Journal of Aeronautical History Draft 2 Paper No.

2016/03

The Aerodynamics of the Spitfire

J. A. D. Ackroyd

Abstract

This paper is a sequel to earlier publications in this Journal which suggested a possible origin
for the Spitfires wing planform. Here, new material provided by Collars drag comparison
between the Spitfire and the Hurricane is described and rather more details are given on the
Spitfires high subsonic Mach number performance. The paper also attempts to bring
together other existing material so as to provide a more extensive picture of the Spitfires
aerodynamics.

1. Introduction

To celebrate the Spitfires eightieth year, the Royal Aeronautical Societys Historical Group
organised the Spitfire Seminar held at Hamilton Place, London, on 19th September 2016.
This celebration offered the author the opportunity to talk about the aerodynamics of the
(1, 2)
Spitfire. Part of that lecture was based on material already published in this Journal
which suggested that the semi-elliptical wing planform adopted could have come from earlier
publications by Prandtl. Section 2 offers further thoughts on this planform choice, in
particular how this interlinked with the structural decision to place the single wing spar at the
quarter-chord point and the selection of the aerofoil section. Since the publication of
References 1 and 2, new material has come to light in the form of an investigation by A. R.
Collar in 1940 which compares the drags experienced by the Spitfire and the Hurricane. This
investigation, described in Section 3, indicates that a significant part of the Spitfires lower
drag can be attributed to the better boundary-layer behaviour produced by its wings thinner
aerofoil section. Section 4 deals with such matters as surface finish, including the well-
known split-pea flight tests, propulsion effects on aerodynamic performance and further data
on the turning radius problem. The paper closes with more detailed information on the
Spitfires performance at high subsonic Mach numbers. This shows an aerodynamically
cleaner version to be slightly superior in this respect to the aircraft which replaced it, the Mk
IV Gloster Meteor. Suggestions are offered as to why this was so.

As a result of the Spitfire Seminar, the author has received two contributions containing
information which has here been incorporated at appropriate points. The contributors are
(3, 4)
Ralph Pegram, the author of two books on Supermarine aircraft , and Michael Salisbury,
from 1947 an aerodynamicist at Supermarine, later Chief Aerodynamicist there and
subsequently on the TSR2 project.

59
Journal of Aeronautical History Draft 2 Paper No. 2016/03

2. Structural and Other Considerations Influencing the Spitfire Wings

Aerodynamic Design Evolution

The explanation of aerofoil behaviour began to emerge in the early years of the twentieth
(5)
century. In particular, Zhukovskiis aerofoil theory of 1910, based on shapes having blunt
noses and streamlined tails, was found to produce good agreement when subjected to, as
(6)
Galileo put it, the ordeal of experiment. Wind-tunnel tests by Betz in 1915, for example,
confirmed this both for the pressure variation around a Zhukovskii aerofoil and for its lift
forces steady increase with incidence. A further correct feature of this theory is its
prediction of an aerofoils centre-of-pressure behaviour with change of incidence. For
symmetric aerofoils this point at which the lift force acts is found throughout the incidence
range below stall to be fixed at the quarter-chord point aft of the leading edge. However, for
cambered aerofoils the centre of pressure moves steadily forward from the aerofoils rear as
incidence increases, the lift force increasing as it does so. Yet this movement occurs in such
a manner that the moment of the lift force about the quarter-chord point does not change with
incidence. As argued in Reference 1, this lift loading can be replaced by one in which the lift
is fixed at the quarter-chord point, the lift increasing with incidence, and a pure torque which,
in contrast, does not change with incidence. The latter behaviour provides an attractive
feature for structural design.

These characteristics of the Zhukovskii aerofoil are fairly common to many of the aerofoil
sections designed in the first thirty or so years of powered flight. A further feature is that
many of these aerofoils have their maximum thicknesses around one-third chord aft of the
leading edge.

During the biplane era, the wing structure often favoured involved the use of two spars. Their
locations can be detected by studying the placement of the interplane struts. Occasionally,
however, a single spar, in effect, was used. One early example is provided by the Fokker Dr.
1 Triplane of 1917. In this case, the single spar within each of the three wings can again be
detected by the placement of the outboard interplane struts running through the wings. In
side view, for the two lower wings this placement is seen to lie at one-third chord aft of the
leading edge. Presumably this location was chosen because the aerofoil section is thickest at
this point and therefore provides maximum spar depth. However, in his presentation to the
Spitfire Seminar, this author mistakenly claimed the quarter-chord point as the spars position.
This error arose from a concentration on the side view of the upper wing alone, in which the
aileron projects aft of the wings trailing edge so as to make the chord appear greater than it
is. For this error the author apologizes. As to the Triplanes spar itself, this is a hollow
wooden rectangular box centred around one-third chord. It is partly the spars twin webs but
mainly the top and bottom flanges booms in aeronautical parlance - which resist the lift
loads direct bending effect whilst the complete box resists the twisting moment created by
that load, the moment in this case varying with incidence.

So Mitchell, in his Spitfire design, was not the first to adopt the structural weight-saving
measure of a single wing spar. As to resisting the moment created by the lift load, Mitchell
accomplished this through the combination of the wings nose skin and the spar web to form
a D-nosed torsion box. In his contribution to Reference 7, Alan Clifton, Supermarines Head

60
Journal of Aeronautical History Draft 2 Paper No. 2016/03

of the Technical Office in the 1930s, points out that Mitchell had introduced this combination
of a single spar and nose torsion box in his design for a six-engine flying boat in early 1930.
(8)
Pegram adds that a patent later that year covered the torsion boxs use as the steam
condenser for evaporatively cooled engines. For such an engine the water coolant emerged
from the cylinder block as steam to be cooled back to the water phase as indicated, thus
avoiding the need for a drag-producing external radiator. Although the flying boat itself was
cancelled before completion, a one-twenty-fourth scale wing was built for testing in the
Duplex wind tunnel at the Aerodynamics Department of the National Physical Laboratory
(NPL). Due to the aircrafts cancellation, the testing programme itself was curtailed but the
early results, for the wings pressure distributions over a range of incidences, were deemed to
(9)
be of general interest and were published in 1934 by Cowley and McMillan . One aim of
these tests was to assess the effect of aileron deflection on the wings pressure distribution
and thus calculate the wing torsional loads created.

A single spar combined with the nose torsion box also appeared in Mitchells fighter design,
the disappointing Supermarine Type 224 of 1934. This was powered by a Rolls-Royce
Goshawk engine which employed evaporative cooling. The fighter was built to the Air
Ministrys Specification F7/30 and became the starting point for the evolution of the Spitfire
design. In the former case, the single-web spar is located at one-third chord from the leading
(10)
edge , presumably on the grounds here again that this location provides maximum spar
depth. The single spar and torsion-box combination was carried over to the Spitfire design
although, as will emerge, the single spars location moved to the quarter-chord point
presumably so as to benefit from the torsional simplification mentioned earlier. It should be
added that the use of metal skinning enabled the skin itself to carry some of the bending and
torsional loads. The skin thickness adds to the spars top and bottom boom areas in resisting
the direct bending stresses and provides a further torsion box aft of the spar which aids in
resisting torsion; hence the term stressed-skin construction.

Aside from the Type 224s problems associated with its evaporative cooling system, which
created difficulties when the aircraft was flown inverted, the other main problem lay in the
aircrafts high drag. Obvious culprits were the fixed, trousered undercarriage and open
cockpit. These were dealt with, together with the removal of the wings anhedral inboard
portion, in the re-design which began in the late spring of 1934. The result is shown in
Figure 1 taken from Cliftons contribution to Reference 7. Moreover, two further members
of Mitchells design team, Ernest Mansbridge (see Reference 10) and Beverley Shenstone
(see References 11 and 12), are both on record as suspecting the aircrafts thick aerofoil
(10)
sections as significant contributors to this high drag. As Mansbridge put it,
We were a bit over-cautious with the wing and made it thicker than it need have been.
We were still very concerned about possible flutter, having encountered that with the
S.4 seaplane.

The Type 224s aerofoil section for the inboard anhedral portion was a symmetric section of
(10)
18% thickness/chord (t/c) ratio (NACA 0018) . For the dihedral outboard portion an RAF 34
section (12.5% t/c ratio) was modified to a t/c ratio of 18% to match the anhedral NACA 0018
(8)
inboard section and tapering linearly to 9.6% at the tip (Pegram ). Thus the Type 224s t/c
distribution was similar to that of the Hurricane (see Table 1 below). A twist, or washout, of

61
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Figure 1 Supermarine F7/30, Drawing No. 30000 Sheet 2. Cliftons contribution

to Reference 7

2.5to improve stalling characteristics was used and this was carried over to the Spitfire
(8, 10)
design . Previously, the Supermarine team had used thinner sections for the Schneider
Trophy floatplanes listed in Table 1. Indeed, for the Stranraer biplane flying boat of 1934 the
(8)
aerofoil used was a similarly thin NACA section ; the Supermarine team were here
remarkably quick off the mark in adopting these new aerofoil sections. However, in all these
cases the wings were externally-braced. In the case of the monoplanes, as Mansbridges
above comment indicates, this measure was adopted following the accident to the unbraced
S.4 Schneider Trophy contender of 1925. The Type 224s thicker wing, in contrast, required
no external bracing.

Table 1 Thickness/Chord Ratios (%)

(8)
Supermarine S. 5 (RAF 30 symmetric section) 12.6
(8)
Supermarine S. 6 (RAF 27 symmetric section) 9.8
(8)
Supermarine Stranraer (NACA 2409 cambered section) 9.0
Hawker Hurricane (Clark YH cambered section) 18 (root) 12 (tip)
(13)
Bf 109 (Resembles NACA 23 series cambered section) 14.8 (root) 10.5 (tip)
(10)
Spitfire (NACA 22 series cambered section) 13 (root) 6 (tip)

So during the second half of 1934, one aim of the Supermarine team became the reduction of
the Type 224s t/c values towards those used on their earlier aircraft. Roger Dicksons

62
Journal of Aeronautical History Draft 2 Paper No. 2016/03

(14) (8)
documents in the Davies archive , Pergram reports, show that this work was initiated
on 1st May 1934 in an instruction issued by Clifton to Dickson. Amongst the questions
asked was How much can the wing area be reduced using a higher lift section, preferably
one we know, say NACA 2412? Reference 10 quotes a report by Shenstone, dated May
1934 and after his visit to the United States, in which he comments favourably on the use
there of NACA four-digit sections from the 22 and 24 series. The data from the NACAs
aerofoil testing programme at high Reynolds numbers in its variable-density wind tunnel had
(15)
begun to emerge in 1930. The data for the 24 series appeared in a technical note dated
(16)
January 1932 and by December of that year a comprehensive NACA Report brought
together the data for a large selection of such four-digit sections. The aerofoil sections finally
selected for the Spitfire were drawn from these NACA designs, in this case the NACA 22
series, 2213 at the root, 2206 at the tip: 2% camber located at 20% chord, t/c 13% at the root,
(10)
6% at the tip . In his contribution to References 11 and 12, Shenstone merely states that
the NACA 22 series was just right. Perhaps the preference for the NACA 22 series lay in
the fact that the aerodynamic moment about the quarter-chord point of that series is roughly
(16)
half that of the NACA 24 series . The point here, as revealed by the original NACA data
of Reference 16, is that it is the camber value and the shape of the camber line which largely
determine the value of the sections aerodynamic moment. Crudely speaking, moving the
point of maximum camber forward from 40% chord to 20% chord whilst retaining the same
camber value reduces the length of the upper surface suction peak immediately aft of the
leading edge, thereby reducing the moment. This reduction would be beneficial for the
torque on the D-nosed torsion box.

As to planform shape, by the early autumn of 1934 straight taper was still being used, as
shown in the Supermarine drawing presented by Joseph Smith, Mitchells eventual successor
(17)
as Chief Designer, in his lecture to this Society in 1947. This was reproduced in
Reference 1 and is here repeated as Figure 2 for ease of reference. The single spar is clearly

(17)
Figure 2 Supermarine F7/30 development, autumn 1934. Smith

63
Journal of Aeronautical History Draft 2 Paper No. 2016/03

seen to be now at the quarter-chord point, although in plan view the spar is, like that of Figure
1, still angled slightly backwards. Yet the wing has now acquired a broader chord at the root,
suggesting that the t/c value is lower than that of Type 224. Straight taper is also used for the
tail surfaces. However, the fuselage is now closer to that of the Spitfire, the Type 224s short
fairing aft of the cockpit, still evident in Figure 1, having been discarded in favour of a perhaps
less drag-prone continuation of the cockpits upper contour to the tail in Figure 2. The engine
is still the steam-cooled Goshawk.

Planform shape changed dramatically around the close of 1934 with the adoption of the semi-
elliptical wing. This particular combination of two semi-ellipses wedded at a common major
axis resulted in the spars quarter-chord location now lying in plan view perpendicular to the
aircrafts fore-and-aft axis. This produced a number of structural and other benefits. In the
layouts of Figures 1 and 2 the overall lift load on the spar would act to the rear of the spars
fuselage joint, creating a twisting moment there which would require a strengthened and
heavier joint; this was eliminated in the straight-spar layout. Reference 10 recounts the
comments of E. J. Davis concerning the difficulties of manufacturing the spar-rib
arrangement had these components not been at right angles to each other; these too were
eliminated in the new layout. A further factor, as Price (11) points out, is that around the time
of the change to the semi-elliptical wing the decision was taken to switch from the Goshawk
engine to the new, more powerful Rolls-Royce PV12 (Merlin) engine, as yet, like the
Goshawk, still steam-cooled. However, the PV12 engine was heavier, forcing a forward
movement of the aircrafts centre of gravity. A compensatory forward shift of the centre of
lift was accomplished by straightening the spar.

Other practical benefits followed from the adoption of the semi-elliptical wing planform. Its
broad root chord, continued outboard with initially only slight reduction, provided a wing
thick enough to accommodate the necessary spar depth, the undercarriage wells, the radiator
(once the switch to water/glycol cooling was adopted) and the outboard placement of the
guns whilst keeping the latter in line with the aircrafts centre of gravity. Yet this broad
chord, whilst allowing all this, also provided a wing thinner, in terms of the t/c ratios of Table
1, than any of its contemporaries. As will be seen in Section 3 below, this provided the
significant aerodynamic benefit of reduced boundary-layer drag.

Also this wing planform offered the further aerodynamic advantage of minimum, or near-
minimum, induced drag. However, to reiterate the point made in Reference 1, one should not
become too transfixed by this feature since the advantage is often slight. The induced drag
(18)
coefficient, CDi, is given by
CDi = k CL2 /(A), (1)
where CL is the lift coefficient and A the wing aspect ratio. The factor k has its minimum
value of unity for elliptic wings alone (excluding the fuselage and tail) of constant aerofoil
(18)
section camber and no twist. However, Glauert shows that for wings alone having
straight tapers of the ratios under discussion the value of k increases to at most 1.03. His
calculations for the effect of twist, when adjusted to the 2.5 value of the Spitfire, suggest a
value of k at most about 1.05. Far more significant are the effects of the fuselage and tail in
(19) (20)
the determination of k for the complete aircraft. Both Anderson and Loftin provide

64
Journal of Aeronautical History Draft 2 Paper No. 2016/03

suggested ranges for k for complete aircraft, although they use an aerodynamic efficiency
factor which is the reciprocal of k. Thus their suggested k values range from 1.05 1.18
(19) (20)
(Anderson ) and 1.33 1.4 (Loftin ) although in the latter case older, less
(21)
aerodynamically efficient aircraft are often of main interest. However, Salisbury recalls
that when he joined Supermarine in 1947 the value of k for the Spitfires successor, the
Spiteful, was taken to be 1.15, the increase above near-unity being largely due to the
fuselage. This value might have been obtained from experience with the Spitfire but he has
no certain knowledge of this.

It is interesting to note that, together with the adoption of the semi-elliptical wing, the
Spitfires tail surfaces also changed from the straight taper of Figures 1 and 2 to elliptic
shapes. These surfaces, when experiencing lifting forces (or side forces for the fin), will
create trailing vortices and hence produce additional induced drag contributions which add to
the value of k. These are likely to have a small effect so that it is surprising that the
Supermarine team appear to have gone to the trouble of trying to minimize them. On the
other hand, the Spitfire design is a prime example of close attention being paid to fine detail;
small percentage improvements add up. A further example of this is contained in Jeffrey
Quills contribution to Reference 7 in which he states that the Spitfire had an unusually thin
(8)
tailplane of 9% t/c ratio. To this Pegram adds the interesting point that the section is so
thin here, and for the wing tip, that Mitchell patented a method for attaching the skin to the
ribs as it had proved impossible to insert conventional riveting tools into the narrow interior.
This illustrates Mitchells keenness to achieve small t/c ratios.

Yet the fact some might call it this authors obsession remains that the Spitfires semi-
elliptical wing planform is geometrically identical to the planform published by Prandtl as
early as 1918. Reference 1 provided a suggestion as to how that might have come about, a
suggestion backed by no hard evidence but merely by what might be called the circumstantial
(10)
evidence given in that reference and in Reference 2. Against this can be set the recollections
of R. J. Fenner of the design office:
When we came to the final version with the Merlin engine RJM (Mitchell) had fixed
the smaller wing area, lower thickness/chord ratio and the optimum single spar position
and I had the job of producing the lay-out to meet his proposals.. The planform of
the wing was, originally, perfectly elliptical and then bent forward along its major axis
until the optimum spar position was straight. I remember clearly making several
drawings of alternative planform and RJM, in his rounds of the drawing boards,
selecting the scheme as described except for minor changes to the wing tips.

That being so, it is nonetheless remarkable that the final shape is so precisely that of Prandtls
design, not only in the 1-to-2 ratio of the ellipses semi-minor axes but also in the 4.1 value
for the ratio of the span (excluding the fuselage) to the root chord. Perhaps both approaches
were used, one confirming the other to create the desired optimum spar position along a
straightened quarter-chord line.

In Reference 1 it was suggested that Shenstone played a significant role in a number of these
key decisions made during the Spitfires aerodynamic design evolution. In commenting on

65
Journal of Aeronautical History Draft 2 Paper No. 2016/03

(8)
Shenstones involvement, Pegram suggests the apt term for him as Mitchells truffle-
hound, seeking out the new ideas in aerodynamics which could be fruitfully incorporated in
that evolution. For example, Shenstones contributions to References 11 and 12 imply that he
had tried to interest Mitchell in some form of elliptic planform. David Faddy, in an article
(22)
on his father, Alfred Faddy, another key member of Mitchells team, suggests that it was
Alfred Faddy who had persuaded Mitchell to take seriously Shenstones suggestion in this
respect. As to the choice of the NACA 22 series aerofoil section, Shenstone again had a hand
(11, 12)
in this as described above. He also pressed for the smooth surface finish discussed in
Section 4.1 below. The appearance of wing root fillets in the Spitfires design development
during the spring of 1935 would also, as suggested in Reference1, have been favoured by
Shenstone after his visit to the United States. There these features had been adopted by, for
example, the Northrop and Douglas companies. However, it appears that it was not he who
instigated this. At the time of the Spitfires design, the published data on fillets came from
(23) (24)
Muttray in Germany and Klein working with von Krmn in California. Both
favoured fairly extensive fillets which begin at the wing leading edge. In contrast, the
Bellanca 28-70 racing monoplane of 1934 used a reduced fillet which begins near the wings
maximum thickness and thereafter flares towards the trailing edge. Reference 10 recounts
that in October 1934 Mitchell had been made aware of the Bellancas fillet and had been
shown associated wind-tunnel data. Flight tests of the Type 224 with such fillets added
occurred in February 1935, wool tufts being used to determine flow patterns at the fillets.
Shenstones photograph of it, plus wool tufts, is reproduced in References 10 and 12. Similar
fillets were used on the Spitfire prototype, K5054, and flight tests to investigate them
occurred in late March 1936 following its first flight near the beginning of that month. In this
case oil was used to indicate flow direction and thereafter no shape changes seem to have
been deemed necessary.

As to the need for wing root fillets, this arises from the fact that a wing roots pressure
distribution does not match that around the fuselage. This mismatch can create local
boundary-layer separation which results in increased drag and an irregular turbulent wake
which can bombard the tailplane and elevator to detrimental effect. According to Reference
10, it was this tailplane buffeting on the Type 224 which Mitchell discussed with the
Bellancas pilot, visiting Supermarine at the time, and it was this conversation which led to
the testing of similar fillets on the Type 224.

Whatever the background to the selection of the Spitfires wing planform may have been, it is
interesting to note that variants of this planform were adopted for later fighter aircraft.
Examples are the Hawker Tempest and Fury, the Fairey Firefly and the Republic P-47
Thunderbolt, although straighter or even straight leading edges were used presumably for
ease of manufacture. However, when it came to Supermarines replacement for the Spitfire,
the Spiteful of 1944 (Specification F1/43), this planform was discarded in favour of a
straight-tapered wing incorporating a so-called laminar flow aerofoil section. Here the
emphasis was on reducing the boundary-layer skin friction drag by maintaining larger regions
of laminar boundary-layer flow over the wing surfaces. The efficacy of this approach to drag
reduction depended on high manufacturing accuracy and the avoidance of the smallest
surface irregularities. However, since much of the inboard wing flow would be contaminated
by turbulence from the propeller wash, it is doubtful that this choice would have provided

66
Journal of Aeronautical History Draft 2 Paper No. 2016/03

(25)
much benefit. Quill , for example, comments on its disappointing performance
improvements compared with the later marks of Spitfire. However, the Spitefuls wing was
adopted for the jet-powered Supermarine Attacker on which, of course, propeller wash was
absent.

3. Collars Investigation

In 1940, at a meeting of the Aeronautical Research Committee (ARC), the question was
asked: why is the Spitfire 40 mph faster than the Hurricane when both are powered by the
same Merlin engine? The person delegated to answer this was Arthur Roderick Collar (1908
(26)
1986), then at the Aerodynamics Department of the NPL. His report , dated June 1940,
to the ARCs Aerodynamics Sub-Committee had remained an internal ARC document
hitherto unremarked in the debate on the Spitfires aerodynamics.

Collars investigation benefitted from three things. Firstly, he had accurate data on the
performance of both aeroplanes: maximum speeds of 325 mph (Hurricane) and 365 mph
(Spitfire) at 18,000 ft, the Merlins power at that altitude being 1,020 hp, both propeller
efficiencies 80%. At these conditions, he states, the CL values are about 0.12 for the Spitfire,
0.15 for the Hurricane. From these data he is able to calculate both the overall drag and the
induced drag for both aircraft at the much lower speed of 100 ft/s; the reason for this choice
of speed will shortly become apparent.

Secondly, a complete Hurricane had been tested in the 24 ft open-jet wind tunnel at the Royal
Aircraft Establishment (RAE), Farnborough, so that data were available on excrescence drags
(tailwheel, aerial post and such) which could also be attributed to the Spitfire, which as yet
had not been tested. Farnboroughs wind-tunnel tests had been conducted at a speed of 100
ft/s so, by suitably scaling his maximum speed data, Collar is able to draw up drag balance
sheets for both aircraft at that lower speed (see Table 2). The basis of these calculations is
given in Appendix 1, the results being that the Spitfires total drag is 59 lb whereas the
Hurricanes is 82 lb. Thus the aim of Collars balance sheets is to provide reasoned values
for individual drag items leading to these totals.

Thirdly, Collar had available methods of proven accuracy for estimating the boundary-layer
drag contributions for both aircraft, contributions which turned out to be major parts of the
(27)
totals (see Table 2). One method , for aerofoil sections and dating from 1937, had been
devised by two Farnborough aerodynamicists, Herbert Brian Squire (1909 1961) and Alec
David Young (1913 - 2005). For a given aerofoil section shape, the method used the
boundary-layer integral momentum equation together with a momentum balance between the
aerofoils trailing edge flow and its subsequent wake so as to calculate both the skin friction
and pressure drags created by the boundary layer. The boundary layer was taken to be
(28, 29)
turbulent with its internal velocity variations following von Krmns logarithmic law .
(30)
Young subsequently extended the method to bodies of revolution. Thus the boundary-
layer drags, or profile drags, of both wings and fuselages could be estimated and are the first
items listed in Table 2.

67
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Table 2 Drag Contributions in lb at 100 ft/s

Spitfire Hurricane
Profile drag (wings, fuselage, and tail) 32.2 40.5
Roughness (including rivets and joints) 2.0 3.3
Induced drag (including washout) 3.0 4.0
Cooling drag (including oil cooler) 7.0 8.0
Air intake 1.0 1.0
Tail wheel 2.0 2.0
Tailplane protection 0.3 ---
Gun holes, aerial post 0.8 0.8
Windscreen 1.2 1.5
Leaks 5.0 11.0
Wing-body interference 4.5 9.9
Total 59.0 82.0

As to the choice of turbulent boundary-layer methods for the estimation of profile drag,
(26)
Collar makes the important point that, due to turbulence in the propeller wash, the
boundary layers on the fuselages and the inboard parts of the wings are likely to be entirely
turbulent. Moreover, the gun holes outboard of the propeller washes, four per wing in both
cases, are also likely to incite turbulence there.

Although Table 2 shows that roughly half of the drag, the profile drag, is created by the
boundary layer in both cases, the Spitfires profile drag is significantly less than that of the
Hurricane. From his calculations, Collar attributes much of this to the Spitfires thinner
wing. Contributions from air intake, tail wheel, gun hole and aerial post drags had been
obtained from the wind-tunnel tests on the Hurricane and are assumed to have the same
values for the Spitfire. The induced drags are calculated using equation (1) above on the
assumption that the CL values in both cases are those small quantities at the maximum speed
conditions quoted above. Table 2 shows that the induced drags are about 5% of the total in
both cases. However, at higher incidence/CL cases, such as the turning problem discussed in
Section 4.3 and its Appendix 2 below, the induced drag contributions will be found to be far
higher. As to Collars estimates for the Spitfires other drag contributions, these are based on
reasoned assessments drawn from close inspections of both aircraft. He judges the Spitfires
better finish to be the main factor in his slightly lower-valued assessments for roughness,
cooling, windscreen, leaks (control gaps etc.) and wing-body interference (the Spitfires wing
root fillet receives particular approval); here again, small percentage reductions add up.
There remains the small-valued item labelled tailplane protection, present on the Spitfire
but absent on the Hurricane, and on this Collar offers no explanation so that this item is
something of a mystery.

68
Journal of Aeronautical History Draft 2 Paper No. 2016/03

(31)
As to the importance of t/c ratio, Young , in reviewing Reference 27s and similar
methods, points out that a good approximation to their results is
(Boundary-layer pressure drag)/(Total boundary-layer drag) t/c.

The total boundary-layer drag is the sum of the skin friction and pressure drags. The pressure
drag arises from the fact that the slower moving boundary layer slightly displaces the flow
field exterior to it. Thus that flows pressure at the aerofoils trailing edge does not return to
the value it would have achieved had the boundary layer been entirely absent, thereby
creating a fore-and-aft pressure imbalance. Crudely speaking, the greater the t/c ratio, the
greater is the flow deceleration aft of the aerofoils maximum thickness, the consequence
being that the boundary layers growth rate increases and so does its displacement effect.

The above underlines the wisdom of choosing as small a value as possible for the t/c ratio in
the Spitfires design. In the discussion following his lecture recorded in Reference 7, Alan
Clifton remarks that this choice went against the advice of the NPL, advice to the effect that
there was no advantage in going below a thickness chord ratio of 15%. It appears that
similar advice was given to Sydney Camm at Hawker during the design of the Hurricane.
Roy Chaplin, a senior member of Camms team at that time, recalls in his contribution to
Reference 32 that wind tunnel tests at the NPL on a Hurricane model produced the advice
that no improvement in drag would be obtained by reducing the thickness-chord ratio of the
wing below 20%. As mentioned in Reference 1, it appears that this advice arose from
measurements obtained in the NPLs Compressed Air Tunnel (CAT), running since 1932,
which, it was later realised, could produce misleading results.

(33, 34)
Two ARC reports published roughly two years apart shed some light on this issue.
Both reports carry the same title, have the same authors and present aerodynamic data for the
same six aerofoils (RAF 28, 34, 38, 48, Clark YH, Gttingen 387) over a similar range of
Reynolds numbers; the influence of the Reynolds number will be discussed later. The
aerofoils have t/c ratios ranging from around 9% to 15%. However, the second report, dated
(33)
April 1936, stresses that it replaces the first report of September 1934. The 1934 report
argues that, whilst the turbulence level in the CAT is still under investigation, it is felt that
this is significantly less than that in the NACA variable-density tunnel (from which the data
of Reference 16 had been obtained) and is therefore more capable of achieving conditions
representative of those at full-scale in free flight. As to a possible drag dependence on t/c
ratio, the authors merely note that, apart from the higher results for the Gttingen 387 section
(one of the thickest), the others lie fairly close together and include the RAF 48 section of
similar t/c ratio to the Gttingen section but of reduced camber. Their graph for the drag
results shows a rather scattered band in which no clear trend is evident. However, the second
(34)
report of 1936 states that corrections to the previous results are necessary due to parasitic
drag in the aerofoils wire suspension system and slight deviation of the flow from the
horizontal, the errors for which are additive and affect, in particular, the earlier drag results.
(34)
After re-testing with better aerofoil supports and correcting for flow deviation, the authors
conclude that the whole series suggests that for normal aerofoils between 9 and 12 per cent
thick there is very little variation of profile drag, but that above 12 per cent thickness the drag

69
Journal of Aeronautical History Draft 2 Paper No. 2016/03

begins to rise appreciably. Since the ARC continued to assert that Reference 34 replaced
Reference 33, it appears that the former paper remained the final word on the matter.

It seems that Sydney Camm had followed the earlier flawed evidence for the Hurricane, the
latter emerging with the significantly higher t/c ratios shown in Table 1 and the consequent
(26)
higher drag. Collar , in commenting on this, can foresee some improvements to the
Hurricanes performance if leaks and such are attended to but predicts that the increase in
maximum speed might be only 15 mph. Otherwise, a major re-design with particular
emphasis on reducing the wings t/c ratio would be necessary in order to produce any marked
improvement.

(11)
Corroboration of Collars Spitfire results is provided by Price in his description of the
Speed Spitfire of 1938, an aerodynamically cleaner version intended for an attempt at the
(11)
world landplane speed record. Price states that, whereas the standard Spitfire Mk I was
estimated to have a drag of 60.2 lb at 100 ft/s (Collars estimate above is 59 lb 3 lb), the
Speed Spitfires drag was 53.3 lb. This was accomplished through a wing-span reduction, a
streamlined skid replacing the tailwheel, a longer, more streamlined windscreen and flat
canopy, the filling of all cracks and a finely polished finish.

(26)
Collars report undertakes a further task, which is to resolve what appeared to be
inconsistencies between the drags measured in Spitfire and Hurricane model tests in the CAT
(26)
and those measured on the full-scale aircraft. Collar notes that the Hurricane model
lacked such features as the air intake, tail wheel and aerial post whereas all these items were
present on the Spitfire model. Moreover, both models were leak-free and some further
adjustments to his drag totals are allowed in the assessment of model roughness in
comparison with that of the full-scale aircraft. Taking all these matters into consideration,
(26)
Collar reduces his drag totals of Table 2 to 52.6 lb for the Spitfire and 59.8 lb for the
Hurricane. These values are then taken to be the drags at 100 ft/s on full-scale aircraft re-
modelled to be close to the configurations of the CAT wind-tunnel models. The drag
(26)
coefficients, CD, calculated by Collar on this basis are
Spitfire CD = 0.0186 Hurricane CD = 0.0190.
It is seen that the Hurricanes drag has been whittled down rather more than that for the
Spitfire. Since the resulting drags are, by chance, almost in the ratio of the wing areas (238
ft2 for the Spitfire, 264 ft2 for the Hurricane), the CD values turn out to be almost the same. It
must be stressed, however, that these CD values are not those for the actual aircraft; this point
will be dealt with later.

70
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Figure 3 Drag coefficient, CD, variations with Reynolds number, R, for Spitfire &
Hurricane models. Collar (26)

The CD values quoted above are shown as circled crosses in the graphs of Figure 3 in which
the CD values obtained from the CAT model tests, marked as plain crosses, are plotted
against the logarithm (to the base of ten) of the test Reynolds number, R. Care is needed in
reading these graphs since two CD ordinates are used, one for each aircraft model, and are
(26)
displaced relative to each other to avoid collision between the graphs. As Collar argues,
7
whilst the test models Reynolds numbers do not reach the full-scale values of over 10 , the
CATs CD model values extrapolate well to the above full-scale CD values. In this respect,
then, there is no inconsistency between the CAT model and the full-scale results.

Figure 3 illustrates the long-running problem of wind-tunnel testing, which is an inability to

achieve the Reynolds numbers of full-scale aeroplanes in flight. Readers might find it helpful
here to study Appendices 6 and 7 of Reference 35 in order to understand this problem and to
see the pitfalls of not achieving full-scale Reynolds number conditions. The essential point is
that, to be truly representative of full-scale conditions, to be precisely dynamically
equivalent, the model test Reynolds number and that for the actual full-scale aircraft must
have exactly the same value. The Reynolds number, R, represents the ratio of inertia to
viscous forces in a flow field and is given by
R = V l/,
where is the air density, V is the airspeed, l is a representative length scale (the wings root
chord, say) and is the airs viscosity coefficient. Clearly, small models tested at the
relatively low speeds achievable in many wind tunnels fall far short of achieving full-scale
Reynolds number conditions. Even Farnboroughs 24 ft open-jet tunnel mentioned above,
whilst accommodating full-scale aircraft, ran at a flow speed of 100 ft/s, roughly 68 mph,

71
Journal of Aeronautical History Draft 2 Paper No. 2016/03

which is far less than the maximum speeds of most aircraft. One method adopted to reduce
this large difference in Reynolds number was to increase the air density by pressurisation.
This was the measure adopted for the CAT which could be pressurised to 25 atmospheres,
raising the air density in that ratio. The NACAs variable-density tunnel used a similar
method to obtain the data for the then-new NACA aerofoil sections (16) mentioned in Section 2.
Even so, as Figure 3 illustrates, the test Reynolds numbers achieved in the CAT were roughly
ten times less than those at full scale. The question then was, would extrapolation of such
wind-tunnel results provide realistic full-scale values?

Whilst Collars extrapolations of the CATs results provided a welcome affirmative answer,
at least in this respect, as noted earlier it must be emphasised that his CD values quoted
immediately above are not those of the actual full-scale aircraft. The latter can be obtained
directly from the totals of Table 2 and according to this authors calculations are
Spitfire CD = 0.021 Hurricane CD = 0.026.
Deducting the induced drag contributions of Table 2, the drag coefficients at zero lift, CD0,
are
Spitfire CD0 = 0.020 Hurricane CD0 = 0.025.
It is intended to include these results in a later paper which will review aerodynamic data
from this and earlier eras.

4. Other Aerodynamic Matters

4.1 Surface Finish

In his article included in References 11 and 12, Shenstone refutes accusations that
Supermarine had simply copied the Heinkel 70s elliptic wing. As he rightly points out, the
Heinkels wing planform is not the semi-elliptic shape chosen for the Spitfire and the
Heinkels thicker aerofoil section differs significantly from the Spitfires thin NACA section.
He admits, however, that the one thing he wished to copy, in a sense, was the Heinkels
smooth surface finish. Impressed by this when he had seen it at the Paris Aero Show in 1934,
he recommended that its flush-rivet construction be adopted for the Spitfire. Thus the
Spitfire prototype, K5054, emerged in March 1936 with this smooth finish. However, this
riveting process was more time-consuming and expensive than the conventional method
using dome-headed rivets. To check on the efficacy of this smooth finish, split peas were
glued to the flush rivet heads using Seccotine, a popular glue at the time. Reference 10
includes Shenstones two photographs of K5054 in this condition, its surface appearing to
have acquired a regularly arrayed chickenpox infection. In January 1937 flight tests were
conducted by Quill in which the split peas were removed in progressive stages from nose to
tail (10). Ernest Mansbridge retained the results which are included in Price (11) and these are
listed in Table 3.

72
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Table 3
The loss of speed due to fixing split peas all over the aircraft is 22 mph
This is made up as follows:
Leading edge aft to spar at chord 8 mph
Transverse rows on wing, fuselage & tail* 4.5 mph
Fore & aft rows on top surface of wing* 5.5 mph
Fore & aft rows on under surface of wing* 3 mph
Fore & aft on fuselage 1 mph
*Aft of spar on wing

Clearly, the choice of flush-riveting was vindicated by this exercise. At first sight, the
detrimental effect of conventional riveting seems surprising, given Collars contention (see
Section 3) that the Spitfires boundary layer would be largely turbulent. However, since the
early 1920s it had been known that surface roughness has a detrimental effect on turbulent
pipe flow. Moreover, in 1934 the investigation by Prandtl and Schlichting (36), in which
various forms of roughness were assessed, established that roughness at the surfaces of flat
plates experiencing turbulent boundary layers significantly increases the boundary-layer drag.

4.2 Propulsion Effects on Aerodynamic Performance

Early flight tests with the Spitfire prototype, K5054, established its maximum speed at
around 335 mph (11, 25). This disappointing result, it was felt, might be due to compressibility
effects on its coarse-pitch twin-bladed wooden propeller. Supermarine already had
experience in this area as a result of its Schneider Trophy work; see Cliftons comments at
the close of Section 5 below. Re-design of the propeller using a smaller t/c ratio and finer
incidence at the tip region improved propeller efficiency, thus raising the maximum speed to
348 mph (11, 25). During subsequent tests with K5054, a number of propeller types were tried
in an attempt to increase further the aircrafts maximum speed. These were mainly fixed-
pitch twin, three and even four-bladed types although a two-pitch De Havilland three-bladed
type was tried. No significant improvements seem to have emerged from this (11).

However, a distinct improvement emerged from work by Rolls-Royce which indicated that
directing the engines exhaust gases directly rearwards would augment thrust. As Price (11)
relates, this thrust augmentation could amount to 70 lb, equivalent to a 70 hp increase at 300
mph. Alternatively, an aeroplane powered by about 1,000 hp flying at 300 mph experiences
around 1,000 lb of thrust. Thus an additional thrust of 70 lb represents a 7% thrust increase,
or nearly 9% when current propeller efficiencies around 80% are taken into account. In
September 1937 ejector exhausts producing this effect were fitted to K5054 and its maximum
speed rose to around 360 mph (11).

Despite the Type 224s Goshawk cooling system having the advantage of requiring no
external radiator, its disadvantages have been mentioned in Section 2. The switch from the

73
Journal of Aeronautical History Draft 2 Paper No. 2016/03

steam-cooling system of the Goshawk and initial PV 12 engines to one based on ethylene
glycol as the coolant now introduced the requirement for a radiator. Since glycol boils at a
little under 200C at normal pressure, its use as a coolant was seen as attractive. However,
glycol was expensive and very prone to leakage. Eventually, a mixture of 70% water and
30% glycol was selected (10).

As to the radiator, cometh the hour, cometh the man, in this case F. W. Meredith of the
RAE (37) with his proposal published in 1935 for a ducted radiator system which would
greatly reduce radiator drag. Merediths proposal rested on two interlinked ideas. The first
reduces the drag of the radiator matrix itself, this varying as the square of the local air speed,
by reducing the latter at the radiator. This is accomplished by capturing the cooling air in a
duct which increases in area up to the radiator, the flow speed dropping progressively
according to the continuity principle (see Appendix 1 of Reference 35) as the duct area
increases. The second idea is to utilise the heat energy released by the radiator. The
combination of the continuity principle and the energy equation indicates that some of the
heat energy can be translated into the kinetic energy of a thrust-producing high-speed jet by
passing the radiator efflux through a convergent duct. This would counteract the residual
drag of the enclosed radiator. The scheme was introduced to the Spitfires design after wind-
tunnel tests on a model duct and wing were conducted at the RAE. These were the only
wind-tunnel tests undertaken for the Spitfires design, and indicated satisfactory performance
(10)
. If Collars estimates of Section 3 are to be relied upon, the combined thrust at 100 ft/s
provided by the ejector exhausts and the radiator, 7.5 lb (see Appendix 1), can be set against
the overall cooling drag of 7 lb (see Table 2), thus making the total engine system virtually
drag-neutral. A further point is that, if this authors estimate of Collars initial value for the
combined exhaust and radiator thrust (Tj 120 lb, see Appendix 1) at maximum speed is
correct, and given the Rolls-Royce value for exhaust thrust of 70 lb (see above), then the
thrust contribution from the radiator is of the order of 50 lb.

The efficacy of the Meredith Effect depends on the area ratios employed in the ducts used. A
greater area ratio for the entry duct produces a greater radiator drag reduction; similarly, a
greater contraction ratio for the jet efflux increases the jet thrust. The drawings of the
Spitfires radiator duct provided by Reference 10 indicate area ratios around 2, these being
limited by the ducts placement beneath the wing. In the case of the P-51 Mustang, in
contrast, the radiators placement at the underbelly of the fuselage offered room for
significantly greater area ratios, perhaps of the order of 4. Consequently the Mustang
benefitted rather more from the Meredith Effect than did the Spitfire. Indeed, Atwood (38),
who was involved in the Mustangs design, claims that the greater implementation of the
Meredith Effect here was more significant in terms of drag reduction than the use of the so-
called laminar flow wing. A further interesting feature of the Mustangs radiator is that its
entry scoop is clear of the fuselage skin so that the entry flow is not contaminated by the
fuselage boundary layer, a feature absent on the Spitfire. Ludwig (39) claims that this
beneficial item in the Mustangs design was suggested by Beverley Shenstone, presumably
when he was working for the British Air Commission in the United States.

Two final points are worth mentioning concerning the Spitfires propulsion system. One is
that production Spitfires did not begin to receive De Havilland variable-pitch propellers until

74
Journal of Aeronautical History Draft 2 Paper No. 2016/03

May 1939 (11) (from the 78th production aircraft onward). These propellers improved take-off
and climbing performance. The second point is that in 1939 the Air Ministry upgraded its
requirement for aviation fuel from 87 Octane to 100 Octane. This enabled higher
supercharger boost pressures to be used, thereby increasing engine performance. In both
cases the year, 1939, is significant here. Whilst the Chain Home radar system was
unquestionably crucial for what was soon to follow, the above two important technical
advances also arrived just in time. This indicates that the preparation for what was to become
the Battle of Britain in the summer of 1940 was a fine-run thing.

4.3 The Turning Radius Problem

Before discussing this, a description of this authors initial involvement might be of interest
and, indeed, cause some amusement. Some years ago, this author and his colleague at
Manchesters Victoria University, Dr Peter Lamont, were approached by Dr Stephen Bungay
who was writing a history of the Battle of Britain (40). Dr Bungay had been puzzled by
certain statements in the popular aviation literature concerning the turning radii of the Spitfire
and the Bf 109. The one source of information immediately to hand quoted limited extracts
from an internal Farnborough report stated to be BA 1604; this turned out to be a report on
seaplane performance. After a fruitless search through the ARCs wartime publications, any
further literature search was abandoned in favour of analysis since a quick response was
needed. The essential steps in this analysis are given in Appendix 2 and lead to the result that
the radius, r, for turning flight at constant speed in the horizontal plane is given by
r = 2w/( g CL sin ). (2)
Here w is the wing loading, the air density, g the gravitational acceleration, CL the lift
coefficient and is the bank angle. The results based on this relation were published in
Reference 41. After publication, entirely by chance it was discovered that the elusive
Farnborough report was BA 1640, not BA 1604; moreover, this report by Morgan and Morris
(13)
had then been published by the ARC. However, although dated September 1940, this had
not been published until after the war, presumably for security reasons.

The important point to note from equation (2) is that turning radius, r, is directly proportional
to w, the wing loading. The Spitfires semi-elliptical wing of substantial area helped in this
respect, producing a wing loading around three-quarters of that for the Bf 109: 24.8 lb/ft2 for
(13) (13, 41)
the Spitfire, 32.2 lb/ft2 for the Bf 109 . The results of both investigations , compared
in Table 4, reflect this; the quantity n in that table is the ratio of lift to weight (see Appendix
2). It should be noted that, whereas the altitude selected in Reference 41 is 10,000 ft, in
Reference 13 this is 12,000 ft. According to equation (2), the decrease in air density, , with
increased altitude will lead to increased turning radii and this is also reflected in Table 4s
results. In both investigations it is assumed that the turning flights occur at maximum power
and at maximum, or near-maximum, CL values. However, Reference 13s investigation
benefitted from the provision of accurate data both for the Spitfire and for the RAEs
captured Bf 109E-3, the latter having been subjected to extensive testing. Reference 41, in
contrast, had to rely on less reliable data coupled to a certain amount of guesswork. For
example, the CL values chosen are less than those of Reference 13 and such differences lead

75
Journal of Aeronautical History Draft 2 Paper No. 2016/03

to differences in the speeds as registered by the pilots air-speed indicator (V(ASI)).

Nonetheless, both investigations lead to the conclusion that the Spitfire is superior to the Bf
109 in turning performance. Moreover, the Spitfire proved superior in other respects, as
Reference 13s extensive investigation of its Bf 109 indicates.

Table 4 Turning Data

Radius/ft n Bank Angle CL V(ASI)/mph
(41)
Spitfire 687 2.7 69 1.35 85
Spitfire (13) 696 2.7 68 1.47 126
(41)
Bf 109 853 2.7 68 1.40 94
(13)
Bf 109 885 2.1 62 1.60 118

The results obtained by Morgan and Morris (13) are derived from analyses by Gates (42, 43). In
1932 Gates (42) produced a general analysis of sustained turning motion which includes
change of altitude (spiralling motion) and the possibility of sideslip. In 1942 he used this
analysis to investigate various dog-fight scenarios (43). For the simpler case of sustained
turning motion at constant altitude discussed here, Gatess analysis reduces to that outlined in
Appendix 2.

Morgan and Morris (44) later considered the influence of altitude on the Spitfires turning
performance, extending their calculations to an altitude of 35,000 ft. As stated earlier, the
turning radius increases with altitude and, to counter this, Morgan and Morris (44) suggest the
use of flaps, settling on an optimum flap angle of 30.

5. The Spitfires Performance at High Subsonic Mach Numbers

Reference 1 included a graph showing the variation of the drag coefficient at zero lift, CD0,
with Mach number for both the Spitfire PR XI and the P-51 Mustang. It is here reproduced
as Figure 4 for ease of reference. The graph was taken from Jeffrey Quills contribution to
Reference 7, the data, dated January 1944, having been obtain from Farnboroughs wartime
high-speed flight test work. However, there is a problem with this graph; one would not
expect variations in CD0 to occur prior to the onset of compressibility effects. Thus the initial
gradual rises in CD0 are suspect and may have arisen from data reduction errors in dealing
with the original flight-test results. This matter is resolved, however, in the large
compendium (45) of Farnboroughs wartime high subsonic research work published in 1950
and edited by William Austyn Mair (1917 - 2008). The rather busy graphs of Figure 5, taken
from that publication, show no increases in CD0 until critical Mach number conditions are
encountered.

Critical Mach number conditions occur when, at some point in the flow field, the local flow
velocity reaches the local speed of sound. These conditions usually occur on wing upper
surfaces, at which point shock waves begin to form there and create an addition drag, wave
drag, causing an increase in CD0. As shown in Figure 5, the onset of this compressibility

76
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Figure 4 Variation of CD0 with Mach number for the Spitfire PR XI and the P-51
Mustang, from Quills contribution to Reference 7

effect occurs well below the point at which the flight Mach number approaches unity, and
depends on the shape of the aircraft. A wings t/c ratio is thus important here and the smaller
its value, the higher is the aircrafts critical Mach number. As Figure 5 shows, the Spitfire
with its thinner wing benefits in this respect compared with the Mustang with its slightly
thicker wing. However, there is a further feature to be seen in both Figures 4 and 5 which is
that the Spitfires rise in CD0 is more gradual than that for the Mustang. At the risk of being

Figure 5 Variation of CD0 with Mach number for various aircraft (45)

77
Journal of Aeronautical History Draft 2 Paper No. 2016/03

over-simplistic, a clue to this difference in behaviour might lie in the basic ideas associated
with the Area Rule.

In the past the discovery of the Area Rule for flight around the speed of sound had been
attributed to Richard Travis Whitcomb (1921 2009) whilst working at the NACA. His first
paper (46) on this appeared in 1952 and the work culminated in his report of 1956 (47).
However, it is now know that the concept originated with Otto Frenzl (1909 1996) around
1943 whilst working at Junkers in Dessau (see Reference 48 for a review of his work). The
Area Rules basic idea is that, from nose to tail, the variation of an aircrafts total cross-
sectional area (including the wing and tail assembly) should be as smooth as possible.
Depicting this variation as a graph, any sudden changes in its gradient should be avoided if
possible since such changes will lead to increased wave drag and an increase in CD0. If such
changes do occur, they should be as small as possible. A frequently cited example is the
behaviour of the prototype Convair F-102 Delta Dagger (1953/54) which proved incapable of
achieving supersonic flight. The remedy provided by Whitcombs work was to waist the
fuselage to compensate for the delta wings cross-sectional area, the result being that, in plan
view, the fuselage acquired a Coke Bottle shape. As a result, the F-102 was able to achieve
supersonic flight.

Applying the above idea to the Spitfire and Mustang, graphs of cross-sectional area variation
aft from the nose would show smooth variations until the wings are reached. At that point
sudden bumps in the curve will occur and thus create increased wave drag. But the bump for
the Spitfire, with its thinner, broader chord wing of elliptic taper, will be more gradual than
that for the Mustang with its slightly thicker, smaller chord wing. Basically the rate of
change of cross-sectional area around the region of the wing is less marked for the Spitfire
than is the case for the Mustang. Furthermore, for the Mustang there will be an additional
small bump due to the underbelly radiator intake. This argument might, it is suggested,
provide a clue to the difference in high subsonic behaviour of the two aircraft. That said, it is
nonetheless remarkable that, as shown in Figure 5, there should be so large a difference in
CD0 behaviour between the aerodynamically cleaner Mk 21 Spitfire with its cannon and
ammunition blisters removed and the standard version.

The switch from the PR XI Spitfire of Figure 4 to the Mk 21 of Figure 5 was due to two
serious accidents involving the former Spitfire mark. On the first occasion, in May 1944 as
part of an instrumentation check, the Spitfire was in a very steep dive up to a Mach number
(45)
of 089 . A loss of oil pressure to the constant speed propeller unit resulted in the
propeller severely over-speeding. The reduction gear and propeller were torn away, thus
preventing any further tests from being made on this aircraft. On the second occasion a
(11)
month later, as Price relates , the replacement PR XI suffered a burst supercharger and
engine fire. On both occasions the pilot, Squadron Leader Martindale, not only survived but
also managed to save the flight data. This stands as a notable tribute to his skill and courage;
clearly the RAEs high subsonic flight test programme involved exacting and dangerous
work. The final replacement for the PR XI, the Mk 21 with guns and blisters removed, is
judged in Reference 45 to be not quite as clean aerodynamically as the PR XI.

78
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Two versions of the Gloster Meteor appear in Figure 5, the Mk I and one labelled Mk IV.
However, the latter is an interim type denoted in Reference 45 as Meteor IV, signifying
that it is a pseudo-Mk IV, aerodynamically close to the production Mk IV but in fact a
modified Mk I. Having experienced buffeting at high subsonic Mach number with the Mk I,
the RAE deduced from in-flight wool tuft tests that the problem lay in the appearance of
shock waves around the short engine nacelles which infected the wing flow there. The RAE
suggested lengthening the nacelles, a measure adopted for the Meteor IV and production
Mk IV, and this alleviated the problem. It is seen in Figure 5 that the Meteor IV
approaches the cleaner Spitfires performance and does rather better than the first British jet-
propelled aircraft, the Gloster E28/39.

As to the other aircraft appearing in Figure 5, the P-38 Lightning scarcely makes it onto the
graph, whilst the twin-engine high-altitude Westland Welkin (maximum t/c ratio 21%)
produced occasional oscillatory behaviour hence the two graphs - which, as far as this
author is aware, was never resolved. There remains the P-47 Thunderbolt and, in
(25)
commenting on this in comparison with the Mustang and the Spitfire, Quill remarks that
this was in the also ran category. According to the Area Rule idea outlined above, the
Thunderbolt, with its blunt nose and consequent sudden jump in cross section area there,
might be expected to do badly.

In the above discussion there is no intention to suggest that aircraft such as the Spitfire and
the Mustang were in any way Area-Ruled. With the benefit of hindsight it could be argued
that the nacelle lengthening on the Meteor was in accord with Area Rule ideas, but it should
be stressed that at that time the RAE, indeed the Allied Powers, had no knowledge of the
concept then being explored by Otto Frenzl. As to the Spitfire specifically, when asked in the
discussion following his lecture recorded in Reference 7 if compressibility effects were
considered in the Spitfires design, Alan Clifton replied, Nothing, Id say! Later he
amplified that remark by stating that in the late 1920s Supermarine had evolved guess-work
corrections for compressibility effects on the tip sections of the Schneider Trophy aircrafts
metal propellers. He concluded with the statement, However, we certainly saw no reason to
apply this data to the Spitfire wing.

6. Concluding Remarks

This paper has attempted to describe the basic aerodynamic ideas incorporated in the
Spitfires design. The emerging picture is one in which few aerodynamic compromises
occurred and, indeed, the latest aerodynamic ideas available in the early 1930s were
successfully incorporated. Yet this picture, aerodynamically speaking, is not complete; as
later marks of Spitfire emerged, changes in their aerodynamics, though often slight, did
occur. Moreover, other aspects of the Spitfire story are equally worthy of investigation:
structure and aeroelasticity, stability, control, general handling qualities, engines and
propellers. All of these contributed to the Spitfires phenomenal development (17) in which its
power and weight both roughly doubled whilst its maximum speed rose by about 100 mph.
Perhaps other writers might take up the challenge of addressing these matters.

79
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Since its first appearance in early March 1936, the Spitfire has now reached the venerable age
of eighty, a circumstance which prompted the Spitfire Seminars celebration mentioned at the
beginning of this paper. Through its longevity and historical associations it has become, one
might say, a National Treasure to be displayed at times of major celebration, a part of the
story the British like to tell about themselves. Its crucial importance, particularly during the
dire circumstances of the summer of 1940, is unquestionable. As Salisbury (21) points out, its
highly distinctive planform provided a considerable boost to civilian morale, particularly in
the south-east of England, but had quite the opposite effect on German pilots. British
airspace was seen to be defended, and effectively so. This alone stands as a considerable
tribute to Mitchell and his design team at Supermarine. A further tribute, and again a
considerable one, is the ability of the Spitfires design to answer the exacting developmental
demands placed upon it. As to Mitchells design team itself, a striking feature is its loyalty to
Mitchell, even more so to his memory. And, whilst individual team members freely gave
credit to others for their contributions, Clifton to Shenstone, Shenstone to Joseph Smith and
so on, none showed any sign of claiming credit for themselves.

Left somewhat dazed after this heady trawl through a limited part of the copious Spitfire
literature, perhaps this academic, now almost as old as the Spitfire itself, might be allowed to
bow out with a touch of levity. To plagiarize two well-known slogans and with an apology
for the repeated classic split-infinitive, the Spitfires achievement of a Mach number of 0.9
toward the end of its wartime career touched those parts (of the flight envelope) which other
aircraft could not reach, enabling pilots to boldly go where no (air)man had gone before.

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 Frank Armstrong FREng, FRAeS for
alerting the author to the existence of Collars paper and to Professor Ian Poll FREng, FRAeS
and John Harrington, Cranfield University Library, for obtaining a copy of it and other ARC
material. The author is also grateful to Gordon Bruce for Mustang data. Finally, particular
gratitude must be expressed to Ralph Pegram and Michael Salisbury MRAeS for their
comments on the Spitfire Seminars lecture and for their generous provision of advice and
information.

References

1. ACKROYD, J. A. D. The Spitfire wing planform: a suggestion. J. Aero. Hist., 2013,

pp 121-135.

2. ACKROYD, J. A. D. Letter to the Editor. J. Aero. Hist., 2014, pp 1-2.

3. PEGRAM, R. The Schneider Trophy Seaplanes and Flying Boats: Victors,

Vanquished and Visions. Fonthill Media, Sussex, 2012.

80
Journal of Aeronautical History Draft 2 Paper No. 2016/03

4. PEGRAM, R. Beyond the Spitfire: The Unseen Designs of R. J. Mitchell. History

Press, Stroud, 2016.

5. ZHUKOVSKII, N. E. ber die Konturen der Tragflchen der Drachenflieger.

Zeitschrift fr Flugtechnik und Motorluftschiffahrt, 1910, 1, pp 281-284.

6. BETZ, A. Untersuchung einer Joukowskyschen Tragflche. Zeitschrift fr

Flugtechnik und Motorluftschiffahrt, 1915, 6, pp 173-179.

7. EAST, R. A. and CHEESEMAN, I. C. Forty Years of the Spitfire. Proceedings of

the Mitchell Memorial Symposium, University of Southampton. Royal Aeronautical
Society Southampton Branch, 1976.

8. PEGRAM, R. Letter to the author, September 2016.

9. COWLEY, W. L. and MCMILLAN, G. A. Pressure distribution on wings with

ailerons. ARC, R & M No. 1625, 1934.

10. MORGAN, E. B. and SHACKLADY, E. Spitfire: The History. Key Publishing,

Stamford, Lincs, 1987.

11. PRICE, A. The Spitfire Story. Janes, London, 1982.

12. PRICE, A. Spitfire: A Documentary History. Macdonald and Janess, London, 1977.

13. MORGAN, M. B. and MORRIS, D. E. Messerschmitt Me. 109. Handling and

manoeuvrability tests. ARC, R & M No. 2361, 1940.

14. DICKSON, R. S. Papers of E. J. Davis, archived at the RAF Museum Hendon.

15. JACOBS, E. N., WARD, K. E. Tests of N. A. C. A. airfoils in the variable density

wind tunnel. Series 24. NACA TN 404, 1932.
http://naca.central.cranfield.ac.uk/reports/1932/naca-tn-404.pdf

16. JACOBS, E. N., WARD, K. E. and PINKERTON, R. M. The characteristics of 78

related airfoil sections from tests in the variable-density wind tunnel. NACA Report
No. 460, 1932.
http://a.moirier.free.fr/A%E9rodynamique/Profils/NACA/Rapports%20NACA/NAC
A_N%B0%20460%20Characteristics%20of%2078%20airfoils%20~%204%20digits.
pdf

17. SMITH, J. The development of the Spitfire and Seafire. JRAeS, 1947, 51, pp 339-
383.

18. GLAUERT, H. Aerofoil and Airscrew Theory. Cambridge University Press, 1926.

81
Journal of Aeronautical History Draft 2 Paper No. 2016/03

19. ANDERSON, J. D. Introduction to Flight, Its Engineering and History. McGraw-

Hill, New York, 1978.

20. LOFTIN, L. Quest for performance. The evolution of modern aircraft. NASA SP
468, NASA, Washington DC, 1985.
https://ntrs.nasa.gov/search.jsp?R=19850023776&hterms=Focke-
Wulf&qs=Ntx%3Dmode%2520matchall%26Ntk%3DAll%26N%3D4294959618%26
Ntt%3DFocke-Wulf

21. SALISBURY, M. Letter to the author, September 2016.

22. FADDY, D. The forgotten Spitfire designer. Aeroplane Monthly, July 2008, 36, (7),
pp 54-58.

23. MUTTRAY, H. Die aerodynamische Zusammenfgung von Tragflgel und Rumpf.

Luftfahrtforschung, 1934, 11, (5), pp 131-139. [The aerodynamic aspect of wing-
fuselage fillets. NACA Tech. Memo. No. 764, 1935.]
https://ntrs.nasa.gov/search.jsp?R=19930094653

24. KLEIN, A. L. Effect of fillets on wing-fuselage interference. Trans ASME, 1934, 56,
(1), pp 1-10.

25. QUILL, J. Spitfire. A Test Pilots Story. John Murray, London, 1983.

26. COLLAR, A. R. The performance of the Hurricane and Spitfire aeroplanes.

ARC, Ae 1674, June 1940.

27. 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

28. KRMN, T. VON. Mechanische hnlichkeit und Turbulenz. Nach der kgl
Gesellschaft der Wiss zu Gttingen, Math-Phys Klasse, 1930, pp 58-76. [Mechanical
similitude and turbulence. NACA Tech. Memo. No. 611, 1931]
https://ntrs.nasa.gov/search.jsp?R=19930094805

29. KRMN, T. VON. Turbulence and skin friction. JAS, 1934, 1, pp 1-20.

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. YOUNG, A. D. Boundary Layers. BSP Professional Books, Oxford, 1989.

32. FOZARD, J. W. (Ed). Sydney Camm and the Hurricane. Airlife, Shropshire, 1991.

82
Journal of Aeronautical History Draft 2 Paper No. 2016/03

33. RELF, E. F., JONES, R. and BELL, A. H. Tests of six aerofoil sections at various
Reynolds numbers in the compressed air tunnel. ARC, R & M No. 1627, 1934.

34. RELF, E. F., JONES, R. and BELL, A. H. Tests of six aerofoil sections at various
Reynolds numbers in the compressed air tunnel. ARC, R & M No. 1706, 1936.

35. ACKROYD, J. A. D. Babinskys demonstration: The theory of flight and its historical
background. J. Aero. Hist., 2015, pp 1-42.

36. PRANDTL, L. and SCHLICHTING, H. Das Widerstandsgesetz rauher Platten.

Werft, Reederei, Hafen, 1934, 14, pp 1-4.

37. 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.

38. ATWOOD, J. L. We can build you a better airplane than the P-40. Aeroplane
Monthly, 1999, 27, (5), pp 30-37.

39. LUDWIG, P. A. P-51 Mustang: Development of the Long-Range Escort Fighter.

Classic Publications, Sussex, 2003.

40. BUNGAY, S. F. The Most Dangerous Enemy. A History of the Battle of Britain.
Aurum Press, London, 2000.

41. ACKROYD, J. A. D. and LAMONT, P. J. A comparison of turning radii for four

Battle of Britain fighter aircraft. Aeronaut J, February 2000, 104, (1032), pp 53-58.

42. GATES, S. B. Aircraft turning performance. Part 1. ARC, R & M No. 1502, 1932.

43. GATES, S. B. Notes on the dog fight. ARC, R & M No. 2381, 1940.

44. MORGAN, M. B. and MORRIS, D. E. Notes on the turning performance of the

Spitfire as affected by altitude and flaps. ARC, R & M No. 2349, 1941.

45. MAIR, W. A. (Ed). Research on high speed aerodynamics at the Royal Aircraft
Establishment from 1942 to 1945. ARC, R & M No. 2222, 1950.

46. WHITCOMB, R.T. A study of the zero-lift drag-rise characteristics of wing-body

combinations near the speed of sound. NACA RM L52H08, 1952.

47. WHITCOMB, R. T. A study of the zero-lift drag-rise characteristics of wing-body

combinations near the speed of sound. NACA Report No. 1273, 1956

48. MEIER, H-U. (Ed). German Development of the Swept Wing 1939-1945. AIAA,
Reston, USA, 2010.

83
Journal of Aeronautical History Draft 2 Paper No. 2016/03

Appendix 1

Collars Calculations
(26)
The following analysis, not present in Collars paper , is this authors attempt to provide
the basis for Collars calculations.

For an aeroplane flying straight and level at constant speed, V, the thrust, T, must equal the
drag, D, and lift, L, must equal the aeroplanes weight, W. Thus
T = D, L = W.
However, there are two contributions to T, one being from the engine/propeller combination,
Te, the other being the jet thrust provided by the engine exhaust and heat regeneration in the
radiator, Tj. Thus
Te + Tj = T = D,
(Te + Tj)V = DV.
For a propeller efficiency, , and engine power, P,
TeV = P.
The drag, D, can be written in terms of the drag coefficient, CD, as
D = V2 S CD,
where is the air density and S the wing planform area. Thus,
P + Tj V = V3 S CD.
It follows that
P /(V3) + Tj/(V2) = S CD.
Writing = /0, 0 being the sea-level air density, then
P /(V3) + Tj/(V2)= 0 S CD.
However,
0 S CD (100)2 = D100,
which is the drag at sea level at a speed of 100 ft/s. Consequently,
104 P/ ( V3) + Tj,100 = D100,
Tj,100 = 104Tj/( V2), (A1)
the latter being the exhaust/radiator thrust at a speed of 100 ft/s. Thus,
104 P/ ( V3) = Te,100 (A2)
is the engine/propeller thrust at 100 ft/s.

84
Journal of Aeronautical History Draft 2 Paper No. 2016/03

This last result (equation A2) is Collars starting point for his calculations (26). He uses it in
what he calls the usual procedure to scale the data for the maximum speed and
corresponding engine power conditions at 18,000 ft altitude to a common speed of 100 ft/s at
sea level. This yields his estimates for engine/propeller thrust in lb for both aircraft, as shown
in the table below. To these he adds his estimates for Tj,100, the thrust due to the exhaust and
the radiators heat regeneration at sea level and at the speed of 100 ft/s. He says these
estimates have been obtained by scaling an unstated value, but common to both aircraft, at
the maximum speed conditions. Using equation (A1), that unstated value can be retrieved by
scaling up the values for Tj,100 shown in the table to the maximum speed conditions. This
yields a Tj value of about 120 lb for both aircraft.

Collars totals given in the table below are then those totals appearing in the drag balance
sheets of Table 2, to which he gives a probable error of 3 lb.

Spitfire Hurricane
Engine/propeller thrust, Te,100 51.5 73.0
Exhaust & heat regeneration, Tj,100 7.5 9.0
Total 59.0 82.0

Appendix 2

Turning Flight

For an aeroplane in straight and level flight at constant speed, the lift, L, opposes the
aeroplanes weight, W = mg, m being the aeroplanes mass, g the gravitational acceleration.
Suppose now that the aeroplane banks through the angle from the horizontal whilst
maintaining its speed, V, and its altitude. Since the lift force continues to be perpendicular to
the plane of the wings, the lift is now directed at the angle to the vertical. The lift forces
vertical component, L cos , is consequently now less than the weight, W. To correct this,
the wing incidence angle must be increased so that the lift force increases to the extent that
L cos = W.
There is also the lift forces horizontal component, L sin , which provides the centripetal
force causing the aeroplane to follow a circular path in the horizontal plane. A body moving
in a circular path of radius r experiences a continual acceleration towards the circles centre, a
centripetal acceleration, equal to V2/ r. Consequently, since force is equal to the product of
mass and acceleration,
L sin = m V2/r.
The lift force can be written in terms of the lift coefficient, CL, as
L = V2 S CL,
being the air density and S the wing planform area. Rearranging, we obtain

85
Journal of Aeronautical History Draft 2 Paper No. 2016/03

r = 2w/( g CL sin ),
where w = W/S is the wing loading. The crucial point is that the radius of turn is directly
proportional to the wing loading.

We wish the aeroplane to retain its altitude so as to turn in a horizontal plane. As explained
earlier, it is then necessary to increase the lift, L, by increasing the wing incidence angle. Yet
this entails an increase in the value of CL. Two potential aerodynamic problems arise here.
The first is that CL might be increased to its stall value, a difficult situation for the pilot. The
second is that the CL increase increases the induced drag, its coefficient being given by
equation (1) of Section 2, that is
CDi = k CL2 /( A),
A being the wing aspect ratio equal to b2/ S, where b is the wings span. The total drag
coefficient for an aeroplane can be written as
CD = CD0 + CDi,
where CD0 is the drag coefficient at zero lift. For the fighters under discussion, a typical
value of CD0 is around 0.02, the aspect ratios being close to six. For straight and level flight
at high speed the induced drag contribution is very small since the aeroplane is flying at a low
incidence/low CL value (see the values in Table 2 in Section 3). However, in the turn, taking
a CL value comfortably below stall, say CL = 1, then CDi has a value around 0.05.
Consequently, in executing the turn whilst maintaining the speed V, the aeroplanes overall
drag coefficient has increased from a little over 0.02 (see the results quoted at the end of
Section 3) to around 0.07 and the drag force has increased in that proportion. The question
then is: has the aeroplane sufficient engine thrust/power to sustain this speed?

There are further difficulties in turning motion and these are associated with the load factor,
n, given by
n = L/W = 1/cos .

(42)
Gates , for example, points out that in a sustained turn a pilot will experience blackout in
vision at a value of n around 6 ( 80) whereas at a value of n around 10 ( 84)
structural failure is likely to occur.

86