achieved in the Open Jet Tunnel being 0.4 of that reached at maximum speed in
Table 14 Heinkel He. 70 in 24 ft tunnel, Drag & Drag Reductions in lbs at 100 ft/s
Drag & drag reduction in lbs at 100 ft/s
Drag with nose radiator* 88.7
Leak sealing reduction 7.1
Removing nose radiator** 5.2
Removing tail skid 4.0
Removing exhausts, air intakes etc. 2.2
Drag completely faired 70.2
* Includes 13.3 lbs added for wings outside jet.
** Removing ventral radiator 7.3 lbs from total of 90.8 lbs.
Source: Shaw and Diprose, Reference 60
60
flight. As evidence for this, they point to the decrease in drag coefficient with increasing Re
shown by flat plate data; Figures 5 and 14 and equation (8) illustrate this behaviour. The
(29)
more accurate but yet to be announced Squire and Young analysis shows a similar
decreasing drag with increasing Re but the drag is higher than that for the flat plate due, for
example, to the additional boundary-layer pressure drag (see equation (11)). Parasitic items
having bluff shapes causing local separation, on the other hand, can show little variation over
wide ranges of Re. With this uncertainty concerning Re behaviour, the value of the data
listed in Table 14 therefore lies mainly in the scale of the drags revealed for individual
parasitic items. In contrast, the initial and final drags, 88.7 lbs and 70.2 lbs, seem of rather
dubious accuracy considering the rather simplistic estimate of 13.3 lbs for the drag addition
(60)
due to the wings projecting beyond the tunnel’s jet. Although Shaw and Diprose do not
do so, calculation of CD0 based on the above drag for the aircraft before modification yields
the value 0.0191 whereas for the completely faired aircraft it is 0.0151.
(60)
The Pitot-static traverse method used by Shaw and Diprose had emerged in 1936 from an
earlier analysis by B. M. Jones supported by experimental work, both in flight and in a wind
(61)
tunnel, at Cambridge University . One of the participants in the wind tunnel tests was
Flight Lieutenant Frank Whittle, then a Cambridge student. Another such was Alec Young,
who later brought his understanding gained from this link between an aerofoil’s trailing-edge
(29)
flow and its wake to the turbulent boundary-layer analysis of Squire and Young . Shaw
(60)
and Diprose carried out their Pitot-static traverses at a distance of 20 ft aft of the He. 70’s
nose, in one case a little aft of the fuselage’s cabin hood, in the other slightly aft of the wing
root. In the latter tests they could detect no sign of wing-fuselage interference. This they
(56)
attributed to the beneficial effects of the wing root anhedral and fillets, as Heinkel had
suggested. The fuselage traverses aft of the cabin hood yielded a drag higher than that
(60)
anticipated from flat plate turbulent boundary-layer data. This Shaw and Diprose
attributed to the effect of the cabin hood.
The use of wool tufts in the Open Jet Tunnel tests revealed flow separation around the
(60)
radiator cowl’s entry. Shaw and Diprose suggested that this could be rectified by keeping
the entry clear of the fuselage boundary layer. As mentioned in Reference 2, it seems this
measure was suggested to the Mustang’s design team by Beverley Shenstone, previously
Mitchell’s aerodynamicist for the Spitfire design, although it may not necessarily be the case
that this advice sprang from the He. 70 tests in the Open Jet Tunnel.
Meanwhile Hufton had been attempting a more detailed analysis of the flight test data
obtained with G-ADZF. These data now included results for the aircraft fitted with a new
radiator arrangement in which a redesigned and slightly larger cowled forward unit used
(62)
water as the coolant. His report , appearing in May 1938, is largely concerned with the
effects of different radiator deployments together with reassessments of propeller efficiency
(62)
and engine power variation with altitude. From this Hufton concludes that there is no
variation of L/D with altitude due to Reynolds number variation and the maximum value of
L/D is 15.8 at CL = 0.46.
61
One consequence of this re-appraisal is that his earlier CD values are now increased. These
cover a range of radiator flap settings and also three radiator deployments: fully retracted,
partially retracted and fully exposed. These CD values have been obtained at various CL
values and therefore at different speeds and thus different Reynolds numbers. In estimating
(62)
CDi, which is deducted from CD so as to give CD0, Hufton uses equation (5) with k = 1, an
adequate approximation since CDi is generally small at high speed. His CD0 values range
from 0.0156 to 0.0191, depending on these various radiator settings and deployments. The
best result, CD0 = 0.0156, is for the nose glycol radiator described as in a level condition, the
worst case being CD0 = 0.0191 with the ventral glycol radiator deployed and its flap down. If
nothing else, though, these results illustrate the importance of radiator design in controlling
drag. However, one puzzling aspect of the results is that the various CD0 values calculated
appear to depend on the CL conditions from which they were obtained. Here one is reminded
(43)
of Courtney’s dictum that drag estimation is by no means an exact science. Indeed, this
(43)
situation is similar to that later encountered by Courtney (see Section 6) in which Reynolds
number variation helped to explain such curious behaviour. It would be interesting to try his
2
procedure in which a graph of CD ~ CL is plotted to estimate CD0 and k. Regrettably, there
being so many different radiator conditions investigated, too few data points at any one
condition are available to make this procedure viable.
(62)
Hufton ends the summary of his main conclusions with the cryptic statement “No further
work is proceeding”. It was time to turn away from the He. 70 and move to the more urgent
assessments of the new military aircraft coming into service. Nonetheless, the RAE had
learned much from its detailed investigation of this extraordinarily low drag aircraft.
Dr J A D Ackroyd CEng FRAeS
Born in Bradford, Yorkshire, in 1938, John Ackroyd graduated in Aeronautical Engineering

in 1960 at Queen Mary College, University of London. After doctoral and postdoctoral
research in shock tubes there, he joined the staff of the Department of the Mechanics of
Fluids (later the Aerospace Division, Manchester School of Engineering) of the Victoria
University of Manchester. He taught in aerodynamics, flight dynamics, aircraft structures
and propulsion whilst carrying out research in boundary-layer theory. He retired in 2000 as
Senior Lecturer. His interest in the technical history of aviation resulted in him giving the
Royal Aeronautical Society's Lanchester (1991) and Cayley (2000) Lectures, the Inaugural
Cody Lecture (2003) and the Cody Lecture (2015). Together with B. P. Axcell and A. I.
Ruban, he is co-author of Early Developments of Modern Aerodynamics published jointly by
the AIAA and Butterworth-Heinemann (now Elsevier) in 2001.
62

Aerodynamics As The Basis of Aviation How Well Did It Do

Uploaded by

Copyright:

Available Formats

Aerodynamics As The Basis of Aviation How Well Did It Do

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Aerodynamics As The Basis of Aviation How Well Did It Do

Uploaded by

Copyright:

Available Formats

Journal of Aeronautical History Paper No.

Aerodynamics as the Basis of Aviation: How Well Did It Do?

2. Aerodynamic Drag and Its Evaluation

monoplane wing alone experiencing an

It is important to understand that, at whatever point an aeroplane reaches in its flight

According to equation (5), CDi varies

3. From the Wright Flyer to Aircraft of the Early 1930s

The Royal Aircraft Factory’s B. E. 2c is a particularly interesting case since it emerged so

Economic considerations in commercial aviation became more of a driving force for

4. B. M. Jones and the ‘Streamline Aeroplane’ (1929)

leading member of Britain’s Aeronautical

Figure 4 Power versus Speed; Jones, Reference 18

5. Piston-Engine Monoplanes (1930s and 1940s)

5.2 Drag measurement by flight testing

Table 2 Figures of merit from flight test measurements

* Engine Power Estimates may be 5 – 10% too high

5.3 Flight test and wind tunnel comparisons

COMPARISON OF C.A.T. AND FLIGHT DRAG MEASUREMENTS

Heinkel He. 70 CD0 = 0.008 Gladiator CD0 = 0.009.

5.4 Sources of parasitic drag

In August 1938 Morgan (Figure 6) used the

Table 3 Drag Contributions in lbs at 100 ft/s

Turbulent Skin Friction (Flat Plate)

Cleanness Ratio 0.576 0.35 0.36 0.35 0.35

Source: Morgan, Reference 25

5.5 Thrust from exhausts and radiators and other improvements

5.6 Drag of the Spitfire and Hurricane

Figure 10 Arthur Roderick Collar FRS (1908 –

Table 4A Thrust/Drag Contributions in lbs at 100 ft/s

5.7 Drag of bombers

Table 5A Thrust/Drag Contributions in lbs at 100 ft/s

Table 5B Thrust/Drag Contributions in lbs at 100 ft/s

Table 5C Zero lift drag coefficients and cleanness ratios

Table 5D Estimated maximum speeds

5.8 Collected drag and thrust data

Table 6 Thrust/Drag Contributions in lbs at 100 ft/s

Source: RAE Staff, Reference 39

Table 7 Drag and other data for a variety of types

Firebrand 0.0228 0.587 6.7 15.2

6. The Early Jet Age

Table 8 Thrust & Drag Contributions in lbs at 100 ft/s

Table 9 Drag of the English Electric Canberra B.1 at M = 0.74

6.2 Swept wing jets

Table 10A Drag of swept wing aircraft

Table 10B Residual Drag, lbs at 100 ft/s

19. PRANDTL, L. Über den Reibungswiderstand strömender Luft. Ergebnisse der

38. GREEN, W. Warplanes of the Third Reich. MacDonald, London, 1970.

Figure 11 Gliding flight in still air conditions

proportional to the L/D ratio. Consequently improvement of that ratio provides a

In order to determine an aeroplane’s maximum L/D ratio it is advantageous to put the

The two results differ only in the form of

As is well-known, Lanchester’s work of

Figure 15 Variation of drag for axisymmetric bodies with Reynolds Number

* World Speed Record S6B with ‘Sprint’ R engine of higher power.

Table 11B Bazzocchi’s Results, Reference 52

CD0 0.049 0.031 0.037 ?

Speed (mph) 226.75** 306 345 407.5**

Table 12 Drag of a number of aircraft

Table 13 Heinkel He. 70 characteristics

Speed (mph) 226.75 306 345 407.5