Annual Review of Fluid Dynamics Vol 1, Goldstein S.

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.
org
Access provided by 179.154.162.28 on 03/01/16. For personal use only.
ANNUAL
REVIEWS
Quick links to online content
Further
FLUID MECHANICS IN THE FIRST HALF

OF THIS CENTURY
By SYDNEY GOLDSTEIN
Harvard University, Cambridge, Massachusetts
Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

THE FIRST DECADE
The opening decade of the twentieth century was a period of progress in

fluid mechanics. M uch was begun that was to continue for many years. Most
of the great discoverers in the latter part of the previous century were to be
with us still during the decade, and other younger men were to come for
ward. Let us start by listing some of the happenings.
In 1900 and 1901 Benard published descriptions of his experiments on
the convective motion of a fl uid heated from below. In 1901 Levi-Civita
published his first note on the explanation of the resistance of a solid body
held in a stream of fluid by postulating the existence of surfaces of discon
tinuity of velocity, even when the body is of rounded form, without sharp
edges. A paper on lifting forces in streams of fluid, by Kutta, appeared in
1902 ; this contained a solution for the two-dimensional flow of an inviscid
fluid past a solid surface in the shape of a circular are, at zero incidence, with
circulation round the surface and a finite velocity at the trailing edge.
Sir George Gabriel Stokes, who was born in 1819, was chosen as Master of
Pembroke College, Cambridge in 1902, and died on February 1, 1903.
C haplygin's 1902 doctoral dissertation on gas j ets, with the hodograph
transformation of the equations of steady two-dimensional gas flows, and
the application to j ets, was published in 1904.
In 1904 Prandtl read his paper Htiber Fltissigkeitsbewegung bei sehr
kleiner Reibung" to the Third International Congress of Mathematicians
at Heidelberg. In the same year Lord Kelvin published three papers o n
"Waves on water" (with o n e more in each o f the years 1905 and 1906) . The
classical papers of Sommerfeld and Michell on the hydrodynamical theory of
l ubrication appeared in 1904 and 1905, and Ekman's paper "On the influence
of the earth's rotation on ocean currents" in 1905 .
Zhukovskii's famous lift theorem, connecting the lift force with t h e cir
culation quite generally for the two-dimensional flow of an inviscid fluid,
was published in two notes in 1906, one in Russian and one in French. I n
1907 and 1908 the works of Orr and Sommerfeld o n the stability of fluid
motions appeared. Lanchester's book Aerodynamics was published in London
in 1907, and a German edition followed in 1909. Near the beginning of 1 9 1 0
Chaplygin formally enunciated t h e postulate that out o f t h e infinite number
of theoretically possible flows (depending on the magnitude of the circula
tion) past an airfoil profile with a sharp trailing edge, the flow that is nearest
to experiment is the one with a finite velocity at the trailing edge. Also near

GOLDSTEIN
the beginning of 1910 Kutta finally published his 1902 dissertation, with
revisions, and toward the end of 1910 Zhukovskii published a note on the
design of airfoil sections, largely devoted to a graphical construction of the
profiles.
In 1910 also Rayleigh and Taylor showed clearly how dissipative pro
cesses removed infinite gradients but left sharp transitions in shock waves i n
gases. Controversy, mystification, and hesitation were ended. I n the same
year Oseen put forward a "cure" for the "paradoxes" of Stokes and White
head.
Each professional worker and scholar may find fault with the items I
have chosen for my list, but none will fail to recognize the meaning and sig
nificance of those I have included. None can deny that it was a fruitful period.
It is an interesting game to try to guess now what a similar list written in
20 1 8 would contain for the happenings of the first eleven years of the second
half of this century, 1950-1960. But perhaps it is an even more interesting
and certainly more instructive-exercise to ru minate on what the list for
1900 to 1910 would have contained if compiled fifty years ago, in 1 9 1 8.
M uch-very much-would have been missing, for not only was there a
grievous slowness of communication, but much of what is listed was so new
and unfamiliar that it escaped attention or merely induced disbelief or doubt,
except in the cases of rare individuals. It was to' be a few more years after
1 9 1 8-when the first world war was j ust ending-before the new discoveries
were to begin to have their full effect.
For the sake of orientation let us take a brief glance ahead from the end of
1910. I t was in 1 9 1 1 , with further publications in 1 9 1 2 , that Karman pub
lished his first paper on the vortex street i n the wake for two-dimensional
flow past a cylinder and the mechanism of resistance. [Benard had shown
evidence of the vortex street in 1908 ; there had been mention of such vor
tices before that by Mallock in 1907, by Ahlborn in 1902 and M arey in 190 1
(with photographs) , and by Reynolds in 1 883 (with drawings) . M allock
published a second paper in 1910.] Osborne Reynolds lived u ntil 1 9 1 2 ,
Rayleigh until 1 9 1 9 , and Zhukovskii until 1 9 2 1 .
RAYLEIGH S R EVIEW O F LAMB S HYDRODYNAMICS-GENERAL REMARKS
'
'
The third edition of Lamb's Hydrodynamics, of which the first edition

appeared in 1879 and the second in 1895, was published in 1906 and the fourth
edition in 1916. Rayleigh reviewed the fourth edition for Nature, and his
review tells much of the state of the science and is worth quotation. Some
excerpts follow.
That this work should have already reached a fourth edition speaks well for the
study of mathematical physics. By far the greater part of it is entirely beyond the
range of the books available a generation ago. And the improvement in the style is
as conspicuous as the extension of the matter. My thoughts naturally go back to the
books in current use at Cambridge in the early sixties. With rare exceptions, such
FLUID MECHANICS-FIRST HALF OF THIS CENTURY
as the notable one of Salmon's Conic Sections and one or two of Boole's books, they
were arid in the extreme, with scarcely a reference to the history of the subject
treated, or an indication to the reader of how he might pursue his study of it . ..

The progressive development of his subject is often an embarrassment to the

writer of a text-book. Prof. Lamb remarks that his "work has less pretensions than
ever to be regarded as a complete account of the science with which it deals. The
subject has of late attracted increased attention in various countries, and it has
become correspondingly difficult to do justice to the growing literature. Some
memoirs deal chiefly with questions of mathematical method and so fall outside the
scope of this book; others though physically important hardly admit of a condensed
analysis; others, again, owing to the multiplicity of publications, may unfortunately
have been overlooked. And there is, 1 am afraid, the inevitable personal equation
of the author, which leads him to take a greater interest in some branches of the
subject than in others. "
Most readers will be of opinion that the author has held the balance fairly. Formal
proofs of "existence theorems" are excluded. Some of these, though demanded by
the upholders of mathematical rigour, tell us only what we knew before, as Kelvin
used to say. Take, for example, the existence of a possible stationary temperature
within a solid when the temperature at the surface is arbitrarily given. A physicist
feels that nothing can make this any clearer or more certain. What is strange is
that there should be so wide a gap between his intuition and the lines of argument
necessary to satisfy the pure mathematician. ...
Naturally a good deal of space is devoted to the motion of a liquid devoid of
rotation and to the reaction upon immersed solids. When the solids are "fair"
shaped, this theory gives a reasonable approximation to what actually occurs; but
when a real liquid flows past projecting angles the motion is entirely different, and
unfortunately this is the case of greatest practical importance. The author, follow
ing Helmholtz, lays stress upon the negative pressure demanded at sharp corners in
order to maintain what may be called the electric character of flow. This explana
tion may be adequate in some cases; but it is now well known that liquids are
capable of sustaining negative pressures of several atmospheres. How too does the
explanation apply to gases, which form jets under quite low pressure differences?
It seems probable that viscosity must be appealed to. This is a matter which much
needs further elucidation. It is one on which Kelvin and Stokes held strongly
d ivergent views. . ..
It would not have accorded with the author's scheme to go into detail upon
experimental matters, but one feels that there is room for a supplementary volume
which should have regard more especially to the practical side of the subject. Per
haps the time for this has not yet come. During the last few years much work has
been done in connexion with artificial flight. We may hope that before long this
may be coordinated and brought into closer relation with theoretical hydrody
namics. In the meantime one can hardly deny that much of the latter science is out
of touch with reality.
Rayleigh's review is a remarkab ly interesting review and short essay. I n

conjunction with m y 1900 to 1910 list and the remark, written with the bene
fit of hindsight, that much in the list was so new and unfamiliar that i t
escaped attention o r merely i n duced disb elief or doubt, it gives an illuminat
ing picture of the general state of affairs in the science in 1916.

GOLDSTEIN
Lamb's textbook was the predominant " high-class" textbook for many
years, certainly in England in the 1920's for students of applied mathematics,
but we were not really happy with it. Rayleigh may have contrasted it with
the "arid" textbooks of the 1 860's, but in the 1920's we were complaining
that it was impossible to remember while reading Lamb that water is wet.
Something of the same dry atmosphere still persists.
By the end of the first half-century there was a stronger and more wide
spread element of physics in thought and research on fluid mechanics than
in the first twenty or thirty years, and this is much more so by now. Several
factors and several research workers contributed to this, but the greatest in
fluence has been the example of G. I. Taylor.
Fluid mechanics is a part of applied mathematics, of physics, of many
branches of engineering, certainly civil, mechanical, chemical, and aeronau
tical engineering, and of naval architecture and geophysics, with astrophysics
and biological and physiological fluid dynamics to be added. Significant
contributions to the theory of airfoils came early in the century, and during
the whole of the first half of the century applied aerodynamics was to be
probably the major incentive, dealing with questions that were important
also in mechanical and civil engineering ; but geophysical q uestions, certainly
not without charm and fascination, received much attention.
Curiosity about at least two of the branches of fluid mechanics and their
applications has a long and distinguished history, for in the Proverbs of
Solomon the son of David, king of Israel, it was stated in the words of Agur
the son of Jakeh that "There be three things which are too wonderful for
me, yea four which I know not," of which two were "The way of an eagle
in the air" and "The way of a ship in the midst of the sea," which I take to
be questions of aerodynamics and naval architecture, questions that con
cern us still.
There is now no dearth of books, some very good and many satisfactory,
on the various branches of our science, in which the details of the work done
u p to 1950 may be read. So I propose not to attempt to be systematic or
even to discuss several significant advances, but to plead, in Lamb's words,
" the inevitable personal equation of the author, which leads h i m to take a
greater interest in some branches of the subject than in others." Let us go
back to 1900.
THE RESISTANCE OF FLUIDS. THE SURFACES-OF-DISCONTINUITY THEORY
Probably the outstanding difficulty was still that of accounting for the
resistance of a solid body in motion relative to a fluid in which it is i mmersed.
The three known theories were not satisfactory, though there was still con
siderable difference of opinion about the last of them. According to the ideas
of Isaac Newton, the force on a flat plate in a two-dimensional motion would
be wholly normal to the plate, and proportional to the square of the sine of
the angle of incidence, i.e., the angle between the relative velocity of fluid

and solid and the trace of the plate. Later it was held that in an inviscid
liquid, because of the theorems of Lagrange and Kelvin, a motion started
from rest would be irrotational, and then there would be no force on an
isolated body moving at a sufficient distance from any boundaries when the
motion is steady, the influence of the motion of the fluid being completely
allowed for by a modification of the inertia of the solid. The next theory in
volved surfaces of discontinuity of tangential velocity beginning at sharp
corners, and in particular from the edges of a flat plate in a stream. Lamb
writes (pp. 641, 642 of the fifth edition of Hydrodynamics, 1924):
The absence of resistance, properly so called, in such cases is often referred to by
continental writers as the 'paradox of d'Alembert. '
on exact theoretical lines, a result less opposed to ordinary experience is contained
in the investigations of Kirchhoff and Rayleigh relating to the two-dimensional
form of the problem of the motion of a plain lamina. It is to be noticed that the
motion of a fluid in such problems is no longer strictly irrotational, a surface of dis
continuity being equivalent to a vortex-sheet.
Rayleigh's 1 876 paper "On the Resistance of Fluids" begins by observing
that "there is no part of hydrodynamics more perplexing to the student
than that which treats of the resistance of fluids." Lamb's paragraph on the
" Resistance of Fluids" begins by observing that :
This subject is important in relation to many practical questions, e.g. the propul
sion of ships, the flight of projectiles, and the effect of wind on structures. Although
it has recently been studied with renewed energy, owing to its bearing on the prob
lems of artificial flight, our knowledge of it is still mainly empirical.
" Resistance" was a term used for the total force, not the drag force,
which was "the resolved part of the resistance in the direction of the stream."
It is interesting that Rayleigh compared, in part, the results of the theory
for a flat plate at angles of incidence between 100 and 900 with measurements
by Vince published in the Philosophical Transactions of the Royal Society in
1798. Vince had measured the "drag," and Rayleigh divided the measured
results by the sine of the angle of incidence to obtain the "resistance," and
compared the ratios of the resistance at an angle of incidence of 900 to the
resistance at other angles (a) down to 10. He then remarked :
The result of Vince's experiments agrees with theory remarkably well and the
contrast with sin2 a is especially worthy of note. The experiments were made with a
whirling machine and appear to have been carefully conducted ; but they were on
too small a scale to be quite satisfactory. The subject might now be resumed with
advantage.
However, the theory was by no means winning complete acceptance. I n
particular Sir William Thomson, later Lord Kelvin, was quite unconvinced.
Rayleigh wrote :
It was observed by Sir William Thomson in Glasgow that motions involving a
GOLDSTEIN
surface of separation are unstable. This is no doubt the case

But it may be
doubted whether the calculations of resistance are materially affected by this cir
cumstance as the pressures experienced must be nearly independent of what hap
pens at some distance in the rear of the obstacle, where the instability would first
begin to manifest itself. . . . The formulae proposed in the present paper are also
liable to a certain amount of modification from friction which it would be difficult
to estimate beforehand, but which cannot be very considerable if the experiments
of Vince are to be at all relied on.

Kelvin seems to have been more and more unconvinced. I n 1894 he pub
lished four notes on the question of resistance in Nature, which are repro
duced in Volu me 4 of his Mathematical and Physical Papers, with a note by
the editor, Sir Joseph Larmor, that "These com m unications formed the sub
j ec t of a prolonged playful controversy between Lord Kelvin and his intimate
friend Sir George Stokes, in a series of letters which have been preserved."
Kelvin showed that the results of the surfaces-of-discontinuity theory for a
flat plate were not in agreement with the experiments of Dines, published in
the Proceedings of the Royal Society in 1 890. The editor of the Mathematical
and Physical Papers enquired from the Director of the National Physical
Laboratory, Dr. T. E. Stanton, about the reliability of Dines's results and
about the results of more recent investigations, before republishing Kelvin's
papers in 19 10. Stanton wrote that, when allowance was made for differ
ences in size of plates, Dines's results compared well with more recent mea
surements at the National Physical Laboratory, and that the excess in the
total resistance over that given by Lord Rayleigh's formula was due to the
suction effect of the eddies on the leeward side.
Levi-Civita's first note on the extension of the surfaces-of-discontinuity
theory to flows past solid bodies with curved boundaries, for which we note
that negative pressures need not enter, was published in 190 1 , and the gen
eral mathematical theory followed in 1906. H e sought to justify the math
ematical investigation by the hope of practical comparisons, for example
to ships, but also by comparison with p ublished experimental results. It is
interesting that the observations to which he referred (by M arey and by
Ahlborn) are among those which have already been mentioned as showing
evidence of vortices in the wake. Levi-Civita in fact remarked that in the
actual wake there are vortical and turbulent motions, and that at a certain
distance from the body the state of motion of the fluid no longer presents any
vestige of discontinuity, but argued that the resistance depends only on the
state of motion in the near part of the wake in contact with the body, and
that this state of motion is perceptibly the same in both the actual wake and
his hypothetical one, so the value of the resistance resulting from his cal
culations should closely approximate the correct value, in spite of the dif
ferences between the actual circumstances and the theoretical model. (This
was essentially Rayleigh's 1876 argument, but Levi-Civita's discussion is
longer and more detailed.) I shall not discuss the applications and further

development of Levi-Civita's method by Cisotti and Villat, and later by

others, except to remark that the mathematics was of both considerable
analytical difficulty and interest, and also that for a curved boundary the
solution is not unique until the positions of separation of the "free stream
lines"-the traces of the surfaces of discontinuity-have been determined,
and that for this there is only one clue, namely, that it may be required that
the curvature of the streamlines at the separation points should be finite,
since for general positions of these points this curvature is infinite. When it
is finite, it is equal to the curvature of the section of the solid boundary.
There was, then, in the first part of this century some confusion about
the validity of the surfaces-of-discontinuity theory for the approximate
calculation of resistance. Whatever else it did, however, it introduced the
notion of attempting to bring the results of the potential theory of the motion
of inviscid fluids into better agreement with observation in fluids of small
viscosity by the introduction of discontinuities in the velocity field, in this
case vortex-sheets, which was to prove a valuable idealization in spite of
instabilities. Of course it exercised also considerable mathematical fascina
tion, and, in England at any rate, it was always learned rather extensively
by students of applied mathematics. I do not know how much time, if any,
is spent on it now. There is still much to be said for teaching it, but if that is
done I hope it is done now, as it was not done earlier, with a full description
of the actual circumstances and discussions of its relations to boundary
layer separation and to the instability of vortex-sheets and layers.
BOUNDARY CONDITIONS. SLIP
I suppose that it is still correct that for practical purposes in most situa
tions our quantitative knowledge of resistance is mainly empirical. However,
much more is understood now of the underlying physical processes, and we
may ask what was the main cause of the difficulties and confusion. Certainly
it did not lie in any lack of intellectual ability of the very distinguished
scientists who wrestled with the problems from Newton to Stokes and
Rayleigh. The real trouble was doubt about the boundary conditions to be
applied at the dividing surface between a solid and a liquid or gas. In the
theory of the irrotational motion of an inviscid fluid, the relative normal
velocity at the surface of an impermeable solid must be zero, and no other
boundary condition is required or can be imposed. For the motion of a
viscous fl uid, on the other hand, according to the dynamical differential
equations published during the period 1 822 to 1 845 ( Navier, 1822 ; Poisson,
1 829 ; Saint-Venant, 1843 ; Stokes, 1845) , another boundary condition is re
quired for a solution. There was doubt and vacillation for a long time. On the
whole, one thing seems to have been agreed: that there is no slip, i.e., no
relative tangential velocity, at the surface of a solid body in the case of a
very slow motion in a viscous fluid ; but all else was in doubt. A note on the
hypotheses and beliefs concerning the conditions at the surface between a

GOLDSTEIN
fluid and a solid body is printed as an appendix at the end of the second
volume of Modern Developments in Fluid Dynamics, and what was written
t here will not be repeated. The discrepancies between the actual motions o f
a real fluid o f small viscosity, when laminar, and t h e results calculated for
the irrotational motion of an inviscid fluid arise mainly, in most cases, from
the condition in a real fluid of no slip at a boundary. If a fluid could slip
freely over the surface of a solid body it would be a very different world.
Those, among them Lamb and Levi-Civita, who have asserted in the past
that viscosity cannot be considered a predominant cause of direct resistance,
were correct in this sense in most ordinary circumstances.
In his 1904 lecture to the International Congress of Mathematicians
Prandtl stated briefly but definitely that by far the most important question
in the problem (of the flow of a fluid of small viscosity past a solid body) is
the behavior of the fluid at the walls of the solid body. H e continued (the
original German has been translated) :
The physical processes in the boundary layer (Grenzschicht) between fluid and
solid body can be calculated in a sufficiently satisfactory way if it is assumed that
the fluid adheres to the wans, so that the total velocity there is zero-or equal to
the velocity of the body. If the viscosity is very small and the path of the fluid along
the wall not too long, the velocity will have again its usual value very near to the
wall. In the thin transition layer (Ubergangsschicht) the sharp changes of velocity,
in spite of the small viscosity coefficient, produce noticeable effects.
I n 1 9 1 2 , in the book on the theoretical bases of aeronautics in which his

lectures were reproduced, Zhukovskii pointed out that for an incompressible
fluid with a uniform coefficient of viscosity the terms in the differential
equations of motion containing the coefficient of viscosity disappear for po
tential motions, so the influence of viscosity when a potential of the velocity
exists can appear only at the walls, where the boundary conditions must be
satisfied. He then remarked, as late as 1912, that there are differing opinions
on the behavior of the flowing fluid at the walls; some investigators thought
there was no motion of the fluid along the walls, while others supposed that
the fluid slips along them. For his part he thought (there was no exact ex
perimental confirmation, but it approximated nearly enough to reality) that
the fluid velocity is zero at the walls and rapidly increases until it becomes
equal to the theoretical velocity, the layer of fluid around the walls being
vortical and also very thin.
At the end of the nineteenth century the most satisfactory theoretical
progress, other than mathematical, had been made in cases where the bound
ary conditions at a solid boundary did not greatly affect the results, as for
waves on water, or for the one case where the condition of zero slip was ac
cepted, slow motion in a viscous fluid, on the theory of Stokes, for a sphere
and also for an ellipsoid , and certain problems of small oscillations. For
waves on water, mention may be made of the effect of a local disturbance
( Poisson, 1 8 1 6 ; Cauchy, 1827 ; Kelvin, 1887), the theory of group velocity
(Stokes in an examination q uestion, Smith's Prize Examination, 1 876;

Rayleigh, 1 88 1 ; Osborne Reynolds, 1877), and the effect of capillarity and
the minim u m wave-velocity (Kelvin, 1 8 7 1 ) . On this subject much of the
later progress has been made somewhat recently.

ROTATING
FLUIDS
The results of the theory of flows of inviscid fluids without vorticity were
widely at variance with observation and practice over most of the important
parts of the velocity field for flow past bluff bodies, because the condition of
zero slip at a solid surface was not satisfied. However, the theory was to prove
its worth for fair-shaped streamlined bodies such as airfoils, nacelles, and
airships, though "improvements" were to be desirable for practical applica
tions after the first calculations. But that is another story, which will be
mentioned later. If we assume the existence of vorticity in a diffused form
and thereafter neglect viscosity, the motion of a solid body in a fluid pos
sessing such vorticity is amenable to calculation only in a few cases. One
such case is motion in a u niformly rotating fluid. I n a series of papers on
this subject between 1 9 1 7 and 1923, G. I. Taylor made certain theoretical
predictions and reported certain experiments of great interest and impor
tance, not only for themselves and their applications but because predic
tions were made that either did not depend on the boundary conditions or
that gave no slip at a boundary. These predictions were verified by experi
ment. In his paper on "Experiments with Rotating Fluids" in 1921 G. 1.
Taylor wrote:
It is well known that predictions about fluid motion based on the classic hy
drodynamical theory are seldom verified in experiments performed with actual
fluids. The explanation of this want of agreement between theory and experiment
is to be found chiefly in the conditions at the surfaces of the solid boundaries of the
fluid.
The classical hydrodynamical theory assumes that perfect slipping takes place,
whereas in actual fluids the surface layers of the fluid are churned up into eddies.
In the case of motions which depend on the conditions at the surface, therefore, no
agreement is to be expected between theory and experiment. This class of fluid
motion, unfortunately, includes all cases where a solid moves through a fluid which
is otherwise at rest.
On the other hand, there are types of fluid motion which only depend to a sec
on dary extent on the slip at the boundaries. For this reason theoretical predictions
about waves and tides, or about the motion of vortex rings, are in much better
agreement with observation than predictions about the motion of solids in fluids.
Some time ago the present writer made certain predictions about the motion of
solids in rotating fluids, or rather about the differences which might be expected
between the motion of solids in a rotating fluid and those in a fluid at rest. The pre
dicted features of the motion did not depend on conditions at the boundaries. It
was therefore to be anticipated that they might be verified by experiment. The
experiments were carried out and the predictions were completely verified.
In view of the interest which attaches to any experimental verification of the-
10
GOLDSTEIN
oretical results in hydrodynamics, and more particularly to verifications of those.

concerning the motion of solids in fluids, it seems worth while to publish photo
graphs showing the experiments in progress. In the second and third part of this
paper further experiments are described in which theoretical predictions are verified

in experiments with water.

The same paper contained a proof, by the consideration of the circulation
round a circuit, that "if any small motion is communicated to a fluid which
is initially rotating steadily like a solid body, the resulting flow must be two
dimensional, though small oscillations about this state of slow motion are
possible," with a footnote that " this is practically the same thing as the
fact previously noted by Proud man, that small steady motions of a rotating
fluid are two-dimensional. " The paper closes with the remark that "In a future
paper the author hopes to discuss what happens in the case when the bound
aries of the fluid move slowly in such a way that three-dimensional motion
must take place."
More, then, was to follow from Taylor himself, and much, much more
from many others. Motion in rotating fluids, partly but not entirely for
geophysical-meteorological and oceanographic-applications, has become
a large branch of the science of fluid dynamics. The experimental results
predicted and observed by Taylor are striking in the extreme, but they are
now adequately described elsewhere, together with further developments.
This subject must be abandoned here for a return to matters previously
introduced.
BOUNDARY LAYERS, B OUNDARY-LAYER SEPARATION,
VORTEX GENERATION
Prandtl's remarks, in his 1904 lecture to the International Congress of
on the boundary condition of no slip and the circumstances
in a thin layer of fluid near a solid wall have already been cited. The lecture
was reprinted in 1927 in a book Vier Abhandlungen zur Hydrodynamik und
Aerodynamik, together with Prandtl's two classic papers of 1 9 1 8 and 1919 on
airfoil theory and Betz's paper of 1919 on the screw propeller with minimum
energy loss. I n this book the paper on the motion of a fluid of very small
viscosity occupies less than eight pages. Much of the lecture was devoted to
showing to the asssembled mathematicians qualitative diagrams of stream
lines and separations for flows past a projecting corner and a circular cylin
der, with the rolling up of vortex-sheets, and then experimentally obtained
photographs of actual flows in similar cases, with boundary-layer separation
and also with boundary-layer suction. Nevertheless, PrandtI found time to
derive the result that in a region of closed streamlines where vorticity had
been produced by the action of a very small viscosity for a very long time,
the vorticity would be constant in a two-dimensional flow, and to announce
that it would be proportional to the distance from the axis of symmetry for
an axisymmetric flow ; to give the boundary-layer equations for steady twoMathematicians,
FLUID MECHAN ICS-FIRST HALF OF THI S CENTURY
11

di mensional flow ; to mention the difficulty for numerical computation arising

from singularities at the wal l ; to state the form of the "similar" solution for
flow along a flat plate, with a first rough computation of the drag; and to
discuss boundary-layer separation in the presence of a pressure i ncreasing in
the stream direction, with a diagram which has been reproduced ever since.
Prandtl wrote (the German has been translated) :
The most important result of the investigation for applications is that in definite
cases the fluid flow will separate from the wall at a place completely determined
by the external conditions. A layer of fluid, which has been set in rotation by the
friction at the wall, makes its way into the free fluid where, causing a complete
transformation in the motion, it plays the same part as a Helmholtz surface of dis
contin uity . When the viscosity coefficient k is altered the thickness of the vortex
layer is altered (it is proportional to v(kl/pu but everything else remains un
changed, so that one can if one will go over to the limit k=O and obtai n al ways the
same flow picture . . . .
Prandtl explained the plausible reason for separation with an unfavorable

pressure gradient, and explained that the consideration of a flow must be
dealt with i n two i nteracting parts, an inviscid flow obeying Helmhol tz's
vortex theorems, and transition layers (boundary layers) at the solid bound
aries in which the motion will be regulated by the free fl uid but which give
the free stream its character by the emission of vortex layers.
This was a most extraordinary paper of less than eight pages. I n 1928 I
asked Prandtl why he had kept it so short, and he replied that he had been
given ten minutes for his lecture at the Congress and that, being still quite
young, he had thought he could publish only what he had had time to say.
The paper will certainly prove to be one of the most extraordinary papers of
this century, and probably of many centuries. Of course, to a limited extent
the existence and nature of a boundary layer and its connection with fric
tional drag had been briefly mentioned before (Rankine, 1 864; Froude, 1874;
Mendeleyev, 1 880) , but it had amounted to very little compared with
Prandtl's contribution ; there were no boundary-layer equations and no
explanation of separation. The influence of Prandtl's boundary-layer theory
has been enormous. It has been used to make clear physical phenomena that
were, or would have been, otherwise baffling or at least murky. It formed the
basis for approximate methods of computation of practical utility. The ideas
were applied to sciences other than fluid dynamics and, in fluid dynamics, to
situations other than those involving a small viscosity. After the conclusion
of the half-century it was extended and generalized and transmuted, es
pecially by Kaplun, Lagerstrom, and their co-workers, into the theory of
singular perturbations for the approximate asymptotic solution of differ
ential equations.
However, for some years after it was published Prandtl's lecture was
almost, if not completely, unnoticed. Perhaps this is not surprising. It was
so very short, and it was published where no one who was likely to appreciate

12
GOLDSTEIN
it might be expected to look for it. In 1908 Blasius published i n a more ac

cessible, more conventional, medium of communication a fuller accou nt of the
derivation of the boundarylayer equations and a detailed i nvestigation of the
flow along a flat plate parallel to a stream, but even after that there was not
exactly a rush of acceptance, exposition, or further investigation of boundary
layer theory. Blasius also began the study of the boundary layer at the
surface of a circular cylinder started from rest. Boltze, in a Gottingen thesis
in 1908, studied boundary layers at the surfaces of bodies of revolution,
especially spheres, and i n 1 9 1 1 H iemenz performed boundary-layer com
putations with an experimentally determined pressure distribution on a
circular cylinder. These also drew little attention.
Lanchester's Aerodynamics had been published in 1907 . In it, among
much else, he found independently that the skin-friction drag would vary
as pl/2U3/2, gave an explanation of separation less detailed than Prandtl's,
gave i ndications that he knew about turbulence in a boundary layer, ex
plained that "a stream-line body is one that in its motion through a fluid
does not give rise to a surface of discontinuity," and expected separation to
be delayed on one side and hastened on the other side of a rotating cylinder
in a stream. This publication presumably drew even less attention.
With the publication in 1921 of Karman's momentum equation and the
Karman-Pohlhausen approximate method of i ntegration, and the publica
tion in 1924 of the experiments of J. M. Burgers and van der H egge Zij nen,
boundary-layer theory at last became the subject of more attention and
a cceptance.
The single reference to Prandtl's boundary-layer theory in Lamb's Hydro
dynamics in 1924 (5th ed.) is interesting. Lamb remarks that in flow past
a cylinder
the central stream-line divides where it meets the surface in front, and then fol
lows the surface for some distance on each side, the motion of the fluid on either
hand being fairly smooth and regular. At a certain stage, however, the stream-line
in question appears to leave the surface, and can no longer be definitely traced, the
space between its apparent continuation and the cylinder being filled with eddies...
An able attempt to trace this phenomenon mathematically has been made by
Prandtl. The region in front of the solid is regarded as made up of two portions, viz.
(1) a thin stratum in contact with the solid, with a rapid variation of relative
(tangential) velocity in the direction of the thickness, and
(2)
an outer region in
which the motion is taken to be irrotational, being practically unaffected by viscos
i ty
Approximate solutions of
the equation
of motion are sought, appropriate to
these two regions, and continuous with one another at the common boundary. The
calculations are necessarily elaborate, but the results, which are represented
graphically, are interesting.
There are references to Prandtl's lecture and to Blasius.

The situation was altered appreciably in the sixth edition of Hydro
dynamics, published in 1932. This contained a new section on boundary-layer
theory, with references to Prandtl and Blasius, as before, and to M ises's

13
treatment in 1927 , and with Lamb's own contribution to the use of the
Karman-Pohlhausen approximate method.
In the late 1920's and early 1930's there was eagerness to use and study
boundary-layer theory, so it was appropriate to print as the motto of
Modern Developments in Fluid Dynamics an extract from the essays of Sir
Francis Bacon ( 1 612) : "For when propositions are denied, there is an end
of them, but if they bee allowed, it requireth a new worke."
With u nderstanding of boundary-layer separation, coupled with knowledge
that when a motion of a solid body is started from rest the fluid does its best
to make the initial motion irrotational without circulation (with a vortex
sheet wrapped round the surface of the solid body) , and also with some
understanding of the rolling-u p of vortex-sheets, the way was clear for
considerable insight into much that was not clear before-the origin of a
large part of the resistance, the process of vortex formation behind a bluff
obstacle, and the origin of the circulation round a lifting airfoil.
Karman pointed olit that a theoretical determination of the velocity
and spacing of the vortices in a vortex street would require an investigation
of the process of vortex formation, and referred to Prandtl's theory of the
motion of fluids with small viscosity for an explanation of the formation of
vortices even in a fluid of vanishingly small viscosity.
I n 1 9 1 2 Zhukovskii had considered the formation of vortices. There was
first the problem of the singularity in the solution at the leading edge of an
airfoil of zero thickness, such as a circular-arc airfoil. Zhukovskii believed
this would simply lead to a vortical thickening. Kutta had also thought a
disturbance arising from the leading edge would not be important. More
generally, there was agreement on the desirability of a rounded nose, and
there does not appear to have been any widespread u neasiness. On the other
hand, Zhukovskii pointed out that vortices could also separate from the
trailing edge, and thought that the main cause of all resistances is to be found
in such separations.
Meanwhile, in 1 9 10, ideas about resistance had been clarified in reports
by Kutta's teacher, Finsterwalder, and by Prandtl himself. Finsterwalder,
in a lecture actually delivered the previous year, had discussed Prandtl's
theory and mentioned the difficulties, but pointed out that from the theory it
could be deduced that i n the case of an airship it is not so much the form of
the nose that matters as the shape of the tail. Prandtl introduced the idea of
the drag being due to two causes, which are not independent of one another,
giving rise to surface resistance (skin-friction) and vortex or form drag.
(Cf. Stanton's reply to Larmor in connection with the publication of Kelvin 's
papers) . Prandtl stated flatly that the form of the after part of a body has in
many cases more importance for drag than the forward part. By suitably
forming the after part the drag of an airship hull could be reduced, so that it
came near to the theoretical value of zero. Prandtl continued that all theories
that try to base drag on what happens in the front must lead to wrong

14
GOLDSTEIN
results, and must be rejected. Moreover, he says, the greater or less turbu
lence of the air is of great importance for the values of the drag.
The concern with the design of airship hulls at that ti m e is important.
Also noteworthy is the reference to turbulence, which will be mentioned later.
Exact, or even rough, calculations, were not possible, and the whole
process was not completely understood, but some time later the theory
began to have considerable influence on design. In fact by the late 1920's and
early 1930's "streamlining" was a fashionable word, even if rather completely
misunderstood by nonprofessionals. Many things, no matter what their
shape in relation to the technical meaning of the term, were referred to as
streamlined-for example, fast motor cars, fast railroad trains, and, accord
ing to advertisers, agreeable young women at any speed if equipped with the
right foundations. The slang use of the term died out, partly perhaps for
sociological reasons, but partly perhaps because of the airship disasters of
1933 to 1937, but the art, science, and technology of streamlining in its tech
nical meaning remained of primary importance, at any rate until science
fiction started to become engineering fact.
The notion of a cast-off vortex, with a vortex being formed and detached
in a short time when a (streamlined) airfoil at lift begins to move, so that a
circulation round the airfoil in an opposite sense is left, gave physical sub
stance to the circulation theory of lift. Many striking experiments have been
made in fluid mechanics and aerodynamics, and this notion of the cast-off
vortex led to one of them. I have never forgotten when William Farren
showed me, on a small scale model in a tank with transparent walls, the
cast-off vortices when an airfoil was started and then stopped impulsively
and the vortex pair with opposite rotations appeared rapidly, and moved
sedately downwards perpendicularly to the line joining them, all in accor
dance with theory. Seeing was certainly believing. I think that a description
of the tank, etc., and the experiment apeared in Walker's thesis ( 1932).
AIRFOIL THEORY
The circulation theory of lift was not widely accepted with any rapidity.
In the early 1920's at least one distinguished aeronautical engineer was still
expressing scepticism, which produced the experiment of Bryant & Williams
( 1925) at the National Physical Laboratory. This not only verified the
existence of a circulation but confirmed the Kutta-Zhukovskii formula for
the lift, even for a real fluid with the presence of a wake, if the contour
around which the circulation is taken does not approach the airfoil too
closely and cuts the trailin g wake at right angles to the direction of the un
disturbed relative motion. The problem was then not to explain the lift, but
to explain why the formula was so nearly correct in the presence of a vortical
wake, and this explanation was i mmediately provided by G. 1. Taylor.
There are three ingredients in inviscid, incompressible, two-dimensional
airfoil theory:. the lift formula, the condition at the trailing edge, and the

FLU I D M ECHANI CS-FI RST HALF OF THIS CENTURY
15
conformal mapping of the airfoil contour. As time went on, the second was
improved by allowances for the boundary layers and the wake. Two-dimen
sional airfoil theory was extended by Mises and by Karman & Trefftz among
others. I t is noteworthy that the theory of the complex variable was increas
ingly used for design (of course with the desired physical, mechanical, and
geometrical criteria in mind) , cul minating in the work of Lighthill ( 1945).
The method of singularities was also developed for application to airfoil
sections.
Prandtl's two classical papers on three-dimensional airfoil theory were
published in 1 9 1 8 and 1 9 19, leading to calculations for airfoils of large but
finite aspect ratio. These papers had been several years in the making, the
ideas dating back in part at least to 1910, with the first published reference in
publications by Foppl in 191 1 . (Foppl also referred to Lanchester's Aero
dynamics, in connection with a pair of trailing vortices which start from the
wing tips and make possible, in simply-connected space, the transition from
flow around the wing.) Prandtl's papers are classical, not only for aerody
namics, but as part of fluid dynamics generally. Moreover, I remember that
when I first read them I formed the strong impression from the way they
were written that Prandtl really knew he was writing classical papers.
M uch research was going on both before and immediately after the
publication of the two papers mentioned above, and anything like a full
description is not possible here. The term "induced drag" appears to be due
to Munk (19 1 8), who also provided what is now known as " M unk's stagger
theorem" ( mentioned also in Prandtl's second paper) and an easier and more
general proof (with generalizations) that wings with elliptic loading have the
smallest possible induced drag. Betz's paper on the screw-propeller with
least energy loss appeared in 1919. Trefftz's method of working in the
"Trefftz plane" and also of using Fourier series appeared in 192 1 .
The first practical triu mph o f the theory was in making sense o f experi
mental results on airfoils of various different aspect ratios.
The news of Prandtl's airfoil theory spread much more rapidly than
the news of his boundary-layer theory. In 1921 the National Advisory
Committee for Aeronautics in the U.S.A. requested and published a report
by Prandtl himself on "Applications of modern hydrodynamics to aero
nautics," and in the same year Pistolesi drew attention to the theory in a
lecture and p ublication in I taly. The same author published a more complete
exposition in the following year. Also in 1922 Roy published a booklet on the
theory in France, and in 1923 the theory was explained, and used, by Glauert
and by Low, and verified experimentally by Fage & Nixon, in England. A
German textbook by Fuchs & Hopf appeared in 1922, and Glauert's text on
The Elements of Aerofoil and Airscrew Theory appeared in England in 1926
and rapidly came into very general use.
An appraisal of the contributions of Lanchester and their influence
would require at least a complete article. Durand , in his " Historical sketch of

16
GOLDSTEIN
the development of aerodynamic theory," spoke of Lanchester's " remarkable

physical insight," and of this there is no doubt. Lanchester read a paper to
the Birmingham Natural History and Philosophical Society in 1 894, sent a
revised version of the paper to a Fellow for publication by the Royal Society
about two years later, and was, rather curiously, advised to send it to the
Physical Society, who rejected it in September, 1897. Apart from taking o u t
a patent, Lanchester made n o further effort a t publication until Aerodynamics
was published in 1907 and Aerodonetics in 1908. His further contributions to
the theory of wings of finite span and airscrews were published in the Pro
ceedings of the Institution of A utomobile Engineers in 1915, and reprinted as a
booklet. A German edition of Aerodynamics was published in 1909, and a
French edition in 19 14, but the publications of 1915 remained largely u n
known in continental Europe for some time, and Prandtl stated that he did
not become aware of them until 1926.
At a very early stage in each case Lanchester had the fundamental ideas
of the circulation theory of lift and of trailing vortices behind an airfoil of
finite span. Some attention was soon paid in Europe to the wing theories in
Aerodynamics. In 1910 Zhukovskii remarked that Lanchester's great merit
was in having illuminated the passage from an airfoil of infinite span, for
which the field occupied by the fluid is doubly-connected, to the finite wing,
for which the field is simply-connected. Foppl's reference in 19 1 1 has al
ready been mentioned. However, no attention was paid until much later to
Lanchester's theories in his native England. H is works contained but little
mathematical development, even though he apologized in the preface to
Aerodynamics to the non mathematical reader, "who may find himself out of
his depth," for the mathematics that was included. Later, in his Wilbur
Wright lecture, he stated that his writings were in plain English, divested of
all mathematical ornament, but in fact, plain English or not, they were by
no means easy to understand. Aerodynamics should still be read, not only for
i ts content but to savor Lanchester's style. Many words he used were his own
coinage, and others were used with what is now an obsolete meaning, so a
glossary is needed , and is in fact provided in the book. For example, the
title of the 1894 paper to the Birmingham Natural History and Philosophical
Society was "Stability of an aerodrome. " Aerodrome was apparently
Langley's word; it was used for a flying machine. The Oxford Universal
Dictionary says that the word with the meaning of an aeroplane dates from
1 89 1 and became obsolete in 1896, which seems rather hard on Lanchester,
who used the word in that sense very firmly in Aerodynamics in 1907 . In fact,
in a footnote he wrote that " the word aerodrome has been grossly misapplied
by Continental writers to denote a balloon shed." The Random House College
Dictionary says that the word (airdrome in the United States) now means an
airport; so the change of meaning in what was called the "stability of an
aerodrome" must be noted . The contents of the early paper, and several
diagrams from it, were reproduced in Aerodynamics.
17

FLUID MECHANICS-F I RST HALF OF T H I S CENTURY
Of course, eventually attention was paid to Lanchester's contributions

in England, and he was invited by the Royal Aeronautical Society to deliver
the Wilbur Wright Memorial Lecture in 1926, a year before Prandtl.
Prand tl's lecture, the beginning of which he read hi msel f with the reading
completed by Major Low, since Prandtl had trouble with the English
language, was on "The generation of vortices in fluids of small viscosity"
by the action of the boundary layer, not on airfoil theory, because, he said,
he thought "it would be preferable to select another subject with which you
are possibly less familiar in England, although its beginnings go further back
than do those of the aerofoil theory. " However, after referring to Froude and
stating that " Lanchester developed an approximate theory for steady
laminar flow" in connection with boundary layers, Prandtl spoke briefly on
the history of airfoil theory. He said :
In England you refer to it as "the Lanchester-Prandtl theory," and quite rightly
so, because Lanchester obtained independently an important part of the results.
He commenced working on the subject before I did, and this no doubt led people to
believe that Lanchester's investigations, as set out in 1907 in his 'Aerodynam ics,'
led me to the ideas upon which the airofoil theory was based. But this was not the
case. The necessary ideas upon which to build up that theory, so far as these ideas
are comprised in Lanchester's book, had already occurred to me before I saw the
book. In support of that statement, I should like to point ou t as a matter of fact we
in Germany were better able to understand Lanchester's book when it appeared
than you in England. English scientific men, indeed, have been reproached for the
fact that they paid no attention to the theories expounded by their own country
man, whereas the Germans studied them closely and derived considerable benefit
therefrom . The truth of the matter, however, is that Lanchester's treatment is
difficult to follow, since it makes a very great demand on the reader's intuitive per
ceptions, and only because we had been working on sim ilar lines were we able to
grasp Lanchester's meaning at once. At the same time, however, I wish it to be
distinctly understood that in many part ic u lar respects Lanchester worked on dif
ferent lines than we did, lines which were new to us, and that we were therefore
able to draw many useful ideas from his book. The volume published in 1 9 1 5 . . in
which Lanchester comes to the same conclusion with regard to the induced drag
as we did, was unknown to us until 1926. As it happens the same formula was pub
lished by us in 1914 (by Betz) . . . .
.
I t remains to add that the vortex-sheet model of Lanchester and Prand tl

formed also the starting point for calculations on "lifting-surface" theory, to
which much attention was paid in the 1930's and 1940's and on which work is
still proceeding. I n 1936 Prandtl introduced the acceleration potential and
reduced the downwash integral to a form in which it is taken over the wing
surface only. Special planforms, in particular elliptic and circular planforms,
were considered [by Kinner ( 1937) , by Kochin (1940) , and by Krienes
( 1940) ] . and perhaps more importantly, numerical methods came more and
more into use.
Prandtl's own presentation was not completely free from i n tuitional

18
GOLDSTEIN
"insights, and it has lately been pointed out [e.g. in Incompressible A ero
dynamics, edited by Thwaites (1960) and in Ktichemann's " Ludwig Prandtl
M emorial Lecture" ( 1967)] that what Prandtl himself called a lifting-line
theory must be considered an approximate lifting-surface theory in which
the chordwise distances involved are small compared with the spanwise
distances. In such a case it can be proved that the chord wise loading is the
same as in the two-dimensional case and the spanwise distribution of lift is
given by Prandtl's lifting-line theory.
The time came when the influence of the compressibility of air at high
subsonic speeds could not be ignored, and also theoretical considerations of
supersonic flight were taken more and more seriously. Swept wings had to be
considered. I t appears that the first published suggestion of the use of swept
wings was made by Betz in 1940. Slender aircraft became matters of serious
consideration only considerably later, but a theory of pointed wings of very
small aspect ratio--in contradistinction to the theory for large aspect ratios
was published by R. T. Jones in 1946. In all cases trailing vortices were still
with us.
STABILITY
Among other matters exciting the curiosity and attention of investigators
probably the most important were stability, turbulence, and gas dynamics.
Theoretical investigations of stability in the period dealt mainly with
stability to infinitesimal disturbances on a linearized theory. Progress in
considering finite disturbances was not to come until later.
In 1 9 1 6 Rayleigh considered , as a question of stability, the convection
currents in a horizontal layer of fluid heated below, in connection with
observations of Benard in 1900 and 190 1 , mentioned in the 1900 to 1910 list
at the beginning of this article. Rayleigh's discussion is based on the ap
proximate equations of Boussinesq. He remarked that M . Benard did not
appear to be acquainted with a paper by James Thomson in the Proceedings
of the Glasgow Philosophical Society for 1881-1882, where a like structure was
described in much thicker layers of soapy water cooling from the surface. I n
t h e Scientific Papers there appears a note added in 1 9 1 8 about his own work,
that "This problem had already been treated by Aichi (Proc. Tokio Math.
Phys. Soc. 1907) . " The problem was later considered by Jeffreys, Low,
Pellew & Southwell, and others. Harold Jeffreys used to call it the " porridge
problem", for obvious reasons. In addition to the consideration of finite
disturbances, other factors were to be added, such as the influence of surface
tension and its variation with temperature, but . there was nothing about
Rayleigh's stability problem that should be called controversy, and the
investigations could be counted as definitely successful.
In 1922 G. I. Taylor published his classical paper on the "Stability of a
viscous liquid contained between two rotating cylinders." The results of the
theory were definite and correct, in agreement with the experimental re-

19
suits. Simplifications of the mathematics, sufficient for the required pur

poses, and many further considerations of the flow between rotating cylinders,
were to follow, but there could be no controversy.
Jeffreys demonstrated in 1928 the mathematical equivalence of the two
stability problems of convection and flow between rotating cylinders, an
equivalence which had been suggested by Taylor and Low.
The kind of instability to three-dimensional disturbances for flow over
concave curved surfaces, exemplified by the work of Taylor in 1922 and later
work on the transition to turbulence for flow between rotating cylinders, -was
once a cause for some concern in airfoil design. Once upon a time, in connec
tion with the design of low-drag suction wings for which the boundary layer
should stay laminar as long as possible, at the National Physical Laboratory
we designed an airfoil section with a single slot on each surface, and with
favorable pressure gradients everywhere except at the slot. The work was
done by Richards. A model was made, and tested in a wind tunnel. In spite
of the favorable velocity gradient behind the slot, the flow in the boundary
layer became turbulent soon after the slot except at low Reynolds numbers.
The airfoil surface was concave to the flow aft of the slot, and the transition
to turbulence was undoubtedly connected with the three-dimensional in
stability in flow over such a surface, and its further development. For
boundary-layer flow this instability was studied theoretically by Gortler in
1940, but because of the war we did not know of Gortler's paper, although in
fact recollection of Taylor's work on the flow between rotating cylinders
should have been enough. Transition to turbulence in the boundary layer
on a concave surface was shortly afterwards studied experimentally by
Liepmann (1943) , of whose results we were informed. As we said at the
time, all the laws of nature always continue to work, including those we
forget.
As regards the definiteness of theoretical discussions of problems of
stability, the state of affairs was different when it involved discussion of the
Orr-Sommerfeld equation for parallel flows. M uch controversy developed
after Heisenberg's discussion in 1924 of the stability of plane PoiseuilIe flow,
which did not die down until C. C. Lin clarified the general theory and gave a
detailed calculation of the neu tral curve.
Doubt and controversy had appeared before 1924. In 1914 Rayleigh,
discussing work by Mises and Hopf on the apparently simpler problem of
plane Couette flow, had remarked that " Doubtless the reasoning employed
was sufficient for the writers themselves, but the statements of it put forward
hardly carry conviction to the mere reader. The problem is indeed one of no
ordinary difficulty."
There were even more causes for doubt and controversy over the calcula
tions by Tollmein of the stability of the flow in a boundary layer on a flat
plate without a pressure gradient, following previous discussions by Prandtl
(1921) and Tietjens [1922 (thesis) and 1925), and themselves foHowed by
20
GOLDSTEI N

further calculations b y Schlichting ( 1933 and 1935) . Flow i n a boundary

layer is not really a parallel shearing flow as assumed in the calculations.
M oreover, for some time experimental observations failed to find the neutral
or amplified oscillations predicted by the theory, and it appeared that the
development of turbulence in a boundary layer depended on the amount of
turbulence in the main stream outside the layer. Then in 1940, in experiments
by Schubauer & Skramstad at the National Bureau of Standards, under the
direction of H. L. Dryden, the theory was verified experimentally. In the
introduction to the English edition of Schlichting's Boundary Layer Theory,
Dryden wrote :
My own interest in the experimental aspects of boundary-layer flow began in the
late twenties. With the appearance of Schlichting's papers intensive attempts were
made to find the amplified disturbances predicted by the theory. For 10 years the
experimental resu lts not only failed to confirm this theory but supported the idea
that transition resulted from the presence of turbulence in the free air stream as
described in a t heory set forth by G. I. Taylor. Then on a well-remembered day in
August, 1940, the predicted waves were seen in the flow near a flat plate in a wind
tunnel of very low turbulence. The theory of stability described in the papers of
Tollmien and Schlichting was soon confirmed quantitatively as well as qualitatively.
In the theories two-dimensional disturbances are considered. In fact, i n

1933 Squire proved that t h e problem of three-dimensional disturbances of a
plane parallel flow is equivalent to a problem with two-dimensional distur
bances at a lower Reynolds number, so the minimu m critical Reynolds
nu mber is given by a two-dimensional analysis.
After the experimental verification of the calculations of Tollmien and
Schlichting, difficulties remained. Instability in a laminar flow, even with
amplified disturbances, is not the same as transition to a fully turbulent flow ;
the gap in understanding was still rather large, and attempts have only
recently been made to further understanding by considerations of finite
disturbances, stability of the disturbed flow, etc. Further, I doubt if it is yet
possible to state with certainty how much the appearance of turbulence in a
boundary layer, especially over a curved surface with a pressure gradient, as
in flow past a sphere, for example, is due to instability in the boundary-layer
flow and how m uch to the effect of impressed disturbances from turbulence
in the main stream, or how much they interact.
TURBULENCE
As previously stated, turbulence in boundary layers was mentioned at a

fairly early stage by Lanchester and by Prandtl. One of the early triumphs of
the ideas of boundary-layer theory was the explanation of the rapid drop of
the drag coefficient, for instance of a sphere, with increasing Reynolds
nu mbers in the neighborhood of a certain critical nu mber, which depends on
everything that can influence the transition to turbulence i n the boundary
layer, such as the degree of turbulence in the main stream, roughness of the

FLUID MECHAN I CS-FI RST HALF OF THIS CENTURY
21
surface, the method of support, and protuberances, hooks, or other attach- ments on the surface. In a turbulent boundary layer, because of the more
vigorous interchange of momentum between different strata, the retarded
fluid can make its way into regions of higher pressure before separation, so
that although the friction drag is increased the form drag is considerably
decreased. The phenomenon itself was first demonstrated by Eiffel in Paris
in 1912 for a sphere, and the explanation was given by Prandtl in 1 9 14, who
obtained small drag coefficients even at fairly low Reynolds numbers by
inducing turbulence with a wire hoop fixed on a sphere. M any years later
stories were still being told in Gottingen about great disagreements in the
results experimentally obtained there with those of Eiffel, and how the
research on the cause began.
The search for a satisfactory method of forecasting frictional resistance
with turbulent flows began early. A complete mathematical theory is not to
be expected, and experiments did not deal with high Reynolds numbers.
Coupled with formulae for the resistance coefficient is a formula for the
distribution of velocity near a wall. In 1913 Blasius put forward interpolation
formulae according to which the resistance coefficient varied as the inverse
of the 1 /4th power of the Reynolds number, and the velocity as the 1/7th
power of the distance from the wall. Attempts were made to show that these
formulae had a theoretical basis, but when experiments were made at
Reynolds numbers above 105, it was found that the index 1/7 had to be
diminished progressively to 1/8, 1/9, etc.
In 1925 Prandtl p u t forward what became known as his " mixture- length "
theory, and also assumed in that connection that momentum is a transferable

property. For forecasts near a wall, this was a valuable advance.
Meanwhile search for satisfactory extrapolation formulae for resistance
and velocity near a wall continued vigorously, and, at the Third I nter
national Congress of Applied M echanics at Stockholm, Karman was the first
to announce the famous logarithmic formulae for wall turbulence. He ob
tained the result from his "similarity theory." Prandtl obtained the results
more simply in a paper published shortly afterwards ( 1933) , and it later
appeared that other "rational" arguments would produce the same result.
However, the announcement in 1 930 in Stockholm by Karman was the first
announcement of this famous " law" of wall turbulence.
The late 1920's and early 1930's was a marvelously exciting time for
"semi-empirical" theories of turbulence.
I n 1932 Taylor published a paper on "The transport of vorticity and heat
through fluids in turbulent motion," in which he took vorticity, not mo
mentum, as the transferable property, pointing out that the assumption that
momentum is a transferable property involves the assumption that the
fluctuations in pressure in the turbulent flow do not affect the mean transport
of momentum . Taylor's vorticity-transfer theory in fact dated back to his
paper on "Eddy motion in the atmosphere" in 1 9 1 5 and to his essay for the

22
GOLDSTEIN
Adams Prize awarded the same year. The equations for the velocity distribu
tion are, in general, different on the momentum-transfer and the vorticity
transfer theories, but, more i mportantly, when there is heat transport,
predicted temperature distributions differ. In an appendix to Taylor's 1932
paper, Fage & Falkner reported on the measured temperature distribution in
the wake of a heated cylinder, and the results on the vorticity-transfer theory
were much closer to the experimental results than those of the momentum
transfer theory. Thereafter many experiments were made in an effort to
assess the relative merits of the two theories.
Taylor published his " modified" vorticity-transfer theory in 1935"and
1937 and the application to flow in pipes i n 1937. I cannot be sure of my
recollection, but I believe he unearthed an old manuscript from among
his papers (again part of his Adams Prize essay) when I showed him a
(somewhat unsatisfactory) paper of mine on the generalized vorticity-trans
fer theory, saying that he had previously thought it too speculative to publish.
I am more certain about a paper of Taylor's on another subject, stability i n
stratified fluids with density gradients, which was published in 193 1 . A
manuscript was taken out of storage on my entreaty ; it was part of Taylor's
Adams Prize essay and had been in storage for some considerable time, its
submission for publication having been hindered by the occurrence of the
first world war.
Meanwhile research was proceeding on turbulent diffusion, which had
its genesis in Taylor's 1921 paper in the Proceedings of the London Mathe
matical Society on " Diffusion by continuous movements."
However, a wholly new direction was given to research by the publica
tion in 1935 of Taylor's "Statistical theory of turbulence" in a series of papers
of striking originality, containing many new notions, among them isotropic
turbulence, curves of correlation and energy dissipation, the decay of
turbulence behind a grid, and correlations and energy spectra as Fourier
transforms ( 1938) . Karman introduced the correlation tensor, depending,
in an incompressible fluid, on one scalar function; and derived an equation
for changes in that scalar, which can be used to obtain information about
the rate of decay, with the assumption that the mean values of triple products
of components of velocities at two points could be neglected. Karman pointed
o u t that if this is incorrect the vortex filaments would have a permanent
tendency to be stretched or compressed along the axis of the vorticity, and
believed that this could not be the case. H owever, Taylor showed that it was,
the term neglected being shown from certain measurements in one case to be
three times a term that is not neglected. There is a tendency for the vortex
filaments to be stretched on the average. Turbulence is essentially dispersive.
Later, during the second world war, I remarked to Taylor that an army
general had j ust explained that the reason why information and supplies did
not always arrive at their destination at the planned time was "the friction
of war." " I suppose," replied Taylor, "that supplies are strewn over the
countryside because of the turbulence of war . "

FLUID MECHAN I CS-FI RST HALF OF TH I S CENTURY
23
Karman & Howarth showed in 1938 that the triple-correlation tensor

also involves one scalar function for isotropic turbulence in an incompressible
fluid, and carried the discussion further with the hypothesis that the graphs
of the correlation functions preserve their shape.
Work continued vigorously. Contributions came in from Loitsyanskii,
Millionshchikov, and Kolmogorov, whose contribution proved particularly
valuable. The name of the distinguished mathematician Kolmogorov was well
known to other distinguished mathematical statisticians, some of whom had
hopes of contributing to the theory of turbulence. When they saw the
physical, rather than mathematical, nature of Kolmogorov's contribution
most of them decided that such research was not for them. Other work,
particularly on spectral analysis, also appeared. Meanwhile, experimental
and semi-empirical work on shear turbulence was also proceeding.
On the whole the experimental and semi-empirical work flourished.
After some time, however, the more fundamental theoretical work served
more, after protracted examination, to exhibit clearly the difficulties than to
solve them. This, too, was a valuable contribution.
It was at a meeting of the British Association in London in 1932 that I
remember that Lamb remarked " I am an old man now, and when I die and
go to H eaven there are two matters on which I hope for enlightenment.
One is quantum electrodynamics, and the other is the turbulent motion of
fluids. And about the former I am really rather optimisitic." ( I have quoted
from memory, so do not guarantee all the actual words. B u t the sense is
correct. I have heard a similar story since repeated with other names than
Lamb and other times and places.) Lamb was correct on two scores. All who
knew him agreed that it was Heaven that he would go to, and he was right to
be more optimisitic about quantum electrodynamics than turbulence.
SHOCK WAVES, GAS DYNAMICS, HIGH-SPEED AERODYNAMICS
Let me return to shock waves and gas dynamics. Rankine and H ugoniot
had published their papers on shock waves in the nineteenth century, and
Ernst M ach and his associates began their experimental observations about
the same time as H ugoniot published ( 1889) . However, there was still some
doubt and confusion about the theoretical explanation. The discontinuity
seemed so very sharp. Was the process to be associated with the theory of an
ideal i nviscid gas, or were dissipative processes essential ? After the publica
tions of Rayleigh and Taylor in 19 10, referred to in the opening 1900 to 1910
list, the theoretical ideas rested on a firm foundation and the way was open
for further advances, both for more and more complicated problems on
shock-wave interactions and on the reflection, diffraction, and refraction of
shock waves, as i n the work of Polachek & Seeger for example, and on the
structure of shock waves, with allowance for bulk viscosity and variations
with temperature of the viscosity coefficients (though the correct variation
of the bulk viscosity is still uncertain) . The investigations of the structure of
shock waves required nu merical methods. I nvestigations of shock-wave

24
GOLDSTEI N
structure by the kinetic theory o f gases also followed later. I n the first half of
the century the work culminated in J. von Neumann's numerical methods.
The early state of affairs may be illustrated by some quotations from the
papers of Rayleigh and Taylor. Rayleigh writes, under the heading "Per
manent regime u nder the influence of dissipative forces, " "The first investi
gation to be considered under this head is a very remarkable one by Rankine
'On the Thermodynamic Theory of Waves of Finite Longitudinal Distur
bance' (Phil. Trans. 1 870) , which (except for a limited part expounded by
Maxwell in his Theory of Heat) has been much neglected. " There is here a
footnote in which Rayleigh says "I must take my share of the blame," but
adds that " Rankine is referred to by Lamb (Hydrodynamics, 1906, p. 466) . "
I n the text h e continues "Conduction of heat i s here for the first time taken
into account and although there are one or two serious deficiencies, not to
say errors, presently to be noticed , the memoir marks a very definite ad
vance." Later he refers to "a long and ably written memoir by Hugoniot"
and after repeating that " Rankine's investigation is expressly based upon
conduction of heat in the gas" remarks that "A wave of this kind is never
possible under the conditions, laid down by Hugoniot, of no viscosity or heat
conduction . . . . A closer examination of the process by which [Equation 85]
was obtained will show that while the first law of thermodynamics has been
observed, the second law has been disregarded . " The remarkable thing was
that both Rankine and Hugoniot ended up with the same equations.
Neither had considered viscosity.
Taylor pointed out that "The possibility of the propagation of a surface
of discontinuity in a gas was first considered by Stokes" ( 1 848) . He later
states flatly that " I t is evident that a plane of absolutely sharp or mathe
matical discontinuity cannot occur in any real gas. " He refers to the kinetic
1 heory
of gases, and continues " This suggests that heat conduction and
viscosity are, in the case of a real gas, the causes of the prod uction of dis
sipative heat; it will be shown that under certain conditions they are also
sufficient to produce permanence of type in the layer of transition. "
The study o f gas dynamics and high-speed aerodynamics proceeded
steadily and at a somewhat increasing pace during, say, the first third of the
century. Some of the most valuable advances were in the provision and de
sign of high-speed wind tunnels and instrumentation. Towards the end of the
1920's the only lectures on gas dynamics I remember being given in Cam
bridge, England were a series of eight lectures by G. I. Taylor.
In 1935 came the Fifth Volta Congress, which was indeed an important
event. The Schneider Trophy context had had its effect on high-speed
aerodynamics, and one of the papers at the Congress was by Wimperis o n
"The British technical preparation for t h e Scl.meider Trophy Contest,
193 1 , " in which he described, as he said, "the combination which achieved
success in the 193 1 Sch neider Trophy Contest." However, in addition,
scientists and engineers were already dreaming of supersonic flight. Among the

FLUI D MECHANICS-F I RST HALF OF THIS CENTURY
25
papers a t the Congress those that are o f interest here were general lectures
on high-speed flow by Prandtl, Taylor, Karm{m, Busemann, and Pistolesi,
on high-speed wind tunnels and experimental technique by Eastman
Jacobs, Ackeret, Luigi Crocco, and Panetti, on model airscrews at high
speeds by Douglas, and research in France by Villat.
In the last decade of the half-century, particularly after the end of the
second world war, the pace of publication and advance in knowledge be
came rapid, almost hectic. Reports delivered in one place of what had j ust
been done at another were greeted by remarks that "Yes, we've j ust done
that, too." It was all exciting, and great fun, and of serious use, too,-and
the war was over.
It seems impossible even to list here the highlights of what was done i n
the half-century, apart from the provision o f high-speed wind tunnels and
equipment, b u t let us bring a few to mind : Prandtl-M eyer expansions;
graphical and numerical use of the mathematical theory of characteristics;
linearization for subsonic and for supersonic flow, and higher approximations ;
the Prandtl-Glauert rule; the linearized solution of Karman and Moore for
bodies of revolution ; the clearing-up of the difficulties about the Prandtl
Glauert rule for bodies of revolution, etc. , by Gathert and Sears, and the
appearance of Gathert's rule ; methods of successive approximation , first
Jansen- Rayleigh and later Hantzsche-Wendt; Taylor's electrical analogy,
with the electric field explored in an electrolyte in a shallow tank of variable
depth which could be changed by successive approximations; boundary
layers in gases at high speeds and frictional resistance ; Taylor's calculations
for a vortex and a source, forecasting the troubles of transsonic theory ; the
Taylor- Maccoll nonlinearized solution for supersonic flow past a cone; coni
cal fields, whose study was started by Busemann ; the slender-body theory of
Jones (and earlier of M unk for low-speed flow) and its development by Ward
and others (and l ater by M ac C. Adams and W. R. Sears) ; the development
of Chaplygin's hodograph method and the alteration in his use of an ap
proximate equation of state, leading to the famous Karmfm-Tsien pressure
formula, which was so extensively used in airfoil design for subsonic flight
both during and soon after the second world war, with the theory later ex
tended by C. C . Lin ; the full analytical, mathematical development of
hodograph theory by Lighthill and by Cherry (and also, differently, by
Bergman) in 1947 ; Karman's transsonic similarity rules ; the similarity rules
for hypersonic flow of Tsien and H ayes ; and the work, done independently by
Sedov, Taylor, and von Neumann, on blast waves.
I n his lecture at the Volta Congress, Karman had remarked " I have the
i mpression that the possibilities provided by the hodograph method are
not yet exploited sufficiently and that it can possibly be used for investiga
tion of the mixed cases, i.e., of flows with partly subsonic, partly supersonic
regions. " I n that lecture he also considered flow at very high M ach numbers,
at what he called " ultra supersonic" speeds, and is now called the hyper-

26
GOLDSTEI N
sonic range, and its connection with Newton's conception o f air resistance.
I n 1933 B usemann had studied flow in a shock layer of vanishing thickness
and obtained a formula for the surface pressures at the base of the shock layer.
I n 1949 Lighthill published papers on the diffraction of blast, and on a
technique for rendering approximate solutions to physical problems uni
formly valid. The technique in the latter paper has been rather widely used,
and there has been considerable discussion of its connection with the theory
of si ngul ar perturbations, boundary-layer theory, the theory of "inner" and
"outer" solutions.
Whitham's paper in the Proceedings of the Royal Society, published i n
1950, on "The behaviour of supersonic flow past a body of revolution, far
from the axis" was first turned down by a referee with a very i nteresting and
rather full discussion of his objection, which, in summary, was the danger of
using isentropic methods to study the decay of a shock wave. This had
previously been done by Friedrichs in the plane case, so, as the referee
pointed out, "Whitham is in good company." The argument of the referee
was serious, and this was the genesis of Lighthill's 1950 paper on "The
energy distribution behind decaying shocks." Whitham's paper was duly
accepted by the Royal Society and published.
Lighthill's first papers were published in 1944, and by 1948 he was in
vited to deliver a general lecture at the Seventh International Congress for
Applied M echanics. The lecture was on " Methods for predicting phenomena
in the high-speed flow of gases. " Lighthill spoke then as an applied mathema
tician or mathematical physicist, and his concluding words are interesting :
In conclusion, I w il l say that our u nderstanding of the high-speed flow of gases
is growing rapidly; as a mathematician I believe that our endeavors in this field are
no waste of effort or of mathematical techniques not only because we are assisting
in a grea t new engineering adventure--supersonic fl ight-but also since we are
getting to grips in this problem with that old bogeyman, the nonlinear partial dif
ferential equation, and smelling out his ways in a manner for which our colleagues,
in the more fundamental parts of physics, may later be grateful.
EXPERIMENTAL FACILITIES AND TECHNIQUES, NUMERICAL METHODS
M any i mprovements in experimental facilities and techniques took place

in the first half-century. There were wind tunnels for special purposes, high
speed tunnels and low-turbulence tunnels (with which the i nvestigations of
Prandtl and of Taylor on the effect of a contraction on turbulence in a wind
tunnel were connected) . As time went on, we were on our way to bigger and
bigger, and faster and faster tunnels. In instruments, electrical methods,
especially for measurements in turbulent flow, came into increasing use :-the
hot-wire anemometer, the hot-wire direction-meter, methods of measuring
speed variations and correlations in turbulent flow, and of determining
energy spectra. The methods of visualizing and photographing fluid motions
produced fascinating results, whether the method was smoke from the

27
rotted wood of an apple tree or a cheap cigarette or titaniu m or stannic

tetrachloride, or the results from the chemical coating of a surface, or hot
wire or spark shadows, or Schlieren or interferometer pictures, or the
ultramicroscope photographs in water of Fage & Townend.
N u merical methods have previously been mentioned in connection with
stability i nvestigations, lifting-surface theory, and the structure of shock
waves. I n the 1930's SouthwelI's relaxation method, and other somewhat
similar methods, were being used. Digital machines came gradualIy more and
more into use. Later, largely under the inspiration of von Neumann and of
Pekeris in the Weizmann I nstitute in Israel, serious use was to be made of
high-speed computing machinery, but that was really a later story, and even
now there is stilI a long way to go.
Prandtl once told me that he had considered building an analogue ma
chine, but came to the conclusion that water itself was the best. It is reported
that Neumann once said that there would be no further need for experiment
-high-speed computation could take over. I suppose both were incorrect ;
we need both experiment and computation.
CONCLUSION
M uch has been omitted. I have steadfastly resisted the temptation to

write about the arguments concerning education, and the relative importance
of mathematics, physics, and engineering. Among other matters omitted are :
(a) the beginnings of magnetohydrodynamics, Hartmann & Lazarus ( 1937)
on the flow of mercury in a channel in the presence of a magnetic field ,
Hoffmann & Teller on magnetohydrodynamic shocks, the connection with
astrophysics and the publication of AHven's Cosmical Electrodynamics j ust
after the close of the first half-century; (b) whole chunks such as mete
orological investigations, oceanography, surface waves and tides, tidal
dissipation and its astronomical application (Taylor, 1 9 1 8 and 1920, and
Jeffreys 1920) , and shallow-water theory, and geophysics ; also the stability
of rotating masses of fluid and astronomical applications; (c) interesting and
important detailed questions, which some of us taught to students with
considerable enthusiasm, such as the method of singularities for discussion
of the aerodynamic properties and design of airship shapes and the use by
IGrman of doublets to allow for crosswinds ( 1927) , and trailing vortices for
two-dimensional airfoils in time-variable motion (Birnbaum, 1924 ; Wagner,
1925 ; Karman & Sears, 1938 ; Sears, 1940 ; and others) ; unsteady airfoil
motion in high subsonic and in supersonic flow; also the general calculation
of virtual inertia, forces and moments in irrotational inviscid motion; and
Carrier's modification of Oseen's method ; (d) the connection of fluid dy
namics and heat transfer ; (e) rarefied gases and relaxation effects-I still
remember vividly Arthur Kantrowitz's first demonstration with CO2 at
Langley Fiel d ; (f) such things as turbulence in round pipes, and the repeti
tions of Reynolds's experiments by B arnes & Coker, Ekman, and Taylor,
28
GOLDSTEI N

and also discussions o f the possible wake behind a sphere corresponding to

a vortex-street behind a cylinder, with some of us discussing this far into the
night in every night-club in Aachen at the time of the Aachen Conference in
1929 and finally "solving" the problem with alcoholically induced euphoria;
(g) j ets and cavities ; and (h) much else.
Some quotations have been included in this article. Perhaps there are
read ers who think the article would have been improved if there were more
quotations and less other matter. I am reminded of a brief review in Nature
of a book on education. The whole review was: " Half of this book is quota
tions; the other half should have been."

Annual Review of Fluid Dynamics Vol 1, Goldstein S.

Uploaded by

Copyright:

Available Formats

Annual Review of Fluid Dynamics Vol 1, Goldstein S.

Uploaded by

Document Information

Original Description:

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

Annual Review of Fluid Dynamics Vol 1, Goldstein S.

Uploaded by

Copyright:

Available Formats

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.

Quick links to online content

FLUID MECHANICS IN THE FIRST HALF

Harvard University, Cambridge, Massachusetts

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

THE FIRST DECADE

The opening decade of the twentieth century was a period of progress in

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

The third edition of Lamb's Hydrodynamics, of which the first edition

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

The progressive development of his subject is often an embarrassment to the

Rayleigh's review is a remarkab ly interesting review and short essay. I n

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

surface of separation are unstable. This is no doubt the case

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

development of Levi-Civita's method by Cisotti and Villat, and later by

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

I n 1 9 1 2 , in the book on the theoretical bases of aeronautics in which his

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

(Stokes in an examination q uestion, Smith's Prize Examination, 1 876;

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

oretical results in hydrodynamics, and more particularly to verifications of those.

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

in experiments with water.

FLUID MECHAN ICS-FIRST HALF OF THI S CENTURY

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

di mensional flow ; to mention the difficulty for numerical computation arising

Prandtl explained the plausible reason for separation with an unfavorable

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

it might be expected to look for it. In 1908 Blasius published i n a more ac

which the motion is taken to be irrotational, being practically unaffected by viscos

of motion are sought, appropriate to

There are references to Prandtl's lecture and to Blasius.

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLU I D M ECHANI CS-FI RST HALF OF THIS CENTURY

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

the development of aerodynamic theory," spoke of Lanchester's " remarkable

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHANICS-F I RST HALF OF T H I S CENTURY

Of course, eventually attention was paid to Lanchester's contributions

I t remains to add that the vortex-sheet model of Lanchester and Prand tl

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHANICS-FIRST HALF OF THIS CENTURY

suits. Simplifications of the mathematics, sufficient for the required pur

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

further calculations b y Schlichting ( 1933 and 1935) . Flow i n a boundary

In the theories two-dimensional disturbances are considered. In fact, i n

As previously stated, turbulence in boundary layers was mentioned at a

Annu. Rev. Fluid Mech. 1969.1:1-29. Downloaded from www.annualreviews.org

FLUID MECHAN I CS-FI RST HALF OF THIS CENTURY