Escolar Documentos
Profissional Documentos
Cultura Documentos
BOOK ONLY
C0>
<
-z 60351 >m
Accession No.
last
marked below
MECHANICS
MACMILLAN &
MECHANICS
BY
PH.D., LL.D.
NEW YORK
COPYRIGHT, 1937,
Reprints! Nov.
1946.
ST UP AND ELECTROTYPED BY J.
PREFACE
a natural science, and like any natural science comprehension the observation and knowledge of a vast fund of individual cases. Arid so the solution of problems is
Mechanics
is
all
But Mechanics is not an empirical subject in the sense in which physics and chemistry, when dealing with the border region of tUe human knowledge of the day are empirical. The latter take cognizance of a great number of isolated facts, which it is not as
yet possible to arrange under a few laws, or postulates. The laws of Mechanics, like the laws of Geometry, so far as first approximations go the laws that explain the motion of the golf ball or the gyroscope or the skidding automobile, and which make possible the calculation of lunar tables and the prediction of eclipses
and will bo as new arid important two thousand years hence, as in the recent past of science when first they emerged into the light of day. Here, then, is the problem of training the student in Mechanics to provide him with a vast fund of case material and to develop in him the habits of thought which refer a new problem back to the few fundamental laws of the subject. The physicist is keenly alive to the first requirement and tries to meet it both by simple
these laws are known,
laboratory experiments and by problems in the part of a general course on physics which is especially devoted to "Mechanics." The interest of the mathematician too often begins with virtual
velocities
and d'Alembert's
Principle,
which Hamilton's Principle is they are dping nothing that is wrong but each takes such a fragmentary view of the whole subject, that his
of
right, in the sense that
;
work
is
ineffectual.
The world
in
girl
have lived
is
the true
laboratory of elementary mechanics. The tennis ball, the golf the automobile ball, the shell on the river good old Model T,
in its day,
vi
PREFACE
the amateur youngsters construct and will continue to construct in which mechanics of the the the body is a games printing press rich with to the student all these laboratory things go provide part
;
experience before he begins a systematic study of mechanics. It is this experience on which the teacher of Mechanics can draw, and
fifty years and more ago has been and the criticism was not without foundation. It was a method which turned out problem solvers so said its opponents. But it turned out a Clerk Maxwell and it
no uncertain terms the debt he owes to just this training, and to Arthur Gordon Webster, through whom he first came to know this method a method which Benjamin Osgood Peirce also in his work as a physicist. And so we make no prized highly for fullest extent of that which ourselves to the apologies availing the old Tripos Papers contributed to training in Mechanics. But we do not stop there. After all, it is the laws of Mechanics, their comprehension, their passing over into the flesh and blood of our scientific thought, and the mathematical technique and theory, that is our ultimate goal. To attain to this goal the mathematical theory, absurdly simple as it is at the start, must be systematically In this respect inculcated into the student from the beginning. the physicists fail us. Because the mathematics is simple, they do not think it important to insist on it. Any way to get an answer is good enough for them. But a day of reckoning comes. The physicist of to-day is in desperate need of mathematics, and at best all he can do is to grope, trying one mathematical expedient after another and holding to no one of these long enough to test it mathematically. Nor is he to be blamed. It is the old (and most useful) method of trial and error he is employing, and must
continue to employ for the present. Is the writer on Mechanics, per contra, to accept the challenge of preparing the physicist to solve these problems? That is too
large a task.
Rather,
it
is
the
wisdom
of Pasteur
who
said
"Fortune favors the prepared mind" that may well be a guide for us now and in the future. What can be done, and what we havo attempted in the present work, is to unite a broad and deep knowv
PREFACE
vii
edge of the most elementary physical phenomena in the field of Mechanics with the best mathematical methods of the present day, treating with completeness, clarity, and rigor the beginnings in scope not restricted, in detail not involved, in of the subject
;
spirit scientific.
The book
is
first
course in Mechanics,
given for sophomores, and culminating in a thorough study of the dynamics of a rigid body in two dimensions. This course may be
by a half-course or a full course which begins with the kinematics and kinetics of a rigid body in three dimensions and proceeds to Lagrange's Equations and the variational principles.
followed
So important are Hamilton's Equations and their solution by means of Jacobi's Equation, that this subject has also been inIt appears that there is a special need for treating this cluded.
theory, for although it is exceedingly simple, the current textbooks are unsatisfactory. They assume an undefined knowledge of the theory of partial differential equations of the first order, but they do not show how the theory is applied. As a matter of fact, no theory of these equations at all is required for understanding the solution just mentioned. What is needed is the fact that Hamilton's Equations are invariant of a contact transformation. A simple proof is given in Chapter XIV, in which the method
most important
the variables,
is
for the physicist, namely, the method of separating set forth with no involved preliminaries. But
even
this proof
may be
may
Chapter XV.
The concept of the vector is essential throughout Mechanics, but intricate vector analysis is wholly unnecessary. A certain minute amount of the latter is however helpful, and has been set
forth in
Appendix A. Appendix D contains a definitive formulation of a class of problems which is most important in physics, and shows how d'Alembert's Principle and Lagrange's Equations apply. It ties together the various detailed studies of the text and gives the
reader a comprehensive view of the subject as a whole. The book is designed as a careful and thorough introduction to
Mechanics, but not of course, in this brief compass, as a treatise. With the principles of Mechanics once firmly established and
clearly illustrated
by numerous examples
the student
is
well
which
may
viii
PREFACE
: :
Routh An Elementary Treatise on Rigid Dynamics Advanced Dynamics, by the same author; particularly valuable for its many problems. Webster, Dynamics good
be mentioned
and
also
material,
and
who
is
rudiments, but hard reading for the beginner through poor presentation and lacunae in the theory Appell, Mecanique rationelle a charming book, which the student may open vols. i and ii at any chapter for supplementary reading and examples. Jeans, Mechanics, may also be mentioned for supplementary exercises; as a text it is unnecessarily hard mathematically for the Sophomore, and it does not go far enough physically for the upperIt is unnecessary to emphasize the importance of classmari. further study by the problem method of more advanced and But to go difficult exercises, such as are found in these books.
;
y
work
would increase
unduly. It is not merely a formal tribute, but one of deep appreciation, which I wish to pay to The Macmillan Company and to The Norwood Press for their hearty cooperation in all the many difficult details of
its size
the typography.
Good composition
is
a distinct
aid in setting forth the thought which the formulas are designed to express. Its beauty is its own reward.
Benjamin Osgood Peirce, who first blazed the Mathematics 4, given at Harvard in the middle of the eighties the Author wishes to acknowledge his profound Out of these beginnings the book has grown, developed gratitude. through the Author's courses at Harvard, extending over more than forty years, and out of courses given later at The National
To
his teacher,
University of Peking. May it prove a help to the beginner in his first approach to the subject of Mechanics.
May
1937
CONTENTS
CHAPTER
I
STATICS OF A PARTICLE
1.
PAGE
1
Parallelogram of Forces
2.
3.
The
Triangle of Forces.
Addition of Vectors
3 4
7
4.
5.
The Polygon
Friction
of Forces
9
Problem
12
15
6.
CHAPTER
II
STATICS OF A RIGID
1.
BODY
21
2. 3.
4. 5.
n Forces
Equilibrium Couples in Space 8. Resultant of Forces in Space. Equilibrium 9. Moment of a Vector. Couples 10. Vector Representation of Resultant Force and Couple.
7.
6.
23 26 28 29 32 34 36
37
Resul-
tant Axis.
11. 12.
13.
Wrench
about a Line
Moment
of a Vector
38 40
41
42 43 46
CHAPTER
III
MOTION OF A PARTICLE
1.
2. 3.
4.
49 50 55 58
CONTENTS
PA OB
5.
Problem of Motion
60 63
6. 7. 8.
9.
Simple Harmonic Motion Motion under the Attraction of Gravitation Work Done by a Variable Force
Kinetic Energy and Work Change of Units in Physics
10.
11. 12. 13.
The Check
Motion
in
of Dimensions
64 69 72 75 76 79
81
;
a Resisting Medium 14. Graph of the Resistance 15. Motion in a Plane and in Space
16.
84
86
Vector Acceleration
17.
18. 19.
90 92 93 95 97 99
101
22.
23.
Centrifugal Force
24.
25.
The The
105
106
108 114
the Force
Central Force
26.
27.
28.
29.
The Two Body Problem to Determine The Inverse Problem Kepler's Laws On the Notion of Mass
114
115 118
CHAPTER
IV
Motion
120
123 126
127
2.
Applications
3.
4.
5.
The Equation
Moments
130
6. 7.
....
8.
9.
No
Point Fixed
139
141
Examples
CONTENTS
14.
xi
15. 16.
........
PAGE 143
145 146
151
CHAPTER V
1.
2. 3. 4.
5.
The Centrodes
Continuation.
....
162
165
6. 7. 8.
9.
The Dancing Tea Cup The Kinetic Energy of a Rigid System Motion of Space with One Point Fixed
Vector Angular Velocity Moving Axes. Proof of the Theorem of Space Centrode and Body Centrode
8
166
10.
11.
12.
13.
14.
176
177 179 182
15.
16.
12
Case.
17.
Continuation.
The General
The Commensurable
Case.
186 Periodicity Professor Sabine's Tracings of Lissajou's Curves between pages 190-191
CHAPTER
VI
ROTATION
1.
Moments
of Inertia
191
2. 3. 4.
5.
194 196
.
Continuation.
Moment
197 199
6.
7.
The Fundamental Theorem of Moments Vector Form for the Motion of the Centre The Invariable Line and Plane
Transformation of Moments about the Centre of Mass Moments about an Arbitrary Point
Mass
....
201
201
8.
9.
202 204
10.
205
xii
CONTENTS
Moments about
Euler's
11.
PAGE 207
12.
13. 14. 15.
...
Dynamical Equations
Moving Axes
20. 21.
22.
23.
24.
Cartwheels
237 241
25.
R&ume*
245
CHAPTER
Work
Continuation
:
VII
2.
3.
Curved Paths
Potential
4. 5.
6. 7.
Vanishing of the Internal Work for a Rigid System Kinetic Energy of a Rigid Body
Final Definition of
....
....
258 260
261
Work
8. 9.
Work Done by
a Moving Stairway
10.
Other Cases in Which the Internal Work Vanishes for a Rigid Body
CHAPTER
Impact of
VIII
IMPACT
1.
Particles
2.
3. 4. 5.
Continuation.
Oblique Impact
CHAPTER IX
RELATIVE MOTION AND MOVING AXES
1.
2.
285
Terms
of
Angular Velocity
285
CONTENTS
3. 4.
5.
xiii
Acceleration
PAGE 287
6.
290
291
292
The Problem
Lagrange's Equations in the Simplest Case Continuation. Particle on a Fixed or Moving Surface
.
. .
297
2. 3. 4.
5. 6. 7. 8.
The
Spherical
Pendulum
Geodesies
Lemma
Lagrange's Equations in the General Case Discussion of the Equations. Holonomic and Non-Holonomic
Systems
9.
Continuation.
Conclusion.
The Forces
Lagrange's Multipliers
10.
11.
Work
TTT
12.
13.
Computation
of
Qr
Virtual Velocities, an Aid in the Choice of the 14. On the Number of the Q T
15.
Forces of Constraint
Euler's Equations, Deduced from Lagrange's Equations Solution of Lagrange's Equations
.
16.
17.
18.
19.
2. 3.
The Problem
Lagrange's Equations d'Alembert's Principle
for a
345
2.
System
of Particles,
Deduced from
348
3.
The
Deduced from
349
riv
CONTENTS
PAGE
4.
5.
Application
6.
Examples
CHAPTER
XIII
Definition of 5
356
The
Energy
4.
5. 6.
7.
Virtual
Work
8. 9.
10. 11.
and Prospect
....
. .
.
379 379
381 386
12. 13.
Hamilton's Integral a Minimum in a Restricted Region Jacobi's Integral a Minimum in a Restricted Region
....
CHAPTER XIV
CONTACT TRANSFORMATIONS
1.
2. 3.
389 392
Theorem
395
4.
5.
6.
Theft-Relations
CHAPTER XV
SOLUTION OF HAMILTON'S EQUATIONS
1.
2.
3.
4.
5.
Its Treatment Reduction to the Equilibrium Problem Example. Simple Harmonic Motion H, Independent of t. Reduction to the Form, H' Examples. Projectile in vacuo
Pi
CONTENTS
6. 7.
xv
PAGE 429
Comparison of the
Two Methods
8.
9.
430
433 434 438
10. 11.
Variation of Constants
12.
Continuation.
A Second
Method
440 444
APPENDIX
A. Vector Analysis z = B. The Differential Equation (du/dt) f(u) C. Characteristics of Jacobi's Equation D. The General Problem of Rational Mechanics
:
447
456 466 476
491
INDEX
MECHANICS
CHAPTER
I
STATICS OF A PARTICLE
1.
Parallelogram of Forces.
By
a force
is
meant a push or a
pull.
spiral spring,
like
those used in the upholstered seats of automobiles, when compressed by a load, exerts a force. The earth exerts a force
of attraction
on a
a given point, 0, depends not on the or magnitude, merely intensity, of the force, but also on the direction in which it acts. Lay off a right line from O in the direction of the force, and mpke the length of ^
effect of a force acting at
The
the line proportional to the intensity of the force for example, if F is 10 Ibs., the length may be taken as
;
pJG
10 in., or 10 cm., or more generally, ten times the length which Then this directed right line, or vector, represents the unit force. Thus if a barrel gives a complete geometric picture of the force.
of flour is
is
be represented by a vector pointing downward and of length W, the weight of the barrel. On the other hand, the force which the rope exerts on the barrel will be represented by an equal and opposite vector, pointing
gravity
will
upward.
For,
action
opposite. forces act at a point, they are equivalent to a single force, which is found as follows. Lay off from the point the two vec-
When two
P and Q, which represent the given forces, and construct the parallelogram, of which the right line segments determined by P and Q are Fia 2 two adjacent sides. The diagonal of the paralfrom determines a vector, R, which represents drawn lelogram
tors,
1
2
the combined effect of
resultant of
MECHANICS
P
1.
P and Q. This force, R, is called the and Q, and the figure just described is known as the
Two
forces of 20 pounds each make an angle of 60 with each other. To find their resultant. Here, it is obvious from the geometry of the figure that the parallelogram is a rhombus, and
parallelogram offerees.
Example
20 N/3
34.64 pounds,
that _the length of the diagonal in question is = 34.64. Hence the resultant is a force of
action bisecting the angle between the
its line of
given forces.
Example 2. Two forces of 7 pounds and 9 pounds act at a point and make an angle of 70 with each other. To find their
resultant.
Graphical Solution.
Draw
angle by means of a protractor. Then complete the parallelogram and measure the diagonal. Find its direction with the
protractor.
the wall
Example 3. by a
picture weighing 15 Ibs. hangs from a nail in wire, the two segments of which make angles of 30
Here, the resultant, 15, of the two unknown tensions, T and T9 is given, and the angles are known. It is evident from the figure that T also has the value 15. So the answer is 15 Ibs.
:
Conversely, a given force can be detwo whatever. All that is needed directions composed along any to construct the is, parallelogram, of which the given force is
Decomposition of Forces.
lie
lines.
FBlTKf,
FIG. 4
FIG. 5
If,
in particular,
F cos
<p,
F sin
<p.
STATICS OF A PARTICLE
EXERCISES
1.
Two
forces of 5 Ibs.
and 12
Ibs.
make a
is
each other.
Show
of
makes an angle
2.
force.
Forces of 5
other.
3.
If
Ibs. and 7 Ibs. make an angle of 100 with each Determine the resultant force graphically. the forces in Question 1 make an angle of 60 with each
Then obtain an analytical however no solution, using trigonometry beyond a table of natural and sines, cosines, tangents.
Give
first
a graphical solution.
4. If two forces of 12 Ibs. and 16 Ibs. have a resultant of 20 Ibs., what angle must they make with each other and with the resultant?
5.
A A A
Resolve
it
into an easterly
into an
Resolve
46.20
Ibs.
;
it
easterly
7.
Ans.
19.14 Ibs.
it
Resolve
into
two forces that make angles of 30 and 40 with Only a graphical solution is required.
2.
Analytic
Treatment by Trigonometry.
The problem
of
finding the resultant calls for the determination of one side of a triangle when the other two sides and the included angle arc
known;
problem
(1)
and
is
also
of
The
:
first
solved
c2
by the Law
Trigonometry
= =
a2
P,
+
b
62
2ab cos C.
c
Here,
=
180
2
Q,
R,
c
C =
and hence
(2)
72
2
w
cos
co.
= P2
+Q +
2PQ
FIG.
Example. Forces of 5 Ibs. and 8 Ibs. make an angle of 120 with each other. Find their resultant. Here,
R =
2
25
64
-2X5 X R = 7 Ibs.
49
To complete the solution and find the remaining angles we can use the Law of Sines :
4
~
(3)
1
MECHANICS
I,
sin
A
sm
sin
B
sm
sin
Thus
^>
(4)
There
is
is
no
difficulty here
= sin co. to) angle In the numerical example above, Equation (4) becomes
used, since sin (180
8
sin
<p
|\/3
<p
Thus
sin
<p
4-V3,
is
cos
= |
<t>
81 47'.
The
third angle
angles of a triangle is
sum
of the
A +B+C =
To sum
up, then
:
180.
Compute
the resultant
by the Law
of Cosines
of Sines.
EXERCISES
Give both a graphical and an analytical solution each time. 1. Forces of 2 Ibs. and 3 Ibs. act at right angles to each other. Find their resultant in magnitude and direction.*
2.
Forces of 4
other.
Ibs.
make an
angle of 70
with each
Equilibrium. The Triangle of Forces. Addition of VecIn order that three forces be in equilibrium, it is clearly necessary and sufficient that any one of them be equal and opposite
3.
tors.
The condition can be expressed conveniently by aid of the idea of the addition of vectors. First of all, two vectors are defined as equal if they have the
to the resultant of the other two.
same magnitude,
direction,
and
lie.
sense,
no matter where
in the
may
* Observe that in this case it is easier to determine the angle from its tangent. Square roots should be computed from a Table of Square Roots. Huntington's Four-Place Tables are convenient, and are adequate for the ordinary cases that arise in practice. But cases not infrequently arise in which more elaborate tables are needed, and Barlow's will be found useful.
STATICS OF A PARTICLE
Vector Addition.
Let
and
its initial
point of
initial
is
the terminal point of B, is defined as the vector sum, or, simply, the sum of A and B
:
C = A
It is
+ B.
obvious that
B
Any number
successively.
+A
= A
+ B.
added by applying the
definition
of vectors can be
(A
+ B) +
A!
C = A
+
-
(B
C).
Consequently the
sum
+A +
2
+ An
independent of the order in which the terms are added. For accuracy and completeness it is necessary to introduce the nil vector. Suppose, for example, that A and B are equal and Then their sum is not a vector in any sense as yet opposite.
is
considered, for the terminal point coincides with the initial point. When this situation occurs, we say that we have a nil vector, and
denote
it
by
+B =
0.
We
write, furthermore,*
=-
A.
sufficient,
Equilibrium. The condition, necessary and three forces be in equilibrium is that their vector
metrically this
that
Geois equivalent to saying that the vectors which represent the forces can be drawn so that the figure will close and form a triangle. From the Law of Sines we have
0.
:
sum be
P
sin
"
FIG. 8
* It
is
Q
p
sin q
E
sin e
+q+e=
180.
not necessary for the present to go further into vector analysis than the Later, the two forms of product will be needed, and the student may be interested even at this stage in reading Chapter XIII of the author's Advanced Calculus, or Appendix A.
MECHANICS
Since sin (180 A) = sin 4, we can state the result in the Let three forces, P, Q, and E, acting on a parform. following Denote the angles between ticle, be in equilibrium.
the forces, as indicated, by p, q, e. Then Equation (1) represents a necessary condition for equilibrium.
We
Conversely, this condition is sufficient. thus obtain a convenient solution in all cases
except the one in which the magnitudes of the forces, but no angles, are given. Here, the Law of Cosines *
FIG. 9
Example
1.
the angles between them. First, solve the problem graphically, measuring the angles. Next, apply the Law of Cosines
:
42
=
cos
52
+ 62 _
=f
,
5
<p
X
=
cos p,
23'.
<p
41
second angle
is
now computed by
5
sin
i
the
Law
of Sines
4
sin
<p
5V7
16
'
55
46'.
82 51'. 40 Ib. weight rests on a smooth horizontal Example cylinder and is kept from slipping by a cord that passes over the cylinder and carries a 10 Ib. weight at its other Find the position of equilibrium. end.
third angle
2.
is
The
The cord
is
it
passes over a smooth surface, the tension in The surface of it is the same at all points.
the cylinder
is
normal to
angle
that
its
surface.
the
drawn
to
the weight
FIG. 10
* If we had a large number of numerical problems to solve, it would pay to use the more elaborate theorems of Trigonometry (e.g. Law of Tangents). But for ordinary household purposes the more familiar law is enough.
STATICS OF A PARTICLE
makes with the we see that
or
sin 6
vertical.
Then from
10
40
sin
sin 6
90'
=
,
=
EXERCISES
14 29'.
1.
Forces of
7, 8,
particle at rest,
Find
the angles they make with one another. 2. Forces of 51.42, 63.81, and 71.93 grs. keep a particle at rest. What angle do the first two forces make with each other? Find
the other angles.
3.
A
is
level
and
weightless string passes over two smooth pegs at the same In the middle, carries weights of and at its ends.
there
knotted a weight W. What the segments of the string do angle make with the vertical, when the
system
is
at rest ?
w
sin 6
Ans.
4.
-^
to stakes at opposite points
boat
is
tied to the
banks.
and
ft.
on the
125
stream is 200 ft. broad. 20 Ibs., what is the tension in the other rope?
6.
long; the other, 150 ft.; and the If the tension in the shorter rope is
inclined planes, back to meet along a horizontal straight back, make and angles of 30 arid 45 with line,
Two smooth
the horizon.
FIG. 12
weight of 10
is
Ibs.
placed
carries
a weight W,
plane passes over the top of the planes and resting on the other plane and attached to
on the
first
held
by a cord that
W must have.
the
The Polygon
Forces.
From
case of three forces the generalization to the case of n forces acting at a point presents no difficulty. Add the forces geometrii.e. the The vector vector law. cally, by
sum
represents the resultant of all n forces. Thus, in the figure, the resultant is given by
FIG. 13
MECHANICS
is
the point
is
The
nil vector,
but not necessarily a polygon in the sense of elementary geometry, since its sides may intersect, as
in Figure 14.
ful-
each of
and only if the sum of the projections two lines that intersect is zero.*
Analytically, the resultant can be represented as follows. Let a Cartesian system of coordinates be assumed, and let the k and Yk components of the force Fk along the axes be If,
by
and Y,
we have
v _ A! v v ~r A A Y= Y + F +
,
,
-\-
v ~r A n + Yn
,
X*,
t=l
or
I
Xk
>
or
*>
The
force
if,
and only
if,
the resultant
is nil,
and
this will
be the case
if
(2)
t-i
2)
Xk
0,
t=i
2Y
0.
EXERCISE
acted on by three forces, all of which lie in the same vertical plane a force of 10 Ibs. making an angle of 30 with
Ib.
A4
weight
is
the vertical, and forces of 8 Ibs. and 12 Ibs. on the other side of the vertical and making angles of 20 with the upward vertical
vertical respectively.
Space of Three Dimensions. If more than two forces act at a But they can be added two point, they need not lie in a plane.
* Cf. Osgood
1-6.
STATICS OF A PARTICLE
at a time
by the parallelogram law, the first two thus being their resultant and this force in replaced by a single force turn compounded with the third force; etc. The broken line that represents the addition of the vectors no longer lies in a plane, but becomes a skew broken line in space, and the polygon of forces becomes a skew polygon. The components of the resultant
force along the three axes are
n
:
(3)
X=
2^
X^
Y=
is
:
n^
2^
Yk,
Z = 2^
Zk.
The
In
all of
being positive
Yk Zk are algebraic quantities, these formulas, k when the component has the sense of the positive
,
,
and negative when tho sense is the opposite. In solving problems in equilibrium it is frequently simpler to single out the components that have one sense along the line in question
axis of coordinates,
and equate their sum, each being taken as positive, to the sum of the components in the opposite direction, each of these being taken as positive, also. The method will be illustrated by the
examples in friction of the next paragraph.
Let a brick be placed on a table let a string be to the let the be and fastened pulled horizontally with brick, string a force F just sufficient to move the brick. Then the law of
5.
Friction.
physics
is
that
F =
pR,
r
R\
J
where R (here, the weight of the brick) is the normal* pressure of the table on the
-p^
15
brick, arid ju (the coefficient of friction) is a constant for the two surfaces in contact. Thus, if a second brick were placed on top would be doubled, and so would F. of the first,
can state the law of friction generally by saying: When two surfaces are in contact and one is just on the point of slipping
* Normal means, at right angles to the surface in question. The normal to a surface at a point is the line perpendicular to the tangent plane of the surface at the point in question.
We
10
MECHANICS
normal pressure
where /*, the coefficient of friction, is independent of F and R, and depends only on the substances in contact, but not on the area of the surfaces which touch each other. For metals on metals n usually lies between 0.15 and 0.25 in the case of statical fricFor sliding friction /* is about 0.15; cf. Rankine, Applied tion.
Mechanics.
determining
simple experiment often performed in the laboratory for Let one of the surfaces be repre/x is the following.
sented by an inclined plane, the angle of which can be varied. Let the other surface be represented by a rider, or small block of the substance in question, placed on the If the plane is gradually tilted from plane. a horizontal position, the rider will not slip for a time. Finally, a position will be reached for which the rider just slips. This angle of the plane is known as the angle of friction and is usually denoted by X. Let us show that
n
Resolve the force of gravity,
tan
X.
W,
into its
W sin
And now
directed
X,
W cos W sin X
W cos
,
X.
the forces acting up the plane (i.e. the components up the plane) must equal the forces down the plane, or
F =
forces
and the forces normal to the .plane and upward must equal the normal to the plane and downward, or
R =
Hence
X.
R
But
Consequently
sin X
cosX
= an x ^
/J2.
X.
F =
tan
STATICS OF A PARTICLE
Example.
11
is placed on a rough inclined plane, cord attached to the weight passes over a smooth pulley at the top of the plane and carries a weight at its lower end. For what values of will the system be in
50
Ib.
weight
angle of elevation,
30.
equilibrium
if
jut
-J
Here, X
slip
down
the plane.
slipping.
weight will just large enough to prevent Suppose Then friction acts up the plane, and the forces which
so, if
is
<
30, and
Ib.
Hence
'F+W
W
Fio. 17
50 sin 30
50 cos 30
Fia. 18
F =
%R.
W=6
If,
~
o
25
17.8 Ibs.
is slightly increased, the 50 Ib. weight will obviously now, be in equilibrium, and this will continue to be the case until the 50 Ib. weight is just on the point of slipping up the plane. = 32.2 Ibs. as the student can now prove This will occur when for himself. for which there is Consequently, the values of for which are those equilibrium
still
17.8
32.2.
EXERCISES
1.
If
2,
3, is
Ans.
rough, /z = 4 49' g
5, 3.
find the
23
44'.
If the
one
on which the 10
of values for
W that
weight
rests is rough,
/z
= T^,
12
3.
MECHANICS
Prove the formula
H
tan X
i.e.
by means
6.
the
Law
of Sines,
3, (1).
50
Ib.
is
cord
fastened to the weight, passes over a smooth pulley 2 ft. above the plane, and carries a weight of 25 Ibs. which hangs freely at To find all the positions of equilibrium. its other end.
25 cos 6
R,
50.
25 sin
R =
the equation
FIG. 19
(1)6 cos
of the
sin
2.
This equation
(2)
is
form
a cos
is
+ b sin =
c,
and
solved as follows.
Divide through by
Va +
2
b2
cos
a cos
+
(0
sin
sin 6
Va +
2
'
62
or
(5)
cos
a)
Va +
2
b2
* The student should observe carefully the trigonometric technique set forth in this paragraph, not merely because equations of this type are important in themselves, but because the practical value of a working knowledge of trigonometry Of far greater scope and importance is not confined to solving numerical triangles. in practice are the purely analytical reductions to other trigonometric foi-ms, and solution of the trigonometric equations. That is one of the reasons why the
harder examples at the end of the chapter are valuable. They not only give needed practice in formulating mathematically physical data they require also the ability to handle analytical trigonometry according to the demands of practice.
;
STATICS OF A PARTICLE
The
(6)
13
angle a
is
most
easily determined
Thus a
which one seen to be one of two angles clear by plotting the point on the unit circle
is
:
is
rendered
z2
2
2/
1,
= __JL__
b
'
7 y
__
The angle from the positive axis of x to the radius drawn to this It is point is a. Thus we have a graphical determination of a.
to observe in
not necessary to compute the coordinates accurately, but merely which quadrant the point lies, so as to know which
Thus if b > 0, a must be root of the equation for tan a to take. an angle of the first or second quadrant. Finally, if c/Va 2 b2
the equation has no solution. In defining a, it would, of course, Jiave answered just as well if sin a and cos a had been interchanged, and if either or both the
is
1,
ratios in (4)
had been replaced by their negative values.* Returning now to the numerical equation above, we see that
tan a
=
,
sin
>
0,
9 28'
cos
(6
9 280
70
48'.
first
Since
6 in
9 28'
impossible, and
80
16'.
friction
have determined the point of the plane at which all the is called into play and the 50 Ib. weight is just on the For other positions, F will not equal pR. point of slipping. Such a position will be one of equilibrium if the amount of friction actually called into play, or F, is less than the amount that could
* Equation (2) might also have been solved by transposing one term from the the right-hand side and squaring. On using the Pythagorean Identity
:
We
left- to
sin 2
in general,
2 -f cos
1,
we should be led to a quadratic equation in the sine or cosine. This equation will, have four roots between and 360, and three of them must be excluded. Moreover, the actual computation by this method is more laborious.
14
MECHANICS
It
seems plausible that such points lie but this conclusion is not im;
F =
0, still,
25 cos
B,
is less
the
Mfiis
also less.
We must
F < nR
or
OK cos 25
-- -amount
g
available, or
50-25 sin
^50-25 < 5
b
sin
This
will
obviously be so
6 cos 6
if
<
sin
0,
80 16'
<
2
<
90,
or
if
6 cos
rif
+
1
sin
<
2,
6
or
if
cos
(0
9 28
< -='
9 28' also starting with the value 80 16', increases, and cos decreases. 9 Conseincreases, (0 28') consequently
0,
As
quently our guess is borne out by the facts, and the 50 Ib. weight be in equilibrium at all points on the table within a circle of radius .343 ft., or a little over 4 in., whose centre is directly under
will
the pulley.
EXERCISES
Solve the same problem if the plane is inclined at an angle with the horizon, and the vertical plane through the weight and the pulley is at right angles to the rough plane.
1.
of 15
Ana.
43 33'
g0g
69
24'.
2. At what angle should the plane be tilted, in order that the region of equilibrium may just extend indefinitely down the plane?
3.
of the third
equation M
3 sin 6
2 cos
1.
STATICS OF A PARTICLE
4.
15
all eight ways of solving Equation (2), the right-hand sides of Equations (4) equal to given by setting cos a, + sin a and sin a, cos a, where the signs are
Show
dx
a s-.
6.
:
cos x
b sin
<p
sin
<p
3 sin 2
<p
5.
Suggestion.
2<p.
EXERCISES ON CHAPTER I*
rope runs through a block, to which another rope attached. The tension in the first rope is
1.
is
120
Ibs.,
is
What
2.
and the angle it includes is 70. the tension in the second rope ?
weighing 160
Ibs.
is
A man
30
lying in a
hammock.
angle of
The rope
at his head
makes an
in the
two
ropes.
load of furniture
is
The rope
;
that binds
round 100.
4.
passes over the round of a chair. The tension on one side of the is 40 Ibs. and on the other side, 50 Ibs. and the angle is What force does the round have to withstand ?
canal boat
is
on the bank.
4
The
400
Ibs.
and
it
makes
is
an angle of 15
What
as
FlG 21
-
shown
and
in the oblique
member?
6.
Three smooth pulleys can be set at pleasure on a horizonThree strings, knotted together, pass over the
* The student should begin each time by drawing an adequate figure, illustrating the physical objects involved, and he should put in the forces with colored ink or A bottle of red ink, used sparingly, contributes tremendously to clear pencil.
thinking.
16
pulleys
MECHANICS
and carry weights
of 7, 8,
and 9
Ibs.
How must
the pulleys be set, in order that the knot may be at rest at the centre of the circle ?
7. A telegraph pole at two crossroads supports a cable, the tension The cable lies in in which is a ton.
FIG. 22
a horizontal plane and is turned through a right angle at the pole. The pole is kept from tipping by a
stay from
top to the ground, the stay making an angle of 45 with the vertical. What is the tension in the stay?
its
8. The figure suggests a stake of a circus tent, with a tension of 500 Ibs. to be held. What is the tension in
FIG. 23
the stay,
9.
if
Two men
by a rope that over two smooth passes pulleys and is knotted at A. How hard are they
pulling?
10. If, in the preceding question, instead of being knotted at A, the two ropes the men have hold of passed over
pulleys at
FIG. 24
A and
is placed in a smooth hemispherical bowl a weight string, attached to the weight, passes over the edge of the bowl and carries a weight P at its other end. Find the position of
11.
equilibrium.
12.
Solve the same problem for a parabolic bowl, the rim being
One end
of a string
is
made
fast to a
peg at A.
The
string
passes over a smooth peg at J?, at the same level as A, and carries a at its free end. A smooth heavy bead, of weight W, can weight slide on the string. Find the position of equilibrium and the
bead weighing
Ibs.
can
is
slide
on a smooth
vertical
circle of radius a.
To
the bead
STATICS OF A PARTICLE
17
over a smooth peg situated at a distance %a above the centre of the circle, and has attached to its other end a weight P. Find
the positions of equilibrium. A 50 Ib. weight rests on a smooth inclined plane (angle with the horizontal, 20) and is kept from slipping by a cord which
all
16.
1 ft.
which
16.
carries a weight of
25
Ibs.
above the top of the plane, and Find the at its other end.
in the
position of equilibrium.
slide
on a smooth wire
form of a
parabola with vertical axis and vertex at the highest point, A string attached to the bead passes over a smooth peg at the focus
of the parabola and carries a weight at its other end. in general there is only one position of equilibrium ;
Show
that
but some-
positions are positions of equilibrium. * can slide on a smooth wire in the form weightless bead of an ellipse whose plane is vertical. A string is knotted to the
all
times
17.
bead and passes over two smooth pegs at the foci, which are at the same horizontal height. Weights of P and Q iro attached to the two ends of the string. Find the positions of equilibrium.
f
18.
An
its
ends
points and
of the
smooth bead.
Show
bead
is
one of equilibrium,
if
properly
directed.
19.
make equal
bead of weight P can slide on a smooth, vertical rod. To the bead is attached an inextensible string of length 2a, and having its other end carrying at its middle point a weight made fast to a peg at a horizontal distance a from the rod. Show
is
P tan
where
0,
<p
<p
(P
+ W} tan
6,
sin
<p
1,
make with
the vertical.
* Questions of this type may be objected to on the ground that a force must act on mass, and so there is no sense in speaking of forces which act on a masslcss ring. But if the ring has minute mass, the difficulty is removed. The problem may be thought of, then, as referring to a heavy bead, whose weight is just supported by a vertical string. Since the weight of the bead now has no influence on the position of equilibrium, the mass of the bead may be taken as very small, and so, physically
negligible.
18
21.
If,
MECHANICS
in the preceding question,
P=
show that
21
V =
long
38
49'.
55',
2P.
22.
A
is
laid over
smooth beads
librium,
if
of
form of a loop 60 in. two smooth pegs 20 in. apart and carries two weight P and W. Find the position of equi-
come
Ans.
W sin 6
+
I (i
cos
(f>
together.
^>,
= P sin = 3 cos 6
2
cos p
cos e
cos e
~) cos
2
hence
9 cos 4 6
6 cos 3 6 X
8\ cos 2
2
+
2
6X cos 6
0,
where
23.
_ (W - P ~ TP
the
Show
10,
that
if,
in
preceding
TF
25
of
8',
and
P=
and
an inextensible string a in. long is made fast to a peg A and at the other end is knotted a weight W. A second string, attached to W, passes over a smooth peg at B, distant b in. from A and at the same level, and carries a weight P at its other end. Find the position of equilibrium. If P = Wj how far below the level of the pegs will the first
24.
One end
25.* Observe the braces that stiffen the frame of a railroad car. Formulate a
by what you
sec,
and solve
it.
bridge of simple type is suggested designing such a structure, the stiffness of the
A
J_
FIG. 26
following four problems are given only in outline, and the student thus has the opportunity of filling in reasonable numerical data and formulating a cleancut question. It is not necessary that he respond to all the problems; but he should demand of himself that he develop a number of them and supplement these by others of like kind which he finds of his own initiative in everyday life. For, imagination is one of the highest of the intellectual gifts, and too much effort
The
it.
STATICS OF A PARTICLE
where these come together
is
19
is
planned as if the members were all pivoted there. Draw such a bridge to scale and find what the tensions and thrusts will be if
it is
Make
27.
to support a weight of 20 tons at each of the points A, B. a reasonable assumption about the weight of the road bed,
tie rods, etc.
_j_
1(j
j 100 lbs<
, Jo o
The
a whiffle-tree 3
is
100
is
Ibs.
The
distance from
is
10
in.
What
the ten-
100 Ibs.
Fm
27
a small passenger car, 28. Have you ever seen a funicular hauled up a steep mountain by a cable? How is the tension in tho cable related to the weight of the car? When the direction
of the cable is
changed by a
is
cable passes,
what
FRICTION
29.
.ire
5,
3.
If
both
rough and
what
is
0.1.
weightless bead can slide on a rough horizontal wire, A cord is attached to the bead and carries a weight at
its
other end, thus forming a simple pendulum. Through what angle can the pendulum swing without causing the bead to slip?
31.
water main 5
ft.
in diameter
is
filled
with water to a
mouse tumbles in and swims to the nearest point depth of 1 ft. on the wall. If the coefficient of friction between her feet and the
pipe
is
-J-,
will she
be drowned?
32.
heavy bead
is
placed on a rough vertical circle, the f If the angle between the radius
.
drawn
to the
vertical
is
will slip
when
rope
33.
zontal plane, and pulled until the weight just moves. Find the tension in the rope, and show that it will be least when the rope
friction.
The same
20
35.
MECHANICS
A
50
Ib.
weight is placed on a rough inclined plane, M = t, 10. A string tied to the weight passes over a smooth peg at the same level as the weight and carries a weight of 7 Ibs. at its lower end. When the system is released from rest,
angle of inclination,
will it slip ?
36.
37. If the parabolic wire described in Question 16 is rough, and the weights are P and W, find all positions of equilibrium. Ans. When W, the limiting position is given by one or
P^
tan i 2
~ - P ~" P+
W.
W W
'
1
/z'
tan ? 2
"
W~P
W+P
'
1
n'
P =
Cast iron rings weighing 1 Ib. each can slide on a rough A string 6 ft. long is attached to each of horizontal rod, M = -J. How far these beads and carries a smooth bead weighing 5 Ibs. apart can the two beads on the rod be placed, if the system is to
to a length of 6
string are
level.
in.
Law,
is
stretched
of the
The ends
made
fast at
two points 6
Ibs. is attached to the mid-point of the string lowered. Find the position of equilibrium, neglectcarefully of the Ans. is given by the equation string. ing the weight
weight of 4
and
cot
40.
120
(1
cos
0,
0).
to one-tenth of a degree.
Ans.
41.
= 14|.
is
The mast
of a derrick
is
40
^20
ft.
and a stay
fastened
riG. oo &Q
4 tons and resting on a paveThe boom is ment, M = f 35 ft. long, and its end is distant 20 ft. from the top of the mast. Is it possible to raise
.
11-
a 5 ton weight, without the derrick's being pulled over, the tance from the stone to the derrick being 120 ft.?
dis-
CHAPTER
II
Q, act on a
which their lines of action meet. These latter forces are now seen to have a resultant,
(1)
R =P+
Q,
and
fr I(3
its line
29
AB
two segments, AC and CB. Let the lengths of the segments be denoted as follows: AC = a, CB = 6, AB = c, DC = h. From similar triangles it is seen that
P
S
Hence
(2)
h
a'
Q
AS'
h
6'
aP =
a
:
bQ.
Moreover,
(3)
+6=
c.
To sum up, then The original forces, P and Q, have a resultant determined by the equations (1), (2), and (3).
Example.
The
familiar gravity balance, in which one arm, a, to be determined from which the weight
suspended, is short, and the other arm, b, from which the rider Q hangs is long, is a
is
FIG. 30
case in point.
21
22
Opposite Forces.
If
MECHANICS
are opposite in direction, and have a resultant. Introduce a unequal (Q force E (Equilibriant) parallel to P and Q and having the sense of P, determining it so that Q will be equal and opposite to the Then resultant of P and E.
and
>
Q = P cP a = b
Thus
1
+ E,
bE,
c.
P and Q
a resultant,
(4)
FlG 31
-
the point
(5)
R = Q - P, cutting AC produced
:
in
(6)
If
aP = bQ, a = b + c.
later,
and Q are equal, they form a couple and, as we shall show cannot be balanced by a single force; i.e. they have no
Consider a pair of nut crackers.
resultant (force).
Example.
The
forces that
;
members
the pressure of the nut; force the hand exerts, balancing the
are
if)
Q,
resultant, R, of
We
which says that the effect of a force on a body is the same, no matter at what
point in its line it acts. Thus a service truck will tow a mired car as effectively
(but no
it is
FIG. 32
is
more
effectively)
long, as
when
is
parallel to the
Moreover,
it is
not necessary to think of the point of application It might be the centre of a ring.
For we can always imagine a rigid weightless truss attached to the body and extending to the desired point. But we always think of a body, i.e. mass, on which the system of forces in question acts.
23
10 ton truck passes over a bridge that is 450 ft. long. is one-third of the way over, how much of the load goes to one end of the bridge, and how much to the other
1.
When
the truck
end?
2.
Ans.
is
to be raised
by a
lever 6
ft.
long, the
fulcrum being at one end of the lever, and the weight distant 9 in. from the fulcrum. What force at the other end is needed, if the
weight of the lever
3.
is
negligible?
two baskets of pottery by a pole 6 ft. long. one basket weighs 50 pounds and the other, 70 pounds, how far arc the ends of the pole from his shoulder?
coolie carries
If 2.
Analytic Formulation
n Forces.
Suppose that n
parallel
forces act.
are not equal and opposite, can be and this in turn combined with a third their resultant, replaced by the one of the given forces, until number has been reduced to two.
These will in general have a resultant, but, in particular, may form a couple or be in equilibrium. Thus the problem could be solved
piecemeal in any given case.
An
arid
with n
can be obtained as follows. Begin 2 and denote the forces by P l and P2 Moreover, let P 1
.
tities,
one direction
AT
1
negative,
if
they act
Next draw a
line
VI* perpendicular*
P
1
XssQ
We
3J=1
-p\
1
**
*2
Fl(J
33
and
2,
and regard
and negative)
numbers,
P
P
2
cut the
The forces
(l)
and
have a resultant,
R =
P,
+P
provided
P +P ^
l
0.
* An oblique direction could be used, but in the absence of generalization, the orthogonal direction is more concrete.
any need
for
such a
24
Suppose,
first,
MECHANICS
that
and
Then, by
1,
R =
where
a
provided Xj
P,
+P
,
x 1?
x2
x,
<
x 2 (algebraically).
(x
Hence
(x2
x,)
P =
l
x)
2,
and from
The
< x is dealt with in a similar manner, as and P2 are both negative. Next, suppose P 1 and P2 have opposite senses, but
case that x 2
is
also
Let
P <
1
|
0,
P + P, * 0. P > 0, P
l
<P <
2,
where
|
absolute
let
value
of
x.
Thus
1,
xl
x2
Then, by
12
-?!
+P
b
2,
x, corresponding to
it is
obtained as follows
=
1,
x
(5)
:
x lt
xtl
(x
x { ) (-
t)
(x
x2 )
2.
(1)
and
(2), as
the
solution of the problem. It remains merely to treat the remaining cases in like manner.
The
(1)
and
(2).
We
are
now ready
1.
THEOREM
lt
Pn
act.
They
will
have a resultant,
R = ^
0,
---P! ^
its line
+ Pn
provided this
sum
and
where
/o\ (3)
_ P\ X ;= x
'
i"
"T
Pn X n
25
tion.
it
The proof can be given by the method of mathematical inducThe theorem is known to be true for n = 2. Suppose
were not true for all values of n. Let m be the smallest value n for which it is false. We now proceed to deduce a conSuppose, then, that P D which the theorem is
, ,
of
tradiction.
Pm
for
3,
false,
1.
By
hypothesis,
Now,
let
it is
us say.
+ + Pm * 0. 1 of the P, whose sum is not possible to find m + Pm-, * 0, Pi + forces have, by hypothesis, a resultant These m R' = P + + Pw Pi
'
's
'
lf
and
its
_/ /I x ~
3*i
'
"
'T~
-*
#w~i
'
/>!
+
.
-TP^;
since the
theorem holds by hypothesis for all values combine this force with Pw Since Next,
R'
of
n < m.
+ Pm
R'
is
= 1\
+ Pm *
-
0,
R =
and
its line
_
+ Pm = P +
1
+ Pw
of action
given by
the equation
R'x' #'
m.
Let
+ Pm x m + PM
P lgl +
+Pw --+P
a,
But
false for
assumption that the theorem Hence the theorem is true for all values of n.
is
Couples.
(4)
P,
+
lf
Thei1
Pn * + Pn = 0, P + + Pn-! ^ 0,
-
0.
and the
forces
Pn_
have a resultant,
i
ft' = P ^
4r ... 4n^
Pn
1>
whose
line of action is
-/
given
by
'
the equation
'
P!
%l H~
26
If,
MECHANICS
in particular, x'
=
n
xn
this resultant,
72',
will
line of action as
Pn
and
since
0,
R'
the
+P
Xn
or
R = - Pn
f
We
r> f n
then have
PI X \
'
I
'
~T 1 n-i
X n -i
y
(5)
Pl*l
conversely,
if
'
PnX n =
0.
And
and
we can
infer equilibrium.
_^
xn
Hence
(6)
Pl*l
'
'
+PnX n ^Q.
We then have a couple. And conversely, if (4) and (6) hold, we can retrace our steps and infer that we have a couple. We have thus proved the following theorem.
THEOREM
arulonlyif
2.
The n
parallel forces
lt
P n form
a couple
if,
PI
+ P. =
0.
Equilibrium. The case of equilibrium includes not only the case above considered (Pn 0), but also the case in which all
forces vanish.
We
THEOREM
if,
3.
The n
ly
P n are in equilibrium
and only
if
PI Pi X l
+ +
Pn =
----h Pn X n =
0.
3. Centre of Gravity. mn Let n particles, of masses Wi, be fastened to a rigid rod, the weight of which may be neglected, and let them be acted on by the force of gravity. If the rod is supported at a suitable point, (?, and is at rest, there will be no
,
,
tendency to turn
of gravity of the
in
any
direction.
This point
position
-
is
particles,
and
l
its
is
determined by the
equation
x + m +
l
+ mn xn + mn
STATICS OF A RIGID
If
BODY
27
the particles lie anywhere in a plane, being rigidly connected a by truss work of weightless rods, and if we denote the coordinates of m k by (xk, yk) the centre of gravity is defined in a similar manner
1
(sec below)
(1)
/
and
its
coordinates,
(x,
?/),
are given
>n
by Equations
and
fl y
(2)
9x
_ ~
*KI y\
H----
n^
+ yn + .-.+m.
'
For, let the plane of the particles be vertical, the axis of x being Then the system is acted on by n parallel forces, horizontal. whose lines of action cut the axis of x at right angles in the points
%!,'', x ny and
tion (1).
is determined in position by Equathe rotating plane through a right angle and repeatthe ing reasoning, Equation (2) is obtained. The centre of gravity of any material system, made up of par-
their resultant
On
ticles
and
line, surface,
if
and volume
distributions,
is
defined as a
the parts of the system be rigidly connected point, (7, if G be supported, there will be no tendency and by weightless rods, of the system to rotate, no matter how it be oriented. We have
such that,
proved tho existence of such a point in the case of n particles For n particles in a plane we have assumed that lying on a line. a centre of gravity exists and lies in the plane, and then we have
computed its coordinates. We shall prove later that n particles always have a centre of gravity, and that its coordinates are given by Equations (1), (2), and
m +
.
+ mn
.
In the case of a continuous distribution of matter, like a triangular lamina or a solid hemisphere, the methods of the Calculus lead to the solution. It is the definite integral, defined as the
limit of a
is
sum, that
in
is
essential
the formulation.
integrals suffice.
But even
some
and
volume
The
integrals simplify the computation. following centres of gravity are given for reference.
fl)
Solid hemisphere
x
:
fa.
a.
b)
c)
x
x x
= f h.
f A.
d)
e)
Conical surface
Triangle
:
28
4.
MECHANICS
Moment
of a Force.
Let
be a point of the plane. By the moment of plane, and let is meant the product of the force by the distance from
line of action, or
hF.
A moment may
furthermore be defined as
an algebraic quantity, being taken as positive when it tends to turn the body in one direction (chosen arbitrarily as the positive lies on Finally, if direction), and negative in the other case.
the line of action of the force, the moment is defined as 0. Let a force F act at a point (x, y), and let the components of F along the axes be denoted by X, Y. Then the moment (taken
algebraically) of
(1)
F about
the origin, 0,
is
xY-yX.
Let the equation of the line of action of
Proof.
in Hesse's
be written
Normal Form
x cos a
+ y sin a =
h.
/
O
i
FIG. 34
FIG. 35
Suppose,
first,
that the
moment
is
positive.
Then
a).
:
it
will
be
hF =
Here,
2ir
x (F cos a)
+
6,
y (F sin
Fig. 34
is
the complement of
a =
Hence
cos
+
cos
2rr.
a =
sin 6,
6,
sin
0,
and
since
X = F cos
the proof
is
Y = F sin 6,
complete.
29 be connected
is
negative,
a.
and
6 will
,- a --Hence
cos
Bt
sin
+ |.
cos
6.
a =
sin
9,
a =
The moment
will
now be
x
(
repressed as
hF =
and thus we
is
F cos a) +
F sin a),
as before.
arrive at the
(1) holds in all cases. From (1) we prove at once that the moment of the resultant of two forces acting at a point is the sum of the moments of the two
moment.
The same
given forces.
Fj,
F2
xY
The components
yX
2,
and
xY
yX
form
X +X
l
and
Y + F
and
its
moment
is
From
is
moment
of a force
(2)
(x-x )Y-(y-y,}X.
is
physical meaning of the moment of a force about a point a measure of the turning effect of the force. Suppose the body were pivoted at 0. Then the tendency to turn about 0, due to
The
the force F,
set of forces
is
expressed quantitatively by the momenta And a augment or reduce one another in their combined
turning effect according to the magnitude and sense of the sum From this point of view of the moments of the individual forces. a moment is often described in physics and engineering as a torque.
A couple has already been defined as 6. Couples in a Plane. a system of two equal and opposite parallel forces. A couple cannot be balanced by a single force, but is an independent mechanical entity the proof is given below. By the moment of a couple, taken numerically, is meant the
;
lines of action
30
MECHANICS
THEOREM.
Two
moment and
sense are
equivalent.
Suppose first that the forces of the one couple are parallel to the forces of the other couple. Then, by proper choice of the axis of x, we can represent the couples as indicated, where
P.
J
P.I
,
I
+ P.-O.
4
0<P,;
P +P =
3
3
0,
0<P
<
l
>
p
*\
"
P,
J
Fxo.36
x^xt + h, = * + *4
J,
0<h;
Now
P P
lf
2,
3t
4.
These are
For,
2
3 4
P,
+P -P -P =
_
/:> *
and
_|_p-r Pr ^l
1
I
** /
_ P
2
*
,. 3 "^S
**/
/r
(P ^JTj
-4- X P2 / ^2 -f
-I
P
i
It
(P
V.^3
-\I
-*
P4/ }
4- P r ^3 *'4
'
7 ^
fl VJ>
Hence the first couple is balanced by the negative of the second couple, and thus the theorem is proved for the case that all the
forces are parallel.
If
OA
and
the forces of the two couples are oblique to each other, let OB be two lines at right angles to the forces of the first
couple
and to those
of the
second
proved the
and P.*
Furthermore,
the
second
couple, reversed in sense, can be repLet resented by the forces Q and Q. the lines of action of at A and Q at
meet
in C,
and carry
four
Then the
* It might seem that there are two cases to be considered, for cannot the vectors that represent the forces of the first couple be opposite in sense? True. But then we can begin with the second couple. Its forces will be represented by the Q and Q of the diagram and the forces of the first couple, reversed in sense, will now appear as and P.
;
31
forces obviously are in equilibrium, for all four are equal in magnitude, since by hypothesis the moments of the two given couples
are equal
line
in such
and the forces make equal angles with the indefinite a manner that the resultant of one pair is equal
of the other pair, the lines of action of these
resultants coinciding.
To sum
is
the
and no matter how large or small the forces may be, provided only that the moment of the couple is preserved both in magnitude and in sense.
same, no matter whore
forces act,
Composition of Couples. From the foregoing it appears that two couples in a plane can be compounded into a single couple, whose moment is the sum of the moments
of the constituent couples
;
all
mo-
ments being taken algebraically. For, assume a system of Cartesian axes in the plane, and mark the
point
(1,0).
The
first
couple
A:(i,o)
can be realized by a force P 1 at A parallel to the axis of y (and either positive or negative) and an equal
and
opposite
force
Pi at
the
along the axis of y. The moment of this couple, taken algebraically, is obviously P,. Dealing with the second couple in a similar manner, we now
origin, acting
have as the result two forces, P 1 and P2 at A parallel to the axis and two equal and opposite forces at along the axis of y', These forces constitute a resultant couple, whose moment is y.
,
of
the
sum
of the
moments
is
This
last
statement
It
may happen
that the second couple is equal and opposite to the first, and then In order that this case may not the resultant forces both vanish.
nil couple
lie
cause an exception, we extend the notion of couple to include a i.e. a couple whose forces are both zero, or whose forces
:
in the
We
same straight line and we define are thus led to the following theorem.
;
its
moment
to be 0.
THEOREM.
to
effect is
is the
sum
of the
moments
32
MECHANICS
Remark.
of a couple is equal to the sum of the about an arbitrary point of the plane. This is seen directly geometrically from the definition of a moment. In be chosen let a at The can point particular, pleasure. couple be realized by two forces, one of which passes through 0. The moment of the couple is then equal to the moment of the other force about 0.
The moment
moments
of its forces
Let any Equilibrium. are equivalent f) to a single force, or ii) to a single couple or, finally, Hi) they are in equilibrium. Let be an arbitrary point of the plane. Beginning
6.
Then they
;
F D let us introduce at two and opposite to Fj. The two forces checked form a couple, and the remaining
is
force
x,
transferred to
the point 0.
'
,
Proceeding in this manner with each of the F n we arrive at remaining forces, F2 a new system of forces and couples equiv,
and consisting of those n These n forces are equivacouples. lent to a single force, R, at 0; or are in equilibrium. And the n couples are equivalent to a single couple, or are in equilibrium.
alent to the original system of forces
forces, all acting at 0, plus
In general, the resultant force, R, will not vanish, nor will the The latter can, in particular, be resultant couple disappear. and acting at 0, and a realized as a force equal and opposite to
second force equal to R, but having a different line of action. Thus the resultant of all n forces is here a single force. Incidentally we have shown that a non-vanishing couple can not be
balanced by a non- vanishing force for, the effect of such a force and such a couple is a force equal to the given force, but transferred to a new line of action, parallel to the old line.
;
It
may happen
ant couple does not. For equilibrium, it is necessary and sufficient that both the resultant force and the resultant couple vanish.
This condition can, with the help of the 5, be expressed in the following form.
Remark
at the close of
librium
will be in equi-
33
they are such as would keep a particle at rest if they all acted at
a point; and
ii)
the
sum
of the
it
moments of
is
enoughj and
may
the forces about a point (one point be chosen anywhere) of the plane is zero.
Let Analytically, the condition can be formulated as follows. as be chosen are to be which moments about the point taken,
the origin, and let the force
F
0,
Then
X
r=l
2) (XrYr- yrXr)
Example.
friction for
is
0.
A ladder rests against a wall, the coefficient of both ladder and wall being the same, M- If the ladder
just on the point of slipping when inclined at an angle of 60 with the horizontal, what is the
value of
Since
ju
all
the friction
is
R and S being
of the verti-
Condition
cal
?)
tells
us that the
sum
components upward must equal the sum of the vertical components downward, or
R+ S=
Furthermore, the
W.
FIG. 40
sum sum of
of the horizontal
or
S =
nR.
Finally, the moments about a point, 0, of the plane must It is convenient to choose as balance. a point through which a
number
of
unknown
2a cos 60
forces pass;
for example,
Thus
2a sin 60
R =
R+
a cos 60
TF,
or
R =V3/
Hence
W
2(1-
S
2
(1
M V3)
34
MECHANICS
We
can now eliminate
and
S.
The
resulting equation
is
Thus
/i
2-V3
0.27.
The other
EXERCISES
1.
If in
is
floor is rough,
2.
and
if /*
=
,
If in the
all
that
3.
show
ladder 12 ft. long and weighing 30 Ibs. rests at an angle with the horizontal against a smooth wall, the floor being = A man weighing 160 Ibs. goes up the ladder. How rough, /u far will he get before the ladder slips?
of
60
4.
the
man
6.
is
Show
that the
sum
that a necessary and sufficient condition for equilibrium of the moments about each of three points, A, B,
in a line, shall vanish for each point separately.
Couples in Space.
THEOREM
I.
couple
may be transferred
its
its effect }
moments
are equal
two couples in parallel planes, whose and opposite, and to show that their forces are
in equilibrium.
Construct a cube
faces in the planes
with two of
its
Then one couple can of. the couples. be represented by the forces marked
P
A
at
and
p+Q
having the same sense, and passing the of the cube. Turn next to P at C and through centre, 0, Q at D. The resultant of these forces is obviously equal and
35
opposite to the resultant just considered and having the same The four forces are, then, in equilibrium. This line of action. the completes proof.
Example.
in the engine
In a certain type of auto (Buick 45-6-23) the last bolt head was so near the cowl that a flat wrench could
not be used. The garage man immediately bent a flat wrench through a right angle, applied one end of the wrench to the nut and, passing a screw driver through
the opening in the other end, turned the nut. Thus the applied couple was transferred from the horizontal
IQ
riut.
'
Vector Representation of Couples. couple can be represented a as vector follows. Construct a vector by perpendicular to the of the of to and the moment of the plane couple length equal
couple.
permissible. plane of the couple and looking down on the plane. If we are on the proper side of the plane, we shall see the couple tending to produce rotation in the clock-wise sense. And now the direction
of the vector, either convention is Let us think of ourselves as standing upright on the
from our
feet to
the vector
be taken as the positive sense of the or, equally well, opposite direction.
our head
may
II. The combined effect of two couples is the same as a single couple represented by the vector obtained by adding geometrically the two vectors which represent respectively the given
THEOREM
that of
couples.
they
line seg-
\R
ment
of
unit
length,
AB,
QX^/
FlQ 43
the
line
AB.
Then
it
is
easily seen that the resultant of the two forces at ant of the two forces at B form a new couple.
A and
the result-
Finally, the vector representations of these three couples are three vectors perpendicular respectively to the three planes of the couples, equal in length to the forces of the couples, and so
36
oriented as to give the
MECHANICS
same
figure yielded
by
forces act
Resultant of Forces in Space. Equilibrium. Let any n on a body in space. Let them be represented by the
,
.
.
F n Let be an arbitrary point of space. Introvectors F,, two forces that are equal and opposite to the force Fk duce at
Then the n
(1)
forces
lf
Fn
F!
at
have a resultant
R =
Fn
acting at 0, or are in equilibrium. And the remaining forces, C n whose combined suitably in pairs, yield n couples, C lt ,
,
resultant couple, C,
(2)
or, in particular,
is
C =
;
C,
Cn
vanishes the couples being then in equilibrium. Thus the n given forces In general, neither R nor C will vanish. reduce to a force and a couple. The plane of the resultant couple, C, will in general be oblique to the line of action of the resultant
force,
Let
C =
where C^
is
C,
C2
is
collinear with R,
and C 2
;
perpendicular to R.
The
can be realized by two forces in a plane couple represented and its forces can be combined the resultant containing force, R
with R, thus yielding a single force R, whose line of action, howThis leaves only the couple Cj. We ever, has been displaced.
have, therefore, obtained the following theorem.
by C 2
THEOREM.
lent to
Any
is
in general equiva-
and to single couple, whose plane is perpendicular to the line of action of the resultant force. In particular, the resultant force may vanish, or the resultant couple
a a
single force
whose
line of action is
uniquely determined,
may
vanish, or both
may
vanish.
Equilibrium.
The given
if
and only
vanish.
if
its line
necessary to show that the resultant of action, and the resultant couple,
t,
For
it is
STATICS OF A RIGID
BODY
37
ent choice, O', of the point O might have led to a different result. Now, the vector R is uniquely determined by (1), and so is the
but C depends on the choice of 0', and so C l be different from CJ, though each would be might conceivably This is, however, not the case. For, reverse R collinear with R.
same
in each case
in
force
the second case, and also the couple C[. Then the reversed and couple must balance the first force and couple. But
perceive.
9.
Moment
//,
of a Vector.
Couples.
is
Given a
along a line
of
in space.
By
with respect
*
about)
product
(1)
/\F
r
F,
where r is a vector drawn from to a point of L. and It is a vector at right angles to the plane of L, and its length is numerically equal to the FIG. 44 in that plane. Its sense moment of F about depends on whether we are using a right-handed or a left-handed
system.
(2)
X
(3)
r
r'
=
i
r'
y
j
rj,
+ z k,
FIG. 45
3 of Appendix A.
is all
definitions that
of
This, together with the mere Vector Analysis which he will need for
the present.
38
MECHANICS
Vector Representation of a Couple.
forces,
F! and F 2
F,
+F =
2
The vector C acting respectively along two lines L and L2 which represents the couple is seen from the definition of the
vector product to be
(4)
:
C =
XP
lf
where r represents any vector drawn from a point of L 2 to a point Let be any point of space. Then the sum of the vector of Lj. moments of F x and F2 with respect to yields the vector couple
:
(5)
C =
r,
P!
r2
XF
2,
to
L and L 2
1
respec-
F!
T!
F!
r2
Fj
T!
Fj
r2
XF
10. Vector Representation of Resultant Force and Couple. F n be any system of Resultant Axis. Wrench. Let Pi, Let P be any point of space, and let equal and forces in space.
,
FA
-
, -
k
,
1,
Fn which
-
R =
F!
Fn
The remaining
PA', (%k, Vk, Zk)
forces yield n couples, consisting each of F* at and F& at P: (x, y, z). Let r k r be the vectors drawn from the origin of coordinates (chosen arbitrarily) to Pk and P respectively. Then the Jfc-th couple, C*, is represented
, :
by the equation
C*
r*
F*
F*.
We
THEOREM.
force, (6)
lf
Fn
R =
P; and
to
F,
+
X
+
-
Fn
acting at
a couple,
(7)
C = 5)
fc-i
r*
F*
R.
39
The
resultant axis
;
is
which
lies
along R
i.e. is
collinear with R.
RXC =
0,
R*
0.
Let
(9)
iXF^Li
R =
L =
Fj
(10)
+ Zk.
Thus
(11)
Then
or:
by
virtue of (7)
(r
RX
(Li
+ M + tf k) = R X
j
R),
(12)
This is the equation of the resultant axis in vector form. reduce to ordinary Cartesian form, equate the coefficients of
To
i,
j,
respectively.
Thus we
find
(13)
- Z (xX
One
of these equations
Y + zZ = YN - ZM - XN zZ) = ZL - YL zZ) = XM
illusory through the vanishbut some two always define intersecting and planes, for the rank of the determinant is 2, since R > between the three equations there exists an identical relation. Let ( ij, f ) be the coordinates of the nearest point of the line to
may become
ing of
all
the coefficients
the origin.
Then
Hence
YN
(15) $
-ZM
>?
ZL-XN
ZMft
2
YL
40
MECHANICS
is
that of R.
and the
is
given
by
(7),
where
r
(16)
Tjj
fk.
wrench is defined as two forces, acting at arbitrary Wrench. points; moreover, neither force shall vanish, and their lines of action shall be skew.
Let the forces be F*, acting at (xk, y^ Zk), k = 1, 2. The reader will do well to compute the resultant force, axis, and couple. Suppose, in particular, that F x is a unit force along the positive
axis of Z,
and
F2
is
a force of
2, parallel to
the axis of y
and acting
at the point
(1, 0, 0).
EXERCISE
Let F! and F2 be two forces, the sum a point is 0. Show that F t F2 and
, ,
of
lie
11.
Moment
vector, F, be given.
points of
and a be the 0, and L' nearest together. Let r be the vector from Let a be to 0', and let r = h.
of
Let a
line,
L,
line of F,
and
let
z
r
-
Assume Then
co-
L^r
By
the moment of
about
a,
L is meant
FlG 46
M M
=
r
= hY
hi,
we have
X F =k
(r
Hence
(1)
F)
hY.
M
may
= {a-(rXF))a.
L
We
be any vector drawn from a point of have thus arrived at the following result.
41
:
about a line
is given
by the formula
M
where r
of F,
attributed to L.
{a-
(r
F)}a,
line
is any vector drawn from a point of L to a point of the and a is a unit vector collinear with L and having the
sign
F about
xZ,
yZ -
zY,
zX -
xY - yX.
EXERCISE
force of 12 kgs. acts at the point ( 2), and its direc1, 3, about the moment tion cosines arc its Find 4, 12). line of i.e. the the unit through the cube; principal diagonal
(3,
origin,
positive axes.
12.
In
Fn
in
terms of the vanishing of the resultant force and the resultant By means of Equations (6) and (7) of 10 we can formucouple. The first, namely, R = 0, late these conditions analytically.
:
gives
and now the second, namely, C = 0, reduces Equation vanishing of the first term on the right, or
(2)
(7) to
the
(y k
Zk -
zk
Yk ) =0,
(z k
X k
xk
Zk ) =
0,
This last condition, which was obtained from the vanishing of a couple, admits two further interpretations in terms of the
vanishing of vector moments, namely with i) The sum of the vector moments of the given forces
:
of space is 0.
moments
about
an arbitrary
line of
space
is 0.
:
The
f
condition
ii) is
ii )
The sum
of the vector
42
MECHANICS
It
is
ness to analyse these conditions further, as to whether they are necessary or sufficient or both.
Condition
is
if
it
i), regarded as a necessary condition, is broadest when taken as any point of space. But Condition ii) is sufficient holds for the lines through just one particular point 0, the
fulfilled.
Condition
ii),
when the
condition,
(1)
line is
regarded as a necessary condition, is broadest taken as any line in space. But as a sufficient
though true as formulated, it is less general than and Condition ii'), which may, therefore, be taken as the
EXERCISES
1.
Show Show
i) is sufficient
for equilibrium.
for equilibrium.
2.
ii) is sufficient
Centre of Gravity of n Particles. Let the n particles w n be acted on by gravity. Thus n parallel forces arise, ^i, and since they have the same sense, they have a resultant not 0. Let the axis of z be vertical and directed downward. Then
13.
,
the resultant
(1)
is
R =
Thus,
g
Wj g
+ mn g =
5)
k,
Furthermore,
n
Xk =
0,
Yk =
0.
0,
Zk = m k g.
(2)
10, (11):
L =
Vm
t-l
k yk,
M =- gVm
*-i
:
xk
The
nates given
(3)
If
nearest point of the resultant axis to the origin has the coordiby (15), 10
*=5-,
of this line
rigidly
is
.,_**>
.-f-0.
any point
connected) will be supported, and the system will remain at rest. In particular, one point on this line has the coordinates
(thought of as
43
If, secondly, we allow gravity to act parallel to the axis of x, the resultant axis now becomes parallel to that axis, and the nearest
found by advancing the letters cyclically Equations (3). Again the point whose coordinates are given by (4) lies on this axis. And, similarly, when gravity acts parallel to the axis of y. It seems plausible, then, that if this point be the supported, system will be at rest, no matter in what direction This acts. To prove the statement, gravity is, in fact, the case. let the point P of 10 be taken as (x, y, 0). Then C = 0. For
point to the origin
in
is
i
k
2
ij
mk a mk p mk y
since the coefficient of each of the unit vectors
i,
S nik y
j,
is
seen at
once to vanish, no matter what values a, 0, 7 may have. Thus the existence of a centre of gravity for n particles is established. It is a point such that, no matter how the system
is nil.
If three non-vanishing forces, acting on 14. Three Forces. a rigid body, are in equilibrium, they lie in a plane and either pass through a point or are parallel.
If two forces in space are in equilibrium, they must be or else each and opposite, and have the same line of action equal must vanish. Exclude the latter case as trivial. Take vector
Proof.
moments about an arbitrary point, O, in the line of action of one Then the vector moment of the other force must of the forces. 12. vanish by Thus the second force either vanishes or passes
through every point of the line of action of the Finally, they must be equal arid opposite. In the case of three forces, no one of which vanishes, and no two of which have the same line of action, take vector moments in the line of action of the first force, but of no about a point
through
;
i.e.
first force.
other force. The sum of the second and about must be zero. Hence the second a plane through 0. They are, therefore, force a couple is impossible, since it could
first force.
third vector
moments
lie
and third
forces
in
Thus the
first
and third
proved.
44
MECHANICS
A
Trigonometric Theorem.
The
is
useful in
many problems
Let a
line
of the equilibrium of a
by
three forces.
dividing the opposite side into two segments of lengths and making angles 6 and <p with these sides. Then
m and
<p,
n,
(m
where
\l/
+
is
n) cot
if/
m cot 6
cot
segment
n.
on this
\(/
line,
produced
6 cos
:
(m
n) cos
a cos
of sines
b
<p t
m
sin
i
sin0'
sn
sn
<p
uniform rod of length 2a is held by a string end of the rod arid to a peg in a smooth vertical wall, the other end of the rod Find all the positions resting against the wall.
Example
1.
W,
T,
and
must pass
through a point, and this must be the mid-point of the string. Hence, applying the above trigonometric theorem to either of the triangles ABO or ABC, we have
:
(1)
2 tan 6
tan
<p.
A
(2)
second relation
is
ations,
namely
*
:
cos
2a cos
<p.
It
Squaring
sec 2
2
<p,
(1)
and reducing,
we have
or:
4 sec 2 6
cos 2
4 cos
<p 2 <p
1+3 cos
=
a sin
<p.
sin
But the
45
we
1
get
2
4Z 2 cos
<p <p
3 cos 2
it
4a 2 cos 2
Since cos
<p
cannot vanish,
cos
follows that
<p
But a and
<
cos
<
1.
Hence
and so
or, the string
<
<
2a,
must be longer than the rod, but not twice as long. Furthermore, there are always two positions of equilibrium, in which the rod is vertical, regardless of I and a.
Remark.
What
the trigonometric theorem has done for us is Without it, we should have been obliged
to write down two or three equations involving T and R, and then eliminate these unknowns, with which we have no concern so far as the position of equilibrium goes.
Example
2.
Suppose
that,
in
is
an upward force of
friction,
is
just on
angle \ with the normal to the wall, and so the problem is reduced to a three-force problem. Applying the trigonometric
force
S making an
ABC we
6
find
=
<p
a cot
a tan
\,
2 cot
cot
/z.
form
sin 8
a sin
(p.
From
Hence
(3)
we now have
esc 2
=
-
4 cot 2
<p
+ 4/i cot + M +
2
<p
1.
sin 2
A 4 cot 2
-+ - -+
72
:
99
4/i
cot
<p
= r-rr 2 +1
sin 2
46
This
last
MECHANICS
equation can be given the form
:
4 cos 2
<p
+ 4/i cos
<p
sin
<p
(1
2
/j
sin 2
<p
This equation, in turn, could be reduced to a quartic in sin <p or cos <p but such procedure would be bad technique. Rather,
;
let
2 cos 2
<p
+ cos
2 sin
2y?,
2
2 cos
1
<p
sin
v?
sin
2<p,
p =
cos
2<p.
:
The equation
is
cos 2<p
+ B sin 2^
of
C,
I,
Chapter
6.
EXERCISE
1.
of
Example
is
and
finding
(K2 J
ii)
2.
2
;
If
If,
p,
EXERCISES ON CHAPTER
1.
II*
Show
that, in a tackle
and
fall
2n
times
What
force
hub of a wheel (at rest) will just cause the wheel to surmount an obstacle of height h?
Two heavy beads of the same weight can on a rough horizontal rod. To the bead is attached a string that carries a smooth heavy bead. How far apart can the beads on the rod be placed if they are to remain at rest when
3.
slide
FIG. 50
4.
released ?
is
gate
raised
on
its
fall
back.
How
showing the
forces,
STATICS OF A RIGID
5.
BODY
47
There has been a light fall of snow on the gate. A cat weighing 5 Ibs. walks along the top of the gate, and the gate
The disconcerted cat springs off. It is observed from drops. her tracks in the snow that she reached a point 2 ft. from the
end of the gate. The distance between the hinges is 2^- ft., and the centre of gravity of the gate is 5 ft. from the vertical
line
If
Ibs.,
what
is
the value of
6.
A rod rests in a smooth hemispherical bowl, one end inside the bowl and the rim of the bowl in contact with the rod. Find
the position of equilibrium.
Ans.
cos 6
-f V32r
+a
2
,
or
where the radius of the bowl is r, the distance of the centre of gravity of the rod from its lower end is a, and the inclination of the rod to the horizon,
7.
0,
provided a
<
2r.
A
all
floor
and the
Find
8.
9.
equilibrium
10.
Formulate the problem of picture hangs on a wall. when the wall is smooth, and solve it.
wall
is
rough.
11. A smooth rod rests with one end against a vertical wall, a peg distant h from the wall supporting the rod. Find the
position of equilibrium.
Ans.
wall
is
cos
\ * a
[ft
12.
smooth.
13.
positions of equilibrium.
A
its
lying
on
its side.
with
floor.
equilibrium.
14.
rough.)
long is stood up against a carpenter's work8 in. high. The coefficient of friction between If the plank either the floor or the bench and the plank is makes an angle of 15 with the vertical, will it slip down when
ft.
let
go?
48
15.
MECHANICS
A
smooth uniform rod
is
rests
in
a test-tube.
Find the
2 tan
16.
cot
^>,
2a cos
0.
table.
To
uniform rod 2 ft. long rests with one end on a rough the other end of the rod is attached a string 1 ft. long,
made
fast to a
peg 2
ft.
above the
table.
Find
all
positions of
equilibrium.
Ans.
One system
:
ju
by the equations
cot
<p
+ 2 cot
0,
2 cos 6
cos
<p
2.
Method
of Successive
Approximations.
100 ft. high and 100 ft. in diameter. in the plates near the base. the tension approximately
17.
A water tower is
Find
18. Water is gradually poured into a tumbler. Show that the centre of gravity of the glass and the water is lowest when it is in the surface of the water.
one attempts to pull out a two-handled drawer by one handle, what is the condition that the drawer will stick fast ?
19.
If
CHAPTER
III
MOTION OF A PARTICLE
Rectilinear Motion.* Tube simplest case of motion of matunder the action of force isHhat in which a rigid body moves without rotation, each point of the body describing a right line, and the forces that act being resolved along that line. Consider, for example, a train of cars, and neglect the rotation of the wheels and axles. The train is moved by the draw-bar pull of the locomotive, and the motion is resisted by the friction of the tracks
1.
ter
pressure. Obviously, it is only the components of the forces parallel to the tracks that count, and the problem of Dynamics, or Kinetics, as it is more specifically called, is to
determine the relation between the forces and the motion; or, if one will: Given the forces, to find the distance traversed as a function of the time.
A more conventional example, coming nearer to possible experimentation in the laboratory, would be that of a block of iron
obliged to finish this chapter before going on. What the treatment of the early problems by the present methods, for these are the general methods of Mechanics, to inculcate which is a prime object of this book. Elementary text-books in Physics sometimes write down three equations :
feel
is
needed
a thorough
drill in
^at
2
,
at,
t?
2as,
and give an unconscionable number of problems to be solved by this device. The pedagogy of this procedure is totally wrong, since it replaces ideas by a rule of thumb, and even this rule is badly chosen, since it disguises, instead of revealing, the mechanical intuition. Now, a feeling for Mechanics is the great object to be obtained, and the habits of thought which promote such intuition are, fortunately, cultivated by just the same mathematical treatment which applies in the more advanced parts of Mechanics. It is a happy circumstance that here there is no conflict, but the closest union, between the physics of the subject and the
1-12 through working each problem by the present general 22 should here be included with, of Moreover, 16 and the statement of course, the definition of vector acceleration given in Newton's Second Law in 17. The student should then turn to Chapter IV, the most revealing chapter in the whole elementary part of the book, and study it ic all detail. The remaining sections of the present chapter should be read casually at an early stage, so as not to impede progress. Ultimately, they are important but they are most useful when the student comes to recognize their importance through his experience gathered from the later work above referred to.
methods
most important.
49
50
MECHANICS
placed on a table and drawn along by cords, so applied that the block does not rotate and that each point of it describes a right
line
It is clear that
i.e.
with varying velocity. a block of platinum having the same mass, the same amount of matter, if acted on by the containing
forces,
would move just like the block of iron, if the two were by side from rest or with the same initial velocities. We can conceive physical substances of still greater density, and the same would be true. On compressing the given amount of matter into smaller and ever smaller volume, we are led to the idea of a particle, or material point, i.e. a geometrical point, to which the property of mass is attached. This conception has the advantage that such a particle would move exactly as the actual body does if acted on by the same forces; but we need say nothing about rotation, since this idea does not enter when we consider only particles. Moreover, there is no doubt about where the forces are applied they must be applied at the one
same
started side
Sir Isaac
Newton
(1642-1727),
FIRST LAW. A body at rest remains moves in a straight line with unchanging
force acts
at rest
velocity, unless
on
it.
SECOND LAW.
is
The
proportional
to the resultant
THIRD LAW.
The meaning
moving
in
1.*
Law
is
clear enough,
if
we
restrict
and
Contoo, is self-explanatory. two particles of unequal mass, connected by a spring, the mass of which is negligible. Then the pull (or push) of the spring on the one particle is equal and opposite to its pull (or push) on the other particle. The Second Law is expressed in terms of momentum, and the momentum of a particle is defined as the product of its mass by
sider, for example,
might consider, furthermore, such material distributions as laminae, i.e. material surfaces; and also wires, or material curves. Finally, rigid combinations of all these bodies.
*We
MOTION OF A PARTICLE
its velocity,
51
or mv. Here, v is not an essentially positive quanmere We must think of the position of the the speed. tity a The latter may as described suitable by coordinate, s. body be the distance actually traversed by the particle; or we may
x,
and
s as the
The
velocity,
v,
will
then be defined
and
is positive when s is increasing negative, when The Second Law can now be stated in the form
; :
s is decreasing.
(2)
Here, / denotes the resultant force, and is positive when to increase s negative, when it tends to decrease s.
;
it
tends
is
constant
and so
d (mv)
~dT
The quantity dv/dt represented by a
:
_ ~
m dv
'dt
is
is
often
It is positive
when
v is increasing, negative
when
in the
:
v is decreasing.
form
The mass
force
ma oc
the proportion
/.
:
From
(5)
we now
pass to an equation
X/,
ma =
is
where X
a physical constant.
If these are
The value
of X
depends on the
the English units, the pound being the unit of mass, the foot the unit of length, the second the unit of time, and the pound the gravitational unit of force, then X has the
units used.
value 32
32.2),
of
A,)
52
MECHANICS
In the decimal system, the gramme being the unit of mass, the and the
gramme the
Second
Law
of
981,
and Newton's
2)
mf = 981
t
/.
In 3 we shall discuss the absolute units. In particular, the units of mass, length, and time having been chosen arbitrarily, " is that unit as in Physics, the so-called "absolute unit of force
which makes X
A
\
1 in
A,)
Three Forms for the Acceleration. The acceleration as dv/dt, and since v = ds/dt, we have
:
is
defined
/!\ (6)
a
form
is
x/2 * U
^
:
A
(
third
(7\
'
_ dsdv ~
di ds
value,
v.
Thus
. - .*.
:
dz s
dv
form
it is
Which s, t, v in pairs in all possible ways. better to use in a given case, will become clear from
practice in solving problems. freight train weighing 200 tons is drawn by Example 1. a locomotive that exerts a draw-
PP-I
jt
/==8000
resistances.
How much
if it
from rest?
MOTION OF A PARTICLE
Here we have
63
m/ = 9
400,000^
or
dt
= 32X8000,
=
25
t,
we
find
Tjngt ~f~
Since v
when
0,
we must have C
0,
and hence
At the end
of a minute,
^
60,
and so
38 4
-
60
ft
P er
sec.
it is
To
convenient
equivalent to 44 ft. a second, as the student can readily verify; or roughly, 2 miles an hour corresponds to 3 ft. a second. Hence the speed in the present
is
an hour
about two-thirds of 38.4, or 26 miles an hour. Example 2. A stone is sent gliding over the ice with an initial If the coefficient of friction between the velocity of 30 ft. a sec.
case
is
is
-fa,
how
we take account
of is the retarding
force of friction, and this amounts to one-tenth of a for every pound of mass there is
in the stone.
pound
of force
m
\
10
Hence,
if
there are
4
s
*
-q
pounds of mass
it
in the stone
lbs.,f
s,
^m
and
it is
FlG 52
tends to decrease
to be taken as negative:
* The student must distinguish carefully between the two meanings of the word pound, namely (a) a mass, and (6) a force two totally different physical objects. Thus a pound of lead is a certain quantity of matter. If it is hung up by a string, the tension in the string is a pound of force. t The student should notice that is neither a mass nor a force, but a number, like all the other letters of Algebra, the Calculus, and Physics.
54
MECHANICS
Now what we want is a relation between v and s, for the quesHow far (s = ?), when the stone stops (v = 0)? So we
:
tion is:
dv
V
16
ds
or
rr Hence
" v dv
2
~-5>
/-ds.
=
16
^-
T>
n + C.
.
To
determine
C we
0,
30.
must hold
^!
and so
=
v
2
+ C,
=
900
0,
C =
450,
- ^s.
s
When
and we have
900
- 3s,
EXERCISES
141
ft.
1. If the train of Example 1 was moving at the rate of 4 m. an hour when we began to take notice, how fast would it be moving half a minute later? Give a complete solution, beginning with Ans. About 17 m. an h. drawing the figure.
2.
He
reaches
small boy sees a slide on the ice ahead, and runs for it. it with a speed of 8 miles an hour and slides 15 feet.
How
3.
Ans.
.15.
the coefficient of friction between a sprinter's that, shoes and the track is TV> n ^ ^ cs ^ possible record in a hundredif
Show
less
than
5 seconds.
4. An electric car weighing 12 tons gets up a speed of 15 miles an hour in 10 seconds. Find the average force that acts on it,
* It is important that the student should work these exercises set forth in th4 e text, beginning each time by drawing a figure and
force,
by means of a directed right and (t'i) the coordinate used, as s or x, of Elementary Physics as
v
line,
or vector,
drawn preferably
etc.
at,
= a2
vz
2as
to present purposes. For, tho object of these simple exercises is to prepare the way for applications in which the force ia not constant, and here the formulas just cited do not hold.
MOTION OF A PARTICLE
i.e.
55
the
same
velocity in
6. In the preceding problem, assume that the given speed is acquired after running 200 feet. Find the time required and the average force.
6. A train weighing 500 tons and running at the rate of 30 miles an hour is brought to rest by the brakes after running 600 feet. While it is being stopped it passes over a bridge. Find the force Ans. 25.2 tons. with which the bridge pulls on its anchorage.
7.
An
and
electric
it
car
is
starting
on an icy track.
The wheels
skid
takes the car 15 seconds to get up a speed of two miles Compute the coefficient of friction between the wheels
3. Absolute Units of Force. The units in terms of which we measure mass, space, time, and force arc arbitrary, as was pointed 2. If we change one of them, we thereby change the out in value of X in Newton's Second Law. Consequently, by changing the unit of force properly, the units of mass, space, and time being held fast, we can make X = 1. Hence the definition above given: DEFINITION. The absolute unit of force _is that unit which makes X = 1 in Newton's_ Second Law of Motion ij*
(1)
moT^f.
In order to determine experimentally the absolute unit of force, we may allow a body to fall freely and observe how far it goes in a
known
time.
It is a physical
is
gravity attracts
any body proportional to the mass of that body. Let the number g be the number of absolute units of force with
* have already met a precisely similar question twice in the Calculus. differentiating the function sin x we obtain the formula x sin x = cos x
We
In
only
when we measure
if
angles in radians.
Dx sin x
In particular,
as follows
:
the unit is a degree, X = Tr/180. We may, therefore, define a radian absolute unit of angle (the radian) is that unit which makes X = 1 in the above equation. Again, in differentiating the logarithm, we found
The
X
This multiplier reduces to unity when we take a = e. Hence the definition The absolute (natural) base of logarithms is that base which makes the multiplier logo e in the above equation equal to unity.
:
56
MECHANICS
which gravity attracts the unit of mass. Then the force, measured which gravity attracts a body of units of mass will be mg. Newton's Second Law A 3 ) gives for this case
in absolute units, with
dv
dv
ds
%gt*
+ K,
0,
and we have the law for freely falling bodies deduced directly from Newton's Second Law of Motion, the hypothesis being
merely that the force of gravity is constant. Substituting in the last equation the observed values s = S, t = T, we get
:
9
If
= 28
?*'
we
two
If
is equal nearly to half an ounce. from 978 to 983 at different parts c.g.s. units, g ranges in has value 980. The absolute and the of the earth, Cambridge in this system is called the dyne. unit of force Since g is equal to the acceleration with which a body falls freely under the attraction of gravity, g is called the acceleration But this is not our definition of g it is a theorem of gravity. about g that follows from Newton's Second Law of Motion. The student can now readily prove the following theorem, which is often taken as the definition of the absolute unit of
we use
The
is
that
on the unit of mass for the unit of time, generof the equations for
Incidentally
falling
freely
body
is
gt,
%gt
The
third
found by setting
a =
dv
2gs.
MOTION OF A PARTICLE
57
Example. A body is projected down a rough inclined plane with an initial velocity of V Q feet per second. Determine the
absolute units, down the plane, and the force of friction, nmg cos 7 up the plane. Hence
mg sin 7 pR =
ma = mg sin 7 dv = g sm 7 -IT
get
v
nmg
cos 7,
cos 7.
we
C,
=
o
g (sin
/z
cos 7)
+
<?>
+
=
g (sin
cos 7)
+v
Q t,
A
B)
\g
(sin
^ cos 7) P
+V
between A) and
for the acceler-
Or we can
:
(8),
2,
ation
v-r
do
g (sin 7
/x
cos 7),
%v
2
t;
g (sin 7
M cos 7)
^
=
1.
=
/z
20 (sin 7
s + K, + X, cos 7) s +
#o-
EXERCISES
If,
in the
text, the
body
is
pro-
jected
2.
up
how
it
far
it will
go up.
1
to
jected vertically
v
2
v
8
= = =
+ vl gt + V ot* + v
2gs
Q
or or
t
= =-
or
+ vl +V =-tf* + M.
2gs
;
gt
58
4.
MECHANICS
On
much
as
moon a pound weighs only one-sixth on the surface of the earth. If a mouse can jump foot on the surface of the earth, how high could she jump
the surface of the
as
Compare
is
in the air
table.
block of iron weighing 100 pounds rests on a smooth cord, attached to the iron, runs over a smooth pulley
and
which
hangs vertically. from the pulley. How long will it be before the iron reaches the pulley, and how fast will it be moving?
Ans.
6.
The system
2.19 sec.
9.1
ft.
sec.
Solve the same problem on the assumption that the table is rough, n = ^, and that the pulley exerts a constant retarding force of 4 ounces.
Regarding the big locomotive exhibited at the World's by the Baltimore and Ohio Railroad the Scientific American said "Previous to sending the engine to St. Louis, the engine was tested at Schenectady, where she took a 63-car train weighing 3150 tons up a one-per-cent. grade." Find how long it would take the engine to develop a speed of 15 m. per h. in the same train on the level, starting from rest, the draw-bar pull being assumed to be the same as on the grade.
7.
Fair in 1905
8.
If
Sir
Isaac
Newton
registered
rest,
balance in an elevator at
and
if,
moving, he weighed only 169 pounds, what inference would he draw about the motion of the elevator?
elevator that
man whose weight is 180 pounds weigh in an descending with an acceleration of 2 feet per second per second ?
9.
What
does a
is
4. Elastic Strings. When an elastic string is stretched by a moderate amount, the tension, T, in the string is proportional to the stretching, i.e. to the difference, s, between the stretched
of the string
T oc
fc
s,
or
T =
ks,
where
is
particular string
MOTION OF A PARTICLE
of 12 Ibs.
59
Suppose, for example, that a string is stretched 6 in. by a force If we measure the force in gravitar to determine k. ;
i.e.
tional units,
pounds, then
T =
=
12
when
(1),
we have:
&,
T =
or
24s.
24,
If
we had chosen
to
units,
i.e.
poundals, then, since it takes (nearly) 32 of these units to make a pound, the given force of 12 pounds would be expressed as
(nearly) 12 X 32, or precisely 120, poundals. Hence, substituting the present value of the force in (1), which, to avoid confusion, we will now write in the form
:
T
we have
(3)
:
k's,
120
k'
\
T'
or
k'
240,
240s.
1
in.,
When
as given
the string
is
stretched
2,
i.e.
by
(2)
is
T =
2 pounds.
expressed in terms of poundals, and this takes 64 half-ounces to make 2 pounds, and so
T'
right;
for
it
we should have
20.*
of strings stated
The law
form of
above
is
Hooke's Law:
rr\
__.
'
where
f
I is the natural, or unstretched, length of the string, and the stretched length; the coefficient being Young's Mod-
ulus.
For a given
string,
E/l
is
constant,
and V
s is
variable.
* It is easy to check an answer in any numerical case. The student has only force in pounds, or have I to ask himself the question: "Have I expressed expressed it in terms of half-ounces?" Just as five dollars is expressed by the number 5 when we use the dollar as the unit, but by the number 500 when we
my
use the cent, so, generally, the smaller the unit, the larger the a given quantity.
60
MECHANICS
EXERCISES
1. An clastic string is stretched 2 in. by a force of .3 Ibs. Find the tension (a) in pounds; (b) in poundals, when it is stretched s ft. Ans. (a) T = 18s ; (6) T = 180s.
2.
When
is
stretched 4
in.,
what
is
the tension (a) in terms of gravitational units; (b) in terms of absolute units? Ans. (a) 6 pounds; (6) 192 poundals.
3.
An
grs.
(6) in
when
it is
stretched
Ans.
(a)
100s;
(b) 98,000s.
One end
is
of
an
elastic string 3
is
at A,
gradually lowered till it is just supported by the string, and it is found that the length of the string has thus been doubled. Find the tension in the string when it is stretched s ft.
weight
Ans.
5.
fs
Ibs.
^s poundals.
A Problem
4
;
of Motion.
is
One end
in the text of
horizontal table
of the string
fastened to a peg at a point of a smooth a weight of 3 Ibs. is attached to the other end
rest
How
the string becomes slack? The weight evidently describes a straight line from the starting point, A, toward the peg 0, and we wish to know its velocity when it has reached a point B, one foot from A.
It is
-- -<
'
The
solution
is
Law
of Motion.
convenient here to take as the coordinate, not the distance that the particle has travelled at
AP
-j
T-,
riG. 54
KA
but its distance s from B. which acts is the tension of the string; measured in absolute
any
instant,
The
Since
it
force
i.i
is
units
it
is
240rs.
tends to decrease
:
s,
it
negative.
To
(2)
___
fJ2 a
tit)
-^ by
its
value v -=-
MOTION OF A PARTICLE
Hence
I
61
v dv
= =
8g
I
v dv
sds,
(3)
To determine
was
released at A,
=
4(7
and
I.
particle
=and
(4)
+
=
C,
C =
40,
(3)
becomes
2
t;
80
(1
).
have now determined the velocity of the particle at an arbitrary point of its path, and thus are in a position to find its velocity at the one point specified in the question proposed,
namely, at B.
v2
We
Here, s
0,
and
v\ s=sQ
Uo =
8g
32,
16
(ft.
per sec.)
EXERCISES
1.
The weight in the problem just discussed is projected from along the table in the direction of OB produced with a velocity Find how far it will go before it begins to return. of 8 ft. per sec.
Ans.
Newton's equation
32,
is
;
same
but
is
initially s
and
8.
Hence C =
2.
If,
6 inches.
in the
example worked
/i,
rough
and the
the
coefficient of friction,
how
fast will
3^=-24g8 + i-3g,
and the answer
3.
is
:
4 Vl5
15.49
ft.
per sec.
Solve the problem of Question 1, for a rough table, M = TAns. The required distance is the positive root of the
equation
16s 2
+s
0,
or
.4698
ft.,
or
about
5fin.
* In the following exercises and examples, it will be convenient to take the value of g as exactly 32 when English units are used. Begin each exercise by drawing a figure showing the coordinate used, and mark the forces in red ink.
62
MECHANICS
if
4. Find where the weight in Question 2 will come to rest the string, after becoming slack, does not get in the way. 6.
The 2
Ib.
at a point
directly
4,
.4
from
4, is released from rest and at a distance of 3 ft. but not stretched. Find
how
rise.
Ans.
ft.
6.
peg at A, find
Ans.
7.
in the last question, the weight is dropped from the how far it descends before it begins to rise.
To a
distance of 6
+ 3\/3 =
11.196
is
ft.
below A.
If the
ft. ft.
below
below
If,
it will rise
8.
ft.
9.
The
string of the
example studied
in the text of
is
placed on a smooth inclined plane making an angle of 30 with the horizon, and one end is made fast to a peg at A in the plane.
If
a weight of
Ibs.
A, find how far down the plane Assume the unstretched length of the string to be 4
rest at
if
10. 11.
the plane
is
rough,
/*
= ^V3.
buoy (specific gravity ^) is anchored submerged at high water. If the cable should break at high tide, show that the spar would jump entirely out
cylindrical spar
so that
it is
just
of the water.
of the water
is
particle of
mass 2
Ibs. lies
on a rough horizontal
table,
fastened to a post by an elastic band whose unstretched is 10 inches. The coefficient of friction is -, and the band length is doubled in length by hanging it vertically with the weight at
is
and
its
lower end.
If the particle
15 inches from the post and then projected directly away from the post with an initial velocity of 5 ft. a sec., find where it will
stop for good.
MOTION OF A PARTICLE
6. Continuation ; the Time. whose motion was studied in
63
found as follows.
From
v
equation
we have
s 2.
(5)
=
t
-^
V8g Vl
is
Since s decreases as
sign holds.
increases, ds/dt
Replacing
V8g by
its
value, 16,
(6)
by separating the so that only the variables, i.e. by transforming the equation variable s occurs on one side of the new equation, and only t on
is
readily solved
the other
(7)
thus
I6dt
==s
2
Hence
16*
= - f ds J Vl , ,
sinsi
+ C.
the
If
we measure
when
released at A, then
arid s
=
C,
body was
determine
=Thus
sin- 1
C = ~
sin- 1
s.
l&t
16t
cos" 1
s,
or
cos 16$.
we might
2
Vl -
We should
16*
=
and the
final result is
cos' 1
= cos- 1 8 - 0, * = 1 in 1 - C", or
C'.
this equation,
we
find
C"
0,
64
MECHANICS
This equation gives the time it takes the body to reach an In particular, the time from A to arbitrary point of its path. B is found by putting s =
:
(9)
W=
Show
that
cos- 1
=
|,
=
J2
.09818 sec.
EXERCISES
1.
if
the body, in the case just discussed, had been any other distance from the peg, the string
being stretched, the time to the point at which the string becomes slack would have been the same.
2.
Show
that
takes the body twice as long to cover the path as it does to cover the remainder.
it
first
Find the time required to cover the entire path in the case
of the following exercises at the close of
3.
4.
5.
Exercise Exercise
t
1.
Ans.
= ^ OA
.09818.
5.
Ans.
= \^K * 04
6.
J/ v/6s
-',
total time,
TrA/^ = .9618
sec.
5.
Exercise
Exercise Exercise
Ans.
sin- 1
77=
=?
6.
7.
7.
9.
Ans.
8.
.9618 sec.
Exercise 10.
9.
Exercise
8.
The simplest and most important 7. Simple Harmonic Motion. ease of oscillatory motion which occurs in nature is that known as Simple, Harmonic Motion. It is illustrated with the least
amount
first
of technical detail
Exercise below.
Example.
hole
is
is
stone
is
bored through the centre of the earth, a inserted, the air is exhausted, and the stone
as a
homogeneous
Its attraction, F, on sphere, at rest in space. the stone diminishes as the stone nears the centre,
FIG. 55
and
it
MOTION OF A PARTICLE
any point
centre:
of the hole,
65
from the
oc r,
or
fc,
F =
kr.
To determine the constant observe that, at the surface, = R (the radius of the earth), and, if we measure F in absolute = wgr, where m denotes the mass of the stone. Hence units, F
r
mg = kR
and
or
~^, /
F =
^r R
As the coordinate
us:
/i\ (1)
of the stone
mg r m d*r = -R
For,
when
r Ls positive,
r,
and so
is
negative.
braically,
When
and so
:
r is negative,
is
positive.
Hence
(1),
w
(2)
<*L=-L r
dp
>
we employ the
by 2dr/dt:
cydrd^r
dtdt*~
The
left-hand
side
=_
2g
dr
T
dt
R
*
thus
becomes
t
:
d
-77
/rfr\
(
at \dt/
-n
Hence each
side
form
a 4?
<&c cte 2
The
~
J
Hence
Integrating,
we have
66
MECHANICS
Cd
or
Jdt\dt
~ --^.Cr^ r
dt
~_
dt
--
R
0.
To determine
at Ay r
(7,
observe that
initially, i.e.
= R and
+ 0=~|ft2 7
Thus
<3 >
c,
C =
!# it
2
.
finally
(I)'- 1 <'-">
2
At the centre of the earth, r = 0, and (dr/dt) = gR. If we take the radius of the earth as 4000 miles, then R = 4000 X 5280, = 32, and the velocity is about 26,000 ft. a sec., or approxig mately 5 miles a second.
The
for
an instant when r = J?, i.e. it just reaches the other side Thus it oscillates throughout of the earth, and then falls back. the whole length of the hole, reaching the surface at the end of each excursion, and continuing this motion forever. The result no is not unreasonable, for there is no damping of any sort,
friction or air resistance.
The Time.
Equation
(3) it
To
find
the time
we proceed
as in
6.
From
follows that
'"
=-
\
9
VR* t
F====^.
r
dr
or
=
^\|
cctfh
I
~+
=
r
0,
C.
Initially,
=
=
*
and
72
thus
(7
and
(4)
\
<7
cos- 1 ~,
/t
or
= R cos
(^ \
VB) A//
'
MOTION OF A PARTICLE
The time from A
to
is
67
:
found by putting
* -'
On computing
and 16
sec.
the value of this expression it is seen to be 21 min. to B is twice the above. Hence
to
and back to
is
This time
inated
A\
'
is
known
is
always dom-
by
~_
~
'
~dfi
where the coordinate x characterizes the displacement from the This equation can be integrated as in the position of no force. special case above, and it is found that
B)
where h denotes the value of x which corresponds to the extreme is numerically nh, and
h.
second integration
C)
provided the time
is
h cos
ntj
h.
The
period, T,
is
inversely proportional to
27T
^
and the amplitude
the amplitude.
is
2h.
is
independent of
The motion represented by Equation C) is known as Simple Harmonic Motion. The graph of the function is obtained from
* In the first equation (4) the principal value of the anti-cosine holds during first passage of the stone from A to B. The second equation (4) holds without restriction.
the
68
MECHANICS
scale
the graph of the cosine curve by plotting the latter to one on the axis of t, and to another scale on the axis of x.
FIG. 56
EXERCISES
4 are strings like the one described in the text of A and on a one of smooth to two end B, each, pegs, fastened, horizontal table, the distance AB being double the length of either string, and the other end of each string is made fast to
1.
Two
a 3 Ib. weight, which is placed at 0, the mid-point of AB. Thus each string is taut, but not stretched. The weight being moved and A and then released from rest, show to a point C between
Find the velocity oscillates with simple harmonic motion. oscillation. It is and of the the which it with period passes assumed that the string which is slack in no wise interferes with
that
it
Ans.
is
The
d2 x
ffi at
differential
Ib.
weight
The
the
required velocity
maximum
the
4,
displacement.
The
period
is
27T/16
.3927 sec.
2.
Work
4,
Question
3.
one of
Show
7,
5, is
simple harmonic
4. If a straight hole were bored through the earth from Boston to London, a smooth tube containing a letter inserted, the air
exhausted from the tube, and the letter released at Boston, long would it take the letter to reach London?
how
MOTION OF A PARTICLE
6.
69
If in the
problem of Question
9,
5,
leased with the string taut, but not stretched, and directed straight down the plane, show that the weight would execute simple
harmonic motion.
6.
Work
8.
moon;
cf.
the data
in
7.
one square millimeter cross-section is hung of 25 kilogrammes is fastened to the lower end of the wire and carefully brought to The weight is then given a slight vertical displacement. rest. Determine the period of the oscillation. Given that the force required to double the length of the wire
steel wire of
up
in
Bunker
Hill
is
21,000 kilogrammes, and that the length of tho wire is 210 feet. Ans. A little over half a second.
8. A number of iron weights are attached to one end of a long round wooden spar, so that, when left to itself, the spar floats A ten-kilogramme weight having become vertically in water. accidentally detached, the spar is seen to oscillate with a period of 4 seconds. The radius of the spar is 10 centimetres. Find
the sum of the weights of the spar and attached what distance docs the spar oscillate ? Ans. (a) About 125 kilogrammes
8.
iron.
Through
(6)
0.64 metre.
of Gravitation.
Problem.
To
A
find the velocity which a stone acquires in falling to the earth from interstellar space.
to
be at
rest
exerts.
and
let r
the centre
of the earth at
Then
*-?
Since
El
F = mg when
mq =
X
-F^
JR,
and
r v
= mgR* ^
70
MECHANICS
Law
of
Newton's Second
d r _ ~ m ~dT*
mgR
~7*~'
Hence
To
we employ
the
method
of
7 and
multiply by 2 dr/dt
drd 2 r
dt dt 2
2gR
r2
dr
r
dt'
t
d/dr\* =
di\dt)
:
2gR*dr
r
2
dt
we
find
Initially dr/dt
and
hence
o.' +
Since dr/dt
is
c,
c~
:
is very great, the last term in the parenthesis is small, and no matter how great I is, V can never quite equal V2gR. Here g = 32, R = 4000 X 5280, and hence the velocity in question is about 36,000 feet, or 7 miles, a second. This solution neglects the retarding effect of the atmosphere; but as the atmosphere is very rare at a height of 50 miles from the earth's surface, the result is reliable down to a point com-
If
so,
it
fall,
con-
from
Hence
MOTION OF A PARTICLE
and
,
71
Vl r
rdr
169,
we
find
dr
Vlr
r2
=
Thus
Initially
t t
V - r + ~ sinlr
2
=
fji
and
r
Finally, then,
fr-H +
is
(3)
EXERCISES*
1. If the earth had no atmosphere, with what velocity would a stone have to be projected from the earth's surface, in order not to come back?
2. it
If
the
it
take
Ans.
4 days, 18
:
hrs.,
10 min.
* In working these exercises, the following data may be used Radius of the moon, that of the earth. Mass of moon, ^T that of earth. distance Mean of moon from earth, 237,000 miles. Acceleration of gravity on the surface of the moon, that on the surface of the
earth.
of sun, 860,000 miles. of sun, 333,000 that of the earth. Mean distance of earth from sun, 93,000,000 miles. Acceleration of gravity on the surface of the sun, 905
Diameter
Mass
ft.
72
3.
MECHANICS
Solve the preceding problem accurately, assuming that the
at their present
moon are released from rest in interstellar space mean distance apart. Their common centre of
for the earth
mean
the earth and the moon were held at rest at their present distance apart, with what velocity would a projectile have to be shot from the surface of the moon, in order to reach the
earth?
the earth and the moon were held at rest at their present distance apart, and a stone were placed between them at the point of no force and then slightly displaced toward the earth,
6.
If
mean
it
If
a hole were bored through the centre of the moon, asrest in interstellar space,
it
sumed spherical, homogeneous, and at and a stone dropped in, how long would
the other side ?
8.
if two spheres, each one foot in diameter and of the earth's mean density (specific gravity, 5.6) to density equal of an inch apart and were acted were placed with their surfaces
Show
that
on by no other forces than their mutual attractions, they would come together in about five minutes and a half. Given that the spheres attract as if all their mass were concentrated at their
centres.
Work Done by a Variable Force. If a force, F, constant magnitude and always acting along a fixed line AB in the same sense, be applied to a particle,* and if the particle be displaced along the line in the direction of the force, the work done by the force on the particle is defined in elementary physics as
9.
in
F
I
W
B
eo
Fl,
p
p
where
the particle has been displaced. Suppose, however, that the force is variable, but varying continuously and always acting along the same fixed line. How
shall the
of a rigid or deformable
MOTION OF A PARTICLE
as the axis of x.
to B:
73
line
Let a coordinate be assumed on the line i.e. think of the Let the particle be displaced from A: x
;
6,
and
let
<
b.
Then
F=f(x),
where f(x) denotes a positive continuous function of x. Divide the interval (a, b) up into n parts by the points x l9 xz zn _i, and let X Q = a, x n = b. Then, if
,
xt +i
Xk
Azfc,
the work, ATF*, done by the force in displacing the particle through the fc-th interval ought, in order to correspond to the general physical conception of work, to lie between the quantities
FiAx
'
and
Fi'Az,
where FJ and F' k denote respectively the smallest and the largest in We have, then this interval.* values of f(x)
:
(1)
FiAx ^ ATFt ^
Fi'te.
I,
,
On
n
that
(2)
writing out the double inequality (1) for k = 0, 1 and adding the n relations thus resulting together,
we
find
W = 2 AWk
lies
F'^x
F'Jte
of these
(3)
+ F(Ax + + F('&x +
)
Each
(4)
/(*
Az
+ f(Xl
A*!
+ /(*_,) Ax n
whose
(5)
lim [/(x
n-oo
A*
+ /(x,) Ax +
h
+ /(*._0 Ax]
:
= f /(*) As.
That
J a
is
(6)
W = Jf(x)dx,
a
follows from
DuhamePs Theorem.
* This statement is pure physics. It is the physical axiom on which the generalMore precisely, it is one of two physical ization of the definition of work is based. axioms, the other being that the total work, W, for the complete interval is the sum of the partial works, ATT*, for the subintervals.
74
MECHANICS
If the force F acts in the direction opposite to that in which the point of application is moved, we extend the definition and say that negative work is done. For the case that F is constant, the work is now defined as follows
:
(7)
W
F
is
= F(b -
a).
Thus
(7) is
provided that a < 6. Will (7) There are in all four possible cases
i)
seen to hold in whichever direction the force acts, It will. still hold if 6 < a?
:
++
done.
ii)
Hi) H
In cases
i)
and
work
is
done.
ii) the force overcomes resistance, and positive In cases in) and iv) the force is overcome, and
negative work
It is laid
is
Hence
now easy to see how the definition of work should be down when F varies in any continuous manner. The consider-
ations are precisely similar to those which led to Equation (6), and that same equation is the final result in this, the most general, case &
:
= Cf(x) dx.
a
Example. To find the work done in stretching a wire. Let the natural (or unstretched) length of the wire be I, the stretched Then the tension, T, is length, V.
__,
r-^-i
T>
A
FIG. 59
given by Hooke's
Law
\
T =
I
~~
where X
ulus.
is
independent of
its
and
V,
and
is
known
as Young's
Mod-
when
stretched,
along the line OA, and being an arbitrary interLet x be measured from A, and let x = h at
natural state,
lie
lie
along OB,
OP
T = \~
and
,
W
TT7
/\ x = I = \jdx
,
x
j
r
I
xdx =
,
-==-
MOTION OF A PARTICLE
This
If
is
75
it.
the wire contracts, the work done by the wire on the body to which its end P is attached will be
Xa 2
/(->!)*<"
1.
21
21
EXERCISES
In the problem of
particle
by the earth
on the
2.
when
A
is
particle of
that the
plane
mass m moves down an inclined plane. Show work done on it by the component of gravity down the the same as the work done by gravity on the particle
descends vertically a distance equal to the change in
when
level
3.
is
it
by a
force
which
inversely proportional to the square of the distance from 0. How much work is done on the particle when it moves from a distance a to a distance b along a right line through 01
4. If the earth and the moon were stopped in their courses and allowed to come together by their mutual attraction, how much work would the earth have done on the moon when they meet?
5. Find the work done by the sun on a meteor when the latter moves along a straight line passing through the centre of the
sun, from
10.
an
initial
distance
to a final distance
r.
ds/dt.
mv 2
'
Let the particle move under the action of any force F which continuously: F = /(s). Then Newton's Second Law can be written in the form
varies
:
dv
fi
-/().
f(s)ds.
f/
Hence
mvdv =
76
MECHANICS
6,
Integrate this equation between the limits a and the corresponding values of v by v l and v2
:
denoting
mv
dv
=
:
f(s) ds.
The
mv
Pa
~2~ n
The right-hand side is, by definition, the work Hence particle by the force F.
(1)
i
done on the
and we
THEOREM.
equal
to the
The change in the kinetic energy of a work done on it by the force which acts on it.
particle is
the Principle of
it
This theorem expresses, in this the simplest case imaginable, Work and Energy in Mechanics. By means of a first integral of the equation arising from Newton's Second
known
amine
(4) in
can be found in the case of a particle, when the force is as a function of the position, and the student will do well to go back over the foregoing problems and exercises, and expoint of view e.g. Equation and 7, Equation (2) in 8 are, save 5, Equation the factor m/2, the Equation of Energy, as (1) is often called.
their solution
this
;
Law
from
new
(3) in
for
EXERCISES
Work
Method
11.
is
the Exercises of
of
5,
7,
8,
so far as possible,
by the
Change
to determine
how many
contained in a given
introduction of this paragraph and the next at this stage seems to require If these two purely physical subjects are sufficiently important to be justification. taken up here, then why not, at the beginning of this chapter, the first time they are needed ? But if they are merely for reference, why break the unity, coherence, of the presentation by placing them here rather than at the end of the chapter? The Author feels that this is about the time when the beginner in Mechanics should turn his attention systematically to these subjects, for until he has some knowledge of the problems studied in this chapter, ho can hardly be expected to recognize the importance of Change of Units and of the Check of Dimensions.
The
MOTION OF A PARTICLE
amount
of the substance.*
lines is to find
77
how many
the unit of length tained in a given right-line segment. The number, s, thus resultIt depends on two things ing is called the length of the line. If a different unit of the particular line and the unit chosen.
length be chosen, the s', assigned to it, and
lines, s'
Thus to measure the length of right times a right line chosen arbitrarily as is cona foot or a centimetre or a cubit
same
its
line will
Now,
for all
will be proportional to s :
s' oc
(1)
or
s'
s,
a constant depending on the units. It is determined in any given case by substituting particular values for s and s', known to correspond. Thus if we wish to transform from feet to yards, consider in particular a line which is a yard long. Here,
where
c is
s' will
equal
and
s will
1
equal
3,
so
c
3c,
'
i,
and,t
(2)
$8.
Example.
If
a yard
is
the unit of length, a minute the unit and a kilogramme the unit of
We
(3)
will start
*<y
32/,
* The word substance here may be too narrow in its connotations, for we want a word that will include every measurable quantity, from the length of a lightwave to the wheat crop of the world. Such a woid obviously does not exist, and so we agree to use substance in this sense as a terminus tcchnicus. t It would seem paradoxical to say that the same line has a length of 6 when the foot is the unit, and a length of 2 when the yard is the unit. But it must bo remembered that the length is a function of two variables, the unit being one of them. The attempt is sometimes made to meet the apparent difficulty by saying "3 ft. = 1 yd." But this makes confusion worse confounded; for 3 = 1 is not true, while on the other hand to try to introduce "concrete numbers," like 3 ft., 10 Ibs., 5 sees., into mathematics, is not feasible. To try to change units in this way leads to blunders and wrong numerical results. There is only one kind of number in elementary mathematics. To attempt to qualify it as abstract, is to qualify that which is unique. The denominate attribute (3 ft., 10 Ibs., etc.) is part of the physical thing conceived it does not pertain to the mathematical counterpart, which is purely arithmetical. this t Compare equation with the attempted form of statement mentioned in " the last footnote 1 yd. = 3 ft." It would seem to follow from that statement that ' yds. = 3s ft. But ' = What a cheerful prospect for getting the right answer by that method 1
; :
78
MECHANICS
in the
form
Then
Next,
the problem
is
to determine
X'.
m =
f
km,
,
_ -
j. ft
v X
9000 ZOOO,
lc lc
20Q()
m m' m 2QO()
$
2.20'
Similarly,
i'
_
60'
2
? =
Thus
_ m ,d "^
s'
60 2
2000*X3
md
^
s
XT *j =
The
__L_\'/ A;-
220
left-hand sides of these equations are equal by (4). equating the right-hand sides and dividing by (3) we find
:
On
60 2 2000 X 3
X'
2.20
X'
422.4.*
32'
On
new
form
EXERCISES
the units of length, time, and mass are respectively a mile, a day, and a ton, compute the absolute unit of force in pounds.
1.
If
If
is
981 in the
c.g.s.
system,
3.
English system,
compute g
*
More
102
two
significant figures.
MOTION OF A PARTICLE
4.
79
If
the unit of force be a pound, the unit of time a second, of length a foot, explain what is meant by the absolute
it is
equal (nearly) to 32
Ibs.
system.
6.
mass
is
12. The Check of Dimensions. The physical quantities that enter in Mechanics can be expressed in terms of the units of Mass [Af], Length [L], and Time [7"]. Thus velocity is of the
dimension length/time, or L/T = LT~l Acceleration has the dimension LT~ 2 and force, the dimension ML/T~*. When, an equation is written in literal form, as
.
each term must have the same dimension. For, such an equation remains true, no matter what the units of mass, length, and time may be; and if two terms had different dimensions in any one
of of units
the fundamental quantities (mass, length, time), a change would lead to a new equation not in general equivalent
Thus,
an
ellipse is
all
the quantities x,
y, a, b are of
of its area
must be L 2
and
it is,
for
A =
:
irab.
the ellipsoid of revolution corresponding to rota3 tion about the axis of x should be of dimension L and it is
,
V =
out
7ra6
2
.
This principle affords a useful check on putting in or leaving Thus in the g, when problems are formulated literally. in our if of we had writing down the forgotten g Example 3, would of dimensions check the immediately right-hand side,
have shown up the oversight. For, the left-hand member is of dimension ML/T~~*\ hence every term on the right must have
80
this
It
MECHANICS
same dimension.
of course, only
is,
when
It does, in the correct equation of the text. all the quantities which enter are in
form, that the check can be used. If some are replaced by numbers, the check does not apply. Observe that in computing the dimension of a derivative, like
literal
d 2 s/dt 2 we may think of the latter as a quotient, the numerator being a difference, and hence of the dimension of the dependent variable, while the denominator is thought of as a power.
,
EXERCISES
Determine the dimension of each of the following quantities:
1.
Kinetic energy.
Ans.
2.
Work.
Ans.
Ans.
3.
4. 6.
ML T~ ML T~ ML
2 2
2
.
2
.
Ans.
MLT~
l
.
Ans.
density.
ML
T' 2
Z
.
6.
7.
Volume
8.
9.
2
.
1
.
Ans. LT~ 2 The acceleration of gravity. 10. The wind resistance can often be assumed proportional 2 what is If it is written as cv to the square of the velocity.
.
the dimension of c?
11.
Ans.
ML~
l
.
sistance
12.
In Question 10, what is the answer when the wind is taken per square foot of surface exposed ?
re-
3.
13.
14.
Equation
(3),
the check
Explain why.
What
Modulus?
15. In the Example treated in 8, we wished to find the velocity But of the stone at the centre of the earth in miles per second. 2 = if we substituted for R, in the formula (dr/df) gR, the value
of
R in miles (i.e. 4000), we obtained a wrong answer, even though the dimensions of both sides of this equation are the same, namely,
L 2/T~ 2
.
Explain why, and show how Formula (2), which is one hundred per cent literal, can be used to yield a correct result,
when
16.
R=
4000.
in
of
Dimensions
applicable.
MOTION OF A PARTICLE
13.
81
When a body moves in a Resisting Medium. air or the these media oppose rethe through water, through on which the velocity, but of the depends magnitude sistance,
Motion
does not follow any simple mathematical law. For low velocities up to 5 or 10 miles per hour, the resistance R can be expressed approximately by the formula
:
(1)
R =
av,
where a is a constant depending both on the medium and on the For higher size and shape of the body, but not on its mass. velocities up to the velocity of sound (1082 ft. a sec.) the formula
(2)
R=
We
shall
cv*
many
speak
graph.
Problem 1. A man is rowing in still water at the rate of 3 miles an hour, when he ships his oars. Determine the subsequent motion of the boat. Here Newton's Second Law gives us
:
/o\ (3)
dv
m^=-av.
at
j*
TT
Hence
/A\
m dv
a
v
,
i
(4)
A t
m =-
VQ
log-^
V Q is the initial velocity,
where
nearj^^Bft. a
sec.
To
v,
observe that
a*
m
Hence
(5)
*>o
log--,
v
or
v _at VQ e
.
It
rest,
might appear from (5) that the boat would never come to but would move more and more slowly, since
_at
lim e~
= 00
0.
We
warn the student, however, against such a conclusion. For the approximation we are using, R = av, holds only for a limited
82
time,
will
MECHANICS
and even
for that time is at best
probably not be
many
if
sidewise,
of a for this aspect of the boat would indeed the approximation K = av could
To determine
we have from
(3)
av,
and consequently
(6)
o-^.
Hence, even if the above law of resistance held up to the limit, the boat would not travel an infinite distance, but would approach a point distant
feet
momentum.
and
t,
integrate (5)
ds
-?
/>7\
(7)
ffiVn /1
-(l
~m\ m
).
From
this result
is
a distance of
ft.
EXERCISE
force of 3 Ibs.
Ibs. and if a steady maintain a speed of 3 miles an hour in still water, show that when the boat has gone 20 ft., the speed has fallen off by a little less than a mile an hour.
If
the
man and
is
Problem
2.
A
ft.
velocity of v
drop of rain falls from a cloud with an a sec. Determine the motion.
is
initial
We
already of
it
its
final
size,
not
proceeds,
2
.
R=
ct;
MOTION OF A PARTICLE
The
and
it
83
forces
i)
if)
of the particle
we
will
AP, Figure
:
60,
which
has
fallen.
= mg
cv 2 .
cv 2
TT
Hence
dv
v~r
as
= mg
cv*
mv dv cv*' mg
s
=-
log (mg
cv 2 )
<7,
and thus
/ON (8) v 7
finally
s
W m = 5log
i
<7
2c
m#
cv
2
FIG. 60
-t'
00
cv 2
Solving for v
we have
mg
(9) v y
^gg-^-^g
s increases indefinitely,
When
its
as
limit,
equal) v
It is
is
arid_hence the velocity v can never exceed (or quite Vmg/c ft. a sec. This is known as the limiting velocity.
initial velocity, and by the rain as it falls, for a rain drop is faster when it reaches the ground than it was
practically attained
EXERCISES
1.
its
Work Problem 2, taking as the coordinate of the rain drop height above the ground.
2. Find the time in terms of the velocity and the velocity in terms of the time in Problem 2.
3.
Show
it will
that, if a charge of shot be fired vertically upward, return with a velocity about 3 times that of rain drops
84
of the
;
MECHANICS
same size and that if it be fired directly downward from a balloon two miles high, the velocity will not be appreciably
greater.
4.
3,
Determine the height to which the shot will and show that the time to the highest point is
rise in
Question
where
is
The resistance which the at14. Graph of the Resistance. mosphere or water opposes to a body of a given size and shape can in many cases be determined experimentally with a reasonable degree of precision and thus the graph
of the resistance
:
JL
'
^
function of
phere,
v.
can be plotted. The mathematical problem then presents itself of representing the curve
with sufficient accuracy by means of a simple In the problem of vertical motion in the atmos2,
Problem
13,
dV = m -^
mg
f/ \ f(v),
according as the body is going up or coming down, s being measNow if we approximate to f(v) by ured positively downward. means of a quadratic polynomial or a fractional linear function,
or
And it is obviintegrate the resulting equation readily. we can so approximate, at least, for a restricted range of values for v.
we can
ous that
is
medium
is
is
MOTION OF A PARTICLE
85
Here a and b are merely arbitrary constants, enabling us to imfor example, to pose two arbitrary conditions on the curve, make it go through two given points, and are to be determined so as to yield a good approximation to the physical law. Sometimes the simple values 6
1, 2,
larly appearing Thus, in geometry, the areas of similar surfaces and the volumes of similar solids are proportional to the squares or cubes of corresponding linear dimensions. This law expresses a fact that
these approximate formulas with simiformulas that represent exact physical laws.
holds to the finest degree of accuracy of which physical measurements have shown themselves to be capable and with no restriction whatever on the size of the bodies. But the law R = av 2
= cv* ceases to hold, i.e. to interpret nature within the limits or of precision of physical measurements, when v transcends certain restricted limits, and the student must be careful to bear this
fact in mind.
EXERCISES
Work
and which
v
1.
I,
and
s,
is
the only force acting is the resistance of the represented by the formula
:
medium,
R =
Show
bv
cv\
2.
R =
~-~V
7
--
3.
dv
R =
av*.
4.
that
it
formulas of Questions
5.
would be feasible mathematically to use the 1 and 2 in the case of the falling rain drop.
A train weighing 300 tons, inclusive of the locomotive, can be just kept in motion on a level track by a force of 3 pounds The locomotive is able to maintain a speed of 60 to the ton. miles an hour, the horse power developed being reckoned as 1300.
Assuming that the
frictional
same
at high
speeds as at low ones and that the resistance of the air is proportional to the square of the velocity, find by how much the speed of the train will have dropped off in running half a mile
if
the steam
6.
is
full
speed.
A man
How
large
must the parachute be that the man may trust himself to it at any height, if 25 ft. a sec. is a safe velocity with which to reach
the ground?
of the air
is
as the square
86
of the velocity
MECHANICS
and
is
sec.
Ans.
7.
About 12
is
ft.
in diameter.
toboggan
is
slide of
fall
constant slope
ft.
a quarter of a mile
coefficient
is
of 200
Q,
proportional
to the square of the velocity and is equal to 2 pounds per square foot of opposing surface for a velocity of 30 ft. a sec., and that a loaded toboggan weighs 300 pounds and presents a surface
find the velocity acquired ft. to the resistance of the air the descent and time the required to reach the bottom. during Find the limit of velocity that could be acquired by a toboggan under the given conditions if the hill were of infinite length.
of 3 sq.
Ans.
(a)
68
ft.
sec.
(b)
30
sees.
(c)
74
ft.
sec.
8. The ropes of an elevator break and the elevator falls without obstruction till it enters an air chamber at the bottom of the shaft. The elevator weighs 2 tons and it falls from a height
of 50
is
ft.
The
If
12
ft.
no
is 6 X 6 ft. and its depth escaped from the well, how far would the What would be the maximum weight of a
man
law
:
of 170
pounds?
of air
const.,
is
9. In the early days of modern ballistics the resistance of the atmosphere to a common ball was determined as follows. A number of parallel vertical screens were set up at equal dis-
tances, the ball was shot through them (with a practically horizontal trajectory), and the time recorded (through the breaking of an electric circuit) at which it cut each screen. Explain the
theory of the experiment, and show how points on tho graph of the resistance as a function of the velocity could be obtained.
16. Motion in a Plane and in Space. Vector Velocity. When a point P moves in a plane or in space, its position at any instant can be represented by its Cartesian coordinates
:
(1)
*-/(*),
2/
MOTION OF A PARTICLE
87
where the functions are continuous, together with any derivatives we shall have occasion to use. The velocity of P has been defined as ds/dt. For, hitherto, we have regarded the path as given, and it was a question merely of the speed and sense of description of the path. But now we need more. We need to put into evidence the direction and sense of the motion, and so we extend the idea, defining velocity more broadly as a vector. Lay off on the tangent to the path, in the sense of the motion, a directed line segment whose length is the speed of the point, and let the vector thus determined be
defined as the vector velocity of the point P. Composition and Resolution of Velocities.
mouse runs
across
To determine
mouse
the velocity of the car is u, and the velocity of the mouse relative
to the car
is v.
Let the mouse start from a point P on one side of the car and run across
the floor in a straight line with constant Let Q v, relative to the car.
ut
velocity,
seconds.
Then
p~ _
FIG. 62
initial position of
let
be the
In Figure 62, the line the vector velocity u of the train, and AB represents the vector velocity v of the mouse relative to the freight car. Their geometric, or vector, sum is
represented by 07?. From similar triangles it appears that the path of the mouse in and 5, space is the right line through
OP = ut. OA represents
velocity in space
is
the vector
v.
velocity in space may be defrom scribed, analogy with the paralleloof of as resultant the two component velocities, the gram forces, u along the direction of OA and v along the direction OC through
Thus her
parallel to
A.
88
Similarly,
MECHANICS
velocities along
;
ponent
any vector velocity may be resolved into two comany two directions complanar with the
given velocity Fig. 63. The extension to space is obvious. Any three non-complanar vector velocities can be composed into a single velocity by the
parallelepiped law.
And
any
divided by
A,
When
this
vector,
PP
divided
by
or,
~P**' -
r hm
A =O
At
=r hm
AJ-=O
As
At
n = = Dts
v,
n numerically.
(PQ) approaches
its limit.
And
v,
as its limit, or
--VAt
AZ =
lim (PQ)
v.
This vector,
defined as the vector velocity of the point P. To prove that the above limit actually consider the exists, components of (PP') and (PQ) along the.
v, is
Cartesian Coordinates.
axes.
These arc
Ax,
Ay,
A*
and
A#
-,
:
Ay
Az
-,
=
Dtz.
of
The
approach limits
Af-0 At
lim -7 A-0 At
Dix,
lim --7
y,
lim
A/=*0
~
At
limit, v,
v along
not feasible to represent vectors by bold face type, the ( ) The student should draw the be used, as: (PP') or, later, (a). figure which represents the vectors (PP') and (PQ)>
is
When it may
MOTION OF A PARTICLE
dx
Vx
89
__ "
~Tt'
Vv
__ ~
dy
Vz
~dt'
dz
'
These equations admit the following physical interpretation. Consider the projections, L, Af, TV, of the point P on the axes of
coordinates.
The
velocities with
along the axes are precisely dx/dt, dy/dt, and dz/dt. And so we can say The projections of the vector velocity v along the axes are
:
to the velocities of the projections. Finally, observe that, just as the average vector velocity approaches the actual vector velocity as its limit, so the projections
equal respectively
of the average vector velocity approach the projections of the actual vector velocity as their limits.
The student may raise the question If vx , vy , and are the components of the vector velocity, v, are they not, therefore, themselves vectors, and should they not be written
Remark.
:
vz
as such, v z
flict
vy vz ?
,
Yes, this
is
correct.
But
it
on the axes of
and
z.
tion (but not sense) is sional vectors, and these are equivalent to directed line segments, since the two systems stand in a one-to-one relation to each
other.
cally
v x , vy ,
and
vz as directed line
by the ordinary
zero.
EXERCISES
1.
Show
that,
if
component
velocities along
:
polar coordinates in the plane are used, the and orthogonal to the radius vector
are respectively
dr
dB
2.
point
moves on the
dB
surface of a sphere.
.
Show
that
A d<p
where
tude.
3.
and
<p
A point moves in
dr
space.
Show
dB
that
where
r, 6, <p
90
16.
MECHANICS
Vector Acceleration.
as
in
(1)
cf.
15.
By
Av =
Fig. 65, p. 96.
v'
v,
The
vector
Av
AT
When A
approaches 0, the average vector acceleration approaches a limiting value, and this limiting vector is defined as the vector
acceleration of the point
:
(a)
Av r = hm
Ar=o At
Cartesian Coordinates.
The components
,
ation along the Cartesian axes, at x a^, and a z arc readily computed. For, the components of the vector (1) along the axes are respectively :
v' x
vx
v'y
'
vy
v' z
'
vz
AJ
AJ
AZ
As in the case of velocities, the components of the limiting vector and the limits approached by the components of the variable vector are respectively equal.* Hence
ax
= hm
,.
At> z
***
~ = Dtv x
A/ =
av =
hm
A/
,.
Avy
&t
~ = Dtv v
az
= hm
..
Av z
At
~ Dtv t
A/
or:
d'2 z
Osculating Plane and Principal Normal.] Let a vector r be of space to the variable drawn from an arbitrary fixed point
point
that
is
(1),
.
15.
Then
v
Let
s
dr
--a
*-
be the
arc,
of the
motion
and
let
ds
* This theorem is true of any vector which approaches a can readily verify. 8. f Cf. the Author's Advanced Calculus, p. 304,
limit, as the
student
MOTION OF A PARTICLE
Then
r'
is
91
Furthermore,
=
ds
is
a vector drawn along the principal normal, toward the concave side of the projection of the curve on the osculating plane,
and
its
On
=
Hence
dv
and
dsdr *-__
EXERCISES
1. A point describes a circle with constant velocity. Show that the vector acceleration is normal to the path and directed
circle,
and that
or
2 or
its
magnitude
is
"
2
,
r.
2. Show that, when a point is describing an arbitrary plane path, the components of the vector acceleration along the tangent 2 and normal are <j s v*
:
at
w'
an
?
<x n
whoro p denotes the radius of curvature, and the component is directed toward the concave side of the curve.
3. A point describes a cycloid, the rolling circle moving forward with constant velocity. Show that the acceleration is constant in magnitude and always directed toward the centre of the circle. 4.
in
where a r a denote the components of the acceleration along and perpendicular to the radius vector. Use the system of -ordinary complex numbers, a + bi, where
,
=V
5.
1, '
and
set
r
re 6 \
92
17.
MECHANICS
Newton's Second Law.
Let a particle
move under
the
Let (a) be action of any forces, and let F be their resultant. Then Newton's Second Law of Motion its vector acceleration.
asserts that the
mass times
if
adopted,
m(a) =
F.
:
= X,
(2)
d?z
t
If, in particular, X, Y, Z are continuous functions of x, y, z, and t y it then follows from the theory dx/dt, dy/dt, dz/dt, of differential equations that the path is uniquely determined
by the
initial
,
conditions;
i.e.
if
the particle
is
projected from a
point (z 2/o> ZG) with a velocity whose components along the axes This remark are (u^ v w ), the path is completely determined. is striking when one considers that the corresponding theorem
,
is
not true
if
of
one determines the motion by means of the principle cf. the Author's Advanced Calculus, p. 351,
Singular
The
essential
point
here
is
that
Equa-
tions (2) never admit a singular solution, whereas the equations of Work and Energy do.
it is
determined by the
initial
conditions.
firmed in the case of each of the examples considered below. For a general treatment, cf. Appendix A.
The force, F, always lies in the osculating Osculating Plane. of the plane path. 16, and Equation (1) above, For, from
Hence we can
resolve
F
o
into a
component
tYllfi
where p
l//c.
MOTION OF A PARTICLE
EXERCISE
Show
that
(r'
93
r")
F =
0.
-(z'x")Y+(x'y")Z =
where
0,
and
ff JU
/ 1
y'z"-z'y",
dx
~;
.
etc.
JU
f fr
etc.
ds
18.
Motion
of
a Projectile.
Problem.
To
find the
path
of a
projectile acted on only by the force of gravity. The degree of accuracy of the approximation to the true motion
obtained in the following solution depends on the projectile and on the velocity with which it moves. For a cannon ball it is crude, though suggestive, whereas for the 16 Ib. shot, used in
putting the shot, it is decidedly good. Hitherto we have known the path of the body;
not.
The path
will
Second
Law
of
Motion becomes
m
dt
Y are the components of the resultant force along the where axes, measured in absolute units. In the present case X = 0, Y = mg, and we have
,
(2)
FIG. 64
with velocity V Q at If we suppose the body projected from an angle a with the horizontal, the integration of these equations
ives:
dx
-jj
(J
=
a
dt
VQ
cos a,
cos
v Q sin
0tf ,
sin
94
Eliminating
(3)
t
MECHANICS
we
get
:
x tan a
at the point
/%
A
-
(z,, t/J,
#?, U
*>
t$ U
sin
a cos a
'
sin 2
a
:
*>
ff
~2T~
y lt
+
*
a?!,
y'
'
n, Cc
This curve
is
a parabola with
its
vertex at A.
The
height
of its directrix
above
of
and hence the height v\ the directrix of the parabola represented by (3) is 2 2 vl sin a v% cos a _ v%
above
is
cos 2 a/2</,
~~2g~
The
it
+
.
"
2i~
27'
and so
appears that all the parabolas traced out by projectiles leaving with the same velocity have their directrices at the same level,
would
rise if
EXERCISES
1.
Show
is
R =
j/
sin 2a,
is
attained
when a = 45
~
g'
The
2.
is
launched with a velocity of V Q ft. a sec. and projectile Show that is to hit a mark at the same level and within range. there are two possible angles of elevation and that one is as much
is
greater than 45
as the other
is less.
j3
3. Find the range on a plane inclined at an angle horizon and show that the maximum range is
to the
~
~g
sin
MOTION OF A PARTICLE
4. is
95
A small boy can throw a stone 100 ft. on the level. He on top of a house 40 ft. high. Show that he can throw the stone 134 ft. from the house. Neglect the height of his hand above the levels in question.
6.
The
one time, 46
Gin.
best collegiate record for putting the shot was, at ft. and the amateur and world's record was 49 ft.
If a man puts the shot 46 ft. and the shot leaves his hand at a height of 6 ft. 3 in. above the ground, find the velocity with which he launches it, assuming that the angle of elevation a is
Am.
35.87.
6. How much better record can the man of the preceding question make than a shorter man of equal strength and skill, the shot leaving the latter's hand at a height of 5 ft. 3 in. ? 7.
Show
that
it is
possible to hit a
mark B
(x6,
2/&),
provided
8.
ft.
sec.
of a mile
9.
It has
piano curve.
been assumed that the path of the projectile is a Prove this assumption to be correct by using all
17.
Constrained Motion.
in a given curve, like
first
move
Consider
the normal be N.
Motion,
17,
(1)
mv 2 = N.
P
in
17
analysis.
We
will
give one for the plane case without the use of vector methods.
Geometric Proof. Compute the components of the acceleration along the tangent and along the normal.
vector
Let
<p
96
MECHANICS
be the angle which the tangent has turned through in passing Then the component of Av along the tangent from P to P' will be
.
v cos
tp
=
<p
(t;
+ Av) cos
v (1
<p
=
By
K
Av cos
cos
<p).
= lim~r,
PP'
lim
PP'
is
&
At
TT
=
<p
Av
-r-
COS
At
<f>
cos (p -
At
0,
Let At approach
first
0.
Then
is
approaches
and the
limit of the
lim
AZA/V
Mlimcos^J
~ = D
v.
To
cos
y?
__
cos
<p
<p
jp_
As
At
At
As
The
first
remain
finite,
of the right
hand
side
is 0.
We
v cos
lim
<p
~ = Dtv,
established.
first
of Equations (1)
is
obtain the second of Equations (1), consider the component of the average acceleration along the normal, or
v
To
sm
At
<p
sin
<p
<p
As (?_ _ As At'
MOTION OF A PARTICLE
where
is
97
s is
t.
The
2
seen to be
= -,
P
It is
and
The component
the centripetal force due to the motion. The foregoing analysis yields the first of Equations (1) for twisted curves.
EXERCISE
Use the present geometric method to obtain the formulas
1
:
where a r
ot e
acceleration along
20.
vector.
lum.
Here
s
^s = m -772 at
=
1 0,
mg
sin
0,
and since
is
Pendulum
of
first
.
integral
by
the
method
2g
dt dt 2
77^
de
-jr.
~T~
I
sin u
at
=
where a
(2)
is
cos
+
;
C,
hence
^ = ^(cos0-cos).
I
in the path at the lowest point = 0, or times the angular velocity for that same would have been acquired and is the cos V20Z (1 a), if the bob had fallen freely under the force of gravity through the
is
The velocity
MECHANICS
same
If
difference in level.
Equation
(2) is virtually
the Integral
of Energy.
we attempt
by
de
we
20
Vcos
cos
This integral cannot be expressed in terms of the functions at When 0, It is an Elliptic Integral.* present at our disposal. however, is small, sin 6 differs from 6 by only a small percentage
of either quantity, and hence we may expect to obtain a good approximation to the actual motion if we replace sin 6 in (1) by
:
(3)
_ -~
is
2 7, A), n having here the value Simple Harmonic Motion, a when through a small swings pendulum Hence, simple g/l. harmonic and its period is its motion approximately amplitude,
is
approximately
9
The Tautochrone.
mathema-
In what curve should ticians of the eighteenth century was this a pendulum swing in order that the period of oscillation may bo
It turns out that absolutely independent of the amplitude? the cycloid has this property. For, the differential equation of
motion
is
dz s
where
'
$ is
point,
and
4a
sin
r,
FIG. 67
we have
d*s
-j-z 2
g -~
at
4a
s.
* Cf. the author's Advanced Calculus, Chapter IX, page 195, where this integral
is
MOTION OF A PARTICLE
This
7,
is
99
9
is
independent of the amplitude. cycloidal pendulum may be constructed by causing the cord
of the
pendulum
to
of the path.
it
The
resist-
would thus be
slight
21. Motion on a Smooth Curve. Let a bead slide on a smooth wire under the force of gravity. Consider first the plane case. Choosing the axes as indicated, we have
:
dx
(1) dt*
ds
Hence
_ ~
2
\A:(xQ ,y
9g
~ds~dt
we
find
If
we suppose
A, then
rest at
=
(2)
2gx Q
+ C,
FIG. 68
But the
x
x In the more general case that the bead passes the point with a velocity v we have
:
velocity that a body falling freely from rest a distance of attains is expressed by precisely the same formula.
eg
2
=
eg
2gx<>
C,
(3)
t>
2g (x
).
Thus
seen that the velocity at is the same that the bead would have acquired at the second level if it had been projected vertically from the first with velocity
it is
.
100
MECHANICS
also asserts that the
;
The theorem
is
equal to the work done on the bead If the bead starts from rest at A,
it
10.
it will
continue to slide
till
reaches the end of the wire or comes to a point A' at the same v* In the latter case it will in gen>^L level as
FIG. 69
then retrace path back to A. But if the tangent to the ' is curve at A horizontal, the bead may
eral just rise to the point
its
A and
1
approach
it.
EXERCISES
1. A bead slides on a smooth vertical circle. It is projected from the lowest point with a velocity equal to that which it would acquire in falling from rest from the highest point. Show that it will approach the highest point as a limit which it will never
reach.
2.
From
(2)
deduce the
first
integral
20.
Space Curves.
dimensions.
based on Cartesian coordinates. Choose the axis of x as before positive downward. Then we have
It is interesting, however, to give a solution
:
~dT*
= "%
"'
D +R
.
*>
(4)
_ ~* ~ D
d*y
where
Rx Rv R
, ,
Since
are the components of the reaction of the wire is normal to the curve, we have
s-S+*-8 + *!To
integrate Equations (4) multiply through respectively by thus find, with the aid dz/dt and add. dy/dt,
dx/dty
We
of (5):
MOTION OF A PARTICLE
But
"2
2
101
(&) +
2
/efo\
2
.
/dz\* U) + b)
(dy\
Hence
(7)
(6)
reduces to
/o\ (8)
mvZ
-7j
is
mv ---
2^
/ = "V (*
\ *<>)
This
It is
written
It could have been precisely the Equation of Energy. at the start from the Principle of Work and Energy. the generalization of (2) for space curves.
down
EXERCISE
bead
vertical.
slides on a smooth wire in the form of a helix, axis Determine the reaction of the wire in magnitude and
direction.
22.
circle
describes Centrifugal Force. When a particle of mass with constant velocity, the acceleration is directed toward
its
magnitude
is
The
Its
force
therefore,
which holds the particle in its path is, normal to the path and directed inward.
is
magnitude
N
Why,
the centre"?
,r
2 = mv =
mco 2r.
FIG. 70
"
centri/u^aif force"
The explanation is a confusion of ideas. If the mass is held in its path by a string fastened to a peg at the centre, in the direction OP away from the 0, does not the string tug at centre and is not this force exerted by the particle in its attempt, or tendency, to fly away from the centre? The answer to the
first
methods
question is, of course, "Yes." Now one of the standard of the sophists is to begin with a question on a non-
controversial point, conceded without opposition in their favor, "and issue in their second question
102
MECHANICS
Matter cannot exert force, for a force is a push or a pull, and matter can neither push nor pull it is inert. The particle does not pull on the string, the string pulls on the particle. But even this statement will be accepted only half-heartedly, if at all, by people who have not yet grasped the basic idea of the science the study of the motion of matter under the of Mechanics What comes first is a material system solid action of forces.
;
bodies, particles, laminae and material surfaces, wires, any combination of these things, including even deformable media (hydro-
dynamics, elasticity)
forces.
and then
this
system
is
acted on by
aere.
think of a smooth table, the particle going round and round in a horizontal circle and being held in its path by a string whose In the of the table. other end is attached to a peg at a point
to
case of the
first
is is
what
"centripetal"
force
the
force
that
the pull of the string now-a-days described " seeks the centre."
particular, this system may as well as gravity, exerts certain forces, all the forces, the table stays at rest.
of
The
varying in direction, causes the forces of the floor to vary. And now, after all is said and done, comes the rejoinder: "But
not have pulled on the peg."
the particle did pull on the string, for otherwise the string would There is no answer to these people.
Some
of
citizens.
They vote
they party that is responsible for the prosperity of the country belong to the only true church ; they subscribe to the Red Cross drive but they have no place in the Temple of Science ; they
profane
it.
sisting of a string 5
1 oz. is shot into a sling, conone end fastened at 0, the other with long end carrying a leather cup. If the velocity of the bullet is 600 ft. a sec., how strong must the string be, not to break?
Example
1.
bullet weighing
ft.
MOTION OF A PARTICLE
The
tension in the string will be
103
5
^
I
mv
FIG. 71
where
m = ^,
600,
or
600 2
16X5
4500 what? pounds?
units, or poundals,
4500
No,
is
measured
in absolute
and
divide
by
32.
is
The
141
get the answer in pounds, we must tension, then, that the string must be able
so, to
to withstand
Ibs.
Example 2. A railroad train rounds a curve of 1000 ft. radius How high should the outer rail be raised, at 30 m. an hour.
if
fc
track?
Standard
in
a plumb bob
is
hung up
then
it
car,
oscillate,
should be
at right angles to the axles of the wheels. It will describe its circular path in space
of
two
forces,
namely,
gravity, mg, downward, and the tension, Let the string make T, of the string.
7
mg
an angle
vertical
is
a.
with the
vertical.
Then the
no
vertical
T cos a. ~
The
horizontal
component
of
yields the
normal force
N which
its circular
path, or
Hence
v2
44 2
is
ft.
in., it
follows that
must be
104
MECHANICS
EXERCISES
weighing 4 oz. is attached to a string which small hole, 0, in a smooth table and carries a passes through If the first weight is projected at its other end. a weight with P a distance of 2 ft. from at from a the table point along
1.
particle
ft. a second in a direction at right angles to OP, the string being taut and the part below the table vertical, how be, that the 4 oz. weight may describe a circular great must
a velocity of 50
path?
Ans.
Ibs.
12 oz.
2. boy on a bicycle rounds a corner on a curve of 60 ft. radius at the rate of 10 m. an hour, and his bicycle slips out from under him. What is the greatest value p, could have had ?
Ans.
3.
Not
quite |.
it is
conical
projected so
vertical one.
pendulum is like a simple pendulum, only that it moves in a horizontal circle instead of
in
Show
that
Zo>
2
g sec a.
4. If the earth were gradually to stop rotating, how much would Bunker Hill Monument be out of plumb? Given, that the height of the monument is 225 ft. and the latitude of CharlesAns. About 4 in. town is 42 22'.
steaming east on the equator she puts about and steams in her apparent weight ? at the same what is increase west the rate, 6. If the earth were held in her course by steel wires attached
5.
An
is
an hour.
If
on the side toward the sun and evenly distributed by a plane at right angles to them, show that they would have to be as close together as blades of It is assumed that their other ends are guided near the grass.
to the surface
as regards a cross-section
earth's surface.
7.
Show
is
made
fast to
the sun and which rotates in a plane with constant velocity, making one rotation in a year, could just about reach to the
earth without breaking.
forces of gravitation.
8.
all
steel
is
wire
sq.
mm.
70 kgs.,
if
strung round the earth along the equator. Show that, the earth gradually stopped rotating, the wire would snap.
MOTION OF A PARTICLE
9.
105
What
is
allel, would not break ? 10. If the earth had a satellite the latter rise and set in a day?
how
often would
times.
Ans.
About 18
boy swings a bucket of water around in a vertical circle without spilling any. Does not the bucket exert a pull on the boy's hand ?
11.
i)
Explain the situation by isolating a suitable system, namely: the bucket of water ii) the boy.
;
Centrifugal Oil Cup. A device once used for deterthe mining speed of a locomotive consisted of a cylindrical cup containing oil and caused to rotate about its axis, which was
23.
The
vertical,
the train.
with an angular velocity proportional to the speed of Let us sec how it worked.
oil
to be rotating like a rigid body, with no cross currents or other internal disturbances. What will be the form
Suppose the
on the
oil.
on by the force of gravity, mg, downward and the buoyancy, B, of the oil normal to the surface. The resultant of these two
forces
must
just
yield
the
centripetal
force
its
path.
Now
N
On
mco 2x.
is
FIG. 73
determined by the
B cos T =
Hence
mg,
B sin
U'X
= N.
or
"
dx
It follows, then, that
X
g
'
0)
Thus
it
v-$*.
appears that the free surface
is
a paraboloid of revolution
106
MECHANICS
To Graduate the Cup. It is easily shown that the volume of a segment of a paraboloid of revolution is always half the volume of the circumscribing cylinder. If, then, we mark the level of
the
oil
level
when it is when it is
at rest, the height, h, to which it rises above this in motion will just equal the depth, h, of the
lowest point of the surface below this point. then, that if a denotes the radius of the cup,
From
(1) it follows,
or
EXERCISES
1.
tomato can 4
It is placed
in.
sealed up.
about its axis, which is Find the pressure on the top of the can. Ans. The weight of a column of water 4 and standing on top of the can.
2.
in diameter is filled with water and on a revolving table and caused to rotate vertical, at the rate of 30 rotations a sec.
ft.
high (nearly)
How
The
great
is
seam?
It is possible to view the cup from a statical standpoint. Imagine very tiny insects crawling slowly round on tho surface of the oil. To them the oil and all they could sec of the walls and top of the cup would appear stationary, and they would refer their motion to the rotating space as if it were at rest. We can reproduce the situation, so far as statical problems
24.
field of force, in
are concerned, in a space that is actually at rest by creating a which the force which acts on a particle of mass
is
of gravity, mg, vertical and downward, and a force mcoV directed away from the axis, where co is a constant. Thus the magnitude of the force would be
+
and
it
(mco r)
<p
tan
= wV
MOTION OF A PARTICLE
To
bring the mechanical situation nearer to our
107
human
intui-
a large round cup, 500 ft. across at the tion, with the constructed flooring in the form of the paraboloid top, in question and rotating with the suitable angular velocity. There would be a small opening at the vertex, through which
of
we might think
The view
of all surrounding
be so nearly perfect that, when we were inside the cup, we should not perceive its motion. Suppose, for example, that the slope of the floor along the rim were 45. Then, since
tan T
it
uPx
,
follows that
~
32
co
'
= T4T
(nearly), or .36.
is
The
given
by the equation
or
27T
jT,
T =
0>
2 co /gr
= ^^,
and upon the rim of the cup, this would amount to mgV2, or 41 per cent greater than gravity on the fixed surface of the earth roughly, two-fifths more. A movie actress who was maintaining her weight in Hollywood, would tip the scales at,
well,
said applies, however, only to bodies that are When a body moves, still other forces enter,
be considered in the chapter on Relative Motion. describe the motion of a projectile directly, Nevertheless, since it would be a parabola in the fixed space we started with. Imagine a tennis court laid out with its centre at the lowest point of the bowl. Lob the ball from the back line to the back
and these
we can
line,
slice
108
MECHANICS
One may reasonably inquire concerning the engineering problems of the construction. There will be a tendency of the cup
to burst to fly apart, due to the "centrifugal force."
Can
it
be held together by reinforcing it with steel bands round the outer rim, or will these have all they can do to hold themselves together? It turns out that only one-seventieth of the breaking strength will be needed to hold the band together, thus leaving sixty-nine seventieth^ for reinforcing.
since at the rim the "centrifugal force" is as great as the force of gravity, any unbalanced load will cause the cup to tug
But
A hundred men weigh approxiits anchorage unmercifully. mately 8 tons, and if they were bunched at a point of the rim, the reaction on the anchorage would be 8 tons. The student
on
will find it interesting to
in case a racing
Let a particle be acted on by a force 25. Central Force. directed toward a fixed point, O, and depending only on the Newton's Second Law distance from O, not on the direction.
of
Motion,
17,
then becomes
(1 >
rdt\
where
d.(r*\ =
dt)
r.
'
R is a
continuous function of
Law
(2)
of Areas.
first
integral
==h
'
This equation admits a striking interpretation. Consider the area, A, swept out by the radius
vector drawn from
ticle.
to the par-
Then
A =
/ r*dd,
~~
FIG. 74
dt
dt
MOTION OF A PARTICLE
and hence
(3)
or, equal areas are
109
A=ft(-g,
swept out in equal times.
(2)
have tacitly assumed that h 5* 0. If h = 0, then reduces to dd = 0, and the path is a straight line through 0. Work and Energy. The kinetic energy of the particle is
We
W
By
" nw^_m(W +r 2 2
virtue of (2) this becomes
:
W
<
i
= ~
2
2
2
On
(6)
cf.
r
Chap. VII,
3.
TF
Cfidr.
Hence
I
_ ~
7>
/ mh*J
2
J?
first
This
r
is
order connecting
and 6, and its integral gives the form of the path. The Law of Nature. Newton discovered the Law of Universal Gravitation, which says that any two particles in the universe attract each other with a force proportional to their masses and inversely proportional to the square of the distance between
them.
This law
is
Law
is
of Nature.
In the present case, then, the particle with a force proportional to 1/r 2 and so
,
attracted toward
(8)
ll^,
^
=-
Thus
(9)
The Law
form
of
Equation
(7),
here becomes
110
MECHANICS
=
m/x,
where X
tions.
and
is
initial
condi-
The form
u =
J-
Thus
/io\ (12)
(10)
becomes
+
i
2/4
-
S;
+
.
~
C'.
Write
Since the left-hand side can never be negative, the right-hand side can be written as 5 2 and B itself may be chosen as either
,
Finally, set
- u ~
?
B".
Then
(14)
*
The
in the
(15)
dr 2
+^-
form
x
is
= B cos
(0
7),
0,
the constant of integration. When 5 = of this statement is obvious, for then (14) reduces to
where y
the truth
de*
* +x
'
x
If
0.
S *
2
0,
* have here an example of a differential equation of the first order, handed to us by physics, whose general integral does not depend on an arbitrary constant, but consists of a unique function of alone.
We
MOTION OF A PARTICLE
where, however, the two
111
But in
all
signs are not necessarily the same. cases this last equation leads to (15).*
We
(16)
Equation
(12),
at the result
u =
~ + B cos (e setting
7).
and
where
(\7\ 1 }
now
is
Thus
(16) yields
=
M
1
ecos(0
7)
This
The path of the particle is given by Equation (17). the equation of a conic referred to a focus as pole and having the eccentricity e. The Case e < 1. If e < 1, the conic is an ellipse, and the
The
Orbit.
is
is
(1
e*)
by a and
.
6,
2
we have
h*
Md-e
)'
M Vf=T'
is
The
(19)
The area
(20)
of the ellipse is
TTOfc
=
(1
e )1
2
Th The
tion:
periodic time
is
A by
the
rela-
A = pr.
T2 =
is
Hence
(21)
* It
result
47r
--
tions, that
worth the student's while to follow through these multiple-valued funche may secure a firmer hold on the Calculus, even though the final is simple. Equation (15)
112
MECHANICS
Determination of the Constants of Integration. Let the body be = (a, 0) with an initial velocity projected from the point (r, 0)
VQ in
o.
ft
We will mention first a general formula. Let be the angle from the radius vector produced to the tangent. Then
\l/
component
v e of
By
becomes
(22)
h
this
is
vr sin ^,
and
the formula
we had
in
mind.
To determine
in the
/f>O\
form
/'-j
//j
\\
Hence
(24)
-^ ah*
(1
e cos 7),
e cos
7 =
an
Furthermore,
du
dO
Since
=
dr
fie -
h'
sin (Q
-y).
du
d6
__
__
dr
__
v cos
\f/
'
r 2 dO
h dt
we have
/rt/\
initially
(26)
M^ sin 7 T^ i
^n
COS p
,
-.
e sin
Vi\ il
COS p
From
(27)
(22),
VU Q
a sin
j8, /
i
cos 2
ft
i
^y^
2
(26),
and adding,
we
find
by
(28)
MOTION OF A PARTICLE
The evaluation
mined
;
113
is
is
now
complete.
By means
and
of (27), h
deter-
e,
and
(24)
(26) yield 7.
From
we
infer that
and
will
this equation contains the interesting result that the orbit be the following conic
:
i)
ellipse,
if
v%
<
=
ii)
parabola,
if
v\
Hi)
hyperbola,
if
v%
>
is
launched.
From
h2
(24)
(30)
=4 an
2
MO.
7r/2,
and so we
infer
from
A2
via*.
(30),
v*
=
Jj.
path be a
Conversely, conditions (30) and (31) are sufficient, that the For from (27) follows that cos 2 /3 = 0, and circle.
The result checks with the fact that the numer0. (29) gives 6 2 ical value of R, or w/*/a , is equal to the centripetal force, or
EXERCISES
1.
Show
that
if
r2
-rr
and
u =
dt
-, r'
then
114
2.
MECHANICS
It has
been assumed that the orbit is a plane curve. Prove be the case by means of a constraint, consisting of a smooth plane through 0, the point of projection, and the tangent to the path at that point. Use Newton's Equations, 17, (2), and
this to
is 0.
If two particles of masses w, 26. The Two Body Problem. m'j attracting each other according to the law of nature, and
acted on by no other forces, be projected in any manner, their centre of gravity, G, will describe a right line, with constant
1. Convelocity, or remain permanently at rest ; cf. Chapter IV, be the fixed point 0, and let the Let sider the latter case. be r, r'. Then the force of their distances of the particles from
mutual attraction
is
/ =
where
(r
r ')2>
is
On
mr =
Hence
r
,
m'r'.
_ m+
~m'
m'
r'
and so
f
(
-m\m + ml
'
-
2 \2 \
,)
m'.
with the force that Thus the particle m is attracted toward fixed at 0, and so the orbit of m would be exerted by a mass In particular, if m describes 25. is determined by the work of an ellipse, m will describe a similar ellipse with the same focus, being turned through an angle of 180.
Let The Inverse Problem to Determine the Force. Then a particle move in a plane according to the Law of Areas.
27.
= h r^ dt
T
ft '
and the component of the force perpendicular to the radius vector, Hence the particle is acted on by a central force, R, 0, is nil.
either attractive or repulsive.
From
Exercise
1,
25,
we have
MOTION OF A PARTICLE
Example. Let the path be an ellipse conic) with the centre of force at a focus.
(or,
115
generally,
more Then
any
u =
e cos
(07)
,
.
p
1
const.,
W* + U
and
d*u
p>
K~
The
mh*l
p
r*
force is, therefore, an attractive force, inversely proportional to the square of the distance from the centre, when r lies between its extreme values for this ellipse. But an arbitrary range of
<
is
<
<
0,
can be included
in
such an
ellipse,
and so
general.
EXERCISE
Show
centre.
that
if
the path
is
is
an
ellipse
28. Kepler's Laws. From observations made by Tycho Brahe, Kepler deduced the laws which govern the motion of the planets.
1.
the
sun according
to the 2.
sun
at
a focus;
3.
to the
The squares of the periodic times of revolution are proportional cubes of the major axes of the ellipses.
y
s Inferences. From Kepler's laws Newton drew the Consider a particular planet. From the following inferences. first law it follows that the force acting on it is a central force, since the component at right angles to the radius vector is nil.
Newton
From
first, it
follows from
116
It has
MECHANICS
been shown in
25, (21) that
T2
47T 2
-,
M
is
p,
or that
/*
is
the same
To sum up, then, Newton inferred that the planets are attracted toward the sun with a force proportional to their masses and inversely proportional to the square of their distances from
the sun.
From
tion.
If
here
it is
Law
of Universal Gravita-
the sun attracts the planets, so must, by the principal of action and reaction, the planets attract the sun. Let denote
Then
of Universal Gravitation
is
evolved
Any two
bodies (particles) in the universe attract each other with a force proportional to their masses and inversely proportional to the
win
The
factor
is
Its value
in c.g.s. units is
= 6.5X
10-*;
cf.
Appell,
l.c.,
pp. 390-405.
EXERCISES
1.
Show
that the
first of
the equations
dt 2
cf.
(1),
25
1,
3d
ed.,
1909,
MOTION OF A PARTICLE
on making the transformation
(11)
:
117
and employing
(2)
r* r
dt
Hence obtain
(16)
u =
2.
~ + B cos
h
v -,
(B
7).
Prove that
3. Show that the earth's orbit, assumed circular, would become parabolic if half the sun's mass were suddenly annihilated, the sun being assumed to be at rest. 4. A smooth tube revolves around one end in a fixed plane with constant angular velocity. A particle is free to move in the tube. Determine the motion.
the preceding question, an elastic string is made and attached to the end of the tube, determine the motion.
5. If,
in
6.
particle is attracted
proportional to the distance. Show that the path curve, and that it can be represented by the equations
x
Is
it
= A
cos (nt
+ a),
= B sin
(nt
a).
an
ellipse?
7. Show that a comet describing a parabolic path cannot remain within the earth's orbit, assumed circular, for more than
(2
8.
- 1-th
shell
describing
attractive force.
an Prove that,
elliptical
if
;
orbit under
all
a central
it
meet again at the same moment between the explosion and the collision, each piece
the pieces will explodes, and that after half the interval
will
be moving
118
MECHANICS
force,
10.
Show that a particle, moving under the action of a central cannot have more than two apsidal distances cf Appendix B.
;
.
circle,
the centre of force being situated on the circumference. Ans. The inverse fifth power.
11.
If
in
equal to the
long would
it
mean
inches apart,
how
take them to come together? How great would the error be if their mutual attraction were taken as constant?
12.
cannon
ball
is fired
vertically
ft.
How
cannon would
13.
it fall, if
Show
varying according to the inverse square, will in general describe a branch of a hyperbola with the centre of force at that focus
which
lies
What
It
is
the excep-
tional case?
29.
On
Matter
is inert.
cannot exert
cannot push or pull. It yields to force, acquiring in the direction in which the force acts we are thinkvelocity of a of virtue its inertness it ing particle. By possesses mass, which may be described as the quantity of matter which a body
contains.
a force;
Mass is measured by the effect which force produces on the motion of a body. We assume that force may be measured by a spring balance. If a force, constant in magnitude and direction, be applied to a body initially at rest, the body will acquire a certain velocity in a given time. If the same force be applied to another body, and if the second body acquire the same velocity in the same time, the two bodies shall be said to have the same mass. Thus different substances can be compared as to their masses and on adopting an arbitrary mass as the unit in the case of one substance, the unit can be determined in the
case of other substances.
MOTION OF A PARTICLE
119
It was proved experimentally by Newton that the forces with which gravity attracts two masses equal according to the above And so one is led to infer the physical definition, are equal. law that the weight of a body is proportional to its mass. This law affords a convenient means of measuring masses, namely,
by weighing.
In abstract dynamics, however (to quote from Maxwell), matter is considered under no other aspect than that under which it can have its motion changed by the application of force. Hence
any two bodies are of equal mass if equal forces applied to these bodies produce, in equal times, equal changes of velocity. This is the only definition of equal masses which can be admitted in dynamics, and it is applicable to all material bodies, whatever
they may be made of.* In Engineering it has become customary to define masses as equal when their weights are equal. We have here a question of a sense of values, and Maxwell has gone on record as declaring
unequivocally for the inertia property. To use weight to define mass is like saying that two lengths are equal when the rods by which we measure them have the same weight. Just as space
elementary impor-
XL VI.
CHAPTER
IV
into
two
i)
classes
i)
ii)
By
or
which
on m a force F z may have any magnitude and any direction whatever, in particular not be present at all, F 12 = 0. The particle And now we exerts a force on w 2 which is denoted by F2l
we mean
exerts
12
and
i.e.
F21
is
Fi2
FIG. 75
F21 =
0.
of
in
a plane.
The
immediate.
of
Denote the components of a vector force F along the axes coordinates by X, Y. Then
*i2
of
+X =
21
0,
F + 721 =
12
0.
Motion gives
Second Law
=
__
X + Xn + X
1
There are in
following
:
all
d*x
d2 x d2 x
~dfi
-^3
120
121
Xu
on the
adding these three equations together, the components right, arising from the internal forces, annul one an-
sum
,
of the
Ci
Xi remains
-*
Xl
+ X>2 + *
.
*r
In a similar manner we
tions in
infer,
three equa-
^
*
"eft
"^ "* 2
^
^ft 2
"^ "* 3
~~
~eft^
^~
* ? ~"~ * 3
'
Precisely the same reasoning shows that if, instead of three, we have any number, n, of particles, the internal forces annul one another in pairs, and thus we obtain the result
:
The left-hand sides of Coordinates of the Centre of Mass. these equations admit a simple interpretation in terms of the motion of the centre of mass of the system. The coordinates, (x y), of the centre of mass are given by the equations
:
x
(2)
V
If
m +
1
mn x n mn mn yn + wn
we denote
by M, then
rn k
xk
Hence we have
d*x a'Xk k
fc
_ ~M ~
l*l/
TUTM
~jfLi H/w
V 2*
j^
(1)
(3)
122
MECHANICS
for a particle of
These equations are precisely Newton's Second Law of Motion mass Af, acted on by the given external forces,
We
THEOREM.
as if
all the
The centre of mass of any system of particles moves mass were concentrated there and all the external forces
a third equation,
acted there.
is
now by
, .
(4)
^
is
one detail in the statement of the theorem We have written down requires explicit consideration. of the differential tbe motion, but we have not inteequations in coinwe them. If do not start the particle of mass grated
Remark.
There
that
cidence with the initial position of the centre of mass, it obviously cannot describe the same path. More than this, we must
Is this the same initial velocity (i.e. vector velocity). in its coincidence with the to insure enough always remaining centre of mass? The answer to this question is a categorical
give
it
Yes
cf.
Chapter
III,
17 and Appendix B.
have proved the theorem of the mass for a system of particles. In the motion case of a rigid body, we can think of the body as divided up into a large number of cells, each of small maximum diameter; the mass of each cell as then concentrated at one of its points, and the n particles thus resulting as connected by masslcss rods,
Generalized
Theorem.
We
of the centre of
after the
ticles
this auxiliary system of parthe theorem as above developed applies. Arid now it is intuitionally evident, or plausible, that the system of particles
will move when the
* It
is
manner
of a truss.*
To
in a
cells
manner
often necessary to use a truss, at some of whose vertices there are no think of minute masses attached at these points and acted on by gravity or by no external forces at all. The effect of these small masses is to in Equations (4). And now it follows from the the value of modify slightly theory of differential equations that the integrals of (4) are thereby also modified only slightly. Hence the physical assumption is made, that Equations (4) hold even when there are no masses at the vertices in question.
masses.
We may
123
body
is
the motion of the system of particles as n grows large and the cells small. And this is, in fact, true. But this is not a mathe-
matical inference
far
from
it
it is
We
to a Principle.*
PRINCIPLE OF THE MOTION OF THE CENTRE OF MASS. The mass of any material system whatsoever moves as if all the mass were concentrated there, and all the external forces acted there :
centre of
A)
The Glass of Water. Suppose a glass of thrown out of a third-story window. As the water falls, it takes on most irregular forms, breaking first into large pieces, and these into smaller ones. The forces that act are gravity and the resistance of the atmosphere, the latter spread out all
2.
Applications.
is
water
over the surfaces of the pieces. And now the Principle of the paragraph tells us that the centre of gravity moves as if all the mass were concentrated there and all these forces translast
ferred bodily
(i.e.
and
rest,
To determine
pressure, F, on the table. idealize the chain as a uniform flexible string, of length / = pi), and think of it as imand density p (hence of mass
We
Let s pinging always at the same fixed point, 0, of the table. be the distance the chain has fallen and let x be the height of
the centre of gravity above the table. Then the Principle of the Motion of the Centre of Mass gives the equation
:
"Principle" in Mechanics is well described in the words of Professor Koopthe Author's Advanced Calculus, p. 430): "According to the usage of the present day the word principle in physics has lost its metaphysical implication, and now denotes a physical truth of a certain importance and generality. Like but whether it is taken as a all physical truths, it rests ultimately on experiment physical law, or appears as a consequence of physical laws already laid down, does not matter."
nrian
(of.
124
MECHANICS
(1)
Now,
s
a
ld*s
'
I
I
G *
'
dt*'
Opl
falling
Fia. 76
On
OP
(s
we
obtain
(2)
1)
+
gps
F-
gpl.
Hence
(3)
F =
pv
2
,
or
F =
3gps.
This means that the pressure of the chain on the table is always just three times the weight of that part of the chain which has
already come to rest on the table. It appears, then, that F is made up of two parts, i) the pressure gps on the table, of that part of the chain already at rest; and ii) a pressure
(4)
P=
pv*,
due to the impact of the chain against the table. A Stream of Water, Impinging on a Wall. Suppose a hose is turned on a wall (or a convict !). To determine the pressure. We idealize the motion by thinking of the stream as hitting the wall at right angles, the water spattering in all directions along the wall and thus giving up all its velocity in the line of motion of the stream.
Dynamically, this is precisely the same case as that of the chain falling on the table, so far as the impact is concerned, and hence
the pressure
is
given by (4)
PA
== ov
FIG. 77
*
125
fire
engine
ft.
is
able to send a 2 in. stream to a Find the pressure if the stream is Ans. 541 Ibs.
is
The Crew on
the River.
The crew
Ob-
how
it
moves, and
why
it
moves
as
it
does.
decide to isolate ?
EXERCISES
1.
If
man
shell
table,
how
If
in its flight,
3.
what could
yo.u
from a gun on the moon and exploded say about the motion of the pieces ?
air tight
goose
form
less
4.
is
scales.
box which
rests
on plat-
more
or
or the
same?
water is placed on some scales. A cork a string tied to the bottom of the pail. by
Do
same?
6. A man, standing in the stern of a row boat at rest, walks forward to the prow. What can you say about the motion of the boat?
6.
When
the
man
stops at the
prow
and man
will
7.
Explain why.
heavy string is laid over a smooth cylinder, horizontal, and kept from slipping by holding one end, A, fast, the part of the string from A up to the cylinder being vertical. The part of the string on the other side of the
uniform
axis
flexible
cylinder is, of course, also vertical, its lower end, /?, being below the level of A, and the whole string lies in a vertical plane per-
The
string
is
released
from rest. Determine the motion, there being a smooth guard which prevents the string from leaving the upper side of the
cylinder.
8.
If,
in Question 7, the
difference in level
between
A
is
and
8
ft.,
is
ft.,
and
if
A up
to the cylinder
126
MECHANICS
of the string
when
it
the cylinder.
10.
ported a
The sporting editor of a leading newspaper recently new stroke which a certain coach had developed,
re-
the
advantage of which was that it gave an even motion to the shell and avoided the jerkiness 01 the old-fashioned strokes. Examine this news item.
3.
The Equation
of
Moments.
moment
namely,
(1)
F about
the origin,
xY-yX.
Consider a system of particles acted on by any external forces whatever, and interacting on one another by FIG. 78 forces that are equal and opposite, but are now assumed each time to lie in the lino joining the two particles in question. Moreover, the particles shall Begin with the case of plane. three particles, as in and 1, write down the six equations that
lie
in a fixed
express
of
Newton's Second Law Motion for these particles.* Next, form the expression
:
FIG. 79
m
and compute
(x,
its
r
Y 12
Y l
Vl
X,)
(x,
Vl
12 )
'+ (x,
Yn -
y,
Xn
).
The parentheses
F 13 about
Now,
represent respectively the moments of F D F 12 the origin. do the same thing for the particle ra2 and finally, for ra 3
,
On
it
is
* It is important that the student do this, and do it neatly, and not merely gaze 1 and try to imagine the three not printed. at the three equations printed in He should write out the full equation derived below from these, neatly on a single line, and then write the other two under this one.
127
of
the internal forces about the origin destroy one side only the sum
the
moments of the applied forces. n > 3 particles, m w2 are thus led to the and we same,
there are
lt
mn
the procedure
is
THEOREM OF MOMENTS:
B)
d*Xk
k
- yA).
refrain from writing down the corresponding theorem in three dimensions, because we shall have no need of it for the
We
present.
Let the 4. Rotation about a Fixed Axis under Gravity. system of particles of 3 be rigidly connected, and let one point, O, of the trass-work be at rest, so that the system rotates about For example, take the case of a uniform rod, one as a pivot. end of which is held fast, and which is released from rest under Divide the rod into n equal parts, x gravity. and concentrate the mass of each part, for defiWe thus niteness, at its most remote point. have a system of n particles, and we connect them rigidly by a massless truss-work as shown
in the figure.*
We
ticles.
are
now ready
to
compute each
side of
Equation B)
Let r
Then
y
x =
r cos
0,
Fia. 80
r sin
t,
0,
where
to
t.
but
r is constant
with respect
Hence
dx
-rr
(2)
dO
-7T,
Sill
dt
dt
$-all
We
generality,
by mere
differentia-
tion, i.e.
purely mathematically,
<L
(3)
di\
( x d-i
~di
4*\
dt)~
d*y
d?
I.
128
MECHANICS
of
and we proceed to compute the parenthesis by means We find tions (1) and (2).
:
Equa-
dy
In the present case we have
dx
~dO
d ( dyk Xk dt\ dt
for r k does not
dx k \
yk
-~
d*8
rt
~dt)
~dT
0.
Hence
here appears
is
Thus the
Moments
reduces to
the expression
k th particle,
Turning now to the right-hand side of B) we see that the m/t, yields a moment about equal to the quantity mk grk sin 0, and so the sum in question becomes
]
t
- mk grk sin 0,
or
(2) t
rk)g
sin
0.
But
where h
is
m k rk = Mh
the distance from
particles.
(?,
of the
system of
(6)
Hence,
I^ =-MghsmO.
2
This
is
Pendulum Motion,
Chapter
III,
20
^\
*
d2 6
Moments
Author's Introduction
of inertia for such bodies as interest us here are treated in the to the Calculus, p. 323.
129
of
a simple pendulum
m
7
or
(8 2 )
I
=
j,
where
MW,
More precisely, what we mean by the last statement is this. Let a simple pendulum be supported at 0, let its length be k*/h, and let it be placed alongside the rod, the bob being at a point If now both be released from rest at the same distant Z from O.
instant, they will oscillate side
by
side,
other.
As n grows
larger
and
rod weighted with the n particles comes nearer and nearer to the actual rod, dynamically. This is not a mathematical statement. It expresses our feeling from physics for the situation
our intuition.
rod
is
And
we
whon we say that the motion of the actual by the motion of the auxiliary rod, a new physical postulate. The result is, that the
so
pendulum
3
of length
-iL -?/
EXERCISES
Do particles, and proceeding to the limit. not try short cuts by attempting to use in part the result of the exercise worked in the text.
1.
A rod
long and weighing 30 Ibs. carries a 20 Ib. weight Ib. weight at the other. It is supported middle point. Find the length of the equivalent simple
10
ft.
and a 30
pendulum.
2.
Ans.
30
ft.
Equal masses are fixed at the vertices of an equilateral If it triangle and the latter is supported at one of the vertices. be allowed to oscillate in a vertical plane, find the length of the
equivalent simple pendulum.
130
3.
MECHANICS
A
rigid
Ibs.
ing 12
is
uniform circular wire * 6 in. in diameter and weighhas a 4 Ib. weight fastened at one of its points and about
its
free to oscillate
centre in
its
own
plane.
Find the
pendulum.
is
gon and connected rigidly by a weightless truss. The system pivoted at one of the particles and allowed to oscillate in a Find the length of the equivalent vertical plane under gravity.
6.
simple pendulum.
Generalize to the case of
n equal
Consider an arbitrary lamina, 6. The Compound Pendulum. Let it be supported at a or plane plate of variable density. arid allowed to oscillate freely in its own plane, assumed point This is essentially the most general vertical, under gravity.
To determine the motion. Divide the lamina up in any convenient manner into small pieces and concentrate the mass of each piece at one of its points.
compound pendulum.
Connect these
particles
with
one
another and with the support at by a truss-work. The auxiliary system can be dealt with by the Principle of
Moments.
COS
6k,
Set
%k
^k
yk
rk sin
0*.
Then
Now, draw a
FIG. 81
and
the
0.
particles,
axis of
x by
Then
where
ct k
varies with
fc,
but
is
Hence
d0k
dt
*
= "
de
dt'
d 2 Bk
=
~~
d*0
dt*
dP
By
a wire
is
131
Moments,
3,
n\
(1)
inertia of the
system of particles
The right-hand
(2)
side of B),
3,
can be written
"
2)
last
The
sum has
by
the value
are denoted
(x, y).
h sin
and
(3)
(2)
becomes
d~n
M0/i sin
6.
On
(4)
we have
/5y[=is
JfffAsinfl.
This
system of particles oscillates a simple pendulum. As we allow n to increase without limit, the maximum diameter of the little pieces approaching 0, it seems plausible that the motion will approximate more and more closely
like
this consider-
ation leads us to lay down the physical law, or postulate, that Equation (4) holds for the compound pendulum, where 7 and
now
Remark.
We
an
pendulum
But
a
this is
Conceive
block
of
you
please.
Mount
axis.
it
on two
this block will obviously oscillate exactly as a plane lamina perpendicular to the axis would, if the mass of the actual block were projected parallel to the axis on a plane at right angles to the axis.
a horizontal
Now
132
MECHANICS
The above "obviously" is not to be taken mathematically, It is true that when is a new physical law, or postulate. we come to treat the general case of motion in three dimensions, this postulate will be merged in more general ones.
but
EXERCISES
compound pendulum
Find the length of the equivalent simple pendulum when the is one of the following.
its
own
I I
plane
Ans.
2.
circular wire,
1,
Ans.
Ans.
= fr. = 2r.
3.
Question
when
=
=
fr.
4.
Question
2,
when the
axis
is
Ans.
6.
6. 7.
-Jr.
A rectangular lamina, about a side. A square lamina, about a vertex. A triangle, about a vertex.
Continuation.
6.
Let
7
be the
Mfc
moment
of inertia of the
parallel axis through the centre of gravity, (?. By the theorem of 10 the moment of inertia about the actual axis will be
pendulum
is
soon from
to be
(6)
^A
2
-
The question
what other
from
arises:
What
(i.e.
x,
* ,_* + x
(7)
x*
Ix
Jk
0,
133
h*
Ih
+k =
0.
(7)
is
h.
_i ^I
ft
j,
* ~
2
We
FIG. 82
THEOREM.
lation is the
The locus of the points 0, for which the time of oscilsame consists of two concentric circles with their centre
y
and
k2
-rfi
EXERCISES
1.
Draw two
Show
is
concentric circles about G, of radii h and k*/h. Z, of the equivalent simple pendulum cor-
on one of these circles responding to an axis through a point obtained by drawing a line from through G, and terminating it whore it meets the other circle.
This theorem
2. is
due to Huygens.
Show
that the locus of the points of support, for which is least, form a circle with G as centre and
7.
Kater's Pendulum.
value of
T =
9
students of physics and mathematics. The chief error in the result arises from the error in determining I.
is
familiar to
all
The bob
is
To
could
attain
not sensibly a particle and the string stretches. greater accuracy, Kater made use of Huygens's
6,
Theorem,
be
Ex.1, constructing a compound pendulum that It consists essentially of a massive rod, or bar, provided with two sets of adjustable knife-edges. These
reversed.
edges
lie
in
two
parallel lines,
of gravity,
(?,
is
134
MECHANICS
The
situated in their plane, at unequal distances, h and A', from them. knife-edges are now so adjusted experimentally that the
as
period
same
when the pendulum oscillates about the one pair is the when it is reversed and allowed to oscillate about the
Since
I
other pair.
+ h',
the determination of the length of the equivalent simple pendulum can now be made with great accuracy by measuring the distance between the knife-edges. Indeed, the accuracy in thus deter-
mining g
to the
is
now
buoyancy
pendulum due
to
changes in temperature, and the give of the supports have to be considered. For an elaborate and interesting account, cf.
98
et seq.
Atwood's Machine.
An Atwood's Machine
consists of a
pulley free to rotate about a horizontal axis, and a string passing over the pulley and carrying weights, and m, at its two ends. It may be used to measure the acceleration of gravity.
M+
Our problem is to determine the motion of the system. The "system" which we choose to isolate is the complete system of pulley and weights, the mass of the string being assumed
This is not a rigid system, but still, if we replace the pulley by a system of particles rigidly connected, the internal * forces of the complete auxiliary system will satisfy the hypothesis
negligible.
of
3, and thus the Equation of Moments will hold. For the auxiliary system of particles due to the wheel the
B),
contribution to the left-hand side of the Equation of Moments, 3, becomes as in the case of the compound pendulum
:
where I denotes the moment of inertia of this system about the axis, and B is the angle through which the wheel has rotated.
* Consider a short interval of time in the duration of the motion. In the auxiliary system, let each vertical segment of the string be fastened to a particle Then is it near the point of taiigericy of the string in the actual case. plausible physically that the motion of the auxiliary system during this short interval differs but slightly from that of the actual system. Hence we may assume that the force of the string always acts at the points of tangency with the wheel, and neglect the rest of the string which is in contact with the wheel. But this is a new physical law.
135
in
Let the radius of the wheel (more precisely, of the groove which the string lies) be a. Observe, too, that
yl
const.
+ a0,
const.
aO.
to the left-
(M + m) a 2 -^
side of
+ Ma -^
2
The right-hand
(3)
B) reduces to
(M + m) ga
:
Mga =
mga.
Thus B) becomes
dt
And now we assume, This, for the auxiliary system of particles. physically, that the limit approached by the motion of the auxiliary
system
is
i.e.
holds for the actual system. Let s denote the distance the weight and rider have descended. Then s = aO, and from (4) it follows that
mga'
(5)
dt 2
(2M
we
2 m) a
On
(6)
=
I
(2M
+ m) a
Corresponding values of
Thus Equation
(6) is
:
~
II
0.
If
is
it is
and so we
have
two independent
unknown
136
MECHANICS
EXERCISES
the text.
1.
In working these exercises use the method, not the Begin each time by drawing a figure.
result,
of
Suppose that the wheel is a uniform circular disc weighing Ibs., and that 5 Ib. weights are fastened to the two ends of the string. What will be the acceleration due to a 1 oz. rider?
10
is a hoop, i.e. a uniform masses of the spokes being negligible; and show that the acceleration of the rider does not depend on the and m. radius, but only on the mass of the hoop, and
2.
Work
3.
4.
6.
Prove the assertion in the text about the taking on different values when m is varied.
6.*
coefficient of g's
How
case, in order
not to slip?
Consider 9. The General Case of Rotation about a Point. an arbitrary rigid body in two dimensions, acted on by any forces in its plane, and free to rotate about a point 0, i.e. about an
axis through
motion
is
say,
its
B)
/ -jT2
5J Moments about 0.
The Principle is rendered plausible by dividing the actual distribution into small pieces, as in the example of the cornpound pendulum and the Atwood's machine, and observing that
the Principle
is
The
is
proached by the motion of the auxiliary system defined by Equation B) of the present paragraph.
And
thus
we
are led to lay down the physical postulate that this is the motion of the actual system. Equation B), then, is an independent physical law, made plausible by the mathematical considerations
set forth above,
The
tion
Whenever gravity acts, the contribuEffect of Gravity. of this force to the right-hand side of Equation B) can
cf.
* This problem
the Calculus
;
is more difficult than the others, and is essentially a problem in 8. the author's Advanced Calculus, Chapter 14,
137
moment
being the attraction of gravity on a single particle of mass equal to the mass of the entire body and situated at the centre of gravity
of the body.
This
is
true in the
of motion,
of the
body
is
permanently at rest.
Here, again,
physical postulate.
EXERCISES
1. turn table consisting of a uniform circular disc is free to rotate without friction about its centre. man walks along the rim of the table. Find the ratio of the angle turned through
by the
start
2.
by the man,
if
man and
table
from
of the table,
10.
the man walks in along a radius the system not being, however, initially at rest.
of
Inertia.
Moments
The moment
of
inertia
of
the
simpler and more importanjb distributions of matter are determined by the methods of the Integral Calculus cf. for example the author's Introduction to the Calculus, p. 323, and the Advanced
;
Ml 2
:
5 o
TI/T
2.
A rod A
A
2a about
its
midpoint
:
3.
circular disc
about
its
centre
5 &
4.
5. 6. 7.
circular disc
about a diameter
:
Mr
:
~T~~'
A square about its centre A square about a side; cf. A scalene triangle about
where h denotes the
A A
i * r * sphere about a diameter
i
f
Example
a side
:
Ma
2
.
1.
altitude.
:
8.
9.
cf.
cube about a
5.
line
Example
is
is
contrary
stated.
138
MECHANICS
The moment of inertia of any distribuan arbitrary axis, is equal to the mothe centre of gravity,
A GENERAL THEOREM.
Mh
begin by proving the theorem for a system of particles. axis be taken as the axis of z in a system of Cartesian and let the second axis be the axis of z' coordinates, (x, y, z)
will
We
Let the
first
in a
Then
7
2m
k (x,?
+
x'
2 2/* ),
= =
2 w* W +
y'
yi
).
Since
x
it
+ x,
+ y,
follows that
)
2
k .
2x
mk xi + 2y
is
<k
yi-
The
2m
It
k x't
o,
2
left,
remains merely to interpret the terms that are is proved for a system of particles.
and
matter,
we have a body consisting of a continuous distribution of we divide it up into small pieces, concentrate the mass
piece at its centre of gravity, form the above sums, r their limits. shall have as before 2mkX t = 0,
of each
and take S mk yi =
We
0,
and hence
lim
n-oo
2
j7
*(**
2/t )
= Km
n = oc
or
since these limits are
by
definition the
moments
continuous distribution.
139
To
find the
moment
/
T
,.
Here, 7
and h
11.
= %Mr 2
r.
Hence
Q = f Mr 2
Let a rod be clamped at its and suspended, the rod horizontal and
Let the rod be displaced slightly in its horizontal plane, the wire remaining vertical, arid then released. To determine the motion. The forces acting on the rod amount to a couple, due to the
torsion of the wire, and the moment of the couple is proportional to the angle through which the rod is displaced such Thus the Principle of Moments, is the law of elasticity. 9,
yields in this case the differential equation,
where /
2 = Ma
is
the
moment
and
is
the
of Simple
Harmonic Motion,
T =
2*
the same, no matter what the initial displacement may have been, provided merely that the distortion of the wire is not so
great as to impair the physical law above stated, and provided
damping
12.
is
neglected.
No
Point Fixed.
Let a rigid
let it
plane lamina be acted on by any forces in its plane, and move in its plane. To determine the motion.
The centre of gravity will move as if all the mass were concentrated there and all the forces were transferred to that point
;
1.
It
PRINCIPLE OF MOMENTS.
The lamina
of gravity were held fast and the same forces acted on the lamina as those applied in the actual case
y
(1)
=
^"77/2
S Moments about
(?,
140
MECHANICS
where I denotes the moment of inertia about the centre of gravity, G; 6 is the angle that a line fixed in the lamina makes with a line fixed
in the plane, and the right-hand side is the the forces about G.
Proof.
sum
of the
moments
of
Consider
first
Let
(x,
(2)
+ a,
77
+ y,
and
0,
rj k
0,
and hence,
too,
Remembering that
dt\dt
we
see that Equation B),
3,
here becomes
Because
xk
= &
x,
yk
ri k
+
:
y,
2 7 fen rjkXk) +
Since
141
follows that
On
(5)
(4),
there remains
*
contained the proof of the theorem for a For, the left-hand side reduces to the particles.
is
In this equation
system
of
(1), since the distance of the point (&, ?/*) from of centre the gravity, G, does not change with t; and the righthand side expresses precisely the sum of the moments of the
left-hand side of
the usual way, laying down a new physical postulate to the effect that Equation (1) shall hold for all rigid distributions of matter
in a plane.
A hoop* rolls down a rough inclined plane Determine the motion. The forces are: the force of gravity and the reaction of the Let the latter force be resolved into a normal complane. ponent, R, and the tangential force of fric13. Examples. without slipping.
Then, for tion, F, acting up the plane. the motion of the centre of gravity, we shall
have:
-.
The second equation
merely
(2)
tells
FIQ 84
for the
us that
R =
Mg cos
a,
a fact that we could have guessed, since the centre of gravity always remains at the same distance from the plane. However, let us formulate the second equation, and prove our guess right.
Let y denote the distance of the centre of gravity from the plane.
* pipe, the thickness of which is negligible, when placed on the plane with its The two problems are dynamically axis horizontal, would move in the same way. identical.
142
MECHANICS
Then
But y =
this
a,
equation
of the hoop,
we
write
down Equa-
(4)
/-^
is
aF,
= Ma 2
Since there
(5)
no
slipping,
5
a0,
where, for convenience, we take as 6 the angle that the radius drawn to the point of contact with the plane at the start has
at the start.
now be
Ma -JTJ = Mg sin a
(6)
Ma*<^
On
eliminating
aF.
find
or
(8)
^=
(2)
sina
Hence it appears that the centre of the hoop moves down the plane with just half the acceleration it would have if the plane
were smooth.
Equation
pR
F ^
ascertain
ju,
R.
for
To
(6) for
of friction,
the coefficient
a, solve Equations
and substitute
^y s n a
*
'
143
M^ cos a,
^
2M
.
. .
tan a
2ju.
EXERCISES
1.
Show
that,
if
~ sm a, = gt
.
~- sm a, = at*
.
v2
2.
0s sin a.
Show furthermore
that
.
at
TJ<&&
sin a,
at = Tsin a,
.
4<z
W
3.
*L
gj
na
Show
that
^5
4.
^ am
5g
d*s
a,
50 ^J^Bina.
.
tan a
6.
JJLI.
Make
a complete study of a
14.
billiard ball
is
struck
full
by The
the cue.
To
forces are:
downward
at the
and the reaction of the billiard table, which vertical a component, R, and a horizontal component, F. yields Let s be the space described by the centre of the ball, and 6, the angle through which the ball has turned.* The Principle of the Motion of the Centre of Gravity, 1,
centre of gravity,
yields the equations
:
__
(1)
dt*
F
FIG. 85
begin each new problem, as here, coordinates used in setting up the too, to note at the same time any
= pR.
R = Mg
of prime importance that the student figure showing the forces and the It is well, differential equations of the motion. auxiliary relations, as in the present instance, F
is
* It
by drawing a
144
MECHANICS
Principle of Rotation about the Centre of Mass,
:
The
12,
r 7
d* e
= nw ^
is
T I
slipping,
F =
(1), (2),
MB.
appears that
r
From Equations
/A \
and
(3) it
&S
i
(4.)
s=
'
~
d< 2
2a
:
The
.
t>
ngt,
2
s
vl
t-
,g
\
and
Thus as the
slowly, while the speed of rotation steadily increases. Finally, in. when This of will set takes the rolling place velocity pure
the point of the ball in contact with the table is nil. Now, the velocity of this point of the ball is made up of two velocities, namely, i) the velocity of translation, or the velocity the point
would have if the ball were not rotating, i.e. v, as given by (50; and ii) the velocity due to rotation, or the velocity the point would have if the ball were spinning about its centre, thought of
The latter is a velocity of ao> in the direction opposite Thus the veto the motion of the centre, and is given by (5 2 ). the table is ball in contact with the of of forward the locity point
as at rest.
(6)
ow.
is
positive,
and
(7)
oo>
0.
145
given
by the equation
The
is
Finally,
(11)
1
|>0, 7
T la
2'
EXERCISES
1.
and the
2. is
Solve the same problem in case the table is slightly tipped ball is projected straight down the plane.
the last problem with the modification that the ball projected straight up the plane.
15.
Work
Continuation.
At the end
of
the stage of the motion just discussed, the ball has both a motion of translation and one of rotation, the point of the ball
in contact with the table being at rest. If from now on the force exerted by the table on the ball consists solely of an upward component R and a tangential component F, the latter force will
vanish,
s
and the
oQ,
without slipping. For, rough enough to prevent slipping. Then and since equations (1) and (2) still hold, we have
ball will continue to roll
is
:
Hence
both
vanishes,
0, too.
be acanswer is, that the reaction of the table is not merely a force, with components R and F. but, in addition, a couplej the moment of which
counted
in practice the ball will slow up. is this to air if is resistance of the the for, negligible? The
But
How
146
MECHANICS
will
we
denote by C.
of the centre of gravity; thus Equations (1), 14, remain as before. But Equation (2) now becomes
O
FIG. 86
(13)
Furthermore,
(14)
Hence
dP
Since
7 Ma'
dt*
7 Ma*'
la
is
up
gradually.
EXERCISES
1.
If
ft.
sec.
the centre of the ball was moving initially at the rate of and if the ball stops after rolling 18 ft., show that
C = IMa.
2.
If
rolled
I ft.,
the initial velocity of the centre was V Q and if the ball show that C is proportional to the initial kinetic energy
rolled.
ON ROUGH STEEPLY IN16. Further Examples, i) HOOP CLINED PLANE. Suppose, in the Example studied in the text 1 of What will the motion then 13, that a does exceed tan" 2/*. released from rest? the be, hoop being Equations (1), (2), and (4) will be as before. But now (5) is
replaced
(i)
all
by the equation
p=
the friction
find
:
&
On
eliminating
now
and R, we
<7(sin<*
(2)
-= =
dt 2
cos a.
constant,
s -
if
a (sin a
-p cos
these differential equations can be written particular, it is seen that the ratio of s to 6
:
u cos a) -
'-
o(tan a cot X
,,
-, 1).
147
last
1
>
when a
r
parenthesis has the value 1 when tan a = 2/z, and is larger. Thus the motion is one in which a cir,
_
cle of radius
a (tan a cot X
1)
and centre at the centre of the hoop rolls without slipping on a line parallel to the plane and beneath it. We have here an illustration of the general theorem that any motion of a lamina in its own plane can be
realized
FIG. 87
by the
rolling
cf.
Chapter V,
4.
First,
draw a
:
figure representing the forces and the coordinates. The three equations of motion thus become
(3)
FIG. 88
"77/2 U/t"
= a^ s n
*
"~
a^ cos
x
five
a cos
0,
a sin
6.
equations determine the five unknown functions x, y, time being the independent variable ; or they deterthe 0, R, S, mine five of the variables x, y, 0, R, /S, t as functions of the
These
sixth.
Eliminate
72,
S between
>
the
first
three equations:
<
5>
'.dx
2
.
-^
dO
Ma cos ~
:
Afgra cos
6.
From
a cos
dt
-jr,
dt
d x
148
MECHANICS
(5)
This differential equation can be integrated by the device of multiplying through by 2d0/dt and then integrating each side with respect to t dO - 30
:
dt
Since
d
it
follows that
2
fdO\ (dt)
=-2H s
is
30
The constant
conditions.
If
of integration, C,
is
determined by the
6
initial
the ladder
=
sin
and
a
=jj
Zd
Hence,
finally,
+ C.
*>
To find where
'-(-..-*.)
the ladder will leave the wall.
This question
is
answered by computing
E = R
R and
setting
2
-
it
M d2 x =
--trr
at
/>d Ma smO-jrz at
2
2
,
(8)
2 sin
a).
Hence
2 sin a
:
0.
Let
ft
<
149
here complete
the wall and slides along with the lower end in contact with the floor. But suppose a person is unwilling to trust his intuition
and says:
still
showing that
in merely a certain value of 0. The ladder might remain in contact with the wall, R increasing as the ladder
R =
for
continues to slide."
The
The
objection can be met as follows. Think of the upper end of the ladder as provided with a ring that slides on a smooth vertical rod. Then the ladder will not
leave the wall.
How
about
;
in this case?
Formula
6
(8)
now
sin" 1 (| sin a), and so the vertical rod has to pull on the ladder instead of pushThis proves that our intuition was correct. ing.
holds clear
down
to the floor
but
R <
when
<
The Time.
From Equation
(7) it
appears that
(9)
Vsin a
sin 6
On making
the substitution
sin
6,
the integral goes over into an Elliptic Integral of the First Kind, and can be treated by well-known methods; cf. the Author's Advanced CakuluSj Chapter IX.
iii)
coin
is
released
from
rest
with one point of the rim touching a smooth horizontal table. To determine the motion.
The
Gravity,
reaction,
Thus the
centre of gravity of the coin Let its height above the table be de-
noted by
y.
Then
become
(10)
=-aRcosO.
at*
is
osinfl.
150
MECHANICS
eliminating
2 (fc
On
(12)
R and
(Zf
y we find
a 2 sin
+a
cos 2 0)
^-
cos
(9
(~f) \ttf /
=-
ag cos
0.
This differential equation comes under a general class, namely, those in which one of the variables fails to appear explicitly.
The
is
to introduce a
new
variable,
And this can be done here. But in the present case there is a short cut, due to the special form of the differential equation. It is observed that, on multiplying the equation through by
2dQ/dt the left-hand side becomes the derivative of a certain function with respect to t so that the equation takes on the form
y y
:
(13)
On
t,
(F
+o
cos 2 0)
\Gfv /
=-2ag sin 9 +
C.
To
and
(14)
determine
C make
2
use of the
initial conditions,
dO/dt
a.
Thus
2
(A;
+a
cos 2
^{^)
:
2a0
(sin
sin 0).
table
the coin
when
it falls flat
on the
_ ~
2ag sin a
But here
is
= 0. initially in contact with the table remains in contact till This is plausible enough physically; but in this gunws, is there
not an
appreciable admixture of unimaginativeness and the question which the moron so frequently asks: "Why shouldn't it?" The angular velocity dd/dt of the coin is steadily increasing,
as
we
increase to such
and from (14). May it not an extent that the lowest point in the coin may
151
down?
mathematics sits as judge over the situation. actual the problem by one equivalent during the early Replace of the stage motion, and see whether this stage lasts through to
It is here that
Let the lowest point of the coin be provided with a on a smooth horizontal rod. Then the coin will that slides ring fall as we guessed. Compute now the reaction, R. The test
the end.
is:
it
Does
We
leave
EXERCISES ON CHAPTER
1.
IV
homogeneous
on a rough inclined
it rolls,
Will
it slip
as
or will
it
without slipping? Ans. It will slip if the angle of inclination of the plane 1 is greater than tan" (|M)-
2.
for
homogeneous spherical
shell
(material surface).
3.
is
released, just touching the cloth of the billiard table. far will it go before pure rolling sets in ?
4.
How
wound round its circumference. end of tho string is fastened to a peg, A, and the disc is released from rest in a vertical plane with its centre below the Show that the level of A, and the string taut and vertical. centre of the disc will descend in a vertical right line with twocircular disc has a string
The
free
The
problem
is
laid flat
horizontal table
the string
is
carried over a
edge of the table, and a weight equal to the weight of the disc The system is released from is attached to the end of the string. rest, the string being taut and the weight hanging straight down.
Show
6.
7.
is
three-fourths that
of gravity.
last question.
152
Discuss in
full
MECHANICS
the case that the rotation of the ball
is
in such
a sense that the ball moves down the plane faster than if it had not been rotating.
8.
is
it
would
when
the rotation
billiard ball is placed on a billiard table inclined to the 9. horizontal at an angle a, and is struck full by the cue, so that it starts off straight down the plane without any initial rotation.
straight
11.
the ball
is
so struck that
it
starts
man
table,
how
12.
table.
A plank can rotate about one end, A man, starting from the other
Determine the motion.
pivot.
13. A smooth tube, the weight of which may be neglected, can turn freely about one end. A rod is placed in the tube and the system is released from rest with the rod horizontal. Deter-
end of the string and the system Determine the motion. Consider first the case in which the string leaves the axle from the top then, the case that the string leaves the axle from the bottom. In each case, the segment of the string between the axle and the pulley shall be horizontal and at right angles to the axis, and the part below the pulley, vertical.
to the lower
is
A weight is attached
15.
The
is
above the track and midway between the axles, which are 8 ft. apart. The coefficient of friction between the wheels (when they are locked) and the track is ^-. If the car is running at the rate of 30 m. an h., in how short a distance can it be stopped by
5
ft.
applying the brakes to the rear wheels only? brakes are applied to the fronfr wheels only?
How
far,
if
the
153
16. A uniform rod is suspended in a horizontal position by two vertical strings attached to its ends. One string is cut. Find the initial tension in the other one.
17.
it
hoop
is
released.
Find whether
will slip.
18. A uniform circular disc, of radius 1 ft. and weight 10 Ibs., can rotate freely about its centre, its plane being vertical. There is a particle weighing 1 Ib. fixed in the rim, and a fine inextensible weightless string, wound round the rim of the disc, has a weight of P Ibs. fastened to it. The system is released from rest with
the
freely at the
is
weight at the lowest point and the other weight hanging same level. How great may P be, if the 1 Ib. weight not to be pulled over the top?
1 Ib.
19.
billiard
ball
rolls
in
a punch bowl.
Determine the
motion.
20.
slightly displaced,
gravity.
21.
uniform rod is released from rest, inclined at an angle, with its lower end in contact with a rough horizontal plane. Will it slip at the start? Determine the motion.
22.
packing box
Will
is it
sliding over
tip
an icy
side walk.
It
comes
to bare ground.
up?
CHAPTER V
KINEMATICS IN
TWO DIMENSIONS
1. The Rolling Wheel. When a wheel rolls over a level road without slipping, the nature of the motion is particularly accessible to our intuition, for the points of the wheel low down move slowly, the point in contact with the ground actually being at rest for the instant, and it is much as if the whole wheel were pivoted at this point and rotating about it as an axis. This is, in fact, precisely the case, the velocity of each point of the wheel
at the instant being the same as if the wheel were rotating permanently about that point. If the wheel is skidding, it is not so easy to see that a similar situation exists, and yet it does. No matter how the wheel is moving, provided it is rotating at all, there is at each instant a
away
it
may
is
This point
be), about which the wheel rotates called the instantaneous centre.
To prove
plane.
It
lation of the
we will begin by giving a general formuproblem of the motion of any plane lamina in its makes the problem more concrete to think of an actual
this assertion,
But we lamina, like a disc or a triangle or a finite surface, S. are really dealing with the motion of the whole plane, thought of as rigid.
described
mathematically as follows.
Draw a
moving plane;
i.e.
think
X
FIG. 90
Assume
155
/)-axes.
of the
Then the (, ^-coordinates of an arbimoving plane are connected with the (x, y)same point by the relations
:
x y
XQ
2/0
0,
0-
The
as 0',
The how
moving plane is known when one point, the orientation, as given by 0, is known. motion may, therefore, be completely described by stating XQ, 2/0) vary with the time; i.e. by saying what funcposition of the
is
known and
tions x
2/o>
are
Wo
shall assume at the outset that these functions are continuous and possess continuous derivatives of the first order. Later, it will bo dosirable to restrict them further by requiring that they have continuous derivatives of the second order.
EXERCISE
Express
tions
and
rj
in
terms of x and y
;
i)
geometrically,
ii)
analytically,
:
A)
for
77.
The formulas
f
rj
are cos
= =
(x
(x
XQ) sin
+ +
(y (y
2/0)
sin
0,
0.
y Q ) cos
The Instantaneous Centre. Let P be a point fixed in the mark it with a dot of red ink on the sheet of Lot the coordinates of P be (x, y). Then they are deterpaper.
2.
moving plane
mined as functions
nates of
TJ
of t by Equations A), (, rj) being the coordiand with reference to the moving axes. Of course,
not
are constants with respect to the time, for the red ink dot does move in tho paper it moves in space.
The vector velocity, v, of P in space can be determined by moans of its components along the axes of x and y which are fixed
y
in space,
Chapter
III,
15
-^
-,
-^-
156
MECHANICS
These derivatives can be computed in terms of the known functions (1), namely, x y Q 0, and of their derivatives, by means of Equations A). Thus
,
dx
(2)
___
dx Q
dy
dt
dy*
dt
The parentheses
to
-(yHence
dx
dt
XQ.
_ ~
~dt
dy
dB
These equations express the components of the vector velocity v of the point P along the axes fixed in space, in terms of the
coordinates
(x, y) of
(1).
New
Notation.
(4)
_dx ~
'
=
dt*'
etc.
Thus
final
form
B)
y
= =
z
2/0
(y
(*
+
We
2/0) 0>
#o) *
any) of the
body
0,
:
or
A
is,
point that x
is
"at rest"
if its
and y
velocity or
0.
(5)
= x = y +
(y
(*
j/
) 6,
6.
z n)
157
for the
unknown x
C)
- -
Xn
A
THEOREM.
At any
6 is not 0, there is
one and only one point of the body at rest. This point is called the instantaneous centre, and
(
its
coordinates^
xu
!/i)i
If 6
=
is
no point
of the
body
is
there
others.
When
are
0,
all
points of the
same
direction
of translation.
EXERCISES
1.
Show
moving
^o
si
6
a
"
?y n
cos &
1/1
x Q cos
+
^
2/0
sin 6
2.
A
A
circle rolls
is
on a
line
without slipping.
Show
that the
point of contact
3.
at rest.
by the
cue.
name
is
The very Rotation about the Instantaneous Centre. " instantaneous centre implies that the motion of the body
7 '
Let us
make
this state-
ment
o>.
What
rotating about the origin, 0, with angular will be the vector velocity of an arbitrary
158
MECHANICS
*"""*
(1)
f
I
xu.
The
6, its
Vx +
2
ft
ru,
<
6.
If 6
<
0, its
sense
is
reversed.
form of Equations (1) that is important. We say that any motion of the points of the (x, 2/)-plane such that, at a given instant, the velocity of each point is given by (1), is one The velocities of rotation of the plane as a rigid body about 0. of the points in the actual motion before and after the instant in question may be different from those of the rigid body that is But for a short space of time rotating permanently about 0. before and after the instant, the discrepancy will be small because of the continuity of the motion, and at the one instant, the velociIt is the
ties will all tally exactly.
If the point about which the body is permanently rotating had been the point (a, 6) instead of the origin, Equations (1) would have been the following
:
*=-&-&)*,
y
(2)
1
(x
a)6.
We
t,
are
now ready
to state
THEOREM.
at
The motion of
?* 0, is
which 6
the actual body at an arbitrary instant one of rotation about the instantaneous centre.
To prove
the theorem
we have
to
show
2,
(x lt y^ are given by Equations C), 2, and (#, y} are given by Equations B) of the same paragraph. To do this, eliminate X Q and 2/ between Equations B) and C). This can be done most
where
= * = 2/0 +
(Vi
(x l
2/oH
)6,
159
in this form, subtracting them respectively from EquaThe result is Equations (3) of this paragraph, and the
theorem
From the foregoing result a new theorem about the motion of the plane can be derived at once. Let A be an arbitrary point, and let its vector velocity be denoted by V. Impress on each point of the plane, as it moves under the given law, a vector velocity equal and opposite to V. Then A is reduced to rest, and the new motion is one of rotation about A with the same angular velocity as before. We thus have the
THEOREM.
of
The
sum
of the
due
an arbitrary
point,
A;
and
ii)
as
centre.
In
an
arbitrary point, A,
A.
follows immediately
if
at A.
without slipping on a right curve a wheel rolling on the ground. Consider, more generally, the motion of a lamina
circular disc
rolls
We
when an
arbitrary curve drawn in it rolls without slipping on an arbitrary curve fixed Wo may think of a brass cylinder, in space.
or cam, as cut with its face corresponding to the first curve, and attached to the body;
its
And now
the
first
cam
is
allowed
on the second cam. Thus a great variety of motions of the lamina can be realized, and now the remarkable fact is that all motions can be generated
way, with the single exception of the translations, 1 have continuous derivaprovided that the functions (1) of tives of the second order, and the space centrode is traced out by the instantaneous centre with non-vanishing velocity. A necessary condition for the truth of this statement is evident the point of contact of the two cams from intuition, namely
in
this
:
160
MECHANICS
of the actual motion.
This fact
the faces of the cams, i.e. the curves, must suggests the proof be the loci of the instantaneous centres in the body and in space. The locus of the instantaneous centre in the body Definition.
is
centre in space
called the body centrode, and the locus of the instantaneous is called the space centrode.
THEOREM.
Any
is not transla-
tion can be generated by the rolling of the body centrode (without slipping) on the space centrodc, provided the space centrode is traced
the
functions (1) of
Before
selves
wo can prove the theorem, we must make how to formulate mathematically the rolling
of one curve
variable,
As the independent
the *timo most naturally suggests itself; but it is better at the outset not to choose it, but to take, rather, a variable X which
(i.e.
variable)
we
And now
i)
ii) that the arc of the one curve corresponding to any two different values of X, namely, Xj and X 2 and the arc of the other
;
X,
may
be moving
general case than the one that interests us here. Let the equation of the one curve, C, referred to a system of Cartesian axes, (x, y), be
:
(1)
ff(A),
A(X),
where the functions #(X), h(\) are continuous together with their first derivatives, and the latter do not vanish simultaneously
(2)
:
<<7'(X)
/i'(X)
Let the second curve, F, referred to a second system of Cartesian axes, (, r;), be represented by similar equations,
(3) (4)
161
coordinates of any point of the plane, referred to the one connected with the coordinates of the same point,
by the equations
% cos 6
TI
x
(5)
x
2/o
+ +
,
sin
0,
0.
sin 9
rj
cos
And now we
continuous
(6)
first
*o=/00,
/'(X),
2/o=* 00,
^'(X)
exist
where
Since
^'(X),
always meet in a point P, whose coordinates are expressed by the equations (1) and (3), it follows that Equations (5) will hold identically in X if the values of x y from (1), and those of 77 from (3), be substituted therein.
}
,
C and F
The
_ dx ~
d\'
its
Vy
__ ~~
dy d\
is
tangent to
at
and
length
is
The
vector
u whose components
arc
is
tangent to
at
and
its
length
is
The requirements i) and ii) demand that these two vectors be identical. This condition is both necessary and sufficient. The analytical formulation of tho condition is as follows
:
vx
(7)
=
=
u$ cos B
u% sin B
all
Uy sin
0,
B.
vy
u^ cos
We now
proof.
have
the material
the
it
out of which
follow? that
to
construct
(5)
From Equations
FIG. 92
162
MECHANICS
dx
The
the
first line
in these equations
(7),
is
nothing more or
less
than
set
first of
Equations
and the
latter equations
we have
out to prove. Hence the second line must vanish, if the equation is to be true, and so, by the aid of (5), we obtain the first of Equations (8)
:
<8)
is
second of the above equations. Equations (8) represent a new form of necessary and cient condition for the fulfilment of Conditions i) and ii).
5.
of the Fundamental Theorem. It 4. The two curves, C theorem of and F, are here the space centrode and the body centrode, and we will now take as our parameter X, the time t. Equations (1), and y lt of the instanta4, thus represent the coordinates, x neous centre in space, and in Equations (8), the (x, y) are the
Continuation.
Proof
is
now easy
to prove the
same point, (x l9 yj. The other quantities that enter into (8) are the functions (6) that determine the position of the moving body; and X = t. Thus Equations (8) go
coordinates of this
*o
2/o
(2/i
(x l
2/o)
*
6
= =
0,
0.
But
instantaneous centre.
these are precisely Equations (5) of 2, which determine the are thus shown to be true. Equations (8)
163
From Equations
(8)
4, it
that a necessary condition for the truth of the theorem the coordinates of a point P of C satisfy the equations
:
X
(10)
= =
Xn
dy dX
/d0
/ dX'
I dB / 3T-
2/o
+ -3r
dx a
But
y thus defined
these conditions are not sufficient, since the functions x will not in general admit derivatives.
this latter
and
To meet
the functions (6) possess, furthermore, continuous second derivatives. But this is not enough, even if the case that x and y
reduce to constants
It
is,
is
4,
de
be not both
the equations
In other words,
_
"
dX dX 2
(12)
""
d^cteo dX 2 dX
_,
"*"
^!^o =
dX 2 dX
U|
"~
=
dX 2 dX
t
shall
dX'dX 2
dX*"d\
the case in
also includes
more complicated
If,
hood of a point, X = X and if the case of permanent rotation about a fixed point be excluded, the curve C will at most have a cusp and otherwise be smooth in the neighborhood of the point and the same will be true of P.
;
Acceleration of the Point of Contact. Let the point (X Q y ), at a given instant, t, be taken at the point of contact of C and T. Then it follows from (10) since or, more simply, from (8)
,
x Q and y
y Q> that
dX
164
MECHANICS
,
Let the origin, furthermore, be taken at this point (x y ), and C be tangent to the x-axis here. Now, the derivatives of x and y in (10) cannot both vanish. On computing them it is seen that they reduce respectively to
let
d\ 2
~d\'
d\ 2
0,
d\
The
since
is
of x at the origin.
Hence we
infer that
the time, t, these derivatives become the components along the axes of the acceleration of the point of contact, thought From (13) it appears that of as a point fixed in the moving body.
If
is
this acceleration
is
never
0,
but
is
billiard ball is projected along a smooth horiExample. zontal table with an initial spin about the horizontal diameter which is perpendicular to the line of motion of the centre. Deter-
mine the two centrodes. Take the path described by the centre of the ball as the axis of x, and the centre of the ball as (x y ). Then
,
= XQ
=--.
is a horizontal straight line at a distance below the centre of the ball, and the instantaneous centre This means that the is always beneath the centre of the ball. ball rolls without slipping on a right line distant c/w below the Hence the body centrode is a circle of radius c/co about centre.
EXERCISE
A
cf.
billiard
ball
is
struck
full
by the
cue.
Determine the
;
space centrode and the body centrode during the stage of slipping
Chapter IV,
14.
165
SC
~~
*C rt
~~ Cv
""""
_ ~
2a
5
2acl
5/
t'
where a denotes the radius of the ball, and c the initial velocity of its centre. The time that elapses during the stage of slipping is 2c/7/A<7 seconds. The space centrode meets the billiard table at
the angle
The equations
coordinates, are
:
of the
body centrode,
=-
2ac
2a
5
'
When an empty tea cup is set 6. The Dancing Tea Cup. down on a saucer, the cup sometimes will dance for a long time Two features of this phenomenon attract before coming to rest.
attention;
dissipated
first,
by damping, and
that the energy, obviously slight, is not earlier secondly, that we can hear a noise
The second point can be in which so little energy is involved. disposed of easily because of the physical fact that the energy
of
sound waves
is
surprisingly small.
first critically
To examine
of
the
the motion.
The
results
The
highly sug-
Consider the motion of a lamina, in which the body centrode is a right line making a small (variable) angle with the horizontal. For the space centrode take a curve suggested
by the
figure.
suggestively as follows.
(1)
166
MECHANICS
This curve gives satisfactorily the part of the figure not too near the lines y = 1, but it is tangent to these lines, whereas it
should have cusps on them. The desired modification
the curve
is
simple.
y=f(X )=
X*
from one which is tangent to the axis of x into one which has a cusp on the axis, it is enough to replace f(x) by [/(z)]*
:
Apply
(2)
Now
goes on as the tea cup dances. and smaller angles as its point of intersection with the axis of x descends.
the tangent line to descend here a We have picture of what The line oscillates through smaller
Tyndall,* in his popular lectures, showed an experiment with a coal shovel illustrating the same phenomenon. The all-metal shovel was heated near its centre of gravity and laid across two thin lead plates clamped in a vise, with their edges horizontal.
expanded with the heat, throwing the shovel onto the other Then the process was reversed. Thus vibrations like plate. those of the tea cup arose, and died down.
7.
The
ergy of
The
2
2
.
kinetic en-
T = i 2) m^\
v k2
xk2
+ yi +
zt
We
restrict ourselves to
xj?
yj?.
Suppose, now, that the particles are rigidly connected. In Equations A), 1, let the point (x y ) be taken at the centre
,
Thus Equations
Mode
(2),
2,
become
167
= =
(&
sin
+
-
*;*
cos 0)
sin 0)
0,
0.
Vk
+
:
(& cos
77*
mk
2)
and each
of these
is 0,
of the (, 7y)-axes.
(1)
Hence
is
at the origin
T =
|MF + i/fl
2
2
,
where
moment
velocity.
of inertia
denotes the velocity of the centre of gravity and I is the about the centre of gravity, 12 being the angular
Second Proof.
The
result
may
also
be obtained by means of
the instantaneous centre, 0. For the motion, so far as the velocities that enter into the definition of T are concerned, is
Hence
T =
/'
i/'ft
where
denotes the
moment
/'
of inertia
about 0.
Now,
+ Mh\
and
Aft.
where h
(3)
is
V =
On substituting this value of /' in (2) and then making T takes on the form (1), and this completes the proof.
Generalization.
u$e of
(3),
The most general rigid bodies with which we made up of particles and material distribu-
tions spread out continuously along curves, over surfaces, and throughout regions of space. When such a body rotates about
axis, the kinetic energy is defined by Equation can state the result in the form
an.
(1).
We
The kinetic energy of any rigid material system which is rotating about an axis, is given by the formula:
168
MECHANICS
Remark. The formula holds even for the most general case of motion of any rigid distribution of matter in space. For, such motion is helical, i.e. due to the composition of two vector fields of velocity, i) a field corresponding to rotation about an axis; and ii) a field of translation along that axis cf 12 below.
;
.
EXERCISES
1.
ball rolls
mine the
2.
down a rough plane without slipping. Deterkinetic energy in terms of the velocity of its centre.
A
A
A
floor.
3.
ladder slides down a wall, the lower end sliding on the Find the kinetic energy in terms of the angular velocity.
ellipse is rotating in
its
Compute
is
4.
homogeneous cube
Determine
Consider any 8. Motion of Space with One Point Fixed. motion of rigid space, one point, 0, being fixed. We shall show that there is an instantaneous axis, i.e. a line through 0, the velocity of each point of which is 0; and that the velocities of all the points of the moving space, considered at an arbitrary instant, form a vector field which coincides with the vector field arising from the permanent rotation of space about this axis. We give first a geometrical proof which appeals strongly to the intuition. The refinements which a critical examination
of the details calls for are best given through a
new proof by
vector methods.
be a point of the fixed space, distinct from O. If its then the velocity of every point of the indefinite velocity line and Q is 0, since a variable right line is eviright through Let
is 0,
dently at rest if two of its points are at rest. If, on the other hand, Q is moving, pass a sphere, with centre at 0, through Q and consider the field of vector velocities cor-
responding to the points of this sphere. The vectors are evidently all tangent to the sphere, and they vary continuously, together with their first derivatives, for we are not concerned with discontinuous motions.
Pass a great
velocity of Q.
circle,
Let
(7, through Q perpendicular to the vector be a point of C near Q. Then the vector
169
will also be at right angles to the plane of C and on velocity of the same side of C as the vector at Q. For, since the vector of is at to the chord Q velocity right angles QP, the vector velocity
of
it
must
there,
lie
in the plane
through
perpendicular to QP.
But
I
And now
velocity is 0.
of For, otherwise, all the vectors that represent the velocities of the points of C would be directed toward the same
say,
must be a point
side of C.
In particular, then, the point Q' diametrically opposite But that would mean velocity.
that the mid-point of the diameter Q'Q, i.e. the centre of the is at From not rest. this contradiction follows the truth sphere,
is a point A of C which is at rest. Hence and A is at rest, and the existthe whole indefinite line through ence of an instantaneous axis, 7, is established.
It
remains to prove
that the vector field of the actual velocities coincides with the
the vector velocities due to a rotation about I. Conan arbitrary point, P, not on /. Then P cannot be at For, if three points, not in a line, rest, unless all space is at rest. of moving space are at rest, all points must be at rest. Pass a plane, M, through P and the axis. Then the vector velocity For let Q be any point of /. of P must be perpendicular to M. Since Q is at rest, the vector velocity of P must lie in a plane
field of
sider
through P perpendicular to QP. Consider next the circle, C, through P with / as its axis. The vector velocity of P is tangent to C. For it is perpendicular to any line joining P with a point of /. Moreover, the vector
velocities of all points of
wise two
points of
C are of the same length. For otherbe approaching each other, or recedwould C
ing from each other.* is proporLastly, the magnitude of the vector velocity of be from /. Let the determined its distance tional to plane
M
1
Consider two points, P and P 2 in but not on from be and h 7. Let uh the magni7, distant A, 2 respectively is Then coA the tude of the vector velocity of P magnitude of 2 P For otherwise the vector velocity of P 2 2 would issue from
by 7 and P.
M.f
* Exercise 4 below.
170
MECHANICS
EXERCISES
1. Give a rigorous analytic proof that if two points of a straight line are at rest, every point of the line is at rest.
moving
2.
is
point
is
moving
so
moving
point, P, neither approaching Q Give a rigorous analytic proof that the orthogonal to the line QP, if the vector
it is
in
velocity of
3.
is
orthogonal to that
line.
If three
lie
and
if
these points do
not
on a
space
is
at rest.
ically.
4. Prove analytically the statement of the text which refers to this Exercise.
5.
The same
Let space rotate as a rigid body 9. Vector Angular Velocity. about a fixed axis, L, with angular velocity o>. Let P be an
Then
arbitrary point fixed in the moving space. will be represented by the velocity of a vector v perpendicular to the plane deter-
FIG. 95
vector
by
and
(co).
mined by P and the line L, and of length hw, where h is the distance of P from L. be any point of L. Lay off from Let along L a vector of length w and denote this Let r be the vector drawn from O to P. Then
is
Xr;
Appendix A.
as
Let a system of Cartesian axes (or, y, z) be assumed with Write origin, and let i, j, k be unit vectors along these axes.
(2)
co 2 k.
Then
i
(3)
171
= =
=
XUg
ZO)
(4)
Vy Vg
Composition of Angular Velocities. Consider two rotations about axes which pass through 0. Let them be represented by the vectors (o>) and (a/)- An arbitrary point P of space has a vector
velocity v given
by
(1)
v = ()
due to the
first rotation,
r,
v'
V=
(')
r,
due to the second rotation. Let these vectors, v and v', be added. Then a third vector one in which to the point P is assigned the vector field results v + v'. It is not obvious that this third vector field can be realized by a motion of rigid space far less, then, that it is the field of velocities due the to precisely angular velocity represented by the vector
(5)
(0)
this is in fact true
Velocities.
is
= ()
that
That
is
the
(0Law
of the Composition of
Angular
The
(6)
proof
immediate.
We
have
r
V = () X
X
r
(')
r.
Now,
(7)
is
distributive
r
{()
v
(')}
= () X
(a/)}
+
=
(')
r.
Hence
(8)
+ v' =
{()
(12)
r,
and we
are through.
The
result
fol-
lowing theorem.
THEOREM.
Angular
velocities
can be compounded by
the
Law
of Vector Addition.
EXERCISE
Prove the law of composition for angular
of Equations (4).
velocities
by means
172
10.
MECHANICS
Moving Axes.
Proof of the Theorem of
8.
Let space
Let
i,
be moving as a rigid body with one point, 0, fixed. be three mutually orthogonal unit vectors drawn from
in space,
j,
and
:
the body.
following
and fixed 7 be a second set of such vectors fixed in The scheme of their direction cosines shall be the
let a, 0,
a
'l
j8
rj
'2
(1)
Thus
a
with similar expressions for /3, 7, where the direction cosines are any functions of the time, t, continuous with their first (and for later purposes their second or even third) derivatives, and
satisfying the familiar identities
;
cf.
Appendix A.
Observe that
187
=
aa =
07
0,
(2)
etc.
0,
+ 7/3 =
etc.
We
Then
(3)
are
now
an instantaneous axis. Let P be an arbitrary point fixed body, and let r be the vector drawn from the fixed point
r
to P.
= fa
+ vp +
,
f 7.
Since
is
77,
to the time,
(4)
and
so
= $a
+ 40 + fy.
P
A necessary and sufficient condition that P be at rest is, that the projections of f on three non-complanar axes all vanish. be at rest can be exHence, in particular, the condition that
pressed in the form
(5)
:
at
0,
fit
0,
7* =
(4),
0.
Applying
we
ordinary equations
173
+ f ay =
(6)
tfa+
f 07
=0
Let
(7)
a
(2) it
yp,
ay,
pa.
From
follows that
pj
=
=
7,
f
aft.
Equations
(6) arc
now
a,
b,
c.
These cannot all be unless the body is at rest, since the vanishing of the above scalar products would mean that
pa
and
of course
0,
ya =
P and
aa =
0.
at rest,
and
The
tions
(8)
:
Xa,
ij
X6,
Xc,
oo
<
<
oo
These points, and these only, are at rest. They form the instantaneous axis, and it remains to show that the* latter deserves its name. Instantaneous Axis. Let a vector (co) be defined as follows:
(9)
o>
7/3,
a,
ay,
co$
fta
(10)
(CO)
C00!
+ ^0 + W^7.
whose equa:
Then
(co)
is
now
1
CO^
JL
COr,
= L.
CO^
have seen that the vector velocity v of an arbitrary point fixed in the body is given by (4). The components of v along the (, 77, f )-axes can be written in the form
:
We
V{
= at = %aa = yt = %ya
+ yap + f ay
+ rjyp +
yy
174
MECHANICS
ff
Hence
From
(13)
COf
W,
CO^
(CO)
T,
and so we
v
if
of
is
the same
would have
rigid space
were
rotating about the instantaneous axis with angular velocity co. coinThus the actual field of vector velocities of the points cides with the field of vector velocities due to rotation about the
instantaneous axis represented by the vector angular velocity (co), and the proof is complete.
11.
The
locus of the
instantaneous axis in fixed space is called the space centrode, and The actual its locus in the moving space, the body centrode.
motion consists of the rolling of the one cone (the body centrode) without slipping on the other cone (the space centrode). To prove this statement consider the path traced out by a
specified point in
the terminal point of the vector (co), the initial point being at 0. The locus of this point is a certain curve C on the space centrode
(co)
co x i
+ coj +
cosk,
:
of the
body centrode
It is sufficient to
corresponding arcs are equal. This will surely be the case if d(u>)/dt for C is equal to d(co)/cft for F. Now, the first vector
The
is
a*,/?
co$
175
line vanishes
because
it
the point
(o>)
axis.
The
first
line is the
vector
This completes
the proof.
EXERCISE
Treat the motion of the plane by analogous vector methods.
Let
_
P
let p,
be the vector drawn from the fixed to the moving origin, and and rj. a- be unit vectors drawn along the positive axes of Let r be the vector from the fixed origin to an arbitrary point P.
r
Then
+
:
r?<7.
The
is
=
is
fP
+ ^=
0.
The instantaneous
centre
given by setting t
On
e",
= C + D'.
diffi-
Motion
in
of
Space.
General Case.
subject to
any manner, about continuity. Reduce a point A all space a motion of translation whose vector is equal and oppoThe vector field of the velocities site to the vector velocity of A. in the original motion is compounded by the parallelogram law of vector addition out of the two vector fields i) of translation and if) of rotation about the instantaneous axis, 7. Let the vector that represents the translation be resolved into two vectors, one, T, collinear with 7, the The velocity ~H other, A, at right angles to 7.
moving
of any point, P, distant h from the axis, is, in the case of pure rotation, hu its direc;
tion
and the axis, and its at right angles to the plane through sense is a definite one of the two possible senses. Hence it is seen
is
176
that
it is
MECHANICS
possible to find a point, B, whose vector velocity due to is equal and opposite to the vector velocity A. (Draw
the rotation
of the axis, perpendicular to / and A, and a line from a point measure off on it, in the proper direction, a distance h = A/u.) All points in the line L through B parallel to / will also be at It thus appears that the original motion is one of rotation rest. about L compounded by the law of vector addition with a motion
of translation parallel to L tiad represented by the vector T. This vector field is, in general, the same as that of the vector
velocities of the points of a nut which moves along a fixed machine screw (or of the points of a machine screw which moves through a fixed nut). The two exceptional cases are those of rotation,
of the threads, and translation, the corresponding to a pitch as the becomes infinite. limiting case, pitch
13.
The Ruled
is
;
Surfaces.
We
have seen
in
12 that the
the
sum
of
two vector
fields
the other, translation parallel to L. The locus of L in axis, a ruled surface S, the space centrode, and the locus of L space is in the moving space is also a ruled surface, S, the body centrode.
we should
anticipate Jhe
The surface
is
tangent to
along L, and
it
rolls
slides
on S.
An
So
clear
First of all, it intuitional proof can be given as follows. from the very definition of L that S slides on S along L.
it is necessary to prove only the tangency of the two surfaces. and let P be a point of L Let L be the line L at time t = Pass a plane through P perpendicular to L cutting S in the curve Cj and let P be the point in which L at time t = t Q + At cuts C. Let Q be the point fixed in S, which will coincide with
.
at time t Q A. The vector velocity of Q at the instant t has a component, c, parallel to L and a component hu at right angles to the plane through L and Q. Obviously h is infinitesimal with AJ. In time the point Q will, then, have been displaced,
t
save as to an infinitesimal of higher order, parallel to distance cAt. But it will have reached P.
L by
The
an
proof
is
now
clear.
at
PQ
because
177
contains a point
of
infinitely
near to
but not on
P makes an P because it
Z/
.
And
of S make
an infinitesimal angle with each other, because when Q is displaced parallel to L by a distance cA, its distance from P is an infinitesimal of higher order than the distance of P from P
.
Instead of developing the details needed to make the intuitive proof rigorous, we will treat the whole question by vector methods.
First,
14.
in space,
and
let its
axes.
Consider,
origin, 0.
first, the case that the moving axes have a fixed Let a system of axes (x, y, z), fixed in space, with be chosen let the moving axes be denoted by (, 77, f),
by the scheme
to
of direction cosines
of
10.
0) Then
+ 4/9 +
[ 7.
or
(3)
= v
+ v,
The
df
4>
drj _
represents the velocity of P relative to the moving axes; i.e. what its absolute velocity would be if the (, t;, f)-axes were fixed and P moved relative to them just as it actually does move.
ve
= **
+ it + r 7
represents the velocity in space of that point fixed in the body, which at the instant t coincides with P. To say the same thing Let us consider the point P at an arbitrary in other words
:
instant of time,
t.
Let
178
MECHANICS
Then ve
is
the vector
Q is
the vitesse d'entrainement, the velocity with being transported by the body at the instant t.
The
form
(6)
ve we know
r
all
about.
In vector
it is
ve
= () X
or
(7)
Its
axes,
if
we
write v'
vej are
(8)
Thus we have
components of
as the final solution of our problem this: The along the axes of % ry, f are :
(9)
General Case.
(x, y, z)
Let (,
TJ,
f)
i/o,
).
Then
r'.
Hence
(11)
+ v'.
Here,
FIG. 97
dx Q
(12)
and
v' is given
by
(9).
The
v' of (11)
is,
of (8).
179
>
Denoting the components of v along the ( 17, f )-axes by v*y vl> show that the components of v along these axes are
, :
dy
(13)
'dt
+ ^~
4+
Here,
=
Proof of the Theorem of
12.
7*0.
16.
axes fixed in space, (x, y, 2), with origin in be assumed. Let O' (x y^ ZQ) be a point fixed in the body, the motion of 0'
: ,
being
(1)
known
*o=/(0,
2/o
v(0,
*o
lKO.
Finally, let
(2)
cf.
Then
rc
r',
14, (10)
(3)
dt
Then
of
the components of the absolute velocity of P (i.e. its velocity are given by the formulas ( 77, f )
,
(4)
We can formulate the problem ^(i> i/D fi) fixed in the moving
velocity
(5)
is *,
is
To find a point space whose absolute vector collinear with the vector (o>), or is
as follows
:
(<>).
Here,
(co)
whose rotation
the vector angular velocity of the moving space, is defined by the direction cosines of 10.
180
MECHANICS
virtue of (4) the vector equation (5)
:
By
is
equivalent to the
(6)
Since
(7)
0,
is
utat
+ co^f
;
We
in
for then the space can dispose at once of the case o> = which 0' is at rest, is stationary, and so the motion of the given
space is translation (unless it be at rest). Thus all lines parallel to the vector that represents the translation are axes such as
we
seek.
o)
If
0,
we obtain from
(8)
a unique determination of
k.
On
two
chosen, determine uniquely two of the three unknowns lf rj l} f t as linear functions of the third, and then the remaining equation (6) is true because of (8).
=
i
i>
i7i
61,
fi
ci
be
i
particular
solution
of
(6),
:
'Ju Ti>
equations
+
(1
(fi
fli)
(f i
- c iK = - c^cu* =
Hence
(9)
T^,
ft
is
seen to be any point of the line through This line we define as L. These con(<*>).
181
its
locus in the
moving space) in common. We wish to show that they are S and that slides over S in the direction of L. tangent along L, The last fact is clear from the definition of L.
have the
line
the
the surface S.
These surfaces
f t ) is not uniquely determined by the time, rj lf We will, for our purposes, select it be any point of L. may as follows. Let Z/ be a particular L, and let P be an arbitrary once chosen and then held fast. Pass a plane point of L Then (x lf y lt zj shall be the interP orthogonal to L through
The
point (&,
but
L with M
shall
t
and
its
by
by
F:
Its
C.
The point
(f t , y ly
fj
be the point of
Its locus in
S which
coin-
yi, 2i)
at time
t.
shall
:
be denoted
ti=F(t),
77!
$(0,
is
fi=*(0-
dt
#, k a + *?i* * + 7 dt dt
1
'
provided this vector 7* 0. To show that two surfaces which intersect at a point P are tangent it is sufficient to show i) that they have a common tangent
and ii) that a tangent vector tj to the one surface and vector, t a tangent vector t 2 to the other surface, neither collincar with t, are complanar with t, all three vectors, emanating from P
;
.
because they are both tangent to L. to the point Secondly, consider the vector ^ drawn from of intersection of C and F at time t t. Its derivative is a vector
surfaces
satisfy
i)
The
S and 2
tangent to C, provided it 5^ 0. On the other hand, consider the point (| t 77^ J\) of F, for which t = t. Let the vector drawn from 0' to this point be denoted
,
by
r[.
Then
*i
TO
i{,
where
is
given by
(1),
and
182
MECHANICS
Hence
dt
dt
4-fo
+ ^d +
7/^
^7.
This last line is precisely the vector velocity of that point fixed in S, which at the instant in question, t = t, coincides with
(
x \>
V\y z i)-
(5),
unless
This vector, it be 0.
t,
let
us
along
because of
The other
vector on the right of (11) is the vector (10); i.e.. (, rj ly f,) or (x l9 y lt zj. Equation
t,
1.
tx
will
not
lie
along L.
t
Hence
t. 2
Consequently Condition
ii) is satisfied,
arid the
;
surfaces
it
The
case
is
included
EXERCISE
Let a cylinder of revolution
roll
and
slide
on a second cylinder
which
is
and there being no slipping oblique to the element. Choose the point (x i/ z ) in the axis of the moving cylinder, and discuss the whole problem by the method of this paragraph.
element,
,
16.
Lissajou's Curves.
It
of freedom, the
monic Motion.
(1)
a cos
(nt
+ 7),
where
(2) '
T =
is
n
is
is
determined
In two dimensions, or with two degrees of freedom, an important case of oscillatory motion about a fixed point is that in which the projections of the moving point on two fixed axes
183
itself,
simple har-
monic motion
\u)
I
1 I
x
y
+ 7) = b cos (mi + e)
a cos
(nt
It is possible to generalize at
once to n dimensions
(4)
xk
ak cos (n k t
y k),
fc
Z,
n.
Let us study
first
the two-dimensional rase, beginning with We may set 7 = if, as usually hapis
measured
is
unimpor-
n. Dynamically, this case can bo realized approximately by the small oscillations of a spherical pendulum. Let 7 = 0,
Example
m=
<p
nt.
Then
mt
+ =
e
<f>
a cos
cos
<p
(x
A =
Assume that
will
= A
e,
<p
B sin
b sin
<p
b cos
B =
c.
neither a nor b vanishes, since otherwise we should be thrown back on right line motion along one of the axes.
We
take a
>
0,
b > 0. B ^ 0. The
its
is
then
(6)
cos
<p
=
^,
sin
<p
-^
^ y.
On
(7)
we
find
x2
(Ax
ay)
a2 B2
and
to infinity,
an
ellipse.
184
MECHANICS
analytic geometry,
or
The axes can be found by the methods of computed directly by making the function
COS 2
__
sin
+ B* sin
Here,
e
^
or
We
TT,
B = have omitted the special case and the motion is rectilinear, along the line
0.
:
In
all cases,
the path
is
=
an
h,
y
four
=b,
sides,
sometimes being a
If
Example
2.
m=
where h
is
small.
:
we
write the
x
(9)
a cos nt
b cos (n
+ A/ +
e)
2w/n,
nearly constant, and the path is an nearly ellipse which, however, does
is
not quite
slightly different orientation
close.
And now,
circumscribing rectangle
will
be described,
all
inscribed
6.
By
this is
tem that
consists of
cos 2* =
or
6 2 sin 2e
cos 2y
where y denotes the angle from the axis of x to an axis of the conic.
of the axes are found to be
:
The length*
where
A 8 = a4
-f 2a*&* cos
+ b*.
185
which is made fast at the mid-point of an inextensible string whose two ends are fastened at the same level. When the bob oscillates
in
remains at
of the
the vertical plane through the supports, the second string rest, and we have simple pendulum motion, the length
pendulum being Z, the length of the first string. Secondly, let the bob oscillate in a vertical plane at right angles to the line through the points of support, and mid-way between
but the length in where d denotes the the second string. + d, sag For small oscillations, the coordinates of the bob will evidently be given approximately by Equations (3), and by suitably choosing I and d, we can realize an arbitrary choice of m and n.
these.
Again,
is
now
I'
If the
bob
of the
pendulum
is
a tunnel
of small opening, filled with fine sand, the sand, as it issues from the tunnel, will trace out a curve on the floor which shows ad-
mirably the whole phenomenon of the Lissajou's Curves. In particular, if the second string is drawn as taut as is feasible, so that d is small, the two periods will be nearly, but not quite, and it is possible to observe the near-ellipses steadily equal
;
(the
diagonals of
Example
<p
3.
m
x
2n.
0*
0,
and
set
nt:
(10)
a cos
<p,
b cos
2<p.
Hence
(11)
*/
I**
an arc of a parabola, passing through the vertices the circumscribing rectangle and tangent to the opposite side at the mid-point. The sand pendulum may be released from rest at the point (a, 6), and it then traces reis
6),
(a,
6) of
a cos
<p,
y
b cos
= A
cos 2<p
B sin 2<p
e.
A =
6,
B =
b sin
* This experiment should be shown in the course. It is not necessary to have a physical laboratory. A tunnel can be bought at the Five and Ten, and string is still available, even in this age of cellophane and gummed paper.
186
MECHANICS
e is
When
give x'
Fig. 100.
It is
small, the curve runs along near to the parabola; <p symmetric in the axis of y, since ^ and <p'
x, y'
cf.
TT
=
x
=
of
y.
It is tangent
a,
the
circumscribing
and
twice
?r/2,
When
a cos
<p,
has increased to
A =
(13)
and
x
b sin
2p
or
This curve
FIG. 99
is
(15)
y
is
which
readily plotted.
c
parabola
we have again an arc of a TT, the former arc, turned upside down. As e continues to increase, the new curves are the mirrored images of the old in All TT reverses the signs of A and B. the axis of x, for e' = e
has reached the value
When
these curves except the arcs of parabolas are quartics, inscribed in the fixed rectangle, and having symmetry in the axis of y.
Example
f
4.
m=
= =
2n
h,
where h
x y
a cos nt
b cos [2n*
(16)
I
+ ht + excursion, M is
is small.
Here,
e]
nearly
0.
an but as time
close to
FIG. 100
advances, too.
readily trace these curves with the sand tunnel. If he does his best to
will
make d =
h.
J,
there
17.
Case.
Let
Then
ap,
m=
aq.
187
at,
7 =
0.
Then
(x =
y
(? +
e).
These functions are periodic with the primitive periods 2ir/p and 27r/<7, and evidently have the common period 2ir. The smallest positive value of co for which
a cos p(<p
6 cos {(/(p
is co
co)
+ +
=
co)
e}
= =
a cos p^>
fe
cos {</p
e}
27T.
For,
if to is
to be a period of the
CO
first
function, then
x A
2?r
And
if co is
27T
M
.
Hence
X -
/*
-,
Xg
,,
is
of X, n in natural numbers which satisfy = = q. X arc p, n equation From the periodicity of the functions it appears that the curve closed, arid thus, as t increases, the curve is traced out re-
For a non-specialized value of e, the curve is tanpeatedly. a of the circumscribing recgent to each of the sides x = a, tangle p times, corresponding to the solutions of the equations
cos p<p
1,
of the sides y
6,
;
6.
line
x',
<
x'
<
a,
line
in 2q points. b < y' < 6, y = y'j These curves are all algebraic, and rational, or unicursal. = tan -J-p, the variables x and y appear, by For, on setting The curves de Moivre's Theorem, as rational functions of are all symmetric in the axis of y. The Incommensurable Case. Aperiodic. If on the other hand
.
n/m
incommensurable, the curve never closes. It courses every region contained within the rectangle. If P be an arbitrary point of the rectangle, the curve will not in general pass not merely through P but it will come indefinitely near to P but often once, infinitely possibly, occasionally passing through P.
is
;
188
MECHANICS
as follows.
<f>
at the centre.
Let
= 2wa
2, 3, (denote Hence they must have at least one point of condensation, P. But from this follows that of be a condensation. must For, let P n and every point point Pm be two points near P. Then the point corresponding to n w must be near the point corresponding to <p = 0. Havthem by
incommensurable with 2?r; i.e. let a be points of the circle which correspond to
Then
the
P P
t,
2,
are
all distinct.
ing thus obtained an arc of arbitrarily small length, we have but to take multiples of it, i.e. to construct the points P*( n -m)i k = 1, 2, 3, to come arbitrarily near to any point on the
,
circumference.
=
x
nt,
m=
n
<f>,
a.
Then
1
y Let
= =
a cos
b cos (cup
<p
17).
x'
a,
and
let
<p'
a cos
The curve
y
which
<?'
b cos
+ 2kw) +
6 cos
{(-
2kw)
y}.
since the angles 2kair lead to points on the circle which are everywhere dense, the corresponding values of the cosine 1 and 1. factor are also everywhere dense between
And now,
It is of interest to
These
occur
t)
when
<f>'
kw
+ ^, + ^~a'
*
(1
^0;
it)
^=
)br
fc?0;
provided
(19)
-n
a)ir.
the inequality (19) holds, there is a one-to-one correspondence between the values of <p and the points of the curve,
When
189
provided the multiple points (which are always double points with distinct tangents) are counted multiply. If, however,
(20)
77
(1
+k
+
Q a)7r,
then
a<p
f\
= a (<p =
<f>
fc
7r)
i 7r.
Set
(21)
+k
:
Q w.
(18)
become
x
I
=
=
a! cos
where
a!
a or
a,
and
likewise &'
6 or
6.
Let
^
Then
there
is
<
oo.
a one-to-one correspondence between the values on the curve. The point for which 9 =
:
x
is
a',
&',
It is simple,
through
The double
kw
> +a
0,
I
0,
where
/CTT
-->
a.
^
k
or where
far
+ - > 0,
0.
and n-Dimensions.
x
(23)
a cos
<p
y
z
=
=
b cos (ay
c cos ($v
+ +f
??)
The
and
case that a,
/3
before.
ft
The curve
the
motion
is
is
and
their ratio
190
MECHANICS
region,
however small,
of
a^xga,
6^2/^6,
c,
and has no multiple points, the correspondence between the oo < <p < oo points of the curve and the values of <p when one-to-one without Whether former the being exception property is present for all such values of a and 0, provided furthermore that or, /3, and /3/a are not connected by a linear non-homoare not geneous equation with integral coefficients, and that
77,
J"
cannot say, though I surmise it to be. property, however, can be established. The same statements hold in the general case,
specialized, I
The
latter
xk
If; may dom and
,
ak cos
(oL k <p
+
,
rik),
1,
n.
qm 1 g burn's Pendulum suspended in a moving elevator will have its projection on a horizontal plane executing a Lissajou's motion,
happen, in a dynamical system with n degrees of freecoordinates q lt q n that only a sub-set, <ft, m < n, execute a Lissajou's motion. Thus a Black,
is
not periodic at
all.
Wallace Clement Sabine drew mechanically some very beautiful curves, which are here reproduced in half-tone. So far as I I still have the half-tone which Dr. Sabine gave me.
late Professor
The
have been able to ascertain, the curves were never published. The figures here shown were made from lantern slides in possession of
the Jefferson Physical Laboratory, and it is through the courtesy of the Laboratory that I have been enabled to reproduce them here.
CHAPTER
VI
ROTATION
1.
w,-,
Moments
of Inertia.
The moment
is
of inertia of
particles,
defined as the
sum
(1)
/
r
10.
<=i
i><r<,
whore
IV,
cf.
Chapter
with
be an arbitrary point of space, and let Cartesian axes as origin be assumed. Let the moments of inertia about the axes be denoted as follows
Let
A =
(2)
The
(3)
sums
-0
= 5 m
Vi z <>
E =
in the usual
way by
the methods
it is
possible to express
about an arbitrary axis through 0. the direction cosines of the axis be a, 0, y and let P (x, be an arbitrary point in space. Then
of inertia
:
moment
Let
y, z)
r2
=
z2
p2
<^
or
r2
x2
tf
(ax
2
+ py + yz)
1,
2
.
Since
of
+P+ +
* 2 )(
2
7 =
FlG
:
2
2/
+P+
191
(ox
+ fry + yz)*.
192
MECHANICS
Hence
Thus
/
or:
(4)
2yazx
2
etc.
=a 5) m;(^
2
+ z> +
2 )
5J
2
t -(*i
+ a* +
)
7
is
= 4a 2
50 2
+ CV -
2D07 ~ 2#y
2Fa/3.
This
The meaning
of the
formula can
Ax
By
Cz*
- 2Dyz -
2Ezx
2Fxy =
1.
It is
known
Let an
arbitrary line
through
the surface in the point (X, Y, Z), and let p be the length of the segment of the axis included between the centre of the
meet
ellipsoid
and
its surface.
Then
X
Since X, Y,
(6)
ap,
(5), it
Y =
C7 2 -
0p,
Z =
yp.
Z
2
satisfy
follows that
p (^la
+ Bp +
(4)
2D/37
find
:
- 2#ya =
2Fa/3)
1.
On
(7)
combining
and
(6)
we
2 P /
1,
and /
is
property
If all
seen to be the square of the reciprocal of p. From this it appears that the Ellipsoid of Inertia is invariant of
n particles m lie on a line, Equation (5) no longer reprean ellipsoid. Let the axis of z be taken along this line. Then A = B and all the other coefficients vanish. Thus (4) becomes =
z
sents
4 (' +
*).
A ^
and
lies
at
cases but this one, the quadric surface (5) the exists, being cylinder of revolution
In
all
(8)
A(*
2 2/ )
1,
in (7) is
still
true.
ROTATION
Suppose, conversely, that (5)
If
all
193
to represent a true
fails
(i.e.
,
coefficients A, B, non-degenerate) ellipsoid. F are 0, the system of particles evidently reduces to a single In all other cases, (5) represents a central particle situated at 0.
the
quadric surface, S. If this is not a true ellipsoid, then there is a line, L, which meets 8 at infinity; i.e. which does not meet S in any proper point, but is such that a suitably chosen variable
line
definitely as
approaches L.
The moment
0.
of inertia
about
L'
is
given by
(7)
and approaches
the
of a
all
moment
system
of inertia about
L is
obvious that
the particles must lie on this axis. We see, then, that (5) represents a true ellipsoid in all cases except the one in which the particles lie on a line, and that (7) holds in all the latter cases, too, except the one in which the
We
recall, finally,
axes
Chapter IV,
10
THEOREM. The moment of inertia, I, about any axis, L, is equal to the moment of inertia, 7 about a parallel axis, L through
, ,
mass times
+ MW.
EXERCISE
Show
is
that the
moment
:
of inertia
about any
line,
L, in space
= {A
+ M(y\ + *?)} a + {B + M(z\ + x$\ + {C + M(x\ + tf)} 7 - 2(Z> + MVlzJ 0y - 2(E + MZ.X,) ya - 2(F + Mx.y,) aft
2 2
/3 2
arc
^ ie
z l respectively
by
*>
y\
y>
*i
*>
194
MECHANICS
#,
?/,
where
x u Vn
the
2.
and
z \ are
all
referred to
new
axes.
Principal
Axes
of
a Central Quadric.
:
Ax 2
+ By* +
Cz*
2/)7/*
2Ezx
2Fxy =
1,
coefficients are arbitiary subject to the sole restriction The problem is, so to rotate the that they shall not all vanish. axes that the new equation contains only the square terms. Let
(2) (3)
where the
F(x,
y, z)
= Ax 2
Eif
+
=
Cz*
x*
*(*, P,s)
+ 2Dyz + + y* +
z*.
2Ezx
2Fxy,
y, z)
on the surface
(4)
x*
2
?/
z2
a2
or
*(x,
y, z)
a2
z) is
must
continuous and the sphere is a closed surface, attain a maximum value there, and also a
minimum.
Let the axes be so rotated that the
on the axis of
of
Equations
and
(3)
F+
\$,
;
the independent variables being x, y, z, with X as a parameter and we then sot each of tho first partial derivatives equal to
:
(5)
b\
X*,
0,
X$ 2 =
0,
t\
X$3 =
0.
These three equations, combined with (4), form a necessary condition on the four unknowns x, y> Zj X for a maximum
:
'
(6)
+ Fy + Ez + Fx + By + Dz + Ex + Dy + Cz +
Ax
(0, 0, f),
\x
= = =
0,
\y
\z
^
0,
yields a
maximum.
0,
E=
Chapter VII,
5.
ROTATION
and the new F(x,
.
195
y, z)
y, z)
F(x,
Ax*
+ 2Fxy + By* +
Cz\
the coefficient of the term in xy does not vanish, it can be made to do so by a suitable rotation of the axes about the axis
If
2. Thus F(x,y,z) is cf. Analytic Geometry, Chap. XII, reduced finally by at most two rotations (these may be combined into a single rotation, but that is unessential) to the desired form
of z;
(7)
F(x,
y, z)
= Ax*
By*
Cz\
C may be any three numbers, positive, negative, or 0, except that we have excluded as trivial the case that The original equation (1) will obviously repreall three vanish. sent an ellipsoid if and only if the new coefficients A, B, C in (7)
Here, A, B,
are all positive.
We
THEOREM.
F(x,
be
y, z)
An arbitrary homogeneous quadratic function can be reduced by a suitable rotation of the axes of coordisum
of squares.
nates to a
The new
coefficients of
r
,
y', z
may
any numbers,
positive, negative, or 0.
EXERCISE
Show, by the method of mathematical induction and Lagrange's Multipliers, that an arbitrary homogeneous quadratic function in n variables,
can be reduced to a sum of squares by a suitable rotation. By a rotation is meant a linear transformation
:
x{
an x1
ai n x n
x' n
a n i#i
+
J_
\
+ a nn x n
,
x n ) and
r '2
_L
\
4. *n r '2
\
*\
4. An r 2)
\
of the transformation,
A = S
which necessarily has the value
an
1, is
a nn
equal to
1.
196
MECHANICS
It is easy to write down the conditions that must hold between the coefficients of the transformation, but these conditions are
not needed for our present purpose. any two rotations is a rotation.
3.
Continuation.
In the fore-
going paragraph we have been content to show the existence of at least one rotation, whereby the given function is reduced
to a
sum
of squares.
new
coefficients,
We have not computed the values of the nor have we determined the lengths of tho axes.
Now, any
into itself
*'(*',
*/',
z')
s
fl
*(*',
',
z')
*(x, y,
z).
The function
= F
\3>,
F'
X*',
:
a?
is
'
'
00 ~
'
since
Hence Equations (6) of the preceding paragraph be of the same form for the transformed functions. When
vanish.
F'(x', y', *')
A'x'*
+ B'y'* + C'z">,
:
+ X)(B' + X)(C" + X) =
A
0.
Thus the
+\
F
(8)
As
B+
E D
C
ROTATION
197
arc seen to be the negatives of the coefficients A', B' t C', and so the axes of the quadric are found. If the roots of the deter-
minant
axes are
(8)
are denoted
by \ 19 X 2 X 3 the lengths
,
,
of the semi-
In case a X
cially, to
0,
two planes.
the quadric reduces to a cylinder, or more speAll three X's will vanish if and only if the
When tho X t have once been determined, Equations (6) give In general, the three the equations of the axes of the quadric. are and then \i represent a right line distinct, Equations (6)
for each X.
4.
Let
By
(1)
a Localized Vector. be given, and let Let r be the vector drawn from to P.
meant the
vector, or outer
product
A.
We
have met
the vector
M
=
r
F.
is
The moment of momentum of a particle with respect to a point defined as the vector |
(2)
ff
mv,
FIG. 102
whore
the vector drawn from the point to the particle, and v is the vector velocity of the
r
is
particle.
The moment
to
a point
is
(3)
ff
1=1
* Cf. Appendix A. f Contrary to tho usual notation of writing vectors in boldface, as a, x, i, etc., or by parentheses, as (co), it seems here expedient to denote the vector moment of momentum by a, the vector momentum by p, and the vector angular velocity
J)
rk
X m k vk
by
o>.
198
MECHANICS
where rk is drawn from the point in question to nik, and v^ is tin vector velocity of m,k. In the case of a continuous distribution of matter the extensioi
of the definition
is
made
in the usual
way by
1
definite integrals.
:
y
(4)
dx
'dt
m dy m dz dt dt
dz
dz
di
dx
dx
And
so,
for a
:
system
of particles, th(
dx k
(5)
dz
dx
Rate of Change of
(&}
*L (
<r.
Since
^y.
dt \
it is
dt
di dt
dz x y-^~v
(7)
These equations,
(8)
in vector form,
become
=
ljt
^ mkTk X a
*'
fc-th particle.
ROTATION
The
from
result,
If
199
Equation
(8),
(3).
we
differentiate
Equation
.
(2),
we
.
find
da = di dv _ mrx __ + w xv
^,
^,
._*
Now,
-77
=
do-
and
at
X
r
0.
Hence
dt
y dv = = mr X
.
dt
>
Similarly,
from
(3)
we
derive
(8).
5.
of
Moments.
any system
In Chap. IV,
of particles in a
3, it
plane such that the internal forces between any two particles are equal and opposite and lie along the lino through the particles,
the
moments
and the
equation of rotation
becomes
(1)
I;
m k (x
^
its
d
yk
ji*)
fc=
2
1
(Xk
Yk
three dimensions.
proof can be generalized at once to space Newton's Second Law of Motion is ex-
mk
by the equations
_i_
J
d*Xk _ v '<~W~ Xk +
= Yk
where F*/ denotes the internal force which is exerted on the particle w* from the particle my. Multiplying the third of these
z k and adding, and observing equations by y^ the second by that the moments of the internal force cancel in pairs, since the
forces
F/jfc
and
line
of action, the first of the following three equations The other two are deduced in a similar manner.
obtained.
200
MECHANICS
(2)
ments.
(3)
S-i>XF,
^7
or:
2) (Moments
about 0).
Equation (3) can be deduced more simply by vector methods. Write Newton's Second Law in the vector form
:
(4)
mk ak =
F*
+
F*
2) Fw
m k Tk X
a*
fc
2)
r*
F*/,
The sum on the left is equal to d<r/dt by 4, (8). On the right, the vector moments of the internal forces cancel in pairs, and there remains the right-hand side of (3).
and add.
The
rate
of change
is
system of particles
moment
and opposite
and in
or,
(2).
foregoing result applies to the most general system of particles, subject merely to internal forces of the very general
The
nature indicated.
we extend
By the usual physical postulate of continuity the theorem to the case of continuous distributions
ROTATION
of matter, or to
201
set.
system
6.
is
we
will
Vector
Form for
be an arbitrary fixed point in space, and let f be the vector drawn Let from to the centre of gravity, (?, of a material system. F n be the forces that act; i.e. the applied, or external, Fi,
,
forces.
is
Then
Mass
(1)
M
=
.,
df/dt. Equation (1) is merely the vector form of 1. It can be derived by vector Equations A), Chapter IV, methods, by adding Equations (4), 5, and observing that
where v
Mi =
Let p denote the
2)
=
1
momentum,
p
v.
:
Equation
(2)
(1)
now
Thus we have for any system of particles, rigid or deformable, and even for rigid bodies and fluids, the two equations of momen-
tum
\
:
7.
The
Invariable Line
and Plane.
act,
<
"
202
MECHANICS
cr
remains constant.
The
line
through
collinear
is
and a plane
perpendicular to
the invariable plane with respect to 0. a case in point, if we may neglect any force system the stars may exert. We may consider the actual distribution
The
solar
is
and velocities, and then, on choosing a fixed point, the 0, corresponding value of a will be constant. Or we may replace the sun and each planet by an equal mass
of matter
concentrated at
its
centre of gravity,
a-
and consider
this system.
will
be
constant, and obviously nearly equal to the former a. Let us choose one of these cases arbitrarily and discuss it further. The vector a depends on the choice of 0. Can we normalize The centre of mass of the solar system is not at this choice? since and we are neglecting any force exerted by the rest, so, of the system, p = Mv, is constant and the momentum stars*, The direction of this vector, p or v, does not depend on 5^ 0. The point 0' can be so chosen that a' is collinear the choice of 0.
with
p.
We
(2)
shall
show
Equation
V,
(5),
that
a'
+ Mr X
where
collinear with
If a is not already 0' respectively. let <r be resolved into two components; one, the other, cr at right angles. We wish, then,
,
that
Mr X
Q
of
(7
<T
0.
Since
this
The point O will be any point of a is known as the invariable line of the
discussion,
8.
cf.
line collinear
solar system.
I,
For a further
p. 242.
Transformation of
and
P
o
r
<r. Let be a point fixed in space, O' be a second point, moving or fixed. Let be the position of a particle of the system. Then
let
CD
f
I
Fia. 103
= =
r'
+r v' + V
where
relative to 0',
and V
is
the
ROTATION
For a
to
is
203
moment
of
momentum
mr
with respect
the vector
<r
=
<r
rav
mr'
(v'
+v +
)
X
v
.
v,
or
(2)
mr'
+ mr X
=
r'
mr'
The
first
wv',
referred to 0' as a
or the relative
point.
moment
of
momentum, we
moving
For a system of
particles
*
infer that
x
<r' r
v*
or
(3)
<r
+ Mr X
moment
v
of
Mi'
where
o> is
the relative
;
momentum
referred to 0' as
;
a moving point v is the velocity of the centre of mass is the vector drawn from 0' to the centre of mass.
and
'
the right,
Mr X
can be interpreted as the
of the total
X Mv,
to O,
M
is
mass,
concentrated at
third term,
O and moving
1
v.
The
the
j
Mr'
f'
X Mv
moment of momentum, relative to 0', of the total mass, concentrated at the centre of gravity and moving with veV
.
locity
If,
in particular, 0'
be taken at
<r
(?,
then
v,
f'
0,
v,
and
(4)
<r' r
+ Mr X
moment
where
to
o>
of
momentum,
referred
as a
moving
point,
and
Mr X
is
X Mv
the
moment
of
at
G and moving
momentum, Mv, of the total mass, concentrated with the velocity of G, referred to 0.
204
Let
point;
cr'
MECHANICS
denote the value of a referred to the point 0' as a fixed
i.e.
X
Then
</
vk
5J
m^i X
=
=
v'r
vi
mk ri X
X
v
.
or
</
Mr'
:
Thus Equation
(5)
(3)
+ Mr X
in (3).
v.
v =
Moments about
of
We
mental Equation
(I)'
Moments,
S
the Equation of Transformation,
8, (4)
:
And we have
(2)
a
o> is
*' r
+ Mf X
of
v,
where
the relative
moment
momentum,
referred to
as
a moving point.
Differentiate this last equation, observing that
since df/dt
/ON (6)
is
cither
v.
Thus we
find
da
dt
_ ~
dff'r
Ht
r
+M X
r'
d? Tt
On
where
(4)
=
;
+f
X
r,
f are
drawn from
5)
t
r'
from 0.
F*
Thus
5)
Ic
r*
X F =
g
k
:
r;
+fX
F*.
and
Substituting in Equation (1) the values found in Equations (3) (4), we obtain the result
ROTATION
Recall the Equation of the Motion of the Centre of Mass,
(6)
205
6
:
"
it
From
follows that
Mr X
Thus
^=
X
fc
F*.
In this equation
as the
is
embodied the
result
which
may
be described
PRINCIPLE OF MOMENTS WITH RESPECT TO THE CENTRE OF The rate of change of the relative vector moment of momentum, referred to the centre of mass G regarded as a moving point, is equal to the sum of the vector moments of the applied forces with
MASS.
respect to
G:
EXERCISE
is dynamically equivalent, in general, masses on the axis of or, a second pair on the axis of y, and a third pair on the axis of z, each pair being situated symmetrically with respect to the origin, and all six disit being assumed that the tances from the origin being the same
Show
to a pair of equal
Discuss
Moments about an
*
Arbitrary Point.
:
general transformation,
(1)
8, (3)
<j' T
+ Mr X
+ Mr' X
and
differentiate
On
(3)
2
t
r"
x F* =
2
I
ri
x F*
206
MECHANICS
the Equation of Linear
From
(4)
Momentum,
ro
6, (1) follows
that
MIO X
%-2
G/v
xF
V
*-
^^T
Moreover,
V
-jj
-f-
For,
hence
=
v
X
v'
v'
+v X
v
=
0.
v',
and
+ v' X
Moment
of
Momentum,
B):
g=
and reducing, we
find
:
2r*XF*, t
This equation is general, covering all cases of taking moments about a moving point 0', relative to that point. When, however, one uses the expression: " taking moments about a point O'"
the meaning ordinarily attached to these words
tion
(6)
is,
f=?
be true.
f
ri><
F*
shall
(7)
'X
=
Let 0' be a point which
r ft
r).
Then
ROTATION
At the
instant
t
207
r,
/ dv *\ (-5-),.,
a2f
<A
(IF;,.;
Then Equation
t
(7) is to
T.
Thus we have
and
(7)
by 0'
case of
curves,
cf.
Consider the 11. Moments about the Instantaneous Centre. motion of a lamina, i.e. a rigid plane system, in its own plane. Let Q be the instantaneous centre at a given instant, t = T. Then <r, referred to the point Q, is a vector perpendicular to the plane, and its length is
T 1
de
dt'
where /
is
What
does
it
it
foregoing
moment of inertia of the lamina about Q. mean to "take moments about Q"? From the means to take moments about a point 0' describing
the
TO
a curve
= *o&
T)
which at the instant t = r passes through Q. There is an unlimited set of such curves. Let us select, in particular, the curve C which is the path of that point fixed in the lamina, which passes through Q at the instant t = r. Observe that this is an arbitrary choice of C. This curve C is known in terms of the rolling of the body centrode on the space centrode. The velocity of 0' at Q is 0, but its acceleration, if Q is an ordinary point, is normal to the centrodes at Q and does not vanish 10 is to be satisfied, the Chapter V, 5. If, then, Equation (7), centre of gravity, G, must lie in the normal to the centrodes. In the normal to the centrode must particular, pass through body the centre of gravity. Hence the body centrode must be a circle
;
with the centre of gravity at the centre, if the condition is to be permanently satisfied. The equation of moments now becomes
:
~
I-JJ2
(Moments about
Inst. Centre).
208
MECHANICS
case, then, of
The only
permanently thought of as a circle rolls on an arbitrary curve, the centre of gravity being at the centre of the circle; and the limiting case, namely, that the point Q is permanently at rest. This last case corresponds
to the identical vanishing of dvjdt.
take
motion in a plane, in which we may moments about the instantaneous centre, a point fixed in the moving body, is that in which
Point.
Consider
it
now an
arbi-
trary point 0' fixed in the body. t = r, and let C be the curve,
TO
Let
be at
at the instant
(J,
T),
which
it
is
describing.
reference
Then
d?0
'
dt 2
If,
furthermore, Equation
:
(7),
10
is
satisfied,
the equation of
moments becomes
/-72/)
=
^77/2"
(7) here means, in general, that the acceleration of 0' In particular, collinear with the line determined by Q and G. at Q ; or if the equation is satisfied if the acceleration of O' is
Equation
is
coincides with G.
EXERCISE
by the
that
cue.
it
is
while there
slipping.
Show
moments about
it
Find the points of zero acceleration and verify the fact that is possible to take moments about them, explaining carefully
acceleration passes through the on a circle through the centre of the ball,
what you mean by these words. Show that the points whose
centre of the ball
lie
above
ball.
12. Evaluation of <r for a Rigid System; One Point Fixed. Consider a rigid system of particles with one point, 0, fixed. * Ed ward V. Huntington has discussed this question, Amer. Math. Monthly,
vol.
XXI
(1914) p. 315.
ROTATION
The motion
]
209
cf
then one of rotation about an axis passing through 8. Let the vector angular velocity be Chapter V,
is
denoted by
w,
and
let a, p,
7 be a system
of mutually perpen-
dicular unit vectors lying along Cartesian axes with the origin at 0. When we wish these axes to be fixed, we shall use the
In the general (x, y z) and replace QJ, j3, 7 by i, j, k. be shall the coordinates case, 17, f Let P be any point fixed in the body, and let r be the vector
coordinates
y
drawn from
to
a)
r
velocity of P,
lift
+ f y.
The
(2)
"
co
is
as follows (Chapter V,
r.
9)
(3)
co
In Cartesian form,
(4)
or
(5)
For a
(6)
rav,
y
(7)
=m
Hence
(8)
210
MECHANICS
:
(9)
m = m m
[(rj*
) C0
(f
&CO,,
- rft + - fcof -
ftw,
:
2 (
2
i;
and
so, finally,
(10)
or
=-
77,
when the
origin
is fixed.
It is
obviously im-
Euler's
rigid
Moments,
(1)
Moments about
),
referred to this point, admits a simple expression in terms of the Let the (, r/, f)-axes be fixed angular velocity, &, of the body.
in the body, Let law.
and
let
<*
+
dr
dt
T70
+ f%
to P.
where r
is
Then we have
seen
(Chapter V,
cr$
(Ty 13
<TS
7,
side of (1).
ROTATION
On
the right-hand side of (1) let
211
X
Thus we have
:
F*
= La
+ M/3 + Ny.
=
dt
L,
(2)
dt
dfft
~
at
_j."
C0
O"TJ
== WTJ (7A
xV.
-LJ
12, the substituting for <T, o^,, a^ their values from (10), In result. known as Euler's Dynamical Equations equations
On
77,
and Equations
(2)
dp -
(3)
where
When
(4) 1
dt
dt
dt
(En,
(C
B)
w^
L,
and two others obtained by advancing the letters cyclically. Eider's Dynamical Equations also apply to the rotation of a rigid body about its centre of mass 9. Here, there is no restriction whatsoever on the motion.
;
212
14.
MECHANICS
Motion about a Fixed Point.
:
the action of no forces, save the reaction at 0. Equations become the following
Then
Euler's
first
integral
is
spectively
by
p, q,
and
(2)
Ap*
Bq*
Cr*
h.
This
is the Equation of Energy, Chapter VII, 5, 6. second integral is found by multiplying Equations (1) spectively by Ap, Bq, and Cr, and adding
re-
dr
(3)
A 2p2
+B
+C
r2
=
dt
0,
and so
a
is
= Apa
and
Bqf$
Cry
2 p and and then,
constant.
From Equations
#
2
,
(2)
(3),
two
of the variables, as
can, in general, be determined in terms of the third, in the third equation (1), a differential equation It is seen that t is expressed as an elliptic for r alone is found.
on substituting
r.
Thus,
p, q t
and
r are
found as
functions of
Exercise.
t.
Let
A =
1.
3,
B =
2,
C =
and
t
let p, q, r all
have
in-
Work
in
terms of the
ROTATION
The Body Cone.
tracting,
(4)
213
(3)
On
multiplying (2) by
Z,
by
r2
A,
and sub-
we
find
A(l-
Ah) p*
+ B(l-*
Bh) g
C(l
Ch)
0.
The equations
(5) '
p
p, q, r are the
r'
t.
when
above functions of
is
2
T,
Hence the
:
locus of
A(l-
Ah)
? + B(lry,
Bh)
C(l
Ch) f
(3)
0.
p, q, r satisfy (2)
f) of (5) satisfies
let
and
(6),
Conversely,
TJ,
(,
77,
f)
f)
(0, 0, 0),
/*>
= M,
/if,
Thus
(3)
is
satisfied.
r/,
And
(4)
holds, too.
Hence
axis.
(2)
is
true.
Consequently, (,
f ) lies
on an instantaneous
Consider the ellipsoid of inertia, 2. of the space centrode. Let is a surface fixed in the body.
be the point in which the ray drawn from and collinear with co cuts S.
Then
to S at m is the tangent plane a plane fixed in space, the same for all The motion is seen to be one points, m. of rolling of the surface S on the plane without slipping.
p IG ^4
statement, it is is perpendicular to sufficient to show that the tangent plane at not O distance from does that its and depend on m. The cr,
To prove
the
first
equation of
is
S:
4
coordinates of
+ 5T; +
2
:
Cf 2 =
1.
The
m are
PQ,
PP,
pr,
where
214
MECHANICS
of
is
P Ap^
pBqv
pCrf
1.
of its normal are Ap, Bq, Cr. But of on the <r Hence the axes. these are precisely projections is perpendicular to a. Moreover, the distance of the plane
The
direction
components
M M
from
is
1
P^l
and so
ment.
is
h J *
l
'
constant.
first state-
To prove
much
of the
Let F be the curve on S which cone, (6), as lies in S. the intersection of these two surfaces. Then F rolls on
body marks
M withM
oujb slipping, and the curve of contact, C, can servo as a directrix For the body cone, (6), rolls without of the space centrode. on the space cone, and the curves F, C are two curves slipping on these cones, which curves aro always tangent at the point
The angular
for
velocity, w,
is
propor-
Om
Euler introduced as co15. Euler's Geometrical Equations. ordinates describing the position of a rigid body, ono point of which is fixed at 0, the three angles, 6, p, and ^ represented in the figure. Between the components of the angular velocity
axis,
V =
Sin
COS
<p
d$ rr
dt
+
,
dd
SHI
(p -r.
at
(1)
...
sin
+
,
de
cos tp-r
Ctt
'
dt
dt
ROTATION
215
FIG. 105
with the point C in which the positive axis of f pierces the surface As we look down on the sphere from above of the unit sphere. from the figure that it evident is this point,
dO
dt
= p
sin
<p
+
<p
q cos
<p
sin 6
=
(it
p cos
q sin
<p.
FIG. 106
first
two
of Equations (1).
obtain the third equation, consider the motion of E. Its velocity is made up of a velocity o> tangent to the arc EA, and
To
two
its
velocities
is
perpendicular to this arc. On the other hand, composed of the velocity tp tangent to the arc EA
cos
0,
\ft
also tangent to
EA
and
perpendicular
to
EA.
Hence
and
Equation
(1).
216
If p,
<7,
MECHANICS
r
have onc6" been determined as functions of the time, Equations (1) yield a system of three simultaneous differential
equations of the first order for determining 0, Thus in the problem of of the time. 14,
rigid body under no forces, or acted straint that holds the point O fixed,
p, q, r
<p,
as functions
the motion of a
problem
16.
is
Continuation.
The
Moving Axes.
The moving
by
the scheme,
y
z
is,
to express the nine direction cosines in terms This can be done conveniently by vector
methods, if we effect the displacement one step at a time. Let k be unit vectors along the original axes, and a, 7 unit i, j, Let the first displacement vectors along the displaced axes. be a rotation about the axis of z through the angle ^, and let i 1; jj. Then i, j go over into
/ft,
it
J!
= =
cos
i
+
^
sin
j
^
^
sin
cos
k *V|
ix
JEL.
jt
:
through an angle
6,
whereby
and k x into k 2
= 7
it
= = J k =
12
2 2
cos
sin
Ji
ij
sin
+
=
<p
kj cos
6.
i2
k 2 through an
j3
:
angle ^, whereby
goes
over into
i3
= a and
J2
goes into
13
= = =
i2
cos
i2
j3
j2
k,
k,.
ROTATION
From
these equations
Zj
217
it
appears that
cos
<p
cos
cos
sin
v?
sin
<p
sin
m = = n!
l
cos 6 cos
sin
(f>
+ sin ^ cos ^
^
^
cos
<p
cos
Z2
= =
<p
cos
sin
sin
cos
p cos ^
n2 =
= w3 = n3 =
Z3
sin
cos
^ ^
sin 6 sin
cos
0.
17. The Gyroscope. It is now possible to set forth in simplest terms the essential characteristics of the motion of a rotating rigid body, which is the basis of gyroscopic action. By a gyro-
scope
body spinning at high velocity about an axis passing through the centre of gravity, which is at rest, and acted on by a couple whose representative vector is perpendicular
is
meant a
rigid
to the axis.
Let A = J?, Consider, in particular, the following motion. 7* 0, and let the axis of f be caused to rotate with constant
c,
in the plane
^ =
0.
What
de
will
be the couple?
=
P
are unknown, where dd/dt = c. The components L and = 0. The third of the Dynamical Equations becomes but
~ = 0;
dt
hence
v,
and
P is
Since here
.
first
two equations,
c sin
vtj
p =
c cos
vt.
218
MECHANICS
substituting these values in the
On
find:
Dynamical Equations, we
Ccv sin
of
vt.
L =
Ccv cos
vt,
M=
We may
at the point
made up
a force
acting
(, 77, f) (0, 0, 1) in Fig. 105, and an equal and opposite force at 0. Then F will be tangent to the sphere at Let it be resolved into two components, one perpendicular C.
:
to the (, f )-plane ; the other, in that plane. The first will have the value L, taken positive in the sense of the negative ry-axis ; the second will equal M, taken positive in the sense of the -axis.
When
0,
L =
Ccv,
M=
0,
and at any later time, the result is the same. This can be seen directly from the nature of the problem, since the motion of the axis of f in the plane \f/ = is uniform, and hence the force which
at one instant as at
the constraint exerts will be the same force relative to the body any other instant.
It is easy to verify analytically the truth of the last statement. will always be For, the force normal to the plane \l/ =
L
and the
cos
<p
M
+
sin
<p
Ccv
L sin
essential
if>
M cos p =
0.
phenomenon
to
in
if
exerted
standing on the gyroscope and moving with it, one's body along the positive axis of f and facing in the
direction of the motion of the axis, the force
ap-
on the
FIG. 107
constraint will be directed toward the right, the gyroscope spinning in the clockwise sense as ono looks
down on
it. Of course, if the sense of the rotation were reversed, the sense of the reaction would be reversed also.
ROTATION
EXERCISES
1.
219
Show
r
that,
if
no assumption regarding
-
6 is
made, but ^
and
v,
then
T = A L A
sin vt -rz
at*
d*e
de Cv cos vt -T: +n at
.
TM M
A = A
COS
vt -T7T 1
d* e
at
dd ~ n Cv Sin vt -rr<
at
the axis presses against a rough plane, \l/ * gential force being p, times the normal force, then
2.
If
0,
the tan-
dB
and furthermore the point of the axis in provided dO/dt > contact with the plane moves backward, i.e. in the sense of the
decreasing 6. On the other hand, the axis must have a sufficiently large radius so that the requirement below relating to the motion of
the point of contact can be
fulfilled.
Hence
Q 9
dO 37
at
c^ ce A
and
,
= cA
-/ Y
^
A
CIJLV
whore
denotes the
initial
cA/Cp,v.
Moreover,
VSS
-T 8
'
where
refer to the point in which the sphere v = dO/dt and s = is radius cut axis. the 1) (of by
3. Prove that, in the problem of the preceding question, the normal reaction of the constraint is
acts 4. Show that, if a (small) constant couple, of moment on the gyroscope, the vector that represents the couple being = and directed in the proper at right angles to the plane
,
\l/
* If we think of the material axis as a cylinder of small radius, there will be a small couple about the axis, tending to reduce r. But as this couple approaches when the radius of the cylinder approaches 0, we may consider the ideal case of an axis that is a material wire of nil cross section, the couple now vanishing.
220
sense,
MECHANICS
and
if
in the plane
move
This
6.
last
equation
is
true,
even when
scope always being constrained to move in the plane ^ the reaction on the constraining plane \l/ = is numerically
0,
its
\l/
when
is
dO/dt
>
0,
but the
opposite
6.
when
dd/dt
<
0.
It has
equivalent dynami-
\l/
\l/
7.
point of contact
Discuss the problem of Question 2 for the case that the is allowed to slip forward. Consider also all
<
initially.
is
material
a point
the case of a fixed peg, supported at Let the positive axis of f pass through the
centre of gravity, (7, distant h from 0. The third of Euler's Dynamical Equations becomes, since the both pass applied forces gravity and the reaction of the peg
C%
CV>
0.
A (p 2 +
H-
2Mgh
cos
6.
ROTATION
Furthermore, the vertical component of the vector a For, the applied forces giving a vector moment at
is
221
constant.
reduce to
gravity, which is vertical, and so its vector moment with respect to O is horizontal. Now, the components of o- along the moving
Hence the
vertical
component
of a
16)
Bqn 2
On
(3)
substituting for n lt
n 2 nz
,
their values
sin 6 sin
<p
from
16
we have
Ap sin 6
cos
<p
Aq
Cv cos
= K.
:
p =
q
sin
cos <p-~
dt
+
,
dO
<p-jr
sin
(4)
dO
cos
-=r
at
On
p and q
in (2)
and
(3)
we
find
(5)
2 sin e
-- =
bi>
cos
(6)
= C
the constants
motion;
i.e.
a and depending on the initial conditions of the they are constants of integration; whereas a and
<p
after
found from
(5) as functions of
(7)
<p
vt
cos 6
dt
(5)
dt.
and eliminating
d\l//dt,
we
sin 2 6
(~) =
sin 2 e(a
a cos
6)
(J8
6i>
cos
0)
222
MECHANICS
result is
6.
The
able,
(9)
It
a differential equation for the single dependent can be improved in form by the substitution
vari-
u =
2
cos 9
(10)
(1
u*)(a
au)
CJ
bvuY
f(u).
Thus f(u)
Equation
seen to be a cubic polynomial, which we will presently discuss in detail. But first observe that the second
is
(5) gives
d*
""
bvu
Hence
by a quadrature
after
To sum up, then, we have reduced Retrospect and Prospect. the problem to the solution of Equation (10) for u as a function
of
t.'
Equation
<p.
(9) gives 0;
We may
Equation
19.
(1)
(10).
Continuation.
The Polynomial
fM
u
u*)(a
for
au)
(0
- bmY
It is
becomes positively
for
infinite
=+
oo.
negative or
will
=+
1,
1.
Hence
in general the
graph
be as indi-
cated, or
</(M),
u,
<u<u
2
fM
Moreover,
1,
and
roots
>
1.
The
w^ u 2
will, therefore,
be simple roots.
The
differential
equation
in
Appendix B.
In particular,
u =
<*>(0
values of
ROTATION
223
(4)
or
(5)
<p(t
if
+
*!
=
Furthermore,
(6)
then
And
//w\
similarly,
if
/, \
then
/*7'\
/*
\t )
^C*2
T)
/j
^(^2
~T~
Ty.
Let a sphere /S be placed about Physical Interpretation. as centre, and let be the point of intersection of the positive axis of f with S. Lot C be the curve that describes on S. The
show that C
u
lies
parallels
of latitude corresponding to
(8)
= u
lt
= u2
parallel,
For convenience let t be measured from a point on the upper u = u 2 Then there are three cases according as initially
.
III.
<
0.
FIG. 109
CASE
I.
Since
bvu
eft
u2
when u has its greatest value, u 2j d^/dt will remain when Let ^ = and so ^ will steadily increase with t. positive, increase t = 0. t to As increases to iT, will
is
positive
\!/
224
MECHANICS
=
T, \p will have in^(0, Equation (3). When t of curve C will have arch the and one complete by ^, been described. The arch is symmetric in the plane ^ = -J-^. The rest of C is obtained by rotating this arch about the polar
where u
creased
axis of
S through
II.
CASE
Here,
at the start,
and hence
bvu 2
0.
Since
decreases,
it
motion
<
and so
\[/
bvu,
steadily increases.
parallel of latitude.
CASE
III.
Here
is
if
bvu
<
at the start,
forever.
this
conceivable that this relation should persist were not the case, it is still conceivable
when
is
should be
less
^ =
0.
\l/
That
corre-
sponding to the first return of P to the upper circle is positive has been shown by Haclarnard.* The curve C has double points in this case, but it proceeds with increasing t in the sense of the
advancing
discussed,
^, as indicated.
Special Cases.
There
:
is
still
one of which
is
<
MI
tig
<
1.
Since
/(I)
in all cases,
and
since
there
must be a
third root
^
(w
1.
Thus u
u,
1
is
f(u)
=
0,
where
x(u)
<
<
<
1.
ROTATION
The only
on the value
225
when
0, is
u = wt
when u =
There
in
The point north or it may the with a may pass through pole velocity ; a as north the pole limit; or the gradually climb, approaching the rotate about axis. top may permanently polar Similarly,
1 is
a root.
is
a great wealth of literature on the gyroscope and The reader can refer to the article on the Gyroscope
EXERCISE
peg
Treat the case of a top on a smooth table. Assume that the The distance, then, from the is a surface of revolution.
centre of gravity to the vertical through the point of contact will be a function of the angle of inclination of the axis.
Assume axes
of gravity.
fixed in the
Write down i) the equation of energy; ii) the equation that says that the vertical component of or is constant. From this point on the procedure is precisely as before, and
the result again a differential equation of the type treated in Discuss all cases, and show that in general the axis oscillates between two inclinations, both oblique to the
is
Appendix B.
vertical.
is
a point.
Having
studied this case in detail, proceed to the general case and study Then derive the special case as a particular it in detail, also. case under the general case.
The most general 20. Intrinsic Treatment of the Gyroscope.* case of motion of a gyroscope reduces to one in which a single couple acts on the body, and this couple can be broken up into
*The results of this paragraph arc contained in a paper by the Author: the Gyroscope," Trans. Amer. Math. Soc., vol. 23, April, 1922, p. 240.
"On
226
MECHANICS
two couples one, represented by a vector at right angles to the axis of the gyroscope ; the other, by a vector collinear with the axis. In the most important applications that arise in practice,
rise to
the latter couple vanishes. But in the general case, the third of the Dynamical Equations in the form
it
:
gives
The former couple can be realized by a single force F perin which the pendicular to the axis and acting at the point the other force of the positive f-axis cuts the unit sphere,
Let
on the unit sphere by P, and let S be the cone which is the locus of the axis of the gyroscope, and of which C is the directrix. Consider the rate at which the tangent plane to S is turning when P This quantity shall be denoted describes C with unit velocity. as the bending of the cone and represented by the number K. It is also the rate at which the terminal point of a unit vector drawn from at right angles to the tangent plane traces out its path on the unit sphere. K shall be taken positive when an observer,
to his right.
Let
with the sense of the increasing ^ to the tangent to C with the sense of the increasing s. Then it appears form an infinitesimal treatment that
'
__
ds
ds
Since
tan
V =
d\f/
sin B
-,
or
V=
tan" 1 -77-;
-,
and
since
d6*
+ dj* sin
0'
0,
or
/2
+
/2
V* sin 2
1,
follows that
jc
(3)
*
It
= WB" -
iH
sin
(1
V cos
0.
The
may
point O need not be the centre of gravity in the following treatment. be any point fixed in the axis of material symmetry.
ROTATION
227
From the definition it follows at once that the bending of a cone of revolution must be constant. To find its value, let the
Then
coordinates be so chosen that the equation of the cone the length of the arc of C is
5
is
a.
= ^ sin a
is
and so
\l/'
sin
a =
1.
From
(4)
(3) it
now
seen that
K
cot
oc.
lies to the right of the observer, as he travels along C. he reverses his sense, the sign of K will be changed. But both cases are embraced in the single formula (4), the second corresponding to a cone whose angle is TT a, or
The cone
If
s is
replaced
by
s.
cone.
nating
ntt
(5)
=
whenever
^'
the
sign holding
<
0.
Hence
If
is
one solution,
constant, set K = = a, or
cot a.
Then Equation
a;
(6)
admits
7r
theory of differential equations, this is takes on the value a, or at a point s = s derivative vanishes there.
,
a,
and whose
that
From
2
sin B
,~
(7)
-J-T^
cos 6
sin 2 6 cos
where the
sign holds
whenever
\f/'
<
0.
* These results are inserted for completeness. They will not follows, and the student may pass on without studying them. of interest to the student of Differential Geometry.
be used in what
They
are chiefly
228
If K is
MECHANICS
known, or given, as a function of s, then Equation (6) is then found by a quadas a function of s, and
\l/
determines
rature
(8)
:
The bending,
K, is
K* =
cf.
K2
+
i
1.
Furthermore,
Fig. Ill
below:
j
k
z
z'
(10)
n = a
x
x'
y
y'
(11)
hence
=
(12)
KZ
'
yz"
zx"
xy
- xz" _ yX ".
Since
K\ =
1'
and
(13)
^
(9) follows at
x"*
y"*
z"*,
formula
(13).
Moreover, from
x
(14)
y
y'
z
z'
*=C
x'
connected with
/c
by the
relation
d
(
Ts
v *=
l
T,
ROTATION
229
21. The Relations Connecting v, F, and *. The physical phenomenon which it is most important to bring home to one's
F on
scope. Any such explanation must take account of all three But many popular explanations claimquantities, v, F, and K. to be but ing correctly "non-mathematical," incorrectly to be fail because they are unaware accurate in their mechanics
of
Thus, for example, the statement often made that when a couple is applied to a rotating gyroscope, the forces of the
K.
"
couple intersecting the axis of the gyroscope at right angles, the move in a plane perpendicular to the plane of the forces In fact, the axis will begin to move of the couple" is false.
axis will
if
it
starts
from
rest,
initial
inter-
of the
F,
and
K,
(1)
and requires no further comment than i) that it is perfectly general, applying to the motion of the gyroscope under any forces whatever; and ii) that in the case which most interests us, namely that in which there is only the force F (and the reac= 0, and so r = a constant. tion at 0) we have
:
*>,
Let F, then, be resolved, in the tangent plane, into a component along the positive tangent, and a component Q, taken positive
the
left
of the observer;
i.e.
is
posi-
Then
(2)
AKV*
Crv
Q,
where
r
ds/dt
and
motion of P,
"On
230
Proof.
is
MECHANICS
Let the unit vector from
to
;
let t
and
\p X
t
with regard to them as ft is with regard to 7 and a. These are principal axes of and the moments of inertia about inertia,
them are
L =
In
A,
Ia
C.
velocity
FIG. Ill
o>t
0,
wn =
I),
Now,
(T
In
C0 n
It
CO*
Ia
C0 a
a.
Hence
(3)
From
this equation
we can compute
-rr
do-
dv
dn
da
It is clear that
da
(4)
vt.
dt
it
appears that
(5)
dn =
-j-r
KVt
Hence,
(6)
finally,
d<r
= Av-n
(Aw*
+ C)t + C
:
which represents the resultant moment of be written in the form the applied forces about
M
Since
n
da
+M
a a.
-.
ROTATION
we have
(7)
:
231
Av^us
n,
A Kv* +
in
Crv
t,
C% =
dt
a.
acts at
OP
F=-Qn+Tt,
we
see
that
T,
established.
Equation
(1) is
We
form:
(8)
AKV*
Crv
= Q
22.
The
first
of
Equa-
tions (8),
AV
=
ds
r, '
admits a simple interpretation. It shows that the point scribes the curve C exactly as a smooth bead of mass
de-
m=A
in the
form of
if it
were acted on by
C~ dt
^
shows that the component
permanently at
acted.
rest
= *' N
& about the
would
the axis were
The second
A)
equation,
AKV*
Crv
Q,
expresses the sole relation which holds between the four variables In the applications, however, r is constant, r = v, K, v, r, and Q.
AKV*
+ Cw = Q
K, v,
and Q.
232
MECHANICS
F =
is
The Case
but the axis
gives
i)
0.
vanishes,
not at
Here,
Q = =
0,
0,
T =
0.
Equation A)
AKV
Cr
or,
1/| K
|,
and choos-
ing r
>
Av
p==
If r is
'Cr' v
a positive constant, r
>
Cv
0,
then
and
since v
is
constant, for
A
__
n
'
FIG. 112
ds~
and negative.
tan
K is also constant,
The
is
a,
where
|,
or
a =
p,
of the description is such that the observer, walking along C in the positive sense, has the cone on his right.
is
P
FIG. 113
and
Q =
The
r
is
Crv,
or
Q = Cw,
the motion of
no matter what
T and
along
its
path
may
be.
pressure of the axis against the constraint, in a normal direcor, if tion, is to the right, and is proportional to r and to v ;
constant, to
v,
the coefficient then being Cv. Thus we obtain minimum of effort, the main result of 17.
We
A)
:
can
now
give a
Am*
The
left-hand side
is
Crv
of
Q.
the
sum
two terms.
Q2 =
Crv,
ROTATION
the sphere;
left, for
233
move
tangent to C.
This force,
Q2
l
is
Q2 >
0.
The
first
term,
~ Q =
Its
AKV*,
numerical value,
can be interpreted as the centripetal force exerted on a particle, of mass m = A, to make it describe a circle of radius p with
velocity
v.
Wlien
and so
is
and vice
versa.
n at P, which along the normal (combined with the smooth constraint of the surface of the sphere) would be required to hold a particle of mass m = A in the path C. Let the vector a be written in the form
Consider
the force
now
xi
yj
+ zk.
Then
v
where
x'
dx/ds
t
x
f
n = a
(yz
yx')k.
The
acceleration, (a), of
P in
zi
space
is,
of course
() =
Now, the component
to the plane of a
+ ^k.
of the acceleration
t
is
and
form
x
x'
y
f
z
z'
y
y
Since x
x
vx',
it
follows that
v*x"
+
x
vx',
etc.,
and so
x
x'
y
y'
z
z'
234
MECHANICS
Thus
mw
is
(7, which would be required to hold a particle of mass w, describing (7, in its path; the component along t being mvdv/ds, and the third component, along a, being the reaction normal to the sphere, in which we are not interested. It is natural to think of the point pn on the line through P
normal to
If
we draw
of revolution through P, this is the point Q in which that line meets the axis of the cone. An obvious
mw
is
the centripetal
force of a particle describing a circle of radius p, with centre at Q, tangent to C at P, the velocity being v.
_,
.
The
force
Let an electro-magnetic
a north-pole situated at 0, and let the particle carry a charge, The force exerted on e by the field will be at e, of electricity.
the path and tangent to the sphere, and, proportional to the velocity, v, of m. Hence e can be so chosen that this force will be precisely equal to Q 2 = CW. In the more general, but less interesting, case that r is variable,
right
angles to
finally,
still
Summary of the Results. To sum up, then, we can say The point P, in which the axis of the gyroscope meets the unit sphere = A constrained to about 0, moves like a particle of mass
:
on the sphere and carrying a charge of electricity, e. The forces that act on m are supplied by the electromagnetic force of the field, Q 2 = Cw, acting on e and a force F acting on m, the components of F along the tangent and normal at P being
lie
y
T and Q
respectively.
e.
The
met by
a variable charge,
particle
less
lie
of the sphere,
we may think
of a
is
mass-
rod of unit length, free to turn about one end which at 0, and carrying the particle at the other end.
*
pivoted
idea of using the above electro-magnetic field to obtain #2 was suggested colleague, Professor Kemble, to whom I had just communicated the results of the text, down to this point. (Note of Jan. 23, 1933.)
to
The
me by my
ROTATION
EXERCISES
1. Suppose the axle P of the gyroscope is caused to a smooth slot in the form of a meridian circle, which is
235
move made
in
to
rotate in
any manner.
The
force
will
Show
that
d2 6
Suggestion
Euler's
2.
Combine
Euler's
Geometrical
Equations
with
Dynamical Equations.
Let the components of F along the meridian in the sense and along the parallel of latitude in the sense and Show of the increasing ^ be denoted respectively by
of the increasing 6
that
If
and
are
known
as functions of
6,
\f/ }
t,
these equations
suffice to
Consider small oscillations of the axis of the gyroscope in Let the neighborhood of the axis 6 = v/2, $ = 0.
3.
Show
equations
4.
,
Generalize the equations of Question 2 to the case that A, are all distinct.
236
5.
MECHANICS
Intrinsic Equations.
From
the equations
T, '
= Av% ds
AKV*
Cvv
Q,
if
T,
are
known
as functions of
and
6.
.
v.
The gyroscope can be used to reduce the massive gyroscope is mounted in a cage, or frame, its axis being fixed with reference to the frame, and The frame is mounted on trunnions, with axis horivertical.
Ship's Stabilizer.
rolling of
a ship.
zontal
and at
a brake to
in the plane
dampen
right angles to the keel, and it is provided with its oscillations about this axis. Thus the
axis of the gyroscope has two degrees of freedom; it can rotate through the keel and the masts, and this plane rotates
Hi)
of the gyroscope
and frame
The
where the
first
is
water; the second, to the righting moment produced by the buoyancy ; and the third, to the force exerted by the trunnions.
of as rotating about the point, 0, in as which the axis of the gyroscope cuts the regarded fixed,
*A picture and an account of the ship's gyroscope is found in the article on the "Gyroscope" in the Encyclopaedia Britannica and in Klein-Sommerfeld, Theorie des Kreisels, vol. iv., p. 797. For the discussion which follows the reader also needs, however, the theory and practice of Oscillatory Motion with Damping cf. the Author's Advanced Calculus, Chap. XV.
;
ROTATION
axis of the trunnions.
237
0,
^ =
0,
is
the
plane
^ =
being vertical.
Moreover,
be replaced by $,
where
The motion
point
(9,
will
of the gyroscope about its centre of gravity, the be governed by the approximate equations of Ques-
tion 3.
Finally, the motion of the frame is governed by the equations These equations are modified by called for in Question 4 above.
the condition
sin
<p
0,
1.
still
further
by
setting
& =
0, cos
# =
Thus
dt
where the first term on the right is due to the brake and other damping, and the second, to gravity, since the frame is so constructed that its centre of gravity is appreciably below O.
A + A'
in
They
are
be projected along the the centre, 0, and an initial determine the motion. of To rotation. arbitrary velocity Let the (x, y)-plane of the axes fixed in space be horizontal. Let moving axes of (, 17, f) be chosen parallel to (#, y, z), but
Billiard Ball.
Let a
billiard ball
initial velocity of
with the origin at the centre of the ball. The point of the ball P, in contact with the table, shall be slipping, and the angle from the positive direction of the axis
of
x or
motion
shall
be ^.
238
MECHANICS
are
at
and
M -jp
CD
rf2
pMg cos
\f/
dt
a-,
moving axes
where
The moment
f Ma
equation,
d<r
2
.
thus gives
/u-6
nMga sin
(2)
--
= nMga cos ^
-
'ifHence
(3)
=
is
:
const.
it
The angle ^
tions (1)
unknown.
Eliminate
by combining Equa-
and
(2)
dt*
5 dt
(4)
^y.
dt 2
dt
Hence
2a
(5)
2a
where A,
conditions.
are constants of integration depending on the initial They may have any values whatever.
ROTATION
Let
239
ball.
Then
Vx = V cos $ =
(6)
-7-
aco,
Vv = V sin ^ =
Combining these equations with
(5)
+ au(.
get
:
we
2\dt
where
^
=B'
A'
Equations
ation let
A',
(1)
=- fB.
For abbrevi-
now take on
dt
Then
du
dt
(8)
dv
Hence
du
v -7T dt
u -77
dt
dv
^
0, y
and consequently
av,
/3
Moreover, u and
v are
0.
Suppose u
>
0,
>
0.
Then
av
av
The proof
two cases
:
7*
ii)
which
= 0. w ^ 0.
240
It thus appears that
(9)
MECHANICS
cos
sn
ft
are constants, and consequently the centre of the ball describes in general a parabola; in particular, a straight line. The direction, however, in which the point
is
always the same; cf. Equations (9). This result comprises the main interest of the problem, so long
slipping, is
is
as there
slipping.
Slipping ceases
when
V=
0,
or
(10)
From
this instant
we
other damping.
the angular velocity is related to the linear velocity of the centre of gravity as follows. Let the centre of the ball be at the origin and let its velocity be directed along the positive axis of x. Then
(11)
dx Tt
au(
|
c>
t-O
#
Jt
t
\
0;
c,
t-o
0,
aco,
-o
wf
t-o
y,
where 7 can have any value, positive, negative, or 0. Let us consider the motion which consists in pure rolling and pivoting, and see what force at P is necessary. First, we have
Mw =
(12)
-
dt*
~ Y Y>
where X,
are the components of the unknown reaction at P. Next, taking moments about the centre of gravity, we find
:
(13)
UOi) ___.
dt
f UWf) i ___. dt
_ ax
-w*
dt
dt
ROTATION
Finally,
Tr
241
y.
(14)
- - a*, - A
(12), (13), (14), together with the conditions (11), formulate the problem completely, and determine the seven unknown functions, x, y, co$, w^, o^, F, as we will now show.
initial
From
it
appears that
first
equation (12),
we
find
But the
equation (14)
siderations
left-hand side of this equation vanishes because the first = 0. Similar conis an identity in t. Hence
show that
Y =
0.
On substituting these values in (12) and (13), these five equations can be solved subject to the five initial conditions (11), and
the other condition, that initially x = 0, y = 0. The centre of the ball describes the positive axis of x with constant velocity, c.
The angular
by their initial values (11). Since 7 is arbitrary, w may be any vector whatever in the (77, f)-plane, whose component along the r;-axis is c/a. The foregoing discussion may be abbreviated by means of the
Principle of
is
components along
VII.
The motion
no
force to be exerted
Cart Wheels.
ized they
To determine
242
MECHANICS
will
We
have
or disc,
mounted
that of a single wheel, begin with a still simpler case so that it can turn and roll freely, but will always
its plane perpendicular to the plane on which it rolls. The frame which guides it may be thought of as smooth. Its mass can be taken into account, but we will disregard it, in order not to obscure the main points of the problem.
We
will
in the disc
choose the coordinates as indicated, the axis of being and always parallel to the plane the axis of 77, being the axis of the disc, is also
;
The
and
is
horizontal.
The
axis of x
directed
down
the plane.
Let
v? be the angle through which the disc has turned about the
axis of
rj ;
let s
scribed
tact,
P
ZT
F = Xa
Yj9
where a, 0, y are unit vectors along the moving axes. Z = Mg cos e, where e is the inclination of the plane, and
Then
Md x
-;-
X cos
sin 6
+ Mg sin
(1)
= Xsin0
The angular
has the value
velocity,
Ycos0
Moreover,
=-
da,
7 =
0.
centre of gravity
(2)
dt
ROTATION
Since
243
=
and
0,
co,,
=
2
,
<p,
o>
A = C =
a2
# = pfa
we have
Hence
or
(3)
=-
Beta.
5^j3
In computing the right-hand side of Equation (2), the couple which keeps the axis of the disc parallel to the plane must be taken into account. The vector which represents it is collinear Hence the couple may be realized by the two with the axis of
.
forces
at
r2
=-0.
Thus
r*
X
:
F*
j8
X F l7 + (-
18)
X (-
F,y)
(- ay)
F;
or, finally
(4)
r*
F,
(2F,
aY) a
(4)
aX0.
find
(3)
and
we
(5)
the form
dx
~dt
v cos 0,
dy
dt
v sin ^,
244
MECHANICS
where
and so
/n\ (6) v '
dx
-TT
dt
d<p n = a-cos 0,
dy -~
dt
a-
d<p
sin
0.
dt
dt
The formulation
functions,
is
now
complete.
X, Y,
namely
x, y, 0, p,
l9
To
(7)
by determining
9
from
(5)
f
t
X,
\t
Next, eliminate Y in
cos<9
(1)
at 1
di L
sin ^
=
J
(5),
+ Mg sin
y
cos
6.
And now X
can be eliminated by
and x y by
sin
e
(6).
Thus
Ma or
(8)
i Ma
~ + Mg
cos
0,
where
Hence
(9)
and
v?
fc
^5
(cos
^ = *(sin-slnM) + M+ M ^ -.
*o,
fc
cos 0)
sin M\
Jt
v?o-
From
X, Y,
(6),
Fu
x and y can now be found as functions of can be determined from (1) and (5).
of
t;
and
finally
(5), (6) is an example of equaHertz because some of them, by involve of the first order only time-derivatives namely (6), and cannot be replaced by geometric equations between the
The system
Equations
(1),
coordinates.
An interesting case of a non-holonomic problem is that of a coin rolling on a rough table. It is studied in detail by Appell,
ROTATION
Mecanique
explicit
raiionelle, Vol. I, p. 242, of
is
245
solution
obtained in
function.
EXERCISES
The student should first, without reference to the book, reproduce the treatment just given in the text, arranging in his mind
the procedure: i) figure, forces, coordinates; ii) motion of the centre of gravity; Hi) moments about the centre of gravity; iv) conditions of constraint v) the solution of the equations.
;
1.
in
the
text.
2. Coin Read casually Appell, rolling on a rough table. adopting his system of coordinates. Then construct independently the solution, following the method used in the problem
of the text.
3.
The problem
of the text,
when
is
taken into account. Begin with the case that the bottom of the frame is smooth and its centre of mass is at the centre of the disc.
4.
in the text,
Study the motion of the centre of gravity of the disc treated by means of the explicit solution of x, y in terms of t.
In dealing with the motion of a rigid body, 25. Resume. there are the two vector equations
:
equivalent to six ordinary equations. It is always possible to take moments about the centre of
gravity.
Principle of Work and Energy frequently gives a useful integral of the equations of motion.
The
If
the right-hand side of the Moment Equation is a vector lying in a fixed plane, the component of a normal to this plane is constant, and thus an integral of the equations of motion is
obtained.
Sometimes there are conditions which are expressed by equabetween time-derivatives of the first order, t = t, but which cannot be expressed by equations between the coordinates only.
tions
246
MECHANICS
first
The
is
to
draw the
figure,
mark
;
the forces,
like nature,
and pass
which
mentioned
may
With the
forces
the centre
of gravity, a Cartesian system with its origin there is usually the solution. These axes may be fixed in the body, coinciding
with the principal axes of inertia. Or their directions may be Or they may move in the body and in space fixed in space. some to condition subject peculiar to the problem in hand.
It remains to write down the equations each from of the above considerations. They must be arising
Final Formulation.
in
number equal
to the
Besides the
second order, these may also include differential equations of the first order, not reducible to equations between the coordinates.
differential equations of the
The
lem.
Go back
solution of these equations is a purely mathematical probfrequently over familiar problems and recall the
mathematical technique, writing the equations down on paper, In this neatly, and carrying through all details of the solution. is consciousness is it way, analytical developed; composed of
experience and common sense. Further Study. There is a vast fund of interesting problems in Rigid Dynamics, of all orders of difficulty, and two invaluable
Mecanique Routh, Rigid Dynamics, Vols. I, theory is execrable, but his lists
old
rationelle,
II.
Vols.
and
II,
and
Cambridge Tripos Papers, are capital. earth is a top, and the study of the precession and nutation of the polar axis is a good subject for the student to take up next. Webster's Dynamics is also useful in the important applicaThe text is hard reading; but the student tions it contains. who once dominates the method as set forth, for example, in the foregoing treatment, can and should construct his own solution
The
problem in hand. Finally, Klein-Sommerfeld, Theorie des Kreisels in four volumes. This is a classic treatment of the subject. The first three volof the
y
ROTATION
247
umes treat the theory of the top by modern mathematical methods. The fourth volume, devoted to the applications in engineering, can be studied directly through the theory which we have deis
veloped above, without reference to the earlier volumes. There a detailed study of the gyroscopic effect in the case of railroad wheels, the Whitehead torpedo, the ship's stabilizer, the
stability of the bicycle, the gyro-compass, the turbine of Leval,
and a
large
number
of further topics.
CHAPTER
VII
Work.
In
is
defined
as
the
Fl,
and
the understanding being that a force F, constant in magnitude fixed in a direction, acts on a particle, P, or at a point rigid or elastic body, and displaces P a distance I in the direction
of tho force.
The
force,
definition shall
still
now be extended
acting on a
body.
Let
g
Let
F=f(x)
be the
force,
where f(x)
is
a continuous function.
,
interval into
sub-interval
.
FlG 116
_,
xk
Az*
xk
_!.
work
shall
sum of
shall lie between the work corvalue of the force in that interval, and the work corresponding to the minimum force
ii)
the
work
for
any interval
responding to the
maximum
AW ^
k
248
249
Fi'
Now,
mum
and
on
its
mini-
similarly, its
maximum
value
FZ
Hence
=/(*;'),
**:
lies
But each
longest
of these
Axjb
sums approaches a limit as n increases, the the definite integral approaching 0, and this limit
is:
i>
Hm
Hence the requirements,
sufficient to
i.e.
f J
i)
physical postulates
and
ii)
:
are
(2)
W=Jf(x)dx. a
;
The foregoing definition applies to a negative force, and also to the case that 6 < a the work now being considered as an algeif a force, instead of overcoming resistance, overcome i.e. yields, it does negative work. The work which corresponds to a variable displacement, x, where a ^ x ^ 6, is by definition
braic quantity.
Thus
;
is itself
(3)
W=Jf(x)dx. a
Hence
(O
f-
,.
* Strictly speaking, we have shown that (2) is a necessary condition for the definition of work according to the postulates i) and ii) . It is seen at once, however, that conversely Equation (2) affords a sufficient condition, also.
250
MECHANICS
EXERCISES
Show
that the work done in stretching an elastic string proportional to the square of the stretching.
1.
is
2.
falls
directly into
3.
x to
&
(5)
W=ff(x)dx. X
is
What
the value of
:
dW/dxl
2. Continuation Curved Paths. Suppose the particle describes a curved path C in a plane, and that the force, F, varies in magnitude and direction in any continuous manner. What will be the work done in this case? Suppose the path C is a right line and the force, though oblique to the line, is constant in magnitude and direction; Fig. 117.
Resolve
the
force
into
its
tett
<
>f
j*a
i \
r^
\ii
-~-
two components along the line and normal to it. Surely, we must lay down our definition of work so that the work done
by F
FlG 117
forces.
by the component along the line has already been defined, namely, Fl cos ^, where F = F is the intensity of the force. It is an essential part of the idea of work that the force over|
|
comes resistance through distance (or is overcome through disNow, the normal component does neither; it merely tance). sidles off and sidesteps the whole question. It is natural, thereno to define it as work. we Thus arrive at our final fore, doing definition: The work done by F in the particular case in hand shall be
(6)
W
and
Fl cos ^.
second form of the expression on the right is as follows. Let Y be the components of F along the axes. Let T be the that the path AB makes with the positive axis of x. Then angle
251
of the projections
F on A B
or
is
equal to the
sum
X and Y on AB,
On
F cos ^ =
the other hand,
X cos T + Y sin
T,
2/ 2
T.
x2
x
TF
=
l
cos
2/i
sin
r.
Hence
(7)
= X(x z
arO H
it
If C be any regular curve, divide Let the sn -i, s n = J. by points s = 0, s t the value of F at an arbitrary point of the /b-th arc, and let ^i be the angle
General Case.
into
arcs
F be
(s t -i, s k )
to the vector
F.
Then
the
sum
Jb-l
FIG. 118
where
Zfc
denotes
the
length
of
the
chord, gives us approximately what we should wish to understand by the work, in view of our physical feeling for this quantity. The limit of this sum, when the longest h approaches 0, shall be
W
~^As*
is
Since
1, '
it is
above
n
limit
the same as *
/
lim
V Fk cos fa
/*
Asfc
IF cos
\l/
ds.
We
of
(9)
a curved path
W
o
iF cos ^ ds.
is
* Cf. the author's Advanced Calculus, p. 217. It learn thoroughly what is meant by a line integral.
252
MECHANICS
second formula for the work
<
A
(10)
is
W= Ax COST +
J
7 SUITES =
= Cxdx
^ as JC(x^
Yt
or
(11)
Ydy.
immediate.
The
is
The
defini-
tion (9) applies at once without even a formal change. (11) is replaced by the following
:
Formula
(12)
= Cxdx
6'. c')
+ Ydy +
Zdz
or
(',
Cxdx+Ydy + Zdz.
(a.b.c)
Example. To find the work done by gravity on a particle mass m which moves from an initial point (X Q y Q Z Q ) to a final point (x lt y lt zj along an arbitrary twisted curve, C.
of
,
Then
Y=
0,
Z = mg
W=
mg(z l
).
is
the initial
(taken algebraically), and depends only on and final points, but not on the path joining them.
EXERCISES
1. A well is pumped out by a force water at the mouth of a pipe which
pump which
is
delivers the
fixed.
Show
that the
equal to the weight of the water initially in the well, multiplied by the vertical distance of the centre of gravity below the mouth of the pipe.
is 2.
work done
The components
2x
of the force
5,
which acts
x
or?
a particle are x
3y + 4z
Y=
+ 8,
Z =
+y+ +
z
l2.
253
Find the work done when the particle describes the arc of the
x
for
cos
2*.
is
0,
sin
0,
70,
which
If
3.
the
g curve C
represented parametrically
*(X),
C:
x= /(X),
y
is
= f (X),
integral
:
Xi,
given by the
3.
Field of Force.
Force Function.
Potential.
particle
neighborhood of the solar system is attracted by all the other particles of the system with a force F that varies in magnitude and direction from point to point. Thus F is a vector pointin the
Its function throughout the region of space just mentioned. components along Cartesian axes, namely, X, Y y Z, are ordinary
z,
of the particle.
In
(1)
F = Xi
+Y +
j
Zk.
The example
of force.
We may
serves to illustrate the general idea of a field have an electro-magnetic field, as when a
If
the north pole, P, of a magnet is brought into the neighborhood of the wire, it will be acted on to the wire by a force F at right angles to any line drawn from
straight wire carries a current.
and
from
the wire, the sense of the force depending on the sense of the
current.
If
Z-0,
where (r, 0, z) are the cylindrical coordinates of P, Thus in vector form positive or negative constant.
and
is
and
254
Force Function.
(4)
MECHANICS
It
may happen
u =
<p(x,
that there
y,z)
is
a function
such that
,
(5)
Y _ du X ~te>
w, is
v_
du
__
du
~d~z
Such a function,
In vector form
Let
V be a
(7)
V =
is
+'
Then Vu
defined as
Hence
(9)
F =
Vu.
field generated by a a force function
:
Gravitational Field.
(10)
=
is
where r
particle,
and X
the distance from the given fixed particle to the variable is a positive constant.
(n)
2T
k
>
Ekctro-Magnetic Field.
described,
(12)
field
above
CO.
u = C tan~ l -; X
9
is treacherous, since only certain values of the function are admissible. However, since the
by
additive con-
255
we can
shall
and we
' (14)
have
= |H = dx
When
C-=gx + v
2
2'
Y=
2/
f* ?/
C^jhi, x + y
2
Z=
0.
a particle describes an arbitrary path in a field particle by the field is given by of 2. If is a force function, this formula there Equation (12)
Work.
of force, the
becomes
&u
du
i.e.
u
is
region in
which
lies
W W
is
:
= u
const.
Thus
at
its final
independent of the path by which the particle arrived destination, and depends only on the starting point
W
if
u(x, y, z)
u(a,
6, c).
It is true conversely the field represented by the vector (1) is conservative, then there always is a force function, u. For then the integral
is
0.
called conservative.
that
(*.*.
(18)
u= IXdx +
(a.b.c)
Ydy
+ Zdz
is
**,
dx
*y,
dy
a
*
dz
Potential Energy.
When
field of force
is
a constant,
energy
(20)
=- u
<p
+ C.
exists,
256
MECHANICS
EXERCISES
Show
1.
by the vector
(3) is
be any region of space such that an in can be drawn together continuously closed curve R arbitrary to a point not on the axis, without ever meeting the axis, though passing out of R, then the field of force defined in R by (3) is
not conservative.
if
But
conservative.
A meteor, which may be regarded as a particle, is attracted the sun (considered at rest) and by all the rest of the matter by in the solar system. It moves from a point A to a point B.
2.
Show
it
by the sun
is
W = Km(K
respectively,
where r and r x represent the distances of A and B, is the gravitational constant. from the sun, and
4.
Conservation of Energy.
force whatever.
any
The motion
Second
Law
'
(i)
U/l/
U>C/
U/l/
and add
}
\dt~dt 2
~dt~dt
dzd^z\
dt ~dfi)
_ ydx ~
dt
ydy
dt
:
-dz
dt
The
~2dt
v *'
where
v2
W+
Y dx
dt
~dP
+
~dt
'
Hence
md
2
dt
y,
dy
dt
^ dz
dt
Each
t,
we
integrate
257
the results must
limits,
<
and
t lt
J
o
f*. Vdt dt =
2dt
vt
J
'o
The
The right-hand
side
is
nothing more or
less
than
Cxdx + Ydy +
;
Zdz,
taken over the path of the particle But this is pre2, (13). force Hence the work done on the acts. the that cisely particle by
The quantity
mv
defined as the kinetic energy of the particle. Equation (3) the following theorem.
is
We
have, then,
in
THEOREM.
to the
particle is equal
If
we have a system
of particles,
true.
:
d xk v mt-^r-Xt,
infer that
*-- =
we
2( X
The
kinetic energy of the system
is
^ (v
xk dx
+^
.
dy k
7 dz k \
defined as
258
MECHANICS
integrating, then,
On
89
<
and
< 1;
we have
r.-'.t
of the works done on work done on the system. the individual particles, or the total The result is the Law of Work and Energy in its most general
The right-hand
sum
particles.
THEOREM.
particles is
The change in the kinetic energy of any system equal to the total work done on the system.
the
forces
of
Conservative
i.e. if
are
conservative;
such that
C/j
Ug
and so
1\
potential energy,
-T
is
U,
- U
:
The
(8)
<l>,
defined as
$ =- U
(7)
const.
Hence
(9)
can be written
7
1 !
+ * = T + *ot
(10)
E = T + *.
have, then
:
We
(11)
E,
= Em
or the total energy remains constant. This is the Law of the Conservation of Energy in its most general form for a system of
particles.
6.
Vanishing of the Internal Work for a Rigid System. Conwhich form a rigid system. Let them be
-
held together by massless rods connecting them in pairs. Thus the internal forces with which any two particles, ra t and m/, react on each other are equal arid opposite
:
(1)
ty
Fn =
0,
259
lie along the line joining the particles, and furthermore the distance between the particles is constant i.e.
(2)
is
rl
(Xi
xtf
( yi
2/y)
(Zi
ztf
rods with
In general, however, if each particle is connected by these all the others, there will be redundant members, so
that the stresses in the individual rods will be indeterminate. that case, let the superfluous rods be suppressed. Consider the work done on the particle mi
In
necting
it
with mj.
It is
/
C( Y
By
the
{
Y^ dyi
of the
Z/ dzi
t
parameter
in the
form
^Mi -u
^A
m
,//
rn,j
by
is
Since
Xu + Xn =
the
0,
Ya
Y,-t
0,
Za
Za =
:
0,
sum
i
of these
in the
form
i
'
ij
~~
dy\ ~
(dz {
"
dz ~
/(
This
i;
is
collinear with
identically
by
(3).
260
MECHANICS
have thus obtained the
result that the
We
It follows, internal forces of a rigid system of particles is nil. in that the kinetic of the such a then, change energy system is the work the or to done forces. Lookby external, equal applied,
we can now
THEOREM.
whatever is equal
The change in the kinetic energy of any rigid system to the work done by the applied forces.
Before
to rigid bodies,
we can extend
it
we must
and work.
Consider a rigid body. 6. Kinetic Energy of a Rigid Body. Let the volume density, p, be a continuous function. Denote by v the velocity of a variable point P of the body. Then the
kinetiQ energy is defined as
(i)
T =
extended throughout the region T of space, occupied by the body. The vector velocity v of P is the vector sum i) of the velocity
ii)
to that axis.
(2)
Hence
v*
V*
+ r*a>*
P
from the
axis,
where
find:
and
co
is
the
we
Hence
(3)
i-Tr+TT
;
Let v denote the velocity of the centre of gravity, G h be the distance of G from the axis of rotation. Then
0t
and
lot
= F2 =
7
+A
co
2
.
Moreover,
I
+ Mh\
about a parallel axis through
where 7
G.
(4)
is
the
moment
of inertia
Hence
261
and
(4)
is
THEOREM.
The
kinetic energy of
sum
and
of
the
kinetic energy of rotation about the instantaneous axis. It can also be expressed as the sum of the kinetic energy of a particle
of like mass, moving with the velocity of the centre of gravity, and the kinetic energy of rotation about an axis through the centre of gravity,
parallel to the instantaneous axis.
Let One Point Fixed. Let a point of the body be at rest. the (, 17, f)-axes lie along the principal axes of inertia, being Then the components of the vector velocity v of any the origin.
point fixed in the
body
are
770^
fa?,/
fw
Vt
~~
Hence
(5)
If
T = %(Aw$ 2
+ BuJ +
Cco^
).
then
(6)
T = i
From
Mv*
(4)
and
(Ap*
(5)
we
infer that
T =
+ Bq* +
Cr 2 ),
about the principal
where A, B,
are the
moments
of inertia
axes of inertia through the centre of gravity, and p, q, r are the components of the vector angular velocity w along these axes.
7.
Final Definition of
Work.
We
the point of application, P, of the force is fixed in the body. describes a curve C either in the body or in space. pose
How
t
shall the
work now be defined? Take the time as a parameter. Divide the = r < t < into n parts by the points
v
interval TO
-
<
tn -\
the body,
which at time
^ t ^ r < t n = rv t = tk will
262
reach
MECHANICS
C
;
let
it
Tk be
its
space when
reaches
path in space, and let Vk be its velocity in For the interval of time cf Fig. 120.
;
.
we may take
the force as constant, F = FA the value of F at the intersection of I\ with C, and let F*
,
act on the point Qk throughout the inThen J?k will do work equal terval Atffc.
approximately to
(1)
Fk v k cos
\l/
\{/
A^,
where
FIG. 119
k is is
Pk
v k kt k ,
If
Qk
to
fc
a distance
will
work
precisely.
We
now
define the
work
as
Km
or,
cos
FIG. 120
(2)
W=
Fig. 119.
Fvcostdt;
is
cf.
(3)
/Fvttt,
?r
where v
is
at P,
and Fv
is
the scalar
have used the time as the independent variable, or the parameter, in terms of which to define the displacement. But the result is in no wise dependent on the time in which the displacement takes place. Any other parameter, X, would have done equally well, provided d\/dt is continuous and positive (or For negative) throughout.
j. v at
We
263
This formulation of the definition of work in the general case due to Professor E. C. Kemble.
Example
the plane.
1.
down a rough
^
inclined plane
without slipping.
Here,
circle;
is
each
cusp at
v
0.
W=
Example
FIG. 121
0.
2. The same, except that the ball slips. The curve C shall be taken along the plane.
The normal
;
component R =
F =
does.
FIG. 122
Let
by
0,
through which
the
ball
has
\l/
The curve T
is
a trochoid tangent to
ds
at P.
Hence
or *,
dd
and
'
cos
{(!
<*
(0i
0o)l0.
It
Observe that in the definition, Equation (2), v is positive or would not, therefore, be right in this example to write
v
_ ds ~Tt
d8
dt
EXERCISES
1.
of the ball
2.
Check the result in Example 2 by determining the motion and computing the change in kinetic energy.
train
is
car
is
264
MECHANICS
ing himself on the floor. He runs forward, then slides. If he was running at the rate of 6 m. an h. when he began to slide, and slid
ft.,
how much work did he do on the car? Compute by the definition and check your work by
3.
solving
A
**!
rope
^~^~"^
^
is
,
The
free
end
is
B
Take the
through
is
it.
frozen part as
R = 20 Ibs. to free the frozen part. How much work is done ? straight, and P in the vertical plane
F
4. Extend the definition of work to a body force, F, where a continuous vector, defined at each point of the body
:
5.
Show
is
machine
6.
that the internal work due to the rope in an At wood 's nil. Would this be the case if the rope stretched?
of rigid bodies are
number
them
Show that the sum of the works done by the; cords slipping. on the system and the system on the cords is nil. First, extend the definition of work so as to include the case of the work done on the system by the part of a cord which is in contact with a body.
8. Work Done by a Moving Stairway. Consider the work which an escalator, or moving stairway, does on a man as he walks up. The forces that act on the man are /, S, and Mg, where R, S are the components of the force which the escalator exerts on his foot, and Mg acts at his centre of The curve T is always a gravity.
and
ds
dt
FIG. 124
where
moved
since the
man came
aboard
265
ir/2.
force
it
^ =
The whole
due to
S = F cos
^,
and
is
The speed
of the escalator
is
constant
denote
it
by
c.
Thus
(2)
W
I
And
(3)
c* t ,
where
is
is
moved
while the
man
running up.
On
ity of the
the other hand, consider the motion of the centre of gravman. Let the axis of x be taken up the plane. Then
M --f = S
(4)
Mg sin a,
S dt
Mu^
dx/dt.
Mu =
(2)
l
sn
a,
where u
It follows, then,
(5)
If
from
and
(4) that
W
the
= c(Mu
MV,Q)
Mgct^
sin a.
steps off with the same velocity with which he w then, with the help of (3), stepped on, u^
man
(6)
W = Mgl sin
h
a.
Now
is
sin
in
by which the man would have been raised tho time he was on the escalator if he had not run, but stood
Hence,
finally,
still.
(7)
W = Mgh.
makes no
difference, then,
It
whether the
man
runs fast or
slowly,
is
up
or down.
on the
escalator.
The one thing that counts is how long he Thus when small boys play on the escalator,
266
MECHANICS
it,
proportion to the time they are on leave with the same velocity.
9.
i)
Other
Cases
in
Which the
Rolling
Internal
Work
Vanishes.
Two Rigid
Bodies,
without
Slipping.
action
and reaction are equal and opposite, general normal to the surfaces.
Moreover,
Vi' v2
---^[
^p^f >^
the vector velocity of the point of contact, regarded as a point fixed in the one body, must be the same as the vector velocity of the
point of contact, regarded as a point fixed in the other body.
F on
2
W
But
l
CF.V, cos ft
dt,
W
=
TT.
=
to
(*F 2 v 2 cos ft
dt.
F = F
2,
i\
v2,
ft
ft
Hence
0.
W,
ii)
Two Smooth Rigid Bodies, Rolling and Slipping. F and F 2 are equal and opposite, and normal surfaces at the point of contact. The velocities Vj and v 2 are not equal when there is
the forces
l
Here
to the
on the normal
v l cos ft
+
l
vz
cos ft
2)
0,
:
Since furthermore
F = F we W + W 2 = 0.
1
have
Fio. 126
have already mentioned the case of rigid bodies on which wind and unwind, 7, Exercise 6: and massless rods were shown in 5 to do no work. Thus syj? terns of rigid bodies connected by inextensible strings and rods, even though the point of application of the force exerted by the string or rod be variable, show no internal work.
inextensible massless strings
10.
We
for
a Rigid Body.
THEOREM.
The
change in the kinetic energy of a rigid body, acted on by any forces, is equal to the work done by these forces.
267
the theorem
first for
two
special cases.
is
No
Rotation.
m w
i.e.
^i _
2
-fl^o
'
the change in the kinetic energy of a particle of mass M, moving as the centre of gravity is moving. On the other hand, consider the work done by one of the
forces,
(2)
W=
is
Since there
no rotation, v
v,
^ =
^,
and
(3)
w=
W
is
CFvcos$dt.
Hence
by
the work done on a particle at the centre of gravity the same force, and the theorem is true by 4.
II.
tions,
Here, Eulcr's Dynamical EquaConsider a force 13, determine the motion. Chapter VI, which acts on the body at P. Let r be the vector drawn from Then the v of the definition of work, 7, (3) is to P.
CASE
= & X
-
r.
Hence
(4)
Fv
cos
^ = F v = F
(w
of
r).
is
On
From
moment
F about
Euler's Equations,
A
I.e.,
.
we have
~ dr ~
/it
dp
^TT/
""
n dq
'5/
"
,,
,r
'
'
'
Hence
(5)
$(Ap*
Bq*
+
Now
F),
Lp
+ Mq +
Nr =
w =
co
(r
268
MECHANICS
Fv
it is
and so
(7)
cos
^ = Lp
+ Mq + Nr.
(b
For
true of
c)
+c
=
a)
0,
since
a1 a
a (b
-
c)
Moreover,
w
Hence
=-
(r
o>).
F-(w Xr) =
-(r
F).
From (5) it follows, then, that for a single force, the change in kinetic energy is equal to the work done. For the case of
n
The extension to the case of forces the proof is now obvious. forces and forces out body spread continuously over surfaces or
Remark. We have shown incidentally that the work done on a rigid body with one point fixed is
h
J
where
o>
t,
M
is
= La
+ Mp + Nj
first
is
Consider
a single
force, F.
The work
does
W
Here,
= /Fvd*. =
V
+ V',
where v
is
the velocity of the centre of gravity and v' is the veQ relative to the centre of gravity, as it flashes
through P.
Hence
W=
269
Mv*
The second integral is equal to the right-hand side of Equation (5). Thus the theorem is proved for one force. For a number of forces the proof is now obvious.
EXERCISES
1.
ball is placed
on a rough
and
slightly displaced
fixed sphere of the same size near the highest point. Find where it Let p, have any value.
A
is
rod
smooth weightless tube can turn freely about one end. inserted in the tube and the system is released from rest
How
when
it
vertical ?
3.
mounted
cylinder.
It is cylindrical can is filled with water and sealed up. so that it can rotate freely about an element of the
Show
is
that
it
oscillates like
smooth.
to
its
In the preceding problem, the height of the can is equal diameter, and the can weighs 5 Ibs. The water weighs
31
Ibs.
5.
tube,
smooth
is
inside,
filled
with water.
The tube
placed,
rest.
Show
is
that
it
oscillates
latter.
it
like
slide freely
on a smooth
table.
The
vertical
mounted arm
so that
is
can
filled
with
water, and the system is released from rest. How fast will it be moving where the vertical arm has just been emptied? Assume the tube smooth inside; and also take the weight of
its
mount equal
from
The can
CHAPTER
VIII
IMPACT
1.
Impact of Particles.
Let two
particles, of
masses
and
2,
,
w2
be moving in the same straight line with velocities Ui and and let them impinge on each other. To find their velocities
two
particles.
act,
w2
xgfr
external
forces
^f
FIG. 128
u2
momentum
Hence
remains
(1)
m^
yet, nothing has
m^!
+m
u2
w2
0.
As
ticles.
The extreme
balls)
cases are:
two
bil-
liard
and perfect
inelasticity
two
balls
of
putty).
will
for now we In each case there is deformation of the bodies no longer think of particles, but, say, of spheres, and the
,
u lt w 2 etc. refer to their centres of gravity. During the deformation the mutual pressures mount high, and even if other (ordinary) forces act, their effect is negligible,
velocities
compared with the pressures in question. In the case of perfect inelasticity, there is no tendency toward a restitution of shape, and so, when the maximum deformation has been reached, the mutual pressures drop to nothing at all. At this point, the velocities of the two centres of gravity are the same,
u(
= u
2,
and hence
this
common
:
velocity,
which we
"*"
will
denote by
f7, is
u =
is
miUl
m m +m
l
* 2
u*
solved for perfect inelasticity. For partial The helpful to picture the impact as follows.
270
IMPACT
motion of the centre of gravity
equation
:
271
by the
i.e. up to the time of greatest on deformation, T, have, integrating each side of each and T the limits between equation
For the
first
we
(4)
mu
l
<-T
l
T
/
t=T
/
/
I
J Rdt.
i-o
J
an impulse,* and
T
is
e/
The
integral is called
denoted by
(5)
P = CR dt.
o
Hence
(6)
m U
2
= =
P
P.
The second stage of the impact now begins, as the balls are kicked apart by their mutual pressures. On integrating the V equations (3) between the limits T and 7 we have
:
T'
(7)
m^
CR'dt'.
fftdt',
Now it is easily intelligible physically if we assume that, in the case of partial elasticity, the value of R' stands in a constant ratio to the value of R at corresponding instants of time, or that
(8)
R'
=
=
eR,
when
T-
t'
T.
e
is
t
t'
between
and
FIG. 129
<
P
<
1,
* Sometimes spoken of as an impulsive force but this nomenclature is unforis not of the nature of a force, which is a push or a pull, but rather tunate, since is expressed by a change of momentum. Moreover, the dimensions of impact are
;
ML/T,
not
ML/T*.
272
being
ticity.
MECHANICS
in the case of perfect inelasticity
and
Hence
7"
T
f
(10)
P'
/V dt
CR dt =
:
eP.
Equations
(7)
(U)
{
^-Ilf/
U
df
u2
The
problem.
four equations, (6) and (11), contain the solution of the Between them, and P can be eliminated, and the
u[,
.
The
result is
Uf\ *
(12)'
U2
The Case
without
m 4+ m, e) m u + (ra
:
vn..
m =
2
oo.
If in
to increase
limit,
we obtain
the equations
U2 = U 2
These equations do not prove that, when the mass m 2 is held fast, or is moving with unchanging velocity u 2 the velocity of the mass m l after the impact will be given by the first equation (13), but they suggest it. The proof is given by means of the first of the equations (6) and (11), resulting as they do respectively from the first of the equations (4) and (7), combined with (10); U having here the known value u 2 = 0, we have If, in particular, u 2
,
.
(14)
u{
-,.
*= 1. Equation (14) becomes in this case Perfect Elasticity, e = and the recedes with the same velocity as that ball -MI, u[
with which
If the
it
impinged.
m =w
l
2,
IMPACT
and the
balls
273
balls
interchange
illustrated
their
velocities.
nomenon can be
of
suggestively
side
pendulums.
If
at rest, and the other is released from an angle with the vertical, the second
ball will
pact, and the first will rise to the same height on its side of the vertical as that F from which the second ball was released. Thus the velocities will be successively interchanged at the lowest
In this hypothesis we have Critique of the Hypothesis (8). in the phenomenon before of for an amount detail taken granted
us far in excess of what the physicist will admit as reasonable in viewing the actual situation, and he may easily be repelled by so dogmatic an assumption in a case that cannot bo submitted to
direct physical experiment and which, after all, is far less simple than we have led the reader to suppose, since the problem is
essentially
one in the
elasticity
of
three-dimensional distribuis
tions of matter.
The
:
objection,
however,
eP,
easily met.
We
may
P'
EXERCISES
1.
overtakes a ball of 4
h.
h.
Determine their
cient of restitution
2.
directions.
3.
elastic
perfectly elastic sphere impinges on a second perfectly sphere of twice the mass. Find the velocity of each after
the impact.
4.
is
Newton found
.
a glass marble is dropped from a height of two feet ^f on a glass slab, how high will it rise?
If
274
5.
MECHANICS
In the
last question,
what
will
it
rest.
text, the impinging ball will not be quite reduced to rest, because no two material substances are quite elastic. If, for given balls, e 0.9, show that the ball which is at rest should be about 11
per cent heavier than the other one, in order to attain complete rest for the latter. What per cent larger should its radius be ?
8.
If,
,
of glass.
= ^f
9.
If
two perfectly
elastic balls
equal velocities, show that one of them will be brought to rest if it is three times as heavy as the other.
Determine the coefficient of restitution for a tennis ball it and comparing the height of riie rebound with the height from.which it was dropped.
10.
by dropping
11. Some pitchers used to deliver a slow ball to Babe Ruth, believing that he could not make a home run so easily as on a fast Discuss the mechanics of the situation. ball.
J 2.
Continuation.
Oblique Impact. Let two spheres impinge them to be perfectly smooth. To determine the velocities after the
impact. Let the line of centres be taken
as the axis of x.
The deforma-
Fia 131
tion of each sphere is slight, and the force exerted by the other
sphere, spread out as it ^s over a very small area and acting normally at each point of this area, will yield a resultant force, K, nearly parallel to the axis of x.
For the
first
sphere
we have
* The physics of the second part of this problem (and of the next) is altogether After a few rebounds we pass beyond the domain within which the phantastic. physical hypothesis of the text applies, and the further motion becomes a purely mathematical fiction. It is amusing for those who have a sense of humor in science. But for the literal-minded person, be he physicist or mathematician, it is dangerous.
IMPACT
(15)
275
where
X
e
= #
cos
6,
F =
72 sin
c,
being numerically small and x lt y l referring to the centre of gravity, and for the second sphere,
(16)
rn^X,
we obtain
:
*,
F.
On
,.,,_>.
integrating (15)
(17)
*-o
=-
/-Yd*,
m,^ dl
is
:
=
*
=o
J/Yrf*.
down
the
And now we
denote the
first
(18)
Thus the
integrals of (15)
and
m^U
(19)
1
=
l
m F
t
m^^
= P
first epoch of the impact, the equations for the second epoch being, as in the corresponding case of 1, the
:
u\
U= m U =
m,
2
eP
{
1.
m* v\
eP
m^
:
m V= m F=
}
0.
P'
eP.
The result at which we have arrived is seen to be the following. The component of the velocity of each sphere perpendicular to the line of centres has been unchanged by the impact,
(22)
v[
v l9
v' 2
v2
276
MECHANICS
of the velocity along the line of centres are
1,
The components
tions (12)
:
Equa-
u\
= =
mi
*--
(23)
Mj
m
e
lt
Kinetic Energy.
for then
When
i.e.
is
~~2
_ "
>\ ~~
as
is
shown by
e
direct
all cases,
0, i.e. when the spheres are totally inelastic, an easy shows that the kinetic energy has been diminished. computation The intermediate case, < e < 1, is treated in the same way. It follows from direct computation that the left-hand side of Equation (24) has the value
:
When
(m^ + w
and
this
is
M2) 2
+mm m
l
(u
u 2) z e 2
2)
at once
shown
The terms
the result.
EXERCISES
1.
smooth
an equal
their
If e
ball
line
travelling north-east with one-quarter the velocity, of centres at the time of impact being east and west.
find the velocities of the balls after impact.
2.
^,
of
45.
smooth ball strikes a horizontal pavement at an angle Find the angle of rebound if the coefficient of restitution
1 is
is f.
3. Show that the kinetic energy of the balls of Question diminished in the ratio of 245/272 by the impact.
4.
IMPACT
Rigid Bodies. impulse. By that
3.
is
277
Let a rigid body be acted on by a single meant the postulates about to be laid
picture.
force
at a point
(x,
y)
for
val of time,
T,
only
FrG 132
The
dt*
= Y *'
On
we
find
M(u'
(2)
u)
= Cxdt,
T
M(v'
v)
= CY dt
U
T
x)
7(0'
0)
= Ax will
Ydt
f(y
y)Xdt.
assume that the vector changes conand direction during the interval of magnitude tinuously time in question, and that the point of application, (x, T/), also moves continuously, remaining near a fixed point (a, b) throughConcerning
in
F we
Let
=
Let
17.
Then
rj
(4)
P=
Cxdt,
Q =
CYdt.
278
MECHANICS
last
The
(5)
Equation
(2)
now becomes
/(' -
0)
=
T
(a
*)
Q T
(b
y)
+
We
esis
CtYdt -
Cr,Xdt.
that the integrals (4) approach limits when T approaches 0, the integrals in the last line of (5) converge toward 0; for then
the equations
:
we should have
(6)
M(u'-u)=P,
'
7(12'
$2)
(a
x)
Q -
(b
y) P,
(4).
P and Q
This inference can in fact be drawn, provided the angle through which the vector F ranges is less than 180. Equations (6) then hold, and, in particular, it follows, on eliminating P and
Q
or
(7)
7(Q
2
fl)
= M(a Q)
x)(v'
v)
- M(b -
y)(u'
u)
(8)
fc
(12'
(a
is
x)(v'
-v)- (b-
y)(u'
u).
rotating about one end, and it strikes an obstruction, which brings it suddenly to rest without any What point of the rod comes into reaction on the support.
An
Example.
rod
{Let
be
h,
'
the distance from the stationary end and let I be the length of the rod.
k
FIG. 133
Let
= C;
then
ft
=
/
- C.
Since
u =
0,
fi'
0,
^ =
0,
12'
0,
M/ 2
^-,
we have
Hence
The point
is
IMPACT
4.
279
proof
is
The
given by
means
Mean, which is as follows. Let f(x), <p(x) be two functions which are continuous in the closed interval a ^ x g b, and let <p (x) not change sign there. Then
of the of the
(9)
Law
//*
ft
f(x) <p(x) dx
f(x')
<p(x) dx,
<
a:'
<
6.
e/
g
Hence
(10)
Y.
C^Ydt =
'
'
f*Ydt,
t
where
is
the value of
.
at a suitable point,
',
in the interval
g
i.e.
g T? approach Now, by hypothesis, uniformly, the largest numerical value that either has in the interval
7
and
g T
approaches 0;
and furthermore,
also
the integral on the right approaches a limit, Q. gral on the left approaches 0.
If the
range of the angle of F does not exceed 90, the axes of nor Y changes sign coordinates can be so chosen that neither in the interval ^ t ^ 7 and then it can be shown as above
that both integrals in the second line of (5) approach 0. In the more general case that F is contained merely within
an angle less than TT, the axes can be chosen in more ways than one so that Y will not change sign. If (x, y) refer to one such choice and (x', y') to a second, then
x'
ax
ex
(11)
+ by + dy
b
where
a
c
= =
cos 7,
sin 7,
sin 7,
d =
cos 7.
FIG. 134
* It is in such a case as the present one that the scientific importance of the proper definition of (-artesian coordinates, laid down in Analytic Geometry, ia revealed. That definition begins with directed line segments on a line, proceeds to the theorem that the sum of the projections of two broken lines having the same
280
f
MECHANICS
X'
= aX + bY
fl'
(12 >
I
r-T +
jj)
:
and
r =
Hence
r
(14)
of
+ 6,
T
A'*" dt
ac
The integral on the left, and the last two integrals on the right, with 7 as has been shown above. We approach the limit will show that this is true also of the other integrals, and hence To do this, write down in particular of the last integral in (5).
7
,
0000
r T
Ax A + 6c Ax dt + bd
Cr,Y dt
+ ad ft-Y dt.
Equation (14) for two choices of axes (x y') subject to the above conditions and characterized by two values of y: y and 7 2 where y l ^ 0, y 2 ^ 0, y l ^ 7 2 and solve the resulting equaThe determinant of tions for the two integrals in question.
,
l
,
the equations,
yl
sin
y2
sin (y z
7^,
Thus the
integrals for
which we are
solving are seen to be linear functions of integrals that are to approach 0, and this completes the proof.
known
The
Restriction.
The theorem
terval
is
t
required
in the in-
T,
shows.
Let
X
FIG. 135
= F cos = p cos ^,
<?j
Y = F sin = p sin
77
<p
\l/.
Then
on an arbitrary line, is the same for 'both lines, and ends by declaring the coordinates of a point as the projections on the axes of the vector whose initial point is the origin and whose terminal point is the point in question. With that definition, Equations (12) and (13) are merely particular cases of Equations (11), since both sets of equations express the projections of a vector on the coordinate
extremities,
IMPACT
Let
281
let
_
Then
r
2wt
TT
_
T
2wt
P-fx*
Ffa*-0,
If,
now, we set
P
__
/TT
>
__
*_
rpy
the integrals (4), being always 0, each approach limits, and so and Q of Formulas (6) have each the value 0. But the the
But may
it
not
still
be
sufficient, in
/(*rto
demand
that
Y^
0?
That
shown by
Let
0.
X
The
= F cos ?,
Y =
it
etc.
EXERCISES
1.
A A
is
sliding over
rotate.
3. If, in the preceding question, the pavement is icy, and if the box, before it reaches the curb, comes to a bare spot, M = 1> what is the condition that it should not tip?
282
4.
MECHANICS
If the
line.
box
tips, find
whether
it will
slide,
or rotate about
a fixed
6.
that greater braking power is available when brakes are applied to the wheels of the forward truck of a
Show
the
rail-
road
car.
6. If all four wheels of an automobile are locked, compare the pressure of the forward wheels on the ground with that of the rear wheels.
A lamina is rotating in its own plane about a point 0, when suddenly brought to rest by an obstruction at a point P situated in the line OG produced. Show that OP is equal to the of the length equivalent simple pendulum, when the lamina
7.
it
is
is
supported at
and allowed to
oscillate
under gravity in a
vertical plane.
5.
Tennis
Ball,
Consider a tennis
ball,
that the lowest point of the ball is moving backward. The ground thus exerts a forward force, and we will assume that this state of affairs holds throughout the impact. We shall have,
then, the following formulation of the problem
:
M(U - M = Q M(V -V = Q
)
Q)
Mfa M(v,
{/)
'==
eQ
F)
= eQ
where
^^^^
no
F=pR
Fid. 136
is
the impulse.
First of
all,
V =
0,
for the point of greatest deformation is marked Thus of gravity of the ball's ceasing to descend. equations for the seven unknowns, u , v lt co t , [/, V,
{
Q.
It is
now easy
to solve.
VQ
Observe that
co
<
0,
<
0,
>
0.
IMPACT
283
We
have, then
Q =
M(-
t>
),
vl
= e(-
V Q),
The value
of c^
^
where X
_ _ __
;
is
not interesting.
What we do want
e\
to
know
is
i.e.
e(
VQ)
(1
__ ~
e)
M (-t,
(1
e)/i\
= ( VQ)/U Q is the numerical value of the slope before the impact. As the ball has a drop due to the cut, X will be considerably larger numerically than the slope in the part of the trajectory
.
just preceding the last ten feet or so before touching the ground. The value of It might conceivably have a value as great as
e is
about
0.8.
ju
varies considerably
Thus
&
*i
.15,
as against X = .20, or the steepness of the rebound is only threefourths the steepness of the incident path. Not only does the ball rise at a smaller angle, but the horizontal
velocity
is
increased
u,
for
1.09t*
.
= w
(1
<0/*A]
it
=
is
For
this discussion to
its
be correct
maintain
This explains spin throughout the whole impact. The racquet has a high upward velocity the nature of the stroke. while the ball is on the guts.
The ball loses spin during the flight before the impact, due to the air resistance causing the drop, and this loss may easily be comparable with the loss during the impact. It would be interesting to take motion pictures of the ball, showing the trajectory just before and just after the impact.
EXERCISES
1. A billiard ball, rotating about a horizontal axis, falls on a partially elastic table. Find the direction of the rebound if
ju
= % and
.9.
284
2.
MECHANICS
A
rod,
moving
in
elastic
smooth
3.
table.
The preceding
rough, M
4.
iis
rough.
5.
rigid
lamina
it
is
oscillating in
point
is
when
strikes
an obstacle at
that
equal to the length of the equivalent simple pendulum. Show it will be brought to rest without any reaction on the axis.
this reason
For
P is
6. rigid lamina, at rest, is struck a blow at a point 0. the point about which it will begin to rotate.
A
A
Find
7%
About what
line will it
begin to rotate ?
CHAPTER IX
RELATIVE MOTION AND MOVING AXES
1. Relative Velocity. It is sometimes convenient to refer the motion of a system to moving axes. Let O be a point fixed in space. Let 0' be a point moving in any manner, like the centre of gravity of a material body, or the centre of geometric symmetry of a body whose centre of gravity is not at 0'; it is a point whose motion is known, or on which we wish particularly to focus our attention. Finally, let P be any point of the system whose motion we are studying. Then
(1)
+ ^
r',
dr
dt
_ " dr
dt
di^ dt'
v'.
-
or
(2)
v
,
The
choice of notation
important
FlG 137 here particularly because boldface letters denote as usual vectors
i ,
is
we have
i)
the breaking
v';
up
the
of the velocity v into the two velocities breaking up of v' into the two velocities
:
and
and
if)
(3)
v'
,
= vr
+ v.,
vitesse
where v r the
relative velocity,
and v e the
d'entrainement are
first recall
we must
In Chap-
ter V,
we have
17,
moving axes (,
(4)
Here,
= a
+ iff + f y
and
+ r/4 +
285
f 7-
286
MECHANICS
=
namely,
of the point
(7)
P into two
velocities,
vr
+v
e,
where
(%}
4-
#4-^
the relative velocity of P with respect to the moving axes ; i.e. the absolute velocity which P would have if the ( r;, f )-axes were at rest and the point P moved relatively to them just as it does
is
,
(9)
f(t),
??
<p(t),
\l/(i).
Secondly,
(10)
is
Ve
vitesse
rift
f7
the
d'entrainement,
the
Schleppgeschwitidigkeit,
the
velocity with which that point Q fixed in the moving space and Let flashing through P at the one instant, t, is moving in space.
(w)
= pa
qp
ry,
where
Then
FIG. 138
(12)
ve
= () X
r,
or
(13)
(&
:
f p)
j8
(77??
y.
The
The components
of v, or dr/dt,
(14)
287
repeat
have their
3.
These are the formulas when the moving axes r = 0, v = 0, v = v'. origin, 0', fixed
: :
Acceleration.
(1)
of
and
Equations
and
we
Hence
(16)
=
a
+
=
a
or
(17)
+
,
a'.
The
It is
The
requires no further comment. merely the acceleration in fixed space of the known point 0'. second term, a', relates to the rotation and admits of a numfirst
term on the
right, a
is
as follows.
Let
a'
be
denoted by
<
a.
Then
-T
;
r of
identify the variable vector v' with the variable vector 2, Formula (4) for, of course, r was any vector, moving we law wish. to according Now, we have evaluated the any
We may
right-hand side of (18) by means of Equations (14). Hence the components of the right-hand side of (18) are obtained by substituting in the right-hand side of (14) for
v
>
rj f
f respectively
v^ v
On
(19)
a$
+ a^
ft ~f~ fl
<7-
Thus wo
functions
known
__
dv*
__
(20)
=
a$
dVrt ~
jT-
+ pity
qv$
288
Second Evaluation.
for the vector
MECHANICS
The Theorem of Coriolis. Another form a can be obtained by differentiating (5), 2
:
+f +
Thus
77/3
+ f y.
y*
dt
<*
c^
r/j3
_
l
<ft
e&y
+
This result
is
f 7.
The
first
interpretations.
For,
is
referred to the
(23)
is
ae==
^
=
77
d^
+ f W~
the acceleration with which the point Q, fixed in the moving space and coinciding at the instant t with P, is being carried
ae
&+
tip
+ fy,
as
is
Since
geometrically,
or
a =
:
r/3
qy,
= py -
ra,
y = qa
p/3,
we have
(25)
-J-}
+ rp qy.
last line
The
289
= pa
+ q@ + ry.
f
>.dp\
Hence
/oc\ (26)
- /\dQ r
w2 (
'
dr\ a
f.dr
'
"
dp
or
(27)
i?/3
+ r 7) +
r
(p$
+
2
qr>
ae
= (0 X
where
This vector
(w') is
77,
f),
with which the terminal point of (co) is moving when the initial it is the relative angular acceleration, referred to the point is at 0'
;
(>
*?>
f )-axes as fixed.
dtdt^ dtdt^
:
dt dt
(co)
vr
or:
given by (8), the righthand side of (10) comes to coincide with the right-hand side of (29). With the aid of (12), this vector can be written in the form of the right-hand side of (30), and this completes the proof.
To sum
(31)
up, then
From
a
(17),
a',
where
is
a'
a,
and a
is
given by (20).
A
e,
second evaluation of a
given
by
(21),
(32)
ar
2a
+a
where a r
the latter, in a differis given by (22) and a c by (23) ent form, by (26) or (27). Finally, a t is given by (29) or (30).
;
290
4.
MECHANICS
The Dynamical
of
(1)
it
ma =
F,
follows that
(2)
ma
a
is
+ ma =
ar
F,
origin, 0',
where
moving
and
2a,
a*.
vector a r is the relative acceleration and is given by Formula (22), 3. The vector a c is the acceleration d' entrainement and is defined by (23) it is represented by (26) or (27). Finally, a* is defined by (29) and is represented by (30).
;
The
If
Q O is'a
(11),
(3)
known
2.
function of
(2)
moving axes is regarded as known, then and (o>), i.e. p, q, r are known from t
t
Equation
r
ma = F
substituting for a*
(26),
,
ma Q
2ma
ma
e.
On
its
its
value
from
mining
77,
rf
\dt
dt
dt
(4)
/,
<p,
\l/
can be thrown into the form by using a suitable form for a, a a r a^ a as pointed out at the end of Each one of these 3. accelerations must be studied in the particular case. There is no single choice of sufficient importance to justify writing down
generally, Equation required in a given problem
, ,
More
will
do well to make
hi^
foi
own
a,,
syllabus, writing
.
down
a{
291
6?
+&
fp
tfop)
where
Vy
5.
_ ~ V
'
__
-ft-
-fa
+ "I -5rj,
The
Centrifugal Field.
in
is
through 0, the (,
f)-axes
velocity
moving space. Then the vector angular The vector ae constant, and a = 0. 3, (27)
the
,
reduces to
(1)
a.=-*r+(().r)(),
is
and
Kinematically, it is, of course, the acceleration ; centripetal geometrically, it is a vector drawn from the point toward the axis and of length o> 2 p, where p is the to the axis. distance from
easily interpreted.
ma =
where a
a
is
F,
given by
3, (32),
and thus
(2)
2q
2r
+ pfa + 9, + rf
292
MECHANICS
f,
Axis of
r
the
Axis of Rotation.
In this case,
:
0,
w,
(3)
axis of f
is
the same as
it
would
space were not rotating. The projection of the path on the (, 77)-plane is the same as the path of a particle in fixed, or
stationary, space,
when acted on
i)
by the applied
rao>
2
/>
force
ii)
by
;
a force
directed
away from
and Hi) by a force at right angles to the path, equal in magnitude to 2ma)V, and so oriented to the vector velocity v as the positive axis of is, with respect to the positive axis of 17. This third force is known as the
Fia. 139
Coriolis force.
sponding revolving tennis court, Chapter III, 23 it was enough, u for problems in statics, to take into account the centrifugal But for problems in motion, this force," or the force ii) above.
not sufficient. There is the Coriolis force Hi) at right angles to the path, like the force an electro-magnetic field exerts on a moving charge of electricity.
is
6.
Foucault Pendulum.
when the
think, then, of the earth as rotating about a fixed axis through the poles, which we will take as the axis of z, the axes of x and y
lying in the plane of the equator. Let be a point of the northern hemisphere, and let its distance from the axis be p. By the vertical through is meant
at rest, or,
more
precisely,
upward;
293
through the point of support of the pendulum; then 77 will be tangent to the parallel of latitude through the point of support, and directed west. Let X be the latitude of P; i.e. the angle that f makes with the plane of the
equator. The earth rotates about
its axis
from west to
vector
directed
east,
and so the
(co),
angular
velocity,
is
downward.
co
Thus
q
p =
cos
X,
0,
=
2rr
sin X,
co
=
24
can
=
-
60
60
.000727.
FIG. 140
We
of
now
write
down
the motion.
4
:
ma + ma =
be the point of support, and
Let
let (f,
f ) be the coordinates
of the
First,
pendulum compute F
;
Z,
:
its
length,
F =
-
G + N,
,
where
au
w.c>u
;
is
and
N = - j^a-Jtfflis
j-#7
co
p(
sin
7 cos
X).
5. For, although given by the formulas (2) of down for the particular case a = 0, they apply is arbitrary, provided the vector angular generally, where a velocity of the moving space is constant.
Finally,
is
Thus we can write down explicitly the three equations of motion. These we replace by approximate equations obtained as
294
follows.
field. field,
MECHANICS
Approximate,
first,
by the gravity
U = - mg
2 Next, suppress those terms which contain co as a factor, or are Thus of the corresponding order of small quantities.
Finally,
?+
T?
=
r?
= -I
We
introduce the further approximations which consist in suppressing the term in d/dt in the second equation, and setting = mg. The first two equations thus become
:
A)
Discussion of the Equations. Multiply the first equation A) through by di/dt, the second by dq/dt, and add. The resulting
equation,
,
dtdt*~^dtdt*
integrates into the equation of energy
:
2~
or,
295
by
and
respectively,
and
%drj
17
^- cor
=
'r
C +r
'
or
(2)
fJ
co'
+C
where
sin X.
Special Case.
Then
initially r
=
;
dB
di
It follows, then, that
=
=
"'
w't.
This means that, if the motion be referred to moving axes, so ' chosen that f' coincides with f, but makes an angle u't with
,
the
pendulum
determine r
motion.
swing in the (', f ')-plane. It is now easy to r executes simple harmonic as a function of t from (1)
will
;
Returning now to the general case, let (the point moving plane through of equilibrium of the pendulum), perpendicular to the f-axis, and rotating with constant angular velocity a/ about 0. Then
(3)
is
<f>
u't
new
plane.
Equation
(2)
now
becomes
(4)
r-l-C,
this is the equation of areas in its usual form.
(1)
and
Equation
or
(5)
^ + r^ + 2'C + rV = - |r
296
MECHANICS
2 suppressing the term r
'2
On
because of
its
smallness,
we
find
+--!-+*
But
this is precisely the equation of
energy corresponding to an
i.e. the pendulum, once released, at the centre with in the a fixed describes ellipse moving plane. The axes of this in the positive sense, i.e. the clockwise sense, as one ellipse rotate
looks
down on
the earth.
in either sense, according to the initial conditions, the degenerate case of the straight line lying between the description in positive
in the
In the Foucault experiment sense and that in negative sense. Pantheon the pendulum was slightly displaced from the
position of equilibrium and released from rest relative to the earth. It then described the ellipse in the negative sense. For
initially dr/dt
was
0,
so that
it
(obviously the major axis) and its initial motion relative to the moving plane was in the negative sense of rotation ; i.e.
an
axis
counter clockwise.
value
:
24
60
has the
=
The
North
Pole, 2w.
u't
27rsinX.
should be
0,
and at the
EXERCISE
Obtain the equations of motion A), directly from Lagrange's Equations, Chapter X.
CHAPTER X
LAGRANGE'S EQUATIONS AND VIRTUAL VELOCITIES
INTRODUCTION
In the preceding chapters, the treatment has been based on
Newton's Second
Law
of
Motion.
It
is
theorems have been established, whereby some of the forces of constraint have been eliminated, like the theorem relating to the motion of the centre of gravity, and the theorem of rotation of a rigid body. But in the last analysis, when there have been forces
of constraint
pairs, the
setting up of the problem has involved explicitly any unknown forces of constraint, as well as the known forces, and the former
have then been eliminated analytically, anew in each new problem. We turn now to methods whereby, in certain important cases, the forces of constraint can be eliminated once for all, so that they will not even enter in setting up the equations on whose
solution the problem depends.
coordinates
and
intrinsic functions.
The
minimum number
of independent variables
completely the system. They are often called generalized coordiThe intrinsic functions are qm nates, and are denoted by q lt the kinetic energy, the work function or its negative, the potential energy, and the Lagrangean function L. These we have called
,
.
because they do not depend on any special coordinate system, or on any special choice of the q's. Later, we shall consider intermediate cases in which the number of q's, though highly reintrinsic
stricted, is
forces, or constraints,
1.
its
not a minimum, and in which, moreover, the unknown have not been wholly eliminated.
A material system may be determined in one or more coordinates, q lt position by </,* and the
The Problem.
-
to denote the number of the q's. But we choose, in general, the letter that refers replace it by n in these early examples to avoid confusion with the
We
to the
mass of the
particle.
297
298
time,
t.
MECHANICS
circular wire,
smooth
For example, let a bead of mass m slide freely on a which rotates in a horizontal plane about one of its points, 0, with constant angular
.
velocity.
The
is
angle
<p
that
the
radius
drawn from
zontal line
makes with a
given explicitly,
fixed hori-
Let
the position of
set
OQ
t.
produced to
the radius,
QP
y
drawn
6
to the bead.
is
fully determined by
and
Thus
if
Then we
q,
f(q,
t),
\(/(q, t).
The problem
as a function of
of determining the
t.
motion
is
that of finding q
generally, let a smooth wire, carrying a bead, move according to any law, and let the bead be acted on by any forces.
More
To
lem
We
prob-
As the second
masses,
illustrative
l9
nj
one end, 0, of which is held fast, and let the system be slightly displaced from the position of equilibrium. To determine the
extensible flexible
vstring,
oscillation.
Finally,
we may think
of a rigid
body, acted on
it will
by any
forces.
If there are
,
no constraints,
qB9 , the position. These may be the three coordinates of one of the points of the body, as the centre of gravity, x, ?/, 2; and the three Eulerian angles,
0,
tp 9
\f/ 9
to determine
body.
We may,
one point
fixed, there are three degrees of freeand so three coordinates, q l9 </ 2 q Z9 for exdom, the are Eulerian ample, angles required. Or,
is
,
again, the
Then n =
sufficient.
would be
LAGRANGE'S EQUATIONS.
a fixed line and rotate about
coordinates.
it.
VIRTUAL VELOCITIES
Here n
line to
two examples may be varied by causing the an altogether specified manner. Then, beside = q, or q 1 and q2 the time, t, would enter explicitly. ql In all such cases, the motion is determined by Lagrange's Equations, which, when there is a force function U, take the
of the last
in
Each
move
form:
A;
where
A\
1(1T dtdb
_^-^ dq~,~dq
r
'
_ r-1, ...,m,
i
now
T denotes the kinetic energy, and q r turn dq r /dt. to the establishment of these equations, beginning with the
Let a bead Lagrange's Equations in the Simplest Case. on a smooth wire whose form as well as position varies with
:
We
simplest cases.
2.
slide
the time
(1)
x=f(q,f),
*/
*>(<?,
0,
*(</,*),
where the functions f(q, t), <p(q, t), \f/(q, t) are continuous gether with whatever derivatives we need to use, and where
fa
to-
V
do not
all
8y
'dq'
dz
dq
vanish simultaneously.
:
The motion
is
determined by
the equations
"
(2)
where X, F, Z refer to the given, or applied forces, other than the reaction, (X, Y, Z), of the wire, and
(3)
i.e.
forces
XX
(X,
/z,
/*Y
v1
0,
where
v)
the wire
300
MECHANICS
re-
and add
^+
the remaining terms, namely
y A
:
8y
Pz8z\ ~ _
dt*
y>
dq)
dx
TT~
dy y "o~
dq
*- ~^r~i
dz
dq
dq dy/dq,
vanishing because of
rection
(3), since
dx/dq,
components
The
follows.
(6)
can be transformed as
We
T =~
tYI
(x*
+ y* +
z*) 9
dx
Tt>
^Tv
~
dqdt
dx
.
.
dq
ctc
From
(1)
dx
dt
"
dV
or
/\
(7)
^ITqO
+ Jt'
z.
dx
On
t.
q, q,
hand
side of
Equation
(4) turns
<L
<W W
For,
first,
L-?L
dq
dt dq
we have
-z-r
dq/
From
(7) it
follows that
dx
dq
= ~
dx
dq
LAGRANGE'S EQUATIONS.
VIRTUAL VELOCITIES
Thus
301
dT
Next, differentiate with respect to the time
.V
d dT
__
'
dtdq
.
/d x dx \dt 2 dq
.
d y dy
dt
2
d2
dt 2 dq,
.
dq
d dx
d dy
<
On
, 1AN (10)
dT _ _ = m ^
/
.
dx
dq
_j_
dy L
dq
dz
dq
^-.
dq/
Now,
dx
'dq
_ ~
d dx
dt~dq'
/ft IT C/ JU
For,
from
(7),
f)'V UJU
~
=
s)^ (/ IF JU
q
~dif
~dq
dqdt'
d2 x dtdq'
and, of course,
d dx
Jt~dq
d 2 xdq
d^~dt
Substituting the value given by (11) and the corresponding values for dy/dq, dz/dq in the right-hand side of (10), we see that dT/dq is equal to the last half of the right-hand side of (9),
is
complete
We
dt^j
This
is
~
8q
'
The
precisely Lagrange's Equation for the present case. case that the applied forces have a force function, U,
in practice.
is
of
prime importance
Here
~>
and thus
dV
dU v Y = ''
~'
8U
Q becomes
=
dx dq
4.
4. "*"
dy
~dq
dz dq
302
Lagrange's Equation
MECHANICS
now
takes the form
:
dt dq
dq
dq
Example.
paragraph.
U=
const.
Furthermore,
x
y
= =
a cos ut
a sin ut
+ a cos + a sin
a (0
(6
+ (0 + ui)
coJ)
where q
=
;
and
x
2/
aco sin ut
aco
cos
to
+
+
a (^
+ +
co)
sin (0
co)
cos
+ co) (^ +
+
co
co<)
T =
orn
^- ((^ +
-f
co
a?)
2co(0
/^T
1
+
=
co)
cos
= ma 2 (^
co
cos
0),
ma 2 co(0
.
co)
sin
0,
d dT
or
3T
a
5ay"-8
"
ma
/d*B
"(3i
+w8i
,
g=-.. dn ,
This
last is the
-&
Thus
pendulum
of length
}**'
moving
line
OQ
as a simple
^~co 2
Z
would
oscillate
about the
vertical.
EXERCISES
1. A bead slides on a smooth circular wire which is rotating with constant angular velocity about a fixed vertical diameter.
Show
that
tl^fi
= -^ at*
co
sin 6 cos B
s n +2 a
i
0.
303
If
the bead
is
released with
no
vertical velocity
from a
it will
point on the
show that
3. A bead slides on a smooth rod which is rotating about one end in a vertical plane with uniform angular velocity. Show
that
d2r
= w2 r
+ g sin wt.
4.
tion.
6. A bead slides on a smooth rod, one end of which is fixed, and the inclination of which does not change. Determine the
motion,
if
angular velocity.
6.
tion,
7.
N,
In the problem discussed in the text, determine the reac2 = ma 2 [w 2 cos 6 of the wire. Ans. (0 w) ]. A smooth circular wire rotates with constant angular
velocity about a vertical axis which lies in the plane of the circle.
bead
8.
slides
on the wire.
that
if,
Show
in Question 1, the
is
a horizontal di-
is
~~
where
<p
f w 2 cos
~ cos coH
sin
<p
0,
is
can slide on a smooth circular wire which is expandalways remaining in a fixed plane. One point of the wire is fixed, and the centre describes a right line with constant velocity. Determine the motion of the bead. 8 Ans. 6 = a
ing,
.
A bead
+
,
10.
if
the centre
is
at rest
increases at
3.
rate.
particle, of
face,
(1)
Particle on a Fixed or Moving Surface. Let mass w, be constrained to move on a smooth surwhich can vary in size and shape,
= f(q
l9
q2 0,
,
304
MECHANICS
where the functions on the right are continuous, together with whatever derivatives we wish to use, and the Jacobians
do not vanish simultaneously. The motion is given as before by Equations (2) of 2, where now, however, the reaction, due to
the surface,
is
known
in direction completely.
Multiply
dy/d<li,
these
dz/d<li>
equations through respectively by Sx/dq l9 and add. On the right-hand side there remains
only
ponents of
components of a certain line (x, y, z), and X, Y, Z are the coma force normal to the surface. Hence
drawn from
dq,
dt* dq,
is similar to the reduction however, just as easy to carry this reduction through for a system with n degrees of freedom, and this is done in the following paragraph. Thus we see that
The
dt dq t
l '
dq l
and, similarly
dt dq%
dq%
i.e.
These are Lagrange's Equations for the case of two variables; the case of (q l9 qz ). If, in particular, a force function, [/, then exists, Lagrange's Equations take the form
:
(/
uL
uLJ
uL
uL
dt dqi
dq l
dq
dt dq2
^2
uU ^2
Example. A particle is constrained to move, without frica plane which is rotating with constant angular velocity about a horizontal axis. Determine the motion.
tion, in
305
Let the axis of rotation be taken as the axis of x, and let r denote the distance of the particle from the axis. The coordinates of the particle are
(z,
y, z),
where
y
r cos
0,
z
l
r sin
x,
0,
=
r.
co,
and we take q
y
z
q2
Then
r cos r sin
w
o>2
7*w
sin co,
ait,
rw cos
a result that
of y,
z.
may
be read
off directly,
Furthermore,
U =
Lagrange's Equations
-r-r
mgz =
dT =
dx
mgr sin
:
ut.
now become
~^~
dT = mx, dx
.
n
0, J
dU = n 0; dx
Next,
dT
dT
mco 2 r
= =
ing sin
co/,
mg
sin
<
ttr*
or
^2
o>
=-^sm^.
:
special solution of this equation is found, either by the method of the variation of constants or, more simply, by inspection, to be
r
"
sni
2co
solution
is
Since
w^,
306
MECHANICS
B
= = 0. particular case of interest is that in which A conditions of the the initial to This corresponds launching particle
from a point
the axis
is
with a velocity whose projection along the projection normal to the axis being g/2w. arbitrary,
in the axis
The path
is
then a
helix.
EXERCISES
is rotating with constant angular about a vertical axis, exterior to the cylinder, the axis velocity A particle is projected of the cylinder being always vertical. which is the inner smooth. Determine the motion. along surface,
1.
cylinder of revolution
Use Lagrange's Equations to determine the motion of a particle in a plane, referred to polar coordinates
2.
:
d*r
dd*\
md
4.
The
;
lum
i.e.
Consider the spherical penduSpherical Pendulum. a particle moving under gravity and constrained to lie
on a smooth sphere.
the longtitude,
<p,
Take
6,
and
equilibrium
Then
i
T =
O/TI
fY)f]1
^- (02 + p S n
PT
2
7
0)^
jj
00
= ma
first
6,
= ma 2
2
<f>
sin 6 cos
Cv
Cv
= mga
sin
0,
and the
of Lagrange's Equations
becomes
ma
or
ma
2
(f>
sin 6 cos 6
= mga sin 0,
a
Proceeding
now
we have
Hence
307
^sin*0
is
h,
of dimension
(1)
1 in
Combining Equations
and
(2),
we
This
is
the motion
In case the Equation of Spherical Pendulum Motion. is to be studied for small oscillations near the lowest
it is
well to replace 6
by
its
supplement,
6.
Equation
(3)
then becomes
This
last
Motion when h =
fA\ (4)
equation reduces to the Equation of Simple Pendulum 0. Any differential equation of the form
:
d2 6
r at 2
A = A
cos0
--T-T 3
sm
D sm +B
.
6,
and B > are arbitrary constants, can obviously where A > in be interpreted terms of spherical pendulum motion. A first integral of (3) can be obtained in the usual manner
:
2A cosgdfl
2g
"*"
dtdP
sin 3
dt
0,
m
.
dfi
dt'
:
we obtain
<
--;-?i,
+*
cf.
Appell, Mecanique
277.
EXERCISES
1.
oscillations near
Here,
is
an
ellipse
with
as centre.
308
2.
MECHANICS
Treat the motion of a particle constrained to
surface,
move on a
smooth
22
a2
2
2/
2
,
Show
when
is
pendulum.
18,
v
More
precisely,
we mean
0.
6. Geodesies. Let a particle be constrained to move on a smooth surface under no applied forces. The path is a geodesic.* Let the surface be given by the equations
:
f(u,
i;),
<p(u, 0),
t(u,
v),
where these functions are continuous together with their derivatives, and not all the Jacobians
first
d(u,vy
vanish.
d(u,v)'
is
d(u,v)
The element
ds*
of arc
given
by
the formula
= Edu*
2Fdudv
Gdv\
where the
The
T =
Lagrange's Equations
% (Eu* +
2Fui>
+
U =
:
since
(Eu
(6)
+ Fit) + Gv) -
+ +
2FU uv
2Fv uv
+ Gu
2
t-
=
=
0,
(Fu
2 (Ev u
+G
v 2)
0.
On the other hand, the geodesies, in their capacity of being the shortest lines on the surface, are given as extremals of the
integral
A
,
2 L = fVEu' / VEu' *0
2Fu'v'
+ Gv'
;
d\
By
a geodesic
the points
is
least, if
it
line of minimum length on a surface minimum, at connects are not too far apart cf. Advanced Calculus, p. 411.
meant a
309
411)
by
the equations
Eu'
(7)
+ Fv'
u v'
d
d\
Fu'
+ Gv'
r
u'*
VEu'^'2Fu'v
+
M
Gv' 2
2V'Eu' 2
/u
/(A),
its first
t
provided that/(X) is continuous, together with and/'(X) 9* 0. In particular, then, the choice
one.
derivative,
/z
is
a possible
h,
But
T =
v'
or
(8)
i
(#w'
+ 2*W + Gv'*) =
h,
u'
u,
*,
it
Since the velocity along the path is constant, the only force being normal to the path, t is proportional to 5. In fact, (8) says that
mds
Thus the transformation
in
of the
parameter from X to
amounts
(6)
:
substance to a transformation to s; i.e. Equations virtually the intrinsic differential equations of the geodesic
are
~ (Eu' + Fv') as
2 (E. w'
+ 2FU u'v' + G + 2F
.
u v"*)
(9),
f
J- (Fu as
+ Gv') .
2 (E, w'
u'v'
+ G, v' =
2 )
U*
du = -J-, n
rl
'
V'
dv
-7ft n
EXERCISES
Obtain the geodesies on a cylinder of revolution. Observe when the cylinder is rolled out on a plane, the geodesies must go over into straight lines. 2. The same problem for a cone of revolution.
1.
that,
310
3.
MECHANICS
Show
that the geodesies on an anchor ring, or torus, are the differential equations
:
given by
d
(10)
2
.
h sin a(b
6
3
"
acosfl)
>
ring,
and A
6
is
constant of integration.
4.
Show
that
if,
ir/2,
or,
= A
then
a 2 (6
a cos0) 2
2
a2
(ID
dt
6.
(b
a cos
0)
Lemma.
dt dq
We have seen in
dq
#,
(Py dy
dt
2
d^z
dt 2
dq
It is
important to recognize this equation as a purely analytic any physical meaning to be attached to
T = ~(x 2
and
*
+y +
2
z 2)
f(q, 0,
v(<7,
0,
^(<7,
0,
where these functions arc any functions subject merely to the ordinary requirements of continuity, then Equation (1) is true.
We
turn
now
n
n.
particles,
with the
coordinates
(a?, yt,Zi) 9
1, 2,
m
t
?!,,
*i
/ifoi,
<Pi(q ly
'
'
',
?m, ?m,
A)
yi= = 2i
/,-,
-,
ti(qi,
t,
gm,
(p,
are continuous together with their two orders, and where the rank of
311
dql
The
kinetic energy,
q m q ly
,
qm
t,
for
with similar formulas for y iy z^ Our aim is to establish the general fundamental formula corresponding to
(1)
:
"
-^~'
i
\
r
1,
The independent
' "
variables
>
in
the
partial
Zi
differentiations
are
(<7i>
(7m, (7i>
:
qm, 0>
and x^ y^
are given
by
(3).
We
have
From
(3)
_
dq r
etc.
'dqr
Hence
(A\
*^
<L?L
r
V mi (tfxjdxj V V"^
2
d 2 Zj
dz
0?r
^r
.
^
2
'
07
d dxi
d^dy/i
d[
dzj\
312
MECHANICS
the other hand,
On
(5)
=
*
dq r
dqr
dqr<
Now,
^
dt
dq r dt'
while
d
dtdq r
Similar relations hold for diji/dq r
these values in (5), it
,
and dZi/dq r
On
substituting
is
sum
is
established.
Lagrange's Equations in the General Case. Let a system y^ Zi) be acted on by any
forces whatever,
X, F
t,
Zi.
of
Motion
v = Xi
(i)
6,
by
ifei,
<7
A)
(<7i,
'
?w
'
'
i
i(^li
(7m,
/-, ^,-,
first
two
dq l
313
being of rank m. Multiply the first of Equations (1) by 8xi/dq r the second by dy t /dq r , the third by dzi/dq r , and add
:
/0 , (2)
^
r
dx <
4.
v ^<
^
,
j.
^'
:
^A
1,
ra.
The
left-hand side has the value expressed by the Fundamental Equation I. of 6. Let the right-hand side be denoted by Q r
(3)
"
appears that
It thus
d dT
We
Law
For
if
of Motion.
have deduced Lagrange's Equations from Newton's Second They include Newton's Law as a particular case.
we
set
t
then T becomes
and Q3 <, Q3 <+i, Qa<+2 are now the components of the Thus Newton's Equations result at once. acts on m
-. t
force
which
8. Discussion of the Equations. Holonomic and NonHolonomic Systems. We have before us the most general case. No restrictions have been made on the forces. These may, then,
comprise dissipative forces, like those of friction or air resistance. On the other hand, there may be one or more equations of the form:
(4)
F does not
depend on the
initial conditions.
314
MECHANICS
Moreover it may happen, whether there are relations of the form (4) or not, that the <?/s and their time derivatives are bound together by one or more equations
:
(5)
$(?!,
'
,?m,
?m,
0.
An
<f>
would be an example.
in practice,
is
The
is
case which
is
:
most important
that in which
linear in the q r
(6)
AM +
may happen
--
+A m qm + A
(q lt
,
=
qm
,
0,
C),
independent of
is
equivalent to
(4).
Thus
ds
if
dO
ad
c,
which is essentially of the form (4). If no relations (5) or (6) are present or if such relations (5) or (6) as may have entered in the formulation of the problem are all capable of being replaced by equations of the form (4), the
;
Examples of non-holonomic systems 24 infra, and the Billiard Ball on the rough table, rolling and pivoting without slipping, p. 240 also the coin on the rough table, and the bicycle.* But when the Billiard
system
is
;
said to be holonomic.
Ball slips, p. 237, the system is holonomic, for the unknown reaction of the table can be computed explicitly, as the reader can
easily verify, in
and thus
its
terms of the velocity of the point of contact, components are expressed in terms of the time deriv-
atives of the generalized coordinates. are still leaving in abeyance the question of whether Lagrange's Equations admit a unique solution. Our conditions are
We
necessary for a solution of the mechanical problem, but not always The study of sufficient conditions will be taken up in sufficient.
17
and
in
Appendix D.
rationelle, Vol. II,
* Appell, Micaniqae
315
The Forces. The question of holonomic or 9. Continuation. non-holonomic has to do with the left-hand side of Lagrange's Equations, i.e. with conditions on the q r q r t which do not involve
,
,
the right,
It
and
it
is
to
these
that
we
may
x,
= xi
+ xi
Y<
Yi
17,
z =
t
z(
z;
a manner that the X{, Y^ Z\ will be essentially simpler than the X, Yi, Z iy and that the X*, F*, Z* disappear altogether from Lagrange's Equations. For example, the X{, F{, Z\ may be
in such
v _ ~ Xi
,
till
~
_ v Yi "
,
d(J
r/r
'
_ dU "
where
is
known
explicitly in
terms of
X*, 2/,
,-,
t.
Z* the problems discussed in 1-6 have afforded ample illustration. These were the so-called forces of constraint, and they did not appear in
of the A7, Y*,
Lagrange's Equations.
it
may
fulfil
the condition
1,
m.
When
this
is
true, the
Q r on
:
This case
rid of
is
it
enables us to get
some of the unknown problem arising from constraints cf 1 5. But even when all of the latter forces cannot be eliminated in this way, their number can be reduced to a minimum and then the method of multipliers set forth in the
or
forces of the
;
316
10.
MECHANICS
Conclusion.
ddT_dT^ = Q
It
r 's
can be
r
split in
two
,
(14)
in such a
Q =
r
Qr
&*,
1,
m,
known
example
property that
(15)
Q*TT, H
irlf
,
+
a ms Trm
Qiirm
0,
:
where
(16)
wm are any
disTTi
+
an
1,
-,/*< m.
(17)
be
u.
From Equations
(13)
it
follows that
d dT
dT
no matter what numbers the TT, may be. If, in particular, the Q r and the jr, are subject to the condition expressed in (14), (15), and (16), then
d dT
(18)
Multiply the
bers, Xi,
,
/z
num-
X^,
(18)
Of the
set of
(16).
numbers v lt
p,
m
For
arbitrarily,
definiteness,
suppose
317
(20)
Then
ir^+i,
,irm
Now
(21)
let
the
Xj,
TT>
are determined.
This
is
/z
linear equa-
tions in
XM
is
not zero.
of X
Equa-
ju
equations
(22)
The determination of the X's by (21) is independent of any The numbers TJ>+I, TTOT are wholly arbiHence each coefficient in (22) must vanish. We have trary.
choice of the TT/S.
,
THEOREM.
(23)
w/ie
When Q r
1,
(24)
(25)
the
a ls *-!
+ ama
7r
0,
1,
rank of
the
matrix
am \
(26)
'
'
'
Ctl/x
iwgf
/z,
find p numbers \ 19
X^ swcA
dT
(27)
the
318
MECHANICS
Applications of this theorem occur in practice in a variety of problems in which the q r q r and t are connected by relations of 7 the form (6),
,
(28)
ai,^
a,
+ am
qm
a,
0,
1,
,/*,
depend on the
conditions.
As a matter
cases the coefficients a ra in (28) do lead to a system of equations wm for which an (16) which control a set of numbers TT^
,
analysis (14) of Q r with the resulting relation (15) is possible. In other problems, however, the a rs of equations (16) have nothing to do with any such equations as (28), if indeed the latter
exist,
but may even themselves depend on q r as well as q r and t. For a complete discussion cf Appendix D. We turn now to a direct determination of the Q r from purely
,
.
mechanical considerations.
Let a system of be given (i (x %1 ?/ particles 1, be be 3n Let numbers, and let any ft). dxi, dyt, dzi Then tha carried to the point (xi 8x fyi, Zi yi dZi). a virtual is said to system displacement (5xi, tyi, dZifl experience " the word virtual" expressing the fact that the actual system may not be capable of such a displacement, even approximately. Thus a particle constrained to move on a curve or a surface would
11.
Virtual Velocities
ra
with
the
coordinates
z t)
l,
in general be
taken
and not
lie
gent
line or plane.
If forces (Xi,
Y^ Z$
(1)
W* =
J (X,
t
dxi
Yi d yi
Zi
dzi)
is
defined as the virtual work due to the virtual displacement. It is convenient in many applications to restrict the virtual
displacements admitted to consideration by linear homogeneous equations between the 5x 6?/ 5zi. Consider, in particular, a system of particles whose coordinates are given by Equations
t -, t,
A),
7:
^=
fi(qi,
qm
t)
A)
*,
319
dx
t
dqm
a)
^
dq l
is
dZ n
dqm
m.
Let
(2)
where dq
toi,
8y i}
tions.
If < 3n, the arbitrary quantities. 3q m are or linear one more are dzi subject to homogeneous equaThus only a limited number of them can now be chosen
} ,
then determined.
displacement
it
Consider
the
actual
(Ax
t -,
A?/*,
system
It is
:
experiences in time At as
AXt
(3)
A?/t
fi (q l
<pi
(q l
+ Ag,, +
At/!,
(q ly
qm gm
t)
pi (q lf
^)
Since the 5# r are arbitrary, it is possible to choose them equal to It does not follow, however, that the corthe Aq r or dq r = Aq r
,
.
responding tot, diji, dzi will differ from Ax i} Ay iy Az by infinitesThis will, in fact, be imals of higher order with respect to A. the case if the functions / <p \f/i do not contain the time, i.e.
t -,
t ,
if
dfi/dt
0, etc.
But otherwise
:
in general not.
The
(4)
virtual
where
all
of the
numbers
up
Ft
If,
in particular,
Xi = XI
+ XI
Y(
F?,
Zt
Z\
Zt,
320
so that (11) holds:
MECHANICS
(6)
jjj(ir*
1,
,
+17* + *
:
m, then
(4)
(7)
r,-
In either case,
(8)
W
t,
Q.dq,
+ ...+Qm 8qm
in
Consider the actual displacement (Ax, Ayi, AZi) of the system time At as it describes its natural path. If /,-, ^? t \f/i do not
-,
contain
will differ
In Equation (8), 11 the dq r are Computation of Q r numbers. We set arbitrary may, then, dqk 0, k ^ r; the value and of We shall compute corresponding 5q r 7* 0, 6 have then
on a moving
tions
(10)
:
Consider, for example, a particle that is constrained to lie surface. Its coordinates are subject to the condi-
f(q ly q2 0,
,
<p(qi,
0,
iKfc, q*
t).
virtual displacement means that we fix our attention on an arbitrary instant of time, t, and consider the surface represented
by
this
t.
(x, y, z)
of
displacement (&r, dy, dz) point means an arbitrary displacement in the tangent plane to In particular, if we set the surface at the point in question. = then the virtual and take ^ dq 0, displacement takes dq2 in that curve surface the to the whose coortangent place along
1
surface.
Then a
virtual
of
this
dinates are represented by (10) when </ 2 and t are held fast. Now, the natural path of the particle under the forces that
it
act does not in general lie in the surface just considered, nor is tangent to the surface. If the surface is smooth, the reaction
particle will be
on the
321
But the actual work s due to the reaction will be 0. done by the reaction in A seconds along the natural path will in general be an infinitesimal of the same order as At. It is now easy to see how to compute Q l and Q z m case the The virtual work of the reaction of the surface surface is smooth. is nil, and so we need consider only the other forces. The virtual
and we can compute
and
13.
5ft
displacement takes place in the tangent plane to the surface, directly the virtual work corresponding
to the successive virtual displacements given
by
8q l
0,
8q2
=
In
0,
8q z
cf.
further
16 infra.
in the Choice of the
ir r .
Virtual Velocities,
an Aid
the general theorem of 10 there was no indication as to how In certain cases which arise in practice, the 7r r may bo chosen.
the motion being subject to Lagrange's Equations
/
:
d 3T
dT
happens that there are geometric or kinematical relations between the q r 's of the form (28), 10
it
:
(2)
li(f
ft
+
,
'
'
+ dmsqm +
CL 8
0,
1,
-,/*< 1,
,
where the a r8 a 8 are known functions of q lt q m t,* which do not depend on the initial conditions, and where the rank of
,
the matrix
an
(3)
IS
fJL.
If, now, the possible virtual displacements corresponding to an arbitrary choice of 5ft, 8qm are so restricted that
,
(4)
it
a ls
8qi
+ am8
8qm
0,
1,
/*,
constraint ")
of
work of certain forces ("forces of Hence by identifying these "forces " constraint with the Q* and setting 8q r = 7r r the hypotheses
vanish.
,
of that
* It
theorem arc
fulfilled.
form: but it
may happen that some or all of these equations may be integrated in the F (qi, m t) = 0, where F does not depend on the initial conditions;
-,
</
is
322
MECHANICS
24 as free to Example. Consider the disc of Chapter VI, without slipping on a rough horizontal plane which is moving for example, rotatin its own plane according to any given law with The force which fixed constant a about velocity. point ing
roll
;
on the
on the
a
disc.
But the
is
virtual
disc at the point of contact will do work work of this force, when the virtual
is nil.
displacement
restricted as above,
7r r ,
Thus we
are led to
On
the
Number
:
of the
qr
For a system
of particles
the q r 's, as has already been pointed out, can always be identified with the coordinates
m = 3w, and Lagrange's Equations become identical with Newton's Equations. In 'theory, then, there is no difference between the two systems. In practice, Lagrange's Equations provide in many cases an elimination of forces in which we arc not interested. Consider a ladder sliding down a wall. If the wall and floor are smooth, we may take m = 1, q = 0, and all the forces in which we are not interested will be eliminated. s Here,
Here,
(2)
a cos
0,
a sin
0.
MgeM K
FIG. 144
Hence
(3 )
r =
:
^(02
1)0.
where
U=
Thus we
Mg a sin 0.
motion
:
M (a +
2
A; )
-^
= - Mga cos
ag
or
fK\
d?0
323
maximum
we think
of elimination has
of the ladder as
made up
been atof a
huge number of particles connected by weightless rods, the forces have been eliminated, and also the forces exerted by
Suppose, however, that the floor and the wall We can still write down a single rough.
Lagrangean equation,
But now
2a cos Op,S
(7)
W Q = _J = - Mga cos
dq
2a sin
R.
We
(8)
The
have not equations enough to solve the problem. = 3 and setting difficulty can be met by taking m
0i
x,
02
27,
03
0-
is
given by
(1).
And now
Q!
(9)
1
=S -
M#,
=
;
a(sin
Lagrange
three equations
(10)
=
The
first
a (s in
+ n cos 0) S + aO
sin
cos 0)
two
of gravity ; of gravity.
of these are the equations of motion of the centre the third, the equation of moments about the centre
did in Chapter when we the theorem the motion of the about IV, 1, proved centre of gravity; and similarly, when we proved the theorem of moments, Chapter IV, 3 and 9.
What Lagrange's method here has done, is the internal forces between the particles, just as
first
to eliminate
we
324
Let
MECHANICS
Qr
+ Qf,
where
Q*
We wish
(11)
?r2 , 7r 3 ,
such that
Ql^
TT I
+
/i7T 2
Q*T2
Q,*T,
0.
+ +
7r 2
+ +
?r 2
a(sin 6
(M sin
:
/x
cos cos
0) 0)
7r
= = =
7r 3
sin (0
2X),
cos
(0
2X),
7r 3
I/a.
But a mechanical derivation is easy, too. Consider the resultant of the forces R and p,R. Draw a perpendicular to it through the lower end of the rod, and displace this end along this line.
Do
the same thing at the upper end of the rod, and displace the upper end along this line. The result is, that
f
'
Sq l dq2
lo)
=-asin = a cos
(9
(6
Corresponding to such a displacement the virtual work of the "constraints" must vanish. And now it is merely a question of trigonometry to show that our expectation is fulfilled
:
(14)
Qj>i
Ql fy,
,*?,
0.
Equation (11) corresponds to Equation (15) of 10, and Equations (12) are the Equations (16) of that paragraph. But Equations (13), though corresponding to Equations (4), 13, do not have their origin in Equations (2), 13. The latter would arise
from differentiating (2). Returning now to Equations (10), we see that the unknown reactions R and S are eliminated by (14), where 5q lf 8q.2 8q s sat,
isfy (13).
Hence
Jflfift
(15)
+ (M +
(0
Mg)
dq,
Mk*0dq z =
cos (0
0,
or
(16)
- MX a sin
2X)
2X)
325
(2),
leads at once
problem
(17)
a 2 cos 2X)
~~ +
a 2 sin 2X
^~
+ ag cos (0 +
of
2X)
=
of
0.
15.
Forces of Constraint.
definition
"forces
con-
straint" from the point of view of physics, which shall be both accurate and comprehensive, has, so far as the author knows,
in
such forces as
;
I.e., Equa*, 9, which disappear from the Q r 10 arise from unknown Arid still again, the Q* of
Z* of
by the method
of multipliers.
Perhaps these two cases are comprehensive in Rational MeAre there problems in this science not included here? chanics.
If not,
straint
cf
Appendix D.
16.
When
(1)
Euler's Equations, Deduced from Lagrange's Equations. a rigid body rotates about a fixed point, the kinetic energy is
T = i(4p
p, q, r
+ #? +
2
).
Let
15:
p
(2)
6 sin 6 cos
<j>
(p
\}/
sin
cos
<p
q
r
<p
+
cos
\l/
sin Q sin
<p
-{-
\l/
The second
of Lagrange's
Equations
.
is
readily
dr
computed
Cr;
dT =
dp dq + CV + Bq~~ Ap~
A
,,
.
v<P
O(D
V(D
(/<p
TT-
8 cos
<p
sin 6 sin
<p
g,
=
Hence
6 sin
v> -\-
$ sin
cos
^>
p,
326
MECHANICS
<$>,
To compute
may
act,
and a couple. they can be replaced by a force at can be realized by means of three forces
:
The
latter
o
ii)
r
r T
Ma
"
"
Hi)
= = =
a.
virtual
displacement
80
0,
5<p
0,
d\l/
gives
as
the
virtual
work
N5<p
and
Hence
(A
<t>
N and we have
N.
Cj|This
is
-B)pq =
the third of Euler's Dynamical Equations. The other this one by symmetry, and are obtained by
letters cyclically.
EXERCISE
Obtain the
six
Lagrange's Equations.
17.
Solution
if
of
10 that
the
paragraph, then
(i)
dt dq r
:
=
dq r
Qr
2)
a r8 \ 9
1,
m,
(2)
is
of rank
ju
<
m.
By
is
the
"
point r
"
is
of the vector r
when
the
initial
point
at 0.
327
ju,
qi
+
is
ams qm
,
=
,
aa
1,
where the
qm
t.
The
the ql9
kinetic energy
,
,
T = T2 + T,+ T
where
and T19
ish identically.
The
(4)
2
*,./
</*<**
is
For
T can
be
/tu,
ju
(7
are functions of
known
/*
t. #i, -, ^ m function of q r q r t.
,
THEOREM.
The
m+
d dT
Equations:
8T
A
a.,
H---p,
+ o.g =
(ft,
1,
/z,
m+
functions
</, X x ,
:
XM
The
ir
first
m of
A
1,
qi
A mr qm
r
a rl Xj
a r/A X M
</,.,
= Br
,
m, where
is
a function of the
:
qr
and
The
remaining
(6)
/*
----
+ amsqm = C
,
a,
1,
/*,
where C.
is
m+M
Xj,
m+
:
X^.
Their determinant
328
MECHANICS
""
an
aii
Oi,A
OmM
m+n
+
linear
homogeneous
u ti
+
* '
AmiZm
an*?!
+ ai^ =
'
' '
lmti
+ Ammtm +
Oml^
+ Omplfc =
= =
+
But the rank of the matrix Hence all the (2), is /x.
contradiction.
i) l9
,
0.
of these equations,
Next, multiply the r-th equation (8) by r r = 1, m, and add. The terms in i^, 77^ drop out because of the last p. of
,
, ,
the equations
(8),
and so there
where not all the j, m are 0. This is impossible, since is a positive definite quadratic form. Equations (5) and (6) admit, therefore, a solution
,
:
(4)
0,
(9)
329
determinant
an
<v
I/*'**
Uft.fi.
wo
of
given by
or
where, in particular,
ticular integral
/, is linear in <fo+i,
is
qm
This
which
initial
is a parconditions of
qa
Ka ,
1,
m.
The system of
alent,
m differential equations (9) is now seen to be equivunder the restrictions of the dynamical problem, to the
:
system
dq8
'
,
'
'
'
'
j\
'
(12)
d Ka
,
Km,
0,
tt
/*
+ 1,
,
.
is a system of 2m p differential equations of the first order Their M unknown functions q r K a determining the 2m desired functions, q lt The X 8 are solution yields the qm now uniquely determined as functions of t and the initial condi-
Here
for
tions.
on which
gr
of q r
q r are restricted
* *
I
by the equation
fl\sf]\
*
I
drnsqm
$.s-
It
is
important here, as in so
many
dis-
cussed in this chapter, to distinguish between constants that are connected with the choice of coordinates and constants that arise
from the
initial
Thus in the
of ('hap. IV,
different level,
ad
c.
330
18.
MECHANICS
n
Let a dynamical system be given, with Equilibrium. the motion being subject to Newton's Law particles m,,
:
(1)
Z
forces are said to
it
1,
n.
The
(2)
be in equilibrium
t
if
0,
Y =
0,
0,
A
(3)
is,
that
0,
0,
0.
are not interested in the general case, which, in accord with the definition just given, relates to a single instant of time, the forces not in general being in equilibrium at any other instant.
We
have concern rather with a permanent state of rest of a system capable of certain motions which are subject to geometric conditions.
We
We
of particles
A unibut also of form circular disc has a particle attached to its rim. The disc rests on a smooth ellipsoid and a rough table which contains two axes of the ellipsoid. Find the positions of equilibrium.
:
and rigid bodies as were studied in Chapters more general problems, like the following
and
II
More
precisely, the
system
shall
:
positions defined
by
the equations
=
(4)
fi(q\,
,?*)
\f/i
do not depend on
t,
(5)
s ra.
last
has
the least value for which the x iy y^ Zi can be represented by equations of the form (4), satisfying the above requirements. It is
331
qm
may
j
be connected by relations
(6)
Ft
(ft,
qm )
0,
1,
<
m.
it does imply that if the x^ y iy 2, all vanish, true of the q r and conversely; and if, furthermore, both the ij #, z, and the x iy #, z all vanish, then this is true of the q r and the q r and conversely.
On
then this
The motion
tions
(7)
:
of the
system
is,
first of all,
[71,
= Qr
1,
<
m,
where by definition
\T\ lr 1
=<L2L-VL.
dt dq r
dq r
:
To
(8)
these
may
a la qi
+ am*qm =
0,
1,
,/*< m,
(9)
is
(1.
It
is
possible that
some or
all
A
(10)
first
(6), but this is unimportant. necessary and sufficient condition for equilibrium
is
that
Qr =
0,
1,
m.
For, a necessary and sufficient condition for the vanishing of the is that ft = 0, left-hand side of (7) for ft = 0, ,&, =
-
qm
0.
the form
:
(8), it
may happen
that the
can be
split
up
as follows
(11)
Qr
Qr
Q?>
where
(12)
Q*ft
+ Q* m *qm =
=
s
0,
+ Om^qm
0,
1,
/*.
332
MECHANICS
sufficient
condi-
that
<?>/!
m dqm = +Q'
0,
provided the 8q r satisfy (13). That the condition is necessary appears from the fact that (10) is true, and hence
(15)
(Q[
<#)
& +
.
(<&
Qi) ?*
true,
and
(14)
now
follows.
Suppose conversely that (12) and (14) hold when the 8q r are Then the system is in equilibrium. Suppose subject to (13). the statement false. From (12) and (14) it follows that (15) holds, provided (13) is true, and hence from (7) it follows that
(16)
i)[r]r5<y r
T~.\
0,
Lot
qr
= =
cr
=
0.
1,
m,
Then not
all
the c r are
cr
,
Now,
1,
,
(18)
dq r
m,
is a system of values satisfying (13). For, on differentiating (8) = 0, these relations with respect to t and then setting t = / qr
,
follow,
namely
0>lsC\
'
'
'
OmsCm
0,
1,
.
(JL.
cr
Now,
(19)
T =
f)T
v(]r
^A^q
a,
ft
a qft .
Hence
=
-
Ai r
,
(ji
'
'
'
+
. .
A m rqmj
.
d BT
^ -QT
and so
m
.... = Air q\ +
m
A +A mr qm +
..
terms in
,.
(q l9
qm ),
initially
[T] r dq r
-!
r=l
^ MirCj H
A mr cn )c =
r
^ A pc
a
a,/3
a cft .
333
a contradiction, since (19), being a positive definite quadratic form, can vanish only when all the arguments are 0. We can state the result as a
THEOREM.
system, the
motion of which
_rf
d_T __
8T
(20)
dt'dq r
i7i
Wr~ + +
of the
V"
Om. (7m
- 1 '"-> m;
0,
=
p,
1,
,/*< m,
/*,
matrix
of these last
qr
equations is
1,
,
be in
that
equilibrium and at
(21)
where. 5q
-
rest, is that
0,
m, and
Qiffi
} ,
+
a m5 <$tf w
+Qmdq m =
=
0,
(22)
a\8qi
7
T
0,
=
definite
1,
-,/*.
is
homogeneous
positive
quadratic form
in
//,
in particular,
(23)
Qr =
?/
Qr
Qr*,
1,
m,
and
(24)
is
known
that
Q*fy,
,
+
+
-
+ +
n dq m Q*
= =
0,
where dq l9
(25)
19.
can 6e replaced by
QlSft
m dq m Q'
0.
Small Oscillations. Two equal masses are knotted to a Determine string, one end of which is made fast to a peg at 0. the motion in the case of small vibrations. Here,
T =
^
<p,
(202
^2
^ cos
<p).
fc,
)),
U = mga(2
Since
6,
cos
cos
d,
<p
placed
by the approximations
9X
FIG. 146
U =-
mga(Q*
+ ^) +
const.
334
Equations
MECHANICS
(1) arc typical for
an important class of problems system about a position of stable equiforces the being derived from a force function, applied librium, for Let T and U both be independent of t\ let q r = (7. r = 1, be of let the and be reU m, position equilibrium, T,
in small oscillations of a
,
when q r
Then
T = =
-
V
r.s
dra qr<ls
r,
1,
ra,
where the
coefficients
o r,
is
:
a ar
b r,
b, r
are
constants and each of the quadratic forms Lagrange's Equations now take the form
(3)
Orltfi
definite.
'
'
'
Cl rm
qm
=~
-
(&rl<7l
-
+ &rm?m),
1,
m.
*To integrate these equations, it is convenient to introduce new * It is a theorem of algebra variables as follows. that by means
qr
riq(
Mmrf,,
1,
ro,
sum
of squares
T = q(* + U =- n2
t
+
2
ft'
m\ q'
-
nm 2
2
.
Lagrange's Equations
now become
$-V*
:
r-l,...,.
=
,
= C
cos (n r t
7r),
1,
m.
we
(8)
qr
CiMri cos
(nj
TI)
+
-
7m),
1,
m,
where the
C y
r,
are the
2m
constants of integration.
LAGRANGE'S EQUATIONS.
In this
result,
VIRTUAL VELOCITIES
in theory, there appear,
335
complete as
/* r,
it is
howThese
can be determined by the following consideration. in (8) when r 9* s; and let C 8 = 1. Let C r =
Thus we have
a special solution
(9)
q?
\ r cos
a,
(nt
8
.
7),
where X r
(10)
=
(bn
/ir
and n
n*arl )
= n
7 = y
(b
-
Substitute q? in (3)
Xi H
-----h
m,
n 2 a rm ) \ m
0,
1,
m.
A necessary condition that (9) be a solution is, that the m linear Equations (10) admit a solution in which the X/s are not all 0. Hence the determinant of these equations must vanish
:
n an
(ii)
=o.
6mm
n a
2
If
the
nr
are
all distinct,
.
roots of
2 Moreover, each n? leads to a unique deterequation in n mination of the ratios of the X's through the Equations (10),
this
of (11) coincide.
still
In that
have k
linearly independent solutions (9) corresponding to such a root n2 More precisely, let n 2 be a multiple root of order fcj n 22
.
fc
etc.
Let n2 be set
lt
,
= n^
it
in
Equa-
unknown \
all
Xm
is
possible to
rest will
uniquely determined.
Moreover these solutions are obvi2 ously linearly independent. Proceeding to n 2 we determine in the same manner fc 2 further sets of (X n X m ), each set giv.
,
be but one of these A^ X's be set X w ), and each set gives a sets of (X x
arbitrarily,
Let
these solutions are likewise linearly indeing a solution of (3) and also of the earlier solutions. And of one another pendent
;
Thus
linearly independent solutions (9). The variables q' normal coordinates of the r are known as the
problem.
Each
is
336
MECHANICS
(11)
are distinct.
But
in
number
of different choices
are possible.
EXERCISE
Carry through the example given at the beginning of the paragraph.
Ans.
Two
:
sets
following
f
0j
cos fat
7j),
cos fat
72)
^ = A/2
where
cos fat
+ 7i),
- V2) 2
Cv
02
= ~ ^2
(2
cos fat
7 2 ),
n* =
The
(2
n2 2 =
+ \/2) ?
C*
general solution
is
tf?
= =
<?!
cos (n^
TJ)
+C
TI)
cos fat
2
7 2 ),
fat
- C V2 cos
72 ).
EXERCISES ON CHAPTER X
smooth wedge rests on a table. A block is placed on Determine the wedge, and the system is released from rest.
1.
the motion.
2.
Two
rest, slightly
A
A
uniform rod
Will
it
is
gravity.
4.
move
uniform rod of length 2a and mass 3m can turn freely about its mid-point. A mass m is attached to one end of the If the rod is_rotating about a vertical axis with an angular rod.
velocity of \/2ng/a,
will dip
till
the rod
and so released, show that the heavy end makes an angle of cos~ 1 (Vn 2 + 1 ri) with
Determine the motion of a top whose peg, considered as a point, slides on a smooth horizontal plane.
6.
7.
is
taken into
account.
337
8. The ladder of p. 322, the initial position being oblique to the line of intersection of the wall and the floor.*
9. Two equal rods are hinged at their ends and project over a smooth horizontal plane. Determine the motion.
SUGGESTION.
of gravity
Take
; (2) the inclination 6 of the line through the centre of and the gravity hinge ; (3) the angle a between this line and either of the rods.
Two
of Lagrange's
of gravity. third expresses the fact that the total moment of momentum with respect to the centre of gravity, is constant. And fourthly there is the equation of energy, f
10. A rough table is rotating about a vertical axis. Study the motion of a billiard ball on the table, assuming that there is
no
slipping.
11.
12.
slipping.
19, p. 333,
Work
the problem of
when the
particles are
not required to
13.
move
in a vertical plane.
Two
equal uniform rods arc hinged at one of their ends, of one rod is pivoted. Find the motion for
lying in a plane.
is supported by two strings of equal length, ends, their other ends being made fast at two points on the same level, whose distance apart is equal to the smooth vertical wire passes through a length of the rod.
16.
uniform rod
its
attached to
small hole at the middle of the rod and bisects the line joining the fixed points. Determine the motion.
16. If in the preceding question the wire is absent, study the small oscillations of the rod about the position of equilibrium.
bead can slide on a circular wire, no external forces the motion in two and in three dimensions. Determine acting. of no friction. the case with Begin
17.
* Routh, Elementary Rigid Dynamics,
p.
329.
t Appell, Mtcanique rationelle, vol. ii, chap. 24, of the present kind are found in this chapter.
446.
Many
other problems
CHAPTER XI
HAMILTON'S CANONICAL EQUATIONS
1.
The Problem.
The problem
of this chapter
:
is
the deduc-
dqr_m "
dt
'
dpr__W
dt
'
i'
...
'
'
dp r
:
dq r
d dL -- dL
Q~
^"~
n
0,
1,
m.
cftd</ r
dq r
is purely analytical, involving no physical confor that reason it is well to set the theorem and cepts whatever, and proof apart in a separate chapter. The problem can be stated as follows. We start out with a Lagrangean System. Such a system is defined as a material system which can be located by means of m generalized coordinates q ly qm and whose motion is determined by Lagrange's
The
transition
Equations.
tions
:
If
we
set q r
Kr,
2m
equa-
A)
d ML
dt dK r
r
_
dq r
1,
<
,
m, where
' '
(1)
is
L = L(q
function
l9
tfm, *i,
Km,
i(0r,
*r,
<.
a function of the
2m
is
The
there
is
called the
In case
a work function,
is
(2)
L = T+
338
U,
339
qr
and
T = T(q r
?r
is
the Jacobian
/\
(6)
d(L 1?
Lm )
0(*i,---,O'
where
The (2m
shall
S2m +i
of the variables (q r * r
,
t)
of the variables (q r
/ M \
pr
t)
by means
Pr
R2m +i
m,
of
(4)
= 8L
^-,
1,
the
f/
The system
2m
dependent variables q r K r namely, Equations A), thereby goes over into a system of 2m like equations in the 2m dependent variables These last equations are the following: qr pr
, ,
.
2m
fr . (5)
-*=& -*=-*?
=
dq r
811
dp r
8H
r
'-"-^
where II
H (q
r,
pr
t) is
(6)
r=l
the K r being functions of (q r ,
,
I) prKr
L,
pr defined by (4). This is the theorem which is the subject of this chapter and to the proof of which we now turn. The converse is true under
suitable restrictions.
2.
General Theorem.
Let F(x l9
x n ) be any function,
:
continuous together with its derivatives of the first two orders, and such that its Hessian Determinant, the Jacobian
<"
Let a transformation, T, be defined by the equations
:
T:
340
Let
MECHANICS
G (y
lt
,
(2)
G(yi
,
j/n)
= J
r=l
xr yr
- F(x
l>
-,xm)
>
where x l9
1
.
y n defined by
Then
(3)
-%,
r-
!,...,.
dx s
The right-hand
and the proof
is
side,
by the equation^
defining T, reduces to x r ,
complete.
Furthermore,
For, on performing first the transformation T, then the transformation T~~ l the result is the identical transformation. Hence
j
d(x u
or:
x n)
'
d(y l9
yn)
,
G.)
3(*i,
*n)
In particular, then,
f^i
/r v
d(Gi.
Gn )
r=
r$(ri,
I
/n)"T~
I
These
results
may
THEOREM
condition:
I.
Let
F(x ly
x n ) be a function satisfying
the
Perform
the transformation
T ^
11
2/r
- dF
1,
w n.
341
G(ylt
y n ) be defined by
n
,
the equation:
(6)
G(yly
is
y) = 5)
xr yr
- F(x
l9
x n ),
r=l
where x r
determined as a function of (y lt
T~
l
,
of T.
.
Then
the inverse of
T
-
is represented
T~ 1 l
xr
^,
' '
1 l,
n n.
Moreover,
d(Gi,
^(2/1,
'
>
Gn)
-,2/n)
In
particular,
The
(7)
identical relation
can be written in
the
symmetric form :
F(xlt
,x n )
r-1
where
3F
and
the
dG
Hessian Determinants of F,
are
0.
is
We now
of importance
THEOREM
II.
// F,
%
,
G
x^
and
if each
depends
on a parameter,
(8)
the relation
J'ft; a?,,..-,
+
T
Oft; yu
or
1
2/n)
T=l
Jxy
r
being
(
;
an
identity, because of
T~
l
,
either in the
(
;
+
,
arguments
yly
arguments yn
Let
!,, X
B)
(8).
Then
=
y r ft
But
and the proof
&
ft'
is
complete.
342
3.
MECHANICS
We
fulfils
start
the condition
w
(2)
'"' L
n(if V \ K li
) if } K m/
pr
-,
1,
m.
// (q r ,
pr
f)
is
L
then,
+H=
,
%p
T
,
Kr
If,
we
all
parameters,
K r y r = p r and regard the q r and t as set x r 2 will be met. the conditions of the theorems of
by the equation
+
dt
~
g^
dt
t
,--,
m.
is
Turn now
to Lagrange's Equations
dq,
~dt
<9L
_
(4),
cuu
dt d* r
dq r
The
first
of these,
combined with
dqr
,
gives
m
\OJ
m
,
r-\ 1,, m
/
III.
From
(9)
(5),
we
infer that
-37-
-r-
1,
m.
343
dqr_9H ~
dt
>
dpr__3H dt
>
A'
'
'
dp r
dq r
which we set out to establish. The mathematical converse with the condition
is
simple.
Given Equations
(10)
Equations
(4)
by
(3).
Then
first of
define a transformation, and then L (2) arid (5) follow from the theorems of
is
defined
2.
And
now
the
Equations
(10),
combined with
(4),
gives
dq r
=
(2)
= l,...,m.
.
(10),
combined with
d dL
and
(5),
leads to
~7i ~n
dt 8K r
dL _ ~^
dq r
"
"
"
1>
W"
Thus wo
is sufficient that a knowledge of the function complete mathematical formulation of the motion. But what can we say of the physical meaning of // in the general
We
see, then,
for a
an important restricted class of cases in which simple. Suppose there is a force function U on t and furthermore that the kinetic enr q depending alone, T is a definite positive ergy homogeneous quadratic form in the
case?
is
There
the definition
is
(/!,--, qm
(11)
>
Let
L = T
Kr
U.
Let
qr
>
Then
The transformation
8L
*-aT
now becomes
:
344
MECHANICS
Thus
?,,r
=l
2
r
" ,j-2T.
c/
Hence
(12)
ff
p r *r J) T
if
-L
T
=
nor
T-U.
[7
Still
more
specially,
total
neither
depends on
t,
then //
energy of the system. along any given path appears as follows. case the relation
:
becomes the
That
is
here constant
We
dH
as
is
m
:
seen at once
by
differentiating
*H =
dt
$,
VM^: + T?dH_dpr ^ dp
d(Ir
BH
'
"*"
dt
dt
dt
of Hamilton's Equations.
But
in the gen-
0,
and so
H
:
is
path.
since
(14)
If,
however,
H
0,
now dH/dt =
we have
h.
CHAPTER
XII
D'ALEMBERT'S PRINCIPLE
1.
The Problem.
it
The
so far as
i) in
relates to a
system of
can be formulated
Law
-
of
Motion
A)
miXi
Xi,
rmi/i
= Y
i}
iy
wz = Z
it
1,
,n
ii) in terms of further conditions equivalent to 3n relations A postubetween the 6n + 1 variables (x y i} z iy if Yi, Zi, t). lational treatment of these conditions will be found below in Appendix D.
be given as an explicit function if and their first derivatives with respect to the
may be Z may
First,
time,
and
X, =
YJ =
*/ fa, y t z if
,
Xi,
i,
y iy y i}
z {,
t)
*/ fa, y iy
Zi,
Zi, t)
Thus A) reduces
tions for determining Xi, y iy Zi as functions of the time, and with the solution of this problem the determination of the Xi, Yi, Zi is given by substitution.
Secondly, at the other extreme, the path of each particle and the velocity of the particle in its path may be given. Thus #ij y\> Zi become known functions of t, and again the Xi, F t Zi
,
are found
a class of problems in which constraints occur which can be eliminated by a general We shall not attempt to give principle due to d'Alembert. a general definition of " constraints/' for no such definition exists ; but we can formulate a requirement which embraces the ordinary cases that arise in practice. Let 8x it dyi, 6z be any 3n quantities
is
t
345
346
whatsoever.
MECHANICS
Then
true
:
it is
equation
(1)
is
2)
-i
m& ~ x
is
*>
*x *
+ m *y* ~ r
(
fy<
+ m <*< - z
(
*>
dZi
This equation
of Dynamics.
Now
(2)
it
may happen
forces
:
X^ F,
Zi
Zi can be broken
up into two
Xi = XI
X!,
Yi
forces,
Y(
Yl
Z(
+ ZI
X*
i,
F|*,
it F, Z\ and the namely, the X' than the old for the following reasons.
i)
The
Zi,
yi f
known
ii)
Yi, Z{ are either explicit functions of the x^ ?/, Zi, or they involve in addition a restricted set of unfunctions arising from forces which are not given as func-
tions of these
&n
variables.
The Xf,
Yf,
(3)
Yfdyi
ZfdZi
=
\i
!--=!
which
dyi
satisfy the
equations
(4)
<=1
^A
/
ia dx i
+B
ia
CieSz^
0,
a
6n
=
1
1,
where the
Xi y yij
Zi,
variables
Xi f
"11
of the matrix
"
-"nl
'
(5)
is
/x
and
conversely.
of- Equations (4) the 3n quantities X*, Yf, Z* can be expressed in terms of /x unknowns as follows. Multiply the
By means
i-l
2) (ft
a-l
D'ALEMBERTS PRINCIPLE
Now,
347
in Equations (4), a certain set of 3n n of the &c, By*, dZi can be chosen at pleasure and then the remaining /* of these quantities will be uniquely determined, for at least one At-rowed determinant from the matrix (5) does not vanish. It follows,
then, that \,
set of
ju
X^ can be determined uniquely from a suitable chosen from the 3n equations equations
,
:
H (7)
M
,
f-
X*
a= 1
2^ Aia \a
Yi
And now
/z Equations (7) will be satisfied For Equation (6) has become an equation in which only those terms appear for which 6x,, 8yi, 6z are arbitrary, and hence their coefficients must each vanish. Equations A) can now be written in the form
by these values
(8)
mix*
= X(
5)
A ia \ a
nnyi
Y(
fi ia
X,
where the X t
of
/z
X^ have
9
come to us
as linear combinations
On
in Equations (8) merely as /z unknown functions, which can be determined from /* of these equations and then eliminated from
the remainder.
To put first things first was never more imporVirtual Work. tant than in the statement of d'Alembert's Principle. The 3n
8xi, Sy iy dZi are to begin with 3n arbitrary numbers, and we then proceed to restrict them by the equations (4). Nevertheless, whatever values they may have, they determine by defi-
quantities
(Xi, t/,
2,-),
W
is
2)
dxi
Yi
By*
+ Z, dZi
=i
by
displacement. Thus Equation (3) says that the force X* Y*, Z* is such that it does no virtual work when the virtual displacement
is
(4).
348
MECHANICS
Given
and
The discovery of an analysis of Xi Yi, # by (2) Equations A). the most of general virtual displacement 5Xi, dy^ 6z if whereby the
y
virtual work of the force Xf, Yf, Z* vanishes, this virtual displacement being expressed by (4); finally the elimination of dXi, dy if dz ir and X*j Yf, Z*, as above set forth, whereby 3n n equations free this is the spirit and content of from these unknowns result; d 'A lembcrt's Principle
.
origin,
This enunciation of the Principle does not represent its historic but rather its interpretation in the science today cf
;
.
Appendix
2.
1).
Lagrange's Equations for a System of Particles, Deduced Let the coordinates x, yi, z of the 1 be expressible in terms of system of particles considered in
i?m,
(D
<im,
tfm,
where
the
(</!,
(q l}
,
q m ) is an arbitrary point of a certain region of Let m g )-spacc and the rank of the matrix is m.
,
yields
dT
where
8T
ZdtWr'Wr
and
(4)
Hence quantities 6q l9 dqm are wholly arbitrary. the coefficient of each term in (3) must vanish, and so we arrive
Now, the
D'ALEMBERT'S PRINCIPLE
at Lagrange's Equations in their
of particles
,,,
:
349
for a system
d 8T
8T
Here, no restriction whatever is placed on the forces nor is the number of q r '$ required to be a minimum. In an important class of cases which arise in practice,
(6)
X, F
t,
x, =
x'<
+ xi
F,
F;
17,
Zi
z;
+ z*,
where
Hence
1,
TW.
Equations
case.
(5)
The
included.
3. The Six Equations for a System of Particles, Deduced from d'Alembert's Principle. Let the system of particles of I be subject to internal forces such that the action and reaction between
any two
in the line
through
the particles
X*
And
ij
__
let
any other
forces
X' tj
dXi
fyi
dZi
Y[, Z( act.
Let
where
a,
6,
c,
a,
/?,
Since the
internal forces destroy one another in pairs, and likewise, their moments, Equation (1) goes over into the following:
350
MECHANICS
=
a
J (m,< - X'
()
+ b 2 (*<#< -
F|)
+ c 2 (<*< -
Zf)
18
2) ( x
Now, set any five of the six quantities a, 6, c, a, #, 7 equal to 0, and the sixth equal to 1. Thus the six equations of motion, from which the internal reactions have been eliminated, emerge
:
n
)
2J
(1)
Momentum
^=
and as the Equation
of
Moment
dt
of
Momentum
We have used
conditions.
going choice of dx if 8y iy
4.
deduce a set of necessary because the forenot in general the most general one.
sufficient,
Lagrange's Equations in the General Case, and d'Alembert's Consider an arbitrary system of masses, to which Principle. Lagrange's Equations, on the basis of suitable postulates, apply
/i\ (1)
d dT
5a
set
--"
dT
r-i,..-,*.
If
we
by way
of abbreviation
(2) (2)
ddT _dT ~ _
D'ALEMBERT'S PRINCIPLE
then
(3)
351
2 ([71 - Qr)
,
Sq r
0,
Qr
It
may
Qr
Q'r
Q],
where Q' r
(4)
is
for
+ +
' '
Q;s?m
0,
provided
(5)
daidqi
:
+ a>am8qm =
0,
a =
1,
/I,
being /*. By reasoning precisely similar to that used in seen that the Q* can be represented in the form
:
1,
it is
Qr
2L a <*r^a,
1,
W,
where the X can be interpreted physically as certain linear combinations of a suitable set of /* of the quantities Q*
.
d dT
dT
where now the X a are thought of as unknown functions, which can be determined by /* of these equations and then eliminated from the remaining m n equations.
Virtual Work.
In
all
can be interpreted as the virtual work done on the system by the forces which correspond to the Q r In particular, then, the condition (4) means that the virtual work of the forces which
.
lead to the Q* is nil, provided that the virtual displacement corresponds to the condition expressed by Equations (5).
352
5.
MECHANICS
Application:
forces.
Euler's
by any
Its position
may
q2
be described geometrically in
15
:
0,
t,
q*
<p.
by Chapter VII,
-i
T =
p
(Ap*
+ Bq* +
<p
Cr 2 ),
where
=
=
\j/
sin 6 cos
(3)
q
r
\l/
sin 6 sin
(p
+ 6 sin + 6 cos
<p
<p
cos
By
d'Alembert's Principle,
where
all
Let 5q
0, dq 2
0,
Compute
The value
is
seen at once to be
On
Denote
the vector
moment
M
Then the
virtual
(5q ly 8q 2) 5g 3 )
= La
is
+ Mf3 + Ny.
to the virtual displacement
:
work corresponding
dq z )
(0, 0,
seen to be
Hence
Q = N, and we
3
find
Thus one of Euler's Dynamical Equations is obtained, and the other two follow by symmetry, through advancing the letters
cyclically.
D'ALEMBERT'S PRINCIPLE
The reader
will
353
But this is precisely the same solution 16." as that given earlier by Lagrange's Equations, Chap. X, the so far as the and this is details of solution True, analytic go ;
say
:
"
is
usually the case with applications of d'Alembert's Principle. It the approach to the problem through the General Equation
of
Dynamics, 1, which here yields (4), and the concept and use of virtual work, that brings the treatment under d'Alembert's
Principle.
6.
Examples.
of
the
ladder sliding
down a smooth
Fig. 88, p.
147.
problem as the motion of a lamina, moving in its own plane. AH the generalized coordinates of the lamina we may take the
coordinates of the centre of gravity
?i
:
x,
q2
y,
qz
8.
Then
(1)
T = %M(fr
is
+ p) + pf F
2
,
where k
By
(2)
d'Alembert's Principle,
([T\ r-l V
-Q
8q r ) '
0.
Q^ = S5x,
Q
and thus
3 5<7 3
2 6</ 2
(R
- Mg) by,
a (S sin
- R cos 6) d6,
is
determined.
Let
Qr
Q;
+ QM
a OS sin
9
where
Q*
s,
Ql
R,
* Q3 =
- R
cos
8).
Now,
(3)
If,
x, y, 6
are connected
by the
y
relations:
x
then,
a cos
0,
a sin
6.
we
a sin
6 60,
6?y
a cos
60,
we
(4)
see that
Qr^i
Ql 5<7 2
Q,*ff,
0.
354
MECHANICS
virtual
Thus the
work
:
and so Equation
(2) is
replaced
r-l
i;
(m, V
$W
'
== o.
Hence
M^j(On
is
a sin
0) 66
66
= 0.
from
replacing these second derivatives of x and y by their values (3) a differential equation in the single dependent variable
obtained
and
this
Rough Wall.
p. 323.
is
Equations
and
x,
(2) still
hold.
But now
Qi*fc
= =
(S
R)
Q 2 6q 2 = (R
/z
S - Mg)
8ff,
Cs ^3
Let
a [S (sin 6
+
r
cos 6)
+R
(/x
sin 6
cos 0)]
66.
Q =
where
Q;
+ Q;,
Q*
Qr Q*
= S - /,
= B
+ MS,
cos
0).
aS(sin
,
The
force
values of 5^ D 6q 2 Q* vanish
:
QI^ +
are found
c;?i
cr?i =
o,
by making
/z&7i
the coefficients of
and S zero
in this last
equation
+
.
6q 2
a (M sin 6
cos 0) 6q z
3
.. 4-
(sin
+ M cos 0) 6q =
2
/i
Hence
f
(1
(1
+M +M
)
2
fyi
= =
a a
(1
sin
(1
5^ 2
[-
2M sin
D'ALEMBERT'S PRINCIPLE
Equation
(5)
355
now becomes
Sq,
+ Mk^dq, =
0,
and it remains only to substitute the values of 5qlf 8q 2 from and the values of x, y from (3), and reduce. The result is
:
(7),
((1
a*
(1
+ M *) JJ +
2
ag
[2/x sin 6
(1
2
/*
cos 0].
The
in
elimination of the
virtual displacement (dq lt 5g 2 dq^) which here led to the unknown reactions R, S was not one which
,
any wise conformed to the constraints in the sense of the floor and the wall. If we replace R and /z# by their resultant and draw a line L through the bottom of the ladder perpendicular to it, and then do the same thing at the top of the ladder, thus obtaining a line L 2 the above virtual displacement corresponds to an actual displacement in which the bottom of the ladder is moved along L and the top along L 2
,
"
"
CHAPTER
XIII
1.
Definition of
is
Mechanics
:
Integral
V+
j
t/) dt,
extended along the natural path of the system, and is considered for a neighboring, or varied, path. The Principle asserts that the integral is a minimum for the
is
typical, is
its
then
value
natural path, or at least that the integral path, i.e. that its variation vanishes
:
is
(T
U)
dt
0.
It
is
ously we must begin by defining and by a variation 8. x' Let F(x l9 x n x[, n u) be a function
what
is
>
2n
variables indicated.
Here,
(x l9
x n ) shall
lie
region R of the n-dimensional space of the variables (x ly x' the variables x[ n shall be wholly unrestricted
9
a certain xn)
,
and u
lie in the interval a g u g b. The function F shall be continuous, together with its partial derivatives of the first and second orders.* Let
shall
*As regards assumptions of continuity, we lay down on GO and for all the requirement that whatever arbitrary functions are introduced shall be continuous, together with whatever derivatives we may wish to use, unless the contrary is
stated.
For an introductory treatment of the Calculus of Variations Advanced Calculus, Chap. XVII.
cf.
the author's
356
357
Xi
Xi(u),
in R.
a
Let
^
,
6,
1,
n,
be a path lying
Xl
_ ~
dxj(u)
'
du
l)-dimensional space of the arguments
following.
Thus a path T
of
in the (2n
F', is
meant the
:
Let
be a curve
equations
e),
Xi
Xi(u,
^ u g =
6,
1,
n,
where
a?t(w, 0)
Xi(u),
|
is considered only in a region for which and note partial differentiation with respect to u by ~~
'
,
e
|
is
small.
De-
d.
Let
e)
//
'
\
'
-i
, ,
,
du
The
is
is
what
#(,-,
is
F' in the
curve.
Since x l (u,
e) is
to the general
requirements of continuity,
we
dxi
see that
IH(U),
1,
n,
is
quirements of continuity.
The
defined
is
is
by the equation
dude
>o
Thus
Hence
.
-r--oa;
du
It
is
o-j
<
du
:
now
natural to lay
down
(4)
358
Definition of SF.
MECHANICS
By
the variation of
(a;,-,
x' t , w) is
meant
"=()..,
(u,
where Xi and xj on the right-hand side are set equal to x and ZI'(M, e). Hence
(6)
^-|(J>' + I
obvious that
It is
8(F
+ $) =
8F
+ d$;
and
also that
where
,
,*,,,
:
:
x'n,
u),
1,
m.
ddF = ddF,
dF
d8F
5?-"5T
*
b
And
(9)
similarly,
6
J
a
FSd* =
a
x n (u) play a r61e in the The dependent variables x^u), foregoing definitions analogous to that of the independent variables But the analogy holds only up to a in partial differentiation.
,
certain point,
and to assume
it
beyond theorems
like the
above
formulas which
physics.
we can
and
error in
Variation of an Integral.
b
F(Xi,
'
' >
X H9 X19
Xn U) du,
,
HAMILTON'S PRINCIPLE.
taken along the path F.
LEAST ACTION
:
359
By
8J
F'.
is
Extend
is
defined.
and now, by
(10)
definition
^
once as a theorem that
It follows at
(ID
The
integral
is
if
8
a
CFdu =
8x*
0,
=
is
m(u)
may
be.
The
condition
u = a
and for u
(12)
=
|
dxi
in (a)
0;
&e<
Ussb
^(6)
0.
For:
d ( $F d^ V^J
^ \ 5X
" dF 8Xi
~d$
* + dH W< 8Xi
i
'
Hence
03)
to vanish for an arbitrary choice of &C;, it is easily seen that each parenthesis in the integrand of the last integral must vanish, or
If,
is
(15)
jFdu=
j8Fdu.
360
MECHANICS
limits of integration
a'
The
may
),
be varied, too.
6'
6.
Let
^(a,
0)
= *(&,),
Let
where
^?(a, 0)
a,
\f/(b,
6'
J (c)
Then
= IF [Xi (u,
a'
e),
x'i
(u, c),
u\ du.
is
defined as before,
by
(10).
It follows that
6
b
f
(16)
i9
where
.
A*
z(o),
A'<
x(a),
<
EXERCISE
Since
l/,
6)
it
follows (under the ordinary hypotheses of continuity), on letting e approach 0, that the right-hand side approaches
f. du
The
left-hand side approaches 8&.
,eM>
5 -j-
Hence
aw
-7- 6$.
d aw
Thus Equation
is
necessary.
new
definition
2. The Integral of Rational Mechanics. All that has gone before merely leads up to the definition of the variation of the
following integral
ti >
(1)
jF(x
lt
---,x n ,x
,-",xn ,()dt,
HAMILTON'S PRINCIPLE.
LEAST ACTION
361
where the limits of integration may be constant or variable. The answer would seem to be simple, since t is the variable of integration and hence the independent variable in each of the functions
Xi
Xi(t).
is
taken in Physics
mean something
t
totally different.
Let
t(u),
^u
(0, 1),
-f. 0<f du
Then
the symbol (1)
is
(3)
/ =
JF (x
lt
,*,
|f,
..-,
|?)
t'du,
where x[ = dxi/du, and the "variation of the integral (1)" understood to be the variation of this last integral. Thus
is
(4)
dJ
=
o
Cd(Ft')du,
where u
is the variable of integration, and the independent variable in each of the functions Xi = Xi(u), t = t (u).
earlier definition
(4)
In particular, Equation
may
8J
= fbFt'du+ CF8t
du.
In each of these integrals the variable of integration changed back from u to t, and thus
t
may
be
<t
(5)
SJ
= CdFdt+
foddt.
362
MECHANICS
variation of
t,
The
and
namely
St, is
an arbitrary function of u :
T(M),
T'(U)
t'(u)'
U =
dSt
dt
where u 6F in (5)
is is
by
(2).
Moreover, by
m
where
8
/z,'\
(7)
t'5x' t
is
xW
.
?5
*,
= d
..,
Tu $x<,
st
a.
We see,
'
then, that
Now, when
/o\ (8)
d dXi = -
ir
show that
should
5i
it
The two
of
(8),
is
not invariant
Similarly,
the
independent
Why
be?
the variation of an integral is not invariant of the variable of Much of the confusion in the literature arises from integration. of this fact. The " variation of the independent losing sight
variable"
is
only when
analysis.
supposed to cover this case. It does so when and it becomes identical in substance with the above
By
defini-
Hence
n
(1)
&T = 5} mi(idXi
i-i
+
is
yifoji
may
be.
If the
former
u then
}
363
8T =
,[*,
(7) of
() + fr S () +
:
By Formula
y
iy
Ziy
this equation
a,
w.
J =
(4)
Xi
Xi(u),
1,
and
let
The
by
2, (5)
the value
(5)
f*Tdt=
*o
(*8Tdt+ CTdd
to
t
By
the aid of
(3),
The
term on the right can be transformed by integration by parts, the integrand obviously having the value
first
:
Hence
*t
<i
(7)
<
'l tl
5) wiifofa,
<
+ yi% + ^fe
{)
<o
2 TlTd^.
364
Finally, then
/i
:
MECHANICS
(8)
CT dt = - /
5)
i:sl
(ft fa,
The
variations
subject merely to
in particular,
we
1 arbitrary functions, dy^ bz^ dt are 3n the ordinary conditions of continuity. If, impose on &c-, dy iy 5z t the condition that they
dxi,
i.e.
for
*,
/t
<i
(9)
/o
Ardt = - f 5) ~
'o
mi(ft fa*
+ *< + 2*820 * ~
J
*o
and
''
/!
(10)
(*Tdt / /O
fa,-
+ y< fyi + 5f
4.
Virtual
Work.
By =
:
F,-,
t -,
Xi
Xi(u),
yi
y(w),
2.
ti(u),
UQ-^U ^ u ly
is
(2)
where
fai, 5i/i, 5^^ are 3n arbitrary functions of u, subject merely to the ordinary conditions of continuity. This quantity is often denoted by but it is not, in general, ;
dW
is
it is
5W only when
3 with Equation
of
HAMILTON'S PRINCIPLE.
LEAST ACTION
365
(1)
Here the 3n
that the
first
variations &r t
-,
6?y t ,
5z t
t
-,
5
tQ,
3n
If these are the total forces Xi, Y{, Zi are any forces whatever. acting on the particles, the right-hand side of (1) will vanish since
(2)
J(ST h
Xi = xi
+ 2T ~ +
<
W ) dt =
t
0.
+ xi
Y>
= =
y;
Y;,
z,
z;
+ zi
Then
(4)
W
W
8*
Wv + W
+
Yt8y {
s *.
=
t=i
X;txt
Ztdz<
0,
when
the variations
dX{,
8yi,
dzi
conditions.
Then Equation
form
(6)
these conditions,
dt
being
still
wholly
W9
t=i
2) XJ
:
366
, ,
MECHANICS
/<,
where A ta Bta Cia are functions of x,-, rank of the matrix An Cni
:
Zi,
x it y f z it
,
t,
and the
(9)
* 1 I/A
*
*
v> n/A
is M.
nomic
But
it
is
in general
no
Generalized Coordinates.
of each
' ' '
i}
q.u
an<i th e time
Xi
(10)
Vi
Zi
= ~ =
fi(q\j
'
>
qm,
q*n,
f)
<?i(qu
^i(q^
qmj
t)
is
m.
(5) is satisfied
,
when
"
OQ'i ~i~
dfi
^Q'm
O^m
^^j
*.
(11)
fa *
+
the 5^,
/1O\ (12)
/^
,
I*-'**.
8q m being arbitrary.
( (
Let
> Q = ''O
r
A<
V ^i +
I
-r
V/ Yi
^2/
-r
+
I
1,
m.
* The definition of the variation of a function, it will be recalled, is based on the dependence of the latter on certain arbitrary functions, whose variations may also be taken as arbitrary. These arbitrary functions are analogous, let us repeat, to the independent variables in the case of partial differentiation. And so further assumptions (i.e. postulates or definitions) are needed before dT can again mean a
variation.
367
W? =
the other hand,
Q!
&
+
,.,*,
+Qm Sqm
On
(14)
T = T(q
l9
,&,,<)
(2),
this latter
the independent , qm (u), t(u) of the present paragraph now takes the Equation (6)
form:
it
W
It
s is
the
Wy of
says
the variation of
and where
(13).
will denote this equation as the Fundamental Equation. embraces Equation (2) above, for the Xi, t/, z can always be taken as m = 3n generalized coordinates. This equation is sometimes written in the form
:
We
ti
X)
where
(15)
C(dT
5W)
dt
+ 2T 8dt =
0,
dW = Q
dq l
+ -- +Qm dqm
Let us see just what this means. First of all, the equation is These were, true under the hypotheses which led to Equation I. that the path is the natural path of the system, given by the
equations
(16)
:
qr
(0)
= = =
ffr(tO,
t lt
t(u),
where
t(l)
8q r
ir(u),
(w),
:
(0)=0,
0, 1,
...,m;
r?
(l)
0,
1,
m,
368
MECHANICS
to a further restriction
(u)
:
and possibly
I
rj r
<
A,
j (u) <
r
|
h,
0, 1,
m,
where h
is
8W = W$ is not in general the of function q qm t. The Q r have definite lt any values at each point of the natural path, and so are definite functions of u; but they do not in general have any meaning at a
Furthermore, in Equation X),
variation of
, ,
point (q r
t)
Finally, (17)
^SfV
^
where
\du du
du du
:
\du
The
last
term
/27
If q r and t And now the meaning of Equation X) is this are set equal to the functions (16) which define the natural path and if dq r 8t are chosen arbitrarily, subject merely to the general
:
,
X) will be fulfilled. Thus Equation X) expresses a necessary condition for the motion of the system and this in all cases, be they holonomic or nonconditions above imposed, Equation
holonomic.
Since Equation
X)
is
true for
all
when
any
special restrictions
it
:
we may choose
that the
example,
functions
may happen
Qr
Q
r
Qr
Qr*,
m,
such that
QiiQi
+
+
Qlfym
0,
provided that
a*i8qi
+ Oamfyw =
0,
1,
/i,
369
:
a ar
(q l9
qm q lf
,
is
IJL.
Here,
dW
is
replaced
by
(18)
5W
=
,
and dt, can now be chosen arbitrarily. but only m /i of the &/ r In what sense is 5T now a "variation"? Emphatically, in no for no definition has been laid down which reaches out sense to this case, and it is only from a definition that 5T can derive
;
Nevertheless, Equations (17) and (18) continue to define the values of the terms dT, bW that appear in Equation X), and thus this equation continues to have a meaning, and
its
meaning.
to hold
when a
certain set of
/*
variations 8q r ,
and
dt,
are
chosen arbitrarily.
Force Function.
Finally, there
may
be a force function,
(19)
where U is a function of the #,-, y^ z, and mental Equation (2) becomes in this case
:
t.
II.
F<,
Zi) of (19)
is
t ;
m,-,
we must understand
Again, there
in (15)
may
be a function U(ql ,
qm>
<)
such that
SU
Then
1,
m.
&W = BU
370
MECHANICS
II.,
and the Fundamental Equation takes on the same form, provided U does not depend on t otherwise,
;
'
~W
oq\
~'+W m
oq
*"
6.
The
Variational Principle.
The
pressed
Equation
or physical quantity,
5
I
does not assert that the integral of some function, is a minimum, or even stationary
:
(something)
=0
or
Id (something)
0.
For the integrand is not a variation, in the sense of the Calculus of Variations; nor are the forces of the problem varied; they are considered only along the natural path of the system.* The Principle expresses a necessary condition for the motion of the
system.
sufficient,
In the non-holonomic case, the condition cannot be since the first-order differential equations have not
We
turn
now
ton's Integral or an analogous integral does and in fact, in a restricted region, a minimum.
7.
Hamilton's Principle.
t(u, c)
If
we
set
t
U,
8t
^U^
:
t lt
then
s
I.
becomes
(1)
We
is
not to be
varied.
* In its leading ideas this treatment was given by Holder, Gottinger Naehrickten, 1896, p. 122. Unfortunately Holder felt impelled to defer to the primitive view of " variations as infinitely small quantities" in the sense of little zeros, i.e. infinitely small constants or functions of Xi, y lt z, t. In the foot-notes on pp. 130, 131 the in fact, vitiates "neglect of infinitesimals of higher order" renders obscure the treatment, so far as clean-cut definitions go. The writer cannot but feel that the inner Holder would have preferred such a treatment as that of the text, but that he did not have the courage to break with the unsound traditions of the little zeros, for fear of losing his clientele.
371
Suppose that a force function U exists, which depends only on the Xij 2/, z and t, or on the q r and t
:
(2)
U=U
t
(x if y<, z iy
f)
or
U = U (q
r , t).
Since
5,
is
we have
W
*i
= 6U
and
(1)
becomes
(3)
C(ST
0.
It
bZi
t
must be remembered, however, that the variations &r,, fly,-, or 5q r satisfy the condition of vanishing when t = t and when
t
1
.
first
example of an
integral,
(4)
to
(5)
f(T
h
U)
dt
0.
we may
a system The
kinetic energy of
U=
U(q r
t)
exist.
(6)
is stationary:
l(T
U)dt,
(7)
d I
(T
U)
dt
0.
Here
is the
when
tQ
and when
372
MECHANICS
is
A
by
necessary and sufficient condition that (7) be true Euler's Equations, 1, which here become
:
afforded
d dT
_
__
dT ~ dU
dq r
______
'
rp
, 1
'
i*jj
'
dtdq r
dq r
But these are precisely Lagrange's Equations for the system. Incidentally we have a new proof of Lagrange's Equations, in case we make Hamilton's Principle our point of departure. have proved the Principle for systems of particles with
We
it
general cases, e.g. for systems of rigid bodies; provided each time that a force function exists. The case is also included, in which
relations of the
form
*>(?!>
exist
;
.
'
'
'
>
?,
0,
1,
Ml
case
is
The most general cf Bolza, Variationsrechnung, p. 554. that of a system having a Lagrangean Function, or kinetic
In the above cases,
potential, L.
L = T
+
it
U.
When
body
it is
motion
is
taken as
itself
of the system.
The
(8)
fldt
to
;
be stationary
or that
i
Cldt = V
0.
14 that Hamilton's Integral (8) is actually It will be shown in a minimum for a path lying within a suitably restricted region but the minimum property does not necessarily hold for un;
restricted paths.
8.
is
Lagrange's Principle of Least Action. Our point of departure the Fundamental Equation II., 5, in which U now does not
t
:
depend on
373
T =
Thus we have
(1)
<0
:
T(qi,
qm q iy
,
,?m).
ti
C(ST
5
SU)
dt
ZTddt =
0.
Here each
of
represents a variation in the sense of the Calculus t Variations, the independent functions being q l9 </ m
-
is
is
Nevertheless, the variations, q lt q m t, are Let us examine more minutely the meaning
integral.
some
1
m+
<
(u,),
t(u,0)
?r(l,
such that
q r (u,0)
=g
(u),
t(u).
r
Moreover
?r(0,)
<?r(0),
=g
(l)j
j(0, c )
<(0, 0)
=
t
<
const.
i(l, 0)
^.
But
in general
^(1, c)
^ v 5|-o =
Thus
0;
^|
tt=1
0.
In particular, then, the functions (2) may be restricted by any further conditions which are compatible merely with the general conditions of continuity. Such a condition is the one that, not = 0, but also for merely for the natural path corresponding to
all
varied paths
(3)
or,
T = U
more
explicitly
:
ft,
(3')
T
fe
[ ?r (u,
),
^|] = U
Since
,
,
[ ?r (u, c)]
A,
where
is
a constant.
in q lf
-
is
it is
clear that
r
1,
m, are chosen
arbi-
374
Let
be any function.
(4)
MECHANICS
By dF we
SF
shall
now mean
the following
= u
where q r (u,
e),
:
by the
relation (3),
i.e.
(3').
And
(5)
similarly
J\Fdu
=
-j-
iFdu
e-O
0*J
where the integrand on the left is formed for the arguments q r (u), etc., and the integrand on the right, for q r (u, c), etc.;
Equation
(6.)
(3') still
holding.
Thus
it
bT = 8U.
ones,
Although these definitions are in form identical with the earlier where the m + 1 functions (2) were arbitrary, they are in
m+
functions are
now
:
related
by
(3')-
Equation
(1)
now becomes, on
r,
(7)
CdTdt
Tddt =
0.
new
definition of
rn jj^' -f- 7 Ot,
i
5,
xT 7 'I tf 01
and
since
5t'
(8)
o
d(Tt')du
0.
Hence,
(9)
finally
=
lj*Tdt
0.
We are
LAGRANGE'S PRINCIPLE OF LEAST ACTION. Let a system of particles have the kinetic energy T and a force function U, where U
depends only on the position of the system, not on
its velocity
or the
375
suffi-
and where
is
independent of
t.
cient condition for the natural path of the system is, that
(10)
subject to the hypothesis that all varied paths fulfil the requirement
that
(11)
T = U
+ h.
t
The
auxiliary condition
CTM =
(12)
o,
in
that
theory*
outlined
employs
it
Briefly
is
as
Set
F = T+
where X
the
/*o\
is
\v,
a function of
equations
t,
and
let
by
m+
*fl?
d &F
< 13 >
f\
'
Wr'dtWr^
(12).
'-I.'".*.
(13) it follows
From Equation
X
d
^ + ^_^r^ +
X
tyr
dq r
dt
Ldq r
or
(12), give
=-J-.
* Cf. Bolza, Variationarechnung, p. 586, where the case is considered that there are, in addition, relations between the coordinates, not involving the time.
376
MECHANICS
resulting equations,
8T
dU
d dT
The latter are Lagrange's Equations. Incidentally we have a new deduction of them, based on Lagrange's Principle of Least
Action.
As in the case of Hamilton's Principle, so here we can give a direct proof of Lagrange's Principle of Least Action by means
of the Calculus of Variations.
Principle
is
For, as above pointed out, the equivalent to the Lagrangean problem represented
by
'
m
'
(12).
Recurring to the condition (3) we see that the functions q l (u, e), Qm(u, e) may be chosen arbitrarily, and the function t (w, e) then determined by (3'). If the function t(u, e) thus determined
9
(16)
fft'du,
t is
then
(17)
For
T=%A
r. s
.q r q.,
AT =
.
-
A. r
the coefficients
r8l
qm
Now,
(18)
/
to
/2r
'
Let
(19)
where, as usual,
r q'
= dq
(u)/du.
Then
or
(20)
From
(20)
and
Tt'
377
is
the simplest type of all, considered at simpler type the outset. It is the integral (21), formed for the functions q r (u), that is to be stationary, and these functions are all arbitrary.
(20), or
much
is
determined from
(22)
t-t, = f-j-JL
J Vu
This
is
in the next
Jacobi's Principle of Least Action, which we will treat paragraph as an independent Principle. But it is
interesting to see how it can be derived from the Fundamental Equation of 5, and proved as a particular case under Lagrange's
EXERCISE
Show that Equation thrown into the form
:
(1)
<i
dT
-x-
dU
o
d dT\
(It
17 TT-) &Qr dt
G(j[r'
,,
0.
O(l r
0q r
where the conditions imposed on the coordinates do not contain the Then a necessary and sufficient condition for the time explicitly.
natural path of the system
is,
(1)
fVU + hVTdt
be stationary:
(2)
378
MECHANICS
is given
The time
(3)
by the equation:
T = U
or
r
(4)
vir+T
=Vrdt.
can give a direct proof as follows.
:
where
(5)
We
(6)
The
the value
JVu + hVSdu,
,
where <h(u
stationary
(7)
:
q m (u)
i
It is to
be
VU + h VS du
hold, or
:
0.
(V(7
+ h V5)
T
(V[J
m.
+ h VS) =
0,
1,
Hence
.
/vTT+1
/
'
du\
Equations
the solution.
(9)
Equations
and
(5)
now determine
is
given
by
(4).
It follows
furthermore that
dq r
\du/
dq r
(4)
dq'r
du dq r
(9)
and
we
find
.__
at oq r
J4
O^,
O^.
oq r
oqr
379
Thus we arrive at Lagrange's Equations. If we assume them, then we have a proof of Jacobi's Principle. Conversely, if we assume Jacobi's Principle, we have a new proof of Lagrange's
Equations.
10.
and on Mechanics are not careful to say what they mean by it. In d'Alembert's Principle the dxt, 8y if dZi began life by being 3n arbitrary numbers. In their youth they were disciplined to conform to certain linear homogeneous equations. Thus still a number of them were arbitrary quantities; the rest had no choice, they were uniquely determined.
symbol
writers
Critique of the Methods. Retrospect and Prospect d is treacherous. It can and does mean many things,
The
and
Enter, the Calculus of Variations. And now the dxij dy^ dzi, 8t become the variations of functions of a parameter, or inde-
From now on
of
these
<$'s
must be
dealt with
at least,
tho findings of that branch of mathematics are to be adopted. The Future. As the physicist fares forth over the uncharted
ocean of his ever-expanding science, his compass is the Principles. He seeks an integral which in the new domain will do for him what
Hamilton's Principle achieved in classical mechanics. There is mysticism about this integral. Imagination must guide him, and ho will try many guesses. But he will not be helped by an
He must make a cloan-cut postulate defining the d. and then lay down a clean-cut definition of what he moans by tho variation. There is no short cut. A thoroughundefined
integral,
is
in art.
11. Applications. Lot a particle be acted on by a central attracting force inversely proportional to the square of the distance. Then
(1)
r =
it is
where the pole is at the centre of force, and motion takes place in a plane (cf. Exercise
integral
(2)
:
4,
below).
Then
the
f\
r*0'*du
380
MECHANICS
Set
F (r,
Then
B, r', 0')
>(? +
20
)(
'2
r2 *' 2 )'
dw
Since dF/dO
A^-^ = n
00'
.
0, it
follows that
(6)
W'-Vr + h
.
^F
- ^/M
r*0'
_ "
r*0'*
Vr'*
If c
0,
then
const.
line.
and the motion takes place in a right be taken as the variable of integration
*
:
u =
Hence
cdr
*/;rVhr 2
=
""
+
:
c2
of integration
1
Then
*=+/
_
1
/T
Al
~|
/it/*
22 1L
y)
EXERCISES
1.
0.
e,
K, y
in
terms
u was
.
(0, 1)
but
it
might equally
well
have
^b
HAMILTON'S PRINCIPLE.
3. 4.
LEAST ACTION
381
its
path
is
Suggestion
5.
Discuss the motion of a particle in vacuo under the force of Assume the path to lie in a plane. gravity.
6.
In Question
5,
lie
in a plane.
7. Explain the case of motion in a circle under the solution given in the text.
12.
Hamilton's Integral a
Minimum
in
a Restricted Region.*
THEOREM.
(1)
The
integral
fldt
to
is
minimum for
tQ
too far
apart.
?m, ft,
(3)
2
1,1
is
Moreover,
r,
(4)
H+
L =
J Pr q
T
where
(5)
Pr
=
J|
-,pm,
1,
m,
H(q
has the properties
:
ly
,?,?!,
* Oarathfodory has given a proof of this theorem Riemann-Weber, Partielle Differenlialgleichungen der mathematischen Physik, 8. cd. 1930, Vol. I, Chap. V.
:
382
MECHANICS
(1)
be a
minimum is,
that
8 I
fL
L dt =
0.
The extremals
are given
by
Kuler's equations
By
(7),
the transformation
(5),
Lagrange's equations
(8)
equations, Chap.
(9)
XI
dqr^M ~
'
dpr___M *
'
~ -i 1
;
...
'
dt latter
dp r
dt
'
'
dq r
cf.
The
/im
(10)
Chap.
XV and Appendix C
.
W + uf _
ff
(,
Q
,
dv ,...,,m ,_
,
...,_
:
dv
as follows.
Let
(q r
pr
t )
which Equations
(11)
(9) are to
V =
S(q l9
qm
!,,
f ^
)
t),
(12)
^(ll
*
' '
'
istics
in question follows at onoo from the theory of character^ That theory tells us thut there exists a solution as applied to Equation (10).
of (10)
V
such that, when
t
-Stai,
-,
Q m 0,
,
to,
function
<f>
^
,
(71,
Qm)
(qi,
qm ).
its first
Here, <p (q\, qm ) is any function which, together with continuous in the neighborhood of the point (qi, qm ).
, ,
derivatives,
is:
is
Such a function
<P (Ql,
'
'
'.
Qm)
= Sr7r,
r
where the oti, are arbitrary constants, or parameters. The function *S thus resulting is the function required in the text. is analytic in the point (q r If, as we may assume, the function // (q r p r p r Jo), and if, as is here the case, <f> (q r ) is analytic in the point (g r ), then the fundamental existence theorem of the classical Cauchy Problem, formulated for the simplest case, applies at once, and the theory of characteristics is not needed.
, , , , ,
383
(13)
pr
Wf
fir
8S -^,
,
= l,.-.,m
t
the first set, when (qv, p r J ) are given, by the ar = ar and then the second set determines /3 r By means of this function S Equations (9) are solved. The solution is contained in (13) and is obtained explicitly by solving
are satisfied
values
(13) for q r ,
.
pr
?r
r
/r(i,
grfai,
'
'
'
'
m, ft,
ftn, t)
j8m,
(
I
p =
OW, ft,
If
'
'
'
r/ r
</ r
(0 represents an
:
extremal, and
if
dq
/dt,
then by
(5)
and
(13)
05)
Moreover
(16)
:
^
2)
r
=
r
f,
l,
.-.,n.
'M' +
it
&
=
by the
aid of (13), that
For, since
(17)
is
a solution of (10),
t
follows,
H(q
l9
,?,?!,
,p*,0 =0.
On
in (4), and replacing p r in the substituting this value of resulting equation by its value from (13), Equation (16) results.
T, q'
qr
t).
V
,
= L(q
where (</ r </J, t) are 2n + 1 independent variables. Let (q r q r be an arbitrary point, and develop L' about this point by Taylor's Theorem with a Remainder. We have
,
(18)
= L
2)
r
L ir (q' r
q r)
E(q r
&g
r,
0,
(q ri q r ,
t),
and
E(q r
Qr,
<jr,
= i 2 liri
r,
, (q'r
qr
MHence
?.)
the coefficient I^ Q 8 being the value of L<j r ^ for a mean value of < 8 < 1. the arguments q r namely, q r r 0(q' q r ), where
r
,
384
(20)
if (q'i,
>
MECHANICS
0<E(q
m)
is
r ,q r ,t) ,q'
distinct
from
(ft,
gm ).
Proof of the
(O Q
Minimum Property.
P
=
(q r
(t),
t<>),
Qr
l
:
qr
1,
m.
Let P!
(q r
t\)
be a second point on
<~
near by.
Connect
C:
and
let q' r (t)
qr
qr
(t),
1,
m,
however, be a weak
dq r /dt.
The curve C
<>7,
shall,
variation,
l?r(0-tfr|
Let
L = L(q
r ,t). ,q'
be an arbitrary point on C. Through this point Let (<?r, there passes an m-parameter family of extremals, (13) or (14). a m retain the We select one of them as follows. Let a lt
,
but let ft, values they have for <~ ftn have new values, the second those of the namely, equations (13), when given by = = The i t. of q r will be given value r q corresponding q r)
;
,
a r $ r in question, by difthe first of the ferentiating equations (14) with respect to t. These values of r q n q r t satisfy Equations (15) and (16); the p r do
by
(7).
not enter explicitly in these equations, and so the fact that they depend on t does not complicate the equations.
We now
(21)
apply Equation
(18),
L = L(q
first
r,
qr
t)
% L^
r
r (q rj q ry t)(q'
q r)
+ E(q
r , q'r,
q r 0,
The
follows.
First,
(22)
^^
:
S,
r (q r , f)q'
+S
(q r ,
t).
2s
r
W"
9r
+ S, ($
- L (g
r , (/
0.
385
we have
r (q r t)(q'
,
^jj^
Finally, since
2) S,
qj
+ L(q
r,
q r>
f).
from
(15)
the
first
of (21)
t)/dt,
and
(24)
(21)
L =
+ E(q
r,
n q'
qr
f).
We now
serve that
to
Ob-
to
CL
<u,
taken along the natural path of the system. tion (16) says that
r>
Q ~
and so
ti
t.
/Lett
= 5fa ,0
r
to
But
in the
:
g r (0-
We
final
result
(25)
to
CL dt
= /L dt
to
+
o
CE(q r $,
,
qr
t) dt.
If,
then,
0,
differs
from
cF
E >
and so the
integral of
left) will
integral on the
be greater than the integral of L over (" right) and our theorem is proved.
the
first
386
MECHANICS
Minimum
in
a Restricted Region.
In
(1)
the functions
T and
:
'U
do not contain
explicitly,
and
T is homo-
geneous in the q r
(2)
T=
varied functions, q r (u, ), are arbitrary, subject merely to in each end-point, t the condition that dq r It is t t^ obvious that the integral (1) lias the same value as the integral
,
The
rt
(3)
JTdt,
:
T = U
+ h.
This condition shall hold for the varied paths, too. Thus q r (u, c) but t(u, c) is determined by (4). To prove, is still arbitrary; then, that the integral (1) is a minimum for the natural path, it
is
sufficient to
if
(4)
(5) (6)
L = T
U. U.
H
(4)
= T -
From
(7)
and
(5),
L = 2T (n Q
h.
Let
(8)
in
12,
and
let
C'
7r(w, 0>
?=<(",),
be a varied path.
From
(7)
(9)
/2? dl = fl dt
h(t l
* ),
387
The right-hand side of Equation (9) can be computed as follows. From the analysis used in 12, Equation (24), we see that
Li'
du
Hence
Ldt
"'
S(q r
I)
+ CEdl
<b
Since q r value
:
q r for
u = w
MJ,
the
first
StorSZD-StorVo)Hence
*
*i
(10)
to
/2T df
is
= S fer
1
,
?i)
- S far
t
o)
+h
(t,
+ CE dl
to
Since //
independent of
it
follows as in Chap.
XIV,
l9
4,
that a function
of the
form
S =-
ht
+ W(q
:
-,?*,
A,
2,
-,
a*)
Using
this
(11)
<0
ArdT =
I^to,
- Fto
then
+ Csdl
<0
If
we allow
J?
0,
and
J*2T<lt
= W(q
- W(q
r ).
388
MECHANICS
(11)
Thus
becomes:
C2Tdt = C2Tdt
to
t
be positive for some parts of the interval of integration, and hence the integral (3), extended over C', will exceed in value the same integral extended over the
J5,
For,
will
if
C"
is
distinct
from
co'o,
then
natural path, as was to be proved. The case U = const, leads to the geodesies on a manifold for which the differential of arc is given by the equation
:
ds 2
5} A rs dq dq
t
r,
Thiis
we have a proof
is
above
that a geodesic on a manifold obtained ar the shortest line connecting two points which are not too
far apart.
CHAPTER XIV
CONTACT TRANSFORMATIONS
1. Purpose of the Chapter.* The final problem before us the integration of Hamilton's Canonical Equations
:
is
..
A)
dq r
= 8H
Wr'
df
dp r dT
8H
-W,'
p ry
t)
r=l '-' m
,
The method
r, (q'
and important
into
class of
new
variables
t'),
system of
A/ , A)
or, as
form
W=W
r dq'
3H'
dpr
dr
~W
3H'
- 1 '-'
t
'
we
which Hamilton's
The most
we
shall con-
Canonical Transformations.
one-to-one
transformation
'
'
'
'
'
'
01,
tfm,
Pi,
Pm,
is
said to be canonical
' ' ,
if
there exist
Pm,
two functions,
-
H(<li,
7m, Pi,
and H'
(q(,
m , p(, q'
Pm,
t')
paragraph is designed to give an outline of the treatment contained in the following chapter. The student should read it carefully, not, however, expecting to comprehend its full meaning, but rather regarding it as a guide, to which, in his study of the detailed developments, he will turn back time and again for purposes of orientation.
389
390
MECHANICS
(1)
/(2 V'M
1"
H'df)
J(2 Prdqr
1'
Hdt),
an arbitrary closed curve of the (2m l)-dimensional ail(l V' * s its image in the transformed (#J, pj, <')space, these spaces being thought of as simply connected. To a canonical transformation there corresponds a function such that Pm, <7m, Pi,
is
where F
(gv, ?>r,
0- s P a(
'
>
'
'
'
(2)
2 M*H
?'
'
'
dt
^ Prdq
- Hdt
And
conversely, when three functions H', H, ^ exist, for which the latter relation is true, the transformation is canonical.
An important sub-set of these caContact Transformations. nonical transformations consists in those for which the last Equation
(3)
I. is
t.
On
we
(2)
'
Since
/'
=
it
/,
the variable
I.
and
treat
as a parameter.
<?'
Equations
'
'
'
q'rfal,
'
iQmiPi,
,
'
' ,
Pm,
Pm,
t)
Pr
Pr(q\y
'
'
Qm, Pi,
' ,
t)
with
^(<7?>
'
'
'
<?m,
,
P'
'
' 9
} ,
Pm) Pm)
_^
'
'
'
d(9l9
(7m, Pi,
modification
(q rj
of
Equation
,
(2).
,
t),
of
the form
(5)
r %(p'dq'
r
p r dq r ) =d*.
in (2)
Of course, d$ has
different
meanings
and
(5).
In
(2),
^\
(6)
^T,
l
CONTACT TRANSFORMATIONS
since here the independent variables are q r
/*\ (7)
,
391
(5),
pr
t,
whereas in
*
JT
,
.
since here the independent variables are g r p r ; a similar remark This is not an exception, r applying to the other differentials, dq' or contradiction, in principle, but only in practice, since the
differential of
pendent variables
__
<Nf
and
it is
i.e.
defined
our function that we can speak of its differential. A transformation we will henceforth change the notation
from
,
m to n
Qr
= =
'
'
' ,
'
'
'
?'(<?!,
9n, Pi,
' ' '
Pn)
Pn)
(
I
Pr
Pifei,
'
'
'
tfn,
Pi,
3 (ft',
'
'
'
,
'
'
'
,
0i>
Pl'
'
Pn)
#ft>
'
'
'
'
'
0n, Pi,
such that
(9)
=
Jpfdtf
r'
jprdq
r
r,
where F
is an arbitrary closed curve of the (QV, p r )-space, thought of as simply connected, and F' is. the curve into which it is transformed, shall be called a contact transformation. There cor-
q nj
'
Pi>
>
(10)
-prdq
r)
d*.
is
And
fies (9)
true satis-
contact transformation may, of course, depend on certain thus becoming functions of these parar and parameters, p', q'
meters as well.
The transformation
II.
above
is
mation
may
itself
392
MECHANICS
II. is
transformations
I.
The approach
is
through the
Integral Invariants of Poincarg. The contact transformations, as defined generally by (8) and (9), are of especial importance in Mechanics because any such
transformation carries an arbitrary system A) of Hamiltonian cf. infra, 4. Equations over into a second such system, A') We shall treat the application of these transformations to the
;
integration of Hamilton's Equations at length in Chapter XV. If the student is willing to take this one property of contact
and he
thajt
2.
transformations for granted, he can turn at once to Chapter XV, will find no other assumptions needed in the study of
chapter. Integral Invariants.
(1)
j*L(q
to
r ,q'
{)dt,
and the extremals, which are the path curves, given by Lagrange's
Equations
/o\ (2)
:
**L
^_n
Wr~
'
:
diWr
r-V-.,n,
1
where
The
(3)
L is the Lagrangean function, or the kinetic general solution can be written in the form
qr
,
potential.
q r (t
qf,
qJ, qf,
tf n
),
i.e.
1,
n,
where g r
qr
fo.
qr
their
values
when
*
'
t
*
=
9
In the (2n
*
(<7u
q*>
'
' >
<ii>
:
q*9
further equations
(4)
<?r
<?r(';
1,
,?,
:
?1,
,<7n)>
or more properly, a 2n-parameter family represent a curve C, Let a closed curve, F of curves C.
(5)
?r
?r(X),
tf r
g r (X),
.
l,
n,
X,,
The curves C which pass through be drawn in the plane t = the points of F form a tube of solutions, which we will denote by S.
CONTACT TRANSFORMATIONS
which form
393
Let the action integral, (1), be extended along the curves Its value is a function of X S.
:
(6)
(3)
(5).
Differentiate
/ (X)
FIG. 147
- C "
On
integrating
by
d dq r
di~d\'
we have
C'^^LM 3\ J
d</r
d(j r
Hence
t\
'(\\
- C ()
~
V (^ J $ \Wr
J
d 3L\ fyr,, CU
dtWr'
8\
+
|
dL8q r
-f 2q r d\
The
integral vanishes,
because q r
is
by hypothesis a solution
3
:
of (2).
We now make
(7)
dL
Thus
(8)
Since F
is
= ^r(
and
(9)
(6) gives
Hence
0,
394
MECHANICS
(8)
:
and so from
*i
P/y
<l
eJX
0.
Let
as variable
is
also arbitrary,
This equation represents the theorem in which the whole inIn substance it can vestigation of this paragraph culminates. be stated as follows. may regard Equation (7), along with
We
the
qr
t,
as repre(q r ,
senting a transformation of the (q r q r , )-space on the Observe that the Jacobian space.
p rj 0"
?j
Chapter XI,
3.
into curves C" of the second space, and TO, S being transformed into a tube S'.
F goes over
into a curve
the second space the space of the variour point of departure and, dropping the primes, consider a closed curve in the plane t = t Q of that space
Let us
(<?r,
now make
,
ables
pr
t)
Consider furthermore curves C through its points, which are obtained by transforming the curves C of the earlier space. The integrals (11) now become line integrals in the present space.
If
we change
(13)
=
:
r,
Pr
Pr,
/5J Prdq
r
=
*J
^0rda
r
r,
where F
is
the curve of intersection of the arbitrary plane / = t S determined by T an arbitrary closed curve of
,
CONTACT TRANSFORMATIONS
the plane
t
395
a contact
tQ .
t
transformation,
,
.
But
ft,
,
Pr(i,
3.
' i
n , ft,
fti,
Equations.
of the Theorem, a) Hamilton's Canonical 2 form a system of n Lagrange's Equations (2), simultaneous total differential equations of the second order.
Consequences
By means
a simultaneous system of 2n total differential equations of the first order in the (q r p rj 0-space. Let these be written in the
,
form
(16)
= Qrfop,0,
Pr(?,P,0,
is
independent of
t
t,
the deriva-
must vanish.
Hence
'
dX dij 2,P'^
or
^-\
dq r j\
_ -
(\ u '
dx ?1T a\ + pr aTax)
Integrate
'
by
parts
d\
Since
n pr
is
a closed curve,
0,
*The geometric picture is here slightly different from the earlier one, since the variables (a ri /3 r) and (q rt p r) are interpreted in different planes. But of course one may think of a cylinder on T as directrix, with its elements parallel to the On cutting this cylinder with the plane t = t, we have a curve F lying in t-axis. the same plane with F. Or, to look at the situation from another angle, t is only a parameter, and it is the spaces of (a r Pr) and (q r p r) which concern us.
,
,
396
MECHANICS
:
and we have
Here,
dq r
8t
__
dqr
dt
_ n ~ Wr>
tyr
dt
_ p __ dp " ~
r
dt
Thus Equation
(17)
may
(Prd
-Q
r )
0.
But r may be any closed curve of the plane t = /, since to any such curve in that plane corresponds a F in the plane t = t It follows, then, that we can define a function // by moans of
.
the integral
(19)
:
,
f
'
-Prdqr
a n b lf &, ) of the plane t = t (a lt connected with the variable point (q ly q ni Pi, pn of this same plane by a curve lying in the plane. Because of (18) the value of the integral does not depend on the path,
and thus
value of
t.
//
is
t),
for dofinitcness,
lie
new
will
Then H becomes a function of (q rj p r 0is ) replaced by a different point (a", /3", t Q ), the differ from the old H by an additive term which is a
but not of
(q r
,
function of
t,
p r ).
More
(20)
by the equation
= li+f(i),
77 just defined,
where
f(t)
is
and
From
follows that
dn
CONTACT TRANSFORMATIONS
Thus the system
, ,
397
:
dg r
(22)
_SH
dp r
8H
r=l,.
..,n.
Hf-Wr'
~dt~~Wr'
grows naturally out of Equation (14) ; for (18) is but another form of (14), and (18) at once suggests the definition of // by
(J
9)
and
(20).
Thus
if
the transit
by the
can be
theorem of
this paragraph.
The Function
and
:
H.
Equation
(14)
f
To
,
.
lr-
Prdctr)
=0,
where g r p r are given by (15), the curve F being as before any closed curve in the plane t = t It follows, then, that the integral
:
(<*,$)
(23)
in the
(a,
/?),
is
independent
t
plane of the
joining the
path
:
and
thus
entering as a parameter
(24)
f
('>)
g (Prdqr '
Prdctr)
= V (<*,
/
ft t).
dq r \
'~
__
:
dV
where the
(15).
Trans-
/o c x (25)
rj
r
dv _
398
MECHANICS
integral^
The
on the
,
left is precisely
f).
(19), or
H(q r p r
Hence
(26)
H= H
is
2 PrQr ~ ^,
where
the other hand, the Lagrangean Function L (q r q r connected with (q r p r t) by the relation (cf Chapter XI,
On
t)
is
:
3)
(27)
L
it
+H =
-
]T p r q r
r
Hence
appears that
<*>
Just as
so
is
was defined only save as to an additive function of t, can be modified by adding any function t and the same true of L. But it is convenient to restrict these additive func-
and
From
clusion,
general con-
it
THEOREM
Let
r
be
l,---, n,
differential equations t
or n ,
and
let
<p r (t
!,--,
ft,
j9)
ar
to
t
tQ
pr
t).
Let
l)-dimensional space of the variof the (2n S be a tube consisting of the curves ii) which
and
let
be the section of
by the plane
-.
//
\
5) Pr dq r
CONTACT TRANSFORMATIONS
is
399
an
i.e.
if
iv)
^Prdq
'
=
r
^LfPrdotr,
dq r -/*_
dt
"'
= dH
'
dp rr =
.
dH
>
r T
8p r
dt
lf
8q r
y),
ti >n
'
Conversely, if Equations i)
Hi) will be
an
integral
then
Observe, however, that Theorem I. is more general than its origin from the action integral (1) and the transformation (7) would indicate. It applies to any functions Q r P r for which
,
Hi)
an integral invariant or, in the converse, to any function H, provided that the determinant
is
;
Hu ".H nn *
0,
i3
02 TI
-
But a system of Equations v) may conceivably not lead to a mechanical problem why should it ?
L can
THEOREM
a)
II.
Let
qr
9r(a D
hr(ot lt
'
'
'
'
'
<*n, ft,
' '
ft, t)
fti,
'
'
'
' ,
n, ft,
where r
1,
n,
Pi,
0^,
a?id
f
-,.,
ft,
--^ftr
,
'
ar
^r
=
=
gfrC^,
an an
ft,
'
ft,
ft,
< )
A r (a!,
the
ft,
O
the
fee
(q r ,
2n-dimensional
(<x r ,
p r ) -space on
400
MECHANICS
a)
;
i.e.
defined by a).
//
a) is
a contact transformation;
i.e. if
^Prdq -
j ^Prdar,
or
(p r dq r 2J r
Prdoir)
= dV (a,
/3,
/),
then b)
C'
z's
a Hamiltonian System:
^
'
='"~
'
r==
~dt
r ~dp~
~dt
>'"> n
>
~dq r
and
4.
conversely.
formations.
Transformation of Hamilton's Equations by Contact TransIf we start out with a given system of Hamiltonian
:
Equations
,
,
()
dqr
~dt
_ dH ~
'
dpr
~dt
"
:
dH_
'
1 r-I,---,n,
__
r 8p~
r ~dq~
W
/0 ,
f
1
9r
,
= =
/rfe,
,
'
'
'
'
^n, Pi,
'
Pn,
Pr
Vr(q\y
'
'
'
'
'
A
,
?n, Pi,
Pn,
(3)
1,
n, will
(1)
i.e.
they
will
where /f
'
= H' (q'
T,
r p'
t) is
some function
of the
arguments
i.e.
T , p' r q'
sufficient condition that (3) be Hamiltonian, (4), is that (2) be a contact transformation.
of the
form
The proof is based on Theorem II., 3 and the fact that the contact transformations form a group. Let (2), then, be a contact transformation. Denote it by T. Let (aj, #) be the initial val-
CONTACT TRANSFORMATIONS
ues of
*
401
T,
(?', p'r)
J
for
tQ.
They
it.
arise
from (a r
r)
by
formed
for
*o
^o> let
us write
(a r
,
Thus, symbolically,
or
:
(*;,
#) = T
r ),
r,
r)
= T Q-*(cl,
ft).
Again,
we may
write symbolically
(#, PJ)
= T
(g r ,
p r ).
(a r
Finally, consider the solution of (1), whereby the space of the is carried over into the space of the (g r p r ). This , p r)
,
transformation
is
3,
it
and
by
D
On
r,
r)
(q r ,
p r ).
by
the differential equations (3) is to carry the space of the (a' Ty ft) over into the space of the (q' Denote it by A r p r ).
,
:
And now we
I TQ on the
this transformation
A
:
can
be obtained as follows
(aj,
Perform
first
/3 r
)-space
0J = 57
(;,
#)
Next, perform the contact transformation space, thus obtaining the (q r p r )-space
: ,
on the (a r
fi r )-
on the
latter space,
r , p'J (q'
fer',
Pr)
= T
A
(q r ,
p r)
We
have in
this
:
way
obtained
transformations
A = TDT?.
Hence A
is is itself
(3)
Hamiltonian, by Theorem II., 3. This is the result on which the developments of Chapter depend. It may be stated as follows.
*
XV
The
is
obviously
itself
a contact transfor-
mation.
402
MECHANICS
// a system of Hamiltonian Equations (1) be transa contact transformation (2), the result is a Hamiltonian formed by is sufficient, but not necessary. The condition system (4).
THEOREM.
Computation of H'. The original system of Hamiltonian Equations (1) leads to the contact transformation Z), for which
the relation
(5)
:
^p
r
dq r
2 ^ dar = dV (a
r
ri
r , t),
is
characteristic,
where
JT 77
(6)
= "V
/n
Prtfr
__ ~
dt
The transformed Hamiltonian Equations (4) lead likewise to a contact transformation D' = A, for which the relation
(7)
2 M - 2 # da
P'
r r
'
= dV> ('.
'>
is
characteristic,
where
j jr/
Let
Then
(10)
2 # da
r
'
"
S^
r
dc* r
= dW (<*"
&r
>
*o)
will
be the characteristic relation corresponding to T of the differentials on the right is taken on the supposition that t is a parameter, and so a constant. Moreover, (q r> p r)
.
Each
by equations
3.
From
we
infer that
d[-
W(ct r
ftr, < )
V(a r
0r,
CONTACT TRANSFORMATIONS
Hence
403
~
dt dt
, ,
dt'
(a r $, f) and W(q r p r 0, which arc determined only provided save as to additive functions of t, are chosen properly. From
,
(6)
and
(8)
we now
infer,
by means
of (12), that
(13)
H' =
H-
(p'^
p r q r).
Each
we have developed it will be convenient to denote the Thus a formed variables by Q rj P r instead of by q' r n p'
formation
:
trans-
Qr
r
/rfoi,
'
<7, Pi,
p,
Pn,
P =
<7r(<7i,
Qn, Piy
'
' ,
where
' '
"
(jft,
l)'*'?
* '
n)
'
is
a contact transformation
if
(3)
2 (P
r
,
dQ r -
Pr .rf7r)
^^
(<7r,
Pr,
0,
is regarded as a parameter and the differentials are taken with respect to (q r p r ) as the independent variables. If such a transformation be applied to the Hamiltonian system
where
W
(4\
dqr__<M
dt
'
^Pr__^ff
dt
'
-_!... n lj
' :
'
dp r
dq r
W
where H' (Q r
(13),
(6)
4,
,
dQ,_ff' ~
<tt
dPr__W ~
rf<
_j
,
...
' '
_
'
P,'
3Q,'
r)
is
connected with
H (q
-
r,
pr
t)
by Equation
or:
H' =
H-
(P r Q r
p r q r ).
404
MECHANICS
(q r ,
The
r , t)
when
(q r ,
(3) hold.
p are chosen as the independent variables, (2) and It may be possible to choose the 2n variables (q rj Q r )
r)
meter
(7)
in (3).
Write
W(q
(3)
,pr,t)
= W'(q
,Qr,t).
Thus
(8)
becomes
(PrdQr
PrdQr)
dW (q
r,
r , t)
On
dQ r
dq, in (8)
we
find:
Pr =
(9)
Equation
(6)
Since
dW
dt
dW
dt
'
dW'dQr
SW'dq r
8W
'
'dQ r dt
8q r
dt
dt
we have
Hence
(10)
(6)
becomes
ff
'
ff
OO
OO
The Transformation:
pr
-^-,
P =
r
Oq r
^T' Glflr
We
down a particular contact transformation, in which be taken as the independent variables. Let
' ,
(q r , Qr)
qn
!,
a ny
t)
be a function of the 2n
CONTACT TRANSFORMATIONS
Set a r
/irk\
405
(12)
Pr
= &S Wr
Pr=~ Wr
r>
&S
= l,...,n.
i
The
last
(2)
of the (q r ,
n of these equations can be solved for the Q r in terms p r) because of (11), and then the P r are given by the n equations. Thus a transformation (1) results, the Jacobian
first
not vanishing.*
will
The transformation
(PrdQ r
and we
becomes
(13)
Pr dq r )
=~
(||
S, since Tf
dQ r
dq r
=~
)
d3,
may
:
set
W=
and hence
t.
H' =
such a function
H+
How
XV.
S can be
shown
in
Chapter
is
given by (12).
only to set
the
*
For, Equations (9) must be true, and it remains r of (9) must It is seen at once that the
W.
would be a
relation
between
r.
The
proof
is
as follows.
If
-,
Vr
=
=
d
fr (Xl,
Xn),
1,
U,
(l/Ii
',
2/n),
=
,
1,
n,
where fr
<f> r
are
all
first
derivatives, then
-,
-
yn)
d
.
(x\,
xn )
-,*) d(yi,--,yn) Consequently neither Jacobian can vanish. In the present case, the q r Pr can be expressed in terms of the Q r P r since the value of the determinant (11) is unchanged if the q r a r are interchanged.
d(xi,
,
406
MECHANICS
EXERCISES
OO
^Cf
,
1.
The Transformation:
pr = T
,
oq r
Q r = ^pOf r
Study the
analogous case, in which (q r P r ) can be taken as the independent variables, t being, as usual, a parameter. Show that, if be chosen as before, and if we set Qn> <*i> <*n, (#u
' ' * *
>
>
P =
r
/-.
a r then
,
^\
(14)
will
p,
v*S
Qr=
s^
O&
l,..-,n,
Observe that
(3)
can be
d(P r Q r ) = PrdQr
so that
it
+
:
QrdPr,
rdPr
Choose
prdq r )
,
= d(~ W"
I\,
t),
+
-so
Qr).
W(q p
r,
r,
= W"(q r
therefore,
that
S=- W"
Compute dW"/dt and show by
(15)
ff'
can be expressed.
Show
that
06)
where q r
,
-,
Q and
r,
(TT*,
0-
3.
The Transformation : q r
P =
r
OC|
Gpr
if
-^r-
#V r
If (p r , Qr)
can
we
set
S =
W+
Pr?r,
CONTACT TRANSFORMATIONS
where q r
,
407
t),
W, and S
are
now
:
functions of (p r
Qr
then
the
(17)
^=
(6) yields
:
'
d P r'
and
(18)
H' =
if
r
H,
Conversely,
if
S(q lt
QL r ,
we
6.
set
Q =
The fi-Relations. There is one case of importance still be considered, namely, that in which is a function of be cannot chosen as the independbut the (q r Qr, t), (q rj Q r t) ent variables. The extreme case would be that in which
to
Qr
Wr
n,
Tho
tions
m^
n independent more
:
rela-
of the
matrix
X^
(2)
Xli
Wn
is
m.
Thus
m Q
f\
fc
's
be
QD
ff
Qm
/")
___
f\
/\
__
/yyj
Among
TTJ,
the 2n
,
r ) it
shall
to choose
*
' ' >
m
<7i,
variables,
,
Qn,
^n,
independent
408
variables.
MECHANICS
But the function W(q r p r
,
t),
when expressed
ir ly
,
in
ir
W(q
(3),
f,
Pr ,t)
IF*
for,
Or,
fl.
(1).
Equation
(5)
now
2) ( Pr dQ r
will rewrite it in the
p r cfy r) =
:
dW* (q Q
r,
r,
0.
We
form
I (^-CK -?(*
It is not,
differentials vanish.
the m equations (1) the first m differentials be eliminated, the resulting equation being can dQm dQu of the form
By means
,
:
(7)
+ XndQn +
Y.dq,
+Y
n dq n
0.
The
infer that
0,
. ,
Xn
0,
Y,
0,
Yn =
0.
The actual elimination can be conveniently performed by means of Lagrange's multipliers. From Equations (1) we infer
that
n
(8)
Multiply the fc-th of these equations by X& and add to (6). Then determine the X/t's so that the coefficients of dQ dQ m vanish.
} ,
The
resulting equation
is
of the
form
vanish automatically.
'
We
thus arrive
_ ~ =
Wr + Wr
'
(9)
r
...
l
CONTACT TRANSFORMATIONS
The
first
409
the \ k 's. The remainder shown above. The result is symmetric and holds, no matter what set of m Q k 's is determined by (1) i.e. no matter what ?n-rowed determinant out of the matrix (2) is different
are satisfied as
from
0.
It is
now easy
to determine
by means
of (13),
On
that
replacing
r,
p r here by
their values
from
(9)
and observing
dW* =
dt
dtt r
dQ r
dt
dQr
d{
dt
>
""
dttr
dQr
dt
dt
2? 3Q r
^
.
^ d&r ^ dq
^r
dt
^ ^r dt
,
^'
we
-J
do not contain
t
explicitly, this
equa-
(11)
H' =
CHAPTER XV
SOLUTION OF HAMILTON'S EQUATIONS
1.
Its
Treatment.
We
have considered a
great variety of problems in mechanics, the solution of which depends, or can be made to depend, on Hamilton's Canonical
Equations
dqr
(i)
- dH
, ,
dpr
~dt
-~
m
Wr
r-1
..'
~dt~~Wr'
'
'
is a function of (q r p r The object of this chapter is ). to solve these equations explicitly in the important cases which arise in practice.
where
The method
is
that of transformation.
:
By means
,
of a suitably
chosen transformation
= F Pr = G
Qr
(q 19
(q ly
' -
q n Pi,
,
P,
Pn,
q n , Pi,
'
' ,
same type
'
~
dt
d/V
dt
,
but more easily solved. Here, H' is a function of (Q T P r , <)> 11(>t in general equal to The determination of a convenient transformation (2) depends
on a
(A\ (4)
due
to Jaeobi *
dV --
_i_
It is not the theory of this equation, however, but the practice, that concerns us, for all we need is a single explicit solution,
(5)
V = V(q
.
lf
g,
i,
n,
0,
depending in a suitable manner on n arbitrary constants, or Such a solution is found in practice by parameters, !,-, n
means
set forth is
due to Jaeobi.
411
differentiation
,
function (5) once found, the further work consists merely and the solution of equations defining the
implicitly.
to
qr
pr
Two
the
namely
a) Reduction
(5)
Equilibrium Problem.
0.
Equations
(3)
can
now be
inte-
grated at sight
where a r
ues in
($ r
(2),
On
PI,
qr
/r(Q!,
-,,
,P,0
qr
(8)
=
=
/r(i,
'
'
'
'
'
n, ft,
fti,
Pr
0r(i,
'
'
'
<*, ft,
'
'
'
ft,
shown
in
2, is
^
.
Pr
_ dV
~Wr
pr _
dV
~~Wr'
,
t r-l,..-,^
where
is
Qr
Thus the
solution (8)
is
pr
= dV
Wr'
&T
=~^
h.
dV
r=l,..-,n,
(5).
b) Constant Energy,
H (q
l9
r,
r)
The second
:
case
is
that
which
,g,
Pi,
,pn).
(2) in
a transformation
which
r,
G>
do not depend on
(10)
Q = F
r
(q lf
'
g w p lf
,
p n)
'
'
^n, Pi,
'
412
such that the new
MECHANICS
will
rj
but not on
rj
t.
In particular,
#' = P^
:
Equations
(3)
now
=
Thus*
Qi
r
0,
r=l,...,n.
+
,
/3 }
Q, =ft,
r
5
n.
2,
P =
ar
1,
(12)
Then the
solution of (1)
is
(13)
1
ft, ft,
The transformation
given by
(14)
be shown in
4, is
the equations
HA\
pfSS
dW Wr
n Qr =
:
where
is
or:
^'-'^
iff
dW
2,
Tf
= W(q
l9
,?n, A,
,)
"
*
' t
:
Here TF depends on the arbitrary constant A, and also, in a suitable 1 further constants, or parameters, a 2 <** manner, on n These are set equal respectively to the P r
Pi
*
P,
= a
a,
2,
n.
is
The change
of notation
made
for the
413
combined with
(11),
thus yield
(15)
,w
last
The
;
in terms of q ly as will be
of these equations can be solved for q^ qn shown in 4, thus giving the form of the
,
first
equation (15) in
We have characterized this case by the caption: "Constant Energy," but this is not a physical hypothesis. Our hypothesis does not depend explicitly on 2, and this is all we need is, that for the mathematical development. then turns out to That be constant along the curves of the natural path, is an important consequence; but our treatment does not depend on this
hypothesis.
b),
A
in
The transformations used in a) and Contact Transformations. namely, (9) and (14), are examples of contact transformations. transformation (2) with non-vanishing Jacobian was defined
Chapter XIV,
1,
to be a contact transformation
if
(16)
(PrdQr 2} r
Prdq r )
dW (q
r,
pr
t),
where the
differentials are taken with respect to the (q r p r ) as the independent variables, t being regarded as a parameter. Such a transformation always carries a Hamiltonian System (1) into
,
a Hamiltonian System
(3).
(9)
and
(14) satisfy the condition (16) is seen at once in (16), observing in the case of (14) that
by substituting
d(PrQr) = PrdQr
This
is all
QrdP r
the theory the student need know from Chapter XIV, on the study of the present chapter, and this amount to enter 1-4 of that chapter. of theory was all developed in
We
have seen
in
414
MECHANICS
S = 8(q
-, fr,
where
if
-,
is
and where
Qr
is
set
= ar
(2)
/r-ff + f.
:
if
The transformed function H' can be made to vanish identically we can find a solution V of the partial differential equation
an
V =
and
is
V(qlJ
?,
!,-,
a n 0>
,
such that
^.......vj^
3(ll
>)
On
setting
formation
(1),
equal to this function F, and making the transH' as now determined vanishes identically. Thus
:
the transformation
(5)
p,
is
g,
by Q r
in
Pr
jfc
r=l,...,n,
where a r
replaced
System
t-oThe
1.
T'-,
'- 1 '-'"'
solution of these equations is the system of equations (6), These are the values of Q r r to be substituted in the
transformation (1)
(7)
i.e.
p,
last
g,
0,=~g,
give the
r-l,...
The
of (4),
n equations
pr
415
ar
of course, a further requirement in the large, namely, $r can be so determined as to correspond to the initial
t
conditions
Pr
=
r
tQ,
qr
-
=
,
qr
gn
Q
,
p r = Pr.
= V
!,,
,
,
Q,
,
=
,
1,
n,
,
must admit a solution, a r = a r and V (ft, q n a l9 must satisfy all the conditions of continuity, notably neighborhood of the point (q r a r ) = (<7r, a r ).
e* n ,
(4), in the
EXERCISE
tion studied in
dS
*'~W
Here,
n Qr =
dS
"
>
= V--,".
1
Let
V =
V(QI,
qn ,
a lf
an
then,
we
:
'
will
vanish
~W
If
dQ r =
>
dP =
r
n
'
~di:
l,...,n.
we
Q =r
Pr
P =
r
Qf r ,
1,
tt,
we
same
solution
of
Equations as before
3.
orgy
Example. Simple Harmonic Motion. Here the kinetic enT and tho work function U are expressible respectively in the form
:
(1)
T=
^q\
L =
V --
\q*,
0<X.
Thus
(2)
T+V-?-f
(3)
416
MECHANICS
(4)
ff-rt-L.-Lp. + lj..
:
dq
di
- p
dp
~m'
Tt~
--
\n A9
'
We
for determining V,
2.
The equation
We
(6)
which satisfies this equation.* One such function Let us see if we cannot find one in the form
:
is
enough.
(8)
F =
12
fl
W,
and
where
12 (f)
is
a function of
alone,
W
o.
=
:
(q) is
a func-
tion of q alone.
If this
+
X
V=
:
d!2
The
on
t
Hence each
is
a constant
denote
it
by a
it
is
obvious that a
* Let the student disembarasa himself of any fears due to his ignorance of tho theory of partial differential equations. No such theory is needed in the kind of application in Physics which we are about to consider it would not even be helpful in practice. The single function V(q, a) is obtained by a simple device fully explained in the text. There is, of course, a most intimate relation between the theory of Hamilton's Equations arid the theory of this partial differential equation, as is indicated, for " example, by the theory of characteristics" cf. Appendix C. The point is, that For the this theory is not employed in such applications as those illustrated here. is all that is required, latter purpose, a single solution V(q\, q nt ai, and such a solution is obtained by ingenious devices of a homely kind, as set forth
;
; ,
, ,
in this Chapter.
417
J_/dTF\
2==a
The
first
equation gives
Q =
ai
no constant of integration being added because we need only a From the second particular integral, and so choose the simplest.
equation,
2
-j (dW\
= 2ma
is
:
m\q 2
One
W=
Thus
(10)
V2ma -
m\q*dq.
V=-o* +
/
(7),
Equations
2 here become
dq
(11)
__^_ fl_-^--<m
u
^_
Xg2
This
last
equation gives
and thus
(12)
q
the
first
From
(13)
Equation
(11),
p = V2m^ cos
\-
0).
Equations (12) and (13) constitute a solution of Hamilton's Equations, which, however, is at present restricted for we have not paid heed to Condition (7) on the one hand or, on the other,
;
418
MECHANICS
considered that the second equation (11) is restricted. Here then is a difficulty.* Either we must follow the theory as hitherto and V ; then t is condeveloped, using single-valued functions
fined
Or
else
we must
introduce
multiple-valued functions F, and then we must go back and revise and supplement the general theory. a way out, whereby we There is, however, a third choice
can remain within the restrictions of the present theory. According to that theory the solution given by (12), (13) is valid so long
as
Equawhole
yield
tions a) admit a solution single-valued and analytic for the Equations (12), (13) range of values
oo<2<+oo.
EXERCISES
1.
of
(14)
p =
by choosing
as
V2ma sin
:
^~
(t
W the function
W=
o
2.
V2m<x
2 m\q dq
+ C(a),
of integration C(a).
* There is also a further difficulty, since the first equation (11) may not admit a solution (suppose p < 0), but this difficulty can be met by choosing the negative
radical,
V;2ma
419
nit
-fi,
A.
Then
,
ml m T =
-
2
.
9 2
Tr = U
mgl cos q
Obtain the equation for motion near the point of stable equilibrium:
f
where
dq
t is restricted. Hence discuss the two cases a) oscillatory motion (libration) b) quasi-periodic motion, when the pendulum describes continually complete circles (limitation).
:
Observe that, when t passes beyond the restricted interval, the sign of the radical changes, and q changes from increasing to decreasing, or vice versa.
4.
Freely
Falling
Here, q shall
Body, or vertical motion under gravity. be measured downward from the initial position.
mq,
p*-
420
MECHANICS
dW = V2ma 2m +
Since
gq.
_ dV
dq
dW
dq'
option as to which radical shall be taken. If the body projected upward, q will bo negative for a while, and so we must choose the negative radical for this stage of the motion. At the
is
we have no
V/dqda does not exist. have now a new problem, as the body descends. The choice of must be made on the basis of the positive radical. Nevertheless, both stages of the motion are covered by the soluturning point, (7)
is
not
fulfilled, since d
We
ft
ft
o\ 2
2^y
/9i
"%/_____ // ___
p = mg (t
/3)
V2om.
Why?
4.
We
,
Reduction to the Form, H' = Pi. have seen in Chap. XIV, 5, Ex. 1, that if S be an arbitrary
H, Independent
,
of
/.
function of the q r
of
a r and
where
of
*'*'
'"'*S *
and
fn\ (2)
if
o,
we
set
ar =
r,
dS pr =
^,
5,
Qr-
W
-
dS
.,
l,...,n,
go over into
(5),
and
(3)
H' =
H+f
t
If
(4)
on
t,
H' =
//.
t
:
H
-
is
also independent of
,
= H(q
lf
qn
p lt
p n ).
421
,".,Q*,P ,---,P n ).
l
will
depend
H' = H'(Plt
,Pn\
of the
(5)
and, in fact, that //' will be an arbitrarily preassigned function P r Begin with the case
.
H'(P19
find such a function
:
Pn)
,
=P
qn
,
To
,R
v
S(qlt
(xlt
a n ), consider
the equation
(6)
Suppose
it is
W
depending on n course on h, which
= W(q
1
is
ly
g n A,
,
2,
,)
2,
,
arbitrary constants
also arbitrary
an
and
of
such that*
(7)
It
then follows, as
we
will
show
later,
that
2,
This
(9)
is
we
will
choose as
,
S(q
l ,
qn
a,,
a) = TT^,
set
:
tfn,
A,
2,
where a t
(10)
=
Pj
A.
If
now we
ai
P.
a.,
2,
:
n,
then
(6)
becomes, because of
//(ft,
(4) gives
:
(2), (9),
Pi,
and
,
(10)
(H)
and hence
as
<?n,
Pn)
PI,
#' =
was
desired.
* In practice this
desired, obtained
PI,
is done by writing down an explicit function by such artifices as the separation of variables.
of the nature
422
MECHANICS
:
-dt^ dt
(12)
r
1'
dQ.
dt
0, 2,
o,
1,
,n.
The
is
obviously
Ci
(13)
r
<
+ ft,
,
Q>
r
P;
,
s
n.
2,
-,;
which now
P =
:
ar
1,
(2),
pr
:
{T
dP r
'
n,
we liave
(15)
I of these equations can be solved for q2 The last n qn as functions of q 1 because of (7), thus determining the form of the curves of the natural path of the system. And then the This last statefirst equation can be solved for q l in terms of t.
, ,
ment
is
conveniently
substantiated
,
indirectly.
q n in
terms of
1
last
so the earlier solution of these equations for q2 of q l become identities in t when q r is replaced
by
q r (t) given
by
using
all
n equations.
(8).
Proof of Relation
in the h,
a a as
a n we
,
find
+W^ +
(16)
423
of these equations
0, it
is
were
Xn
,
=
(17)
=
are true,
pleasure.
and
by
hypothesis,
\ may
be chosen at
Now
multiply the
The
coefficient of
each
Hp
side reduces to 0.
trary.
true,
But the right-hand side is X u which is arbiThis contradiction arises from supposing that (8) is not
and the proof is complete. The Equation of Energy. When the kinetic energy T and the work function U are both independent of t, H is also independent of t, and H represents the total energy (sum of the kinetic energy T and the potential energy Hence* // is constant and we U).
may
write
= H(q
is.
i9
qn
p l9
p).
be derived from the physics of not helpful in the For we are dealing with contact transformapresent theory. tions which reduce Hamilton's equations to a desired form, and
this equation appears to
Thus
the problem.
It
But
this derivation is
Equation
transformation possible. Nevertheless, the physics of the situation throws a side light on the situation, which it is well to note.
in
The Symmetric Form. We have set, unsymmetrically, h = Pl Equations (10). We might equally well replace (10) by the
:
equations
(100
*(Pi,
,
Pn)
A,
P*
.,
2,
n,
where <i> (a lt a n ) is any function such that cfa/d^ 5^ 0. The above reasoning, with an obvious modification in detail, shows that the determinant
:
*That
IT
'
424
MECHANICS
d(W ~^
q
,
-
W
,
Q )
(8')
3(alf
<*2 ,
-^ a)
,
0,
= W(qly where q* h, a lt from (6), and h = $(!, 2 tion (14) is justified and Equations
, 9
n ) is
determined as before
dQ r
dt
(120
-
1,
n.
dt
The
First,
is
P =
r
a rt
1,
Next,
1,
,
= Ur t +
Pr,
=
r
H,
-
where
o> r
$r(<*i,
a n ),
1,
n,
and the
f} r
are
arbitrary constants.
Thus we have,
finally
(19)
ft,
a wholly symmetric solution of Hamilton's Equations. If we should wish to use a function $(0^, a n ), for which
,
some other derivative, as d$/da z is ^ 0, then we should need a solution W(q lt a n ) such that q n a lt
, ,
Examples. Projectile in vacuo. Let a particle of mass be acted on solely by gravity, and let it be launched so that it will rise for a time. Let qlt q2 </ 3 be its Cartesian coordinates,
5.
,
Then
I/
T=
& +
2
ft
2 </s
),
425
Cpl2
+ p + P^
*
:
The equation
for
W becomes
'
0(a, a,)
f
by
setting
W.
= W,
+W +
2
3,
where
(q
r)
is
a function of q r only.
i
j
Thus
n 0.
Vd^/
Since
it is
"j
Vrf?2 /
Vrf
W that
4,
is
(8),
suffice
^J
Herc, h
is
= 2m(/l-a
2
2
-a
2
3
)-
determined by the
2>
A,
and
<*3
are
initial conditions from the equation anv ^ wo parameters such that initially
c*2
2m (h
2
3
2
)
+ 2m
A /O
gq l
>
0.
We now may
JTTf
*
choose
-v/O
TTZ
O Q
where
is
tion of inequality.
merely to the
rela-
lf
it is
a,*
a,*)
_
Pl
dW
l
J
426
MECHANICS
1
and p
<
in the stage
0i
we
are considering.
We may take
2m
W
where
= - CV2m(h Cl
2 2
a 3 2)
c x is the initial
Q\
value of q r
Thus,
finally,
W =c,
A/2m(A -
<*2
The
condition
are
(7),
4, is satisfied.
We
now
in a position to write
problem.
It is given
by Equations
ff
=^=- m Jf
dh
_
V2m(h :
down
4
:
(15),
^i
2 2
2 3 )
'
2m*gq l
= A = |^ 8.
2ma
Jf
The first of the equations in each of these sets of three is in substance identical with the one which governs the vertical
motion of a
falling
body,
h
3,
<* 2
2
Exercise 4, where
now
;
a
and hence
:
2
,
]8
=-
ft
2 3 )
(7
initially,
and so
ct
mnst be
:
set
0.
The
last
two equations
427
2, 3.
The method we have employed gives the solution of the probno longer for when it is lem so long as the body is rising dWl /dq1 = dW/dqt cannot becomes and descending, p positive, be expressed by the negative radical. This second stage of the motion, in which the body is falling, could be dealt with by applyin particular, ing the method afresh with suitable modifications for the radical dW this But by taking positive t /dq^ step can be eliminated if we observe that the equations we are integrating, Hamilton's Equations, here become
;
di
~m Pr
>
1,2,3;
dt
di
2,3.
The
functions of
solution of these equations is unique, and is expressed by Hence the t which are analytic for all values of L
analytic continuation of the restricted solution found above gives the general solution, and the formulas found for q r p r are true
,
generally.
EXERCISES
1.
law of nature.
Let q l
r,
q2
<p.
Then
= R
2m\dr
428
MECHANICS
n =2mA + (W)
,
.
/dR\*
2wX
-^
a2
Thus
W=
the
first
J
r*
2mh
+ <*>, -^dr T
where either the plus sign or the minus sign holds throughout
stage.
Hence
r
=+ ~ m f
dr
2mh Vo
z,
2mX
H
,
02
=
2mh
r r
Discuss the case that the radicand vanishes for two distinct
positive values of r, expressing r as a periodic function of evaluate the integral that expresses t cf. 9.
;
^>,
and
2.
in space.
2/
Let q
=
l
r,
q%
= =
0,
qz
;
<p
r cos 6 sin
^?,
r sin
~2m
TT
fl
+ *;
2m\
J VO
Complete the solution and discuss the cases that the radicands have distinct roots.
429
4 when n
:
1.
Show
that
is
and integrating
ef "
^
dh
W
Then
I
ff(q,h)dq.
y
Thus
_8W
dP
"-
8q-
6. Comparison of the Two Methods. We have studied two methods of solving Hamilton's Equations, a) Reduction to the b), when // does not depend on t, ReducEquilibrium Problem
;
tion to the
Form, //' = P v The first method, being general, must apply to the second case. It does. Let us treat this case by the first method, as set forth
2.
in the Exercise of
(1)
>
>
n) is
8.
the function of
4,
P P
lt
The transformation
r
of
that Exercise,
(2)
pr an
r)V
?)V
Qr
yields
'
(Ti
-dT-'
^
dt
n ~'
r r
. .
lj
'
430
MECHANICS
in the
...
Departing from the notation of the Exercise, write their integrals form
:
I
1 I
Qr
P,
= =
Pr,
(4)
h,
P.
= !,--,; = a., s = 2,
r
:
,n.
is
The solution of the original Hamiltonian Equations by substituting these values in (2)
now
given
= 8V
(5)
*
2,
n.
But
3V_
^
= 3W ~
'
W =_
dh
3W
dh
'
3V ~ =
da,
da,'
dq r
dq r
(5) agree not only in substance, but even in 4. The one save for form, exception, with Equations (15), in from with to differentiation h the equation arising respect
Hence Equations
Here
it is
7.
Cyclic Coordinates.
t,
= H(q p ---,p n ).
l9 lt
and hence
(3)
?>
f-f-0.
=
,
-.-.,
,
2,
n.
It is
not
difficult to
ton's Equations
(4)
and the
h
= H(q p
19
,p n );
431
We
/*\ (5)
desire a discussion
:
not the form of solution in which we are interested. by the methods of 4 ; in particular, by
the transformation
dW
where
(6)
is
= W( qi1
:
?,
h,
an )
l,
,...,
\
J,
,,,...,
-
and
P =
l
A,
P,
= a
a,
2,
n.
To
Let
'(9)
we turn
to the
Method
of Separation
in the past.
of Variables,
w=W
r
+ -- + Wn
(5)
where
r (<?r)
,
is
,
a function of q r alone
.
and
of
the
parameters, A,
a2
From
a"
and
2,
-
(3)
we
,
see that
n,
and so we try
TT,
g.,
=
:
2,
n.
W, = ^(fc^afc-'-.aO
(7)
=t>.
Then
(11)
If
becomes
fl (g 1
,^,a
,---,n)=A.
we assume
that
fiff
(12)
l^ffp.feuPi."*'--.*..)* ^Pl
:
'
432
for p,
(14)
:
MECHANICS
p,
:
=*(?
=
A,
2,
',),
-,).
we have
(15)
-^
choose as
: i
*(?i, A, a,,
Now
(16)
J *(<?i> A,
C
a2
a.) dqlt
where
c is
a numerical constant.
We
(17)
otf,
+a
n qn is satisfied,
"
dh'
(13).
'
dq.dh
dh'
'
where p 1
is
determined by
On
M!!
dp, dh
We
=
the solution of which
ft
x
0,
r-l,-..,n,
is
+ fc,
A,
Q.
A,
a,
= =
2,
2,
..-,n;
-
P =
These values
for
P,
r,
r
= a
n.
Q P
and the
result-
pr
433
aw
dh
dh
dh
l9
(18)
Pi
8W = r =*Wi,
,
h,
a2
a n),
n.
p9 =
-7T--
dW =
a,,
2,
The equations
of ft:
(W
/'
The
first
~^~dqly
8
2,
'--,n.
equation gives
ft
as a function of
t.
EXERCISE
Obtain the
8.
1
Continuation.
v
depend
on
<
<n
arguments
(1)
#(ft,
is
-,??!,
similar,
,pn).
The method
of treatment
in general be obtained
by quadratures.
s
of
now become
(2)
p,
a.,
I,
n.
:
By
(3) (4)
analogy we
now
seek to determine
W in the form
+ an q
2,
,
W = v + a* +l qr+1 +
v
n,
*(?
,<?, h,
),
(5)
a^T
434
MECHANICS
(7),
7,
Equation
/c \
for
W now becomes
,
(b)
L n
rj
//^ft,
q,, TT-,
Tjjp
a^+i,
This
is
variables q r
As
is sought, and such a solution may notably the method of separation of variables. A function v once found, the solution proceeds as before.
(7)
/Si
qi,
1,
n.
From
terms of
the equations of the second line g* can be found in = 2, p. From the first equation ft is now ft, k
,
found in terms of
I
t.
Finally,
qi
is
line,
=
9.
1,
n.
of
Examples. The Two-Body Problem. Consider the motion two bodies (particles) that attract each other according to the law of nature and are acted on by no other forces. Their
centre of gravity travels in a right line with constant velocity, will assume the latter or else remains permanently at rest.
We
as
case.
Then each
of the bodies
moves
if
attracted
by a
force
at 0, the centre of gravity, which is inversely proportional to the square of the distance of the body from 0.
We
Let
g2
later, in space.
the
particle
be
to
polar
coordinates,
ft
r,
(p.
Then
"
2\dt 2
Hence
dt 2
Let
W=v+aq
2
2.
435
by the equation
1
f/^\
/ l
2
,
2
2 2
(7j
2m \\dq
or
1 J
^
ql
_
'
rt
2mA
2mX
r
<* 2
2
,
12
dr. '
where a
definite
first
stage of
the motion.
now
dv
/*
Sv
or
,=/c
dr
2mh
2m\
ft
The
is
evaluation of the second complicated eliminations. r as a can be function of <p. But trigonometric integral expressed the discussion of r in terms of t is more complicated cf below
;
.
To
treat the
motion
sin
^>,
in three dimensions
z
r cos
cos
<p t
cos
r sin
0.
Then
Let
Pi
0,
Then
Ps
436
MECHANICS
H
Since
2m
( vI \
+ Si
2
tf,
P
2
?,
cos 2 ?2 /
ql
= H(qv
q^,
p 1( p a p 3 ), we
,
see that
p3
a3
(const.)-
Thus
W = v + ci^s,
where
v is
4.
q,
cos 2
i
q.2 J
x
ql
t-hcro
arc only
two independent
variables,
:
q^
and
6.
0.
cos
On
setting
v
= R
:
Hence
r*(~^ +
2mhr*
2m\r =
ft
=
y
rv^ _ a
2
.,
where the signs are determined for a particular stage of the motion, and c, T are arbitrary numerical constants. Adding the further term a 3 g 3 we have:
,
W
We
by
(7),
ag^,
= R
+ 0.
are thus led to the solution of the problem in the form given
8:
437
+A
-.A
J
c
.
*
r
&=(2)
dr
r2
9^/0
\2rnh *
d6
T,
-\
2raX
r
a,
r2
v a,
a32
sec 2
a32
sec 2
The discussion of this solution on the hand of the explicit evaluation of the integrals and the inverse functions thus arising presents practical difficulties. The problem is of so great importance in Astronomy that it has been treated at length by Charlier,
Mcchanik des Himmels, Vol.
Failure of the Method.
I,
Chap.
4,
p.
167.
On
p.
171,
in
Consider, for example, motion in a plane. Suppose the body is projected from a point A, distant a from the centre of force, 0, at right angles to the line OA and with a velocity
v
(}
breaks down.
such that
It
will
or (2)
then describe a circle, r = a. But the Equations can obviously never yield this solution. Why?
function v was determined from the equation
:
(1)
The
= 2mA
In the present case,
h
2raX
=2a'
mav
and hence
0.
438
MECHANICS
no reason why the method on which it depends do not should apply, since the hypotheses
is
not
fulfilled,
and
hold.
10.
Continuation.
The Top.
We
VI,
By
T =
Let
tfl
Q>
<?2
<P>
Then, since
= dT
we have
:
Pi
Ad,
ps =
C<p cos
+ A
(
sin 2
+
and
C cos 2
<?'s,
Thus
becomes
Furthermore,*
U=
Thus
Mgb
cos
0.
Hence
it
First, then,
* It is necessary to change from the earlier notation h for the distance from the peg to the centre of gravity, since h plays so important a rdle in the present Let the distance be denoted by 6. theory.
Irl
dv*
2 Li 55?
sin 2 ft
^2 ^2 =
2 = r+ V(2^ A La2
t/
c
where is an arbitrary numerical constant, not a parameter, and the sign of the radical must be chosen with respect to the Moreover, for special stage of the motion under consideration.
brevity,
_
we have
1
9
.
439
+ C^ + l(
,
l/a2 cosq
a 3\ 2 1
sin gl
) J
,_ +M * 6 COS * .
(24 A
- La - N cos ft)
2 2
sin 2 q l
cos q1
2
3)
,
cos
?,
^
?1>
L = 4, v/
The
form:
= 2AMgb.
7,
now
takes on the
+*-*
Thus
4-
C'~e/~H
sin ft
v (2Ah
2 2
~
=
A^ cos
gj
sin 2 gj
cos
gj
a s) 2
Let
cos
#,.
Then
this equation
becomes
9l
= r
e/
c
^
+V/<W
cose,
where
This
is
the same
18.
Chap. VI,
result obtained by elementary methods, But compare the technique. With only Euler's
440
MECHANICS
Dynamical and Geometrical Equations to work with,* eliminations had to be made by ingenious devices, whereas the present advanced methods free the treatment from all artifice. The fundamental equation in desired form is evolved naturally, directly, from the general theory, not untangled from a snarl of equations. Instead of having to solve three equations for 6, by more or less ingenious methods of elimination, the functions 7 [7, and
<j>,
\j/
7
,
hence
any
\f/
artifice
<p
and
immediately.
EXERCISE
Study the motion of a top with hemispherical peg, spinning and sliding on a smooth table. Show that
where
F(u)
(2h
^-
In the problem Variation of Constants. motion which the system would execute if only the major forces acted is regarded as fundamental, and then the variation from this motion due to the disturbing forces,
11.
Perturbations.
of perturbations the
is
mirrored mathematically
motion
~dt
~Wr
dt"
___
' '
-.-
dq r
''
and then setting the characteristic function H of the actual problem equal to the H of the problem due to the major forces, plus
a remainder,
(2)
* It
H
+
= H,
+H
is true that in the earlier treatment we had two integrals of the differenequations of motion to work with at the outset, namely the equation of energy, T = U h, and the equation arising from the fact that the vector moment of momentum <r is always horizontal. But even so there were three equations in 0, $,
tial
<f>
to integrate.
441
Transformation of the Major Problem to the Equilibrium Problem. First, the major problem, represented by Hamilton's Equations
in the
form
solved by reducing it, through a contact transformation, to the Equilibrium Problem. The contact transformation is given by the equations
is
:
fA\ (4)
Pr
r
where
(5)
is
F =
obtained
as
F(<7i,
q,
P,,
Pn,
t)
follows.
Let
V = Vfe,
-,?, a lf
",
an
0(a l9
/
-
v
,
an )
.
7~
VJ.
The resulting function is the In this function, replace a r by P r [In practise, the function F ^, q ny a lf (5). a n t) is obtained, not from an elaborate theory of partial differfunction
,
ential equations,
but by means of simple devices, ad hoc.] (4) be written in the explicit form
,.
= Fr(p P r = G (p
Qr
r
'
'
lt
Pn
'
* 9
qi,
'
qn,
'
lt
jpniQi,
"
,qn,
or
f
= pr =
9r
fr(P
g r (P ly
'
P, Q -^Qn, Pn Q On,
,
19
To say
is
442
MECHANICS
transfor:
form the solution of Equations (3) are subjected to the mation (4), the resulting Hamiltonian Equations become
(7)
f=
Qr =
,
0,
^=
r,
0,
r=l,..-,n.
:
The
(8)
Pr
1,
ft,
where a r Pr are constants. Now transform the variables Q r P r that are the solution of Equations (7), namely, the functions given by (8), back by means of the transformation (4"), and we have the solution of Equations (3) in the form
,
:
Qr
Pr
= =
/r(i,
0r(a lf
'
'
'
n, ft,
'
0n,
" -,,
;
ft,
',0,
and
(9),
(4
or
(4"),
identical
two
distinct things:
a) In the form (4") these equations represent the Contact Transformation (4). b) In the form (9) they represent the Solution of the Hamiltonian Equations of the Major Problem, or (3).
Transformation of the Actual Problem by the Same Contact Transnow proceed to apply the contact transformation formation. (4), not to the variables (q r p r ) which satisfy Equations (3),
We
(q r ,
p r)
which satisfy
Since this
(1) will
(W)
Here H'
(15):
(11)
r,
dQ,_8ir ~
dt
dP r
'
dPr__9ff' ~
dt
cf.
Wr
j.
'
'
= H'(Q P
r , <)
Chap. XIV,
5,
Ex.
1,
H' =
(6)
:
H+
But from
Hence
H' =
H-
443
from
(2) it follows
that
H
(10) take the
ff t
:
Thus Equations
form
dQr^dH,
The
result
dP r _
m,
When
the variables q rj
may
be stated as follows.
pr
which form the solution of the actual problem represented by Equations (1) are transformed by the contact transformation (4) or (4'), the transformed equations take the form (13), where is the given, l
now
Q P
r,
r,
t.
The Final Solution. It is now but a step to the solution of Equations (1), which represent the actual problem. Solve Equations (13), thus determining Q r P r as functions of t. Then transform these functions, the solution of (13), back by means
,
The latter satisfy of (4) or (4") to the variables q r p r tions (1). The result can be expressed in the form
,
.
Equa-
(14)
Pr =flTr(P,,
,Pn,Qi,
'
'
'
Qn,
where
Q P
r,
arc determined
by Equations
(13).
The method above set forth has been Variation of Constants. This expression is a mathecalled the "variation of constants."
matical pun. It is a pun on the letters a r r These, in Equations (9), are constants the equations there representing the soluOn the other hand, they can be tion of the major problem, (3).
,
.
r,
Qr of
being
determined by
(13),
tion of the actual problem, (1). can attain complete confusion of ideas, as is done in the literature, by changing the notation in (13) and (14) from
We
Q Pr
r,
to
r,
oL r .
(13) is replaced
into the
form of
(9),
and
da r _ ~ 3(- g,)
dt
df)r
'
_ ~
B(- H,)
da r
'
d0 r
dt
L >'"'
>
444
MECHANICS
= 1 (Q r P rj t) is now written as l (ft1) a1) t) the where 1 instead of ff,. Hamiltonian function now being 1 but it is a poor pun that has to be Thus the pun is explained
,
explained.
It is possible to treat Continuation. A Second Method. problem of perturbations in still another manner. Let a n ) be any given function whose first partial deriva<f>(a lf Let the Hamiltonian Equations for the undistives are not all 0. turbed motion, namely, (3), be transformed by a new contact 12.
the
transformation
where
/i/r\
S
/
is
defined as follows.
\
(16)
*(,, -..,.)
Let
,
qn
!,--, a n
t)
(17)
Pr
ll
^
Wr
,
,
Pr=~
r>
where
W
,
5 = 8(q
q^
Q,,
Qn, 0-
h
where h
is
OS
qi ,...,
dS
=H(
qn
,~,...,--,t)
:
+ -,
,
dS
S = S
such that
(<?,,
qn
'
h,
2,
On, Of
'
gt ,
',
Qn
)
(h, as,
-,
where
as,
On are arbitrary.
h
Substitute
<p
(ai,
an )
in
iS.
If d<p/dai
^ 0,
this will
445
these over
same type
dt
dP r
5
:
'
dt
8Q r
'
'
'
'
where,
(19)
by Chap. XIV,
H' =
:
H + ?jVI
(2)
the arguments
transformed by
now being the Q ry P r into which Thus Equations (18) take (17).
dQr "
0/7,
dP _ _
r
3/7,
3jp
"' n
Solve these equations and substitute the functions of t thus in (17). The functions q r p r of t r obtained, namely, the Q r obtained from these equations are the solution of the actual prob,
(20)
as follows.
He
writes
He
<
22 >
= P =
Qr
r
Ct r
\Cl r
'
'
,
where
C*r,
Dkr
are functions of
set
t
t,
vanishing
when
(for sim-
substituting these values for 0). plicity of like powers of X, the P in and coefficients r r Q (21) equating coefficients C* n Dk r can then be obtained by quadratures.
,
we have
On
The page
in
R.-W.
is
211.
APPENDIX A
VECTOR ANALYSIS
In Rational Mechanics only a slight knowledge of Vector Analysis is needed. It is important that this knowledge be based on a postulational treatment of vectors. The system of
Certain vectors is a set of elements, forming a logical class. functions of these elements are defined, whereby two elements
are transformed into a third element.
addition, multiplication by
real
enters as the independent variable), the inner product (scalar multiplication), and the outer product (vector multiplication). The functions obey certain functional, or formal, laws, which
happen to be a subset
+B=B+A
AB = BA
A(BC) = (AB)C
A (B +
(B
C)
= AB
C)A
+ AC = BA + CA
A
in
brief, systematic treatment such as is here required is given the Author's Advanced Cakulus, Chap. XIII. For a first
approach to the subject the Hamiltonian notation of S and V for the scalar and vector products has the great advantage of clearness in emphasizing the functional idea the concept transformation. On the other hand the notation pretty generally adopted at the present day is the designation of vectors by Clarendon or boldface, the scalar product being written as a b or ab (read a dot b), and the vector product as a X b (read a cross b). It
:
is
useful, therefore, to
have a syllabus of
447
definitions
and
essential
448
1.
APPENDIX A
Vectors and Their Addition.
By
vector is
meant a directed
segment, situated anywhere in space. Vectors will usually be denoted by boldface letters a, A, or by parentheses; thus
line
may
be written
(w).
and B, are defined as equal if they are parallel vectors, and have the same sense, and moreover are of equal length
:
Two
A =
By
denoted by
Addition.
B.
is
meant
its
length
it is
|,
or
by A.
By
the
sum
of
two vectors,
and B,
is
meant
their
A+B =
In
order
C.
definition
it
is
that
this
all
may
apply in
cases,
necessary to enlarge the system of vectors above defined by a nul vector, represented by the
symbol
If
0.
is
parallel to
A and
or
of the
same
length,
A+B =
any
0,
B =-
A.
B
FIG. 149
as long its sense being the same as that of A, or opposite, according as = 0, then raA is a nul vector If is positive or negative.
;
m is m times
also 6
,
m
:
OA =
0.
The
notation
Am
aA
a
6B + ro +j
means
+o
rr B.
A+B =B+A
(B
A+
Subtraction.
C)
(A
+ B) +
C
which added
By A
is
meant that
vector, X,
to
will give
B+ X =
A,
X =A-
B.
VECTOR ANALYSIS
To
obtain
449
geometrically, construct A and B with the same A B is the vector whose initial point is the point ; terminal point of B, and whose terminal point is the terminal
initial
then
point of
Fig. 149.
Cartesian Representation of a Vector. Let a system of Cartesian axes be chosen, and let i, j, k be three unit vectors lying along these axes. Let A be an arbitrary vector, whose components
A A A
lt
2J
A = AJ + A 2 j + A E =
then
B,i
A+B =
Also
:
(A,
BJi
A ^
Resultant.
resultant, F,
If
is
_
.
Then evidently
z
k.
+ B,j + B,k
(A,
B,)j
(A,
)k.
forces,
F2
Fn
:
sum
F = F
If
+F +
2
,
+ Fn
:
couples,
couple,
M,
is
M M M
M
2
= MJ
M +
+ Mn
Two
is
or
more vectors
if
there
is
line
in space to
any
vector.
arc said to be complanar if there is a plane in space to which they are all parallel. In particular, a nul vector is said to be parallel If three vectors, A, B, and C, are non-complanar, to any plane.
(i.e.
X =
where
Z,
ZA
+ mB + nC,
Acceleration.
Differentiation.
Osculating Plane.
:
A =
450
APPENDIX A
i,
where
/(Oi
j,
<p(t),
If are three fixed vectors mutually perpendicular. a deA will have have the vector derivatives, ^(0
rivative defined as
lim-rr
Its value is
:
Moreover,
If
m is
a function of x and
is
mA will
ax
if
+ m ~T' ax
.
of
a point
P move
in
:
any manner
where
/,
<p,
xi
yj
+ zk,
by
P is
represented
/i
<f>,
derivatives,
the
vector
acceleration of
P is given by P
(on the assumption that neither is a nul vector) is the osculating Thus the vector acceleration always lies in the osculating plane.
plane.
2.
The
by
of
two
vectors,
values
The scalar or inner product defined as the product of their absolute the cosine of the angle between them. It is denoted
and B, and
-
is
by
A-B
or
AB
is
read:
"AdotB."
| |
Thus
c.
A B = AB = A
cos
VECTOR ANALYSIS
If
451
is
is
defined
asO.
AB = BA A(B + C) = AB +
The associative law has no meaning. The scalar product vanishes when otherwise when and only when the
each other.
AC.
either factor
is
a nul vector ;
Furthermore
norm
:2
of the vector.
i
ii,
jk
;2
j
i,
k2 _ K
ij
i,
0,
ki
0,
0.
Cartesian
Form
=
Differentiation :
AB
If
is
a unit vector,
i.e. if
|
=
|
1,
then
aa'
a2
1,
and
0.
3. The Vector or Outer Product. Let two vectors, A and B, be drawn from the same initial point. Then they determine a plane, M, and a parallelogram in that plane. The vector
or outer product
of
is
is
length
to
equal to
It
and
sense
arbitrary.
is
AXB
erence
the
denoted by
AXB
and
is
read
"
:
cross B."
FIG. 150
452
If
APPENDIX A
one of these vectors
is 0,
or
is
if
defined as
Otherwise, let
Then
|
A XB = A
| |
sin
|
c.
in general, for
AXB=-BXA.
The associative law does not hold But the distributive law is true
:
e.g. (i
j)
5^
(j
j).
C)=AXB
and
as c^n be proved geometrically, or still more simply, analytically, by means of the Cartesian form ; cf infra.
.
It is
that
Xj =
k,
k =
i,
j.
In any case
A XA =
i
0,
and
so, in particular,
X =
i
0,
0,
k =
0.
Cartesian
Form
A XB =
J
-^-
**l
Differentiation:
-
dx
4.
dx
dx
General Properties. Let A, B, C be three non-complanar drawn from the same point. The volume of the parallelepiped determined by these vectors is numerically
vectors
(B
C).
VECTOR ANALYSIS
A
C be complanar
is
453
(B
C)
0.
Linear Velocity in Terms of Angular Velocity. Let space be rotating about an axis / with vector angular velocity (w). Then
the velocity v of an arbitrary
point
will
be
= () X
r,
Fia. 151
v =
x
y
zwy
Xtl)g
and
vx
Vy
= =
ywz
Z<l) X
If it
(a, 6, c),
then
(y
(z
c)
co y
6)
wz
vz
(y
6)
o) x
(x
a)
uy
In tho general case of motion of a rigid body (i.e. motion of rigid space), lot 0' (z y z ) be a point fixed in the body, and let (f, ??, f ) be the coordinates of any point P fixed in the body, the origin being at 0' but otherwise the (, rj, f )-axes may
:
move
where
in
any manner.
V
Then
V',
O'
FIG. 152
P
COf
7
C0
CO,
454
Localized Vectors.
initial
APPENDIX A
It is
point of a vector, or the line in which the vector shall lie, as in the case of a force acting on a particle, or a force acting on a rigid body. It is with reference to such vectors that the following definitions are framed. By the moment of a vector
with respect
r
to
a point
is
meant
the vector
M
the vector
F,
where r which F
or
is
lies.
F may
drawn from to any point of the line in In practice, F may be a force acting on a rigid body, be the vector momentum, mv, of a particle.
of a couple can be expressed as
TI
The moment
F!
+ r, X
F2
where F D F 2 are the forces of the couple and rt r 2 are vectors of space to any points P lt P 2 of the lines drawn from any point
,
is
meant the
= Ma,
M=a
(r
F),
where a
in
its
is
r is the vector
a unit vector having the direction and sense of L, and drawn from any point of L to any point of the line
lies.
which
Thus
if
is
let
nearest to point of application be transferred to the point is F the be of L to common the nearest i.e. and let ; L, point F. L line of action of of and the Decompose F perpendicular
OP
at
vector
is
The into a force parallel to L and one perpendicular to L. with respect to moment of the latter component at
Ma.
5.
Direction Cosines.
transforma-
tion from one set of Cartesian axes to a second having the same origin (both systems being right-handed, or both left-handed)
is
characterized
by the scheme
of direction cosines
X y
z n^
n%
7i 3
fc
HI
w2
VECTOR ANALYSIS
Between the nine direction cosines there
relations
:
455
exist the following
IS
+ mS +
n
mS + m + m =
2 2
2
3
1 1
n*
+
3
n2
+
2
a
n3
+ tn w + n n = + m^m + n n = -f w,m + n n =
2
l
==
n^j
JiW,
+ +
n2 Z2
Z2
+ m +
2
n3
J
i!
= =
n2
m =
l
n2 Z3
n 3 Z2
n2
Z3
Z^j
Z3
1.
APPENDIX B
^
(dij\ ~jj)
Differential equations of the
= /()
form
where
I.
f(u)
/(w)
is
= (t*-a)(6 -*) =
(w
or
II.'
a) (6
tO
and ^ (w)
^u ^
bj
CASE
I.
found by extracting
du
du
<
2>
'-Ajsr
^u ^
b.
Geometrically, the function on the right of (2) can be interpreted as the area under the curve,
(3) \ /
w
of the function
u a
FIG. 153
b
The graph
(4)
y
456
= V(u -
a) (6
A DIFFERENTIAL EQUATION
is
457
represented by Fig. 153. The reciprocal of an ordinate of this curve gives the corresponding ordinate of the graph of the function (3), Fig. 154
/r\
:
..
V(u-a)(b -u)t(u)
The area under the curve
(5),
shaded in
v
(6)
~~
du_
t
J V^-~Mb~^
and brings out
t
Thus
the
fact that
increases as
increases.
u
FIG. 154
r
(7)
du^
V(1T^~ci)(6~the graph of u, regarded as a functy
Then
tion of
(8)
FIG. 155
is
=
in Fig. 155.
Its slope is
as
shown
at
each extremity and positive in between. The definite integral, (2) or (6), has now served its purpose.
It has yielded for a restricted interval,
^
a particular solution of
(1).
A,
Continuation by Reflection. Reflect the graph of the function (8), Fig. 155, in the axis of ordinates, and let the curve thus
obtained define a continuation of the function ^(0 throughout A ^ t rg 0. Analytically the reflection is reprethe interval
sented by the transformation
:
Thus
= *(-
0,
-A
458
APPENDIX B
:
is
fdu\ =
(w)
Hence the function
<p(t)
A, A), or
u =
is
<*(0,
- A ^t ^
<p(f)
A,
a solution of
(1).
To complete
oo
the definition of
for all
values of
t,
i.e.
<
<
oo,
we could repeat
^ ut
**
*s
s^
m P^er
to
/i\
now be extended
-2A -A
FIG. 156
values
of
by the requirement
.
of peri-
odicity
(9)
<f>(t
+ 2A) =
<p(f),
oo
<
<
oo.
(1) in
u =
is
<p(t
+ 7),
Observe that
<f>(i).
where 7
an arbitrary constant.
*>('/
t\
(U) Hence
/i
o\
ft\
To an
for
arbitrary value u
of
u such that a
t
<
<
(
6 there
in the interval
A, A),
which
(13)
UQ =
^>(0,
namely
If
=
,
there
t
is
&,
then
only one value, namely, t = 0; and if A. But only one should be counted,
A DIFFERENTIAL EQUATION
an open
interval,
c
459
<t g
2A
or
<
+ 2A,
signs in
J
where c is arbitrary. Moreover, du/dt has opposite and t' Q because of (12).
,
We
ential
can
now prove
that there
is
differ-
tions
condi-
H!
6.
Suppose that
du/dt
tQ
it is
is
UQ =
and determine
^(a
:
*'(*o)
<0.
Finally, define
7 by the equation
*i
+7=
u
tQ,
7i
<o
'i-
Thus 7 =
(14)
is
7i
is
*(
+ 7i)
still
the solution
we
But
is
other solutions
which satisfy the same initial conditions? If a < u^ < 6, the answer is affirmative for values of t near ^ but for remote values, the question of singular solutions arises, to which we now turn.
;
Singular Solutions. The given differential equation admits, furthermore, singular solutions. The functions
u =
a,
6
:
each being considered in any interval for t, finite or infinite. Such a solution, moreover, may be combined with a solution (10) at
any
it
point.
The
solution
now may
may
These solutions do not, however, have any validity in the problems of mechanics, for which the above study has been made. The mechanical problems depend each time on differential equa-
460
tions of the second order,
APPENDIX B
and these have unique solutions, dependor initial on the boundary conditions. Equation (14) repreing But the converse is not sents an integral of these equations. of that (15) is an integral of the true, namely, every integral
that
second order equations why should it be ? We see, then, we may be on dangerous ground when we replace the latter
equations, in part, by the integral of energy, for example ; since the modified system may have solutions other than that of the given mechanical problem. Cf. the Author's Advanced Calculus, p. 349.
Does
this
remark not
call
treatment in Chap.
XV,
Not if we apply that method essentially the integral of energy? as set forth in the text. For in a suitably restricted region there is only one solution yielded by those methods, and we were careful
is
to point out that it is the analytical continuation of this solution that yields the solution of the mechanical problem beyond this
region.
Thus the
II.
CASE
(16)
is
This case
=
(J~J
with.
(u
a) (6
u)*t(u),
is
given
dM
f-(b _ a
*>(0,
===
<
<
The
(18)
inverse function,
u =
oo,
represents an integral of (16) in the interval indicated. now this solution can be completed by the definition
:
And
(19)
<p(u= b
t)
<p(t).
solution
u
oo
,(<),
t
<
<
oo.
It
is
now shown
as before that
is
A DIFFERENTIAL EQUATION
(21)
461
u = ?(
further case,
+ 7).
A
(22)
namely
can be treated in a similar manner; or, more simply, be thrown back on the case just considered by a linear transformation. Finally, the case (not mentioned above)
:
(23)
=+
(u
a) (6
t)vV(tt);
b)
f =-(u-a)(b-u)Vt(d.
t
Each
of these
is
solved at once
by a quadrature.
Case
is
I. There is another treatment of which brings out the important fact that the function <p(t) essentially a sine or cosine function
(24)
u = C
0,
cos
C',
:
where
is
=
j<,
form
:
and
where h (/)
is
2A
2A)
h(t).
may
The given
(25)
differential
equation
462
APPENDIX B
:
a U= 2u b-a
,
6
'
Make
(27)
the substitution
u =
cos
6,
<
<
IT.
(28)
This equation
(29)
is
(30)
=
solution of (30)
t.
The
is
A
(3D
"/: ^(cosfl)'
oo
<e <
oo.
Write
(32)
where
(33)
_J V^(cose)
d9
.
24.
Then
g(6)
is
For,
(34)
* In so doing
we
A DIFFERENTIAL EQUATION
But the value of the integral, because of the integrand, is 2A for all values of 0. Hence
(35)
463
periodicity of the
g(e
+ 2r)
=
Equation
(36)
^+
defines 6 as a single- valued function of t, since the integral (31) Let be written in the represents a monotonic function of 0.
form:
(37)
Then
(38)
h(t
2A) =
h(t).
in (36)
and
let
the correspond-
Let
27r,
and
let
t'
,,*('. + 20 +
By
virtue of (35),
+ ar)
or
t'
<
+ 24.
From
(37)
we now
infer
Hence
and the proof
If
is
complete.
we
multiply (36)
by v and
(37)
by
and add, we
find
(39)
0()+AA(0 =0.
464
APPENDIX B
are
restricted.
We
B
was
of
/.
In Equation (27)
(40)
This function
If
is
it
seen
by
we denote
by
<p(f) y
(41)
*>(*
7).
The other
If it is
same
result.
a question actually of computing h(t), then the integral from to TT, the reckoning can tabulated for values of be (31) methods for the ordinary evaluating definite being performed by
integrals
Simpson's Rule,
etc.
f(x
+ A) = /(*),
A
oo
<
<
oo,
where
is
Let
c =
Then
Jf(x)dx.
L J+ A
f(x) dx
C,
where x
is
arbitrary.
X+ A
For,
dx
//(*)
Let
f(x
+ A)- /(*)
-
0.
Then
For,
/V
X
f(x)dx
jx.
<p(x) is periodic
<p(x
+ A) =
<p(x).
x+A
v(x
+ A) -
*(x)
=
ff(x)dx
-(
+ A)+
A DIFFERENTIAL EQUATION
Hence
465
where X
= C/A. The
result
may
be stated as follows.
is the
THEOREM.
sum
of a
where
and
A
C=ff(x)dx,
Instead of the linear function \x
=
|write
we may
X(z
Xz
+7
or
x n ),
constant.
by an additive
In
if
and only
A
if
=
ff(x)dx
o
0.
APPENDIX C
CHARACTERISTICS OF JACOBI'S EQUATION
Although Jacobi's partial
differential equation of the first order
:
A)
AN
rn *>
where //
d(Jr
_ dH
,
dp r _
m
t),
~dt~Wr
= H(q
-
~dt~~Wr'
,
r-l,..-,m,
it
ly
qm
ply
pm
necessary to refer to the theory of characteristics, partly because we have sought certain explicit solutions by means of ingenious
devices (separation of variables, for example)
;
when we have needed an existence theorem, it was supplied once by reference to the Cauchy Problem. Nevertheless it
partly because, at
is
The
Analytic Theorem.
dif-
..
l' '
-
n>
>
'
dx
Let F(x ly
arguments
(n
for
x n z, 2/i, 2/n), together with its partial derivatwo orders, be continuous for those values of the x n z) is an interior point of an which (x l9
,
l)-dimensional
,
:
(x 19
xn
z),
* of the space of the variables region are and the yk wholly unrestricted. Use the
notation
/i\ (1)
y Xk
the Yk not
all
~^'
:
__
dF
7 Z
,
__
dF
b 19
-~te> an
,
v Yk
~^
,
3F
At a given point A
let *
(a,,
;
c,
b n)
=
0.
(a, c, b)
of
vanish
in particular, let
its
Yn ^
R shall
467
differential equations
II
dxk
Yk
dz
S y k Yk
dyk
Xk + yk Z*
=l
'
'
The
solution of II. shall go through the point (x, 2, y Q ), which neighborhood of A and moreover on the manifold
0,
F =
or
F (x
'2/^^=0
the (2, 2, y) Although there are 2n + 1 initial values there is only a 2n-parameter family of solutions of II., for we may without loss of generality set x n = a n once for all. The solution of II. can now be written in the form
:
T. ^t
f-(r i \^n
-r
,
-^i
-^n
1)
7 *
11
9
Ml
11
9 9
if* /9
= l,---,n-l;
(2)
(2)
xn
z,
y^
yn)
constant, since
dF =
%X
dx k
Zcfe
+5)
y*d*by
II.
it
On
(3)
is
appears that
dF =
0.
Hence
,x n ,z,y
l
F(x l9
,.-,y n )=C
which
,
F(x,
4-i,
On,
2,
2/i,
0,
C =
in
(3).
Thus
there
is
Consider
now
a (2n l)-parameter family of characteristic strips. the (n l)-dimensional space of the variables
(&!,,
(5)
x ny
z), in
which a solution
z
*(*!,
-,*),
468
APPENDIX C
=
:
plane x n equation
(6)
a n of this space
2
let
Cdfo
'
' ,
Xn-l),
where
is
x n -\), together with its first partial derivatives, o)(x lt a n ~0, continuous in the neighborhood of the point (a,,
, ,
and
Furthermore,
let
and
let
y n be given by
(4) or (4').
If,
as
set
Xn
the
first
Un
(2), combined with this last equation, will a surface represent (hyper-) parametrically, the equation of which w n ), can be thrown into the form (5) by eliminating the (u lf and this function (5) is a solution of the given partial differential
,
n equations
equation
solution
I.
Moreover
such that
it is
(5),
satisfies
continuity, can be obtained in this manner.* This is the general theorem of the solution of
characteristics.
by means
to
of
We
proceed
to
apply
the
result
Jacobi's
Equation A).
2.
Jacobi's Equation.
Xr
Let
q"
y'
(8)
(
1
z=F,
,
xn
see
= =
Pr >
*.
r ==
1,
m=
As regards y n we
tion:
(9)
from
I.
and A) that
-
it is
ss
H(x
II.
lt
xm *
,
Vi,
2/m,
xn )
+ yn =
0.
Equations
*For
Calculus, Chap.
now
cf.
the proof
XIV,
469
m
initial
m
:
dp r
dy n
dff
dt
d#
dq r
The
(H)
values are
Xr
/7
r </r
?/
3 >
Jf r
^o>
2/w
= 7i Pr = "" "
,
r r
>
=s
1,
*n 771
^0*
:
From
(12)
Hamilton's Equations
dq r
dt
'
dp r
(9),
dt
'
m.
dq r
(
\
dtf^m.
at
finally, since
fit
ot
QJ
and
from (12)
(14)
dV =
dt
Z PrQr is
the La-
H+L
and so
(15)
pr qr 5) r
~-L L
dV
results of the general theory and the of facts the case in Equations (12) (13), although obtaining and (14), we have not brought out the direct testimony of the
;
Let us turn back, then, to general theory in the present case. and or Condition It appears that the Equations (2) (4) (4').
solution of Equations (10) takes the form
:
n Qr
7)
__
Jr(J>
//
-<) Pi
>
Pr
(16)
~ Jn+r / (t \f
n Q.m
<7
' 0>
0)
Pm
470
APPENDIX C
=
1,
where r
(17)
m= =- H
,
and
,
(<?,,
qn
Pl
-,
pm
,
t )
In the case before us the functions frt /w + r r = 1, m, arising as they do from the solution of Equations (12), do not Q depend on F and y n is given by (17). Thus we have
,
,
(18)
V
2/n
'
'
and
(19)
by
,
(9)
=-
//(ft,
g,
.
p,,
is
?m, 0:
But
V in
(16)
does depend on
It
given by (14)
(30)
or
t
V =
to
So much, then,
II., i.e. (10).
Equations
I., i.e.
solution of Equation
A)
(21)
+
subject to the conditions under (6)
:
we choose w
(22)
F =
set
n
cofoo,
qm
and
'
have seen in Chap. XV, 3. Application. 2, that ton's Equations can be solved by a contact transformation
/f\n\
We
Hamil:
(23)
Pr
= d$
w
Pr
n Pr=~
W
1,
d$
= l,--,m,
-i
which transforms the given dynamical problem into the Equilibrium Problem, the solution of which is
Qr = ar
r,
m,
471
far
dQ r _ Q
"rfT"
are concerned.
'
~3T~'
-
d^-n
r-1 m r-l,---,m,
Qm , , qm , Q if possible to solve the
,
The demands
fulfil
must be alf
equations
(24)
br
=S
qr
(a lt
Om,
ct
for the
ar
(25)
oj
aj
am
= am
where a r
of
</r,
#r
6r
= pr
-
r.
1?
am
shall
be continuous,
dS
'
dS
dctr
d*S
'
dq r
in the
dq r doL 8
,
t Q ),
and
!,
,,
V =
S(q r
,
<x r ,
t)
Equa-
tion A).
The
theorem of
(27)
is
given by the
qm
= a^i
+
,
+
,
<*
m qm
of Jacobi's
Equation A)
V =
S(qly
-,q m a n
,
a w 0,
is seen to have the value and we are through. 1, We have obtained this existence theorem for the function by means of the theorem of 2, the proof of which is based on
472
characteristics.
APPENDIX C
But it might equally well have been derived from the existence theorem which is usually referred to directly 5 below, provided we are willing to assume as Cauchy's Problem,
that
H(q r p r
,
is
pr
t Q ).
t.
Consider the
(1)
= H(q
l9
,<?m ,
p lf
:
,pm ).
Jacobi's
A/ ,
Equation now
A')
_+
*
We seek
(2)
V =
in
3.
S(q l9
-,?,
-,ob,0
defnanded
It is possible to obtain
as follows.
found by setting
(3)
V =-
ht
W,
where
does not depend on
w=
t.
:
w( qi ..-,?)
,
Then
will satisfy
equation
r,x
C)
The
derivatives of
all
l}
qm
p l9
pm ) with respect to
0.*
Let
*
Then
(5)
the equation
ff fe,
l
: '
'
' ,
?m, Pi,
'
'
Pm)
ft
Pi
is
x(7i,
",?,
.
A, P2i
:
'
Pm),
and C)
^ c)
dW =
dW
dW
3 or because
H is
a positive
473
W= W
l
W( qi ,*--,qm ,h,a
2,
-,)
a>(<72
fr)
set h
!,
(9)
S(ql9
qmy
!,--, a,
=
,
- o^ +
For,
TF(ft,
qm
a l9
am ).
dW
Hence
pr = ar
r
2,
m,
ar
2,
Since
= (J1S\ /
\dardqa
to
0,
{ I
r r
s
s
r, s
1,
2,
m,
we have only
show that
Now,
(12)
x
i
is
given
by the equation
(5)
of (4)
is
complete.
474
APPENDIX C
of Results.
Summary
To sum
is
up,
then
the solution of
:
Hamilton's Equations B)
98
(14)
8S
or
dW
(15)
The a r
p) by
;
the equations
=
(16)
Pr,
2,
The # r are now given by (15) on setting q r = q r Q and substituting for a r the value given by (16). The Function W. The total differential equations which determine the
characteristics of
C) are
dq r
(17)
dW
2 Pr
8p r
.
-dp
dq r
^L
Sp r
r-1,
m.
Since
m_ ~
dP r
9"
we have
dW
(18)
or
(19)
W
a homogeneous quadratic function
of
If ff is
pM
pm then
,
(20)
Hdt
Q.
475
,
Let
c,
b)
F (xi, = (a
1?
xn an
z, y\,
c,
yn )
&n )
b l9
Consider
fa
the
partial
differential
,...,x n
Let
2,
,..-,
2Z\
)=o.
a n ) and
let
^(
2,
,a n )
,
c,
^*(oa,
o) =
6*,
A;
2,
M.
Then
^(Xj,
X n ),
,
which
is
a n ), has
c,
^(i,
^/( a i>
'
On)
n)
=
==
'
'
>
6/,
1,
n,
and
satisfies
of the point (o 1;
This
is
known
as the
Cauchy Problem.
446.
APPENDIX D
PATHS
forces (Xi, Yi, Zi).
Consider* a system of n particles m (x, y^ Zi) acted on by Their motion is governed by Newton's Law:
t
:
A)
niiXi
Xi
niiiji
Yi
mi'Zi
it
Z,-
Here are Qn dependent variables, the x,, y z,-, X^ Y Z connected by 3n equations. The problem of motion is to find 3n supplementary conditions whereby these 6n variables will be determined as functions of the time, t, and suitable initial condiEach member of the tions, and to solve for these functions. family which forms the solution, namely the curve
i}
t,
:
Xi
Xi
\l))
y\
yi
\'/j
Zi
Zi \tj
Xi =
determines a curve
Xi
(2)
Xi(t),
:
Yi
Yi(t),
Zi
Zi(t)
=
_ ~
xt(t),
yi
.
=
__ ~
2/ t
(0,
*<
.
=
__ "
*;(0
,
Xi
~dT'
Vi
"df
'
Zi
~dt
in the (6n
l)-dimensional space of the (a;,-, 2/ t 2, ft, ?/, is called a path. Obviously the paths
-,
0>
(2)
stand in a one-to-one relation to the curves (1). The problem of motion as so formulated transcends the domain In order to restrict our attention to the of Rational Mechanics. latter field, we now lay down the further postulate which, be
it
viz.,
noted, is not satisfied by certain systems which occur in nature, certain systems in which electro-magnetic phenomena are
present.
* The following treatment is the result of a joint study of the problem Professor Bernard Osgood Koopman and myself.
4.76
by
477
DYNAMICAL DETERMINATENESS.
when 6n
In a given
Xi,
7/,, z,-,
J
dynamical system,
constants (x^, yi
()
,
z t-,
are arbitrarily assigned, not more than one path (2) exists which
THE DOMAIN ZX
(6n
Those points
space,
(#;,
1) -dimensional
through
stitute the
domain D.
This domain
;
may
but in general neither of those things space, or of a region of it will be the case. It is a point set, concerning the constitution of which we need make no hypothesis at the present moment. It
will
THEOREM
The
variables
X F
i}
t,
in the points of
Xi = Xi
(4)
fa,
(x/,
(Xj,
y,, Zj,
y,-,
x h y it
Zj, t)
Yi
Zi
= =
,
Yi Zi
is
yh
Z 3; Xj,
7/y,
Zj, /)
where
(xj,
yh
z it Xj,
y,-,
z}
t)
any point of D.
D
A).
mined as functions
by
Hence Xi,
Yi,
Zi are uniquely
in question,
THE DOMAIN
variables
(x^
yi,
R.
z^
In the (3n
f)
those
set R,
which
domain D.
In particular
R may
consist of the whole space, or of a (3n l)-dimensional region in it. But in general neither of these things will be the case.
Let
be a point of R.
(2).
path given by
first line of (2),
The
:
To P
namely
(5)
all
Xi
Xi (0,
Vi
Vi (0,
(5) lies
Zi
= ^ (0,
belong to R.
wholly in R.
478
APPENDIX D
Consider an arbitrary line through P, but not perpendicular t. Let its direction components be a,-, ft, 7,-, K, where K ^ 0. There may be a path corresponding to P, such
to the axis of
yi
Zi
<*<
ft
7*.
When
through
correspond
and so certain relations between the direction components (ca, pi, 7,, K) must exist. Thus we are led to a second
to paths,
postulate.
POSTULATE
II.
The
(6)
ii
where
2} (A 3i ai
+B
D
8
8i
ft
+C
ai
t)
+D
=
yl
,
0,
1,
cr,
8 i,
8i ,
C,,
:
are functions* of
(Xi,
z t , t)
rank of
the matrix
All
(7)
'
'
'
Aln
Bn
'
Bin Cll
'
Cln
is
<r.
~~
~~
ft
7i
at the point P,
n
it
follows that
B)
(Ati Xi
+B
8i
yi
C,
Zi)
+D =
8
0,
1,
<r.
These equations form a necessary and sufficient condition for (, yi, Zi) if (Xi, y iy z iy x iy y it z it t) is to be a point of D. It may happen that the system of Equations B) (a system of Pfaffians) admits certain integrals
:
* Throughout the whole treatment, the continuity of the functions which enter, and the existence and continuity of such derivatives as it may be convenient to use,
are assumed.
479
(8)
is I. Since the system B) may obviously be replaced by any non-specialized linear combination of these equations, it is clear that Equations B) may be so chosen that the last I of them are
:
(9)
=
dt
0,
The
constants Ck
of integrals
come
C) system toward determining the particular dynamical system we are defining, different choices of the Ck leading to separate dynamical systems. They are not to be confused with constants of integration that are determined by the initial conditions within a particular
of the
dynamical system. Holonomic and Non-Holonomic Systems. If, in particular, = or, Equations B) can be replaced by Equations C) and thus
become completely
remain a
shall
I
in process of defining is
we
if
are
there
Equations B), which then are nonsaid to be non-holonomic. Equations B) now be replaced by the first p of them, and Equations C)
/*
:
>
B')
(Aaii
a=
1,
,/*;
C)
/x
I.
II
force
X
F',
t,
X' iy
of
iy
it t,
and
of further forces XJ/, F{,-, Z/, where j = 1, 2, p, the , forces of or in these latter being wholly part uncomponents
480
APPENDIX D
Denote
the
,
SK unknown components by S19 for known must enough equations postulates provide between the S's and the Xi, y^ z Xi, 2,-, t to make possible the elimination of the $'s between these equations and Equations A), with the result that the equations thus obtained, combined with Equations B') and C), will just suffice to determine x y*, z> as funcWe proceed to the details. tions of t and the initial conditions.
known.
.
Then our
t -,
?/,-,
if
D'ALEMBERT'S PRINCIPLE
In practice the equations which the Si, SK satisfy are be linear. Our shall restricted to usually problem systems which
,
POSTULATE
(10) 'Xi
III.
it
Z{
is the
sum
Zi
of two forces:
= xi
y;
17;
z;
z;,
Xi,
where X\, FJ, Z\ are known in terms of the coordinates t f an arbitrary point of D, ami where ilij Zi,
a:,-,
y^
z,-,
(11)
i=l
ijxrfc
rrih.
z;
$-,
(>
for all
fi,
r/i,
f,-
such that
(12)
i=l
ai Here, A'
2J
Abb + Shu +
C'pift
0,
j8
1,-
i/.
Xi 9
tit*
*i) t>
and
B' a i, C^i are known functions of the above the rank of the matrix :
#,-, 7/ t -,
z^
(13)
v.
Conversely,
when Equations
is
known
as the
(14)
0,
481
are 3n arbitrary quantities. where Under the sanction T/,-, f i of Postulate III. this equation can be replaced by the following
:
(15)
<-l
2) (m t
f
XI)
fc
(m<fr
- Ffl* + (w2 -
Z')f
0,
where
(12).
,-,
?7 t ,
,-
are
sulting equation
Multiply the 0-th equation (12) by X^ and subtract the from (15)
:
re-
(16)
2) (mtfi
<=1
X,'
5) A'fiiljh
0-1
(mtfr
Y'<
% B'^n
/3-1
(m i z
-Z' -^C'e
i
\p)t<
0.
0=1
Suppose for definiteness that the determinant whose matrix columns of the matrix (13) is ^ 0. Then the X's can be so determined that the coefficients of the first v
the
n i?i, quantities 1, Substitute these values of X 1?
,
,
of
*7n,
fi,
,
fn in (16) will
vanish.
\ v in the remaining
-
Thus a new linear equation in the f in which v of these quantities appear. arises, only the last 3n But the latter are arbitrary. Hence each coefficient must vanish. v equations thus obtained express the result of elimThe 3n SK i.e. the X*, Yf, Zf, from the inating the unknown S They contain only a?, y z Xi, #, z^ x y iy z t, and problem.
coefficients of (16).
,-,
17,-,
iy
iy
it
i}
E)
<-i
Gyi
2,)
+ #* =
Gyi,
0,
1,
3n
v,
where the
Zt,
IJi,
yil
Fyi
Hy
are
known
functions of
at each point of D.
Equations E) and B) form a necessary condition for the funcHence if P tions Xi(f), yi(t), Zi(t) which define a path (2). Q Q t of an is Z), Equations E) (xfj yfj Zi Xi yfj zP, Q ) arbitrary point and B) admit a solution having as its initial values the coordinates
:
Furthermore, by virtue of Postulate I., this solution is unique. We have now arrived at a complete analytical formulaLet tion of the problem, for we can retrace our steps.
of JP
.
482
APPENDIX D
be a curve lying on C) and satisfying B'). Then a) gives rise to a curve T which lies on D. Consequently all the coefficients in B'), Let a) also (12), and E) are determined in the points of a).
satisfy E).
Since E) holds,
as to
it
make each
X*
by the equations
Y*
(11)
X* = 2*
0=1
Afii
X/3,
j Bfii Xj3,
Z*
j Cfti X/3.
/3=i
0=1
(12).
and
On
substituting
them
in
F, Z{ are obtained for which A) is true, because each parenthesis in (16) vanishes, and so T is a path. But there
is
of
D.
Hence
a)
is
unique.
These Postulates complete the formulation of the problems in Rational Mechanics which we set out to The role which d'Alembert's Principle * plays is twoisolate. fold. First, it requires that the relations between the unknown &* shall b e linear. Secondly, it performs the elimination S\, such that the a multipliers {,-, T/;, f can always be technique by
class of
'
by
setting
bXi
(17)
fc
byt
is
??i,
bZi
ft.
Remark.
In general there
8i
coefficients
of Equations B) and the Ap it B^i, C'^ of EquaHence the virtual displacements to,-, 6r/, fai of (17) tions (12). will not coincide save as to infinitesimals of higher order with any
i,
,
A9
B9i C
possible displacement A# t Ay iy As* due to an actual motion of the system in time A/. In a sub-class of cases it happens, however, that the 4, J3, t Ci in B) and the Apt, B' in (12) are respectively equal to each ftiy C'pi
,
,
* Historically d'Alembcrt's Principle took its start in the assumption of a condition, necessary and sufficient, that a system of forces, acting on a system of particles, be in equilibrium, namely, that the virtual work corresponding to a virtual velocity be nil. When a system of forces not in equilibrium acts on a system of particles, the former can be replaced by a system of forces in equilibrium through the introduction of "counter effective forces" or "forces of inertia"
(sic),
483
other. But even so, if the D a are not all 0, the virtual displacement will not tally save as to infinitesimals of higher order with any possible actual displacement. Finally it can happen that, in addition, the D, are all 0. Then the virtual displacement corresponds to a possible displacement. But this is a very special, though highly important, case.
Ill
LAGRANGE'S EQUATIONS
Let
', 0m,
(18)
-, ffm,
-,ff,
of the
matrix
dq,
(19)
w, and where, moreover, the region of the (x, y^ z, 0-space which corresponds to the points (qly qm f) in which /, ??,, ^i
is
, ,
are defined, at least includes the points of R. Let T denote the kinetic energy
:
484
Consider a path
Zi
APPENDIX D
(2).
Xi(t),
is
yi
yi(t),
:
Zi
(0
all lie in
R, a curve of the
qr
0-space
thus defined
r
(20)
qr
:
(t),
1,
...,m.
__
Sq r
4-
dtdqr
~ Q"
8yi 4-
'
where
<+ Yi+ 7
8xi
8z
Equations (21) are always true under the foregoing restrictions. will be sufficient to determine the motion if the system is holonomic and if
They
(22)
r
1,
m.
For then
'
dXi
dyi
'
dzr "
and thus
is
known
,
first,
,
in
terms of
x<,
y^
z lt
x iy y
iy
z iy
t,
and
That Equations (21) can be solved so finally in terms of q r qr t. from fact that T is a positive definite the follows for q l9 m q in the form m q q ly quadratic
,
,
.
f t in
Equation
(12) is
vhere the 8qr are arbitrary. Equations C), if present, are all when the Xi, y^ Zi are expressed in terms of " tte q r and t by Equation (18). Equations B') are not present in the to be The sure, holonomic, but it is not the system is, problem.
satisfied identically
is so.
t,
from the
We
that X' F{, Z( and that the S's coincide with the now divide the S's into two categories
/S's,
:
fulfil
the
485
shall
a second sub-set,
19
RT
>
holds,
r
where
Qr
is
given
by
(23).
Let
-
Q =
in
Q?,
,
r
t
1,
-,
m,
as correspond
(18), it
namely, Equations
(26)
is
Qi*
all
*-!
Qi
*
ir r
=
for
true for
equations hold
(27)
Oft!
+ a'tmKm =
,
0,
1,
v l9
where
of D,
n'p r
t,
(28)
is v l
' 9
4"
"
"
"
~T~
ttam^wt
==
~f"
&a
==
0,
Oi == 1,
', /Zj
Cq)
<&k (fir)
T*k)
1,
:
',
lu
where
^ ^
Z,
(29)
is
/z t
(30)
Fg~
being ^.
486
APPENDIX D
,
R r can be eliminated Finally it can happen that the R v between these equations, thus leaving a system of equations beSuch a system yields a unique solution, cortween the q rj qr, q r each to path of the dynamical system with which we responding
-
set out.
is
completely formulated by
"
means
of Lagrange's Equations.
It All of the foregoing assumptions are in tentative form " At one choice of the the the functions extreme, may happen
fit
9* ti can always be made so that all these things do happen be identified with the x y i} z<
i}
:
though
be eliminated
such' a choice
namely, the
T.
of the functions
(18),
the
-,
f,
The &, rj if f. of Equations (11) are quanwhich must be able to take on every set of values which The which here figure, given by (24), are {* satisfy (12). not free under the condition (24), but arc unwarrantably restricted
(24) are too restricted.
tities
.-,
77,-,
by
(24).
In a given problem the desideratum usually is, to choose m as small as possible, subject to the requirement that the same degree
of elimination of the'S's through (24) shall as if Equations (11) and (12) had been used.
IV
NOTES
Consider the dynamical system that consists of a bead sliding on a fixed circular wire and acted on by no other forces than the reaction of the wire. Equations A) take the form
:
mx = X,
Let the wire be a
tions
circle
my =
whose
xx
Y,
is
mz =
Z.
2.
axis
the axis of
Then Equa-
B) become
+ yy =
o
:
B)
This system of Pfaffians
f
is
completely integrable
2
+
z
2
2/
=
c
487
Different values of the constants of integration, a and c, give but a path of one such system
;
has no point in
so
common
Z =
0, since 3
0,
and
is
smooth.
Then
X
and
provided
X*,
Y =
Y*,
X*+ F% =
^
y t]
0.
set
m=
q.
and take
a cos
#,
a sin
Then
and
it remains merely to integrate this differential equation. The Rough Wire. Suppose, however, the wire is rough.
Let
>
0.
Then
X* =
Y =
*
Lagrange's Equation
Jt
is still
Q'
~dij
true.
But
488
APPENDIX D
and Lagrange's Equation becomes
:
have not enough equations to solve the problem. This is the case in which Lagrange's Equations are said to "fail" or be
"inapplicable."
We
The
failure
lies,
We
by the methods
of Parts
I., II.,
before
Here, then,
m at* =
m-j% ut
r72rr
X* = - R cos
+ R sin
= F* =-juftcos0 - Rsin0.
:
x**
or
(
7*77
0,
R cos +
(
p,R sin 6)
+
+
(- nR
cos 6
- R sin 0)
=
0,
rj
0,
or, finally,
y)
MZ
(12).
rj
y) v
and
this is the
form of Equation
=
On
substituting
:
IJLX
+ y,
values
2
=- x + ny.
the
these
in
General
Equation of
~
0.
Dynamics we have
/
(i*x
+ y* =
a2
provide us with two equations for determining x and y as functions of I, and thus the problem is reduced to a purely mathematical problem in differential equations. Observe, however, that the
virtual displacement used in this solution
:
489
wire
is compatible with the constraints, i.e. the circular even save as to infinitesimals of higher order than e. It corresponds to a displacement along a line at right angles to the
R and pR. Turning now to Lagrange's Equations let us choose q l and qz as the polar coordinates of the mass m. Then Lagrange's Equations (21) become
resultant of
:
(31)
7<S
Now, Equations C q ) here become
:
fl
a.
On
here becomes
0,
form
+
/i,
M^2
0-
If,
then,
we
set
TTi
=
yield
^2
Cq
-.
ma
_
dO*
7ri
+
,
ma
<W
d*0
or
#8
and it remains merely to integrate this equation. As a further illustration of the use and abuse of Lagrange's Equations may be mentioned the Ladder Problems of pages 322 and 323.
INDEX
Absolute unit of force, 52 of mass, 79 Absolute value, 24 Acceleration, 50, 52, 287
d'entrainement, 288
of gravity,
Change
466
Charlier, 437 Check of dimensions, 79 Coefficient of friction, 10 of restitution, 271
56
d'Alembert's Principle, 345, 480 Angle of friction, 10 Angular velocity, Vector, 170, 285
Appell, 225, 244, 246, 307, 337 Areas, Law of, 108
Component
of force, 2
of velocity, 87
B
226 of, 234 Billiard ball, with slipping, 143, 237, 314 without slipping, 145, 240, 314 Blackburn's pendulum, 184 Bocher, 334 Bolza, 372, 375 Brah6, Tycho, 115
Bending, Centre
K,
130 Cone, Body, Space, 213 Conservation of energy, 256 Conservative field of force, 255, 258 Constrained motion, 95 Constraint, Forces of, 315, 325 Contact transformations, 390, 399 Particular, 403 Coordinates, Cyclic, 430 Generalized or intrinsic, 297 Normal, 335 Coriolis, 288 Couples, 25, 29, 34, 37 Composition of, 31
Compound pendulum,
Nil, 31
of,
38
Canonical equations, 338, 395 transformations, 389 Carathe'odory, 381, 445 Cart wheels, 241, 314 Cauchy problem, 475 Central force, 108, 379, 427, 434 Centre of bending, 234 Centre of gravity, 26, 27, 42 Motion of the, 120 Centre of mass, Motion of the, 120, 123
Centrifugal force, 101
field of force, 106, oil cup, 105
5,
Definition
of,
356
Dyne, 56
E
Elastic strings, 58 Elasticity, Perfect, 272
291
491
Electromagnetic
field,
254
492
INDEX
Generalized coordinates, 297 Geodesies, 308 Goursat, 468 Gravitation, Motion under the attraction of, 69
Energy, Kinetic, 75, 260 Conservation of, 256 Potential, 255 Work and, for a rigid body, 266 Equation, Solution of a trigonometric,
12
.
Law
of universal, 116
Gravitational constant, 116 Gravity, Acceleration of, 56 Gyration, Radius of, 129
H
Hadamard, 224
Hamilton's Canonical Equations, 338, 395 Proof of, 342 Solution of, 410, 432 Reduction of, to the Equilibrium Problem, 411, 413 for constant energy, 411, 420 Hamiltonian Function, 342 Hamilton's Principle, 371 Integral, 356 Integral a minimum, 381 Harmonic Motion, Simple, 64, 415 Haskins, 228 Hedrick, 468 Helical motion, 168 Hertz, 244 Holder, 370 Holonomic, 313, 479 Hooke's Law, 59, 74 Huntington, 208
Problem, 413
Escalator, 265 Euler's Angles, 214, 215
Component
of,
2
5,
43;
cf.
Huygens, 133
Impact
;
of particles,
270
Oblique, 274
in space,
36
of rigid bodies,
277
of,
Moment
191
Angle
of,
10
464 360
INDEX
Internal work, 258
Intrinsic treatment of the gyroscope,
493
225
coordinates, functions, 297 equations of the gyroscope, 236
Invariable line and plane, 201 Inverse problem, 114 Isolate the System, 102
Material point, 50 Maxwell, 119 Moment of a force, 28 of a couple, 29 of a vector, 37 of a localized vector, 197 of momentum, 197, 205
of inertia, 128, 137, 191
Moments about
centre,
the instantaneous
207 Moments about an arbitrary point, 205, 208 of a vector about a line, 40 Momentum, 50, 201, 350 Moment of, 197, 200, 350 Motion under the attraction of gravitation, 69 Newton's Laws of, 50 Simple Harmonic, 64, 415 Constrained, 95 Simple Pendulum, 97 Spherical Pendulum, 306
in a resisting medium, 81 in a plane and in space, 86 of a projectile, 93, 424 on a smooth curve, 99
Koopman,
123,
Kreisel, 236,
476 246
Ladder, 147, 322, 323, 353 Lagrange's Equations, 299, 304, 312,
348, 350, 482
377
Moving
Equations, 326
Lagrangean function, 338 Lagrangean integral, 372 Lagrangean integral a minimum, 381 Lagrangean system, 338
about a fixed point, 212 axes, 172, 216 curve, 299 surface, 303
N
Newton's Laws of Motion, 50 Second Law, 92, 290 Non-holonomic, 244, 313, 479 Normal, 9 Principal, 90 Normal coordinates, 335 Nul vector (or nil vector), 5, 447
couple, 31
Law
Numerical value, 24
M
Mass, Absolute unit of, 79 Moments about centre of, 139, 205 Motion of the centre of, 120, 201 Notion of, 118
Operator, Symbolic vector, 254 Orbit of a planet, 111, 113, 435
Oscillations, Small,
333
494
INDEX
Rotation of the axes, 454 about a fixed axis, 127, 136 of a plane lamina, 139 of a rigid body, Chap. VI Routh, 134, 202, 225, 246, 337 Ruled surfaces, 176
Parabolic motion, 93
Parallel forces in in space, 36
Particle,
a plane, 21, 23
S
<r,
197 Evaluation
of, for
rigid system.
of,
202
Poundal, 56 Principal axes of a central quadric, 194 Principle of the motion of the centre of mass, 123 of moments, 139 of moments with respect to the centre of mass, 205 d'Alembert's, 345, 480 Hamilton's, 371 of Least Action, 374 Variational, 370 Product of inertia, 191
Projectile,
Hamilton's Equations, 410, 432 Sommerfeld, 236, 246 Space curve, Motion on a, 100 Spherical pendulum, 306 Stabilizer, Ship's, 236, 247 Stationary, 359 Strings, Elastic, 58 Symbolic vector operator, 254
Motion
of a, 93,
424
Tautochrone, 98 Tennis ball, 282 Top, 220, 438 Torque, 29 Torsion pendulum, 139 Transformation of <r, 202 Contact, 390, 399, 413 Canonical, 389 of Hamilton's Equations by contact transformations, 400, 413 Translation, 159 Transmissibility of force, 22 Triangle of forces, 4
Trigonometric equation, 12 theorem, 44 Two body problem, 114, 879, 427, 434 TychoBrah6, 115 Tyndall, 166
39
Units, Absolute, 52, 55, 79
Change
of,
76