An Elementary Treatise On Fourier Series

The Project Gutenberg EBook of An Elementary Treatise on Fourier’s Series
and Spherical, Cylindrical, and Ellipsoidal Harmonics, by William Elwood

Byerly
This eBook is for the use of anyone anywhere at no cost and with
almost no restrictions whatsoever. You may copy it, give it away or
re-use it under the terms of the Project Gutenberg License included
with this eBook or online at www.gutenberg.org
Title: An Elementary Treatise on Fourier’s Series and Spherical,

Cylindrical, and Ellipsoidal Harmonics
With Applications to Problems in Mathematical Physics
Author: William Elwood Byerly
Release Date: August 19, 2009 [EBook #29779]
Language: English
Character set encoding: ISO-8859-1
*** START OF THIS PROJECT GUTENBERG EBOOK TREATISE ON FOURIER’S SERIES

***
Produced by Laura Wisewell, Carl Hudkins, Keith Edkins and

the Online Distributed Proofreading Team at
http://www.pgdp.net (The original copy of this book was
generously made available for scanning by the Department
of Mathematics at the University of Glasgow.)
AN ELEMENTARY TREATISE
ON
FOURIER’S SERIES
AND
SPHERICAL, CYLINDRICAL, AND ELLIPSOIDAL

HARMONICS,
WITH
APPLICATIONS TO PROBLEMS IN MATHEMATICAL PHYSICS.
BY
WILLIAM ELWOOD BYERLY, Ph.D.,
PROFESSOR OF MATHEMATICS IN HARVARD UNIVERSITY.
GINN & COMPANY

BOSTON · NEW YORK · CHICAGO · LONDON
Copyright, 1893,
By WILLIAM ELWOOD BYERLY.
ALL RIGHTS RESERVED.
Transcriber’s Note: A few typographical errors have been corrected - these are noted at the end
of the text.
i
PREFACE.
About ten years ago I gave a course of lectures on Trigonometric Series,

following closely the treatment of that subject in Riemann’s “Partielle Differen-
tialgleichungen,” to accompany a short course on The Potential Function, given
by Professor B. O. Peirce.
My course has been gradually modified and extended until it has become an
introduction to Spherical Harmonics and Bessel’s and Lamé’s Functions.
Two years ago my lecture notes were lithographed by my class for their own
use and were found so convenient that I have prepared them for publication,
hoping that they may prove useful to others as well as to my own students.
Meanwhile, Professor Peirce has published his lectures on “The Newtonian Po-
tential Function” (Boston, Ginn & Co.), and the two sets of lectures form a
course (Math. 10) given regularly at Harvard, and intended as a partial intro-
duction to modern Mathematical Physics.
Students taking this course are supposed to be familiar with so much of
the infinitesimal calculus as is contained in my “Differential Calculus” (Boston,
Ginn & Co.) and my “Integral Calculus” (second edition, same publishers), to
which I refer in the present book as “Dif. Cal.” and “Int. Cal.” Here, as in the
“Calculus,” I speak of a “derivative” rather than a “differential coefficient,” and
δ
use the notation Dx instead of δx for “partial derivative with respect to x.”
The course was at first, as I have said, an exposition of Riemann’s “Partielle
Differentialgleichungen.” In extending it, I drew largely from Ferrer’s “Spherical
Harmonics” and Heine’s “Kugelfunctionen,” and was somewhat indebted to
Todhunter (“Functions of Laplace, Bessel, and Lamé”), Lord Rayleigh (“Theory
of Sound”), and Forsyth (“Differential Equations”).
In preparing the notes for publication, I have been greatly aided by the
criticisms and suggestions of my colleagues, Professor B. O. Peirce and Dr.
Maxime Bôcher, and the latter has kindly contributed the brief historical sketch
contained in Chapter IX.
W. E. BYERLY.
Cambridge, Mass., Sept. 1893.
ii
ANALYTICAL TABLE OF CONTENTS.
CHAPTER I.
pages
Introduction 1–29
Art. 1. List of some important homogeneous linear partial differential equa-

tions of Physics.—Arts. 2–4. Distinction between the general solution and a
particular solution of a differential equation. Need of additional data to make
the solution of a differential equation determinate. Definition of linear and of
linear and homogeneous.—Arts. 5–6. Particular solutions of homogeneous lin-
ear differential equations may be combined into a more general solution. Need of
development in terms of normal forms.—Art. 7. Problem: Permanent state of
temperatures in a thin rectangular plate. Need of a development in sine series.
Example.—Art. 8. Problem: Transverse vibrations of a stretched elastic string.
A development in sine series suggested.—Art. 9. Problem: Potential function
due to the attraction of a circular ring of small cross-section. Surface Zonal Har-
monics (Legendre’s Coefficients). Example.—Art. 10. Problem: Permanent
state of temperatures in a solid sphere. Development in terms of Surface Zonal
Harmonics suggested.—Arts. 11–12. Problem: Vibrations of a circular drum-
head. Cylindrical Harmonics (Bessel’s Functions). Recapitulation.—Art. 13.
Method of making the solution of a linear partial differential equation depend
upon solving a set of ordinary differential equations by assuming the dependent
variable equal to a product of factors each of which involves but one of the inde-
pendent variables. Arts. 14–15 Method of solving ordinary homogeneous linear
differential equations by development in power series. Applications.—Art. 16.
Application to Legendre’s Equation. Several forms of general solution obtained.
Zonal Harmonics of the second kind.—Art. 17. Application to Bessel’s Equa-
tion. General solution obtained for the case where m is not an integer, and
for the case where m is zero. Bessel’s Function of the second kind and zeroth
order.—Art. 18. Method of obtaining the general solution of an ordinary lin-
ear differential equation of the second order from a given particular solution.
Application to the equations considered in Arts. 14–17.
CHAPTER II.
Development in Trigonometric Series 30–55
Arts. 19–22. Determination of the coefficients of n terms of a sine series so

that the sum of the terms shall be equal to a given function of x for n given val-
ues of x. Numerical example.—Art. 23. Problem of development in sine series
treated as a limiting case of the problem just solved.—Arts. 24–25. Shorter
TABLE OF CONTENTS iii
method of solving the problem of development in series involving sines of whole

multiples of the variable. Working rule deduced. Recapitulation.—Art. 26. A
few important sine developments obtained. Examples.—Arts. 27–28. Develop-
ment in cosine series. Examples.—Art. 29. Sine series an odd function of the
variable, cosine series an even function, and both series periodic functions.—
Art. 30. Development in series involving both sines and cosines of whole mul-
tiples of the variable. Fourier’s series. Examples.—Art. 31. Extension of the
range within which the function and the series are equal. Examples.—Art. 32.
Fourier’s Integral obtained.
CHAPTER III.
Convergence of Fourier’s Series 56–69
Arts. 33–36. The question of the convergence of the sine series for unity
considered at length.—Arts. 37–38. Statement of the conditions which are
sufficient to warrant the development of a function into a Fourier’s series. His-
torical note. Art. 39. Graphical representation of successive approximations to
a sine series. Properties of a Fourier’s series inferred from the constructions.—
Arts. 40–42. Investigation of the conditions under which a Fourier’s series can
be differentiated term by term.—Art. 43. Conditions under which a function
can be expressed as a Fourier’s Integral.
CHAPTER IV.
Solution of Problems in Physics by the Aid of Fourier’s Inte-

grals and Fourier’s Series 70–135
Arts. 44–48. Logarithmic Potential. Flow of electricity in an infinite plane,

where the value of the Potential Function is given along an infinite straight line;
along two mutually perpendicular straight lines; along two parallel straight lines.
Examples. Use of Conjugate Functions. Sources and Sinks. Equipotential lines
and lines of Flow. Examples.—Arts. 49–52. One-dimensional flow of heat.
Flow of heat in an infinite solid; in a solid with one plane face at the tempera-
ture zero; in a solid with one plane face whose temperature is a function of the
time (Riemann’s solution); in a bar of small cross section from whose surface
heat escapes into air at temperature zero. Limiting state approached when the
temperature of the origin is a periodic function of the time. Examples.—Arts.
53–54. Temperatures due to instantaneous and to permanent heat sources
and sinks, and to heat doublets. Examples. Application to the case where
there is leakage.—Arts. 55–56. Transmission of a disturbance along an infinite
stretched elastic string. Examples.—Arts. 57–58. Stationary temperatures
in a long rectangular plate. Temperature of the base unity. Summation of a
Trigonometric series. Isothermal lines and lines of flow. Examples.—Art. 59.
TABLE OF CONTENTS iv
Potential Function given along the perimeter of a rectangle. Examples.—Arts.

60–63. One-dimensional flow of heat in a slab with parallel plane faces. Both
faces at temperature zero. Both faces adiathermanous. Temperature of one face
a function of the time. Examples.—Art. 64. Motion of a stretched elastic string
fastened at the ends. Steady vibration. Nodes. Examples.—Art. 65. Motion of
a string in a resisting medium.—Art. 66. Flow of heat in a sphere whose surface
is kept at a constant temperature.—Arts. 67–68. Cooling of a sphere in air.
Surface condition given by a differential equation. Development in a Trigono-
metric series of which Fourier’s Sine Series is a special case. Examples.—Arts.
69–70. Flow of heat in an infinite solid with one plane face which is exposed
to air whose temperature is a function of the time. Solution for an instanta-
neous heat source when the temperature of the air is zero. Examples.—Arts.
71–73. Vibration of a rectangular drumhead. Development of a function of two
variables in a double Fourier’s Series. Examples. Nodal lines in a rectangular
drumhead. Nodal lines in a square drumhead.
Miscellaneous Problems 135–143
I. Logarithmic Potential. Polar Coördinates.—II. Potential Function in

Space. III. Conduction of heat in a plane.—IV. Conduction of heat in Space.
CHAPTER V.
Zonal Harmonics 143–195
Art. 74. Recapitulation. Surface Zonal Harmonics (Legendrians). Zonal

Harmonics of the second kind.—Arts. 75–76. Legendrians as coefficients in
a Power Series. Special values.—Art. 77. Summary of the properties of a
Legendrian. List of the first eight Legendrians. Relation connecting any three
successive Legendrians.—Arts. 78–81. Problems in Potential. Potential Func-
tion due to the attraction of a material circular ring of small cross section.
Potential Function due to a charge of electricity placed on a thin circular disc.
Examples: Spheroidal conductors. Potential Function due to the attraction
of a material homogeneous circular disc. Examples: Homogeneous hemisphere;
Heterogeneous sphere; Homogeneous spheroids. Generalisation.—Art. 82. Leg-
endrian as a sum of cosines.—Arts. 83–84. Legendrian as the mth derivative
of the mth power of x2 − −1.—Art. 85. Equations derivable from Legendre’s
Equation.—Art. 86. Legendrian as a Partial Derivative.—Art. 87. Legen-
drian as a Definite Integral. Arts. 88–90. Development in Zonal Harmonic
Series. Integral of the product of two Legendrians of different degrees. Integral
of the square of a Legendrian. Formulas for the coefficients of the series.—
Arts. 91–92. Integral of the product of two Legendrians obtained by the aid
of Legendre’s Equation; by the aid of Green’s Theorem. Additional formulas
for integration. Examples.—Arts. 93–94. Problems in Potential where the
value of the Potential Function is given on a spherical surface and has circular
TABLE OF CONTENTS v
symmetry about a diameter. Examples.—Art. 95. Development of a power

of x in Zonal Harmonic Series.—Art. 96. Useful formulas.—Art. 97. Devel-
opment of sin nθ and cos nθ in Zonal Harmonic Series. Examples. Graphical
representation of the first seven Surface Zonal Harmonics. Construction of suc-
cessive approximations to Zonal Harmonic Series. Arts. 98–99. Method of
dealing with problems in Potential when the density is given. Examples.—Art.
100. Surface Zonal Harmonics of the second kind. Examples: Conal Harmonics.
CHAPTER VI.
Spherical Harmonics 196–219
Arts. 101–102. Particular Solutions of Laplace’s Equation obtained. As-

sociated Functions. Tesseral Harmonics. Surface Spherical Harmonics. Solid
Spherical Harmonics. Table of Associated Functions. Examples.—Arts. 103–
108. Development in Spherical Harmonic Series. The integral of the product
of two Surface Spherical Harmonics of different degrees taken over the surface
of the unit sphere is zero. Examples. The integral of the product of two Asso-
ciated Functions of the same order. Formulas for the coefficients of the series.
Illustrative example. Examples.—Arts. 109–110. Any homogeneous rational
integral Algebraic function of x, y, and z which satisfies Laplace’s Equation is
a Solid Spherical Harmonic. Examples.—Art. 111. A transformation of axes
to a new set having the same origin will change a Surface Spherical Harmonic
into another of the same degree.—Arts. 112–114. Laplacians. Integral of the
product of a Surface Spherical Harmonic by a Laplacian of the same degree.
Development in Spherical Harmonic Series by the aid of Laplacians. Table of
Laplacians. Example.—Art. 115. Solution of problems in Potential by direct
integration. Examples.—Arts. 116–118. Differentiation along an axis. Axes
of a Spherical Harmonic.—Art. 119. Roots of a Zonal Harmonic. Roots of a
Tesseral Harmonic. Nomenclature justified.
CHAPTER VII.
Cylindrical Harmonics (Bessel’s Functions) 220–238
Art. 120. Recapitulation. Cylindrical Harmonics (Bessel’s Functions) of the

zeroth order; of the nth order; of the second kind. General solution of Bessel’s
Equation.—Art. 121. Bessel’s Functions as definite integrals. Examples.—
Art. 122. Properties of Bessel’s Functions. Semi-convergent series for a Bessel’s
Function. Examples.—Art. 123. Problem: Stationary temperatures in a cylin-
der (a) when the temperature of the convex surface is zero; (b) when the convex
surface is adiathermanous; (c) when the convex surface is exposed to air at the
temperature zero.—Art. 124. Roots of Bessel’s functions.—Art. 125. The in-
tegral of r times the product of two Cylindrical Harmonics of the zeroth order.
TABLE OF CONTENTS vi
Example.—Art. 126. Development in Cylindrical Harmonic Series. Formulas

for the coefficients. Examples.—Art. 127. Problem: Stationary temperatures
in a cylindrical shell. Bessel’s Functions of the second kind employed. Example:
Vibration of a ring membrane.—Art. 128. Problem: Stationary temperatures
in a cylinder when the temperature of the convex surface varies with the distance
from the base. Bessel’s Functions of a complex variable. Examples.—Art. 129.
Problem: Stationary temperatures in a cylinder when the temperatures of the
base are unsymmetrical. Bessel’s Functions of the nth order employed. Miscel-
laneous examples. Bessel’s Functions of fractional order.
CHAPTER VIII.
Laplace’s Equation in Curvilinear Coördinates. Ellipsoidal

Harmonics 239–266
Arts. 130–131. Orthogonal Curvilinear Coördinates in general. Laplace’s

Equation expressed in terms of orthogonal curvilinear coördinates by the aid of
Green’s theorem.—Arts. 132–135. Spheroidal Coördinates. Laplace’s Equation
in spheroidal coördinates, in normal spheroidal coördinates. Examples. Condi-
tion that a set of curvilinear coördinates should be normal. Thermometric Pa-
rameters. Particular solutions of Laplace’s Equation in spheroidal coördinates.
Spheroidal Harmonics. Examples. The Potential Function due to the attrac-
tion of an oblate spheroid. Solution for an external point. Examples.—Arts.
136–141. Ellipsoidal Coördinates. Laplace’s Equation in ellipsoidal coördinates.
Normal ellipsoidal coördinates expressed as Elliptic Integrals. Particular solu-
tions of Laplace’s Equation. Lamé’s Equation. Ellipsoidal Harmonics (Lamé’s
Functions). Tables of Ellipsoidal Harmonics of the degrees 1, 2, and 3. Lamé’s
Functions of the second kind. Examples. Development in Ellipsoidal Har-
monic series. Value of the Potential Function at any point in space when its
value is given at all points on the surface of an ellipsoid.—Art. 142. Conical
Coördinates. The product of two Ellipsoidal Harmonics a Spherical Harmonic.—
Art. 143. Toroidal Coördinates. Laplace’s Equation in toroidal coördinates.
Particular solutions. Toroidal Harmonics. Potential Function for an anchor ring.
CHAPTER IX.
Historical Summary 267–274
APPENDIX.
Tables 274–285
Table I. Surface Zonal Harmonics. Argument θ 276
Table II. Surface Zonal Harmonics. Argument x 278
TABLE OF CONTENTS vii
Table III. Hyperbolic Functions 280

Table IV. Roots of Bessel’s Functions 284
Table V. Roots of Bessel’s Functions 284
Table VI. Bessel’s Functions 285
1
CHAPTER I.
INTRODUCTION.
1. In many important problems in mathematical physics we are obliged

to deal with partial differential equations of a comparatively simple form.
For example, in the Analytical Theory of Heat we have for the change of
temperature of any solid due to the flow of heat within the solid, the equation
Dt u = a2 (Dx2 u + Dy2 u + Dz2 u), 1 [I]
where u represents the temperature at any point of the solid and t the time.
In the simplest case, that of a slab of infinite extent with parallel plane
faces, where the temperature can be regarded as a function of one coördinate,
[I] reduces to
Dt u = a2 Dx2 u, [II]
a form of considerable importance in the consideration of the problem of the
cooling of the earth’s crust.
In the problem of the permanent state of temperatures in a thin rectangular
plate, the equation [I] becomes
Dx2 u + Dy2 u = 0. [III]
In polar or spherical coördinates [I] is less simple, it is
a2

1 1
Dt u = 2 Dr (r2 Dr u) + Dθ (sin θDθ u) + Dφ
2
u . [IV]
r sin θ sin2 θ
In the case where the solid in question is a sphere and the temperature at any
point depends merely on the distance of the point from the centre [IV] reduces
to
Dt (ru) = a2 Dr2 (ru). [V]
In cylindrical coördinates [I] becomes
1 1
Dt u = a2 [Dr2 u + Dr u + 2 Dφ2 u + Dz2 u]. [VI]
r r
In considering the flow of heat in a cylinder when the temperature at any
point depends merely on the distance r of the point from the axis [VI] becomes
1
Dt u = a2 (Dr2 u + Dr u). [VII]
r
In Acoustics in several problems we have the equation
Dt2 y = a2 Dx2 y; [VIII]

1 For the sake of brevity we shall often use the symbol ∇2 for the operation D 2 + D 2 + D 2 ;
x y z
and with this notation equation [I] would be written Dt u = a2 ∇2 u.
INTRODUCTION. 2
for instance, in considering the transverse or the longitudinal vibrations of a

stretched elastic string, or the transmission of plane sound waves through the
air.
If in considering the transverse vibrations of a stretched string we take ac-
count of the resistance of the air [VIII] is replaced by
Dt2 y + 2kDt y = a2 Dx2 y. [IX]
In dealing with the vibrations of a stretched elastic membrane, we have the
equation
Dt2 z = c2 (Dx2 z + Dy2 z), [X]
or in cylindrical coördinates
1 1
Dt2 z = c2 (Dr2 z + Dr z + 2 Dφ2 z). [XI]
r r
In the theory of Potential we constantly meet Laplace’s Equation
Dx2 V +Dy2 V + Dz2 V = 0 [XII]

2
or ∇ V =0
which in spherical coördinates becomes

1 2 1 1 2
rD r (rV ) + Dθ (sin θDθ V ) + D φ V = 0, [XIII]
r2 sin θ sin2 θ
and in cylindrical coördinates
1 1
Dr2 V + Dr V + 2 Dφ2 V + Dz2 V = 0. [XIV]
r r
In curvilinear coördinates it is

h1 h2 h3
h1 h2 h3 Dρ1 Dρ V + Dρ2 Dρ V + Dρ3 Dρ V = 0;
h2 h3 1 h3 h1 2 h1 h2 3
[XV]
where f1 (x, y, z) = ρ1 , f2 (x, y, z) = ρ2 , f3 (x, y, z) = ρ3
represent a set of surfaces which cut one another at right angles, no matter what
values are given to ρ1 , ρ2 , and ρ3 ; and where
h21 = (Dx ρ1 )2 + (Dy ρ1 )2 + (Dz ρ1 )2
h22 = (Dx ρ2 )2 + (Dy ρ2 )2 + (Dz ρ2 )2
h23 = (Dx ρ3 )2 + (Dy ρ3 )2 + (Dz ρ3 )2 ,
and, of course, must be expressed in terms of ρ1 , ρ2 , and ρ3 .
If it happens that ∇2 ρ1 = 0, ∇2 ρ2 = 0, and ∇2 ρ3 = 0, then Laplace’s
Equation [XV] assumes the very simple form
h21 Dρ21 V + h22 Dρ22 V + h23 Dρ23 V = 0. [XVI]
INTRODUCTION. 3
2. A differential equation is an equation containing derivatives or differ-

entials with or without the primitive variables from which they are derived.
The general solution of a differential equation is the equation expressing the
most general relation between the primitive variables which is consistent with
the given differential equation and which does not involve differentials or deriva-
tives. A general solution will always contain arbitrary (i. e., undetermined)
constants or arbitrary functions.
A particular solution of a differential equation is a relation between the
primitive variables which is consistent with the given differential equation, but
which is less general than the general solution, although included in it.
Theoretically, every particular solution can be obtained from the general
solution by substituting in the general solution particular values for the arbitrary
constants or particular functions for the arbitrary functions; but in practice it is
often easy to obtain particular solutions directly from the differential equation
when it would be difficult or impossible to obtain the general solution.
3. If a problem requiring for its solution the solving of a differential equa-

tion is determinate, there must always be given in addition to the differential
equation enough outside conditions for the determination of all the arbitrary
constants or arbitrary functions that enter into the general solution of the equa-
tion; and in dealing with such a problem, if the differential equation can be
readily solved the natural method of procedure is to obtain its general solu-
tion, and then to determine the constants or functions by the aid of the given
conditions.
It often happens, however, that the general solution of the differential equa-
tion in question cannot be obtained, and then, since the problem if determinate
will be solved if by any means a solution of the equation can be found which
will also satisfy the given outside conditions, it is worth while to try to get par-
ticular solutions and so to combine them as to form a result which shall satisfy
the given conditions without ceasing to satisfy the differential equation.
4. A differential equation is linear when it would be of the first degree

if the dependent variable and all its derivatives were regarded as algebraic un-
known quantities. If it is linear and contains no term which does not involve
the dependent variable or one of its derivatives, it is said to be linear and ho-
mogeneous.
All the differential equations collected in Art. 1 are linear and homogeneous.
5. If a value of the dependent variable has been found which satisfies a

given homogeneous, linear, differential equation, the product formed by multi-
plying this value by any constant will also be a value of the dependent variable
which will satisfy the equation.
For if all the terms of the given equation are transposed to the first member,
the substitution of the first-named value must reduce that member to zero;
substituting the second value is equivalent to multiplying each term of the result
of the first substitution by the same constant factor, which therefore may be
INTRODUCTION. 4
taken out as a factor of the whole first member. The remaining factor being
zero, the product is zero and the equation is satisfied.
If several values of the dependent variable have been found each of which
satisfies the given differential equation, their sum will satisfy the equation; for if
the sum of the values in question is substituted in the equation each term of the
sum will give rise to a set of terms which must be equal to zero, and therefore
the sum of these sets must be zero.
6. It is generally possible to get by some simple device particular solutions

of such differential equations as those we have collected in Art. 1. The object of
the branch of mathematics with which we are about to deal is to find methods of
so combining these particular solutions as to satisfy any given conditions which
are consistent with the nature of the problem in question.
This often requires us to be able to develop any given function of the variables
which enter into the expression of these conditions in terms of normal forms
suited to the problem with which we happen to be dealing, and suggested by
the form of particular solution that we are able to obtain for the differential
equation.
These normal forms are frequently sines and cosines, but they are often
much more complicated functions known as Legendre’s Coefficients, or Zonal
Harmonics; Laplace’s Coefficients, or Spherical Harmonics: Bessel’s Functions,
or Cylindrical Harmonics; Lamé’s Functions, or Ellipsoidal Harmonics, &c.
7. As an illustration, let us take Fourier’s problem of the permanent state

of temperatures in a thin rectangular plate of breadth π and of infinite length
whose faces are impervious to heat. We shall suppose that the two long edges of
the plate are kept at the constant temperature zero, that one of the short edges,
which we shall call the base of the plate, is kept at the temperature unity, and
that the temperatures of points in the plate decrease indefinitely as we recede
from the base; we shall attempt to find the temperature at any point of the
plate.
Let us take the base as the axis of X and one end of the base as the origin.
Then to solve the problem we are to find the temperature u of any point from
the equation
Dx2 u + Dy2 u = 0 [III] Art. 1
subject to the conditions
u=0 when x = 0 (1)

u=0 “ x=π (2)
u=0 “ y=∞ (3)
u=1 “ y = 0. (4)
We shall begin by getting a particular solution of [III], and we shall use a

device which always succeeds when the equation is linear and homogeneous and
has constant coefficients.
INTRODUCTION. 5
Assume2 u = eαy+βx , where α and β are constants, substitute in [III] and

divide by eαy+βx , and we have α2 + β 2 = 0. If, then, this condition is satisfied
u = eαy+βx is a solution.
Hence u = eαy±αxi 3 is a solution of [III], no matter what value may be given
to α.
This form is objectionable, since it involves an imaginary. We can, however,
readily improve it.
Take u = eαy eαxi , a solution of [III], and u = eαy e−αxi , another solution
of [III]; add these values of u and divide the sum by 2 and we have eαy cos αx.
(v. Int. Cal. Art. 35, [1].) Therefore by Art. 5
u = eαy cos αx (5)
is a solution of [III]. Take u = eαy eαxi and u = eαy e−αxi , subtract the second
value of u from the first and divide by 2i and we have eαy sin αx. (v. Int. Cal.
Art. 35, [2]). Therefore by Art. 5
u = eαy sin αx (6)
is a solution of [III].
Let us now see if out of these particular solutions we can build up a solution
which will satisfy the conditions (1), (2), (3), and (4).
Consider u = eαy sin αx. (6)
It is zero when x = 0 for all values of α. It is zero when x = π if α is a whole

number. It is zero when y = ∞ if α is negative. If, then, we write u equal
to a sum of terms of the form Ae−my sin mx, where m is a positive integer, we
shall have a solution of [III] which satisfies conditions (1), (2) and (3). Let this
solution be
u = A1 e−y sin x + A2 e−2y sin 2x + A3 e−3y sin 3x + A4 e−4y sin 4x + · · · (7)
A1 , A2 , A3 , A4 , &c., being undetermined constants.

When y = 0 (7) reduces to
u = A1 sin x + A2 sin 2x + A3 sin 3x + A4 sin 4x + · · · . (8)
If now it is possible to develop unity into a series of the form (8), our problem
is solved; we have only to substitute the coefficients of that series for A1 , A2 ,
A3 , &c. in (7).
It will be proved later that

4 1 1 1
1= sin x + sin 3x + sin 5x + sin 7x + · · ·
π 3 5 7
2 This assumption must be regarded as purely tentative. It must be tested by substituting
in the equation, and is justified if it leads to√a solution.

3 We shall regularly use the symbol i for −1.
INTRODUCTION. 6
for all values of x between 0 and π; hence our required solution is

4 −y 1 1 1
u= e sin x + e−3y sin 3x + e−5y sin 5x + e−7y sin 7x + · · · (9)
π 3 5 7
for this satisfies the differential equation and all the given conditions.
If the given temperature of the base of the plate instead of being unity is a
function of x, we can solve the problem as before if we can express the given
function of x as a sum of terms of the form A sin mx, where m is a whole number.
The problem of finding the value of the potential function at any point of a
long, thin, rectangular conducting sheet, of breadth π, through which an electric
current is flowing, when the two long edges are kept at potential zero, and one
short edge at potential unity, is mathematically identical with the problem we
have just solved.
Example.
Taking the temperature of the base of the plate described above as 100◦
centigrade, and that of the sides of the plate as 0◦ , compute the temperatures
of the points
π π π
(a) , 1 ; (b) , 2 ; (c) ,3 ,
6 3 2
correct to the nearest degree. Ans. (a) 26◦ ; (b) 15◦ ; (c) 6◦ .
8. As another illustration, we shall take the problem of the transverse

vibrations of a stretched string fastened at the ends, initially distorted into
some given curve and then allowed to swing.
Let the length of the string be l. Take the position of equilibrium of the
string as the axis of X, and one of the ends as the origin, and suppose the string
initially distorted into a curve whose equation y = f (x) is given.
We have then to find an expression for y which will be a solution of the
equation
Dt2 y = a2 Dx2 y [VIII] Art. 1,
while satisfying the conditions
y=0 when x=0 (1)
y=0 “ x=l (2)
y = f (x) “ t=0 (3)
Dt y = 0 “ t = 0, (4)
the last condition meaning merely that the string starts from rest.
As in the last problem let4 y = eαx+βt and substitute in [VIII]. Divide by
αx+βt
e and we have β 2 = a2 α2 as the condition that our assumed value of y
shall satisfy the equation.
y = eαx±aαt (5)
4 See note on page 5.
INTRODUCTION. 7
is, then, a solution of (VIII) whatever the value of α.

It is more convenient to have a trigonometric than an exponential form to
deal with, and we can readily obtain one by using an imaginary value for α in
(5). Replace α by αi and (5) becomes y = e(x±at)αi , a solution of [VIII]. Replace
α by −αi and (5) becomes y = e−(x±at)αi , another solution of [VIII]. Add these
values of y and divide by 2 and we have cos α(x ± at). Subtract the second value
of y from the first and divide by 2i and we have sin α(x ± at).
y = cos α(x + at)
y = cos α(x − at)
y = sin α(x + at)
y = sin α(x − at)
are, then, solutions of [VIII]. Writing y successively equal to half the sum of the
first pair of values, half their difference, half the sum of the last pair of values,
and half their difference, we get the very convenient particular solutions of [VIII].
y = cos αx cos αat
y = sin αx sin αat
y = sin αx cos αat
y = cos αx sin αat.
If we take the third form
y = sin αx cos αat
it will satisfy conditions (1) and (4), no matter what value may be given to α,
mπ
and it will satisfy (2) if α = where m is an integer.
l
If then we take
πx πat 2πx 2πat 3πx 3πat
y = A1 sin cos + A2 sin cos + A3 sin cos + · · · (6)
l l l l l l
where A1 , A2 , A3 · · · are undetermined constants, we shall have a solution of
[VIII] which satisfies (1), (2), and (4). When t = 0 it reduces to
πx 2πx 3πx
y = A1 sin + A2 sin + A3 sin + ··· (7)
l l l
If now it is possible to develop f (x) into a series of the form (7), we can solve
our problem completely. We have only to take the coefficients of this series as
values of A1 , A2 , A3 · · · in (6), and we shall have a solution of [VIII] which
satisfies all our given conditions.
In each of the preceding problems the normal function, in terms of which
a given function has to be expressed, is the sine of a simple multiple of the
variable. It would be easy to modify the problem so that the normal form
should be a cosine.
We shall now take a couple of problems which are much more complicated
and where the normal function is an unfamiliar one.
INTRODUCTION. 8
9. Let it be required to find the potential function due to a circular wire

ring of small cross section and of given radius c, supposing the matter of the
ring to attract according to the law of nature.
We can readily find, by direct integration, the value of the potential function
at any point of the axis of the ring. We get for it
M
V =√ (1)
c2+ x2
where M is the mass of the ring, and x the distance of the point from the centre
of the ring.
Let us use spherical coördinates, taking the centre of the ring as origin and
the axis of the ring as the polar axis.
To obtain the value of the potential function at any point in space, we must
satisfy the equation
1 1
rDr2 (rV ) + Dθ (sin θDθ V ) + Dφ2 V = 0, [XIII] Art. 1,
sin θ sin2 θ
subject to the condition
M
V = 1 when θ = 0. (1)
(c2 + r2 ) 2
From the symmetry of the ring, it is clear that the value of the potential
function must be independent of φ, so that [XIII] will reduce to
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0. (2)
sin θ
We must now try to get particular solutions of (2), and as the coefficients
are not constant, we are driven to a new device.
Let5 V = rm P , where P is a function of θ only, and m is a positive integer,
and substitute in (2), which becomes
rm
m(m + 1)rm P + Dθ (sin θDθ P ) = 0.
sin θ
Divide by rm and use the notation of ordinary derivatives since P depends upon
θ only, and we have the equation
dP
1 d sin θ
m(m + 1)P + dθ = 0, (3)
sin θ dθ
from which to obtain P .
Equation (3) can be simplified by changing the independent variable. Let
x = cos θ and (3) becomes

d dP
(1 − x2 ) + m(m + 1)P = 0. (4)
dx dx
INTRODUCTION. 9
Assume6 now that P can be expressed as a sum or as a series of terms

involving whole
P powers of x multiplied by constant coefficients.
Let P = an xn and substitute this value of P in (4). We get
[n(n − 1)an xn−2 − n(n + 1)an xn + m(m + 1)an xn ] = 0,

P
(5)
P
where the symbol indicates that we are to form all the terms we can by
taking successive whole numbers for n.
As (5) must be true no matter what the value of x, the coefficient of any
given power of x, as for instance xk , must vanish. Hence
(k + 2)(k + 1)ak+2 − k(k + 1)ak + m(m + 1)ak = 0 (6)

m(m + 1) − k(k + 1)
and ak+2 = − ak . (7)
(k + 1)(k + 2)
ak xk
P
If now any set of coefficients satisfying the relation (7) be taken, P =
will be a solution of (4).
If k = m, ak+2 = 0, ak+4 = 0, &c.
Since it will answer our purpose if we pick out the simplest set of coefficients
that will obey the condition (7), we can take a set including am .
Let us rewrite (7) in the form
(k + 2)(k + 1)
ak = − ak+2 . (8)
(m − k)(m + k + 1)
We get from (8), beginning with k = m − 2,
m(m − 1)
am−2 = − am
2.(2m − 1)
m(m − 1)(m − 2)(m − 3)
am−4 = am
2.4.(2m − 1)(2m − 3)
m(m − 1)(m − 2)(m − 3)(m − 4)(m − 5)
am−6 =− am , &c.
2.4.6.(2m − 1)(2m − 3)(2m − 5)
If m is even we see that the set will end with a0 , if m is odd, with a1 .

m m(m − 1) m−2 m(m − 1)(m − 2)(m − 3) m−4
P = am x − x + x − ···
2.(2m − 1) 2.4.(2m − 1)(2m − 3)
where am is entirely arbitrary, is, then, a solution of (4). It is found convenient

to take am equal to
(2m − 1)(2m − 3) · · · 1
m!
and it can be shown that with this value of am P = 1 when x = 1.
INTRODUCTION. 10
P is a function of x and contains no higher powers of x than xm . It is usual

to write it as Pm (x).
We proceed to compute a few values of Pm (x) from the formula

(2m − 1)(2m − 3) · · · 1 m m(m − 1) m−2
Pm (x) = x − x
m! 2.(2m − 1)

m(m − 1)(m − 2)(m − 3) m−4
+ x − ··· . (9)
2.4.(2m − 1)(2m − 3)
We have:
P0 (x) = 1 or P0 (cos θ) = 1



P1 (x) = x “ P1 (cos θ) = cos θ





2
1
P2 (cos θ) = 12 (3 cos2 θ − 1)

P2 (x) = 2 (3x − 1) “ 



3

P3 (x) = 1
2 (5x − 3x) “ P3 (cos θ) = 12 (5 cos3 θ − 3 cos θ)


1 4 2
(10)
P4 (x) = 8 (35x − 30x + 3) or 


P4 (cos θ) = 18 (35 cos4 θ − 30 cos2 θ + 3)






P5 (x) = 18 (63x5 − 70x3 + 15x) or





P5 (cos θ) = 18 (63 cos5 θ − 70 cos3 θ + 15 cos θ).

We have obtained P = Pm (x) as a particular solution of (4) and P =

Pm (cos θ) as a particular solution of (3). Pm (x) or Pm (cos θ) is a new function,
known as a Legendre’s Coefficient, or as a Surface Zonal Harmonic, and occurs
as a normal form in many important problems.
V = rm Pm (cos θ) is a particular solution of (2) and rm Pm (cos θ) is sometimes
called a Solid Zonal Harmonic.
We can now proceed to the solution of our original problem.
V = A0 r0 P0 (cos θ) + A1 rP1 (cos θ) + A2 r2 P2 (cos θ) + A3 r3 P3 (cos θ) + · · · (11)
where A0 , A1 , A2 , &c., are entirely arbitrary, is a solution of (2) (v. Art. 5).
When θ = 0 (11) reduces to
V = A0 + A1 r + A2 r 2 + A3 r 3 + · · · ,
since, as we have said, Pm (x) = 1 when x = 1, or Pm (cos θ) = 1 when θ = 0.

By our condition (1)
M
V = 1
(c + r2 ) 2
2
when θ = 0.
By the Binomial Theorem
1 r2 1.3 r4 1.3.5 r6

M M
1 = 1− + − + ···
(c2 + r2 ) 2 c 2 c2 2.4 c4 2.4.6 c6
INTRODUCTION. 11
provided r < c. Hence
1 r2 1.3 r4

M
V = P0 (cos θ) − P 2 (cos θ) + P4 (cos θ)
c 2 c2 2.4 c4
1.3.5 r6

− P6 (cos θ) + · · · (12)
2.4.6 c6
is our required solution if r < c; for it is a solution of equation (2) and satisfies
condition (1).
Example.
Taking the mass of the ring as one pound and the radius of the ring as one
foot, compute to two decimal places the value of the potential function due to
the ring at the points
π
(a) (r = .2, θ = 0) ; (d ) (r = .6, θ = 0) ; (f ) r = .6, θ = ;
3
π π π
(b) r = .2, θ = ; (e) r = .6, θ = ; (g) r = .6, θ = ;
4 6 2
π
(c) r = .2, θ = ; Ans. (a) .98; (b) .99; (c) 1.01; (d ) .86;
2
(e) .90; (f ) 1.00; (g) 1.10.
The unit used is the potential due to a pound of mass concentrated at a point
and attracting a second pound of mass concentrated at a point, the two points
being a foot apart.
10. A slightly different problem calling for development in terms of Zonal

Harmonics is the following:
Required the permanent temperatures within a solid sphere of radius 1, one
half of the surface being kept at the constant temperature zero, and the other
half at the constant temperature unity.
Let us take the diameter perpendicular to the plane separating the unequally
heated surfaces as our axis and let us use spherical coördinates. As in the last
problem, we must solve the equation
1 1
rDr2 (ru) + Dθ (sin θDθ u) + Dφ2 u = 0 [XIII] Art. 1
sin θ sin2 θ
which as before reduces to
1
rDr2 (ru) + Dθ (sin θDθ u) = 0 (1)
sin θ
from the consideration that the temperatures must be independent of φ.
Our equation of condition is
π π
u = 1 from θ = 0 to θ = and u = 0 from θ = to θ = π, (2)
2 2
INTRODUCTION. 12
when r = 1.
As we have seen u = rm Pm (cos θ) is a particular solution of (1), m being
any positive whole number, and
u = A0 r0 P0 (cos θ) + A1 rP1 (cos θ) + A2 r2 P2 (cos θ) + A3 r3 P3 (cos θ) + · · · (3)
where A0 , A1 , A2 , A3 · · · are undetermined constants, is a solution of (1).

When r = 1 (3) reduces to
u = A0 P0 (cos θ) + A1 P1 (cos θ) + A2 P2 (cos θ) + A3 P3 (cos θ) + · · · (4)
If then we can develop our function of θ which enters into equation (2) in a
series of the form (4), we have only to take the coefficients of that series as the
values of A0 , A1 , A2 , &c., in (3) and we shall have our required solution.
11. As a last example we shall take the problem of the vibration of

a stretched circular membrane fastened at the circumference, that is, of an
ordinary drumhead. We shall suppose the membrane initially distorted into
any given form which has circular symmetry7 about an axis through the centre
perpendicular to the plane of the boundary, and then allowed to vibrate.
Here we have to solve

2 2 2 1 1 2
Dt z = c Dr z + Dr z + 2 Dφ z [XI] Art. 1
r r
z = f (r) when t=0 (1)

Dt z = 0 “ t=0 (2)
z=0 “ r = a. (3)
From the symmetry of the supposed initial distortion z must be independent

of φ, therefore [XI] reduces to

2 2 2 1
Dt z = c Dr z + Dr z (4)
r
and this is the equation for which we wish to find a particular solution.
We shall employ a device not unlike that used in Art. 9.
Assume8 z = R.T where R is a function of r alone and T is a function of t
alone. Substitute this value of z in (4) and we get

1
RDt2 T = c2 T Dr2 R + Dr R
r
7 A function of the coördinates of a point has circular symmetry about an axis when its
value is not affected by rotating the point through any angle about the axis. A surface has
circular symmetry about an axis when it is a surface of revolution about the axis.
INTRODUCTION. 13
1 d2 T d2 R 1 dR

1
or 2 2
= + . (5)
c T dt R dr2 r dr
The second member of (5) does not involve t, therefore its equal the first member
must be independent of t. The first member of (5) does not involve r, and
consequently since it contains neither t nor r, it must be constant. Let it equal
−µ2 , where µ of course is an undetermined constant.
Then (5) breaks up into the two differential equations
d2 T
+ µ2 c2 T = 0 (6)
dt2
d2 R 1 dR
+ + µ2 R = 0. (7)
dr2 r dr
(6) can be solved by familiar methods, and we get T = cos µct and T = sin µct
as simple particular solutions (v. Int. Cal. p. 319, § 21).
To solve (7) is not so easy. We shall first simplify it by a change of indepen-
x
dent variable. Let r = . (7) becomes
µ
d2 R 1 dR
+ + R = 0. (8)
dx2 x dx
Assume, P
as in Art. 9, that R can be expressed in terms of whole powers of
x. Let R = an xn and substitute in (8). We get
[n(n − 1)an xn−2 + nan xn−2 + an xn ] = 0,
P
an equation which must be true no matter what the value of x. The coefficient
of any given power of x, as xk−2 , must, then, vanish, and
k(k − 1)ak + kak + ak−2 = 0
or k 2 ak + ak−2 = 0
whence we obtain ak−2 = − k 2 ak (9)
as
P the konly relation that need be satisfied by the coefficients in order that R =
ak x shall be a solution of (8).
If k = 0, ak−2 = 0, ak−4 = 0, &c.
We can then begin with k = 0 as our lowest subscript.
ak−2
From (9) ak = − 2 .
k
a0
Then a2 = − 2
2
a0
a4 = 2 2
2 .4
a0
a6 = − 2 2 2 , &c.
2 .4 .6
x2 x4 x6

Hence R = a0 1 − 2 + 2 2 − 2 2 2 + · · ·
2 2 .4 2 .4 .6
INTRODUCTION. 14
where a0 may be taken at pleasure, is a solution of (8), provided the series is

convergent.
Take a0 = 1, and then R = J0 (x) where
x2 x4 x6 x8
J0 (x) = 1 − + − + − ··· (10)
22 22 .42 22 .42 .62 22 .42 .62 .82
is a solution of (8).
J0 (x) is easily shown to be convergent for all values real or imaginary of x,
since the series made up of the moduli of the terms of J0 (x) (v. Int. Cal. Art. 30)
r2 r4 r6
1+ + + + ··· ,
22 22 .42 22 .42 .62
where r is the modulus of x, is convergent for all values of r. For the ratio of
r2
the n + 1st term of this series to the nth term is and approaches zero as
4n2
its limit as n is indefinitely increased, no matter what the value of r. Therefore
J0 (x) is absolutely convergent.
J0 (x) is a new and important form. It is called a Bessel’s Function of the
zeroth order, or a Cylindrical Harmonic.
Equation (8) was obtained from (7) by the substitution of x = µr, therefore
(µr)2 (µr)4 (µr)6
R = J0 (µr) = 1 − 2
+ 2 2 − 2 2 2 + ···
2 2 .4 2 .4 .6
is a solution of (7), no matter what the value of µ, and z = J0 (µr) cos µct or
z = J0 (µr) sin µct is a solution of (4).
z = J0 (µr) cos µct satisfies condition (2) whatever the value of µ. In order
that it should also satisfy condition (3) µ must be so taken that
J0 (µa) = 0; (11)
that is, µ must be a root of (11) regarded as an equation in µ.

It can be shown that J0 (x) = 0 has an infinite number of real positive roots,
any one of which can be obtained to any required degree of approximation
without serious difficulty. Let x1 , x2 , x3 , · · · be these roots. Then if
x1 x2 x3
= µ1 , = µ2 , = µ3 , &c.
a a a
z = A1 J0 (µ1 r) cos µ1 ct + A2 J0 (µ2 r) cos µ2 ct + A3 J0 (µ3 r) cos µ3 ct + · · · , (12)
where A1 , A2 , A3 , &c., are any constants, is a solution of (4) which satisfies

conditions (2) and (3).
When t = 0 (12) reduces to
z = A1 J0 (µ1 r) + A2 J0 (µ2 r) + A3 J0 (µ3 r) + · · · . (13)
If then f (r) can be expressed as a series of the form just given, the solution of
our problem can be obtained by substituting the coefficients of that series for
A1 , A2 , A3 , &c., in (12).
INTRODUCTION. 15
Example.
The temperature of a long cylinder is at first unity throughout. The convex

surface is then kept at the constant temperature zero. Show that the tempera-
ture of any point in the cylinder at the expiration of the time t is
2
µ21 t 2
µ22 t 2
µ23 t
u = A1 e−a J0 (µ1 r) + A2 e−a J0 (µ2 r) + A3 e−a J0 (µ3 r) + · · ·
where µ1 , µ2 , &c., are the roots of J0 (µc) = 0, and where
1 = A1 J0 (µ1 r) + A2 J0 (µ2 r) + A3 J0 (µ3 r) + · · · ,
c being the radius of the cylinder.
12. Each of the five problems which we have taken up forces upon us the
consideration of the development of a given function in terms of some normal
form, and in two of them the normal form suggested is an unfamiliar function.
It is clear, then, that a complete treatment of our subject will require the inves-
tigation of the properties and relations of certain new and important functions,
as well as the consideration of methods of developing in terms of them.
13. In each of the problems just taken up we have to deal with a homo-
geneous linear partial differential equation involving two independent variables,
and we are content if we can obtain particular solutions. In each case the as-
sumption made in the last problem, that there exists a solution of the equation
in which the dependent variable is the product of two factors each of which in-
volves but one of the independent variables, will reduce the question to solving
two ordinary differential equations which can be treated separately.
If these equations are familiar ones their solutions can be written down at
once; if unfamiliar, the device used in problems 3 and 5 is often serviceable,
namely, that of assuming that the dependent variable can be expressed as a
sum or series of terms involving whole powers of the independent variable, and
then determining the coefficients.
Let us consider again the equations used in the first, second and third prob-
lems.
(a) Dx2 u + Dy2 u = 0 (1)
Assume u = X.Y where X involves x but not y, and Y involves y but not x.
Substitute in (1), Y Dx2 X + XDy2 Y = 0,
or, since we are now dealing with functions of a single variable,

1 d2 X 1 d2 Y
2
+ = 0,
X dx Y dy 2
1 d2 Y 1 d2 X
or 2
=− . (2)
Y dy X dx2
INTRODUCTION. 16
Since the first member of (2) does not contain x, and the second member
does not contain y, and the two members must be identically equal, neither of
them can contain either x or y, and each must be equal to a constant, say α2 .
d2 Y
Then − α2 Y = 0 (3)
dy 2
d2 X
and + α2 X = 0; (4)
dx2
and if (3) and (4) can be solved, we can solve (1). They have for their complete
solutions
Y = Aeαy + Be−αy
and X = C sin αx + D cos αx. (v. Int. Cal. p. 319, § 21.)
Hence Y = eαy and Y = e−αy are particular solutions of (3), X = sin αx and
X = cos αx are particular solutions of (1), and consequently
u = eαy sin αx, u = eαy cos αx, u = e−αy sin αx, and u = e−αy cos αx
are particular solutions of (1). These agree with the results of Art. 7.
(b) Dt2 y = a2 Dx2 y (1)
Assume y = T.X where T is a function of t only and X a function of x only;

substitute in (1) and divide by a2 T X. We get
1 d2 T 1 d2 X
= ; (2)
a2 T
dt 2 X dx2
1 d2 X
hence as in the last case is a constant; call it −α2 , and (2) breaks up
X dx2
into
d2 X
+ α2 X = 0 (3)
dx2
d2 T
+ α2 a2 T = 0. (4)
dt2
The complete solutions of (3) and (4) are
X = A sin αx + B cos αx
and T = C sin αat + D cos αat, (v. Int. Cal. p. 319, § 21).
y = sin αx cos αat, y = sin αx sin αat, y = cos αx cos αat, y = cos αx sin αat
are particular solutions of (1), and agree with the results of Art. 8.
1
(c) rDr2 (rV ) + Dθ (sin θDθ V ) = 0. (1)
sin θ
INTRODUCTION. 17
Assume V = R.Θ where R involves r alone, and Θ involves θ alone; substi-

tute in (1), divide by R.Θ, and transpose; we get
dΘ
2
r d (rR) 1 d sin θ
=− dθ . (2)
R dr2 Θ sin θ dθ
Since by the reasoning used in (a) and (b) each member of (2) must be a
constant, say α2 , we have
d2 (rR)
r = α2 R (3)
dr2
dΘ
1 d sin θ
and dθ + α2 Θ = 0. (4)
sin θ dθ
(3) can be expanded into
d2 R dR
r2 + 2r − α2 R = 0. (5)
dr2 dr
(5) can be solved (v. Int. Cal. p. 321, § 23), and has for its complete solution
R = Arm + Brn ,
q q
where m = − 21 + α2 + 1
4 and n = − 12 − α2 + 41 .
Hence n = −m − 1, and α2 may be written m(m + 1), m being wholly arbitrary;
and
R = Arm + Br−m−1 .
1
R = rm , and R=
rm+1
are, then, particular solutions of
d2 R dR
r2 + 2r − m(m + 1)R = 0. (6)
dr2 dr
With the new value of α2 (4) becomes
dΘ
1 d sin θ
dθ + m(m + 1)Θ = 0. (7)
sin θ dθ
which has been treated in Art. 9 for the case where m is a positive integer, and
the particular solution Θ = Pm (cos θ) has been obtained.
Hence V = rm Pm (cos θ)
1
and V = m+1 Pm (cos θ),
r
m being a positive integer, are particular solutions of (1). The first of these was
obtained in Art. 9, but the second is new and exceedingly important.
INTRODUCTION. 18
14. The method of obtaining a particular solution of an ordinary linear

differential equation, which we have used in Articles 9 and 11, is of very extensive
application, and often leads to the general solution of the equation in question.
As a very simple example, let us take the equation Art. 13 (a) (4), which we
shall write
d2 z
+ α2 z = 0. (1)
dx2
Assume that there isPa solution which can be expressed in terms of powers
of x; that is, let z = an xn , where the coefficients are to be determined.
Substitute this value for z in (1) and we get
[n(n − 1)an xn−2 + α2 an xn ] = 0.

P
Since this equation must be true from its form, without reference to the value
of x, that is, since it must be an identical equation, the coefficient of each power
of x must equal zero, and we have
(n + 1)(n + 2)an+2 + α2 an = 0;
(n + 1)(n + 2)
whence an = − an+2
α2
is the only relation that need hold between the coefficients in order that z =
an xn should be a solution of (1).
P
If n + 2 = 0 or n + 1 = 0, an will be zero and an−2 , an−4 , &c., will be zero.
In the first case the series will begin with a0 , in the second with a1 .
α2
an+2 = − an .
(n + 1)(n + 2)
If we begin with a0 we have
α2 α4 α6
a2 = − a0 , a4 = a0 , a6 = − a0 , &c., · · ·
2! 4! 6!
α 2 x2 α 4 x4 α 6 x6

and z = a0 1− + − + ··· (2)
2! 4! 6!
or z = a0 cos αx (3)
is a particular solution of (1).

If we begin with a1 we have
α2 α4 α6
a3 = − a1 , a5 = a1 , a7 = − a1 , &c., · · ·
3! 5! 7!
α2 x3 α 4 x5 α 6 x7

and z = a1 x − + − + ··· (4)
3! 5! 7!
INTRODUCTION. 19
is a solution of (1); a1 can be taken at pleasure. Let a1 = α, (4) becomes
α 3 x3 α 5 x5 α 7 x7
z = αx − + − + ···
3! 5! 7!
or z = sin αx
which, then, is a particular solution of (1).
z = A sin αx + B cos αx (5)
is, then, a solution of (1), and since it contains two arbitrary constants it is the
general solution.
15. As another example we will take the equation

d2 z dz
x2 + 2x − m(m + 1)z = 0, (1)
dx2 dx
which is in effectPequation (6), Art. 13 (c), and let m be a positive integer.
Assume z = an xn and substitute in (1). We get
[n(n + 1) − m(m + 1)]an xn = 0.

P
This is an identical equation, therefore
[n(n + 1) − m(m + 1)]an = 0.
Hence an = 0 for all values of n except those which make
n(n + 1) − m(m + 1) = 0,
that is, for all values of n except n = m and n = −m − 1. Then
z = Axm + Bx−m−1 (2)
is the general solution of (1) and

1
z = xm and z =
xm+1
are particular solutions. If m is not a positive integer this method will not lead
to a result, and we are driven back to that employed in Art. 13 (c).
16. Let us now take the equation

d 2 dz
(1 − x ) + m(m + 1)z = 0 (1)
dx dx
which is in effect equation (4), Art. 9, and is known as Legendre’s Equation. (1)
may be written
d2 z dz
(1 − x2 ) 2 − 2x + m(m + 1)z = 0. (2)
dx dx
INTRODUCTION. 20
an xn and substitute in (2). We get

P
Assume z =
{n(n − 1)an xn−2 + [m(m + 1) − n(n + 1)]an xn } = 0.

P
Hence (n + 1)(n + 2)an+2 + [m(m + 1) − n(n + 1)]an = 0,
(n + 1)(n + 2)
or an = − an+2 . (3)
m(m + 1) − n(n + 1)
If an = 0, then an−2 = 0, an−4 = 0, &c.; but an = 0 if n = −2 or n = −1. For

the first case we have the sequence of coefficients
m(m + 1)
a2 = − a0
2!
m(m − 2)(m + 1)(m + 3)
a4 = a0
4!
m(m − 2)(m − 4)(m + 1)(m + 3)(m + 5)
a6 = − a0 , &c.
6!
Let us take a0 , which is arbitrary, as 1. Then z = pm (x) where

m(m + 1) 2 m(m − 2)(m + 1)(m + 3) 4
pm (x) = 1 − x + x − ··· (4)
2! 4!
is a solution of Legendre’s Equation if pm (x) is a finite sum or a convergent
series.
For the second case we have the sequence of coefficients
(m − 1)(m + 2)
a3 = − a1
3!
(m − 1)(m − 3)(m + 2)(m + 4)
a5 = a1
5!
(m − 1)(m − 3)(m − 5)(m + 2)(m + 4)(m + 6)
a7 = − a1 , &c.
7!
Let us take a1 , which is arbitrary, as 1. Then z = qm (x) where

(m − 1)(m + 2) 3 (m − 1)(m − 3)(m + 2)(m + 4) 5
qm (x) = x − x + x − · · · (5)
3! 5!
is a solution of Legendre’s Equation if qm (x) is a finite sum or a convergent
series.
If m is a positive even whole number, pm (x) will terminate with the term
containing xm , and is easily seen to be identical with
h m i2
m
2 Γ + 1
m
(−1) 2 2 Pm (x). [v. Art. 9 (9)]
Γ(m + 1)
INTRODUCTION. 21
For all other values of m, pm (x) is a series.

The ratio of the (n + 1)st term of pm (x) to the nth, when m is not a positive
even integer, is
(2n − 2 − m)(2n − 1 + m) 2
x .
(2n − 1)(2n)
Its limiting value, as n is increased, is x2 , and the series is therefore convergent
if −1 < x < 1. It is divergent for all other values of x.
If m is a positive odd whole number qm (x) will terminate with the term
containing xm , and is easily seen to be identical with
h m + 1 i2
m−1
2 Γ
m−1
(−1) 2 2 Pm (x).
Γ(m + 1)
For all other values of m, qm (x) is a series, and can be shown to be convergent
if −1 < x < 1, and divergent for all other values of x.
z = Apm (x) + Bqm (x) (6)
is the general solution of Legendre’s Equation if −1 < x < 1, no matter what

the value of m. From Art. 13 (c) it follows that
V = rm pm (cos θ)



1


V = m+1 pm (cos θ)



r (7)
V = rm qm (cos θ) 


1



V = m+1 qm (cos θ) 
r
are particular solutions of
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0,
sin θ
no matter what the value of m, provided cos θ is neither one nor minus one.
In the work we shall have to do with Laplace’s and Legendre’s Equations, it
is generally possible to restrict m to being a positive integer, and hereafter we
shall usually confine our attention to that case.
With this understanding let us return to (3), which may be rewritten
(m − n)(m + n + 1)
an+2 = − an .
(n + 1)(n + 2)
If an+2 = 0, then an+4 = 0, an+6 = 0, &c.;
but an+2 = 0 if n = m, or n = −m − 1.
INTRODUCTION. 22
If in (3) we begin with n = m − 2, we get the sequence of coefficients already

obtained in Art. 9, and we have z = Pm (x), where

(2m − 1)(2m − 3) · · · 1 m m(m − 1) m−2
Pm (x) = x − x
m! 2(2m − 1)
m(m − 1)(m − 2)(m − 3) m−4
+ x
2.4.(2m − 1)(2m − 3)

m(m − 1)(m − 2)(m − 3)(m − 4)(m − 5) m−6
− x + ··· , (8)
2.4.6.(2m − 1)(2m − 3)(2m − 5)
as a particular solution of Legendre’s Equation.
If, however, we begin with n = −m − 3, we have
(m + 1)(m + 2)
a−m−3 = a−m−1
2(2m + 3)
(m + 1)(m + 2)(m + 3)(m + 4)
a−m−5 = a−m−1
2.4.(2m + 3)(2m + 5)
(m + 1)(m + 2)(m + 3)(m + 4)(m + 5)(m + 6)
a−m−7 = a−m−1 , &c.
2.4.6.(2m + 3)(2m + 5)(2m + 7)
m!
a−m−1 may be taken at pleasure, and is usually taken as ,
1.3.5. · · · (2m + 1)
and z = Qm (x) where

m! 1 (m + 1)(m + 2) 1
Qm (x) = +
(2m + 1)(2m − 1) · · · 1 xm+1 2.(2m + 3) xm+3

(m + 1)(m + 2)(m + 3)(m + 4) 1
+ + ··· (9)
2.4.(2m + 3)(2m + 5) xm+5
is a second particular solution of Legendre’s Equation, provided the series is
convergent. Qm (x) is called a Surface Zonal Harmonic of the second kind. It is
easily seen to be convergent if x < −1 or x > 1, and divergent if −1 < x < 1.
Hence if m is a positive integer,
z = APm (x) + BQm (x) (10)
is the general solution of Legendre’s Equation if x < −1 or x > 1.
We have seen that for −1 < x < 1
m Γ(m + 1)
Pm (x) = (−1) 2 h m i2 pm (x) (11)
2m Γ +1
2
if m is an even integer, and
m−1 Γ(m + 1)
Pm (x) = (−1) 2
h m + 1 i2 qm (x) (12)
2m−1 Γ
2
INTRODUCTION. 23
if m is an odd integer.
If now we define Qm (x) as follows when −1 < x < 1
h m + 1 i2
m−1
2 Γ
m+1
Qm (x) = (−1) 2 2 pm (x) (13)
Γ(m + 1)
if m is an odd integer, and

h m i2
m
2 Γ + 1
m
Qm (x) = (−1) 2 2 qm (x) (14)
Γ(m + 1)
if m is an even integer, then (10) will be the general solution of Legendre’s

Equation if m is a positive integer when −1 < x < 1, as well as when x < −1
or x > 1.
17. Let us last consider the equation
d2 z m2

1 dz
+ + 1− 2 z =0 (1)
dx2 x dx x
which is known as Bessel’s Equation, and which reduces to (8) Art. 11, that is,
to
d2 z 1 dz
+ +z =0
dx2 x dx
when m = 0;9 (1) can be simplified by a change of the dependent variable. Let
z = xm v and we get
d2 v 2m + 1 dv
+ +v =0 (2)
dx2 x dx
to determine v. P
Assume v = an xn , and substitute in (2). We get
[n(2m + n)an xn−2 + an xn ] = 0;

P
whence an−2 = −n(2m + n)an .
If we begin with n = 0, then an−2 = 0, an−4 = 0, &c., and we have the set of
values
a0 a0
a2 = − =− 2
2(2m + 2) 2 (m + 1)
a0 a0
a4 = = 4
2.4(2m + 2)(2m + 4) 2 .2!(m + 1)(m + 2)
9 This equation was first studied by Fourier in considering the cooling of a cylinder. We
shall designate it as “Fourier’s Equation.”

INTRODUCTION. 24
a0 a0
a6 = − =− 6 ;
2.4.6(2m + 2)(2m + 4)(2m + 6) 2 .3!(m + 1)(m + 2)(m + 3)
x2 x4

whence z = a0 xm 1 − + 4
22 (m + 1) 2 .2!(m + 1)(m + 2)
x6

− 6 + ··· (3)
2 .3!(m + 1)(m + 2)(m + 3)
1
is a solution of Bessel’s Equation. a0 is usually taken as m if m is a positive
2 m!
1
integer, or as m if m is unrestricted in value, and the second member
2 Γ(m + 1)
of (3) is represented by Jm (x) and is called a Bessel’s Function of the mth order,
or a Cylindrical Harmonic of the mth order.
If m = 0, Jm (x) becomes J0 (x) and is the value of z obtained in Art. 11 as
the solution of equation (8) of that article.
If in equation (1) we substitute x−m v in place of xm v for z, we get in place
of (2) the equation
d2 v 1 − 2m dv
+ +v =0
dx2 x dx
and in place of (3)
x2 x4

−m
z = a0 x 1− 2 + 4
2 (1 − m) 2 .2!(1 − m)(2 − m)
x6

− 6 + ··· (4)
2 .3!(1 − m)(2 − m)(3 − m)
1
If a0 is taken equal to the second member of (4) is the same
2−m Γ(1 − m)
function of −m and x that Jm (x) is of +m and x and may be written J−m (x).
Therefore z = AJm (x) + BJ−m (x) (5)
is the general solution of (1) unless Jm (x) and J−m (x) should prove not to be
independent.
It is easily seen that when m = 0, J−m (x) and Jm (x) become identical and
(5) reduces to
z = (A + B)J0 (x)
and contains but a single arbitrary constant and is not the general solution of
Fourier’s Equation (8) Art. (11).
It can be shown that J−m (x) = (−1)m Jm (x) whenever m is an integer, and
consequently that the solution (5) is general only when m if real is fractional or
incommensurable.
INTRODUCTION. 25
The general solution for the important case where m = 0 is, however, easily
obtained. Let F (m, x) be the value which the second member of (3) assumes
when a0 = 1; then the value which the second member of (4) assumes when
a0 = 1 will be F (−m, x), and it has been shown that z = F (m, x) and z =
F (−m, x) are solutions of Bessel’s Equation; z = F (m, x) − F (−m, x) is, then,
a solution, as is also
F (m, x) − F (−m, x)
z= , (6)
2m
F (m, x) − F (−m, x)
but the limiting value which approaches as m approaches
2m
zero is [Dm F (m, x)]m=0 and consequently
z = [Dm F (m, x)]m=0 (7)
is a solution of the equation
d2 z 1 dz
2
+ + z = 0, (8)
dx x dx
and the general solution of (8) is
z = AJ0 (x) + B[Dm F (m, x)]m=0 .
x2 x4

F (m, x) = xm 1 − 2 + 4
2 (m + 1) 2 .2!(m + 1)(m + 2)
x6

− 6 + ···
2 .3!(m + 1)(m + 2)(m + 3)
x2 x4

m
Dm F (m, x) = x log x 1 − 2 + − ···
2 (m + 1) 24 .2!(m + 1)(m + 2)
x2 x4

m
+x Dm 1 − 2 + + ··· .
2 (m + 1) 24 .2!(m + 1)(m + 2)
The general term of the last parenthesis can be written
x2k
(−1)k ,
22k .k!(m + 1)(m + 2) · · · (m + k)
and its partial derivative with respect to m is
x2k 1
(−1)k Dm .
22k .k! (m + 1)(m + 2) · · · (m + k)
1
log = −[ log(m + 1) + log(m + 2) + · · ·
(m + 1)(m + 2) · · · (m + k)
+ log(m + k)].
INTRODUCTION. 26
Take the Dm of both members and we have

1
Dm
(m + 1)(m + 2) · · · (m + k)

1 1 1 1
=− + + ··· .
(m + 1)(m + 2) · · · (m + k) m + 1 m + 2 m+k
x2 x4 x6

Dm 1 − 2 + 4 − 6
2 (m + 1) 2 .2!(m + 1)(m + 2) 2 .3!(m + 1)(m + 2)(m + 3)
x2 x4

1 1 1 1
+ ··· = 2 − 4 +
2 (m + 1)2 2 .2! (m + 1)(m + 2) m + 1 m + 2
x6

1 1 1 1
+ 6 + + + ···
2 .3! (m + 1)(m + 2)(m + 3) m + 1 m + 2 m + 3
and we have
x2 1 x4

1 1
[Dm F (m, x)]m=0 = J0 (x) log x + 2 − +
2 (1!)2 1 24 (2!)2 1 2
x6 x8

1 1 1 1 1 1 1
+ 6 + + − + + + + ··· ;
2 (3!)2 1 2 3 28 (4!)2 1 2 3 4
and z = AJ0 (x) + BK0 (x), (9)

2 4

x x 1 1
where K0 (x) = J0 (x) log x + 2 − 4 2
+
2 2 (2!) 1 2
x6 x8

1 1 1 1 1 1 1
+ 6 + + − + + + + · · · (10)
2 (3!)2 1 2 3 28 (4!)2 1 2 3 4
is the general solution of Fourier’s Equation (8).
K0 (x) is known as a Bessel’s Function of the Second Kind.
18. It is worth while to confirm the results of the last few articles by
getting the general solutions of the equations in question by a different and
familiar method.
The general solution of any ordinary linear differential equation of the second
order can be obtained when a particular solution of the equation has been found
[v. Int. Cal. p. 321, § 24 (a)].
The most general form of a homogeneous ordinary linear differential equation
of the second order is
d2 y dy
2
+P + Qy = 0 (1)
dx dx
where P and Q are functions of x. Suppose that
y=v (2)
is a particular solution of (1). Substitute y = vz in (1) and we get
d2 z

dv dz
v 2+ 2 + Pv = 0. (3)
dx dx dx
INTRODUCTION. 27
dz
Call = z 0 . Then (3) becomes
dx
dz 0

dv
v + 2 + P v z 0 = 0, (4)
dx dx
a differential equation of the first order in which the variables can be separated.
Multiply by dx and divide by vz 0 and (4) reduces to
dz 0 dv
0
+2 + P dx = 0.
z v
Integrate and we have
w
log z 0 + log v 2 + P dx = C
r r
or z0 v2 = e C− P dx
= Be− P dx
,
r
dz e− P dx
z0 = =B ,
dx vr2
w e− P dx
z =A+B dx;
vr2
w e− P dx
and y = v A+B dx (5)
v2
is the general solution of (1), the only arbitrary constants in the second member
of (5) being those explicitly written, namely, A and B.
(a) Apply this formula to (1) Art. 14,
d2 z
+ α2 z = 0; (1)
dx2
given: z = cos αx, as a particular solution. Substituting in (5) we have since
P =0
w dx

z = cos αx A + B
cos2 αx

B
= cos αx A + tan αx
α
= A cos αx + B1 sin αx, (2)
as the general solution of (1), and this agrees perfectly with (5) Art. 14.
(b) Take equation (1) Art. 15.
d2 z dz
x2 + 2x − m(m + 1)z = 0; (1)
dx2 dx
given: z = xm , as a particular solution.
INTRODUCTION. 28
2 w r 1
Here P = , P dx = 2 log x = log x2 , and e− P dx = 2 . Hence by (5)
x x
w dx
B 1

z = xm A + B = x m
A + ,
x2m+2 −2m − 1 x2m+1
B1
that is z = Axm + m+1 (2)
x
is the general solution of (1), and agrees with (2) Art. 15.
(c) Take Legendre’s Equation, (2) Art. 16.

d2 z dz
(1 − x2 ) 2
− 2x + m(m + 1)z = 0; (1)
dx dx
given: z = Pm (x), as a particular solution.
−2x w r 1
Here P = 2
, P dx = log(1 − x2 ), and e− P dx = .
1−x 1 − x2
w dx

Hence by (5) z = Pm (x) A + B (2)
(1 − x2 )[Pm (x)]2
is the general solution of (1) and must agree with (10) Art. 16, if m is an integer,
and therefore w dx
Qm (x) = CPm (x) (3)
(1 − x )[Pm (x)]2
2
where C is as yet undetermined, and no constant term is to be understood with

the integral in the second member.
(d ) Take Bessel’s Equation, (1) Art. 17.

d2 z m2

1 dz
+ + 1 − z = 0; (1)
dx2 x dx x2
given: z = Jm (x), as a particular solution.
1 w r 1
Here P = , P dx = log x, and e− P dx = . Hence by (5)
x x
w dx

z = Jm (x) A + B (2)
x[Jm (x)]2
is the general solution of Bessel’s Equation.
If m = 0 (2) becomes
w dx

z = J0 (x) A + B (3)
x[J0 (x)]2
and must agree with (9) Art. 17. Therefore
w dx
K0 (x) = CJ0 (x) , (4)
x[J0 (x)]2
INTRODUCTION. 29
where C is at present undetermined, and no constant term is to be taken with

the integral.
The first considerable subject suggested by the problems which we have
taken up in this introductory chapter is that of development in Trigonometric
Series (v. Arts. 7 and 8).
30
CHAPTER II.
DEVELOPMENT IN TRIGONOMETRIC SERIES.
19. We have seen in Chapter I. that it is sometimes important to be able

to express a given function of a variable x, in terms of the sines or of the cosines
of multiples of x. The problem in its general form was first solved by Fourier in
his “Analytic Theory of Heat” (1822), and its solution plays a very important
part in most branches of modern Physics. Series involving only sines and cosines
of whole multiples of x, that is series of the form
b0 + b1 cos x + b2 cos 2x + · · · + a1 sin x + a2 sin 2x + · · ·
are generally known as Fourier’s series.

Let us endeavor to develop a given function of x in terms of sin x, sin 2x,
sin 3x, &c., in such a way that the function and the series shall be equal for all
values of x between x = 0 and x = π.
To fix our ideas let us suppose that we have a curve,
y = f (x),
given, and that we wish to form the equation,
y = a1 sin x + a2 sin 2x + a3 sin 3x + · · · ,
of a curve which shall coincide with so much of the given curve as lies between
the points corresponding to x = 0 and x = π. It is clear that in the equation
y = a1 sin x (1)
a1 may be determined so that the curve represented shall pass through any
given point. For if we substitute in (1) the coördinates of the point in question
we shall have an equation of the first degree in which a1 is the only unknown
quantity and which will therefore give us one and only one value for a1 .
In like manner the curve
y = a1 sin x + a2 sin 2x
may be made to pass through any two arbitrarily chosen points whose abscissas
lie between 0 and π provided that the abscissas are not equal; and
y = a1 sin x + a2 sin 2x + a3 sin 3x + · · · + an sin nx
may be made to pass through any n arbitrarily chosen points whose abscissas
lie between 0 and π provided as before that their abscissas are all different.
If, then, the given function f (x) is of such a character that for each value
of x between x = 0 and x = π it has one and only one value, and if between
DEVELOPMENT IN TRIGONOMETRIC SERIES. 31
x = 0 and x = π it is finite and continuous, or if discontinuous has only finite

discontinuities (v. Int. Cal. Art. 83, p. 78), the coefficients in
y = a1 sin x + a2 sin 2x + a3 sin 3x + · · · + an sin nx (2)
can be determined so that the curve represented by (2) will pass through any n
arbitrarily chosen points of the curve
y = f (x) (3)
whose abscissas lie between 0 and π and are all different, and these coefficients
will have but one set of values.
For the sake of simplicity suppose that the n points are so chosen that their
projections on the axis of X are equidistant.
π
Call = ∆x; then the coördinates of the n points will be [∆x, f (∆x)],
n+1
[2∆x, f (2∆x)], [3∆x, f (3∆x)], · · · [n∆x, f (n∆x)]. Substitute them in (2) and
we have
f (∆x) = a1 sin ∆x + a2 sin 2∆x + a3 sin 3∆x + · · · + an sin n∆x 


f (2∆x) = a1 sin 2∆x + a2 sin 4∆x + a3 sin 6∆x + · · · + an sin 2n∆x 




f (3∆x) = a1 sin 3∆x + a2 sin 6∆x + a3 sin 9∆x + · · · + an sin 3n∆x

(4)
.. .. .. .. .. 

. . . . .





2
f (n∆x) = a1 sin n∆x + a2 sin 2n∆x + a3 sin 3n∆x + · · · + an sin n ∆x,

n equations of the first degree to determine the n coefficients a1 , a2 , a3 , · · · an .

Not only can equations (4) be solved in theory, but they can be actually
solved in any given case by a very simple and ingenious method due to Lagrange.
Let us take as an example the simple problem to determine the coefficients
a1 , a2 , a3 , a4 , and a5 , so that
y = a1 sin x + a2 sin 2x + a3 sin 3x + a4 sin 4x + a5 sin 5x (5)
shall pass through the five points of the line
y=x
π 2π 3π 4π 5π π
which have the abscissas , , , , and , here being ∆x.
6 6 6 6 6 6
We must now solve the equations
π π 2π 3π 4π 5π 

= a1 sin + a2 sin + a3 sin + a4 sin + a5 sin 
6 6 6 6 6 6 

2π 2π 4π 6π 8π 10π 

= a1 sin + a2 sin + a3 sin + a4 sin + a5 sin



6 6 6 6 6 6 

3π 3π 6π 9π 12π 15π

= a1 sin + a2 sin + a3 sin + a4 sin + a5 sin (6)
6 6 6 6 6 6 
4π 4π 8π 12π 16π 20π 

= a1 sin + a2 sin + a3 sin + a4 sin + a5 sin



6 6 6 6 6 6 

5π 5π 10π 15π 20π 25π 

= a1 sin + a2 sin + a3 sin + a4 sin + a5 sin .

6 6 6 6 6 6
π 2π
Multiply the first equation by 2 sin , the second by 2 sin , the third by
6 6
3π 4π 5π
2 sin , the fourth by 2 sin , the fifth by 2 sin and add the equations.
6 6 6
The coefficient of a2 is
π 2π 2π 4π 3π 6π 4π 8π
2 sin sin + 2 sin sin + 2 sin sin + 2 sin sin
6 6 6 6 6 6 6 6
5π 10π
+ 2 sin sin ;
6 6
π 2π π 3π
but 2 sin sin = cos − cos , &c.
6 6 6 6
Hence the coefficient of a2 becomes
π 2π 3π 4π 5π 

cos + cos + cos + cos + cos 
6 6 6 6 6 (7)
3π 6π 9π 12π 15π 
− cos − cos − cos − cos − cos 
6 6 6 6 6
and this may be reduced by the aid of an important Trigonometric formula
which we proceed to establish.
20. Lemma.
θ
1 1 sin(2n + 1) 2
cos θ + cos 2θ + cos 3θ + · · · + cos nθ = − + . (1)
2 2 θ
sin
2
For let S = cos θ + cos 2θ + cos 3θ + · · · + cos nθ and multiply by 2 cos θ.
2S cos θ = 2 cos2 θ + 2 cos θ cos 2θ + 2 cos θ cos 3θ + · · · + 2 cos θ cos nθ

= 1 + cos θ + cos 2θ + · · · + cos(n − 1)θ
+ cos 2θ + cos 3θ + cos 4θ + · · · + cos(n + 1)θ
= 2S + 1 + cos(n + 1)θ − cos θ − cos nθ. Hence
1 cos nθ − cos(n + 1)θ

S=− +
2 2(1 − cos θ)
θ
1 1 sin(2n + 1) 2
or S=− + . Q.E.D.
2 2 θ
sin
2
21. Applying (1) Art. 20 to (7) Art. 19 the coefficient of a2 reduces to

11π 33π
sin sin
12 − 12 ;
π 3π
2 sin 2 sin
12 12
11π π 33π 3π
but = π − , and = 3π − ;
12 12 12 12
π 3π
sin π − sin 3π −
therefore 12 − 12 = 1 − 1 = 0,
π 3π 2 2
2 sin 2 sin
12 12
and a2 vanishes.
In like manner it may be shown that the coefficients of a3 , a4 , and a5 vanish.
The coefficient of a1 is
π 2π 3π 4π 5π
2 sin2 + 2 sin2 + 2 sin2 + 2 sin2 + 2 sin2
6 6 6 6 6
=1 + 1 + 1 + 1 + 1
2π 4π 6π 8π 10π
− cos − cos − cos − cos − cos
6 6 6 6 6
11π π
1 sin sin 2π −
=5+ − 6 = 51 − 6 = 6.
2 2
π π
2 sin 2 sin
6 6
The first member of the final equation is
2π π 2π 2π 3π 3π 4π 4π 5π 5π
sin + 2 sin +2 sin +2 sin +2 sin . Hence
6 6 6 6 6 6 6 6 6 6
k=5
2 X kπ kπ π √
a1 = sin = (2 + 3) = 2 approximately.
6 6 6 6
k=1
2π
If we multiply the first equation of (6) Art. 19 by 2 sin , the second by
6
4π 6π 8π 10π
2 sin , the third by 2 sin , the fourth by 2 sin , the fifth by 2 sin , add
6 6 6 6
and reduce as before we shall find
k=5
2 X kπ 2kπ π√
a2 = sin =− 3 = −0.9;
6 6 6 6
k=1
and in like manner we get

k=5
2 X kπ 3kπ π
a3 = sin = = 0.5
6 6 6 6
k=1
k=5 √
2 X kπ 4kπ π 3
a4 = sin =− = −0.3
6 6 6 18
k=1
k=5
2 X kπ 5kπ π √
a5 = sin = (2 − 3) = 0.1.
6 6 6 6
k=1
Therefore
y = 2 sin x − 0.9 sin 2x + 0.5 sin 3x − 0.3 sin 4x + 0.1 sin 5x (1)
π 2π 3π 4π
cuts the curve y = x at the five points whose abscissas are , , , , and
6 6 6 6
5π
.
6
22. The equations (4) Art. 19 can be solved by exactly the same device.
To find any coefficient am multiply the first equation by 2 sin m∆x, the second
by 2 sin 2m∆x, the third by 2 sin 3m∆x, &c. and add.
The coefficient of any other a as ak in the resulting equation will be
2 sin k∆x sin m∆x + 2 sin 2k∆x sin 2m∆x + 2 sin 3k∆x sin 3m∆x + · · ·
+ 2 sin nk∆x sin nm∆x
= cos(m − k)∆x + cos 2(m − k)∆x + cos 3(m − k)∆x + · · · + cos n(m − k)∆x
− cos(m + k)∆x − cos 2(m + k)∆x − cos 3(m + k)∆x − · · · − cos n(m + k)∆x
2n + 1 2n + 1
sin (m − k)∆x sin (m + k)∆x
= 2 − 2 ; by (1) Art. 20.
(m − k)∆x (m + k)∆x
2 sin 2 sin
2 2
2n + 1 1
=n+1− and (n + 1)∆x = π.
2 2
Hence the coefficient of ak may be written
h (m − k)∆x i h (m + k)∆x i
sin (m − k)π − sin (m + k)π −
2 − 2
(m − k)∆x (m + k)∆x
2 sin 2 sin
2 2
1 1 1 1
but this is equal to − or − + according as m − k is odd or even and so
2 2 2 2
is zero in either case.
The coefficient of am will be
2 sin2 m∆x + 2 sin2 2m∆x + 2 sin2 3m∆x + · · · + 2 sin2 nm∆x

= 1 + 1 + 1 + ··· + 1
− cos 2m∆x − cos 4m∆x − cos 6m∆x − · · · − cos 2nm∆x
1 sin(2n + 1)m∆x
=n+ − , by (1) Art. 20.
2 2 sin m∆x
But (2n + 1)m∆x = 2m(n + 1)∆x − m∆x = 2mπ − m∆x,

sin(2n + 1)m∆x sin(2mπ − m∆x) 1
therefore = =− ,
2 sin m∆x 2 sin m∆x 2
and the coefficient of am is n + 1.

The first member of our final equation will be
k=n
X
2 f (k∆x) sin km∆x.
k=1
Hence
k=n
2 X
am = f (k∆x) sin km∆x, (1)
n+1
k=1
and the curve

y = a1 sin x + a2 sin 2x + · · · + an sin nx, (2)
where the coefficients are given by (1) will pass through the n points of the
π
curve y = f (x) whose abscissas are ∆x, 2∆x, 3∆x, · · · n∆x. ∆x being .
n+1
It should be noted that since the n equations (4) Art. 19 are all of the first
degree there will exist only one set of values for the n quantities a1 , a2 , a3 , · · ·
an that can satisfy these equations. Consequently the solution which we have
obtained is the only solution possible.
23. The result just obtained obviously holds good no matter how great a
value of n may be taken.
If now we suppose n indefinitely increased the two curves (2) Art. 22 and
y = f (x) will come nearer and nearer to coinciding throughout the whole of
their portions between x = 0 and x = π, and consequently the limiting form
that equation (2) Art. 22 approaches as n is indefinitely increased will represent
a curve absolutely coinciding between the values of x in question with y = f (x).
Let us see what limiting value am approaches as n is indefinitely increased.
k=n
2 X
am = f (k∆x) sin km∆x (1) Art. 22.
n+1
k=1
k=n
2∆x X
= f (k∆x) sin km∆x
π
k=1
" #
2 f (∆x) sin m∆x.∆x + f (2∆x) sin 2m∆x.∆x + · · ·
=
π + f (n∆x) sin nm∆x.∆x
" #
2 f (∆x) sin m∆x.∆x + f (2∆x) sin 2m∆x.∆x + · · ·
=
π +f (π − ∆x) sin m(π − ∆x).∆x
π
since ∆x = .
n+1
As n is increased indefinitely ∆x approaches zero as a limit. Hence the

limiting value of am as n increases indefinitely is
" #
2 f (∆x) sin m∆x.∆x + f (2∆x) sin 2m∆x.∆x + · · · 1
limit
.
π ∆x=0 +f (π − ∆x) sin m(π − ∆x).∆x
2w
π
= f (x) sin mx.dx. [v. Int. Cal. Arts. 80, 81.]
π
0
Hence f (x) = a1 sin x + a2 sin 2x + a3 sin 3x + · · · , (2)
where any coefficient am is given by the formula
2w
π
am = f (x) sin mx.dx, (3)
π
0
is a true development of f (x) for all values of x between x = 0 and x = π

provided that the series (2) is convergent, for it is in that case only that we
can assume that the limiting value of the second member of (2) Art. 22 can be
obtained by adding the limiting values of the several terms.
When x = 0 and when x = π every term in the second member of (2) is
zero, and the second member is zero and will not be equal to f (x) unless f (x)
is itself zero when x = 0 and x = π; but even when f (x) is not zero for x = 0
and x = π the development given above holds good for any value of x between
zero and π no matter how near it may be taken to either of these values.
24. Instead of actually performing the elimination in equations (4) Art. 19

and getting a formula for am in terms of n, and then letting n increase indefi-
nitely, we might have saved labor by the following method.
Return to equations (4) Art. 19 and multiply the first by ∆x sin m∆x, the
second by ∆x sin 2m∆x, and so on, that is multiply each equation by ∆x times
the coefficient of am in that equation, and then add the equations.
We get as the coefficient of ak
sin k∆x sin m∆x.∆x + sin 2k∆x sin 2m∆x.∆x + · · · + sin nk∆x sin nm∆x.∆x.
Let us find its limiting value as n is indefinitely increased. It may be written,

since (n + 1)∆x = π,
" #
sin k∆x sin m∆x.∆x + sin 2k∆x sin 2m∆x.∆x + · · ·
limit
.
∆x=0 + sink(π − ∆x) sin m(π − ∆x).∆x
wπ
= sin kx sin mx.dx;
0
1 We . .
shall use the sign = for approaches. ∆x = 0 is read ∆x approaches zero.
wπ wπ
1
but sin kx sin mx.dx = 2 [cos(m − k)x − cos(m + k)x]dx
0 0
= 0 if m and k are not equal.
The coefficient of am is
∆x(sin2 m∆x + sin2 2m∆x + sin2 3m∆x + · · · + sin2 nm∆x).
Its limiting value

2 2 2
limit
. sin m∆x.∆x + sin 2m∆x.∆x + · · · + sin m(π − ∆x)∆x
∆x=0
wπ π
= sin2 mx.dx = .
2
0
The first member is
f (∆x) sin m∆x.∆x + f (2∆x) sin 2m∆x.∆x + · · · + f (n∆x) sin mn∆x.∆x
and its limiting value is

wπ
f (x) sin mx.dx.
0
Hence the limiting form approached by the final equation as n is increased

is
wπ π
f (x) sin mx.dx = am .
2
0
2w
π
Whence am = f (x) sin mx.dx as before.
π
0
This method is practically the same as multiplying the equation
f (x) = a1 sin x + a2 sin 2x + a3 sin 3x + · · · (1)
by sin mx.dx and integrating both members from zero to π.

It is exceedingly important to realize that the short method of determining
any coefficient am of the series (1) which has just been described in the italicized
paragraph, is essentially the same as that of obtaining am by actual elimination
from the equations (4) Art. 19, and then supposing n to increase indefinitely,
thus making the curves (3) Art. 19 and (2) Art. 19 absolutely coincide between
the values of x which are taken as the limits of the definite integration.
25. We see, then, that any function of x which is single-valued, finite,

and continuous between x = 0 and x = π, or if discontinuous has only finite
discontinuities each of which is preceded and succeeded by continuous portions,
can probably be developed into a series of the form
f (x) = a1 sin x + a2 sin 2x + a3 sin 3x + · · · (1)
2w 2w
π π
where am = f (x) sin mx.dx = f (α) sin mα.dα; (2)
π π
0 0
and the series and the function will be identical for all values of x between x = 0
and x = π, not including the values x = 0 and x = π unless the given function
is equal to zero for those values.
An elaborate investigation of the question of the convergence of the series
(1), for which we have not space, entirely confirms the result formulated above2
and shows in addition that at a point of finite discontinuity the series has a
value equal to half the sum of the two values which the function approaches as
we approach the point in question from opposite sides.
The investigation which we have made in the preceding sections establishes
the fact that the curve represented by y = f (x) need not follow the same
mathematical law throughout its length, but may be made up of portions of
entirely different curves. For example, a broken line or a locus consisting of
finite parts of several different and disconnected straight lines can be represented
perfectly well by y = a sine series.
26. Let us obtain a few sine developments.
(a) Let f (x) = x. (1)
We have x = a1 sin x + a2 sin 2x + a3 sin 3x + · · · (2)

2w
π
where am = x sin mx.dx (3)
π
0
w 1
x sin mx.dx = (sin mx − mx cos mx),
m2
wπ (−1)m π
x sin mx.dx = − ,
m
0

sin x sin 2x sin 3x sin 4x
and x=2 − + − + ··· (4)
1 2 3 4
2 Provided the function has not an infinite number of maxima and minima in the neighbor-
hood of a point. v. Arts. 37–38.

(b) Let f (x) = 1. (1)
2w
π
am = sin mx.dx; (2)
π
0
w cos mx
sin mx.dx = − ,
m
wπ 1 1
sin mx.dx = (1 − cos mπ) = [1 − (−1)m ]
m m
0
= 0 if m is even
2
= if m is odd.
m

4 sin x sin 3x sin 5x sin 7x
Hence 1= + + + + ··· . (3)
π 1 3 5 7
It is to be noticed that (3) gives at once a sine development for any constant
c. It is,
4c sin x sin 3x sin 5x
c= + + + ··· . (4)
π 1 3 5
π
If we substitute x = in (4) (a) or (3) (b) we get a familiar result, namely
2
π 1 1 1 1
= − + − + ··· , (5)
4 1 3 5 7
a formula usually derived by substituting x = 1 in the power series for tan−1 x.
(v. Dif. Cal. Art. 135.)
(4) (a) does not hold good when x = π, and (3) (b) fails when x = 0 and
when x = π, for in all these cases the series reduces to zero.
π
(c) Let f (x) = x from x = 0 to x = and
2
π
f (x) = π − x from x = to x = π. That is, let
2
y = f (x) represent the broken line in the figure.
As the mathematical expression for f (x) is
different in the two halves of the curve we must
break up
π
wπ w2 wπ
f (x) sin mx.dx into f (x) sin mx.dx + f (x) sin mx.dx.
0 0 π
2
We have, then,
π
2w 2w
2 π
am = x sin mx.dx + (π − x) sin mx.dx (1)
π ππ
0 2
4 π
= 2 sin m .
m π 2
π
But sin m =1 if m=1 or 4k + 1
2
=0 “ m=2 “ 4k + 2
= −1 “ m=3 “ 4k + 3
=0 “ m=4 “ 4k.
Hence if y = f (x) represents our broken line,

4 sin x sin 3x sin 5x sin 7x
f (x) = − + − + · · · . (2)
π 12 32 52 72
π π
When x = f (x) = and we have
2 2
π2 1 1 1 1
= 2 + 2 + 2 + 2 + ··· (3)
8 1 3 5 7
(d ) As a case where the function has a finite discontinuity, let
π
f (x) = 1 from x=0 to x = and
2
π
f (x) = 0 “ x= “ x = π.
2
y = f (x) will in this case represent the locus in the figure.
As before
π
wπ w2
f (x) sin mx.dx = f (x) sin mx.dx
0 0
wπ
+ f (x) sin mx.dx.
π
2
2w 2w
2 π
am = sin mx.dx + 0. sin mx.dx. (1)
π ππ
0 2
2w
2
2 1 π
am = sin mx.dx = 1 − cos m .
π πm 2
0
π
But cos m = 0 if m=1 or 4k + 1
2
= −1 “ m=2 “ 4k + 2
= 0 “ m=3 “ 4k + 3
= 1 “ m=4 “ 4k.
Hence

2 sin x 2 sin 2x sin 3x sin 5x 2 sin 6x sin 7x
f (x) = + + + + + + ··· . (2)
π 1 2 3 5 6 7
π 1
If x = the second member of (2) reduces to , for
2 2

2 1 1 1 1 1
− + − + ··· = by (5) (b);
π 1 3 5 7 2
and we see that the series represents the function completely for all values of
π
x between x = 0 and x = π except for x = and there it has a value which
2
π
is the mean of the values approached by the function as x approaches from
2
opposite sides.
EXAMPLES.
Obtain the following developments:—

2
π2
2
π2

2 π 4 π 4
(1) x2 = − 3 sin x − sin 2x + − 3 sin 3x − sin 4x
π 1 1 2 3 3 4
2
π 4
+ − 3 sin 5x − · · · .
5 5
3 3 3
3 2 π 6π π 6π π 6π
(2) x = − 3 sin x − − 3 sin 2x + − 3 sin 3x
π 1 1 2 2 3 3
3
π 6π
− − 3 sin 4x + · · · .
4 4

2 sin x π sin 3x 2π sin 5x
(3) f (x) = + 2 sin 2x − − 2 sin 4x +
π 12 2 32 4 52

3π
+ 2 sin 6x − · · · ,
6
π π
if f (x) = x from x = 0 to x = and f (x) = 0 from x = to x = π.
2 2

2 sin x 2 sin 2x 3 sin 3x 4 sin 4x
(4) sin µx = sin µπ 2 − + − + · · ·
π 1 − µ2 22 − µ2 32 − µ2 42 − µ2
if µ is a fraction.

2 1 2 3
(5) ex = (1 + eπ ) sin x + (1 − eπ ) sin 2x + (1 + eπ ) sin 3x
π 2 5 10

4
+ (1 − eπ ) sin 4x + · · · .
17

2 sinh π 1 2 3 4
(6) sinh x = sin x − sin 2x + sin 3x − sin 4x + · · · .
π 2 5 10 17

2 1 2
(7) cosh x = (1 + cosh π) sin x + (1 − cosh π) sin 2x
π 2 5

3
+ (1 + cosh π) sin 3x + · · · .
10
27. Let us now try to develop a given function of x in a series of cosines.

As before suppose that f (x) has a single value for each value of x between
x = 0 and x = π, that it does not become infinite between x = 0 and x = π,
and that if discontinuous it has only finite discontinuities.
Assume
f (x) = b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · · (1)
To determine any coefficient bm multiply (1) by cos mx.dx and integrate each
term from 0 to π.
wπ
b0 cos mx.dx = 0.
0
wπ bk w
π
bk cos kx cos mx.dx = [cos(m − k)x + cos(m + k)x]dx
2
0 0
= 0 if m and k are not equal.
w bm
bm cos2 mx.dx = (mx + cos mx sin mx),
2m
wπ π
bm cos2 mx.dx = bm , if m is not zero.
2
0
2w 2w
π π
Hence bm = f (x) cos mx.dx = f (α) cos mα.dα, (2)
π π
0 0
if m is not zero.
To get b0 multiply (1) by dx and integrate from zero to π.

wπ
b0 dx = b0 π,
0
wπ
bk cos kx.dx = 0.
0
1 wπ 1w
π
Hence b0 = f (x)dx = f (α)dα, (3)
π π
0 0
which is just half the value that would be given by formula (2) if zero were
substituted for m.
To save a separate formula (1) is usually written
f (x) = 21 b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · · (4)
and then the formula
2w 2w
π π
bm = f (x) cos mx.dx = f (α) cos mα.dα (2)
π π
0 0
will give b0 as well as the other coefficients.

It is important to see clearly that what we have just done in determining
the coefficients of (1) is equivalent to taking n + 1 terms of (4), substituting in
y = 21 b0 + b1 cos x + b2 cos 2x + · · · + bn cos nx (5)
in turn the coördinates of the n + 1 points of the curve
y = f (x)
whose projections on the axis of X are equidistant, determining b0 , b1 , b2 , · · · bn
by elimination from the n + 1 resulting equations, and then taking the limiting
values they approach as n is indefinitely increased. (v. Art. 24.)
π
If ∆x = the abscissas of the n + 1 points used are 0, ∆x, 2∆x, 3∆x,
n+1
· · · n∆x, so that we should expect our cosine development to hold for x = 0 as
well as for values of x between zero and π.
28. Let us take one or two examples:

(a) Let f (x) = x. (1)
2w
π
2 π2
b0 = x dx = = π.
π π 2
0
2w
π
2 2
bm = x cos mx.dx = 2 (cos mπ − 1) = 2 [(−1)m − 1].
π m π m π
0

π 4 cos 3x cos 5x cos 7x
Hence x= − cos x + + + + · · · . (2)
2 π 32 52 72
(2) holds good not only for values of x between zero and π but for x = 0
and x = π as well, since for these values we have

π 4 1 1 1
0= − 1 + 2 + 2 + 2 + ··· (3)
2 π 3 5 7

π 4 1 1 1
and π= + 1 + 2 + 2 + 2 + ··· (4)
2 π 3 5 7
which are true by Art. 26 (c)(3).
(b) Let f (x) = x sin x. (1)
2w
π
2
b0 = x sin x.dx = π = 2,
π π
0
2w 1w
π π
1
b1 = x sin x cos x.dx = x sin 2x.dx = − ,
π π 2
0 0
2w 1w
π π
bm = x sin x cos mx.dx = [x sin(m + 1)x − x sin(m − 1)x]dx
π π
0 0
2
= if m is odd
(m − 1)(m + 1)
2
=− if m is even.
(m − 1)(m + 1)
Hence
cos x 2 cos 2x 2 cos 3x 2 cos 4x
x sin x = 1 − − + − + ··· (2)
2 1.3 2.4 3.5
π
If x = we have
2
π 1 1 1 1
= + − + − ··· . (3)
4 2 1.3 3.5 5.7
EXAMPLES.
Obtain the following developments:

π 2 cos 2x cos 6x cos 10x cos 14x
(1) f (x) = − + + + + · · ·
4 π 12 32 52 72
π π
if f (x) = x from x = 0 to x = and f (x) = π − x from x = to x = π.
2 2
1 2 cos x cos 3x cos 5x cos 7x
(2) f (x) = + − + − + ··· ,
2 π 1 3 5 7
π π
if f (x) = 1 from x = 0 to x = and f (x) = 0 from x = to x = π.
2 2
π2

2 cos x cos 2x cos 3x cos 4x
(3) x = −4 − + − + ··· ,
3 12 22 32 42
π3
2
π2
2
6 π 4 π 4
(4) x3 = − − cos x − cos 2x + − cos 3x
4 π 12 14 22 32 34
π2
2
π 4
− 2 cos 4x + − cos 5x − · · · ,
4 52 54

π 2 π 2 1 3π
(5) f (x) = + − 1 cos x − 2 cos 2x − 2 + 1 cos 3x
8 π 2 2 3 2

1 5π 2
+ 2 − 1 cos 5x − 2 cos 6x − · · · ,
5 2 6
π π
if f (x) = x from x = 0 to x = and f (x) = 0 from x = to x = π.
2 2
2 1 1 1
(6) ex = (eπ − 1) − (eπ + 1) cos x + (eπ − 1) cos 2x
π 2 1 + 12 1 + 22

1 π
− (e + 1) cos 3x + · · · ,
1 + 32

2 sinh π 1 1 1 1
(7) cosh x = − cos x + cos 2x − cos 3x
π 2 2 5 10

1
+ cos 4x − · · · ,
17

2 1 1
(8) sinh x = (cosh π − 1) − (cosh π + 1) cos x
π 2 2

1 1
+ (cosh π − 1) cos 2x − (cosh π + 1) cos 3x + · · · ,
5 10

2µ sin µπ 1 cos x cos 2x cos 3x
(9) cos µx = − 2 + 2 − 2
π 2µ2 µ − 12 µ − 22 µ − 32

cos 4x
+ 2 − ··· ,
µ − 42
if µ is a fraction.
29. Although any function can be expressed both as a sine series and as
a cosine series, and the function and either series will be equal for all values of
x between zero and π, there is a decided difference in the two series for other
values of x.
Both series are periodic functions of x having the period 2π. If then we let
y equal the series in question and construct the portion of the corresponding
curve which lies between the values x = −π and x = π the whole curve will
consist of repetitions of this portion.
Since sin mx = − sin(−mx) the ordinate corresponding to any value of x
between −π and zero in the sine curve; will be the negative of the ordinate
corresponding to the same value of x with the positive sign. In other words the
curve
y = a1 sin x + a2 sin 2x + a3 sin 3x + · · · (1)
is symmetrical with respect to the origin.
Since cos mx = cos(−mx) the ordinate corresponding to any value of x
between −π and zero in the cosine curve will be the same as the ordinate
belonging to the corresponding positive value of x. In other words the curve
y = 21 b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · · (2)
is symmetrical with respect to the axis of Y .

If then f (x) = −f (−x), that is if f (x) is an odd function the sine series
corresponding to it will be equal to it for all values of x between −π and π,
except perhaps for the value x = 0 for which the series will necessarily be zero.
If f (x) = f (−x), that is if f (x) is an even function the cosine series corre-
sponding to it will be equal to it for all values of x between x = −π and x = π,
not excepting the value x = 0.
As an example of the difference between the sine and cosine developments
of the same function let us take the series for x.

sin 2x sin 3x sin 4x
y = 2 sin x − + − + ··· (3)
2 3 4

y= − cos x + + + + · · · (4)
2 π 32 52 72
[v. Art. 26(a) and Art. 28(a)]. (3) represents the curve
and (4) the curve
Both coincide with y = x from x = 0 to x = π, (3) coincides with y = x from

x = −π to x = π, and neither coincides with y = x for values of x less than −π
or greater than π. Moreover (3), in addition to the continuous portions of the
locus represented in the figure, gives the isolated points (−π, 0) (π, 0) (3π, 0)
&c.
30. We have seen that if f (x) is an odd function its development in sine
series holds for all values of x from −π to π, as does the development of f (x)
in cosine series if f (x) is an even function.
Thus the developments of Art. 26(a), Art. 26 Exs. (2), (4), (6); Art. 28(b),
Art. 28 Exs. (3), (7), (9) are valid for all values of x between −π and π.
Any function of x can be developed into a Trigonometric series to which it
is equal for all values of x between −π and π.
Let f (x) be the given function of x. It can be expressed as the sum of an
even function of x and an odd function of x by the following device.
f (x) + f (−x) f (x) − f (−x)

f (x) = + (1)
2 2
f (x) + f (−x)
identically; but is not changed by reversing the sign of x and is
2
f (x) − f (−x)
therefore an even function of x; and when we reverse the sign of x,
2
is affected only to the extent of having its sign reversed and is consequently an
odd function of x.
Therefore for all values of x between −π and π
f (x) + f (−x) 1
= b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · ·
2 2
2 w f (x) + f (−x)
π
where bm = cos mx.dx; and
π 2
0
f (x) − f (−x)
= a1 sin x + a2 sin 2x + a3 sin 3x + · · ·
2
2 w f (x) − f (−x)
π
where am = sin mx.dx.
π 2
0
bm and am can be simplified a little.
2 w f (x) + f (−x)
π
bm = cos mx.dx
π 2
0
wπ wπ
1
= f (x) cos mx.dx + f (−x) cos mx.dx ,
π
0 0
but if we replace x by −x, we get
wπ w
−π w0
f (−x) cos mx.dx = − f (x) cos mx.dx = f (x) cos mx.dx,
0 0 −π
1 w
π
and we have bm = f (x) cos mx.dx.
π −π
In the same way we can reduce the value of am to
1 w
π
f (x) sin mx.dx.
π −π
Hence
 f (x) = 1 b + b cos x + b cos 2x + b cos 3x + · · ·

 
0 1 2 3
2 (2)
+ a1 sin x + a2 sin 2x + a3 sin 3x + · · ·
 
1 w 1 w
π π
where bm = f (x) cos mx.dx = f (α) cos mα.dα. (3)
π −π π −π
1 w 1 w
π π
and am = f (x) sin mx.dx = f (α) sin mα.dα. (4)
π −π π −π
and this development holds for all values of x between −π and π.

The second member of (2) is known as a Fourier’s Series.
EXAMPLES.
1. Obtain the following developments, all of which are valid from x = −π

to x = π:—

2 sinh π 1 1 1 1 1
(1) ex = − cos x + cos 2x − cos 3x + cos 4x + · · ·
π 2 2 5 10 17

2 sinh π 1 2 3 4
+ sin x − sin 2x + sin 3x − sin 4x + · · · .
π 2 5 10 17

(2) f (x) = − cos x + + + + · · ·
4 π 32 52 72
sin x sin 2x sin 3x sin 4x
+ − + − + ··· ,
1 2 3 4
where f (x) = 0 from x = −π to x = 0 and f (x) = x from x = 0 to x = π.

3π 1 1 2 1 1
(3) f (x) = − + 2
cos x + 2 cos 2x + 2 cos 3x + 2 cos 5x
16 π 1 2 3 5

2
+ 2 cos 6x + · · ·
6

1 3π 3π 3π 1
+ − 1 sin x − sin 2x + + 2 sin 3x
π 2 4 6 3

3π 3π 1
− sin 4x + − 2 sin 5x − · · · ,
8 10 5
π
where f (x) = x from x = −π to x = 0, f (x) = 0 from x = 0 to x = ,
2
π π
and f (x) = x − from x = to x = π.
2 2
2. Show that formula (2) Art. 30 can be written
1
f (x) = c0 cos β0 + c1 cos(x − β1 ) + c2 cos(2x − β2 ) + c3 cos(3x − β3 ) + · · ·
2
1 am
where cm = (a2m + b2m ) 2 and βm = tan−1 .
bm

1
f (x) = c0 sin β0 + c1 sin(x + β1 ) + c2 sin(2x + β2 ) + c3 sin(3x + β3 ) + · · ·
2
1 bm
am
31. In developing a function of x into a Trigonometric series it is often

inconvenient to be held within the narrow boundaries x = −π and x = π. Let
us see if we cannot widen them.
Let it be required to develop a function of x into a Trigonometric series
which shall be equal to f (x) for all values of x between x = −c and x = c.
Introduce a new variable
π
z = x,
c
which is equal to
c −π when x = −c and to π when x = c.
f (x) = f z can be developed in terms of z by Art. 30 (2), (3), and (4).
π
We have
c 1 
f z = b0 + b1 cos z + b2 cos 2z + b3 cos 3z + · · ·
π 2 (1)
+ a1 sin z + a2 sin 2z + a3 sin 3z + · · ·

1 w c
π
where bm = f z cos mz.dz. (2)
π −π π
1 w c
π
and am = f z sin mz.dz. (3)
π −π π
and (1) holds good from z = −π to z = π.

Replace z by its value in terms of x and (1) becomes
1 πx 2πx 3πx

f (x) = b0 + b1 cos + b2 cos + b3 cos + · · ·

2 c c c (4)
πx 2πx 3πx
+ a1 sin + a2 sin + a3 sin + · · ·

c c c
The coefficients in (4) are the same as in (1), and (4) holds good from x = −c
to x = c.
Formulas (2) and (3) can be put into more convenient shape.
1 w c 1 w
π c
mπx π
bm = f z cos mz.dz = f (x) cos dx
π −π π π −c c c
1w 1w
c c
mπx mπλ
or bm = f (x) cos dx = f (λ) cos dλ. (5)
c −c c c −c c
In like manner we can transform (3) into
1w 1w
c c
mπx mπλ
am = f (x) sin dx = f (λ) sin dλ. (6)
c −c c c −c c
By treating in like fashion formulas (1) and (2) Art. 25 and formulas (4) and
(2) Art. 27 we get
πx 2πx 3πx
f (x) = a1 sin + a2 sin + a3 sin + ··· (7)
c c c
2w 2w
c c
mπx mπλ
where am = f (x) sin dx = f (λ) sin dλ. (8)
c c c c
0 0
1 πx 2πx 3πx
and f (x) = b0 + b1 cos + b2 cos + b3 cos + ··· (9)
2 c c c
2w 2w
c c
mπx mπλ
where bm = f (x) cos dx = f (λ) cos dλ. (10)
c c c c
0 0
and (7) and (9) hold good from x = 0 to x = c.

EXAMPLES.
1. Obtain the following developments:

4 πx 1 3πx 1 5πx
(1) 1 = sin + sin + sin + ···
π c 3 c 5 c
from x = 0 to x = c.

2c πx 1 2πx 1 3πx 1 4πx
(2) x= sin − sin + sin − sin + ···
π c 2 c 3 c 4 c
from x = −c to x = c.

c 4c πx 1 3πx 1 5πx 1 7πx
x = − 2 cos + 2 cos + 2 cos + 2 cos + ···
2 π c 3 c 5 c 7 c
2c2
2
πx π 2
2
π 4 2πx π 4 3πx
(3) x2 = 3 − 3 sin − sin + − 2 sin
π 1 1 c 2 c 3 3 c
2
2
π 4πx π 4 5πx
− sin + − 3 sin + ···
4 c 5 5 c
c2 4c2

πx 1 2πx 1 3πx 1 4πx
x2 = − 2 cos − 2 cos + 2 cos − 2 cos + ···
3 π c 2 c 3 c 4 c
1 + ec πx 2(1 − ec )

2πx
(4) ex = 2π 2 2
sin + 2 sin
c +π c c + 4π 2 c
c
4(1 − ec )

3(1 + e ) 3πx 4πx
+ 2 sin + sin + · · · ,
c + 9π 2 c c2 + 16π 2 c
c
1e −1 ec + 1 πx ec − 1 2πx
ex = 2c 2
− 2 2
cos + 2 cos
2 c c +π c c + 4π 2 c
c

e +1 3πx
− 2 cos + ···
c + 9π 2 c

4c πx 1 3πx 1 5πx
(5) f (x) = sin − sin + sin + · · ·
π2 c 32 c 52 c
from x = 0 to x = c,
c c
where f (x) = x from x = 0 to x = and f (x) = c − x from x = to x = c.
2 2

1 πx 2πx
f (x) = c0 cos β0 + c1 cos − β1 + c2 cos − β2
2 c c

3πx
+ c3 cos − β3 + · · ·
c
1 am
bm

1 πx 2πx
f (x) = c0 sin β0 + c1 sin + β1 + c2 sin + β2
2 c c

3πx
+ c3 sin + β3 + · · ·
c
1 am
bm
32. In the formulas of Art. 31 c may have as great a value as we please, so

that we can obtain a Trigonometric Series for f (x) that will represent the given
function through as great an interval as we may choose to take. If, then, we can
obtain the limiting form approached by the series (4) Art. 31 as c is indefinitely
increased the expression in question ought to be equal to the given function of
x for all values of x. Equation (4) Art. 31 can be written as follows if we replace
b0 , b1 , b2 , · · · a1 , a2 , · · · by their values given in Art. 31 (5) and (6).
1 1w
c
f (x) = f (λ) dλ
c 2 −c
wc πλ πx wc 2πλ 2πx
+ f (λ) cos
cos dλ + f (λ) cos cos dλ + · · ·
−c
c c −c
c c
wc πλ πx wc 2πλ 2πx

+ f (λ) sin sin dλ + f (λ) sin sin dλ + · · ·
−c
c c −c
c c
1w
c
1 πλ πx πλ πx
= f (λ) dλ + cos cos + sin sin
c −c 2 c c c c

2πλ 2πx 2πλ 2πx
+ cos cos + sin sin + ···
c c c c
1w
c
1 π 2π
f (x) = f (λ) dλ + cos (λ − x) + cos (λ − x) + · · ·
c −c 2 c c
1 w
c
π 2π
= f (λ) dλ 1 + cos (λ − x) + cos (λ − x) + · · ·
2c −c c c
π 2π
+ cos − (λ − x) + cos − (λ − x) + · · ·
c c
since cos(−φ) = cos φ.
1 w
c
π 2π π π
f (x) = f (λ) dλ · · · + cos − (λ − x) + cos − (λ − x)
2π −c c c c c
π 0π π π
+ cos (λ − x) + cos (λ − x)
c c c c
π 2π
+ cos (λ − x) + · · · (1)
c c
As c is indefinitely increased the limiting value approached by the parenthesis
in (1) is
w∞
cos α(λ − x).dα.
−∞
Hence the limiting form approached by (1) is
1 w w∞
∞
f (x) = f (λ) dλ cos α(λ − x).dα, (2)
2π −∞ −∞
and the second member of (2) must be equal to f (x) for all values of x.
The double integral in (2) is known as Fourier’s Integral, and since it is a
limiting form of Fourier’s Series it is subject to the same limitations as the
series.
That is, in order that (2) should be true f (x) must be finite, continuous,
and single valued for all values of x, or if discontinuous, must have only finite
discontinuities.3
(2) is sometimes given in a slightly different form.
w∞ w0 w∞
Since cos α(λ − x).dα = cos α(λ − x).dα + cos α(λ − x).dα
−∞ −∞ 0
and
w0 w0 w0
cos α(λ − x).dα = cos(−α)(λ − x).d(−α) = − cos α(λ − x).dα
−∞ ∞ ∞
w∞ w∞
cos α(λ − x).dα = 2 cos α(λ − x).dα
−∞ 0
and (2) may be written
1 w w∞
∞
f (x) = f (λ) dλ cos α(λ − x).dα. (3)
π −∞
0
If f (x) is an even function or an odd function (3) can be still further simpli-
fied.
Let f (x) = −f (−x).
Since the limits of integration in (3) do not contain α or λ the integrations
may be performed in whichever order we choose. That is
w∞ w∞ w∞ w∞
f (λ) dλ cos α(λ − x).dα = dα f (λ) cos α(λ − x).dλ.
−∞ 0 0 −∞
Now
w∞ w0 w∞
f (λ) cos α(λ − x).dλ = f (λ) cos α(λ − x).dλ + f (λ) cos α(λ − x).dλ.
−∞ −∞ 0
w0 w0
f (λ) cos α(λ − x).dλ = f (−λ) cos α(−λ − x).d(−λ)
−∞ ∞
w∞
=− f (λ) cos α(λ + x).dλ
0
and (3) becomes
1w w∞
∞
f (x) = dα f (λ)[cos α(λ − x) − cos α(λ + x).dλ
π
0 0
2w w∞
∞
= dα f (λ) sin αλ sin αx.dλ
π
0 0
2w w∞
∞
or f (x) = f (λ)dλ sin αλ sin αx.dα. (4)
π
0 0
If f (x) = f (−x) (3) can be reduced in like manner to
2w w∞
∞
f (x) = f (λ) dλ cos αλ cos αx.dα. (5)
π
0 0
Although (4) holds for all values of x only in case f (x) is an odd function,
and (5) only in case f (x) is an even, function, both (4) and (5) hold for all
positive values of x in the case of any function.
EXAMPLE.
(1) Obtain formulas (4) and (5) directly from (7) and (9) Art. 31.
56
CHAPTER III.
CONVERGENCE OF FOURIER’S SERIES.
33. The question of the convergence of a Fourier’s Series is altogether

too large to be completely handled in an elementary treatise. We will, however,
consider at some length one of the most important of the series we have obtained,
namely

4 sin 3x sin 5x sin 7x
sin x + + + + ··· , [v. (3) Art. 26(b).]
π 3 5 7
and prove that for all values of x between zero and π its sum is absolutely equal
to unity; that is, that the limit approached by the sum of n terms of the series
2
wπ wπ wπ
sin x sin α.dα + sin 2x sin 2α.dα + sin 3x sin 3α.dα + · · · ,
π
0 0 0
as n is indefinitely increased, is 1, provided that x lies between zero and π.

Let
2
wπ wπ wπ
Sn = sin x sin α.dα + sin 2x sin 2α.dα + sin 3x sin 3α.dα + · · ·
π
0 0 0
wπ
+ sin nx sin nα.dα . (1)
0
Then
2w
π
Sn = [sin α sin x + sin 2α sin 2x + sin 3α sin 3x + · · · + sin nα sin nx]dα
π
0
1w
π
= [ cos(α − x) − cos(α + x) + cos 2(α − x) − cos 2(α + x) + · · ·
π
0
+ cos n(α − x) − cos n(α + x)] dα

1 w
π
= [cos(α − x) + cos 2(α − x) + cos 3(α − x) + · · · + cos n(α − x)]dα
π
0
1w
π
− [cos(α + x) + cos 2(α + x) + cos 3(α + x) + · · · + cos n(α + x)]dα.
π
0
CONVERGENCE OF FOURIER’S SERIES. 57
Therefore by Art. 20 (1)

α−x
1w
π
1 1 sin(2n + 1) 2
Sn = − + α−x dα
π 2 2 sin
0
2
α+x
1w
π
1 1 sin(2n + 1) 2

− − + α+x dα.
π 2 2 sin
0
2
α − x α+x
1 w sin(2n + 1) 2 1 w sin(2n + 1) 2
π π
Sn = α−x dα − α+x dα.
2π sin 2π sin
0 0
2 2
α−x
In the first integral substitute β for , and in the second integral sub-
2
α+x
stitute β for .
2
We get
π x π x
2−2 2+2
1 w sin(2n + 1)β 1 w sin(2n + 1)β
Sn = dβ − dβ. (2)
π x sin β π x sin β
−2 2
It remains to find the limit approached by Sn as n is indefinitely increased.

π
34. w2 sin(2n + 1)β π
dβ = . (1)
sin β 2
0
For
sin(2n + 1)β 1
= 2 + cos 2β + cos 4β + · · · + cos 2nβ, by Art. 20.
2 sin β
π
w2
and cos 2kβ.dβ = 0.
0
Let us construct the curve

sin(2n + 1)x
y= .
sin x
We have only to draw the curve y = sin(2n + 1)x and then to divide the
length of each ordinate by the value of the sine of the corresponding abscissa.
In y = sin(2n + 1)x the successive arches into which the curve is divided by
the axis of X are equal, and consequently their areas are equal.
Each arch has for its altitude unity and for its base
π
and is symmetrical with respect to the ordinate
2n + 1
of its highest or lowest point.
sin(2n + 1)x
If now we form the curve y = from
sin x
the curve y = sin(2n + 1)x, it is clear that, since sin x
π
increases as x increases from 0 to , the ordinate of
2
any point of the new curve will be shorter than the
ordinate of the corresponding point in the preceding
arch, and that consequently the area of each arch y =
sin(2n + 1)x
will be less than that of the arch before
sin x
it.
If a0 , a1 , a2 , · · · an−1 are the areas of the successive
arches and an that of the incomplete arch terminated
π
by the ordinate corresponding to x =
2
π
w2 sin(2n + 1)x
dx = a0 − a1 + a2 − a3 + · · · .
sin x
0
But
π π
w2 sin(2n + 1)x w2 sin(2n + 1)β π
dx = dβ = by (1).
sin x sin β 2
0 0
Hence
π
= a0 − a1 + a2 − a3 + a4 − · · · + an if n is even,
2
or
π
= a0 − a1 + a2 − a3 + a4 − · · · − an if n is odd.
2
These equations can be written
π
= a0 + (−a1 + a2 ) + (−a3 + a4 )
2
+ (−a5 + a6 ) + · · · + (−an−1 + an )
if n is even, and
π
= a0 + (−a1 + a2 ) + (−a3 + a4 )
2
+ (−a5 + a6 ) + · · · + (−an−2 + an−1 ) + (−an )
if n is odd.
In either case each parenthesis is a negative quantity since

a0 > a1 > a2 > a3 · · · > an ,
π
and it follows that a0 is greater than .
2
Again
π
= a0 − a1 + (a2 − a3 ) + (a4 − a5 ) + · · · + (an−2 − an−1 ) + an
2
if n is even and
π
= a0 − a1 + (a2 − a3 ) + (a4 − a5 ) + · · · + (an−1 − an )
2
if n is odd.
In either case each parenthesis is positive and it follows that a0 − a1 is is less
π
than .
2
Since
π
a0 > > a0 − a1 ,
2
π
a0 and a0 − a1 differ from by less than they differ from each other, that is,
2
by less than a1 .
π
In like manner we can show that a0 − a1 and a0 − a1 + a2 differ from by
2
π
less than a2 ; and in general that a0 − a1 + a2 − a3 + · · · ± ak differs from by
2
less than ak ; or even that
ak
a0 − a1 + a2 − a3 + · · · ±
p
π
differs from by less than ak no matter the value of p, provided p is greater
2
than unity.
35. From what has been proved in the last article it follows that
wb sin(2n + 1)x
dx,
sin x
0
π π π
where b is some value between and , differs from by less than the
2n + 1 2 2
sin(2n + 1)x
area of the arch in which the ordinate of y = corresponding to
sin x
x = b falls if this ordinate divides an arch, or by less than the area of the arch
next beyond the point (b, 0) if the curve crosses the axis of X at that point.
π
The area of the arch in question is less than , its base, multiplied by
2n + 1
1
π , a value greater than the length of its longest ordinate.
sin b −
2n + 1
wb sin(2n + 1)x
Therefore dx
sin x
0
π π 1
differs from by less than .
2n + 1 sin b − π

2
2n + 1
π 1
If now n is indefinitely increased approaches zero
2n + 1 sin b − π

2n + 1
as its limit, and we get the very important result
wb
sin(2n + 1)x π
limit dx = (1)
n=∞ sin x 2
0
π
if 0 < b < .
2
π x π x
2−2 2+2
1 w sin(2n + 1)β 1 w sin(2n + 1)β
36. Sn = dβ − dβ. [Art. 33. (2)]
π x sin β π x sin β
−2 2
π x
2−2
1 w sin(2n + 1)β 1 w
0
sin(2n + 1)β
= dβ + dβ
π x sin β π sin β
−2 0
π x
1 w sin(2n + 1)β 1 w sin(2n + 1)β

2 2
− dβ + dβ
π sin β π sin β
0 0
π x
2+2
1 w sin(2n + 1)β
− dβ.
π π sin β
2
This last value for Sn can be somewhat simplified.

Substituting γ = −β we get
x
w0 sin(2n + 1)β w0 sin(2n + 1)γ w2 sin(2n + 1)β
dβ = − dγ = dβ.
x
sin β x
sin γ sin β
−2 2
0
Substituting γ = π − β in
π x
2+2w sin(2n + 1)β
dβ we have
π
sin β
2
π x π x π
2+2 −2
w sin(2n + 1)β
2w
sin(2n + 1)γ w2 sin(2n + 1)β
dβ = − dγ = dβ
π
sin β π
sin γ π x
sin β
2 2 2 −2
π π x
w sin(2n + 1)β
2 w 2−2
sin(2n + 1)β
= dβ − dβ.
sin β sin β
0 0
Hence
x π x π
2−2
2 w sin(2n + 1)β 2 w sin(2n + 1)β 2 w sin(2n + 1)β
2 2
Sn = dβ + dβ − dβ.
π sin β π sin β π sin β
0 0 0
π
w sin(2n + 1)β
2
π
dβ = by (1) Art. 34.
sin β 2
0
wx2
sin(2n + 1)β π
limit dβ = if 0<x<π by (1) Art. 35
n=∞ sin β 2
0
and
π2w− x2
sin(2n + 1)β π
limit dβ = if 0<x<π by (1) Art. 35.
n=∞ sin β 2
0
Therefore limit[Sn ] = 1 + 1 − 1 = 1 if 0<x<π and

n=∞

4 sin 3x sin 5x sin 7x
sin x + + + + ··· = 1
π 3 5 7
for all values of x between zero and π.
37. By a somewhat long but not especially difficult extension of the

reasoning just given it can be shown that if f (x) is single-valued and finite
between x = −π and x = π, and has only a finite number of discontinuities and
of maxima and minima between x = −π and x = π the Fourier’s Series
1
b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · ·
2
1 w
π
where am = f (α) sin mα.dα
π −π
1 w
π
and bm = f (α) cos mα.dα,
π −π
and that Fourier’s Series only is equal to f (x) for all values of x between x = −π
and x = π, excepting the values of x corresponding to the discontinuities of f (x),
and the values π and −π if f (π) is not equal to f (−π); and that if c is a value
of x corresponding to a discontinuity of f (x), the value of the series when x = c
is
1
. [f (c − ) + f (c + )];
limit
2 =0
and that if f (π) is not equal to f (−π) the value of the series when x = −π and
when x = π is
1
[f (−π) + f (π)].
2
If f (x) while satisfying the conditions named in the preceding paragraph
except for a finite number of values of x, becomes infinite for those values, the
series is equal to the function except for the values of x in question provided
wπ
that f (x) dx is finite and determinate. (v. Int. Cal. Arts. 83 and 84.)
−π
38. The question of the convergency of a Fourier’s Series and the condi-
tions under which a function may be developed in such a series was first attacked
successfully by Dirichlet in 1829, and his conclusions have been criticised and ex-
tended by later mathematicians, notably by Riemann, Heine, Lipschitz, and du
Bois Reymond. It may be noted that the criticisms relate not to the sufficiency
but to the necessity of Dirichlet’s conditions.
An excellent résumé of the literature of the subject is given by Arnold Sachse
in a short dissertation published by Gauthier–Villars, Paris, 1880, entitled “Es-
sai Historique sur la Représentation d’une Fonction Arbitraire d’une seule vari-
able par une Série Trigonométrique.”
39. A good deal of light is thrown on the peculiarities of trigonometric

series by the attempt to construct approximately the curves corresponding to
them.
If we construct y = a1 sin x and y = a2 sin 2x and add the ordinates of the
points having the same abscissas we shall obtain points on the curve
y = a1 sin x + a2 sin 2x.
If now we construct y = a3 sin 3x and add the ordinates to those of y =

a1 sin x + a2 sin 2x we shall get the curve
y = a1 sin x + a2 sin 2x + a3 sin 3x.
By continuing this process we get successive approximations to
y = a1 sin x + a2 sin 2x + a3 sin 3x + a4 sin 4x + · · ·
Let us apply this method to a few of the series which we have obtained in
Chapter II.
Take
1 1
y = sin x + sin 3x + sin 5x + · · · (1)
3 5
π
= 0 when x = 0, from x = 0 to x = π, and 0 when x = π,
4
v. Art. 26 [b](3).

1 1 1
y = 2 sin x − sin 2x + sin 3x − sin 4x + · · · (2)
2 3 4
= x from x = 0 to x = π, and 0 when x = π,
Art. 26 [a](4).

4 1 1 1 1
y= sin x − 2 sin 3x + 2 sin 5x − 2 sin 7x + · · · (3)
π 12 3 5 7
π π
= x from x = 0 to x = , and π − x from x = to x = π,
2 2
Art. 26 [c](2).
1 2 1 1 2 1
y= sin x + sin 2x + sin 3x + sin 5x − sin 6x + sin 7x + · · · (4)
1 2 3 5 6 7
π π π
= 0 when x = 0, from x = 0 to x = , and 0 from x = to x = π,
2 2 2
v. Art. 26 [d ](2).
It must be borne in mind that each of these curves is periodic having the
period 2π, and is symmetrical with respect to the origin.
The following figures I, II, III, and IV represent the first four approximations
to each of these curves.
In each figure the curve y = the series, and the approximation in question
are drawn in continuous lines, and the preceding approximation and the curve
corresponding to the term to be added are drawn in dotted lines.
Figs. I, II, III, and IV immediately suggest the following facts:

(a) The curve representing each approximation is continuous even when the
curve representing the series is discontinuous.
(b) When the curve representing the series is discontinuous the portion of
each successive approximate curve in the neighborhood of the point whose ab-
scissa is a value of x for which the series curve is discontinuous approaches more
and more nearly a straight line perpendicular to the axis of X and connecting
the separate portions of the series curve.
(c) The curves representing successive approximations do not necessarily

tend to lose their wavy character, since each is obtained from the preceding one
by superposing upon it a wave line whose waves are shorter each time but do
not necessarily lose their sharpness of pitch. This is the case in Figures I, II,
and IV. In Fig. III the waves of the superposed curves grow rapidly flatter.
It follows from this that in such cases as those represented in Figures I, II, and
IV the direction of the approximate curve at a point having a given abscissa does
not in general approach the direction of the series curve at the corresponding
point, or indeed, approach any limiting value, as the approximation is made
closer and closer; and that the length of any portion of the approximate curve
will not in general approach the length of the corresponding portion of the series
curve.
Analytically this amounts to saying that the derivative of a function of x
cannot in general be obtained by differentiating term by term the Fourier’s
Series which represents the function.
(d ) The area bounded by a given ordinate, the approximate curve, the axis
of X, and any second ordinate will approach as its limit the corresponding area
of the series curve if the series curve is continuous between the ordinates in
question; and will approach the area bounded by the given ordinate, the series
curve, the axis of X, any second ordinate, and a line perpendicular to the axis
of X, and joining the separate portions of the series curve if the latter has a
discontinuity between the ordinates in question.
Analytically this amounts to saying that the Fourier’s Series corresponding
to any given function can be integrated term by term and the resulting series will
represent the integral of the function even when the function is discontinuous
(v. Int. Cal. Art. 83).
We may note in passing that if the function curve is continuous a curve
representing the integral of the function will be continuous and will not change
its direction abruptly at any point; while if the function curve is discontinuous
the curve representing the integral will still be continuous but will change its
direction abruptly at points corresponding to the discontinuities of the given
function.
40. The facts that the derivative of a Fourier’s Series cannot in general
be obtained by differentiating the series term by term and that its integral can
be obtained by integrating the series term by term are so important that it is
worth while to look at the matter a little more closely. Let us consider the
differentiation of the series represented in Art. 39 Figure I.
Let
1 1 1
Sn = sin x + sin 3x + sin 5x + · · · + sin(2n + 1)x.
3 5 2n + 1
dSn
Then = cos x + cos 3x + cos 5x + · · · + cos(2n + 1)x.
dx
π
If x =
2
dSn
=0
dx
π
and the curve is parallel to the axis of X for x = no matter what the value
2
of n.
If x = 0 or x = π
dSn
= 1 + 1 + 1 + 1 + ··· + 1 = n + 1
dx
and the curve y = Sn becomes more nearly perpendicular to the axis of X at

the origin and for x = π as we increase n.
π
If x =
3
dSn 1 1 1 1
= −1+ + − 1 + + ···
dx 2 2 2 2
dSn 1
That is = if n=0 or n = 3k
dx 2
1
=− “ n=1 “ n = 3k + 1
2
= 0 “ n=2 “ n = 3k + 2.
π dSn
Consequently when x = does not approach any limiting value as n is
3 dx
indefinitely increased. Indeed, in the successive approximations the point whose
π
abscissa is is successively on the rear, on the front, and on the crest or in the
3
trough of a wave, and although the waves are getting smaller they do not lose
their sharpness of pitch.
dSn
If x has any other value between 0 and π will change abruptly as n is
dx
changed and will not approach any limiting value as n is increased.
41. In general if we differentiate a Fourier’s Series

1
S= b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · ·
2
we get
− b1 sin x − 2b2 sin 2x − 3b3 sin 3x − · · ·
+ a1 cos x + 2a2 cos 2x + 3a3 cos 3x + · · · .
Differentiate again and we get
− b1 cos x − 22 b2 cos 2x − 32 b3 cos 3x − · · ·

− a1 sin x − 22 a2 sin 2x − 32 b3 sin 3x − · · · .
We see that each time we differentiate we multiply the coefficient of sin kx

and of cos kx by k while the term still involves cos kx or sin kx.
Since the series
cos x + cos 2x + cos 3x + · · ·

+ sin x + sin 2x + sin 3x + · · ·
is not convergent, and a Fourier’s Series converges only because its coefficients
decrease as we advance in the series, the differentiation of a Fourier’s Series must
make its convergence less rapid if it does not actually destroy it, and repetitions
of the process will usually eventually make the derived series diverge.
It is to be observed that the derived series are Fourier’s Series, but of some-
what special form, that is they lack the constant term. (v. Art. 30.)
If now we integrate a Fourier’s Series
1
b0 + b1 cos x + b2 cos 2x + b3 cos 3x + · · ·
2
1 1 1
we get C + b0 x + b1 sin x + b2 sin 2x + b3 sin 3x + · · ·
2 2 3
1 1
− a1 cos x − a2 cos 2x − a3 cos 3x − · · · ,
2 3
a Trigonometric Series which converges more rapidly than the given series.
It is to be observed that the series obtained by integrating a Fourier’s Series
is not in general a Fourier’s Series owing to the presence of the term 12 b0 x. (v.
Art. 30.)
42. We are now ready to consider the conditions under which a function
of x can be developed into a Fourier’s Series whose term by term derivative shall
be equal to the derivative of the function.
Let the function f (x) satisfy the conditions stated in Art. 37. Then there is
one Fourier’s Series and but one which is equal to it. Call this series S.
Let the derivative f 0 (x)1 of the given function also satisfy the conditions
stated in Art. 37. Then f 0 (x) can be expressed as a Fourier’s Series. By Art. 39
(d) the integral of this latter series will be equal to the integral of f 0 (x), that is
to f (x) plus a constant, and one integral will be equal to f (x).
If this integral which is necessarily a Trigonometric Series is a Fourier’s Series
it must be identical with S. It will be a Fourier’s Series only in case the Fourier’s
Series for f 0 (x) lacks the constant term 12 b0 .
1 w 0
π
But b0 = f (x)dx by (3) Art. 30.
π −π
1
Therefore b0 = [f (π) − f (−π)];
π
and will be zero if f (π) = f (−π).
In order that f 0 (x) shall satisfy the conditions stated in Art. 37 f (x) while
satisfying the same conditions must in addition be finite and continuous between
x = −π and x = π.
If, then, f (x) is single-valued, finite, and continuous, and has only a finite
number of maxima and minima, between x = −π and x = π, (the values x = −π
and x = π being included), and if f (π) = f (−π) f (x) can be developed into a
df (x)
1 We shall regularly use the notation f 0 (x) for . v. Dif. Cal. Art. 124.
dx
Fourier’s Series whose term by term derivative will be equal to the derivative of
the function.
It will be observed that in this case the periodic curve y = S is continuous
throughout its whole extent.
43. Since a Fourier’s Integral is a limiting case of a Fourier’s Series the

conclusions stated in this chapter hold, mutatis mutandis for a Fourier’s Integral.
For example if a function of x is finite and single-valued for all values of x
and has not an infinite number of discontinuities or of maxima and minima in
the neighborhood of any value of x it will be equal to the Fourier’s Integral
1w w∞
∞
dα f (λ) cos α(λ − x).dλ
π −∞
0
and to that Fourier’s Integral only, w and the integral with respect to x of this
Fourier’s Integral will be equal to f (x)dx.
If in addition f (x) is finite and continuous for all values of x the derivative
df (x)
of the Fourier’s Integral with respect to x will be equal to .
dx
70
CHAPTER IV.
SOLUTION OF PROBLEMS IN PHYSICS BY THE AID OF FOURIER’S
INTEGRALS AND FOURIER’S SERIES.
44. In Art. 7 we have already considered at some length a problem in

Heat Conduction which required the use of a Fourier’s Series. We shall begin
the present chapter with a problem closely analogous in its treatment to that
of Art. 7, but calling for the use of a Fourier’s Integral.
Suppose that electricity is flowing in a thin plane sheet of infinite extent and
that the value of the potential function is given for every point in some straight
line in the sheet, required the value of the potential function at any point of the
sheet.
Let us take the line as the axis of X and consider at first only those points
for which y is positive:
We have, then, to satisfy the equation
Dx2 V + Dy2 V = 0 (1)
V =0 when y=∞ (2)

V = f (x) “ y=0 (3)
where f (x) is a given function, and we are not concerned with negative values
of y.
As in Art. 7 we have e−αy sin αx and e−αy cos αx as particular values of V
which satisfy (1) and (2). We must multiply them by constant coefficients and
so combine them as to satisfy condition (3).
By (3) Art. 32
1w w∞
∞
f (x) = dα f (λ) cos α(λ − x).dλ. (4)
π −∞
0
We wish to build up a value of V which will reduce to (4) when y = 0. This

requires a little care but not much ingenuity.
Take e−αy cos αx and e−αy sin αx and multiply the first by cos αλ, and the
second by sin αλ; they are still values of V which satisfy (1). Add these and we
get
e−αy cos α(λ − x),
still a value of V which satisfies (1), no matter what the values of α and λ.
Multiply by f (λ)dλ and we have
e−αy f (λ) cos α(λ − x).dλ (5)

SOLUTION OF PROBLEMS IN PHYSICS. 71
as a value of V which satisfies (1).

w∞
V = e−αy f (λ) cos α(λ − x).dλ (6)
−∞
is still a solution of (1) since it is the limit of the sum of terms covered by the
form (5); and finally
1w w∞
∞
V = dα e−αy f (λ) cos α(λ − x).dλ (7)
π −∞
0
1
is a solution of (1) as it is multiplied by the limit of the sum of terms formed
π
by multiplying the second member of (6) by dα and giving different values to
α.
But (7) must be our required solution since while it satisfies (1) and (2), it
reduces to (4) when y = 0 and therefore satisfies condition (3).
If f (x) is an even function we can reduce (7) to the form
2w w∞
∞
V = dα e−αy f (λ) cos αx cos αλ.dλ (8)
π
0 0
and if f (x) is an odd function to the form
2w w∞
∞
V = dα e−αy f (λ) sin αx sin αλ.dλ. (9)
π
0 0
(7), (8), and (9) are valid only for positive values of y, but as the problem
is obviously symmetrical with respect to the axis of X, (7), (8), and (9) enable
us to get the value of the potential function at any point of the plane.
EXAMPLES.
1. Obtain forms (8) and (9) directly by the aid of (5) and (4) Art. 32.
2. State a problem in statical electricity of which the solution given in

Art. 44 is the solution.
45. As a special case under Art. 44 let us consider the problem:—To find
the value of the potential function at any point of a thin plane sheet of infinite
extent where all points of a given line which lie to the left of the origin are kept
at potential zero, and all points which lie to the right of the origin are kept at
potential unity.
Here f (x) = 0 if x < 0 and f (x) = 1 if x > 0.
(7) Art. 44 gives us the required solution. It is
1w w∞
∞
V = dα e−αy cos α(λ − x).dλ; (1)
π
0 0
but this can be much simplified.

We have
1w w∞
∞
V = dλ e−αy cos α(λ − x).dα.
π
0 0
w∞ a
Now e−ax cos mx.dx =
a2 + m2
0
if a > 0. (Int. Cal. Art. 82, Ex. 8.)

w∞ y
Hence e−αy cos α(λ − x).dα = ,
+ (λ − x)2 y2
0
1w
∞
y dλ 1 π −1 x
and V = = + tan .
π y 2 + (λ − x)2 π 2 y
0

π x x y
tan − tan−1 = ctn tan−1 = ;
2 y y x
and consequently

1 π x 1 y
V = + tan−1 =1− tan−1 . (2)
π 2 y π x
1 y
Since log z = log(x + yi) = log(x2 + y 2 ) + i tan−1 ,
2 x
[Int. Cal. Art. 33 (2)],

1 1 1 2 2 1 −1 y
i − log z = i − log(x + yi) = − log(x + y ) + i 1 − tan
π π 2π π x
1 y 1
and 1 − tan−1 and − log(x2 + y 2 ) are conjugate functions. (v. Int. Cal.
π x 2π
Arts. 209 and 210.) Hence
1
V1 = − log(x2 + y 2 ) (3)
2π
Dx2 V1 + Dy2 V1 = 0; (4)

and the curves

1 π x
+ tan−1 =a (5)
π 2 y
1
and − log(x2 + y 2 ) = b (6)
2π
cut each other at right angles.
If we construct the curves obtained by giving different values to a in (5)
we get a set of equipotential lines for the conducting sheet described at the
beginning of this article, and the curves obtained by giving different values to
(b) in (6) will be the lines of flow.
Moreover since
1
V1 = − log(x2 + y 2 ) (3)
2π
is a solution of Laplace’s Equation (4), the lines of flow just mentioned will be
equipotential lines for a certain distribution of potential, for which the equipo-
tential lines above mentioned will be lines of flow.
V = a, that is

1 π −1 x
+ tan = a, (5)
π 2 y
reduces to y = −x tan aπ. (7)
If now we give to a values differing by a constant amount we get a set of

straight lines radiating from the origin and at equal angular intervals.
V1 = b, that is
1
− log(x2 + y 2 ) = b, (6)
2π
reduces to
x2 + y 2 = e−2πb . (8)
If we give to b a set of values differing

by a constant amount we get a set of circles
whose centres are at the origin and whose
radii form a geometrical progression. They
are the equipotential lines for a thin plane
sheet of infinite extent where the potential
function is kept equal to given different con-
stant values on the circumferences of two given concentric circles or where we
have a source at the origin; and for this system the lines (7) are lines of flow,
and (3) is the complete solution.
The figure gives the equipotential lines and lines of flow for either system,
but only for positive values of y. The complete figure has the axis of X as an
axis of symmetry.
EXAMPLES.
1. Solve the problem of Art. 44 for the case where
f (x) = −1 if x < 0 and f (x) = 1 if x > 0.

2 x
Ans., V = tan−1 .
π y
2. Solve the problem of Art. 44 for the case where
f (x) = a if x<0 and f (x) = b if x > 0.

1 1 x
Ans., V = (a + b) + (b − a) tan−1 .
2 π y
3. Reduce (7), (8), and (9) Art. 44 to the forms
1 w
∞
yf (λ)dλ
V = ,
π −∞ y 2 + (λ − x)2
1w
∞
1 1
V = yf (λ)dλ 2 + ,
π y + (λ − x)2 y 2 + (λ + x)2
0
1w
∞
1 1
V = yf (λ)dλ 2 − 2 ,
π y + (λ − x)2 y + (λ + x)2
0
respectively.
46. An especially interesting case of Art. 44 is the following where
f (x) = 0 if x < −1, f (x) = 1 if − 1 < x < 1, and f (x) = 0 if x > 1.

1 1+x 1−x
Here V = tan−1 + tan−1 . (1)
π y y
1 1 1
Now log[(1 − z)i] = log[(1 − x − yi)i] = log[y + (1 − x)i]
π π π
1 i 1−x
= log[(1 − x)2 + y 2 ] + tan−1 ,
2π π y
and
1 1 1
− log[(−1 − z)i] = − log[(−1 − x − yi)i] = − log[y − (1 + x)i]
π π π
1 i 1+x
=− log[(1 + x) + y ] + tan−1
2 2
.
2π π y
(1 − x)2 + y 2

1 1−z 1 i −1 1 + x −1 1 − x
log = log + tan + tan .
π −1 − z 2π (1 + x)2 + y 2 π y y
Hence
(1 − x)2 + y 2

1 1+x 1−x 1
tan−1 + tan−1 and log
π y y 2π (1 + x)2 + y 2
are conjugate functions: 1 and

1 1+x 1−x
tan−1 + tan−1 =a (2)
π y y
is any equipotential line, and
1 (1 − x)2 + y 2
log =b (3)
2π (1 + x)2 + y 2
any line of flow for the system described at the beginning of this article; and
1 (1 − x)2 + y 2
V1 = log (4)
2π (1 + x)2 + y 2
is the solution of a new problem for which (3) represents any equipotential line
and (2) any line of flow.
1 The function conjugate to
» –
1 1+x 1−x
tan−1 + tan−1
π y y
might have been found as follows. If φ is the required function and ψ the given function we
have by Int. Cal. Arts. 211, 212, and 213 the relations
Dx φ = Dy ψ and Dy φ = −Dx ψ.
» –
1 1+x 1−x
Here Dy ψ = − +
π (1 + x)2 + y 2 (1 − x)2 + y 2
» –
1 y y
and −Dx ψ = − − .
π (1 + x)2 + y 2 (1 − x)2 + y 2
If now we integrate Dy ψ with respect to x treating y as a constant and add an arbitrary
function of y we shall have φ. So that
 ff
1
φ=− log[(1 + x)2 + y 2 ] − log[(1 − x)2 + y 2 ] + f (y).
2π
» –
1 y y df (y)
Dy φ = − − +
π (1 + x)2 + y 2 (1 − x)2 + y 2 dy
df (y)
Comparing this with its equal −Dx ψ above we find = 0 and f (y) = C a constant
dy
1 (1 − x)2 + y 2
therefore log + C,
2π (1 + x)2 + y 2
where C may be taken at pleasure, is our required conjugate function.
(2) reduces to
2y
= tan aπ
x2 + y2 − 1
or x2 + (y − ctn aπ)2 = csc2 aπ; (5)
e2bπ + 1
and (3) to x2 + y 2 + 2 x+1=0
e2bπ − 1
2 2
ebπ + e−bπ e + e−bπ
bπ
2
or x + bπ + y = −1
e − e−bπ ebπ − e−bπ
or (x + ctnh bπ)2 + y 2 = csch2 bπ. (6)
(5) and (6) are circles. The circles (5) have their centres in the axis of Y , and
pass through the points (−1, 0) and (1, 0); and the circles (6) have their centres
in the axis of X.
(4) is the complete solution, (6) is any equipotential line and (5) any line
of flow for a plane sheet in which the points in the circumferences of two given
circles whose centres are further apart than the sum of their radii are kept at
different constant potentials, or where a source and a sink of equal intensity are
placed at the points (−1, 0) and (1, 0). An important practical example is where
two wires connected with the poles of a battery are placed with their free ends
in contact with a thin plane sheet of conducting material. The figure shows the
equipotential lines and lines of flow of either system.
The complete figure would have the axis of X for an axis of symmetry.
EXAMPLES.
1. Show that if f (x) = a1 when x < −b, f (x) = a2 when −b < x < b,
f (x) = a3 when x > b,

a1 + a3 1 b+x b−x
V = + (a2 − a1 ) tan−1 + (a2 − a3 ) tan−1 .
2 π y y
2. Show that if f (x) = 0 if x < 0, f (x) = a1 if 0 < x < b1 , f (x) = a2 if

b1 < x < b2 , f (x) = a3 if b2 < x < b3 , &c.,

1 x b1 − x b2 − x
V = a1 tan−1 + (a1 − a2 ) tan−1 + (a2 − a3 ) tan−1
π y y y

−1 b3 − x
+ (a3 − a4 ) tan + ··· .
y
3. Show that if f (x) = −1 if x < −1, f (x) = x if −1 < x < 1, f (x) = 1 if

x > 1,
(1 − x)2 + y 2

1 −1 1 + x −1 1 − x y
V = (1 + x) tan − (1 − x) tan + log .
π y y 2 (1 + x)2 + y 2
4. Show that if f (x) = −1 if x < −1, f (x) = 0 if −1 < x < 1, f (x) = 1 if

x > 1,
1 −1 1 + x −1 1 − x
V = tan − tan .
π y y
Show that the equipotential lines are equilateral hyperbolas passing through
the points (−1, 0) and (1, 0), and that the lines of flow are Cassinian ovals hav-
ing (−1, 0) and (1, 0) as foci. The lines of flow are equipotential lines and the
equipotential lines are lines of flow for the case where the points (−1, 0) and
(1, 0) are kept at the same infinite potential, or where very small ovals surround-
ing these points are kept at the same finite potential. The case is approximately
that of a pair of wires connected with the same pole of a battery whose other
pole is grounded, and then placed with their ends in contact with a thin plane
conducting sheet.
5. Show that if f (x) = 0 if x < 0, f (x) = −1 if 0 < x < a, f (x) = 0 if

a < x < b, and f (x) = 1 if x > b,

1 π −1 a − x −1 b − x −1 x
V = − tan − tan − tan .
π 2 y y y
The conjugate function
1 x2 + y 2
V = log
2π [(a − x) + y 2 ][(b − x)2 + y 2 ]
2
is the solution for the case where a sink and two sources of equal intensity lie
on the axis of X, the sink at the origin and the sources at the distances a and b
to the right of the origin. One of the lines of flow is easily seen to be the circle
x2 + y 2 = ab.
47. If the plane conducting sheet has two straight edges at right angles
with each other and one is kept at potential zero while the value of the potential
function is given at each point of the second, that is if V = 0 when x = 0 and
V = f (x) when y = 0, the solution is readily obtained. It is
2w w∞
∞
V = dα e−αy f (λ) sin αx sin αλ.dλ. (1)
π
0 0
v. (9) Art. 44.

This reduces to
1w
∞
y y
V = f (λ)dλ 2 − . (2)
π y + (λ − x)2 y 2 + (λ + x)2
0
v. Ex. 3 Art. 45.
EXAMPLES.
1. If V = 0 when y = 0 and V = F (y) when x = 0 show that
2w w∞
∞
V = dα e−αx F (λ) sin αy sin αλ.dλ
π
0 0
1 w
∞
x x

= F (λ)dλ 2 − 2 .
π x + (λ − y)2 x + (λ + y)2
0
2. If V = f (x) when y = 0 and V = F (y) when x = 0 show that
1w
∞
y y
V = f (λ) −
π y 2 + (λ − x)2 y 2 + (λ + x)2
0

x x
+ F (λ) − 2 dλ.
x2 + (λ − y)2 x + (λ + y)2
3. If F (y) = b the result of Ex. 2 reduces to
1w
∞
2b y y y
V = tan−1 + f (λ)dλ 2 − .
π x π y + (λ − x)2 y 2 + (λ + x)2
0
4. If F (y) = 1 for 0 < y < 1 and F (y) = 0 for y > 1 while f (x) = 1 for
0 < x < 1 and f (x) = 0 for x > 1

1 1−x 1+x y
V = tan−1 − tan−1 + 2 tan−1
π y y x

1 − y 1 + y x
+ tan−1 − tan−1 + 2 tan−1 .
x x y
5. If one edge of the conducting sheet treated in Art. 47 is insulated, so

that Dx V = 0 if x = 0 and V = f (x) when y = 0
2w w∞
∞
V = dα e−αy f (λ) cos αx cos αλ.dλ
π
0 0
1 w
∞
y y

= f (λ)dλ 2 + 2 .
π y + (λ + x)2 y + (λ − x)2
0
48. If the conducting sheet is a long strip with parallel edges one of which
is at potential zero while the value of the potential function is given at all points
of the other, that is if V = 0 when y = 0 and V = F (x) when y = b the problem
is not a very difficult one.
Since we are no longer concerned with the value of V when y = ∞ V =
eαy sin ax and V = eαy cos ax are available as particular solutions of the equation
Dx2 V + Dy2 V = 0 (1)
as well as V = e−αy sin αx and V = e−αy cos αx.

eαy + e−αy
Consequently sin αx = cosh αy sin αx [Int. Cal. Art. 43 (2)]
2
eαy − e−αy
and sin αx = sinh αy sin αx [Int. Cal. Art. 43 (1)]
2
and cosh αy cos αx and sinh αy cos αx
are now available values of V and can be used precisely as e−αy cos αx and
e−αy sin αx are used in Art. 44.
Following the same course as in Art. 44 we get
1w w∞ sinh αy
∞
V = dα F (λ) cos α(λ − x).dλ (2)
π −∞
sinh αb
0
as a solution of (1) which will reduce to V = F (x) when y = b

1−1
and to V =0 when y = 0, since sinh 0 = = 0,
2
and (2) is therefore our required solution.
If V is to be equal to zero when y = b and to f (x) when y = 0 we have only
to replace y by b − y and F (x) by f (x) in (2). We get
1w w∞ sinh α(b − y)
∞
V = dα f (λ) cos α(λ − x).dλ. (3)
π −∞
sinh αb
0
If V = f (x) when y = 0 and V = F (x) when y = b then
∞
V = dα f (λ) cos α(λ − x).dλ
π −∞
sinh αb
0
1w w∞ sinh αy
∞
+ dα F (λ) cos α(λ − x).dλ.
π −∞
sinh αb
0
This can be considerably simplified by the aid of the formula

pπ
w∞ sinh px sin
π q
cos rx.dx =
sinh qx 2q cos pπ + cosh rπ
0
q q
if p2 < q 2 . [Bierens de Haan, Tables of Def. Int. (7) 265] and becomes
1 π w∞ dλ
V = sin (b − y) f (λ)
2b b π(b − y) π
−∞ cos + cosh (λ − x)
b b
πy w
∞
1 dλ
+ sin F (λ) πy π or
2b b −∞ cos + cosh (λ − x)
b b
πy w
∞
1 fλ Fλ
V = sin + πy dλ. (5)
2b b −∞ cosh π (λ − x) − cos πy π
cosh (λ − x) + cos
b b b b
EXAMPLES.
1. Given the formula

w dx 2
r
b−a x

−1
= √ tan tanh if b > a,
a + b cosh x b2 − a2 b+a 2
1
show that if V = 1 when y = 0 and V = 0 when y = b V = (b − y).
b
2. Show that if V = 0 when y = b, V = −1 when y = 0 and x < 0, and
V = 1 when y = 0 and x > 0
πx #
tanh
"
2 2b
V = tan−1 πy
π tan
2b
The solution for the conjugate system, that is, for a strip having a source at
(0, 0) and an infinitely distant sink is

1 πx πy
V = − log cosh2 − cos2 .
π 2b 2b
3. Show that if V = −1 when y = 0 and x < 0, V = 1 when y = 0 and

x > 0, V = −1 when y = b and x < 0, and V = 1 when y = b and x > 0,

2 −1 π πx 2 −1 π πx
V = tan tan (b − y) tanh + tan tan y tanh
π 2b 2b π 2b 2b
πx #
sinh
"
2 b
= tan−1 πy .
π sin
b
The solution for the conjugate system, that is, for a strip having a source and
a sink at the points (0, 0) and (0, b) is
πx πy #
cosh + cos
"
1 b b
V = log πx πy .
π cosh − cos
b b
4. If V = 0 when x = 0, V = f (x) when y = 0 and x > 0, and V = 0 when
y = b and x > 0.
∞
V = dα [cos α(λ − x) − cos α(λ + x)]f (λ)dλ
π sinh αb
0 0
πy w
∞
1 1 1
=
2b
sin
b π πy − π πy f (λ)dλ
0 cosh (λ − x) − cos cosh (λ + x) − cos
b b b b
for positive values of x and for values of y between 0 and b.
5. If V1 = 0 when x = 0, V1 = F (x) when y = b and x > 0, and V1 = 0

when y = 0 and x > 0
πy w
∞
1 1 1
V1 =
2b
sin
b π πy − π πy F (λ)dλ
0 cosh (λ − x) + cos cosh (λ + x) − cos
b b b b
for positive values of x and values of y between 0 and b.
6. If V2 = 0 when x = 0, V2 = f (x) when y = 0 and x > 0, and V2 = F (x)

when y = b and x > 0
V2 = V + V1 for x > 0 and 0 < y < b. (v. Exs. 4 and 5)
7. If one edge of the strip described in Art. 48 is insulated so that we have

V = f (x) when y = 0 and Dy V = 0 when y = b show that
1w w∞ cosh α(b − y)
∞
V = dα f (λ) cos α(λ − x).dλ.
π −∞
cosh αb
0
By the aid of the formula

rπ pπ
w∞ cosh px cosh cos
π 2q 2q
cos rx.dx = pπ rπ if p < q,
cosh qx q cos + cosh
0
q q
[Bierens de Haan, Def. Int. Tables (6) 265],
reduce this to
π
1 πy w
∞ f (λ) cosh (λ − x)
V = sin 2b dλ.
b 2b −∞ cosh π (λ − x) − cos πy
b b
8. If V = 0 when y = 0 or b and x < −a, V = 1 when y = 0 or b and
−a < x < a, and V = 0 when y = 0 or b and x > a
π(a − x) π(a + x) #
sinh sinh
"
1 −1 b −1 b
V = tan πy + tan πy .
π sin sin
b b
9. If V = 0 when y = 0 or b and x < −a, V = 1 when y = 0 and
−a < x < a, V = 0 when y = 0 or b and x > a, and V = −1 when y = b and
−a < x < a
π(a − x) π(a + x) #
tanh tanh
"
1 −1 b −1 b
V = tan πy + tan πy .
π tan tan
b b
10. A system conjugate to that of Ex. 9 is V = +∞ when y = 0 or b and
x = −a, V = −∞ when y = 0 or b and x = a. In this case
πy π(a − x)
1 sin2 + sinh2
V = log b b .
2π 2 πy 2 π(a + x)
sin + sinh
b b
49. Let us take now a problem in the flow of heat. Suppose we have an
infinite solid in which heat flows only in one direction, and that at the start the
temperature of each point of the solid is given. Let it be required to find the
temperature of any point of the solid at the end of the time t.
Here we have to solve the equation
Dt u = a2 Dx2 u (1)
[v. Art. 1 (II)] subject to the condition
u = f (x) when t = 0. (2)

As the equation (1) is linear with constant coefficients we can get a particular
solution by the device used in Arts. 7 and 8.
Let u = eβt+αx and substitute in (1). We get
β = a2 α2
as the only relation which need hold between β and α.
2
α2 t 2
α2 t αx
Hence u = eαx+a = ea e (3)
is a solution of (1) no matter what value is given to α.
To get a trigonometric form replace α by αi.
2
α2 t αxi
Then u = e−a e .
If in (3) we replace α by −αi we get
2
α2 t −αxi
u = e−a e .
As in Arts. 7 and 8 we get from these values
2
α2 t 2
α2 t
u = e−a sin αx and u = e−a cos αx
as particular solutions of (1), α being wholly unrestricted.
From these values we wish to build up a value of u which shall reduce to
f (x) when t = 0 and shall still be a solution of (1).
1w w∞
∞
We have f (x) = dα f (λ) cos α(λ − x).dλ (4)
π −∞
0
v. Art. 32 (3), and by proceeding as in Art. 44 we get

1w w∞
∞
2 2
u= dα e−a α t f (λ) cos α(λ − x).dλ (5)
π −∞
0
as our required value of u.

This can be considerably simplified.
Changing the order of integration
1 w w∞
∞
2 2
u= f (λ)dλ e−a α t cos α(λ − x).dα. (6)
π −∞
0
w
∞
1
r
π − (λ−x) 2
−a2 α2 t
e cos α(λ − x).dα = .e 4a2 t (7)
2a t
0
by the formula
w∞ √
2
x2 π − b22
e−a cos bx.dx = e 4a [Int. Cal. Art. 94 (2)]
2a
0
1 w∞ (λ−x)2
Hence u= √ f (λ)e− 4a2 t dλ. (8)
2a πt −∞
λ−x
Let now β= √ ,
2a t
√
then λ = x + 2a t.β
1 w
∞
√ 2
and u= √ f (x + 2a t.β)e−β dβ. (9)
π −∞
EXAMPLES.
1. Let the solid be of infinite extent and let the temperature be equal to a
constant c at the time t = 0.
c w −β 2 2c w −β 2
∞ ∞
Then u= √ e dβ = √ e dβ = c.
π −∞ π
0
v. Int. Cal. Art. 92 (2).
2. Let u = x when t = 0.
1 w
∞
√ 2
Then u= √ (x + 2a t.β)e−β dβ = x.
π −∞
3. Let u = x2 when t = 0.
Then u = x2 + 2a2 t.
4. Let u = 0 if x < −b, u = 1 if −b < x < b, and u = 0 if x > b, when t = 0.
Then
b−x
√
1 w −β 2
2a t
b3 + 3bx2 b5 + 10b3 x2 + 5bx4

2 b
u= √ e dβ = √ √ − √ + √ − · · · .
π b+x π 2a t 3(2a t)3 5.2!(2a t)5
− 2a√t
5. Let u = 0 if x < 0 and u = 1 if x > 0 when t = 0.

Then
x√ x√
1 w −β 2
2aw t w∞ 1 w −β 2
∞ 2a t
1 2 2 1
u= √ e dβ = √ e−β dβ + e−β dβ = √ e dβ +
π x√ π π 2
− 2a t
0 0 0
x3 x5 x7

1 1 x
= +√ √ − √ + √ − √ + · · · .
2 π 2a t 3.(2a t)3 5.2!(2a t)5 7.3!(2a t)7
6. An iron slab 10 c. m. thick is placed between and in contact with two

very thick iron slabs. The initial temperature of the middle slab is 100◦ , and of
each of the outer slabs 0◦ . Required the temperature of a point in the middle
of the inner slab fifteen minutes after the slabs have been put together. Given
a2 = 0.185 in C.G.S. units. Ans., 21◦ .6.
7. Two very thick iron slabs one of which is at the temperature 0◦ and the
other at the temperature 100◦ throughout are placed together face to face. Find
the temperature of each slab 10 c. m. from their common face fifteen minutes
after they have been placed together. Ans., 70◦ .8, 29◦ .2.
8. Find a particular solution of Dt u = a2 Dx2 u on the assumption that it is

of the form u = T.X where T is a function of t alone and X is a function of x
alone.
50. If our solid has one plane face which is kept at the constant tem-
perature zero, and we start with any given distribution of heat, the problem is
somewhat modified.
Take the origin of coördinates in the plane face. Then we have as before the
equation
Dt u = a2 Dx2 u, (1)
but our conditions are
u=0 when x=0 (2)
u = f (x) “ t=0 (3)
and we are concerned only with positive values of x.
We may then use the form (4) Art. 32
2w w∞
∞
f (x) = dα f (λ) sin αx sin αλ.dλ, (4)
π
0 0
and proceeding as in the last section we get

2w w∞
∞
2 2
u= dα e−a α t f (λ) sin αx sin αλ.dλ (5)
π
0 0
as our required solution. This may be reduced considerably.

1w w∞
∞
2 2
u= f (λ)dλ e−a α t [cos α(λ − x) − cos α(λ + x)]dα,
π
0 0
1 w
∞
(λ−x)2 (λ+x)2
or u = √ f (λ)(e− 4a2 t − e− 4a2 t )dλ (6)
2a πt 0
by (7) Art. 49, and this may be reduced to the form
w∞ w∞
√ √

1 −β 2 −β 2
u= √ e f (x + 2a t.β)dβ − e f (−x + 2a t.β)dβ . (7)
π x√ x√
− 2a t 2a t
EXAMPLES.
1. Let the initial temperature be constant and equal to c.

Then
w∞ w∞
c 2 2
u= √ e−β dβ − e−β dβ
π x√ x√
− 2a t 2a t
x√
2c w −β 2
2a t
=√ e dβ
π
0
x3 x5 x7

2c x
=√ √ − √ + √ − √ + · · · .
π 2a t 3.(2a t)3 5.2!(2a t)5 7.3!.(2a t)7
2. Assuming that the earth was originally at the temperature 7000◦ Fahren-
heit throughout, and that the surface was kept at the constant temperature 0◦ ,
find (1) the temperature 10 miles below the surface 10,000,000 years after the
cooling began; (2) the temperature 1 mile below the surface at the same epoch;
(3) the temperature 10 miles below the surface 100,000,000 years after the cool-
ing began; (4) the temperature 1 mile below the surface at the same epoch; (5)
the rate at which the temperature was increasing with the distance from the
surface at each point at each epoch.
Neglect the convexity of the earth’s surface and take Sir Wm. Thomson’s
value of a2 (400) the foot, the Fahrenheit degree, and the year being taken as
units. (Thomson and Tait’s Nat. Phil. Vol. II. Appendix.)
Ans., (1) 3114◦ ; (2) 329◦ .5; (3) 1036◦ ; (4) 103◦ ; (5) 1◦ for every 20 feet, 3◦
for every 50 feet, 1◦ for every 50 feet, 1◦ for every 50 feet.
3. Let the initial temperature be constant and equal to −b, then by Ex. 1
x√
2b w −β 2
2a t
u = −√ e dβ.
π
0
4. Let the temperature of the plane face be b instead of zero, and let the
initial temperature be zero.
Then we have only to add b to the second member of the solution in Ex. 3,
as we may since u = b is a solution of (1) Art. 49, and we get
x√
2 w −β 2
2a t
u=b 1− √ e dβ .
π
0
5. Let u = b when x = 0 and u = f (x) when t = 0.

Then
x√
2 w −β 2 1 w
2a t ∞
(λ−x)2 (λ+x)2
u=b 1− √ e dβ + √ f (λ)[e− 4a2 t − e− 4a2 t ]dλ
π 2a πt 0
0
by (6) Art. 50.
6. Let u = b when x = 0 and u = c when t = 0.

x√
2 w −β 2
2a t
Then u = b + (c − b) √ e dβ.
π
0
7. If the earth has been cooling for 200,000,000 years from a uniform tem-
perature, prove that the rate of cooling is greatest at a depth of about 76 miles,
and that at a depth of about 130 miles the rate of cooling has reached its max-
imum value for all time. Let a2 = 400.
8. Show that if the plane face of the solid considered in Art. 50 instead of
being kept at temperature zero is impervious to heat
1 w
∞
(λ−x)2 (λ+x)2
u= √ f (λ)(e− 4a2 t + e− 4a2 t )dλ. v. (6) Art. 50.
2a πt 0
51. If the temperature of the plane face of the solid described in Art. 50
is a given function of the time and the initial temperature is zero, the solution
of the problem can be obtained by a very ingenious method due to Riemann.
Here we have to solve the equation
Dt u = a2 Dx2 u (1)
)
u = F (t) when x=0
(2)
u=0 “ t = 0.
We know that x√
2 w −β 2
2a t
u= √ e dβ
π
0
is a solution of (1), v. Ex. 1 Art. 50. It is easily shown that
√x
2 w −β 2
2a t−c
u= √ e dβ, (3)
π
0
where c is any constant, is a solution of (1).

For
2 x2
2 x 1 − 4a2x(t−c) x 3 −
Dt u = − √ 3 e = − √ (t − c)− 2 e 4a2 (t−c)
π 2a 2(t − c) 2 2a π
2
2 1 − x
Dx u = √ √ e 4a2 (t−c)
π 2a t − c
2 1 2x x2 x x2
− 3 −
Dx2 u = − √ √ 2
e 4a2 (t−c) = − 3 √ (t − c)− 2 e 4a2 (t−c)
π 2a t − c 4a (t − c) 2a π
and Dt u = a2 Dx2 u.
Let φ(x, t) be a function of x and t which shall be equal to zero if t is negative
and shall be equal to
x√
2 w −β 2
2a t
1− √ e dβ
π
0
if t is equal to or greater than zero; so that if x = 0 φ(x, t) = 1 and if t =

0 φ(x, t) = 0.
We shall now attack the following problem, to solve equation (1) subject to
the conditions
u=0 if t = 0
u = F (0) “ x=0 and 0<t<τ
u = F (kτ ) “ x=0 “ kτ < t < (k + 1)τ,
where k is any whole number and τ is any arbitrarily chosen interval of time.
If we form the value
u = F (kτ )[φ(x, t − kτ ) − φ(x, t − (k + 1)τ )] (4)
u will satisfy equation (1) since zero, unity and
√x
2 w
2a t−kτ
2
√ e−β dβ
π
0
are values of u which satisfy (1). u will be zero if t < kτ by the definition of the
function φ(x, t); if x = 0 u = 0 if t > (k+1)τ and u = F (kτ ) if kτ < t < (k+1)τ .
Therefore
k=∞
X
u= F (kτ )[φ(x, t − kτ ) − φ(x, t − (k + 1)τ )] (5)
k=0
is the solution of the problem stated above.

(5) can be simplified somewhat from the consideration that for a given value
of t φ(x, t − kτ ) = 0 if kτ > t. If, then, nτ is the greatest whole multiple of τ
not exceeding t,
k=n
X
u= F (kτ )[φ(x, t − kτ ) − φ(x, t − (k + 1)τ )]. (6)
k=0
If now we decrease τ indefinitely the limiting form of (6) will be the solution
of the problem stated at the beginning of this article.
(6) may be written
k=n
X φ(x, t − kτ ) − φ(x, t − (k + 1)τ )
u= F (kτ ) τ (7)
τ
k=0
and if τ is indefinitely decreased the limiting form of (7) is

wt
u=− F (λ)Dλ φ(x, t − λ)dλ. (8)
0
Since t − λ is positive between the limits of integration

√x
2 w
2a t−λ
2
φ(x, t − λ) = 1 − √ e−β dβ,
π
0
x x2
− 3
and Dλ φ(x, t − λ) = − √ e 4a2 (t−λ) (t − λ)− 2 ;
2a π
x w
t
x2
− 3
u= √ F (λ)e 4a2 (t−λ) (t − λ)− 2 dλ, (9)
2a π
0
x
or if we let β= √
2a t − λ
2 w
∞
x2

−β 2
u= √ e F t− dβ. (10)
π x√
4a2 β 2
2a t
EXAMPLES.
1. If u = nt when x = 0 and u = 0 when t = 0

x√
√
2 w −β 2
2a t
x2

nx t x2
u=n t+ 2 1− √ e dβ − √ e− 4a2 t .
2a π a π
0
2. A thick iron slab is at the temperature zero throughout, one of its plane
faces is then kept at the temperature 100◦ Centigrade for 5 minutes, then at the
temperature zero for the next 5 minutes, then at the temperature 100◦ for the
next 5 minutes, and then at the temperature zero. Required the temperature of
a point in the slab 5 c.m. from the face at the expiration of 18 minutes. Given;
a2 = .185. Ans., 20◦ .1.
3. If u = F (t) when x = 0 and u = f (x) when t = 0, then
2 w −β 2 1 w − (λ−x)
∞ ∞
x2
2 (λ+x)2
u= √ e F t − 2 2 dβ + √ (e 4a2 t − e− 4a2 t )f (λ)d(λ).
π x√ 4a β 2a πt 0
2a t
v. (6) Art. 50.

4. If in Art. (51) F (t) is a periodic function of the time of period T it can

be expressed by a Fourier’s series of the form
m=∞
1 X 2π
F (t) = b0 + [am sin mαt + bm cos mαt], where α= ,
2 m=1
T
m=∞
1 X
or F (t) = + ρm sin(mαt + λm ),
2b0 m=1
where ρm cos λm = am and ρm sin λm = bm . v. Art. 31 Ex. 3.
Show that with this value of F (t) (10) Art. 51. becomes
1 w∞ 2 X
m=∞ w∞ mαx2
−β 2 2
u = √ b0 e dβ + √ ρm sin(mαt + λm ) e−β cos 2 2 dβ
π x√
π m=1 x√
4a β
2a t 2a t
w∞ 2 mαx2

− cos(mαt + λm ) e−β sin dβ
x√
4a2 β 2
2a t
and that as t increases u approaches the value
1
m=∞
X x
√ mα x
r
mα
u= b0 + ρm e− a 2 sin(mαt − + λm ).
2 m=1
a 2
Given that
w∞ √ w∞ √
−x2 b2 π −b√2 √ −x2 b2 π −b√2 √
e sin 2 dx = e sin b 2; e cos 2 dx = e cos b 2.
x 2 x 2
0 0
v. Riemann, Lin. par. dif. gl. § 54.
5. If we are dealing with a bar of small cross-section where the heat not
only flows along the bar but at the same time escapes at the surface of the bar
into air at the temperature zero we have to solve the differential equation
Dt u = a2 Dx2 u − b2 u. v. Fourier, Heat § 105.
Show that for this case

2
+a2 α2 )t 2
+a2 α2 )t
u = e−(b sin αx and u = e−(b cos αx
are particular solutions, and that if u = f (x) when t = 0
e−b t w − (λ−x) e−b t w −β 2

2 ∞ 2 ∞
2 √
u= √ e 4a2 t f (λ)dλ = √ e f (x + 2a t.β)dβ.
2a πt −∞ π −∞
cf. (8) and (9) Art. 49.

If u = 0 when x = 0 and u = f (x) when t = 0
2 w∞ w∞
e−b t √ √

−β 2 −β 2
u= √ e f (x + 2a t.β)dβ − e f (−x + 2a t.β)dβ .
π
− 2ax√t x√
2a t
cf. (7) Art. 50.

bx
If u = −e− a when t = 0 and u = 0 when x = 0
1

bx
w∞ √ 2 bx
w∞ √ 2

u= √ ea e−(b t+β) dβ − e− a e−(b t+β) dβ ,
π x√ x√
2a t
− 2a t
bx
and if u = 1 when x = 0 and u = 0 when t = 0 we have only to add e− a to the
bx
second member of the last equation, since u = e− a satisfies the equation
Dt u = a2 Dx2 u − b2 u.
If u = F (t) when x = 0 and u = 0 when t = 0 we can employ the method of

Art. 51.
1
w∞ √ w∞ √

− bx bx
−(b t−λ+β)2 − bx −(b t−λ+β)2
φ(x, t − λ) = e a +√ e a e dβ − e a e dβ ,
π √x √x
2a t−λ
− 2a t−λ
− 23
x(t − λ) −b 2 2
(t−λ)− 4a2x(t−λ)
−Dλ φ(x, t − λ) = √ e ;
2a π
x w
t
3 −b2 (t−λ)− x2
and u= √ (t − λ)− 2 e 4a2 (t−λ) F (λ)dλ,
2a π
0
cf. (9) Art. 51,
2 w −β 2 − 4a
∞
x2

b2 x2
or u= √ e 2 β2
F t − 2 2 dβ,
π x√ 4a β
2a t
cf. (10) Art. 51.

If F (t) is periodic and has the value taken in Ex. 4, show that the value
approached by u as t increases is
m=∞ √
√
1 − bx X
− x2a2 p x 2
u = b0 e a + ρm e sin mαt − q + λm ,
2 m=1
2a
1 1
p p
where p = (b2 + b4 + m2 α2 ) 2 and q = (−b2 + b4 + m2 α2 ) 2 .
w∞ √
−x2 − x
a2 π −2a
Given e 2
dx =
e
2
0
w∞ √
2 a2 b2 π −2c
e−x − x2 sin 2 dx = e sin 2d
x 2
0
w∞ √
a2
−x2 − x b2 π −2c
and e 2
cos 2 dx = e cos 2d,
x 2
0
where
√ √
2 2 p 4 1 2 p 1
c= (a + a + b4 ) 2 and d = (−a2 + a4 + b4 ) 2 .
2 2
Ängstrom’s method of determining the conductivity of a metal is based
on the result just given (v. Phil. Mag. Feb. 1863), and is described by Sir
Wm. Thomson (Encyc. Brit. Article “Heat”) as by far the best that has yet
been devised.
52. If u is a periodic function of the time when x = 0 as in Art. 51 Ex. 4

and we are concerned with the limiting value approached by u as t increases
we can avoid evaluating a complicated definite integral if we take the following
course.
Since as we have seen in Art. 49 u = eβt+αx is a solution of
Dt u = a2 Dx2 u (1)
provided only that β = a2 α2 we have

x
√
u = eβt± a β
as a solution.
Replacing β by ±βi this becomes
x
√ √
u = e±βti± a β ±i
x
√β
or u = e±βti± a 2 (1±i)
√ 1√
since i=± 2(1 + i)
2
√ 1√
and −i = ± 2(1 − i).
2
Hence
√β r
x β √β r
x β
−x −x
u=e a 2sin βt − , u=e a cos βt −
2 , (2)
a 2 a 2
x
√β r
x β

x
√β r
x β

u = e a 2 sin βt + , u = e a 2 cos βt + , (3)
a 2 a 2
are particular solutions of (1).

From these we get readily
√ mα r
x mα

−x
u = ρm e a 2 sin mαt − + λm (4)
a 2
as a solution. (4) reduces to
u = ρm sin(mαt + λm ) when x = 0
√ mα r
x mα

−x
and to u = ρm e a 2 sin λm − when t = 0.
a 2
If we add a term which

satisfies
r (1)
and which is equal to zero when x = 0
√ mα x mα
−x
and to −ρm e a 2 sin λm − when t = 0 (v. Art. 50) we shall have
a 2
a solution of (1) which is zero when t = 0 and which is
ρm sin(mαt + λm ) when x = 0.
The term in question approaches zero as t increases [v. (7) Art. 50] and we have
at once the solution given in Art. 51 Ex. 4, as our required result.
EXAMPLE.
Show that u = eβt+αx is a solution of Dt u = a2 Dx2 u − b2 u if β = a2 α2 − b2 ,

and hence that
√2 √2 x
x x ±βti± a√ (p±qi)
u = eβt± a b +β , u = e±βti± a b ±βi , u = e 2 ,

± px
√ qx ± px
√ qx
u = e a 2 sin βt ± √ , and u = e a 2 cos βt ± √ ,
a 2 a 2
where p 1 p 1
p = [ β 2 + b4 + b2 ] 2 and q = [ β 2 + b4 − b2 ] 2 ,
are solutions. Hence

− apx
√ qx
u = ρm e 2 sin βt − √ + λm
a 2
is a solution.
If β = mα this last result reduces to u = ρm sin(mαt + λm ) when x = 0 and
by the reasoning of Art. 52 it must be the value u approaches as t increases if
we have the same conditions as in the last part of Art. 51 Ex. 5.
53. The whole problem of the flow of heat is treated by Sir William
Thomson (v. Math. and Phys. Papers, Vol. II), and other recent writers from a
different and decidedly interesting point of view, which we shall briefly sketch
in connection with the problem of Linear Flow.
Suppose we are dealing with a bar having a small cross-section and an
adiathermanous surface, and take as our unit of heat the amount required to
raise by a unit the temperature of a unit of length of the bar. If at a point
of the bar a quantity Q of heat is suddenly generated the point is called an
instantaneous heat source of strength Q.
If the heat instead of being suddenly generated is generated gradually and
at a rate that would give Q units of heat per unit of time the point is called a
permanent heat source of strength Q.
The temperature at any point of the bar at any time due to an instantaneous
source of strength Q at the point x = λ is easily found by the aid of formula (8)
Art. 49 as follows:—
If a quantity of heat Q is suddenly generated along the portion of the bar
from x = λ to x = λ + ∆λ, where ∆λ is any arbitrary length, the temperature
Q
of that portion will be suddenly raised to , and we shall have by (8) Art. 49
∆λ
1 w
λ+∆λ
Q (λ−x)2
u= √ e− 4a2 t dλ (1)
2a πt ∆λ λ
as the temperature of any point of the bar at any time t thereafter.

If now we write u equal to the limiting value approached by the second
member of (1) as ∆λ is made to approach zero we get
Q (λ−x)2
u= √ e− 4a2 t (2)
2a πt
as the solution for the case where we have an instantaneous source at the point
x = λ.
Q
It is to be observed that in (2) u = 0 when t = 0 and u = √ when
2a πt
x = λ and t > 0.
If we have several sources we have only to add the temperatures due to the
separate sources.
Formula (8) Art. 49 may now be regarded as the solution for the case where
we start with an instantaneous heat source of strength f (λ)dλ in every element
of length of the bar.
A source of strength −Q is called a sink of strength Q; and (6) Art. 50 may be
regarded as the solution for the case where we have at the start an instantaneous
source of strength f (λ)dλ in every element of the bar whose distance to the right
of the origin is λ, and an instantaneous sink of strength f (λ)dλ in every element
of the bar whose distance to the left of the origin is λ.
If we have an instantaneous source at the origin (2) reduces to
Q x2
u= √ e− 4a2 t (3)
2a πt
For a permanent source of constant strength Q at the origin (3) gives
Q w − 4a2x(t−τ
t 2
1
u= √ e ) (t − τ )− 2 dτ (4)
2a π
0
and for a permanent source of variable strength f (t)
1 w − 4a2x(t−τ
t 2
1
u= √ e ) (t − τ )− 2 f (τ )dτ. (5)
2a π
0
In (4) and (5) u obviously reduces to zero when t = 0 and x > 0, but its value
when x = 0 is not easily determined. We can avoid the difficulty by introducing
the conception of a doublet.
54. If a source and a sink of equal strength Q are made to approach each
other while Q multiplied by their distance apart is kept equal to a constant P
the limiting state of things is said to be due to a doublet of strength P whose
axis is tangent to the line of approach and points from sink to source. A doublet
of strength −P differs from a doublet of strength P only in that its axis has the
opposite direction.
Let us find the temperature due to an instantaneous doublet of strength P
placed at the origin. For a source of strength Q at x = η and an equal sink at
x = −η we have
Q (η−x)2 (η+x)2
u= √ (e− 4a2 t − e− 4a2 t ),
2a πt
or if 2ηQ = P ,
P (η 2 +x2 ) ηx ηx
u= √ e− 4a2 t (e 2a2 t − e− 2a2 t )
4aη πt
P (η 2 +x2 ) ηx
= √ e− 4a2 t sinh 2 .
2aη πt 2a t
If η is made to approach zero

1 ηx x
limit sinh 2 = 2 ,
η 2a t 2a t
Px x2
and u= √ e− 4a2 t (1)
4a3 πt3
is the solution for the temperature at any time and place due to an instantaneous
doublet of strength P placed at the origin. For a doublet at any other point
x = λ we have
P (x − λ) − (x−λ)2
u= √ e 4a2 t . (2)
4a3 πt3
For a permanent doublet of constant strength P placed at the origin we have
P x w − 4a2x(t−τ
t 2
3
u= √ e ) (t − τ )− 2 dτ ; (3)
4a3 π
0
and for a permanent doublet of variable strength f (t)
x w − 4a2x(t−τ
t 2
3
u= √ e ) (t − τ )− 2 f (τ )dτ, (4)
3
4a π
0
1 w
∞
2

x2

or u = 2√ e−β f t − 2 2 dβ (5)
a π x√ 4a β
2a t
if x > 0, and
1 w
−∞
x2

−β 2
u= √ e f t − dβ (6)
a2 π x√ 4a2 β 2
2a t
x
if x < 0, if we let β = √ .
2a t − τ
f (t)
From (5) and (6) we see readily that u = 0 when t = 0 and that u =
2a2
f (t)
when x = 0 if we approach the origin from the right and that u = − 2 when
2a
x = 0 if we approach the origin from the left.
If the point x = 0 is kept at the constant temperature b and we are concerned
only with positive values of x we can get from (5) the solution given in Art. 50
Ex. 4 by supposing a permanent doublet of strength 2a2 b placed at the origin.
To solve the problem treated in Art. 51 we have only to suppose a permanent
doublet of strength 2a2 F (t) placed at x = 0 and from (5) we get at once (10)
Art. 51.
EXAMPLE.
Show that if Dt u = a2 Dx2 u − b2 u and an instantaneous source of strength Q

is placed at x = λ
Q 2 (λ−x)2
u= √ e−b t− 4a2 t v. Art. 51, Ex. 5.
2a πt
Show that if an instantaneous doublet of strength P is placed at the point
x=0
Px 2 x2
u= √ e−b t− 4a2 t .
4a3 πt3
If a permanent doublet of strength f (t) is placed at x = 0
x w −b2 (t−τ )− 4a2x(t−τ

t 2
3
u= √ e ) (t − τ )− 2 f (τ )dτ
3
4a π
0
1 w
±∞
b2 x2

x2

−β 2 − 4a
= 2√ e 2 β2
f t − 2 2 dβ,
a π x√ 4a β
2a t
f (t)
whence u = 0 when t = 0 and x > 0 or x < 0 and u = ± when x = 0.
2a2
Hence if we place at x = 0 a permanent doublet of strength 2a2 F (t) we get
the solution given in Art. 51 Ex. 5 for the case where u = F (t) when x = 0 and
u = 0 when t = 0 provided we are concerned only with positive values of x.
If F (t) = c this reduces to
2c w −β 2 − 4a
∞
b2 x2
u= √ e 2 β2
dβ.
π x√
2a t
55. As another example of the use of Fourier’s Integral we shall consider

the transmission of a disturbance along a stretched elastic string.
Suppose we have a stretched elastic string so long that we need not consider
what happens at its ends, that is so long that we may treat its length as infinite.
Let the string be initially distorted into some given form and then released; to
investigate its subsequent motion.
Let us take the position of equilibrium of the string as the axis of X and
any given point as origin.
We have, then, to solve the differential equation
Dt2 y = a2 Dx2 y (1)
[v. (VIII) Art. 1] subject to the conditions
y = f (x) when t = 0 (2)

Dt y = 0 “ t = 0. (3)
As in Art. 8 we find
y = cos α(x ± at) and y = sin α(x ± at)
as particular solutions of (1).

From these we must build up a value that will reduce to
1w w∞
∞
f (x) = dα f (λ) cos α(λ − x).dλ (4)
π −∞
0
when t = 0 and will at the same time satisfy (3).

y = cos αλ cos α(x + at) + sin αλ sin α(x + at)
or y = cos α(λ − x − at)
is a solution of (1).
1w w∞
∞
Hence y= dα f (λ) cos α(λ − x − at).dλ (5)
π −∞
0
is also a solution of (1).

(5) reduces to y = f (x) when t = 0 but it gives
aw w∞
∞
Dt y = αdα f (λ) sin α(λ − x).dλ
π −∞
0
when t = 0 and consequently does not satisfy equation (3).

If in forming (5) we use cos α(x−at) and sin α(x−at) instead of cos α(x+at)
and sin α(x + at) we get
1w w∞
∞
y= dα f (λ) cos α(λ − x + at).dλ (6)
π −∞
0
which is a solution of (1), and reduces to y = f (x) when t = 0, but it gives
aw w∞
∞
Dt y = − αdα f (λ) sin α(λ − x).dλ
π −∞
0
when t = 0 and does not satisfy (3).

If, however, we take one-half the sum of the values of y in (5) and (6) we get
1 1w w∞
" ∞
y= dα f (λ) cos α(λ − x − at).dλ
2 π −∞
0
1w w∞
∞
#
+ dα f (λ) cos α(λ − x + at).dλ , (7)
π −∞
0
a solution of (1) which satisfies both (2) and (3), and is, therefore, our required
solution.
This result can be very much simplified.
If we substitute z = x + at
1w w∞
∞
dα f (λ) cos α(λ − x − at).dλ
π −∞
0
1w w∞
∞
= dα f (λ) cos α(λ − z).dλ = f (z) = f (x + at);
π −∞
0
and in like manner we can show that

1w w∞
∞
dα f (λ) cos α(λ − x + at).dλ = f (x − at).
π −∞
0
Hence our solution becomes

1
y = [f (x + at) + f (x − at)]. (8)
2
This result is of great importance in the theory of elastic strings and it
shows that the initial disturbance splits into two equal waves which run along
the string, one to the right and the other to the left, with a uniform velocity a,
and that there is nothing like a periodic motion or vibration of any sort unless
the ends of the string produce some effect.
56. If the string is not initially distorted but starts from its position of
equilibrium with a given initial velocity impressed upon each point we have to
solve the equation
Dt2 y = a2 Dx2 y (1)
y=0 when t = 0 (2)
Dt y = F (x) “ t= 0. (3)
We get by the process used in Art. 55
1 w w∞
∞
sin α(λ − x + at) sin α(λ − x − at)
y= dα F (λ) − dλ
2πa −∞
α α
0
1 w w∞ sin α(λ − x + at) sin α(λ − x − at)

∞
= F (λ)dλ − dα;
2πa −∞ α α
0
w∞ sin α(λ − x + at) w∞ sin α(λ − x − at)

but dα − dα = π
α α
0 0
if x − at < λ < x + at, and is equal to zero for all other values of λ; since
w∞ sin mx π
dx = if m > 0
x 2
0
π
=− if m < 0
2
= 0 if m = 0.
v. Int. Cal. Art. 92 (3).
1 w
x+at
Hence y= F (λ)dλ (4)
2a x−at
is our required solution.

EXAMPLES.
1. If the string is initially distorted and starts with initial velocity so that
y = f (x) and Dt y = F (x) when t = 0
1 w
x+at
1
y= [f (x + at) + f (x − at)] + F (λ)dλ.
2 2a x−at
2. If the initial disturbance is caused by a blow, as from the hammer in a

piano, which impresses upon all the points in a portion of the string of length c
an equal transverse velocity b show that the front of the wave which will be seen
to run to the left along the string will be a straight line having a slope equal to
b c p 2
and a length equal to 4a + b2 . Of course a wave having a front of the
2a 2a
b
same length with a slope equal to − will be seen to run to the right along
2a
the string, and the effect of the two waves will be to lift the string bodily and
bc
permanently to a distance above its original position.
2a
57. We shall now take up a few examples of the use of Fourier’s Series.
In the problem of Art. 7 let the temperature of the base of the plate be a
given function of x, the other conditions remaining unchanged.
m=∞
X
Since f (x) = (am sin mx)
m=1
2 wπ
where am = f (α) sin mα.dα
π
0
m=∞
2 X −my wπ
we have u= e sin mx f (α) sin mα.dα . (1)
π m=1
0
If the breadth of the plate is a instead of π
mπx w
m=∞ a
2 X − mπy mπλ
u= e a sin f (λ) sin dλ . (2)
a m=1 a a
0
58. If the temperature of the base is unity and the breadth of the plate
is π the solution is, as we have seen in Art. 7,

4 −y 1 −3y 1 −5y
u= e sin x + e sin 3x + e sin 5x + · · · . (1)
π 3 5
This series can be summed without difficulty. We have the development
z z2 z3 z4
log(1 + z) = − + − + ···
1 2 3 4
if the modulus of z is less than 1. Int. Cal. Art. 221 (4).
z z2 z3 z4
Hence log(1 − z) = − − − − − ···
1 2 3 4
if mod. z < 1.
1 z z3 z5
and [log(1 + z) − log(1 − z)] = + + + ··· (2)
2 1 3 5
if mod. z < 1.
But
log(1 + z) = log[1 + r(cos φ + i sin φ)]

1 r sin φ
= log[(1 + r cos φ)2 + (r sin φ)2 ] + i tan−1
2 1 + r cos φ
1 2 −1 r sin φ
= log (1 + 2r cos φ + r ) + i tan ,
2 1 + r cos φ
and
1 r sin φ
log(1 − z) = log(1 − 2r cos φ + r2 ) − i tan−1 ,
2 1 − r cos φ
[Int. Cal. Art. 33 (2)],
and (2) becomes
1 + 2r cos φ + r2

1 1 −1 2r sin φ
log + i tan
2 2 1 − 2r cos φ + r2 1 − r2
r(cos φ + i sin φ) r3 (cos 3φ + i sin 3φ)
= + + ··· (3)
1 3
From (3) we get two equations
1 1 + 2r cos φ + r2 r cos φ r3 cos 3φ r5 cos 5φ

log = + + + ··· (4)
4 1 − 2r cos φ + r2 1 3 5
1 2r sin φ r sin φ r3 sin 3φ r5 sin 5φ
tan−1 = + + + ··· (5)
2 1 − r2 1 3 5
both valid for all values of φ provided r < 1.
e−y is less than 1 if y is positive.
Hence from (5)
e−y sin x e−3y sin 3x e−5y sin 5x 1 2e−y sin x

+ + + · · · = tan−1
1 3 5 2 1 − e−2y
1 2 sin x 1 sin x
= tan−1 y = tan−1 ,
2 e − e−y 2 sinh y

2 sin x
u= tan−1 . (6)
π sinh y
If we replace r by e−y and φ by x in
log[1 + r( cos φ + i sin φ)]
it becomes log[1 + e−y cos x + ie−y sin x]
or log[1+ cos z + i sin z]
v. Int. Cal. Art. 35 (3) and (4)
a function of z as a whole; and
log[1 − r( cos φ + i sin φ)]
becomes log(1− cos z − i sin z);
hence by Int. Cal. Arts. 212 and 213,
1 1 + 2e−y cos x + e−2y 1 2e−y sin x
log −y −2y
and tan−1
4 1 − 2e cos x + e 2 1 − e−2y
1 cosh y + cos x 1 sin x
or log and tan−1
4 cosh y − cos x 2 sinh y
are conjugate functions, and
1 cosh y + cos x
u1 = log (7)
π cosh y − cos x
is the solution for the problem where the isothermal lines are the lines of flow of
the present problem and the lines of flow are the isothermal lines of the present
problem.
For our problem, then, the isothermal lines are given by the equation
2 sin x
tan−1 =a
π sinh y
sin x aπ
or = tan (8)
sinh y 2
and the lines of flow by
1 cosh y + cos x
log = b,
π cosh y − cos x
cosh y + cos x
or = eπb . (9)
cosh y − cos x
EXAMPLES.
1. If Dx2 u + Dy2 u = 0, and u = 1 when y = 0, and u = 0 when x = 0 and
when x = a,

4 − πy πx 1 − 3πy 3πx 1 − 5πy 5πx
u= e a sin + e a sin + e a sin + ···
π a 3 a 5 a
πx
2 sin
= tan−1 a
π πy .
sinh
a
2. If u = φ(x) when y = 0, u = f (y) when x = 0, and u = F (y) when
x=a
mπx w
m=∞ a
2 X − mπy mπλ
u= e a sin φ(λ) sin dλ
a m=1 a a
0
πx w
∞
1 1 1
+ sin π πx − π πx f (λ) dλ
2a a cosh (λ − y) − cos cosh (λ + y) − cos
0
a a a a
πx w
∞
1 1 1
+
2a
sin
a π πx − π πx F (λ) dλ
0 cosh (λ − y) + cos cosh (λ + y) + cos
a a a a
v. Art. 48, Exs. 4, 5, and 6.
59. If three sides of a plane rectangular sheet of conducting material be

kept at potential zero and the value of the potential function at every point of
the fourth side be given; to find the value of this potential function at any point
of the sheet.
To formulate:—
Dx2 V + Dy2 V = 0. (1)
V =0 when x = 0. (2)
V =0 “ x = a. (3)
V =0 “ y = b. (4)
V = f (x) “ y = 0. (5)
Working as in Art. 48 we get
mπ
sinh (b − y) mπx
a sin
mπb a
sinh
a
as a value of V which satisfies equations (1), (2), (3), and (4) if m is an integer.
Therefore
mπ
m=∞ sinh
2 X (b − y) mπx w
a
mπλ

V = a sin f (λ) sin dλ (6)
a m=1 mπb a a
sinh 0
a
EXAMPLES.
1. If f (x) = 1 Eq. (6) Art. 59 reduces to

π 3π
4 sinh a (b − y) πx 1 sinh a (b − y)
"
3πx
V = sin + sin
π πb a 3 3πb a
sinh sinh
a a
5π
1 sinh a (b − y)
#
5πx
+ sin + ···
5 5πb a
sinh
a
2. If V = 0 when x = 0, V = 0 when x = a, V = 0 when y = 0, and
V = F (x) when y = b, then
mπy
2 X sinh a mπx w
a
m=∞
" #
mπλ
V = sin F (λ) sin dλ .
a m=1 mπb a a
sinh 0
a
3. If F (x) = 1 the answer of Ex. 2 reduces to
πy 3πy 5πy
4 sinh a πx 1 sinh a 3πx 1 sinh a
" #
5πx
V = sin + sin + sin + ··· .
π πb a 3 3πb a 5 5πb a
sinh sinh sinh
a a a
4. If V = 0 when x = 0, V = 0 when x = a, V = f (x) when y = 0, and
V = F (x) when y = b, then
mπ
sinh (b − y) wa
m=∞
"
2 X mπx a mπλ
V = sin f (λ) sin dλ
a m=1 a mπb a
sinh 0
a
mπy a
sinh w
!#
a mπλ
+ F (λ) sin dλ .
mπb a
sinh 0
a
5. If f (x) = F (x) the answer of Ex. 4 reduces to
mπ b
cosh −y wa
m=∞
" #
2 X
a 2 mπx mπλ
V = sin f (λ) sin dλ .
a mπb a a
m=1 cosh 0
2a
6. If f (x) = F (x) = 1 the answer of Ex. 5 reduces to

πb 3π b
cosh − y cosh − y
"
4 a 2 πx 1 a 2 3πx
V = sin + sin
π πb a 3 3πb a
cosh cosh
2a 2a
5π b
1 cosh a 2 − y
#
5πx
+ sin + ··· .
5 5πb a
cosh
2a
7. If V = f (x) when y = 0, V = F (x) when y = b, V = φ(y) when x = 0,
and V = χ(y) when x = a, then
mπ
mπx sinh a (b − y) w
a
m=∞
"
2 X mπλ
V = sin f (λ) sin dλ
a m=1 a mπb a
sinh 0
a
mπy a
sinh w
!#
a mπλ
+ F (λ) sin dλ
mπb a
sinh 0
a
mπ
mπy sinh b (a − x) w
b
m=∞
"
2 X mπλ
+ sin mπa φ(λ) sin dλ
b m=1 b sinh b
0
b
mπx b
sinh w
!#
b mπλ
+ mπa χ(λ) sin b dλ .
sinh
b 0
8. If f (x) = φ(y) = 0 and F (x) = χ(y) = 1 the answer of Ex. 7 may be
reduced to
π a 2π a
sinh − x cosh − x
"
2 πy b 2 πy 1 b 2 2πy
V = − πa sin + sin
π 2b sinh b 2 2πa b
2b cosh
2b
3π a 4π a
sinh − x cosh − x
#
1 b 2 3πy 1 b 2 4πy
− sin + sin − ··· .
3 3πa b 4 4πa b
sinh cosh
2b 2b
9. Find the temperature of the middle point of a thin square plate whose
faces are impervious to heat; 1st, when three edges are kept at the temperature
0◦ and the fourth edge at the temperature 100◦ ; 2d, when two opposite edges
are kept at the temperature 0◦ and the other two at the temperature 100◦ ; 3d,
when two adjacent edges are kept at the temperature 0◦ and the other edges at
the temperature 100◦ . See examples 3, 6, and 8.
Ans., (1) 25◦ ; (2) 50◦ ; (3) 50◦ .
60. Let us pass on to the consideration of the flow of heat in one dimen-
sion.
Suppose that we have an infinite solid with two parallel plane faces whose
distance apart is c.
Take the origin in one face and the axis of X perpendicular to the faces. Let
the initial temperature be any given function of x and let the two faces be kept
at the constant temperature zero; to find the temperature at any point of the
slab at any time.
We have to solve the equation
Dt u = a2 Dx2 u (1)
u=0 when x=0 (2)

u=0 “ x=c (3)
u = f (x) “ t = 0. (4)
In Art. 49 we have found

2
α2 t
u = e−a sin αx
−a2 α2 t
and u=e cos αx
as particular solutions of (1).

2 2
u = e−a α t sin αx satisfies (2) whatever value is given to α. It satisfies (3) if
mπ
α= provided m is an integer. Let us try to build a value of u out of terms
c
a2 m2 π 2 t mπx
of the form Ae− c2 sin which shall satisfy (4).
c
We have
mπx w
m=∞ c
2 Xh mπλ i
f (x) = sin f (λ) sin dλ . (5)
c m=1 c c
0
mπx w
m=∞ c
2 X h − m2 a22 π2 t mπλ i
u= e c sin f (λ) sin dλ , (6)
c m=1 c c
0
reduces to (5) when t = 0 and is our required solution.
EXAMPLES.
1. If f (λ) = b, a constant, (6) Art. 60 reduces to

4b h − a2 π22 t πx 1 − 9a22π2 t 3πx 1 − 25π22a2 t 5πx i
u= e c sin + e c sin + e c sin + ··· .
π c 3 c 5 c
2. An iron slab 10 cm. thick is placed between and in contact with two
other iron slabs each 10 cm. thick. The temperature of the middle slab is at
first 100◦ throughout, and of the outside slabs 0◦ throughout. The outer faces
of the outside slabs are kept at the temperature 0◦ . Required the temperature
of a point in the middle of the middle slab fifteen minutes after the slabs have
been placed in contact. Given a2 = 0.185 in C.G.S. units. Ans., 10◦ .3.
3. Two iron slabs each 20 cm. thick one of which is at the temperature 0◦
and the other at the temperature 100◦ throughout, are placed together face to
face, and their outer faces are kept at the temperature 0◦ . Find the temperature
of a point in their common face and of points 10 cm. from the common face
fifteen minutes after the slabs have been put together.
Ans., 22◦ .8; 15◦ .1; 17◦ .2.
4. One face of an iron slab 40 cm. thick is kept at the temperature 0◦ and
the other face at the temperature 100◦ until the permanent state of tempera-
tures is set up. Each face is then kept at the temperature 0◦ . Required the
temperature of a point in the middle of the slab, and of points 10 cm. from the
faces fifteen minutes after the cooling has begun. Ans., 22◦ .8; 15◦ .6; 16◦ .7.
61. If the faces of the slab treated in Art. 60 instead of being kept at the
temperature zero are rendered impervious to heat, the solution of the problem
is easy.
In this case we have to solve the equation
Dt u = a2 Dx2 u
Dx u = 0 when x=0
Dx u = 0 “ x=c
u = f (x) “ t = 0.
We have only to use the particular solution

2
α2 t
u = e−a cos αx
−a2 α2 t
as we used u=e sin αx
in Art. 60. We get
2 1w mπx w
c m=∞ c
X − m 2 a2 π 2 t mπλ
u= f (λ)dλ + e c2 cos f (λ) cos dλ . (1)
c 2 m=1
c c
0 0
EXAMPLES.
1. Solve example 2 Art. 60 supposing that the outer surfaces are blanketed
after the slabs are placed together so that heat can neither enter nor escape.
Find in addition the temperature of the outer surfaces fifteen minutes after the
slabs are placed in contact. Ans., 33◦ .3; 33◦ .3.
2. Solve example 3 Art. 60 on the hypothesis just stated, getting in addi-

tion the temperatures of points on the outer surfaces.
Ans., 50◦ ; 33◦ .9; 66◦ .1; 27◦ .2; 72◦ .8.
3. Solve example 4 Art. 60 supposing that heat neither enters nor escapes
at the outer surfaces after the permanent state of temperatures has been set up.
Find also the temperatures of points in the outer surfaces.
Ans., 50◦ ; 39◦ .7; 60◦ .3; 35◦ .5; 64◦ .5.
4. Show that if u = 0 when x = 0, Dx u = 0 when x = c, and u = f (x)

when t = 0,
(2m + 1)πx w
m=∞ c
2 X −
(2m+1)2 a2 π 2 t (2m + 1)πλ
u= e 4c 2
sin f (λ) sin dλ .
c m=0 2c 2c
0
Suggestion: Assume u = 0 when x = 2c and f (2c − x) = f (x), and see (6)

Art. 60.
62. If the temperature of the right-hand face of the slab considered in

Art. 60 is a constant γ instead of zero we have only to add to the second
member of (6) Art. 60 a term u1 which shall satisfy the conditions
Dt u1 = a2 Dx2 u1 (1)
u1 = 0 when x = 0 (2)
u1 = 0 “ t=0 (3)
u1 = γ “ x = c. (4)
γx
u1 = obviously satisfies (1), (2), and (4); to make it satisfy (3) as well
c
we must add a term u2 which shall be equal to zero when x = 0 and when x = c
γx
and to − when t = 0, while always satisfying (1). It is given immediately by
c
(6) Art. 60 and is
mπx w
m=∞ c
2γ X 2 2 2
− m ac2 π t mπλ
u2 = − e sin λ sin dλ . (5)
c2 m=1 c c
0
wc mπλ c2 c2
λ sin dλ = − cos mπ = (−1)m+1 ,
c mπ mπ
0
m=∞
2γ X (−1)m − m2 a22 π2 t

mπx
and u2 = e c sin . (6)
π m=1 m c
m=∞
2 X (−1)m − m2 a22 π2 t

x mπx
Hence u1 = γ + e c sin . (7)
c π m=1 m c
If the left-hand face of the slab considered in Art. 60 is to be kept at a

constant temperature β and the right-hand face at the temperature zero we can
get the term u3 which must be added to the second member of (6) Art. 60 by
replacing γ by β and x by c − x in (7). We then have
m=∞
c−x 2 X 1 − m2 a22 π2 t mπx
u3 = β − e c sin . (8)
c π m=1 m c
EXAMPLES.
1. Show that if u = β when x = 0, u = γ when x = c, and u = f (x) when

t=0
m=∞
2 X (−1)m − m2 π22 a2 t

x mπx
u = β + (γ − β) + e c sin
c π m=1 m c
mπx w
m=∞ c
2 X 2 2 2
− m ac2 π t mπλ
+ e sin [f (λ) − β] sin dλ .
c m=1 c c
0
2. Show that if u = β when x = 0, u = 0 when t = 0, and Dx u = 0 when

x=c
m=∞
4 X 1 (2m+1)2 a2 π 2 t (2m + 1)πx
u=β 1− e− 4c2 sin
π m=0 2m + 1 2c

4 2 2
− a 4cπ2 t πx 1 − 9a2 π22 t 3πx 1 − 25a22π2 t 5πx
=β 1− e sin + e 4c sin + e 4c sin + ··· .
π 2c 3 2c 5 2c
63. If the temperature of the right-hand face of the slab just considered is
a function of the time instead of a constant and the temperature of the left-hand
face is zero the problem can be solved by a method nearly identical with that
of Art. 51.
Let φ(x, t) be a function of x and t which shall be zero if t is less than zero
and shall be equal to
m=∞
2 X (−1)m − m2 a22 π2 t

x mπx
+ e c sin
c π m=1 m c
[v. (7) Art. 62] if t is equal to or greater than zero. So that
φ(x, t) = 0 if t<0
φ(x, t) = 0 “ t=0 unless x = c
φ(x, t) = 1 “ t=0 and x = c
φ(x, t) = 1 “ x=c
φ(x, t) = 0 “ x = 0.
Precisely as in Art. 51 we get

k=n
X
[φ(x, t − kτ ) − φ(x, t − (k + 1)τ )]τ
u = limit
. F (kτ ) (1)
τ =0 τ
k=0
as the required solution of our problem, n being as in Art. 51 the largest integer
t
in where t is any given value of the time.
τ
On our hypothesis the last term of (1), that is, −F (nτ )φ[x, t − (n + 1)τ ] = 0;
the next to the last term F (nτ )φ(x, t − nτ ) has for its limiting value
" m=∞ #
x 2 X (−1)m mπx
F (t)φ(x, 0) = F (t) + sin ,
c π m=1 m c
while as in Art. 51 the limiting value of the rest of the sum is
wt
− F (λ)Dλ φ(x, t − λ)dλ.
0
m=∞
2a2 π X

m
2 2 2
− m ca2 π (t−λ) mπx
Dλ φ(x, t − λ) = 2 (−1) me sin .
c m=1
c
Hence
m=∞
2 X (−1)m

x mπx
u = F (t) + sin
c π m=1 m c
mπx w
m=∞ t
2a2 π X
2 2 2

m − m ca2 π (t−λ)
− 2 (−1) m sin F (λ)e dλ ,
c m=1
c
0
m=∞
2 X (−1)m

x mπx
u= F (t) + sin F (t)
c π m=1 m c
m2 a2 π 2 w
t
2 2 2
− m ca2 π (t−λ)
− F (λ)e dλ . (2)
c2
0
m2 a2 π 2
If we substitute β = (t − λ) we get
c2
m a π t 2 2 2
x
m=∞
2 X (−1)m mπx
wc2
βc2

−β
u = F (t)+ sin F (t)− e F t− 2 2 2 dβ . (3)
c π m=1 m c m a π
0
EXAMPLES.
1. If the temperature of the left-hand face is a function of t and the tem-

perature of the right-hand face is zero and the initial temperature is zero
m a π t 2 2 2

x
m=∞
2 X 1 mπx
wc2
βc2

u= 1− F (t) − sin F (t) − e−β F t − 2 2 2 dβ .
c π m=1 m c m a π
0
2. If the temperature of the left-hand face is a function of t, the initial

temperature is zero, and the right-hand face is impervious to heat
m=∞
4 X 1 (2m + 1)πx
u = F (t) − sin F (t)
π m=0 2m + 1 2c
(2m + 1)2 a2 π 2 w
t
(2m+1)2 a2 π 2
− (t−λ)
− F (λ)e 4c2 dλ .
4c2
0
3. If in Arts. 60–63 we are dealing with a bar of small cross-section and of

length c and heat is radiating from the surface of the bar into air at the temper-
ature zero so that Dt u = a2 Dx2 u − b2 u, show that: (a) the second members of
2
(6) Art. 60 and (1) Art. 61 must be multiplied by e−b t ; (b) equation (7) Art. 62
becomes
bx
sinh
( m=∞ )
X m 2 2 2 mπx
u1 = γ a +2a2 πe−b2 t m
(−1) 2 2 e − m ac2 π t
sin ;
bc b c + m2 a2 π 2 c
sinh m=1
a
(c) equation (2) Art. 63 becomes
bx
sinh m=∞
X (−1)m m mπx

u= a F (t) + 2a2 π sin F (t)
bc b2 c2 + m2 a2 π 2 c
sinh m=1
a
b2 c2 + m2 a2 π 2 w − b2 c2 +m22 a2 π2 (t−λ)
t
− e c F (λ)dλ .
c2
0
64. The problem of the motion of a finite stretched elastic string of length
l fastened at the ends and distorted at first into some given curve y = f (x), and
then allowed to swing, has been treated and partially solved in Art. 8.
The complete solution is easily seen to be
mπat w
m=∞ l
2 X mπx mπλ
y= sin cos f (λ) sin dλ. (1)
l m=1 l l l
0
2l
The second member of (1) is a periodic function of t having the period .
a
The motion, then, unlike that in the case of an infinite string (Art. 55) is a true
2l
vibration, a periodic motion. The period is the time it takes a disturbance
a
to travel twice the length of the string (v. Art. 55).
A careful examination of (1) will show that the actual motion is a good deal
like that in the case considered in Art. 55. The original disturbance breaks up
into two waves one of which runs to the right until it reaches the end of the
string and is then reflected, and runs back to the left or the under side of the
string, while the other wave runs to the left and is reflected at the left-hand end
of the string and runs back to the right under the string and is again reflected,
runs back to the left over the string and so on indefinitely.
If the curve into which the string is distorted at the start is of the form
mπx
y = b sin the solution is
l
mπx mπat
y = b sin cos . (2)
l l
No matter what value t may have the curve is always of the form
mπx
y = A sin ;
l
that is, for different values of t we have a set of sine curves differing only in
the amplitude and not at all in the period of the curve. In this case either the
whole string if m = 1, or each mth of the string if m is not equal to one, rises
and falls, and there is no apparent onward motion. When this is the case we
are said to have a steady vibration.
If m = 1 we get steady motion of the string as a whole and if the vibration is
rapid enough to give a musical note the note is said to be the pure fundamental
note of the string. If m = 2 the vibration is twice as rapid as when m = 1, the
middle point of the string does not move and is called a node, the two halves
of the string are in opposite phases of vibration at any instant, and the note
given is an octave higher than the fundamental note and is called its pure first
harmonic.
If m = 3 the vibration is three times as rapid as in the first case, there are
l 2l
two nodes x = and x = , and the note is the pure second harmonic of the
3 3
fundamental note.
For any value of m the vibration is m times as rapid as when m = 1, there

l 2l m−1
are m − 1 nodes at the points x = , x = , · · · x = l, and we get the
m m m
m − 1st harmonic of the fundamental note.
It is clear from (1) that no matter what the original form of the string
the resulting vibration can be regarded as a combination of steady vibrations
each of which alone would give the fundamental note of the string or one of
its harmonics, and that the complex note resulting is really a concord of the
fundamental note and some of its harmonics.
A finely trained ear can often recognize in a complex note the fundamental
note of the string and some of its harmonics and is capable of analyzing a
complex note into its component pure notes precisely as Fourier’s Theorem
enables us to analyze the complex function representing the initial form of the
string into the simpler sine-functions which must be combined to form it.
EXAMPLES.
1
1. Show that if a point whose distance from the end of a harp string is th
n
the length of the string is drawn aside by the player’s finger to a distance b from
its position of equilibrium and then released, the form of the vibrating string at
any instant is given by the equation
m=∞
2bn2 X 1
mπ mπx mπat
y= sin sin cos .
(n − 1)π 2 m=1 m2 n l l
Show from this that all the harmonics of the fundamental note of the string
which correspond to forms of vibration having nodes at the point drawn aside
by the finger will be wanting in the complex note actually sounded.
2. If a stretched string starts from its position of equilibrium, each of its

points having a given initial velocity, so that we have
y=0 when t=0

Dt y = F (x) “ t=0
y=0 “ x=0
y=0 “ x = l,
the solution of the problem of its vibration is easy and gives
mπat w
m=∞ l
2 X 1 mπx mπλ
y= sin sin F (λ) sin dλ .
aπ m=1 m l l l
0
3. Write down the solution for the case where the string is initially distorted
and each point has a given initial velocity.
65. If we do not neglect the resistance of the air in the problem of the
vibration of a stretched string the differential equation is rather more compli-
cated and the solution is not so easily obtained. The equation is given as (IX)
Art. 1.
Let us solve the problem for the case where there is no initial velocity.
Here we have Dt2 y + 2kDt y = a2 Dx2 y. (1)

y=0 when x=0 (2)
y=0 “ x=l (3)
y = f (x) “ t=0 (4)
Dt y = 0 “ t = 0. (5)
We get particular solutions of (1) in the usual way. Assume y = eαx+βt and
substitute in (1). We have
β 2 + 2kβ = a2 α2
as the only necessary relation between β and α. This gives
p
β = −k ± a2 α2 + k 2 .
√
a2 α2 +k2
Hence y = eαx−kt±t (6)
is a solution of (1) no matter what the value of α.

To throw it into Trigonometric form replace α by αi, and since in actual
problems k, which is proportional to the resistance, is very small, take −1 out
as a factor of the radical. We have
√ 2 2 2
y = e−kt e(αx±t a α −k )i .
Since α may be positive or negative we can get

p
y = e−kt sin(αx ± t a2 α2 − k 2 )
p
and y = e−kt cos(αx ± t a2 α2 − k 2 )
as solutions of (1), or by combining these

p
y = e−kt sin αx cos t a2 α2 − k 2 (7)
p
y = e−kt sin αx sin t a2 α2 − k 2 (8)
p
y = e−kt cos αx cos t a2 α2 − k 2 (9)
p
y = e−kt cos αx sin t a2 α2 − k 2 (10)
mπ
(7) and (8) satisfy (1) and (2) for all values of α. They satisfy (3) if α =
.
l
Let us see if out of them we cannot build up a value that will satisfy (4) and
(5) as well.
mπx w
m=∞ l
2 X mπλ
f (x) = sin f (λ) sin dλ . (11)
l m=1 l l
0
m=∞ wl
r
m2 π 2 a2

2 −kt X mπx 2
mπλ
y= e sin cos t − k . f (λ) sin dλ (12)
l m=1
l l2 l
0
reduces to (11) when t = 0 and therefore satisfies (4).

m=∞
r r
2 X m2 π 2 a2 2
mπx m2 π 2 a2
Dt y = − kt − k . sin sin t − k2
le m=1 l2 l l2
wl mπλ

. f (λ) sin dλ
l
0
m=∞ wl
r
X m2 π 2 a2

2k mπx 2.
mπλ
− sin cos t − k f (λ) sin dλ . (13)
lekt m=1
l l2 l
0
When t = 0 the first line of the second member of (13) vanishes but the
second line reduces to
mπx w
m=∞ l
2k X mπλ
− sin f (λ) sin dλ .
l m=1 l l
0
We must, then, introduce into (12) an additional term which shall equal zero
when t = 0 and whose derivative with respect to t shall cancel the term above
when t = 0.
This is easily seen to be
m=∞ wl
r
2k −kt X 1 mπx m2 π 2 a2 2.
mπλ
e r sin sin t 2
− k f (λ) sin dλ.
l m=1
2 2 2
m π a l l l
2 0
−k
l2
Hence our complete solution is
m=∞
r
2 −kt X m2 π 2 a2
y= e cos t − k2
l m=1
l2
mπx w
r l
m2 π 2 a2

k 2 sin
mπλ
+r sin t − k f (λ) sin dλ . (14)
m2 π 2 a2 l2 l l
2 0
−k
l2
Here the fact that e−kt , which decreases rapidly as t increases, is a factor
of the whole second member shows that the amplitude of the vibration rapidly
decreases.
Comparing this solution with that given in Art. 64 for the case where there
is no resistance we see that the period of any given term
r
mπx m2 π 2 a2
A sin cos t − k2 ,
l l2
mπx mπat
is greater than that of the corresponding term A1 sin cos in Art. 64.
l l
In other words the effect of the resistance of the air is to flatten somewhat
each component part of the note given by the string. More than this since the
periods of the different terms of (14) are no longer exact submultiples of the
period of the first term, the component notes are no longer in perfect harmony
with the fundamental note of the string, and the ideal perfect harmony between
the fundamental note and its harmonics is not quite realized in any actual case.
When k is very small, as in the case of a fine string, the departure from
perfect harmony is very slight; but in the case of a coarse string or worse still of
an elastic ribbon, where the resistance of the air is considerable, the unmusical
character of the sound is very noticeable.
EXAMPLES.
1. Solve Ex. 1 Art. 64 allowing for the resistance of the air.
2. Solve Ex. 2 Art. 64 allowing for the resistance of the air;

m=∞
r
2 −kt X 1 mπx m2 π 2 a2
y= e sin sin t − k2
l2 l
r
l 2 2 2
m π a l
m=1
− k2 w mπλ

l2 . F (λ) sin dλ .
l
0
3. Find a particular solution of (1) Art. 65 on the assumption that it is of

the form y = T.X, where T is a function of t alone and X a function of x alone.
66. We pass on now to a couple of problems that require the modification

and extension of Fourier’s Theorem, the cooling of a sphere in air, and the
vibration of a stretched rectangular membrane, but as an introduction to the
former we shall first consider the following very simple problem; to find the
temperature of any point of a sphere whose initial temperature is any given
function of r the distance of the point from the centre, and whose surface is
kept at the constant temperature b.
Here we are to solve
Dt (ru) = a2 Dr2 (ru), (1)
see [V] Art. 1, subject to the conditions
u = f (r) when t=0 (2)

u=b “ r=c (3)
if c is the radius.
Let v = ru, then our equations become
Dt v = a2 Dr2 v (4)
v = rf (r) when t=0 (5)
v = bc “ r=c (6)
v=0 “ r = 0. (7)
Our problem is now precisely that of Art. 62 and we have as our solution
mπr w
m=∞ c
2 X m2 a2 π 2 mπλ
ru = e− c2 t sin λf (λ) sin dλ
c m=1 c c
0
m=∞
2c X (−1)m − m2 a22 π2 t

mπr
+b r+ e c sin . (8)
π m=1 m c
EXAMPLES.
1. If f (r) = b (8) Art. 66 reduces to u = b and there is no change of
temperature.
2. If the initial temperature is constant and equal to β

2c a2 π 2 πr 1 − 4a22π2 t 2πr
u=b+ (β − b) e− c2 t sin − e c sin
πr c 2 c

1 − 9a22π2 t 3πr
+ e c sin − ··· .
3 c
3. An iron sphere 40 cm. in diameter is heated to the temperature 100◦
centigrade throughout; its surface is then kept at the constant temperature 0◦ .
Find the temperature of a point 10 cm. from the centre, and find the temperature
of the centre, 15 minutes after cooling has begun. Given a2 = 0.185 in C.G.S.
units. Ans., 2◦ .1; 3◦ .3.
67. If instead of having the temperature of the surface of the sphere con-
stant, the sphere is placed in air which is kept at the constant temperature zero,
the problem is much more complicated. For in this case the surface temperature
can no longer be simply expressed but is given by a new differential equation
Dr u + hu = 0 when r = c, (1)
where h is an experimental constant depending upon what is called the surface
conductivity of the sphere.
Our equations, then, are
Dt (ru) = a2 Dr2 (ru) (2)
u = f (r) when t=0 (3)
Dr u + hu = 0 when r = c. (4)
As in Art. 66 let v = ru; then we have
Dt v = a2 Dr2 v (5)
v = rf (r) when t=0 (6)
v=0 “ r=0 (7)

1
Dr v + h − v=0 when r = c. (8)
c
2 2 2
α2 t
v = e−a α t cos αr and v = e−a sin αr have already been found as partic-
ular solutions of (5) (see Art. 60).
2
α2 t
v = e−a sin αr (9)
satisfies (7) for all values of α.

Substitute this value of v in (8) and we have
αc cos αc + (hc − 1) sin αc = 0. (10)
If αk is a value of α which is a root of the transcendental equation (10)

2
α2k t
v = e−a sin αk r (11)
will satisfy (5), (7), and (8).

It remains to see whether out of terms of the form given in (11) we can build
up a value of v which will satisfy (6).
When t = 0 the second member of (11) reduces to sin αk r. If then we can
express rf (r) as a sum of terms of the form bk sin αk r where αk is a root of (10)
X 2 2
v= bk e−a αk t sin αk r (12)
will satisfy all of the equations (5), (6), (7), and (8), and will be the required
solution.
Here, then, we have a new problem analogous to that of developing in a
Fourier’s Series, but rather more
X complicated, namely, to develop any function
of x in a series of the form am sin αm x where αm is a root of the equation
φm
(10); or if we call ac = φ and hc − 1 = p, where am = , φm being a root of
c
the equation
φ cos φ + p sin φ = 0 (13)
or more simply of
φ + p tan φ = 0; (14)
remembering that the series and the function must be equal for all values of x
between zero and c.
If φm is a root of (14) −φ
m is also
a root.
φm φm
Since sin x = − sin − x the terms of the required development
c c
which correspond to negative roots may be combined with those corresponding
to positive roots, and therefore we need consider only positive roots.
φ = 0 is a root of (14) but as sin 0 = 0 there will be no corresponding term
in the development. If we construct the curve
1
y=− x (15)
p
and the curve
y = tan x (16)
x
the abscissas of their points of intersection are values of x which satisfy +
p
tan x = 0, that is, are roots of equation (14). It is easy to see that there will
always be an infinite number of real positive roots, one for each of the branches
of the periodic curve y = tan x which lie to the right of the origin. The numerical
values of these roots can be obtained by an easy computation. The construction
suggested above shows that as m increases φm will rapidly approach the value
π
(2m − 1) if p is positive or if p is negative and numerically less than unity, and
2
π
(2m + 1) if p is negative and numerically greater than unity.
2
There exist, then, an infinite number of positive real roots of φ + p tan φ = 0
and consequently of
αc cos αc + (hc − 1) sin αc = 0.
68. The development called for in the last article can be obtained very
easily from a simpler one which we shall now consider, namely, to develop f (x)
into a series of the form
f (x) = a1 sin φ1 x + a2 sin φ2 x + a3 sin φ3 x + · · · (1)
where φ1 , φ2 , φ3 · · · are roots of the equation
φ cos φ + p sin φ = 0, (2)
the development to hold good for all values of x between x = 0 and x = 1.

1
Let us proceed as in Arts. 24 and 27. Call = ∆x and form n equations
n+1
by substituting for x in turn in the equation
f (x) = a1 sin φ1 x + a2 sin φ2 x + a3 sin φ3 x + · · · + an sin φn x (3)

the values ∆x, 2∆x, 3∆x, · · · n∆x; this being equivalent to making the values
of the sum and the function coincide for the n values of x substituted.
To determine any coefficient am multiply the first equation by ∆x. sin(φm
∆x), the second by ∆x. sin(2φm ∆x), the third by ∆x. sin(3φm ∆x), and so on,
the nth equation by ∆x. sin(nφm ∆x); add the equations and compute the lim-
iting values of the terms of the resulting equation as n is indefinitely increased.
This as in Art. 24 is seen to be equivalent to multiplying (3) by sin φm x.dx and
integrating between the limits x = 0 and x = 1.
The first member of the resulting equation is
w1
f (x) sin φm x.dx;
0
The coefficient of ak is
w1
sin φk x sin φm x.dx,
0
and of am is
w1
sin2 φm x.dx.
0
w1 1w
1
sin φk x sin φm x.dx = [cos(φk − φm )x − cos(φk + φm )x]dx
2
0 0

1 sin(φk − φm ) sin(φk + φm )
= −
2 φk − φm φk + φ m
φk cos φk sin φm − φm sin φk cos φm
=− (4)
φ2k − φ2m
But φk cos φk + p sin φk = 0

and φm cos φm + p sin φm = 0 by (2).
Hence the numerator of the second member of (4) is zero, and the coefficient
of ak vanishes if k is not equal to m.
w1 1 1

sin 2φm

sin2 φm x.dx = [φm − sin φm cos φm ] = 1− . (5)
2φm 2 2φm
0
2 w1
Therefore am = f (x) sin φm x.dx. (6)
sin 2φm
1− 0
2φm
The coefficient of the integral in (6) can be transformed as follows so as not

to involve trigonometric functions.
φm cos φm + p sin φm = 0, by (2)

p
φm cos2 φm + sin 2φm = 0,
2
sin 2φm cos2 φm
=− . (7)
2φm p
φ2m cos2 φm = p2 sin2 φm ,
(φ2m + p2 ) cos2 φm = p2 ,
cos2 φm p
= 2 . (8)
p φm + p2
Hence by (7) and (8)
sin 2φm φ2 + p(p + 1)

1− = m 2 ,
2φm φm + p 2
2(φ2m + p2 ) w
1
and am = 2 f (α) sin φm α.dα. (9)
φm + p(p + 1)
0
Therefore our required development is

m=∞
X 2(φ2m + p2 ) w1
f (x) = sin φ m x f (α) sin φ m α.dα . (10)
m=1
φ2m + p(p + 1)
0
From (10) it easily follows that for values of x between 0 and c
f (x) =a1 sin a1 x + a2 sin a2 x + a3 sin a3 x + · · · (11)

2 α2 c2 + p2 wc
where am = . 2 2m f (λ) sin αm λ.dλ, (12)
c αm c + p(p + 1)
0
and αm is a root of the equation
αc cos αc + p sin αc = 0. (13)
It is to be observed that if p is infinite (13) reduces to sin αc = 0, αm becomes

mπ
and (11) and (12) give our regulation Fourier sine series (v. Art. 31), and
c
therefore the ordinary Fourier development in sine series is merely a special case
of the problem just solved.
Moreover since the Fourier method of determining the coefficients of such a
series requires that
wc
sin αm x sin αn x.dx = 0,
0
sin(αm − αn )c sin(αm + αn )c
that is that − =0
αm − αn αm + αn
αm c cos αm c αn c cos αn c
or reducing, that = ,
sin αm c sin αn c
or that αm and αn should be roots of the equation
αc cos αc
=p
sin αc
where p is some constant, it follows that we have obtained in (11) the most
general sine development that can be obtained by Fourier’s method.
EXAMPLES.
1. Show that the solution of the problem of Art. 67 is

m=∞
2
α2m t
X
ru = bm e−a sin αm r,
m=1
c + (hc − 1)2 w
2 2 c
2 αm
where bm = . 2 2 λf (λ) sin αm λ.dλ
c αm c + hc(hc − 1)
0
and αm is a root of
αc cos αc + (hc − 1) sin αc = 0.
2. If the initial temperature of the sphere is constant and equal to β
m=∞
2
α2m t
X
ru = bm e−a sin αm r
m=1
2 2
αm c + (hc − 1)2 sin αm c
where bm = 2βh. 2 c2 + hc(hc − 1)
. 2
αm αm
1
2 2
2βhc [αm c + (hc − 1)2 ] 2
= . 2 2 .
αm αm c + hc(hc − 1)
3. If the temperature of the air is a constant γ instead of zero the surface

equation of condition is
Dr u + h(u − γ) = 0 when r = c.
The substitution of u1 = u − γ, however, brings the problem under Ex. 1

and we get
m=∞
X 2 2
r(u − γ) = bm e−a αm t sin αm r
m=1
2 αm c + (hc − 1)2 w
2 2 c
where bm = . 2 2 λ[f (λ) − γ] sin αm λ.dλ.
c αm c + hc(hc − 1)
0
4. An iron sphere 40 cm. in diameter is heated to the temperature 100◦
centigrade throughout; it is then allowed to cool in air which is kept at the
constant temperature 0◦ . Find the temperature at the centre; at a point 10 cm.
from the centre; and at the surface; 15 minutes after cooling has begun. Given
1
a2 = 0.185 and h = in C.G.S. units. (v. Ex. 3, Art. 66.)
800
Ans., 97◦ .67; 97◦ .36; 96◦ .46.
5. Show that if in the slab considered in Art. 60 one face is exposed to air
at the temperature zero, so that we have Dt u = a2 Dx2 u, u = 0 when x = 0,
u = f (x) when t = 0, and Dx u + hu = 0 when x = c, then
m=∞
2
α2m t
X
u= am e−a sin αm x
m=1
2
αm + h2 w c
where am = 2 2 c + h(hc + 1)
f (λ) sin αm λ.dλ,
αm
0
αm being a root of αc cos αc + hc sin αc = 0.
6. If in the problem of Art. 57 heat escapes from one side of the plate into
air at the temperature zero so that we have Dx2 u + Dy2 u = 0, u = 0 when x = 0,
u = f (x) when y = 0, and Dx u + hu = 0 when x = a, then
m=∞
X
u= am e−αm y sin αm x
m=1
2
αm + h2 w a
where am = 2 2 a + h(ha + 1)
f (λ) sin αm λ.dλ,
αm
0
αm being a root of αa cos αa + ha sin αa = 0.
7. If in the problem of Art. 59 there is leakage at one side of the sheet so

that we have Dx2 V + Dy2 V = 0, V = 0 when x = 0, V = 0 when y = b, V = f (x)
when y = 0, and Dx V + hV = 0 when x = a, then
m=∞
X sinh αm (b − y)
V = am sin αm x,
m=1
sinh αm b
where am has the value given in Ex. 6.
69. If we have an infinite solid with one plane face which is exposed to air
at the temperatures U = F (t) and heat can flow only at right angles to this face,
we can solve the problem readily for the case where the initial temperatures are
zero. We have
Dt u = a2 Dx2 u
u=0 when t = 0
and Dx u + h(U −u) = 0 when x = 0.
1
Let v = u − Dx u. (1)
h
Then v will satisfy the equation
Dt v = a2 Dx2 v,
and we shall also have v = U when x = 0.
2 w −β 2
∞
x2

Since U = F (t) v=√ e F t − 2 2 dβ (2)
π x√ 4a β
2a t
by Art. 51 (10).
Dx u − hu = −hv by (1).
w
Hence ue−hx = −h e−hx vdx + C;
v. Int. Cal. § 4, page 314.

Determining C by the fact that ue−hx = 0 when x = ∞ we have
w∞
u = hehx e−hx vdx. (3)
x
Substituting the value of v from (2) we have
2hehx w −hx w −β 2
∞ ∞
x2

u= √ e dx e F t − 2 2 dβ, (4)
π x x√
4a β
2a t
as our required solution.

For an extension of this method to the flow of heat in two and three dimen-
sions and for the interpretation of the results by the aid of the theory of Images,
see E. W. Hobson, Proc. Lond. Math. Soc., Vol. XIX.
EXAMPLES.
1. If the temperature of the air is a periodic function of the time, say

ρm sin(mαt + λm ) and we care only for the limiting value of u as t increases,
show that this value is
√
x mα
hρm e− a 2
r r
1 mα x mα
r h + sin mαt − + λ m
1 mα 2 mα a 2 a 2
h+ + 2
a 2 2a
r r
1 mα x mα
− cos mαt − + λm .
a 2 a 2
v. Art. 52 and Art. 51 Ex. 4.
w eax (a sin bx − b cos bx)
Note that eax sin bx.dx =
a2 + b2
w e ax
(a cos bx + b sin bx)
and eax cos bx.dx =
a + b2
2
v. Int. Cal. Table of Int. (235) and (236).
2. If Dx2 V + Dy2 V = 0, V = 0 when y = 0 and Dx V + h[F (y) − V ] = 0

when x = 0 show that
hehx w −hx w
∞ ∞
x x
V = e dx F (λ)dλ 2 − ;
π x x + (λ − y)2 x2 + (λ + y)2
0
v. Art. 47 Ex. 1.
70. The solution for an instantaneous heat source of strength Q at the

point x = λ if heat escapes at the origin into air at the temperature zero, so
that Dx u − hu = 0 when x = 0, can be obtained by the aid of Art. 53.
Let u = u1 + u2 where u1 is the temperature that would be due to the given
source if we had no boundary at the origin, so that
Q (λ−x)2
√ e− 4a2 t .
u1 = [Art. 53 (2)]
2a πt
Dx u − hu = Dx u1 − hu1 + Dx u2 − hu2 = 0 when x = 0.
Therefore Dx u2 − hu2 = −(Dx u1 − hu1 ) (1)
when x = 0.

Q λ−x (λ−x)2
But −(Dx u1 − hu1 ) = − √ 2
− h e− 4a2 t
2a πt 2a t

Q λ λ2
=− √ 2
− h e− 4a2 t
2a πt 2a t
when x = 0.
This is easily seen to be the value to which

Q λ+x (λ+x)2
− 4a2 t
− √ − h e
2a πt 2a2 t
reduces when x = 0, and this last expression is
Q (λ+x)2
(Dx + h) √ e− 4a2 t
2a πt
and therefore satisfies the equation
Dt u = a2 Dx2 u; (2)
Q (λ+x)2
since √ e− 4a2 t is the temperature due to a source at x = −λ.
2a πt
If, then, we determine u2 from the condition that

Q λ+x (λ+x)2
Dx u2 − hu2 = − √ 2
− h e− 4a2 t (3)
2a πt 2a t
taking care not to introduce any arbitrary constant or arbitrary function of t in
our integration, u2 will satisfy equation (2) and condition (1).
Integrating (3) [v. Int. Cal. § 4, page 314] and determining the constants of
integration suitably we get
Q

(λ+x)2
w∞ (λ+x)2

u2 = √ e− 4a2 t − 2hehx e−hx− 4a2 t dx . (4)
2a πt x
Therefore the solution of our problem is
Q

(λ−x)2 (λ+x)2
w∞ (λ+x)2

u= √ e− 4a2 t + e− 4a2 t − 2hehx e−hx− 4a2 t dx . (5)
2a πt x
If we replace Q by f (λ)dλ and integrate from 0 to ∞ we get as the solution

for the case where u = f (x) when t = 0 and x > 0, and Dx u − hu = 0 when
x=0
1 w w∞
∞
(λ−x)2 (λ+x)2 (λ+x)2
u= √ f (λ)dλ e− 4a2 t + e− 4a2 t − 2hehx e−hx− 4a2 t dx . (6)
2a πt 0 x
For an interpretation of this result by the theory of Images and the extension
of the method to the conduction of heat in n dimensions see G. H. Bryan, Proc.
Lond. Math. Soc., Vol. XXII.
EXAMPLE.
Show that if u = f (x) when t = 0 and Dx u + h[F (t) − u] = 0 when x = 0

we must take u equal to the sum of the second members of (6) Art. 70 and of
(4) Art. 69.
71. As another problem requiring a slight extension of Fourier’s Theo-

rem let us consider the vibration of a rectangular stretched elastic membrane
fastened at the edges, that is of a rectangular drumhead.
If two of the sides are taken as axes and the plane of equilibrium of the
membrane as the plane of XY the equation for the motion of the membrane is
Dt2 z = c2 (Dx2 z + Dy2 z) (1)
see [X] Art. 1.

Let the membrane be distorted at the start into some given form z = f (x, y)
and then allowed to swing. Our equations of conditions are then
z=0 when x=0 (2)

z=0 “ x=a (3)
z=0 “ y=0 (4)
z=0 “ y=b (5)
z = f (x, y) “ t=0 (6)
Dt z = 0 “ t = 0. (7)
We can get a particular solution of (1) by our usual device. Assume
z = eαx+βy+γt
and substitute in (1). We get γ 2 = c2 (α2 + β 2 ) as the only relation that need
hold between α, β, and γ, in order that z = eαx+βy+γt may be a solution. This
gives
p
γ = ± c α2 + β 2 .
√ 2 2
Therefore z = eαx+βy±ct α +β
is a solution of (1) no matter what values are given to α and β.

Replace α and β by αi and βi and we have
√ 2 2
z = e(αx+βy±ct α +β )i
as a solution, and from this we get

p
z = sin(αx + βy ± ct α2 + β 2 ) (8)
p
and z = cos(αx + βy ± ct α2 + β 2 ) (9)
as particular solutions of (1), α and β being unrestricted.

From (8) and (9) we can get solutions of the following forms
p 
z = sin αx sin βy sin ct α2 + β 2 


p 
2 2

z = sin αx sin βy cos ct α + β  


p 
2 2

z = sin αx cos βy sin ct α + β  

p 

2
z = sin αx cos βy cos ct α + β 2

p (10)
z = cos αx sin βy sin ct α2 + β 2 


p 

z = cos αx sin βy cos ct α2 + β 2 



p 

2
z = cos αx cos βy sin ct α + β 2 

p 


2
z = cos αx cos βy cos ct α + β ,2 
each of which will satisfy equation (1). The second of these will satisfy also (2),
(4) and (7) whatever values be taken for α and β. It will satisfy (3) and (5) if
mπ nπ
α and β are equal and respectively.
a b
If, then, we can so combine terms of the form
r
mπx nπy m2 n2
sin sin cos cπt +
a b a2 b2
as to satisfy (6) our problem will be completely solved.
This can be done if we can express f (x, y) as a sum of terms of the form
mπx nπy
A sin sin , the sum and the function being equal when x lies between
a b
0 and a and y between 0 and b.
mπx
f (x, y) can be expressed in terms of sin by Fourier’s Theorem if we
a
regard y as constant. We have
m=∞
X mπx
f (x, y) = am sin (11)
m=1
a
2w
a
mπλ
where am = f (λ, y) sin dλ. (12)
a a
0
f (λ, y) in (12) is a function of y and may be developed by Fourier’s Theorem.

n=∞
X nπy
We have f (λ, y) = bn sin (13)
n=1
b
2w
b
nπµ
where bn = f (λ, µ) sin dµ. (14)
b b
0
Substituting for f (λ, y) in (12) the value just obtained we have
22 X w wb
n=∞ a
mπλ nπµ nπy
am = dλ f (λ, µ) sin sin dµ sin
a b n=1 a b b
0 0
and
nπy w wb
m=∞ n=∞ a
4 X X mπx mπλ nπµ
f (x, y) = sin sin dλ f (λ, µ) sin sin dµ .
ab m=1 n=1 a b a b
0 0
(15)
m=∞
X n=∞
r
X m2 n2

mπx nπy
Hence z= Am,n sin sin cos cπt + , (16)
m=1 n=1
a b a2 b2
4 w w
a b
mπλ nπµ
where Am,n = dλ f (λ, µ) sin sin dµ. (17)
ab a b
0 0
EXAMPLES.
1. Show that if the membrane starts from its position of equilibrium but
with a given initial velocity impressed upon each point so that z = 0 when t = 0
and Dt z = F (x, y) when t = 0 the solution is
m=∞ n=∞
r
m2 n2

1 X X 1 mπx nπy
z= Am,n sin sin sin cπt + 2
a2
r
cπ m=1 n=1 m2 n2 a b b
2
+ 2
a b
4 w w
a b
mπλ nπµ
where Am,n = dλ F (λ, µ) sin sin dµ.
ab a b
0 0
2. If there is both initial distortion and initial velocity

m=∞ n=∞
r
m2 n2

4 X X mπx nπy
z= sin sin Am,n cos cπt +
ab m=1 n=1 a b a2 b2
r
m2 n2

+Bm,n sin cπt +
a2 b2
wa wb mπλ nπµ
where Am,n = dλ f (λ, µ) sin sin dµ,
a b
0 0
1 wa wb mπλ nπµ
and Bm,n = r dλ F (λ, µ) sin sin dµ.
m2
n 2 a b
0 0
cπ + 2
a2 b
3. Obtain a particular solution of (1) Art. 71 by assuming z = T.X.Y where
T is a function of t alone, X of x alone, and Y of y alone.
72. A number of interesting conclusions can be drawn from the results of

Art. 71 and Exs. 1 and 2.
(a) No one of the three values of z is in general a periodic function of t,
and consequently a vibrating rectangular membrane will not in general give a
musical note.
(b) A stretched rectangular membrane can be made to give a musical note
by starting the vibration properly. For if the initial circumstances are such that
the solution reduces to a single term, as will be the case if the initial distortion
mπx nπy
in the problem of Art. 71 be such that f (x, y) = Am,n sin sin , or the
a b
mπx nπy
initial velocity in Ex. 1 be such that F (x, y) = Bm,n sin sin , or the
a b
initial distortion and initial velocity in Ex. 2 be the values just given, then the
vibration will be periodic and will have the period
2
T = r . (1)
2
m n2
c +
a2 b2
Since T is a function of m and n and m and n are any whole numbers, the
same membrane is capable of giving a great variety of musical notes of different
pitches. If m and n are both unity we get the lowest note the membrane can
give, which is called its fundamental note. Its period
2 2ab
T1 = r = √ (2)
1 1 c a2 + b2
c + 2
a2 b
If m and n are both equal to k we get
2ab
Tk = √ ; (3)
kc a2 + b2
therefore the membrane can be made to give any harmonic of its fundamental
note.
More than this, since as we have seen
2
Tm,n = r
2
m n2
c 2
+ 2
a b
is the period of any note the membrane can give, and since if m and n are
replaced by mk and nk we get
2
Tmk,nk = r
m2 n2
ck 2
+ 2
a b
the membrane can sound all the harmonics of any note which it can give.
(c) In the case considered above, where the solution reduces to the single
term
" r r #
mπx nπy m2 n2 m2 n2
z = sin sin Am,n cos cπt + 2 + Bm,n sin cπt + 2 ,
a b a2 b a2 b
a 2a 3a (m − 1)a
if x = , or , or · · · or , z = 0 for all values of t, and the lines
m m m m
a 2a (m − 1)a
x= ,x= , ··· x = remain at rest during the whole vibration
m m m
and are nodes. The same thing is true of the lines
b 2b 3b (n − 1)b
y= , y = , y = , ··· y = .
n n n n
73. If the membrane is square it may have much more complicated nodes
than if the length and breadth are unequal, as in this case the period of any
term of the general solution reduces to
2a
T = √ (1)
c m2 + n2
and there will in general be two terms having the same period, and a musi-
cal note of the pitch corresponding to that period may be produced by initial
circumstances that bring in both terms. Thus

mπx nπy cπt p 2 cπt p 2
z = sin sin Am,n cos m + n2 + Bm,n sin m + n2
a a a a

nπx mπy cπt p cπt p 2
+ sin sin An,m cos m2 + n2 + Bn,m sin m + n2
a a a a
is a form of vibration that will give a musical note. Let us write this

cπt p 2 2
mπx nπy nπx mπy
z = cos m + n A sin sin + B sin sin
a a a a a

cπt p 2 2
mπx nπy nπx mπy
+ sin m + n C sin sin + D sin sin (2)
a a a a a
and in studying the forms of musical vibration of which the membrane is capable
we may take A, B, C, and D at pleasure. Consider the simple case where A = C
and B = D; then (2) reduces to

mπx nπy nπx mπy cπt p 2
z= A sin sin + B sin sin cos m + n2
a a a a a

cπt p 2
+ sin m + n2 . (3)
a
Values of x and y that will reduce the first parenthesis in (3) to zero will corre-
spond to points of the membrane remaining motionless during the vibration.
Let us consider a few cases at length.
(a) If m = 1 and n = 1, the first parenthesis in (3) becomes
πx πy
(A + B) sin sin ,
a a
which is equal to zero only when x = 0 or y = 0, or x = a or y = a, that
is, for the four edges of the membrane. If, then, the membrane is sounding its
fundamental note it has no nodes.
(b) If m = 1 and n = 2, we have
πx 2πy 2πx πy
A sin sin + B sin sin =0
a a a a
to give the nodes.
πx 2πy a
Let B = 0, then sin sin = 0, which is satisfied by y = ; and in
a a 2
a
addition to the edges the line y = is at rest and is a node.
2
a
If A = 0 x = is a node.
2
If A = B
πx 2πy 2πx πy
sin sin + sin sin =0
a a a a
πx πy πy πx πx πy
2 sin sin cos + 2 sin cos sin =0
a a a a a a
πx πy πy πx
sin sin cos + cos = 0.
a a a a
The first factor gives the four edges of the membrane. The second written equal
to zero gives
πy πx πx
cos = − cos = cos π −
a a a
πy πx
=π−
a a
x + y = a,
which is a diagonal of the square.

If B = −A
πx 2πy 2πx πy
sin sin − sin sin =0
a a a a
πy πx
cos = cos
a a
x − y = 0,
which is the other diagonal of the square.

Other relations between A and B will give Trigonometric curves of the form
πy B πx
cos = − cos
a A a
which are easily constructed and which obviously all agree in passing through
the middle point of the square.
We give the figures for a few of the cases
(c) If m = n = 2 we have
2πx 2πy
(A + B) sin sin =0
a a
to give the nodes, which are merely the lines
a a
x= , and y = .
2 2
This form gives the octave of the fundamental note.
(d ) If m = 1 and n = 3 we have
πx 3πy 3πx πy
A sin sin + B sin sin =0
a a a a
to give the nodes.
a 2a
If A=0 we get x= and x = (1)
3 3
a 2a
If B=0 we get y= and y = (2)
3 3
If A = −B we get
πx 3πy 3πx πy
sin sin − sin sin =0
a a a a
πx πy h πy πx i
sin sin 4 cos2 − 1 − 4 cos2 +1 =0
a a a a
πy πx
cos2 − cos2 =0
a a
πy πx πy πx
cos − cos cos + cos =0
a a a a
or x−y =0 and x + y = a. (3)
πy πx 1
If A=B we get cos2 + cos2 =
a a 2
2πy 2πx
or cos + cos = −1, (4)
a a
a Trigonometric curve easily constructed.
For other relations between A and B we get more complicated Trigonometric
curves coming under the general form
2πy 2πx A+B
A cos + B cos =− (5)
a a 2
which all agree in containing the points
a a a 2a 2a a 2a 2a
, , , , , , and , .
3 3 3 3 3 3 3 3
135
MISCELLANEOUS PROBLEMS.
I. Logarithmic Potential. Polar Coördinates.
1. Show that Dx2 V + Dy2 V = 0 becomes

1 1
Dr2 V + Dr V + 2 Dφ2 V = 0
r r
if we transform to Polar Coördinates.
1 1
2. If in Dr2 V + Dr V + 2 Dφ2 V = 0 (1)
r r
we let V = R.Φ we get
Φ = A cos αφ + B sin αφ Φ = Aeαφ + Be−αφ

or
R = A1 rα + B1 r−α R = A1 cos(α log r) + B1 sin(α log r);
whence
V = rα cos αφ V = eαφ cos(α log r) V = cosh αφ cos(α log r)
V = rα sin αφ V = eαφ sin(α log r) V = cosh αφ sin(α log r)
1
V = α cos αφ V = e−αφ cos(α log r) V = sinh αφ cos(α log r)
r
1
V = α sin αφ V = e−αφ sin(α log r) V = sinh αφ sin(α log r)
r
3. Show that if V satisfies (1) Ex. 2 and V = f (φ) when r = a

m=∞
X r m
1
V = b0 + (bm cos mφ + am sin mφ) for r < a
2 m=1
a
m=∞
X a m
1
and V = b0 + (bm cos mφ + am sin mφ) for r > a,
2 m=1
r
1 w 1 w
π π
where bm = f (φ) cos mφ.dφ and am = f (φ) sin mφ.dφ
π −π π −π
4. Show that if V satisfies (1) Ex. 2 and V = f (r) when φ = 0 and r > 0
1 w w∞ cosh α(π − φ)
∞
λ
V = f (e )dλ cos α(λ − log r).dα
π −∞ cosh απ
0
1
1 φ w
∞ cosh (λ − log r)
= sin λ
f (e ) 2 dλ.
π 2 −∞ cosh(λ − log r) − cos φ
MISCELLANEOUS PROBLEMS. 136
5. If V = 1 when φ = 0 and 0 < r < 1, and V = 0 when φ = 0 and r > 1

log r #)
sinh
( " " !#
1 π −1 2 1 π r−1
V = − tan = − tan−1 .
π 2 φ π 2 √ φ
sin 2 r. sin
2 2
6. If V = f (r) when φ = 0 and V = 0 when φ = β
1 w w∞ sinh(β − φ)α
∞
V = f (eλ )dλ cos α(λ − log r).dα
π −∞ sinh βα
0
πφ w
∞
1 f (eλ )dλ
= sin ,
2β β −∞ cosh π (λ − log r) − cos π φ
β β
if 0 < φ < β.
7. If V = 0 when φ = 0 and V = F (r) when φ = β
1 w w∞ sinh φα
∞
λ
V = F (e )dλ cos α(λ − log r).dα
π −∞ sinh βα
0
πφ w
∞
1 F (eλ )dλ
= sin .
2β β −∞ cosh (λ − log r) + cos π φ
π
β β
8. If V = χ(r) when φ = 0 and r < a, V = 0 when φ = β, and V = 0 when

r=a
πφ w
0
"
1 dλ
V = sin χ(aeλ )
2β β −∞ π r πφ
cosh λ − log − cos
β a β
#
dλ
− .
π r πφ
cosh λ + log − cos
β a β
π
9. If V = 0 when r = 1, V = 1 when φ = 0, V = 0 when φ =
2
1 − r2

2
V = tan−1 ctn φ .
π 1 + r2
π
10. If V = 0 when r = 1, V = 1 when φ = 0, V = 1 when φ =
2
1 − r4

2
V = tan−1 .
π 2r2 sin 2φ
11. If V = f (φ) when r = a, V = 0 when φ = 0, and V = 0 when φ = β

m=∞ r mπ
X β mπφ
V = am sin if r < a
m=1
a β
m=∞ a mπ
X β mπφ
V = am sin if r > a
m=1
r β
2w
β
mπφ
where am = f (φ) sin dφ and 0 < φ < β.
β β
0
12. If V = f (φ) when r = a, V = 0 when r = b, V = 0 when φ = 0, and

V = 0 when φ = β, then if a < r < b and 0 < φ < β
X a mπ
m=∞ mπ
β b β
mπ mπ
r β b β mπφ

V = 2mπ 2mπ − am sin
m=1 a β −b β b r β
2w
β
mπφ
where am = f (φ) sin dφ.
β β
0
13. If V = F (φ) when r = b, V = 0 when r = a, V = 0 when φ = 0, and

V = 0 when φ = β, then if a < r < b and 0 < φ < β
X a mπ
m=∞ mπ
β b β
mπ mπ
r β a β mπφ

V = 2mπ 2mπ − am sin
m=1 b β −a β a r β
2w
β
mπφ
where am = F (φ) sin dφ.
β β
0
14. If V = χ(r) when φ = 0, V = 0 when φ = β, V = 0 when r = a, and

V = 0 when r = b, then if a < r < b and 0 < φ < β
mπ(β − φ)
m=∞
( sinh )
X log b − log a mπ(log r − log a)
V = am sin
mπβ log b − log a
m=1 sinh
log b − log a
b
log a
2 w mπx
where am = χ(aex ) sin dx.
log b − log a log b − log a
0
15. If V = ψ(r) when φ = β, V = 0 when φ = 0, V = 0 when r = a, and

V = 0 when r = b, then if a < r < b and 0 < φ < β
mπφ
m=∞
( sinh )
X log b − log a mπ(log r − log a)
V = am sin
mπβ log b − log a
m=1 sinh
log b − log a
b
log a
2 w mπx
where am = ψ(aex ) sin dx.
log b − log a log b − log a
0
II. Potential Function in Space.

1. Show that
1 w w∞ w∞ w∞
∞
f (x, y) = dα dβ dλ f (λ, µ) cos α(λ − x) cos β(µ − y).dµ,
π2
0 0 0 0
for all values of x and y.
2. Find particular solutions of Dx2 V + Dy2 V + Dz2 V = 0 in the forms

√ 2 2
V = e±z α +β cos(αx ± βy)
√ 2 2
V = e±z α +β sin(αx ± βy)
p
V = sinh z α2 + β 2 . sin(αx ± βy)
p
V = cosh z α2 + β 2 . sin(αx ± βy)
&c.
3. Given Dx2 V + Dy2 V + Dz2 V = 0, and V = f (x, y) when z = 0, solve for
positive values of z.
1 w w∞
∞
zf (λ, µ)dµ
Result: V = dλ .
2π −∞ −∞ [z + (λ − x)2 + (µ − y)2 ] 23
2
4. Confirm the result of the last example by showing that if f (x, y) is

independent of y
1 w zf (λ, µ)dλ
∞
V = (v. Ex. 3 Art. 45).
π −∞ z 2 + (λ − x)2
5. If Dx2 V + Dy2 V + Dz2 V = 0, and V = 1 when z = 0 for all points within

the rectangle bounded by the lines x = a, x = −a, y = b, and y = −b; and
V = 0 when z = 0 for all points outside of this rectangle, then
2πV =
(a − x)2 (b − y)2 − z 2 [(a − x)2 + (b − y)2 + z 2 ]

b−y π 1
+ sin−1
(a − x)2 (b − y)2 + z 2 [(a − x)2 + (b − y)2 + z 2 ]
p
(b − y)2 2 2
(a + x)2 (b − y)2 − z 2 [(a + x)2 + (b − y)2 + z 2 ]

1
sin−1 +
2 (a + x)2 (b − y)2 + z 2 [(a + x)2 + (b − y)2 + z 2 ]
(a − x)2 (b + y)2 − z 2 [(a − x)2 + (b + y)2 + z 2 ]

b+y π 1
+ + sin−1
(a − x)2 (b + y)2 + z 2 [(a − x)2 + (b + y)2 + z 2 ]
p
(b + y) 2
2 2
(a + x)2 (b + y)2 − z 2 [(a + x)2 + (b + y)2 + z 2 ]

1
+ sin−1
2 (a + x)2 (b + y)2 + z 2 [(a + x)2 + (b + y)2 + z 2 ]
if −a < x < a, and
(a − x)2 (b − y)2 − z 2 [(a − x)2 + (b − y)2 + z 2 ]

b−y
4πV = p sin−1
(b − y)2 (a − x)2 (b − y)2 + z 2 [(a − x)2 + (b − y)2 + z 2 ]
(a + x)2 (b − y)2 − z 2 [(a + x)2 + (b − y)2 + z 2 ]

− sin−1
(a + x)2 (b − y)2 + z 2 [(a + x)2 + (b − y)2 + z 2 ]
(a − x)2 (b + y)2 − z 2 [(a − x)2 + (b + y)2 + z 2 ]

b+y
+p sin−1
(b + y)2 (a − x)2 (b + y)2 + z 2 [(a − x)2 + (b + y)2 + z 2 ]
(a + x)2 (b + y)2 − z 2 [(a + x)2 + (b + y)2 + z 2 ]

−1
− sin ;
(a + x)2 (b + y)2 + z 2 [(a + x)2 + (b + y)2 + z 2 ]
if x < −a or x > a.
6. If the value of the potential function V is given at every point of the base
of an infinite rectangular prism and if the sides of the prism are at potential
zero the value of V at any point within the prism is
m=∞ n=∞
4 X X −πz m22 + n22 mπx nπy
q
V = e a b sin sin
ab m=1 n=1 a b
wa wb mπλ nπµ
dλ f (λ, µ) sin sin dµ.
a b
0 0
If V = 1 on the base of the prism this reduces to
(2m + 1)πx (2n + 1)πy

m=∞ n=∞
16 X X −πz (2m+1)
q
2 (2n+1)2
sin sin
V = 2 e a2
+ b2 a b .
π m=0 n=0 (2m + 1)(2n + 1)
7. If the value of the potential function on five faces of a rectangular paral-

lelopiped, whose length, breadth, and height are a, b, and c, is zero, and if the
value of V is given for every point of the sixth face, then for any point within
the parallelopiped
r
m2 n2
m=∞
X n=∞ sinh π(c − z) + 2
V =
X
Am,n r a2 b sin mπx sin nπy
m=1 n=1 m 2
n 2 a b
sinh πc 2
+ 2
a b
4 w w
a b
mπλ nπµ
where Am,n = dλ f (λ, µ) sin sin dµ.
ab a b
0 0
8. If the value of the potential function is given on two opposite faces of a

rectangular parallelopiped and is zero on the four remaining faces, then within
the parallelopiped
r
m2 n2
m=∞
X n=∞ sinh π(c − z) +
V =
X
Am,n r a2 b2 sin mπx sin nπy
m=1 n=1 m 2
n 2 a b
sinh πc +
a2 b2
r
2
m n2
m=∞
X n=∞ sinh πz +
+
X
Bm,n ra
2 b2 sin mπx sin nπy
m=1 n=1 m 2
n2 a b
sinh πc 2
+ 2
a b
4 w w
a b
mπλ nπµ
where Am,n = dλ f (λ, µ) sin sin dµ
ab a b
0 0
4 w wb
a
mπλ nπµ
and Bm,n = dλ F (λ, µ) sin sin dµ.
ab a b
0 0
9. If the value of the potential function is given at every point on the

surface of a rectangular parallelopiped, what is its value at any point within the
parallelopiped?
III. Conduction of Heat in a Plane.

1. Find particular solutions of Dt u = a2 (Dx2 u + Dy2 u) of the forms
2
(α2 +β 2 )t
u = e−a sin(αx ± βy)
−a2 (α2 +β 2 )t
u=e cos(αx ± βy).
2. Given the initial temperature of every point in a thin plane plate, find
the temperature of any point at any time,
1 w∞ w∞ (λ−x)2 +(µ−y)2
u= 2
dλ e− 4a2 t f (λ, µ)dµ
4a πt −∞ −∞
1 w −β 2 w∞
∞
2 √ √
= e dβ e−γ f (x + 2a t.β, y + 2a t.γ)dγ.
π −∞ −∞
3. For an instantaneous source of strength Q at (λ, µ)

Q − (λ−x)2 +(µ−y)2
u= 2
e 4a2 t v. Art. 53.
4πa t
For an instantaneous doublet of strength P at (0, µ) with its axis perpendic-
ular to the axis of Y
P x − x2 +(µ−y) 2
u= e 4a2 t v. Art. 54.

8πa4 t2
For a permanent doublet of strength P at (0, µ) with its axis perpendicular

to the axis of Y
P x −
x2 +(µ−y)2
u= e 4a2 t .
2πa2 x2 + (µ − y)2
If the strength of the doublet were P dµ and the heat were uniformly gener-
ated and absorbed along the element dµ of the axis of Y beginning at (0, µ) we
should have
P − x2 +(µ−y) 2
xdµ P − x2 +(µ−y) 2
µ−y
u= e 4a2 t = e 4a2 t d tan−1 ,
2πa2 x2 + (µ − y)2 2πa2 x
µ−y
and since d tan−1 is the angle ARA0 , where A and A0 are the points (0, µ)
x
and (0, µ+dµ) and R is the point (x, y), u = 0 when x = 0 unless µ < y < µ+dµ,
P
in which case u = if x approaches zero from the positive side; and u = 0
2a2
when t = 0 except in the element dµ. If then u = 0 when t = 0 and u = f (y)
when x = 0 we have only to suppose a doublet of strength 2a2 f (x)dx placed in
each element of the axis of Y and then to integrate; we get
1 w − x2 +(µ−y)
∞ 2
xf (µ)
u= e 4a2 t dµ.
π −∞ x + (µ − y)2
2
For a permanent doublet of strength F (t) at (0, µ) we have
x w − x4a
t 2 +(µ−y)2
u= e 2 (t−τ )
(t − τ )−2 F (τ )dτ.
8πa4
0
1

xF (0) x2 +(µ−y)2
wt xF 0 (τ ) x2 +(µ−y)2

− − 4a2 (t−τ )
= e 4a 2t
+ e dτ .
2πa2 x2 + (µ − y)2 x2 + (µ − y)2
0
From the reasoning above this must be zero when t = 0 except at the point
(0, µ), must be 2a2 F (t) at the point (0, µ), and 0 at every other point of the
axis of Y when t is not zero.
Hence if u = 0 when t = 0 and u = F (y, t) when x = 0
1 w 1 w wt xD F (µ, τ )
∞ ∞
x2 +(µ−y)2
xF (µ, 0) −
x2 +(µ−y)2
τ − 4a2 (t−τ )
u= e 4a2 t dµ + dµ e dτ.
π −∞ x2 + (µ − y)2 π −∞ x2 + (µ − y)2
0
For an extension of this solution by the method of images to the case where
there are other rectilinear boundaries and for its application to the correspond-
ing problems in the flow of heat in three dimensions see E. W. Hobson in Vol.
XIX. Proc. Lond. Math. Soc.
4. If the perimeter of a thin plane rectangular plate is kept at the temper-

ature zero and the initial temperatures of all points of the plate are given, then
for any point of the plate

m=∞ n=∞
4 X X −a2 π2 mb22 + nc22 t mπx nπy
“ ”
u= e sin sin
bc m=1 n=1 b c
wb wc mπλ nπµ
dλ f (λ, µ) sin sin dµ.
b c
0 0
if b is the length and c the breadth of the plate.
5. A large mass of iron at the temperature 0◦ contains an iron core in the

shape of a long prism 40 cm. square. The core is removed and heated to the
temperature of 100◦ throughout and then replaced. Find the temperature of
a point in the axis of the core fifteen minutes afterward. Given a2 = .185 in
C.G.S. units. Ans., 52◦ .9.
6. If the prism described in Ex. 5 after being heated to 100◦ has its lateral
faces kept for 15 minutes at the temperature 0◦ find the temperature of a point
in its axis. Ans., 20◦ .8.
IV. Conduction of Heat in Space.
1. Show that
1 w w∞ w∞ w∞ w∞ w∞
∞
dα dβ dγ dλ dµ f (λ, µ, ν)
π3 −∞ −∞ −∞
0 0 0
cos α(λ − x) cos β(µ − y) cos γ(ν − z).dν = f (x, y, z)
for all values of x, y, and z.
2. Show that
X p=∞
X n=∞
m=∞ X mπx nπy pπz
f (x, y, z) = Am,n,p sin sin sin
m=1 n=1 p=1
a b c
8 wa wb wc mπλ nπµ pπν

where Am,n,p = dλ dµ f (λ, µ, ν) sin sin sin dν,
abc a b c
0 0 0
for 0 < x < a, 0 < y < b, 0 < z < c.
3. Obtain particular solutions of Dt u = a2 (Dx2 u + Dy2 u + Dz2 u) of the forms

2
(α2 +β 2 +γ 2 )t
u = e−a sin(αx ± βy ± γz).
2 2 2 2
u = e−a (α +β +γ )t
cos(αx ± βy ± γz).
4. Given the initial temperature of every point in an infinite homogeneous

solid find the temperature of any point at any time.
1 w∞ w∞ w∞ (λ−x)2 +(µ−y)2 +(ν−z)

u= 3 dλ dµ e− 4a2 t f (λ, µ, ν)dν
8a3 (πt) 2
−∞ −∞ −∞
1 w w∞ w∞
∞
2 2 2 √ √ √
= 3 e−β dβ e−γ dγ e−δ f (x + 2a t.β, y + 2a t.γ, z + 2a t.δ)dδ.
π 2
−∞ −∞ −∞
5. If the surface of a rectangular parallelopiped is kept at the temperature

zero and the initial temperatures of all points of the parallelopiped are given,
then for any point of the parallelopiped
X p=∞
X n=∞
m=∞
−a2 π 2
“
m2
+n
2 p2
+d
”
mπx nπy pπz
2 t
X
u= Am,n,p e b2 c2 sin sin sin
m=1 n=1 p=1
b c d
8 w w w
b c d
mπλ nπµ pπν
where Am,n,p = dλ dµ f (λ, µ, ν) sin sin sin dν.
bcd b c d
0 0 0
6. An iron cube 40 cm. on an edge is heated to the uniform temperature

of 100◦ Centigrade and then tightly enclosed in a large iron mass which is at
the uniform temperature of 0◦ . Find the temperature of the centre of the cube
fifteen minutes afterwards. Ans., 38◦ .4.
7. An iron cube 40 cm. on an edge is heated to the uniform temperature

of 100◦ and then its surface is kept for fifteen minutes at the temperature 0◦ .
Required the temperature of its centre. Ans., 9◦ .5.
144
CHAPTER V.1
ZONAL HARMONICS.
74. In Art. 16 we obtained
z = Apm (x) + Bqm (x) (1)
[v. (6) Art. 16] as the general solution of Legendre’s Equation
d2 z dz
(1 − x2 ) − 2x + m(m + 1)z = 0, (2)
dx2 dx
m being wholly unrestricted in value and x lying between −1 and 1; where
m(m + 1) 2 m(m − 2)(m + 1)(m + 3) 4

pm (x) = 1 − x + x
2! 4!
m(m − 2)(m − 4)(m + 1)(m + 3)(m + 5) 6
− x + ··· (3)
6!
and
(m − 1)(m + 2) 3 (m − 1)(m − 3)(m + 2)(m + 4) 5

qm (x) = x − x + x
3! 5!
(m − 1)(m − 3)(m − 5)(m + 2)(m + 4)(m + 6) 7
− x + ··· ; (4)
7!
and we found V = rm pm (cos θ) 



1


V = m+1 pm (cos θ) 


r (5)
V = rm qm (cos θ) 


1



V = m+1 qm (cos θ),
r
m being unrestricted in value, as particular solutions of the special form assumed
by Laplace’s Equation in spherical coördinates when V is independent of φ; that
is, of the equation
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0. (6)
sin θ
For the important case where m is a positive integer we found
z = APm (x) + BQm (x) (7)

1 Before reading this chapter the student is advised to re-read carefully articles 9, 10, 13(c),
15, 16, and 18(c).

ZONAL HARMONICS. 145
[v. (10) Art. 16] as the general solution of Legendre’s Equation (2), whence
V = rm Pm (cos θ) 


1


V = m+1 Pm (cos θ) 


r (8)
V = rm Qm (cos θ)  

1



V = m+1 Qm (cos θ)
r
are particular solutions of (6) if m is a positive integer.

(2m − 1)(2m − 3) · · · 1 m m(m − 1) m−2
Pm (x) = x − x
m! 2(2m − 1)

m(m − 1)(m − 2)(m − 3) m−4
+ x − ··· (9)
2.4.(2m − 1)(2m − 3)
[v. (8) Art. 16] and is a finite sum terminating with the term which involves x
if m is odd and with the term involving x0 if m is even.
It is called a Surface Zonal Harmonic, or a Legendre’s Coefficient, or more
briefly a Legendrian.

m! 1 (m + 1)(m + 2) 1
Qm (x) = m+1
+
(2m + 1)(2m − 1) · · · 1 x 2.(2m + 3) xm+3

(m + 1)(m + 2)(m + 3)(m + 4) 1
+ + ··· (10)
2.4.(2m + 3)(2m + 5) xm+5
if x < −1 or x > 1. [v. (9) Art. 16.]
It is called a Surface Zonal Harmonic of the second kind.
h m + 1 i2
m−1
2 Γ
m+1
Qm (x) = (−1) 2 2 pm (x)
Γ(m + 1)
m+1 2.4.6. · · · (m − 1)
= (−1) 2 pm (x) (11)
3.5.7. · · · m
[v. (13) Art. 16] if m is odd and −1 < x < 1.
h m + 1 i2
m
2 Γ
m
Qm (x) = (−1) 2 2 qm (x)
Γ(m + 1)
m 2.4.6. · · · m
= (−1) 2 qm (x) (12)
1.3.5. · · · (m − 1)
[v. (14) Art. 16] if m is even and −1 < x < 1.
In most of the work that immediately follows we shall regard x in Pm (x)as
equal to cos θ and therefore as lying between −1 and 1.2
2 English writers on Spherical Harmonics generally use µ in place of x for cos θ. We shall
follow them, however, only when we should thereby avoid confusion.

75. In Article 9 the undetermined coefficient am of xm in Pm (x) was

(2m − 1)(2m − 3) · · · 1
arbitrarily written in the form for reasons which shall
m!
now be given.
In Articles 9 and 16 z = Pm (x) was obtained as a particular solution of
Legendre’s Equation
d2 z dz
(1 − x2 ) 2
− 2x + m(m + 1)z = 0 (1)
dx dx
by the device of assuming that z could be expressed as a sum or a series of terms
of the form an xn and then determining the coefficients. We can, however, obtain
a particular solution of Legendre’s Equation by an entirely different method.
The potential function due to a unit of mass concentrated at a given point
(x1 , y1 , z1 ) is
1
V =p (2)
(x − x1 )2 + (y − y1 )2 + (z − z1 )2
and this must be a particular solution of Laplace’s Equation
Dx2 V + Dy2 V + Dz2 V = 0, (3)
as is easily verified by direct substitution.

If we transform (2) to spherical coördinates using the formulas of transfor-
mation
x = r cos θ
y = r sin θ cos φ
z = r sin θ sin φ we get
1
V =p (4)
r2 − 2rr1 [cos θ cos θ1 + sin θ sin θ1 cos(φ − φ1 )] + r12
as a solution of Laplace’s Equation in Spherical Coördinates
1 1
rDr2 (rV ) + Dθ (sin θDθ V ) + Dφ 2 V = 0 [XIII] Art. 1.
sin θ sin2 θ
If the given point (x1 , y1 , z1 ) is taken on the axis of X, as it must be that
(4) may be independent of φ, θ1 = 0, and
1
V =p (5)
r2 − 2rr1 cos θ + r12
is a solution of
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0. (6)
sin θ
Equation (5) may be written

1 1
V = r (7)
r r1 r2
1 − 2 cos θ + 12
r r
1 1
or V = s . (8)
r1 r r2
1 − 2 cos θ + 2
r1 r1
√
1 − 2z cos θ + z 2 is finite and continuous for all values real or complex of
z. It is double-valued but the two branches of the function are distinct except
for the values of z which make 1 − 2z cos θ + z 2 = 0 namely z = cos θ + i sin θ
and z = cos θ − i sin θ, both of which have the modulus unity and which are
critical values.
1
√ is finite and continuous except for the values of z =
1 − 2z cos θ + z 2
cos θ − i sin θ and z = cos θ + i sin θ for which it becomes infinite; it is double-
valued but has as critical values only these values of z. It is then holomorphic
within a circle described with the origin as centre and the radius unity, and can
be developed into a power series which will be convergent for all values of z
having moduli less than one. (Int. Cal. Arts. 207, 212, 214, 220.)
1
If then r > r1 r can be developed into a convergent series
2r1 r12
1− cos θ + 2
r r
r1
involving whole powers of .
r
X rm
Let pm 1m be this series, pm , of course, being a function of cos θ. Then
r
1X rm
V = pm 1m
r r
[v. (7)] is a solution of (6). Substitute this value of V in (6) and we get
X rm r1m 1 d

dpm

1
m(m + 1)pm + m+1 sin θ = 0.
rm+1 r sin θ dθ dθ
As this must hold whatever the value of r provided r > r1 the coefficient of each
power of r must be zero, and hence the equation

1 d dpm
sin θ + m(m + 1)pm = 0 (9)
sin θ dθ dθ
must be true.
But as we have seen in Art. 9 the substitution of x = cos θ in (9) reduces it
to
d2 pm dpm
(1 − x2 ) 2
− 2x + m(m + 1)pm = 0,
dx dx
and therefore z = pm
is a solution of Legendre’s Equation (1).
1
If r < r1 s can be developed into a convergent series
2r r2
1− cos θ + 2
r1 r1
2
r
involving whole powers of 2 .
r1
X rm
Let pm m be this series. Then
r1
1 X rm
V = pm m
r1 r1
(v. 8) is a solution of (6); substituting in (6) we get
X rm rm 1 d

dpm

m(m + 1)pm + m+1 sin θ = 0,
r1m+1 r1 sin θ dθ dθ
whence it follows as before that
z = pm
is a solution of Legendre’s Equation.
r
But pm is the coefficient of the mth power of in the development of
r1
1
−2
r2

r r r1
1 − 2 cos θ + 2 according to powers of , or of the mth power of
r1 r1 r1 r
1
2 −2
r1 r r1
in the development of 1 − 2 cos θ + 12 according to powers of , or
r r r
more briefly it is the coefficient of the mth power of z in the development of
1
(1 − 2xz + z 2 )− 2 according to powers of z, x standing for cos θ.
1 1
(1 − 2xz + z 2 )− 2 = [1 − z(2x − z)]− 2
and can be developed by the Binomial Theorem; the coefficient of z m is easily
picked out and is

(2m − 1)(2m − 3) · · · 1 m m(m − 1) m−2
x − x
m! 2(2m − 1)

m(m − 1)(m − 2)(m − 3) m−4
+ x − ··· .
2.4.(2m − 1)(2m − 3)
But this is precisely Pm (x). [v. Art. 74 (9)]
Hence Pm (x) is equal to the coefficient of the mth power of z in the devel-
1
opment of [1 − 2xz + z 2 ]− 2 into a power series, the modulus of z being less than
unity.
1
76. If x = 1 Pm (x) = 1. For if x = 1 (1 − 2xz + z 2 )− 2 reduces to
1
(1 − 2z + z 2 )− 2 that is to (1 − z)−1 , which develops into
1 + z + z2 + z3 + z4 + · · · ,
and the coefficient of each power of z is unity. Therefore
Pm (1) = 1. (1)
We have seen that if m is even Pm (x) contains only even powers of x and
terminates with the term involving x0 , that is with the constant term.
The value of this constant term can be picked out from the formula for Pm (x)
m 1.3.5. · · · (m − 1)
[v. Art. 74 (9)]. It is (−1) 2 ; or it can be found as follows:—It
2.4.6. · · · m
is clearly the value Pm (x) assumes when x = 0; it is, then, the coefficient of z m
1
in the development of (1 + z 2 )− 2 ; but
1 1 1.3 4 1.3.5 6 1.3.5.7 8
(1 + z 2 )− 2 = 1 − z 2 + z − z + z − ···
2 2.4 2.4.6 2.4.6.8
m1.3.5 · · · (m − 1)
and the coefficient of z m , m being an even number, is (−1) 2 .
2.4.6 · · · m
If m is odd Pm (x) contains only odd powers of x and terminates with the
term involving x to the first power. The coefficient of this term can be picked
m−1 3.5.7. · · · m
out from (9) Art. 74 and is (−1) 2 ; or it can be found as
2.4.6. · · · (m − 1)
dPm (x)
follows:—It is clearly the value assumed by when x = 0.
dx
z
It is, then, the coefficient of z m in the development of 3 .
(1 + z 2 ) 2
z 3 3.5 5 3.5.7 7
3 = z − z3 + z − z + ···
(1 + z2) 2 2 2.4 2.4.6
m−1 3.5.7 · · · m
and the coefficient of z m in this development is (−1) 2 , m
2.4.6 · · · (m − 1)
being an odd number.
77. To recapitulate:

1.3.5 · · · (2m − 1) m m(m − 1) m−2
Pm (x) = x − x
m! 2(2m − 1)
m(m − 1)(m − 2)(m − 3) m−4
+ x
2.4.(2m − 1)(2m − 3)

m(m − 1)(m − 2)(m − 3)(m − 4)(m − 5) m−6
− x + ··· , (1)
2.4.6.(2m − 1)(2m − 3)(2m − 5)
m being a positive integer, is a Surface Zonal Harmonic or Legendrian of the
mth order. It is a finite sum terminating with the first power of x if m is odd,
and with the zeroth power of x if m is even.
Pm (x) is the coefficient of the mth power of z in the development of (1 −

1
2xz + z 2 )− 2 into a power series. Hence if z < 1
1
(1 − 2xz + z 2 )− 2 = P0 (x) + P1 (x).z + P2 (x).z 2 + P3 (x).z 3
+ P4 (x).z 4 + P5 (x).z 5 + · · · + Pm (x).z m + · · · . (2)
Whence
r12

1 1 r1
p = P0 (cos θ) + P1 (cos θ) + 2 P2 (cos θ) + · · · 

r2 − 2rr1 cos θ + r12 r r r





m

r1 

+ m Pm (cos θ) + · · · if r > r1 
r

(3)
r2

1 r 
= P0 (cos θ) + P1 (cos θ) + 2 P2 (cos θ) + · · · 


r1 r1 r1 



m

r


+ m Pm (cos θ) + · · · if r < r1 .


r1
z = Pm (x)
is a solution of Legendre’s Equation
d2 z dz
(1 − x2 ) − 2x + m(m + 1)z = 0
dx2 dx
when m is a positive integer.
V = rm Pm (cos θ)
1
and V = m+1 Pm (cos θ)
r
are solutions of the form of Laplace’s Equation in Spherical Coördinates which
is independent of φ, namely
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0. (4)
sin(θ)
Pm (1) = 1. (5)
P2m (−x) = P2m (x). (6)
P2m+1 (−x) = −P2m+1 (x). (7)
P2m+1 (0) = 0. (8)
1.3.5. · · · (2m − 1)
P2m (0) = (−1)m . (9)
2.4.6. · · · 2m

dP2m+1 (x) 3.5.7. · · · (2m + 1)
= (−1)m . (10)
dx x=0 2.4.6. · · · 2m
For convenience of reference we write out a few Zonal Harmonics. They are
obtained by substituting successive integers for m in formula (1).

P0 (x) = 1 



P1 (x) = x





1 2


P2 (x) = (3x − 1)


2




1 3


P3 (x) = (5x − 3x)


2




1 4 2


P4 (x) = (35x − 30x + 3)

8 (11)
1


5 3
P5 (x) = (63x − 70x + 15x)



8 


1


6 4 2
P6 (x) = (231x − 315x + 105x − 5)



16 


1


7 5 3
P7 (x) = (429x − 693x + 315x − 35x)



16 


1


8 6 4 2
P8 (x) = (6435x − 12012x + 6930x − 1260x + 35).

128
Any Surface Zonal Harmonic may be obtained from the two of next lower
orders by the aid of the formula
(n + 1)Pn+1 (x) − (2n + 1)xPn (x) + nPn−1 (x) = 0 (12)
which is easily obtained and is convenient when the numerical value of x is

given.
Differentiate (2) with respect to z and we get
−(z − x)
3 = P1 (x) + 2P2 (x).z + 3P3 (x).z 2 + · · ·
(1 − 2xz + z 2 ) 2
whence
−(z − x)
1 = (1 − 2xz + z 2 )(P1 (x) + 2P2 (x).z + 3P3 (x).z 2 + · · · ).
(1 − 2xz + z 2 ) 2
Hence by (2)
(1 − 2xz + z 2 )(P1 (x) + 2P2 (x).z + 3P3 (x).z 2 + · · · )

+ (z − x)(P0 (x) + P1 (x).z + P2 (x).z 2 + · · · ) = 0 (13)
(13) is identically true, hence the coefficient of each power of z must vanish.
Picking out the coefficient of z n and writing it equal to zero we have formula
(12) above.3
3 For tables of Surface Zonal Harmonics v. Appendix Tables I and II.

78. We are now able to solve completely the problem considered in Art. 9.
We were to find a solution of the differential equation
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0 (1)
sin θ
subject to the condition
M
V = 1 when θ = 0. (2)
(c2 + r2 ) 2
We know (v. Art. 77) that
V = rm Pm (cos θ)
1
and V = m+1 Pm (cos θ)
r
are solutions of (1).
For values of r < c
M Mh 1 r2 1.3 r4 1.3.5 r6 i
1 = 1− + − + · · · . (3)
(c2 + r2 ) 2 c 2 c2 2.4 c4 2.4.6 c6
Therefore for values of r < c
Mh 1 r2
V = P0 (cos θ) − P2 (cos θ)
c 2 c2
1.3 r4 1.3.5 r6 i
+ P 4 (cos θ) − P 6 (cos θ) + · · · (4)
2.4 c4 2.4.6 c6
is our required solution; because each term satisfies equation (1), and therefore
the whole value satisfies (1), and when θ = 0
Pm (cos θ) = Pm (1) = 1
[v. (5) Art. 77], and hence (4) reduces to (3) and (2) is satisfied.
For values of r > c
M Mh 1 c2 1.3 c4 1.3.5 c6 i
1 = 1− 2
+ 4
− 6
+ ··· (5)
(c2 + r2 ) 2 r 2r 2.4 r 2.4.6 r
h 1 1 c2 1.3 c 4
1.3.5 c6 i
=M − + − + · · · .
r 2 r3 2.4 r5 2.4.6 r7
Therefore for values of r > c
M hc 1 c3
V = P0 (cos θ) − P2 (cos θ)
c r 2 r3
1.3 c5 1.3.5 c7 i
+ P 4 (cos θ) − P 6 (cos θ) + · · · (6)
2.4 r5 2.4.6 r7
is our required solution. For it satisfies (1) and reduces to (2) when θ = 0.
79. As another example let us suppose a conductor in the form of a thin

circular disc charged with electricity, and let it be required to find the value of
the potential function at any point in space.
If the magnitude of the charge is M and the radius of the plate is a the
surface density at a point of the plate at a distance r from the centre is
M
σ= √
4aπ a2 − r2
πM
and all points of the conductor are at the potential . (v. Peirce’s Newtonian
2a
Potential Function, § 61.)
The value of the potential function at a point in the axis of the plate at the
distance x from the plate is easily seen to be
Mw
a
rdr
V = p
a (a2 − r 2 )(x2 + r 2 )
0
M x2 − a 2
= cos−1 2 .
2a x + a2
d M x2 − a2 M
cos−1 2 2
=− 2
dx 2a x +a a + x2
Mh x2 x4 x6 i
=− 2 1− + − + · · ·
a a2 a4 a6
if x < a,
Mh a2 a4 a6 i
=− 2 1− + − + · · ·
x x2 x4 x6
if x > a.
Integrating and then determining the arbitrary constant we have
M x2 − a2 M hπ x x3 x5 x7 i
cos−1 2 = − + − + − · · ·
2a x + a2 a 2 a 3a3 5a5 7a7
if x < a,
M ha a3 a5 a7 i
= − 3 + 5 − 7 + ···
a x 3x 5x 7x
if x > a.
We have, then, to solve the equation

1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0
sin θ
r3 r5 r7

M π r
V = − + 3 − 5 + 7 − ···
a 2 a 3a 5a 7a
when θ=0 and r < a
3
a5 a7

M a a
and V = − 3 + 5 − 7 + ···
a r 3r 5r 7r
when θ=0 and r > a.
The required solution is easily seen to be
1 r3 1 r5

M π r
V = − P1 (cos θ) + P3 (cos θ) − P5 (cos θ) + · · ·
a 2 a 3 a3 5 a5
π
if r < a and θ < ,
2
M a 1 a3 1 a5 1 a7

and V = − P 2 (cos θ) + P 4 (cos θ) − P 6 (cos θ) + · · ·
a r 3 r3 5 r5 7 r7
if r > a.
EXAMPLES.
1. Given that if a charge M of electricity is placed on an ellipsoidal con-

Mp
ductor the surface density at any point P of the conductor is equal to ,
4πabc
where p is the distance from the centre of the conductor to the tangent plane
at P (v. Peirce, New. Pot. Func. § 61); find the value of the potential function
at any external point when the conductor is the oblate spheroid generated by
x2 y2
the rotation of the ellipse 2 + 2 = 1 about its minor axis.
a b
Ans. (1) If the point is on the axis of revolution
bx + a2 − b2 bx − a2 + b2

M −1 −1
V = √ sin √ − sin √
2 a2 − b2 a x2 + a2 − b2 a x2 + a2 − b2
x being the distance from the centre.
(2) If the point is on the surface of the spheroid
2
2b − a2

M π −1 M π −1 b
V = √ − sin =√ − tan √ .
2 a2 − b2 2 a2 a2 − b2 2 a2 − b2
√
(3) If the distance r of the point from the centre is less than a2 − b2 and
π
θ<
2

M π r
V =√ − 1 P1 (cos θ)
2
a −b 2 2 (a − b2 ) 2
2
r3 r5

+ 3 P3 (cos θ) − 5 P5 (cos θ) + · · · .
3(a2 − b2 ) 2 5(a2 − b2 ) 2
√
(4) If the distance r of the point from the centre is greater than a2 − b2
2 1 3
M (a − b2 ) 2 (a2 − b2 ) 2
V =√ − P2 (cos θ)
a2 − b2 r 3r3
5 7
(a2 − b2 ) 2 (a2 − b2 ) 2

+ P 4 (cos θ) − P 6 (cos θ) + · · · .
5r5 7r7
2. If the conductor is the prolate spheroid generated by the rotation of the
x2 y2
ellipse 2 + 2 = 1 about its major axis, show that if the point is an external
a b
point and is on the axis at a distance x from the centre,
√
M x + a2 − b2
V = √ log √ .
2 a2 − b2 x − a2 − b2
√
If the point is not on the axis and r > a2 − b2
2 1 3
M (a − b2 ) 2 (a2 − b2 ) 2
V =√ + P2 (cos θ)
a2 − b2 r 3r3
5 7
(a2 − b2 ) 2 (a2 − b2 ) 2

+ P 4 (cos θ) + P 6 (cos θ) + · · · .
5r5 7r7
80. As a third example we will find the value of the potential function
due to a thin homogeneous circular disc, of density ρ, thickness k, and radius a.
The value of V at a point in the axis of the disc at a distance x from its
centre is readily found and proves to be
p 2M p
V0 = 2πρk( x2 + a2 − x) = 2 [ x2 + a2 − x].
a
If x > a
1
a2 2
p
2 2
x +a =x 1+ 2
x
1 a2 1.1 a4 1.1.3 a6 1.1.3.5 a8

=x 1+ − + − + · · ·
2 x2 2.4 x4 2.4.6 x6 2.4.6.8 x8
3 5
1.1.3.5 a7

2M 1 a 1.1 a 1.1.3 a
and V0 = − + − + · · · .
a 2 x 2.4 x3 2.4.6 x5 2.4.6.8 x7
If x < a
1
x2 2
p
x2 + a2 = a 1 + 2
a
1 x2 1.1 x4 1.1.3 x6

=a 1+ − + + · · ·
2 a2 2.4 a4 2.4.6 a6
x 1 x2 1.1 x4 1.1.3 x6 1.1.3.5 x8

2M
and V0 = 1− + − + − + ··· .
a a 2 a2 2.4 a4 2.4.6 a6 2.4.6.8 a8
Hence the solution for any external point is
1 a 1.1 a3

2M
V = − P2 (cos θ)
a 2r 2.4 r3
1.1.3 a5 1.1.3.5 a7

+ P 4 (cos θ) − P 6 (cos θ) + · · ·
2.4.6 r5 2.4.6.8 r7
if r > a, and
2M h r
V = 1 − P1 (cos θ)
a a
1 r2 1.1 r4 1.1.3 r6

+ P2 (cos θ) − P4 (cos θ) + P6 (cos θ) − · · ·
2 a2 2.4 a4 2.4.6 a6
π
if r < a and θ < .
2
EXAMPLES.
1. The potential function due to a homogeneous hemisphere whose axis is

taken as the polar axis, is
a 3.1 a2

M
V = + P1 (cos θ)
a r 2.4 r2
3.1.1 a4 3.1.1.3 a6

− P 3 (cos θ) + P5 (cos θ) − · · ·
2.4.6 r4 2.4.6.8 r6
if r > a, and is
r2

M 3 3r
V = + P1 (cos θ) + 2 P2 (cos θ)
a 2 2a a
3.1 r3 3.1.1 r5

+ P3 (cos θ) − P5 (cos θ) + · · ·
2.4 a3 2.4.6 a5
π
if r < a and θ > .
2
2. The potential function due to a solid sphere whose density is proportional

to the distance from a diametral plane is, at an external point,
8 M 5.3 a 5.3.1 a3

V = + P2 (cos θ)
15 a 2.4 r 2.4.6 r3
5.3.1.1 a5 5.3.1.1.3 a7

− P4 (cos θ) + P6 (cos θ) − · · · .
2.4.6.8 r5 2.4.6.8.10 r7
3. The potential function due to the homogeneous oblate spheroid gener-
x2 y2
ated by the rotation of 2 + 2 = 1 about its minor axis is, at an external
a b
point,
2
x + a2 − b2 (a2 − b2 + bx)

3 M
V = 2 2 1 sin−1 √
2 (a − b ) 2(a2 − b2 ) 2 a x2 + a2 − b2
2 2

−1 (a − b − bx)
+ sin √ −x
a x2 + a2 − b2
if the point is on the axis of the spheroid at a distance x from its centre.
" 1 3
3M 1 (a2 − b2 ) 2 1 (a2 − b2 ) 2
V = 1 − P2 (cos θ)
(a2 − b2 ) 2 1.3 r 3.5 r3
5
#
1 (a2 − b2 ) 2
+ P4 (cos θ) − · · ·
5.7 r5
1
if r > (a2 − b2 ) 2 , and
r2

3M π r π
V = 1 − 1 P 1 (cos θ) + P2 (cos θ)
(a2 − b2 ) 2 4 (a2 − b2 ) 2 4 (a2 − b2 )
r3 r5

1 1
− P 3 (cos θ) + P 5 (cos θ) − · · ·
1.3 (a2 − b2 ) 32 3.5 (a2 − b2 ) 52
1 π
if r < (a2 − b2 ) 2 and θ < .
2
4. The potential function due to the homogeneous prolate spheroid gener-
x2 y2
ated by the rotation of 2 + 2 = 1 about its major axis is, at an external
a b
point,
" 1 3
3M 1 (a2 − b2 ) 2 1 (a2 − b2 ) 2
V = 1 + P2 (cos θ)
(a2 − b2 ) 2 1.3 r 3.5 r3
5
#
1 (a2 − b2 ) 2
+ P4 (cos θ) + · · ·
5.7 r5
1
if r > (a2 − b2 ) 2 .
81. The method employed in the last three articles may be stated in
general as follows:—Whenever in a problem involving the solving of the special
form of Laplace’s Equation
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0,
sin θ
the value of V is given or can be found for all points on the axis of X and
this value can be expressed as a sum or a series involving only whole powers
positive or negative of the radius vector of the point, the solution for a point
not on the axis can be obtained by multiplying each term by the appropriate
Zonal Harmonic, subject only to the condition that the result if a series must
be convergent.
It will be shown in the next article that Pm (cos θ) is never greater than one
nor less than minus one. Hence the series in question will be convergent for all
values of r for which the original series was absolutely convergent.
82. In addition to the form given in (1) Art. 77 for Pm (x) other forms are
often useful.
It ought to be possible to develop Pm (cos θ), which may be regarded as a
function of θ, into a Fourier’s Series, and such a development may be obtained,
though with much labor, by the methods of Chapter II.
The development in terms of cosines of multiples of θ may be obtained much
more easily by the following device.
We have seen in Art. 75 that Pm (cos θ) is the coefficient of the mth power
1
of z in the development of (1 − 2z cos θ + z 2 )− 2 in a power series, and that if
1
mod z < 1 (1 − 2z cos θ + z 2 )− 2 can be developed into such a series. We know
by the Theory of Functions that only one such series exists, so that the method
by which we may choose to obtain the development will not affect the result.
1 1
(1 − 2z cos θ + z 2 )− 2 = (1 − z(eθi + e−θi ) + z 2 )− 2
1 1
= (1 − zeθi )− 2 (1 − ze−θi )− 2 .
1
(1−zeθi )− 2 may be developed into an absolutely convergent series if mod z < 1,
by the Binomial Theorem. We have
1 1 1.3 2 2θi 1.3.5 3 3θi 1.3.5.7 4 4θi
(1 − zeθi )− 2 = 1 + zeθi + z e + z e + z e + ···
2 2.4 2.4.6 2.4.6.8
1 1 1.3 2 −2θi
(1 − ze−θi )− 2 = 1 + ze−θi + z e
2 2.4
1.3.5 3 −3θi 1.3.5.7 4 −4θi
+ z e + z e + ···
2.4.6 2.4.6.8
1
The product of these series will give a development for (1 − 2z cos θ + z 2 )− 2
in power series. The coefficient of z m is easily picked out, and must be equal to
Pm (cos θ). We thus get

1.3.5. · · · (2m − 1) mθi
Pm (cos θ) = e + e−mθi
2.4.6. · · · 2m
1 2m
+ . (e(m−2)θi + e−(m−2)θi )
2 2m − 1

1.3 2m(2m − 2)
+ . (e(m−4)θi + e−(m−4)θi ) + · · ·
2.4 (2m − 1)(2m − 3)

1.3.5 · · · (2m − 1) 1.m
Pm (cos θ) = 2 cos mθ + 2 cos(m − 2)θ
2.4.6. · · · 2m 1.(2m − 1)
1.3 m(m − 1)
+2 cos(m − 4)θ
1.2(2m − 1)(2m − 3)

1.3.5 m(m − 1)(m − 2)
+2 cos(m − 6)θ + · · · . (1)
1.2.3 (2m − 1)(2m − 3)(2m − 5)
If m is odd the development runs down to cos θ; if m is even to cos(0),
but in that case the coefficient of cos(0), that is, the constant term, will not
contain the factor 2 which is common to all the other terms, but will be simply
2
1.3.5 · · · (m − 1)
.
2.4.6. · · · m
We write out the values of Pm (cos θ) for a few values of m

P0 (cos θ) = 1 




P1 (cos θ) = cos θ 



1



P2 (cos θ) = (3 cos 2θ + 1) 

4 


1



P3 (cos θ) = (5 cos 3θ + 3 cos θ) 

8 


1



P4 (cos θ) = (35 cos 4θ + 20 cos 2θ + 9) 

64 
1 (2)
P5 (cos θ) = [63 cos 5θ + 35 cos 3θ + 30 cos θ] 

128 


1



P6 (cos θ) = [231 cos 6θ + 126 cos 4θ + 105 cos 2θ + 50] 

512 


1



P7 (cos θ) = [429 cos 7θ + 231 cos 5θ + 189 cos 3θ + 175 cos θ]

1024 


1



P8 (cos θ) = [6435 cos 8θ + 3432 cos 6θ + 2772 cos 4θ 

16384 



+ 2520 cos 2θ + 1225].

Since all the coefficients in the second member of (1) are positive, and since
each cosine has unity for its maximum value it is clear that Pm (cos θ) has its
maximum value when θ = 0; but we have shown in Art. 76 that Pm (1) = 1.
Therefore Pm (cos θ) is never greater than unity if θ is real. It is also easily seen
from (1) that Pm (cos θ) can never be less than −1.
83. Pm (x) can be very simply expressed as a derivative. We have

(2m − 1)(2m − 3) · · · 1 m m(m − 1) m−2
Pm (x) = x − x
m! 2.(2m − 1)

m(m − 1)(m − 2)(m − 3) m−4
+ x − ···
2.4.(2m − 1)(2m − 3)
wx
(2m − 1)(2m − 3) · · · 1 m+1 (m + 1)m m−1
Pm (x)dx = x − x
(m + 1)! 2.(2m − 1)
0

(m + 1)m(m − 1)(m − 2) m−3
+ x − ···
2.4.(2m − 1)(2m − 3)
wx wx wx
2
Pm (x)dx2 = dx Pm (x)dx
0 0 0

(2m − 1)(2m − 3) · · · 1 m+2 (m + 2)(m + 1) m
= x − x
(m + 2)! 2.(2m − 1)

(m + 2)(m + 1)m(m − 1) m−2
+ x − ···
2.4.(2m − 1)(2m − 3)
wx
(2m − 1)(2m − 3) · · · 1 2m 2m(2m − 1) 2m−2
m
Pm (x)dxm = x − x
(2m)! 2(2m − 1)
0

2m(2m − 1)(2m − 2)(2m − 3) 2m−4
+ x − ···
2.4.(2m − 1)(2m − 3)

(2m − 1)(2m − 3) · · · 1 2m m(m − 1) 2m−4
= x − mx2m−2 + x
(2m)! 2!

m(m − 1)(m − 2) 2m−6
− x + ··· .
3!
The quantity in brackets obviously differs from (x2 − 1)m by terms involving
lower powers of x than the mth.
1.3.5 · · · (2m − 1) dm 2
Hence Pm (x) = (x − 1)m ,
(2m)! dxm
1 dm 2
or Pm (x) = m (x − 1)m . (1)
2 m! dxm
This important formula is entirely general and holds not merely when x =
cos θ, but for all values of x.
84. The last result is so important that it is worth while to confirm it by

obtaining it directly from Legendre’s Equation
d2 z dz
(1 − x2 ) − 2x + m(m + 1)z = 0 (1)
dx2 dx
v. (1) Art. 75.
dz
Let us differentiate (1) with respect to x a few times representing by z 0 ,
dx
d2 z 3
00 d z
by z , by z 000 , &c. We get
dx2 dz 3
d2 z 0 dz 0
(1 − x2 ) 2
− 2.2x + [m(m + 1) − 2]z 0 = 0,
dx dx
d2 z 00 dz 00
(1 − x2 ) 2 − 2.3x + [m(m + 1) − 2(1 + 2)]z 00 = 0,
dx dx
d2 z 000 dz 000
(1 − x2 ) − 2.4x + [m(m + 1) − 2(1 + 2 + 3)]z 000 = 0,
dx2 dx
and in general
d2 z (n) dz (n)
(1 − x2 ) 2
− 2(n + 1)x + [m(m + 1) − 2(1 + 2 + 3 + · · · + n)]z (n) = 0
dx dx
d2 z (n) dz (n)
or (1 − x2 ) − 2(n + 1)x + m(m + 1) − n(n + 1)]z (n) = 0. (2)
dx2 dx
Following the analogy of these steps it is easy to write equations that will
differentiate into (1).
dz1 d2 z2 d3 z3
Let = z, = z, = z, &c. Then
dx dx2 dx3
d2 z1
(1 − x2 ) + m(m + 1)z1 = 0
dx2
will differentiate into (1),
d2 z2 dz2
(1 − x2 ) 2
+ 2.1x + [m(m + 1) − 2.1]z2 = 0
dx dx
if differentiated twice will give (1),
d2 z3 dz3
(1 − x2 ) + 2.2x + [m(m + 1) − 2(1 + 2)]z3 = 0
dx2 dx
if differentiated three times will give (1), and in general
d2 zn dzn
(1 − x2 ) + 2(n − 1)x + [m(m + 1) − n(n − 1)]zn = 0 (3)
dx2 dx
if differentiated n times with respect to x will give (1).
If n = m + 1 (3) reduces to
d2 zm+1 dzm+1
(1 − x2 ) 2
+ 2mx = 0, (4)
dx dx
and the (m+1)st derivative with respect to x of any function of x which satisfies
(4) will be a solution of (1). (4) can be written
dzm
(1 − x2 ) + 2mxzm = 0
dx
and can be readily solved by separating the variables and integrating. v. Int.
Cal. (1) page 314. It gives
zm = C(x2 − 1)m .
dm zm dm (x2 − 1)m
Hence z= = C (5)
dxm dxm
is a solution of Legendre’s Equation (1) and agrees with the value of Pm (x)
obtained in Art. 83.
85. The equations obtained in Art. 84 are so curious and so simply related
that it is worth while to consider them a little more fully.
We have seen that
d2 z dz
(1 − x2 ) + 2mx =0 (1)
dx2 dx
differentiates into
d2 z dz
(1 − x2 ) 2
+ 2(m − 1)x + 2mz = 0; (2)
dx dx
that if we differentiate (2) m times we get Legendre’s Equation
d2 z dz
(1 − x2 ) 2
− 2x + m(m + 1)z = 0; (3)
dx dx
that if we differentiate (2) 2m times we get
d2 z dz
(1 − x2 ) − 2(m + 1)x = 0; (4)
dx2 dx
that if we differentiate (2) m − n times we have
d2 z dz
(1 − x2 ) + 2(n − 1)x + [m(m + 1) − n(n − 1)]z = 0; (5)
dx2 dx
and that if we differentiate (2) m + n times we have
d2 z dz
(1 − x2 ) − 2(n + 1)x + [m(m + 1) − n(n + 1)]z = 0. (6)
dx2 dx
By the aid of (1) we found in the last article a particular solution of (2), namely
z = (x2 − 1)m .
If we substitute in (2) z = u(x2 − 1)m following the method illustrated fully

in Art. 18, we get as the general solution of (2)
w dx
z = A(x2 − 1)m + B(x2 − 1)m , (7)
(x2 − 1)m+1
A and B being arbitrary constants.

w dx
is easily written out [v. formula (42) page 6. Table of Inte-
(x − 1)m+1
2
grals. Int. Cal. Appendix]. If x < 1 it vanishes when x = 0. If x > 1 it vanishes
when x = ∞. If then x < 1 (7) can be written
wx dx
z = A(x2 − 1)m + B(x2 − 1)m (8)
(x2 − 1)m+1
0
and if x > 1
w∞ dx
z = A(x2 − 1)m + B(x2 − 1)m (9)
x
(x2 − 1)m+1
and in these forms unnecessary arbitrary constants are avoided.
From (7) we can get the general solutions of (3), (4), (5), and (6).
dm (x2 − 1)m dm
w dx

2 m
z=A + B (x − 1) (10)
dxm dxm (x2 − 1)m+1
is the general solution of (3).
d2m (x2 − 1)m d2m

w dx

2 m
z=A + B (x − 1) (11)
dx2m dx2m (x2 − 1)m+1
dm−n (x2 − 1)m dm−n

w dx

2 m
z=A + B (x − 1) (12)
dxm−n dxm−n (x2 − 1)m+1
dm+n (x2 − 1)m dm+n

w dx

2 m
z=A + B (x − 1) (13)
dxm+n dxm+n (x2 − 1)m+1

In each of these forms A and B are arbitrary constants and the integral is
to be taken from 0 to x if x < 1 and from x to ∞ if x > 1.
Of course (10) must be identical with the forms already obtained in Arts. 16
and 18 as general solutions of Legendre’s Equation.
Equation (4) is so simple that it can be solved directly, and we get its solution
in the form w dx
z = A1 + B 1 (14)
(x − 1)m+1
2
which must be equivalent to (11).

Comparing (14) with (7), the solution of (2), we see that every solution of
(4) can be obtained from a solution of (2) by dividing the latter by (x2 − 1)m ,
or in other words that if we write (2)

d2 z dz
(1 − x2 ) + 2(m − 1)x + 2mz = 0, (2)
dx2 dx
d2 z1 dz1
and (4) as (1 − x2 ) 2 − 2(m + 1)x =0 (4)
dx dx
z = z1 (x2 − 1)m ; and the substitution of this value in (2) will give (4), and
z
the substitution of z1 = 2 in (4) will give (2).
(x − 1)m
We have, then, two ways of obtaining (4) from (2); we may differentiate (2)
2m times with respect to x, or we may replace z in (2) by z1 (x2 − 1)m .
If we use the first method we have seen that Legendre’s Equation (3) is
midway between (2) and (4). That is if we differentiate (2) m times we get (3)
and if we then differentiate (3) m times we get (4). Let us see if the half-way
equation in our second process is Legendre’s Equation.
m
If z = y(x2 − 1) 2
m
and y = z1 (x2 − 1) 2
z = z1 (x2 − 1)m .
m
So that if in (2) we replace z by y(x2 − 1) 2 and then repeat the operation
on the resulting equation we shall get (4). Making the first substitution we find,
d2 y m2

dy
(1 − x2 ) 2 − 2x + m(m + 1) − y = 0, (15)
dx dx 1 − x2
not Legendre’s Equation but a somewhat more general form. Of course its
solution is
m m w dx
y = A(x2 − 1) 2 + B(x2 − 1) 2 . (16)
(x − 1)m+1
2
(2) and (4) are special forms of (5) and (6). Let us try the experiment of
n y
substituting in (5) z = y(1 − x2 ) 2 and in (6) z = n . We find that both
(1 − x2 ) 2
substitutions give the same equation
2
n2

2 d y dy
(1 − x ) 2 − 2x + m(m + 1) − y = 0. (17)
dx dx 1 − x2
The solution of (17) can be obtained from either (12) or (13) and is
1
m−n 2
d (x − 1)m dm−n
w dx

2 m
y= n A + B m−n (x − 1)
(1 − x2 ) 2 dxm−n dx (x2 − 1)m+1
(18)
or
n

dm+n (x2 − 1)m dm+n
w dx

2 2 2 m
y = (1 − x ) A1 + B1 m+n (x − 1)
dxm+n dx (x2 − 1)m+1
(19)
which of course must be equivalent.
86. In addition to the value of Pm (x) given in (1) Art. 83 there is another
important derivative form which we shall proceed to obtain. It is
(−1)m m+1 m 1

Pm (cos θ) = r Dx . (1)
m! r
1 1
We have seen in Art. 75 that r can be developed into
r r1 r12
1 − 2 cos θ + 2
r r
a convergent series if r1 < r and that the (m + 1)st term of that series is
Pm (cos θ)r1m
. Let us obtain this term by Taylor’s Theorem.
rm+1
1 1 1 1
r =p =p
r r1 r 2 2
r − 2r1 r cos θ + r1 2 x + y + z 2 − 2xr1 + r12
2 2
1 − 2 cos θ + 12
r r
1
=p
(x − r1 )2 + y 2 + z 2
Regarding this as a function of (x − r1 ) and developing according to powers

of r1 by Taylor’s Theorem we get as the (m + 1)st term
(−1)m m m (−1)m m m 1

1
r1 Dx p or r1 Dx .
m! x2 + y 2 + z 2 m! r
(−1)m m 1

Pm (cos θ)
Hence = Dx .
rm+1 m! r
87. We have now obtained four different forms for our zonal harmonic,
a polynomial in x, an expression involving cosines of multiples of θ, a form
involving an ordinary mth derivative with respect to x, and a form involving
a partial mth derivative with respect to x. We shall now get a form due to
Laplace, involving a definite integral.
wπ dφ π
= 1 (1)
a − b cos φ (a − b2 ) 2
2
0
if a2 > b2 [v. Int. Cal. page 68].

1 1
1 can be expressed in the form 1 by taking a =
(1 − 2xz + z 2√)2 (a2 − b2 ) 2
1 − zx and b = z x2 − 1 and no matter what value x may have z can be taken
so small that a2 will be greater than b2 . Then by (1)
1w
π
1 dφ
1 = √
(1 − 2xz + z 2 ) 2 π 1 − zx − z x2 − 1. cos φ
0
1w
π
dφ
= √
π 1 − z(x + x2 − 1. cos φ)
0
1 w
π p p
= [1 + (x + x2 − 1. cos φ)z + (x + x2 − 1. cos φ)2 z 2
π
0
p
+ (x + x2 − 1. cos φ)3 z 3 + · · · ]dφ
√
if z is taken so small that the modulus of z(x + x2 − 1. cos φ) is less than
1. But by Art. 77 (2) Pm (x) is the coefficient of z m in the development of
1
1 ,
(1 − 2xz + z 2 ) 2
1w
π p
hence Pm (x) = [x + x2 − 1. cos φ]m dφ. (2)
π
0
By replacing φ by π − φ in (2) we get
1w
π p
Pm (x) = [x − x2 − 1. cos φ]m dφ. (3)
π
0
1 1 1 1
1 = 12 and if mod z < 1 or in other words
(1 − 2xz + z2) 2 z 1 1
1 − 2x + 2
z z
1
if mod z > 1 can be developed into a convergent series
1 1 21
1 − 2x + 2
z z 1 m
1
involving powers of , and the coefficient of will be Pm (x); but this will
z z
1
be the coefficient of z −m−1 in the development of 1 according to
(1 − 2xz + z 2 ) 2
descending powers of z, mod z being greater
√ than 1.
If now we let a = zx − 1 and b = z x2 − 1, a2 − b2 = 1 − 2xz + z 2 and z
may be taken so great that a2 − b2 > 0. Then by (1)
1w
π
1 dφ
1 = √
(1 − 2xz + z 2 ) 2 π zx − 1 − z x2 − 1. cos φ
0
1 wπ dφ
= √
π
h 1 i
0 z(x − x2 − 1. cos φ) 1 − √
z(x − x2 − 1. cos φ)
1 wπ 1

1
= √ z −1 + √ z −2
π (x − x 2 − 1. cos φ) (x − x2 − 1. cos φ)
0

1 −3
+ √ z + · · · dφ
(x − x2 − 1. cos φ)2
1w
π
dφ
and the coefficient of z −m−1 is √ .
π [x − x2 − 1. cos φ]m+1
0
1w
π
dφ
Hence Pm (x) = √ . (4)
π [x − x2 − 1. cos φ]m+1
0
Replace φ by π − φ and we get
1w
π
dφ
Pm (x) = √ . (5)
π [x + x2 − 1. cos φ]m+1
0
88. In the problems in which we have already used Zonal Harmonics

(v. Arts. 78–81) we have been able to start with the value of the Potential
Function at any point on the axis of X, and it has been necessary to develop
the expression for V on that axis in terms of ascending or descending powers of
x. If, however, we start with the value of V in terms of θ for some given value
of r, that is on the surface of some sphere, we must develop the function of θ
in terms of zonal harmonics of cos θ (v. Art. 10), and our problem becomes the
following:—To develop a given function of cos θ in terms of zonal harmonics of
cos θ, or to develop a given function of x in terms of the functions Pm (x), x
lying between 1 and −1.
The problem resembles closely that of developing in a Fourier’s series, which
we have already considered at such length.
Let f (x) = A0 P0 (x) + A1 P1 (x) + A2 P2 (x) + A3 P3 (x) + · · · (1)
for all values of x from −1 to 1 and let it be required to determine the coefficients.
If f (x) is single-valued and has only finite discontinuities between x = −1
and x = 1 we may proceed as in Art. 19.
Let us take n + 1 terms of (1) and attempt to determine the coefficients.
Take n + 1 values of x at equal intervals ∆x between x = −1 and x = 1 so that
(n + 2)∆x = 2; f (−1 + ∆x), f (−1 + 2∆x), f (−1 + 3∆x), · · · f [−1 + (n + 1)∆x]

will be the corresponding values of f (x). Substitute these values in (1) and we
have

f (−1 + ∆x) = A0 P0 (−1 + ∆x) + A1 P1 (−1 + ∆x) 


+ A2 P2 (−1 + ∆x) + · · · + An Pn (−1 + ∆x)  



f (−1 + 2∆x) = A0 P0 (−1 + 2∆x) + A1 P1 (−1 + 2∆x) 




+ A2 P2 (−1 + 2∆x) + · · · + An Pn (−1 + 2∆x) (2)
.. .. .. .. .. .. .. .. 

. . . . . . . .





f (1 − ∆x) = A0 P0 (1 − ∆x) + A1 P1 (1 − ∆x) + A2 P2 (1 − ∆x) + · · · 




+ An Pn (1 − ∆x),

that is, n + 1 equations from which in theory the n + 1 coefficients A0 , A1 , · · ·

An can be determined.
Following the analogy of Art. 24 let us multiply the first equation by Pm (−1+
∆x).∆x, the second by Pm (−1+2∆x).∆x, the third by Pm (−1+3∆x).∆x, &c.,
and add the equations. The first member of the resulting equation is
k=n+1
X
f (−1 + k∆x)Pm (−1 + k∆x).∆x, (3)
k=1
and the coefficient of any A as Al in the second member is

k=n+1
X
Pm (−1 + k∆x)Pl (−1 + k∆x).∆x. (4)
k=1
If now n is indefinitely increased (3) approaches as its limiting value
w1
f (x)Pm (x)dx (5)
−1
w1
and (4) approaches Pm (x)Pl (x)dx. (6)
−1
We have now to find the value of the integral (6) or as we shall write it for
the sake of greater convenience
w1
Pm (x)Pn (x)dx.
−1
w1 1 w1 dm (x2 − 1)m dn (x2 − 1)n

89. Pm (x)Pn (x)dx = . dx by
2m+n m!n! dxm dxn
−1 −1
(1) Art. 83.
w1 dm (x2 − 1)m dn (x2 − 1)n m 2 1

d (x − 1)m dn−1 (x2 − 1)n
. dx = .
dxm dxn dxm dxn−1
−1
−1
w1 dm+1 (x2 − 1)m dn−1 (x2 − 1)n
− . dx (1)
dxm+1 dxn−1
−1
by integration by parts.
Now if z = X(x2 − 1)n
dz dX
= 2nxX(x2 − 1)n−1 + (x2 − 1)n
dx dx
2 n−1 2 dX
= (x − 1) 2nxX + (x − 1) . (2)
dx
Hence the pth derivative with respect to x of any function of x containing
(x2 − 1)n as a factor will contain (x2 − 1)n−p as a factor if p < n.
dn−1 (x2 − 1)n
, then, contains (x2 − 1) as a factor and is zero when x = 1
dxn−1
and when x = −1, so that (1) reduces to
w1 dm (x2 − 1)m dn (x2 − 1)n w1 dm+1 (x2 − 1)m dn−1 (x2 − 1)n
. dx = − . dx.
dxm dxn dxm+1 dxn−1
−1 −1
It follows that
w1 dm (x2 − 1)m dn (x2 − 1)n
. dx
dxm dxn
−1
w1 dm+p (x2 − 1)m dn−p (x2 − 1)n
= (−1)p . dx
dxm+p dxn−p
−1
w1 dm−p (x2 − 1)m dn+p (x2 − 1)n
= (−1)p . dx. (3)
dxm−p dxn+p
−1
If m < n we get from (3)

w1 dm (x2 − 1)m dn (x2 − 1)n w1 d2m (x2 − 1)m dn−m (x2 − 1)n
m
. dx = (−1) . dx
dxm dxn dx2m dxn−m
−1 −1
n−m−1 2 n
1
d (x − 1)
= (−1)m (2m)! = 0,
dxn−m−1
−1
d2m (x2 − 1)m

since = (2m)!.
dx2m
If m > n
w1 dm (x2 − 1)m dn (x2 − 1)n w1 dm−n (x2 − 1)m d2n (x2 − 1)n
n
. dx = (−1) . dx
dxm dxn dxm−n dx2n
−1 −1
m−n−1 2 1
n d (x − 1)m
= (−1) (2n)! = 0.
dxm−n−1
−1
If, then, m is not equal to n

w1
Pm (x)Pn (x)dx = 0. (4)
−1
w1
If m = n we have to find [Pm (x)]2 dx.
−1
w1 1 w1 dm (x2 − 1)m dm (x2 − 1)m

2
[Pm (x)] dx = 2m . dx.
2 (m!)2 dxm dxm
−1 −1
w1 dm (x2 − 1)m dm (x2 − 1)m w1 d2m (x2 − 1)m

m
. dx = (−1) .(x2 − 1)m dx
dxm dxm dx2m
−1 −1
w1
by (3), = (−1)m (2m)! (x2 − 1)m dx.
−1
w1 w1 m w
1
(x2 − 1)m dx = (x − 1)m (x + 1)m dx = − (x − 1)m−1 (x + 1)m+1 dx
m+1
−1 −1 −1
m! w1
= (−1)m (x + 1)2m dx
(m + 1)(m + 2) · · · 2m
−1
22m+1 m!
= (−1)m .
(m + 1)(m + 2) · · · (2m + 1)
w1 1 (−1)m (2m)!(−1)m m!22m+1

Hence [Pm (x)]2 dx =
22m (m!)2 (m + 1)(m + 2) · · · (2m + 1)
−1
w1 2
or [Pm (x)]2 dx = . (5)
2m + 1
−1
90. The solution of the problem in Art. 88 is now readily obtained, and
we have
f (x) =A0 P0 (x) + A1 P1 (x) + A2 P2 (x) + · · · (1)
2m + 1 w
1
where Am = f (x)Pm (x)dx. (2)
2
−1
The function and the series are equal for all values of x from x = −1 to
x = 1, and f (x) is subject to no conditions save those which would enable us to
develop it in a Fourier’s Series. [v. Chapter III.]
Of course (1) can be written
f (cos θ) = A0 P0 (cos θ) + A1 P1 (cos θ) + A2 P2 (cos θ) + · · ·
2m + 1 w
1
where Am = f (cos θ)Pm (cos θ)d(cos θ)
2
−1
or if f (cos θ) = F (θ)
F (θ) =A0 P0 (cos θ) + A1 P1 (cos θ) + A2 P2 (cos θ) + · · · (3)

2m + 1 w
π
where Am = F (θ)Pm (cos θ) sin θ.dθ (4)
2
0
and the development holds good from θ = 0 to θ = π.

If f (x) is an even function, that is, if f (−x) = f (x) (1) and (2) can be
somewhat simplified. For in that case it can be easily shown (v. Art. 77) that
w1 w1
f (x)P2k (x)dx = 2 f (x)P2k (x)dx,
−1 0
w1
and that f (x)P2k+1 (x)dx = 0;
−1
so that if f (−x) = f (x)
f (x) = A0 P0 (x) + A2 P2 (x) + A4 P4 (x) + A6 P6 (x) + · · · (5)

w1
where A2k = (4k + 1) f (x)P2k (x)dx. (6)
0
If f (x) is an odd function, that is, if f (−x) = −f (x) it can be shown in like
manner that
f (x) = A1 P1 (x) + A3 P3 (x) + A5 P5 (x) + A7 P7 (x) + · · · (7)
w1
where A2k+1 = (4k + 3) f (x)P2k+1 (x)dx. (8)
0
If it is only necessary that the development should hold for 0 < x < 1 any
function may be expressed in form (5) or (7) at pleasure.
w1
91. We can establish the fact that Pm (x)Pn (x)dx = 0 by a more general
−1
method than that used in Art. 89.
Let Xm be any solution of Legendre’s Equation

d dz
(1 − x2 ) + m(m + 1)z = 0 [v. (1) Art. 16].
dx dx
which with its first derivative with respect to x is finite, continuous, and single-
valued for values of x between −1 and 1, −1 and 1 being included.

d dXm
Then (1 − x2 ) + m(m + 1)Xm = 0 (1)
dx dx

d dXn
and (1 − x2 ) + n(n + 1)Xn = 0 (2)
dx dx
Multiply (1) by Xn and (2) by Xm and subtract and integrate and we get
w1 w1 d

dXn

[m(m + 1) − n(n + 1)] Xm Xn dx = Xm (1 − x2 ) dx
dx dx
−1 −1
w1 d

dXm

− Xn (1 − x2 ) dx.
dx dx
−1
Integrate by parts,
x=1
w1
2 dXn 2 dXm
[m(m + 1) − n(n + 1)] Xm Xn dx = Xm (1 − x ) − Xn (1 − x )
dx dx
−1
x=−1
w1 dXn dXm w1 dXm dXn
− (1 − x2 ) dx + (1 − x2 ) dx. (3)
dx dx dx dx
−1 −1
w1
Whence Xm Xn dx = 0 (4)
−1
unless m = n.
(3) gives at once the important formula
h dX dXn i
m
w1 (1 − x2 ) Xn − Xm
Xm Xn dx = dx dx (5)
x
m(m + 1) − n(n + 1)
from which come as special cases

h dPm (x) dPn (x) i
w1 (1 − x2 ) Pn (x) − Pm (x)
Pm (x)Pn (x)dx = dx dx (6)
x
m(m + 1) − n(n + 1)
and since P0 (x) = 1

dPm (x)
w1 (1 − x2 )
Pm (x)dx = dx , (7)
x
m(m + 1)
unless m = 0.
EXAMPLES.
1. Show that
w1
Pm (x)dx = 0 if m is even and is not zero.
0
m−1 1 3.5.7. · · · m
= (−1) 2 if m is odd.
m(m + 1) 2.4.6. · · · (m − 1)
v. Art. 91 (7) and Art. 77 (10).
2. Show that
w1
Pm (x)Pn (x)dx = 0 if m and n are both even or both odd.
0
m+n+1 m! n!
= (−1) 2
m 2 n − 1 2
2m+n−1 (m − n)(m + n + 1) ! !
2 2
if m is even and n odd. v. Art. 91 (6) and Art. 77 (8), (9), and (10). cf. J. W.
Strutt (Lord Rayleigh) Lond. Phil. Trans. 1870, page 579.
w1 1
3. Show that [Pm (x)]2 dx = v. Art. 89 (5).
2m + 1
0
92. Formula (4) Art. 91 can be obtained directly from Laplace’s Equation
by the aid of Green’s Theorem (v. Peirce’s Newt. Pot. Func. § 48).
Take the special form of Green’s Theorem, [(148) § 48 Peirce’s Newt. Pot.
Func.] y w
(U ∇2 V − V ∇2 U )dxdydz = (U Dn V − V Dn U )ds (1)
where ∇2 stands for (Dx2 + Dy2 + D22 ), Dn is the partial derivative along the
external normal, and the left-hand member is the space-integral through the
space bounded by any closed surface, and the right-hand member is the surface
integral taken over the same surface. (v. Int. Cal. Chapter XIV.)
If U and V are solutions of Laplace’s Equation ∇2 V = ∇2 U = 0 and (1)
reduces to w
(U Dn V − V Dn U )ds = 0. (2)
Now rm Xm and rn Xn are solutions of Laplace’s Equation if x = cos θ
(v. Art. 16).
If the unit sphere is taken as the bounding surface and U = rm Xm and
V = rn Xn (1) and (2) will hold good.
Dn U = Dr (rm Xm ) = mrm−1 Xm ,
Dn V = nrn−1 Xn ,
ds = sin θ.dθdφ,
w
2π wπ
and (2) becomes dφ (nXm Xn − mXm Xn ) sin θ.dθ = 0
0 0
wπ
or 2π(n − m) Xm Xn sin θ.dθ = 0. (3)
0
Since x = cos θ, sin θ.dθ = −dx and (3) reduces to
w1
Xm Xn dx = 04 (4)
−1
unless m = n.
93. We can now solve completely the problem of Art. 10 which was in
that article carried to the point where it was only necessary to develop a certain
function of θ in the form
A0 P0 (cos θ) + A1 P1 (cos θ) + A2 P2 (cos θ) + · · ·

π
given that f (θ) = 1 from θ=0 to θ =
2
π
and f (θ) = 0 from θ= to θ = π.
2
This amounts to the same thing as developing F (x) into the series
F (x) = A0 P0 (x) + A1 P1 (x) + A2 P2 (x) + A3 P3 (x) + · · ·

4 It should be noted that this proof is no more general than that of the last article, for,
in order that Green’s Theorem should apply to rm Xm this function and its first derivatives
must be finite continuous and single-valued within and on the surface of the unit sphere. (v.
Peirce, Newt. Pot. Func. § 48.)
where F (x) = 0 from x = −1 to x = 0

and F (x) = 1 from x=0 to x = 1.
By Art. 90 (1) and (2)
1w 1w
1 1
1
A0 = P0 (x)dx = dx = ,
2 2 2
0 0
(2m + 1) w
1
and any coefficient Am = Pm (x)dx.
2
0
By Art. 91, Ex. 1

w1
Pm (x)dx = 0 if m is even
0 m−1 1 3.5.7. · · · m
= (−1) 2 if m is odd.
m(m + 1) 2.4.6. · · · (m − 1)
Hence Am = 0 if m is even
2m + 1 1.3.5. · · · (m − 2)
m−1
= (−1) 2 . if m is odd.
2m + 2 2.4.6. · · · (m − 1)
1 3 7 1 11 1.3
Then F (x) = + P1 (x) − . P3 (x) + . P5 (x) − · · · (1)
2 4 8 2 12 2.4
1 3 7 1 11 1.3 5
and u= + rP1 (cos θ) − . r3 P3 (cos θ) + . r P5 (cos θ) + · · · (2)
2 4 8 2 12 2.4
for any point within the sphere.
94. If in a problem on the Potential Function the value of V is given at

every point of a spherical surface and has circular symmetry5 about a diameter
of that surface the value of V at any point in space can be obtained.
We have to solve Laplace’s Equation in the form
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0 (1)
sin θ
V = f (θ) when r = a
V =0 “ r = ∞.
We have f (θ) = A0 P0 (cos θ) + A1 P1 (cos θ) + A2 P2 (cos θ) + · · ·

(2m + 1) w
π
where Am = f (θ)Pm (cos θ) sin θ.dθ. v. Art. 90 (4).
2
0
Hence
r r 2 r 3
V = A0 + A1 P1 (cos θ) + A2 P2 (cos θ) + A3 P3 (cos θ) + · · · (2)
a a a
is the required solution for a point within the sphere, and
a a 2 a 3 a 4
V = A0 + A1 P1 (cos θ) + A2 P2 (cos θ) + A3 P3 (cos θ) + · · ·
r r r r
(3)
is the required solution for an external point.
EXAMPLES.
1. If on the surface of a sphere of radius c V is constant and equal to a

ac
show that V = a for any point within the sphere and V = for any external
r
point.
2. Two equal thin hemispherical shells of radius c placed together to form

a spherical surface are separated by a thin non-conducting layer. Charges of
statical electricity are placed on the two hemispheres one of which is then found
to be at potential a and the other at potential b. Find the value of the potential
function at any point.
7 1 r3

a+b 3r
V = + (b − a) P1 (cos θ) − . 3 P3 (cos θ)
2 4c 8 2c
11 1.3 r5

+ . P 5 (cos θ) − · · ·
12 2.4 c5
for an internal point

2
a+b c 3c 7 1 c4
V = . + (b − a) P 1 (cos θ) − . P3 (cos θ)
2 r 4 r2 8 2 r4
11 1.3 c6

+ . P 5 (cos θ) − · · ·
12 2.4 r6
for an external point.
3. If V1 = f (cos θ) when r = a and V1 = 0 when r = b show that for

a<r<b
m=∞ m+1 −1
rm
m+1
X b b am
V1 = Am − m − m Pm (cos θ)
m=0
rm+1 b am+1 b
2m + 1 w
1
where Am = f (x)Pm (x)dx.
2
−1
4. If V2 = F (cos θ) when r = b and V2 = 0 when r = a show that for

a<r<b
m=∞ m −1
am+1
m
X r b am+1
V2 = Bm − m+1 − m+1 Pm (cos θ)
m=0
am r am b
2m + 1 w
1
where Bm = F (x)Pm (x)dx.
2
−1
5. If the value of the potential function is given arbitrarily on the surfaces

of a spherical shell but has circular symmetry6 about a diameter V = V1 + V2
(v. Exs. 3 and 4).
6. Two concentric hollow spherical conductors are insulated and charged.

The inner one of radius a is at potential p, and the outer one of radius b is at
potential q. Find V for any point in space.
V =p if r < a,

pa b qb a
V = −1 + 1− if a < r < b,
b−a r b−a r
qb
V = if r > b.
r
7. If V = 0 on the base of a hemisphere and V = f (cos θ) on the convex
surface, show that for a point within the hemisphere
k=∞ 2k+1
X r
V = A2k+1 P2k+1 (cos θ)
a
k=0
w1
where A2k+1 = (4k + 3) f (x)P2k+1 (x)dx [v. Art. 90 (8)].
0
8. If the convex surface of a solid hemisphere of radius a is kept at the

constant temperature unity and the base at the constant temperature zero show
that after the permanent state of temperatures is set up the temperature of any
internal point is
3r 7 1 r3 11 1.3 r5
u= P1 (cos θ) − . 3 P3 (cos θ) + . P5 (cos θ) − · · ·
2a 4 2a 6 2.4 a5
9. A sphere of radius a and with blackened surface is exposed to the direct
rays of the sun in air at the temperature zero. Find the stationary temperature
of any internal point.
6 See note on page 12

Suggestion: Dr u + hu − M f (θ) = 0 when r = a.

X rm X
Let u= Am m Pm (cos θ), and f (θ) = Bm Pm (cos θ).
a
Then we have
X Am X X
m Pm (cos θ) + h Am Pm (cos θ) − M Bm Pm (cos θ) = 0,
a
M Bm
whence Am = m.
h+
a
π π
Here f (θ) = cos θ if 0 < θ < and f (θ) = 0 if < θ < π.
2 2
1 1 5 3
f (θ) = + P1 (cos θ) + P2 (cos θ) − P4 (cos θ) + · · ·
4 2 16 32
k+1 (4k + 1)(2k)!
+ (−1) P2k (cos θ) + · · ·
(4k + 4)(2k − 1)22k (k!)2
v. Art. 91 Exs. (2) and (3). cf. J. W. Strutt (Lord Rayleigh), Lond. Phil. Trans.
vol. 160, page 587.
95. The formulas of Art. 90 enable us to develop a given function of x

in terms of Zonal Surface Harmonics, the development holding true for values
of x between −1 and +1. If, however, we can show by outside considerations
that a given function of x can be expressed in Zonal Surface Harmonics, the
development holding true for all values of x, the formulas of Art. 90 will give us
the development in question.
For example if n is a positive integer xn can be expressed in terms of
Zonal Surface Harmonics no matter what the value of x, and no Harmonic
of higher order than n will enter. For the formulas giving the values of P1 (x),
P2 (x), · · · Pn (x) (v. Art. 77) may be regarded as n algebraic equations of the
first degree in terms of x, x2 , x3 , · · · xn and P1 (x), P2 (x), · · · Pn (x).
From these equations the n − 1 quantities x, x2 , x3 , . · · · xn−1 , can be elim-
inated, and there will result an equation of the first degree in xn and P1 (x),
P2 (x), · · · Pn (x), which will enable us to express xn in the form
A0 + A1 P1 (x) + A2 P2 (x) + · · · + An Pn (x),
no matter what the value of x, and we shall have the same formula when −1 <
x < 1 as when x > 1 or x < −1.
Let us obtain this development. By Art. 90 (1) and (2)
xn = A0 P0 (x) + A1 P1 (x) + A2 P2 (x) + · · · (1)
2m + 1 w n
1
where Am = x Pm (x)dx. (2)
2
−1
2m + 1 1 w n dm (x2 − 1)m
1
Then Am = x dx by (1) Art. 83.
2 2m m! dxm
−1
By integration by parts we get
w1 dm (x2 − 1)m w 1
xn dx = n(n − 1)(n − 2) · · · (n − m + 1) xn−m (1 − x2 )m dx,
dxm
−1 −1
if m < n + 1, (3)
=0 if m > n.
By integration by parts we readily obtain the reduction formula
w1 2q w p+2
1
xp (1 − x2 )q dx = x (1 − x2 )q−1 dx whence
p+1
−1 −1
w1 2m m! w1
n−m 2 m
x (1 − x ) dx = xn+m dx.
(n − m + 1)(n − m + 3) · · · (n + m − 1)
−1 −1
w1 2
xn+m dx = if n + m is even,
(n + m + 1)
−1
=0 if n + m is odd.
(2m + 1)n(n − 1)(n − 2) · · · (n − m + 1)

Hence Am =
(n − m + 1)(n − m + 3)(n − m + 5) · · · (n + m + 1)
if m < n + 1 and m + n is even,
=0 if m > n or if m + n is odd.
Therefore

n n! (2n + 1)
x = (2n + 1)Pn (x) + (2n − 3) Pn−2 (x)
1.3.5 · · · (2n + 1) 2
(2n + 1)(2n − 1)
+ (2n − 7) Pn−4 (x)
2.4
(2n + 1)(2n − 1)(2n − 3)
+ (2n − 11) Pn−6 (x) + · · · (4)
2.4.6
1
the second member ending with the term P0 (x) if n is even and with the
n+1
3
term P1 (x) if n is odd.
n+2
For convenience of reference we write out a few powers of x.
x0 = 1 = P0 (x)




x = P1 (x) 



2 1


x2 = P2 (x) + P0 (x)



3 3 


2 3


3 
x = P3 (x) + P1 (x) 

5 5 


8 4 1


4 
x = P4 (x) + P2 (x) + P0 (x) 
35 7 5 (5)
5 8 4 3 

x = P5 (x) + P3 (x) + P1 (x) 

63 9 7 


16 24 10 1


6 
x = P6 (x) + P4 (x) + P2 (x) + P0 (x) 

231 77 21 7 


16 8 14 1


7 
x = P7 (x) + P5 (x) + P3 (x) + P1 (x) 

429 39 33 3 


128 64 48 40 1


8 
x = P8 (x) + P6 (x) + P4 (x) + P2 (x) + P0 (x).
6435 495 143 99 9
If a given function of x can be expressed as a terminating power series it
can be developed into a Zonal Harmonic Series by the aid of (4). Given that
f (x) = a0 + a1 x + a2 x2 + a3 x3 + · · · ,
let f (x) = B0 + B1 P1 (x) + B2 P2 (x) + B3 P3 (x) + · · · ;
then picking out carefully the coefficient of Pm (x) we have

m! (m + 1)(m + 2)
Bm = am + am+2
1.3.5. · · · (2m − 1) 2.(2m + 3)

(m + 1)(m + 2)(m + 3)(m + 4)
+ am+4 + · · · . (6)
2.4.(2m + 3)(2m + 5)
dPn (x)
96. The development of is useful and is easily obtained.
dx
dPn (x)
Let =A0 P0 (x) + A1 P1 (x) + A2 P2 (x) + · · ·
dx
2m + 1 w
1
dPn (x)
Then Am = Pm (x) dx (1)
2 dx
−1
by Art. 90 (2);
w1 dPn (x) h ix=1 w1 dPm (x)

Pm (x) dx = Pm (x)Pn (x) − Pn (x) dx. (2)
dx x=−1 dx
−1 −1
h ix=1
Pm (x)Pn (x) =0 if m + n is even
x=−1
=2 if m + n is odd.
dPn (x)
Since Pn (x) is an algebraic polynomial of the nth degree in x, is an
dx
algebraic polynomial of the n − 1st degree in x. Therefore in (1) m is less than
dPm (x)
n; consequently is an algebraic polynomial in x of lower degree than n
dx
and
w1 dPm (x)
Pn (x) dx = 0 by Art. 95 (3).
dx
−1
We get then Am = 2m + 1 if m + n is odd and m < n,

=0 if m + n is even or m > n − 1; and
dPn (x)
= (2n − 1)Pn−1 (x) + (2n − 5)Pn−3 (x) + (2n − 9)Pn−5 (x) + · · · (3)
dx
the second member ending with the term 3P1 (x) if n is even and with the term
P0 (x) if n is odd.
From (3) a number of simple formulas are readily obtained. For example
dPn+1 (x) dPn−1 (x)

− = (2n + 1)Pn (x) (4)
dx dx
w1 1
Pn (x)dx = [Pn−1 (x) − Pn+1 (x)]. (5)
x
2n +1
dPn (x) dPn+1 (x) dPn−1 (x)
(2n + 1)x =n + (n + 1) (6)
dx dx dx
[v. (4) and Article 77 (12)].
dPn (x)
(x2 − 1) = nxPn (x) − nPn−1 (x) (7)
dx
[v. (5) and Article 91 (7).]
97. By the aid of the formulas of Art. 96 a number of valuable develop-

ments can be obtained.
Let us get cos nθ and sin nθ, n being any positive real.
z = cos nθ and z = sin nθ are solutions of the equation
d2 z
+ n2 z = 0
dθ2
or if we let x = cos θ, of the equation

d2 z dz
(1 − x2 ) −x + n2 z = 0. (1)
dx2 dx
Let a0 P0 (x) + a1 P1 (x) + a2 P2 (x) + · · ·

be the required development of cos nθ or of sin nθ.
m=∞
d2 Pm (x)

X dPm (x)
Then am (1 − x2 ) − x + n 2
P m (x) =0 by (1).
m=0
dx2 dx
z = Pm (x) is a solution of Legendre’s Equation (v. Art. 77). Hence

d2 Pm (x) dPm (x) dPm (x)
(1 − x2 ) 2
−x =x − m(m + 1)Pm (x),
dx dx dx
and (1) becomes
m=∞
X dPm (x) 2
am x + [n − m(m + 1)]Pm (x) = 0. (2)
m=0
dx
Formulas (4) and (6) of Art. 96 enable us to throw (2) into the form
m=∞ 2
n − m2 dPm+1 (x) n2 − (m + 1)2 dPm−1 (x)
X
am − = 0. (3)
m=0
2m + 1 dx 2m + 1 dx
dPm+1 (x)
(3) must be identically true. Therefore the coefficient of must
dx
equal zero, and we have
2m + 5 n2 − m2
am+2 = . 2 am . (4)
2m + 1 n − (m + 3)2
If we are developing cos nθ
1w
π
a0 = cos nθ sin θ.dθ by Art. 90 (4),
2
0
1 wπ
= [sin(n + 1)θ − sin(n − 1)θ]dθ,
4
0
1 1 + cos nπ
a0 = − . ; (5)
2 n2 − 1
3w
π
and a1 = cos nθ cos θ sin θ.dθ by Art. 90 (4),
2
0
3 1 − cos nπ
a1 = − . . (6)
2 n2 − 4
(4), (5), and (6) give us
1 + cos nπ h n2
cos nθ = − P 0 (cos θ) + 5 P2 (cos θ)
2(n2 − 1) n2 − 3 2
n2 (n2 − 22 ) i
+9 2 P 4 (cos θ) + · · ·
(n − 32 )(n2 − 52 )
1 − cos nπ h n2 − 12
− 2 2
3P1 (cos θ) + 7 2 P3 (cos θ)
2(n − 2 ) n − 42
(n2 − 12 )(n2 − 32 ) i
+ 11 2 P 5 (cos θ) + · · · . (7)
(n − 42 )(n2 − 62 )
If n is a whole number 1 + cos nπ or 1 − cos nπ will vanish and the series will
end with the term involving Pn (cos θ). For this case (7) may be rewritten
1 2.4.6. · · · 2n h
cos nθ = . (2n + 1)Pn (cos θ)
2 3.5.7. · · · (2n + 1)
n2 − (n + 1)2
+ (2n − 3) 2 Pn−2 (cos θ)
n − (n − 2)2
[n2 − (n + 1)2 ][n2 − (n − 1)2 ] i
+ (2n − 7) 2 P n−4 (cos θ) + · · · . (8)
[n − (n − 2)2 ][n2 − (n − 4)2 ]
If we are developing sin nθ
1w
π
1 sin nπ
a0 = sin nθ sin θ.dθ = − . 2 ,
2 2 n −1
0
3w
π
3 sin nπ
a1 = sin nθ cos θ sin θ.dθ = . and
2 2 n2 − 22
0
1 sin nπ h n2
sin nθ = − . 2 P0 (cos θ) + 5 2 P2 (cos θ)
2 n −1 n − 32
n2 (n2 − 22 ) i
+9 2 P 4 (cos θ) + · · ·
(n − 32 )(n2 − 52 )
1 sin nπ h n2 − 12
+ . 2 3P 1 (cos θ) + 7 P3 (cos θ)
2 n − 22 n2 − 42
(n2 − 12 )(n2 − 32 ) i
+ 11 2 2 2 2
P5 (cos θ) + · · · . (9)
(n − 4 )(n − 6 )
If n is a whole number sin nπ = 0, and all the terms of (9) vanish except
those involving Pn−1 (cos θ), Pn+1 (cos θ), Pn+3 (cos θ) &c., which become inde-
terminate. For this case it is necessary to compute an−1 independently.
We have
2n − 1 w
π
an−1 = sin nθPn−1 (cos θ) sin θ.dθ
2
0
2n − 1 w
π
= [cos(n − 1)θ − cos(n + 1)θ]Pn−1 (cos θ)dθ.
4
0
2n − 1 1.3.5. · · · (2n − 3)
Hence an−1 = . π [v. Art. 82 (1)],
4 2.4.6. · · · (2n − 2)
and
π 1.3. · · · (2n − 3) h
sin nθ = . (2n − 1)Pn−1 (cos θ)
4 2.4. · · · (2n − 2)
n2 − (n − 1)2
+ (2n + 3) 2 Pn+1 (cos θ)
n − (n + 2)2
[n2 − (n − 1)2 ][n2 − (n + 1)2 ] i
+ (2n + 7) 2 P n+3 (cos θ) + · · · . (10)
[n − (n + 2)2 ][n2 − (n + 4)2 ]
EXAMPLES.
1. Show that
πh 1 2 1.3 2 1.3.5 2 i
csc θ = 1+5 P2 (cos θ) + 9 P4 (cos θ) + 13 P6 (cos θ) + · · ·
2 2 2.4 2.4.6
whence
1 πh 1 2 1.3 2 1.3.5 2 i
√ = 1+5 P2 (x) + 9 P4 (x) + 13 P6 (x) + · · ·
1 − x2 2 2 2.4 2.4.6
[v. Art. 90 (4) and Art. 82].
2. Show that
π h 1 3 1 2 5 1.3 2 i
ctn θ = 3 P1 (cos θ) + 7 P3 (cos θ) + 11 P5 (cos θ) + · · ·
2 2 4 2 6 2.4
whence
1 πh 1 3 1 2 5 1.3 2 i
√ = 3 P1 (x) + 7 P3 (x) + 11 P5 (x) + · · ·
1 − x2 2 2 4 2 6 2.4
[v. Art. 90 (4) and Art. 82].
3. By integrating the result of Ex. 1 and simplifying by the aid of Art. 96

(5), obtain the development
π h 1 2 1 2
sin−1 x = 3 P1 (x) + 7 P3 (x)
2 2 2.4
1.3 2 1.3.5 2 i
+ 11 P5 (x) + 15 P7 (x) + · · ·
2.4.6 2.4.6.8
π h 1 2
1 2
whence θ= P0 (cos θ) − 3 P1 (cos θ) − 7 P3 (cos θ)
2 2 2.4
1.3 2 i
−11 P5 (cos θ) − · · · .
2.4.6
4. By integrating the result of Ex. 2 and simplifying by the aid of Art. 96
(5) obtain
p πh1 1 1 2 3 1 2
1 − x2 = −5 P2 (x) − 9 P4 (x)
2 2 4 2 6 2.4
5 1.3 2 i
−13 P6 (x) + · · ·
8 2.4.6
whence
πh1 1 1 2 3 1 2 i
sin θ = P0 (cos θ) − 5 P2 (cos θ) − 9 P4 (cos θ) − · · · .
2 2 4 2 6 2.4
To make clearer the analogy of development in Zonal Harmonic Series with
development in Fourier’s Series we give on page 186 a cut representing the first
seven Surface Zonal Harmonics P1 (cos θ), P2 (cos θ), · · · P7 (cos θ), which are of
course somewhat complicated Trigonometric curves resembling roughly cos θ,
cos 2θ, · · · cos 7θ; and on page 187, the first four successive approximations to
the Zonal Harmonic Series
1 3 7 1 11 1.3
+ P1 (cos θ) − . P3 (cos θ) + . P5 (cos θ) − · · · [I]
2 4 8 2 12 2.4
[v. (1) Art. 93], and
πh 1 2 1 2
P0 (cos θ) − 3 P1 (cos θ) − 7 P3 (cos θ)
2 2 2.4
1.3 2 i
− 11 P5 (cos θ) − · · · [II]
2.4.6
(v. Ex. 3 Art. 97).
π π
[I] is equal to 1 from θ = 0 to θ = , and to 0 from θ = to θ = π; and [II]
2 2
is equal to θ from θ = 0 to θ = π.
The figures on page 187 are constructed on precisely the same principle as
those on pages 64 and 65, with which they should be carefully compared.
The curves y = P0 (cos θ), y = P1 (cos θ), . . . y = P7 (cos θ). (v. page 185.)
98. By applying Gauss’s Theorem (B. O. Peirce, Newt. Pot. Func. § 31)
or the special Form of Green’s Theorem,
y w y
∇2 V dxdydz = Dn V ds = −4π ρdxdydz,
[Peirce, N. P. F. § 49 (149)] to a box cut from an infinitely thin shell of attracting

matter by a tube of force whose end is an element of the surface of the shell we
readily obtain the important result
4πρκ = Dn V1 − Dn V2 . (1)
where ρ is the density and κ the thickness of the shell, V1 the value of the
potential function due to the shell at an internal point and V2 its value at an
external point, and where Dn is the partial derivative along the external normal
to the outer surface of the shell.
If we have to deal with a surface distribution of matter we have only to
replace ρκ in (1) by σ where σ is the surface density, whence
4πσ = Dn V1 − Dn V2 (2)
(v. Peirce, N. P. F. §§ 45, 46, and 47).

Formulas (1) and (2) enable us to solve problems in attraction when we know
the density of the attracting mass, and problems in Statical Electricity when we
know the distribution of the charge, by methods analogous to that of Art. 94.
For example let us find the value of the potential function due to a thin
material spherical shell of density ρ and radius a.
Since V must be a solution of Laplace’s Equation and must be finite both
when r = 0 and r = ∞ we have

X
V1 = Am rm Pm (cos θ)
X 1
V2 = Bm m+1 Pm (cos θ).
r
V1 and V2 must approach the same limiting values as r approaches a. Hence
Bm
= Am am
am+1
or Bm = Am a2m+1 .
X
Dn V1 = Dr V1 = mrm−1 Am Pm (cos θ),
X Am a2m+1
Dn V2 = Dr V2 = − (m + 1) Pm (cos θ).
rm+2
Therefore by (1)
X
4πρκ = (2m + 1)Am am−1 Pm (cos θ)
if κ is the thickness of the shell.

X
Let ρ = f (cos θ) = Cm Pm (cos θ)
2m + 1 w
1
where Cm = f (x)Pm (x)dx by Art. 90 (2).
2
−1
Then 4πκCm = (2m + 1)Am am−1 , and

4πκCm 4πκ
Am = m−1
, and Bm = Cm am+2 ,
(2m + 1)a 2m + 1
X Cm rm
and V1 = 4πaκ Pm (cos θ), (3)
2m + 1 am
X Cm am+1
and V2 = 4πaκ Pm (cos θ). (4)
2m + 1 rm+1
99. We can now get the value of the potential function due to a spherical
shell of finite thickness, provided that its density can be expressed as a sum of
terms of the form Crk Pm (cos θ).
Let a be the radius of the outer surface and b be the radius of the inner
surface of the shell.
1st.—Let ρ = Crk Pm (cos θ). Then for the shell of radius s and thickness ds
Csk rm
V1 = 4πsds Pm (cos θ) by (3) Art. 98,
2m + 1 sm
Csk sm+1
and V2 = 4πsds Pm (cos θ) by (4) Art. 98.
2m + 1 rm+1
Then if r < b
wa 4πC (ak−m+2 − bk−m+2 ) m
V = V1 = r Pm (cos θ), (1)
(2m + 1) (k − m + 2)
b
if r > a
wa 4πC (ak+m+3 − bk+m+3 ) Pm (cos θ)
V = V2 = , (2)
(2m + 1) (k + m + 3) rm+1
b
and if b < r < a

wr wa 4πC rk+m+3 − bk+m+3

V = V2 + V1 =
r
2m + 1 (k + m + 3)rm+1
b
ak−m+2 − rk−m+2 m

+ r Pm (cos θ). (3)
(k − m + 2)
X
2d.—If ρ = Cm rk Pm (cos θ) the solutions will consist of sums of terms of
the forms given in (1), (2), and (3).
EXAMPLES.
1. If the shell is homogeneous
V = 2πρ(a2 − b2 ) if r < b,
4 1 M
V = πρ(a3 − b3 ) = if r > a,
3 r r
2b3 r2
V = 2πρ a2 − − if b < r < a.
3r 3
2. If the density is any given function of the distance from the centre
M
V = if r > a, and V = a constant if r < b.
r
3. If the density at any point of a solid sphere is proportional to the square
of the distance from a diametral plane
M a 2 a3

V = + P2 (cos θ) if r > a.
a r 7 r3
4. If the density at any point of a solid sphere is proportional to its distance
from a diametral plane
M a 1 a3 1.1 a5 1.1.3 a7

V = + P 2 (cos θ) − P 4 (cos θ) + P 6 (cos θ) − · · ·
a r 6 r3 6.8 r5 6.8.10 r7
if r > a. Compare Ex. 2 Art. 80.
100. We have seen in Art. 18 (c) (3) that

w dx
Qm (x) = CPm (x) , (1)
(1 − x2 )[Pm (x)]2
w dx
no constant term being understood with .
(1 − x )[Pm (x)]2
2
1
is a rational fraction and becomes infinite only for x = 1,
(1 − x2 )[Pm (x)]2
x = −1, and for the roots of Pm (x) = 0, all of which are real and lie between −1
1 dm (x2 − 1)m
and 1, as can be proved by the aid of the relation Pm (x) = m .
2 m! dxm
w∞ dx
If x2 > 1 is finite and determinate and contains no
x
2
(1 − x )[Pm (x)]2
constant term. Hence if x2 > 1
w∞ dx w∞ dx
Qm (x) = −Pm (x) 2 2
= P m (x) (2)
x
(1 − x )[Pm (x)] x
(x − 1)[Pm (x)]2
2
for the constant factor of Qm (x) has been chosen so that C = −1.
If x2 < 1 the second member of (2) is not finite and determinate, and we are
thrown back to the form (1), and C proves to be unity.
(1) gives us readily
1 1+x
Q0 (x) = log (3)
2 1−x
x 1+x
Q1 (x) = −1 + log (4)
2 1−x
if x2 < 1. 1 x+1
(2) gives us Q0 (x) = log (5)
2 x−1
x x+1
Q1 (x) = −1 + log (6)
2 x−1
if x2 > 1.
From Art. 85 (10) it follows that
dm
wx dx

Qm (x) = C m (x2 − 1)m if x2 < 1,
dx (x2 − 1)m+1
0
d m
w
∞
dx

2 m
= C m (x − 1) 2 − 1)m+1
if x2 > 1.
dx x
(x
(−1)m+1 2m m!
C can be determined and is equal to if x2 < 1, and is equal to
(2m)!
(−1)m 2m m!
if x2 > 1.
(2m)!
(−1)m+1 2m m! dm
wx dx

2 m
Hence Qm (x) = (x − 1) (7)
(2m)! dxm (x2 − 1)m+1
0
if x2 < 1,
(−1)m 2m m! dm
w∞ dx

2 m
and Qm (x) = (x − 1) (8)
(2m)! dxm x
(x2 − 1)m+1
if x2 > 1.
(7) and (8) give us for Q0 (x) and Q1 (x) the values already written in (3),
(4), (5), and (6).
By the repeated application of the formula
(m + 1)Qm+1 (x) − (2m + 1)xQm (x) + mQm−1 (x) = 0, (9)
which may be obtained for the case where x2 < 1 from Art. 16 (13) and (14),
and for the case where x2 > 1 from Art. 16 (9), any Surface Zonal Harmonic
of the Second Kind can be obtained from Q0 (x) and Q1 (x) as given in (3), (4),
(5), and (6).
Analogous formulas for pm (x) and qm (x) can be obtained without difficulty
from Art. 16 (4) and (5). They are
(m + 1)2 qm+1 (x) − (2m + 1)xpm (x) − m2 qm−1 (x) = 0 (10)

and pm+1 (x)+(2m + 1)xqm (x) − pm−1 (x) = 0 (11)
and they hold good for any value of m.
EXAMPLES.
1. Confirm the values of Q0 (x) and Q1 (x) given in Art. 100 (3), (4), (5),
and (6) by expanding them and comparing them with Art. 16 (13), (14), and (9).
2. If the value of V on the surface of a cone of revolution can be expressed

in terms of whole powers positive or negative of r, V can be found for any point
in space, cf. Art. 81.
X Bm

If V = Am rm + m+1 when θ = α then
r
X Bm

Pm (cos θ)
m
V = Am r + m+1 .
r Pm (cos α)
X Bm

m
3. If V = Am r + m+1 when θ = α, and V = 0 when θ = β,
r
X Bm

Qm (cos β)Pm (cos θ) − Pm (cos β)Qm (cos θ)

V = Am rm + m+1 .
r Pm (cos α)Qm (cos β) − Pm (cos β)Qm (cos α)
4. Find V for points corresponding to values of θ between α and β when

V can be given in terms of whole powers of r for θ = α and for θ = β.
5. Find by the method of Art. 16 solutions of Legendre’s Equation of the

form
m(m + 1) (m − 1)m(m + 1)(m + 2)
z = 1 Pm (x) = 1 + (x − 1) + (x − 1)2
2 22 (2!)2
(m − 2)(m − 1)m(m + 1)(m + 2)(m + 3)
+ (x − 1)3 + · · · ,
23 (3!)2
m(m + 1) (m − 1)m(m + 1)(m + 2)
z = −1 Pm (x) = 1 − (x + 1) + (x + 1)2
2 22 (2!)2
(m − 2)(m − 1)m(m + 1)(m + 2)(m + 3)
+ (x + 1)3 + · · · .
23 (3!)2
If m is a whole number, 1 Pm (x) = Pm (x) and −1 Pm (x) = (−1)m Pm (x). No
matter what the value of m, 1 Pm (x) is absolutely convergent for −1 < x < 3,
and −1 Pm (x) is absolutely convergent for −3 < x < 1.
6. By the aid of (7) Art. 16 show that

1 1
V = √ sin(n log r)kn (cos θ), V = √ sin(n log r)ln (cos θ),
r r
1 1
V = √ cos(n log r)kn (cos θ), V = √ cos(n log r)ln (cos θ),
r r
are solutions of Laplace’s Equation
1
rDr2 (rV ) + Dθ (sin θDθ V ) = 0, if
sin θ
1 2 h 1 2 ih 5 2 i
n2 + n2 + n2 +
kn (x) = p− 12 +ni (x) = 1 + 2 2
x + 2 2 x4
2! 4!
h 1 2 ih 5 2 ih 9 2 i
n2 + n2 + n2 +
+ 2 2 2 x6 + · · ·
6!
and 3 2 h 3 2 ih 7 2 i
n2 + n2 + n2 +
ln (x) = −q− 21 +ni (x) = x + 2 x3 + 2 2 x5
3! 5!
h 3 2 ih 7 2 ih 11 2 i
n2 + n2 + n2 +
+ 2 2 2 x7 + · · ·
7!
kn (x) and ln (x) are convergent if x2 < 1, but are divergent if x2 = 1.
7. Show by the aid of Example 5 that

1 1
V = √ sin(n log r)Kn (cos θ), V = √ sin(n log r)Kn (− cos θ),
r r
1 1
V = √ cos(n log r)Kn (cos θ), V = √ cos(n log r)Kn (− cos θ),
r r
1
are solutions of rDr2 (rV ) + Dθ (sin θDθ V ) = 0
sin θ
1 2
n2 +
if Kn (x) =1 P− 12 +ni (x) = 1 − 2 (x − 1)
2
h 1 2 ih 3 2 i
n2 + n2 +
+ 2 2 (x − 1)2
22 (2!)2
h 1 2 ih 3 2 ih 5 2 i
n2 + n2 + n2 +
− 2 2 2 (x − 1)3 + · · ·
22 (3!)2
and 1 2
n2 +
Kn (−x) =−1 P− 12 +ni (x) = 1 + 2 (x + 1)
2
h 1 2 ih 3 2 i
n2 + n2 +
+ 2 2 (x + 1)2
22 (2!)2
h 1 2 ih 3 2 ih 5 2 i
n2 + n2 + n2 +
+ 2 2 2 (x + 1)3 + · · · .
23 (3!)2
Kn (cos θ) is convergent except for θ = π, and Kn (− cos θ) is convergent except

for θ = 0.
kn (x), ln (x), Kn (x), and Kn (−x) are sometimes called Conal Harmonics.
They are particular values of z which satisfy Legendre’s Equation written in the
form
d2 z dz 2 1
(1 − x2 ) 2 − 2x − n + z = 0.
dx dx 4
For an elaborate treatment of them see E. W. Hobson on “A Class of Spherical
Harmonics of Complex Degree.” Trans. Camb. Phil. Soc., Vol. XIV.
8. If V = f (r) when θ = β,
1 w∞ w∞ λ Kα (cos θ)
V = √ dλ e 2 f (eλ ) cos[α(λ − log r)]dα; if θ < β.
π r −∞ Kα (cos β)
0
9. If V = f (r) when θ = β and r < a, and V = 0 when r = a,
a w w∞ λ
r 0
2 Kα (cos θ) r
V = dλ e 2 f (aeλ ) sin αλ sin α log dα; if θ < β.
π r −∞ Kα (cos β) a
0
10. If V = f (r) when θ = β and a < r < b, and V = 0 when r = a and

when r = b,
m=∞
X Km0 (cos θ) mπ(log r − log a)
V = Am sin
m=1
Km0 (cos β) log b − log a
mπ
where m0 = and
log b − log a
b
log a
a w
r
2 x mπx
Am = e 2 f (aex )sin dx; if θ < β.
log b − log a r log b − log a
0
11. If θ > β cos θ must be replaced by (− cos θ) in examples 8, 9, and 10.
12. If V = f (r) when θ = β, and V = 0 when θ = γ,
1 w∞ w∞ λ
V = √ dλ e 2 f (eλ )
π r −∞
0
kα (cos θ)lα (cos γ) − kα (cos γ)lα (cos θ)
cos[α(λ − log r)]dα;
kα (cos β)lα (cos γ) − kα (cos γ)lα (cos β)
if β < θ < γ.
13. If V = f (r) when θ = β and a < r < b, V = 0 when θ = γ and

a < r < b, and V = 0 when r = a and when r = b,
m=∞
X km0 (cos θ)lm0 (cos γ) − km0 (cos γ)lm0 (cos θ) mπ(log r − log a)
V = Am sin ,
m=1
km (cos β)lm (cos γ) − km (cos γ)lm (cos β)
0 0 0 0 log b − log a
mπ
where m0 = and
log b − log a
b
log a
2
r
a w x mπx
Am = e 2 f (aex ) sin dx;
log b − log a r log b − log a
0
if β < θ < γ and a < r < b.
14. If V = f (r) when θ = β and a < r < b, and V = 0 when r = a and

Dr V + hV = 0 when r = b,
m=∞
X Kαm (cos θ) r
V = Am sin αm log , where
m=1
Kαm (cos β) a
b
log a
2(αm2
+ h2 b2 ) w x
Am = 2 e 2 f (aex ) sin αm x.dx
αm (log b − log a) + hb[hb(log b − log a) + 1]
0
and αm is a root of the equation

b b
α cos α log + hb sin α log =0 v. Art. 68 Ex. 5.
a a
196
CHAPTER VI.
SPHERICAL HARMONICS.
101. When we are dealing with problems in finding the potential function
due to forces which have not circular symmetry1 about an axis and are using
Spherical Coördinates, we have to solve Laplace’s Equation in the form
1 1
rDr2 (rV ) + Dθ (sin θDθ V ) + Dφ2 V = 0 (1)
sin θ sin2 θ
[v. (XIII) Art. 1].
To get a particular solution of (1) we shall assume as usual that V is a
product of functions each of which involves but a single variable.
Let V = R.Θ.Φ; where R involves r only, Θ involves θ only, and Φ φ only.
Substitute in (1) and we get
dΘ
2
r d (rR) 1 d sin θ 1 d2 Φ
+ dθ + =0 (2)
2 2
R dr Θ sin θ dθ Φ sin θ dφ2
dΘ
2 2
r sin θ d (rR) sin θ d sin θ 2
or + dθ = − 1 d Φ .
R dr2 Θ dθ Φ dφ2
As the first member does not contain φ the second member cannot contain
φ, and as it contains no other variable it must be constant; call it n2 . Equation
(2) is then equivalent to the two equations
d2 Φ
+ n2 Φ = 0 (3)
dφ2
h dΘ i
2
r d (rR) 1 d sin θ 2
and + dθ − n = 0 (4)
R dr2 Θ sin θ dθ 2
sin θ
(3) has been solved before and gives us
Φ = A cos nφ + B sin nφ (5)
[v. Art. 13(a)].

The first term of (4) does not involve θ and the second and third terms do
not involve r.
r d2 (rR)
must, then, be a constant; we shall call it m(m+1) as in Art. 13(c).
R dr2
Then (4) breaks up into
d2 (rR)
r = m(m + 1)R (6)
dr2
1 See note, page 12.
SPHERICAL HARMONICS. 197
h dΘ i
1 d sin θ 2

and dθ + m(m + 1) − n Θ = 0. (7)
sin θ dθ sin2 θ
(6) was solved in Art. 13(c) and gives
R = A1 rm + B1 r−m−1 . (8)
If in (7) we replace cos θ by µ we get
n2

d 2 dΘ
(1 − µ ) + m(m + 1) − Θ = 0, (9)
dµ dµ 1 − µ2
the equivalent of
d2 z n2

dz
(1 − x2 ) − 2x + m(m + 1) − z = 0, (10)
dx2 dx 1 − x2
[v. (17) Art. 85], which was solved in Art. 85 for the case where m and n are
positive integers and n < m + 1. v. (18) and (19) Art. 85.
From (19) Art. 85 we get as a particular solution of (9)
n dn Pm (µ) dn Pm (µ)
Θ = (1 − µ2 ) 2 n
= sinn θ , (11)
dµ dµn
if we restrict ourselves to whole positive values of m and n, as we shall do
hereafter unless the contrary is explicitly stated, and suppose m not less than
n.
A second but less useful particular solution of (9) is
n dn Qm (µ)
Θ = (1 − µ2 ) 2 .
dµn
Combining our results we have as important particular solutions of (1)
dn Pm (µ)
V = rm (A cos nφ + B sin nφ) sinn θ , (12)
dµn
1 dn Pm (µ)
and V = m+1 (A cos nφ + B sin nφ) sinn θ , (13)
r dµn
where m and n are positive integers and n < m + 1.
dn Pm (µ) n dn Pm (µ)
102. sinn θ n
or (1 − µ2 ) 2 is a new function of µ, that
dµ dµn
n
is of cos θ, and we shall represent it by Pm (µ)2 and shall call it an associated
function of the nth order and mth degree. It is a value of Θ satisfying equation
(9) Art. 101.
2 Most n (µ).
of the English writers represent this function by Tm
By differentiating the value of Pm (x) given in (9) Art. 74 we get the formula
(2m)! sinn θ

n (m − n)(m − n − 1) m−n−2
Pm (µ) = m µm−n − µ
2 m!(m − n)! 2.(2m − 1)

(m − n)(m − n − 1)(m − n − 2)(m − n − 3) m−n−4
+ µ − ··· (1)
2.4.(2m − 1)(2m − 3)
the expression in the parenthesis ending with the term involving µ0 if m − n is

even and with the term involving µ if m − n is odd.
For convenience of reference we give on the next page a table from which
n
Pm (µ) can be readily obtained for values of m and n from 1 to 8.
n n
cos nφPm (µ) and sin nφPm (µ), that is,
dn Pm (µ) dn Pm (µ)
cos nφ sinn θ and sin nφ sinn θ
dµn dµn
are called Tesseral Harmonics of the mth degree and nth order, and are values
of V which satisfy the equation
1 1
m(m + 1)V + Dθ (sin θDθ V ) + Dφ2 V = 0 (2)
sin θ sin2 θ
or its equivalent
1
m(m + 1)V + Dµ [(1 − µ2 )Dµ V ] + D2 V = 0. (3)
1 − µ2 φ
There are obviously 2m + 1 Tesseral Harmonics of the mth degree, namely
dPm (µ) dPm (µ)

Pm (µ), cos φ sin θ , sin φ sin θ
dµ dµ
d2 Pm (µ) d2 Pm (µ)
cos 2φ sin2 θ , sin 2φ sin2 θ
dµ2 dµ2
d3 Pm (µ) 3
d Pm (µ)
cos 3φ sin3 θ , sin 3φ sin3 θ
dµ3 dµ3
. . . . . . . . . . . . .
dm Pm (µ) dm Pm (µ)
cos mφ sinm θ , sin mφ sinm θ
dµm dµm
If each of these is multiplied by a constant and their sum taken, this sum
is called a Surface Spherical Harmonic of the mth degree, and is a solution of
equations (2) and (3). We shall represent it by Ym (µ, φ) or by Ym (θ, φ).
dn Pm (µ)
Table for cscn θPm
n
(µ) = .
dµn
m n = 1. n = 2.
1 1
2 3µ 3
3
3 (5µ2 − 1) 15µ
2
5 15
4 (7µ3 − 3µ) (7µ2 − 1)
2 2
15 105
5 (21µ4 − 14µ2 + 1) (3µ3 − µ)
8 2
21 105
6 (33µ5 − 30µ3 + 5µ) (33µ4 − 18µ2 + 1
8 8
7 63
7 (429µ6 − 495µ4 + 135µ2 − 5) (143µ5 − 110µ3 + 15µ)
16 8
9 315
8 (715µ7 − 1001µ5 + 385µ3 − 35µ) (143µ6 − 143µ4 + 33µ2 − 1)
16 16
m n = 3. n = 4.
3 15
4 105µ 105
105
5 (9µ2 − 1) 945µ
2
315 945
6 (11µ3 − 3µ) (11µ2 − 1)
2 2
315 3465
7 (143µ4 − 66µ2 + 3) (13µ3 − 3µ)
8 2
3465 10395
8 (39µ5 − 26µ3 + 3µ) (65µ4 − 26µ2 + 1)
8 8
m n = 5. n = 6. n = 7. n = 8.
5 945
6 10395µ 10395
10395
7 (13µ2 − 1) 135135µ 135135
2
135135 135135
8 (5µ3 − µ) (15µ2 − 1) 2027025µ 2027025
2 2
1
rm Ym (µ, φ) and Ym (µ, φ) are called Solid Spherical Harmonics of the
rm+1
mth degree, and are solutions of Laplace’s Equation (1) Art. 101.
To formulate:—
n=m
Xh dn Pm (µ) n
n d Pm (µ)
i
Ym (µ, φ) = An cos nφ sinn θ + B n sin nφ sin θ (4)
n=0
dµn dµn
n=m
X
n n
or Ym (µ, φ) = A0 Pm (µ) + [An cos nφPm (µ) + Bn sin nφPm (µ)] (5)
n=1
is a Surface Spherical Harmonic of the mth degree.
A Tesseral Harmonic is a special case of a Surface Spherical Harmonic, and a
Zonal Harmonic a special case of a Tesseral Harmonic; Pm (µ) being the Tesseral
0
Harmonic of the zeroth order and the mth degree; it might be written Pm (µ).
EXAMPLES.
1. Show that
d2 z n2

2 dz
(1 − x ) 2 − 2x + m(m + 1) − z=0
dx dx 1 − x2
reduces to
d2 y dy
(1 − x2 ) − 2(n + 1)x + [m(m + 1) − n(n + 1)]y = 0
dx2 dx
n
if we substitute (1 − x2 ) 2 y for z, even when m and n are unrestricted.
ak xk we get
P
2. Show that if in the second equation of Ex. 1 we let y =
(m − n − k)(m + n + 1 + k)
ak+2 = − ak (v. Art. 16)
(k + 1)(k + 2)
whence z = pnm (x) and z = qm

n
(x) are solutions of the first equation of Ex. 1, no
matter what the values of m and n, if

n (m − n)(m + n + 1) 2
pnm (x) = (1 − x2 ) 2 1 − x
2!

(m − n)(m − n − 2)(m + n + 1)(m + n + 2) 4
+ x − ···
4!
and

n n (m − n − 1)(m + n + 2) 3
qm (x) = (1 − x2 ) 2 x − x
3!

(m − n − 1)(m − n − 3)(m + n + 2)(m + n + 4) 5
+ x − ··· .
5!
n
If m − n is a positive integer, pnm (x) or qm (x) will terminate with the term
m−n
involving x , and in that case

n (m − n)(m − n − 1) m−n−2
z = (1 − x2 ) 2 xm−n − x
2.(2m − 1)

(m − n)(m − n − 1)(m − n − 2)(m − n − 3) m−n−4
+ x − ··· .
2.4.(2m − 1)(2m − 3)
the parenthesis ending with a term involving x0 if m − n is even and x if m − n

is odd, is a solution of the first equation of Ex. 1. If m and n are integers this
2m m!(m − n)! n
value of z is Pm (x).
(2m)!
103. We have seen in the last chapter that in many problems it is im-
portant to be able to express a given function of cos θ, that is of µ, in terms of
Zonal Harmonics of µ. So it is often desirable to express a given function of µ
and φ in terms of Tesseral Harmonics of µ and φ.
If, for example, we are trying to find the Potential Function due to certain
forces and have the value of the function given for some given value of r, that is,
on the surface of some given sphere whose centre is at the origin of coördinates,
of course the given value will be a function of θ and φ and if we can express it
in terms of Spherical Harmonics of θ and φ we have only to multiply each term
by the proper power of r to get the required solution of the problem. For we
shall then have a value of V satisfying Laplace’s Equation and reducing to the
given function of θ and φ on the surface of the given sphere.
104. Suppose that we have a function of µ and φ given for all points on
the unit sphere, that is, for all values of µ from −1 to 1 and for all values of φ
from 0 to 2π, µ and φ being independent variables, and that we wish to express
it in terms of Surface Spherical Harmonics.
Assume that
m=∞
Xh n=m
X i
n n
f (µ, φ) = A0,m Pm (µ)+ An,m cos nφPm (µ) + Bn,m sin nφPm (µ) . (1)
m=0 n=1
Let us consider first a finite case, and attempt to determine the coefficients
so that
m=p
Xh n=m
X i
n n
f (µ, φ) = A0,m Pm (µ)+ An,m cos nφPm (µ) + Bn,m sin nφPm (µ) (2)
m=0 n=1
shall hold good at as many points of the sphere as possible. The expression in
brackets in the second member of (2) is a Surface Spherical Harmonic of the
mth degree and contains 2m + 1 constant coefficients. The whole number of
coefficients to be determined is then the sum of an Arithmetical Progression of
p + 1 terms the first term of which is 1 and the last is 2p + 1, and is therefore
equal to (p + 1)2 .
Let the interval from µ = −1 to µ = 1 be divided into p + 2 parts each of
which is ∆µ so that (p + 2)∆µ = 2, and let the interval from φ = 0 to φ = 2π
be divided into p + 2 parts each of which is ∆φ so that (p + 2)∆φ = 2π.
Then if we substitute in equation (2) in turn the values (−1 + ∆µ, ∆φ),
(−1 + 2∆µ, ∆φ), · · · [−1 + (p + 1)∆µ, ∆φ]; (−1 + ∆µ, 2∆φ), (−1 + 2∆µ, 2∆φ),
· · · [−1 + (p + 1)∆µ, 2∆φ]; · · · [−1 + ∆µ, (p + 1)∆φ], [−1 + 2∆µ, (p + 1)∆φ],
· · · [−1+(p+1)∆µ, (p+1)∆φ]; since the first member in each case will be known
we shall have (p+1)2 equations of the first degree containing no unknown except
the (p+1)2 coefficients, and from them the coefficients can be determined. When
they are substituted in equation (2) it will hold good at the (p+1)2 points of the
unit sphere where p + 1 circles of latitude whose planes are equidistant intersect
p + 1 meridians which divide the equator into equal arcs. If now p is indefinitely
increased the limiting values of the coefficients will be the coefficients in equation
(1), and (1) will hold good all over the surface of the unit sphere.
To determine any particular constant we multiply each of our (p + 1)2 equa-
tions by ∆µ∆φ times the coefficient of the constant in question in that equation
and add the equations and then investigate the limiting form approached by the
resulting equation as p is indefinitely increased.
As p is indefinitely increased the summation in question will approach an
integration; and since dµdφ = − sin θ.dθdφ is the element of surface of the unit
sphere, and as the limits −1 and 1 of µ correspond to π and 0 of θ the integration
is a surface integration over the surface of the unit sphere.
In determining any coefficient as An,m in (1) the first member of the limiting
form of our resulting equation will be
w
2π w1
n
dφ f (µ, φ) cos nφPm (µ)dµ.
0 −1
In the second member we shall come across terms of the forms

w
2π w1 w
2π w1
dφ sin lφ cos nφPml (µ)Pm
n
(µ)dµ, dφ cos lφ cos nφPml (µ)Pm
n
(µ)dµ,
0 −1 0 −1
w
2π w1 w
2π w1
n
dφ sin nφ cos nφ[Pm (µ)]2 dµ, dφ cos2 nφ[Pm
n
(µ)]2 dµ,
0 −1 0 −1
and other terms all of which come under the form

w
2π w1
dφ Yl (µ, φ)Ym (µ, φ)dµ,
0 −1
where Ym (µ, φ) and Yl (µ, φ) are Surface Spherical Harmonics of different de-
grees.
If we are determining a coefficient Bn,m the only difference is that sin nφ
and cos nφ will be interchanged in the forms just specified.
105. The integral over the surface of the unit sphere of the product of two
Surface Spherical Harmonics of different degrees is zero.
w
2π w1
That is dφ Yl (µ, φ)Ym (µ, φ)dµ = 0. (1)
0 −1
For as we have seen U = rl Yl (µ, φ) and V = rm Ym (µ, φ) are solutions of

Laplace’s Equation. Hence by Green’s Theorem
w
(U Dn V − V Dn U )ds = 0 v. Art. 92.
Dn V =Dr V = mrm−1 Ym (µ, φ),
Dn U = Dr U = lrl−1 Yl (µ, φ);
U Dn V − V Dn U = (m − l)rl+m−1 Yl (µ, φ)Ym (µ, φ),
= (m − l)Yl (µ, φ)Ym (µ, φ)
on the surface of the unit sphere; and
w w
2π w1
(m − l) Yl (µ, φ)Ym (µ, φ)ds = (m − l) dφ Yl (µ, φ)Ym (µ, φ)dµ = 0.
0 1
Hence unless l = m
w
2π w1
dφ Yl (µ, φ)Ym (µ, φ)dµ = 0.
0 −1
EXAMPLES.
1. Obtain (1) Art. 105 directly from the equation
1
m(m + 1)Ym (µ, φ) + Dµ [(1 − µ2 )Dµ Ym (µ, φ)] + D2 Ym (µ, φ) = 0
1 − µ2 φ
v. (3) Art. 102, and Art. 91.
2. Show that the integral over the surface of the unit sphere of the product
of two Tesseral Harmonics of the same degree but of different orders is zero.
Suggestion:
w
2π w
2π w
2π
sin kφ cos lφ.dφ = sin kφ sin lφ.dφ = cos kφ cos lφ.dφ = 0.
0 0 0
w1
Pln (µ)Pm
n
(µ)dµ = 0 unless l = m
106. −1
2 (m + n)!
= if l = m.
2m + 1 (m − n)!
For
w1 w1 dn Pl (µ) dn Pm (µ)
Pln (µ)Pm
n
(µ)dµ = (1 − µ2 )n . dµ
dµn dµn
−1 −1
1
dn Pm (µ) dn−1 Pl (µ) i
= (1 − µ2 )n .
dµn dµn−1
−1
w1 dn−1 Pl (µ) d
n

2 n d Pm (µ)
− . (1 − µ ) dµ,
dµn−1 dµ dµn
−1
w1 dn−1 P (µ) d n

l 2 n d Pm (µ)
=− . (1 − µ ) dµ;
dµn−1 dµ dµn
−1
by integration by parts.
Replacing n by n − 1 in equation (2) Art. 84 and remembering that
dn−1 Pm (x)
is a possible value of z (n−1) we get
dxn−1
dn+1 Pm (µ) dn Pm (µ) dn−1 Pm (µ)
(1 − µ2 ) − 2nµ + [m(m + 1) − n(n − 1)] = 0,
dµn+1 dµn dµn−1
or if we multiply by (1 − µ2 )n−1
dn+1 Pm (µ) dn Pm (µ)

(1 − µ2 )n n+1
− 2nµ(1 − µ2 )n−1
dµ dµn
dn−1 Pm (µ)
+ (m + n)(m − n + 1)(1 − µ2 )n−1 = 0,
dµn−1
or
dn Pm (µ) n−1

d 2 n−1 d Pm (µ)
(1 − µ2 )n n
= −(m + n)(m − n + 1)(1 − µ ) .
dµ dµ dµn−1
Hence follows the reduction formula
w1 dn Pl (µ) dn Pm (µ)
(1 − µ2 )n . dµ
dµn dµn
−1
w1 dn−1 Pl (µ) dn−1 Pm (µ)
= (m + n)(m − n + 1) (1 − µ2 )n−1 . dµ.
dµn−1 dµn−1
−1
Using this formula n times we get

w1 (m + n)! w
1
Pln (µ)Pm
n
(µ)dµ = Pl (µ)Pm (µ)dµ
(m − n)!
−1 −1
=0 unless l = m
2 (m + n)!
= if l = m
2m + 1 (m − n)!
v. Art. 89 (4) and (5).
107. We are now able to complete the solution of the problem in Art. 104
w
2π w2π w
2π
2 2
and since cos nφ.dφ = sin nφ.dφ = π and dφ = 2π we get as the
0 0 0
coefficients in (1) Art. 104
2m + 1 w w1
2π
A0,m = dφ f (µ, φ)Pm (µ)dµ, (1)
4π
0 −1
2m + 1 (m − n)! w w1
2π
n
An,m = . dφ f (µ, φ) cos nφPm (µ)dµ, (2)
2π (m + n)!
0 −1
2m + 1 (m − n)! w w1
2π
n
Bn,m = . dφ f (µ, φ) sin nφPm (µ)dµ, (3)
2π (m + n)!
0 −1
whence
m=∞
X n=m
X
n
f (µ, φ) = A0,m Pm (µ) + (An,m cos nφ + Bn,m sin nφ)Pm (µ) (4)
m=0 n=1
and the development holds good for all values of µ and φ corresponding to points
on the unit sphere, provided only that the given function satisfies the conditions
that would have to be satisfied if it were to be developed into a Fourier’s Series.
If we use µ1 and φ1 in place of µ and φ in (1), (2), and (3), we can write (4)
in the form
m=∞ w2π w1
1 X 1
f (µ, φ) = (2m + 1) dφ1 f (µ1 , φ1 )Pm (µ)Pm (µ1 )dµ1
2π m=0 2
0 −1
(m − n)! w w1
n=m
X 2π
n n
+ dφ1 f (µ1 , φ1 )Pm (µ)Pm (µ1 ) cos n(φ − φ1 )dµ1 . (5)
(m + n)!
n=1 0 −1
Formulas (1), (2), (3), and (4) are convenient for actual work; (5) is rather
more compactly written.
108. As an example let us express sin2 θ cos2 θ sin φ cos φ in terms of Sur-
face Spherical Harmonics.
1 2
Here f (µ, φ) = µ (1 − µ2 ) sin 2φ.
2
2m + 1 w 2 w
1 2π
2
A0,m = µ (1 − µ )Pm (µ)dµ sin 2φ.dφ = 0,
8π
−1 0
2m + 1 (m − n)! w 2 w
1 2π
2 n
An,m = . µ (1 − µ )Pm (µ)dµ sin 2φ cos nφ.dφ = 0,
4π (m + n)!
−1 0
2m + 1 (m − n)! w 2 w
1 2π
Bn,m = . µ (1 − µ2 )Pm
n
(µ)dµ sin 2φ sin nφ.dφ,
4π (m + n)!
0 0
=0 unless n = 2.
w
2π w
2π
If n = 2 sin 2φ sin nφ.dφ = sin2 2φ.dφ = π, and
0 0
2m + 1 (m − 2)! w 2
1
d2 Pm (µ)
B2,m = . µ (1 − µ2 )2 dµ
4 (m + 2)! dµ2
−1
1 2m + 1 (m − 2)! w 2
1
dm+2 (µ2 − 1)m
= m
µ (1 − µ2 )2 dµ.
2 m! 4 (m + 2)! dµm+2
−1
w1 dm+2 (µ2 − 1)m w dm−4 (µ2 − 1)m

1
µ2 (1 − µ2 )2 m+2
dµ = 720 dµ
dµ dµm−4
−1 −1
by repeated integration by parts,

=0 if m > 4,
w1 4096
= 720 (µ2 − 1)4 dµ = if m = 4,
7
−1
1 9 2! 4096 1
and B2,4 = . . . = .
24 4! 4 6! 7 105
By a like process we find
1
B2,3 = 0 and B2,2 = . Hence
42
1 2 1 2
sin2 θ cos2 θ sin φ cos φ = P (µ) sin 2φ + P (µ) sin 2φ, (1)
42 2 105 4
1 d2 P2 (µ) 1 2
2 d P4 (µ)
= sin 2φ sin2 θ + sin 2φ sin θ , (2)
42 dµ2 105 dµ2
1 1
= sin2 θ sin 2φ + sin2 θ(7µ2 − 1) sin 2φ. (3)
14 14
The required expression might have been obtained without using the formu-
las of Art. 107, by a very simple device, as follows:
1 2 2
sin2 θ cos2 θ sin φ cos φ =
µ sin θ sin 2φ. (4)
2
X d2 Pm (µ)
If now we can express µ2 in the form the work will be done.
dµ2
1 d2 (µ4 )
µ2 = ,
4.3 dµ2
8 4 1
µ4 = P4 (µ) + P2 (µ) + P0 (µ), (5) Art. 95.
35 7 5
d2 (µ4 ) 8 d2 P4 (µ) 4 d2 P2 (µ)
2
= + ;
dµ 35 dµ2 7 dµ2
2 d2 P4 (µ) 1 d2 P2 (µ)
whence µ2 = 2
+ ,
105 dµ 21 dµ2
and substituting this value in (4) we get (2).
EXAMPLES.
1. Show that

1 1
cos3 θ sin3 θ sin φ cos2 φ = P63 (µ) + P43 (µ) sin 3φ
6930 1540

2 1 1 1 1
− P6 (µ) − P4 (µ) − P21 (µ) sin φ.
693 770 63
2. Show that

5 2 9.2! 2 13.4! 2
cos 2φ = 2 cos 2φ P2 (µ) + P4 (µ) + P6 (µ) + · · · .
4! 6! 8!
3. If in a problem on the Potential Function V = f (µ, φ) when r = a, we

shall obviously have
m=∞ n=m
rm
X X
n
V = A0,m Pm (µ) + (An,m cos nφ + Bn,m sin nφ)Pm (µ)
m=0
am n=1
at an internal point and

m=∞ n=m
am+1
X X
n
V = A0,m Pm (µ) + (An,m cos nφ + Bn,m sin nφ)Pm (µ)
m=0
rm+1 n=1
at an external point, where A0,m , An,m , and Bn,m have the values given in (1),
(2), and (3) Art. 107.
4. Solve problems (3), (4), and (5) of Art. 94 for the case where V is not
symmetrical with respect to an axis.
109. Any Solid Spherical Harmonic rm Ym (µ, φ) being a value of V that

satisfies Laplace’s Equation in Spherical Coördinates will transform into a func-
tion of x, y, and z satisfying ∇2 V = 0 if we change to a set of rectangular axes
having the same origin and the same axis of X as the polar system. Moreover
the new function will be a homogeneous rational integral Algebraic function of
x, y, z, of the mth degree.
For each term of rm cos nφPm n
(µ) is of the form
Crm cosn−2k φ sin2k φ sinn θ cosm−2l−n θ

where 2k < n + 1 and 2l < m − n + 1.
This may be written
Cr2l .rm−2l−n cosm−2l−n θ.rn−2k sinn−2k θ cosn−2k φ.r2k sin2k θ sin2k φ
which becomes C(x2 + y 2 + z 2 )l xm−2l−n y n−2k z 2k ,
and is a homogeneous rational integral Algebraic function of x, y, and z of the

mth degree. The same thing may be shown of each term of rm sin nφPm n
(µ).
m
Consequently r Ym (µ, φ) is a homogeneous rational integral Algebraic function
of the mth degree in x, y, and z.
110. Any homogeneous rational integral Algebraic function Sm (x, y, z) of

the mth degree in x, y, and z, which is a value of V satisfying ∇2 V = 0 contains
2m + 1 arbitrary constant coefficients.
(m + 1)(m + 2)
For Sm (x, y, z) will in general consist of terms and will there-
2
(m + 1)(m + 2)
fore contain coefficients.
2
2
∇ Sm (x, y, z) will be homogeneous of the (m − 2)d degree and will contain
m(m − 1)
coefficients, which, of course, will be functions of the coefficients
2
in Sm (x, y, z). Since ∇2 Sm (x, y, z) = 0 independently of the numerical values
m(m − 1)
of x, y, and z the coefficients in ∇2 Sm (x, y, z) must be separately
2
m(m − 1)
zero, and that fact will give us equations of condition between the
2
(m + 1)(m + 2) (m + 1)(m + 2) m(m − 1)
original coefficients and will leave −
2 2 2
or 2m + 1 of them undetermined. Sm (x, y, z) contains, then, the same number
of arbitrary coefficients as rm Ym (µ, φ).
We can then choose the coefficients in rm Ym (µ, φ) so that it will transform
into any given Sm (x, y, z).
Consequently a Solid Spherical Harmonic of the mth degree might be defined
as a homogeneous rational integral Algebraic function of x, y, and z, Sm (x, y, z),
of the mth degree satisfying the equation ∇2 Sm (x, y, z) = 0; and a Surface
Spherical Harmonic of the mth degree as such a function divided by (x2 + y 2 +
m
z 2 ) 2 , that is by rm .
EXAMPLES.
1. Show that if Sm (x, y, z) is a Solid Spherical Harmonic of the mth degree
∇2 [rn Sm (x, y, z)] = n(2m + n + 1)rn−2 Sm (x, y, z).
Suggestion:
2 mSm
∇2 Sm = 0. ∇2 r = . Dr Sm = . (Dx r)2 + (Dy r)2 + (Dz r)2 = 1.
r r
2. Show that if fn (x, y, z) is a rational integral homogeneous function of x,
y, and z of the nth degree it can be expressed in the form
fn (x, y, z) = Sn (x, y, z) + r2 Sn−2 (x, y, z) + r4 Sn−4 (x, y, z) + · · · , (1)
terminating with rn−1 S1 (x, y, z) if n is odd, and with rn S0 (x, y, z) if n is even.
Suggestion: If a term rSn−1 were present in the second member of (1), and
we were to operate with ∇2 on both members we should by Ex. 1 have a term
2n
Sn−1 which would be irrational when all the other terms of the resulting
r
equation were rational. No such term, then, could occur. In the same way it
may be shown by operating twice on (1) with ∇2 that there can be no term
r3 Sn−3 in (1); and thus step by step we can reach the result formulated in (1).
3. Express x2 yz in the form S4 + r2 S2 + r4 S0 .
Suggestion: let x2 yz = S4 + r2 S2 + r4 S0
and take ∇2 of both members we get
2yz = 14S2 + 20r2 S0 .
Operate again with ∇2 . 0 = 120S0 . Whence

1 1
S0 = 0, S2 = yz, and S4 = (6x2 − y 2 − z 2 )yz.
7 7
4. Express sin2 θ cos2 θ sin φ cos φ in terms of Surface Spherical Harmonics.
x2 yz
Suggestion: sin2 θ cos2 θ sin φ cos φ = .
r4
For result v. Art. 108 (3).
111. A transformation of coördinates to a new set of axes having the

same origin as the old set will change a given Surface Spherical Harmonic into
another of the same degree. For such a transformation does not change the
form of Laplace’s Equation ∇2 V = 0 if both sets of axes are rectangular, and it
is effected by replacing x, y, and z in the Solid Harmonic corresponding to the
given Surface Harmonic by x cos α1 + y cos α2 + z cos α3 , x cos β1 + y cos β2 +
z cos β3 and x cos γ1 + y cos γ2 + z cos γ3 respectively where the cosines are the
direction cosines of the new axes, and it will leave the function a homogeneous
function of the mth degree in the new variables, and on dividing this by the
mth power of the unchanged radius vector we shall have a Surface Spherical
Harmonic of the mth degree.
112. We have seen in Art. 75 that if (x1 , y1 , z1 ) are the coördinates of a

given point
1
V =p (1)
(x − x1 ) + (y − y1 )2 + (z − z1 )2
2
is a solution of Laplace’s Equation ∇2 V = 0, and transforming to spherical

coördinates that
1
V =p (2)
r2 − 2rr1 [cos θ cos θ1 + sin θ sin θ1 cos(φ − φ1 )] + r12
is a solution of
1 1
rDr2 (rV ) + Dθ (sin θDθ V ) + Dφ2 V = 0. (3)
sin θ sin2 θ
If γ is the angle between the radii vectores r and r1 of the points (x, y, z)
and (x1 , y1 , z1 ) (1) can be written
1
V =p (4)
r − 2rr1 cos γ + r12
2
which must be equivalent to (2), and hence
cos γ = cos θ cos θ1 + sin θ sin θ1 cos(φ − φ1 ).
(4) which is a solution of (3) is of the same form as (5) Art. 75 and by developing
it as we developed (5) Art. 75 we find that
V = Pm (cos γ)

1 1
sin θ sin2 θ
1
and that V = rm Pm (cos γ) and V = m+1 Pm (cos γ)
r
are solutions of (3).
If we transform our coördinates keeping the origin unchanged and taking as
our new polar axis the radius vector of (x1 , y1 , z1 ) γ becomes our new θ and
Pm (cos γ) reduces to Pm (cos θ), a Surface Zonal Harmonic, or a Legendrian,3
of the mth degree. It is then a Legendrian having for its axis not the original
polar axis but the radius vector of (x1 , y1 , z1 ). Since a Legendrian is a Surface
Spherical Harmonic,
Pm (cos γ) = Pm [cos θ cos θ1 + sin θ sin θ1 cos(φ − φ1 )]
is a Surface Spherical Harmonic of the mth degree.

It is, however, of very special form, since being a determinate function of
µ, φ, µ1 , and φ1 it contains but two arbitrary constants if we regard it as a
function of µ and φ, instead of containing 2m + 1.
It is known as a Laplace’s Coefficient, or briefly as a Laplacian, of the mth
degree.
We shall soon express it in the regulation form of a Surface Spherical Har-
monic.
The radius vector of (x1 , y1 , z1 ) is called the axis of the Laplacian and the
point where the axis cuts the surface of the unit sphere is the pole of the Lapla-
cian.
We shall represent the Laplacian Pm (cos γ) by Lm (µ, φ, µ1 , φ1 ). Of course
Lm (µ, φ, 1, φ1 ) = Pm (µ) = Pm (cos θ), and is really independent of φ.
3 v. Art. 74.
113. If the product of a Surface Spherical Harmonic of the mth degree by

a Laplacian of the same degree is integrated over the surface of the unit sphere,
4π
the result is equal to multiplied by the value of the Spherical Harmonic
2m + 1
at the pole of the Laplacian.
That is,
w
2π w1 4π
dφ Ym (µ, φ)Lm (µ, φ, µ1 , φ1 )dµ = Ym (µ1 , φ1 ). (1)
2m + 1
0 −1
Transform to the axis of the Laplacian as a new polar axis, and let Zm (µ, φ)
be the transformed Spherical Harmonic. Lm (µ, φ, µ1 , φ1 ) will become Pm (µ),
and (1) will be proved if we can show that
w
2π w1 4π
dφ Zm (µ, φ)Pm (µ)dµ = Zm (1, 0). (2)
2m + 1
0 −1
n=m
X
Zm (µ, φ)Pm (µ) = A0 [Pm (µ)]2 + n
(An cos nφ + Bn sin nφ)Pm (µ)Pm (µ)
n=1
(v. (5) Art. 102).
w
2π
Zm (µ, φ)Pm (µ)dφ = 2πA0 [Pm (µ)]2 and
0
w1 w
2π
4π
dµ Zm (µ, φ)Pm (µ)dφ = A0 (v. (5) Art. 89).
2m + 1
−1 0
n
n
But Zm (1, 0) = A0 , since Pm (1) = 1 and Pm (1) contains (1 − 1) 2 as a factor
and is equal to zero.
Hence (2) is proved.
114. We can now express a Laplacian in the regulation form as a Spherical

Harmonic, by the formulas of Art. 107.
Lm (µ, φ, µ1 , φ1 ) =Pm (cos γ) = Pm [cos θ cos θ1 + sin θ sin θ1 cos(φ − φ1 )]

k=∞
X n=k
X
n
= A0,k Pk (µ) + (An,k cos nφ + Bn,k sin nφ)Pk (µ)
k=0 n=1
2m + 1 w w1
2π
where A0,m = dφ Lm (µ, φ, µ1 , φ1 )Pm (µ)dµ.
4π
0 −1
2m + 1 4π
= Pm (µ1 ) = Pm (µ1 ) by (1) Art. 113,
4π 2m + 1
2m + 1 (m − n)! w w1
2π
n
An,m = dφ Lm (µ, φ, µ1 , φ1 ) cos nφPm (µ)dµ
2π (m + n)!
0 −1
2(m − n)! n
= cos nφ1 Pm (µ1 ) by (1) Art. 113, and
(m + n)!
2m + 1 (m − n)! w w1
2π
n
Bn,m = dφ Lm (µ, φ, µ1 , φ1 ) sin nφPm (µ)dµ
2π (m + n)!
0 −1
2(m − n)! n
= sin nφ1 Pm (µ1 ) by (1) Art. 113,
(m + n)!
and A0,k = An,k = Bn,k = 0 by Art. 105 unless k = m. Hence
Lm (µ, φ, µ1 , φ1 ) =
n=m
X
(m − n)! n n
Pm (µ)Pm (µ1 ) + 2 P (µ)Pm (µ1 ) cos n(φ − φ1 ) . (1)
n=1
(m + n)! m
Each term of a Laplacian involves a numerical coefficient, a factor which is a

function of µ, a second factor which is the same function of µ1 , and a third
factor which is of the form cos k(φ − φ1 ). We give below a table of the first few
Laplacians, taken from Minchin’s Statics, omitting in each term for the sake of
brevity the function of µ1 .
By the aid of (1) we can write (5) Art. 107 more compactly. It becomes
m=∞
1 X w
2π w1
f (µ, φ) = (2m + 1) dφ1 f (µ1 , φ1 )Lm (µ, φ, µ1 , φ1 )dµ1 (2)
4π m=0
0 −1
m=∞
1 X w
2π wπ
or F (θ, φ) = (2m + 1) dφ1 F (θ1 , φ1 )Pm (cos γ) sin θ1 dθ1 . (3)
4π m=0
0 0
LAPLACIANS.
coef. of cos 0(φ − φ1 ) coef. of cos(φ − φ1 ) coef. of cos 2(φ − φ1 )
L0 1
1
L1 µ (1 − µ2 ) 2
1 1 3
L2 (3µ2 − 1) 3µ(1 − µ2 ) 2 (1 − µ2 )
4 4
1 3 1 15
L3 (5µ3 − 3µ) (1 − µ2 ) 2 (5µ2 − 1) µ(1 − µ2 )
4 8 4
1 5 1 5
L4 (35µ4 − 30µ2 + 3) (1 − µ2 ) 2 (7µ3 − 3µ) (1 − µ2 )(7µ2 − 1)
64 8 16
coef. of cos 3(φ − φ1 ) coef. of cos 4(φ − φ1 )
L0
L1
L2
5 3
L3 (1 − µ2 ) 2
8
35 3 35
L4 µ(1 − µ2 ) 2 (1 − µ2 )2
8 64
EXAMPLE.
Work the problems of Art. 108 and Art. 108 Exs. 1 and 2 by the aid of (3)
Art. 114.
115. Such problems as we have handled in Arts. 98 and 99, and also
problems differing from them in not having circular symmetry about an axis,
can now be solved by direct integration.
For instance let it be required to find the value at an external point of the
potential function due to the attraction of a solid sphere whose density at any
point is proportional to the product of any power of the radius vector by a
Surface Spherical Harmonic.
Let ρ = Cr1k Ym (µ1 , φ1 ).
Then using our ordinary notation we have
wa w
2π w1 Crk Y (µ , φ )r2 dµ
V = dr1 dφ1 p 1 m 1 1 1 1 .
0 0 −1
r2 − 2rr1 cos γ + r12
But by (3) Art. 77

1 1 r1
p = P0 (cos γ) + P1 (cos γ)
2 2
r − 2rr1 cos γ + r1 r r
r12 r1m

+ P 2 (cos γ) + · · · + Pm (cos γ) + · · ·
r2 rm
if r > r1 .
Consequently since
w
2π w1
dφ1 Ym (µ1 , φ1 )Yn (µ1 , φ1 )dµ1 = 0,
0 −1
V reduces to the single term
C wa w
2π w1
V = r1m+k+2 dr1 dφ1 Ym (µ1 , φ1 )Pm (cos γ)dµ1
rm+1
0 0 −1
C wa
4π

= r1m+k+2 Ym (µ, φ) dr1 by Art. 113.
rm+1 2m + 1
0
4πC am+k+3 Ym (µ, φ)

∴ V = . . .
2m + 1 m + k + 3 rm+1
EXAMPLES.
1. Solve by direct integration the problems worked in Arts. 98 and 99 and

Examples 1, 2, 3, and 4 of Art. 99.
2. The density of a solid sphere is proportional to the product of the squares

of the distances from two mutually perpendicular diametral planes; find the
value of the potential function at an external point.
Ans. ρ = kr14 cos2 θ1 sin2 θ1 cos2 φ1

4 1 1 1
= kr1 P0 (µ1 ) + P2 (µ1 ) + cos 2φ1 P22 (µ1 )
15 21 42

4 1 2
− P4 (µ1 ) + cos 2φ1 P4 (µ1 ) .
35 105
M a a3 1

1
V = + 3 P2 (µ) + cos 2φP22 (µ)
a r r 9 18
a5 4

1
− 5 P4 (µ) − cos 2φP42 (µ) .
r 33 99
3. Solve Example 2 by an extension of the method of Arts. 98 and 99.
4. A conducting sphere of radius a connected with the ground by a wire

is placed in the field of force due to an electrified point at which m units of
electricity are concentrated. Find the value of the potential function due to the
induced charge.
Suggestion: Let V1 be the potential function due to the point, and V2 that
due to the induced charge, and let b be the distance of the point from the centre
of the sphere. Then

m
V1 = √
b2
− 2br cos θ + r2
r2

m r
= P0 (cos θ) + P1 (cos θ) + 2 P2 (cos θ) + · · · if r < b.
b b b
b2

m b
= P0 (cos θ) + P1 (cos θ) + 2 P2 (cos θ) + · · · if r > b.
r r r
r r2
V2 = A0 P0 (cos θ) + A1 P1 (cos θ) + A2 2 P2 (cos θ) + · · · if r < a.
a a
a a2 a3
= A0 P0 (cos θ) + A1 2 P1 (cos θ) + A2 3 P2 (cos θ) + · · · if r > a.
r r r
When r = a V1 + V2 = 0. Hence
m ma ma2
A0 = − , A1 = − , A2 = − ,···
b b2 b3
and
r2

m r
V2 = − P0 (cos θ) + P1 (cos θ) + 2 P2 (cos θ) + · · · if r < a
b b b
a2 a4

ma
=− P0 (cos θ) + P1 (cos θ) + 2 2 P2 (cos θ) + · · · if r > a.
br br b r
Hence the effect of the induced charge is precisely the same at an external
ma
point as if the sphere were replaced by units of negative electricity concen-
b
a2
trated at the point r = , θ = 0. v. Peirce, Newt. Pot. Func., § 66.
b
116. If the two points P and P 0 are taken on the line OH whose direction
cosines are λ, µ, and ν, and if u and u0 are the valuesat P and 0
P of any
0
u −u
continuous function of the space coördinates, then limit . is called the
P P 0 =0 PP0
partial derivative of u along the line OH and will be represented by Dh u.
Let x, y, z be the coördinates of P and x+∆x, y+∆y, z+∆z the coördinates
of P 0 ; then
u0 − u = Dx u.∆x + Dy u.∆y + Dz u.∆z +
where is an infinitesimal of higher order than the first if ∆x, ∆y, and ∆z are
infinitesimal (v. Dif. Cal. Art. 198).
u0 − u ∆x ∆y ∆z
Hence = Dx u. + Dy u. + Dz u. + .
PP0 PP0 PP0 PP0 PP0
∆x ∆y ∆z
But = λ, = µ, and = ν.
PP0 PP0 PP0
Therefore Dh u = λDx u + µDy u + νDz u. (1)
If ∇2 u = 0, Dxp Dyq Dzr u is a solution of Laplace’s Equation.
For ∇2 (Dxp Dyq Dzr u) = Dxp Dyq Dzr (∇2 u) = 0.
Hence if ∇2 u = 0 Dh u is a solution of Laplace’s Equation, and if OH1 , OH2 ,

OH3 , · · · are a set of lines through the origin Dh1 Dh2 Dh3 · · · u is a solution of
Laplace’s Equation.
117. If Hk is a rational integral homogeneous Algebraic function of x, y,

and z of the kth degree

Hk Hk 1
Dx l
= Dr l
Dx r + l Dx (Hk )
r r r
lxHk Hk−1 lxHk r2 Hk−1
= − l+2 + l
= − l+2 + ,
r r r rl+2
Hk+1
and is of the form .
rl+2 H H
k k
The same thing can be proved of Dy l
and Dz l
and therefore holds
H r r
k
good of Dh .
rl
If u is a homogeneous function of x, y, and z of the degree −m − 1 and
∇2 u = 0 then ∇2 (r2m+1 u) = 0.
∇2 (r2m+1 u) = (2m + 1)(2m + 2)r2m−1 u

+ 2(2m + 1)r2m−1 (xDx u + yDy u + zDz u) + r2m+1 ∇2 u
= 0,
since xDx u + yDy u + zDz u = −(m + 1)u
by Euler’s Theorem (v. Dif. Cal. Art. 220).
M M
118. =p is a solution of Laplace’s Equation (v. Art. 75)
r x + y2 + z2
2
H0
and is of the form .
r
M
Dh1 Dh2 Dh3 · · · Dhm is then a solution of Laplace’s Equation by Art.
r
Hm
116; it is of the form 2m+1 by Art. 117 and is a homogeneous function of the
r
degree −m − 1.
M
Therefore r2m+1 Dh1 Dh2 Dh3 · · · Dhm is a solution of Laplace’s Equa-
r
tion, and is a rational integral homogeneous Algebraic function of x, y, and z
of the mth degree, and is consequently a Solid Spherical Harmonic of the mth
M
degree (v. Art. 110); and rm+1 Dh1 Dh2 Dh3 · · · Dhm is a Surface Spherical
r
Harmonic of the mth degree.
Moreover since the direction of each of the lines OH1 , OH2 , · · · OHm de-
pends upon two angles which may be taken at pleasure, these angles and M
are 2m + 1 arbitrary constants and may be so chosen that rm+1 Dh1 Dh2 Dh3 · · ·
M
Dhm may be any given Surface Spherical Harmonic.
r
Consequently any given Surface Spherical Harmonic may be regarded as
M
formed by differentiating successively along m determinate lines OH1 , OH2
r
· · · OHm , and is given except for the undetermined factor M when these lines
are given.
The lines OH1 , OH2 , OH3 , · · · OHm are called the axes of the Harmonic,
and the points where they meet the surface of the unit sphere the poles of the
Harmonic. The m axes of a Zonal Harmonic coincide with the axis of coördinates
(v. Art. 86) and consequently the m axes of a Laplacian coincide with what we
have called the axis of the Laplacian (v. Art. 112).
119. Any Surface Zonal Harmonic Pm (µ) is equal to zero for m real and
distinct values of µ which lie between −1 and 1; and any Associated Function
n
Pm (µ) is equal to zero for m − n real and distinct values of µ, which lie between
−1 and 1.
1 dm (µ2 − 1)m
Pm (µ) = . . v. Art. 83 (1).
2m m! dµm
dk (µ2 − 1)m
contains (µ2 − 1)m−k as a factor. v. Art. 89.
dµk
From Rolle’s Theorem, “If f (x) is continuous and single-valued and is equal
df (x)
to zero for the real values a and b of x, is equal to zero for at least one
dx
real value of x between a and b,” (v. Dif. Cal. Art. 126) it follows that since
d(µ2 − 1)m
(µ2 − 1)m = 0 when µ = −1 and when µ = 1 = 0 for at least one
dµ
d(µ2 − 1)m
value of µ between −1 and 1. cannot be equal to zero for more than
dµ
one value of µ between −1 and 1, for it contains (µ2 − 1)m−1 as a factor and is
a rational Algebraic polynomial of the 2m − 1st degree.
d2 (µ2 − 1)m
In like manner we can show that = 0 has m − 2 roots equal to
dµ2
−1, m − 2 roots equal to 1 and two real roots between −1 and 1 which separate
d(µ2 − 1)m
the three distinct roots of = 0; and in general if k < m + 1 that
dµ
k 2 m
d (µ − 1)
= 0 has m − k roots equal to −1, m − k roots equal to 1, and k
dµk
dk−1 (µ2 − 1)m
real roots separating the k + 1 distinct roots of = 0.
dµk−1
1 dm (µ2 − 1)m
Hence Pm (µ) = 0 or m . = 0 has m real and distinct roots
2 m! dµm
between −1 and 1, and it has no more since it is of the mth degree.
dm+n (µ2 − 1)m

The same reasoning shows that = 0 has m − n distinct real
dµm+n
n
roots between −1 and 1, and therefore that Pm (µ) is equal to zero for m − n
distinct real values of µ between −1 and 1. Since Pm n
(µ) contains sinn θ as a
factor it is also equal to zero when µ = −1 and when µ = 1.
cos nφ is equal to zero for 2n equidistant values of φ, and sin nφ is equal
n
to zero for 2n values of φ. Hence any Tesseral Harmonic sin nφPm (µ) or
n
cos nφPm (µ) is equal to zero for 2n equidistant values of φ, for µ = 1, for
µ = −1, and for m − n real and different values of µ between −1 and 1.
It follows that the value of any Surface Zonal Harmonic Pm (µ) at a point
on the surface of the unit sphere will have the same sign so long as the point
remains on one of the zones into which the surface of the sphere is divided by
the m circles of latitude corresponding to the m roots of Pm (µ) = 0, and will
change sign whenever the point passes from one of these zones into an adjoining
n
one; and that the value of any Tesseral Harmonic sin nφPm (µ) at a point on the
surface of the unit sphere will have the same sign so long as the point remains
on any one of the tesserae into which the surface of the sphere is divided by
n
the m − n circles of latitude corresponding to the roots of Pm (µ) = 0 and the
2n meridians corresponding to the roots of sin nφ = 0, and will change sign
whenever the point passes from one of these tesserae into an adjoining one.
220
CHAPTER VII.1
CYLINDRICAL HARMONICS (BESSEL’S FUNCTIONS).
120. In Arts. 11 and 17 we obtained
z = AJ0 (x) + BK0 (x) (1)
as the general solution of Fourier’s Equation
d2 z 1 dz
+ + z = 0, (2)
dx2 x dx
x2 x4 x6
where J0 (x) = 1 − 2 + 2 2 − 2 2 2 + · · · (3)
2 2 .4 2 .4 .6
and is called a Cylindrical Harmonic or Bessel’s Function of the zeroth order;
and where
x2 x4 x6

1 1 1 1 1
K0 (x) = J0 (x) log x + 2 − 2 2 + + 2 2 2 + + − · · · (4)
2 2 .4 1 2 2 .4 .6 1 2 3
and is called a Cylindrical Harmonic or Bessel’s Function of the Second Kind,

and of the zeroth order.
In Art. 17 we found that z = Jn (x)
is a particular solution of Bessel’s Equation
d2 z n2

1 dz
+ + 1 − z = 0, (5)
dx2 x dx x2
where if n is unrestricted in value
xn x2 x4

Jn (x) = n 1− 2 + 4
2 Γ(n + 1) 2 (n + 1) 2 .2!(n + 1)(n + 2)
x6

− 6 + ··· (6)
2 .3!(n + 1)(n + 2)(n + 3)
and is called a Cylindrical Harmonic or Bessel’s Function of the nth order; and
that unless n is an integer
z = AJn (x) + BJ−n (x)
is the general solution of Bessel’s Equation.

If n is an integer it can be shown that
Jn (x) = (−1)n J−n (x),

1 The student should re-read carefully Arts. 11, 17, and 18(d) before beginning this chapter.
CYLINDRICAL HARMONICS. 221
(v. Forsyth’s Diff. Eq. Art. 102), and then

z = AJn (x) + B{Kn (x)}
is the general solution of Bessel’s Equation and
k=n−1
1 x −n X (n − k − 1)! x 2k
{Kn (x)} = Jn (x) log x −
2 2 k! 2
k=0
k=∞
1 x n X (−1)k

1 1 1
− 1 + + + ··· +
2 2 (n + k)!k! 2 3 k
k=0

1 1 1 x 2k
+1 + + + · · · + (7)
2 3 n+k 2
v. M. Bôcher, Ann. Math. Vol. VI, No. 4.
121. A useful expression for Jn (x) as a definite integral can be obtained

without difficulty from Bessel’s Equation [(5) Art. 120] by a slight modification
of the method given by Forsyth (Diff. Eq. Art. 136).
It was shown in Art. 17 that z = xn v is a solution of Bessel’s Equation if v
satisfies the equation
d2 v 2n + 1 dv
2
+ + v = 0. (1)
dx x dx
wb
Assume v = T cos(xt)dt (2)
a
where x and t are independent, T is an unknown function of t, and a and b are

at present undetermined.
dv wb
Then = − tT sin(xt)dt
dx a
d2 v wb
and = − t2 T cos(xt)dt.
dx2 a
Substituting in (1) after multiplying through by x, we have

wb wb
(1 − t2 )T x cos(xt)dt − (2n + 1)tT sin(xt)dt = 0. (3)
a a
By integration by parts we find that

wb h ib
(1 − t2 )T x cos(xt)dt = (1 − t2 )T sin(xt)
a a
wb h dT i
− (1 − t2 ) − 2tT sin(xt)dt,
a
dt
and (3) reduces to
h ib wb h dT i
(1 − t2 )T sin(xt) − (1 − t2 ) + (2n − 1)tT sin(xt)dt = 0. (4)
a
dt
a
If we determine T so that
dT
(1 − t2 ) + (2n − 1)tT = 0, (5)
dt
h ib
and a and b so that (1 − t2 )T sin(xt) = 0 (6)
a
(4) will be satisfied and our problem will be solved. (5) gives
1
T = C(1 − t2 )n− 2 , (7)
and (6) will obviously be satisfied if a = −1 and b = 1.
w1 (1 − t2 )n cos(xt)dt
Hence v=C √ is a solution of (1),
−1
1 − t2
w1 (1 − t2 )n cos(xt)dt
and z = Cxn √ (8)
−1
1 − t2
is a solution of Bessel’s Equation.

If we let t = cos φ in (8) we get
wπ
z = Cxn sin2n φ cos(x cos φ)dφ.
0
Expand cos(x cos φ) into a series involving powers of x cos φ, integrate term
by term by the aid of the formulas
√ Γ n+1
π
w2 π
sinn x.dx = . 2 [Int. Cal. (1) Art. 99],
2 Γ n +1
0
2
π m + 1 n + 1
w2 Γ Γ
sinn x cosm x.dx = 2m + n 2
0 2Γ +1
2
(Int. Cal. Art. 99 Ex. 2), and compare with (6) Art. 120, and we get
xn w π
Jn (x) = √ sin2n φ cos(x cos φ)dφ. (9)
n
1
2 π Γ n+ 0
2
If n is a positive integer (9) reduces to
1 xn w π
Jn (x) = . sin2n φ cos(x cos φ)dφ. (10)
π 1.3.5. · · · (2n − 1)
0
Let n = 0 in (9) or (10) and we get
1w
π
J0 (x) = cos(x cos φ)dφ. (11)
π
0
EXAMPLES.
1. Obtain Formula (11) directly from Fourier’s Equation, (2) Art. 120.
1
2. Prove by integration by parts that if n > −
2
wπ x w
π
sin2n φ cos φ sin(x cos φ)dφ = sin2n+2 φ cos(x cos φ)dφ.
2n + 1
0 0
1
3. Prove by integration by parts that if n >
2
wπ
sin2n φ cos φ sin(x cos φ)dφ
0
1w
π
= [2n sin2n φ − (2n − 1) sin2n−2 φ] cos(x cos φ)dφ.
x
0
122. We can now readily obtain a number of useful formulas.

Differentiate (11) Art. 121 with respect to x and we get
1w
π
dJ0 (x)
=− cos φ sin(x cos φ)dφ
dx π
0
xw
π
=− sin2 φ cos(x cos φ)dφ by Ex. 2 Art. 121.
π
0
dJ0 (x)
Hence by (10) Art. 121 = −J1 (x). (1)
dx
In like manner by the aid of Exs. 3 and 2, Art. 121, we can obtain the
relations
d[xn Jn (x)]
= xn Jn−1 (x) (2)
dx
1
if n > ,
2
d[x−n Jn (x)]
= −x−n Jn+1 (x) (3)
dx
1
if n > − .
2
(2) can be written
wx
xn Jn−1 (x)dx = xn Jn (x) (4)
1 0
if n > .
2
(2) and (3) can be written
dJn (x)
xn + nxn−1 Jn (x) = xn Jn−1 (x)
dx
dJn (x)
and x−n − nx−n−1 Jn (x) = −x−n Jn+1 (x),
dx
dJn (x) n
or = Jn−1 (x) − Jn (x) (5)
dx x
dJn (x) n
and = − Jn+1 (x) + Jn (x); (6)
dx x
dJn (x)
whence 2 = Jn−1 (x) − Jn+1 (x) (7)
dx
2n
and Jn (x) = Jn−1 (x) + Jn+1 (x). (8)
x
The repeated use of formula (8) will enable us to get from J0 (x) and J1 (x)
any of Bessel’s Functions whose order is a positive integer. For example, we
have
2
J2 (x) = J1 (x) − J0 (x)
x
8 4
J3 (x) = 2
− 1 J1 (x) − J0 (x).
x x
From a table giving the values of J0 (x) and J1 (x), then, tables for the func-
tions of higher order are readily constructed. Such a table taken from Rayleigh’s
Sound (Vol. I., page 265) will be found in the Appendix (Table VI.).
By the aid of (5) and (6) any derivative of Jn (x) can be expressed in terms
of Jn (x) and Jn+1 (x). For example
d2 Jn (x)

n(n − 1) 1
= − 1 Jn (x) + Jn+1 (x).
dx2 x2 x
If we write J0 (x) for z in Fourier’s Equation [(2) Art. 120], then multiply
through by xdx and integrate from zero to x, simplifying the resulting equation
by integration by parts, we get
dJ0 (x) w
x
x + xJ0 (x)dx = 0;
dx
0
wx
whence by (1) xJ0 (x)dx = xJ1 (x). (9)
0
If we write J0 (x) for z in Fourier’s Equation, then multiply through by

2 dJ0 (x)
x dx and integrate from zero to x, simplifying by integration by parts
dx
we get
x2

dJ0 (x)
2 wx
2
+ (J0 (x)) − x(J0 (x))2 dx = 0;
2 dx
0
wx x2
whence by (1) x(J0 (x))2 dx = [(J0 (x))2 + (J1 (x))2 ]. (10)
2
0
In like manner we can get from Bessel’s Equation [(5) Art. 120] the formula
wx
1 2 dJn (x)
2
2 2 2 2
x(Jn (x)) dx = x + (x − n )(Jn (x)) (11)
2 dx
0
which (6) enables us to reduce to the form

wx x2
x(Jn (x))2 dx = [(Jn (x))2 + (Jn+1 (x))2 ] − nxJn (x)Jn+1 (x). (12)
2
0
Formulas (9), (10), (11), and (12) will prove useful when we attempt to
develop in terms of Cylindrical Harmonics.
Values of Jn (x) for larger values of x than those given in Table VI., Appendix,
may be computed very easily from the formula
r
(12 − 4n2 )(32 − 4n2 )

2
Jn (x) = 1−
πx 2!(8x)2
(12 − 4n2 )(32 − 4n2 )(52 − 4n2 )(72 − 4n2 )
π π
+ 4
− · · · cos x − − n
4!(8x) 4 2
r 2 2
2 1 − 4n
+
πx 1!8x
(12 − 4n2 )(32 − 4n2 )(52 − 4n2 )
π π
− + · · · sin x − − n . (13)
3!(8x)3 4 2
v. Lommel, Studien über die Bessel’schen Functionen, page 59.
The series terminates if 2n is an odd integer, but otherwise it is divergent. It

can be proved, however, that in any case the sum of m terms differs from Jn (x)
by less than the last term included, and consequently the formula can safely be
used for numerical computation.
EXAMPLES.
1. Confirm (1), (2), and (3), Art. 122, by obtaining them from (3) and (6),
Art. 120.
2. Confirm (1), Art. 122, by showing that Fourier’s Equation will differen-
tiate into the special form assumed by Bessel’s Equation when n = 1.
3. Show that (9), Art. 122, is a special case of (4), Art. 122.
4. Show that the limit approached by Jn (x) as n increases indefinitely is

zero, and by the aid of this fact and of (8), Art. 122, prove that
2
Jn−1 (x) = [nJn (x) − (n + 2)Jn+2 (x) + (n + 4)Jn+4 (x) + · · · ].
x
5. Prove that
dJn (x) 2
= [ 12 nJn (x) − (n + 2)Jn+2 (x) + (n + 4)Jn+4 (x) − · · · ].
dx x
12
y2

6. Show that the substitution of 1− 2 for x in Legendre’s Equation
n
will reduce it to the form
y 2 d2 z

1 2y dz 1
1− 2 + − + 1+ z = 0,
n dy 2 y n2 dy n
and that the limiting form approached by this equation as n is indefinitely

increased is Fourier’s Equation, and hence that J0 (x) can be regarded as some
1
x2 2

constant factor multiplied by the limiting value approached by Pn 1 − 2
n
as n is indefinitely increased.
123. To complete the solution of the drumhead problem taken up in Art.

11, we found that it would be necessary to develop a given function of r in the
form
f (r) = A1 J0 (µ1 r) + A2 J0 (µ2 r) + A3 J0 (µ3 r) + · · ·
where µ1 , µ2 , µ3 , &c., are the roots of the transcendental equation J0 (µa) = 0;
and in Art. 11, Ex. the development of unity in a series of precisely the same
form was needed.
(a) Let us consider another problem.

The convex surface and one base of a cylinder of radius a and length b are
kept at the constant temperature zero, the temperature at each point of the
other base is a given function of the distance of the point from the centre of the
base; required the temperature of any point of the cylinder after the permanent
temperatures have been established.
Here we have to solve Laplace’s Equation in Cylindrical Coördinates ([XIV]
Art. 1).
1 1
Dr2 u + Dr u + 2 Dφ2 u + Dz2 u = 0 (1)
r r
u=0 when z = 0
u=0 “ r=a
u = f (r) “ z = b,
and from the symmetry of the problem we know that Dφ2 u = 0.

Assuming as usual u = R.Z we break (1) up into the equations
d2 Z
− µ2 Z = 0
dz 2
d2 R 1 dR
+ + µ2 R = 0,
dr2 r dr
whence u = sinh(µz)J0 (µr) (2)
and u = cosh(µz)J0 (µr) (3)
If µk is a root of J0 (µa) = 0 (4)
u = sinh(µk z)J0 (µk r)

satisfies (1) and two of the three equations of condition.
If then f (r) = A1 J0 (µ1 r) + A2 J0 (µ2 r) + A3 J0 (µ3 r) + · · · (5)
µ1 , µ2 , µ3 , &c., being roots of (4),

sinh(µ1 z) sinh(µ2 z) sinh(µ3 z)
u = A1 J0 (µ1 r)+A2 J0 (µ2 r)+A3 J0 (µ3 r)+· · · (6)
sinh(µ1 b) sinh(µ2 b) sinh(µ3 b)
satisfies (1) and all of the equations of condition, and is the required solution.
(b) If instead of keeping the convex surface of the cylinder at the tem-
perature zero we surround it by a jacket impervious to heat, the equation of
condition u = 0 when r = a will be replaced by Dr u = 0 when r = a, or if
u = sinh(µz)J0 (µr),
dJ0 (µr)
by =0 when r = a,
dr
0
that is by µJ0 (µa) = 02 or (v. (1) Art. 122)
by J1 (µa) = 0. (7)
If now in (5) and (6) µ1 , µ2 , µ3 , &c., are roots of (7), (6) will be the solution
of our new problem.
(c) If instead of keeping the convex surface of the cylinder at the tempera-
ture zero we allow it to cool in air at the temperature zero, the condition u = 0
when r = a will be replaced by Dr u + hu = 0 when r = a, or if
u = sinh(µz)J0 (µr)
by µJ00 (µr) + hJ0 (µr) = 0 when r=a
that is by µaJ00 (µa) + ahJ0 (µa) = 0 or (v. (1) Art. 122)
by µaJ1 (µa) − ahJ0 (µa) = 0. (8)
If now in (5) and (6) µ1 , µ2 , µ3 , &c., are roots of (8), (6) will be the solution
of our present problem.
124. It can be shown that J0 (x) = 0 (1)

J1 (x) = 0 (2)
and xJ00 (x) + λJ0 (x) = 0 (3)
have each an infinite number of real positive roots (v. Riemann, Par. Dif. Gl.,
§ 97). The earlier roots of these equations can be computed without serious
difficulty from the table for the values of J0 (x) (Table VI., Appendix).
The first twelve roots of J0 (x) = 0 and J1 (x) = 0 are given in Table IV.,
Appendix, a table due to Stokes. Large roots of J0 (x) = 0 and of J1 (x) = 0
may be very easily computed from the formulas
(s)
x 0 .050661 .053041 .262051
= s − .25 + − + − ··· (4)
π 4s − 1 (4s − 1)3 (4s − 1)5
(s)
x 1 .151982 .015399 .245270
= s + .25 − + 3
− + ··· (5)
π 4s + 1 (4s + 1) (4s + 1)5
df (x)
2 We shall find it convenient to use the familiar notation of f 0 (x) = (v. Dif. Cal., p.
dx
119).
(s)
given by Stokes in Camb. Phil. Trans., Vol. IX., x 0 representing the sth root
(s)
of J0 (x) = 0, and x1 the sth root of J1 (x) = 0.
125. We have seen in Art. 123 that U = sinh(µk z)J0 (µk r) and V =
sinh(µl z)J0 (µl r) are solutions of ∇2 U = 0 and ∇2 V = 0 if we express Laplace’s
Equation in terms
r of Cylindrical Coördinates (v. (1) Art. 123).
Hence, if dS represents the surface integral over any closed surface, we
have w
(U Dn V − V Dn U )dS = 0
by Green’s Theorem (v. Art. 92).
If we take the cylinder of Art. 123 as our surface, and perform the integra-
tions and simplify the resulting equation, we find
wa −1
rJ0 (µk r)J0 (µl r)dr = [µk aJ0 (µl a)J00 (µk a) − µl aJ0 (µk a)J00 (µl a)]
µk 2 − µl 2
0
−1
= [µk aJ0 (µl a)J1 (µk a) − µl aJ0 (µk a)J1 (µl a)]. (1)
µl − µk 2
2
Hence if µk and µl are different roots of

J0 (µa) = 0,
or of J1 (µa) = 0,
or of µaJ1 (µa) − λJ0 (µa) = 0,
wa
then rJ0 (µk r)J0 (µl r)dr = 0. (2)
0
EXAMPLE.
Obtain (1) Art. 125 directly from Fourier’s Equation
d2 J0 (µr) 1 dJ0 (µr)
+ + µ2 J0 (µr) = 0.
dr2 r dr
126. We are now able to obtain the developments called for in Art. 123.
Let f (r) = A1 J0 (µ1 r) + A2 J0 (µ2 r) + A3 J0 (µ3 r) + · · · (1)
µ1 , µ2 , µ3 , &c., being roots of J0 (µa) = 0, or of J1 (µa) = 0, or of
µaJ1 (µa) − λJ0 (µa) = 0.
To determine any coefficient Ak multiply (1) by rJ0 (µk r)dr and integrate
from zero to a. The first member will become
wa
rf (r)J0 (µk r)dr.
0
Every term of the second member will vanish by (2) Art. 125 except the
term
wa
Ak r(J0 (µk r))2 dr.
0
wa 1
µwk a
a2
r(J0 (µk r))2 dr = x(J0 (x))2 dx = [(J0 (µk a))2 + (J1 (µk a))2 ]
µ2k 2
0 0
by (10) Art. 122.
2 w a
Hence Ak = 2 rf (r)J0 (µk r)dr. (2)
a [(J0 (µk a))2 + (J1 (µk a))2 ]
0
The development (1) holds good from r = 0 to r = a (v. Arts. 24, 25, and 88).
If µ1 , µ2 , µ3 , &c., are roots of J0 (µa) = 0, (2) reduces to
2 wa
Ak = rf (r)J0 (µk r)dr. (3)
a2 (J1 (µk a))2
0
If µ1 , µ2 , µ3 , &c., are roots of J1 (µa) = 0, (2) reduces to
2 wa
a2 (J0 (µk a))2
0
If µ1 , µ2 , µ3 , &c., are roots of µaJ1 (µa) − λJ0 (µa) = 0, (2) reduces to
2µ2k wa
(λ2 + µ2k a2 )(J0 (µk a))2
0
For the important case where f (r) = 1

wa wa 1 w
µk a
a
rf (r)J0 (µk r)dr = rJ0 (µk r)dr = 2 xJ0 (x)dx = J1 (µk a) (6)
µk µk
0 0 0
by (9) Art. 122, and (3) reduces to
2
Ak = , (7)
µk aJ1 (µk a)
(4) reduces to Ak = 0 except for k = 1 when µk = 0 and we have A1 = 1,
2λ
(5) reduces to Ak = . (8)
(λ2 + µ2k a2 )J0 (µk a)
EXAMPLES.
1. Show that in (12) Art. 11 any coefficient Ak has the value given in (3)
Art. 126; and in the answer to Art. 11, Ex. the value given in (7) Art. 126.
2. Show that if a drumhead be initially distorted so that it has circular

symmetry, it will not in general give a musical note; that it may be initially
distorted so as to give a musical note; that in this case the vibration will be a
steady vibration; that the frequencies of the various musical notes that can be
given when the distortion has circular symmetry are proportional to the roots
of J0 (x) = 0; that the possible nodes for such vibrations are concentric circles
whose radii are proportional to the roots of J0 (x) = 0.
3. A cylinder of radius one meter and altitude one meter has its upper
surface kept at the temperature 100◦ , and its base and convex surface at the
temperature 15◦ , until the stationary temperature is set up. Find the tempera-
ture at points on the axis 25 cm., 50 cm., and 75 cm. from the base, and also
at a point 25 cm. from the base and 50 cm. from the axis.
Ans., 29◦ .6; 47◦ .6; 71◦ .2; 25◦ .8.
4. An iron cylinder one meter long and twenty centimeters in diameter has
its convex surface covered with a so-called non-conducting cement one centime-
ter thick. One end and the convex surface of the cylinder thus coated are kept
at the temperature zero, the other end at the temperature of 100◦ . Find to
the nearest tenth of a degree the temperature of the middle point of the axis,
and of the points of the axis twenty centimeters from each end after the tem-
peratures have ceased to change. Given that the conductivity of iron is 0.185
and of cement 0.000162 in C. G. S. units. Find also the temperature of a point
on the surface midway between the ends, and of points on the surface twenty
centimeters from each end. Find the temperatures of the three points of the
axis, supposing the coating a perfect non-conductor, and again, supposing the
coating absent. Neglect the curvature of the coating.
Ans., 15◦ .4; 40◦ .85; 72◦ .8; 15◦ .3; 40◦ .7; 72◦ .5; 0◦ .0; 0◦ .0; 1◦ .3.
127. If instead of considering the cooling of a cylinder as in Art. 123 we

have to deal with a cylindrical shell whose curved surfaces are co-axial cylinders,
we are obliged to use the Bessel’s Functions of the second kind. Let our equations
of condition be
u=0 when z = 0, u=0 when r = a,

u = f (r) “ z = b, u=0 “ r = c.
Then (v. Art. 123)

J0 (µk c)
u = sinh(µk z) J0 (µk r) − K0 (µk r)
K0 (µk c)
where µk is a root of the equation

J0 (µc)
J0 (µa) − K0 (µa) = 0 (1)
K0 (µc)
will satisfy Laplace’s Equation [(1) Art. 123] and all of the equations of condition
except the second.
k=∞
X sinh(µk z) J0 (µk c)
Hence u = Ak J0 (µk r) − K0 (µk r) (2)
sinh(µk b) K0 (µk c)
k=1
is the required solution if

k=∞
X J0 (µk c)
f (r) = Ak J0 (µk r) − K0 (µk r) . (3)
K0 (µk c)
k=1
The development (3) is easily obtained.

Call the parenthesis for the sake of brevity B0 (µk r). Then by the method
of Art. 125 we get if we integrate over our cylindrical shell
wc
rB0 (µk r)B0 (µl r)dr = 0 (4)
a
if µk and µl are roots of (1); and by an easy extension of (10) Art. 122
wc
r[B0 (µk r]2 dr = 21 {c2 [B00 (µk c)]2 − a2 [B00 (µk a)]2 }. (5)
a
Determining the coefficients in (3) as in Art. 124 and simplifying by the aid
of (4) we have
wc h J0 (µk c) i
2 rf (r) J0 (µk r) − K0 (µk r) dr
a
K0 (µk c)
Ak = h 2 i2 .
J0 (µk c) 0 i h J0 (µk c) 0
c2 J00 (µk c) − K0 (µk c) − a2 J0 0 (µk a) − K0 (µk a)
K0 (µk c) K0 (µk c)
(6)
EXAMPLE.
If a membrane bounded by concentric circles of radius a and radius b, and
fastened at the edges, is initially distorted into a form symmetrical with respect
to the centre, and then allowed to vibrate
k=∞
X J0 (µk b)
y= Ak cos(µk ct) J0 (µk r) − K0 (µk r)
K0 (µk b)
k=1
where Ak is obtained from (6) Art. 127 by replacing c by b.

128. If in the cooling of a cylinder u = 0 when z = 0, u = 0 when z = b,

and u = f (z) when r = a, the problem is easily solved.
If in (2) and (3) Art. 123 µ is replaced by µi we can readily obtain
z = sin(µz)J0 (µri)
and z = cos(µz)J0 (µri)
as particular solutions of Laplace’s Equation [(1) Art. 123]; and
x2 x4 x6
J0 (xi) = 1 + + + + ··· (1)
22 22 .42 22 .42 .62
and is real.
k=∞
X kπz
f (z) = Ak sin
b
k=1
2w
b
kπz
where Ak = f (z) sin dz (2)
ba b
by Art. 31 (7) and (8).

kπri
k=∞ J
kπz 0 b
X
Hence u= Ak sin kπai (3)
b
k=1 J0
b
EXAMPLES.
1. If the cylinder is hollow and we have u = 0 when z = 0, u = 0 when

z = b, u = 0 when r = c, and u = f (z) when r = a; then
 kπri kπri 
k=∞
kπz  J0 K0
b b
X
u= Ak sin kπci −

kπci 
b

k=1 J0 K0
b b
 kπai kπai 
J0 K0
÷ b b
kπci −
 
kπci 
J0 K0
b b
where Ak has the value given in (2) Art. 128, and

K0 (xi) = K0 (xi) − J0 (xi) log i
x2 x4 x6
= J0 (xi) log x − 2
− 2 2 ( 11 + 21 ) − 2 2 2 ( 11 + 1
2 + 31 ) − · · ·
2 2 .4 2 .4 .6
[v. (4) Art. 120], and is real.
2. A hollow cylinder 6 feet long whose inner surface has the radius 3 inches,
and whose outer surface has the radius 1 foot, has its bases and outer surface
kept at the temperature 0◦ , and its inner surface at the temperature 100◦ , until
the permanent state of temperatures is established; find the temperatures of
two points in a plane parallel to the bases and half-way between them, one of
which is 6 inches and the other 9 inches from the axis. Ans., 49◦ .6; 20◦ .2.
129. If in the problem of Art. 123 the temperatures of the points of the
upper base of the cylinder are unsymmetrical so that u = f (r, θ) when z = b,
we have to get particular solutions of Laplace’s Equation [(1) Art. 123] for the
case where Dφ2 u is not equal to zero. We readily find that
u = sinh (µz)[A cos nφ + B sin nφ]Jn (µr)

and u = cosh (µz)[A cos nφ + B sin nφ]Jn (µr)
are such solutions, and that
n=∞
X k=∞
X sinh µk z
u= [An,k cos nφ + Bn,k sin nφ]Jn (µk r) (1)
n=0 k=1
sinh µk b
is the solution of the given problem if

X k=∞
n=∞ X
f (r, φ) = (An,k cos nφ + Bn,k sin nφ)Jn (µk r) (2)
n=0 k=1
where µk is a root of the equation

Jn (µa)
= 0. (3)
µn an
EXAMPLES.
1. Show that
wa
rJn (µk r)Jn (µl r)dr
0 a
= [µl Jn (µk a)Jn0 (µl a) − µk Jn (µl a)Jn0 (µk a)]
µk 2 − µl 2
a
= 2 [µk Jn (µl a)Jn+1 (µk a) − µl Jn (µk a)Jn+1 (µl a)].
µk − µl 2
2. Show that
wa
r[Jn (µk r)]2 dr
0
1h 2 0 n2 i
= a (Jn (µk a))2 + a2 − 2 (Jn (µk a))2
2 µk
2
a na
= [(Jn (µk a))2 + (Jn+1 (µk a))2 ] − Jn (µk a)Jn+1 (µk a).
2 µk
3. Show that in Art. 129
w
2π wa
dφ rf (r, φ)J0 (µk r)dr
1 0 0
A0,k = ,
π a2 [J1 (µk a)]2
B0,k = 0,
w
2π wa
dφ rf (r, φ) cos nφJn (µk r)dr
2 0 0
An,k = ,
π a2 [Jn+1 (µk a)]2
w
2π wa
dφ rf (r, φ) sin nφJn (µk r)dr
2 0 0
Bn,k = .
π a2 [Jn+1 (µk a)]2
4. Obtain the coefficients for the case where the convex surface of the cylin-
der is impervious to heat.
5. Obtain the coefficients for the case where the convex surface of the cylin-
der is exposed to air at the temperature zero.
6. Show that if in a drumhead problem of Art. 11 the initial distortion is

unsymmetrical, so that we have to solve the equation [XI] Art. 1 subject to the
conditions z = f (r, φ) when t = 0, Dt z = 0 when t = 0, z = 0 when r = a, the
solution is
n=∞
X k=∞
X
z= cos(µk ct)(An,k cos nφ + Bn,k sin nφ)Jn (µk r)
n=0 k=1
where A0,k , B0,k , An,k , and Bn,k have the values given in Ex. 3.
7. What modifications do the statements made in Ex. 2, Art. 126, need to

make them apply to the unsymmetrical case treated in Ex. 6?
Show that any possible nodal system in Ex. 6 is composed of concentric cir-
cles and of radii whose outer extremities are equidistant. v. Rayleigh’s Sound,
Vol. I., Arts. (202-207).
8. Solve the problem of Art. 127 and of Art. 127. Ex. for the unsymmetrical
case. Suggestion: AJn (x) + BKn (x) is a solution of Bessel’s Equation.
9. Solve the problem of Art. 128 and of Art. 128. Ex. 1. for the case where
u = f (z, φ) when r = a. Suggestion: u = sin µz(A cos nφ + B sin nφ)Jn (µri)
is a solution of Laplace’s Equation, and f (z, φ) can be developed into a double
Fourier’s Series [v. (15) Art. 71].
10. Show that in dealing with a wedge cut from a cylinder by planes passed
through the axis, or with a membrane in the form of a circular sector, it may
be necessary to use Bessel’s Functions of fractional or incommensurable orders.
11. Bernouilli’s Problem (v. Chapter IX). In considering small transverse

vibrations of a uniform, heavy, flexible, inelastic string fastened at one end
and initially distorted into some given curve, we have to solve the equation
Dt2 y = c2 (xDx2 y +Dx y), subject to the conditions Dt y = 0 when t = 0, y = f (x)
when t = 0, y = 0 when x = a; the origin being taken at the distance a below
the point of suspension and the axis of X taken vertical.
k=∞
X
Show that y= Ak cos µk ct B0 (µ2k x),
k=1
x x2 x3
where B0 (x) = 1 − + − + ···
12 12 .22 12 .22 .32
√
= J0 (2 x)
and µk is a root of the equation

√
B0 (µ2 a) = J0 (2µ a) = 0,
wa wa √
f (x)B0 (µ2k x)dx f (x)J0 (2µk x)dx
0 0
and Ak = = √ .
µ2 a2 [B00 (µ2k a)]2 a[J1 (2µk a)]2
.
12. As a simple case under Example 10 consider the vibrations of a circular

membrane fastened at the perimeter and also along a radius and then initially
distorted (v. Rayleigh’s Sound, Art. 207). In this case we must modify the
formula given in Ex. 6 by dropping out the terms involving cos nφ and by taking
m
n = . The required solution is
2
m=∞
X k=∞
X mφ
z= Bm,k cos µk ct sin J m (µk r)
m=1 k=1
2 2
J m2 (µa)
where µk is a root of m m =0
µ2a2
w
2π wa mφ
dφ rf (r, φ) sin J m (µk r)dr
2 2 2
0 0
and Bm,k = .
π a2 [J m2 0 (µk a)]2
For the terms in which m is odd, J m2 (x) can be readily obtained from (13)
Art. 122, which will become a finite sum.
For example, (13) Art. 122 gives the values
r r
2 2 1
J 12 (x) = sin x; J 32 (x) = sin x − cos x ;
πx πx x
r
2 3 3
J 25 (x) = − 1 + 2 sin x + cos x ; &c.
πx x x
13. The question of the flow of heat in three dimensions involves a problem
not unlike the last.
Suppose the initial temperatures of all points in a sphere of radius c given,
and let the surface be kept at the temperature zero. Then we have to solve the
equation
a2

2 1 1 2
Dt u = 2 Dr (r Dr u) + Dθ (sin θDθ u) + Dφ u (1)
r sin θ sin2 θ
([IV] Art. 1) subject to the conditions
u=0 when r = c,
u = f (r, θ, φ) when t = 0.
If we assume u = T.R.V where T is a function of t only, R of r only, and V of

θ and φ only, (1) can be broken up into
dT
+ a2 α2 T = 0 (2)
dt
1 1
sin θ sin2 θ
d2 R 2 dR

2 m(m + 1)
and + + α − R = 0. (4)
dr2 r dr r2
2 2
Hence T = e−a α t , V = Ym (µ, φ)√[v. Art. 102 (2)], and R is still to be found.
If in (4) we let x = αr and z = R αr it becomes
d2 z (m + 21 )2

1 dz
+ + 1 − z=0
dx2 x dx x2
which is satisfied by z = Jm+ 12 (x). (v. Art. 17.)
1
Therefore R = √ Jm+ 12 (αr).
αr
m=∞
1 X w
2π wπ
f (r, θ, φ) = (2m + 1) dφ1 f (r, θ1 , φ1 )Pm (cos γ) sin θ1 dθ1
4π m=0
0 0
by (3) Art. 114,
m=∞
X n=m
X
n
= [Am,n fm,n (r) cos nφ + Bm,n Fm,n (r) sin nφ]Pm (µ).
m=0 n=0
k=∞
√ X
rfm,n (r) = Cm,n,k Jm+ 12 (αk r)
k=0
where αk is a root of the equation
Jm+ 12 (αc)
1 =0
(αc)m+ 2
wc 3
2 r 2 fm,n (r)Jm+ 12 (αk r)dr
0
and Cm,n,k = 0 .
c2 [Jm+ 1 (αk c)]
2
2
m=∞
√ X
rFm,n (r) = Dm,n,k Jm+ 12 (αk r)
m=0
wc 3
2 r 2 Fm,n (r)Jm+ 12 (αk r)dr
0
where Dm,n,k = 0 .
c2 [Jm+ 1 (αk c)]
2
2
The final solution is

m=∞ n=m k=∞
1 X X n
X
u= √ Pm (µ) (Am,n Cm,n,k cos nφ
r m=0 n=0
k=1

−a2 α2k t
+ Bm,n Dm,n,k sin nφ)e Jm+ 12 (αk r)
cf. Riemann, Par. Dif. Gl., §§ 72 and 73.

239
CHAPTER VIII.
LAPLACE’S EQUATION IN CURVILINEAR COÖRDINATES.
ELLIPSOIDAL HARMONICS.
130. Orthogonal Curvilinear Coördinates.


If F1 (x, y, z) = ρ1 

F2 (x, y, z) = ρ2 (1)

F3 (x, y, z) = ρ3

are the equations in rectangular coördinates of three surfaces that are mutually
perpendicular no matter what the values of ρ1 , ρ2 , and ρ3 , the parameters ρ1 ,
ρ2 , and ρ3 , may be regarded as a set of coördinates for a point of intersection
of the three surfaces, in the sense that when ρ1 , ρ2 , ρ3 are given the point in
question is determined, and when the point is given the corresponding values of
ρ1 , ρ2 , ρ3 , can be found.
From equations (1) x, y, and z can be expressed in terms of ρ1 , ρ2 , and ρ3 .
Suppose this done. If now x, y, z are the rectangular coördinates of the point
ρ1 = a, ρ2 = b, ρ3 = c, the rectangular coördinates of the points ρ1 = a + dρ1 ,
ρ2 = b, ρ3 = c, are obviously x+Dρ1 x.dρ1 +1 , y+Dρ1 y.dρ1 +2 , z+Dρ1 z.dρ1 +3 ,
where 1 , 2 , and 3 are infinitesimals of higher order than dρ1 . Hence the square
of the distance between the points will differ by an infinitesimal of higher order
than that of dρ21 from dn21 where
dn21 = [(Dρ1 x)2 + (Dρ1 y)2 + (Dρ1 z)2 ]dρ21 .
1 
Let 2 = (Dρ1 x)2 + (Dρ1 y)2 + (Dρ1 z)2 
h1





1 2 2 2

= (Dρ2 x) + (Dρ2 y) + (Dρ2 z) (2)
h22 


1 2 2 2 

= (Dρ x) + (Dρ y) + (Dρ z) . 

h23 3 3 3
Then if dn1 is the element of length normal to the surface ρ1 = a, dn2 normal
to ρ2 = b, and dn3 normal to ρ3 = c
dρ1 dρ2 dρ3
dn1 = , dn2 = , dn3 = . (3)
h1 h2 h3
The element of surface dS1 on the surface ρ1 = a is easily seen to be
dρ2 dρ3
dS1 = ; (4)
h2 h3
and the element of volume dv is
dρ1 dρ2 dρ3
dv = . (5)
h1 h2 h3
ELLIPSOIDAL HARMONICS. 240
EXAMPLE.
Show that h21 = (Dx ρ1 )2 + (Dy ρ1 )2 + (Dz ρ1 )2

h22 = (Dx ρ2 )2 + (Dy ρ2 )2 + (Dz ρ2 )2
h23 = (Dx ρ3 )2 + (Dy ρ3 )2 + (Dz ρ3 )2 .
Dx ρ1 Dy ρ1 Dz ρ1
Suggestion: If h1 has the value just given , , are the direc-
h1 h1 h1
tion cosines of the normal at any given point of ρ1 = a. (v. Int. Cal. page 161.)
Then
Dx ρ1 D y ρ1 D z ρ1 1
dn1 = dx + dy + dz = dρ1 .
h1 h1 h1 h1
131. Laplace’s Equation in orthogonal curvilinear coördinates.

If we apply the special form of Green’s Theorem
y w
∇2 V dxdydz = Dn V dS (v. Art. 98)
to the space bounded by the surfaces ρ1 = a, ρ2 = b, ρ3 = c, ρ1 = a + dρ1 ,

ρ2 = b + dρ2 , ρ3 = c + dρ3 , we have
∇2 V dρ1 dρ2 dρ3
=
h1 h2 h3

dρ2 dρ3 dρ2 dρ3 h1
−h1 Dρ1 V + h1 Dρ1 V + Dρ1 Dρ V dρ1 dρ2 dρ3
h2 h3 h2 h3 h2 h3 1

−h2 Dρ2 V + h2 Dρ2 V + Dρ2 Dρ V dρ1 dρ2 dρ3
h3 h1 h3 h1 h3 h1 2

−h3 Dρ3 V + h3 Dρ3 V + Dρ3 Dρ V dρ1 dρ2 dρ3 ;
h1 h2 h1 h2 h1 h2 3
whence

h1 h2 h3
∇2 V = h1 h2 h3 Dρ1 Dρ1 V + Dρ2 Dρ2 V + Dρ3 Dρ3 V ,
h2 h3 h3 h1 h1 h2
(6)
and Laplace’s Equation in our curvilinear system is

h1 h2 h3
h1 h2 h3 Dρ1 Dρ V + Dρ2 Dρ V + Dρ3 Dρ V = 0. (7)
h2 h3 1 h3 h1 2 h1 h2 3
If it happens
that ∇2 ρ1 = 0, V = ρ1 will satisfy (7) and we
shall have
h1 h2
h1 h2 h3 Dρ1 = 0. In like manner if ∇2 ρ2 = 0 we have Dρ2 = 0,
h2 h3 h3 h1
h3
and if ∇2 ρ3 = 0 we have Dρ3 = 0; and therefore (7) reduces to
h1 h2
h21 Dρ21 V + h22 Dρ22 V + h23 Dρ23 V = 0 (8)
when ∇2 ρ1 = 0, ∇2 ρ2 = 0, and ∇2 ρ3 = 0.
132. If instead of having the value of the Potential Function V given on

the surface of a sphere as in our Spherical Harmonic problem, we have it given
at all the points on the surface of an oblate spheroid, and are required to find its
value at any internal or external point, we can easily get a solution by methods
in no essential respect different from those already employed, if only we rightly
choose our system of coördinates.
If we take an ellipse and an hyperbola having the same foci, and revolve
them about the minor axis of the ellipse, we shall get a pair of surfaces which
are mutually perpendicular; a plane through the axis of revolution will cut both
the spheroid and the hyperboloid orthogonally.
The equations of the three surfaces can be written:—
x2 y2 z2
F1 (x, y, z, λ) = + + −1=0 (1)
λ2 λ2 − b2 λ2
x2 y2 z2
F2 (x, y, z, µ) = 2 + 2 2
+ 2 −1=0 (2)
µ µ −b µ
F3 (x, y, z, ν) = z − νx = 0, (3)
where λ2 > b2 > µ2 , 2b being the distance between the foci.

For all values of λ, µ, and ν consistent with the inequality above written the
surfaces (1), (2), (3) intersect in real points and cut orthogonally.
λ, µ, and ν can be so chosen that the surfaces will intersect in any given point,
and therefore can be taken as a set of curvilinear coördinates, and Laplace’s
Equation can be expressed in terms of them by the aid of Formula [XV] Art. 1.
From (1), (2), and (3) we readily get
λ2 µ2

2
x = 2


b (1 + ν 2 )





2 2 2 2 
2 (λ − b )(b − µ )
y = (4)
b2 

λ2 µ2 ν 2



z2 = 2 ;


b (1 + ν 2 )

r
µ λ b2 − µ2 µν
whence Dλ x = √ , Dλ y = 2 − b2
, Dλ z = √ ;
b 1+ν 2 b λ b 1 + ν2
1 λ2 − µ2
and 2 = 2 (5)
h1 λ − b2
[v. 130 (2)]. In like manner we get
1 λ2 − µ2
2 = 2 (6)
h2 b − µ2
1 λ2 µ2
and 2 = 2 , (7)
h3 b (1 + ν 2 )2
and [XV] Art. 1 becomes

µ p
p Dλ [λ λ2 − b2 .Dλ V ]
b(1 + ν 2 ) b2 − µ2
λ p
+ √ Dµ [µ b2 − µ2 .Dµ V ]
b(1 + ν 2 ) λ2 − b2
b(λ2 − µ2 )
+ p Dν [(1 + ν 2 )Dν V ] = 0, (8)
λµ (λ2 − b2 )(b2 − µ2 )
which is Laplace’s Equation in terms of our Spheroidal Coördinates λ, µ, and

ν.
If now in place of λ, µ, and ν we can introduce some function of λ, some
function of µ, and some function of ν which, therefore, will represent the same
set of orthogonal surfaces, and if we can choose these functions α, β, and γ,
which of course are functions of x, y, and z, so that ∇2 α = 0, ∇2 β = 0, and
∇2 γ = 0, equation (8) must reduce to the simple and symmetrical form given
in [XVI] Art. 1.
These functions α, β, and γ are easily found. Equation (8) is ∇2 V = 0
expressed in terms of λ, µ, and ν. Assume that V is a function of λ only; then
Dµ V = 0, and Dν V = 0, and (8) reduces to
p
Dλ [λ λ2 − b2 .Dλ V ] = 0
p dV
whence λ λ2 − b2 = c1 ,
dλ
c1 dλ
dV = √ ,
λ λ2 − b2
c1 λ
and V = sec−1 ,
b b
and is a function of λ which satisfies Laplace’s Equation.
Take this as α leaving c1 at present undetermined, so that
c1 dλ c1 λ
dα = √ and α = sec−1 .
λ λ2 − b2 b b
In the same way we get

c dµ c2 µ
dβ = p2 and β = sech−1 ,
µ b2 − µ2 b b
(v. Int. Cal. Art. 46, Ex.)
c3 dν
dγ = , and γ = c3 tan−1 ν.
1 + ν2
Substituting these values in (8) and taking c1 = −c2 = b, and c3 = 1, (8)

reduces at once to
Dα2 V Dβ2 V λ2 − µ2 2
+ + Dγ V = 0, (9)
λ2 µ2 λ2 µ2
or since λ = b sec α, µ = b sech β, and ν = tan γ, (10)
to cos2 αDα2 V + cosh2 βDβ2 V + (cosh2 β − cos2 α)Dγ2 V = 0 (11)
which is Laplace’s Equation in terms of what we may call Normal Oblate

Spheroidal Coördinates.
In using (11) it is to be noted that the point whose coördinates are (α, β, γ)
is the point of intersection of an oblate spheroid whose semi-axes are b sec α and
b tan α, an unparted hyperboloid of revolution whose semi-axes are b sech β and
b tanh β, and a plane containing the axis of the system and making the angle γ
with a fixed plane; and that if the axis of revolution is the axis of Y and the
fixed plane is the plane of XY , the rectangular coördinates of (α, β, γ) are
x = b sec α sech β cos γ, y = b tan α tanh β, z = b sec α sech β sin γ (12)
[v. (4)].
π
If now we let α range from 0 to , β from −∞ to ∞, and γ from 0 to 2π, we
2
shall be able to represent all points in space; and if we agree that negative values
of β shall belong to points below a plane through the origin and perpendicular
to the axis of revolution and positive values of β to points above that plane,
not only shall we have no ambiguity, but also the rectangular coördinates of any
point as given in (12) will have their proper signs.
EXAMPLES.
1. If the spheroid is a prolate spheroid, the ellipse and confocal hyperbola

must be revolved about the major axis of the ellipse, and the plane must contain
that axis. In place of equations (1), (2), and (3) of Art. 132 we have, then,
x2 y2 z2
+ + −1=0
λ2 λ2 − b2 λ2 − b2
x2 y2 z2
2
+ 2 2
+ 2 −1=0
µ µ −b µ − b2
z − νy = 0
where λ2 > b2 > µ2 .
λ2 − b2 b2 − µ2 b2 (1 + ν 2 )2
h21 = , h22 = , h23 = .
λ2 − µ2 λ2 − µ2 (λ2 − b2 )(b2 − µ2 )
Laplace’s Equation becomes

1 1
Dλ [(λ2 − b2 )Dλ V ] + 2 Dµ [(b2 − µ2 )Dµ V ]
b2 (1 + ν 2 ) b (1 + ν 2 )
λ2 − µ2
+ 2 Dν [(1 + ν 2 )Dν V ] = 0. (1)
(λ − b2 )(b2 − µ2 )
Dα2 V Dβ2 V λ2 − µ2
(1) reduces to + 2 + 2 D2 V = 0, (2)
2
λ −b 2 b −µ 2 (λ − b2 )(b2 − µ2 ) γ
bdλ bdµ dν
where dα = − , dβ = , dγ = ,
λ 2 − b2 b2 − µ2 1 + ν2
λ µ
α = ctnh−1 , β = tanh−1 , and γ = tan−1 ν.
b b
Since λ = b ctnh α, µ = b tanh β, and ν = tan γ
(2) can be reduced to
sinh2 αDα2 V + cosh2 βDβ2 V + (sinh2 α + cosh2 β)Dγ2 V = 0. (3)
In using (3) it is to be noted that the point (α, β, γ) is the point of intersec-
tion of a prolate spheroid whose semi-axes are b ctnh α and b csch α, a biparted
hyperboloid of revolution whose semi-axes are b tanh β and b sech β, and a plane
containing the axis of revolution and making the angle γ with a fixed plane.
If the fixed plane is that of (XY ) the rectangular coördinates of any point
(α, β, γ) are
x = b ctnh α tanh β, y = b csch α sech β cos γ, z = b csch α sech β sin γ,
and α may range from ∞ to 0, β from −∞ to ∞, and γ from 0 to 2π. Negative
values of β are to be taken for points lying to the left of a plane through the
origin perpendicular to the axis of revolution.
2. Transform Laplace’s Equation in Spherical Coördinates [XIII] Art. 1 to

the symmetrical form
α2 Dα2 V + cosh2 βDβ2 V + cosh2 βDγ2 V = 0
1 θ
where α=, β = log tan , and γ = φ.
r 2
3. Transform Laplace’s Equation in Cylindrical Coördinates [XIV] Art. 1 to
the symmetrical form
Dα2 V + Dβ2 V + e2α Dγ2 V = 0
where α = log r, β = φ, and γ = z.

133. In each of the cases we have considered, it has been easy to pass
from Laplace’s Equation in terms of the chosen coördinates representing an
orthogonal system of surfaces to the symmetrical form [XVI] Art. 1; and it is
evident that our new coördinate α is a value of V corresponding to such a
distribution that the surfaces obtained by giving particular values to ρ1 are
equipotential surfaces; that β is a value of V corresponding to such a distribution
that the surfaces obtained by giving particular values to ρ2 are equipotential
surfaces; and that γ is a value of V corresponding to such a distribution that the
surfaces obtained by giving particular values to ρ3 are equipotential surfaces.
α, β, and γ are called by Lamé “thermometric parameters.”
The condition that these values should exist, for a given system of surfaces,
that is, that the distribution described above should be possible, is readily
obtained. We shall work it out for α. It is merely the condition that V in
Laplace’s Equation may be a function of ρ1 alone.
If V is a function of ρ1 alone
dV dV dV
Dx V = Dx ρ1 , Dy V = D y ρ1 , Dz V = Dz ρ1 ,
dρ1 dρ1 dρ1
d2 V dV 2
Dx2 V = 2
2 (Dx ρ1 ) + dρ Dx ρ1
dρ1 1
2
d V dV 2
Dy2 V = (Dy ρ1 )2 + D ρ1
dρ1 2 dρ1 y
d2 V dV 2
Dz2 V = (Dz ρ1 )2 + D ρ1 .
dρ1 2 dρ1 z
d2 V dV
Therefore [(Dx ρ1 )2 + (Dy ρ1 )2 + (Dz ρ1 )2 ] 2
+ [Dx2 ρ1 + Dy2 ρ1 + Dz2 ρ1 ] =0
dρ1 dρ1
Dx2 ρ1 + Dy2 ρ1 + Dz2 ρ1 d2 V dV
whence 2 2 2
=− ÷ .
(Dx ρ1 ) + (Dy ρ1 ) + (Dz ρ1 ) dρ1 2 dρ1
∇2 ρ 1
or = F1 (ρ1 )
h21
where F1 (ρ1 ) may be any function of ρ1 alone. Our required conditions are then
∇2 ρ1

= F 1 (ρ 1 ) 

h21




2

∇ ρ2 
= F 2 (ρ 2 ) (1)
h22 


∇2 ρ3



2 = F 3 (ρ 3 ) 

h3
and when they are fulfilled the original curvilinear coördinates ρ1 , ρ2 , ρ3 , corre-
spond to possible equipotential or isothermal surfaces, thermometric parameters
α, β, and γ exist, and the reduction of Laplace’s Equation to the symmetrical

form [XVI] Art. 1 is possible.
134. Returning to our Oblate Spheroid problem of Art. 132 we can pro-
ceed as usual to break up our equation (11) Art. 132.
Assume that V = L.M.N , where L is a function of α only, M of β only, and
N of γ only. (11) Art. 132 becomes
cos2 α d2 L cosh2 β d2 M [cosh2 β − cos2 α] d2 N
+ + =0
L dα2 M dβ 2 N dγ 2
1 cos2 α d2 L 1 cosh2 β d2 M 1 d2 N
or 2 + 2 = − .
L cosh β − cos2 α dα2 M cosh β − cos2 α dβ 2 N dγ 2
The first member is independent of γ, and the second member is independent
of α and β, and the two members are identically equal. The second member
is then independent of α, β, and γ and must be constant; call it n2 . We have,
then,
d2 N
+ n2 N = 0 (1)
dγ 2
cos2 α d2 L cosh2 β d2 M
and + − n2 (cosh2 β − cos2 α) = 0. (2)
L dα2 M dβ 2
(1) gives us N = A cos nγ + B sin nγ. (3)
(2) can be written
cos2 α d2 L 2 2 2 2 cosh2 β d2 M
+ n cos α = n cosh β − = m(m + 1),
L dα2 M dβ 2
d2 L
whence cos2 α 2 + [n2 cos2 α − m(m + 1)]L = 0 (4)
dα
d2 M
and cosh2 β + [m(m + 1) − n2 cosh2 β]M = 0. (5)
dβ 2
If we introduce x = tanh β in (5) it becomes
2
n2

2 d M dM
(1 − x ) 2 − 2x + m(m + 1) − M =0 (6)
dx dx 1 − x2
where since x = tanh β and β may have any value from −∞ to ∞, x may have
any value between −1 and 1. (6) is a familiar equation having for a particular
solution n
n d Pm (x)
M = (1 − x2 ) 2 = Pmn n
(x) = Pm (tanh β). (7)
dxn
(v. Arts. 101 and 102).
If we introduce in (4) x = tan α it reduces to
d2 L n2

dL
(1 + x2 ) 2 + 2x + − m(m + 1) L = 0. (8)
dx dx 1 + x2
(8) is an unfamiliar equation, but it can be treated as (6) was treated if we take
the pains to go back to the beginning and follow the steps of the treatment of
Legendre’s Equation.
y
This labor can be saved, however, by noting that if we let x = (8) becomes
i
d2 L n2

dL
(1 − y 2 ) 2 − 2y + m(m + 1) − L=0
dy dy 1 − y2
and is identical in form with (6). Hence
n n dn Qm (y)
L = Pm (y) and L = (1 − y 2 ) 2 (v. Art. 101),
dy n
where y = i tan α, are particular solutions of (4).

We can avoid imaginaries if we use the values
n dn Qm (y)
L = (−i)m−n Pm
n
(y) and L = im+n+1 (1 − y 2 ) 2 . (9)
dy n
Since we assumed V = L.M.N we have
n
(tanh β)(−i)m−n Pm
n 
V = (A cos nγ + B sin nγ)Pm (i tan α) 
n dn Qm (i tan α) (10)
and V = (A cos nγ + B sin nγ)Pm (tanh β)im+n+1 secn α 
(d(i tan α))n
as particular solutions of (11) Art. 132.
If the problem is symmetrical with respect to the axis of the spheroid Dγ2 V =
0, n2 = 0 and our particular solutions (10) reduce to
)
V = (−i)m Pm (i tan α)Pm (tanh β)
(11)
and V = im+1 Qm (i tan α)Pm (tanh β).
If, then, V is given on the surface of a spheroid as a function of β and γ, we

must express it as a function of tanh β and γ, and shall be obliged to develop it
in terms of Spherical Harmonics of tanh β and γ by the formulas of Chapter VII,
using the first equation in (10) for the value of V at an internal point, and the
second for the value of V at an external point. If the problem is symmetrical,
we must develop in Zonal Harmonics of tanh β by the formulas of Chapter VI.
A convenient form for Qm (i tan α) is obtained from (2) Art. 100; it is
w∞ dx
Qm (i tan α) = −iPm (i tan α) . (12)
tan α
(1 + x2 )[Pm (xi)]2
w∞ dx π
Hence Q0 (i tan α) = −i = −i − α . (13)
tan α
1 + x2 2
EXAMPLES.
1. A conductor in the form of an oblate spheroid whose semi-axes are

b sec α0 and b tan α0 is charged with electricity and is found to be at poten-
tial V0 ; find the value of the potential function at any internal or external point.
Here V0 = V0 P0 (tanh β). Hence at an internal point
P0 (i tan α)
V = V0 P0 (tanh β) = V0 , (1)
P0 (i tan α0 )
and at an external point
π
Q0 (i tan α) −α
V = V0 P0 (tanh β) = V0 π2 . (2)
Q0 (i tan α0 ) − α0
2
Since V in (2) involves α only, the equipotential surfaces are all spheroids con-
focal with the conductor.
2. The upper half of an oblate spheroid whose semi-axes are b sec α0 and
b tan α0 is kept at the temperature unity, and the lower half at the temperature
zero. Find the permanent temperature at any internal point.
1 3 P1 (i tan α) 7 1 P3 (i tan α)
Ans. u= + P1 (tanh β) − . P3 (tanh β) + · · ·
2 4 P1 (i tan α0 ) 8 2 P3 (i tan α0 )
(v. Art. 93). u may be expressed in terms of x, y, and z without serious difficulty
[v. (12) Art. 132].
1 3 y 7 1 1 [25y 3 − 15y(x2 + y 2 + z 2 − b2 ) − 9b2 y]
u= + − . . + ···
2 4c 8 2 2 5c3 + 3b2 c
if 2c = 2b tan α0 = minor axis of spheroid.
135. Let us now find the potential function at an external point due to the
attraction of a solid homogeneous oblate spheroid, using the method employed
in Arts. 98 and 99.
Consider first the potential function due to a shell bounded by the spheroids
for which α = φ and α = φ + dφ.
By (1) Art. 98 we have
4πρκ = [Dn V1 − Dn V2 ]α=φ , (1)
where ρ is the density and κ the thickness of the shell, V1 the value of the
potential function at an internal point, and V2 the value of the potential function
at an external point.
X
Let V1 = Am (−i)m Pm (i tan α)Pm (tanh β)
X
and V2 = Bm im+1 Qm (i tan α)Pm (tanh β) [v. (11) Art. 134].
Since V1 and V2 must have the same value when α = φ
Qm (i tan φ) w∞ dx
Am = Bm i2m+1 = (−1)m Bm (2)
Pm (i tan φ) (1 + x )[Pm (xi)]2
2
tan φ
[v. (12) Art. 134].

w∞

X dx
im Bm Pm (tanh β)Pm (i tan α)

Hence V1 =


(1 + x2 )[Pm (xi)]2



tan φ
(3)
X w∞ dx 
m

and V2 = i Bm Pm (tanh β)Pm (i tan α) 2 2
.

(1 + x )[Pm (xi)] 

tan α
Dn V1 = Dα V1 .Dn α. Dn V2 = Dα V2 .Dn α
[Dn V1 − Dn V2 ]α=φ = [Dα V1 − Dα V2 ]α=φ (Dn α)α=φ

= [Dα (V1 − V2 )]α=φ [Dn α]α=φ .
X wα
tan
dx
V1 − V2 = im Bm Pm (tanh β)Pm (i tan α) .
(1 + x2 )[P m (xi)]
2
tan φ
sec2 α
X
m
Dα (V1 − V2 ) = i Bm Pm (tanh β) Pm (i tan α) 2
(1 + tan α)[Pm (i tan α)]2
dPm (i tan α) w
tan α
dx
+ .
dα (1 + x2 )[Pm (xi)]2
tan φ
X Pm (tanh β)
Dα [V1 − V2 ]α=φ = im Bm .
Pm (i tan φ)
dα
Dn α =
dn
p
dρ1 dλ λ2 − µ2
q
dn = = = √ dλ = b sec α tan2 α + tanh2 β.dα (4)
h1 h1 λ2 − b2
v. Art. 130 (3), and Art. 132 (5) and (10).
1
[Dn α]α=φ = p .
b sec φ tan2 φ + tanh2 β
Hence
1 X Pm (tanh β)
[Dn V1 − Dn V2 ]α=φ = p im Bm .
2 2
b sec φ tan φ + tanh β Pm (i tan φ)
q
κ = [dn]α=φ = b sec φ tan2 φ + tanh2 β.dφ
by (4), and (1) may be written

X Pm (tanh β)
4πρb2 sec2 φ(tan2 φ + tanh2 β)dφ = im Bm . (5)
Pm (i tan φ)
Since tanh2 β = 31 P0 (tanh β) + 23 P2 (tanh β)
by (5) Art. 95, to satisfy (5) we must give m the values 0 and 2 and
B0 = 43 πρb2 sec2 φ(3 tan2 φ + 1)dφ

and B2 = 43 πρb2 sec2 φ(3 tan2 φ + 1)dφ.
So that by (3)
w∞
4 2 2 2 dx
V1 = 3 πρb sec φ(3 tan φ + 1)dφ
1 + x2
tan φ
w∞ dx

− P2 (tanh β)P2 (i tan α) (6)
(1 + x2 )[P2 (xi)]2
tan φ
and V2 = 34 πρb2 sec2 φ(3 tan2 φ + 1)dφ[iQ0 (i tan α)

+ i3 P2 (tanh β)Q2 (i tan α)]. (7)
The potential function at an external point due to the solid spheroid for which
α = α0 is
w 0
φ=α
V = V2 = 34 πρb2 sec2 α0 tan α0 [iQ0 (i tan α) + i3 P2 (tanh β)Q2 (i tan α)]. (8)
φ=0
If 2a is the major axis and 2c the minor axis of the spheroid
2 πρa2 c M
4
3 πρb sec2 α0 tan α0 = 4
3 =
b b
where M is the mass of the spheroid. Therefore
M
V = [iQ0 (i tan α) + i3 P2 (tanh β)Q2 (i tan α)] (9)
b
is the required value. (9) can be reduced to

M π 1 h π
2
i
2
V = −α+ − α (3 tan α + 1) − 3 tan α [3 tanh β − 1] . (10)
b 2 4 2
EXAMPLES.
1. Break up the equation (3) Ex. 1, Art. 132, for the prolate spheroid, and
obtain particular solutions of the term
n n
V = (A cos nγ + B sin nγ)Pm (tanh β)Pm (ctnh α),
n dn Qm (ctnh α)
n
V = (A cos nγ + B sin nγ)Pm (tanh β)(−1) 2 cschn α .
(d ctnh α)n
2. Break up and solve the equations of Exs. 2 and 3, Art. 132, and show
that they lead to familiar forms.
3. If in Ex. 1, Art. 132, the conductor is a prolate spheroid whose semi-axes

are b ctnh α0 and b csch α0 show that
α
V = V0 at an internal point. V = V0 at an external point.
α0
4. Show that the potential function at an external point due to the attrac-
tion of a homogeneous solid prolate spheroid is
M
V = [Q0 (ctnh α) − P2 (tanh β)Q2 (ctnh α)].
b
Ellipsoidal Harmonics.
136. If we are dealing with an ellipsoid instead of a spheroid, we can take

as our orthogonal system of surfaces a set of confocal quadrics;
x2 y2 z2

+ + − 1 = 0 
λ2 λ2 − b2 λ2 − c2




2 2 2

x y z

+ + − 1 = 0 (1)
µ2 µ2 − b2 µ2 − c2 


x2 y2 z2


−

+ + 1 = 0 
ν2 ν 2 − b2 ν 2 − c2
where λ2 > c2 > µ2 > b2 > ν 2 . Here the first surface is an ellipsoid, the second
an unparted hyperboloid, and the third a biparted hyperboloid. Each of the
three principal sections of the system consists of confocal conics, and it is well
known and is easily shown that the surfaces cut orthogonally. λ, µ, and ν will
be our curvilinear coördinates, and are known as Ellipsoidal Coördinates.
We find without difficulty that
λ2 µ2 ν 2 (λ2 − b2 )(µ2 − b2 )(b2 − ν 2 )

x2 = , y2 = ,
b2 c 2 b2 (c2 − b2 )
(λ2 − c2 )(c2 − µ2 )(c2 − ν 2 )
z2 = , (2)
c2 (c2 − b2 )
(λ2 − b2 )(λ2 − c2 ) (µ2 − b2 )(c2 − µ2 )

h21 = , h22 = ,
(λ2 − µ2 )(λ2 − ν 2 ) (µ2 − ν 2 )(λ2 − µ2 )
(b2 − ν 2 )(c2 − ν 2 )
h23 = 2 . (3)
(λ − ν 2 )(µ2 − ν 2 )
p To avoid ambiguity, we shall suppose that of the nine semi-axes in (1)

c2 − µ2 is to be taken with the positive sign for a point on the half of the
unparted hyperboloid on which z is positive,
√ and with the negative sign for a
point on the half on which z is negative; b2 − ν 2 is to be taken with the positive
sign for a point on the half of the biparted hyperboloid on which y is positive,
and with the negative sign for a point on the half on which y is negative; ν is to
be taken positive for a point on the half of the biparted hyperboloid on which
x is positive, and negative for a point on the half on which x is negative, and
that the remaining six are to be always positive. It follows that our Ellipsoidal
Coördinates have the disadvantage that to fully fix a point we need
p to know not
merely the values of its coördinates λ, µ, and ν, but the signs of c2 − µ2 , and
√
b2 − ν 2 as well.
We shall see later, Art. 139, when we come to introduce what we may call
the Normal Ellipsoidal Coördinates α, β, and γ that they are free from this
disadvantage.
It is to be observed that λ may range from c to ∞, µ from b to c, and ν from
−b to b.
The element of length perpendicular to the Ellipsoid is
s
dλ (λ2 − µ2 )(λ2 − ν 2 )
dn = = .dλ. (4)
h1 (λ2 − b2 )(λ2 − c2 )
The element of Ellipsoidal surface is
s
dµdν (λ2 − µ2 )(λ2 − ν 2 )
dS = = (µ2 − ν 2 ) .dµdν, (5)
h2 h3 (µ2 − b2 )(c2 − µ2 )(b2 − ν 2 )(c2 − ν 2 )
and the element of volume is
dλdµdν
dv =
h1 h2 h3
(λ2 − µ2 )(λ2 − ν 2 )(µ2 − ν 2 )
=p dλdµdν. (6)
(λ2 − b2 )(λ2 − c2 )(µ2 − b2 )(c2 − µ2 )(b2 − ν 2 )(c2 − ν 2 )
The surface integral of any given function of µ and ν taken over the ellipsoid
is
w wb wc
f (µ, ν)dS = dν [f1 (µ, ν) + f2 (µ, ν) + f3 (µ, ν)
−b b
s
(λ2 − µ2 )(λ2 − ν 2 )
+ f4 (µ, ν)](µ2 − ν 2 ) .dµ, (7)
(µ2 − b2 )(c2 − µ2 )(b2 − ν 2 )(c2 − ν 2 )
where f1 (µ, ν), f2 (µ, ν), f3 (µ, ν) and f4 (µ, ν) are the values of the given function
on the four quarters of the ellipsoid into which it is divided by the planes of
(XY ) and (XZ).
Laplace’s Equation proves reducible to
(µ2 − ν 2 )Dα2 V + (λ2 − ν 2 )Dβ2 V + (λ2 − µ2 )Dγ2 V = 0 (8)
wλ dλ wµ dµ
where α=c p , β=c p ,
c
(λ2 − b2 )(λ2 − c2 ) b
(c2 − µ2 )(µ2 − b2 )
wν dν
γ=c p . (9)
0
(b2 − ν 2 )(c2 − ν 2 )
α, β, and γ can be expressed as Elliptic Integrals of the first class and are
v
b2
u
u1 − 2
r u
b π b 2
c b µ
− F , sin−1 1 − 2 , sin−1 u
u
α=F , , β=F ,
c 2 c λ c t b2
1− 2
c
b
−1 ν

γ = F , sin ; (10)
c b
c b dn α b
whence λ= mod =c mod ,
sn(K − α) c cn α c
b b2 12 b
µ= mod 1 − 2 , ν = b sn γ mod (11)
dn β c c
(v. Int. Cal. Arts. 179, 192, and 196).
137. If in (8) Art. 136 we assume V = L.M.N where L involves α only,

M involves β only, and N involves γ only, (8) can be written
µ2 − ν 2 d2 L λ2 − ν 2 d2 M λ2 − µ2 d2 N
+ + = 0. (1)
(1) is too complicated to be broken up by our usual method.

If, however, we let
1 d2 L P 1 d2 M 1 d2 N
= ak λk , = bk µk , = ck ν k ,
P P
substitute in (1) and make use of the fact that the result must be identically
zero, we find that the coefficients are zero for all values of k except k = 0 and
k = 2, and that a0 = −b0 = c0 , and a2 = −b2 = c2 .
Therefore (1) can be broken up into the three equations
d2 L
= (a0 + a2 λ2 )L
dα2
d2 M
= −(a0 + a2 µ2 )M
dβ 2
d2 N
= (a0 + a2 ν 2 )N.
dγ 2
We shall find it convenient to take a2 as m(m + 1) and a0 as −(b2 + c2 )p;

whence
d2 L

2 2 2
− [m(m + 1)λ − (b + c )p]L = 0


dα2





2

d M 2 2 2

+ [m(m + 1)µ − (b + c )p]M = 0 (2)
dβ 2 



d2 N


2 2 2
− −

[m(m + 1)ν (b + c )p]N = 0. 

dγ 2
If now in (2) we replace α, β, and γ by their values in terms of λ, µ, and ν,
we get
d2 L dL

(λ2 − b2 )(λ2 − c2 ) 2 + λ(λ2 − b2 + λ2 − c2 ) 

dλ dλ 


2 2 2

−[m(m + 1)λ − (b + c )p]L = 0 




2
d M dM


2 2 2 2 2 2 2 2
(µ − b )(µ − c ) 2 + µ(µ − b + µ − c )


dµ dµ (3)
−[m(m + 1)µ2 − (b2 + c2 )p]M = 0 





2
d N dN


2 2 2 2 2 2 2 2
(ν − b )(ν − c ) 2 + ν(ν − b + ν − c )



dν dν 


2 2 2 
−[m(m + 1)ν − (b + c )p]N = 0.
p p p
Whence if L = Em (λ), it follows that M = Em (µ) and N = Em (ν), and that
p p p
V = Em (λ)Em (µ)Em (ν) (4)
is a solution of Laplace’s Equation, (8) Art. 136.

The equation
d2 z dz
(x2 − b2 )(x2 − c2 ) 2
+ x(x2 − b2 + x2 − c2 )
dx dx
− [m(m + 1)x2 − (b2 + c2 )p]z = 0 (5)
p
is known as Lamé’s Equation, and Em (x) as a Lamé’s Function or an Ellipsoidal
Harmonic. We shall suppose m a positive integer.
ak xk . Substitute in (5) and

P
To get a particular solution of (5) let z =
reduce and we get
[k(k + 1) − m(m + 1)]ak − (b2 + c2 )[(k + 2)2 − p]ak+2

+ b2 c2 (k + 3)(k + 4)ak+4 = 0. (6)
We have now only to choose a sequence of coefficients satisfying (6), and we
may take any two consecutive coefficients arbitrarily.
(6) which is ordinarily a relation connecting three consecutive coefficients
reduces to a relation between two when k = m, when k = −3, and when
k = −4. If we take am+2 = 0, am+4 , am+6 , &c., will vanish. Let am = 1. If m
is even the coefficient of a0 in (6) will be zero; if p has such a value that a−2
is zero, a−4 , a−6 , &c., will be zero, and there will be no terms in the solution
involving negative powers of x.
If we write the values of am−2 , am−4 , &c., by the aid of (6) we see that am−2
is of the first degree in p, am−4 of the second degree in p, &c., and a−2 of the
m m
degree + 1 in p. There are then + 1 values of p which we shall call p1 ,
2 2
p2 , p3 , &c., for which a−2 will vanish, and for which our solutions will be of the
form
p
Em (x) = xm + am−2 xm−2 + am−4 xm−4 + · · · + a0
if m is even.
If m is odd, the coefficient of a1 in (6) will vanish and we can choose p so
that a−1 shall be zero, and then all coefficients of lower order will vanish. a−1
m+1 m+1
is of the degree in p, and there will be values p1 , p2 , p3 , &c., of p
2 2
for which
p
Em (x) = xm + am−2 xm−2 + am−4 xm−4 + · · · + a1 x.
p
Following Heine we shall call the solution just obtained Km (x) so that
p
Km (x) = xm + am−2 xm−2 + am−4 xm−4 + · · · (7)
terminating with a0 if m is even, and with a1 x if m is odd. If m is even, there
m p1 p2 m+1
are + 1 of these functions Km (x), Km (x), &c., and there are of them
2 2
if m is odd. The coefficients can be computed√ by the aid of (6).
If in Lamé’s Equation (5) we let z = v x2 − b2 we get the equation
d2 v dv
(x2 − b2 )(x2 − c2 ) + x[x2 − b2 + 3(x2 − c2 )]
dx2 dx
− [(m + 2)(m − 1)x2 + c2 − (b2 + c2 )p]v = 0. (8)
ak xk we obtain the relation

P
Letting v =
[k(k + 3) − (m + 2)(m − 1)]ak − {(b2 + c2 )[(k + 2)2 − p] + c2 (2k + 5)}ak+2

+ b2 c2 (k + 3)(k + 4)ak+4 = 0. (9)
m
Proceeding exactly as before, we find that there are values q1 , q2 , q3 , &c.,
2
m+1
of p for which v = xm−1 + am−3 xm−3 + · · · + a1 x if m is even, and values
2
m−1
for which v = √x + am−3 xm−3 + · · · + a0 if m is odd.
Calling v x2 − b2 Lpm (x) so that
p
Lpm (x) = x2 − b2 [xm−1 + am−3 xm−3 + am−5 xm−5 + · · · ], (10)
m
terminating with a1 x if m is even and with a0 if m is odd, we have values
2
p m+1
of Em (x), namely Lqm1 (x), Lqm2 (x), &c., of the form (10) if m is even and
2
values if m is odd.
By interchanging b and c in (8), (9), and (10) we may show that if
p
p
Mm (x) = x2 − c2 [xm−1 + am−3 xm−3 + am−5 xm−5 + · · · ] (11)
m p r1 r2 r3
there are values of Em (x), namely Mm (x), Mm (x), Mm (x), &c., of the form
2
m+1
(11) if m is even and values if m is odd.
2 p
Finally if in Lamé’s Equation (5) we let z = v (x2 − b2 )(x2 − c2 ) we get
d2 v dv
(x2 − b2 )(x2 − c2 ) + 3x(x2 − b2 + x2 − c2 )
dx2 dx
− [(m + 3)(m − 2)x2 − (b2 + c2 )(p − 1)]v = 0. (12)
ak xk we obtain the relation

P
If now we let v =
[k(k + 5) − (m − 2)(m + 3)]ak

− (b2 + c2 )[(k + 2)(k + 4) + 1 − p]ak+2 + b2 c2 (k + 3)(k + 4)ak+4 = 0. (13)
m
Proceeding as before we find that there are values s1 , s2 , s3 , &c., of p for
2
m+1
which v = xm−2 + am−4 xm−4 + am−6 xm−6 + · · · + a0 if m is even, and
2
m−2 m−4
values for which
p v = x + a m−4 x + · · · + a1 x if m is odd.
p
Calling v (x2 − b2 )(x2 − c2 ) Nm (x) so that
p
p
Nm (x) = (x2 − b2 )(x2 − c2 )[xm−2 + am−4 xm−4 + am−6 xm−6 + · · · ] (14)
m
terminating with a0 if m is even and with a1 x if m is odd, we have values of
p s1 s2 s3
2
Em (x), namely Nm (x), Nm (x), Nm (x), &c., of the form (14) if m is even and
m−1
values if m is odd.
2
Summing up our results we see that there are 2m + 1 Ellipsoidal Harmonics
√
p 2 2
Em (x) each √
of which is a finite sum√ √ in x, or in x and x − b ,
of the mth degree
or in x and x2 − c2 , or in x and x2 − b2 and x2 − c2 .
It was proved by Lamé that the 2m + 1 values of p, namely p1 , p2 , p3 , &c.,

q1 , q2 , q3 , &c., r1 , r2 , r3 , &c., s1 , s2 , s3 , &c., were all real, and by Liouville that
they were all different.
We give tables of the Ellipsoidal Harmonics for m = 0, m = 1, m = 2, and
m = 3. The coefficients were obtained by the aid of formulas (6), (9), and (13).
E0 (x) E1 (x)
K0 (x) = 1 K1 (x) = x
√
L0 (x) = 0 L1 (x) = x2 − b2
√
M0 (x)= 0 M1 (x)= x2 − c2
N0 (x) = 0 N1 (x) = 0
E2 (x)
K2p1 (x)= x2 − 31 [b2 + c2 −

p
(b2 + c2 )2 − 3b2 c2 ]
K2p2 (x)= x2 − 31 [b2 + c2 + (b2 + c2 )2 − 3b2 c2 ]
p
√
L2 (x) = x x2 − b2
√
M2 (x) = x x2 − c2
p
N2 (x) = (x2 − b2 )(x2 − c2 )
E3 (x)
x
K3p1 (x) = x3 − [2(b2 + c2 ) − 4(b2 + c2 )2 − 15b2 c2 ]
p
5
p2 x p
K3 (x) = x3 − [2(b2 + c2 ) + 4(b2 + c2 )2 − 15b2 c2 ]
5
q1
√ p
L3 (x) = x − b2 [x2 − 15 (b2 + 2c2 − (b2 + 2c2 )2 − 5b2 c2 )]
2
√
Lq32 (x) = x2 − b2 [x2 − 15 (b2 + 2c2 + (b2 + 2c2 )2 − 5b2 c2 )]
p
√
M3r1 (x)= x2 − c2 [x2 − 15 (2b2 + c2 − (2b2 + c2 )2 − 5b2 c2 )]
p
√
M3r2 (x)= x2 − c2 [x2 − 15 (2b2 + c2 + (2b2 + c2 )2 − 5b2 c2 )]
p
p
N3 (x) = x (x2 − b2 )(x2 − c2 )
It is to be noted that since in the solution (4) of Laplace’s Equation,

p p p
V = Em (λ)Em (µ)Em (ν),
we have the same m and p in each of the three factors, we shall have to deal
merely with products made up of factors of the same form, for example,
pk pk pk
Km (λ)Km (µ)Km (ν), Lqmk (λ)Lqmk (µ)Lqmk (ν), &c.;
and that in a solution of the form
X
p p p
V = Am,p Em (λ)Em (µ)Em (ν)
we shall have for a given m just 2m + 1 terms.
138. From the particular solution of Lamé’s Equation [(5) Art. 137] z =
p
Em (x), we can get by formula (5), Art. 18, the general solution.
w∞ dx
p p
It is z = AEm (x) + BEm (x) p p . (1)
x
(x − b )(x2 − c2 )[Em
2 2 (x)]2
Making A = 0 and B = 2m + 1 we get a second form of particular solution of

p
Lamé’s Equation, z = Fm (x) where
w∞ dx
p p
Fm (x) = (2m + 1)Em (x) p p . (2)
x
(x − b )(x2 − c2 )[Em
2 2 (x)]2
p
We shall call Fm (x) a Lamé’s Function of the second kind.
It is easily seen to approach the value zero as x is indefinitely increased.
EXAMPLES.
1. If an ellipsoidal conductor is charged with electricity, and is found to be
at potential V0 , show that since V0 = V0 K0 (λ),
V = V0 K0 (λ)K0 (µ)K0 (ν) = V0
at an internal point, and
w∞ dx
V = V0 K0 (µ)K0 (ν) K0 (λ) p
λ
(x2 − b2 )(x2 − c2 )[K0 (x)]2
w∞ dx

÷ K0 (λ0 ) p
λ0
(x2 − b2 )(x2 − c2 )[K0 (x)]2
b c
w∞
dx w∞ dx F , sin−1

= V0 ÷ = V0 c λ ,
b −1 c
p p
(x2 − b2 )(x2 − c2 ) (x2 − b2 )(x2 − c2 )

λ λ0 F , sin
c λ0
b π
F , −α
whence V = V0 c 2 v. (10) Art. 136.
b π
F , − α0
c 2
2. Find the value of the potential function at an external point due to the
attraction of a solid homogeneous ellipsoid (v. Art. 135).
Observe that
(l2 − µ2 )(l2 − ν 2 ) = 13 [3l4 − 2(b2 + c2 )l2 + b2 c2 ]K0 (µ)K0 (ν)

b2 + c2 − 3l2

+ 21 1 + p K2p1 (µ)K2p1 (ν)
(b2 + c2 )2 − 3b2 c2
b2 + c2 − 3l2

1
+ 2 1− p K2p2 (µ)K2p2 (ν);
2 2 2
(b + c ) − 3b c 2 2
and that
wλ0 3l 4
− 2(b2 + c2 )l2 + b2 c2
q
4
3 πρ
p dl = 43 πρλ0 (λ20 − b2 )(λ20 − c2 ) = M
0
(l2 − b2 )(l2 − c2 )
where M is the mass of the ellipsoid.

w∞
dx 3
Ans. V =M p − p
λ
(x2 − b2 )(x2 − c2 ) 2 (b2 + c2 )2 − 3b2 c2
w∞ dx
K2p1 (µ)K2p1 (ν)K2p1 (λ) p
λ
(x2 − b2 )(x2 − c2 ).(K2p1 (x))2
w∞ dx

− K2p2 (µ)K2p2 (ν)K2p2 (λ) .
(x2 − b2 )(x2 − c2 ).(K2p2 (x))2
p
λ
b b2 21
139. If for the sake of brevity we represent by k, and 1 − 2 by k 0
c c
in the formulas (11) Art. 136 we have
dn α b
λ=c (modk), µ= , ν = b sn γ(modk) (1)
cn α dn β(modk 0 )
and from these we get without difficulty (v. Int. Cal. Art. 192)
ck 0 bk 0 sn β
p p 
λ2 − b2 = , µ2 − b2 = (modk 0 ), 

cn α(modk) dn β 



0
ck sn α
p p 
b2 − ν 2 = b cn γ(modk), 2 2
λ −c = (modk), (2)
cn α 

ck 0 cn β
p p 

(modk 0 ),

c2 − µ2 = 2 2
c − ν = c dn γ(modk).


dn β
If we let α range from 0 to K, and β from 0 to 2K 0 ,and

γ from 0to 4K,
π π
where K and K 0 are the complete Elliptic Integrals F k, and F k 0 ,
2 2
respectively, (α, β, γ) may represent any point in space, and there will be no
ambiguity in sign (v. Art. 136).
We may note that if 0 < β < K 0 , z is positive; if K 0 < β < 2K 0 , z is negative;
if 0 < γ < K, x and y are both positive; if K < γ < 2K, x is positive and y
negative; if 2K < γ < 3K, x and y are both negative; and if 3K < γ < 4K, x
is negative and y positive (v. Art. 136).
We can write the values in (4), (5), (6), and (7), Art. 136, more neatly by
bringing in α, β, and γ. We get
1p 2
dn = (λ − µ2 )(λ2 − ν 2 )dα, (3)
c
1 p
dS = 2 (µ2 − ν 2 ) (λ2 − µ2 )(λ2 − ν 2 )dβdγ, (4)
c
1
dv = 3 (λ2 − µ2 )(λ2 − ν 2 )(µ2 − ν 2 )dαdβdγ. (5)
c
For the integral of any function of α, β, and γ over the ellipsoid α = α0 , we
shall have
w 2K 0
1 w w
4K
p
F (α, β, γ)dS = 2 dβ F (α0 , β, γ)(µ2 − ν 2 ) (λ2 − µ2 )(λ2 − ν 2 )dγ. (6)
c
0 0
140. If we make use of the formula (2) Art. 92

w
(U Dn V − V Dn U )dS = 0 (1)
and take as our closed surface any given ellipsoid, we can get a very important
result.
p p p
If U = Em (λ)Em (µ)Em (ν) and V = Enq (λ)Enq (µ)Enq (ν)
then ∇2 U = ∇2 V = 0.
p
p p dEm (λ) c
Dn U = Dα U Dn α = Em (µ)Em (ν) p ,
dα (λ − µ )(λ2 − ν 2 )
2 2
dEnq (λ) c
and Dn V = Dα V Dn α = Enq (µ)Enq (ν) p ,
dα (λ − µ )(λ2 − ν 2 )
2 2
p p
U Dn V − V D n U = Em (µ)Em (ν)Enq (µ)Enq (ν)
dEnq (λ) p

p q dEm (λ) c
Em (λ) − En (λ) p .
dα dα (λ − µ )(λ2 − ν 2 )
2 2
Integrating U Dn V − V Dn U over the whole ellipsoid, and writing the result

equal to zero, we have

0
1 w w
2K 4K
p p
dβ Em (µ)Em (ν)Enq (µ)Enq (ν)
c
0 0
dEnq (λ) p

p q dEm (λ)
Em (λ) − En (λ) (µ2 − ν 2 )dγ = 0.
dα dα
0
w
2K w
4K
p p
Hence dβ Em (µ)Em (ν)Enq (µ)Enq (ν)(µ2 − ν 2 )dγ = 0 (2)
0 0
dEnq (λ) dE p (λ)
unless p
Em (λ) − Enq (λ) m = 0. (3)
dα dα
But as our ellipsoid may be taken at pleasure, λ and α are unrestricted, and
if (3) is true it must be true identically.
p
If we divide (3) by [Em (λ)]2 it becomes
d Enq (λ) Enq (λ)

p = 0 and p = a constant;
dα Em (λ) Em (λ)
and this obviously cannot be true unless n = m and q = p.
EXAMPLES.
1. Show that it follows from (2) Art. 140 that

0
w
K wK
p p
dβ Em (µ)Em (ν)Enq (µ)Enq (ν)(µ2 − ν 2 )dγ = 0.
−K 0 −K
Suggestion:
0 0
w
2K w
K
p
Em (µ)Enq (µ)(µ2 2
− ν )dβ = p
Em (µ)Enq (µ)(µ2 − ν 2 )dβ
0 0
0
w
2K
p
+ Em (µ)Enq (µ)(µ2 − ν 2 )dβ.
K0
If in the last integral we replace β by β + 2K 0 it becomes
w0
p
± Em (µ)Enq (µ)(µ2 − ν 2 )dβ
−K 0
v. Arts. 136 and 139 and Int. Cal. Art. 196.

2. Show that
0 0
w
2K w
4K w
K wK
p p
dβ [Em (µ)Em (ν)]2 (µ2 2
− ν )dγ = 8 p
dβ [Em p
(µ)Em (ν)]2 (µ2 − ν 2 )dγ.
0 0 0 0
141. We can now solve the problem of finding the value of V at any point
in space when it is given at all the points on the surface of the ellipsoid α = α0 .
We have first to develop in Ellipsoidal Harmonics a function of µ and ν or
rather of α and β given at all points on the surface of the ellipsoid in question;
and this is now easily accomplished by our usual method, which leads us to the
result
m=∞
X k=2m+1
X
pk pk
f (α0 , β, γ) = Am,pk Em (µ)Em (ν), (1)
m=0 k=1
0
w
2K w
4K
pk pk
dβ f (α0 , β, γ)Em (µ)Em (ν)(µ2 − ν 2 )dγ
0 0
where Am,pk = . (2)
K0
w wK
pk pk
8 dβ [Em (µ)Em (ν)]2 (µ2 − ν 2 )dγ
0 0
Our final solution is

m=∞
X k=2m+1 pk
X Em (λ) pk pk
V = Am,pk pk E (µ)Em (ν) (3)
m=0
Em (λ0 ) m
k=1
at an internal point;
X k=2m+1
m=∞ pk
X Fm (λ) pk pk
V = Am,pk pk E (µ)Em (ν) (4)
m=0
Fm (λ0 ) m
k=1
at an external point.
Lamé has proved rather ingeniously that
0
w
K wK
pk pk
dβ [Em (µ)Em (ν)]2 (µ2 − ν 2 )dγ
0 0
π
can always be found and that it is equal to multiplied by a rational integral
2 b 2
pk
function of the coefficients of Em (x) and of c2 and .
c
Of course the labor of obtaining even a few terms of the development of a
function that is in the least complicated is enormous.
p
142. If in Laplace’s Equation (8) Art. 136 we let V = Em (λ)U supposing
1 d2 Em p
(λ)
U to be a function of β and γ only, we get after replacing p by
Em (λ) dα2
2 2 2
its value m(m + 1)λ − (b + c )p [v. (2) Art. 137]
(λ2 − ν 2 )Dβ2 U + (λ2 − µ2 )Dγ2 U + (µ2 − ν 2 )[m(m + 1)λ2 − (b2 + c2 )p]U = 0; (1)
and since by hypothesis U is independent of λ, the coefficient of λ2 in (1) must
vanish. Hence
Dβ2 U + Dγ2 U + (µ2 − ν 2 )m(m + 1)U = 0. (2)
p p
Of course U = Em (µ)Em (ν) will satisfy (2).
EXAMPLES.
p p
1. Substitute U = Em (µ)Em (ν) in (2) Art. 142 and by the aid of (2)
Art. 137 show that the equation (2) Art. 142 is satisfied.
2. Obtain (2) Art. 140 directly from (2) Art. 142.
3. Conical Coördinates. Consider the system of coördinates defined by the

equations
x2 + y 2 + z 2 = r 2




2 2 2
x y z



+ + = 0 (1)
µ2 µ2 − b2 µ2 − c2 
2 2 2 
x y z 

+ 2 + 2 =0 
ν2 ν − b2 ν − c2
where c2 > µ2 > b2 > ν 2 .
Show that
r2 µ2 ν 2 r2 (µ2 − b2 )(ν 2 − b2 ) r2 (µ2 − c2 )(ν 2 − c2 )
x2 = , y2 = , z2 = ;
b2 c 2 b2 (b2 − c2 ) c2 (c2 − b2 )
(µ2 − b2 )(c2 − µ2 ) (b2 − ν 2 )(c2 − ν 2 )

h21 = , h22 = , h23 = 1.
r2 (µ2 − ν 2 ) r2 (µ2 − ν 2 )
Laplace’s Equation is
Dα2 V + Dβ2 V + (µ2 − ν 2 )Dr (r2 Dr V ) = 0 (2)
wµ dµ wν dν
where α= p and β = p ,
b
(µ2 − b2 )(c2 − µ2 ) 0
(b − ν 2 )(c2 − ν 2 )
2
If V = U.R (2) breaks up into

d 2 dR
r = m(m + 1)R, (3)
dr dr
Dα2 U + Dβ2 U + m(m + 1)(µ2 − ν 2 )U = 0. (4)
(3) gives R = Arm + Br−m−1 .

p p
(4) gives U = Em (µ)Em (ν) (v. Art. 142).
So that a solution of (2) is
V = Arm Em
p p
(µ)Em (ν).
But since (2) is Laplace’s Equation, V = Arm Ym (µ, φ), if expressed in Con-
p p
ical Coördinates, must satisfy it, consequently Em (µ)Em (ν) must be simply a
Spherical Harmonic of the mth degree.
Toroidal Coördinates.
143. Any pair of circles belonging to the orthogonal system obtained and
figured in Art. 46 can be represented by the equations
x2 + y 2 + a2 

2ax
= 
sinh α cosh α

(1)
2ay x2 + y 2 − a2 
= 

sin β cos β
if we take 2a instead of 2 as the distance between the points common to the
second set of circles.
If we rotate the system about the axis of y we get a set of spheres and a set of
anchor rings which cut orthogonally. These and a set of planes through the axis
of revolution will form an orthogonal system of surfaces, and the parameters
corresponding to them may be taken as a set of curvilinear coördinates and
may be called Toroidal Coördinates.
If we take the axis of the system as the axis of Z, the equations of a set of
the surfaces may be written
4a2 (x2 + y 2 ) [x2 + y 2 + z 2 + a2 ]2 


= 
sinh2 α cosh2 α




2 2 2 2
2az x +y +z −a (2)
=
sin β cos β





y = x tan γ

α, β, and γ being regarded as the coördinates of a point of intersection of

the three surfaces.
Finding Laplace’s Equation in the usual manner we get

a sinh α cos γ a sinh α sin γ a sin β
x= , y= , z= ,
cosh α ∓ cos β cosh α ∓ cos β cosh α ∓ cos β
p a sinh α a cosh α
r= x2 + y 2 = , a + z ctn β = ;
cosh α ∓ cos β cosh α ∓ cos β
cosh α ∓ cos β cosh α ∓ cos β cosh α ∓ cos β
h1 = , h2 = , h3 = ;
a a a sinh α
and Laplace’s Equation becomes

a sinh α a sinh α
Dα Dα V + Dβ Dβ V
cosh α ∓ cos β cosh α ∓ cos β

a
+ Dγ Dγ V = 0, (1)
sinh α(cosh α ∓ cos β)
1
or Dα (rDα V ) + Dβ (rDβ V ) + rDγ2 V = 0. (2)
sinh2 α
We cannot proceed further by our usual method, for the assumption that
V is a function of α alone, or that V is a function of β alone, proves to be
inadmissible. Indeed, not only are α, β, and γ not thermometric parameters (v.
Art. 133), but no thermometric parameters exist, and no possible distribution
can make our anchor rings or our spheres a set of equipotential surfaces.
We can, however, simplify (2). It can be written
√ 1 √ √ √ √
Dα2 (V r) + Dβ2 (V Dγ2 (V r) − V (Dα2 r + Dβ2 r) = 0.
r) + (3)
sinh2 α
√
2
√ 2
√ r √
Dα r + Dβ r proves equal to − ; hence if U = V r (3) becomes
4 sinh2 α
sinh2 α(Dα2 U + Dβ2 U ) + Dγ2 U + 41 U = 0, (4)
for which particular solutions can readily be found by our usual process.
(4) can be broken up into the three equations
d2 N
+ (m + 12 )2 N = 0 (5)
dγ 2
d2 M
+ n2 M = 0 (6)
dβ 2
d2 L
sinh2 α 2 − [m(m + 1) + n2 sinh2 α]L = 0. (7)
dα
N = A cos(m + 21 )γ + B sin(m + 12 )γ
M = A1 cos nβ + B1 sin nβ.
If we introduce into (7) x = ctnh α it becomes
d2 L n2

dL
(1 − x2 ) 2 − 2x + m(m + 1) − L = 0,
dx dx 1 − x2
a solution of which is
n n dn Pm (x)
L = Pm (x) = (1 − x2 ) 2 (v. Art. 102).
dxn
It is to be noted that since ctnh α is greater than 1
n dn Pm (ctnh α)
n
Pm (ctnh α) = i 2 cschn α .
(d ctnh α)n
n
The constant coefficient i 2 can be rejected and we get
dn Pm (ctnh α)
U = [A cos(m+ 12 )γ+B sin(m+ 12 )γ](A1 cos nβ+B1 sin nβ) cschn α
(d ctnh α)n
as a particular solution of (4).

n
1 n n d Pm (ctnh α)
n Pm (ctnh α) = csch α
i2 (d ctnh α)n
has been called a Toroidal Harmonic.
EXAMPLES.
1. Given the value of the potential function at all points on the surface of
an anchor ring; find its value at any point within the ring.
Suggestion: If V = f (β, γ) when α = α0 , the function to be developed is
12
√

a sinh α0
r.f (β, γ) i.e. f (β, γ)
cosh α0 ∓ cos β
and the development will be in a double Fourier’s Series (v. Art. 71).
2. Show that if we let α range from 0 to ∞, β from −π to π, and γ from 0

to 2π, each of the double signs on page 264 may be replaced by the minus sign
without loss of generality.
267
CHAPTER IX.1
HISTORICAL SUMMARY.
The method of development in series which has enabled us in the preceding

chapters to solve problems in various branches of mathematical physics, had its
origin, as might have been expected, in the theory of the musical vibrations of a
stretched string. It was in the year 17532 that Daniel Bernoulli enunciated the
principle of the coexistence of small oscillations, which, in connection with Tay-
lor’s and John Bernoulli’s theory of the vibrating string, led him to believe that
the general solution of this problem could be put in the form of a trigonometric
series. This principle also led him and Euler to treat in a similar manner the
problems of the vibration of a column of air and of an elastic rod. The problem
of the vibration of a heavy string suspended from one end was also treated in
the same manner by these mathematicians and deserves special mention here
as in it Bessel’s functions of the zeroth order appear for the first time.3 In none
of these cases, however, was any method given for determining the coefficients
of the series.
This last remark also applies to the more complicated problems of the vi-
bration of rectangular and circular membranes, which were discussed by Euler4
in 1764, and in the last of which the general Bessel’s functions of integral orders
occur.
It is in problems connected with astronomy that the first completely suc-
cessful application of the method here considered occurs. Legendre in a paper
published in the Mémoires des Savants Étrangers for 1785, first introduced the
zonal harmonics Pm and applied them to the determination of the attraction
of solids of revolution. He was followed by Laplace, who in one of the most
remarkable memoirs ever written5 determined the potential of a solid differing
but little from a sphere by means of the development according to the spherical
harmonics Ym .
Very closely related to this problem is Gauss’s celebrated treatment of the
theory of terrestrial magnetism,6 which we will for that reason mention here,
although it was not published until more than half a century later. This paper
is particularly noteworthy as it contains a numerical application of the method
on a larger scale than has ever been attempted before or since.
After the researches of Legendre and Laplace there was a pause of a quarter
of a century until in 1812 Fourier’s extensive memoir: Théorie du mouvement
1 See preface.
2 See two articles by Bernoulli and one by Euler in the Memoirs of the Academy of Berlin
for this year.
3 See the Transactions of the Academy of St. Petersburg for 1732-33, 1734 and 1781.
4 Transactions of the Academy of St. Petersburg.
5 “Théorie des attractions des sphéroı̈des et de la figure des Planètes” Mémoires de
l’académie des sciences 1782. This article, although bearing an earlier date than that of Leg-
endre, was really inspired by it. It is here that “Laplace’s equation” first appears, occurring,
however, only in polar coördinates.
6 Resultate aus den Beobachtungen des magnetischen Vereins im Jahre 1838. Leipzig, 1839.
Reprinted in Gauss’s collected works, Vol. V., p. 121.

HISTORICAL SUMMARY. 268
de la chaleur dans les corps solides was crowned by the French Academy. Al-
though not printed until the years 1824-26,7 the manuscript of this work was
in the meantime accessible to the other French mathematicians presently to be
mentioned. The first part of this memoir, which was reproduced with but few
alterations in the Théorie analytique de la chaleur (1822), contains a treatment
of the following problems and of practically all of their special cases:
(a) The one dimensional flow of heat. (b) The two dimensional flow of
heat in a rectangle. (c) The three dimensional flow of heat in a rectangular
parallelopiped. (d ) The flow of heat in a sphere when the temperature depends
only on the distance from the centre. (e) The flow of heat in a right circular
cylinder when the temperature depends only on the distance from the axis. In
these problems not merely the simpler boundary conditions are considered but
also the question of radiation into an atmosphere. In special cases of the first
three problems just mentioned (when one or more dimensions become infinite)
the series degenerate into “Fourier’s integrals.”
More important even than any of these special problems is the great advance
which Fourier caused the theory of trigonometric series to make. In a posthu-
mous paper Euler had given the formulae for determining the coefficients,8 but
Fourier was the first to assert and to attempt to prove that any function, even
though for different values of the argument it is expressed by different analytical
formulae, can be developed in such a series. The fact that the real importance of
trigonometric series was thus for the first time shown justifies us in associating
Fourier’s name with them, although, as we have seen, they were known long
before his day.
Fourier’s results were extended by Laplace in 18209 to the general (unsym-
metrical) case of the flow of heat in a sphere, and by Poisson10 (1821) to the
unsymmetrical flow of heat in a cylinder.
In 1835 Green published a paper11 in which the method we are considering
is employed to determine the potential of a heterogeneous ellipsoid. This paper,
in which the analysis is performed at once for space of n dimensions, anticipates
much that was subsequently done by others, but has failed to exert an influence
proportional to its importance.
At about this time Lamé began a series of publications which have connected
his name inseparably with the problem of the permanent state of temperature
of an ellipsoid. In the first of these12 the equation ∇2 V = 0 is transformed
to ellipsoidal coördinates and is then broken up into three ordinary differential
7 Mémoires de l’académie des sciences for 1819-20 and 1821-22.
8 Lagrange had practically determined these coefficients long before but failed to notice
what he had got.
9 Connaissance des Temps pour l’an 1823.
10 Journal de l’École Polytechnique, 19e Cahier. Although the final forms to which Poisson
reduces his results are similar to Fourier’s, his methods are very different.
11 “On the determination of the exterior and interior attraction of ellipsoids of variable
densities.” Transactions of the Cambridge Philosophical Society.

12 Mémoires des Savants Étrangers, Vol. V. Although the volume is dated 1838 this paper
(which was reprinted in Liouville’s Journal, 1837) must have appeared at least as early as
1835.
equations. The rest of the solution, however, is hardly touched upon. Lamé’s
most important work on this subject13 was published in Liouville’s Journal in
1839, and in it the complete solution of the problem is given. Lamé clearly
shows in this paper how he arrived at his solution, by considering first the
simpler case of a sphere where, instead of the polar coördinates θ and φ, the
parameters of two families of confocal cones of the second degree are used as
coördinates. This system of curvilinear coördinates, which, when applied to
the complete sphere, merely gives the old results of Laplace in a new form, is
barely mentioned in Lamé’s later publications. In the same volume of Liouville’s
Journal Lamé published a second paper in which he applies his results to the
special cases of ellipsoids of revolution.
These two papers form the starting-point for a series of articles on the same
subject by Heine and Liouville. Heine in his doctor dissertation14 (1842) deter-
mined the potential not merely for the interior of an ellipsoid of revolution when
the value of the potential is given on the surface, but also for the exterior of
such an ellipsoid and for the shell between two confocal ellipsoids of revolution.
Even in the first of these problems, which is equivalent to that of Lamé, he sim-
plified Lamé’s solution materially by showing that the functions used may be
reduced to spherical harmonics, while in the other two problems he introduced
spherical harmonics of the second kind, which were then new. Shortly after-
wards15 Heine and Liouville published simultaneously two papers in which they
arrived independently of each other at about the same results. In each of these
papers attention is called to the fact that the product of two Lamé’s functions
is a spherical harmonic, and this fact is made use of to throw Lamé’s solution
of the problem of the permanent state of temperatures of an ellipsoid into a
more elementary form. Besides this the second solution of Lamé’s equation is
introduced for the sake of solving the potential problem for the exterior of the
ellipsoid.
In thus following up the theory of heat and the related potential problems,
we have lost sight of the question of small vibrations, to which during the early
part of the century a great deal of attention had been devoted by Poisson, who
frequently made use of the method of development in series. In his memoirs16
most of the problems left unfinished by Bernoulli and Euler are thoroughly
treated, as well as various slight modifications of them. When, however, he
attacked the problem of the vibration of an elastic plate he was unable to make
much progress, owing in part to the erroneous form of his boundary conditions.
13 “Sur l’équilibre des Températures dans un ellipsoı̈de à trois axes inégaux.” An article by
the same author on the two dimensional potential will be found in Vol. I. of this Journal.
14 Reprinted in Crelle’s Journal, Vol. 26 (1843).
In the same Journal for 1847 F. Neumann discussed the related problem of the magnetisation
of a soft iron ellipsoid of revolution.
15 Heine: Crelle’s Journal, Vol. 29, 1845. Liouville: Liouville’s Journal, Vol. X., 1845, and
Vol. XI., 1846. For a treatment of the problem of the potential of an ellipsoidal shell by means
1
of a development of in terms of Lamé’s functions, see a paper by Heine in Crelle’s Journal,
r
Vol. 42, 1851.
16 See especially the one in the Mémoires de l’académie des sciences, Vol. VIII., 1829.
He was, nevertheless, able to solve the problem of the symmetrical vibration

of a free circular plate. The complete theory of the vibration of a free circular
plate was first given by Kirchhoff.17
Passing now to a new subject, the theory of the equilibrium of an elastic
spherical shell, we find a solution by Lamé in Liouville’s Journal for 1854, and
by Sir William Thomson (1862) in the Philosophical Transactions for 1863. Both
of these papers consist of an application of the spherical harmonic analysis to
this rather complicated problem. Thomson, however, considers besides Lamé’s
problem certain related questions and the form of his analysis is very different
from Lamé’s, being of the same nature as that used in the Appendix B of his
Natural Philosophy of which we shall have to speak presently. These investi-
gations form the starting point for a number of recent memoirs among which
those of G. H. Darwin on cosmographical questions deserve special mention.
Closely related to this last mentioned problem is the theory of the small
vibrations of an elastic sphere. While the simplest case of this problem was
treated by Poisson in the memoir referred to above, the general solution has been
only recently obtained by Jaerisch (1879)18 and Lamb (1882).19 The functions
involved are the same as those which occur in the problem of the non-stationary
flow of heat in a sphere as solved by Laplace.
The Appendix B of Thomson and Tait’s Natural Philosophy,20 to which we
have already referred, deserves to be regarded as one of the most important
contributions to the general theory. The way in which spherical harmonics are
introduced (as homogeneous functions of the rectangular coördinates) was then
new21 and the solution of the potential problem for a variety of new solids was
indicated; viz., for solids whose boundaries consist of concentric spheres, cones
of revolution, and planes. We shall have more to say presently concerning the
method employed for the solution of these problems.
Although connected only indirectly with the theory we are discussing, it will
be well to mention at this point the method of electrical images which is also due
to Sir William Thomson (1845). This method enables us to solve many potential
problems for the inverse of any solid when once we have solved it for the solid
itself. By means of this method most of the solutions of potential problems
obtained by our method may be applied at once with very little modification
to systems of curvilinear coördinates derived by inversion from those we have
used. It will not be necessary to mention separately problems of this sort, as it
is clearly immaterial whether they be solved directly or by means of the method
of inversion.22
Returning now to the Continent, we find as the next important question
17 Crelle’s Journal, Vol. 40, 1850.
18 Crelle’s Journal, Vol. 88.
19 Proc. Lond. Math. Soc.
20 First edition, 1867. This appendix was evidently written as early as 1862, as Thomson
refers to it in the memoir quoted above.

21 The same method was used at about the same time by Clebsch.
22 A case in point would be the potential problem for the shell between two non-intersecting
eccentric spheres, since these spheres can be inverted into concentric spheres. This problem
was treated directly by C. Neumann in a monograph published in Halle in 1862.
taken up the problem of the potential of an anchor ring. The first publication
on this subject is a monograph by C. Neumann23 (1864), but in Riemann’s
posthumous papers which were not published until 1876, ten years after his
death, will be found a short fragment on this subject, which (cf. the last page of
Hattendorf’s edition of Riemann’s lectures: “Partielle Differentialgleichungen”)
would appear to date back to the winter 1860-61. This fragment is of peculiar
interest, as the opening paragraphs clearly show that Riemann had in mind an
extended article on the fundamental principles of our subject.
We will next mention two papers by Mehler in which the functions known
as “conal harmonics,” which had already been introduced by Thomson in the
Appendix B above mentioned, were applied to the solution of two problems in
electrostatics. The first of these papers24 (1868) deals with the solid bounded
by two intersecting spheres, while in the second25 (1870) the infinite cone of
revolution is treated. Both of these problems are essentially different from those
discussed in the “Appendix B,” inasmuch as the infinite series which we usually
have degenerate in these cases into definite integrals, just as they do in some
simpler cases treated by Fourier. The later of the two papers just quoted also
contains valuable information concerning the nature of the solution of similar
problems for the hyperboloids and paraboloids of revolution. The solutions of
these problems are not, however, given.
It remains, in order to close the history of this part of the subject, to mention
a number of memoirs which although treating entirely new problems are of far
less importance than most of those considered up to this point, partly because
the solution is not brought to a point where it can be of much immediate use,
and partly because most of the methods employed are such as could not fail to
present themselves to any one attacking these problems.
Of these the first is a paper by Mathieu26 on the vibration of an elliptic
membrane (1868), in which the functions of the elliptic cylinder occur for the
first time.
This was followed in the same year by a paper on closely allied subjects
by H. Weber,27 in which not merely the case of the complete ellipse is briefly
considered, but also that in which the boundary consists of two arcs of confocal
ellipses and two arcs of hyperbolas confocal with them. The special case in
which the ellipses and hyperbolas become confocal parabolas is also considered,
whereby the functions of the parabolic cylinder are for the first time introduced.
In Mathieu’s “Cours de physique mathématique” (1873) the problem of the
non-stationary flow of heat in an ellipsoid is touched upon, and an elaborate
though not very satisfactory treatment of the special cases where we have ellip-
soids of revolution is given. New functions appear in all of these problems.
Of late years C. Baer has supplied a number of missing links in the chain
23 “Theorie der Elektricitäts- und Wärme-Vertheilung in einem Ringe.” Halle.
24 Crelle’sJournal, Vol. 68, 1868.
25 Jahresbericht des Gymnasiums zu Elbing.
26 Liouville’s Journal, Vol. XIII.
27 “Ueber die Integration der partiellen Differentialgleichung δ 2 u + δ 2 u + k 2 u = 0.” Math.
δx2 δy 2
Ann., Vol. I. No physical problem is mentioned in this paper.
of problems here considered by treating in succession the potential problem

for the paraboloid of revolution,28 the parabolic cylinder29 and the general
paraboloid.30 In the first of these problems Bessel’s functions occur, as had
already been stated by Mehler, while in the last we find the functions of the
elliptic cylinder. For each of the three systems of coördinates employed the same
author also touches upon the more general problem of the non-stationary flow
of heat, in which new functions occur.
Except in the case of the anchor ring we have found so far only such solids
treated by our method as are bounded by surfaces of the first or second degree.
Wangerin31 (1875-76) considered in connection with the theory of the potential,
more general systems of curvilinear coördinates than had previously been used in
physical questions, namely, cyclidic coördinates.32 He showed, however, merely
how to break up Laplace’s equation into three ordinary differential equations.33
An important branch of our theory which we have not yet touched upon
dates back to the year 1836, when Sturm published a series of fundamentally
important papers in the first two volumes of Liouville’s Journal. The physical
question which lies at the basis of these papers is the problem of the flow of heat
in a heterogeneous bar.34 The method here employed depends upon the fact
that the functions which occur are characterized by the number of times they
vanish in a certain interval. This same idea reappears in Thomson and Tait’s
Appendix B already referred to, but first finds its full expression in this more
general field of the three dimensional potential in an article by Klein: “Ueber
Körper welche von confocalen Flächen zweiten Grades begrenzt sind”35 (1881).
Still more recently (1889-90) Klein has in his lectures extended this theory to
the treatment of solids bounded by six confocal cyclids, and has indicated how
all the potential problems heretofore treated by our method are special cases of
this one.36
Of late years, especially since the year 1880, the younger English mathemati-
cians have done a vast amount of work in the theory we are here considering.
Although much of this work is of great value, hardly any of it can be regarded
as being a real development of the method; it is rather an application of it to
28 “Ueber das Gleichgewicht und die Bewegung der Wärme in einem Rotationsparaboloid.”
Dissertation, Halle, 1881.

29 “Die Funktion des parabolischen Cylinders,” Gymnasialprogramm Cüstrin, 1883.
30 “Parabolische Coordinaten,” Frankfurt, 1888. See also a paper by Greenhill in the Proc.
Lond. Math. Soc., Vol. XIX., 1889 (read Dec. 8, 1887). Also a posthumous paper by Lamé in
Liouville’s Journal for 1874, Vol. XIX.
31 Preisschriften der Jablanowski’schen Gesellschaft, No. XVIII., and Crelle’s Journal, Vol.
82. See also, concerning a still further extension, the Berliner Monatsberichten for 1878.
32 Cyclids are a kind of surface of the fourth order (see Salmon’s Geom. of three Dimensions,
p. 527). In his first memoir Wangerin considers only cyclids of revolution.

33 See also a paper by this author in Grünert’s Archiv for 1873, where the problem of the
equilibrium of elastic solids of revolution is treated.

34 The similar problem of the vibration of a heterogeneous string under the action of an
external force was treated by Maggi (Giornale di Matematiche, 1880). Several special cases
are also considered here in detail.
35 Math. Ann., 18.
36 For an exposition of this theory see the treatise: Ueber die Reihenentwickelungen der
Potentialtheorie, Leipsic, Teubner, 1894, by the writer of the present chapter.

a great variety of problems. We must therefore content ourselves with giving a

mere list of a few of the more important of these papers.
Niven: On the Conduction of Heat in Ellipsoids of Revolution. Phil. Trans.,
1880.
Niven: On the Induction of Electric Currents in Infinite Plates and Spherical
Shells. Phil. Trans., 1881.
Hicks: On Toroidal Functions. Phil. Trans., 1881.
Hicks: On the Steady Motion and Small Vibrations of a Hollow Vortex.
Phil. Trans., 1884, 1885.
Lamb: On Ellipsoidal Current Sheets. Phil. Trans., 1887.
Chree: The Equations of an Isotropic Elastic Solid in Polar and Cylindrical
Coördinates, their Solution and Application. Camb. Phil. Soc. Trans., XIV.,
1889.
Hobson: On a Class of Spherical Harmonics of Complex Degree with Appli-
cations to Physical Problems. Camb. Phil. Soc. Trans., XIV., 1889.
Chree: On some Compound Vibrating Systems. Camb. Phil. Soc. Trans.,
XV., 1891.
Niven: On Ellipsoidal Harmonics. Phil. Trans., 1892.
The historical sketch we have just given would naturally require as a sup-
plement some account of the work that has been done on the question of the
convergence of the various series which occur. This, however, would carry us
too far, and we will content ourselves with mentioning the two fundamental
memoirs by Dirichlet in Crelle’s Journal, one in 1829 on Fourier’s series, and
one, which has been criticised to some extent by subsequent mathematicians,
in 1837 on Laplace’s spherical harmonic development.
Another subject which naturally presents itself here is the theory of the
various new functions we have met. Those properties of these functions, how-
ever, which the physicist needs have usually been investigated by the physicists
themselves in the papers mentioned above; while any thorough account of the
development of the theory of these functions would lead us into the vast region
of the modern theory of linear differential equations.
We will therefore close by merely giving a list of books which will be found
useful by those wishing to continue their study of the subject further.
We begin with the books relating directly to physical questions:
Fourier: Théorie Analytique de la Chaleur, 1822.
Lamé: Leçons sur les Functions inverses des Transcendantes et les Surfaces
isothermes, 1857.
Lamé: Leçons sur les Coordonnées Curvilignes et leurs diverses Applications,
1859.
Mathieu: Cours de Physique Mathématique, 1873.
Riemann: Partielle Differentialgleichungen, und deren Anwendung auf phys-
ikalische Fragen (edited by Hattendorf), third edition. 1882.
F. Neumann: Theorie des Potentials und der Kugelfunktionen (edited by C.
Neumann), 1887.
Thomson and Tait: Natural Philosophy, second edition, 1879.
Rayleigh: Theory of Sound, 1877.
Basset: Hydrodynamics, 1888.

Love: Theory of Elasticity, 1892.
Heine: Handbuch der Kugelfunktionen (second edition), 1878-81.
Ferrers: Spherical Harmonics, 1881.
Haentzschel: Reduction der Potentialgleichung auf gewöhnliche Differential-
gleichungen, 1893.
These last three books would also belong in the following list of books relat-
ing to the theory of the various functions we use:
Todhunter: The Functions of Laplace, Lamé and Bessel, 1875.
Lommel: Studien über die Bessel’schen Funktionen, 1868.
F. Neumann: Beiträge zur Theorie der Kugelfunktionen, 1878.
And finally concerning the question of convergence:
C. Neumann: Über die nach Kreis-, Kugel- und Cylinder-Functionen fort-
schreitenden Entwickelungen, 1881.
275
APPENDIX.
TABLES.
Table I., a table of Surface Zonal Harmonics (Legendrians), gives the values
of the first seven Harmonics P1 (cos θ), P2 (cos θ), · · · P7 (cos θ) for the argument
θ in degrees. It is taken from the Philosophical Magazine for December, 1891,
and was computed by Messrs. C. E. Holland, P. R. James, and C. G. Lamb,
under the direction of Professor John Perry.
Table II., a table of Surface Zonal Harmonics (Legendrians), gives the values
of the first seven Harmonics P1 (x), P2 (x), · · · P7 (x) for the argument x. It is
reduced from the Tables of Legendrian Functions computed under the direction
of Dr. J. W. L. Glaisher, and published in the Report of the British Association
for the Advancement of Science for the year 1879.
Table III., the table of Hyperbolic Functions, gives the values of ex , e−x ,
sinh x, cosh x, and gd x (Gudermannian of x) for values of x from 0.00 to 1.00;
and the values of log sinh x and log cosh x for values of x from 1.00 to 10.0. The
values of gd x, log sinh x, and log cosh x are taken from the Mathematical Tables
prepared by Professor J. M. Peirce (Boston: Ginn & Co.).
The log sinh x and log cosh x for values of x between 0.00 and 1.00 can be
obtained from the values given for the Gudermannian of x in the table by the
aid of the relations
log sinh x = log tan(gd x) log cosh x = log sec(gd x).
Table IV. gives the first twelve roots of J0 (x) = 0 and J1 (x) = 0 each divided
by π. The table is taken from Lord Rayleigh’s Sound, Vol. I., page 274, and is
due to Professor Stokes, Camb. Phil. Trans., Vol. IX., page 186.
Table V. gives the first nine roots of J0 (x) = 0, J1 (x) = 0, · · · J5 (x) = 0. The
table is taken from Rayleigh’s Sound, Vol. I., page 274, and is due to Professor
J. Bourget, Ann. de l’Ecole Normale, T. III., 1866, page 82.
Table VI., the table of Bessel’s Functions, gives the values of the Bessel’s
Functions J0 (x) and J1 (x) for the argument x from x = 0 to x = 15. It is
taken from Rayleigh’s Sound, Vol. I., page 265, and from Lommel’s Bessel’sche
Functionen.
APPENDIX 276
TABLE I. — Surface Zonal Harmonics
θ P1 (cos θ) P2 (cos θ) P3 (cos θ) P4 (cos θ) P5 (cos θ) P6 (cos θ) P7 (cos θ)
0◦ 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

1 .9998 .9995 .9991 .9985 .9977 .9967 .9955
2 .9994 .9982 .9963 .9939 .9909 .9872 .9829
3 .9986 .9959 .9918 .9863 .9795 .9713 .9617
4 .9976 .9927 .9854 .9758 .9638 .9495 .9329
5 .9962 .9886 .9773 .9623 .9437 .9216 .8961

6 .9945 .9836 .9674 .9459 .9194 .8881 .8522
7 .9925 .9777 .9557 .9267 .8911 .8476 .7986
8 .9903 .9709 .9423 .9048 .8589 .8053 .7448
9 .9877 .9633 .9273 .8803 .8232 .7571 .6831
10 .9848 .9548 .9106 .8532 .7840 .7045 .6164

11 .9816 .9454 .8923 .8238 .7417 .6483 .5461
12 .9781 .9352 .8724 .7920 .6966 .5892 .4732
13 .9744 .9241 .8511 .7582 .6489 .5273 .3940
14 .9703 .9122 .8283 .7224 .5990 .4635 .3219
15 .9659 .8995 .8042 .6847 .5471 .3982 .2454

16 .9613 .8860 .7787 .6454 .4937 .3322 .1699
17 .9563 .8718 .7519 .6046 .4391 .2660 .0961
18 .9511 .8568 .7240 .5624 .3836 .2002 .0289
19 .9455 .8410 .6950 .5192 .3276 .1347 −.0443
20 .9397 .8245 .6649 .4750 .2715 .0719 −.1072

21 .9336 .8074 .6338 .4300 .2156 .0107 −.1662
22 .9272 .7895 .6019 .3845 .1602 −.0481 −.2201
23 .9205 .7710 .5692 .3386 .1057 −.1038 −.2681
24 .9135 .7518 .5357 .2926 .0525 −.1559 −.3095
25 .9063 .7321 .5016 .2465 .0009 −.2053 −.3463

26 .8988 .7117 .4670 .2007 −.0489 −.2478 −.3717
27 .8910 .6908 .4319 .1553 −.0964 −.2869 −.3921
28 .8829 .6694 .3964 .1105 −.1415 −.3211 −.4052
29 .8746 .6474 .3607 .0665 −.1839 −.3503 −.4114
30 .8660 .6250 .3248 .0234 −.2233 −.3740 −.4101

31 .8572 .6021 .2887 −.0185 −.2595 −.3924 −.4022
32 .8480 .5788 .2527 −.0591 −.2923 −.4052 −.3876
33 .8387 .5551 .2167 −.0982 −.3216 −.4126 −.3670
34 .8290 .5310 .1809 −.1357 −.3473 −.4148 −.3409
35 .8192 .5065 .1454 −.1714 −.3691 −.4115 −.3096

36 .8090 .4818 .1102 −.2052 −.3871 −.4031 −.2738
37 .7986 .4567 .0755 −.2370 −.4011 −.3898 −.2343
38 .7880 .4314 .0413 −.2666 −.4112 −.3719 −.1918
39 .7771 .4059 .0077 −.2940 −.4174 −.3497 −.1469
40 .7660 .3802 −.0252 −.3190 −.4197 −.3234 −.1003

41 .7547 .3544 −.0574 −.3416 −.4181 −.2938 −.0534
42 .7431 .3284 −.0887 −.3616 −.4128 −.2611 −.0065
43 .7314 .3023 −.1191 −.3791 −.4038 −.2255 .0398
44 .7193 .2762 −.1485 −.3940 −.3914 −.1878 .0846
45◦ .7071 .2500 −.1768 −.4062 −.3757 −.1485 .1270

APPENDIX 277
TABLE I. — Surface Zonal Harmonics
θ P1 (cos θ) P2 (cos θ) P3 (cos θ) P4 (cos θ) P5 (cos θ) P6 (cos θ) P7 (cos θ)
45◦ .7071 .2500 −.1768 −.4062 −.3757 −.1485 .1270

46 .6947 .2238 −.2040 −.4158 −.3568 −.1079 .1666
47 .6820 .1977 −.2300 −.4252 −.3350 −.0645 .2054
48 .6691 .1716 −.2547 −.4270 −.3105 −.0251 .2349
49 .6561 .1456 −.2781 −.4286 −.2836 .0161 .2627
50 .6428 .1198 −.3002 −.4275 −.2545 .0563 .2854

51 .6293 .0941 −.3209 −.4239 −.2235 .0954 .3031
52 .6157 .0686 −.3401 −.4178 −.1910 .1326 .3153
53 .6018 .0433 −.3578 −.4093 −.1571 .1677 .3221
54 .5878 .0182 −.3740 −.3984 −.1223 .2002 .3234
55 .5736 −.0065 −.3886 −.3852 −.0868 .2297 .3191

56 .5592 −.0310 −.4016 −.3698 −.0510 .2559 .3095
57 .5446 −.0551 −.4131 −.3524 −.0150 .2787 .2949
58 .5299 −.0788 −.4229 −.3331 .0206 .2976 .2752
59 .5150 −.1021 −.4310 −.3119 .0557 .3125 .2511
60 .5000 −.1250 −.4375 −.2891 .0898 .3232 .2231

61 .4848 −.1474 −.4423 −.2647 .1229 .3298 .1916
62 .4695 −.1694 −.4455 −.2390 .1545 .3321 .1571
63 .4540 −.1908 −.4471 −.2121 .1844 .3302 .1203
64 .4384 −.2117 −.4470 −.1841 .2123 .3240 .0818
65 .4226 −.2321 −.4452 −.1552 .2381 .3138 .0422

66 .4067 −.2518 −.4419 −.1256 .2615 .2996 .0021
67 .3907 −.2710 −.4370 −.0955 .2824 .2819 −.0375
68 .3746 −.2896 −.4305 −.0650 .3005 .2605 −.0763
69 .3584 −.3074 −.4225 −.0344 .3158 .2361 −.1135
70 .3420 −.3245 −.4130 −.0038 .3281 .2089 −.1485

71 .3256 −.3410 −.4021 .0267 .3373 .1786 −.1811
72 .3090 −.3568 −.3898 .0568 .3434 .1472 −.2099
73 .2924 −.3718 −.3761 .0864 .3463 .1144 −.2347
74 .2756 −.3860 −.3611 .1153 .3461 .0795 −.2559
75 .2588 −.3995 −.3449 .1434 .3427 .0431 −.2730

76 .2419 −.4112 −.3275 .1705 .3362 .0076 −.2848
77 .2250 −.4241 −.3090 .1964 .3267 −.0284 −.2919
78 .2079 −.4352 −.2894 .2211 .3143 −.0644 −.2943
79 .1908 −.4454 −.2688 .2443 .2990 −.0989 −.2913
80 .1736 −.4548 −.2474 .2659 .2810 −.1321 −.2835

81 .1564 −.4633 −.2251 .2859 .2606 −.1635 −.2709
82 .1392 −.4709 −.2020 .3040 .2378 −.1926 −.2536
83 .1219 −.4777 −.1783 .3203 .2129 −.2193 −.2321
84 .1045 −.4836 −.1539 .3345 .1861 −.2431 −.2067
85 .0872 −.4886 −.1291 .3468 .1577 −.2638 −.1779

86 .0698 −.4927 −.1038 .3569 .1278 −.2811 −.1460
87 .0523 −.4959 −.0781 .3648 .0969 −.2947 −.1117
88 .0349 −.4982 −.0522 .3704 .0651 −.3045 −.0735
89 .0175 −.4995 −.0262 .3739 .0327 −.3105 −.0381
90◦ .0000 −.5000 .0000 .3750 .0000 −.3125 .0000

APPENDIX 278
TABLE II.—Surface Zonal Harmonics.
x P1 (x) P2 (x) P3 (x) P4 (x) P5 (x) P6 (x) P7 (x)
0.00 0.0000 −.5000 0.0000 0.3750 0.0000 −.3125 0.0000

.01 .0100 −.4998 −.0150 .3746 .0187 −.3118 −.0219
.02 .0200 −.4994 −.0300 .3735 .0374 −.3099 −.0436
.03 .0300 −.4986 −.0449 .3716 .0560 −.3066 −.0651
.04 .0400 −.4976 −.0598 .3690 .0744 −.3021 −.0862
.05 .0500 −.4962 −.0747 .3657 .0927 −.2962 −.1069

.06 .0600 −.4946 −.0895 .3616 .1106 −.2891 −.1270
.07 .0700 −.4926 −.1041 .3567 .1283 −.2808 −.1464
.08 .0800 −.4904 −.1187 .3512 .1455 −.2713 −.1651
.09 .0900 −.4878 −.1332 .3449 .1624 −.2606 −.1828
.10 .1000 −.4850 −.1475 .3379 .1788 −.2488 −.1995

.11 .1100 −.4818 −.1617 .3303 .1947 −.2360 −.2151
.12 .1200 −.4784 −.1757 .3219 .2101 −.2220 −.2295
.13 .1300 −.4746 −.1895 .3129 .2248 −.2071 −.2427
.14 .1400 −.4706 −.2031 .3032 .2389 −.1913 −.2545
.15 .1500 −.4662 −.2166 .2928 .2523 −.1746 −.2649

.16 .1600 −.4616 −.2298 .2819 .2650 −.1572 −.2738
.17 .1700 −.4566 −.2427 .2703 .2769 −.1389 −.2812
.18 .1800 −.4514 −.2554 .2581 .2880 −.1201 −.2870
.19 .1900 −.4458 −.2679 .2453 .2982 −.1006 −.2911
.20 .2000 −.4400 −.2800 .2320 .3075 −.0806 −.2935

.21 .2100 −.4338 −.2918 .2181 .3159 −.0601 −.2943
.22 .2200 −.4274 −.3034 .2037 .3234 −.0394 −.2933
.23 .2300 −.4206 −.3146 .1889 .3299 −.0183 −.2906
.24 .2400 −.4136 −.3254 .1735 .3353 .0029 −.2861
.25 .2500 −.4062 −.3359 .1577 .3397 .0243 −.2799

.26 .2600 −.3986 −.3461 .1415 .3431 .0456 −.2720
.27 .2700 −.3906 −.3558 .1249 .3453 .0669 −.2625
.28 .2800 −.3824 −.3651 .1079 .3465 .0879 −.2512
.29 .2900 −.3738 −.3740 .0906 .3465 .1087 −.2384
.30 .3000 −.3650 −.3825 .0729 .3454 .1292 −.2241

.31 .3100 −.3558 −.3905 .0550 .3431 .1492 −.2082
.32 .3200 −.3464 −.3981 .0369 .3397 .1686 −.1910
.33 .3300 −.3366 −.4052 .0185 .3351 .1873 −.1724
.34 .3400 −.3266 −.4117 −.0000 .3294 .2053 −.1527
.35 .3500 −.3162 −.4178 −.0187 .3225 .2225 −.1318

.36 .3600 −.3056 −.4234 −.0375 .3144 .2388 −.1098
.37 .3700 −.2946 −.4284 −.0564 .3051 .2540 −.0870
.38 .3800 −.2834 −.4328 −.0753 .2948 .2681 −.0635
.39 .3900 −.2718 −.4367 −.0942 .2833 .2810 −.0393
.40 .4000 −.2600 −.4400 −.1130 .2706 .2926 −.0146

.41 .4100 −.2478 −.4427 −.1317 .2569 .3029 .0104
.42 .4200 −.2354 −.4448 −.1504 .2421 .3118 .0356
.43 .4300 −.2226 −.4462 −.1688 .2263 .3191 .0608
.44 .4400 −.2096 −.4470 −.1870 .2095 .3249 .0859
.45 .4500 −.1962 −.4472 −.2050 .1917 .3290 .1106

.46 .4600 −.1826 −.4467 −.2226 .1730 .3314 .1348
.47 .4700 −.1686 −.4454 −.2399 .1534 .3321 .1584
.48 .4800 −.1544 −.4435 −.2568 .1330 .3310 .1811
.49 .4900 −.1398 −.4409 −.2732 .1118 .3280 .2027
.50 .5000 −.1250 −.4375 −.2891 .0898 .3232 .2231

APPENDIX 279
TABLE II.—Surface Zonal Harmonics.
x P1 (x) P2 (x) P3 (x) P4 (x) P5 (x) P6 (x) P7 (x)
.50 .5000 −.1250 −.4375 −.2891 .0898 .3232 .2231

.51 .5100 −.1098 −.4334 −.3044 .0673 .3166 .2422
.52 .5200 −.0944 −.4285 −.3191 .0441 .3080 .2596
.53 .5300 −.0786 −.4228 −.3332 .0204 .2975 .2753
.54 .5400 −.0626 −.4163 −.3465 −.0037 .2851 .2891
.55 .5500 −.0462 −.4091 −.3590 −.0282 .2708 .3007

.56 .5600 −.0296 −.4010 −.3707 −.0529 .2546 .3102
.57 .5700 −.0126 −.3920 −.3815 −.0779 .2366 .3172
.58 .5800 .0046 −.3822 −.3914 −.1028 .2168 .3217
.59 .5900 .0222 −.3716 −.4002 −.1278 .1953 .3235
.60 .6000 .0400 −.3600 −.4080 −.1526 .1721 .3226

.61 .6100 .0582 −.3475 −.4146 −.1772 .1473 .3188
.62 .6200 .0766 −.3342 −.4200 −.2014 .1211 .3121
.63 .6300 .0954 −.3199 −.4242 −.2251 .0935 .3023
.64 .6400 .1144 −.3046 −.4270 −.2482 .0646 .2895
.65 .6500 .1338 −.2884 −.4284 −.2705 .0347 .2737

.66 .6600 .1534 −.2713 −.4284 −.2919 .0038 .2548
.67 .6700 .1734 −.2531 −.4268 −.3122 −.0278 .2329
.68 .6800 .1936 −.2339 −.4236 −.3313 −.0601 .2081
.69 .6900 .2142 −.2137 −.4187 −.3490 −.0926 .1805
.70 .7000 .2350 −.1925 −.4121 −.3652 −.1253 .1502

.71 .7100 .2562 −.1702 −.4036 −.3796 −.1578 .1173
.72 .7200 .2776 −.1469 −.3933 −.3922 −.1899 .0822
.73 .7300 .2994 −.1225 −.3810 −.4026 −.2214 .0450
.74 .7400 .3214 −.0969 −.3666 −.4107 −.2518 .0061
.75 .7500 .3438 −.0703 −.3501 −.4164 −.2808 −.0342

.76 .7600 .3664 −.0426 −.3314 −.4193 −.3081 −.0754
.77 .7700 .3894 −.0137 −.3104 −.4193 −.3333 −.1171
.78 .7800 .4126 .0164 −.2871 −.4162 −.3559 −.1588
.79 .7900 .4362 .0476 −.2613 −.4097 −.3756 −.1999
.80 .8000 .4600 .0800 −.2330 −.3995 −.3918 −.2397

.81 .8100 .4842 .1136 −.2021 −.3855 −.4041 −.2774
.82 .8200 .5086 .1484 −.1685 −.3674 −.4119 −.3124
.83 .8300 .5334 .1845 −.1321 −.3449 −.4147 −.3437
.84 .8400 .5584 .2218 −.0928 −.3177 −.4120 −.3703
.85 .8500 .5838 .2603 −.0506 −.2857 −.4030 −.3913

.86 .8600 .6094 .3001 −.0053 −.2484 −.3872 −.4055
.87 .8700 .6354 .3413 .0431 −.2056 −.3638 −.4116
.88 .8800 .6616 .3837 .0947 −.1570 −.3322 −.4083
.89 .8900 .6882 .4274 .1496 −.1023 −.2916 −.3942
.90 .9000 .7150 .4725 .2079 −.0411 −.2412 −.3678

.91 .9100 .7422 .5189 .2698 .0268 −.1802 −.3274
.92 .9200 .7696 .5667 .3352 .1017 −.1077 −.2713
.93 .9300 .7974 .6159 .4044 .1842 −.0229 −.1975
.94 .9400 .8254 .6665 .4773 .2744 .0751 −.1040
.95 .9500 .8538 .7184 .5541 .3727 .1875 .0112

.96 .9600 .8824 .7718 .6349 .4796 .3151 .1506
.97 .9700 .9114 .8267 .7198 .5954 .4590 .3165
.98 .9800 .9406 .8830 .8089 .7204 .6204 .5115
.99 .9900 .9702 .9407 .9022 .8552 .8003 .7384
1.00 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

APPENDIX 280
TABLE III.—Hyperbolic Functions.
x ex e−x sinh x cosh x gd x
◦
0.00 1.0000 1.0000 0.0000 1.0000 0.0000
.01 1.0100 0.9900 .0100 1.0000 0.5729
.02 1.0202 .9802 .0200 1.0002 1.1458
.03 1.0305 .9704 .0300 1.0004 1.7186
.04 1.0408 .9608 .0400 1.0008 2.2912
.05 1.0513 .9512 .0500 1.0012 2.8636

.06 1.0618 .9418 .0600 1.0018 3.4357
.07 1.0725 .9324 .0701 1.0025 4.0074
.08 1.0833 .9231 .0801 1.0032 4.5788
.09 1.0942 .9139 .0901 1.0040 5.1497
.10 1.1052 .9048 .1002 1.0050 5.720

.11 1.1163 .8958 .1102 1.0061 6.290
.12 1.1275 .8869 .1203 1.0072 6.859
.13 1.1388 .8781 .1304 1.0085 7.428
.14 1.1503 .8694 .1405 1.0098 7.995
.15 1.1618 .8607 .1506 1.0113 8.562

.16 1.1735 .8521 .1607 1.0128 9.128
.17 1.1853 .8437 .1708 1.0145 9.694
.18 1.1972 .8353 .1810 1.0162 10.258
.19 1.2092 .8270 .1911 1.0181 10.821
.20 1.2214 .8187 .2013 1.0201 11.384

.21 1.2337 .8106 .2115 1.0221 11.945
.22 1.2461 .8025 .2218 1.0243 12.505
.23 1.2586 .7945 .2320 1.0266 13.063
.24 1.2712 .7866 .2423 1.0289 13.621
.25 1.2840 .7788 .2526 1.0314 14.177

.26 1.2969 .7711 .2629 1.0340 14.732
.27 1.3100 .7634 .2733 1.0367 15.285
.28 1.3231 .7558 .2837 1.0395 15.837
.29 1.3364 .7483 .2941 1.0423 16.388
.30 1.3499 .7408 .3045 1.0453 16.937

.31 1.3634 .7334 .3150 1.0484 17.484
.32 1.3771 .7261 .3255 1.0516 18.030
.33 1.3910 .7189 .3360 1.0549 18.573
.34 1.4049 .7118 .3466 1.0584 19.116
.35 1.4191 .7047 .3572 1.0619 19.656

.36 1.4333 .6977 .3678 1.0655 20.195
.37 1.4477 .6907 .3785 1.0692 20.732
.38 1.4623 .6839 .3892 1.0731 21.267
.39 1.4770 .6771 .4000 1.0770 21.800
.40 1.4918 .6703 .4108 1.0811 22.331

.41 1.5068 .6636 .4216 1.0852 22.859
.42 1.5220 .6570 .4325 1.0895 23.386
.43 1.5373 .6505 .4434 1.0939 23.911
.44 1.5527 .6440 .4543 1.0984 24.434
.45 1.5683 .6376 .4653 1.1030 24.955

.46 1.5841 .6313 .4764 1.1077 25.473
.47 1.6000 .6250 .4875 1.1125 25.989
.48 1.6161 .6188 .4986 1.1174 26.503
.49 1.6323 .6126 .5098 1.1225 27.015
◦
0.50 1.6487 0.6065 0.5211 1.1276 27.524
APPENDIX 281
x ex e−x sinh x cosh x gd x
◦
0.50 1.6487 0.6065 0.5211 1.1276 27.524
.51 1.6653 .6005 .5324 1.1329 28.031
.52 1.6820 .5945 .5438 1.1383 28.535
.53 1.6989 .5886 .5552 1.1438 29.037
.54 1.7160 .5827 .5666 1.1494 29.537
.55 1.7333 .5770 .5782 1.1551 30.034

.56 1.7507 .5712 .5897 1.1609 30.529
.57 1.7683 .5655 .6014 1.1669 31.021
.58 1.7860 .5599 .6131 1.1730 31.511
.59 1.8040 .5543 .6248 1.1792 31.998
.60 1.8221 .5488 .6367 1.1855 32.483

.61 1.8404 .5433 .6485 1.1919 32.965
.62 1.8589 .5379 .6605 1.1984 33.444
.63 1.8776 .5326 .6725 1.2051 33.921
.64 1.8965 .5273 .6846 1.2119 34.395
.65 1.9155 .5220 .6967 1.2188 34.867

.66 1.9348 .5169 .7090 1.2258 35.336
.67 1.9542 .5117 .7213 1.2330 35.802
.68 1.9739 .5066 .7336 1.2402 36.265
.69 1.9937 .5016 .7461 1.2476 36.726
.70 2.0138 .4966 .7586 1.2552 37.183

.71 2.0340 .4916 .7712 1.2628 37.638
.72 2.0544 .4867 .7838 1.2706 38.091
.73 2.0751 .4819 .7966 1.2785 38.540
.74 2.0959 .4771 .8094 1.2865 38.987
.75 2.1170 .4724 .8223 1.2947 39.431

.76 2.1383 .4677 .8353 1.3030 39.872
.77 2.1598 .4630 .8484 1.3114 40.310
.78 2.1815 .4584 .8615 1.3199 40.746
.79 2.2034 .4538 .8748 1.3286 41.179
.80 2.2255 .4493 .8881 1.3374 41.608

.81 2.2479 .4449 .9015 1.3464 42.035
.82 2.2705 .4404 .9150 1.3555 42.460
.83 2.2933 .4360 .9286 1.3647 42.881
.84 2.3164 .4317 .9423 1.3740 43.299
.85 2.3396 .4274 .9561 1.3835 43.715

.86 2.3632 .4232 .9700 1.3932 44.128
.87 2.3869 .4190 .9840 1.4029 44.537
.88 2.4109 .4148 .9981 1.4128 44.944
.89 2.4351 .4107 1.0122 1.4229 45.348
.90 2.4596 .4066 1.0265 1.4331 45.750

.91 2.4843 .4025 1.0409 1.4434 46.148
.92 2.5093 .3985 1.0554 1.4539 46.544
.93 2.5345 .3946 1.0700 1.4645 46.936
.94 2.5600 .3906 1.0847 1.4753 47.326
.95 2.5857 .3867 1.0995 1.4862 47.713

.96 2.6117 .3829 1.1144 1.4973 48.097
.97 2.6379 .3791 1.1294 1.5085 48.478
.98 2.6645 .3753 1.1446 1.5199 48.857
.99 2.6912 .3716 1.1598 1.5314 49.232
◦
1.00 2.7183 0.3679 1.1752 1.5431 49.605
APPENDIX 282
x l sinh x l cosh x x l sinh x l cosh x x l sinh x l cosh x
1.00 0.0701 0.1884 1.50 0.3282 0.3715 2.00 0.5595 0.5754

1.01 .0758 .1917 1.51 .3330 .3754 2.01 .5640 .5796
1.02 .0815 .1950 1.52 .3378 .3794 2.02 .5685 .5838
1.03 .0871 .1984 1.53 .3426 .3833 2.03 .5730 .5880
1.04 .0927 .2018 1.54 .3474 .3873 2.04 .5775 .5922
1.05 .0982 .2051 1.55 .3521 .3913 2.05 .5820 .5964

1.06 .1038 .2086 1.56 .3569 .3952 2.06 .5865 .6006
1.07 .1093 .2120 1.57 .3616 .3992 2.07 .5910 .6048
1.08 .1148 .2154 1.58 .3663 .4032 2.08 .5955 .6090
1.09 .1203 .2189 1.59 .3711 .4072 2.09 .6000 .6132
1.10 .1257 .2223 1.60 .3758 .4112 2.10 .6044 .6175

1.11 .1311 .2258 1.61 .3805 .4152 2.11 .6089 .6217
1.12 .1365 .2293 1.62 .3852 .4192 2.12 .6134 .6259
1.13 .1419 .2328 1.63 .3899 .4232 2.13 .6178 .6301
1.14 .1472 .2364 1.64 .3946 .4273 2.14 .6223 .6343
1.15 .1525 .2399 1.65 .3992 .4313 2.15 .6268 .6386

1.16 .1578 .2435 1.66 .4039 .4353 2.16 .6312 .6428
1.17 .1631 .2470 1.67 .4086 .4394 2.17 .6357 .6470
1.18 .1684 .2506 1.68 .4132 .4434 2.18 .6401 .6512
1.19 .1736 .2542 1.69 .4179 .4475 2.19 .6446 .6555
1.20 .1788 .2578 1.70 .4225 .4515 2.20 .6491 .6597

1.21 .1840 .2615 1.71 .4272 .4556 2.21 .6535 .6640
1.22 .1892 .2651 1.72 .4318 .4597 2.22 .6580 .6682
1.23 .1944 .2688 1.73 .4364 .4637 2.23 .6624 .6724
1.24 .1995 .2724 1.74 .4411 .4678 2.24 .6668 .6767
1.25 .2046 .2761 1.75 .4457 .4719 2.25 .6713 .6809

1.26 .2098 .2798 1.76 .4503 .4760 2.26 .6757 .6852
1.27 .2148 .2835 1.77 .4549 .4801 2.27 .6802 .6894
1.28 .2199 .2872 1.78 .4595 .4842 2.28 .6846 .6937
1.29 .2250 .2909 1.79 .4641 .4883 2.29 .6890 .6979
1.30 .2300 .2947 1.80 .4687 .4924 2.30 .6935 .7022

1.31 .2351 .2984 1.81 .4733 .4965 2.31 .6979 .7064
1.32 .2401 .3022 1.82 .4778 .5006 2.32 .7023 .7107
1.33 .2451 .3059 1.83 .4824 .5048 2.33 .7067 .7150
1.34 .2501 .3097 1.84 .4870 .5089 2.34 .7112 .7192
1.35 .2551 .3135 1.85 .4915 .5130 2.35 .7156 .7235

1.36 .2600 .3173 1.86 .4961 .5172 2.36 .7200 .7278
1.37 .2650 .3211 1.87 .5007 .5213 2.37 .7244 .7320
1.38 .2699 .3249 1.88 .5052 .5254 2.38 .7289 .7363
1.39 .2748 .3288 1.89 .5098 .5296 2.39 .7333 .7406
1.40 .2797 .3326 1.90 .5143 .5337 2.40 .7377 .7448

1.41 .2846 .3365 1.91 .5188 .5379 2.41 .7421 .7491
1.42 .2895 .3403 1.92 .5234 .5421 2.42 .7465 .7534
1.43 .2944 .3442 1.93 .5279 .5462 2.43 .7509 .7577
1.44 .2993 .3481 1.94 .5324 .5504 2.44 .7553 .7619
1.45 .3041 .3520 1.95 .5370 .5545 2.45 .7597 .7662

1.46 .3090 .3559 1.96 .5415 .5587 2.46 .7642 .7705
1.47 .3138 .3598 1.97 .5460 .5629 2.47 .7686 .7748
1.48 .3186 .3637 1.98 .5505 .5671 2.48 .7730 .7791
1.49 .3234 .3676 1.99 .5550 .5713 2.49 .7774 .7833
1.50 0.3282 0.3715 2.00 0.5595 0.5754 2.50 0.7818 0.7876

APPENDIX 283
x l sinh x l cosh x x l sinh x l cosh x x l sinh x l cosh x
2.50 0.7818 0.7876 2.75 0.8915 0.8951 3.0 1.0008 1.0029

2.51 .7862 .7919 2.76 .8959 .8994 3.1 1.0444 1.0462
2.52 .7906 .7962 2.77 .9003 .9037 3.2 1.0880 1.0894
2.53 .7950 .8005 2.78 .9046 .9080 3.3 1.1316 1.1327
2.54 .7994 .8048 2.79 .9090 .9123 3.4 1.1751 1.1761
2.55 .8038 .8091 2.80 .9134 .9166 3.5 1.2186 1.2194

2.56 .8082 .8134 2.81 .9178 .9209 3.6 1.2621 1.2628
2.57 .8126 .8176 2.82 .9221 .9252 3.7 1.3056 1.3061
2.58 .8169 .8219 2.83 .9265 .9295 3.8 1.3491 1.3495
2.59 .8213 .8262 2.84 .9309 .9338 3.9 1.3925 1.3929
2.60 .8257 .8305 2.85 .9353 .9382 4.0 1.4360 1.4363

2.61 .8301 .8348 2.86 .9396 .9425 4.1 1.4795 1.4797
2.62 .8345 .8391 2.87 .9440 .9468 4.2 1.5229 1.5231
2.63 .8389 .8434 2.88 .9484 .9511 4.3 1.5664 1.5665
2.64 .8433 .8477 2.89 .9527 .9554 4.4 1.6098 1.6099
2.65 .8477 .8520 2.90 .9571 .9597 4.5 1.6532 1.6533

2.66 .8521 .8563 2.91 .9615 .9641 4.6 1.6967 1.6968
2.67 .8564 .8606 2.92 .9658 .9684 4.7 1.7401 1.7402
2.68 .8608 .8649 2.93 .9702 .9727 4.8 1.7836 1.7836
2.69 .8652 .8692 2.94 .9746 .9770 4.9 1.8270 1.8270
2.70 .8696 .8735 2.95 .9789 .9813 5.0 1.8704 1.8705

2.71 .8740 .8778 2.96 .9833 .9856 6.0 2.3047 2.3047
2.72 .8784 .8821 2.97 .9877 .9900 7.0 2.7390 2.7390
2.73 .8827 .8864 2.98 .9920 .9943 8.0 3.1733 3.1733
2.74 .8871 .8907 2.99 .9964 .9986 9.0 3.6076 3.6076
2.75 0.8915 0.8951 3.00 1.0008 1.0029 10.0 4.0419 4.0419

APPENDIX 284
TABLE IV.—Roots of Bessel’s Functions.
x x x x
for J0 (x) = 0 for J1 (x) = 0 for J0 (x) = 0 for J1 (x) = 0
π π π π
1 0.7655 1.2197 7 6.7519 7.2448

2 1.7571 2.2330 8 7.7516 8.2454
3 2.7546 3.2383 9 8.7514 9.2459
4 3.7534 4.2411 10 9.7513 10.2463
5 4.7527 5.2428 11 10.7512 11.2466
6 5.7522 6.2439 12 11.7511 12.2469
TABLE V.—Roots of Jn (x) = 0.
n=0 n=1 n=2 n=3 n=4 n=5
1 2.405 3.832 5.135 6.379 7.586 8.780

2 5.520 7.016 8.417 9.760 11.064 12.339
3 8.654 10.173 11.620 13.017 14.373 15.700
4 11.792 13.323 14.796 16.224 17.616 18.982
5 14.931 16.470 17.960 19.410 20.827 22.220
6 18.071 19.616 21.117 22.583 24.018 25.431
7 21.212 22.760 24.270 25.749 27.200 28.628
8 24.353 25.903 27.421 28.909 30.371 31.813
9 27.494 29.047 30.571 32.050 33.512 34.983
APPENDIX 285
TABLE VI.— Bessel’s Functions.
x J0 (x) J1 (x) x J0 (x) J1 (x) x J0 (x) J1 (x)
0.0 1.0000 0.0000 5.0 −.1776 −.3276 10.0 −.2459 .0435

0.1 .9975 .0499 5.1 −.1443 −.3371 10.1 −.2490 .0184
0.2 .9900 .0995 5.2 −.1103 −.3432 10.2 −.2496 −.0066
0.3 .9776 .1483 5.3 −.0758 −.3460 10.3 −.2477 −.0313
0.4 .9604 .1960 5.4 −.0412 −.3453 10.4 −.2434 −.0555
0.5 .9385 .2423 5.5 −.0068 −.3414 10.5 −.2366 −.0789

0.6 .9120 .2867 5.6 .0270 −.3343 10.6 −.2276 −.1012
0.7 .8812 .3290 5.7 .0599 −.3241 10.7 −.2164 −.1224
0.8 .8463 .3688 5.8 .0917 −.3110 10.8 −.2032 −.1422
0.9 .8075 .4060 5.9 .1220 −.2951 10.9 −.1881 −.1604
1.0 .7652 .4401 6.0 .1506 −.2767 11.0 −.1712 −.1768

1.1 .7196 .4709 6.1 .1773 −.2559 11.1 −.1528 −.1913
1.2 .6711 .4983 6.2 .2017 −.2329 11.2 −.1330 −.2039
1.3 .6201 .5220 6.3 .2238 −.2081 11.3 −.1121 −.2143
1.4 .5669 .5419 6.4 .2433 −.1816 11.4 −.0902 −.2225
1.5 .5118 .5579 6.5 .2601 −.1538 11.5 −.0677 −.2284

1.6 .4554 .5699 6.6 .2740 −.1250 11.6 −.0446 −.2320
1.7 .3980 .5778 6.7 .2851 −.0953 11.7 −.0213 −.2333
1.8 .3400 .5815 6.8 .2931 −.0652 11.8 .0020 −.2323
1.9 .2818 .5812 6.9 .2981 −.0349 11.9 .0250 −.2290
2.0 .2239 .5767 7.0 .3001 −.0047 12.0 .0477 −.2234

2.1 .1666 .5683 7.1 .2991 .0252 12.1 .0697 −.2157
2.2 .1104 .5560 7.2 .2951 .0543 12.2 .0908 −.2060
2.3 .0555 .5399 7.3 .2882 .0826 12.3 .1108 −.1943
2.4 .0025 .5202 7.4 .2786 .1096 12.4 .1296 −.1807
2.5 −.0484 .4971 7.5 .2663 .1352 12.5 .1469 −.1655

2.6 −.0968 .4708 7.6 .2516 .1592 12.6 .1626 −.1487
2.7 −.1424 .4416 7.7 .2346 .1813 12.7 .1766 −.1307
2.8 −.1850 .4097 7.8 .2154 .2014 12.8 .1887 −.1114
2.9 −.2243 .3754 7.9 .1944 .2192 12.9 .1988 −.0912
3.0 −.2601 .3391 8.0 .1717 .2346 13.0 .2069 −.0703

3.1 −.2921 .3009 8.1 .1475 .2476 13.1 .2129 −.0489
3.2 −.3202 .2613 8.2 .1222 .2580 13.2 .2167 −.0271
3.3 −.3443 .2207 8.3 .0960 .2657 13.3 .2183 −.0052
3.4 −.3643 .1792 8.4 .0692 .2708 13.4 .2177 .0166
3.5 −.3801 .1374 8.5 .0419 .2731 13.5 .2150 .0380

3.6 −.3918 .0955 8.6 .0146 .2728 13.6 .2101 .0590
3.7 −.3992 .0538 8.7 −.0125 .2697 13.7 .2032 .0791
3.8 −.4026 .0128 8.8 −.0392 .2641 13.8 .1943 .0984
3.9 −.4018 −.0272 8.9 −.0653 .2559 13.9 .1836 .1166
4.0 −.3972 −.0660 9.0 −.0903 .2453 14.0 .1711 .1334

4.1 −.3887 −.1033 9.1 −.1142 .2324 14.1 .1570 .1488
4.2 −.3766 −.1386 9.2 −.1367 .2174 14.2 .1414 .1626
4.3 −.3610 −.1719 9.3 −.1577 .2004 14.3 .1245 .1747
4.4 −.3423 −.2028 9.4 −.1768 .1816 14.4 .1065 .1850
4.5 −.3205 −.2311 9.5 −.1939 .1613 14.5 .0875 .1934

4.6 −.2961 −.2566 9.6 −.2090 .1395 14.6 .0679 .1999
4.7 −.2693 −.2791 9.7 −.2218 .1166 14.7 .0476 .2043
4.8 −.2404 −.2985 9.8 −.2323 .0928 14.8 .0271 .2066
4.9 −.2097 −.3147 9.9 −.2403 .0684 14.9 .0064 .2069
5.0 −.1776 −.3276 10.0 −.2459 .0435 15.0 −.0142 .2051

ANNOUNCEMENTS
APPLICATIONS OF THE CALCULUS
TO MECHANICS
By E. R. Hedrick, Professor of Mathematics in the University of Missouri,
and O. D. Kellogg, Assistant Professor of Mathematics
in the University of Missouri
8vo, cloth, 116 pages, with diagrams, $1.25
T HIS book presents a completed summary of those parts of mechanics which

occur as applications of the calculus.
Although intended primarily as a supplement to the usual standard course
in calculus, it may be used independently as a text for a short course on the
mathematical side of mechanics, if the time allotted to the former study is not
sufficient to include this work.
As a review it fastens in the student’s mind the notions of mechanics previ-
ously gained. It aims also to present these topics in a new light, as articulated
portions of one general theory, and thus to make mechanics seem an integral
subject.
As a preparation for more extended courses in mechanics, or indeed for
courses dealing with any applications of the calculus, the material presented is
valuable in showing concretely how the theoretical ideas of this subject are used
in specific practical applications.
The course outlined in the book is the result of a number of years’ experience
in presenting this material to classes in the calculus, both at the University of
Missouri and elsewhere. The text itself is a modification of a similar text written
by Professor Hedrick and published in mimeograph for the use of students at
the Sheffield Scientific School.
GINN AND COMPANY Publishers

ELECTRICAL PROBLEMS
By William L. Hooper, Professor of Electrical Engineering,
Tufts College, Mass., and Roy T. Wells, in Charge of
Electrical Engineering, State University of Iowa
8vo, cloth, 170 pages, with diagrams, $1.25
T HE work contains sets of problems typical of those met with in electrical

laboratory and engineering practice, with very brief treatment of the methods of
solution. It is intended to meet the need for a collection of numerical problems
in the classes of colleges and technical schools. It is made up as follows:
1. Twelve sets of problems and calculations on combinations of electro-motive
forces and resistances in series and multiple grouping; distribution and fall of po-
tential in railway and lighting circuits, inductance of coils, capacity of condensers,
thermo-electricity, electro-chemistry; and output and efficiency of batteries, genera-
tors, motors, etc.
2. Four sets of problems on combinations of alternating electromotive forces and
currents and the impedance of circuits with constant and with varying values of resis-
tance, inductance, capacity, and frequency.
3. Five sets of problems on calculating and making winding tables and drawings
for direct and alternating current armatures, armature reactions, field windings, etc.
4. Problems on winding and operation of transformers, rotary converters, and
induction motors, and on testing of dynamos and transmission of power.
Answers are given to all problems, many in the form of curves showing the
effect of varying the various constants involved, such as temperature, frequency,
capacity, resistance, and inductance. The text contains about forty explanatory
diagrams.
GINN & COMPANY Publishers

THEORETICAL MECHANICS
By Percey F. Smith, Professor of Mathematics in the Sheffield Scientific
School, Yale University, and William R. Longley, Assistant Professor of
Mathematics in the Sheffield Scientific School, Yale University
8vo, semiflexible cloth, 288 pages, with diagrams, $2.50
T HIS book is intended for use in courses in mechanics, which, as in many

colleges and technical schools, are based upon the calculus. For the convenience
of the student, formulas from analytic geometry and the calculus, and a table
of integrals, are included.
The first chapter deals with centers of gravity and moments of inertia. This
is followed by chapters on kinematics and kinetics of a particle (including im-
pact), motion in various fields of force (constant field, central field, harmonic
field), kinetics of a system of particles, potential, motion in a resisting medium,
dynamics of a rigid body, including uniplanar motion, and equilibrium of copla-
nar forces.
The fundamental problem—to determine the motion due to a given force
under given initial conditions—is thoroughly discussed.
The equations of motion obtained by integration of the force equations have,
however, been studied in a previous chapter, and the student is therefore cog-
nizant immediately of the significance of his results.
Emphasis is laid everywhere in the solution of problems upon the general
application of the force equations, the energy equation, and the impulse equa-
tion.
The problems are carefully selected, and numerous illustrative examples are
worked out in the text.

BOOKS ON HIGHER MATHEMATICS
BYERLY: ELEMENTARY TREATISE ON FOURIER’S SERIES, AND
SPHERICAL, CYLINDRICAL, AND ELLIPSOIDAL HARMONICS,
WITH APPLICATIONS TO PROBLEMS IN MATHEMATICAL
PHYSICS. $3.00.
An introduction to the treatment of some of the important linear partial

differential equations which lie at the foundation of modern theories in physics.
HEDRICK: GOURSAT’S COURSE IN MATHEMATICAL ANALYSIS, VOL-

UME I. $4.00.
The French edition of this work at once attracted widespread attention on

account of the clearness of its style and the thoroughness and rigor with which
the subject matter was presented.
HEDRICK AND KELLOGG: APPLICATIONS OF THE CALCULUS TO ME-

CHANICS. $1.25.
A completed summary of those parts of mechanics which occur as appli-

cations of the calculus.
PEIRCE, B. O.: ELEMENTS OF THE THEORY OF THE NEWTONIAN PO-

TENTIAL FUNCTION (Third, Revised, and Enlarged Edition). $2.50.
Intended for general students who wish to acquire a sound knowledge of

the properties of the Newtonian Potential Function. Nearly four hundred mis-
cellaneous problems have been added in this new edition.
PEIRCE, J. M.: ELEMENTS OF LOGARITHMS. 50 cents.
Designed to give the student a complete and accurate knowledge of the

nature and use of logarithms.
TAYLOR, J. M.: ELEMENTS OF THE DIFFERENTIAL AND INTEGRAL

CALCULUS (Revised Edition, Enlarged and Entirely Rewritten). With
Examples and Applications. $2.00.
Presents clearly, scientifically, and in their true relations the three common
methods in the calculus.
MATHEMATICAL TEXTS. Edited by Percey F. Smith, of the Sheffield Scien-
tific School of Yale University.
Eisenhart: Differential Geometry of Curves and Surfaces. $4.50.
Granville: Plane Trigonometry and Tables. $1.00.
Granville: Plane and Spherical Trigonometry, without Tables. $1.00.
Granville: Plane and Spherical Trigonometry, and Tables. $1.25.
Granville: Elements of Differential and Integral Calculus. $2.50.
Granville: Logarithmic Tables. 50 cents.
Hawkes: Advanced Algebra. $1.40.
Smith and Gale: Elements of Analytic Geometry. $2.00.
Smith and Gale: Introduction to Analytic Geometry. $1.25.

Typographical Errors corrected in Project Gutenberg edition
wπ
p. 43 Equation (3), the second was not printed.
0
2
p. 103 Example 1, the = sign on the second line (beginning ) was not
π
printed.
mπy nπy
p. 139 Example 6, second sin term sin in original, amended to sin .
b b
p. 240 Equation (8), Dρ23 V was printed as Dρ3 V .
End of the Project Gutenberg EBook of An Elementary Treatise on Fourier’s
Series and Spherical, Cylindrical, and Ellipsoidal Harmonics, by William
Elwood Byerly
*** END OF THIS PROJECT GUTENBERG EBOOK TREATISE ON FOURIER’S SERIES ***
***** This file should be named 29779-pdf.pdf or 29779-pdf.zip *****

This and all associated files of various formats will be found in:
http://www.gutenberg.org/2/9/7/7/29779/
Produced by Laura Wisewell, Carl Hudkins, Keith Edkins and

the Online Distributed Proofreading Team at
http://www.pgdp.net (The original copy of this book was
generously made available for scanning by the Department
of Mathematics at the University of Glasgow.)
Updated editions will replace the previous one--the old editions

will be renamed.
Creating the works from public domain print editions means that no
one owns a United States copyright in these works, so the Foundation
(and you!) can copy and distribute it in the United States without
permission and without paying copyright royalties. Special rules,
set forth in the General Terms of Use part of this license, apply to
copying and distributing Project Gutenberg-tm electronic works to
protect the PROJECT GUTENBERG-tm concept and trademark. Project
Gutenberg is a registered trademark, and may not be used if you
charge for the eBooks, unless you receive specific permission. If you
do not charge anything for copies of this eBook, complying with the
rules is very easy. You may use this eBook for nearly any purpose
such as creation of derivative works, reports, performances and
research. They may be modified and printed and given away--you may do
practically ANYTHING with public domain eBooks. Redistribution is
subject to the trademark license, especially commercial
redistribution.
*** START: FULL LICENSE ***
THE FULL PROJECT GUTENBERG LICENSE

PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK
To protect the Project Gutenberg-tm mission of promoting the free

distribution of electronic works, by using or distributing this work
(or any other work associated in any way with the phrase "Project
Gutenberg"), you agree to comply with all the terms of the Full Project
Gutenberg-tm License (available with this file or online at
http://gutenberg.org/license).
Section 1. General Terms of Use and Redistributing Project Gutenberg-tm
electronic works
1.A. By reading or using any part of this Project Gutenberg-tm

electronic work, you indicate that you have read, understand, agree to
and accept all the terms of this license and intellectual property
(trademark/copyright) agreement. If you do not agree to abide by all
the terms of this agreement, you must cease using and return or destroy
all copies of Project Gutenberg-tm electronic works in your possession.
If you paid a fee for obtaining a copy of or access to a Project
Gutenberg-tm electronic work and you do not agree to be bound by the
terms of this agreement, you may obtain a refund from the person or
entity to whom you paid the fee as set forth in paragraph 1.E.8.
1.B. "Project Gutenberg" is a registered trademark. It may only be

used on or associated in any way with an electronic work by people who
agree to be bound by the terms of this agreement. There are a few
things that you can do with most Project Gutenberg-tm electronic works
even without complying with the full terms of this agreement. See
paragraph 1.C below. There are a lot of things you can do with Project
Gutenberg-tm electronic works if you follow the terms of this agreement
and help preserve free future access to Project Gutenberg-tm electronic
works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation ("the Foundation"

or PGLAF), owns a compilation copyright in the collection of Project
Gutenberg-tm electronic works. Nearly all the individual works in the
collection are in the public domain in the United States. If an
individual work is in the public domain in the United States and you are
located in the United States, we do not claim a right to prevent you from
copying, distributing, performing, displaying or creating derivative
works based on the work as long as all references to Project Gutenberg
are removed. Of course, we hope that you will support the Project
Gutenberg-tm mission of promoting free access to electronic works by
freely sharing Project Gutenberg-tm works in compliance with the terms of
this agreement for keeping the Project Gutenberg-tm name associated with
the work. You can easily comply with the terms of this agreement by
keeping this work in the same format with its attached full Project
Gutenberg-tm License when you share it without charge with others.
1.D. The copyright laws of the place where you are located also govern
what you can do with this work. Copyright laws in most countries are in
a constant state of change. If you are outside the United States, check
the laws of your country in addition to the terms of this agreement
before downloading, copying, displaying, performing, distributing or
creating derivative works based on this work or any other Project
Gutenberg-tm work. The Foundation makes no representations concerning
the copyright status of any work in any country outside the United
States.
1.E. Unless you have removed all references to Project Gutenberg:
1.E.1. The following sentence, with active links to, or other immediate
access to, the full Project Gutenberg-tm License must appear prominently
whenever any copy of a Project Gutenberg-tm work (any work on which the
phrase "Project Gutenberg" appears, or with which the phrase "Project
Gutenberg" is associated) is accessed, displayed, performed, viewed,
copied or distributed:
This eBook is for the use of anyone anywhere at no cost and with
almost no restrictions whatsoever. You may copy it, give it away or
re-use it under the terms of the Project Gutenberg License included
with this eBook or online at www.gutenberg.org
1.E.2. If an individual Project Gutenberg-tm electronic work is derived

from the public domain (does not contain a notice indicating that it is
posted with permission of the copyright holder), the work can be copied
and distributed to anyone in the United States without paying any fees
or charges. If you are redistributing or providing access to a work
with the phrase "Project Gutenberg" associated with or appearing on the
work, you must comply either with the requirements of paragraphs 1.E.1
through 1.E.7 or obtain permission for the use of the work and the
Project Gutenberg-tm trademark as set forth in paragraphs 1.E.8 or
1.E.9.
1.E.3. If an individual Project Gutenberg-tm electronic work is posted

with the permission of the copyright holder, your use and distribution
must comply with both paragraphs 1.E.1 through 1.E.7 and any additional
terms imposed by the copyright holder. Additional terms will be linked
to the Project Gutenberg-tm License for all works posted with the
permission of the copyright holder found at the beginning of this work.
1.E.4. Do not unlink or detach or remove the full Project Gutenberg-tm

License terms from this work, or any files containing a part of this
work or any other work associated with Project Gutenberg-tm.
1.E.5. Do not copy, display, perform, distribute or redistribute this

electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1 with
active links or immediate access to the full terms of the Project
Gutenberg-tm License.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form, including any
word processing or hypertext form. However, if you provide access to or
distribute copies of a Project Gutenberg-tm work in a format other than
"Plain Vanilla ASCII" or other format used in the official version
posted on the official Project Gutenberg-tm web site (www.gutenberg.org),
you must, at no additional cost, fee or expense to the user, provide a
copy, a means of exporting a copy, or a means of obtaining a copy upon
request, of the work in its original "Plain Vanilla ASCII" or other
form. Any alternate format must include the full Project Gutenberg-tm
License as specified in paragraph 1.E.1.
1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg-tm works
unless you comply with paragraph 1.E.8 or 1.E.9.
1.E.8. You may charge a reasonable fee for copies of or providing

access to or distributing Project Gutenberg-tm electronic works provided
that
- You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg-tm works calculated using the method
you already use to calculate your applicable taxes. The fee is
owed to the owner of the Project Gutenberg-tm trademark, but he
has agreed to donate royalties under this paragraph to the
Project Gutenberg Literary Archive Foundation. Royalty payments
must be paid within 60 days following each date on which you
prepare (or are legally required to prepare) your periodic tax
returns. Royalty payments should be clearly marked as such and
sent to the Project Gutenberg Literary Archive Foundation at the
address specified in Section 4, "Information about donations to
the Project Gutenberg Literary Archive Foundation."
- You provide a full refund of any money paid by a user who notifies
you in writing (or by e-mail) within 30 days of receipt that s/he
does not agree to the terms of the full Project Gutenberg-tm
License. You must require such a user to return or
destroy all copies of the works possessed in a physical medium
and discontinue all use of and all access to other copies of
Project Gutenberg-tm works.
- You provide, in accordance with paragraph 1.F.3, a full refund of any

money paid for a work or a replacement copy, if a defect in the
electronic work is discovered and reported to you within 90 days
of receipt of the work.
- You comply with all other terms of this agreement for free
distribution of Project Gutenberg-tm works.
1.E.9. If you wish to charge a fee or distribute a Project Gutenberg-tm

electronic work or group of works on different terms than are set
forth in this agreement, you must obtain permission in writing from
both the Project Gutenberg Literary Archive Foundation and Michael
Hart, the owner of the Project Gutenberg-tm trademark. Contact the
Foundation as set forth in Section 3 below.
1.F.
1.F.1. Project Gutenberg volunteers and employees expend considerable

effort to identify, do copyright research on, transcribe and proofread
public domain works in creating the Project Gutenberg-tm
collection. Despite these efforts, Project Gutenberg-tm electronic
works, and the medium on which they may be stored, may contain
"Defects," such as, but not limited to, incomplete, inaccurate or
corrupt data, transcription errors, a copyright or other intellectual
property infringement, a defective or damaged disk or other medium, a
computer virus, or computer codes that damage or cannot be read by
your equipment.
1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except for the "Right

of Replacement or Refund" described in paragraph 1.F.3, the Project
Gutenberg Literary Archive Foundation, the owner of the Project
Gutenberg-tm trademark, and any other party distributing a Project
Gutenberg-tm electronic work under this agreement, disclaim all
liability to you for damages, costs and expenses, including legal
fees. YOU AGREE THAT YOU HAVE NO REMEDIES FOR NEGLIGENCE, STRICT
LIABILITY, BREACH OF WARRANTY OR BREACH OF CONTRACT EXCEPT THOSE
PROVIDED IN PARAGRAPH F3. YOU AGREE THAT THE FOUNDATION, THE
TRADEMARK OWNER, AND ANY DISTRIBUTOR UNDER THIS AGREEMENT WILL NOT BE
LIABLE TO YOU FOR ACTUAL, DIRECT, INDIRECT, CONSEQUENTIAL, PUNITIVE OR
INCIDENTAL DAMAGES EVEN IF YOU GIVE NOTICE OF THE POSSIBILITY OF SUCH
DAMAGE.
1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If you discover a

defect in this electronic work within 90 days of receiving it, you can
receive a refund of the money (if any) you paid for it by sending a
written explanation to the person you received the work from. If you
received the work on a physical medium, you must return the medium with
your written explanation. The person or entity that provided you with
the defective work may elect to provide a replacement copy in lieu of a
refund. If you received the work electronically, the person or entity
providing it to you may choose to give you a second opportunity to
receive the work electronically in lieu of a refund. If the second copy
is also defective, you may demand a refund in writing without further
opportunities to fix the problem.
1.F.4. Except for the limited right of replacement or refund set forth
in paragraph 1.F.3, this work is provided to you ’AS-IS’ WITH NO OTHER
WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
WARRANTIES OF MERCHANTIBILITY OR FITNESS FOR ANY PURPOSE.
1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of damages.
If any disclaimer or limitation set forth in this agreement violates the
law of the state applicable to this agreement, the agreement shall be
interpreted to make the maximum disclaimer or limitation permitted by
the applicable state law. The invalidity or unenforceability of any
provision of this agreement shall not void the remaining provisions.
1.F.6. INDEMNITY - You agree to indemnify and hold the Foundation, the
trademark owner, any agent or employee of the Foundation, anyone
providing copies of Project Gutenberg-tm electronic works in accordance
with this agreement, and any volunteers associated with the production,
promotion and distribution of Project Gutenberg-tm electronic works,
harmless from all liability, costs and expenses, including legal fees,
that arise directly or indirectly from any of the following which you do
or cause to occur: (a) distribution of this or any Project Gutenberg-tm
work, (b) alteration, modification, or additions or deletions to any
Project Gutenberg-tm work, and (c) any Defect you cause.
Section 2. Information about the Mission of Project Gutenberg-tm
Project Gutenberg-tm is synonymous with the free distribution of

electronic works in formats readable by the widest variety of computers
including obsolete, old, middle-aged and new computers. It exists
because of the efforts of hundreds of volunteers and donations from
people in all walks of life.
Volunteers and financial support to provide volunteers with the

assistance they need, are critical to reaching Project Gutenberg-tm’s
goals and ensuring that the Project Gutenberg-tm collection will
remain freely available for generations to come. In 2001, the Project
Gutenberg Literary Archive Foundation was created to provide a secure
and permanent future for Project Gutenberg-tm and future generations.
To learn more about the Project Gutenberg Literary Archive Foundation
and how your efforts and donations can help, see Sections 3 and 4
and the Foundation web page at http://www.pglaf.org.
Section 3. Information about the Project Gutenberg Literary Archive

Foundation
The Project Gutenberg Literary Archive Foundation is a non profit

501(c)(3) educational corporation organized under the laws of the
state of Mississippi and granted tax exempt status by the Internal
Revenue Service. The Foundation’s EIN or federal tax identification
number is 64-6221541. Its 501(c)(3) letter is posted at
http://pglaf.org/fundraising. Contributions to the Project Gutenberg
Literary Archive Foundation are tax deductible to the full extent
permitted by U.S. federal laws and your state’s laws.
The Foundation’s principal office is located at 4557 Melan Dr. S.

Fairbanks, AK, 99712., but its volunteers and employees are scattered
throughout numerous locations. Its business office is located at
809 North 1500 West, Salt Lake City, UT 84116, (801) 596-1887, email
[email protected]. Email contact links and up to date contact
information can be found at the Foundation’s web site and official
page at http://pglaf.org
For additional contact information:

Dr. Gregory B. Newby
Chief Executive and Director
[email protected]
Section 4. Information about Donations to the Project Gutenberg

Literary Archive Foundation
Project Gutenberg-tm depends upon and cannot survive without wide

spread public support and donations to carry out its mission of
increasing the number of public domain and licensed works that can be
freely distributed in machine readable form accessible by the widest
array of equipment including outdated equipment. Many small donations
($1 to $5,000) are particularly important to maintaining tax exempt
status with the IRS.
The Foundation is committed to complying with the laws regulating

charities and charitable donations in all 50 states of the United
States. Compliance requirements are not uniform and it takes a
considerable effort, much paperwork and many fees to meet and keep up
with these requirements. We do not solicit donations in locations
where we have not received written confirmation of compliance. To
SEND DONATIONS or determine the status of compliance for any
particular state visit http://pglaf.org
While we cannot and do not solicit contributions from states where we

have not met the solicitation requirements, we know of no prohibition
against accepting unsolicited donations from donors in such states who
approach us with offers to donate.
International donations are gratefully accepted, but we cannot make

any statements concerning tax treatment of donations received from
outside the United States. U.S. laws alone swamp our small staff.
Please check the Project Gutenberg Web pages for current donation
methods and addresses. Donations are accepted in a number of other
ways including checks, online payments and credit card donations.
To donate, please visit: http://pglaf.org/donate
Section 5. General Information About Project Gutenberg-tm electronic

works.
Professor Michael S. Hart is the originator of the Project Gutenberg-tm

concept of a library of electronic works that could be freely shared
with anyone. For thirty years, he produced and distributed Project
Gutenberg-tm eBooks with only a loose network of volunteer support.
Project Gutenberg-tm eBooks are often created from several printed

editions, all of which are confirmed as Public Domain in the U.S.
unless a copyright notice is included. Thus, we do not necessarily
keep eBooks in compliance with any particular paper edition.
Most people start at our Web site which has the main PG search facility:
http://www.gutenberg.org
This Web site includes information about Project Gutenberg-tm,

including how to make donations to the Project Gutenberg Literary
Archive Foundation, how to help produce our new eBooks, and how to
subscribe to our email newsletter to hear about new eBooks.

An Elementary Treatise On Fourier Series

Uploaded by

Copyright:

Available Formats

An Elementary Treatise On Fourier Series

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

An Elementary Treatise On Fourier Series

Uploaded by

Copyright:

Available Formats

The Project Gutenberg EBook of An Elementary Treatise on Fourier’s Series

and Spherical, Cylindrical, and Ellipsoidal Harmonics, by William Elwood

Title: An Elementary Treatise on Fourier’s Series and Spherical,

Author: William Elwood Byerly

Release Date: August 19, 2009 [EBook #29779]

Character set encoding: ISO-8859-1

*** START OF THIS PROJECT GUTENBERG EBOOK TREATISE ON FOURIER’S SERIES

Produced by Laura Wisewell, Carl Hudkins, Keith Edkins and

SPHERICAL, CYLINDRICAL, AND ELLIPSOIDAL

APPLICATIONS TO PROBLEMS IN MATHEMATICAL PHYSICS.

GINN & COMPANY

ALL RIGHTS RESERVED.

About ten years ago I gave a course of lectures on Trigonometric Series,

ANALYTICAL TABLE OF CONTENTS.

Art. 1. List of some important homogeneous linear partial differential equa-

Development in Trigonometric Series 30–55

Arts. 19–22. Determination of the coefficients of n terms of a sine series so

method of solving the problem of development in series involving sines of whole

Convergence of Fourier’s Series 56–69

Solution of Problems in Physics by the Aid of Fourier’s Inte-

Arts. 44–48. Logarithmic Potential. Flow of electricity in an infinite plane,

Potential Function given along the perimeter of a rectangle. Examples.—Arts.

Miscellaneous Problems 135–143

I. Logarithmic Potential. Polar Coördinates.—II. Potential Function in

Art. 74. Recapitulation. Surface Zonal Harmonics (Legendrians). Zonal

symmetry about a diameter. Examples.—Art. 95. Development of a power

Spherical Harmonics 196–219

Arts. 101–102. Particular Solutions of Laplace’s Equation obtained. As-

Cylindrical Harmonics (Bessel’s Functions) 220–238

Art. 120. Recapitulation. Cylindrical Harmonics (Bessel’s Functions) of the

Example.—Art. 126. Development in Cylindrical Harmonic Series. Formulas

Laplace’s Equation in Curvilinear Coördinates. Ellipsoidal

Arts. 130–131. Orthogonal Curvilinear Coördinates in general. Laplace’s

Historical Summary 267–274

Table III. Hyperbolic Functions 280

1. In many important problems in mathematical physics we are obliged

Dt u = a2 (Dx2 u + Dy2 u + Dz2 u), 1 [I]

Dx2 u + Dy2 u = 0. [III]

In polar or spherical coördinates [I] is less simple, it is

Dt2 y = a2 Dx2 y; [VIII]

for instance, in considering the transverse or the longitudinal vibrations of a

Dx2 V +Dy2 V + Dz2 V = 0 [XII]

2. A differential equation is an equation containing derivatives or differ-

3. If a problem requiring for its solution the solving of a differential equa-

4. A differential equation is linear when it would be of the first degree

5. If a value of the dependent variable has been found which satisfies a

6. It is generally possible to get by some simple device particular solutions

7. As an illustration, let us take Fourier’s problem of the permanent state

u=0 when x = 0 (1)

We shall begin by getting a particular solution of [III], and we shall use a

Assume2 u = eαy+βx , where α and β are constants, substitute in [III] and

u = eαy cos αx (5)

u = eαy sin αx (6)

Consider u = eαy sin αx. (6)

It is zero when x = 0 for all values of α. It is zero when x = π if α is a whole

u = A1 e−y sin x + A2 e−2y sin 2x + A3 e−3y sin 3x + A4 e−4y sin 4x + · · · (7)

A1 , A2 , A3 , A4 , &c., being undetermined constants.

u = A1 sin x + A2 sin 2x + A3 sin 3x + A4 sin 4x + · · · . (8)

in the equation, and is justified if it leads to√a solution.