% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % % 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 *** % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \def\ebook{29779} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% An Elementary Treatise on Fourier's Series, by William Elwood Byerly %% %% %% %% Packages and substitutions: %% %% %% %% book : Document class. %% %% amsmath: Basic AMS math package. %% %% amssymb: Basic AMS symbols %% %% tabularx Tabulation. %% %% graphicx Basic graphics for images. %% %% inputenc: Encoding %% %% needspace Conditional new page %% %% wrapfig textflow round images %% %% verbatim Preformated text %% %% wasysym WASY2 fonts (upright integrals) %% %% %% %% PDF Pages: 309 %% %% %% %% 144 underfull hboxes %% %% %% %% 14 includegraphics calls included as nnn.png %% %% %% %% Compile sequence: %% %% pdflatex x 2 %% %% %% %% Compile History: %% %% %% %% Jul 09: Laverock. Compiled with pdflatex: %% %% [pdfeTeX, Version 3.1415926-1.40.9 (MiKTeX 2.7)] %% %% %% %% %% %% August 2009: pglatex. %% %% Compile this project with: %% %% pdflatex 29779-t.tex ..... TWO times %% %% %% %% pdfTeXk, Version 3.141592-1.40.3 (Web2C 7.5.6) %% %% %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \documentclass[oneside]{book} \usepackage[reqno]{amsmath}[2000/07/18] % basic math support \usepackage{amssymb}[2002/01/22] % math symbols \usepackage{tabularx}[1999/01/07] % tabulation \usepackage{graphicx}[1999/02/16] % basic graphics for images. \usepackage[latin1]{inputenc}[2006/05/05] % encoding system \usepackage{needspace}[2003/02/18] % support testing for enough room at end of page for new section \usepackage{wrapfig}[2003/01/31] % textflow round images \usepackage{verbatim}[2003/08/22] % preformated text \usepackage[integrals]{wasysym}[2003/10/30] % WASY2 integrals \DeclareMathOperator*{\limit}{limit} \DeclareMathOperator{\ctn}{ctn} \DeclareMathOperator{\ctnh}{ctnh} \DeclareMathOperator{\csch}{csch} \DeclareMathOperator{\sech}{sech} \DeclareMathOperator{\sn}{sn} \DeclareMathOperator{\cn}{cn} \DeclareMathOperator{\dn}{dn} \DeclareMathOperator{\gd}{gd} \newcommand{\tablestrut}{\vphantom{\fbox{$\dfrac{0}{0}$}}} \newcommand\Example[1]{\Needspace*{5\baselineskip}\vspace{1ex}\begin{center}\textsc{Example#1.}\end{center}} \newcommand\EXAMPLE[1]{\Needspace*{5\baselineskip}\vspace{1ex}\begin{center}{EXAMPLE#1.}\end{center}} \newcommand\smallrom[1]{[{\scriptsize #1}]} \newcommand\smallromr[1]{({\scriptsize #1})} \newcommand\mypara[1]{\vspace{\baselineskip}\Needspace*{3\baselineskip}\textbf{#1}\quad} \newcommand{\strxx}{\rule{0in}{1.9ex}} % strut for sqrt with exponents inside \newcounter{wrapwidth} \newcount \Zw \newcount \Zh % text flowed image - file name, width, height pixels, vertical offset (normally -24, -12 for top of page) \newcommand\pngright[4]{ \Zw=#2 \divide \Zw by 5 \Zh=#3 \divide \Zh by 60 \advance\Zh by 1 \setcounter{wrapwidth}{\Zw} \begin{wrapfigure}[\Zh]{r}{\value{wrapwidth}pt} \begin{center} \vspace{#4pt} \includegraphics*[width=\Zw pt]{images/#1} \end{center} \end{wrapfigure}} % text flowed image - file name, width, height lines, vertical offset (normally -24, -12 for top of page) \newcommand\pngrightl[4]{ \Zw=#2 \divide \Zw by 5 \Zh=#3 \setcounter{wrapwidth}{\Zw} \begin{wrapfigure}[\Zh]{r}{\value{wrapwidth}pt} \begin{center} \vspace{#4pt} \includegraphics*[width=\Zw pt]{images/#1} \end{center} \end{wrapfigure}} % centred image - file name, width pixels \newcommand\pngcent[3]{ \Zw=#2 \divide \Zw by 5 \begin{center} \includegraphics*[width=\Zw pt]{images/#1} \end{center}} %simulated chapter heading & null markright command \newcommand\mychap[2]{ \setcounter{footnote}{0} \newpage \medskip\begin{center}{\large #1}\par\bigskip{#2}\end{center} \markright{} } % Macro & command to spread a slightly short line right across page. \def\sspreadout#1 {\def\tmp{#1}% \ifx\tmp\empty \def\next{} \else \def\next{ {{\hfill}#1}\sspreadout}% \fi \next} \newcommand\spreadout[1]{\sspreadout#1 } % For sensible insertion of boilerplate/licence % overlong lines will wrap and be indented 0.25in % and text is set in "small" size \makeatletter \def\@makeschapterhead#1{% \vspace*{10\p@}% {\parindent \z@ \centering \normalfont \interlinepenalty\@M \huge \bfseries #1\par\nobreak \vskip 20\p@ }} \renewcommand*\l@section{\@dottedtocline{1}{0pt}{2.3em}} \renewcommand\@pnumwidth{2.55em} \def\@xobeysp{~\hfil\discretionary{}{\kern\z@}{}\hfilneg} \renewcommand\verbatim@processline{\leavevmode \null\kern-0.25in\the\verbatim@line\par} \addto@hook\every@verbatim{\@totalleftmargin0.25in\small} \makeatother \pagestyle{empty} \listfiles \begin{document} \begin{verbatim} 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.) \end{verbatim} % -----File: 001.png \newpage % title page \begin{center} {\Large AN ELEMENTARY TREATISE}\\[3ex] {\scriptsize ON}\\[3ex] {\Huge FOURIER'S SERIES}\\[3ex] {\scriptsize AND}\\[3ex] {\Large SPHERICAL, CYLINDRICAL, AND ELLIPSOIDAL\\ HARMONICS,}\\[3ex] {\scriptsize WITH}\\[3ex] APPLICATIONS TO PROBLEMS IN MATHEMATICAL PHYSICS.\\[5ex] {\scriptsize BY}\\ WILLIAM ELWOOD BYERLY, \textsc{Ph.D.},\\ {\scriptsize PROFESSOR OF MATHEMATICS IN HARVARD UNIVERSITY.}\\[4ex] \hspace{1.5in} \hrulefill \hspace*{1.7in} \vspace{5ex} {\Large GINN \& COMPANY}\\ BOSTON · NEW YORK · CHICAGO · LONDON\\ \end{center} % -----File: 002.png \bigskip\bigskip \begin{center} {\footnotesize Copyright, 1893,\\ By WILLIAM ELWOOD BYERLY.} \par\rule{5em}{0.5pt}\par {\scriptsize ALL RIGHTS RESERVED.} \end{center} \bigskip\bigskip \scriptsize \noindent \textsc{Transcriber's Note:} \emph{A few typographical errors have been corrected - these are noted at the end of the text.} \normalsize % -----File: 003.png \frontmatter \pagestyle{myheadings} \newpage \setcounter{footnote}{0} \newpage \medskip\begin{center}{\Large PREFACE.}\par\rule{5em}{0.5pt}\par\end{center} \markright{} About ten years ago I gave a course of lectures on Trigonometric Series, following closely the treatment of that subject in Riemann's ``Partielle Differentialgleichungen,'' 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 Potential Function'' (Boston, Ginn \& Co.), and the two sets of lectures form a course (Math.~10) given regularly at Harvard, and intended as a partial introduction 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 use the notation $D_x$ instead of $\frac{\delta}{\delta x}$ 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. \rightline{W.~E. BYERLY.\qquad} {\small \textsc{Cambridge, Mass.}, Sept.\ 1893.} % -----File: 004.png %[Blank Page] % -----File: 005.png \setcounter{footnote}{0} \newpage \medskip\begin{center}{\Large ANALYTICAL TABLE OF CONTENTS.}\par\rule{5em}{0.5pt}\par\end{center} \markright{} \begin{center}CHAPTER I.\end{center} \hfill \textsc{pages} \textsc{Introduction} \hfill \pageref{ch1start}--\pageref{ch1end}\\ \textsc{Art.}\ 1. List of some important homogeneous linear partial differential equations of Physics.---\textsc{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.---\textsc{Arts.}\ 5--6. Particular solutions of homogeneous linear differential equations may be combined into a more general solution. Need of development in terms of normal forms.---\textsc{Art.}\ 7. Problem: Permanent state of temperatures in a thin rectangular plate. Need of a development in sine series. Example.---\textsc{Art.}\ 8. Problem: Transverse vibrations of a stretched elastic string. A development in sine series suggested.---\textsc{Art.}\ 9. Problem: Potential function due to the attraction of a circular ring of small cross-section. Surface Zonal Harmonics (Legendre's Coefficients). Example.---\textsc{Art.}\ 10. Problem: Permanent state of temperatures in a solid sphere. Development in terms of Surface Zonal Harmonics suggested.---\textsc{Arts.}\ 11--12. Problem: Vibrations of a circular drumhead. Cylindrical Harmonics (Bessel's Functions). Recapitulation.---\textsc{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 independent variables. \textsc{Arts.}\ 14--15 Method of solving ordinary homogeneous linear differential equations by development in power series. Applications.---\textsc{Art.}\ 16. Application to Legendre's Equation. Several forms of general solution obtained. Zonal Harmonics of the second kind.---\textsc{Art.}\ 17. Application to Bessel's Equation. 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.---\textsc{Art.}\ 18. Method of obtaining the general solution of an ordinary linear differential equation of the second order from a given particular solution. Application to the equations considered in Arts.~14--17.\\[1ex] \begin{center}CHAPTER II.\end{center} \textsc{Development in Trigonometric Series} \hfill \pageref{ch2start}--\pageref{ch2end}\\ \textsc{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 values of $x$. Numerical example.---\textsc{Art.}\ 23. Problem of development in sine series treated as a limiting case of the problem just solved.---\textsc{Arts.}\ 24--25. Shorter method of solving the problem of development in series involving sines of whole multiples of the variable. Working rule deduced. Recapitulation.---\textsc{Art.}\ 26. A % -----File: 006.png few important sine developments obtained. Examples.---\textsc{Arts.}\ 27--28. Development in cosine series. Examples.---\textsc{Art.}\ 29. Sine series an odd function of the variable, cosine series an even function, and both series periodic functions.---\textsc{Art.}\ 30. Development in series involving both sines and cosines of whole multiples of the variable. Fourier's series. Examples.---\textsc{Art.}\ 31. Extension of the range within which the function and the series are equal. Examples.---\textsc{Art.}\ 32. Fourier's Integral obtained.\\[1ex] \markright{TABLE OF CONTENTS} \begin{center}CHAPTER III.\end{center} \textsc{Convergence of Fourier's Series} \hfill \pageref{ch3start}--\pageref{ch3end}\\ \textsc{Arts.}\ 33--36. The question of the convergence of the sine series for unity considered at length.---\textsc{Arts.}\ 37--38. Statement of the conditions which are sufficient to warrant the development of a function into a Fourier's series. Historical note. \textsc{Art.}\ 39. Graphical representation of successive approximations to a sine series. Properties of a Fourier's series inferred from the constructions.---\textsc{Arts.}\ 40--42. Investigation of the conditions under which a Fourier's series can be differentiated term by term.---\textsc{Art.}\ 43. Conditions under which a function can be expressed as a Fourier's Integral.\\[1ex] \begin{center}CHAPTER IV.\end{center} \textsc{Solution of Problems in Physics by the Aid of Fourier's Integrals and Fourier's Series } \hfill \pageref{ch4start}--\pageref{ch4end}\\ \textsc{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.---\textsc{Arts.}\ 49--52. One-dimensional flow of heat. Flow of heat in an infinite solid; in a solid with one plane face at the temperature 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.---\textsc{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.---\textsc{Arts.}\ 55--56. Transmission of a disturbance along an infinite stretched elastic string. Examples.---\textsc{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.---\textsc{Art.}\ 59. Potential Function given along the perimeter of a rectangle. Examples.---\textsc{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.---\textsc{Art.}\ 64. Motion of a stretched elastic string fastened at the ends. Steady vibration. Nodes. Examples.---\textsc{Art.}\ 65. Motion of a string in a resisting medium.---\textsc{Art.}\ 66. Flow of heat in a sphere whose surface is kept at a constant temperature.---\textsc{Arts.}\ 67--68. Cooling of a sphere in air. Surface condition given by a differential equation. Development in a Trigonometric series of which Fourier's Sine Series is a special case. Examples.---\textsc{Arts.}\ % -----File: 007.png 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 instantaneous heat source when the temperature of the air is zero. Examples.---\textsc{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.\\[1ex] \textsc{Miscellaneous Problems} \hfill \pageref{probstart}--\pageref{probend}\\ I. Logarithmic Potential. Polar Coördinates.---II. Potential Function in Space. III. Conduction of heat in a plane.---IV. Conduction of heat in Space.\\[1ex] \begin{center}CHAPTER V.\end{center} \textsc{Zonal Harmonics} \hfill \pageref{ch5start}--\pageref{ch5end}\\ \textsc{Art.}\ 74. Recapitulation. Surface Zonal Harmonics (Legendrians). Zonal Harmonics of the second kind.---\textsc{Arts.}\ 75--76. Legendrians as coefficients in a Power Series. Special values.---\textsc{Art.}\ 77. Summary of the properties of a Legendrian. List of the first eight Legendrians. Relation connecting any three successive Legendrians.---\textsc{Arts.}\ 78--81. Problems in Potential. Potential Function 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.---\textsc{Art.}\ 82. Legendrian as a sum of cosines.---\textsc{Arts.}\ 83--84. Legendrian as the $m$th derivative of the $m$th power of $x^2--1$.---\textsc{Art.}\ 85. Equations derivable from Legendre's Equation.---\textsc{Art.}\ 86. Legendrian as a Partial Derivative.---\textsc{Art.}\ 87. Legendrian as a Definite Integral. \textsc{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.---\textsc{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.---\textsc{Arts.}\ 93--94. Problems in Potential where the value of the Potential Function is given on a spherical surface and has circular symmetry about a diameter. Examples.---\textsc{Art.}\ 95. Development of a power of $x$ in Zonal Harmonic Series.---\textsc{Art.}\ 96. Useful formulas.---\textsc{Art.}\ 97. Development of $\sin n\theta$ and $\cos n\theta$ in Zonal Harmonic Series. Examples. Graphical representation of the first seven Surface Zonal Harmonics. Construction of successive approximations to Zonal Harmonic Series. \textsc{Arts.}\ 98--99. Method of dealing with problems in Potential when the density is given. Examples.---\textsc{Art.}\ 100. Surface Zonal Harmonics of the second kind. Examples: Conal Harmonics.\\[1ex] \begin{center}CHAPTER VI.\end{center} \textsc{Spherical Harmonics} \hfill \pageref{ch6start}--\pageref{ch6end}\\ \textsc{Arts.}\ 101--102. Particular Solutions of Laplace's Equation obtained. Associated Functions. Tesseral Harmonics. Surface Spherical Harmonics. Solid Spherical Harmonics. Table of Associated Functions. Examples.---\textsc{Arts.}\ 103--108. Development in Spherical Harmonic Series. The integral of the product of two % -----File: 008.png Surface Spherical Harmonics of different degrees taken over the surface of the unit sphere is zero. Examples. The integral of the product of two Associated Functions of the same order. Formulas for the coefficients of the series. Illustrative example. Examples.---\textsc{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.---\textsc{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.---\textsc{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.---\textsc{Art.}\ 115. Solution of problems in Potential by direct integration. Examples.---\textsc{Arts.}\ 116--118. Differentiation along an axis. Axes of a Spherical Harmonic.---\textsc{Art.}\ 119. Roots of a Zonal Harmonic. Roots of a Tesseral Harmonic. Nomenclature justified.\\[1ex] \Needspace*{4\baselineskip} \begin{center}CHAPTER VII.\end{center} \textsc{Cylindrical Harmonics (Bessel's Functions)} \hfill \pageref{ch7start}--\pageref{ch7end}\\ \textsc{Art.}\ 120. Recapitulation. Cylindrical Harmonics (Bessel's Functions) of the zeroth order; of the $n$th order; of the second kind. General solution of Bessel's Equation.---\textsc{Art.}\ 121. Bessel's Functions as definite integrals. Examples.---\textsc{Art.}\ 122. Properties of Bessel's Functions. Semi-convergent series for a Bessel's Function. Examples.---\textsc{Art.}\ 123. Problem: Stationary temperatures in a cylinder (\emph{a}) when the temperature of the convex surface is zero; (\emph{b}) when the convex surface is adiathermanous; (\emph{c}) when the convex surface is exposed to air at the temperature zero.---\textsc{Art.}\ 124. Roots of Bessel's functions.---\textsc{Art.}\ 125. The integral of $r$ times the product of two Cylindrical Harmonics of the zeroth order. Example.---\textsc{Art.}\ 126. Development in Cylindrical Harmonic Series. Formulas for the coefficients. Examples.---\textsc{Art.}\ 127. Problem: Stationary temperatures in a cylindrical shell. Bessel's Functions of the second kind employed. Example: Vibration of a ring membrane.---\textsc{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.---\textsc{Art.}\ 129. Problem: Stationary temperatures in a cylinder when the temperatures of the base are unsymmetrical. Bessel's Functions of the $n$th order employed. Miscellaneous examples. Bessel's Functions of fractional order.\\[1ex] \begin{center}CHAPTER VIII.\end{center} \textsc{\spreadout{Laplace's Equation in Curvilinear Coördinates. Ellipsoidal} }\\ \textsc{Harmonics} \hfill \pageref{ch8start}--\pageref{ch8end}\\ \textsc{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.---\textsc{Arts.}\ 132--135. Spheroidal Coördinates. Laplace's Equation in spheroidal coördinates, in normal spheroidal coördinates. Examples. Condition that a set of curvilinear coördinates should be normal. Thermometric Parameters. Particular solutions of Laplace's Equation in spheroidal coördinates. Spheroidal Harmonics. Examples. The Potential Function due to the attraction of an oblate spheroid. Solution for an external point. Examples.---\textsc{Arts.}\ 136--141. % -----File: 009.png \emph{Ellipsoidal Coördinates.} Laplace's Equation in ellipsoidal coördinates. Normal ellipsoidal coördinates expressed as Elliptic Integrals. Particular solutions 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 Harmonic 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.---\textsc{Art.}\ 142. \emph{Conical Coördinates.} The product of two Ellipsoidal Harmonics a Spherical Harmonic.---\textsc{Art.}\ 143. \emph{Toroidal Coördinates.} Laplace's Equation in toroidal coördinates. Particular solutions. Toroidal Harmonics. Potential Function for an anchor ring.\\[1ex] \begin{center} CHAPTER IX. \end{center} \textsc{Historical Summary} \hfill \pageref{ch9start}--\pageref{ch9end}\\ \begin{center}APPENDIX.\end{center} \textsc{Tables} \hfill\pageref{tablestart}--\pageref{tableend}\\ Table I\@. Surface Zonal Harmonics. Argument $\theta$ \hfill\pageref{tableI}\\ Table II\@. Surface Zonal Harmonics. Argument $x$ \hfill\pageref{tableII}\\ Table III\@. Hyperbolic Functions \hfill\pageref{tableIII}\\ Table IV\@. Roots of Bessel's Functions \hfill\pageref{tableIV}\\ Table V\@. Roots of Bessel's Functions \hfill\pageref{tableV}\\ Table VI\@. Bessel's Functions \hfill\pageref{tableVI} % -----File: 010.png %[Blank Page] % -----File: 011.png \mainmatter %\pagestyle{myheadings} \mychap{CHAPTER I.}{INTRODUCTION.} \label{ch1start} \mypara{1.} In many important problems in mathematical physics we are obliged to deal with \emph{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 \[ \tag*{\smallrom{I}} D_t u = a^2(D_x^2 u + D_y^2 u + D_z^2 u),\footnote {For the sake of brevity we shall often use the symbol $\nabla^2$ for the operation $D_x^2 + D_y^2 + D_z^2$; and with this notation equation~\smallrom{I} would be written $D_t u = a^2\nabla^2 u$.} \] 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, \smallrom{I} reduces to \[ \tag*{\smallrom{II}} D_t u = a^2 D_x^2 u, \] 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~\smallrom{I} becomes \[ \tag*{\smallrom{III}} D_x^2u + D_y^2u = 0. \] In \emph{polar} or \emph{spherical coördinates} \smallrom{I} is less simple, it is \[ \tag*{\smallrom{IV}} D_t u = \dfrac{a^2}{r^2} \left[ D_r(r^2D_r u) + \dfrac{1}{\sin\theta} D_\theta (\sin\theta D_\theta u) + \dfrac{1}{\sin^2\theta} D_\phi^2 u \right]. \] 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 \smallrom{IV} reduces to \[ \tag*{\smallrom{V}} D_t(ru) = a^2D_r^2(ru). \] In \emph{cylindrical coördinates} \smallrom{I} becomes \[ \tag*{\smallrom{VI}} D_t u = a^2[D_r^2 u + \dfrac{1}{r} D_r u + \dfrac{1}{r^2} D_\phi^2 u + D_z^2 u]. \] 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 \smallrom{VI} becomes \[ \tag*{\smallrom{VII}} D_t u = a^2(D_r^2 u + \dfrac{1}{r} D_r u). \] % -----File: 012.png In Acoustics in several problems we have the equation \[ \tag*{\smallrom{VIII}} D_t^{2}y = a^{2}D_{x}^{2}y; \] 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. \markright{INTRODUCTION.} If in considering the transverse vibrations of a stretched string we take account of the resistance of the air \smallrom{VIII} is replaced by \[\tag*{\smallrom{IX}} D_t^2 y + 2kD_t y = a^2 D_x^2 y. \] In dealing with the vibrations of a stretched elastic membrane, we have the equation \[\tag*{\smallrom{X}} D_t^{2}z = c^2(D_x^{2}z + D_y^{2}z), \] or in \emph{cylindrical coördinates} \[\tag*{\smallrom{XI}} D_{t}^2 z = c^2 (D_{r}^2 z + \dfrac{1}{r} D_{r}z + \dfrac{1}{r^2}D_{\phi}^2 z). \] In the theory of \emph{Potential} we constantly meet Laplace's Equation \begin{flalign*}\tag*{\smallrom{XII}} &\hfill & D_x^{2}V + &D_y^{2}V + D_z^{2}V=0&& \\ &\text{or}\hfill & &\nabla^2 V = 0&&\hfill\phantom{or} \end{flalign*} which in \emph{spherical coördinates} becomes \[\tag*{\smallrom{XIII}} \dfrac{1}{r^2} \left[ rD_r^{2}(rV) + \dfrac{1}{\sin\theta} D_{\theta}(\sin\theta D_{\theta}V) + \dfrac{1}{\sin^{2}\theta}D_{\phi}^{2}V \right] = 0, \] and in \emph{cylindrical coördinates} \[\tag*{\smallrom{XIV}} D_r^{2}V + \dfrac{1}{r} D_r V + \frac{1}{r^2} D_{\phi}^{2}V + D_z^{2} V = 0. \] In \emph{curvilinear coördinates} it is \[\tag*{\smallrom{XV}} h_{1}h_{2}h_{3} \left[ D_{\rho_1} \left( \frac{h_1}{h_2 h_3} D_{\rho_1} V \right) + D_{\rho_2} \left( \frac{h_2}{h_3 h_1} D_{\rho_2} V \right) + D_{\rho_3} \left( \frac{h_3}{h_1 h_2} D_{\rho_3} V \right) \right] = 0;\\ \] where \hfill $f_1 (x,y,z) = \rho_{1}$, $f_2 (x,y,z) = \rho_{2}$, $f_3 (x,y,z) = \rho_{3}$ \hfill\phantom{where}\\[1ex] represent a set of surfaces which cut one another at right angles, no matter what values are given to $\rho_1$, $\rho_2$, and $\rho_3$; and where \begin{align*} h_1^2 &= (D_{x}\rho_1)^2 + (D_{y}\rho_{1})^2 + (D_{z}\rho_1)^2 \\ h_2^2 &= (D_{x}\rho_2)^2 + (D_{y}\rho_{2})^2 + (D_{z}\rho_2)^2 \\ h_3^2 &= (D_{x}\rho_3)^2 + (D_{y}\rho_{3})^2 + (D_{z}\rho_3)^2, \end{align*} and, of course, must be expressed in terms of $\rho_1$, $\rho_2$, and $\rho_3$. If it happens that $\nabla^{2}\rho_1 = 0$, $\nabla^{2}\rho_{2}= 0$, and $\nabla^{2}\rho_3=0$, then Laplace's Equation~\smallrom{XV} assumes the very simple form \[\tag*{\smallrom{XVI}} h_1^2 D_{\rho_1}^2 V + h_2^2 D_{\rho_2}^2 V + h_3^2 D_{\rho_3}^2 V= 0. \] % -----File: 013.png \mypara{2.} A \emph{differential equation} is an equation containing derivatives or differentials with or without the primitive variables from which they are derived. The \emph{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 derivatives. A general solution will always contain arbitrary (\emph{i.\,e.}, undetermined) \emph{constants} or \emph{arbitrary functions}. A \emph{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. \mypara{3.} If a problem requiring for its solution the solving of a differential equation is \emph{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 equation; 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 solution, 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 equation in question cannot be obtained, and then, since the problem \emph{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 \emph{particular solutions} and so to combine them as to form a result which shall satisfy the given conditions without ceasing to satisfy the differential equation. \mypara{4.} A differential equation is \emph{linear} when it would be of the first degree if the dependent variable and all its derivatives were regarded as algebraic unknown 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 \emph{homogeneous}. All the differential equations collected in Art.~1 are linear and homogeneous. \mypara{5.} \emph{If a value of the dependent variable has been found which satisfies a given homogeneous, linear, differential equation, the product formed by multiplying this value by any constant will also be a value of the dependent variable which will satisfy the equation.} % -----File: 014.png 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 taken out as a factor of the whole first member. The remaining factor being zero, the product is zero and the equation is satisfied. \emph{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. \mypara{6.} It is generally possible to get by some simple device \emph{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 \emph{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 \emph{Legendre's Coefficients,} or \emph{Zonal Harmonics; Laplace's Coefficients,} or \emph{Spherical Harmonics: Bessel's Functions,} or \emph{Cylindrical Harmonics; Lamé's Functions,} or \emph{Ellipsoidal Harmonics,} \&c. \mypara{7.} As an illustration, let us take Fourier's problem of the permanent state of temperatures in a thin rectangular plate of breadth $\pi$ 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 \[ D_x^{2}u +D_y^{2}u= 0 \tag*{[III] Art.~1} \] subject to the conditions \begin{alignat*}{3} u &= 0\quad &&\text{when}\quad & x &= 0 \tag{1} \\ u &= 0 &&\quad\text{``} & x &= \pi \tag{2} \\ u &= 0 &&\quad\text{``} & y &= \infty \tag{3} \\ u &= 1 &&\quad\text{``} & y &= 0. \tag{4} \end{alignat*} % -----File: 015.png We shall begin by getting a particular solution of \smallrom{III}, and we shall use a device which always succeeds when the equation is \textit{linear} and \textit{homogeneous} and has \textit{constant coefficients}. Assume\label{notep5}\footnote {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.} $u = e^{\alpha y + \beta x}$, where $\alpha$ and $\beta$ are constants, substitute in \smallrom{III} and divide by $e^{\alpha y + \beta x}$, and we have $\alpha^2 + \beta^2 = 0$. If, then, this condition is satisfied $u = e^{\alpha y + \beta x}$ is a solution. Hence $u = e^{\alpha y \pm \alpha xi}$ \footnote {We shall regularly use the symbol $i$ for $\sqrt{-1}$.} is a solution of \smallrom{III}, no matter what value may be given to $\alpha$. This form is objectionable, since it involves an imaginary. We can, however, readily improve it. Take $u=e^{\alpha y} e^{\alpha xi}$, a solution of \smallrom{III}, and $u = e^{\alpha y} e^{-\alpha xi}$, another solution of \smallrom{III}; add these values of $u$ and divide the sum by 2 and we have $e^{\alpha y} \cos\alpha x$. (v.~Int.\ Cal.\ Art.~35,\ [1].) Therefore by Art.~5 \[ u = e^{\alpha y} \cos \alpha x \tag{5} \] is a solution of \smallrom{III}. Take $u=e^{\alpha y} e^{\alpha xi}$ and $u = e^{\alpha y} e^{-\alpha xi}$, subtract the second value of $u$ from the first and divide by $2i$ and we have $e^{\alpha y} \sin\alpha x$. (v.~Int.\ Cal.\ Art.~35,\ [2]). Therefore by Art.~5 \[ u = e^{\alpha y} \sin\alpha x \tag{6} \] is a solution of \smallrom{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). \begin{flalign*} &\text{\indent Consider }& & u = e^{\alpha y} \sin\alpha x. &&\phantom{\indent Consider } \tag{6} \end{flalign*} It is zero when $x = 0$ for all values of $\alpha$. It is zero when $x = \pi$ if $\alpha$ is a whole number. It is zero when $y = \infty$ if $\alpha$ 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 \smallrom{III} which satisfies conditions (1), (2) and (3). Let this solution be \[ u = A_1e^{-y} \sin x + A_2e^{-2y} \sin 2x + A_3e^{-3y} \sin 3x + A_4e^{-4y} \sin 4x + \cdots \tag{7} \] $A_1$, $A_2$, $A_3$, $A_4$, \&c., being undetermined constants. When $y = 0$ (7) reduces to \[ u = A_1 \sin x + A_2 \sin 2x + A_3 \sin 3x + A_4 \sin 4x + \cdots . \tag{8} \] 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 $A_1$, $A_2$, $A_3$, \&c.\ in (7). % -----File: 016.png It will be proved later that \[ 1 = \frac{4}{\pi} \left( \sin x + \frac{1}{3} \sin 3x + \frac{1}{5} \sin 5x + \frac{1}{7} \sin 7x + \cdots \right) \] for all values of $x$ between 0 and $\pi$; hence our required solution is \[ u = \frac{4}{\pi} \left[ e^{-y}\sin x + \frac{1}{3} e^{-3y} \sin 3x + \frac{1}{5} e^{-5y} \sin 5x + \frac{1}{7} e^{-7y} \sin 7x + \cdots \right] \tag{9} \] 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 $A \sin m x$, where $m$ is a whole number. The problem of finding the value of the \textit{potential function} at any point of a long, thin, rectangular conducting sheet, of breadth $\pi$, 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 \begin{center} (\textit{a}) $\displaystyle \left( \frac{\pi}{6}, 1 \right)$; (\textit{b}) $\displaystyle \left( \frac{\pi}{3}, 2 \right)$; (\textit{c}) $\displaystyle \left( \frac{\pi}{2}, 3 \right)$, \end{center} correct to the nearest degree. \hfill \emph{Ans.}\ (\textit{a}) 26°; (\textit{b}) 15°; (\textit{c}) 6°. \mypara{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 $\text{X}$, 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 \[ D_t^2 y = a^2 D_x^2 y \tag*{\smallrom{VIII}\ Art.~1,} \] while satisfying the conditions \begin{alignat}{4} y &= 0 && \text{when}\quad& x &= 0 \tag{1} \\ y &= 0 && \quad\text{``} & x &= l \tag{2} \\ y &= f(x) && \quad\text{``} & t &= 0 \tag{3} \\ D_t y &= 0 && \quad\text{``} & t &= 0, \tag{4} \end{alignat} the last condition meaning merely that the string starts from rest. % -----File: 017.png As in the last problem let\footnote{See note on page \pageref{notep5}.} $y = e^{\alpha x+\beta t}$ and substitute in \smallrom{VIII}. Divide by $e^{\alpha x +\beta t}$ and we have $\beta^2=a^2\alpha^2$ as the condition that our assumed value of $y$ shall satisfy the equation. \[ y= e^{\alpha x \pm a\alpha t} \tag{5} \] is, then, a solution of \smallromr{VIII} whatever the value of $\alpha$. 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 $\alpha$ in (5). Replace $\alpha$ by $\alpha i$ and (5) becomes $y=e^{(x\pm at)\alpha i}$, a solution of \smallrom{VIII}. Replace $\alpha$ by $-\alpha i$ and (5) becomes $y=e^{-(x\pm at)\alpha i}$, another solution of \smallrom{VIII}. Add these values of $y$ and divide by 2 and we have $\cos\alpha(x\pm at)$. Subtract the second value of $y$ from the first and divide by $2i$ and we have $\sin\alpha(x\pm at)$. \begin{align*} & y = \cos \alpha(x + at)\\ & y = \cos \alpha(x - at)\\ & y = \sin \alpha(x + at)\\ & y = \sin \alpha(x - at) \end{align*} are, then, solutions of \smallrom{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 \smallrom{VIII}. \begin{align*} &y = \cos \alpha x \cos \alpha at\\ &y = \sin \alpha x \sin \alpha at\\ &y = \sin \alpha x \cos \alpha at\\ &y = \cos \alpha x \sin \alpha at. \end{align*} If we take the third form \begin{align*} &y = \sin \alpha x \cos \alpha at \end{align*} it will satisfy conditions (1) and (4), no matter what value may be given to $\alpha$, and it will satisfy (2) if $\alpha = \dfrac{m\pi}{l}$ where $m$ is an integer. If then we take \[ y=A_1\sin\frac{\pi x}{l}\cos\frac{\pi at}{l} + A_2\sin\frac{2\pi x}{l}\cos\frac{2\pi at}{l} + A_3\sin\frac{3\pi x}{l}\cos\frac{3\pi at}{l} + \cdots \tag{6} \] where $A_{1}$, $A_{2}$, $A_{3}$ $\cdots$ are undetermined constants, we shall have a solution of \smallrom{VIII} which satisfies (1), (2), and (4). When $t=0$ it reduces to \[ y=A_1\sin\frac{\pi x}{l} + A_2\sin\frac{2\pi x}{l} + A_3\sin\frac{3\pi x}{l} + \cdots \tag{7} \] 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 $A_1$, $A_2$, $A_3$ $\cdots$ in (6), and we shall have a solution of \smallrom{VIII} which satisfies all our given conditions. % -----File: 018.png In each of the preceding problems the \emph{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 \emph{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. \mypara{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 \[ V=\frac{M}{\sqrt{c^2 + x^2}} \tag{1} \] 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 \[ rD_r^2(rV) + \frac{1}{\sin\theta}D_\theta(\sin\theta D_\theta V) + \frac{1}{\sin^2\theta}D_\phi^2V=0, \tag*{\smallrom{XIII} Art.~1,} \] subject to the condition \[ V=\frac{M}{(c^2+r^2)^{\frac{1}{2}}} \quad \text{when} \quad \theta=0. \tag{1} \] From the symmetry of the ring, it is clear that the value of the potential function must be independent of $\phi$, so that \smallrom{XIII} will reduce to \[ rD_r^2(rV) + \frac{1}{\sin\theta} D_\theta(\sin\theta D_\theta V)=0. \tag{2} \] We must now try to get particular solutions of (2), and as the coefficients are not constant, we are driven to a new device. Let\footnote{See note on page \pageref{notep5}.} $V=r^m P$, where $P$ is a function of $\theta$ only, and $m$ is a positive integer, and substitute in (2), which becomes \[ m(m+1)r^mP + \frac{r^m}{\sin\theta}D_\theta(\sin\theta D_\theta P) =0. \] % -----File: 019.png Divide by $r^m$ and use the notation of ordinary derivatives since $P$ depends upon $\theta$ only, and we have the equation \[ m(m+1)P + \frac{1}{\sin\theta}\frac{d\Big(\sin\theta\dfrac{dP}{d\theta}\Big)}{d\theta}=0, \tag{3} \] from which to obtain $P$. Equation (3) can be simplified by changing the independent variable. Let $x=\cos\theta$ and (3) becomes \[ \frac{d}{dx}\left [(1-x^2)\frac{dP}{dx}\right ] + m(m+1)P = 0. \tag{4} \] Assume\footnote{See note on page \pageref{notep5}.} now that $P$ can be expressed as a sum or as a series of terms involving whole powers of $x$ multiplied by constant coefficients. Let $P=\sum a_nx^n$ and substitute this value of $P$ in (4). We get \[ \textstyle\sum[n(n-1)a_n x^{n-2} - n(n+1) a_n x^n + m(m+1) a_nx^n] = 0, \tag{5} \] where the symbol $\sum$ 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 $x^k$, must vanish. Hence \begin{flalign*} &&(k+2)(k+1)a_{k+2}-k(k+1)a_k+m(m+ {}& 1)a_k = 0 & \tag{6}\\ &\text{and} & a_{k+2} = -\frac{m(m+1)-k(k+1)}{(k+1)(k+2)}a_k.& & \tag{7} \end{flalign*} If now any set of coefficients satisfying the relation (7) be taken, $P = \sum a_k x^k$ will be a solution of (4). \begin{flalign*} &\indent\text{If }& & k=m,\quad a_{k+2}=0,\quad a_{k+4}=0,\quad \text{\&c.} &&\phantom{\indent If } \end{flalign*} 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 $a_m$. Let us rewrite (7) in the form \[ a_k=-\frac{(k+2)(k+1)}{(m-k)(m+k+1)}a_{k+2}. \tag{8} \] We get from (8), beginning with $k=m-2$, \begin{align*} a_{m-2}& = -\frac{m(m-1)}{2.(2m-1)}a_m\\ a_{m-4}& = \frac{m(m-1)(m-2)(m-3)}{2.4.(2m-1)(2m-3)}a_m\\ a_{m-6}& = -\frac{m(m-1)(m-2)(m-3)(m-4)(m-5)}{2.4.6.(2m-1)(2m-3)(2m-5)}a_m, \text{\quad\&c.} \end{align*} % -----File: 020.png If $m$ is even we see that the set will end with $a_0$, if $m$ is odd, with $a_1$. \[ P = a_m \left[ x^m - \frac{m(m-1)}{2.(2m-1)} x^{m-2} + \frac{m(m-1)(m-2)(m-3)} {2.4.(2m-1)(2m-3)} x^{m-4} - \cdots\,\right] \] where $a_m$ is entirely arbitrary, is, then, a solution of (4). It is found convenient to take $a_m$ equal to \[ \frac{(2m - 1)(2m - 3)\cdots 1}{m!} \] and it can be shown that with this value of $a_m$ $P = 1$ when $x = 1$. $P$ is a function of $x$ and contains no higher powers of $x$ than $x^m$. It is usual to write it as $P_m(x)$. We proceed to compute a few values of $P_m(x)$ from the formula \begin{align*} P_m(x) &= \frac{(2m-1)(2m-3) \cdots 1}{m!} \left[x^m - \frac{m(m-1)}{2.(2m-1)}x^{m-2} \right. \\[1ex] &\left. + \frac{m(m-1)(m-2)(m-3)} {2.4.(2m-1)(2m-3)} x^{m-4} - \cdots \,\right]. \tag{9} \end{align*} We have: { \[ \left.\begin{alignedat}{4} P_0(x) &= 1 && \text{or}\qquad & P_0(\cos\theta) &= 1 \\ P_1(x) &= x && \:\text{``} & P_1(\cos\theta) &= \cos\theta \\ P_2(x) &= \tfrac{1}{2}(3x^2 - 1) && \:\text{``} & P_2(\cos\theta) &= \tfrac{1}{2}(3 \cos^2\theta - 1) \\ P_3(x) &= \tfrac{1}{2}(5x^3 - 3x) \quad && \:\text{``} & P_3(\cos\theta) &= \tfrac{1}{2}(5 \cos^3\theta - 3 \cos\theta) \\ P_4(x) &= \rlap{$\tfrac{1}{8}(35x^4 - 30x^2 + 3)$ or} \\ &&&& \llap{$P_4(\cos\theta) = \tfrac{1}{8}(35$} &\cos^4\theta - 30 \cos^2\theta + 3) \\ P_5(x) &= \rlap{$\tfrac{1}{8}(63x^5 - 70x^3 + 15x)$ or} \\ &&&& \llap{$P_5(\cos\theta) = \tfrac{1}{8}(63 \cos^5\theta $} &-70 \cos^3\theta + 15 \cos\theta). \end{alignedat} \tag{10} \quad\right\} \]} We have obtained $P = P_m(x)$ as a particular solution of (4) and $P = P_m(\cos\theta)$ as a particular solution of (3). $P_m(x)$ or $P_m(\cos\theta)$ is a new function, known as a \textit{Legendre's Coefficient}, or as a \textit{Surface Zonal Harmonic}, and occurs as a normal form in many important problems. $V = r^mP_m(\cos\theta)$ is a particular solution of (2) and $r^mP_m(\cos\theta)$ is sometimes called a \textit{Solid Zonal Harmonic}. We can now proceed to the solution of our original problem. \[ V=A_0r^0P_0(\cos\theta) + A_1rP_1(\cos\theta) + A_2r^2P_2(\cos\theta) + A_3r^3P_3(\cos\theta) + \cdots \tag{11} \] where $A_0$, $A_1$, $A_2$, \&c., are entirely arbitrary, is a solution of (2) (v.\ Art.~5). When $\theta = 0$ (11) reduces to \[ V=A_0 + A_1 r + A_2 r^2 + A_3 r^3 + \cdots\,, \] since, as we have said, $P_m(x) = 1$ when $x = 1$, or $P_m(\cos\theta) = 1$ when $\theta = 0$. By our condition (1) \[ V=\frac{M}{(c^2 + r^2)^{\frac{1}{2}}} \] when $\theta = 0$. % -----File: 021.png By the Binomial Theorem \[ \frac{M}{(c^2+r^2)^{\frac{1}{2}}} = \frac{M}{c} \left[ 1 - \frac{1}{2} \frac{r^2}{c^2} + \frac{1.3}{2.4} \frac{r^4}{c^4} - \frac{1.3.5}{2.4.6} \frac{r^6}{c^6} + \cdots \,\right] \] provided $r < c$. Hence % recast to fit line \begin{align*} V = \frac{M}{c} \left[ P_0(\cos\theta) - \frac{1}{2} \frac{r^2}{c^2} P_2(\cos\theta)\right. & + \frac{1.3}{2.4} \frac{r^4}{c^4} P_4(\cos\theta) \\ & \left. - \frac{1.3.5}{2.4.6} \frac{r^6}{c^6} P_6(\cos\theta) + \cdots\, \right] \tag{12} \end{align*} 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 {\def\bit#1#2#3{\text{(\textit{#1})\ }\left(r = #2, \, \theta = #3 \right);} \begin{align*} & \bit{a}{.2}{0} & & \bit{d}{.6}{0} & & \bit{f}{.6}{\frac{\pi}{3}} \\ % & \bit{b}{.2}{\frac{\pi}{4}} & & \bit{e}{.6}{\frac{\pi}{6}} & & \bit{g}{.6}{\frac{\pi}{2}} \\ % & \bit{c}{.2}{\frac{\pi}{2}} \end{align*} \hfill\smash{\raisebox{3ex}{\begin{tabular} {l@{\hfill}} \llap{\emph{Ans.} }(\emph{a}) .98; (\emph{b}) .99; (\emph{c}) 1.01; (\emph{d}) .86;\\ (\emph{e}) .90; (\emph{f}) 1.00; (\emph{g}) 1.10. \end{tabular}}}}\break 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. \mypara{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 \[ rD_r^2(ru) + \frac{1}{\sin\theta} D_\theta (\sin \theta D_\theta u) + \frac{1}{\sin^2\theta} D_\phi^2 u = 0 \tag*{\smallrom{XIII} Art.~1} \] which as before reduces to \[ r D_r^2 (ru) + \frac{1}{\sin\theta} D_\theta (\sin\theta D_\theta u) = 0 \tag{1} \] from the consideration that the temperatures must be independent of $\phi$. Our equation of condition is \[ u = 1 \text{ from } \theta = 0 \text{ to } \theta = \frac{\pi}{2} \text{ and } u = 0 \text{ from } \theta = \frac{\pi}{2} \text{ to } \theta = \pi, \tag{2} \] when $r = 1$. % -----File: 022.png As we have seen $u = r^mP_m(\cos \theta)$ is a particular solution of (1), $m$ being any positive whole number, and \[ u = A_0r^0P_0 (\cos \theta) + A_1r P_1 (\cos \theta) + A_2r^2P_2 (\cos \theta) + A_3r^3P_3 (\cos \theta) + \cdots \tag{3} \] where $A_0$, $A_1$, $A_2$, $A_3 \cdots$ are undetermined constants, is a solution of (1). When $r = 1$ (3) reduces to \[ \tag{4} u = A_0P_0 (\cos \theta) + A_1P_1 (\cos \theta) + A_2P_2(\cos \theta) + A_3P_3 (\cos \theta) + \cdots \] If then we can develop our function of $\theta$ 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 $A_0$, $A_1$, $A_2$, \&c., in (3) and we shall have our required solution. \mypara{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 symmetry\label{notep12}\footnote {A function of the coördinates of a point has \textit{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.} about an axis through the centre perpendicular to the plane of the boundary, and then allowed to vibrate. Here we have to solve \[ \tag*{\smallrom{XI} Art.~1} D_t^2z=c^2\left(D_r^2z + \frac{1}{r}D_rz + \frac{1}{r^2} D_\phi^2z \right) \] subject to the conditions \begin{alignat*}{3} \tag{1} z &= f(r) \quad && \text{when}\quad & t &= 0 \\ \tag{2} D_t z &= 0 && \quad\text{``} & t &= 0 \\ \tag{3} z &= 0 && \quad\text{``} & r &= a. \end{alignat*} From the symmetry of the supposed initial distortion $z$ must be independent of $\phi$, therefore \smallrom{XI} reduces to \[ \tag{4} D_t^2 z = c^2\left(D_r^2 z + \frac{1}{r} D_r z\right) \] 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. Assume\footnote {See note on page \pageref{notep5}.} $z = R.T$ 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 \[ R D_t^2 T = c^2 T\left(D_r^2 R + \frac{1}{r} D_r R\right) \] \begin{flalign*} \tag{5} &\text{or }&& \frac{1}{c^2T} \frac{d^2T}{dt^2} = \frac{1}{R} \left(\frac{d^2R}{dr^2} + \frac{1}{r} \frac{dR}{dr}\right). && \end{flalign*} 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 % -----File: 023.png $r$, and consequently since it contains neither $t$ nor $r$, it must be constant. Let it equal $-\mu^2$, where $\mu$ of course is an undetermined constant. Then (5) breaks up into the two differential equations \begin{gather} \frac{d^2T}{dt^2} + \mu^2c^2T=0 \tag{6} \\ \frac{d^2R}{dr^2} + \frac{1}{r}\frac{dR}{dr} + \mu^2R = 0. \tag{7} \end{gather} (6) can be solved by familiar methods, and we get $T = \cos \mu ct$ and $T = \sin \mu ct$ 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 independent variable. Let $r = \dfrac{x}{\mu}$. (7) becomes \[ \frac{d^2R}{dx^2} + \frac{1}{x}\frac{dR}{dx}+R=0. \tag{8} \] Assume, as in Art.~9, that $R$ can be expressed in terms of whole powers of $x$. Let $R = \sum a_n x^n$ and substitute in (8). We get \[ \textstyle\sum [n(n - 1)a_nx^{n-2} + na_nx^{n-2} + a_nx^n] = 0, \notag \] an equation which must be true no matter what the value of $x$. The coefficient of any given power of $x$, as $x^{k-2}$, must, then, vanish, and \begin{flalign*} && k(k - 1)a_k + k&a_k + a_{k-2} = 0 &&\phantom{whence we obtain } \\ &\text{or }& k^2a_k + a&_{k-2} = 0 && \\ &\text{whence we obtain }& a_{k-2} ={}& - k^2a_k && \tag{9} \end{flalign*} as the only relation that need be satisfied by the coefficients in order that $R = \sum a_k x^k$ shall be a solution of (8). \[ \hbox to \displaywidth {\rlap{\indent If}\hfil $k = 0$,\quad $a_{k-2}= 0$,\quad $a_{k-4}=0$,\quad \&c.\hfil} \] We can then begin with $k = 0$ as our lowest subscript. \begin{flalign*} &\text{\indent From (9) }& a_k &=-\frac{a_{k-2}}{k^2}. &&\phantom{\indent From\ (9) }\\ &\text{Then }& a_{2} &= -\frac{a_0}{2^{2}} \\ && a_{4} &= \frac{a_0}{2^2.4^2} \\ && a_{6} &= -\frac{a_0}{2^2.4^2.6^2}, \text{ \&c.} \\ &\text{Hence }& R = a_0\left[1-\frac{x^2}{2^2} \right.+{}& \frac{x^4}{2^2.4^2}- \left. \frac{x^6}{2^2.4^2.6^2} + \cdots\,\right] \end{flalign*} where $a_0$ may be taken at pleasure, is a solution of (8), provided the series is convergent. % -----File: 024.png Take $a_0=1$, and then $R=J_0(x)$ where \[ J_0(x)=1-\frac{x^2}{2^2} + \frac{x^4}{2^2.4^2} - \frac{x^6}{2^2.4^2.6^2} + \frac{x^8}{2^2.4^2.6^2.8^2} - \cdots \tag{10} \] is a solution of (8). $J_0(x)$ 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 $J_0(x)$ (v.\ Int.\ Cal.\ Art.~30) \[ 1 + \frac{r^2}{2^2} + \frac{r^4}{2^2.4^2} + \frac{r^6}{2^2.4^2.6^2} + \cdots, \] where $r$ is the modulus of $x$, is convergent for all values of $r$. For the ratio of the $n + 1$st term of this series to the $n$th term is $\dfrac{r^2}{4n^2}$ and approaches zero as its limit as $n$ is indefinitely increased, no matter what the value of $r$. Therefore $J_0(x)$ is \textit{absolutely convergent}. $J_0(x)$ is a new and important form. It is called a \textit{Bessel's Function} of the zeroth order, or a \textit{Cylindrical Harmonic}. Equation (8) was obtained from (7) by the substitution of $x=\mu r$, therefore \[ R = J_0(\mu r) = 1 - \frac{(\mu r)^2}{2^2} + \frac{(\mu r)^4}{2^2.4^2} - \frac{(\mu r)^6}{2^2.4^2.6^2} + \cdots \] is a solution of (7), no matter what the value of $\mu$, and $z=J_0(\mu r)\cos\mu ct$ or $z=J_0(\mu r)\sin\mu ct$ is a solution of (4). $z=J_0(\mu r)\cos\mu ct$ satisfies condition~(2) whatever the value of $\mu$. In order that it should also satisfy condition~(3) $\mu$ must be so taken that \[ J_0(\mu a)=0; \tag{11} \] that is, $\mu$ must be a root of (11) regarded as an equation in $\mu$. It can be shown that $J_0(x)=0$ 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 $x_1$, $x_2$, $x_3$, $\cdots$ be these roots. Then if \begin{gather*} \frac{x_1}{a}=\mu_1, \quad \frac{x_2}{a}=\mu_2, \quad \frac{x_3}{a}=\mu_3, \quad \text{\&c.\,}\\ z=A_1J_0(\mu_1r)\cos\mu_1ct + A_2J_0(\mu_2r)\cos\mu_2ct + A_3J_0(\mu_3r)\cos\mu_3ct + \cdots, \tag{12} \end{gather*} where $A_1$, $A_2$, $A_3$, \&c., are any constants, is a solution of (4) which satisfies conditions (2) and (3). When $t=0$ (12) reduces to \[ z=A_1J_0(\mu_1 r) + A_2J_0(\mu_2 r) + A_3J_0(\mu_3 r) + \cdots. \tag{13} \] 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 $A_1$, $A_2$, $A_3$, \&c., in (12). % -----File: 025.png \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 temperature of any point in the cylinder at the expiration of the time $t$ is \[ u=A_{1}e^{-a^2 \mu_1^2t}J_{0}(\mu_1r) + A_{2}e^{-a^2 \mu_2^2t}J_{0}(\mu_2r) + A_{3}e^{-a^2 \mu_3^2t}J_{0}(\mu_3r) + \cdots \] where $\mu_{1}$, $\mu_{2}$, \&c., are the roots of $J_{0}(\mu c) = 0$, and where \[ 1=A_{1}J_{0}(\mu_{1}r) + A_{2}J_{0}(\mu_{2}r) +A_{3}J_{0}(\mu_{3}r) + \cdots, \] $c$ being the radius of the cylinder. \mypara{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 \emph{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 investigation of the properties and relations of certain new and important functions, as well as the consideration of methods of developing in terms of them. \mypara{13.} In each of the problems just taken up we have to deal with a homogeneous linear partial differential equation involving two independent variables, and we are content if we can obtain particular solutions. In each case the assumption 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 involves 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 problems. \begin{flalign} &\text{\indent(\textit{a})}&& D_x^2 u + D_y^2 u = 0 && \quad \tag{1} \end{flalign} Assume $u = X.Y$ where $X$ involves $x$ but not $y$, and $Y$ involves $y$ but not $x$. \begin{flalign*} &\text{Substitute in (1), }&& YD_x^2 X + XD_y^2 Y = 0, &&\phantom{Substitute in (1), } \end{flalign*} or, since we are now dealing with functions of a single variable, \begin{flalign} && \frac{1}{X}\frac{d^2X}{dx^2} &+ \frac{1}{Y}\frac{d^2Y}{dy^2}=0, &&\notag\\ &\text{or}& \frac{1}{Y}\frac{d^2Y}{dy^2} &= -\frac{1}{X}\frac{d^2X}{dx^2}. \tag{2} \end{flalign} % -----File: 026.png 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 $\alpha^2$. \begin{flalign*} &\text{Then}& \frac{d^2Y}{dy^2} - \alpha^2 Y& = 0 &\tag{3}\\ &\text{and}& \frac{d^2X}{dx^2} + \alpha^2 X &= 0; &\tag{4} \end{flalign*} and if (3) and (4) can be solved, we can solve (1). They have for their complete solutions \begin{flalign*} & & Y &= Ae^{\alpha y} + Be^{-\alpha y}\\ &\text{and}& X &= C \sin {\alpha x} + D \cos {\alpha x}. \quad\text{ (v.\ Int.\ Cal.\ p.~319, \S~21.)} \end{flalign*} Hence $Y = e^{\alpha y}$ and $Y=e^{-\alpha y}$ are particular solutions of (3), $X=\sin {\alpha x}$ and $X = \cos {\alpha x}$ are particular solutions of (1), and consequently \[ u =e^{\alpha y} \sin{\alpha x},\quad u = e^{\alpha y} \cos {\alpha x}, \quad u = e^{-\alpha y} \sin {\alpha x}, \text{ and }u =e^{-\alpha y} \cos {\alpha x} \] are particular solutions of (1). These agree with the results of Art.~7. \begin{flalign*} &\text{\indent(\textit{b})} & D_t^2 y &= a^2 D_x^2 y &\quad\tag{1} \end{flalign*} 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 $a^2 TX$. We get \begin{flalign*} &\phantom{(2)}&& \frac{1}{a^2 T}\frac{d^2T}{dt^2} = \frac{1}{X}\frac{d^2X}{dx^2}; \tag{2}\\ \intertext{hence as in the last case $\dfrac{1}{X}\dfrac{d^2X}{dx^2}$ is a constant; call it $- \alpha^2$, and (2) breaks up into} && &\frac{d^2X}{dx^2} + \alpha^2 X = 0 &\tag{3}\\ &\phantom{(4)}& &\frac{d^2T}{dt^2} + \alpha^2a^2T = 0.&\tag{4} \end{flalign*} The complete solutions of (3) and (4) are \begin{flalign*} && &X = A \sin {\alpha x} + B \cos {\alpha x} \\ &\text{and }& &T = C \sin {\alpha at} + D\cos {\alpha at}, \quad\text{ (v.\ Int.\ Cal.\ p.~319, \S~21).} \end{flalign*} \[ y = \sin {\alpha x}\cos {\alpha at}, \ y = \sin {\alpha x}\sin {\alpha at}, \ y = \cos {\alpha x}\cos {\alpha at}, \ y = \cos {\alpha x}\sin {\alpha at} \] are particular solutions of (1), and agree with the results of Art.~8. \begin{flalign*} &\text{\indent (\textit{c})}& rD_r^2(rV) + \dfrac{1}{\sin{\theta}} D_\theta (\sin{\theta} D_{\theta} V)& = 0. &\quad\tag{1} \end{flalign*} Assume $V= R.\Theta$ where $R$ involves $r$ alone, and $\Theta$ involves $\theta$ alone; substitute in (1), divide by $R.\Theta$, and transpose; we get \[ \frac{r}{R} \frac{d^2(rR)}{dr^2} = - \frac{1}{\Theta \sin \theta}\frac{d\Big(\sin \theta \dfrac{d\Theta}{ d\theta}\Big)}{d\theta}. \tag{2} \] % -----File: 027.png Since by the reasoning used in (\textit{a}) and (\textit{b}) each member of (2) must be a constant, say $\alpha^2$, we have \begin{flalign*} && r\frac{d^2(rR)}{dr^2} ={}& \alpha^2 R \tag{3} \\ &\text{and} &\frac{1}{\sin\theta}\frac{d\Big(\sin\theta \dfrac{d\Theta}{d\theta}\Big)}{d\theta} &+ \alpha^2\Theta = 0. &&\phantom{and}\tag{4} \end{flalign*} (3) can be expanded into \[ r^2 \frac{d^2 R}{dr^2} + 2r \frac{dR}{dr} - \alpha^2 R = 0. \tag{5} \] (5) can be solved (v.\ Int.\ Cal.\ p.~321, \S~23), and has for its complete solution \[ R = Ar^m + Br^n, \] where \hfill $m = -\frac{1}{2} + \sqrt{\alpha^2 + \frac{1}{4}}$ \quad and \quad $n = -\frac{1}{2} - \sqrt{\alpha^2 + \frac{1}{4}}$. \hfill\phantom{where }\\[1ex] Hence $n = - m - 1$, and $\alpha^2$ may be written $m(m + 1)$, $m$ being wholly arbitrary; and \begin{gather*} R = Ar^m + Br^{-m-1}. \\ R = r^m,\quad \text{ and }\quad R = \frac{1}{r^{m + 1}} \end{gather*} are, then, particular solutions of \[ r^2 \frac{d^2R}{dr^2} + 2r\frac{dR}{dr} - m(m + 1) R = 0. \tag{6} \] With the new value of $\alpha^2$ (4) becomes \[ \frac{1}{\sin\theta}\frac{d\Big(\sin\theta \dfrac{d\Theta}{d\theta}\Big)}{d\theta} + m(m + 1) \Theta = 0. \tag{7} \] which has been treated in Art.~9 for the case where $m$ is a positive integer, and the particular solution $\Theta = P_m (\cos\theta)$ has been obtained. \begin{flalign*} &\indent\text{Hence }& V &= r^m P_m (\cos \theta) &&\phantom{\indent Hence }\\ &\text{and }& V &= \frac{1}{r^{m + 1}} P_m (\cos \theta), && \end{flalign*} $m$ 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. \mypara{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. % -----File: 028.png As a very simple example, let us take the equation Art.~13 (\emph{a}) (4), which we shall write \[ \frac{d^2z}{dx^2}+\alpha^2z = 0. \tag{1} \] Assume that there is a solution which can be expressed in terms of powers of $x$; that is, let $z = \sum{a_nx^n}$, where the coefficients are to be determined. Substitute this value for $z$ in (1) and we get \[ \textstyle \sum{[n(n - 1)a_nx^{n-2} + \alpha^2a_nx^n]} = 0. \] 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 \begin{flalign*} && (n + 1)(n + 2)a_{n+2} + \alpha^2a_n = 0; & &&\phantom{whence }\\ &\text{whence }& a_n = -\frac{(n+1)(n+2)}{\alpha^2}a_{n + 2}& && \end{flalign*} is the only relation that need hold between the coefficients in order that $z = \sum{a_nx^n}$ should be a solution of (1). If $n + 2 = 0$ or $n + 1 = 0$, $a_n$ will be zero and $a_{n-2}$, $a_{n-4}$, \&c., will be zero. In the first case the series will begin with $a_0$, in the second with $a_1$. \[ a_{n+2} = - \frac{\alpha^2}{(n+1)(n+2)}a_n. \] If we begin with $a_0$ we have \begin{align*} a_2 &= -\frac{\alpha^2}{2!}a_0, & a_4 &= \frac{\alpha^4}{4!}a_0, & a_6 &= -\frac{\alpha^6}{6!}a_0, \text{ \&c}., \:\cdots\\[-5ex] \end{align*} \begin{flalign*} &\text{and }& z = a_0 \left( 1 - \frac{\alpha^2x^2}{2!} \right.& + \frac{\alpha^4x^4 }{4!} - \left. \frac{\alpha^6x^6}{6!} + \cdots \right) &&\tag{2} \\ &\text{or }& z = {}& a_0 \cos\alpha x &&\tag{3} \end{flalign*} is a particular solution of (1). If we begin with $a_1$ we have \begin{align*} a_3 &= -\frac{\alpha^2}{3!}a_1, & a_5 &= \frac{\alpha^4}{5!}a_1, & a_7 &= -\frac{\alpha^6}{7!}a_1, \text{ \&c}., \:\cdots\\[-5ex] \end{align*} \begin{flalign*} &\text{and }& z &= a_1 \left( x - \frac{\alpha^2x^3}{3!} + \frac{\alpha^4x^5}{5!} - \frac{\alpha^6x^7}{7!} + \cdots \right) &&\tag{4} \end{flalign*} % -----File: 029.png is a solution of (1); $a_1$ can be taken at pleasure. Let $a_1 = \alpha$, (4) becomes \begin{flalign*} &&z &= \alpha x - \frac{\alpha^3x^3}{3!} + \frac{\alpha^5x^5}{5!} - \frac{\alpha^7x^7}{7!} + \cdots &&\phantom{or}\\ &\text{or} & z &= \sin\alpha x& \end{flalign*} which, then, is a particular solution of (1). \[ z = A\sin\alpha x + B\cos\alpha x \tag{5} \] is, then, a solution of (1), and since it contains two arbitrary constants it is the general solution. \mypara{15.} As another example we will take the equation \[ x^2\frac{d^2z}{dx^2} + 2x\frac{dz}{dx} - m(m+1)z = 0, \tag{1} \] which is in effect equation (6), Art.~13~(\emph{c}), and let $m$ be a positive integer. Assume $z = \sum a_nx^n$ and substitute in (1). We get \[ \textstyle\sum [n(n+1) - m(m+1)]a_nx^n = 0. \] This is an identical equation, therefore \[ [n(n+1) - m(m+1)]a_n = 0. \] Hence $a_n=0$ 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 \[ z = Ax^m + Bx^{-m-1} \tag{2} \] is the general solution of (1) and \[ z=x^m \quad \text{and} \quad z=\frac{1}{x^{m+1}} \] 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~(\emph{c}). \mypara{16.} Let us now take the equation \[ \frac{d}{dx}\left [(1-x^2)\frac{dz}{dx}\right ] + m(m+1)z = 0 \tag{1} \] which is in effect equation (4), Art.~9, and is known as \emph{Legendre's Equation}. (1) may be written \[ (1-x^2) \frac{d^2z}{dx^2}-2x\frac{dz}{dx} + m(m+1)z = 0. \tag{2} \] % -----File: 030.png Assume $z=\sum a_nx^n$ and substitute in (2). We get \[ \textstyle\sum \{n(n-1)a_nx^{n-2} + [m(m+1) - n(n+1)]a_nx^n\} = 0. \] \begin{flalign*} &\text{Hence } & (&n+1)(n+2)a_{n+2} + [m(m+1)-n(n+1)]a_n = 0, &&\phantom{Hence } \end{flalign*} \begin{flalign*} &\text{or} & a_n &= -\frac{(n+1)(n+2)}{m(m+1)-n(n+1)}a_{n+2}. && \tag{3} \intertext{If $a_n=0$, then $a_{n-2}=0$, $a_{n-4}=0$, \&c.; but $a_n=0$ if $n=-2$ or $n=-1$. For the first case we have the sequence of coefficients} &&a_2&=-\frac{m(m+1)}{2!}a_0\\ &&a_4&= \frac{m(m-2)(m+1)(m+3)}{4!}a_0\\ &&a_6&=-\frac{m(m-2)(m-4)(m+1)(m+3)(m+5)}{6!}a_0, \quad \text{\&c.} \end{flalign*} Let us take $a_0$, which is arbitrary, as 1. Then $z=p_m(x)$ where \[ p_m(x) = \left[ 1 - \frac{m(m+1)}{2!}x^2 + \frac{m(m-2)(m+1)(m+3)}{4!}x^4 - \cdots \right] \tag{4} \] is a solution of Legendre's Equation if $p_m(x)$ is a finite sum or a convergent series. For the second case we have the sequence of coefficients \begin{align*} &a_3=-\frac{(m-1)(m+2)}{3!}a_1\\ &a_5= \frac{(m-1)(m-3)(m+2)(m+4)}{5!}a_1\\ &a_7=-\frac{(m-1)(m-3)(m-5)(m+2)(m+4)(m+6)}{7!}a_1, && \text{\&c.} \end{align*} Let us take $a_1$, which is arbitrary, as 1. Then $z=q_m(x)$ where \[ q_m(x) = \left[ x - \frac{(m-1)(m+2)}{3!}x^3 + \frac{(m-1)(m-3)(m+2)(m+4)}{5!}x^5 - \cdots \!\right]\!\text{(5)} \] is a solution of Legendre's Equation if $q_m(x)$ is a finite sum or a convergent series. % -----File: 031.png If $m$ is a positive even whole number, $p_m(x)$ will terminate with the term containing $x^m$, and is easily seen to be identical with \[ (-1)^\frac{m}{2} \frac{ 2^m\Big [\Gamma \Big(\dfrac{m}{2}+1\Big)\Big]^2 } { \Gamma(m+1) } P_m(x). \tag*{[v.\ Art.~9~(9)]} \] For all other values of $m$, $p_m(x)$ is a series. The ratio of the $(n+1)$st term of $p_m(x)$ to the $n$th, when $m$ is not a positive even integer, is \[ \frac{(2n-2-m)(2n-1+m)}{(2n-1)(2n)}x^2. \] Its limiting value, as $n$ is increased, is $x^2$, 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 $q_m(x)$ will terminate with the term containing $x^m$, and is easily seen to be identical with \[ (-1)^\frac{m-1}{2} \frac{ 2^{m-1} \Big[\Gamma\Big(\dfrac{m+1}{2} \Big) \Big]^2 } { \Gamma(m+1)} P_m(x). \] For all other values of $m$, $q_m(x)$ is a series, and can be shown to be convergent if $-1 < x < 1$, and divergent for all other values of $x$. \[ z=Ap_m(x)+Bq_m(x) \tag{6} \] is the general solution of Legendre's Equation if $-1 < x < 1$, no matter what the value of $m$. From Art.~13~(\emph{c}) it follows that \[ \left. \begin{aligned} V &= r^mp_m(\cos\theta)\\ V &= \frac{1}{r^{m+1}}p_m(\cos\theta)\quad\\ V &= r^mq_m(\cos\theta)\\ V &= \frac{1}{r^{m+1}}q_m(\cos\theta)\\ \end{aligned} \right\} \tag{7} \] are particular solutions of \[ rD_r^2(rV) + \frac{1}{\sin\theta}D_\theta(\sin\theta D_\theta V)=0, \] no matter what the value of $m$, provided $\cos\theta$ 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. % -----File: 032.png With this understanding let us return to (3), which may be rewritten \[ a_{n+2}=-\frac{(m-n)(m+n+1)}{(n+1)(n+2)}a_n.\\[-3ex] \] \begin{flalign*} &\text{If} && \text{$a_{n+2}=0$, then $a_{n+4}=0$,\; $a_{n+6} = 0$, \&c.;} &&\qquad \\[1ex] &\text{but} && \text{$a_{n+2}=0$\phantom{, }if\; $n=m$,\; or\; $n=-m-1$.} && \end{flalign*} If in (3) we begin with $n=m-2$, we get the sequence of coefficients already obtained in Art.~9, and we have $z=P_m(x)$, where \begin{align*} P_m(x) &= \frac{(2m-1)(2m-3)\cdots 1}{m!} \left[ x^m - \frac{m(m-1)}{2(2m-1)} x^{m-2} \right. \\[1ex] &+ \frac{m(m-1)(m-2)(m-3)}{2.4.(2m-1)(2m-3)} x^{m-4} \\[1ex] &- \left.\frac{m(m-1)(m-2)(m-3)(m-4)(m-5)} {2.4.6.(2m-1)(2m-3)(2m-5)} x^{m-6} + \cdots \right], \tag{8} \end{align*} as a particular solution of Legendre's Equation. If, however, we begin with $n=-m-3$, we have \begin{align*} a_{-m-3} &= \frac{(m+1)(m+2)}{2(2m+3)}a_{-m-1} \\[1ex] a_{-m-5} &= \frac{(m+1)(m+2)(m+3)(m+4)} {2.4.(2m+3)(2m+5)}a_{-m-1} \\[1ex] a_{-m-7} &= \frac{(m+1)(m+2)(m+3)(m+4)(m+5)(m+6)} {2.4.6.(2m+3)(2m+5)(2m+7)} a_{-m-1}, \quad\text{\&c.} \end{align*} $a_{-m-1}$ may be taken at pleasure, and is usually taken as $\dfrac{m!}{1.3.5.\cdots(2m+1)}$, and $z=Q_m(x)$ where \begin{align*} Q_m(x) &= \frac{m!}{(2m+1)(2m-1)\cdots1} \left[ \frac{1}{x^{m+1}} + \frac{(m+1)(m+2)}{2.(2m+3)} \frac{1}{x^{m+3}} \right. \\[1ex] &+ \left. \frac{(m+1)(m+2)(m+3)(m+4)} {2.4.(2m+3)(2m+5)} \frac{1}{x^{m+5}} + \cdots \right] \tag{9} \end{align*} is a second particular solution of Legendre's Equation, provided the series is convergent. $Q_m(x)$ is called a \emph{Surface Zonal Harmonic} of the \emph{second kind}. % -----File: 033.png 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, \[ z = A P_m(x) + B Q_m(x) \tag{10} \] is the general solution of Legendre's Equation if $x < - 1$ or $x > 1$. We have seen that for $- 1 < x < 1$ \begin{align*} P_m(x) &= (-1)^\frac{m}{2} \frac{ \Gamma (m + 1) } { 2^m\left[\Gamma\left( \dfrac{m}{2} + 1 \right) \right]^2 } p_m(x) \tag{11} \intertext{if $m$ is an even integer, and } P_m(x) &= (- 1)^\frac{m-1}{2} \frac{ \Gamma (m + 1) } { 2^{m-1}\Big[\Gamma\Big( \dfrac{m+1}{2} \Big) \Big]^2 } q_m(x) \tag{12} \end{align*} if $m$ is an odd integer. If now we define $Q_m(x)$ as follows when $- 1 < x < 1$ \begin{align*} Q_m(x) &= (-1)^\frac{m+1}{2} \frac{ 2^{m-1} \Big[\Gamma\Big( \dfrac{m+1}{ 2} \Big)\Big]^2 } { \Gamma(m+1) } p_m (x) \tag{13} \intertext{if $m$ is an odd integer, and } Q_m(x) &= (-1)^\frac{m}{2} \frac{ 2^m \Big[\Gamma\Big( \dfrac{m}{2}+1 \Big) \Big]^2 } { \Gamma(m+1) } q_m(x) \tag{14} \end{align*} 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 $x > 1$. \mypara{17.} Let us last consider the equation \[ \frac{d^2z}{dx^2} + \frac{1}{x} \frac{dz}{dx} + \left (1 - \frac{m^2}{ x^2} \right ) z = 0 \tag{1} \] which is known as Bessel's Equation, and which reduces to (8) Art.~11, that is, to \[ \frac{d^2 z}{ dx^2} + \frac{1}{x} \frac{dz}{dx} + z = 0 \] when $m$ = 0;\footnote {This equation was first studied by Fourier in considering the cooling of a cylinder. We shall designate it as ``Fourier's Equation.''} (1) can be simplified by a change of the dependent variable. % -----File: 034.png Let $z = x^m v$ and we get \[ \frac{d^2 v}{ dx^2} + \frac{2m +1 }{ x}\frac{ dv}{dx} + v = 0 \tag{2} \] to determine $v$. Assume $v = \sum a_n x^n$, and substitute in (2). We get \[ \textstyle\sum [n(2m + n)a_n x^{n-2} + a_n x^n] = 0; \] \begin{flalign*} &\text{whence }& &a_{n-2} = -n(2m + n)a_n. &&\phantom{whence } \end{flalign*} If we begin with $n = 0$, then $a_{n-2} = 0$, $a_{n-4} = 0$, \&c., and we have the set of values \begin{align*} a_2 &= - \frac{a_0}{ 2(2m+2)} = - \frac{a_0}{ 2^2 (m+1)}\\[1ex] a_4 &= \frac{a_0}{ 2.4(2m + 2)(2m + 4)} = \frac{a_0}{ 2^4.2!(m + 1)(m + 2)} \end{align*}\\[-3ex] \[ a_6 = - \frac{a_0} {2.4.6(2m + 2)(2m + 4)(2m + 6)} = - \frac{a_0}{ 2^6.3!(m + 1)(m + 2)(m + 3)}; \]\\[-2ex] \begin{flalign*} &\text{whence }& z = a_0 x^m \left[ 1 \vphantom{\dfrac{x^2}{(0)}}\right. &- \frac{x^2}{ 2^2 (m+1)} + \frac{x^4}{ 2^4.2!(m + 1)(m + 2)} &&\\[1ex] && &- \left. \frac{x^6}{ 2^6.3!(m + 1)(m + 2)(m + 3)} + \cdots \right ] && \tag{3} \end{flalign*} is a solution of Bessel's Equation. $a_0$ is usually taken as $\dfrac{1}{2^m m!}$ if $m$ is a positive integer, or as $\dfrac{1}{ 2^m \Gamma (m + 1)}$ if $m$ is unrestricted in value, and the second member of (3) is represented by $J_m(x)$ and is called a \textit{Bessel's Function} of the $m$th order, or a \textit{Cylindrical Harmonic} of the $m$th order. If $m = 0$, $J_m(x)$ becomes $J_0(x)$ 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 $x^m v$ for $z$, we get in place of (2) the equation \[ \frac{d^2 v}{ dx^2} + \frac{1-2m}{x} \frac{dv}{dx} + v = 0 \] and in place of (3) \begin{align*} z = a_0 x^{-m} \left[ 1 \vphantom{\dfrac{x^2}{(0)}}\right. &- \frac{x^2}{ 2^2 (1-m)} + \frac{x^4}{ 2^4.2!(1-m)(2-m)} \\[1ex] &- \left. \frac{x^6}{ 2^6.3!(1 - m)(2 - m)(3 - m)} + \cdots \right ] \tag{4} \end{align*} % -----File: 035.png If $a_0$ is taken equal to $\dfrac{1}{2^{-m} \Gamma (1 - m)}$ the second member of (4) is the same function of $-m$ and $x$ that $J_m (x)$ is of $+m$ and $x$ and may be written $J_{-m}(x)$. \begin{flalign*} &\text{\indent Therefore }& z &= A J_m(x) + B J_{-m}(x) &&\phantom{Therefore } \tag{5} \end{flalign*} is the general solution of (1) unless $J_m(x)$ and $J_{-m}(x)$ should prove not to be independent. It is easily seen that when $m = 0$, $J_{-m}(x)$ and $J_m(x)$ become identical and (5) reduces to \[ z = (A + B) J_0(x) \] 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 $J_{-m} (x) = (-1)^m J_m(x)$ whenever $m$ is an integer, and consequently that the solution~(5) is general only when $m$ if real is fractional or incommensurable. 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 $a_0 = 1$; then the value which the second member of (4) assumes when $a_0 = 1$ 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 \[ z = \frac{F(m,x) - F(-m,x)}{ 2m}, \tag{6} \] but the limiting value which $\dfrac{F(m,x) - F(-m,x)}{ 2m}$ approaches as $m$ approaches zero is $[D_m F(m,x)]_{m=0}$ and consequently \[ z=[D_m F(m,x)]_{m=0} \tag{7} \] is a solution of the equation \[ \frac{d^2 z}{dx^2} + \frac{1}{x} \frac{dz}{dx} + z = 0, \tag{8} \] and the general solution of (8) is \begin{align*} z = {}& A J_0(x) + B[D_m F(m,x)]_{m=0}. \\[1ex] F(m,x) = x^m \bigg[ 1 &- \frac{x^2}{ 2^2 (m + 1)} + \frac{x^4}{ 2^4.2! (m + 1)(m + 2)} \\[1ex] &- \frac{x^6}{2^6.3!(m+1)(m+2)(m+3)} + \cdots \bigg]\\ % -----File: 036.png D_mF(m,x) = x^m\log x \bigg[ 1&-\frac{x^2}{2^2(m+1)} + \frac{x^4}{2^4.2!(m+1)(m+2)} - \cdots \bigg] & \\ +x^mD_m \bigg[ 1 &- \frac{x^2}{2^2(m+1)} + \frac{x^4}{2^4.2!(m+1)(m+2)} + \cdots \bigg].& \end{align*} The general term of the last parenthesis can be written \[ (-1)^k\frac{x^{2k}}{2^{2k}.k!(m+1)(m+2)\cdots(m+k)}, \] and its partial derivative with respect to $m$ is \begin{gather*} (-1)^k\frac{x^{2k}}{2^{2k}.k!} D_m\frac{1}{(m+1)(m+2)\cdots(m+k)}.\\ \log\frac{1}{(m+1)(m+2)\cdots(m+k)} = \begin{aligned}[t] -[ &\log(m+1)+\log(m+2)+\cdots\\ &+ \log(m+k)]. \end{aligned} \end{gather*} Take the $D_m$ of both members and we have \begin{align*} D_m & \frac{1}{(m+1)(m+2)\cdots(m+k)}\\ &=-\frac{1}{(m+1)(m+2)\cdots(m+k)} \left[ \frac{1}{m+1} + \frac{1}{m+2} + \cdots\frac{1}{m+k} \right]. \\ D_m\left[1\vphantom{\frac{x^2}{(0)}}\right. & -\frac{x^2}{2^2(m+1)} + \frac{x^4}{2^4.2!(m+1)(m+2)} - \frac{x^6}{2^6.3!(m+1)(m+2)(m+3)} \\ & +\cdots\left.\vphantom{\frac{x^2}{(0)}}\right] = \frac{x^2}{2^2}\frac{1}{(m+1)^2} - \frac{x^4}{2^4.2!}\frac{1}{(m+1)(m+2)} \left[ \frac{1}{m+1} + \frac{1}{m+2} \right] \\ & +\frac{x^6}{2^6.3!}\frac{1}{(m+1)(m+2)(m+3)} \left[ \frac{1}{m+1} + \frac{1}{m+2} + \frac{1}{m+3}\right] + \cdots \end{align*} and we have % recast to fit line \begin{align*} [D_mF(m,x)]_{m=0} &= J_0(x)\log x + \frac{x^2}{2^2(1!)^2}\frac{1}{1} - \frac{x^4}{2^4(2!)^2} \left(\frac{1}{1} + \frac{1}{2}\right) \\ &+ \frac{x^6}{2^6(3!)^2} \left( \frac{1}{1} + \frac{1}{2} + \frac{1}{3} \right) -\frac{x^8}{2^8(4!)^2} \left( \frac{1}{1} + \frac{1}{2} + \frac{1}{3} + \frac{1}{4} \right) + \cdots; \end{align*} \begin{flalign*} &\text{and} && \hspace{6em} z = AJ_0(x) + BK_0(x), \tag{9} \\ % recast to fit line &\text{where} & K_0&(x) = J_0(x)\log x + \frac{x^2}{2^2} - \frac{x^4}{2^4(2!)^2} \left( \frac{1}{1}+\frac{1}{2}\right) \\ &&& + \frac{x^6}{2^6(3!)^2}\left(\frac{1}{1} + \frac{1}{2} + \frac{1}{3}\right) -\frac{x^8}{2^8(4!)^2} \left( \frac{1}{1} + \frac{1}{2} +\frac{1}{3} + \frac{1}{4} \right) + \cdots \tag{10} \end{flalign*} is the general solution of Fourier's Equation (8). $K_0(x)$ is known as a \emph{Bessel's Function} of the \emph{Second Kind}. % -----File: 037.png \mypara{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~(\emph{a})]. The most general form of a homogeneous ordinary linear differential equation of the second order is \[ \frac{d^2y}{dx^2} + P \frac{dy}{dx} + Qy = 0 \tag{1} \] where $P$ and $Q$ are functions of $x$. Suppose that \[ y = v \tag{2} \] is a particular solution of (1). Substitute $y = vz$ in (1) and we get \[ v\frac{d^2z}{dx^2}+\left(2\frac{dv}{dx} + Pv\right)\frac{dz}{dx} = 0. \tag{3} \] Call $\dfrac{dz}{dx} = z'$. Then (3) becomes \[ v\frac{dz'}{dx}+\left(2\frac{dv}{dx}+ Pv\right)z'=0, \tag{4} \] a differential equation of the first order in which the variables can be separated. Multiply by $dx$ and divide by $vz'$ and (4) reduces to \[ \frac{dz'}{z'} + 2\frac{dv}{v} + Pdx = 0. \] Integrate and we have \begin{flalign*} &&\log{}&z' + \log v^2 + \int Pdx = C \\ &\text{or } & z'v^2 &= e^{C-\int\!Pdx}=Be^{-\int\!Pdx}, &&\phantom{or }\\ &&z'&=\frac{dz}{dx}=B\frac{e^{-\int\!Pdx}}{v^2},&\\ &&z &=A+B\int\frac{e^{-\int\!Pdx}}{v^2}dx;&\\ &\text{and } & y=&\ v\bigg(A+B\int \frac{e^{-\int\!Pdx}}{v^2}dx\bigg) && \tag{5} \end{flalign*} is the general solution of (1), the only arbitrary constants in the second member of (5) being those explicitly written, namely, $A$ and $B$.\\ (\emph{a})\quad Apply this formula to (1) Art.~14, \[ \frac{d^2z}{dx^2} + \alpha^2z=0; \tag{1} \] % -----File: 038.png given: $z=\cos\alpha x$, as a particular solution. Substituting in (5) we have since $P=0$ \begin{align*} z &= \cos\alpha x\left(A+B\int\frac{dx}{\cos^2\alpha x}\right) \\ &= \cos\alpha x\left(A+\frac{B}{\alpha}\tan\alpha x\right) \\ &= A\cos\alpha x + B_1\sin\alpha x, \tag{2} \end{align*} as the general solution of (1), and this agrees perfectly with (5) Art.~14.\\ (\emph{b})\quad Take equation (1) Art.~15. \[ x^2\frac{d^2z}{dx^2}+2x\frac{dz}{dx}-m(m+1)z=0; \tag{1} \] given: $z=x^m$, as a particular solution. \noindent Here $P=\dfrac{2}{x}$, $\displaystyle\int Pdx=2\log x =\log x^2$, and $e^{-\int\!Pdx}=\dfrac{1}{x^2}$. Hence by (5) \[ z = x^m \left( A + B\int\frac{dx}{x^{2m+2}} \right) = x^m \left( A + \frac{B}{-2m-1} \frac{1}{x^{2m+1}} \right),\\[-3ex] \] \begin{flalign*} &\text{that is} & z= &\ Ax^m+\frac{B_1}{x^{m+1}} &&\phantom{that is} \tag{2} \end{flalign*} is the general solution of (1), and agrees with (2) Art.~15.\\ (\emph{c})\quad Take Legendre's Equation, (2) Art.~16. \[ (1-x^2)\frac{d^2z}{dx^2}-2x\frac{dz}{dx}+m(m+1)z=0; \tag{1} \] given: $z=P_m(x)$, as a particular solution. Here $P=\dfrac{-2x}{1-x^2}$, $\displaystyle\int Pdx=\log(1- x^2)$, and $e^{-\int\!Pdx}=\dfrac{1}{1-x^2}$. \begin{flalign*} &\text{Hence by (5)} && z = P_m(x) \left( A + B\int\frac{dx}{(1-x^2)[P_m(x)]^2} \right) &&\phantom{Hence by } \tag{2} \end{flalign*} is the general solution of (1) and must agree with (10) Art.~16, if $m$ is an integer, and therefore \[ Q_m(x)=CP_m(x)\int\frac{dx}{(1-x^2)[P_m(x)]^2} \tag{3} \] where $C$ is as yet undetermined, and no constant term is to be understood with the integral in the second member.\\ (\emph{d})\quad Take Bessel's Equation, (1) Art.~17. \[ \frac{d^2z}{dx^2} + \frac{1}{x}\frac{dz}{dx} + \left(1-\frac{m^2}{x^2}\right)z = 0; \tag{1} \] given: $z=J_m(x)$, as a particular solution. % -----File: 039.png Here $P=\dfrac{1}{x}$, $\displaystyle\int Pdx=\log x$, and $e^{-\int\!Pdx}=\dfrac{1}{x}$. Hence by (5) \[ z=J_m(x)\left(A+B\int\frac{dx}{x[J_m(x)]^2}\right) \tag{2} \] is the general solution of Bessel's Equation. If $m=0$ (2) becomes \[ z=J_0(x)\left(A+B\int\frac{dx}{x[J_0(x)]^2}\right) \tag{3} \] and must agree with (9) Art.~17. Therefore \[ K_0(x)=CJ_0(x)\int\frac{dx}{x[J_0(x)]^2}, \tag{4} \] 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). \label{ch1end} % -----File: 040.png \mychap{CHAPTER II.}{DEVELOPMENT IN TRIGONOMETRIC SERIES.} \label{ch2start} \mypara{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 \[ b_0 + b_1\cos x + b_2\cos 2x + \cdots + a_1\sin x + a_2\sin 2x + \cdots \] 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 $x=\pi$. To fix our ideas let us suppose that we have a curve, \[ y=f(x), \] given, and that we wish to form the equation, \[ y = a_1\sin x + a_2\sin 2x + a_3\sin 3x + \cdots, \] of a curve which shall coincide with so much of the given curve as lies between the points corresponding to $x=0$ and $x=\pi$. It is clear that in the equation \[ y = a_1 \sin x \tag{1} \] $a_1$ 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 $a_1$ is the only unknown quantity and which will therefore give us one and only one value for $a_1$. In like manner the curve \[ y = a_1 \sin x + a_2 \sin 2x \] may be made to pass through any two arbitrarily chosen points whose abscissas lie between 0 and $\pi$ provided that the abscissas are not equal; and \[ y = a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + \cdots + a_n \sin nx \] may be made to pass through any $n$ arbitrarily chosen points whose abscissas lie between 0 and $\pi$ 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 $x=\pi$ it has one and only one value, and if between $x=0$ and $x=\pi$ it is finite and continuous, or if discontinuous has only \emph{finite discontinuities} (v.\ Int.\ Cal.\ Art.~83, p.~78), the coefficients in \[ y = a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + \cdots + a_n \sin nx \tag{2} \] % -----File: 041.png can be determined so that the curve represented by (2) will pass through any $n$ arbitrarily chosen points of the curve \[ y = f(x) \tag{3} \] whose abscissas lie between $0$ and $\pi$ 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. \markright{DEVELOPMENT IN TRIGONOMETRIC SERIES.} Call $\dfrac{\pi}{ n+1} = \Delta x$; then the coördinates of the $n$ points will be $[\Delta x,f(\Delta x)],$ $[2\Delta x,f(2\Delta x)]$, $[3\Delta x,f(3\Delta x)]$, $\cdots\ [n\Delta x,f(n\Delta x)]$. Substitute them in (2) and we have \begin{equation*} \left.\begin{alignedat}{5} f( \Delta x) &= {}& a_1 \sin \Delta x &+{}& a_2 \sin 2 \Delta x &+{} & a_3 \sin 3 \Delta x &+ \cdots +{}& a_n \sin n \Delta x\phantom{,} & \\ f(2\Delta x) &= {}& a_1 \sin 2\Delta x &+{}& a_2 \sin 4 \Delta x &+{} & a_3 \sin 6 \Delta x &+ \cdots +{}& a_n \sin 2n \Delta x\phantom{,} & \\ f(3\Delta x) &= {}& a_1 \sin 3\Delta x &+{}& a_2 \sin 6 \Delta x &+{} & a_3 \sin 9 \Delta x &+ \cdots +{}& a_n \sin 3n \Delta x\phantom{,} & \\ \vdots \quad && \vdots \qquad && \vdots \qquad & & \vdots \qquad && \vdots \qquad & \\ f(n\Delta x) &= {}& a_1 \sin n\Delta x &+{}& a_2 \sin 2n \Delta x &+{} & a_3 \sin 3n \Delta x &+ \cdots +{}& a_n \sin n^2 \Delta x, & \end{alignedat} \right\} \tag{4} \end{equation*} $n$ equations of the first degree to determine the $n$ coefficients $a_1$, $a_2$, $a_3$, $\cdots$ $a_n$. 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 $a_1$, $a_2$, $a_3$, $a_4$, and $a_5$, so that \[ y = a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + a_4 \sin 4x + a_5 \sin 5x \tag{5} \] shall pass through the five points of the line \[ y = x \] which have the abscissas $\dfrac{\pi}{6}$, $\dfrac{2\pi}{6}$, $\dfrac{3\pi}{6}$, $\dfrac{4\pi}{6}$, and $\dfrac{5\pi}{6}$, $\dfrac{\pi}{6}$ here being $\Delta x$. We must now solve the equations \begin{equation*} \left.\begin{alignedat}{6} \frac{\pi}{6}\: &= {}& a_1 \sin \:\frac{ \pi}{6}\: &+{} & a_2 \sin \:\frac{2\pi}{6}\: &+{}& a_3 \sin \:\frac{3\pi}{6}\: &+{} & a_4 \sin \:\frac{4\pi}{6}\: &+{}& a_5 \sin \:\frac{5\pi}{6}\:\phantom{.} \\ \frac{2\pi}{6} &= & a_1 \sin \frac{2\pi}{6} &+{} & a_2 \sin \:\frac{4\pi}{6}\: &+{}& a_3 \sin \:\frac{6\pi}{6}\: &+{} & a_4 \sin \:\frac{8\pi}{6}\: &+{}& a_5 \sin \frac{10\pi}{6}\phantom{.} \\ \frac{3\pi}{6} &= & a_1 \sin \frac{3\pi}{6} &+{} & a_2 \sin \:\frac{ 6\pi}{6}\: &+{}& a_3 \sin \:\frac{ 9\pi}{6}\: &+{} & a_4 \sin \frac{12\pi}{6} &+{}& a_5 \sin \frac{15\pi}{6}\phantom{.} \\ \frac{4\pi}{6} &= {}& a_1 \sin \frac{4\pi}{6} &+{} & a_2 \sin \:\frac{ 8\pi}{6}\: &+{}& a_3 \sin \frac{12\pi}{6} &+{} & a_4 \sin \frac{16\pi}{6} &+{}& a_5 \sin \frac{20\pi}{6}\phantom{.} \\ \frac{5\pi}{6} &= {}& a_1 \sin \frac{5\pi}{6} &+{} & a_2 \sin \frac{10\pi}{6} &+{}& a_3 \sin \frac{15\pi}{6} &+{} & a_4 \sin \frac{20\pi}{6} &+{}& a_5 \sin \frac{25\pi}{6}. \end{alignedat} \right\} \tag{6} \end{equation*} % -----File: 042.png Multiply the first equation by $2 \sin \dfrac{\pi}{6}$, the second by $2 \sin \dfrac{2 \pi }{6}$, the third by $2\sin\dfrac{3 \pi}{6}$, the fourth by $2\sin\dfrac{4\pi}{6}$, the fifth by $2 \sin \dfrac{5 \pi }{6}$ and add the equations. The coefficient of $a_2$ is \begin{flalign*} &&2 \sin \frac{\pi}{6} \sin \frac{2 \pi}{ 6} &+ 2 \sin \frac{2 \pi}{6} \sin \frac{4 \pi}{6} + 2 \sin \frac{3 \pi }{6} \sin \frac{6 \pi}{6} + 2 \sin \frac{4 \pi}{6} \sin \frac{8 \pi}{6} & \phantom{\text{but}}\\ &&&+ 2 \sin \frac{5 \pi}{6} \sin \frac{10 \pi}{6}; \\ &\text{but} && \qquad 2 \sin \dfrac{\pi }{6} \sin \dfrac{2 \pi}{6} = \cos \dfrac{\pi}{6} - \cos \dfrac{3 \pi }{6}, \text{\&c.} \end{flalign*} Hence the coefficient of $a_2$ becomes \begin{equation*} \left.\begin{alignedat}{5} &\cos \:\frac{\pi}{6} &+{}& \cos \frac{2 \pi}{6} &+{}& \cos \frac{3 \pi}{6} &+{}& \cos \:\frac{4 \pi}{6} &+{}& \cos \:\frac{5 \pi}{6}\\ -{}&\cos \frac{3 \pi}{6} &-{}& \cos \frac{6 \pi}{6} &-{}& \cos \frac{9 \pi}{6} &-{}& \cos \frac{12 \pi}{6} &-{}& \cos \frac{15 \pi}{6} \end{alignedat} \right\} \tag{7} \end{equation*} and this may be reduced by the aid of an important Trigonometric formula which we proceed to establish. \mypara{20.} \textsc{Lemma.} \begin{align*} \cos \theta &+ \cos 2 \theta + \cos 3 \theta + \cdots + \cos n \theta = - \frac{1}{2} + \frac{1}{2} \cfrac{\sin(2n+1) \dfrac{\theta}{2}}{\sin \dfrac{\theta}{2}}. \tag{1} \end{align*} For let $S=\cos \theta + \cos 2 \theta + \cos 3 \theta + \cdots + \cos n \theta$ and multiply by $2 \cos \theta$. \begin{align*} 2S \cos \theta &= 2 \cos^2 \theta + 2 \cos \theta \cos 2 \theta + 2 \cos \theta \cos 3 \theta + \cdots + 2 \cos \theta \cos n \theta&&\hspace{1em}\\ &= 1 + \cos \theta + \cos 2 \theta + \cdots + \cos (n - 1) \theta\\ &\hspace{1.85em}+ \cos 2\theta + \cos 3\theta + \cos 4\theta + \cdots + \cos (n + 1)\theta\\ &= 2S + 1 + \cos (n + 1) \theta - \cos \theta - \cos n \theta. \qquad \text{Hence}\\[-5.5ex] \end{align*} \begin{flalign*} && S &= - \frac{1}{2} + \frac{\cos n \theta - \cos (n + 1) \theta }{ 2(1- \cos \theta )} && \\ &\text{or }& S &= - \frac{1}{2} + \frac{1}{2} \cfrac{\sin (2n + 1) \dfrac{\theta}{2}}{ \sin \dfrac{\theta}{ 2 }}. && \tag*{\scriptsize Q.E.D.} \end{flalign*} % -----File: 043.png \mypara{21.} Applying (1) Art.~20 to (7) Art.~19 the coefficient of $a_2$ reduces to \[ \cfrac{\sin\dfrac{11\pi}{12}}{2\sin\dfrac{ \pi}{12}} - \cfrac{\sin\dfrac{33\pi}{12}}{2\sin\dfrac{3\pi}{12}}; \] \begin{flalign*} &\text{but} && \frac{11\pi}{12} = \pi - \frac{\pi}{12},\text{ and }\frac{33\pi}{12} = 3\pi - \frac{3\pi}{12}; &&\phantom{\text{but}} \end{flalign*} \begin{flalign*} &\text{therefore} && \frac{\sin \Big( \pi- \dfrac{\pi}{12} \Big) }{2\sin\dfrac{\pi}{12}} - \frac{\sin \Big(3\pi- \dfrac{3\pi}{12} \Big) }{2\sin\dfrac{3\pi}{12}} = \frac12 - \frac12 = 0, &&\phantom{\text{therefore}} \end{flalign*} and $a_2$ vanishes. In like manner it may be shown that the coefficients of $a_3$, $a_4$, and $a_5$ vanish. The coefficient of $a_1$ is \begin{align*} &\ 2\sin^2 \frac{ \pi}{6} + 2\sin^2 \frac{2\pi}{6} + 2\sin^2 \frac{3\pi}{6} + 2\sin^2 \frac{4\pi}{6} + 2\sin^2 \frac{5\pi}{6} \\[1ex] = {}& \makebox[20.5em]{1 \hfill + \hfill 1 \hfill + \hfill 1 \hfill + \hfill 1 \hfill + \hfill 1} \\ &\!\!- \cos \frac{2\pi}{6} - \cos \frac{4\pi}{6} - \cos \frac{6\pi}{6} - \cos \frac{8\pi}{6} -\cos \frac{10\pi}{6} \\[1ex] = {}& 5 + \frac12 - \cfrac{\sin \dfrac{11\pi}{6}}{2 \sin \cfrac{\pi}{6}} = 5\tfrac12 - \cfrac{ \sin \left( 2\pi-\dfrac{\pi}{6} \right) } { 2\sin \cfrac{\pi}{6} } = 6. \end{align*} The first member of the final equation is \begin{gather*} \frac{2\pi}{6} \sin \frac{ \pi}{6} + 2 \frac{2\pi}{6} \sin \frac{2\pi}{6} + 2 \frac{3\pi}{6} \sin \frac{3\pi}{6} + 2 \frac{4\pi}{6} \sin \frac{4\pi}{6} + 2 \frac{5\pi}{6} \sin \frac{5\pi}{6}. \quad\text{Hence}\\ a_1 = \frac26 \sum_{k=1}^{k=5} \frac{k\pi}{6} \sin \frac{k\pi}{6} = \frac{\pi}{6} (2 + \surd3) = 2\text{\quad approximately.} \end{gather*} If we multiply the first equation of (6) Art.~19 by $2\sin\dfrac{2\pi}{6}$, the second by $2\sin\dfrac{4\pi}{6}$, the third by $2\sin\dfrac{6\pi}{6}$, the fourth by $2\sin\dfrac{8\pi}{6}$, the fifth by $2\sin\dfrac{10\pi}{6}$, add and reduce as before we shall find \begin{align*} a_2 &= \frac26 \sum_{k=1}^{k=5} \frac{k\pi}{6} \sin \frac{2k\pi}{6} = - \frac{\pi}{6} \surd3 = -0.9; % -----File: 044.png \intertext{and in like manner we get} a_3 &= \frac{2}{6}\sum_{k=1}^{k=5} \frac{k\pi}{6}\sin\frac{3k\pi}{6} = \frac{\pi}{6} = 0.5\\ a_4 &= \frac{2}{6}\sum_{k=1}^{k=5} \frac{k\pi}{6}\sin\frac{4k\pi}{6} =-\frac{\pi\surd3}{18} = -0.3\\ a_5 &= \frac{2}{6}\sum_{k=1}^{k=5} \frac{k\pi}{6}\sin\frac{5k\pi}{6} = \frac{\pi}{6}(2-\surd3) = 0.1. \end{align*} Therefore \[ y=2\sin x - 0.9\sin 2x + 0.5\sin 3x - 0.3\sin 4x + 0.1\sin 5x \tag{1} \] cuts the curve $y=x$ at the five points whose abscissas are $\dfrac{\pi}{6}$, $\dfrac{2\pi}{6}$, $\dfrac{3\pi}{6}$, $\dfrac{4\pi}{6}$, and $\dfrac{5\pi}{6}$. \mypara{22.} The equations (4) Art.~19 can be solved by exactly the same device. To find any coefficient $a_m$ multiply the first equation by $2\sin m\Delta x$, the second by $2\sin 2m\Delta x$, the third by $2\sin 3m\Delta x$, \&c.\ and add. The coefficient of any other $a$ as $a_k$ in the resulting equation will be \begin{align*} 2&\sin k\Delta x\sin m\Delta x + 2\sin 2k\Delta x\sin 2m\Delta x + 2\sin 3k \Delta x\sin 3m\Delta x + \cdots\\ &\qquad+ 2\sin nk\Delta x\sin nm\Delta x\\ ={}& \cos(m-k)\Delta x + \cos 2(m-k)\Delta x + \cos 3(m-k)\Delta x + \cdots + \cos n(m-k)\Delta x\\ -{}&\cos(m+k)\Delta x - \cos 2(m+k)\Delta x - \cos 3(m+k)\Delta x - \cdots - \cos n(m+k)\Delta x\\ ={}& \cfrac{\sin\dfrac{2n+1}{2}(m-k)\Delta x}{2\sin\dfrac{(m-k)\Delta x}{2}} -\cfrac{\sin\dfrac{2n+1}{2}(m+k)\Delta x}{2\sin\dfrac{(m+k)\Delta x}{2}}; \text{\qquad by (1) Art.~20.} \end{align*} \[ \frac{2n+1}{2}=n+1-\frac{1}{2} \quad \text{and} \quad (n + 1)\Delta x = \pi . \] Hence the coefficient of $a_k$ may be written \[ \cfrac{ \sin \Big[ (m-k)\pi - \dfrac{(m-k)\Delta x}{2} \Big] } {2\sin\dfrac{(m-k)\Delta x}{2} } -\cfrac{ \sin \Big[ (m+k)\pi - \dfrac{(m+k)\Delta x}{2} \Big] } {2\sin\dfrac{(m+k)\Delta x}{2} } \] but this is equal to $\dfrac{1}{2}-\dfrac{1}{2}$ or $-\dfrac{1}{2}+\dfrac{1}{2}$ according as $m-k$ is odd or even and so is zero in either case. % -----File: 045.png The coefficient of $a_m$ will be \begin{align*} &\;2\sin^2m\Delta x + 2\sin^22m\Delta x + 2\sin^23m\Delta x + \cdots + 2\sin^2nm\Delta x\\ &= \qquad 1 \qquad + \qquad 1 \qquad + \qquad 1 \qquad + \quad \cdots\quad + \qquad 1\\ &\quad-\cos2m\Delta x - \cos4m\Delta x - \cos6m\Delta x - \;\cdots\; - \cos2nm\Delta x\\ &=n+\frac{1}{2}-\frac{\sin(2n+1)m\Delta x}{2\sin m\Delta x}, \text{ by (1) Art.~20.} \end{align*} \begin{flalign*} &\text{\indent But } & (2n+1)m\Delta x = 2m(n & +1)\Delta x-m\Delta x = 2m\pi-m\Delta x, &&\phantom{therefore } \\ &\text{therefore } & \frac{\sin(2n+1)m\Delta x}{2\sin m\Delta x} & = \frac{\sin(2m\pi-m\Delta x)}{2\sin m\Delta x} = -\frac{1}{2}, && \end{flalign*} and the coefficient of $a_m$ is $n+1$. The first member of our final equation will be \[ 2\sum_{k=1}^{k=n}f(k\Delta x)\sin km\Delta x. \] Hence \[ a_m=\frac{2}{n+1}\sum_{k=1}^{k=n}f(k\Delta x)\sin km\Delta x, \tag{1} \] and the curve \[ y = a_1\sin x + a_2\sin 2x + \cdots + a_n\sin nx, \tag{2} \] where the coefficients are given by (1) will pass through the $n$ points of the curve $y =f(x)$ whose abscissas are $\Delta x$, $2\Delta x$, $3\Delta x$, $\cdots$ $n\Delta x$. $\Delta x$ being $\dfrac{\pi}{n+1}$. 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 $a_1$, $a_2$, $a_3$, $\cdots$ $a_n$ that can satisfy these equations. Consequently the solution which we have obtained is the only solution possible. \mypara{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 $x=\pi$, 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 $y=f(x)$. % -----File: 046.png Let us see what limiting value $a_m$ approaches as $n$ is indefinitely increased. \begin{align*} \phantom{(1) Art.~22.}a_m &=\frac{2}{n+1}\sum_{k=1}^{k=n}f(k\Delta x)\sin km\Delta x \tag*{(1) Art.~22.}\\ &=\frac{2\Delta x}{\pi}\sum_{k=1}^{k=n}f(k\Delta x)\sin km\Delta x \end{align*} % recast to fit line \begin{flalign*} = \frac{2}{\pi} \left[ \begin{alignedat}{1} f( \Delta x)\sin m\Delta x.\Delta x + f(2\Delta x)\sin 2m\Delta x.\Delta x & + \cdots \\ & + f(n\Delta x)\sin nm\Delta x.\Delta x \end{alignedat} \right] \end{flalign*} \begin{flalign*} = \frac{2}{\pi} \left[ \begin{alignedat}{1} f( \Delta x)\sin m\Delta x.\Delta x + f( 2\Delta x)\sin2m\Delta x.\Delta &x + \cdots\\ + f(&\pi-\Delta x)\sin m(\pi-\Delta x).\Delta x \end{alignedat} \right] \end{flalign*} since $\Delta x=\dfrac{\pi}{n+1}$. As $n$ is increased indefinitely $\Delta x$ approaches zero as a limit. Hence the limiting value of $a_m$ as $n$ increases indefinitely is \begin{flalign*} & \frac{2}{\pi}\limit_{\Delta x\doteq 0}\left[ \begin{alignedat}{1} f( \Delta x)\sin m\Delta x.\Delta x + f( 2\Delta x)\sin 2m\Delta x.\Delta x &+ \cdots\\ + f(\pi-\Delta x)&\sin m(\pi-\Delta x).\Delta x \end{alignedat} \right]\footnotemark &&\\[-8ex] \end{flalign*} \footnotetext{We shall use the sign $\doteq$ for \emph{approaches}. $\Delta x\doteq 0$ is read $\Delta x$ approaches zero.} \begin{flalign*} &&= \frac{2}{\pi}\int\limits_0^\pi f(x)\sin mx.dx. \quad\text{[v.\ Int.\ Cal.\ Arts.~80, 81.]}\\[-5ex] \end{flalign*} \begin{flalign*} &\text{\indent Hence} & f(x)=a_1\sin x + a_2\sin 2x + a_3\sin 3x + \cdots, & \tag{2}& \end{flalign*} where any coefficient $a_m$ is given by the formula \[ a_m=\frac{2}{\pi}\int\limits_0^\pi f(x)\sin mx.dx, \tag{3} \] is a true development of $f(x)$ for all values of $x$ between $x=0$ and $x=\pi$ \emph{provided that the series} (2) \emph{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 $x=\pi$ 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 $x=\pi$; but even when $f(x)$ is not zero for $x=0$ and $x=\pi$ the development given above holds good for any value of $x$ between zero and $\pi$ no matter how near it may be taken to either of these values. \mypara{24.} Instead of actually performing the elimination in equations (4) Art.~19 and getting a formula for $a_m$ in terms of $n$, and then letting $n$ increase indefinitely, we might have saved labor by the following method. % -----File: 047.png Return to equations (4) Art.~19 and multiply the first by $\Delta x\sin m\Delta x$, the second by $\Delta x\sin 2m\Delta x$, and so on, that is multiply each equation by $\Delta x$ times the coefficient of $a_m$ in that equation, and then add the equations. We get as the coefficient of $a_k$ \[ \sin k\Delta x\sin m\Delta x.\Delta x +\sin 2k\Delta x\sin 2m\Delta x.\Delta x + \cdots +\sin nk\Delta x\sin nm\Delta x.\Delta x. \] Let us find its limiting value as $n$ is indefinitely increased. It may be written, since $(n+1)\Delta x=\pi$, \begin{flalign*} \limit_{\Delta x\doteq 0} \left[ \begin{alignedat}{1} \sin k\Delta x\sin m\Delta x.\Delta x + \sin2k\Delta x\sin2m&\Delta x.\Delta x +\cdots\\ + \sin &k(\pi-\Delta x)\sin m(\pi-\Delta x).\Delta x \end{alignedat} \right] \end{flalign*} \[ = \int\limits_0^\pi\sin kx\sin mx.dx; \] \begin{flalign*} &\text{but } & \int\limits_0^\pi\sin kx\sin mx.dx &=\tfrac{1}{2}\int\limits_0^\pi[\cos(m-k)x-\cos(m+k)x]dx &&\qquad{}\\ &&& = 0 \text{ if $m$ and $k$ are not equal.} && \end{flalign*} The coefficient of $a_m$ is \[ \Delta x(\sin^2 m\Delta x + \sin^2 2m\Delta x + \sin^2 3m\Delta x + \cdots + \sin^2 nm\Delta x). \] Its limiting value \begin{gather*} \limit_{\Delta x\doteq 0} \bigg[ \sin^2 m\Delta x.\Delta x + \sin^2 2m\Delta x.\Delta x + \cdots + \sin^2 m(\pi - \Delta x)\Delta x \bigg] \\ = \int\limits_0^\pi\sin^2 mx.dx = \frac{\pi}{2}. \end{gather*} The first member is \[ f( \Delta x)\sin m\Delta x.\Delta x + f(2\Delta x)\sin 2m\Delta x.\Delta x + \cdots + f(n\Delta x)\sin mn \Delta x.\Delta x \] and its limiting value is \[ \int\limits_0^\pi f(x)\sin mx.dx. \] Hence the limiting form approached by the final equation as $n$ is increased is \[ \int\limits_0^\pi f(x)\sin mx.dx = \frac{\pi}{2}a_m. \] \begin{flalign*} &\text{\indent Whence } & a_m = \frac{2}{\pi}\int\limits_0^\pi f(x)\sin mx.dx && \text{ as before.}& \end{flalign*} This method is practically the same as \emph{multiplying the equation} \[ f(x) = a_1\sin x + a_2\sin 2x + a_3\sin 3x + \cdots \tag{1} \] \emph{by $\mathop\textit{sin} mx.dx$ and integrating both members from zero to $\pi$}. % -----File: 048.png It is exceedingly important to realize that the short method of determining any coefficient $a_m$ of the series (1) which has just been described in the italicized paragraph, is essentially the same as that of obtaining $a_m$ 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. \mypara{25.} We see, then, that any function of $x$ which is single-valued, finite, and continuous between $x = 0$ and $x = \pi$, 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 \[ f(x) = a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + \cdots \tag{1} \] \begin{flalign*} &\text{where }& a_m = \frac{2}{\pi}\int\limits_{0}^{\pi} f(x) \sin mx.dx = \frac{2}{\pi}\int\limits_{0}^{\pi} f(\alpha) \sin m\alpha.d\alpha; &&\phantom{where } \tag{2} \end{flalign*} and the series and the function will be identical for all values of $x$ between $x = 0$ and $x = \pi$, not including the values $x = 0$ and $x = \pi$ 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 above\label{notep38}\footnote {Provided the function has not an infinite number of maxima and minima in the neighborhood of a point. v.\ Arts.~37--38.} 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. \mypara{26.} Let us obtain a few sine developments. \begin{flalign*} &\text{\indent (\emph{a})\quad Let }&f(x)=x.& &&\phantom{\indent (a)\quad Let\ } \tag{1} \end{flalign*} \begin{flalign*} &\text{We have }&x = a_1 \sin x + {}& a_2 \sin 2x + a_3\sin 3x + \cdots &&\phantom{We\ have } \tag{2}\\ &\text{where }&a_m = \frac{2}{ \pi} & \int\limits_{0}^{\pi} x \sin mx.dx&& \tag{3} \end{flalign*} % -----File: 049.png \[ \int x\sin mx.dx = \frac{1}{m^2}(\sin mx - mx\cos mx), \] \[ \int\limits_0^\pi x\sin mx.dx = -\frac{(-1)^m \pi}{m}, \] \begin{flalign*} &\text{and } & x = 2 \left( \frac{\sin x}{1} - \frac{\sin 2x}{2} + \frac{\sin 3x}{3} - \frac{\sin 4x}{4} + \cdots \right) & \tag{4}&& \end{flalign*} \begin{flalign*} &\indent \text{(\emph{b})\quad Let } & f(x) = 1. & \tag{1} &&\phantom{\indent(b)\quad Let\ } \end{flalign*} \[ a_m = \frac{2}{\pi}\int\limits_0^\pi\sin mx.dx; \tag{2} \] \[ \int\sin mx.dx = -\frac{\cos mx}{m}, \] \begin{align*} \int\limits_0^\pi\sin mx.dx &= \frac{1}{m}(1-\cos m\pi) = \frac{1}{m}[1-(-1)^m]\\ &= 0 \text{ if $m$ is even}\\ &= \frac{2}{m} \text{ if $m$ is odd.} \end{align*} \begin{flalign*} &\text{Hence } & 1 = \frac{4}{\pi} \left( \frac{\sin x}{1} + \frac{\sin 3x}{3} + \frac{\sin 5x}{5} + \frac{\sin 7x}{7} + \cdots \right). & \tag{3} &&\phantom{Hence} \end{flalign*} It is to be noticed that (3) gives at once a sine development for any constant $c$. It is, \[ c = \frac{4c}{\pi} \left( \frac{\sin x}{1} + \frac{\sin 3x}{3} + \frac{\sin 5x}{5} + \cdots \right). \tag{4} \] If we substitute $x=\dfrac{\pi}{2}$ in (4)\,(\emph{a}) or (3)\,(\emph{b}) we get a familiar result, namely \[ \frac{\pi}{4}=\frac{1}{1}-\frac{1}{3}+\frac{1}{5}-\frac{1}{7}+\cdots, \tag{5} \] a formula usually derived by substituting $x=1$ in the power series for $\tan^{-1}x$. (v.\ Dif.\ Cal.\ Art.~135.) (4)\,(\emph{a}) does not hold good when $x=\pi$, and (3)\,(\emph{b}) fails when $x=0$ and when $x=\pi$, for in all these cases the series reduces to zero.\\ %[Illustration] \pngright{001.png}{626}{390}{-24} (\emph{c})\quad Let $f(x)=x$ from $x=0$ to $x=\dfrac{\pi}{2}$ and $f(x)=\pi-x$ from $x=\dfrac{\pi}{2}$ to $x=\pi$. That is, let $y=f(x)$ 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 % -----File: 050.png \[ \int\limits_0^\pi f(x)\sin mx.dx \quad \text{into}\quad \int\limits_0^\frac{\pi}{2}f(x)\sin mx.dx + \int\limits_\frac{\pi}{2}^\pi f(x)\sin mx.dx. \] We have, then, \begin{align*} a_m &= \frac{2}{\pi} \int\limits_0^\frac{\pi}{2}x\sin mx.dx + \frac{2}{\pi} \int\limits_\frac{\pi}{2}^\pi(\pi-x)\sin mx.dx \tag{1}\\ &= \frac{4}{m^2\pi}\sin m \frac{\pi}{2}.\\[-6ex] \end{align*} \begin{flalign*} &\text{\indent But}& \begin{alignedat}[t]{5} \sin m \dfrac{\pi}{2} &= 1 && \text{if}\quad && m=1\quad && \text{or}\quad && 4k+1 \\[-1ex] &= 0 && `` && m=2 && \,`` && 4k+2 \\ &= -1\;\; && `` && m=3 && \,`` && 4k+3 \\ &= 0 && `` && m=4 && \,`` && 4k. \\ \end{alignedat} &&\phantom{\indent But} \end{flalign*} Hence if $y=f(x)$ represents our broken line, \[ f(x)=\frac{4}{\pi}\left(\frac{\sin x}{1^2}-\frac{\sin 3x}{3^2}+\frac{\sin 5x}{5^2}-\frac{\sin 7x}{7^2}+\cdots\right). \tag{2} \] When $x=\dfrac{\pi}{2}$ $f(x)=\dfrac{\pi}{2}$ and we have \[ \frac{\pi^2}{8} = \frac{1}{1^2} + \frac{1}{3^2} + \frac{1}{5^2} + \frac{1}{7^2} + \cdots \tag{3} \] (\emph{d})\quad As a case where the function has a finite discontinuity, let \begin{alignat*}{4} f(x)=1 \quad& \text{from} \quad&& x=0\quad && \text{to}\quad && x=\dfrac{\pi}{2} \quad \text{and}\\ f(x)=0 \quad& \;\;\text{``} && x=\dfrac{\pi}{2} && \;\text{``} && x=\pi. \end{alignat*} $y=f(x)$ will in this case represent the locus in the figure. %[Illustration] \pngrightl{002.png}{675}{3}{-24} As before \begin{align*} \int\limits_0^\pi f(x)\sin mx.dx & = \int\limits_0^\frac{\pi}{2} f(x)\sin mx.dx \\ &+ \int\limits_\frac{\pi}{2}^\pi f(x)\sin mx.dx. \end{align*} \[ a_m = \frac{2}{\pi} \int\limits_0^\frac{\pi}{2}\sin mx.dx + \frac{2}{\pi}\int\limits_\frac{\pi}{2}^\pi 0 .\sin mx.dx. \tag{1} \] % -----File: 051.png \[ a_m=\frac{2}{\pi} \int\limits_0^\frac{\pi}{2}\sin mx.dx =\frac{2}{\pi}\frac{1}{m}\left(1-\cos m\frac{\pi}{2}\right). \] \begin{flalign*} &\text{\indent But}& \begin{alignedat}[t]{5} \cos m \dfrac{\pi}{2} &= \phantom{-}0\quad && \text{if}\quad && m=1\quad && \text{or}\quad && 4k+1 \\[-1ex] &=-1 && `` && m=2 && \,`` && 4k+2 \\ &= \phantom{-}0 && `` && m=3 && \,`` && 4k+3 \\ &= \phantom{-}1 && `` && m=4 && \,`` && 4k. \\ \end{alignedat} &&\phantom{\indent But} \end{flalign*} Hence \[ f(x) = \frac{2}{\pi} \left( \frac{ \sin x}{1} + \frac{2\sin 2x}{2} + \frac{ \sin 3x}{3} + \frac{ \sin 5x}{5} + \frac{2\sin 6x}{6} + \frac{ \sin 7x}{7} + \cdots \right). \tag{2} \] If $x=\dfrac{\pi}{2}$ the second member of (2) reduces to $\dfrac{1}{2}$, for \[ \frac{2}{\pi} \left( \frac{1}{1} - \frac{1}{3} + \frac{1}{5} - \frac{1}{7} + \cdots \right) = \frac{1}{2}\qquad \text{by (5) ($b$)}; \] and we see that the series represents the function completely for all values of $x$ between $x=0$ and $x=\pi$ except for $x=\dfrac{\pi}{2}$ and there it has a value which is the mean of the values approached by the function as $x$ approaches $\dfrac{\pi}{2}$ from opposite sides. \EXAMPLE{S} Obtain the following developments:--- \begin{flalign*} &\indent\text{(1)}\quad x^2 = \dfrac{2}{\pi} \bigg[ \bigg( \dfrac{\pi^2}{1} - \dfrac{4}{1^3} \bigg) \sin x - \dfrac{\pi^2}{2}\sin 2x + \bigg( \dfrac{\pi^2}{3} - \dfrac{4}{3^3} \bigg) \sin 3x - \dfrac{\pi^2}{4}\sin 4x \\ & \hspace{8em} + \bigg( \dfrac{\pi^2}{5} - \dfrac{4}{5^3} \bigg) \sin 5x - \cdots \bigg].\\ % &\indent\text{(2)}\quad x^3 = \dfrac{2}{\pi} \bigg[ \bigg( \dfrac{\pi^3}{1} - \dfrac{6\pi}{1^3} \bigg) \sin x - \bigg( \dfrac{\pi^3}{2} - \dfrac{6\pi}{2^3} \bigg) \sin 2x + \bigg( \dfrac{\pi^3}{3} - \dfrac{6\pi}{3^3} \bigg) \sin 3x \\ &\hspace{8em} - \bigg( \dfrac{\pi^3}{4} - \dfrac{6\pi}{4^3} \bigg) \sin 4x + \cdots \bigg]. \\ % &\indent\text{(3)}\quad f(x) = \dfrac{2}{\pi} \bigg[ \dfrac{\sin x}{1^2} + \dfrac{ \pi}{2^2}\sin 2x - \dfrac{\sin 3x}{3^2} - \dfrac{2\pi}{4^2}\sin 4x + \dfrac{\sin 5x}{5^2} \\ &\hspace{8em} + \dfrac{3\pi}{6^2}\sin 6x - \cdots \bigg], \end{flalign*} % -----File: 052.png if $f(x)=x$ from $x=0$ to $x=\dfrac{\pi}{2}$ and $f(x)=0$ from $x=\dfrac{\pi}{2}$ to $x=\pi$. \begin{flalign*} &\indent\text{(4)}\quad \sin\mu x=\dfrac{2}{\pi}\sin\mu\pi \bigg[ \dfrac{ \sin x}{1^2-\mu^2} - \dfrac{2\sin2x}{2^2-\mu^2} + \dfrac{3\sin3x}{3^2-\mu^2} - \dfrac{4\sin4x}{4^2-\mu^2} + \cdots \bigg]\\ &\text{if $\mu$ is a fraction.}\\ % &\indent\text{(5)}\quad e^x = \dfrac{2}{\pi} \bigg[ \dfrac{1}{ 2}(1+e^\pi)\sin x + \dfrac{2}{5}(1-e^\pi)\sin 2x + \dfrac{3}{10}(1+e^\pi)\sin 3x \\ &\hspace{8em} + \dfrac{4}{17}(1-e^\pi)\sin 4x + \cdots \bigg].\\ % &\indent\text{(6)}\quad \sinh x = \dfrac{2\sinh\pi}{\pi} \bigg[ \dfrac{1}{ 2}\sin x - \dfrac{2}{5}\sin 2x + \dfrac{3}{10}\sin3x - \dfrac{4}{17}\sin4x + \cdots \bigg].\\ % &\indent\text{(7)}\quad \cosh x = \dfrac{2}{\pi} \bigg[ \dfrac{1}{2}(1+\cosh\pi)\sin x + \dfrac{2}{5}(1-\cosh\pi)\sin 2x \\ &\hspace{8em} + \dfrac{3}{10}(1+\cosh\pi)\sin 3x + \cdots \bigg]. \end{flalign*} \mypara{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 $x=\pi$, that it does not become infinite between $x=0$ and $x=\pi$, and that if discontinuous it has only finite discontinuities. Assume \[ f(x) = b_0 + b_1\cos x + b_2\cos 2x + b_3\cos 3x + \cdots \tag{1} \] To determine any coefficient $b_m$ multiply (1) by $\cos mx.dx$ and integrate each term from $0$ to $\pi$. \begin{align*} \int\limits_0^\pi &b_0\cos mx.dx = 0. \\ \int\limits_0^\pi b_k\cos kx\cos mx.dx &= \frac{b_k}{2}\int\limits_0^\pi[\cos(m-k)x+\cos(m+k)x]dx\\ &= 0 \text{ if $m$ and $k$ are not equal.} \end{align*} \[ \int b_m\cos^2 mx.dx = \frac{b_m}{2m}(mx+\cos mx\sin mx), \] \[ \phantom{if\ m\ is\ not\ zero.} \int\limits_0^\pi b_m\cos^2 mx.dx = \frac{\pi}{2}b_m, \tag*{if $m$ is not zero.} \] \begin{flalign*} &\text{Hence} & b_m = \frac{2}{\pi} \int\limits_0^\pi f(x)\cos mx.dx = \frac{2}{\pi} \int\limits_0^\pi f(\alpha)\cos m\alpha.d\alpha, & \tag{2}&& \end{flalign*} if $m$ is not zero. % -----File: 053.png To get $b_0$ multiply (1) by $dx$ and integrate from zero to $\pi$. \begin{flalign*} &&&\int\limits_0^\pi b_0 dx = b_0 \pi,&&\\ &&&\int\limits_0^\pi b_k\cos kx.dx=0.&&\\ \label{err053} &\text{\indent Hence} & b_0=\frac{1}{\pi} &\int\limits_0^\pi f(x)dx=\frac{1}{\pi}\int\limits_0^\pi f(\alpha)d\alpha, & \tag{3} &\phantom{Hence} \end{flalign*} 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 \[ f(x) = \tfrac{1}{2} b_0 + b_1\cos x + b_2\cos 2x + b_3\cos 3x + \cdots \tag{4} \] and then the formula \[ b_m = \frac{2}{\pi} \int\limits_0^\pi f(x)\cos mx.dx = \frac{2}{\pi}\int\limits_0^\pi f(\alpha)\cos m\alpha.d\alpha \tag{2} \] will give $b_0$ 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 \[ y = \tfrac{1}{2} b_0 + b_1\cos x + b_2\cos 2x + \cdots + b_n\cos nx \tag{5} \] 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 $b_0$, $b_1$, $b_2$, $\cdots$ $b_n$ by elimination from the $n+1$ resulting equations, and then taking the limiting values they approach as $n$ is indefinitely increased.\quad (v.\ Art.~24.) If $\Delta x = \dfrac{\pi}{n+1}$ the abscissas of the $n+1$ points used are 0, $\Delta x$, $2\Delta x$, $3\Delta x$, $\cdots$ $n\Delta x$, so that we should expect our cosine development to hold for $x=0$ as well as for values of $x$ between zero and $\pi$. \mypara{28.} Let us take one or two examples: \begin{flalign*} &\text{\indent($a$)\quad Let} & f(x)=x. & \tag{1} &&\phantom{\indent\quad Let} \end{flalign*} \begin{gather*} b_0 = \frac{2}{\pi} \int\limits_0^\pi x\,dx = \frac{2}{\pi} \frac{\pi^2}{2} = \pi. \\ b_m = \frac{2}{\pi}\int\limits_0^\pi x\cos mx.dx = \frac{2}{m^2\pi} (\cos m\pi-1) = \frac{2}{m^2\pi} [(-1)^m-1]. \end{gather*} % -----File: 054.png \begin{flalign*} &\text{\indent Hence} & x &= \frac{\pi}{2} - \frac{4}{\pi} \left( \cos x + \frac{\cos 3x}{3^2} + \frac{\cos 5x}{5^2} + \frac{\cos 7x}{7^2} + \cdots \right). &&\phantom{Hence} \tag{2} \end{flalign*} (2) holds good not only for values of $x$ between zero and $\pi$ but for $x = 0$ and $x = \pi$ as well, since for these values we have \begin{flalign*} && 0 &= \frac{\pi}{ 2} - \frac{4}{\pi} \left( 1 + \frac{1}{3^2} + \frac{1}{5^2} + \frac{ 1}{7^2} + \cdots \right) &&\tag{3}\\ &\text{and }& \pi &= \frac{\pi}{2} + \frac{4}{\pi} \left(1 + \frac{1}{3^2} + \frac{1}{5^2} + \frac{ 1}{7^2} + \cdots \right) && \tag{4} \end{flalign*} which are true by Art.~26 (\emph{c})(3). \begin{flalign*} &\text{\indent($b$)\quad Let}& f(x) &= x\sin x. &&\phantom{\indent Let} \tag{1} \end{flalign*} \begin{gather*} b_0 = \frac{2}{\pi}\int\limits_{0}^{\pi} x\sin x.dx = \frac{2}{\pi}\pi = 2,\\ b_1 = \frac{2}{\pi}\int\limits_{0}^{\pi} x\sin x \cos x.dx = \frac{1}{\pi}\int\limits_{0}^{\pi} x\sin 2x.dx = -\frac{1}{2},\\ b_m = \frac{2}{\pi}\int\limits_{0}^{\pi} x\sin x \cos mx.dx = \frac{1}{\pi}\int\limits_{0}^{\pi} [x\sin (m+1)x - x \sin (m-1)x]dx\\ \begin{split} &= \frac{2}{(m-1)(m+1)} \quad \text{if $m$ is odd}\\ &=-\frac{2}{(m-1)(m+1)} \quad \text{if $m$ is even.}\\ \end{split} \end{gather*} Hence \[ x \sin x = 1 - \frac{\cos x}{2} - \frac{2\cos 2x}{1.3} + \frac{2\cos 3x}{2.4} - \frac{2\cos 4x}{3.5} + \cdots \tag{2} \] If $x = \dfrac{\pi}{2}$ we have \[ \frac{\pi}{4} = \frac{1}{ 2} + \frac{1}{1.3} - \frac{1}{3.5} + \frac{1}{5.7} - \cdots. \tag{3} \] \newpage \EXAMPLE{S} Obtain the following developments: \begin{flalign*} &\text{\indent(1)\quad} f(x) = \frac{\pi}{ 4} - \frac{2}{\pi} \bigg[ \frac{\cos 2x}{1^2} + \frac{\cos 6x}{3^2} + \frac{\cos 10x}{5^2} + \frac{\cos 14x}{7^2} + \cdots \bigg] &\\ &\text{if \quad $f(x) = x$ from $x = 0$ to $x = \dfrac{\pi}{ 2 }$ and $f(x) = \pi - x$ from $x = \dfrac{\pi}{ 2}$ to $x = \pi$.}\\ % -----File: 055.png % &\text{\indent(2)\quad} f(x) = \dfrac{1}{2} + \dfrac{2}{\pi}\bigg[\dfrac{\cos x}{1} - \dfrac{\cos 3x}{3} + \dfrac{\cos 5x}{5} - \dfrac{\cos 7x}{7} +\cdots\bigg], \\ &\text{if \quad $f(x) = 1$ from $x = 0$ to $x = \dfrac{\pi}{ 2}$ and $f(x) = 0$ from $x = \dfrac{\pi }{2}$ to $x = \pi$.}\\ % &\text{\indent(3)\quad} x^2 = \dfrac{\pi^2}{ 3} - 4 \bigg[\dfrac{\cos x}{1^2} - \dfrac{\cos 2x}{2^2} + \dfrac{\cos 3x}{3^2} - \dfrac{\cos 4x}{4^2} +\cdots\bigg], \\ % &\text{\indent(4)\quad} x^3 = \dfrac{\pi^3 }{4} - \dfrac{6}{\pi}\bigg[\bigg(\dfrac{\pi^2}{1^2} - \dfrac{4}{1^4}\bigg)\cos x - \dfrac{\pi^2}{2^2}\cos 2x + \bigg(\dfrac{\pi^2}{3^2} - \dfrac{4}{3^4}\bigg)\cos 3x \\ &\hspace{10em} - \dfrac{\pi^2}{4^2} \cos 4x + \bigg(\dfrac{\pi^2}{5^2} - \dfrac{4}{5^4}\bigg)\cos 5x - \cdots\bigg], \\ % &\text{\indent(5)\quad} f(x) = \dfrac{\pi }{ 8} + \dfrac{2}{\pi}\bigg[\bigg( \dfrac{\pi}{2} - 1\bigg)\cos x - \dfrac{2}{2^2} \cos 2x - \dfrac{1}{3^2} \bigg( \dfrac{3\pi}{2}+1\bigg) \cos 3x \\ &\hspace{11em} +\dfrac{1}{5^2}\bigg(\dfrac{5\pi}{2} -1\bigg)\cos 5x -\dfrac{2 }{6^2} \cos 6x - \cdots\bigg], \\ &\text{if \quad $f(x) = x$ from $x = 0$ to $x = \dfrac{\pi }{2}$ and $f(x) = 0$ from $x = \dfrac{\pi}{ 2}$ to $x = \pi$.}\\ % &\text{\indent(6)\quad} e^x = \dfrac{2}{\pi}\bigg[\dfrac{1}{2}(e^\pi-1)- \dfrac{1}{1+1^2} (e^\pi+1)\cos x + \dfrac{1}{1+2^2} (e^\pi-1)\cos 2x \\ &\hspace{7em} - \dfrac{1}{1+3^2} (e^\pi+1)\cos 3x + \cdots\bigg], \\ % &\text{\indent(7)\quad} \cosh x = \dfrac{2 \sinh \pi}{\pi} \bigg[\dfrac{1}{2} - \dfrac{1}{2}\cos x + \dfrac{1}{5}\cos 2x - \dfrac{1}{10}\cos 3x \\ &\hspace{13em} + \dfrac{1}{17}\cos 4x - \cdots\bigg], \\ % &\text{\indent(8)\quad} \sinh x =\dfrac{ 2}{\pi}\bigg[\dfrac{1}{2} (\cosh\pi-1) - \dfrac{1}{2}(\cosh\pi +1)\cos x \\ &\hspace{8em} + \dfrac{1}{5} (\cosh \pi-1)\cos 2x - \dfrac{1}{10}(\cosh\pi +1) \cos 3x + \cdots \bigg], \\ % &\text{\indent(9)\quad} \cos \mu x = \dfrac{2\mu \sin \mu \pi}{\pi} \bigg[ \dfrac{1}{2\mu^2} - \dfrac{\cos x}{ \mu^2-1^2} + \dfrac{\cos 2x}{\mu^2-2^2} - \dfrac{\cos 3x}{\mu^2-3^2} \\ % &\hspace{12.5em} + \dfrac{\cos 4x}{\mu^2-4^2} - \cdots \bigg], \\ &\indent\phantom{(9)} \quad \text{if $\mu$ is a fraction.} \end{flalign*} \mypara{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 $\pi$, there is a decided difference in the two series for other values of $x$. Both series are periodic functions of $x$ having the period $2\pi$. If then we let $y$ equal the series in question and construct the portion of the corresponding % -----File: 056.png curve which lies between the values $x = -\pi$ and $x = \pi$ the whole curve will consist of repetitions of this portion. Since $\sin mx = -\sin(- mx)$ the ordinate corresponding to any value of $x$ between $-\pi$ 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 \[ y = a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + \cdots \tag{1} \] is symmetrical with respect to the origin. Since $\cos mx = \cos (-mx)$ the ordinate corresponding to any value of $x$ between $-\pi$ 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 \[ y = \tfrac{1}{2}b_0 +b_1\cos x + b_2\cos 2x + b_3\cos 3x + \cdots \tag{2} \] is symmetrical with respect to the axis of $Y$. If then $f(x) = -f(-x)$, that is if $f(x)$ is an \textit{odd} function the sine series corresponding to it will be equal to it for all values of $x$ between $-\pi$ and $\pi$, 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 \textit{even} function the cosine series corresponding to it will be equal to it for all values of $x$ between $x = -\pi$ and $x = \pi$, 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 $x$. \begin{align*} y &= 2\left[\sin x - \frac{\sin 2x}{2} + \frac{\sin 3x}{3} - \frac{\sin 4x}{4} + \cdots \right] \tag{3} \\ y &= \frac{\pi}{2} - \frac{4}{\pi} \left[ \cos x + \frac{\cos 3x}{3^2} + \frac{\cos 5x}{5^2} + \frac{\cos 7x}{7^2} + \cdots \right] \tag{4} \end{align*} [v.\ Art.~26(\emph{a}) and Art.~28(\emph{a})]. (3) represents the curve %[Illustration] \pngcent{003.png}{1266}%{346} \newpage\noindent and (4) the curve %[Illustration] \pngcent{004.png}{1266}%{236} % -----File: 057.png Both coincide with $y = x$ from $x = 0$ to $x = \pi$, (3) coincides with $y = x$ from $x = -\pi$ to $x = \pi$, and neither coincides with $y = x$ for values of $x$ less than $-\pi$ or greater than $\pi$. Moreover (3), in addition to the continuous portions of the locus represented in the figure, gives the isolated points $(-\pi,0)$ $(\pi,0)$ $(3\pi,0)$ \&c. \mypara{30.} We have seen that if $f(x)$ is an \textit{odd} function its development in sine series holds for all values of $x$ from $-\pi$ to $\pi$, as does the development of $f(x)$ in cosine series if $f(x)$ is an \textit{even} function. Thus the developments of Art.~26(\emph{a}), Art.~26 Exs.~(2), (4), (6); Art.~28(\emph{b}), Art.~28 Exs.~(3), (7), (9) are valid for all values of $x$ between $-\pi$ and $\pi$. Any function of $x$ can be developed into a Trigonometric series to which it is equal for all values of $x$ between $-\pi$ and $\pi$. 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. \[ f(x)=\frac{f(x)+f(-x)}{2} + \frac{f(x)-f(-x)}{2} \tag{1} \] identically; but $\dfrac{f(x)+f(-x)}{2}$ is not changed by reversing the sign of $x$ and is therefore an \emph{even} function of $x$; and when we reverse the sign of $x$, $\dfrac{f(x)-f(-x)}{2}$ is affected only to the extent of having its sign reversed and is consequently an \textit{odd} function of $x$. Therefore for all values of $x$ between $-\pi$ and $\pi$\\[-1ex] \[ \frac{f(x)+f(-x)}{2} = \frac{1}{2}b_0 + b_1\cos x + b_2\cos 2x + b_3\cos3x + \cdots \\[1ex] \] where \hfill$\displaystyle b_m = \frac{2}{\pi} \int\limits_{0}^{\pi} \frac{f(x) + f(-x)}{2} \cos mx.dx; $ \hfill\quad and \\[1ex] \[ \frac{f(x)-f(-x)}{2} = a_1\sin x + a_2\sin 2x + a_3\sin 3x + \cdots \\[1ex] \] where \hfill$\displaystyle a_m = \frac{2}{\pi} \int\limits_{0}^{\pi} \frac{f(x) - f(-x)}{2} \sin mx.dx. $ \hfill\phantom{where}\\[2ex] \newpage\noindent $b_m$ and $a_m$ can be simplified a little. \\[-1ex] \begin{align*} b_m &= \frac{2}{\pi} \int\limits_{0}^{\pi} \frac{f(x)+f(-x)}{2} \cos mx.dx \\[1ex] &= \frac{1}{\pi} \bigg[ \int\limits_{0}^{\pi} f( x)\cos mx.dx + \int\limits_{0}^{\pi} f(-x)\cos mx.dx \bigg], \end{align*} % -----File: 058.png but if we replace $x$ by $-x$, we get \[ \int\limits_{0}^{ \pi} f(-x) \cos mx.dx = -\int\limits_{0}^{-\pi} f( x) \cos mx.dx = \int\limits_{-\pi}^{0} f( x) \cos mx.dx,\\[-4ex] \] \begin{flalign*} &\text{and we have }&& b_m = \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(x) \cos mx.dx. &&\phantom{and\ we\ have } \end{flalign*} In the same way we can reduce the value of $a_m$ to \[ \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(x) \sin mx.dx. \] Hence \[ \left\{ \begin{aligned} f(x) = \frac{1}{2} b_0 &+ b_1 \cos x + b_2 \cos 2x + b_3 \cos 3x + \cdots\\ &+ a_1 \sin x + a_2 \sin 2x + a_3 \sin 3x + \cdots \end{aligned} \right \}\tag{2}\\[-3ex] \] \begin{flalign*} &\text{where }& b_m &= \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(x) \cos mx.dx = \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(\alpha) \cos m\alpha.d\alpha. &&\phantom{where} \tag{3} \\ &\text{and }& a_m &= \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(x) \sin mx.dx = \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f(\alpha) \sin m\alpha.d\alpha. && \tag{4} \end{flalign*} and this development holds for all values of $x$ between $-\pi$ and $ \pi$. The second member of (2) is known as a Fourier's Series. \EXAMPLE{S} 1.\quad Obtain the following developments, all of which are valid from $x = -\pi$ to $x=\pi$:--- \begin{flalign*} &\indent\text{(1)}& e^x ={}& \frac{2 \sinh \pi}{\pi} \left[ \frac{1}{2} - \frac{1}{2} \cos x + \frac{1}{5} \cos 2x - \frac{1}{10} \cos 3x + \frac{1}{17} \cos 4x + \cdots \right]&& \\ &&&+ \frac{2 \sinh \pi}{ \pi} \left[ \frac{1}{2}\sin x - \frac{2}{5} \sin 2x + \frac{3}{10} \sin 3x - \frac{4}{17} \sin 4x + \cdots \right]. \end{flalign*} \begin{flalign*} &\indent\text{(2)}& f(x) ={}& \frac{\pi}{ 4} - \frac{2}{ \pi} \left[ \cos x + \frac{\cos 3x }{ 3^2} + \frac{\cos 5x}{ 5^2} + \frac{\cos 7x }{ 7^2} + \cdots \right]&&\hspace{5.5em} \\ &&&+ \frac{\sin x}{ 1} - \frac{\sin 2x}{ 2} + \frac{\sin 3x}{ 3} - \frac{\sin 4x}{ 4} + \cdots, \end{flalign*} where $f(x) = 0$ from $x = -\pi$ to $x=0$ and $f(x) = x$ from $x=0$ to $x=\pi$. % -----File: 059.png \begin{flalign*} &\indent\text{(3)} &f(x)={}&- \frac{3\pi}{16} + \frac{1}{\pi}\bigg[\frac{1}{1^2}\cos x + \frac{2}{2^2}\cos2x + \frac{1}{3^2}\cos3x + \frac{1}{5^2}\cos5x\\ &&&\hspace{5em} + \frac{2}{6^2}\cos 6x + \cdots\bigg]&&\hspace{2em}\\ &&&+ \frac{1}{\pi}\bigg[\biggl(\frac{3\pi}{2} - 1\biggr)\sin x - \frac{3\pi}{4}\sin2x + \biggl(\frac{3\pi}{6} + \frac{1}{3^2}\biggr)\sin3x\\ &&&\hspace{3em} - \frac{3\pi}{8}\sin4x + \biggl(\frac{3\pi}{10} - \frac{1}{5^2}\biggr)\sin5x - \cdots\bigg],\\[-6ex] \end{flalign*} \begin{flalign*} &\text{where } &f(x)&=x \text{ from } x=-\pi \text{ to } x=0, \quad f(x)=0 \text{ from } x=0 \text{ to } x=\frac\pi2, &&\\ &\text{and } &f(x)&=x - \frac{\pi}{2} \text{ from } x=\frac{\pi}{2} \text{ to } x=\pi. \end{flalign*} 2.\quad Show that formula~(2) Art.~30 can be written \[ f(x)=\frac{1}{2} c_0\cos \beta_0 + c_1\cos(x - \beta_1) + c_2\cos(2x - \beta_2) + c_3\cos(3x - \beta_3) + \cdots \] \vspace{-4ex} \begin{flalign*} &\text{where} &c_m=(a_m^2 + b_m^2)^\frac{1}{2} \text{ and } \beta_m=\tan^{-1} \frac{a_m}{b_m}.&&\phantom{where} \end{flalign*} 3.\quad Show that formula~(2) Art.~30 can be written \[ f(x)=\frac{1}{2} c_0\sin \beta_0 + c_1\sin(x + \beta_1) + c_2\sin (2x + \beta_2) + c_3\sin (3x + \beta_3) + \cdots\\[-4ex] \] \begin{flalign*} &\text{where} &c_m=(a_m^2 + b_m^2)^\frac{1}{2} \text{ and } \beta_m=\tan^{-1} \frac{b_m}{a_m}.&&\phantom{where} \end{flalign*} \mypara{31.} In developing a function of $x$ into a Trigonometric series it is often inconvenient to be held within the narrow boundaries $x=-\pi$ and $x=\pi$. 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 \[ z=\frac{\pi}{c} x, \] which is equal to $-\pi$ when $x=-c$ and to $\pi$ when $x=c$. $f(x)=f\left(\dfrac{c}{\pi}z\right)$ can be developed in terms of $z$ by Art.~30 (2), (3), and (4). We have \begin{equation} \left. \begin{aligned} f\left(\frac{c}{\pi}z\right)=\frac{1}{2} b_0 &+ b_1\cos z + b_2\cos 2z + b_3\cos 3z + \cdots\\ &+ a_1\sin z + a_2\sin 2z + a_3\sin 3z + \cdots \end{aligned}\right\}\tag{1} \end{equation} \vspace{-3ex} \begin{flalign*} &\text{where} &b_m&=\frac{1}{\pi} \int\limits_{-\pi}^{\pi} f\left(\frac{c}{\pi} z\right)\cos mz.dz.&\phantom{where} \tag{2}\\ % -----File: 060.png &\text{and}& a_m &= \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f\left(\frac{c}{\pi} z\right) \sin mz.dz. & \tag{3} \end{flalign*} and (1) holds good from $z = -\pi$ to $z = \pi$. Replace $z$ by its value in terms of $x$ and (1) becomes \[ \left . \begin{aligned} f(x)=\frac{1}{2}b_0 &+ b_1 \cos\frac{\pi x}{c} + b_2 \cos \frac{2\pi x}{c} + b_3 \cos \frac{3\pi x}{c} + \cdots \\ &+a _1 \sin \frac{\pi x}{c} + a_2 \sin \frac{2\pi x}{c} + a_3 \sin \frac{3\pi x}{c} + \cdots \end{aligned} \right \} \tag{4} \] 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. \begin{flalign*} && b_m &= \frac{1}{\pi} \int\limits_{-\pi}^{\pi} f\left(\frac{c}{\pi} z\right) \cos mz.dz = \frac{1}{\pi} \int\limits_{-c}^{c} f(x) \cos \frac{m\pi x}{c} \frac{\pi}{c} dx&& \\[1ex] &\text{or}& b_m &= \frac{1}{c} \int\limits_{-c}^{c} f(x) \cos \frac{m\pi x}{c} dx = \frac{1}{c} \int\limits_{-c}^{c} f(\lambda) \cos\frac{m\pi\lambda}{c} d\lambda. && \tag{5} \\ \intertext{\indent In like manner we can transform (3) into} && a_m &= \frac{1}{c} \int\limits_{-c}^{c} f(x) \sin \frac{m\pi x}{c} dx = \frac{1}{c} \int\limits_{-c}^{c} f(\lambda) \sin \frac{m\pi \lambda}{c} d\lambda. && \tag{6} \end{flalign*} By treating in like fashion formulas (1) and (2) Art.~25 and formulas (4) and (2) Art.~27 we get \[ f(x) = a_1 \sin \frac{\pi x}{c} + a_2 \sin \frac{2\pi x}{c} + a_3 \sin \frac{3\pi x}{c} + \cdots \tag{7} \] \begin{flalign*} &\text{where }& a_m &= \frac{2}{c} \int\limits_{0}^{c} f(x) \sin \frac{m\pi x}{c} dx = \frac{2}{c} \int\limits_{0}^{c} f(\lambda) \sin \frac{m\pi \lambda}{c} d\lambda. && \tag{8} \\[1ex] &\text{and}& f(x) &= \frac{1}{2 } b_0 + b_1 \cos \frac{\pi x}{c} + b_2 \cos \frac{2\pi x}{c} + b_3 \cos \frac{3\pi x}{c} + \cdots && \tag{9} \\[1ex] &\text{where}& b_m &= \frac{2}{c} \int\limits_{0}^{c} f(x) \cos \frac{m\pi x}{c} dx = \frac{2}{c} \int\limits_{0}^{c} f(\lambda) \cos \frac{m\pi \lambda}{c} d\lambda. && \tag{10} \end{flalign*} and (7) and (9) hold good from $x = 0$ to $x = c$. % -----File: 061.png \EXAMPLE{S} 1.\quad Obtain the following developments: \begin{flalign*} &\rlap{\indent(1)\quad} & 1={}&\dfrac{4}{\pi}\bigg[\sin\dfrac{\pi x}{c}+\dfrac{1}{3}\sin\dfrac{3\pi x}{c}+\dfrac{1}{5}\sin\dfrac{5\pi x}{c}+\cdots\bigg]\\ && &\text{from } x=0 \text{ to } x=c.\\ &\rlap{\indent(2)\quad} & x={}&\dfrac{2c}{\pi}\bigg[\sin\dfrac{\pi x}{c}-\dfrac{1}{2}\sin\dfrac{2\pi x}{c}+\dfrac{1}{3}\sin\dfrac{3\pi x}{c}-\dfrac{1}{4}\sin\dfrac{4\pi x}{c}+\cdots\bigg]\\ && &\text{from } x=-c \text{ to } x=c.\\ && x={}&\dfrac{c}{2}-\dfrac{4c}{\pi^2}\bigg[\cos\dfrac{\pi x}{c}+\dfrac{1}{3^2}\cos\dfrac{3\pi x}{c}+\dfrac{1}{5^2}\cos\dfrac{5\pi x}{c}+\dfrac{1}{7^2}\cos\dfrac{7\pi x}{c}+\cdots\bigg]\\ && &\text{from } x=-c \text{ to } x=c.\\ &\rlap{\indent(3)\quad} & x^2={}&\dfrac{2c^2}{\pi^3}\bigg[\bigg(\dfrac{\pi^2}{1}-\dfrac{4}{1^3}\bigg)\sin\dfrac{\pi x}{c}-\dfrac{\pi^2}{2}\sin\dfrac{2\pi x}{c}+\bigg(\dfrac{\pi^2}{3}-\dfrac{4}{3^2}\bigg)\sin\dfrac{3\pi x}{c}\\ && &\hspace{3em}-\dfrac{\pi^2}{4}\sin\dfrac{4\pi x}{c}+\bigg(\dfrac{\pi^2}{5}-\dfrac{4}{5^3}\bigg)\sin\dfrac{5\pi x}{c}+\cdots\bigg]\\ && &\text{from } x=0 \text{ to } x=c.\\ && x^2={}&\dfrac{c^2}{3}-\dfrac{4c^2}{\pi^2}\bigg[\!\cos\dfrac{\pi x}{c}-\dfrac{1}{2^2}\cos\dfrac{2\pi x}{c}+\dfrac{1}{3^2}\cos\dfrac{3\pi x}{c}-\dfrac{1}{4^2}\cos\dfrac{4\pi x}{c}+\cdots\!\bigg]\:\\ && &\text{from } x=-c \text{ to } x=c.\\ &\rlap{\indent(4)\quad} & e^x={}&2\pi\bigg[\dfrac{1+e^c}{c^2+\pi^2}\sin\dfrac{\pi x}{c}+\dfrac{2(1-e^c)}{c^2+4\pi^2}\sin\dfrac{2\pi x}{c}\\ && &\hspace{3em}+\dfrac{3(1+e^c)}{c^2+9\pi^2}\sin\dfrac{3\pi x}{c}+\dfrac{4(1-e^c)}{c^2+16\pi^2}\sin\dfrac{4\pi x}{c}+\cdots\bigg],\\ && e^x={}&2c\bigg[\dfrac{1}{2}\dfrac{e^c-1}{c^2}-\dfrac{e^c+1}{c^2+\pi^2}\cos\dfrac{\pi x}{c}+\dfrac{e^c-1}{c^2+4\pi^2}\cos\dfrac{2\pi x}{c}\\ && &\hspace{3em}-\dfrac{e^c+1}{c^2+9\pi^2}\cos\dfrac{3\pi x}{c}+\cdots\bigg]\\ && &\text{from } x=0 \text{ to } x=c. \end{flalign*} \begin{flalign*} &\rlap{\indent(5)\quad} & f(x)={}&\dfrac{4c}{\pi^2}\bigg[\sin\dfrac{\pi x}{c}-\dfrac{1}{3^2}\sin\dfrac{3\pi x}{c}+\dfrac{1}{5^2}\sin\dfrac{5\pi x}{c}+\cdots\bigg]\hspace{8em}\\ && &\text{from } x=0 \text{ to } x=c, \end{flalign*} where $f(x)=x$ from $x=0$ to $x=\dfrac{c}{2}$ and $f(x)=c-x$ from $x=\dfrac{c}{2}$ to $x=c$.\\ % -----File: 062.png 2.\quad Show that formula (4) Art.~31 can be written \begin{multline*} f(x) = \frac{1}{2} c_0 \cos\beta_0 + c_1 \cos \bigg( \frac{ \pi x}{c} - \beta_1 \bigg) + c_2 \cos \bigg( \frac{2\pi x}{c} - \beta_2 \bigg) \\ + c_3 \cos \bigg( \frac{3\pi x}{c} - \beta_3 \bigg) + \cdots \end{multline*} \vspace{-3ex} \begin{flalign*} &\text{where}&& c_m = (a_m^2 + b_m^2)^{\frac{1}{2}} \quad \text{and} \quad \beta_m = \tan^{-1}\frac{a_m}{b_m}.&&\phantom{where} \end{flalign*} 3.\quad Show that formula (4) Art.~31 can be written \begin{multline*} f(x) = \frac{1}{2} c_0 \sin\beta_0 + c_1 \sin \bigg( \frac{ \pi x}{c} + \beta_1 \bigg) + c_2 \sin \bigg( \frac{2\pi x}{c} + \beta_2 \bigg) \\ + c_3 \sin \bigg( \frac{3\pi x}{c} + \beta_3 \bigg) + \cdots \end{multline*} \vspace{-3ex} \begin{flalign*} &\text{where}&& c_m = (a_m^2 + b_m^2)^{\frac{1}{2}} \quad \text{and} \quad \beta_m = \tan^{-1}\frac{a_m}{b_m}.&&\phantom{where} \end{flalign*} \mypara{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 $b_0$, $b_1$, $b_2$, $\cdots$ $a_1$, $a_2$, $\cdots$ by their values given in Art.~31 (5) and (6). \begin{align*} f(x) &= \frac{1}{c} \biggl[ \begin{aligned}[t] & \frac{1}{2} \int\limits_{-c}^{c} f(\lambda) \,d\lambda \\ &+ \int\limits_{-c}^{c} f(\lambda) \cos\frac{ \pi\lambda}{c} \cos\frac{ \pi x}{c} d\lambda + \int\limits_{-c}^{c} f(\lambda) \cos\frac{2\pi\lambda}{c} \cos\frac{2\pi x}{c} d\lambda + \cdots \\ &+ \int\limits_{-c}^{c} f(\lambda) \sin\frac{ \pi\lambda}{c} \sin\frac{ \pi x}{c} d\lambda + \int\limits_{-c}^{c} f(\lambda) \sin\frac{2\pi\lambda}{c} \sin\frac{2\pi x}{c} d\lambda + \cdots \biggr] \end{aligned} \\ &= \frac{1}{c} \int\limits_{-c}^{c} f(\lambda) \,d\lambda \biggl[ \begin{aligned}[t] & \frac{1}{2} + \cos\frac{ \pi\lambda}{c} \cos\frac{\pi x}{c} + \sin\frac{ \pi\lambda}{c} \sin\frac{\pi x}{c} \\[1ex] &+ \cos\frac{2\pi\lambda}{c} \cos\frac{2\pi x}{c} + \sin\frac{2\pi\lambda}{c} \sin\frac{2\pi x}{c} + \cdots \biggr ] \end{aligned} \end{align*} % -----File: 063.png \begin{flalign*} f(x) &= \frac{1}{c} \int\limits_{-c}^c f(\lambda) \,d\lambda \biggl[ \frac{1}{2} + \cos\frac{ \pi}{c} (\lambda-x) + \cos\frac{2\pi}{c} (\lambda-x) + \cdots \biggr] \\ &= \frac{1}{2c} \int\limits_{-c}^c f(\lambda) \,d\lambda \biggl[ \begin{aligned}[t] & 1 + \cos\frac{ \pi}{c} (\lambda-x) + \cos\frac{2\pi}{c} (\lambda-x) + \cdots \\ &+ \cos \Bigl( -\frac{ \pi}{c} \Bigr) (\lambda-x) + \cos \Bigl( -\frac{2\pi}{c} \Bigr) (\lambda-x) + \cdots \biggr] \end{aligned} \end{flalign*} since $\cos(-\phi)=\cos\phi$. \begin{flalign*} f(x) = \frac{1}{2\pi} \int\limits_{-c}^c f(\lambda) \,d\lambda \biggl[ \cdots &+ \frac{\pi}{c} \cos \Bigl( -\frac{2\pi}{c} \Bigr) (\lambda-x) + \frac{\pi}{c} \cos \Bigl( -\frac{ \pi}{c} \Bigr) (\lambda-x) \\ &+ \frac{\pi}{c} \cos \frac{0\pi}{c} (\lambda-x) + \frac{\pi}{c} \cos \frac{ \pi}{c} (\lambda-x) \\ &+ \frac{\pi}{c} \cos \frac{2\pi}{c} (\lambda-x) + \cdots \biggr] \tag{1} \end{flalign*} As $c$ is indefinitely increased the limiting value approached by the parenthesis in (1) is \[ \int\limits_{-\infty}^{\infty}\cos\alpha(\lambda-x).d\alpha. \] Hence the limiting form approached by (1) is \[ f(x) = \frac{1}{2\pi} \int\limits_{-\infty}^{\infty} f(\lambda) \,d\lambda \int\limits_{-\infty}^{\infty} \cos\alpha(\lambda-x).d\alpha, \tag{2} \] 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 \emph{Fourier's Integral}, and since it is a limiting form of \emph{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.\footnote{See note on page \pageref{notep38}.} (2) is sometimes given in a slightly different form. Since \quad $\displaystyle \int\limits_{-\infty}^{\infty} \cos\alpha(\lambda-x).d\alpha = \int\limits_{-\infty}^0 \cos\alpha(\lambda-x).d\alpha + \int\limits_0^{\infty} \cos\alpha(\lambda-x).d\alpha$\\ and \begin{gather*} \int\limits_{-\infty}^0 \cos\alpha(\lambda-x).d\alpha = \int\limits_{ \infty}^0 \cos(-\alpha)(\lambda-x).d(-\alpha) =-\int\limits_{ \infty}^0 \cos\alpha(\lambda-x).d\alpha \\ \int\limits_{-\infty}^{\infty} \cos\alpha(\lambda-x).d\alpha =2\int\limits_0^{\infty} \cos\alpha(\lambda-x).d\alpha \end{gather*} % -----File: 064.png and (2) may be written \[ f(x) = \frac{1}{\pi} \int\limits_{-\infty}^{\infty} f(\lambda) \,d\lambda \int\limits_{0}^{\infty} \cos \alpha(\lambda-x).d\alpha. \tag{3} \] If $f(x)$ is an \textit{even} function or an \textit{odd} function (3) can be still further simplified. Let \hfill $ f(x) = -f(-x)$. \hfill\phantom{\indent Let} Since the limits of integration in (3) do not contain $\alpha$ or $\lambda$ the integrations may be performed in whichever order we choose. That is \[ \int\limits_{-\infty}^{\infty} f(\lambda) \,d\lambda \int\limits_{0}^{\infty} \cos \alpha(\lambda-x).d\alpha = \int\limits_{0}^{\infty} d\alpha \int\limits_{-\infty}^{\infty} f(\lambda) \cos\alpha(\lambda-x).d\lambda. \] Now \begin{gather*} \int\limits_{-\infty}^{\infty} \!f(\lambda) \cos\alpha(\lambda-x).d\lambda = \int\limits_{-\infty}^{0} \!f(\lambda) \cos\alpha(\lambda-x).d\lambda + \int\limits_{0}^{\infty} \!f(\lambda) \cos\alpha(\lambda-x).d\lambda. \\ \begin{aligned} \int\limits_{-\infty}^{0} f(\lambda) \cos\alpha(\lambda-x).d\lambda &= \int\limits_{\infty}^{0} f(-\lambda) \cos\alpha(-\lambda-x).d(-\lambda) \\ &=-\int\limits_{0}^{\infty} f(\lambda) \cos\alpha(\lambda+x).d\lambda \end{aligned} \end{gather*} and (3) becomes \begin{align*} f(x) &= \frac{1}{\pi} \int\limits_{0}^{\infty} d\alpha \int\limits_{0}^{\infty} f(\lambda) [\cos\alpha(\lambda-x) - \cos\alpha(\lambda+x).d\lambda \\ &= \frac{2}{\pi} \int\limits_{0}^{\infty} d\alpha \int\limits_{0}^{\infty} f(\lambda) \sin\alpha\lambda \sin\alpha x.d\lambda \end{align*} \begin{flalign*} &\text{or}&& f(x) = \frac{2}{\pi} \int\limits_{0}^{\infty} f(\lambda)d\lambda \int\limits_{0}^{\infty} \sin\alpha\lambda \sin\alpha x.d\alpha. && \tag{4} \end{flalign*} If $f(x) = f(-x)$ (3) can be reduced in like manner to \[ f(x) = \frac{2}{\pi} \int\limits_{0}^{\infty} f(\lambda) \,d\lambda \int\limits_{0}^{\infty} \cos\alpha\lambda \cos\alpha x.d\alpha. \tag{5} \] Although (4) holds for all values of $x$ only in case $f(x)$ is an \textit{odd} function, and (5) only in case $f(x)$ is an \textit{even}, function, both (4) and (5) hold for all \textit{positive} values of $x$ in the case of any function. \EXAMPLE{} (1)\quad Obtain formulas (4) and (5) directly from (7) and (9) Art.~31. \label{ch2end} % -----File: 065.png \mychap{CHAPTER III.}{CONVERGENCE OF FOURIER'S SERIES.} \label{ch3start} \mypara{33.} The question of the \textit{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 \[ \frac{4}{\pi} \biggl[ \sin x + \frac{\sin 3x}{3} + \frac{\sin 5x}{5 } + \frac{\sin 7x}{7} + \cdots\biggr], \tag*{[v.~(3) Art.~26(\emph{b}).]} \] and prove that for all values of $x$ between zero and $\pi$ its sum is absolutely equal to unity; that is, that the limit approached by the sum of $n$ terms of the series \[ \frac{2}{\pi} \biggl[ \sin x \int\limits_{0}^{\pi} \sin \alpha.d\alpha + \sin 2x \int\limits_{0}^{\pi} \sin 2\alpha.d\alpha + \sin 3x \int\limits_{0}^{\pi} \sin 3\alpha.d\alpha + \cdots \biggr ], \] as $n$ is indefinitely increased, is $1$, provided that $x$ lies between zero and $\pi$. Let \begin{align*} S_n = \frac{2}{\pi}\biggl[ & \sin x \int\limits_0^{\pi} \sin \alpha.d\alpha + \sin 2x \int\limits_0^{\pi} \sin 2\alpha.d\alpha + \sin 3x \int\limits_0^{\pi} \sin 3\alpha.d\alpha + \cdots \\ &+ \sin nx \int\limits_0^{\pi} \sin n\alpha.d\alpha \biggr ]. \tag{1} \end{align*} Then \begin{align*} S_{n} &= \frac{2}{\pi} \int\limits_0^{\pi} [ \sin\alpha\sin x + \sin 2\alpha\sin 2x + \sin 3\alpha\sin 3x + \cdots + \sin n\alpha\sin nx ] d\alpha \\ &= \frac{1}{\pi} \int\limits_0^{\pi} \begin{aligned}[t] [ & \cos (\alpha-x) - \cos (\alpha+x) + \cos 2(\alpha-x) - \cos 2(\alpha+x) + \cdots \\[3ex] &+\cos n(\alpha-x) - \cos n(\alpha+x) ] \,d\alpha \end{aligned} \\ &= \frac{1}{\pi} \int\limits_0^{\pi} [ \cos (\alpha-x) + \cos 2(\alpha-x) + \cos 3(\alpha-x) + \cdots + \cos n(\alpha-x) ] d\alpha \\ &-\frac{1}{\pi} \int\limits_0^{\pi} [ \cos (\alpha+x) + \cos 2(\alpha+x) + \cos 3(\alpha+x) + \cdots + \cos n(\alpha+x) ] d\alpha. \end{align*} % -----File: 066.png Therefore by Art.~20 (1) \begin{gather*} \begin{aligned} S_n ={}& \frac{1}{\pi} \int\limits_{0}^{\pi} \biggl[ -\frac{1}{2} + \frac{1}{2} \frac{ \sin (2n+1)\dfrac{\alpha-x}{2} } { \sin \dfrac{\alpha-x}{2} } \biggr] d\alpha \\ &- \frac{1}{\pi} \int\limits_{0}^{\pi} \biggl[ -\frac{1}{2} + \frac{1}{2} \frac{ \sin (2n+1)\dfrac{\alpha+x}{2} } { \sin \dfrac{\alpha+x}{2} } \biggr] d\alpha. \end{aligned} \\ S_n = \frac{1}{2\pi} \int\limits_{0}^{\pi} \frac{ \sin (2n+1)\dfrac{\alpha-x}{2} } { \sin \dfrac{\alpha-x}{2} } d\alpha - \frac{1}{2\pi} \int\limits_{0}^{\pi} \frac{ \sin (2n+1)\dfrac{\alpha+x}{2} } { \sin \dfrac{\alpha+x}{2} } d\alpha. \end{gather*} In the first integral substitute $\beta$ for $\dfrac{\alpha-x}{2}$, and in the second integral substitute $\beta$ for $\dfrac{\alpha+x}{2}$. We get \[ S_n = \frac{1}{\pi} \int\limits_{-\frac{x}{2}}^{\frac{\pi}{2}-\frac{x}{2}} \frac{\sin(2n+1)\beta }{ \sin\beta } \,d\beta - \frac{1}{\pi} \int\limits_{ \frac{x}{2}}^{\frac{\pi}{2}+\frac{x}{2}} \frac{\sin(2n+1)\beta }{ \sin\beta } \,d\beta . \tag{2} \] It remains to find the limit approached by $S_{n}$ as $n$ is indefinitely increased. \markright{CONVERGENCE OF FOURIER'S SERIES.} \mypara{34.}\vspace{-2ex} \[ \int\limits_0^{\frac{\pi}{2}} \frac{ \sin(2n+1)\beta}{ \sin\beta} d\beta = \frac{\pi}{2}. \tag{1} \] For \begin{flalign*} &\phantom{fr}& \frac{\sin(2n+1)\beta }{ 2 \sin\beta} = \tfrac{1}{2 }+ \cos 2\beta + \cos 4\beta &+ \cdots + \cos 2n\beta, && \text{by Art.~20.} \\ &\text{and }& \int\limits_0^{\frac{\pi}{2}} \cos 2k\beta.d\beta &= 0.&& \end{flalign*} Let us construct the curve \[ y = \frac{\sin(2n+1)x}{\sin x}. \] 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. % -----File: 067.png %[Illustration] \pngrightl{005.png}{476}{28}{-12} Each arch has for its altitude unity and for its base $\dfrac{\pi}{2n+1}$ and is symmetrical with respect to the ordinate of its highest or lowest point. If now we form the curve $y = \dfrac{\sin(2n+1)x }{\sin x}$ from the curve $y= \sin(2n + 1)x$, it is clear that, since $\sin x$ increases as $x$ increases from $0$ to $\dfrac{\pi}{2}$, the ordinate of 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 $y=\dfrac{\sin(2n + 1)x}{\sin x}$ will be less than that of the arch before it. If $a_{0}$, $a_{1}$, $a_{2}$, $\cdots a_{n-1}$ are the areas of the successive arches and $a_n$ that of the incomplete arch terminated by the ordinate corresponding to $x =\dfrac{\pi}{2}$ \[ \int\limits_0^{\frac{\pi}{2}} \frac{\sin(2n+1)x}{\sin x} \,dx = a_0-a_1+a_2-a_3+\cdots . \] But \[ \int\limits_0^{\frac{\pi}{2}} \frac{\sin(2n+1)x }{\sin x } \,dx = \int\limits_0^{\frac{\pi}{2}} \frac{\sin(2n+1)\beta}{\sin \beta} \,d\beta = \frac{\pi}{2} \tag*{by (1).} \] Hence \[ \frac{\pi}{2}= a_{0} - a_{1} + a_{2} - a_{3} + a_{4} - \cdots + a_{n} \quad \text{if $n$ is even,} \] or \[ \frac{\pi}{2}= a_{0} - a_{1} + a_{2} - a_{3} + a_{4} - \cdots - a_{n} \quad \text{if $n$ is odd.} \] These equations can be written \begin{align*} \frac{\pi}{2} ={}& a_{0} + (-a_{1} + a_{2}) + (-a_{3} + a_{4})\\ &+ (-a_{5} + a_{6}) + \cdots +(-a_{n-1} + a_{n}) \end{align*} if $n$ is even, and \begin{align*} \frac{\pi}{2} ={}& a_{0} + (-a_{1} + a_{2}) + (-a_{3} + a_{4})\\ &+ (-a_{5} + a{_6}) + \cdots + (-a_{n-2}+a_{n-1})+(-a_{n}) \end{align*} if $n$ is odd. % -----File: 068.png In either case each parenthesis is a negative quantity since \[ a_0 > a_1 > a_2 > a_3 \cdots > a_n, \] and it follows that $a_0$ is greater than $\dfrac{\pi}{2}$. Again \begin{align*} &\frac{\pi}{2} = a_{0} - a_{1} + (a_2-a_3) + (a_4-a_5) + \cdots + (a_{n-2}-a_{n-1}) + a_n \intertext{if $n$ is even and} &\frac{\pi}{2} = a_{0} - a_{1} + (a_2-a_3) + (a_4-a_5) + \cdots + (a_{n-1}-a_{n}) \end{align*} if $n$ is odd. In either case each parenthesis is positive and it follows that $a_0-a_1$ is is less than $\dfrac{\pi}{2}$. Since \[ a_0 > \frac{\pi}{2} > a_0 - a_1, \] $a_{0}$ and $a_{0} - a_{1}$ differ from $\dfrac{\pi}{2}$ by less than they differ from each other, that is, by less than $a_1$. In like manner we can show that $a_0-a_1$ and $a_0-a_1+a_2$ differ from $\dfrac{\pi}{2}$ by less than $a_2$; and in general that $a_0-a_1+a_2-a_3+\cdots \pm a_k$ differs from $\dfrac{\pi}{2}$\vphantom{\fbox{$\dfrac00$}} %[**F2: Spacer added to keep the dfrac from colliding with the line above.] by less than $a_k$; or even that \[ a_0 - a_1 + a_2 - a_3 + \cdots \pm \frac{a_k}{p} \] differs from $\dfrac{\pi}{2}$ by less than $a_k$ no matter the value of $p$, provided $p$ is greater than unity. \mypara{35.} From what has been proved in the last article it follows that \[ \int\limits_0^b \frac{\sin(2n+1)x}{\sin x} \,dx, \] where $b$ is some value between $\dfrac{\pi}{2n+1}$ and $\dfrac{\pi}{2}$, differs from $\dfrac{\pi}{2}$ by less than the area of the arch in which the ordinate of $y= \dfrac{\sin(2n+1)x}{\sin x}$ corresponding to $x = b$ 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. % -----File: 069.png The area of the arch in question is less than $\dfrac{\pi}{2n+1}$, its base, multiplied by $\dfrac{1}{ \sin\Big( b-\dfrac{\pi}{ 2n+1} \Big)}$, a value greater than the length of its longest ordinate. Therefore \hfill $\displaystyle \int\limits_0^b \frac{\sin(2n+1)x }{ \sin x } \,dx $\hfill\hfill\quad\\ differs from $\dfrac{\pi}{2}$ by less than $\dfrac{\pi}{2n+1} \dfrac{ 1}{\sin\Big( b-\dfrac{\pi}{ 2n+1} \Big)}$ . If now $n$ is indefinitely increased $\dfrac{\pi}{2n+1} \dfrac{ 1}{ \sin\Big( b-\dfrac{\pi}{2n+1} \Big)}$ approaches zero as its limit, and we get the very important result \begin{flalign*} && \limit_{n=\infty} \biggl[\int\limits_0^b \frac{\sin(2n+1)x }{ \sin x } \,dx \biggr] = \frac{\pi}{2} && \phantom{\text{if } 00$. (7) Art.~44 gives us the required solution. It is \[ V=\frac{1}{\pi}\int\limits_0^\infty d\alpha\int\limits_0^\infty e^{-\alpha y}\cos\alpha(\lambda-x).d\lambda; \tag{1} \] but this can be much simplified. We have \begin{flalign*} && V= {}& \frac{1}{\pi}\int\limits_0^\infty d\lambda\int\limits_0^\infty e^{-\alpha y}\cos\alpha(\lambda-x).d\alpha. && \\ &\indent\text{Now}& &\int\limits_0^\infty e^{-ax}\cos mx.dx = \frac{a}{a^2+m^2} &&\phantom{\indent Now} \end{flalign*} if $a>0$. (Int.\ Cal.\ Art.~82, Ex.~8.) \begin{flalign*} &\indent\text{Hence} & \int\limits_0^\infty e^{-\alpha y} \cos\alpha(\lambda-x).d\alpha = \frac{y}{y^2+(\lambda-x)^2}, &&\phantom{\indent Hence} \\ &\text{and} & V = \frac{1}{\pi} \int\limits_0^\infty \frac{y\,d\lambda}{y^2+(\lambda-x)^2} = \frac{1}{\pi} \left( \frac{\pi}{2} + \tan^{-1}\frac{x}{y} \right). && \end{flalign*} \[ \tan \left( \frac{\pi}{2} - \tan^{-1}\frac{x}{y} \right) = \ctn \left( \tan^{-1}\frac{x}{y} \right) = \frac{y}{x}; \] and consequently \[ V=\frac{1}{\pi}\left(\frac{\pi}{2}+\tan^{-1}\frac{x}{y}\right) =1-\frac{1}{\pi}\tan^{-1}\frac{y}{x}. \tag{2} \] Since \hfill $\log z=\log(x+yi)=\dfrac{1}{2}\log(x^2+y^2)+i\tan^{-1}\dfrac{y}{x}$, \hfill {\,} \mbox{[Int.\ Cal.\ Art.~33 (2)],} \[ i - \frac{1}{\pi}\log z = i - \frac{1}{\pi}\log(x+yi) = -\frac{1}{2\pi}\log(x^2+y^2) + i\left( 1 - \frac{1}{\pi}\tan^{-1}\frac{y}{x} \right) \] and $1-\dfrac{1}{\pi}\tan^{-1}\dfrac{y}{x}$ and $-\dfrac{1}{2\pi}\log(x^2+y^2)$ are \emph{conjugate functions}. (v.\ Int.\ Cal.\ Arts.~209 and 210.) Hence \[ V_1=-\frac{1}{2\pi}\log(x^2+y^2) \tag{3} \] is a solution of the equation \[ D_x^2V_1 + D_y^2V_1=0; \tag{4} \] % -----File: 082.png and the curves \begin{flalign*} &&&\frac{1}{\pi}\left(\frac{\pi}{2}+\tan^{-1}\frac{x}{y}\right)=a \tag{5}\\ &\text{and}& &-\frac{1}{2\pi}\log(x^2+y^2)=b &&\phantom{and} \tag{6} \end{flalign*} 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 \emph{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 \emph{lines of flow}. Moreover since \[ V_1=-\frac{1}{2\pi}\log(x^2+y^2) \tag{3} \] 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 equipotential lines above mentioned will be lines of flow. $V=a$, that is \begin{flalign*} &&\frac{1}{\pi}&\left(\frac{\pi}{2}+\tan^{-1}\frac{x}{y}\right)=a, \tag{5}\\ &\text{reduces to} & y&=-x\tan a\pi. &&\phantom{reduces\ to} \tag{7} \end{flalign*} 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. $V_1=b$, that is \begin{flalign*} &&-\frac{1}{2\pi}&\log(x^2+y^2)=b, &&\tag{6}\\[-1ex] \intertext{reduces to}\\[-6ex] &&x^2&+y^2=e^{-2\pi b}. \tag{8} \end{flalign*} %[Illustration] \pngright{008.png}{701}{390}{-24} 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 constant values on the circumferences of two given concentric circles or where we have a \emph{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. % -----File: 083.png \EXAMPLE{S} 1.\quad Solve the problem of Art.~44 for the case where \[ f(x)=-1 \quad\text{if}\quad x<0 \quad\text{and}\quad f(x)= 1 \quad\text{if}\quad x>0. \] \begin{flushright} \textit{Ans.}, \quad $V = \dfrac{2}{\pi} \tan^{-1}\dfrac{x}{y}$. \end{flushright} 2.\quad Solve the problem of Art.~44 for the case where \[ f(x)=a \quad\text{if}\quad x<0 \quad\text{and}\quad f(x)=b \quad\text{if}\quad x>0. \] \begin{flushright} \textit{Ans.}, \quad $V = \dfrac{1}{2}(a+b) + \dfrac{1}{\pi}(b-a) \tan^{-1}\dfrac{x}{y}$. \end{flushright} 3.\quad Reduce (7), (8), and (9) Art.~44 to the forms \begin{align*} V &= \dfrac{1}{\pi} \int\limits_{-\infty}^\infty \dfrac{yf(\lambda)d\lambda}{y^2+(\lambda-x)^2}, \\ V &= \dfrac{1}{\pi} \int\limits_0^\infty yf(\lambda)d\lambda \left[ \dfrac{1}{y^2+(\lambda-x)^2} + \dfrac{1}{y^2+(\lambda+x)^2} \right], \\ V &= \dfrac{1}{\pi} \int\limits_0^\infty yf(\lambda)d\lambda \left[ \dfrac{1}{y^2+(\lambda-x)^2} - \dfrac{1}{y^2+(\lambda+x)^2} \right], \end{align*} respectively. \mypara{46.} An especially interesting case of Art.~44 is the following where \[ f(x)=0 \;\;\text{if}\;\; x<-1, \quad f(x)=1 \;\;\text{if}\;\; -1 1. \] \begin{flalign*} &\indent\text{Here} & V = \frac{1}{\pi} \left[ \tan^{-1}\frac{1+x}{y} + \tan^{-1}\frac{1-x}{y} \right]. &&\phantom{\indent Here} \tag{1} \end{flalign*} \begin{flalign*} &\text{Now} & \frac{1}{\pi} \log[(1-z)i] &= \frac{1}{\pi} \log[(1-x-yi)i] = \frac{1}{\pi} \log[y+(1-x)i] &&\phantom{Now} \\ &&&= \frac{1}{2\pi} \log[(1-x)^2+y^2] + \frac{i}{\pi} \tan^{-1}\frac{1-x}{y}, && \end{flalign*} and \begin{align*} -\frac{1}{\pi} \log[(-1-z)i] &= -\frac{1}{\pi} \log[(-1-x-yi)i] = -\frac{1}{\pi} \log[y-(1+x)i] \\ &= -\frac{1}{2\pi} \log[(1+x)^2+y^2] + \frac{i}{ \pi} \tan^{-1}\frac{1+x}{y}. \end{align*} \[ \frac{1}{ \pi} \log\frac{1-z}{-1-z} = \frac{1}{2\pi} \log\frac{(1-x)^2+y^2}{(1+x)^2+y^2} + \frac{i}{\pi} \left[ \tan^{-1}\frac{1+x}{y} + \tan^{-1}\frac{1-x}{y} \right]. \] % -----File: 084.png Hence \[ \frac{1}{\pi} \left( \tan^{-1}\frac{1+x}{y} + \tan^{-1}\frac{1-x}{y} \right) \quad\text{and}\quad \frac{1}{2\pi} \log\frac{(1-x)^2+y^2}{(1+x)^2+y^2} \] are \emph{conjugate functions:}\footnote {The function conjugate to \[ \frac{1}{\pi} \left[ \tan^{-1}\frac{1+x}{y} + \tan^{-1}\frac{1-x}{y} \right] \] might have been found as follows. If $\phi$ is the required function and $\psi$ the given function we have by Int.\ Cal.\ Arts.~211, 212, and 213 the relations \[ D_x\phi = D_y\psi \quad \text{and} \quad D_y\phi = -D_x\psi. \] \begin{flalign*} &\text{Here}& D_y\psi &= -\dfrac{1}{\pi} \left[ \dfrac{1+x}{(1+x)^2+y^2} + \dfrac{1-x}{(1-x)^2+y^2} \right] &&\phantom{Here} \\ &\text{and} & -D_x\psi &= -\dfrac{1}{\pi} \left[ \dfrac{y}{(1+x)^2+y^2} - \dfrac{y}{(1-x)^2+y^2} \right]. &&\phantom{and} \end{flalign*} If now we integrate $D_y\psi$ with respect to $x$ treating $y$ as a constant and add an arbitrary function of $y$ we shall have $\phi$. So that \begin{gather*} \phi = -\dfrac{1}{2\pi} \biggl\{ \log[(1+x)^2 + y^2\,] - \log[(1-x)^2 + y^2\,]\,\biggr\} + f(y). \\ D_y\phi = -\dfrac{1}{\pi} \biggl[ \dfrac{y}{(1+x)^2+y^2 } - \dfrac{y}{(1-x)^2+y^2} \biggr] + \dfrac{df(y)}{dy} \end{gather*} Comparing this with its equal $-D_x\psi$ above we find $\dfrac{df(y)}{dy}=0$ and $f(y) = C$ a constant \begin{flalign*} &\text{therefore} && \dfrac{1}{2\pi} \log{ \dfrac{(1-x)^2+y^2}{(1+x)^2+y^2} } + C, &&\phantom{therefore} \end{flalign*} where $C$ may be taken at pleasure, is our required conjugate function.}\ %[endfootnotetext] and \[ \dfrac{1}{\pi} \biggl( \tan^{-1}\dfrac{1+x}{y} + \tan^{-1}\dfrac{1-x}{y} \biggl) = a \tag{2} \] is any equipotential line, and \[ \dfrac{1}{2\pi} \log \dfrac{(1-x)^2+y^2 }{(1+x)^2+y^2} = b \tag{3} \] any line of flow for the system described at the beginning of this article; and \[ V_1 = \dfrac{1}{2\pi} \log \dfrac{(1-x)^2+y^2}{(1+x)^2+y^2} \tag{4} \] is the solution of a new problem for which (3) represents any equipotential line and (2) any line of flow. % -----File: 085.png (2) reduces to \[ \frac{2y}{x^2+y^2-1}=\tan a\pi \] \begin{flalign*} &\text{or} & x^2+(y-\ctn a\pi)^2 = \csc^2 a\pi;& \tag{5}& \end{flalign*} \begin{flalign*} &\text{and (3) to} & x^2 + y^2 + 2\frac{e^{2b\pi}+1}{e^{2b\pi}-1}x + 1 = 0 &&\phantom{and\ (3)\ to} \end{flalign*} \begin{flalign*} &\text{or} & \left( x + \frac{e^{b\pi}+e^{-b\pi}}{e^{b\pi}-e^{-b\pi}} \right)^2 + y^2 = \left( \frac{e^{b\pi}+e^{-b\pi}}{e^{b\pi}-e^{-b\pi}} \right)^2 - 1 &&\phantom{or} \end{flalign*} \begin{flalign*} &\text{or} & (x+\ctnh b\pi)^2+y^2=\csch^2 b\pi. &&\phantom{or} \tag{6} \end{flalign*} (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. %[Illustration] \pngcent{009.png}{1074} {}\EXAMPLE{S} 1.\quad Show that if $f(x)=a_1$ when $x<-b$, $f(x)=a_2$ when $-b b$, \[ V = \frac{a_1+a_3}{2} + \frac{1}{\pi} \left[ (a_2-a_1)\tan^{-1}\frac{b+x}{y} + (a_2-a_3)\tan^{-1}\frac{b-x}{y} \right]. \] % -----File: 086.png 2.\quad Show that if $f(x) = 0$ if $x < 0$, $f(x) = a_1$ if $0 < x < b_1$, $f(x) = a_2$ if $b_1 < x < b_2$, $f(x) = a_3$ if $b_2 < x < b_3$, \&c., \begin{multline*} V = \frac{1}{\pi}\left [a_1 \tan^{-1} \frac{x}{y }+ (a_1 - a_2) \tan^{-1} \frac{b_1 - x}{y }+ (a_2 - a_3) \tan^{-1} \frac{b_2 - x}{y} \right. \\ \left.{} + (a_3 - a_4) \tan^{-1} \frac{b_3 - x}{y }+ \cdots \right]. \end{multline*} 3.\quad Show that if $f(x) = -1$ if $x < -1$, $f(x) = x$ if $-1 < x < 1$, $f(x) = 1$ if $x > 1$, \[ V = \frac{1}{\pi}\left [(1 + x)\tan^{-1} \frac{1 + x}{y } - (1 - x) \tan^{-1}\frac {1 - x}{y }+ \frac{y}{2} \log \frac{(1 - x)^2 + y^2 }{ (1 + x)^2 + y^2}\right]. \] 4.\quad Show that if $f(x) = -1$ if $x < -1$, $f(x) = 0$ if $-1 < x < 1$, $f(x) = 1$ if $x > 1$, \[ V = \frac{1}{\pi}\left [\tan^{-1} \frac{1 + x }{ y }- \tan^{-1} \frac{1 - x}{y }\right ]. \] 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 having $(-1, 0)$ 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 surrounding 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.\quad 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 $x > b$, \[ V = \frac{1}{\pi}\left[\frac{\pi}{2} - \tan^{-1} \frac{a - x }{y }- \tan^{-1} \frac{b - x }{y} - \tan^{-1} \frac{x}{y}\right]. \] The conjugate function \[ V = \frac{1}{2\pi} \log \frac{x^2 + y^2 }{[(a-x)^2 + y^2] [(b-x)^2 + y^2]} \] 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 $ x^2 + y^2 = ab$. \mypara{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 % -----File: 087.png 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 \[ V = \frac{2}{\pi} \int\limits_0^\infty d\alpha \int\limits_0^\infty e^{-\alpha y} f(\lambda) \sin \alpha x \sin \alpha\lambda.d\lambda. \tag{1} \] v.~(9) Art.~44. \noindent This reduces to \[ V = \frac{1}{\pi} \int\limits_0^\infty f(\lambda)d\lambda \left[ \frac{y}{y^2+(\lambda-x)^2} -\frac{y}{y^2+(\lambda+x)^2} \right]. \tag{2} \] v.\ Ex.~3 Art.~45. \EXAMPLE{S} 1.\quad If $V=0$ when $y=0$ and $V=F(y)$ when $x=0$ show that \begin{align*} V &= \frac{2}{\pi} \int\limits_0^\infty d\alpha \int\limits_0^\infty e^{-\alpha x} F(\lambda) \sin\alpha y \sin\alpha\lambda.d\lambda\\ &= \frac{1}{\pi} \int\limits_0^\infty F(\lambda) d\lambda \left[ \frac{x}{x^2+(\lambda-y)^2} - \frac{x}{x^2+(\lambda+y)^2} \right]. \end{align*} 2.\quad If $V=f(x)$ when $y=0$ and $V=F(y)$ when $x=0$ show that \begin{align*} V = \frac{1}{\pi} \int\limits_0^\infty & \left[ f(\lambda) \left( \frac{y}{y^2+(\lambda-x)^2} - \frac{y}{y^2+(\lambda+x)^2} \right) \right. \\ & \left. {}+ F(\lambda) \left( \frac{x}{x^2+(\lambda-y)^2} - \frac{x}{x^2+(\lambda+y)^2} \right) \right] d\lambda. \end{align*} 3.\quad If $F(y)=b$ the result of Ex.~2 reduces to \[ V = \frac{2b}{\pi} \tan^{-1}\frac{y}{x} + \frac{1}{\pi} \int\limits_0^\infty f(\lambda) d\lambda \left[ \frac{y}{y^2+(\lambda-x)^2} - \frac{y}{y^2+(\lambda+x)^2} \right]. \] 4.\quad If $F(y)=1$ for $01$ while $f(x)=1$ for $01$ \begin{align*} V = \frac{1}{\pi} & \left[ \tan^{-1}\frac{1-x}{y} - \tan^{-1}\frac{1+x}{y} + 2\tan^{-1}\frac{y}{x} \right.\\ & \left. {}+ \tan^{-1}\frac{1-y}{x} - \tan^{-1}\frac{1+y}{x} + 2\tan^{-1}\frac{x}{y} \right]. \end{align*} % -----File: 088.png 5.\quad If one edge of the conducting sheet treated in Art.~47 is insulated, so that $D_{x}V=0$ if $x = 0$ and $V=f(x)$ when $y = 0$ \begin{align*} V &= \frac{2 }{ \pi } \int\limits_0^{\infty} d\alpha \int\limits_0^{\infty} e^{-\alpha y} f(\lambda) \cos\alpha x \cos\alpha\lambda.d\lambda\\ &= \frac{1 }{ \pi } \int\limits_0^{\infty} f(\lambda) d\lambda \left [ \frac{y }{ y^2+(\lambda+x)^2 } + \frac{y }{ y^2+(\lambda-x)^2} \right]. \end{align*} \mypara{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 $y = \infty$ \quad $V= e^{\alpha y}\sin ax$ and $V=e^{\alpha y} \cos ax$ are available as particular solutions of the equation \[ D_x^2V+D_y^2V=0 \tag{1} \] as well as $V=e^{-\alpha y}\sin\alpha x$ and $V=e^{-\alpha y}\cos\alpha x$. \begin{flalign*} &\indent\text{Consequently }&& \frac{e^{\alpha y} + e^{-\alpha y}}{2} \sin \alpha x = \cosh\alpha y \sin\alpha x & \text{[Int.\ Cal.\ Art.~43 (2)]} \\ &\text{and}&& \frac{e^{\alpha y} - e^{-\alpha y} }{2} \sin \alpha x = \sinh\alpha y \sin\alpha x & \text{[Int.\ Cal.\ Art.~43 (1)]}\\[-5ex] \end{flalign*} \begin{flalign*} &\text{and}&& \cosh \alpha y \cos \alpha x \quad \text{and}\quad \sinh\alpha y \cos\alpha x &&\phantom{and} \end{flalign*} are now available values of $V$ and can be used precisely as $e^{-\alpha y}\cos\alpha x$ and $e^{-\alpha y} \sin\alpha x$ are used in Art.~44. Following the same course as in Art.~44 we get \[ V = \frac{1}{ \pi } \int\limits_0^{\infty} d\alpha \int\limits_{-\infty}^{\infty} \frac{\sinh\alpha y}{\sinh\alpha b} F(\lambda) \cos\alpha(\lambda-x).d\lambda \tag{2} \] as a solution of (1) which will reduce to $V= F(x)$ when $y= b$ \begin{flalign*} &\text{and to }&& V = 0 \quad\text{when}\quad y = 0\text{, since } \sinh 0 = \frac{1-1 }{2} = 0, &&\phantom{and\ to} \end{flalign*} 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 \[ V = \frac{1}{ \pi } \int\limits_0^{\infty} d\alpha \int\limits_{-\infty}^{\infty}\frac{\sinh \alpha(b-y) }{\sinh \alpha b} f(\lambda) \cos \alpha(\lambda-x).d\lambda. \tag{3} \] % -----File: 089.png If $V = f(x)$ when $y = 0$ and $V = F(x)$ when $y = b$ then \begin{equation*} \begin{split} V = {}& \frac{1}{\pi} \int\limits^\infty_{0} d\alpha \int\limits^\infty_{-\infty} \dfrac{\sinh\alpha(b - y)}{\sinh\alpha b} f(\lambda)\cos {\alpha(\lambda - x)}.d\lambda \\ &+ \frac{1}{\pi} \int\limits^\infty_{0} d\alpha \int\limits^\infty_{-\infty} \frac{\sinh\alpha y} {\sinh\alpha b} F(\lambda)\cos {\alpha(\lambda - x)}.d\lambda. \end{split} \end{equation*} This can be considerably simplified by the aid of the formula \[ \int \limits^{\infty}_{0} \dfrac{\sinh px}{\sinh qx}\cos{rx.dx} = \dfrac{\pi}{2q} \dfrac{\sin \dfrac{p \pi}{q}} {\cos\dfrac{p \pi}{q} + \cosh \dfrac{r \pi}{q}} \] if $p^2 < q^2$. [Bierens de Haan, Tables of Def.\ Int.\ (7) 265] and becomes \begin{flalign*} && V ={} & \dfrac{1}{2b} \sin \dfrac{\pi}{b} (b - y) \int \limits^\infty_{-\infty}f(\lambda) \dfrac{d\lambda}{\cos \dfrac{\pi(b - y)}{b} + \cosh \dfrac{\pi}{b} (\lambda - x)} && \\ &&\phantom{or} & {}+ \dfrac{1}{2b} \sin \dfrac{\pi y}{b} \int \limits^\infty_{-\infty} F(\lambda) \dfrac{d\lambda} {\cos \dfrac{\pi y}{b} + \cosh \dfrac{\pi}{b} (\lambda - x)} &\text{or}& \end{flalign*} \[\tag{5} V = \dfrac{1}{2b}\sin \dfrac{\pi y}{b} \int\limits^\infty_{-\infty} \biggl [\dfrac{f\lambda}{\cosh \dfrac{\pi}{b} (\lambda - x) - \cos\dfrac{\pi y}{b}} + \dfrac{F\lambda}{\cosh\dfrac{\pi}{b} (\lambda - x) + \cos\dfrac{\pi y}{b}}\biggr] d\lambda. \] \EXAMPLE{S} 1.\quad Given the formula \[ \int \dfrac{dx}{a+b \cosh x} = \dfrac{2}{\sqrt{b^2 - a^2}} \tan^{-1}\biggl(\sqrt{\dfrac{b-a}{b+a}} \tanh \dfrac{x}{2}\biggr) \quad\text{ if } b > a, \] show that if $V = 1$ when $y = 0$ and $V = 0$ when $y = b$\;\;$V = \dfrac{1}{b}(b - y).$\\ 2.\quad Show that if $V = 0$ when $y = b$, $V = -1$ when $y = 0$ and $x < 0$, and $V = 1$ when $y = 0$ and $x > 0$ \[ V=\dfrac{2}{\pi}\tan^{-1}\Biggl[\dfrac{\tanh \dfrac{\pi x}{2b}} {\tan \dfrac{\pi y}{2b}}\Biggr] \] The solution for the conjugate system, that is, for a strip having a source at $(0, 0)$ and an infinitely distant sink is \[ V = -\dfrac{1}{\pi}\log \biggl[\cosh^{2}\dfrac{\pi x}{2b} - \cos^{2}\dfrac{\pi y}{2b}\biggr]. \] % -----File: 090.png 3.\quad 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 $x > 0$, \[ \begin{split} V & = \frac{2}{\pi}\tan^{-1}\bigg ( \tan \dfrac{\pi}{2b} (b -y) \tanh \dfrac{\pi x}{2b}\bigg ) + \frac{2}{\pi}\tan^{-1} \bigg ( \tan \dfrac{\pi}{2b} y \tanh \dfrac{\pi x}{2b}\bigg ) \\ &=\frac{2}{\pi}\tan^{-1} \Bigg[ \frac{\sinh \dfrac{\pi x}{b }} {\sin\dfrac{ \pi y}{b}} \Bigg]. \end{split} \] 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 \[ V = \frac{1}{\pi}\log \Bigg[ \dfrac{\cosh\dfrac{ \pi x}{b} + \cos \dfrac{ \pi y}{b}}{\cosh\dfrac{ \pi x}{b} - \cos \dfrac{ \pi y}{b} } \Bigg]. \] 4.\quad If $V=0$ when $x = 0$, $V = f(x)$ when $y = 0$ and $x>0$, and $V=0$ when $y = b$ and $x > 0$. \[ \begin{split} V &= \frac{1}{\pi} \int\limits_0^{\infty} d\alpha \int\limits_0^{\infty} \frac{\sinh\alpha(b-y)}{\sinh\alpha b} [ \cos\alpha(\lambda-x) - \cos\alpha(\lambda+x) ] f(\lambda) d\lambda\\ &= \frac{1}{2b} \sin \dfrac{\pi y }{b} \int\limits_0^{\infty} \biggl[ \frac{1} {\cosh \dfrac{\pi }{b} (\lambda-x) - \cos \dfrac{\pi y}{ b}} - \frac{1} {\cosh \dfrac{\pi }{b} (\lambda+x) - \cos \dfrac{\pi y}{ b}} \biggr] f(\lambda) d\lambda \end{split} \] for positive values of $x$ and for values of $y$ between $0$ and $b$.\\ 5.\quad If $V_1 = 0$ when $x = 0$, $V_1 = F(x)$ when $y = b$ and $x > 0$, and $V_1 = 0$ when $y = 0$ and $x>0$ \[ V_1 = \frac{1}{2b} \sin \frac{\pi y}{ b} \int\limits_0^{\infty} \biggl[ \frac{1} {\cosh \dfrac{\pi }{b} (\lambda-x) + \cos \dfrac{\pi y}{ b}} - \frac{1} {\cosh \dfrac{\pi }{b} (\lambda+x) - \cos \dfrac{\pi y}{ b}} \biggr] F(\lambda) d\lambda \] for positive values of $x$ and values of $y$ between $0$ and $b$.\\ 6.\quad If $V_2=0$ when $x = 0$, $V_2 =f(x)$ when $y = 0$ and $x > 0$, and $V_2 = F(x)$ when $y = b$ and $x>0$ \[ \phantom{(v.\ Exs.\ 4 and 5)} V_2 = V+ V_1 \quad \text{for} \quad x > 0\quad \text{and} \quad 0 < y < b. \tag*{(v.\ Exs.\ 4 and 5)} \] 7.\quad If one edge of the strip described in Art.~48 is insulated so that we have $V=f(x)$ when $y = 0$ and $D_y V=0$ when $y = b$ show that \[ V = \frac{1}{\pi} \int\limits_0^{\infty}d\alpha \int\limits_{-\infty}^{\infty} \frac{\cosh\alpha(b-y)}{\cosh\alpha b} f(\lambda) \cos\alpha(\lambda-x).d\lambda. \] % -----File: 091.png By the aid of the formula \begin{flalign*} &\phantom{if\ pa$ \[ V = \frac{1}{\pi} \Bigg[ \tan^{-1}\frac{\sinh\dfrac{\pi(a-x)}{b}}{\sin\dfrac{\pi y}{b}} +\tan^{-1}\frac{\sinh\dfrac{\pi(a+x)}{b}}{\sin\dfrac{\pi y}{b}} \Bigg]. \] 9.\quad If $V=0$ when $y=0 \text{\; or \;} b$ and $x<-a$, $V=1$ when $y=0$ and $-aa$, and $V=-1$ when $y=b$ and $-a b$, when $t = 0$.\\ Then \[ u = \frac{1 }{\sqrt{\pi}} \!\!\int\limits_{-\frac{b+x}{2a\sqrt{t}}}^{\frac{b-x}{2a\sqrt{t}}}\!\! e^{-\beta^2} d\beta = \frac{2 }{\sqrt{\pi}} \left [ \frac{b}{ 2a \sqrt{ t }} - \frac{b^3+3bx^2 }{ 3(2a \sqrt{ t })^3} + \frac{b^5 + 10b^3x^2 + 5bx^4 }{5.2!( 2a \sqrt{ t })^5} - \cdots \right]. \] 5.\quad Let $u = 0$ if $x < 0$ and $u = 1$ if $x > 0$ when $t = 0$. Then \begin{align*} u &= \frac{1 }{\sqrt{\pi}} \!\!\int\limits_{-\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} d\beta = \frac{1 }{\sqrt{\pi}} \biggl[ \int\limits_0^{\frac{x}{2a\sqrt{t}}}\!\! e^{-\beta^2} d\beta + \int\limits_0^{\infty} e^{-\beta^2 } d\beta \biggr] = \frac{1 }{\sqrt{\pi}} \!\!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\!\! e^{-\beta^2 } d\beta + \frac{1 }{2} \\ &= \frac{1}{ 2} + \frac{1 }{\sqrt{\pi}} \left [ \frac{ x }{ 2a \sqrt{ t } } - \frac{ x^3}{ 3.(2a \sqrt{ t })^3} + \frac{ x^5}{ 5.2!(2a \sqrt{ t })^5} - \frac{ x^7}{ 7.3!(2a \sqrt{ t })^7} + \cdots \right]. \end{align*} 6.\quad 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 \ $a^2 = 0.185$ \ in C.G.S. units. \hfill \emph{Ans.}, 21°.6.\\ % -----File: 094.png 7.\quad 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. \hfill \emph{Ans.}, 70°.8, 29°.2.\\ 8.\quad Find a particular solution of $D_tu = a^2D_x^2u$ 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. \mypara{50.} If our solid has one plane face which is kept at the constant temperature 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 \[ D_t u = a^2D_x^2 u, \tag{1} \] but our conditions are \begin{alignat*}{3} u &= 0 && \text{when} \quad &x = 0 \tag{2}\\ u &= f(x) && \text {\quad`` } \quad & t = 0 \tag{3} \end{alignat*} and we are concerned only with positive values of $x$. We may then use the form (4) Art.~32 \[ f(x) = \frac{2}{\pi}\int\limits_0^{\infty} d\alpha \int\limits_0^{\infty} f(\lambda) \sin\alpha x \sin\alpha \lambda.d\lambda, \tag{4} \] and proceeding as in the last section we get \[ u = \frac{2}{\pi} \int\limits_0^{\infty} d\alpha \int\limits_0^{\infty} e^{-a^2 \alpha^2 t} f(\lambda) \sin\alpha x \sin\alpha\lambda.d\lambda \tag{5} \] as our required solution. This may be reduced considerably. \[ u = \frac{1}{\pi} \int\limits_0^{\infty} f(\lambda) d\lambda \int\limits_0^{\infty} e^{-a^2 \alpha^2 t}[\cos \alpha (\lambda - x) - \cos \alpha (\lambda + x)]d\alpha, \] \[ \text{or} \quad u = \frac{1}{2a\sqrt{\pi t}} \int\limits_0^{\infty} f(\lambda)(e^{-\frac{(\lambda-x)^2 }{ 4a^2 t}} - e^{-\frac{(\lambda+x)^2 }{ 4a^2 t}} ) d\lambda \tag{6} \] by (7) Art.~49, and this may be reduced to the form \[ u = \frac{1}{\sqrt{ \pi} }\biggl [\!\int\limits_{-\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} f(x + 2a \sqrt{ t}.\beta) d\beta - \!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} f(-x + 2a \sqrt{ t }.\beta) d\beta \biggr ] . \tag{7} \] \EXAMPLE{S} 1.\quad Let the initial temperature be constant and equal to $c$. % -----File: 095.png Then \[ \begin{split} u &= \frac{ c}{\sqrt{ \pi}} \biggl [ \!\int\limits_{-\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} d\beta - \!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} d\beta \biggr ]\\ &= \frac{2c}{\sqrt{ \pi}} \!\int\limits_{0}^{\frac{x}{2a\sqrt{t}}}\!e^{-\beta^2} d\beta\\ &= \frac{2c}{\sqrt{ \pi}} \left [ \frac{x }{ 2a \sqrt{ t}} - \frac{x^3}{3.(2a \sqrt{t})^3} + \frac{x^5}{5.2!(2a \sqrt{t})^5} - \frac{x^7}{7.3!.(2a \sqrt{t})^7} + \cdots \right ]. \end{split} \] 2.\quad Assuming that the earth was originally at the temperature 7000° Fahrenheit 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 cooling 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 $a^2 (400)$ the foot, the Fahrenheit degree, and the year being taken as units. (Thomson and Tait's Nat.\ Phil.\ Vol.\ II. Appendix.) \emph{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.\quad Let the initial temperature be constant and equal to $ -b$, then by Ex.~1 \[ u = - \frac{2b}{\sqrt{\pi}} \!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\! e^{-\beta^2 }d\beta . \] 4.\quad 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 \[ u = b \biggl ( 1 - \frac{2}{\sqrt{ \pi}} \!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\! e^{-\beta^2} d\beta \biggr ) . \] 5.\quad Let $u = b$ when $x = 0$ and $u = f(x)$ when $t = 0$. Then \[ u = b \biggl ( 1 - \frac{2}{\sqrt{ \pi}} \!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\! e^{-\beta^2} d\beta \biggr ) + \frac{1}{ 2a \sqrt{\pi t }}\int\limits_0^{\infty} f(\lambda) [ e^{-{\frac{(\lambda-x)^2 }{4a^2t}}} - e^{-{\frac{(\lambda+x)^2 }{4a^2t }}}] d\lambda \] by (6) Art.~50.\\ % -----File: 096.png 6.\quad Let $u=b$ when $x=0$ and $u=c$ when $t=0$. \begin{flalign*} &\indent \text{Then} && u = b + (c - b) \frac{2}{\sqrt{\pi}}\!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\! e^{-\beta^2} d\beta.&& \phantom{\indent Then} \end{flalign*} 7.\quad If the earth has been cooling for 200,000,000 years from a uniform temperature, 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 maximum value for all time. Let $a^2 = 400$.\\ 8.\quad Show that if the plane face of the solid considered in Art.~50 instead of being kept at temperature zero is impervious to heat \[\phantom{v.\ (6) Art.~50.} u = \frac{1}{2a \sqrt{\pi t}} \int\limits_0^{\infty} f(\lambda)(e^{-\frac{(\lambda - x)^2 }{ 4a^2t}} + e^{-\frac{(\lambda + x)^2}{ 4a^2t}}) d\lambda. \tag*{v.\ (6) Art.~50.} \] \mypara{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 \[ D_t u = a^2 D_x^2 u \tag{1} \] subject to the conditions \[ \left . \begin{aligned} u& = F(t) &&\text{when} &x = 0\phantom{.}\\ u& = 0 &&\text{\quad``} &t = 0. \end{aligned}\right \} \tag{2} \] We know that \[ u= \frac{2}{\sqrt{ \pi}} \!\int\limits_0^{\frac{x}{2a\sqrt{t}}}\! e^{-\beta^2} d\beta \] is a solution of (1), v.\ Ex.~1 Art.~50. It is easily shown that \[ u= \frac{2}{\sqrt{ \pi}} \!\!\int\limits_0^{\frac{x}{2a\sqrt{t-c}}}\!\! e^{-\beta^2} d\beta, \tag{3} \] where $c$ is any constant, is a solution of (1). For \begin{align*} D_t u &= -\frac{2}{\sqrt{ \pi}} \frac{ x}{2a}\frac{ 1}{2(t-c)^\frac{3}{2}} e^{-\frac{x^2}{4a^2(t-c)}} = -\frac{x}{2a \sqrt{ \pi} }(t - c)^{-\frac{3}{2}} e^{- \frac{x^2}{4a^2(t-c)}}\\ D_x u &= \frac{2}{\sqrt{ \pi}} \frac{ 1}{2a\sqrt{t-c}} e^{-\frac{x^2}{4a^2(t-c)}}\\ D_x^2u&= - \frac{2}{\sqrt{ \pi}} \frac{ 1}{2a\sqrt{t-c}}\frac{2x}{4a^2(t-c)}e^{-\frac{x^2}{ 4a^2(t-c)}} = -\frac{x}{2a^3 \sqrt{ \pi}}(t-c)^{-\frac{3}{2}} e^{-\frac{x^2}{4a^2(t-c)}} \end{align*} \begin{flalign*} \text{and} && D_t u &= a^2 D_x^2 u.&& \end{flalign*} % -----File: 097.png Let $\phi(x,t)$ be a function of $x$ and $t$ which shall be equal to zero if $t$ is negative and shall be equal to \[ 1-\frac{2}{\sqrt\pi}\!\int\limits_0^\frac{x}{2a\sqrt t}\! e^{-\beta^2}d\beta \] if $t$ is equal to or greater than zero; so that if $x=0$\;\;$\phi(x,t)=1$ and if $t=0$\;\;$\phi(x,t)=0$. We shall now attack the following problem, to solve equation (1) subject to the conditions \begin{alignat*}{5} &u=0 && \text{if}\quad && t=0 && &&\\ &u=F(0) && \text{``} && x=0\quad && \text{and}\quad && 0(k+1)\tau$ and $u=F(k\tau)$ if $k\taut$. If, then, $n\tau$ is the greatest whole multiple of $\tau$ not exceeding $t$, \[ u=\sum_{k=0}^{k=n}F(k\tau)[\phi(x,t-k\tau)-\phi(x,t-(k+1)\tau)]. \tag{6} \] If now we decrease $\tau$ indefinitely the limiting form of (6) will be the solution of the problem stated at the beginning of this article. (6) may be written \[ u=\sum_{k=0}^{k=n}F(k\tau)\left[\frac{\phi(x,t-k\tau)-\phi(x,t-(k+1)\tau)}{\tau}\right]\tau \tag{7} \] % -----File: 098.png and if $\tau$ is indefinitely decreased the limiting form of (7) is \[ u=-\int\limits_0^t F(\lambda)D_\lambda\phi(x,t-\lambda)d\lambda. \tag{8} \] Since $t-\lambda$ is positive between the limits of integration \begin{flalign*} && \phi(x,t-\lambda)&=1-\frac{2}{\sqrt\pi}\!\!\int\limits_0^\frac{x}{2a\sqrt{t-\lambda}}\!\! e^{-\beta^2}d\beta, &&\\ &\text{and} & D_\lambda\phi(x,t-\lambda)=&-\dfrac{x}{2a\sqrt\pi}e^{-\frac{x^2}{4a^2(t-\lambda)}}(t-\lambda)^{-\frac{3}{2}}; && \phantom{and} \end{flalign*} and (8) may be written \[ u=\frac{x}{2a\sqrt\pi}\int\limits_0^t F(\lambda)e^{-\frac{x^2}{4a^2(t-\lambda)}}(t-\lambda)^{-\frac{3}{2}}d\lambda, \tag{9} \] or if we let \qquad\qquad\qquad\qquad $\beta=\dfrac{x}{2a\sqrt{t-\lambda}}$ \[ u=\frac{2}{\sqrt\pi}\!\int\limits_{\frac{x}{2a\sqrt t}}^\infty\! e^{-\beta^2}F\left(t-\frac{x^2}{4a^2\beta^2}\right)d\beta. \tag{10} \] \EXAMPLE{S} 1.\quad If $u=nt$ when $x=0$ and $u=0$ when $t=0$ \[ u=n\bigg(t+\frac{x^2}{2a^2}\bigg)\bigg[1-\frac{2}{\sqrt\pi}\!\int\limits_0^\frac{x}{2a\sqrt t}\!e^{-\beta^2}d\beta\bigg]-\frac{nx\sqrt t}{a\sqrt\pi}e^{-\frac{x^2}{4a^2t}}. \] 2.\quad 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; $a^2=.185$. \hfill \textit{Ans.}, 20°.1.\\ 3.\quad If $u=F(t)$ when $x=0$ and $u=f(x)$ when $t=0$, then \[ u=\frac{2}{\sqrt\pi}\int\limits_\frac{x}{2a\sqrt t}^\infty e^{-\beta^2}F\bigg(t-\frac{x^2}{4a^2\beta^2}\bigg)d\beta +\frac{1}{2a\sqrt{\pi t}}\int\limits_0^\infty(e^{-\frac{(\lambda-x)^2}{4a^2t}}-e^{-\frac{(\lambda+x)^2}{4a^2t}})f(\lambda)d(\lambda). \] v.~(6) Art.~50.\\ % -----File: 099.png 4.\quad 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 \iffalse \[ F(t) =\frac{1}{2}b_0 + \sum\limits_{m=1}^{m=\infty}[a_m\sin m\alpha t + b_m\cos m\alpha t],\qquad \text{where}\qquad \alpha=\frac{2\pi}{T}, \] or \qquad\qquad $F(t)=\dfrac{1}{2b_0} + \sum\limits_{m=1}^{m=\infty}\rho_m\sin(m\alpha t + \lambda_m)$, \begin{flalign*} &\text{where} & \rho_m\cos\lambda_m=a_m \quad \text{and}\quad \rho_m\sin\lambda_m=b_m. && \text{v.\ Art.~31 Ex.~3.}& \end{flalign*} \fi \begin{flalign*} &&F(t) =\frac{1}{2}b_0 &+ \sum\limits_{m=1}^{m=\infty}[a_m\sin m\alpha t + b_m\cos m\alpha t],\quad \text{where} &&\alpha=\frac{2\pi}{T},\\ &\text{or} & F(t)=\dfrac{1}{2b_0} &+ \sum\limits_{m=1}^{m=\infty}\rho_m\sin(m\alpha t + \lambda_m),\\ &\rlap{where\hfill} && \rho_m\cos\lambda_m=a_m \quad \text{and}\quad \rho_m\sin\lambda_m=b_m. &&\text{v.\ Art.~31 Ex.~3.} \end{flalign*} Show that with this value of $F(t)$ (10) Art.~51.\ becomes \begin{gather*} \begin{split} u=\frac{1}{\sqrt\pi}b_0\!\int\limits_\frac{x}{2a\sqrt t}^{\infty}\! e^{-\beta^2}d\beta + \frac{2}{\sqrt\pi}\sum\limits_{m=1}^{m=\infty}\rho_m\biggl[\sin(m \alpha t + \lambda_m)\int\limits_{\frac{x}{2a\sqrt t}}^{\infty}e^{-\beta^2}\cos\frac{m\alpha x^2}{4a^2\beta^2}d\beta\biggr.\\ \biggl.-\cos(m\alpha t + \lambda_m)\!\int\limits_\frac{x}{2a\sqrt t}^{\infty}\! e^{-\beta^2}\sin\frac{m\alpha x^2}{4a^2\beta^2}d\beta\biggr] \end{split} \end{gather*} and that as $t$ increases $u$ approaches the value \[ u= \frac{1}{2}b_0 + \sum\limits_{m=1}^{m=\infty}\rho_m e^{-\frac{x}{a}\sqrt\frac{m\alpha}{2}}\sin(m\alpha t - \frac{x}{a}\sqrt\frac{m\alpha}{2}+\lambda_m). \] Given that \[ \int\limits_{0}^{\infty}e^{-x^2}\sin\frac{b^2}{x^2}dx=\frac{\sqrt\pi}{2}e^{-b\sqrt2}\sin b\sqrt2;\quad \int\limits_{0}^{\infty}e^{-x^2}\cos\frac{b^2}{x^2}dx=\frac{\sqrt\pi}{2}e^{-b\sqrt2}\cos b\sqrt2. \] v.\ \textit{Riemann, Lin.\ par.\ dif.\ gl.}\ §~54.\\ 5.\quad 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 \[ \phantom{v.\ Fourier, Heat §~105.} D_t u = a^2D_x^2 u - b^2u. \tag*{v.\ Fourier, Heat §~105.} \] Show that for this case \[ u = e^{-(b^2+a^2\alpha^2)t}\sin\alpha x \qquad \text{and}\qquad u=e^{-(b^2+a^2\alpha^2)t}\cos\alpha x \] are particular solutions, and that if $u=f(x)$ when $t=0$ \[ u=\frac{e^{-b^2t}}{2a\sqrt{\pi t}}\int\limits_{-\infty}^{\infty}e^{-\frac{(\lambda-x)^2}{4a^2t}}f(\lambda)d\lambda =\frac{e^{-b^2t}}{\sqrt\pi}\int\limits_{-\infty}^{\infty} e^{-\beta^2}f(x+2a\sqrt t.\beta)d\beta. \] cf.~(8) and (9) Art.~49. % -----File: 100.png If $u = 0$ when $x = 0$ and $u = f(x)$ when $t = 0$ \[ u = \frac{e^{-b^2t}}{\sqrt{ \pi}}\biggl [\!\int\limits_{-\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} f(x+2a \sqrt{ t }.\beta) d\beta - \!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2} f(-x + 2a \sqrt{ t}.\beta) d\beta \biggr ]. \] cf.~(7) Art.~50. If $u = -e^{-\frac{bx}{a}}$ when $t = 0$ and $u = 0$ when $x = 0$ \[ u = \frac{1}{\sqrt{\pi}} \biggl [e^{\frac{bx}{a}} \!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-(b \sqrt{ t} + \beta)^2} d\beta - e^{-\frac{bx}{a}} \!\!\int\limits_{-\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-(b \sqrt{ t} + \beta)^2} d\beta \biggr ], \] and if $u = 1$ when $x = 0$ and $u = 0$ when $t = 0$ we have only to add $e^{-\frac{bx}{a}}$ to the second member of the last equation, since $u = e^{-\frac{bx}{a}}$ satisfies the equation \[ D_t u = a^2 D_x^2 u - b^2u. \] If $u = F(t)$ when $x = 0$ and $u = 0$ when $t = 0$ we can employ the method of Art.~51. \begin{gather*} \phi(x,t- \lambda) = e^{-\frac{bx}{a}} + \frac{1}{\sqrt{ \pi}} \biggl [e^{\frac{bx}{a}} \hspace{-10pt}\int\limits_{\frac{x}{2a\sqrt{t-\lambda}}}^{\infty}\hspace{-10pt} e^{-(b \sqrt{ t - \lambda} + \beta)^2} d\beta - e^{-\frac{bx}{a}} \hspace{-10pt}\int\limits_{-\frac{x}{2a\sqrt{t-\lambda}}}^{\infty}\hspace{-10pt} e^{-(b \sqrt{t-\lambda} + \beta)^2} d\beta \biggr ],\\ -D_{\lambda} \phi(x,t-\lambda) = \frac{x(t-\lambda)^{-\frac{3}{2}}} { 2a \sqrt{ \pi}} e^{-b^2(t-\lambda)- \frac{x^2}{4a^2 (t-\lambda)}}; \end{gather*} \begin{flalign*} &\text{and} &&u = \frac{x}{2a\sqrt{\pi} }\int\limits_0^t (t-\lambda)^{-\frac{3}{2}} e^{-b^2(t-\lambda) -\frac{x^2}{4a^2(t-\lambda)}} F(\lambda)d\lambda,&&\phantom{and} \end{flalign*} cf.\ (9) Art.~51, \begin{flalign*} &\text{or} &&u = \frac{2}{\sqrt{ \pi}} \!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\! e^{-\beta^2 -\frac{ b^2 x^2}{4a^2\beta^2} }F\biggl (t-\frac{x^2}{4a^2\beta^2}\biggr )d\beta,&&\phantom{or} \end{flalign*} 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 \[ u = \frac{1}{2}b_0 e^{-\frac{bx}{a}}+ \sum_{m=1}^{m=\infty} \rho_m e^{-\frac{x \sqrt{ 2}}{ 2a} p} \sin \biggl (m\alpha t - \frac{x \sqrt{2}}{2a }q + \lambda_m \biggr), \] \begin{flalign*} &\text{where} && p = (b^2 + \sqrt{b^4 + m^2\alpha^2})^{\frac{1}{2}} \quad \text{and} \quad q = (-b^2 + \sqrt{b^4 + m^2\alpha^2})^{\frac{1}{2}}. &&\phantom{where} \end{flalign*} % -----File: 101.png \begin{flalign*} &\indent\text{Given} & \int\limits_0^\infty e^{-x^2-\frac{a^2}{x^2}}dx&=\frac{\sqrt\pi}{2}e^{-2a}&\phantom{\indent Given}\\ & & \int\limits_0^\infty e^{-x^2-\frac{a^2}{x^2}}\sin\frac{b^2}{x^2}dx &=\frac{\sqrt\pi}{2}e^{-2c}\sin 2d&\\ &\text{and} & \int\limits_0^\infty e^{-x^2-\frac{a^2}{x^2}}\cos\frac{b^2}{x^2}dx &=\frac{\sqrt\pi}{2}e^{-2c}\cos 2d,& \end{flalign*} where \[ c=\frac{\sqrt2}{2}(a^2+\sqrt{a^4+b^4})^\frac{1}{2}\quad \text{and}\quad d=\frac{\sqrt2}{2}(-a^2+\sqrt{a^4+b^4})^\frac{1}{2}. \] 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. \mypara{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 $u=e^{\beta t+\alpha x}$ is a solution of \[ D_tu = a^2D_x^2u \tag{1} \] provided only that $\beta=a^2\alpha^2$ we have \[ u=e^{\beta t±\frac{x}{a}\sqrt\beta} \] as a solution. Replacing $\beta$ by $±\beta i$ this becomes \begin{flalign*} & & u &=e^{±\beta ti±\frac{x}{a}\sqrt{\beta}\sqrt{±i}}&\\ &\text{or} & u &=e^{±\beta ti±\frac{x}{a}\sqrt{\frac{\beta}{2}}(1±i)}&\\ &\text{since} & \sqrt i &=±\frac{1}{2}\sqrt2(1+i)&\\ &\text{and} & \sqrt{-i} &=±\frac{1}{2}\sqrt2(1-i).& \end{flalign*} Hence \begin{align*} && u&=e^{-\frac{x}{a}\sqrt\frac{\beta}{2}}\sin\biggl(\beta t-\frac{x}{a}\sqrt\frac{\beta}{2}\biggr),& u&=e^{-\frac{x}{a}\sqrt\frac{\beta}{2}}\cos\biggl(\beta t-\frac{x}{a}\sqrt\frac{\beta}{2}\biggr), \tag{2}&\\ && u&=e^{ \frac{x}{a}\sqrt\frac{\beta}{2}}\sin\biggl(\beta t+\frac{x}{a}\sqrt\frac{\beta}{2}\biggr),& u&=e^{ \frac{x}{a}\sqrt\frac{\beta}{2}}\cos\biggl(\beta t+\frac{x}{a}\sqrt\frac{\beta}{2}\biggr), \tag{3}& \end{align*} are particular solutions of (1). % -----File: 102.png From these we get readily \begin{flalign*} &&u&=\rho_me^{-\frac{x}{a}\sqrt\frac{m\alpha}{2}}\sin\biggl( m\alpha t-\frac{x}{a}\sqrt\frac{m\alpha}{2}+\lambda_m\biggr) &\tag{4}& \intertext{as a solution. (4) reduces to} &&u&=\rho_m\sin(m\alpha t+\lambda_m)\quad \text{when}\quad x=0&&\\ &\text{and to} &u&=\rho_me^{-\frac{x}{a}\sqrt\frac{m\alpha}{2}}\sin\biggl (\lambda_m-\frac{x}{a}\sqrt\frac{m\alpha}{2}\biggr) \quad \text{when}\quad t=0.&& \end{flalign*} If we add a term which satisfies (1) and which is equal to zero when $x=0$ and to $-\rho_me^{-\frac{x}{a}\sqrt\frac{m\alpha}{2}}\sin\biggl(\lambda_m-\dfrac{x}{a}\sqrt{\dfrac{m\alpha}{2}}\biggr)$ when $t=0$ (v.\ Art.~50) we shall have a solution of (1) which is zero when $t=0$ and which is \[ \rho_m\sin(m\alpha t+\lambda_m)\quad \text{when}\quad x=0. \] 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 $u=e^{\beta t+\alpha x}$ is a solution of $D_tu=a^2D_x^2u-b^2u$ if $\beta=a^2\alpha^2-b^2$, and hence that \begin{align*} u&=e^{\beta t \pm\frac{x}{a}\sqrt{b^2+\beta}},\quad u =e^{ \pm\beta ti \pm\frac{x}{a}\sqrt{b^2 \pm\beta i}},\quad u =e^{ \pm\beta ti \pm\frac{x}{a\sqrt2}(p \pm qi)},\\ u&=e^{ \pm\frac{px}{a\sqrt2}}\sin\biggl(\beta t \pm\frac{qx}{a\sqrt2}\biggr),\quad \text{and}\quad u =e^{ \pm\frac{px}{a\sqrt2}}\cos\biggl(\beta t \pm\frac{qx}{a\sqrt2}\biggr), \end{align*} where \[ p=[\sqrt{\beta^2+b^4}+b^2]^\frac{1}{2}\quad \text{and}\quad q=[\sqrt{\beta^2+b^4}-b^2]^\frac{1}{2}, \] are solutions. Hence \[ u=\rho_me^{-\frac{px}{a\sqrt2}}\sin\biggl(\beta t-\frac{qx}{a\sqrt2}+\lambda_m\biggr) \] is a solution. If $\beta=m\alpha$ this last result reduces to $u=\rho_m\sin(m\alpha t+\lambda_m)$ 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. \mypara{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 \textit{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 % -----File: 103.png quantity $Q$ of heat is suddenly generated the point is called an \emph{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 \emph{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 $x=\lambda$ 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 $x=\lambda$ to $x=\lambda+\Delta\lambda$, where $\Delta\lambda$ is any arbitrary length, the temperature of that portion will be suddenly raised to $\dfrac{Q}{\Delta\lambda}$, and we shall have by (8) Art.~49 \[ u=\frac{Q}{2a\sqrt{\pi t}}\frac{1}{\Delta\lambda}\int\limits_\lambda^{\lambda+\Delta\lambda}e^{-\frac{(\lambda-x)^2}{4a^2t}}d\lambda \tag{1} \] 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 $\Delta\lambda$ is made to approach zero we get \[ u=\frac{Q}{2a\sqrt{\pi t}}e^{-\frac{(\lambda-x)^2}{4a^2t}} \tag{2} \] as the solution for the case where we have an instantaneous source at the point $x=\lambda$. It is to be observed that in (2) $u=0$ when $t=0$ and $u=\dfrac{Q}{2a\sqrt{\pi t}}$ when $x=\lambda$ 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 $f(\lambda)d\lambda$ 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 $f(\lambda)d\lambda$ in every element of the bar whose distance to the right of the origin is $\lambda$, and an instantaneous sink of strength $f(\lambda)d\lambda$ in every element of the bar whose distance to the left of the origin is $\lambda$. If we have an instantaneous source at the origin (2) reduces to \[ u=\frac{Q}{2a\sqrt{\pi t}}e^{-\frac{x^2}{4a^2t}} \tag{3} \] % -----File: 104.png For a permanent source of constant strength $Q$ at the origin (3) gives \[ u=\frac{Q}{2a\sqrt\pi}\int\limits_0^t e^{-\frac{x^2}{4a^2(t-\tau)}}(t-\tau)^{-\frac{1}{2}}d\tau \tag{4} \] and for a permanent source of variable strength $f(t)$ \[ u=\frac{1}{2a\sqrt\pi}\int\limits_0^t e^{-\frac{x^2}{4a^2(t-\tau)}}(t-\tau)^{-\frac{1}{2}}f(\tau)d\tau. \tag{5} \] 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 \emph{doublet}. \mypara{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 \emph{doublet} of strength $P$ whose axis is tangent to the line of approach and points from sink to source. A \emph{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 $x=\eta$ and an equal sink at $x=-\eta$ we have \begin{align*} u&=\frac{Q}{2a\sqrt{\pi t}}(e^{-\frac{(\eta-x)^2}{4a^2t}}-e^{-\frac{(\eta+x)^2}{4a^2t}}),\\ \intertext{or if $2\eta Q=P$,} u&=\frac{P}{4a\eta\sqrt{\pi t}}e^{-\frac{(\eta^2+x^2)}{4a^2t}} (e^{\frac{\eta x}{2a^2t}}-e^{-\frac{\eta x}{2a^2t}})\\ &=\frac{P}{2a\eta\sqrt{\pi t}}e^{-\frac{(\eta^2+x^2)}{4a^2t}}\sinh\frac{\eta x}{2a^2t}. \end{align*} If $\eta$ is made to approach zero \[ \limit\left[\frac{1}{\eta}\sinh\frac{\eta x}{2a^2t}\right]=\frac{x}{2a^2t}, \] \begin{flalign*} &\text{and} & u=\frac{Px}{4a^3\sqrt{\pi t^3}}e^{-\frac{x^2}{4a^2t}} &\tag{1}& \end{flalign*} 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 $x=\lambda$ we have \[ u=\frac{P(x-\lambda)}{4a^3\sqrt{\pi t^3}}e^{-\frac{(x-\lambda)^2}{4a^2t}}. \tag{2} \] % -----File: 105.png For a permanent doublet of constant strength $P$ placed at the origin we have \begin{flalign*} &&u&=\frac{Px}{4a^3\sqrt\pi}\int\limits_0^t e^{-\frac{x^2}{4a^2(t-\tau)}}(t-\tau)^{-\frac{3}{2}}d\tau;& \tag{3}&\\ \intertext{and for a permanent doublet of variable strength $f(t)$} &&u&=\frac{x}{4a^3\sqrt\pi}\int\limits_0^te^{-\frac{x^2}{4a^2(t-\tau)}}(t-\tau)^{-\frac{3}{2}}f(\tau)d\tau, & \tag{4}&\\ &\text{or} & u&=\frac{1}{a^2\sqrt\pi}\int\limits_\frac{x}{2a\sqrt t}^\infty e^{-\beta^2}f\left(t-\frac{x^2}{4a^2\beta^2}\right)d\beta & \tag{5}&\\ \intertext{if $x>0$, and} &&u&=\frac{1}{a^2\sqrt\pi}\int\limits_\frac{x}{2a\sqrt t}^{-\infty}e^{-\beta^2}f\left(t-\frac{x^2}{4a^2\beta^2}\right)d\beta & \tag{6}& \end{flalign*} if $x<0$, if we let $\beta=\dfrac{x}{2a\sqrt{t-\tau}}$. From (5) and (6) we see readily that $u=0$ when $t=0$ and that $u=\dfrac{f(t)}{2a^2}$ when $x=0$ if we approach the origin from the right and that $u=-\dfrac{f(t)}{2a^2}$ when $x=0$ 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 $2a^2b$ placed at the origin. To solve the problem treated in Art.~51 we have only to suppose a permanent doublet of strength $2a^2F(t)$ placed at $x=0$ and from (5) we get at once (10) Art.~51. \EXAMPLE{} Show that if $D_tu=a^2D_x^2u-b^2u$ and an instantaneous source of strength $Q$ is placed at $x=\lambda$ \[\phantom{v.\ Art.~51, Ex.~5.} u=\frac{Q}{2a\sqrt{\pi t}}e^{-b^2t-\frac{(\lambda-x)^2}{4a^2t}} \tag*{v.\ Art.~51, Ex.~5.} \] Show that if an instantaneous doublet of strength $P$ is placed at the point $x=0$ \[ u=\frac{Px}{4a^3\sqrt{\pi t^3}}e^{-b^2t-\frac{x^2}{4a^2t}}. \] % -----File: 106.png If a permanent doublet of strength $f(t)$ is placed at $x=0$ \begin{align*} u&=\frac{x}{4a^3\sqrt\pi}\int\limits_0^t e^{-b^2(t-\tau)-\frac{x^2}{4a^2(t-\tau)}}(t-\tau)^{-\frac{3}{2}}f(\tau)d\tau\\ &=\frac{1}{a^2\sqrt\pi}\int\limits_{\frac{x}{2a\sqrt t}}^{±\infty} e^{-\beta^2-\frac{b^2x^2}{4a^2\beta^2}}f\left(t-\frac{x^2}{4a^2\beta^2}\right)d\beta, \end{align*} whence $u=0$ when $t=0$ and $x>0$ or $x<0$ and $u=±\dfrac{f(t)}{2a^2}$ when $x=0$. Hence if we place at $x=0$ a permanent doublet of strength $2a^2F(t)$ 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 \[ u=\frac{2c}{\sqrt\pi}\int\limits_{\frac{x}{2a\sqrt t}}^{\infty} e^{-\beta^2-\frac{b^2x^2}{4a^2\beta^2}}d\beta. \] \mypara{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 \[ D_t^2y=a^2D_x^2 y \tag{1} \] [v.\ \smallromr{VIII} Art.~1] subject to the conditions \begin{alignat*}{2} y&=f(x) & \quad\text{when}\quad & t=0 \tag{2}\\ D_ty&=0 & \quad\text{``}\quad\quad & t=0. \tag{3} \end{alignat*} As in Art.~8 we find \[ y=\cos\alpha(x±at)\quad \text{and}\quad y=\sin\alpha(x±at) \] as particular solutions of (1). From these we must build up a value that will reduce to \[ f(x)=\frac{1}{\pi} \int\limits_0^\infty d\alpha\int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x).d\lambda \tag{4} \] % -----File: 107.png when $t=0$ and will at the same time satisfy (3). \begin{flalign*} &&y&=\cos\alpha\lambda\cos\alpha(x+at)+\sin\alpha\lambda\sin\alpha(x+at)&\phantom{or}\\ &\text{or}& y&=\cos\alpha(\lambda-x-at)& \end{flalign*} is a solution of (1). \begin{flalign*} &\indent\text{Hence}& y=\frac{1}{\pi}\int\limits_0^\infty d\alpha\int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x-at).d\lambda & \tag{5}&\hspace{2.5em} \end{flalign*} is also a solution of (1). (5) reduces to $y=f(x)$ when $t=0$ but it gives \[ D_ty=\frac{a}{\pi}\int\limits_0^\infty\alpha d\alpha \int\limits_{-\infty}^\infty f(\lambda)\sin\alpha(\lambda - x).d\lambda \] when $t=0$ and consequently does not satisfy equation (3). If in forming (5) we use $\cos\alpha(x-at)$ and $\sin\alpha(x-at)$ instead of $\cos\alpha(x+at)$ and $\sin\alpha(x+at)$ we get \[ y=\frac{1}{\pi}\int\limits_0^\infty d\alpha \int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x+at).d\lambda \tag{6} \] which is a solution of (1), and reduces to $y=f(x)$ when $t=0$, but it gives \[ D_ty=-\frac{a}{\pi}\int\limits_0^\infty\alpha d\alpha \int\limits_{-\infty}^\infty f(\lambda)\sin\alpha(\lambda-x).d\lambda \] 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 \begin{align*} y=\frac{1}{2}\Bigg[\frac{1}{\pi}&\int\limits_0^\infty d\alpha \int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x-at).d\lambda\\ +&\frac{1}{\pi}\int\limits_0^\infty d\alpha \int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x+at).d\lambda\Bigg], \tag{7} \end{align*} 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 $z=x+at$ \begin{align*} &\frac{1}{\pi}\int\limits_0^\infty d\alpha \int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x-at).d\lambda\\ =&\frac{1}{\pi}\int\limits_0^\infty d\alpha \int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-z).d\lambda =f(z)=f(x+at); \end{align*} % -----File: 108.png and in like manner we can show that \[ \frac{1}{\pi}\int\limits_0^\infty d\alpha\int\limits_{-\infty}^\infty f(\lambda)\cos\alpha(\lambda-x+at).d\lambda=f(x-at). \] Hence our solution becomes \[ y=\frac{1}{2}[f(x+at)+f(x-at)]. \tag{8} \] 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. \mypara{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 \[ D_t^2y=a^2D_x^2y \tag{1} \] subject to the conditions \begin{alignat}{3} y& = 0 & \text{when\quad} & t& = 0 &\tag{2}\\ D_ty& = F(x) & \text{``\qquad} & t& = 0. &\tag{3} \end{alignat} We get by the process used in Art.~55 \begin{align*} y&=\frac{1}{2\pi a}\int\limits_0^\infty d\alpha\int\limits_{-\infty}^\infty F(\lambda) \left[\frac{\sin\alpha(\lambda-x+at)}{\alpha} -\frac{\sin\alpha(\lambda-x-at)}{\alpha}\right]d\lambda\\ &=\frac{1}{2\pi a}\int\limits_{-\infty}^\infty F(\lambda)d\lambda\int\limits_0^\infty \left[\frac{\sin\alpha(\lambda-x+at)}{\alpha} -\frac{\sin\alpha(\lambda-x-at)}{\alpha}\right]d\alpha; \end{align*} but \qquad\qquad $\displaystyle\int\limits_0^\infty\frac{\sin\alpha(\lambda-x+at)}{\alpha}d\alpha -\int\limits_0^\infty\frac{\sin\alpha(\lambda-x-at)}{\alpha}d\alpha=\pi$\\ if $x-at < \lambda < x+at$, and is equal to zero for all other values of $\lambda$; since \begin{alignat*}{2} \int\limits_0^\infty\frac{\sin mx}{x}dx &= \phantom{-}\frac{\pi}{2} &\quad\text{if}\quad & m>0\\ &= - \frac{\pi}{2} &\quad\text{if}\quad & m<0\\ &= \phantom{-}0 &\quad\text{if}\quad & m=0. \end{alignat*} v.\ Int.\ Cal.\ Art.~92~(3). \begin{flalign*} &\indent\text{Hence} & y=\frac{1}{2a}\int\limits_{x-at}^{x+at}F(\lambda)d\lambda &\tag{4}&\hspace{2.5em} \end{flalign*} is our required solution. % -----File: 109.png \EXAMPLE{S} 1.\quad If the string is initially distorted and starts with initial velocity so that $y=f(x)$ and $D_{t}y=F(x)$ when $t=0$ \[ y=\frac{1}{2}[f(x+at)+f(x-at)]+\frac{1}{2a}\int\limits_{x-at}^{x+at}F(\lambda)d\lambda. \] 2.\quad 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 $\dfrac{b}{2a}$ and a length equal to $\displaystyle\frac{c}{2a}\sqrt{4a^2+b^2}$. Of course a wave having a front of the same length with a slope equal to $-\dfrac{b}{2a}$ will be seen to run to the right along the string, and the effect of the two waves will be to lift the string bodily and permanently to a distance $\dfrac{bc}{2a}$ above its original position. \mypara{57.} We shall now take up a few examples of the use of \textit{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. \begin{flalign*} &\indent\text{Since} & f(x)& =\sum_{m=1}^{m=\infty}(a_m\sin mx)&&\phantom{\indent Since}\\[-7ex] \end{flalign*} \begin{flalign*} &\text{where} & a_m =& \frac{2}{\pi} \int\limits_0^\pi f(\alpha)\sin m\alpha.d\alpha&&\phantom{where}\\[-8ex] \end{flalign*} \begin{flalign*} &\text{we have} &u=\frac{2}{\pi}\sum_{m=1}^{m=\infty}\bigg[e^{-my}&\sin mx\int\limits_0^\pi f(\alpha)\sin m\alpha.d\alpha\bigg]. & \tag{1}&\hspace{1.5em} \end{flalign*} If the breadth of the plate is $a$ instead of $\pi$ \[ u=\frac{2}{a}\sum_{m=1}^{m=\infty}\bigg[e^{-\frac{m\pi y}{a}}\sin\frac{m\pi x}{a}\int\limits_0^a f(\lambda)\sin\frac{m\pi\lambda}{a}d\lambda\bigg]. \tag{2} \] \mypara{58.} If the temperature of the base is unity and the breadth of the plate is $\pi$ the solution is, as we have seen in Art.~7, \[ u=\frac{4}{\pi}\bigg[e^{-y}\sin x+\frac{1}{3}e^{-3y}\sin 3x+\frac{1}{5}e^{-5y}\sin 5x+\cdots\bigg]. \tag{1} \] This series can be summed without difficulty. We have the development \[ \log(1+z)=\frac{z}{1}-\frac{z^2}{2}+\frac{z^3}{3}-\frac{z^4}{4}+\cdots \] if the modulus of $z$ is less than 1. Int.\ Cal.\ Art.~221~(4). % -----File: 110.png \begin{flalign*} & \text{Hence} & \log(1 - z) = -\frac{z}{1} -\frac{ z^2}{2} -\frac{ z^3}{3} -\frac{ z^4}{4} - \cdots && \phantom{Hence} \end{flalign*} if $\operatorname{mod.} z<1$. \begin{flalign*} & \text{and} & \frac{1}{2}[\log(1+z)-\log(1-z)]= \frac{ z}{1} +\frac{ z^3}{3} + \frac{ z^5}{5} + \cdots && \,\tag{2} \end{flalign*} if $\operatorname{mod.} z <1$. \\ But \begin{flalign*} \log(1+z) &=\log[1+r(\cos \phi + i \sin \phi)] \\ &=\frac{1}{2} \log[(1+r \cos \phi)^2 + (r \sin \phi)^2] + i \tan^{-1} \frac{r \sin \phi }{ 1+r \cos \phi} \\ &=\frac{1}{2} \log{(1 + 2r \cos \phi + r^2)} + i \tan^{-1}\frac{ r \sin \phi }{1+r \cos \phi}, \intertext{and} \log(1-z) &= \frac{1}{2} \log(1-2r \cos\phi + r^2) - i \tan^{-1} \frac{r \sin \phi}{ 1-r \cos\phi}, \end{flalign*} [Int.\ Cal.\ Art.~33~(2)],\\ and (2) becomes \begin{multline*} \frac{1}{2}\left[\frac{1}{2} \log\frac{ 1+2r \cos\phi+r^2 }{ 1-2r \cos\phi+r^2} + i \tan^{-1} \frac{2r \sin \phi }{1-r^2} \right]\\ = \frac{r(\cos\phi+i \sin\phi)}{1} +\frac{ r^3(\cos 3\phi + i\sin 3\phi) }{3} + \cdots \tag{3} \end{multline*} From (3) we get two equations \begin{gather*} \frac{1}{4}\log {\frac{1+2r \cos \phi+r^2}{ 1-2r \cos \phi+r^2}} = \frac{r\cos \phi}{1} + \frac{r^3 \cos 3 \phi}{3} + \frac{r^5 \cos 5 \phi}{5} + \cdots \tag{4}\\ \frac{1}{2} \tan^{-1}\frac{ 2r \sin\phi}{ 1-r^2} = \frac{r \sin\phi}{1} + \frac{r^3 \sin3\phi}{3} + \frac{r^5 \sin5\phi}{5}+ \cdots \tag{5} \end{gather*} both valid for all values of $\phi$ provided $r < 1$. $e^{-y}$ is less than $1$ if $y$ is positive. Hence from (5) \begin{align*} \frac{e^{-y }\sin x}{1} + \frac{e^{-3y }\sin 3x}{3}+ \frac{e^{-5y }\sin 5x}{5} &+ \cdots = \frac{1}{2} \tan^{-1}\frac{ 2e^{-y }\sin x }{ 1-e^{-2y}}\\ &= \frac{1}{2} \tan^{-1}\frac{ 2\sin x }{e^y-e^{-y} }= \frac{1}{2}\tan^{-1}\frac{ \sin x }{ \sinh y}, \end{align*} and (1) may be written\\ \[ u = \frac{2}{\pi}\tan^{-1}\frac{ \sin x }{ \sinh y}. \tag{6} \] % -----File: 111.png If we replace $r$ by $e^{-y}$ and $\phi$ by $x$ in\\[-4ex] \begin{flalign*} & & \log[1 + r(&\cos \phi + i \sin \phi)]&\\ &\text{it becomes} & \log[1 + e^{-y}&\cos x + i e^{-y} \sin x]\hfil\phantom{it becomes}&\\ &\text{or} & \log[1 + &\cos z + i \sin z]&\\[-5ex] \end{flalign*} v.\ Int.\ Cal.\ Art.~35 (3) and (4)\\ a function of $z$ as a whole; and\\[-4ex] \begin{flalign*} & & \log [1 - r(&\cos \phi + i \sin \phi)]&\phantom{becomes}\\ &\text{becomes} & \log (1 - &\cos z - i\sin z);&\\[-5ex] \end{flalign*} hence by Int.\ Cal.\ Arts.~212 and 213,\\[-4ex] \begin{flalign*} &&\frac{1}{4} \log &\frac{1 + 2e^{-y}\cos x + e^{-2y}}{1 - 2e^{-y}\cos x + e^{-2y}}\quad \text{and}\quad \frac{1}{2}\tan^{-1}\frac{2e^{-y}\sin x}{1 - e^{-2y}}&\\ &\text{or}&& \frac{1}{4}\log\dfrac{\cosh y + \cos x}{\cosh y -\cos x}\quad \text{and}\quad \frac{1}{2}\tan^{-1}\frac{\sin x}{\sinh y}& \end{flalign*} are conjugate functions, and \[ u_1= \frac{1}{\pi}\log\dfrac{\cosh y + \cos x}{\cosh y - \cos x} \tag{7} \] 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 \begin{flalign*} & & \frac{2}{\pi}\tan^{-1} &\frac{\sin x}{\sinh y} = a&&\\ &\text{or} & \dfrac{\sin x}{\sinh y}& = \tan\dfrac{a \pi}{2}& \tag{8}& \end{flalign*} and the lines of flow by \begin{flalign*} && \frac{1}{\pi}&\log\dfrac{\cosh y + \cos x}{\cosh y - \cos x} = b,&&\\ &\text{or}&& \dfrac{\cosh y + \cos x}{\cosh y - \cos x} = e^{\pi b}.&\tag{9}& \end{flalign*} \EXAMPLE{S} 1.\quad If $D_x^2u+D_y^2u=0$, and $u=1$ when $y=0$, and $u=0$ when $x=0$ and when $x=a$, \begin{align*} u &= \frac{4}{\pi}\bigg[e^{-\frac{\pi y}{a}} \sin\frac{\pi x}{a} + \frac{1}{3}e^{- \frac{3 \pi y}{a}} \sin\frac{3\pi x}{a} + \frac{1}{5}e^{-\frac{5 \pi y}{a}} \sin \frac{5\pi x}{a} + \cdots \bigg] \\ \label{err111} &= \frac{2}{\pi}\tan^{-1} \frac{\sin \dfrac{\pi x}{a}}{\sinh \dfrac{\pi y}{a}}. \end{align*} % -----File: 112.png 2.\quad If $u = \phi (x)$ when $y = 0$, $u =f(y)$ when $x = 0$, and $u = F(y)$ when $x = a$ \begin{align*} &u= \frac{2}{a} \sum_{m=1}^{m=\infty} e^{-\frac{m\pi y}{a}} \sin \frac{m\pi x}{a} \int \limits_0^a \phi(\lambda) \sin \frac{m\pi\lambda}{a} d\lambda \\ & + \frac{1}{2a} \sin \frac{\pi x}{a} \int \limits_0^\infty \ \biggl[ \ \frac{1}{ \cosh \dfrac{\pi}{a} (\lambda-y) - \cos \dfrac{\pi x}{a} } - \frac{1}{ \cosh \dfrac{\pi}{a} (\lambda+y) - \cos \dfrac{\pi x}{a} } \ \biggr] \ f(\lambda)\, d\lambda \\ & + \frac{1}{2a} \sin \frac{\pi x}{a} \int \limits_0^\infty \ \biggl[ \ \frac{1}{ \cosh \dfrac{\pi}{a} (\lambda-y) + \cos \dfrac{\pi x}{a} } - \frac{1}{ \cosh \dfrac{\pi}{a} (\lambda+y) + \cos \dfrac{\pi x}{a} } \ \biggr] \ F(\lambda)\, d\lambda \end{align*} v.\ Art.~48, Exs.~4, 5, and 6. \mypara{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:--- \begin{alignat*}{3} &&\llap{$D_x^2V$} &\rlap{$+\;D_y^2V = 0.$} \tag{1}\\ &V = 0 && \text{when}\quad && x = 0. \tag{2}\\ &V = 0 && \quad\text{``} && x = a. \tag{3}\\ &V = 0 && \quad\text{``} && y = b. \tag{4}\\ &V = f(x) && \quad\text{``} && y = 0. \tag{5} \end{alignat*} Working as in Art.~48 we get \[ \frac{ \sinh \dfrac{m\pi }{a} (b-y) } { \sinh \dfrac{m\pi b}{a} } \sin \frac{m\pi x}{a} \] as a value of $V$ which satisfies equations (1), (2), (3), and (4) if $m$ is an integer. Therefore \[ V = \frac{2}{a} \sum_{m=1}^{m=\infty} \ \biggl[ \ \frac{ \sinh \dfrac{m\pi }{a} (b-y) } { \sinh \dfrac{m\pi b}{a} } \sin \frac{m\pi x}{a} \int \limits_0^a f(\lambda) \sin \frac{m\pi\lambda}{a} d\lambda \ \biggr] \tag{6} \] is our required solution. % -----File: 113.png \EXAMPLE{S} 1.\quad If $f(x) = 1$ Eq.~(6) Art.~59 reduces to \[ \begin{split} V=\frac{4}{\pi} \Biggl [ &\frac{\sinh \dfrac{\pi}{a} (b-y) }{ \sinh \dfrac{\pi b}{a }}\sin \frac{\pi x}{a} + \frac{1}{3}\frac{\sinh \dfrac{3\pi}{a} (b-y) }{ \sinh \dfrac{3\pi b}{a} }\sin \frac{3\pi x}{a} \Biggr .\\ &+\Biggl . \frac{1}{5}\frac{\sinh \dfrac{5\pi}{a} (b-y) } {\sinh \dfrac{5\pi b}{a}} \sin \frac{5\pi x}{a}+ \cdots \Biggr ] \end{split} \] 2.\quad If $V=0$ when $x = 0$, $V=0$ when $x = a$, $V= 0$ when $y = 0$, and $V=F(x)$ when $y = b$, then \[ V = \frac{2}{a} \sum_{m=1}^{m=\infty}\Biggl [\frac{\sinh \dfrac{m\pi y }{ a}} {\sinh \dfrac{m\pi b}{a}} \sin \frac{m\pi x}{a }\int \limits_0^a F(\lambda)\sin \frac{m\pi\lambda}{a} d\lambda \Biggr ]. \] 3.\quad If $F(x) = 1$ the answer of Ex.~2 reduces to \[ V = \frac{4}{\pi}\Biggl [\frac{\sinh \dfrac{\pi y}{a }}{\sinh \dfrac{\pi b}{a}} \sin \frac{\pi x}{a }+ \frac{1}{3} \frac{\sinh \dfrac{3\pi y}{a}}{ \sinh \dfrac{3\pi b}{a}} \sin \frac{3\pi x}{a} + \frac{1}{5} \frac{\sinh \dfrac{5\pi y}{a}}{ \sinh \dfrac{5\pi b}{a}} \sin \frac{5\pi x}{a} + \cdots \Biggr ]. \] 4.\quad 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 \[ \begin{split} V = \frac{2}{a} \sum_{m=1}^{m=\infty} \Biggl [\sin \frac{m\pi x }{ a} \Biggl ( \frac{\sinh \dfrac{m\pi}{a} (b-y)}{ \sinh \dfrac{m\pi b}{a}} \int \limits_0^a f(\lambda)\sin \frac{m\pi\lambda}{a} d\lambda \Biggr . \Biggr .\\ +\Biggl . \Biggl . \frac{\sinh\dfrac{ m\pi y }{ a} }{ \sinh \dfrac{m\pi b}{a }}\int \limits_0^a F(\lambda)\sin\frac{ m\pi\lambda}{a} d\lambda \Biggr ) \Biggr ]. \end{split} \] 5.\quad If $f(x) = F(x)$ the answer of Ex.~4 reduces to \[ V = \frac{2}{a }\sum_{m=1}^{m=\infty} \Biggl [ \frac{\cosh \dfrac{m\pi}{a } \Bigl (\dfrac{b}{2}-y \Bigr) }{\cosh \dfrac{m\pi b}{2a}} \sin \frac{m\pi x}{a} \int \limits_0^a f(\lambda)\sin \frac{m\pi \lambda}{a} d\lambda \Biggr ]. \] % -----File: 114.png 6.\quad If $f(x) = F(x) = 1$ the answer of Ex.~5 reduces to \[\begin{split} V = \frac{4}{\pi}\Biggl [ &\frac{\cosh\dfrac{ \pi}{a} \Bigl ( \dfrac{b}{2}-y \Bigr ) }{\cosh \dfrac{\pi b}{2a}} \sin\frac{ \pi x}{a} + \frac{1}{3}\dfrac{\cosh\dfrac{3 \pi}{a} \Bigl ( \dfrac{b}{2}-y \Bigr ) }{\cosh \dfrac{3\pi b}{2a}} \sin\frac{ 3\pi x}{a} \Biggr .\\ &+\Biggl . \frac{1}{5}\dfrac{\cosh\dfrac{5 \pi}{a} \Bigl ( \dfrac{b}{2}-y \Bigr ) }{\cosh \dfrac{5\pi b}{2a}} \sin\frac{ 5\pi x}{a} + \cdots \Biggr ]. \end{split} \] 7.\quad If $V=f(x)$ when $y = 0$, $V=F(x)$ when $y = b$, $V=\phi (y)$ when $x = 0$, and $V=\chi (y)$ when $x=a$, then \[ \begin{alignedat}{2} V &= \frac{2}{a} \sum_{m=1}^{m=\infty}\Biggl [&&\sin \frac{m\pi x}{a} \Biggl (\dfrac{\sinh \dfrac{m\pi}{a}(b-y) }{ \sinh \dfrac{m\pi b}{a}} \int\limits_0^a f(\lambda)\sin \frac{m\pi\lambda}{a} d\lambda \Biggr . \Biggr .\\ &&&+ \Biggl . \Biggl. \dfrac{ \sinh \dfrac{m\pi y}{a }} {\sinh \dfrac{m\pi b}{a}} \int\limits_0^a F(\lambda)\sin \frac{m\pi\lambda}{a} d\lambda\Biggr) \Biggr ]\\ &+ \frac{2}{b}\sum_{m=1}^{m=\infty}\Biggl [&&\sin \frac{m\pi y }{b} \Biggl (\frac{\sinh \dfrac{m\pi}{b}(a-x) }{ \sinh \dfrac{m\pi a}{b}} \int\limits_0^b \phi(\lambda)\sin \frac{m\pi\lambda}{b} d\lambda \Biggr . \Biggr .\\ &&&+ \Biggl . \Biggl . \frac{\sinh \dfrac{m\pi x}{b} }{ \sinh \dfrac{m\pi a}{b}} \int\limits_0^b \chi(\lambda)\sin \frac{m\pi\lambda}{b} d\lambda \Biggr ) \Biggr ]. \end{alignedat} \] 8.\quad If $f(x)=\phi(y)=0$ and $F(x)=\chi(y)=1$ the answer of Ex.~7 may be reduced to \begin{align*} V = \frac{2}{\pi} \Biggl [ &\frac{ \pi y}{2b} - \frac{\sinh \dfrac{\pi}{b} \left (\dfrac{a}{2}-x \right ) }{ \sinh \dfrac{\pi a}{2b }}\sin \frac{\pi y}{b} + \frac{1}{2}\frac{ \cosh \dfrac{2\pi}{b} \left (\dfrac{a}{2}-x \right ) }{\cosh \dfrac{2\pi a}{2b}} \sin \frac{2\pi y}{b} \Biggr .\\ &- \Biggl . \frac{1}{3} \frac{\sinh \dfrac{3\pi}{b} \left (\dfrac{a}{2}-x \right ) }{ \sinh \dfrac{3\pi a}{2b}} \sin \frac{3\pi y}{b} + \frac{1}{4} \frac{\cosh \dfrac{4\pi}{b} \left (\dfrac{a}{2}-x \right ) }{ \cosh \dfrac{4\pi a}{2b}} \sin \frac{4\pi y}{b} - \cdots \Biggr]. \end{align*} % -----File: 115.png 9.\quad 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.\\ \phantom{kindly go to the right hand margin}\hfill \textit{Ans.}, (1) 25°; (2) 50°; (3) 50°. \mypara{60.} Let us pass on to the consideration of the flow of heat in one dimension. 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 \[ D_tu = a^2D_x^2u \tag{1} \] subject to the conditions \begin{alignat*}{3} u& = 0 && \text{when}\quad & x & = 0 \tag{2}\\ u& = 0 && \quad\text{``} & x & = c \tag{3}\\ u& =f(x)&& \quad\text{``} & t & = 0. \tag{4} \end{alignat*} In Art.~49 we have found \begin{flalign*} &&u &= e^{-a^2\alpha^2t}\sin\alpha x&\\ &\text{and} & u&=e^{-a^2\alpha^2t}\cos\alpha x&\\ \end{flalign*} as particular solutions of (1). $u=e^{-a^2\alpha^2t}\sin\alpha x$ satisfies (2) whatever value is given to $\alpha$. It satisfies (3) if $\alpha=\dfrac{m\pi}{c}$ provided $m$ is an integer. Let us try to build a value of $u$ out of terms of the form $Ae^{-\frac{a^2m^2\pi^2t}{c^2}}\sin\dfrac{m\pi x}{c}$ which shall satisfy (4). We have \begin{gather*} f(x)=\frac{2}{c}\sum_{m=1}^{m=\infty}\Bigl[\sin\frac{m\pi x}{c} \int\limits_0^c f(\lambda)\sin\frac{m\pi\lambda}{c}d\lambda\Bigr]. \tag{5}\\ u=\frac{2}{c}\sum_{m=1}^{m=\infty}\Bigl[e^{-\frac{m^2a^2\pi^2t}{c^2}}\sin\frac{m\pi x}{c} \int\limits_0^c f(\lambda)\sin\frac{m\pi\lambda}{c}d\lambda\Bigr], \tag{6} \end{gather*} reduces to (5) when $t=0$ and is our required solution. % -----File: 116.png \EXAMPLE{S} 1.\quad If $f(\lambda)=b$, a constant, (6) Art.~60 reduces to \[ u=\frac{4b}{\pi}\Bigl[e^{-\frac{ a^2\pi^2t}{c^2}}\sin\frac{ \pi x}{c} +\frac{1}{3}e^{-\frac{ 9a^2\pi^2t}{c^2}}\sin\frac{3\pi x}{c} +\frac{1}{5}e^{-\frac{25\pi^2a^2t}{c^2}}\sin\frac{5\pi x}{c} +\cdots\Bigr]. \] 2.\quad 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 $a^2=0.185$ in C.G.S. units. \hfill \textit{Ans.}, 10°.3.\\ 3.\quad 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.\\ \phantom{kindly go to the right hand margin}\hfill \textit{Ans.}, 22°.8; 15°.1; 17°.2.\\ 4.\quad 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 temperatures 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. \hfill \textit{Ans.}, 22°.8; 15°.6; 16°.7. \mypara{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 \[ D_tu = a^2D_x^2u \] subject to the conditions \begin{alignat*}{3} D_xu& = 0 &&\text{when}\quad & x&=0\\ D_xu& = 0 &&\text{\quad``} & x&=c\\ u& = f(x)&&\text{\quad``} & t&=0. \end{alignat*} We have only to use the particular solution \begin{flalign*} & & u& = e^{-a^2\alpha^2t}\cos\alpha x&\\ &\text{as we used}& u& = e^{-a^2\alpha^2t}\sin\alpha x&\phantom{as we used} \end{flalign*} in Art.~60. We get \[ u=\frac{2}{c}\biggl[\frac{1}{2}\int\limits_0^c f(\lambda)d\lambda +\sum_{m=1}^{m=\infty} \Bigl(e^{-\frac{m^2a^2\pi^2t}{c^2}}\cos\frac{m\pi x}{c} \int\limits_0^c f(\lambda)\cos\frac{m\pi\lambda}{c}d\lambda\Bigr)\biggr]. \tag{1} \] % -----File: 117.png \EXAMPLE{S} 1.\quad 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. \hfill \textit{Ans.}, 33°.3; 33°.3.\\ 2.\quad Solve example 3 Art.~60 on the hypothesis just stated, getting in addition the temperatures of points on the outer surfaces.\\ \phantom{kindly go to the right hand margin}\hfill \textit{Ans.}, 50°; 33°.9; 66°.1; 27°.2; 72°.8.\\ 3.\quad 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.\\ \phantom{kindly go to the right hand margin}\hfill \textit{Ans.}, 50°; 39°.7; 60°.3; 35°.5; 64°.5.\\ 4.\quad Show that if $u = 0$ when $x = 0$, $D_xu = 0$ when $x = c$, and $u=f(x)$ when $t = 0$, \[ u = \frac{2}{c}\sum_{m=0}^{m=\infty} \biggl ( e^{-\frac{(2m+1)^2a^2\pi^2t}{4c^2}} \sin\frac{(2m+1)\pi x}{2c} \int\limits_0^c f(\lambda) \sin \frac{(2m+1)\pi\lambda}{2c} d\lambda \biggr ). \] \textit{Suggestion}: Assume $u = 0$ when $x = 2c$ and $f(2c - x) =f(x)$, and see (6) Art.~60. \mypara{62.} If the temperature of the right-hand face of the slab considered in Art.~60 is a constant $\gamma$ instead of zero we have only to add to the second member of (6) Art.~60 a term $u_1$ which shall satisfy the conditions \begin{alignat*}{3} &&\llap{$D_tu_1\:$} &\rlap{$= a^2D_x^2u_1$}\tag{1}\\ u_1 &= 0\quad && \text{when} \quad & x &= 0 \tag{2}\\ u_1 &= 0 && \quad \text{``} & t &= 0 \tag{3}\\ u_1 &= \gamma && \quad \text{``} & x &= c. \tag{4} \end{alignat*} $u_1 = \dfrac{\gamma x}{c}$ obviously satisfies (1), (2), and (4); to make it satisfy (3) as well we must add a term $u_2$ which shall be equal to zero when $x = 0$ and when $x = c$ and to $-\dfrac{\gamma x}{c}$ when $t = 0$, while always satisfying (1). It is given immediately by (6) Art.~60 and is \begin{gather*} u_2=-\frac{2\gamma}{c^2} \sum_{m=1}^{m=\infty} \biggl (e^{-\frac{m^2a^2\pi^2t}{c^2}} \sin \frac{m\pi x}{c} \int\limits_0^c \lambda \sin\frac{m\pi\lambda}{c} d\lambda \biggr ). \tag{5}\\ \int\limits_0^c \lambda \sin\frac{m\pi\lambda}{c} d\lambda =-\frac{c^2}{m\pi}\cos m\pi = (- 1)^{m+1} \frac{c^2}{m\pi}, \end{gather*} % -----File: 118.png \begin{flalign*} &\text{and} &&u_2=\frac{2\gamma}{\pi} \sum_{m=1}^{m=\infty} \left ( \frac{(-1)^m}{m }e^{-\frac{m^2a^2\pi^2t}{c^2}} \sin \frac{m\pi x}{c} \right ). && \tag{6}\\[-5ex] \end{flalign*} \begin{flalign*} &\indent \text{Hence} && u_1 = \gamma \biggl [ \frac{x}{c} + \frac{2}{\pi} \sum_{m=1}^{m=\infty} \left ( \frac{(-1)^m}{m }e^{-\frac{m^2a^2\pi^2t}{c^2}} \sin \frac{m\pi x}{c} \right ) \biggr ]. &&\hspace{2em} \tag{7} \end{flalign*} If the left-hand face of the slab considered in Art.~60 is to be kept at a constant temperature $\beta$ and the right-hand face at the temperature zero we can get the term $u_3$ which must be added to the second member of (6) Art.~60 by replacing $\gamma$ by $\beta$ and $x$ by $c-x$ in (7). We then have \[ u_3= \beta \biggl [\frac{c-x}{c} - \frac{2}{\pi} \sum_{m=1}^{m=\infty} \left ( \frac{1}{m}e^{-\frac{m^2a^2\pi^2t}{c^2}} \sin \frac{m\pi x}{c} \right ) \biggr ]. \tag{8} \] \EXAMPLE{S} 1.\quad Show that if $u = \beta$ when $x = 0$, $u = \gamma$ when $x = c$, and $u=f(x)$ when $t = 0$ \[ \begin{split} u=\beta + (\gamma - \beta) \biggl [ \frac{x}{c} + \frac{2}{\pi} \sum_{m=1}^{m=\infty} \left ( \frac{(-1)^m}{m }e^{-\frac{m^2\pi^2a^2t}{c^2}} \sin \frac{m\pi x}{c} \right ) \biggr ]\\ +\frac{2}{c} \sum_{m=1}^{m=\infty} \biggl ( e^{-\frac{m^2a^2\pi^2t}{c^2}} \sin \frac{m\pi x}{c} \int\limits_0^c [f(\lambda)-\beta] \sin \frac{m\pi\lambda}{c} d\lambda \biggr ). \end{split} \] 2.\quad Show that if $u = \beta$ when $x = 0$, $u = 0$ when $t = 0$, and $D_xu = 0$ when $x = c$ \begin{align*} &u =\beta \biggl [1 - \frac{4}{\pi}\sum_{m=0}^{m=\infty} \left (\frac{1}{2m+1} e^{-\frac{(2m+1)^2a^2\pi^2t}{4c^2}}\sin \frac{(2m+1)\pi x}{2c} \right ) \biggr ]\\ &=\beta \left [1 -\frac{4}{\pi} \left (e^{-\frac{ a^2\pi^2t}{4c^2}} \!\sin \frac{ \pi x}{2c} + \frac{1}{3}e^{-\frac{ 9a^2\pi^2t}{4c^2}} \!\sin \frac{3\pi x}{2c} + \frac{1}{5}e^{-\frac{25a^2\pi^2t}{4c^2}} \!\sin \frac{5\pi x}{2c} + \cdots \!\right ) \right ]. \end{align*} \mypara{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. % -----File: 119.png Let $\phi(x,t)$ be a function of $x$ and $t$ which shall be zero if $t$ is less than zero and shall be equal to \[ \frac{x}{c}+\frac{2}{\pi}\sum_{m=1}^{m=\infty} \left(\frac{(-1)^m}{m}e^{-\frac{m^2a^2\pi^2t}{c^2}}\sin\frac{m\pi x}{c}\right) \] [v.\ (7) Art.~62] if $t$ is equal to or greater than zero. So that \begin{alignat*}{4} \phi(x,t)& =0 \quad&&\text{if}\quad & t &< 0 &&\\ \phi(x,t)& =0 &&\text{``} & t &= 0 & \quad\text{unless}\quad x &= c\\ \phi(x,t)& =1 &&\text{``} & t &= 0 & \text{and}\quad x &= c\\ \phi(x,t)& =1 &&\text{``} & x &= c &&\\ \phi(x,t)& =0 &&\text{``} & x &= 0.&& \end{alignat*} Precisely as in Art.~51 we get \[ u=\limit_{\tau\doteq 0} \sum_{k=0}^{k=n}\left[F(k\tau)\frac{[\phi(x,t-k\tau)-\phi(x,t-(k+1)\tau)]\tau}{\tau}\right] \tag{1} \] as the required solution of our problem, $n$ being as in Art.~51 the largest integer in $\dfrac{t}{\tau}$ where $t$ is any given value of the time. On our hypothesis the last term of (1), that is, $-F(n\tau)\phi[x,t-(n+1)\tau]=0$; the next to the last term $F(n\tau)\phi(x,t-n\tau)$ has for its limiting value \[ F(t)\phi(x,0)=F(t)\left[\frac{x}{c}+\frac{2}{\pi}\sum_{m=1}^{m=\infty}\left(\frac{(-1)^m}{m}\sin\frac{m\pi x}{c}\right)\right], \] while as in Art.~51 the limiting value of the rest of the sum is \begin{gather*} -\int\limits_0^t F(\lambda)D_\lambda\phi(x,t-\lambda)d\lambda.\\ D_\lambda\phi(x,t-\lambda)=\frac{2a^2\pi}{c^2}\sum_{m=1}^{m=\infty}\bigg[(-1)^m me^{-\frac{m^2a^2\pi^2}{c^2}(t-\lambda)}\sin\frac{m\pi x}{c}\bigg]. \end{gather*} Hence \begin{align*} u=F(t)\biggl[\frac{x}{c}&+\frac{2}{\pi}\sum_{m=1}^{m=\infty}\biggl(\frac{(-1)^m}{m}\sin\frac{m\pi x}{c}\biggr)\biggr]\\ &-\frac{2a^2\pi}{c^2}\sum_{m=1}^{m=\infty}\biggl((-1)^m m\sin\frac{m\pi x}{c}\int\limits_0^t F(\lambda)e^{-\frac{m^2a^2\pi^2}{c^2}(t-\lambda)}d\lambda\biggr), \end{align*} % -----File: 120.png \begin{multline*} u=\frac{x}{c}F(t)+\frac{2}{\pi}\sum_{m=1}^{m=\infty}\biggl[ \frac{(-1)^m}{m}\sin\frac{m\pi x}{c}\biggl(F(t)\\ -\frac{m^2a^2\pi^2}{c^2}\int\limits_0^t F(\lambda)e^{-\frac{m^2a^2\pi^2}{c^2}(t-\lambda)}d\lambda\biggr)\biggr]. \tag{2} \end{multline*} If we substitute $\beta=\dfrac{m^2a^2\pi^2}{c^2}(t-\lambda)$ we get \[ u=\frac{x}{c}F(t)+\frac{2}{\pi}\sum_{m=1}^{m=\infty} \biggl[\frac{(-1)^m}{m}\sin\frac{m\pi x}{c} \biggl(F(t)-\hspace{-10pt}\int\limits_0^{\frac{m^2a^2\pi^2t}{c^2}}\hspace{-10pt}e^{-\beta} F\biggl(t-\frac{\beta c^2}{m^2a^2\pi^2}\biggr)d\beta\biggr)\biggr]. \tag{3} \] \EXAMPLE{S} 1.\quad If the temperature of the left-hand face is a function of $t$ and the temperature of the right-hand face is zero and the initial temperature is zero \[ u=\biggl(1-\frac{x}{c}\biggr)F(t)-\frac{2}{\pi}\sum_{m=1}^{m=\infty} \biggl[\frac{1}{m}\sin\frac{m\pi x}{c} \biggl(F(t)-\hspace{-10pt}\int\limits_0^{\frac{m^2a^2\pi^2t}{c^2}}\hspace{-10pt}e^{-\beta} F\biggl(t-\frac{\beta c^2}{m^2a^2\pi^2}\biggr)d\beta\biggr)\biggr]. \] 2.\quad 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 \begin{multline*} u=F(t)-\frac{4}{\pi}\sum_{m=0}^{m=\infty} \biggl[\frac{1}{2m+1} \sin\frac{(2m+1)\pi x}{2c} \biggl(F(t)\\ -\frac{(2m+1)^2a^2\pi^2}{4c^2}\int\limits_0^t F(\lambda)e^{-\frac{(2m+1)^2a^2\pi^2}{4c^2}(t-\lambda)}d\lambda\biggr)\biggr]. \end{multline*} 3.\quad 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 temperature zero so that $D_tu=a^2D_x^2u-b^2u$, show that: (\textit{a}) the second members of (6) Art.~60 and (1) Art.~61 must be multiplied by $e^{-b^2t}$; (\textit{b}) equation (7) Art.~62 becomes \[ u_1=\gamma\Biggl\{\frac{\sinh\dfrac{bx}{a}}{\sinh\dfrac{bc}{a}}+2a^2\pi e^{-b^2t} \sum_{m=1}^{m=\infty}\left[(-1)^m\frac{m}{b^2c^2+m^2a^2\pi^2} e^{-\frac{m^2a^2\pi^2t}{c^2}}\sin\frac{m\pi x}{c}\right]\Biggr\}; \] % -----File: 121.png (\textit{c}) equation (2) Art.~63 becomes \begin{multline*} u=\frac{\sinh \dfrac{bx}{a}} {\sinh \dfrac{bc}{a}} F(t)+2a^2\pi \sum_{m=1}^{m=\infty} \bigg \{ \frac{(-1)^m m}{b^2c^2+m^2a^2\pi^2} \sin\frac{ m\pi x}{c} \bigg [ F(t) \\ - \frac{b^2c^2+m^2a^2\pi^2}{c^2} \int\limits_0^t e^{-\frac{b^2c^2+m^2a^2\pi^2}{c^2}(t-\lambda)}F(\lambda)d\lambda \bigg ] \bigg \}. \end{multline*} \mypara{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 \[ y=\frac{2}{l} \sum_{m=1}^{m=\infty} \sin \frac{m\pi x}{l} \cos \frac{m\pi at}{l }\int\limits_0^l f(\lambda) \sin \frac{m\pi\lambda}{l }d\lambda. \tag{1} \] The second member of (1) is a periodic function of $t$ having the period $\dfrac{2l}{a}$. The motion, then, unlike that in the case of an infinite string (Art.~55) is a true vibration, a periodic motion. The period $\dfrac{2l}{a}$ is the time it takes a disturbance 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 $y=b \sin \dfrac{m\pi x}{l}$ the solution is \[ y=b \sin \frac{m\pi x}{l} \cos\frac{ m\pi at}{l}. \tag{2} \] No matter what value $t$ may have the curve is always of the form \[ y=A \sin \frac{m\pi x}{l}; \] 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 $m$th 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 \textit{steady} vibration. % -----File: 122.png 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 \textit{first harmonic}. If $m = 3$ the vibration is three times as rapid as in the first case, there are two nodes $x=\dfrac{l}{3}$ and $x=\dfrac{2l}{3}$, and the note is the pure \textit{second harmonic} of the fundamental note. For any value of $m$ the vibration is $m$ times as rapid as when $m = 1$, there are ${m -1}$ nodes at the points $x=\dfrac{l}{m}, x=\dfrac{2l}{m}, \cdots x=\dfrac{m-1}{m}l$, and we get the ${m - 1}$st 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. \EXAMPLE{S} 1.\quad Show that if a point whose distance from the end of a harp string is $\dfrac{1}{n}$th 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 \[ y=\frac{2bn^2}{(n-1)\pi^2} \sum_{m=1}^{m=\infty} \left ( \dfrac{1}{m^2} \sin \frac{m\pi}{n} \sin \frac{ m\pi x}{l} \cos \frac{m\pi at}{l} \right ). \] 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.\\ % -----File: 123.png 2.\quad If a stretched string starts from its position of equilibrium, each of its points having a given initial velocity, so that we have \begin{alignat*}{3} y&=0 & \text{when}\quad && t&=0\\ D_ty&=F(x) & \text{``}\qquad && t&=0\\ y&=0 & \text{``}\qquad && x&=0\\ y&=0 & \text{``}\qquad && x&=l, \end{alignat*} the solution of the problem of its vibration is easy and gives \[ y=\frac{2}{a\pi}\sum_{m=1}^{m=\infty}\bigg(\frac{1}{m}\sin\frac{m\pi x}{l}\sin\frac{m\pi at}{l} \int\limits_0^l F(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\bigg). \] 3.\quad Write down the solution for the case where the string is initially distorted and each point has a given initial velocity. \mypara{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 complicated and the solution is not so easily obtained. The equation is given as \smallromr{IX} Art.~1. Let us solve the problem for the case where there is no initial velocity. \begin{flalign*} &\indent\text{Here we have} & D_t^2y+2kD_ty=a^2D_x^2y.\qquad\qquad & \tag{1}& \end{flalign*} \vspace{-6.5ex} \begin{alignat*}{3} y&=0 && \text{when}\quad & x&=0 \tag{2}\\ y&=0 && \quad\text{``} & x&=l \tag{3}\\ y&=f(x) && \quad\text{``} & t&=0 \tag{4}\\ D_ty&=0 && \quad\text{``} & t&=0. \tag{5} \end{alignat*} We get particular solutions of (1) in the usual way. Assume $y=e^{\alpha x+\beta t}$ and substitute in (1). We have \[ \beta^2+2k\beta=a^2\alpha^2 \] as the only necessary relation between $\beta$ and $\alpha$. This gives \begin{flalign*} &&\beta=-k±\sqrt{a^2\alpha^2+k^2}.&&\\ &\indent\text{Hence}& y=e^{\alpha x-kt±t\sqrt{a^2\alpha^2+k^2}}& \tag{6}& \end{flalign*} is a solution of (1) no matter what the value of $\alpha$. To throw it into Trigonometric form replace $\alpha$ by $\alpha i$, 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 \[ y=e^{-kt}e^{(\alpha x±t\sqrt{a^2\alpha^2-k^2)}i}. \] % -----File: 124.png Since $\alpha$ may be positive or negative we can get \begin{flalign*} & &y = e^{-kt}\sin(\alpha x ± t\sqrt{a^2\alpha^2-k^2)}&&\\ &\text{and} &y = e^{-kt}\cos(\alpha x ± t\sqrt{a^2\alpha^2-k^2)}&& \end{flalign*} as solutions of (1), or by combining these \begin{align*} y& = e^{-kt}\sin \alpha x \cos t\sqrt{a^2\alpha^2-k^2} \tag{7}\\ y& = e^{-kt}\sin \alpha x \sin t\sqrt{a^2\alpha^2-k^2} \tag{8}\\ y& = e^{-kt}\cos \alpha x \cos t\sqrt{a^2\alpha^2-k^2} \tag{9}\\ y& = e^{-kt}\cos \alpha x \sin t\sqrt{a^2\alpha^2-k^2} \tag{10} \end{align*} (7) and (8) satisfy (1) and (2) for all values of $\alpha$. They satisfy (3) if $\alpha=\dfrac{m\pi}{l}$. Let us see if out of them we cannot build up a value that will satisfy (4) and (5) as well. \begin{gather*} f(x)=\frac{2}{l}\sum_{m=1}^{m=\infty}\biggl(\sin\frac{m\pi x}{l}\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr). \tag{11}\\ y=\frac{2}{l}e^{-kt}\sum_{m=1}^{m=\infty}\biggl(\sin\frac{m\pi x}{l}\cos t \sqrt{\frac{m^2\pi^2 a^2}{l^2}-k^2}.\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr) \tag{12} \end{gather*} reduces to (11) when $t=0$ and therefore satisfies (4). \begin{align*} % recast to fit line D_t y=&-\frac{2}{le^{kt}}\sum_{m=1}^{m=\infty}\biggl(\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}. \sin\frac{m\pi x}{l}\sin t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}\\ &\hspace{18.5em}.\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr)\\ &-\frac{2k}{le^{kt}}\sum_{m=1}^{m=\infty}\biggl( \sin\frac{m\pi x}{l}\cos t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}.\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr). \tag{13} \end{align*} When $t=0$ the first line of the second member of (13) vanishes but the second line reduces to \[ -\frac{2k}{l}\sum_{m=1}^{m=\infty}\biggl(\sin\frac{m\pi x}{l}\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr). \] 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 $t=0$. % -----File: 125.png This is easily seen to be \[ \frac{2k}{l}e^{-kt}\sum_{m=1}^{m=\infty}\frac{1}{\sqrt{\dfrac{m^2\pi^2a^2}{l^2}-k^2}} \sin\frac{m\pi x}{l}\sin t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}.\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda. \] Hence our complete solution is \begin{multline*} y=\frac{2}{l}e^{-kt}\sum_{m=1}^{m=\infty}\bigg[\bigg(\cos t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}\\ +\frac{k}{\sqrt{\dfrac{m^2\pi^2a^2}{l^2}-k^2}}\sin t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}\bigg) \sin\frac{m\pi x}{l}\int\limits_0^l f(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\bigg]. \tag{14} \end{multline*} 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 \[ A\sin\frac{m\pi x}{l}\cos t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}, \] is greater than that of the corresponding term $A_1\sin\dfrac{m\pi x}{l}\cos\dfrac{m\pi at}{l}$ in Art.~64. 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. \EXAMPLE{S} 1.\quad Solve Ex.~1 Art.~64 allowing for the resistance of the air.\\ 2.\quad Solve Ex.~2 Art.~64 allowing for the resistance of the air; \begin{multline*} % recast to fit line y=\frac{2}{l}e^{-kt}\sum_{m=1}^{m=\infty}\biggl(\frac{1}{\sqrt{\dfrac{m^2\pi^2a^2}{l^2}-k^2}} \sin\frac{m\pi x}{l}\sin t\sqrt{\frac{m^2\pi^2a^2}{l^2}-k^2}\\[-5ex] . \int\limits_0^l F(\lambda)\sin\frac{m\pi\lambda}{l}d\lambda\biggr). \end{multline*} % -----File: 126.png 3.\quad 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. \mypara{66.} We pass on now to a couple of problems that require the modification and extension of Fourier's Theorem, \emph{the cooling of a sphere in air}, and the \emph{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 \[ D_t(ru) = a^2D_r^2(ru), \tag{1} \] see \smallrom{V} Art.~1, subject to the conditions \begin{alignat*}{3} u &=f(r)\ && \text{when}\quad &t&=0 \tag{2}\\ u &=b && \quad\text{``} &r&=c \tag{3} \end{alignat*} if $c$ is the radius. Let $v=ru$, then our equations become \begin{alignat*}{3} &&\llap{$D_tv =\:$} &\rlap{$a^2D_r^2v$} \tag{4}\\ v &=rf(r)\ && \text{when}\quad &t&=0 \tag{5}\\ v &=bc && \quad\text{``} &r&=c \tag{6}\\ v &=0 && \quad\text{``} &r&=0. \tag{7} \end{alignat*} Our problem is now precisely that of Art.~62 and we have as our solution \begin{align*} ru=&\frac{2}{c}\sum_{m=1}^{m=\infty}\biggl(e^{-\frac{m^2a^2\pi^2}{c^2}t}\sin\frac{m\pi r}{c} \int\limits_0^c\lambda f(\lambda)\sin\frac{m\pi\lambda}{c}d\lambda\biggr)\\ &+b\biggl[r+\frac{2c}{\pi}\sum_{m=1}^{m=\infty}\biggl(\frac{(-1)^m}{m}e^{-\frac{m^2a^2\pi^2}{c^2}t}\sin\frac{m\pi r}{c}\biggr)\biggr]. \tag{8} \end{align*} \EXAMPLE{S} 1.\quad If $f(r)=b$ (8) Art.~66 reduces to $u=b$ and there is no change of temperature. 2.\quad If the initial temperature is constant and equal to $\beta$ \begin{multline*} u= b + \frac{2c}{\pi r}(\beta-b) \bigg[e^{-\frac{ a^2\pi^2}{c^2}t}\sin\frac{ \pi r}{c} -\frac{1}{2}e^{-\frac{4a^2\pi^2}{c^2}t}\sin\frac{2\pi r}{c}\\ +\frac{1}{3}e^{-\frac{9a^2\pi^2}{c^2}t}\sin\frac{3\pi r}{c}-\cdots\biggr]. \end{multline*} % -----File: 127.png 3.\quad 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 $a^2 = 0.185$ in C.G.S.\ units. \hfill \textit{Ans.}, 2°.1; 3°.3. \mypara{67.} If instead of having the temperature of the surface of the sphere constant, 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 \[ D_{r}u + hu = 0 \quad \text{when} \quad r = c, \tag{1} \] where $h$ is an experimental constant depending upon what is called the surface conductivity of the sphere. Our equations, then, are \begin{gather*} D_{t}(ru) = a^{2}D_{r}^{2}(ru) \tag{2}\\ u =f(r) \qquad \text{when}\qquad t = 0 \tag{3}\\ D_{r}u + hu = 0 \qquad \text{when} \qquad r = c. \tag{4} \end{gather*} As in Art.~66 let $v = ru$; then we have \begin{alignat*}{3} &&\llap{$D_{t}v =\:$} & \rlap{$a^{2}D_{r}^{2}v$} \tag{5}\\ v &= rf(r) \quad && \text{when} \quad & t &= 0 \tag{6}\\ v &= 0 &&\quad \text{``} & r &= 0 \tag{7}\\[-5ex] \end{alignat*} \[ D_{r}v + \left ( h - \frac{1}{c} \right )v = 0 \quad \text{when} \quad r = c. \tag{8} \] $v = e^{-a^2\alpha^{2}t} \cos \alpha r$ and $v = e^{-a^2\alpha^{2}t} \sin \alpha r$ have already been found as particular solutions of (5) (see Art.~60). \[ v = e^{-a^2\alpha^{2}t} \sin \alpha r \tag{9} \] satisfies (7) for all values of $\alpha$. Substitute this value of $v$ in (8) and we have \[ \alpha c \cos \alpha c + (hc - 1) \sin \alpha c = 0. \tag{10} \] If $\alpha_{k}$ is a value of $\alpha$ which is a root of the transcendental equation (10) \[ v = e^{-a^{2}\alpha_{k}^{2}t} \sin \alpha_{k}r \tag{11} \] 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). % -----File: 128.png When $t = 0$ the second member of (11) reduces to $\sin \alpha_kr$. If then we can express $rf(r)$ as a sum of terms of the form $b_k \sin \alpha_k r$ where $\alpha_k$ is a root of (10) \[ v = \sum b_k e^{-a^2\alpha_k^2t} \sin \alpha_k r \tag{12} \] 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 complicated, namely, to develop any function of $x$ in a series of the form $\displaystyle\sum a_m \sin \alpha_m x$ where $\alpha_m$ is a root of the equation (10); or if we call $ac = \phi$ and $hc-1=p$, where $a_m = \dfrac{\phi_m}{c}$, $\phi_m$ being a root of the equation \begin{gather*} \phi \cos \phi + p \sin \phi = 0 \tag{13} \intertext{or more simply of} \phi + p \tan \phi = 0; \tag{14} \end{gather*} remembering that the series and the function must be equal for all values of $x$ between zero and $c$. If $\phi_m$ is a root of (14) $- \phi_m$ is also a root. Since $\sin \dfrac{\phi_m}{c} x = - \sin \left (- \dfrac{\phi_m}{c} x \right )$ the terms of the required development which correspond to negative roots may be combined with those corresponding to positive roots, and therefore we need consider only positive roots. $\phi = 0$ is a root of (14) but as $\sin 0 = 0$ there will be no corresponding term in the development. If we construct the curve \begin{align*} y &= - \frac{1}{p} x \tag{15} \intertext{and the curve} y &= \tan x \tag{16} \end{align*} the abscissas of their points of intersection are values of $x$ which satisfy $\dfrac{x}{p} + \tan x = 0$, 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 $\phi_m$ will rapidly approach the value $(2m - 1) \dfrac{\pi}{2}$ if $p$ is positive or if $p$ is negative and numerically less than unity, and $(2m + 1) \dfrac{\pi}{2}\tablestrut$ if $p$ is negative and numerically greater than unity. % -----File: 129.png There exist, then, an infinite number of positive real roots of $\phi + p \tan \phi = 0$ and consequently of \[ \alpha c \cos \alpha c + (hc - 1) \sin \alpha c = 0. \] \mypara{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 \[ f(x) = a_1 \sin\phi_1 x + a_2 \sin\phi_2 x + a_3 \sin\phi_3 x + \cdots \tag{1} \] where $\phi_1$, $\phi_2$, $\phi_3 \cdots$ are roots of the equation \[ \phi \cos\phi +p \sin\phi = 0, \tag{2} \] the development to hold good for all values of $x$ between $x = 0$ and $x = 1$. Let us proceed as in Arts.~24 and 27. Call $\dfrac{1}{n+1} = \Delta x$ and form $n$ equations by substituting for $x$ in turn in the equation \[ f(x) = a_1 \sin\phi_1 x + a_2 \sin\phi_2 x + a_3 \sin\phi_3 x + \cdots + a_n \sin\phi_n x \tag{3} \] the values $\Delta x$, $2\Delta x$, $3\Delta x$, $\cdots n\Delta x$; 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 $a_m$ multiply the first equation by $\Delta x.\sin(\phi_m$ $\Delta x)$, the second by $\Delta x.\sin(2\phi_m\Delta x)$, the third by $\Delta x.\sin(3\phi_m \Delta x)$, and so on, the $n$th equation by $\Delta x.\sin(n\phi_m\Delta x)$; add the equations and compute the limiting 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 $\sin\phi_m x.dx$ and integrating between the limits $x = 0$ and $x = 1$. The first member of the resulting equation is \[ \int\limits_0^1 f(x) \sin \phi_m x.dx; \] The coefficient of $a_k$ is \[ \int\limits_0^1 \sin\phi_k x \sin\phi_m x.dx, \] and of $a_m$ is \[ \int\limits_0^1 \sin^2\phi_m x.dx. \] % -----File: 130.png \begin{align*} \int\limits_0^1 \sin\phi_k x \sin\phi_m x.dx &= \frac{1}{2}\int\limits_0^1 [\cos(\phi_k-\phi_m)x - \cos(\phi_k+\phi_m)x]dx\\ &=\frac{1}{2} \left [ \frac{\sin(\phi_k-\phi_m) }{ \phi_k-\phi_m }- \frac{\sin(\phi_k+\phi_m) }{ \phi_k+\phi_m} \right ]\\ &= - \frac{\phi_k \cos\phi_k \sin\phi_m - \phi_m \sin\phi_k \cos\phi_m }{ \phi_k^2 - \phi_m^2 } \tag{4} \end{align*} \begin{flalign*} & \indent \text{But} & \phi_k \cos\phi_k &+ p \sin\phi_k = 0&&\\ & \text{and} & \phi_m \cos\phi_m &+ p \sin\phi_m = 0\qquad\text{by (2)}. \end{flalign*} Hence the numerator of the second member of (4) is zero, and the coefficient of $a_k$ vanishes if $k$ is not equal to $m$. \[ \int\limits_0^1 \sin^2\phi_m x.dx = \frac{1}{2\phi_m }[\phi_m - \sin\phi_m \cos\phi_m] = \frac{1}{2} \left [1 - \frac{\sin 2\phi_m }{ 2\phi_m} \right ]. \tag{5} \] \begin{flalign*} &\text{Therefore}& a_m &= \frac{2 }{1 - \dfrac{\sin 2\phi_m }{ 2\phi_m}} \int\limits_0^1 f(x) \sin\phi_m x.dx. &&\hspace{2em}\tag{6} \end{flalign*} The coefficient of the integral in (6) can be transformed as follows so as not to involve trigonometric functions. \begin{gather*} \phantom{\qquad by (2)}\phi_m \cos\phi_m + p \sin\phi_m = 0, \qquad \text{by (2)}\\ \phi_m \cos^2\phi_m + \frac{p}{2} \sin 2\phi_m = 0,\\ \frac{\sin 2\phi_m }{2\phi_m} = -\frac{\cos^2\phi_m }{p}. \tag{7}\\ \phi_m^2 \cos^2\phi_m = p^2 \sin^2\phi_m,\\[1ex] (\phi_m^2 + p^2) \cos^2\phi_m = p^2,\\[1ex] \frac{\cos^2\phi_m }{p} = \frac{p }{ \phi_m^2 + p^2}. \tag{8} \end{gather*} Hence by (7) and (8) \begin{flalign*} & &&1-\frac{\sin 2\phi_m }{ 2\phi_m} = \frac{\phi_m^2+p(p+1) }{ \phi_m^2 + p^2},&&\\ &\text{and} &a_m = &\frac{2(\phi_m^2+p^2) }{ \phi_m^2+p(p+1)} \int\limits_0^1 f(\alpha)\sin \phi_m \alpha.d\alpha.&& \tag{9} \end{flalign*} Therefore our required development is \[ f(x)=\sum_{m=1}^{m=\infty} \biggl ( \frac{2(\phi_m^2+p^2) }{ \phi_m^2+p(p+1)} \sin\phi_m x \int\limits_0^1 f(\alpha) \sin\phi_m \alpha.d\alpha \biggr ). \tag{10} \] % -----File: 131.png From (10) it easily follows that for values of $x$ between $0$ and $c$ \begin{flalign*} &&f(x) = &a_1 \sin a_1 x + a_2\sin a_2 x + a_3 \sin a_3 x + \cdots \tag{11}\\ &\text{where} && a_m = \frac{2}{c}.\frac{\alpha_m^2 c^2 + p^2}{\alpha_m^2 c^2 + p(p+1)} \int\limits_0^c f(\lambda) \sin \alpha_m \lambda.d\lambda, \hspace{2.5em}&&\tag{12} \end{flalign*} and $\alpha_m$ is a root of the equation \[ \alpha c \cos \alpha c + p \sin \alpha c =0. \tag{13} \] It is to be observed that if $p$ is infinite (13) reduces to $\sin \alpha c=0$, $\alpha_m$ becomes $\dfrac{m \pi}{c}$ and (11) and (12) give our regulation Fourier sine series (v.\ Art.\ 31), and 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\\[-4ex] \begin{flalign*} &&\int\limits_0^c \sin \alpha_m x &\sin \alpha_n x.dx = 0,&&\\ &\text{that is that} &\frac{\sin(\alpha_m-\alpha_n)c}{\alpha_m-\alpha_n} - &\frac{\sin(\alpha_m+\alpha_n)c}{\alpha_m+\alpha_n} = 0 &&\\[1ex] &\text{or reducing, that} &\frac{\alpha_m c \cos \alpha_mc }{ \sin \alpha_mc} &= \frac{\alpha_n c \cos \alpha_n c }{ \sin \alpha_n c}, &&\phantom{or reducing, that} \end{flalign*} or that $\alpha_m$ and $\alpha_n$ should be roots of the equation \[ \frac{\alpha c \cos \alpha c }{ \sin \alpha c} = p \] 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. \EXAMPLE{S} 1.\quad Show that the solution of the problem of Art.~67 is \begin{flalign*} & && ru = \sum_{m=1}^{m=\infty} b_m e^{-a^2\alpha_m^2 t}\sin \alpha_m r, &&\\ &\text{where} &b_m = \frac{2}{c }.&\frac{\alpha_m^2c^2+(hc-1)^2 }{ \alpha_m^2c^2+hc(hc-1)} \int\limits_0^c \lambda f(\lambda) \sin \alpha_m \lambda.d\lambda &&\phantom{where} \end{flalign*} and $\alpha_m$ is a root of \[ \alpha c \cos \alpha c + (hc - 1) \sin \alpha c = 0. \] % -----File: 132.png 2.\quad If the initial temperature of the sphere is constant and equal to $\beta$ \begin{flalign*} &&ru &= \sum_{m=1}^{m=\infty} b_m e^{-a^2\alpha_m^2t} \sin \alpha_m r &&\\ &\text{where} & b_m &= 2\beta h.\frac{ \alpha_m^2c^2+(hc-1)^2}{\alpha_m^2c^2+hc(hc-1)}.\frac{\sin \alpha_mc}{\alpha_m^2} &&\phantom{where}\\ &&& = \frac{2\beta hc}{\alpha_m}.\frac{ [\alpha_m^2c^2+(hc-1)^2]^{\frac{1}{2}}}{\alpha_m^2c^2+hc(hc-1)}. && \end{flalign*} 3.\quad If the temperature of the air is a constant $\gamma$ instead of zero the surface equation of condition is \[ D_r u + h (u - \gamma) = 0 \quad \text{when} \quad r = c. \] The substitution of $u_1 = u - \gamma$, however, brings the problem under Ex.~1 and we get \begin{flalign*} && r(u - \gamma) = \sum_{m=1}^{m=\infty} b_m e^{-a^2\alpha_m^2t} \sin \alpha_m &r && \\ &\text{where} & b_m = \frac{2}{c}.\frac{ \alpha_m^2c^2+(hc-1)^2}{\alpha_m^2c^2+hc(hc-1)} \int\limits_0^c \lambda[f(\lambda) - \gamma] &\sin \alpha_m \lambda.d\lambda. &&\phantom{where} \end{flalign*} 4.\quad 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 $a^2 = 0.185$ and $h = \dfrac{1}{800}$ in C.G.S. units. (v.\ Ex.~3, Art.~66.)\\ \phantom{kindly go to the right hand margin}\hfill \emph{Ans}., 97°.67; 97°.36; 96°.46.\\ 5.\quad Show that if in the slab considered in Art.~60 one face is exposed to air at the temperature zero, so that we have $D_t u = a^2D_x^2u$, $u = 0$ when $x = 0$, $u=f(x)$ when $t = 0$, and $D_x u + hu = 0$ when $x = c$, then \begin{flalign*} &&& u = \sum_{m=1}^{m=\infty} a_m e^{-a^2\alpha_m^2t} \sin \alpha_m x &&\\ &\text{where} & a_m & = 2 \frac{\alpha_m^2+h^2}{\alpha_m^2c+h(hc+1)} \int\limits_0^c f(\lambda) \sin \alpha_m \lambda.d\lambda, && \phantom{where} \end{flalign*} $\alpha_m$ being a root of $\alpha c \cos \alpha c + hc \sin \alpha c = 0$.\\ % -----File: 133.png 6.\quad 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 $D_x^2u+D_y^2u = 0$, $u = 0$ when $x = 0$, $u=f(x)$ when $y = 0$, and $D_{x}u + hu = 0$ when $x = a$, then \begin{flalign*} &&u = \sum_{m=1}^{m=\infty} a_m e^{-\alpha_m y} &\sin \alpha_m x &&\\ &\text{where} & a_m = 2 \frac{\alpha_m^2 + h^2}{\alpha_m^2 a + h(ha + 1)} &\int\limits_0^a f(\lambda) \sin \alpha_m\lambda.d\lambda,&& \phantom{where} \end{flalign*} $\alpha_m$ being a root of $\alpha a \cos \alpha a + ha \sin \alpha a = 0$.\\ 7.\quad If in the problem of Art.~59 there is leakage at one side of the sheet so that we have $D_x^2 V + D_y^2 V = 0$, $V = 0$ when $x = 0$, $V = 0$ when $y = b$, $V=f(x)$ when $y = 0$, and $D_xV+hV=0$ when $x = a$, then \[ V=\sum_{m=1}^{m=\infty} a_m \frac{\sinh \alpha_m(b-y)}{\sinh \alpha_m b }\sin \alpha_m x, \] where $a_m$ has the value given in Ex. 6. \mypara{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 \[ D_tu=a^2D_x^2u\\ \] subject to the conditions \begin{flalign*} &&\multispan{2}{\hfill $u = 0 \quad \text{when} \quad t = 0$ \hfill}\\ &\text{and} &D_xu+ h(U - &u) = 0 \quad \text{when} \quad x = 0. && \phantom{and}\\ &\text{Let} & v ={}&u - \frac{1}{h}D_xu. && \tag{1} \end{flalign*} Then $v$ will satisfy the equation \[ D_tv = a^2D_x^2v, \] and we shall also have $v = U$ when $x = 0$. \begin{flalign*} &\text{\indent Since } U=F(t) \qquad v = \frac{2}{\sqrt{\pi}}\!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\!e^{-\beta^2} F \left (t-\frac{x^2}{4a^2\beta^2} \right )d\beta && \,\tag{2} \end{flalign*} by Art.~51 (10). \begin{flalign*} & &&D_xu - hu = - hv \qquad\text{by (1)}.&&\\ &\indent \text{Hence} \quad &&ue^{-hx} = - h \int e^{-hx} vdx+C; &&\phantom{\indent \text{Hence}} \end{flalign*} v.\ Int.\ Cal.\ \S~4, page 314. % -----File: 134.png Determining $C$ by the fact that $ue^{-hx} = 0$ when $x = \infty$ we have \[ u=he^{hx} \int\limits_x^{\infty} e^{-hx} vdx. \tag{3} \] Substituting the value of $v$ from (2) we have \[ u= \frac{2he^{hx}}{\sqrt{\pi}}\int\limits_x^{\infty} e^{-hx} dx \!\!\int\limits_{\frac{x}{2a\sqrt{t}}}^{\infty}\!\! e^{-\beta^2}F\left ( t- \frac{x^2}{4a^2\beta^2} \right ) d\beta, \tag{4} \] as our required solution. For an extension of this method to the flow of heat in two and three dimensions and for the interpretation of the results by the aid of the theory of \textit{Images}, see E.~W. Hobson, Proc.\ Lond.\ Math.\ Soc., Vol.\ XIX. \EXAMPLE{S} 1.\quad If the temperature of the air is a periodic function of the time, say $\rho_{m}\sin (m\alpha t + \lambda_{m})$ and we care only for the limiting value of $u$ as $t$ increases, show that this value is \begin{align*} \frac{h\rho_m e^{-\frac{x}{a}\frac{\sqrt{m\alpha}}{2}}}{\Big( h +\dfrac{1}{a} \sqrt{\dfrac{m\alpha}{2}}\Big)^2+\dfrac{m\alpha}{2a^2}} &\left [ \left ( h+\dfrac{1}{a} \sqrt{\frac{m\alpha}{2}}\right ) \sin \left (m\alpha t - \frac{x}{a} \sqrt{\frac{m\alpha}{2}}+\lambda_m \right ) \right .\\ & \quad-\left . \frac{1}{a} \sqrt{\frac{m\alpha}{2}}\cos \left (m\alpha t - \frac{x}{a} \sqrt{\frac{m\alpha}{2}}+\lambda_m \right ) \right ]. \end{align*} v.\ Art.~52 and Art.~51 Ex.~4. \begin{flalign*} &\text{\indent Note that } & \int e^{ax}\sin bx.dx &= \frac{e^{ax}(a \sin bx - b \cos bx)}{a^2+b^2} && \phantom{\indent Note that}\\ &\text{and } & \int e^{ax}\cos bx.dx &=\frac{e^{ax}(a \cos bx + b \sin bx)}{a^2+b^2} && \end{flalign*} v.\ Int.\ Cal.\ Table of Int.\ (235) and (236).\\ 2.\quad If $D_{x}^{2}V+D_{y}^{2}V=0$, $V=0$ when $y = 0$ and $D_{x}V+ h[F(y) - V] = 0$ when $x = 0$ show that \[ V=\frac{he^{hx}}{\pi} \int\limits_x^{\infty}e^{-hx}dx \int\limits_0^{\infty}F(\lambda)d\lambda \left [ \frac{x}{x^2+(\lambda-y)^2}-\frac{x}{x^2+(\lambda+y)^2} \right ];\\ \] v.\ Art.~47 Ex.~1. \mypara{70.} The solution for an instantaneous heat source of strength $Q$ at the point $x = \lambda$ if heat escapes at the origin into air at the temperature zero, so that $D_{x}u - hu = 0$ when $x = 0$, can be obtained by the aid of Art.~53. % -----File: 135.png Let $u = u_1 + u_2$ where $u_1$ is the temperature that would be due to the given source if we had no boundary at the origin, so that \[ u_1= \frac{Q}{2a\sqrt{\pi t}}e^{-\frac{(\lambda -x)^2}{4a^2t} }. \tag*{ [Art.~53 (2)]} \] \vspace{-5ex} \begin{flalign*} &&D_{x}u - hu = D_{x}u_{1} - hu_{1} &+ D_{x}u_{2} - hu_{2} = 0 \quad \text{when }\quad x=0. && \\ &\text{Therefore} & D_{x}u_{2} - hu_{2} &= -(D_{x}u_{1} - hu_{1}) && \tag{1}\\ \multispan{3}{when $x = 0$. \hfill}\\[-6ex] \end{flalign*} \begin{flalign*} &\indent \text{But} & -(D_x u_1-hu_1)&=- \frac{Q}{2a\sqrt{\pi t}} \left ( \frac{\lambda -x}{2a^2t} -h \right )e^{-\frac{(\lambda -x)^2}{4a^2t} } && \phantom{But}\\ &&& = -\frac{Q}{2a\sqrt{\pi t}} \left( \frac{\lambda}{2a^2t} -h \right ) e^{-\frac{\lambda^2}{4a^2t}} && \\ \multispan{3}{when $x = 0$.\hfill} \end{flalign*} This is easily seen to be the value to which \[ -\frac{Q}{2a\sqrt{\pi t}} \left( \frac{\lambda+x}{2a^2t} -h \right ) e^{-\frac{(\lambda+x)^2}{4a^2t}} \] reduces when $x = 0$, and this last expression is \[ (D_x+h)\frac{Q}{2a\sqrt{\pi t}}e^{-\frac{(\lambda+x)^2}{4a^2t}} \] and therefore satisfies the equation \[ D_{t}u=a^{2}D_{x}^{2}u; \tag{2} \] since $\dfrac{Q}{2a\sqrt{\pi t}}e^{-\frac{(\lambda+x)^2}{4a^2t}}$ is the temperature due to a source at $x = - \lambda$.\\ If, then, we determine $u_2$ from the condition that \[ D_x u_2-hu_2= -\frac{Q}{2a\sqrt{\pi t}} \left ( \frac{\lambda+x}{2a^2t}-h \right ) e^{-\frac{(\lambda+x)^2}{4a^2t}} \tag{3} \] taking care not to introduce any arbitrary constant or arbitrary function of $t$ in our integration, $u_2$ will satisfy equation (2) and condition (1). Integrating (3) [v.\ Int.\ Cal.\ \S~4, page 314] and determining the constants of integration suitably we get \[ u_2= \frac{Q}{2a\sqrt{\pi t}} \biggl [ e^{-\frac{(\lambda + x)^2}{4a^2t} } - 2he^{hx} \int\limits_x^{\infty} e^{-hx -\frac{(\lambda + x)^2}{4a^2t} } dx \biggr ]. \tag{4} \] Therefore the solution of our problem is \[ u= \frac{Q}{2a\sqrt{\pi t}} \biggl [ e^{-\frac{(\lambda -x)^2}{4a^2t} } +e^{-\frac{(\lambda+x)^2}{4a^2t}}- 2he^{hx} \int\limits_x^{\infty} e^{-hx -\frac{(\lambda + x)^2}{4a^2t} } dx \biggr ]. \tag{5} \] % -----File: 136.png If we replace $Q$ by $f(\lambda)d\lambda$ and integrate from $0$ to $\infty$ we get as the solution for the case where $u=f(x)$ when $t = 0$ and $x > 0$, and $D_{x}u - hu = 0$ when $x = 0$ \[ u=\frac{1}{2a\sqrt{\pi t}}\int\limits_0^{\infty}f(\lambda)d\lambda \biggl [ e^{-\frac{(\lambda-x)^2}{4a^2t}}+ e^{-\frac{(\lambda+x)^2}{4a^2t}} -2he^{hx} \int\limits_x^{\infty} e^{- hx -\frac{(\lambda+x)^2}{4a^2t} } dx \biggr ]. \tag{6} \] 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 $D_{x}u + h[F(t) - u] = 0$ when $x = 0$ we must take $u$ equal to the sum of the second members of (6) Art.~70 and of (4) Art.~69. \mypara{71.} As another problem requiring a slight extension of Fourier's Theorem 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 \[ D_{t}^{2}z= c^{2}(D_{x}^{2}z+D_{y}^{2}z) \tag{1} \] see \smallrom{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 \begin{alignat*}{3} z &= 0 && \text{when}\qquad & x &= 0 \tag{2}\\ z &= 0 &&\quad \text{``} & x &= a \tag{3}\\ z &= 0 &&\quad \text{``} & y &= 0 \tag{4}\\ z &= 0 &&\quad \text{``} & y &= b \tag{5}\\ z &=f(x,y) &&\quad \text{``} & t &= 0 \tag{6}\\ D_{t}z&=0 &&\quad \text{``} & t &= 0. \tag{7} \end{alignat*} We can get a particular solution of (1) by our usual device. Assume \[ z=e^{\alpha x + \beta y + \gamma t} \] and substitute in (1). We get $\gamma^{2} = c^{2}(\alpha^{2} + \beta^{2})$ as the only relation that need hold between $\alpha$, $\beta$, and $\gamma$, in order that $z=e^{\alpha x + \beta y + \gamma t}$ may be a solution. This gives \begin{flalign*} && \gamma = &\pm c \sqrt{ \alpha^2 + \beta^2}. &&\\ &\text{\indent Therefore} &z = e&^{\alpha x + \beta y \pm ct \sqrt{ \alpha^2 + \beta^2}}&&\phantom{Therefore} \end{flalign*} is a solution of (1) no matter what values are given to $\alpha$ and $\beta$. % -----File: 137.png Replace $\alpha$ and $\beta$ by $ \alpha i$ and $\beta i$ and we have \[ z=e^{(\alpha x + \beta y \pm ct \sqrt{\alpha^2+\beta^2 })i} \] as a solution, and from this we get \begin{flalign*} & &z = \sin (\alpha x + \beta y \pm ct \sqrt{ \alpha^{2}+\beta^{2}}) && \tag{8}\\ &\text{and}\qquad &z = \cos (\alpha x + \beta y \pm ct \sqrt{ \alpha^{2}+\beta^{2}}) && \tag{9} \end{flalign*} as particular solutions of (1), $\alpha$ and $\beta$ being unrestricted. From (8) and (9) we can get solutions of the following forms \[ \left . \begin{alignedat}{3} z &= \sin \alpha x &&\sin \beta y &&\sin ct \sqrt{\alpha^2 + \beta^2}\\ z &= \sin \alpha x &&\sin \beta y &&\cos ct \sqrt{\alpha^2 + \beta^2}\\ z &= \sin \alpha x &&\cos \beta y &&\sin ct \sqrt{\alpha^{2} + \beta^{2}}\\ z &= \sin \alpha x &&\cos \beta y &&\cos ct \sqrt{\alpha^{2} + \beta^{2}}\\ z &= \cos \alpha x &&\sin \beta y &&\sin ct \sqrt{\alpha^{2} + \beta^{2}}\\ z &= \cos \alpha x &&\sin \beta y &&\cos ct \sqrt{\alpha^{2} + \beta^{2}}\\ z &= \cos \alpha x &&\cos \beta y &&\sin ct \sqrt{\alpha^{2} + \beta^{2}}\\ z &= \cos \alpha x &&\cos \beta y &&\cos ct \sqrt{\alpha^{2} + \beta^{2}}, \end{alignedat} \right \} \qquad \text{(10)} \] each of which will satisfy equation (1). The second of these will satisfy also (2), (4) and (7) whatever values be taken for $\alpha$ and $\beta$. It will satisfy (3) and (5) if $ \alpha$ and $\beta$ are equal $\dfrac{m\pi}{a}$ and $\dfrac{n\pi}{b}$ respectively. If, then, we can so combine terms of the form \[ \sin \frac{m\pi x}{a}\sin\frac{n\pi y}{b}\cos c\pi t \sqrt{\frac{m^2}{a^2}+\frac{n^2}{b^2}} \] 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 $A \sin \dfrac{m\pi x}{a}\sin \dfrac{n\pi y}{b}$, the sum and the function being equal when $x$ lies between $0$ and $a$ and $y$ between $0$ and $b$. $f(x,y)$ can be expressed in terms of $\sin \dfrac{m\pi x}{a}$ by Fourier's Theorem if we regard $y$ as constant. We have \begin{flalign*} &&f(x,y)&= \sum_{m=1}^{m=\infty} a_m \sin \frac{m\pi x}{a} && \tag{11}\\ % -----File: 138.png &\text{where} &a_m=\frac{2}{a} &\int\limits_0^a f(\lambda,y)\sin \frac{m\pi\lambda}{a} d\lambda. \tag{12}&&\; \end{flalign*} $f(\lambda,y)$ in (12) is a function of $y$ and may be developed by Fourier's Theorem. \begin{flalign*} &\text{We have} & \quad f(\lambda,y)&= \sum_{n=1}^{n=\infty} b_n \sin \frac{n\pi y}{b } \tag{13} &&\\ &\text{where} & b_n=\frac{2}{b} &\int\limits_0^b f(\lambda,\mu)\sin \frac{n\pi\mu}{b} d\mu. \hspace{3em}\tag{14} \end{flalign*} Substituting for $f(\lambda,y)$ in (12) the value just obtained we have \[ a_m =\frac{2}{a}\frac{2}{b}\sum_{n=1}^{n=\infty}\biggl (\int\limits_0^a d\lambda \int\limits_0^b f(\lambda,\mu) \sin\frac{m\pi \lambda}{a}\sin\frac{n\pi \mu}{b} d\mu \biggr ) \sin\frac{n\pi y}{b} \] and \[ f(x,y) =\frac{4}{ab} \sum_{m=1}^{m=\infty}\sum_{n=1}^{n=\infty}\biggl ( \sin\frac{m\pi x}{a}\sin\frac{n\pi y}{b} \int\limits_0^a d\lambda \int\limits_0^b f(\lambda,\mu) \sin\frac{m\pi \lambda}{a}\sin\frac{n\pi \mu}{b}d\mu\biggr ) .\tag{15} \] \begin{flalign*} &\text{Hence} & z=&\sum_{m=1}^{m=\infty}\sum_{n=1}^{n=\infty}\biggl ( A_{m,n} \sin\frac{m\pi x}{a}\sin\frac{n\pi y}{b} \cos c\pi t \sqrt{ \frac{m^2}{a^2}+\frac{n^2}{b^2} }\biggr ), && \tag{16}\\ &\text{where} &&\quad A_{m,n}=\frac{4}{ab} \int\limits_0^ad\lambda \int\limits_0^bf(\lambda,\mu)\sin\frac{m\pi \lambda}{a}\sin\frac{n\pi \mu}{b}d\mu. &&\tag{17} \end{flalign*} is our required solution. \EXAMPLE{S} 1.\quad 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 $D_{t}z= F(x,y)$ when $t = 0$ the solution is \[ z=\frac{1}{c\pi}\sum_{m=1}^{m=\infty}\sum_{n=1}^{n=\infty}\biggl ( A_{m,n} \dfrac{1}{\sqrt{ \dfrac{m^2}{a^2}+\dfrac{n^2}{b^2} }} \sin\frac{m\pi x}{a}\sin\frac{n\pi y}{b} \sin c\pi t \sqrt{ \frac{m^2}{a^2}+\frac{n^2}{b^2} }\biggr )\\[-2ex] \] \begin{flalign*} &\text{where} &A_{m,n}&=\frac{4}{ab} \int\limits_0^ad\lambda \int\limits_0^bF(\lambda,\mu)\sin\frac{m\pi \lambda}{a}\sin\frac{n\pi \mu}{b}d\mu. && \phantom{where} \end{flalign*} % -----File: 139.png 2.\quad If there is both initial distortion and initial velocity \begin{align*} % recast to fit line z = \dfrac{4}{ab}\sum_{m=1}^{m=\infty}\,\sum_{n=1}^{n=\infty}\sin\dfrac{m\pi x}{a} \sin\dfrac{n\pi y}{b}\biggl[&A_{m,n}\cos{c\pi t}\sqrt{\dfrac{m^2}{a^2}+\dfrac{n^2}{b^2}}\\ + &B_{m,n}\sin{c\pi t}\sqrt{\dfrac{m^2}{a^2} + \dfrac{n^2}{b^2}}\biggr]\\[-4ex] \end{align*} \begin{flalign*} &\text{where}\hfil &&A_{m,n} = \int\limits_0^a d\lambda\int\limits_0^bf (\lambda,\mu)\sin\dfrac{m\pi\lambda}{a}\sin\dfrac{n\pi\mu}{b}d\mu,\hfil\phantom{where}&&\\[-1ex] &\text{and}\hfil &B_{m,n} =& \dfrac{1}{c\pi\sqrt{\dfrac{m^2}{a^2}+\dfrac{n^2}{b^2}}} \int\limits_0^a d\lambda \int\limits_0^b F(\lambda,\mu)\sin\dfrac{m\pi\lambda}{a} \sin\dfrac{n\pi\mu}{b}d\mu.\hfil\phantom{and} \end{flalign*} 3.\quad 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. \mypara{72.} A number of interesting conclusions can be drawn from the results of Art.~71 and Exs.~1 and 2. (\emph{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. (\emph{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 in the problem of Art.~71 be such that $f(x,y) = A_{m,n} \sin\dfrac{m\pi x}{a} \sin\dfrac{n\pi y}{b}$, or the initial velocity in Ex.~1 be such that $F(x,y) = B_{m,n} \sin\dfrac{m\pi x}{a}\tablestrut\sin\dfrac{n\pi y}{b}$, or the initial distortion and initial velocity in Ex.~2 be the values just given, then the vibration will be periodic and will have the period \[\tag{1} T = \dfrac{2}{c\sqrt{\dfrac{m^2}{a^2}+\dfrac{n^2}{b^2}}}. \] 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 \[\tag{2} T_1 = \dfrac{2}{c\sqrt{\dfrac{1}{a^2}+\dfrac{1}{b^2}}}=\dfrac{2ab}{c\sqrt{a^2+b^2}} \] If $m$ and $n$ are both equal to $k$ we get \[\tag{3} T_k = \dfrac{2ab}{kc\sqrt{a^2+b^2}}; \] % -----File: 140.png therefore the membrane can be made to give any harmonic of its fundamental note. More than this, since as we have seen \[ T_{m,n}=\frac{2}{c\sqrt{\dfrac{m^2}{a^2}+\dfrac{n^2}{b^2}}} \] is the period of any note the membrane can give, and since if $m$ and $n$ are replaced by $mk$ and $nk$ we get \[ T_{mk,nk}= \frac{2}{ck\sqrt{\dfrac{m^2}{a^2}+\dfrac{n^2}{b^2}}} \] the membrane can sound all the harmonics of any note which it can give. (\emph{c}) In the case considered above, where the solution reduces to the single term \[ z=\sin\frac {m\pi x}{a}\sin \frac {n\pi y}{b} \left [A_{m,n}\cos c\pi t \sqrt{\frac{m^2}{a^2}+\frac{n^2}{b^2}}+B_{m,n}\sin c\pi t \sqrt{\frac{m^2}{a^2}+\frac{n^2}{b^2}} \right ], \] if $x = \dfrac{a}{m}$, or $\dfrac{2a}{m}$, or $\dfrac{3a}{m} \cdots$ or $ \dfrac{(m-1)a}{m}$, $z = 0$ for all values of $t$, and the lines $x = \dfrac{a}{m}$, $x = \dfrac{2a}{m}\tablestrut$, $\cdots$ $x = \dfrac{(m-1)a}{m}$ remain at rest during the whole vibration and are nodes. The same thing is true of the lines \[ y=\frac{b}{n},\; y=\frac{2b}{n},\; y=\frac{3b}{n},\; \cdots\; y=\frac{(n-1)b}{n}. \] \mypara{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 \[ T=\frac{2a}{c\sqrt{m^2+n^2}} \tag{1} \] and there will in general be two terms having the same period, and a musical note of the pitch corresponding to that period may be produced by initial circumstances that bring in both terms. Thus \begin{align*} z=&\sin\frac {m\pi x}{a}\sin \frac{ n\pi y}{a} \left [A_{m,n}\cos\frac{ c\pi t}{a} \sqrt{m^2+n^2}+B_{m,n}\sin \frac{c\pi t}{a} \sqrt{m^2+n^2} \right ]\\ &+ \sin\frac {n\pi x}{a}\sin \frac{ m\pi y}{a} \left [A_{n,m} \cos\frac{c\pi t}{a} \sqrt{m^2+n^2}+B_{n,m} \sin \frac{c\pi t}{a} \sqrt{m^2+n^2}\right ] \end{align*} % -----File: 141.png is a form of vibration that will give a musical note. Let us write this \begin{align*} z = &\cos \frac{c\pi t}{a} \sqrt{m^2+n^2}\bigg [ A \sin \frac{m\pi x}{a} \sin \frac{n\pi y}{a} + B \sin\frac{n\pi x}{a} \sin \frac{m\pi y}{a} \bigg ]\\ &+ \sin\frac{c\pi t}{a} \sqrt{m^2+n^2}\bigg [ C \sin \frac{m\pi x}{a} \sin \frac{n\pi y}{a} + D \sin\frac{n\pi x}{a} \sin \frac{m\pi y}{a} \bigg ] \tag{2} \end{align*} 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 \begin{multline*} z = \bigg ( A \sin\frac{m\pi x}{a} \sin \frac{n\pi y}{a} + B \sin\frac{n\pi x}{a} \sin\frac{m\pi y}{a} \bigg ) \biggl (\cos \frac{c\pi t}{a} \sqrt{m^2+n^2} \biggr .\\ \biggl .+ \sin \frac{c\pi t}{a} \sqrt{m^2+n^2} \biggr ). \tag{3} \end{multline*} Values of $x$ and $y$ that will reduce the first parenthesis in (3) to zero will correspond to points of the membrane remaining motionless during the vibration. Let us consider a few cases at length. (\emph{a})\quad If $m= 1$ and $n = 1$, the first parenthesis in (3) becomes \[ (A + B) \sin\frac{\pi x}{a} \sin \frac{\pi y}{a}, \] 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. (\emph{b})\quad If $m = 1$ and $n = 2$, we have \[ A \sin\frac{\pi x}{a} \sin \frac{2\pi y}{a} + B \sin\frac{2\pi x}{a} \sin \frac{\pi y}{a} = 0 \] to give the nodes. Let $B = 0$, then $\sin\dfrac{\pi x}{a} \sin \dfrac{2\pi y}{a} = 0$, which is satisfied by $y = \dfrac{a}{2}$; and in addition to the edges the line $y = \dfrac{a}{2}$ is at rest and is a node. If $A = 0$ \quad $x = \dfrac{a}{2}$ is a node. If $A = B$ \begin{gather*} \sin\frac{\pi x}{a} \sin \frac{2\pi y}{a} + \sin\frac{2\pi x}{a} \sin \frac{\pi y}{a} = 0\\[1ex] 2 \sin\frac{\pi x}{a} \sin \frac{\pi y}{a} \cos \frac{\pi y}{a} + 2 \sin\frac{\pi x}{a} \cos\frac{\pi x}{a} \sin \frac{\pi y}{a} = 0\\[1ex] \sin\frac{\pi x}{a} \sin \frac{\pi y}{a} \left(\cos \frac{\pi y}{a} + \cos \frac{\pi x}{a} \right) = 0. \end{gather*} % -----File: 142.png The first factor gives the four edges of the membrane. The second written equal to zero gives \begin{gather*} \cos \frac{\pi y}{a} = -\cos \frac{\pi x}{a} = \cos\left (\pi - \frac{\pi x}{a} \right )\\ \frac{\pi y}{a}=\pi-\frac{\pi x}{a}\\ x + y= a, \end{gather*} which is a diagonal of the square. If $B = - A$ \begin{gather*} \sin \frac{\pi x}{a}\sin \frac{2\pi y}{a} - \sin \frac{2\pi x}{a} \sin\frac{\pi y}{a}=0\\ \cos\frac{\pi y}{a} = \cos\frac{\pi x}{a}\\ x-y=0, \end{gather*} which is the other diagonal of the square. Other relations between $A $ and $B$ will give Trigonometric curves of the form \[ \cos \frac{\pi y}{a} =-\frac{B}{A} \cos \frac{\pi x}{a} \] 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 %[Illustration] \pngcent{010.png}{1288} % -----File: 143.png {}(\emph{c})\quad If $m = n = 2$ we have\\[-4ex] \pngrightl{011.png}{344}{10}{-24} \[ (A + B)\sin \frac{2\pi x}{a}\sin \frac{2\pi y}{a} = 0 \] to give the nodes, which are merely the lines \[ x = \frac{a}{2}, \quad \text{and} \quad y = \frac{a}{2}. \] This form gives the octave of the fundamental note. (\emph{d})\quad If $m = 1$ and $n = 3$ we have \[ A \sin \frac{\pi x}{a} \sin \frac{3\pi y}{a} + B \sin \frac{3\pi x}{a} \sin \frac{\pi y}{a} = 0 \] to give the nodes. \begin{flalign*} &\text{\indent If} \quad A = 0 \quad \text{we get} &x &= \frac{a}{3}\quad \text{and} \quad x = \frac{2a}{3} && \phantom{\indent If A = 0 we get} \tag{1} \\ &\text{\indent If} \quad B = 0 \quad \text{we get} &y &= \frac{a}{3} \quad \text{and} \quad y = \frac{2a}{3} && \tag{2} \end{flalign*} If \quad $A = -B$ \quad we get \begin{gather*} \sin \frac{\pi x}{a} \sin \frac{3\pi y}{a} - \sin \frac{3\pi x}{a} \sin \frac{\pi y}{a} = 0 \\[1ex] \sin \frac{\pi x}{a} \sin \frac{\pi y}{a} \left [4 \cos^2 \frac{\pi y}{a} - 1 - 4 \cos^2 \frac{\pi x}{a} + 1 \right ] = 0 \\[1ex] \cos^2 \frac{\pi y}{a} - \cos^2 \frac{\pi x}{a} = 0 \\[1ex] \left ( \cos \frac{\pi y}{a} - \cos \frac{\pi x}{a} \right ) \left ( \cos \frac{\pi y}{a} +\cos \frac{\pi x}{a} \right ) = 0 \\[-5ex] \end{gather*} \begin{flalign*} &\text{or}\hspace{1em} & x - y = 0 \quad &\text{and} \quad x + y = a. \tag{3} && \\[-5ex] \end{flalign*} \begin{flalign*} &\text{\indent If} \quad A = B \quad \text{we get} & \cos^2 \frac{\pi y}{a} &+\cos^2 \frac{\pi x}{a} = \frac{1}{2} && \phantom{\indent If \quad A = B \quad we get}\\[-6ex] \end{flalign*} \begin{flalign*} &\text{or}\hspace{1em} & \cos \frac{2\pi y}{a} +{}& \cos \frac{2\pi x}{a} = -1, \tag{4} && \end{flalign*} a Trigonometric curve easily constructed. For other relations between $A$ and $B$ we get more complicated Trigonometric curves coming under the general form \[ A \cos \frac{2\pi y}{a} + B \cos \frac{2\pi x}{a} = - \frac{A + B}{ 2} \tag{5} \] % -----File: 144.png which all agree in containing the points \[ \Bigl(\dfrac{a}{3}, \dfrac{a}{3}\Bigr), \Bigl(\dfrac{a}{3}, \dfrac{2a}{3}\Bigr), \Bigl(\dfrac{2a}{3}, \dfrac{a}{3}\Bigr), \text{ and } \Bigl(\dfrac{2a}{3}, \dfrac{2a}{3}\Bigr). \] %[Illustration] \pngcent{012.png}{1275} \newpage % -----File: 145.png \label{ch4end} \newpage \begin{center}{MISCELLANEOUS PROBLEMS.}\end{center} \markright{} \label{probstart} \begin{center}{I. \emph{Logarithmic Potential. Polar Coördinates.}}\end{center} \vspace{\baselineskip} 1.\quad Show that $D_x^2V+D_y^2V=0$ becomes \[ D_r^2V + \dfrac{1}{r}D_rV+ \dfrac{1}{r^2}D_\phi^2V = 0 \] if we transform to Polar Coördinates. \begin{flalign*} &\text{\indent 2.\quad If in} && D_r^2V + \dfrac{1}{r}D_rV + \dfrac{1}{r^2}D_\phi^2V = 0 && \phantom{\indent 2.\quad If in}(1) \end{flalign*} we let $V = R.\Phi$ we get \begin{align*} \begin{aligned} \Phi &= A \cos \alpha\phi + B \sin \alpha\phi\ \\ R&= A_1r^\alpha+B_1r^{-\alpha} \end{aligned} \bigg\} \quad\text{or}\quad \begin{aligned} \Phi &= A e^{\alpha\phi} + B e^{-\alpha\phi}\\ R& = A_1 \cos(\alpha \log r) + B_1 \sin(\alpha \log r);\ \end{aligned} \bigg\} \end{align*} whence\rule{0em}{1.3em} \[ \begin{array}{l | l | l} V = r^\alpha \cos \alpha\phi & V = e^{\alpha\phi} \cos(\alpha \log r) & V = \cosh \alpha\phi \cos(\alpha \log r)\\[0.5em] V = r^\alpha \sin \alpha\phi & V= e^{\alpha\phi} \sin(\alpha \log r) & V = \cosh \alpha\phi \sin(\alpha \log r)\\[0.5em] V = \dfrac{1}{r^\alpha} \cos \alpha\phi & V = e^{-\alpha\phi} \cos(\alpha \log r) & V = \sinh \alpha\phi \cos(\alpha \log r)\\[0.5em] V = \dfrac{1}{r^\alpha} \sin \alpha\phi & V = e^{-\alpha\phi}\sin(\alpha \log r) & V = \sinh \alpha\phi \sin(\alpha \log r) \end{array} \] are particular solutions of (1).\\ 3.\quad Show that if $V$ satisfies (1) Ex. 2 and $V = f(\phi)$ when $r = a$ \begin{flalign*} & &V &= \dfrac{1}{2} b_0 + \sum^{m=\infty}_{m=1} \left(\dfrac{r}{a}\right)^m (b_m \cos m\phi + a_m \sin m\phi)\quad\text{for}\quad r < a\rule{3em}{0em}\\[0.5em] &\text{and} & V &= \dfrac{1}{2} b_0 + \sum^{m=\infty}_{m=1} \left(\dfrac{a}{r}\right)^m (b_m \cos m\phi + a_m \sin m\phi)\quad\text{for}\quad r > a,\\[0.5em] &\text{where} &b_m& = \dfrac{1}{\pi} \int \limits ^\pi_{-\pi} f(\phi)\cos m\phi.d\phi\quad\text{and}\quad a_m = \dfrac{1}{\pi} \int \limits ^\pi_{-\pi} f(\phi) \sin m\phi.d\phi && \end{flalign*} % -----File: 146.png 4.\quad Show that if $V$ satisfies (1) Ex.~2 and $V=f(r)$ when $\phi=0$ and $r>0$ \begin{align*} V&=\frac{1}{\pi}\int\limits_{-\infty}^\infty f(e^\lambda)d\lambda \int\limits_0^\infty\frac{\cosh\alpha(\pi-\phi)}{\cosh\alpha\pi}\cos\alpha(\lambda-\log r).d\alpha\\ &=\frac{1}{\pi}\sin\frac{\phi}{2}\int\limits_{-\infty}^\infty f(e^\lambda) \frac{\cosh\dfrac{1}{2}(\lambda-\log r)} {\cosh(\lambda-\log r)-\cos\phi}d\lambda. \end{align*} \markright{MISCELLANEOUS PROBLEMS.} 5.\quad If $V=1$ when $\phi=0$ and $01$ \[ V=\frac{1}{\pi}\Bigg\{\frac{\pi}{2}-\tan^{-1}\Bigg[\frac{\sinh\dfrac{\log r}{2}}{\sin\dfrac{\phi}{2}}\Bigg]\Bigg\} =\frac{1}{\pi}\Bigg[\frac{\pi}{2}-\tan^{-1}\Bigg(\frac{r-1}{2\sqrt{r}.\sin\dfrac{\phi}{2}}\Bigg)\Bigg]. \] 6.\quad If $V=f(r)$ when $\phi=0$ and $V=0$ when $\phi=\beta$ \begin{align*} V &=\frac{1}{\pi}\int\limits_{-\infty}^\infty f(e^\lambda) d\lambda\int\limits_0^\infty\frac{\sinh(\beta-\phi)\alpha}{\sinh\beta\alpha}\cos\alpha(\lambda-\log r).d\alpha\\ &=\frac{1}{2\beta}\sin\frac{\pi\phi}{\beta}\int\limits_{-\infty}^\infty \frac{f(e^\lambda)d\lambda}{\cosh\dfrac{\pi}{\beta}(\lambda-\log r)-\cos\dfrac{\pi}{\beta}\phi}, \end{align*} if $0<\phi<\beta$.\\ 7.\quad If $V=0$ when $\phi=0$ and $V=F(r)$ when $\phi=\beta$ \begin{align*} V &=\frac{1}{\pi}\int\limits_{-\infty}^\infty F(e^\lambda)d\lambda\int\limits_0^\infty\frac{\sinh\phi\alpha}{\sinh\beta\alpha}\cos\alpha(\lambda-\log r).d\alpha\\ &=\frac{1}{2\beta}\sin\frac{\pi\phi}{\beta}\int\limits_{-\infty}^\infty \frac{F(e^\lambda)d\lambda}{\cosh\dfrac{\pi}{\beta}(\lambda-\log r)+\cos\dfrac{\pi}{\beta}\phi}. \end{align*} 8.\quad If $V=\chi(r)$ when $\phi=0$ and $ra &\\ &\text{where} & a_m &= \frac{2}{\beta}\int\limits_0^\beta f(\phi)\sin\frac{m\pi\phi}{\beta}d\phi \quad \text{and}\quad 0<\phi<\beta.& \end{flalign*} 12.\quad If $V=f(\phi)$ when $r=a$, $V=0$ when $r=b$, $V=0$ when $\phi=0$, and $V=0$ when $\phi=\beta$, then if $aa$.\\ 6.\quad 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\label{err149} \begin{multline*} % recast to fit line V = \dfrac{4}{ab} \sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1} e^{-\pi z \sqrt{ \frac{m^2}{a^2} + \frac{n^2}{b^2}}} \sin \dfrac{m\pi x}{a}\sin \dfrac{n\pi y}{b}\\ \int\limits^a_0 d\lambda \int\limits^b_0 f(\lambda,\,\mu) \sin \dfrac{m\pi\lambda}{a}\sin \dfrac{n\pi\mu}{b} d\mu. \end{multline*} If $V=1$ on the base of the prism this reduces to \[ V = \dfrac{16}{ \pi^2}\sum\limits^{m=\infty}_{m=0} \sum\limits^{n=\infty}_{n=0} e^{-\pi z \sqrt{ \frac{(2m+1)^2 }{a^2} + \frac{(2n+1)^2}{b^2}}}\:\dfrac{\sin\dfrac{ (2m+1)\pi x}{a} \sin \dfrac{(2n+1)\pi y}{ b}}{ (2m+1)(2n+1)}. \] 7.\quad If the value of the potential function on five faces of a rectangular parallelopiped, whose length, breadth, and height are $a$, $b$, and $c$, is zero, and % -----File: 150.png if the value of $V$ is given for every point of the sixth face, then for any point within the parallelopiped \[ V =\sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1} A_{m,n} \dfrac{\sinh \pi (c - z) \sqrt{\dfrac{ m^2}{a^2} + \dfrac{n^2}{ b^2}}}{\sinh \pi c \sqrt{\dfrac{ m^2 }{a^2} + \dfrac{n^2 }{ b^2}}} \sin \dfrac{m \pi x}{a} \sin \dfrac{n \pi y}{ b} \] \begin{flalign*} &\text{where} &A_{m,n}&= \dfrac{4}{ab} \int\limits^a_0 d\lambda \int\limits^b_0 f(\lambda,\;\mu) \sin \dfrac{m\pi\lambda}{a} \sin \dfrac{n\pi\mu}{b} d\mu.& \end{flalign*} 8.\quad 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 \begin{align*} V =& \sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1} A_{m,n} \dfrac{\sinh \pi (c - z) \sqrt{\dfrac{ m^2}{a^2} + \dfrac{n^2}{ b^2}}}{\sinh \pi c \sqrt{\dfrac{ m^2 }{a^2} + \dfrac{n^2 }{ b^2}}} \sin \dfrac{m \pi x}{a} \sin \dfrac{n \pi y}{ b}\\ &+ \sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1} B_{m,n} \dfrac{\sinh \pi z \sqrt{\dfrac{ m^2}{a^2} + \dfrac{n^2}{ b^2}}}{\sinh \pi c \sqrt{\dfrac{ m^2 }{a^2} + \dfrac{n^2 }{ b^2}}} \sin \dfrac{m \pi x}{a} \sin \dfrac{n \pi y}{ b} \end{align*} \begin{flalign*} &\text{where} &A_{m,n}&=\dfrac{4}{ab}\int\limits^a_0 d\lambda \int\limits^b_0 f (\lambda,\,\mu) \sin \dfrac{m\pi\lambda}{a}\sin \dfrac{n\pi\mu}{b} d\mu&\\ &\text{and} &B_{m,n}&=\dfrac{4}{ab}\int\limits^a_0 d\lambda \int\limits^b_0 F(\lambda,\,\mu) \sin \dfrac{m\pi\lambda}{a}\sin \dfrac{n\pi\mu }{b} d\mu.& \end{flalign*} 9.\quad 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? \vspace{1ex} \begin{center}{III. \textit{Conduction of Heat in a Plane.}}\end{center} 1.\quad Find particular solutions of $D_tu = a^2(D_x^2 u + D_y^2 u)$ of the forms \begin{align*} &u = e^{-a^2(\alpha^2 + \beta^2)t} \sin(\alpha x\pm \beta y)\\ &u = e^{-a^2(\alpha^2 + \beta^2)t} \cos(\alpha x\pm \beta y). \end{align*} 2.\quad Given the initial temperature of every point in a thin plane plate, find the temperature of any point at any time, % -----File: 151.png \begin{align*} u &= \dfrac{1}{4a^2 \pi t}\int\limits^{\infty}_{-\infty} d\lambda \int\limits^{\infty}_{-\infty} e^{-\frac{(\lambda-x)^2+(\mu -y)^2}{4a^2t}} f(\lambda,\, \mu)d\mu\\ &= \dfrac{1}{\pi}\int\limits^{\infty}_{-\infty}e^{-\beta^2}d\beta\int\limits^{\infty}_{-\infty}e^{-\gamma^2}f(x+2a\sqrt{t}.\beta,\,y+2a\sqrt{t}.\gamma)d\gamma. \end{align*} 3.\quad For an instantaneous \emph{source} of strength $Q$ at $(\lambda, \mu)$ \begin{flalign*} &\phantom{v.\ Art.~53.}&u &= \dfrac{Q}{ 4\pi a^2t}e^{-\frac{(\lambda-x)^2+(\mu -y)^2}{4a^2t}} & \text{v.\ Art.~53}. \end{flalign*} For an instantaneous \emph{doublet} of strength $P$ at $(0, \mu)$ with its axis perpendicular to the axis of $Y$ \begin{flalign*} &\phantom{v.\ Art.~54.}&u &= \dfrac{Px}{ 8\pi a^4t^2}e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} & \text{v.\ Art.~54}. \end{flalign*} For a permanent doublet of strength $P$ at $(0, \mu)$ with its axis perpendicular to the axis of $Y$ \[ u = \dfrac{P}{2\pi a^2} \frac{x}{x^2+(\mu-y)^2}e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} . \] If the strength of the doublet were $Pd\mu$ and the heat were uniformly generated and absorbed along the element $d\mu$ of the axis of $Y$ beginning at (0, $\mu$) we should have \[ u = \dfrac{P }{2\pi a^2} e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} \dfrac{xd\mu}{ x^2+(\mu-y)^2} = \dfrac{P }{ 2\pi a^2} e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} d \tan^{-1}\dfrac{ \mu-y}{x}, \] and since $d \tan^{-1}\dfrac{ \mu-y}{x}$ is the angle $ARA'$, where $A$ and $A'$ are the points (0, $\mu$) and (0, $\mu + d\mu$) and $R$ is the point $(x, y)$, $u = 0$ when $x = 0$ unless $\mu < y < \mu + d\mu$, in which case $u = \dfrac{P}{2a^2}$ if $x$ approaches zero from the positive side; and $u = 0$ when $t = 0$ except in the element $d\mu$. If then $u = 0$ when $t = 0$ and $u=f(y)$ when $x=0$ we have only to suppose a doublet of strength $2a^2f(x)dx$ placed in each element of the axis of $Y$ and then to integrate; we get \[ u = \dfrac{1}{ \pi} \int\limits^{\infty}_{-\infty}e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} \dfrac{xf(\mu)}{x^2+(\mu -y)^2}{}d\mu. \] For a permanent doublet of strength $F(t)$ at (0, $\mu$) we have \begin{align*} u &= \dfrac{x}{ 8\pi a^4} \int\limits^t_0 e^{-\frac{x^2+(\mu-y)^2}{4a^2(t-\tau)}} (t-\tau)^{-2}F(\tau)d\tau.\\ &= \dfrac{1}{2\pi a^2}\bigg[\dfrac{xF(0)}{x^2+(\mu-y)^2}e^{-\frac{x^2+(\mu-y)^2}{4a^2t}}+\int\limits^t_0\dfrac{xF'(\tau)}{x^2+(\mu-y)^2} e^{-\frac{x^2+(\mu-y)^2}{4a^2(t-\tau)}} d\tau\bigg]. \end{align*} % -----File: 152.png From the reasoning above this must be zero when $t = 0$ except at the point (0, $\mu$), must be $2a^2F(t)$ at the point (0, $\mu$), 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 $x = 0$ \[ u = \dfrac{1}{\pi} \!\int\limits^{\infty}_{-\infty}\! \dfrac{xF(\mu,0)}{x^2 + (\mu - y)^2} e^{-\frac{x^2+(\mu-y)^2}{4a^2t}} d\mu + \dfrac{1}{\pi} \!\int\limits^{\infty}_{-\infty}\! d\mu \int\limits^t_0 \dfrac{xD_\tau F(\mu,\tau)}{x^2 + (\mu - y)^2} e^{-\frac{x^2+(\mu-y)^2}{4a^2(t-\tau)}}d\tau. \] 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 corresponding problems in the flow of heat in three dimensions see E.~W. Hobson in Vol.\ XIX. Proc.\ Lond.\ Math.\ Soc.\\ 4.\quad If the perimeter of a thin plane rectangular plate is kept at the temperature zero and the initial temperatures of all points of the plate are given, then for any point of the plate \begin{multline*} % recast to fit line u = \dfrac{4 }{ bc} \sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1}e^{-a ^2\pi^2\left(\frac{m^2}{b^2} +\frac{ n^2}{c^2}\right)t} \sin \dfrac{m\pi x}{b} \sin \dfrac{n\pi y}{c}\\ \int\limits^b_0d\lambda \int\limits^c_0 f(\lambda,\mu) \sin \dfrac{m\pi\lambda}{b} \sin \dfrac{n\pi\mu}{c} d\mu. \end{multline*} if $b$ is the length and $c$ the breadth of the plate.\\ 5.\quad 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 $a^2 = .185$ in C.G.S. units. \hfill \emph{Ans}., 52°.9. 6.\quad 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. \hfill \emph{Ans}., 20°.8. \vspace{1ex} \begin{center}IV.~ \textit{Conduction of Heat in Space.}\end{center} 1.\quad Show that \begin{multline*} % recast to fit line \dfrac{1}{\pi^3}\int\limits^\infty_0 d\alpha \int\limits^\infty_0 d\beta \int\limits^\infty_0 d\gamma \int\limits^\infty_{-\infty} d\lambda \int \limits^\infty_{-\infty} d\mu \int\limits^\infty_{-\infty} f(\lambda,\mu,\nu) \\ \cos \alpha(\lambda - x) \cos \beta(\mu - y) \cos \gamma(\nu - z).d\nu = f(x,y,z) \end{multline*} for all values of $x$, $y$, and $z$.\\ 2.\quad Show that \begin{flalign*} &&f(x,y,z) = \sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1}\sum\limits^{p=\infty}_{p=1}A_{m,n,p} &\sin \dfrac{m\pi x}{a} \sin \dfrac{n\pi y}{b} \sin \dfrac{p\pi z}{c}&&\\ &\text{where}\quad & A_{m,n,p} = \dfrac{8}{abc}\displaystyle \int \limits^a_0 d\lambda \int \limits^b_0 d\mu \int \limits^c_0 f(\lambda,\mu,\nu) &\sin \dfrac{m\pi \lambda}{a} \sin \dfrac{n\pi \mu}{b} \sin \dfrac{p\pi \nu}{c} d\nu, \end{flalign*} for $0 < x < a$, $0 < y < b$, $0 < z < c$.\\ % -----File: 153.png 3.\quad Obtain particular solutions of $D_tu = a^2(D_x^2u + D_y^2u + D_z^2u)$ of the forms \begin{align*} u &= e^{-a^2(\alpha^2 + \beta^2 + \gamma^2)t} \sin(\alpha x \pm\beta y \pm \gamma z).\\ u &= e^{-a^2(\alpha^2 + \beta^2 + \gamma^2)t} \cos(\alpha x \pm \beta y \pm \gamma z). \end{align*} 4.\quad Given the initial temperature of every point in an infinite homogeneous solid find the temperature of any point at any time. \begin{align*} &u=\dfrac{1}{8a^3(\pi t)^{\frac{3}{2}}} \int\limits^{\infty}_{-\infty} d\lambda \int\limits^{\infty}_{-\infty} d\mu \int\limits^{\infty}_{-\infty} e^{-\frac{(\lambda-x)^2+(\mu-y)^2+(\nu-z)}{4a^2t}} f(\lambda,\mu,\nu)d\nu\\ &=\dfrac{1}{\pi^{\frac{3}{2}}} \!\int\limits^{\infty}_{-\infty}\! e^{-\beta^2} d\beta \!\int\limits^{\infty}_{-\infty}\! e^{-\gamma^2} d\gamma \!\int\limits^{\infty}_{-\infty}\! e^{-\delta^2} f(x+2a \sqrt{t}.\beta,\,y+2a \sqrt{t}.\gamma,\,z+2a \sqrt{t}.\delta)d\delta. \end{align*} 5.\quad 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 \[ u=\sum\limits^{m=\infty}_{m=1} \sum\limits^{n=\infty}_{n=1} \sum\limits^{p=\infty}_{p=1} A_{m,n,p}e^{-a^2\pi^2\left(\frac{m^2}{b^2}+\frac{n^2}{c^2}+\frac{p^2}{d^2}\right)t} \sin \dfrac{m\pi x}{b} \sin \dfrac{n\pi y}{c} \sin \dfrac{p\pi z}{d}\\[-5ex] \] \begin{flalign*} &\text{where}&A_{m,n,p}&=\dfrac{8}{bcd}\int \limits^b_0 d \lambda \int \limits^c_0 d\mu \int \limits^d_0 f(\lambda,\,\mu,\,\nu)\sin \dfrac{m\pi\lambda}{b} \sin \dfrac{n\pi\mu}{c} \sin \dfrac{p\pi\nu}{d} d\nu.& \end{flalign*} 6.\quad 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. \hfill \emph{Ans}., 38°.4.\\ 7.\quad 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. \hfill \emph{Ans.}, 9°.5. \label{probend} % -----File: 154.png \label{ch5start} \mychap{CHAPTER V.\footnotemark}{ZONAL HARMONICS.} \footnotetext{ Before reading this chapter the student is advised to re-read carefully articles 9, 10, 13(\emph{c}), 15, 16, and 18(\emph{c}).} \mypara{74.} In Art.~16 we obtained \[ z = Ap_m(x) + Bq_m(x) \tag{1} \] [v.\ (6) Art.~16] as the general solution of Legendre's Equation \[ (1-x^2)\dfrac{d^2z}{dx^2}-2x\dfrac{dz}{dx}+m(m+1)z = 0, \tag{2} \] $m$ being wholly unrestricted in value and $x$ lying between $-1$ and 1; where \begin{multline*} p_m(x)=1-\dfrac{m(m +1)}{2!} x^2 + \dfrac{m(m -2)(m + 1)(m + 3)}{4!} x^4\\ -\dfrac{m(m - 2)(m - 4)(m + 1)(m + 3)(m + 5)}{6!} x^6+ \cdots \tag{3} \end{multline*} and \begin{multline*} q_m(x) = x-\dfrac{(m-1)(m + 2)}{3!} x^3 + \dfrac{(m-1)(m-3)(m + 2)(m + 4)}{5!} x^5\\ -\dfrac{(m - 1)(m - 3)(m - 5)(m + 2)(m + 4)(m + 6)}{7!} x^7 +\cdots; \tag{4} \end{multline*} \begin{flalign*} \left.\!\! \begin{aligned} &\text{and we found}\hspace{5em}& V&=r^mp_m(\cos\theta)\\ && V&=\dfrac{1}{r^{m+1}} p_m(\cos\theta)\\ && V&=r^mq_m(\cos\theta)\\ && V&=\dfrac{1}{r^{m+1}} q_m(\cos\theta), \end{aligned} \right\}&&\hfill\tag{5} \end{flalign*} $m$ being unrestricted in value, as particular solutions of the special form assumed by Laplace's Equation in spherical coördinates when $V$ is independent of $\phi$; that is, of the equation \[ rD_r^2(rV) + \dfrac{1}{\sin\theta} D_\theta(\sin\theta D_\theta V) = 0. \tag{6} \] % -----File: 155.png For the important case where $m$ is a positive integer we found \[ z = AP_m(x) + BQ_m(x) \tag{7} \] [v.\ (10) Art.~16] as the general solution of Legendre's Equation (2), whence \[\left . \begin{aligned} V&= r^mP_m(\cos \theta)\\ V&=\frac{1}{r^{m+1}} P_m(\cos\theta)\\ V&=r^mQ_m(\cos \theta)\\ V&=\frac{1}{r^{m+1}} Q_m(\cos\theta) \end{aligned} \right \} \tag{8} \] are particular solutions of (6) if m is a positive integer. \begin{multline*} P_m(x)=\frac{(2m-1)(2m-3) \cdots 1}{m!} \left [x^m-\frac{m(m-1)}{2(2m-1)}x^{m-2} \right . \\ +\left . \frac{m(m-1)(m-2)(m-3)}{2.4.(2m-1)(2m-3) }x^{m-4} - \cdots \right ] \tag{9} \end{multline*} [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 $x^0$ if $m$ is even. \markright{ZONAL HARMONICS.} It is called a \textit{Surface Zonal Harmonic}, or a \textit{Legendre's Coefficient}, or more briefly a \textit{Legendrian}. \begin{multline*} Q_m(x)=\frac{m!}{(2m+1)(2m-1)\cdots 1} \biggl [\frac{1}{x^{m+1}} + \frac{(m+1)(m+2)}{2.(2m+3)} \frac{1}{x^{m+3}} \biggr . \\ + \biggl . \frac{(m+1)(m+2)(m+3)(m+4)}{2.4.(2m+3)(2m+5)} \frac{1}{x^{m+5}} + \cdots \biggr ] \tag{10} \end{multline*} if $x < -1$ or $x > 1.$ [v.\ (9) Art.~16.] It is called a \textit{Surface Zonal Harmonic} of the \textit{second kind}. \begin{align*} Q_m(x)&=(-1)^\frac{m+1}{2} \frac{2^{m-1}\Big[\Gamma\Big(\dfrac{m+1}{2}\Big)\Big]^2}{\Gamma(m+1)} p_m(x) \\ &=(-1)^\frac{m+1}{2} \frac{2.4.6.\cdots (m-1)}{3.5.7.\cdots m} p_m(x) \tag{11} \end{align*} [v.\ (13) Art.~16] if $m$ is odd and $-1 < x < 1$. \begin{align*} Q_m(x)&=(-1)^\frac{m}{2} \frac{2^m\Big[\Gamma\Big(\dfrac{m+1}{2}\Big)\Big]^2}{\Gamma(m+1)} q_m(x) \\ &=(-1)^\frac{m}{2} \frac{2.4.6.\cdots m}{1.3.5.\cdots (m-1)} q_m(x) \tag{12} \end{align*} [v.\ (14) Art.~16] if $m$ is even and $-1 < x < 1$. % -----File: 156.png In most of the work that immediately follows we shall regard $x$ in $P_m(x) $as equal to $\cos\theta$ and therefore as lying between $-1$ and 1.\footnote{English writers on Spherical Harmonics generally use $\mu$ in place of $x$ for $\cos \theta$. We shall follow them, however, only when we should thereby avoid confusion.} \mypara{75.} In Article 9 the undetermined coefficient $a_m$ of $x^m$ in $P_m(x)$ was arbitrarily written in the form $\dfrac{(2m-1)(2m-3)\cdots 1}{m!}$ for reasons which shall now be given. In Articles 9 and 16 $z = P_m(x)$ was obtained as a particular solution of Legendre's Equation \[ (1-x^2)\dfrac{ d^2z}{dx^2} - 2x\dfrac{ dz}{dx} + m(m+1)z = 0 \tag{1} \] by the device of assuming that z could be expressed as a sum or a series of terms of the form $a_nx^n$ 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 $(x_1, y_1, z_1)$ is \[ V= \dfrac{1}{\sqrt{ (x-x_1)^2 + (y-y_1)^2 + (z-z_1)^2}}\tag{2} \] and this must be a particular solution of Laplace's Equation \[ D_x^2 V+D_y^2V+D_z^2 V= 0, \tag{3} \] as is easily verified by direct substitution. If we transform (2) to spherical coördinates using the formulas of transformation \begin{flalign*} &&x&=r\cos\theta & \\ &&y&=r\sin\theta\cos\phi&\\ &&z&=r\sin\theta\sin\phi& \text{we get} \end{flalign*} \[ V = \dfrac{1 }{\sqrt{ r^2 - 2rr_1[\cos\theta \cos\theta_1 + \sin\theta \sin\theta_1 \cos(\phi-\phi_1)] + r_1^2}} \tag{4} \] as a solution of Laplace's Equation in Spherical Coördinates \begin{flalign*} &&&&rD_r^2(rV)+\dfrac{1}{\sin\theta} D_{\theta}(\sin\theta D_{\theta} V)+\dfrac{1}{\sin^2\theta} {D_\phi}^2V=0 \quad \text{ [{\scriptsize XIII}] Art.~1}. \end{flalign*} If the given point $(x_1, y_1, z_1)$ is taken on the axis of $X$, as it must be that (4) may be independent of $\phi$, $\theta_1 = 0$, and \[ V = \dfrac{1}{ \sqrt{ r^2-2rr_1 \cos\theta + r_1^2}}\tag{5} \] % -----File: 157.png is a solution of \[ rD_{r}^{2}(rV) + \dfrac{1}{\sin\theta} D_\theta(\sin\theta D_{\theta}V) = 0. \tag{6} \] Equation (5) may be written \begin{flalign*} & &V&=\frac{1}{r}\frac{1}{\sqrt {1-2 \dfrac{r_1}{r} \cos\theta + \dfrac{r_1^2}{r^2}} } \tag{7} && \\[1ex] &\text{or} &V&=\frac{1}{r_1}\frac{ 1}{\sqrt{ 1-2 \dfrac{r}{r_1} \cos\theta + \dfrac{r^2}{r_1^2}}}. \tag{8} && \end{flalign*} $\sqrt{ 1- 2z \cos \theta + z^{2}}$ 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 $1 - 2z \cos \theta + z^{2} = 0$ namely $z = \cos \theta + i \sin \theta$ and $z = \cos \theta - i\sin \theta$, both of which have the modulus unity and which are \emph{critical} values.\\ $\dfrac{1}{\sqrt{ 1- 2z\cos\theta + z^2}}$ is finite and continuous except for the values of $z = \cos \theta -i\sin \theta$ and $z=\cos\theta + i\sin\theta$ for which it becomes infinite; it is double-valued but has as critical values only these values of $z$. It is then \textit{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.) If then $r>r_1$ $\displaystyle\frac{1}{\sqrt{1-\dfrac{2r_1}{r}\cos \theta + \dfrac{r_1^2}{r^2}}}$ can be developed into a convergent series involving whole powers of $\dfrac{r_1}{r}$. Let $\displaystyle\sum p_m \dfrac{r_1^m}{r^m}$ be this series, $p_{m}$, of course, being a function of $\cos\theta$. Then \[ V = \frac{1}{r }\sum p_m \frac{r_1^m}{r^m} \] [v.\ (7)] is a solution of (6). Substitute this value of $V$ in (6) and we get \[ \sum \left [\frac{r_1^m}{r^{m+1}} m(m+1)p_m + \frac{r_1^m}{r^{m+1}} \frac{1}{\sin\theta} \frac{d}{d\theta} \left ( \sin\theta \frac{dp_m}{d\theta} \right ) \right ] = 0. \] As this must hold whatever the value of r provided $ r > r_1$ the coefficient of each power of $r$ must be zero, and hence the equation \[ \frac{1}{\sin \theta} \frac{d}{d\theta} \left (\sin \theta \frac{dp_{m}}{d\theta} \right )+m(m+1)p_{m}=0 \tag{9} \] must be true. % -----File: 158.png But as we have seen in Art.~9 the substitution of $x = \cos \theta$ in (9) reduces it to \[ (1 - x^2) \dfrac{d^2p_m}{ dx^2} - 2x \dfrac{dp_m }{ dx} + m(m + 1)p_m = 0, \] \begin{flalign*} &\text{and therefore} & z &= p_m & \phantom{and therefore} \end{flalign*} is a solution of Legendre's Equation (1). If $r < r_1$ $\dfrac{1 }{\sqrt{ 1 - \dfrac{2r}{r_1}\cos \theta + \dfrac{r^2}{r_1^2}} }$ can be developed into a convergent series involving whole powers of $\dfrac{r^2}{r_1^2}$. Let $\displaystyle \sum p_m \dfrac{r^m}{r_1^m}$ be this series. Then \[ V = \dfrac{1}{r_1}\sum p_m\dfrac{r^m}{r_1^m} \] (v.\ 8) is a solution of (6); substituting in (6) we get \[ \sum\left[\dfrac{r^m}{r_1^{m+1}} m(m + 1) p_m + \dfrac{r^m}{r_1^{m+1}}\dfrac{ 1}{\sin \theta} \dfrac{d}{d \theta} \left(\sin \theta \dfrac{dp_m}{d \theta}\right)\right] = 0, \] whence it follows as before that \[ z = p_m \] is a solution of Legendre's Equation. But $p_m$ is the coefficient of the $m$th power of $\dfrac{r}{r_1}$ in the development of $\left(1 - 2 \dfrac{r}{r_1} \cos \theta + \dfrac{r^2}{r_1^2}\right)^{-\frac{1}{2}}$ according to powers of $\dfrac{r}{r_1}$, or of the $m$th power of $\dfrac{r_1}{r }$ in the development of $\left(1 - 2 \dfrac{r_1}{r} \cos \theta + \dfrac{r_1^2}{r^2}\right)^{-\frac{1}{2}}$ according to powers of $\dfrac{r_1}{r }$, or more briefly it is the coefficient of the $m$th power of z in the development of $(1 - 2xz + z^2)^{-\frac{1}{2}}$ according to powers of $z$, $x$ standing for $\cos \theta$. \[ (1 - 2xz + z^2)^{-\frac{1}{2}} = [1 - z(2x - z)]^{-\frac{1}{2}} \] and can be developed by the Binomial Theorem; the coefficient of $z^m$ is easily picked out and is \begin{multline*} \dfrac{(2m - 1)(2m - 3)\cdots 1 }{ m!}\left[x^m - \dfrac{m(m - 1)}{ 2(2m - 1)} x^{m-2}\right.\\ \left.+\dfrac{m(m - 1)(m - 2)(m - 3)}{2.4.(2m - 1)(2m - 3)} x^{m-4} - \cdots\right]. \end{multline*} But this is precisely $P_m(x)$. [v.\ Art.~74 (9)] Hence $P_m(x)$ is equal to the coefficient of the $m$th power of $z$ in the development of $[1 - 2xz + z^2]^{-\frac{1}{2}}$ into a power series, the modulus of $z$ being less than unity. % -----File: 159.png \mypara{76.} If $x=1$\;\;$P_m(x)=1$. For if $x=1$\;\;$(1-2xz+z^2)^{-\frac{1}{2}}$ reduces to $(1-2z+z^2)^{-\frac{1}{2}}$ that is to $(1-z)^{-1}$, which develops into \[ 1 + z + z^2 + z^3 + z^4 + \cdots, \] and the coefficient of each power of $z$ is unity. Therefore \[ P_m(1) = 1. \tag{1} \] We have seen that if $m$ is even $P_m(x)$ contains only even powers of $x$ and terminates with the term involving $x^0$, that is with the constant term. The value of this constant term can be picked out from the formula for $P_m(x)$ [v.\ Art.~74 (9)]. It is $(-1)^{\frac{m}{2}}\dfrac{1.3.5.\cdots(m-1)} {2.4.6.\cdots m}$; or it can be found as follows:---It is clearly the value $P_m(x)$ assumes when $x=0$; it is, then, the coefficient of $z^m$ in the development of $(1+z^2)^{-\frac{1}{2}}$; but \[ (1+z^2)^{-\frac{1}{2}}=1 -\frac{1}{2}z^2 +\frac{1.3}{2.4}z^4 -\frac{1.3.5}{2.4.6}z^6 +\frac{1.3.5.7}{2.4.6.8}z^8 - \cdots \] and the coefficient of $z^m$, $m$ being an even number, is $(-1)^{\frac{m}{2}}\dfrac{1.3.5\cdots(m-1)} {2.4.6\cdots m}$. If $m$ is odd $P_m(x)$ 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 out from (9) Art.~74 and is $(-1)^{\frac{m-1}{2}}\dfrac{3.5.7.\cdots m} {2.4.6.\cdots(m-1)}$; or it can be found as follows:---It is clearly the value assumed by $\dfrac{dP_m(x)}{dx}$ when $x=0$. It is, then, the coefficient of $z^m$ in the development of $\dfrac{z}{(1+z^2)^{\frac{3}{2}}}$. \[ \frac{z}{(1+z^2)^{\frac{3}{2}}} = z -\frac{3}{2}z^3 +\frac{3.5}{2.4}z^5 -\frac{3.5.7}{2.4.6}z^7 +\cdots \] and the coefficient of $z^m$ in this development is $(-1)^{\frac{m-1}{2}}\dfrac{3.5.7\cdots m} {2.4.6\cdots(m-1)}$, $m$ being an odd number. \mypara{77.} To recapitulate: \begin{align*} P_m(x) ={}&\frac{1.3.5\cdots(2m-1)}{m!}\bigg[ x^m-\frac{m(m-1)}{2(2m-1)}x^{m-2}\\ &+\frac{m(m-1)(m-2)(m-3)}{2.4.(2m-1)(2m-3)}x^{m-4}\\ &-\frac{m(m-1)(m-2)(m-3)(m-4)(m-5)}{2.4.6.(2m-1)(2m-3)(2m-5)}x^{m-6} +\cdots\bigg], \tag{1} \end{align*} % -----File: 160.png $m$ being a positive integer, is a \textit{Surface Zonal Harmonic} or \textit{Legendrian} of the $m$th 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. $P_m(x)$ is the coefficient of the $m$th power of $z$ in the development of $(1 - 2xz + z^2)^{-\frac{1}{2}}$ into a power series. Hence if $z < 1$ \begin{multline*} (1 - 2xz + z^2)^{-\frac{1}{2}} = P_0(x) + P_1(x).z + P_2(x).z^2 + P_3(x).z^3\\ + P_4(x).z^4 + P_5(x).z^5 + \cdots + P_m(x).z^m + \cdots. \tag{2} \end{multline*} Whence \begin{flalign*} \left . \begin{aligned} \frac{1}{\sqrt{ r^2 - 2rr_1 \cos\theta + r_1^2}} = \frac{1}{ r } \left [ P_0 (\cos \theta) + \frac{r_1 }{ r }P_1 (\cos \theta) + \frac{r_1^2 }{r^2 }P_2 (\cos \theta) + \cdots \right . \\ + \left . \frac{r_1^m }{ r^m} P_m (\cos \theta) + \cdots \right ] \quad \text{if} \quad r > r_1 \\ = \frac{1}{ r_1} \left [P_0(\cos \theta) +\frac{ r }{ r_1} P_1(\cos \theta) +\frac{ r^2 }{ r_1^2} P_2(\cos \theta) + \cdots \right . \\ +\left . \frac{ r^m }{r_1^m} P_m(\cos \theta) + \cdots \right ]\;\text{ if } \;r < r_1. \end{aligned} \right \} (3) \end{flalign*} \[ z = P_m(x) \] is a solution of Legendre's Equation \[ (1 - x^2) \frac{d^2z }{ dx^2} - 2x \frac{dz}{ dx} + m(m + 1)z = 0 \] when $m$ is a positive integer. \begin{flalign*} & & V&= r^mP_m(\cos \theta) && \\ &\text{and} & V&= \frac{1}{ r^{m+1}} P_m (\cos \theta) && \phantom{and} \end{flalign*} are solutions of the form of Laplace's Equation in Spherical Coördinates which is independent of $\phi$, namely \begin{gather*} rD^2_r (rV) + \frac{1}{ \sin(\theta) }D_\theta (\sin \theta D_\theta V) = 0. \tag{4}\\ P_m(1) = 1. \tag{5}\\[1ex] P_{2m}(-x) = P_{2m}(x). \tag{6}\\[1ex] P_{2m+1} (-x) =-P _{2m+1}(x). \tag{7}\\[1ex] P_{2m+1}(0) = 0. \tag{8}\\ P_{2m}(0) = (-1)^m \frac{1.3.5. \cdots (2m - 1) }{2.4.6. \cdots 2m}. \tag{9}\\[1ex] % -----File: 161.png \left[\frac{dP_{2m+1}(x)}{dx}\right]_{x=0} =(-1)^m\frac{3.5.7.\cdots(2m+1)} {2.4.6.\cdots 2m}. \tag{10} \end{gather*} For convenience of reference we write out a few Zonal Harmonics. They are obtained by substituting successive integers for $m$ in formula (1). \[\left. \begin{aligned} P_0(x) &= 1\\[1ex] P_1(x) &= x\\ P_2(x) &= \frac{1}{2}(3x^2 - 1)\\ P_3(x) &= \frac{1}{2}(5x^3 - 3x)\\ P_4(x) &= \frac{1}{8}(35x^4 - 30x^2 + 3)\\ P_5(x) &= \frac{1}{8}(63x^5 - 70x^3 + 15x)\\ P_6(x) &= \frac{1}{16}(231x^6 - 315x^4 + 105x^2 - 5)\\ P_7(x) &= \frac{1}{16}(429x^7 - 693x^5 + 315x^3 - 35x)\\ P_8(x) &= \frac{1}{128}(6435x^8 - 12012x^6 + 6930x^4 - 1260x^2 + 35). \end{aligned}\right\} \tag{11} \] Any Surface Zonal Harmonic may be obtained from the two of next lower orders by the aid of the formula \[ (n+1)P_{n+1}(x)-(2n+1)xP_n(x)+nP_{n-1}(x)=0 \tag{12} \] which is easily obtained and is convenient when the numerical value of $x$ is given. Differentiate (2) with respect to $z$ and we get \[ \frac{-(z-x)}{(1-2xz+z^2)^{\frac{3}{2}}} =P_1(x) + 2P_2(x).z + 3P_3(x).z^2 + \cdots \] whence \[ \frac{-(z-x)}{(1- 2xz + z^2)^{\frac{1}{2}}} =(1-2xz+z^2)(P_1(x) + 2P_2(x).z + 3P_3(x).z^2 + \cdots). \] Hence by (2) \begin{align*} (1-2xz+z^2)(P_1(x) & +2P_2(x).z + 3P_3(x).z^2 + \cdots)\\ &+(z-x)(P_0(x)+P_1(x).z + P_2(x).z^2 + \cdots) = 0 \tag{13} \end{align*} % -----File: 162.png (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.\footnote {For tables of Surface Zonal Harmonics v.\ Appendix Tables I and II.} \mypara{78.} We are now able to solve completely the problem considered in Art.~9. We were to find a solution of the differential equation \[ rD_r^2(rV) + \frac{1}{\sin\theta}D_\theta(\sin\theta D_\theta V) = 0 \tag{1} \] subject to the condition \[ V = \frac{M}{(c^2 + r^2)^\frac{1}{2}}\quad \text{when}\quad \theta = 0. \tag{2} \] We know (v.\ Art.~77) that \begin{flalign*} &&V = r^mP_m(\cos\theta)&&\\ &\text{and} & V=\frac{1}{r^{m+1}}P_m(\cos\theta)&&&\phantom{and} \end{flalign*} are solutions of (1). For values of $rc$ \begin{align*} \frac{M}{(c^2+r^2)^\frac{1}{2}} &= \frac{M}{r} \Bigl[1-\frac{1}{2}\frac{c^2}{r^2} +\frac{1.3}{2.4}\frac{c^4}{r^4} -\frac{1.3.5}{2.4.6}\frac{c^6}{r^6} +\cdots\Bigr] \tag{5}\\ &=M\Bigl[\frac{1}{r}-\frac{1}{2}\frac{c^2}{r^3} +\frac{1.3}{2.4}\frac{c^4}{r^5} -\frac{1.3.5}{2.4.6}\frac{c^6}{r^7} +\cdots\Bigr]. \end{align*} % -----File: 163.png Therefore for values of $r>c$ \begin{align*} V=\frac{M}{c}\Big[\frac{c}{r}P_0(\cos\theta)-\frac{1}{2}&\frac{c^3}{r^3}P_2(\cos\theta)\\ &+\frac{1.3}{2.4}\frac{c^5}{r^5}P_4(\cos\theta) -\frac{1.3.5}{2.4.6}\frac{c^7}{r^7}P_6(\cos\theta) +\cdots\Big] \tag{6} \end{align*} is our required solution. For it satisfies (1) and reduces to (2) when $\theta=0$. \mypara{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 \[ \sigma = \frac{M}{4a\pi\sqrt{a^2-r^2}} \] and all points of the conductor are at the potential $\dfrac{\pi M}{2a}$. (v.\ Peirce's Newtonian 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 \begin{align*} V &= \frac{M}{a}\int\limits_0^a\frac{r dr}{\sqrt{(a^2-r^2)(x^2+r^2)}}\\ &= \frac{M}{2a}\cos^{-1}\frac{x^2-a^2}{x^2+a^2}. \end{align*} \begin{alignat*}{2} &\frac{d}{dx}\Bigl(\frac{M}{2a}\cos^{-1}\frac{x^2-a^2}{x^2+a^2}\Bigr) &&= -\frac{M}{a^2+x^2}\\ &&&= -\frac{M}{a^2}\Bigl[1- \frac{x^2}{a^2}+\frac{x^4}{a^4}-\frac{x^6}{a^6}+\cdots\Bigr]\\ &\text{if} \quad xa. \end{alignat*} Integrating and then determining the arbitrary constant we have \begin{alignat*}{2} &\frac{M}{2a}\cos^{-1}\frac{x^2-a^2}{x^2+a^2} &&= \frac{M}{a}\Bigl[\frac{\pi}{2}-\frac{x}{a} +\frac{x^3}{3a^3}-\frac{x^5}{5a^5}+\frac{x^7}{7a^7}-\cdots\Big]\\[1ex] &\quad\text{if} \quad xa. \end{alignat*} % -----File: 164.png We have, then, to solve the equation \[ rD^2_r(rV)+\frac{1}{\sin\theta}D_\theta(\sin\theta D_\theta V)=0 \] subject to the conditions \begin{flalign*} &&V&=\frac{M}{a}\bigg[\frac{\pi}{2}-\frac{r}{a} +\frac{r^3}{3a^3}-\frac{r^5}{5a^5} +\frac{r^7}{7a^7}-\cdots\bigg]&\\ &&&\text{when}\quad \theta=0\quad \text{and}\quad ra.& \end{flalign*} The required solution is easily seen to be \begin{flalign*} &&V&=\frac{M}{a}\bigg[ \frac{\pi}{2}-\frac{r}{a}P_1(\cos\theta) +\frac{1}{3}\frac{r^3}{a^3}P_3(\cos\theta) -\frac{1}{5}\frac{r^5}{a^5}P_5(\cos\theta)+\cdots\bigg] \intertext{if $ra$.} \end{flalign*} \EXAMPLE{S} 1.\quad Given that if a charge $M$ of electricity is placed on an ellipsoidal conductor the surface density at any point $P$ of the conductor is equal to $\dfrac{Mp}{4\pi abc}$, 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 the rotation of the ellipse $\dfrac{x^2}{a^2}+\dfrac{y^2}{b^2}=1$ about its minor axis. \textit{Ans.} (1) If the point is on the axis of revolution \[ V=\frac{M}{2\sqrt{a^2-b^2}}\bigg[\sin^{-1}\bigg(\frac{bx+a^2-b^2}{a\sqrt{x^2+a^2-b^2}}\bigg) -\sin^{-1}\bigg(\frac{bx-a^2+b^2}{a\sqrt{x^2+a^2-b^2}}\bigg)\bigg] \] $x$ being the distance from the centre. (2) If the point is on the surface of the spheroid \[ V=\frac{M}{2\sqrt{a^2-b^2}}\bigg[\frac{\pi}{2}-\sin^{-1}\bigg(\frac{2b^2-a^2}{a^2}\bigg)\bigg] =\frac{M}{ \sqrt{a^2-b^2}}\bigg[\frac{\pi}{2}-\tan^{-1}\bigg(\frac{b}{\sqrt{a^2-b^2}}\bigg)\bigg]. \] % -----File: 165.png (3) If the distance $r$ of the point from the centre is less than $\sqrt{ a^2 - b^2}$ and $\theta < \dfrac{\pi}{2}\tablestrut$ \begin{multline*} V = \frac{M }{\sqrt{ a^2 - b^2}} \left [ \frac{\pi }{2} - \frac{r }{(a^2-b^2)^{\frac{1}{2}}} P_1(\cos \theta) \right . \\ \left . + \frac{r^3}{ 3(a^2-b^2)^{\frac{3}{2}} }P_3(\cos \theta) - \frac{r^5 }{5(a^2-b^2)^{\frac{5}{2}}} P_5(\cos \theta)+ \cdots \right ]. \end{multline*} (4) If the distance $ r$ of the point from the centre is greater than $\sqrt{ a^2 - b^2}$ \begin{multline*} V = \frac{M }{\sqrt{ a^2 - b^2}} \biggl [ \frac{(a^2-b^2)^{\frac{1}{2}}}{r} - \frac{(a^2-b^2)^{\frac{3}{2}}}{3r^3} P_2(\cos \theta) \biggr .\\ \biggl . + \frac{(a^2-b^2)^{\frac{5}{2}}}{5r^5} P_4(\cos \theta) - \frac{(a^2-b^2)^{\frac{7}{2}}}{7r^7}P_6(\cos \theta) + \cdots \biggr ]. \end{multline*} 2.\quad If the conductor is the prolate spheroid generated by the rotation of the ellipse $\dfrac{x^2}{a^2} + \dfrac{y^2}{b^2} = 1$ about its major axis, show that if the point is an external point and is on the axis at a distance $x$ from the centre, \[ V = \frac{M }{2\sqrt{ a^2 - b^2}} \log\frac{ x + \sqrt{ a^2 - b^2}} {x - \sqrt {a^2 - b^2}}. \] If the point is not on the axis and $r > \sqrt{a^2 - b^2}$ \begin{multline*} V = \frac{M }{\sqrt{ a^2 - b^2}} \biggl [ \frac{(a^2-b^2)^{\frac{1}{2}}}{r} + \frac{(a^2-b^2)^{\frac{3}{2}}}{3r^3} P_2(\cos \theta) \biggr .\\ \biggl . + \frac{(a^2-b^2)^{\frac{5}{2}}}{5r^5} P_4(\cos \theta) + \frac{(a^2-b^2)^{\frac{7}{2}}}{7r^7}P_6(\cos \theta) + \cdots \biggr ]. \end{multline*} \mypara{80.} As a third example we will find the value of the potential function due to a thin homogeneous circular disc, of density $\rho$, 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 \[ V_0 = 2 \pi \rho k(\sqrt{x^2 + a^2} - x) = \frac{2M}{ a^2} [\sqrt{x^2 + a^2} - x]. \] If $x>a$ \begin{flalign*} % recast to fit line &&\sqrt{x^2+a^2} &= x \left (1+\frac{a^2}{x^2} \right )^{\frac{1}{2}} \\ && &= x \left [1+\frac{1}{2} \frac{a^2}{x^2} - \frac{1.1}{2.4} \frac{a^4}{x^4}+ \frac{1.1.3}{2.4.6} \frac{a^6}{x^6} - \frac{1.1.3.5}{2.4.6.8} \frac{a^8}{x^8} + \cdots \right ] &&\\ &\text{and} &V_0 &=\frac{ 2M}{a} \left [ \frac{1}{2} \frac{a}{x} - \frac{1.1}{2.4} \frac{a^3}{x^3} + \frac{1.1.3}{2.4.6} \frac{a^5}{x^5} - \frac{1.1.3.5}{2.4.6.8} \frac{a^7}{x^7} + \cdots \right ]. && \end{flalign*} % -----File: 166.png If $x < a$ \begin{flalign*} % recast to fit line &&\sqrt{x^2+a^2} &= a\left(1+\dfrac{x^2}{a^2}\right)^{\frac{1}{2}}\\ &&& = a\left[1+\dfrac{1}{2} \dfrac{x^2}{a^2} - \dfrac{1.1}{2.4}\dfrac{ x^4}{a^4} + \dfrac{1.1.3}{2.4.6} \dfrac{x^6}{a^6} +\cdots\right]&\\ &\text{and}& V_0 = \dfrac{2M}{a}{}&\left[1 -\dfrac{ x}{a} + \dfrac{1}{2} \dfrac{x^2}{a^2} - \dfrac{1.1}{2.4}\dfrac{ x^4}{a^4} + \dfrac{1.1.3}{2.4.6} \dfrac{x^6}{a^6} - \dfrac{1.1.3.5}{2.4.6.8}\dfrac{ x^8}{a^8} + \cdots \right].& \end{flalign*} Hence the solution for any external point is \begin{multline*} V = \dfrac{2M}{a} \left[\dfrac{1}{2} \dfrac{a}{r} - \dfrac{1.1}{2.4} \dfrac{a^3}{r^3} P_2(\cos\theta)\right.\\ +\left. \dfrac{1.1.3}{2.4.6} \dfrac{a^5}{r^5}P_4(\cos\theta) - \dfrac{1.1.3.5}{2.4.6.8}\dfrac{ a^7}{r^7}P_6(\cos\theta) + \cdots \right] \end{multline*} if $r > a$, and \begin{multline*} V = \dfrac{2M}{a}\left[1-\dfrac{r}{a}P_1(\cos\theta)\right.\\ +\left.\dfrac{1}{2} \dfrac{r^2}{a^2}P_2(\cos\theta) - \dfrac{1.1}{2.4}\dfrac{ r^4}{a^4} P_4(\cos\theta) + \dfrac{1.1.3}{2.4.6} \dfrac{r^6}{a^6}P_6(\cos\theta) -\cdots\right] \end{multline*} if $r < a$ and $\theta < \dfrac{\pi}{2}$. \EXAMPLE{S} 1.\quad The potential function due to a homogeneous hemisphere whose axis is taken as the polar axis, is \begin{multline*} V = \dfrac{M}{a} \left[\dfrac{a}{r} + \dfrac{3.1}{2.4} \dfrac{a^2}{r^2}P_1(\cos\theta) \right.\\ -\left. \dfrac{3.1.1}{2.4.6} \dfrac{a^4}{r^4} P_3(\cos\theta) + \dfrac{3.1.1.3}{2.4.6.8} \dfrac{a^6}{r^6}P_5(\cos\theta) -\cdots \right] \end{multline*} if $r > a$, and is \begin{multline*} V = \dfrac{M}{a} \left[\dfrac{3}{2} + \dfrac{3}{2} \dfrac{r}{a}P_1(\cos\theta) + \dfrac{r^2}{a^2}P_2(\cos\theta)\right.\\ +\left. \dfrac{3.1}{2.4}\dfrac{ r^3}{a^3} P_3(\cos\theta) - \dfrac{3.1.1}{2.4.6} \dfrac{r^5}{a^5}P_5(\cos\theta)+ \cdots \right] \end{multline*} if $r < a$ and $\theta > \dfrac{\pi}{2}$.\\ 2.\quad The potential function due to a solid sphere whose density is proportional to the distance from a diametral plane is, at an external point, \begin{multline*} V = \dfrac{8}{15}\dfrac{ M}{a} \left[\dfrac{5.3}{2.4}\dfrac{ a}{r} + \dfrac{5.3.1}{2.4.6} \dfrac{a^3}{r^3}P_2(\cos\theta)\right.\\ -\left. \dfrac{5.3.1.1}{2.4.6.8} \dfrac{a^5}{r^5} P_4(\cos\theta) + \dfrac{5.3.1.1.3}{2.4.6.8.10}\dfrac{ a^7}{r^7}P_6(\cos\theta) - \cdots \right]. \end{multline*} % -----File: 167.png 3.\quad The potential function due to the homogeneous oblate spheroid generated by the rotation of $\dfrac{x^2}{a^2} + \dfrac{y^2}{b^2} = 1$ about its minor axis is, at an external point, \begin{multline*} V = \frac{3}{2}\frac{ M}{(a^2-b^2)} \left [ \frac{x^2+a^2-b^2 }{ 2(a^2-b^2)^{\frac{1}{2}}} \left ( \sin^{-1} \frac{(a^2-b^2+bx)}{a \sqrt{ x^2+a^2-b^2}} \right . \right . \\ \left . \left . + \sin^{-1} \frac{(a^2-b^2-bx)}{a \sqrt{ x^2+a^2-b^2}} \right ) - x \right ] \end{multline*} if the point is on the axis of the spheroid at a distance $x$ from its centre. \begin{multline*} V = \frac{ 3M}{(a^2-b^2)^{\frac{1}{2} }} \left [\frac{1}{1.3}\frac{ (a^2-b^2)^{\frac{1}{2}}}{ r} - \frac{1}{3.5}\frac{ (a^2-b^2)^{\frac{3}{2}}} {r^3 }P_2(\cos \theta) \right . \\ \left . + \frac{1}{5.7}\frac{(a^2-b^2)^{\frac{5}{2}}} {r^5} P_4(\cos \theta) - \cdots \right ] \end{multline*} if $r > (a^2 - b^2)^{\frac{1}{2}}$, and \begin{multline*} V = \frac{3M}{(a^2-b^2)^{\frac{1}{2}}} \left [ \frac{\pi}{4} -\frac{ r}{(a^2-b^2)^{\frac{1}{2}}} P_1(\cos \theta) + \frac{\pi}{4} \frac{r^2}{(a^2-b^2)} P_2(\cos \theta) \right .\\ \left . - \frac{1}{1.3} \frac{r^3}{(a^2-b^2)^{\frac{3}{2}}} P_3(\cos \theta) + \frac{1}{3.5}\frac{ r^5}{(a^2-b^2)^{\frac{5}{2}}} P_5(\cos \theta) - \cdots \right ] \end{multline*} if $r<(a^2-b^2)^{\frac{1}{2}}$ and $\theta < \dfrac{\pi }{2}$.\\ 4.\quad The potential function due to the homogeneous prolate spheroid generated by the rotation of $\dfrac{x^2}{a^2} + \dfrac{y^2}{b^2} = 1$ about its major axis is, at an external point, \begin{gather*} V = \frac{3M}{(a^2-b^2)^{\frac{1}{2}}} \left [\frac{1}{1.3}\frac{ (a^2-b^2)^{\frac{1}{2}}}{r } + \frac{1}{3.5}\frac{ (a^2-b^2)^{\frac{3}{2}}}{r^3} P_2(\cos \theta) \right . \\ +\left . \frac{1}{5.7}\frac{ (a^2-b^2)^{\frac{5}{2}}}{ r^5} P_4(\cos \theta) + \cdots \right ] \end{gather*} if $r>(a^2 -b^2)^{\frac{1}{2}}$. \mypara{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 \[ rD^2_r (rV) + \frac{1}{\sin\theta }D_\theta (\sin \theta D_\theta V) = 0, \] 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 % -----File: 168.png 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 $P_m(\cos\theta)$ 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 \textit{absolutely convergent}.\ \mypara{82.} In addition to the form given in (1) Art.~77 for $P_m(x)$ other forms are often useful. It ought to be possible to develop $P_m(\cos\theta)$, which may be regarded as a function of $\theta$, 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 $\theta$ may be obtained much more easily by the following device. We have seen in Art.~75 that $P_m(\cos\theta)$ is the coefficient of the $m$th power of $z$ in the development of $(1 - 2z \cos\theta + z^2)^{-\frac{1}{2}}$ in a power series, and that if mod $z < 1$\;\;$(1 - 2z~\cos\theta + z^2)^{-\frac{1}{2}}$ 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. \begin{align*} (1 - 2z\cos\theta + z^2)^{-\frac{1}{2}} &= (1-z(e^{\theta i} + e^{-\theta i}) + z^2)^{-\frac{1}{2}}\\ &= (1-ze^{\theta i})^{-\frac{1}{2}}(1-ze^{-\theta i})^{-\frac{1}{2}}. \end{align*} $(1-ze^{\theta i})^{-\frac{1}{2}}$ may be developed into an absolutely convergent series if mod $z <1$, by the Binomial Theorem. We have \[ (1-ze^{\theta i})^{-\frac{1}{2} }= 1 + \frac{1}{2} ze^{\theta i} + \frac{1.3}{2.4} z^2e^{2\theta i } + \frac{1.3.5}{2.4.6} z^3e^{3\theta i} + \frac{1.3.5.7}{2.4.6.8 }z^4e^{4\theta i }+ \cdots\\[-4ex] \] \begin{multline*} (1-ze^{-\theta i})^{-\frac{1}{2} }= 1 + \frac{1}{2} ze^{-\theta i} + \frac{1.3}{2.4} z^2e^{-2\theta i } \\ + \frac{1.3.5}{2.4.6} z^3e^{-3\theta i} + \frac{1.3.5.7}{2.4.6.8 }z^4e^{-4\theta i }+ \cdots \end{multline*} The product of these series will give a development for $(1-2z\cos\theta+z^2)^{-\frac{1}{2}}$ in power series. The coefficient of $z^m$ is easily picked out, and must be equal to $P_m(\cos\theta)$. We thus get \begin{align*} % recast to fit line P_m(\cos\theta)={}&\frac{ 1.3.5. \cdots (2m-1) }{ 2.4.6.\cdots 2m} \bigg[e^{m\theta i} + e^{-m\theta i} \\ &\quad + \frac{1}{2}.\frac{2m}{2m-1}(e^{(m-2)\theta i} + e^{-(m-2)\theta i}) \\ &\quad + \frac{1.3}{2.4}.\frac{ 2m(2m-2) }{ (2m-1)(2m-3)} (e^{(m-4)\theta i} + e^{-(m-4)\theta i}) + \cdots \bigg]\\[1ex] % -----File: 169.png P_m(\cos\theta)={}&\frac{1.3.5\cdots(2m-1)}{2.4.6.\cdots 2m} \bigg[2\cos m\theta + 2\frac{1.m}{1.(2m-1)}\cos(m-2)\theta\\ &\quad +2\frac{1.3\,m(m-1)}{1.2(2m-1)(2m-3)}\cos(m-4)\theta\\ &\quad +2\frac{1.3.5}{1.2.3}\frac{m(m-1)(m-2)}{(2m-1)(2m-3)(2m-5)}\cos(m-6)\theta +\cdots\bigg]. \tag{1} \end{align*} If $m$ is odd the development runs down to $\cos\theta$; 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 $\bigg[\dfrac{1.3.5\cdots(m-1)}{2.4.6.\cdots m}\bigg]^2$. We write out the values of $P_m(\cos\theta)$ for a few values of $m$ \[\left. \begin{aligned} P_0(\cos\theta) &=1\\[1ex] P_1(\cos\theta) &=\cos\theta\\ P_2(\cos\theta) &=\frac{1}{4}(3\cos 2\theta+1)\\ P_3(\cos\theta) &=\frac{1}{8}(5\cos 3\theta+3\cos\theta)\\ P_4(\cos\theta) &=\frac{1}{64}(35\cos 4\theta+20\cos 2\theta+9)\\ P_5(\cos\theta) &=\frac{1}{128}[63\cos 5\theta+35\cos 3\theta+30\cos\theta]\\ P_6(\cos\theta) &=\frac{1}{512}[231\cos 6\theta+126\cos 4\theta+105\cos 2\theta+50]\\ P_7(\cos\theta) &=\frac{1}{1024}[429\cos 7\theta+231\cos 5\theta+189\cos 3\theta+175\cos\theta]\\ P_8(\cos\theta) &=\frac{1}{16384}[6435\cos 8\theta+3432\cos 6\theta+2772\cos 4\theta\\ &\qquad\qquad +2520\cos 2\theta+1225]. \end{aligned}\right\} \tag{2} \] 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 $P_m(\cos\theta)$ has its maximum value when $\theta=0$; but we have shown in Art.~76 that $P_m(1)=1$. Therefore $P_m(\cos\theta)$ is never greater than unity if $\theta$ is real. It is also easily seen from (1) that $P_m(\cos\theta)$ can never be less than $-1$. % -----File: 170.png \mypara{83.} $P_m(x)$ can be very simply expressed as a derivative. We have \begin{multline*} \hspace{12pt}P_m(x)=\frac{(2m-1)(2m-3)\cdots 1}{m!}\bigg[x^m-\frac{m(m-1)}{2.(2m-1)}x^{m-2}\hspace{8em}\\ \shoveright{+\frac{m(m-1)(m-2)(m-3)}{2.4.(2m-1)(2m-3)}x^{m-4}-\cdots\bigg]\phantom{.}\quad}\\ % \shoveleft{\int\limits_0^x P_m(x)dx=\frac{(2m-1)(2m-3)\cdots 1}{(m+1)!}\bigg[x^{m+1}-\frac{(m+1)m}{2.(2m-1)}x^{m-1}}\\ \shoveright{+\frac{(m+1)m(m-1)(m-2)}{2.4.(2m-1)(2m-3)}x^{m-3}-\cdots\bigg]}\\ % \shoveleft{\int\limits_0^x{}^2 P_m(x)dx^2= \int\limits_0^x dx \int\limits_0^x P_m(x)dx}\\ \shoveleft{\hspace{62pt}=\frac{(2m-1)(2m-3)\cdots 1}{(m+2)!}\bigg[x^{m+2}-\frac{(m+2)(m+1)}{2.(2m-1)}x^m}\\ \shoveright{+\frac{(m+2)(m+1)m(m-1)}{2.4.(2m-1)(2m-3)}x^{m-2}-\cdots\bigg]}\\ % \shoveleft{\int\limits_0^x{}^m P_m(x)dx^m =\frac{(2m-1)(2m-3)\cdots 1}{(2m)!}\bigg[x^{2m}-\frac{2m(2m-1)}{2(2m-1)}x^{2m-2}}\\ \shoveright{+\frac{2m(2m-1)(2m-2)(2m-3)}{2.4.(2m-1)(2m-3)}x^{2m-4}-\cdots\bigg]}\\ \shoveleft{\hspace{68pt}=\frac{(2m-1)(2m-3)\cdots 1}{(2m)!}\bigg[x^{2m}-mx^{2m-2}+\frac{m(m-1)}{2!}x^{2m-4}}\\ \shoveright{-\frac{m(m-1)(m-2)}{3!}x^{2m-6}+\cdots\bigg].\quad}\\[-4ex] \end{multline*} \indent The quantity in brackets obviously differs from $(x^2-1)^m$ by terms involving lower powers of $x$ than the $m$th. \begin{flalign*} &\text{Hence} & P_m(x)&=\frac{1.3.5\cdots(2m-1)}{(2m)!}\frac{d^m}{dx^m}(x^2-1)^m,&&\phantom{Hence}\\ &\text{or} && P_m(x)=\frac{1}{2^mm!}\frac{d^m}{dx^m}(x^2-1)^m. & \tag{1}& \end{flalign*} This important formula is entirely general and holds not merely when $x=\cos\theta$, but for all values of $x$. % -----File: 171.png \mypara{84.} The last result is so important that it is worth while to confirm it by obtaining it directly from Legendre's Equation \[ (1-x^2)\frac{d^2z}{dx^2}-2x\frac{dz}{dx}+m(m+1)z=0 \tag{1} \] v.\ (1) Art.~75. Let us differentiate (1) with respect to $x$ a few times representing $\dfrac{dz}{dx}$ by $z'$, $\dfrac{d^2z}{dx^2}$ by $z''$, $\dfrac{d^3z}{dz^3}$ by $z'''$, \&c. We get \begin{gather*} (1-x^2)\frac{d^2z'}{dx^2}-2.2x\frac{dz'}{dx}+[m(m+1)-2]z'=0,\\ (1-x^2)\frac{d^2z''}{dx^2}-2.3x\frac{dz''}{dx}+[m(m+1)-2(1+2)]z''=0,\\ (1-x^2)\frac{d^2z'''}{dx^2}-2.4x\frac{dz'''}{dx}+[m(m+1)-2(1+2+3)]z'''=0, \end{gather*} and in general \[ (1-x^2)\frac{d^2z^{(n)}}{dx^2}-2(n+1)x\frac{dz^{(n)}}{dx}+[m(m+1)-2(1+2+3+\cdots+n)]z^{(n)}=0\\[-3ex] \] \begin{flalign*} &\text{or} & (1-x^2)\frac{d^2z^{(n)}}{dx^2}-2(n+1)x\frac{dz^{(n)}}{dx}+m(m+1)-n(n+1)]z^{(n)}=0. \hspace{15pt}(2) \end{flalign*} Following the analogy of these steps it is easy to write equations that will differentiate into (1). Let $\dfrac{dz_1}{dx}=z$, $\dfrac{d^2z_2}{dx^2}=z$, $\dfrac{d^3z_3}{dx^3}=z$, \&c. Then \[ (1-x^2)\frac{d^2z_1}{dx^2}+m(m+1)z_1=0 \] will differentiate into (1), \[ (1-x^2)\frac{d^2z_2}{dx^2}+2.1x\frac{dz_2}{dx}+[m(m+1)-2.1]z_2=0 \] if differentiated twice will give (1), \[ (1-x^2)\frac{d^2z_3}{dx^2}+2.2x\frac{dz_3}{dx}+[m(m+1)-2(1+2)]z_3=0 \] if differentiated three times will give (1), and in general \[ (1-x^2)\frac{d^2z_n}{dx^2}+2(n-1)x\frac{dz_n}{dx}+[m(m+1)-n(n-1)]z_n=0 \tag{3} \] if differentiated $n$ times with respect to $x$ will give (1). If $n=m+1$ (3) reduces to \[ (1-x^2)\frac{d^2z_{m+1}}{dx^2}+2mx\frac{dz_{m+1}}{dx}=0, \tag{4} \] % -----File: 172.png 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 \[ (1-x^2)\frac{dz_m}{dx}+2mxz_m=0 \] and can be readily solved by separating the variables and integrating. v.\ Int.\ Cal.~(1) page 314. It gives \begin{flalign*} &&&\quad z_m = C(x^2-1)^m.&\phantom{Hence}\\ &\text{Hence} & z=&\frac{d^mz_m}{dx^m}=C\frac{d^m(x^2-1)^m}{dx^m} &\tag{5} \end{flalign*} is a solution of Legendre's Equation (1) and agrees with the value of $P_m(x)$ obtained in Art.~83. \mypara{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 \[ (1-x^2)\frac{d^2z}{dx^2}+2mx\frac{dz}{dx}=0 \tag{1} \] differentiates into \[ (1-x^2)\frac{d^2z}{dx^2}+2(m-1)x\frac{dz}{dx}+2mz=0; \tag{2} \] that if we differentiate (2) $m$ times we get Legendre's Equation \[ (1-x^2)\frac{d^2z}{dx^2}-2x\frac{dz}{dx}+m(m+1)z=0; \tag{3} \] that if we differentiate (2) $2m$ times we get \[ (1-x^2)\frac{d^2z}{dx^2}-2(m+1)x\frac{dz}{dx}=0; \tag{4} \] that if we differentiate (2) $m-n$ times we have \[ (1-x^2)\frac{d^2z}{dx^2}+2(n-1)x\frac{dz}{dx}+[m(m+1)-n(n-1)]z=0; \tag{5} \] and that if we differentiate (2) $m+n$ times we have \[ (1-x^2)\frac{d^2z}{dx^2}-2(n+1)x\frac{dz}{dx}+[m(m+1)-n(n+1)]z=0. \tag{6} \] By the aid of (1) we found in the last article a particular solution of (2), namely \[ z=(x^2-1)^m. \] % -----File: 173.png If we substitute in (2) $z=u(x^2-1)^m$ following the method illustrated fully in Art.~18, we get as the general solution of (2) \[ z=A(x^2-1)^m+B(x^2-1)^m\int\frac{dx}{(x^2-1)^{m+1}}, \tag{7} \] $A$ and $B$ being arbitrary constants. $\displaystyle\int\frac{dx}{(x^2-1)^{m+1}}$ is easily written out [v.\ formula (42) page 6. Table of Integrals. Int.\ Cal.\ Appendix]. If $x<1$ it vanishes when $x=0$. If $x>1$ it vanishes when $x=\infty$. If then $x<1$ (7) can be written \[ z=A(x^2-1)^m+B(x^2-1)^m\int\limits_0^x\frac{dx}{(x^2-1)^{m+1}} \tag{8} \] and if $x>1$ \[ z=A(x^2-1)^m+B(x^2-1)^m\int\limits_x^\infty\frac{dx}{(x^2-1)^{m+1}} \tag{9} \] and in these forms unnecessary arbitrary constants are avoided. From (7) we can get the general solutions of (3), (4), (5), and (6). \[ z=A\frac{d^m(x^2-1)^m}{dx^m}+B\frac{d^m}{dx^m}\bigg[(x^2-1)^m\int\frac{dx}{(x^2-1)^{m+1}}\bigg] \tag{10} \] is the general solution of (3). \[ z=A\frac{d^{2m}(x^2-1)^m}{dx^{2m}}+B\frac{d^{2m}}{dx^{2m}}\bigg[(x^2-1)^m\int\frac{dx}{(x^2-1)^{m+1}}\bigg] \tag{11} \] is the general solution of (4). \[ z=A\frac{d^{m-n}(x^2-1)^m}{dx^{m-n}}+B\frac{d^{m-n}}{dx^{m-n}}\bigg[(x^2-1)^m\int\frac{dx}{(x^2-1)^{m+1}}\bigg] \tag{12} \] is the general solution of (5). \[ z=A\frac{d^{m+n}(x^2-1)^m}{dx^{m+n}}+B\frac{d^{m+n}}{dx^{m+n}}\bigg[(x^2-1)^m\int\frac{dx}{(x^2-1)^{m+1}}\bigg] \tag{13} \] is the general solution of (6). 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 $\infty$ 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 \[ z=A_1+B_1\int\frac{dx}{(x^2-1)^{m+1}} \tag{14} \] which must be equivalent to (11). % -----File: 174.png 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 $(x^2 - 1)^m$, or in other words that if we write (2) \begin{flalign*} &&(1 - x^{2})&\dfrac{d^2 z}{dx^2} + 2(m-1)x\dfrac{dz}{dx} + 2mz = 0,\tag{2}\\ &\text{and (4) as} & (1 -{}&x^2)\dfrac{d^2 z_1}{dx^2} - 2(m + 1) x\dfrac{dz_1}{dx} = 0 \qquad\qquad\hfill\phantom{and (4) as}\tag{4} \end{flalign*} $z = z_1 (x^2 - 1)^m$; and the substitution of this value in (2) will give (4), and the substitution of $z_1 = \dfrac{z}{(x^2 - 1)^m}$ in (4) will give (2). 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 $z_1 (x^2 - 1)^m$. 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. \begin{flalign*} &\indent\text{If} &&z = y (x^2 - 1)^{\tfrac{m}{2}}&&\\ &\text{and} &&y = z_1 (x^2 - 1)^{\tfrac{m}{2}}\hfill\phantom{and}&&\\ & &&z = z_1 (x^2 - 1)^{m}.&& \end{flalign*} So that if in (2) we replace $z$ by $y(x^2 - 1)^{\tfrac{m}{2}}$ and then repeat the operation on the resulting equation we shall get (4). Making the first substitution we find, \[\tag{15} (1 - x^2)\dfrac{d^2 y}{dx^2} - 2x\dfrac{dy}{dx} + \biggl[m(m + 1) - \dfrac{m^2}{1 - x^2}\biggr] y = 0, \] not Legendre's Equation but a somewhat more general form. Of course its solution is \[\tag{16} y = A(x^2 - 1)^{\tfrac{m}{2}} + B(x^2 - 1)^{\tfrac{m}{2}}\int\dfrac{dx}{(x^2 - 1)^{m+1}}. \] (2) and (4) are special forms of (5) and (6). Let us try the experiment of substituting in (5) $z = y(1 - x^2)^{\tfrac{n}{2}}$ and in (6) $z = \dfrac{y}{(1 - x^2)^{\tfrac{n}{2}}}$. We find that both substitutions give the same equation \[\tag{17} (1 - x^{2})\dfrac{d^2 y}{dx^{2}} - 2x \dfrac{dy}{dx} + \biggl[m(m + 1) - \dfrac{n^{2}}{1 - x^{2}}\biggr] y = 0. \] % -----File: 175.png The solution of (17) can be obtained from either (12) or (13) and is \[\tag{18} y = \dfrac{1}{(1 - x^2)^{\tfrac{n}{2}}} \biggl\lbrace A \dfrac{d^{m - n}(x^2 - 1)^m}{dx^{m-n}} + B \dfrac{d^{m - n}} {dx^{m - n}} \biggl[(x^2 - 1)^{m} \int\dfrac{dx}{(x^2 - 1)^{m + 1}}\biggr]\biggr\rbrace \] or \[\tag{19} y = (1 - x^2)^{\tfrac{n}{2}} \biggl\lbrace A_1 \dfrac{d^{m+n} (x^2 - 1)^m}{dx^{m + n}} + B_1 \dfrac{d^{m+n}}{dx^{m + n}}\biggl[(x^2 -1)^{m} \int\dfrac{dx}{(x^2 -1)^{m+1}}\biggr]\biggr\rbrace \] which of course must be equivalent. \mypara{86.} In addition to the value of $P_{m}(x)$ given in (1) Art.~83 there is another important derivative form which we shall proceed to obtain. It is \[\tag{1} P_m (\cos \theta) = \dfrac{(-1)^m}{m!} r^{m+1} D_x^m \biggl(\dfrac{1}{r}\biggr). \] We have seen in Art.~75 that $\dfrac{1}{r}\dfrac{1}{\sqrt{1 - 2\dfrac{r_1}{r}\cos\theta + \dfrac{r_1^2}{r^2}}}$ can be developed into a convergent series if $r_1 < r$ and that the $(m + 1)$st term of that series is $\dfrac{P_m (\cos\theta) r_1^m} {r^{m+1}}$. Let us obtain this term by Taylor's Theorem. \begin{align*} \dfrac{1}{r}\dfrac{1}{\sqrt{1-2\dfrac{r_1}{r} \cos\theta + \dfrac{r_1^2}{r^2}}} &= \dfrac{1}{\sqrt{r^2-2r_1 r\cos\theta +r_1^2}} = \dfrac{1}{\sqrt{x^2+y^2+z^2-2xr_1+r_1^2}}\\ &=\dfrac{1}{\sqrt{(x-r_1)^2+y^2+z^2}} \end{align*} Regarding this as a function of $(x - r_1)$ and developing according to powers of $r_1$ by Taylor's Theorem we get as the $(m + 1)$st term \\ \begin{flalign*} &&\dfrac{(-1)^m}{m!} r_1^m D_x^m&\biggl[\dfrac{1}{\sqrt{x^2 + y^2 + z^2}} \biggr]\quad\text{or}\quad \dfrac{(-1)^m}{m!}{ r_1^{m}} D_x^{m} \biggl(\dfrac{1}{r}\biggr).&& \\[1ex] &\text{Hence}&&\dfrac{P_m (\cos \theta)}{r^{m + 1}} = \dfrac{(-1)^m}{m!} D_x^m \biggl(\dfrac{1}{r}\biggr).&&\hfil\phantom{Hence} \end{flalign*} \mypara{87.} We have now obtained four different forms for our \emph{zonal harmonic}, a polynomial in $x$, an expression involving cosines of multiples of $\theta$, a form involving an ordinary $m$th derivative with respect to $x$, and a form involving a partial $m$th derivative with respect to $x$. We shall now get a form due to Laplace, involving a definite integral. \[\tag{1} \int\limits_0^\pi \dfrac{d\phi}{a-b\cos\phi} = \dfrac{\pi}{(a^2-b^2)^{\frac{1}{2}}} \] if $a^2 > b^2$ [v.\ Int.\ Cal.\ page 68]. % -----File: 176.png $\dfrac{1}{(1-2xz+z^2)^{\frac{1}{2}}}$ can be expressed in the form $\dfrac{1}{(a^2-b^2)^{\frac{1}{2}}}$ by taking $a = 1 - zx$ and $b = z \sqrt{ x^2 - 1}$ and no matter what value $x$ may have $z$ can be taken so small that $a^2$ will be greater than $b^2$. Then by (1) \begin{align*} % recast to fit line \dfrac{1}{(1-2xz+z^2)^{\frac{1}{2}}}&= \frac{1}{\pi} \int\limits_0^{\pi} \frac{d\phi}{1-zx-z\sqrt{ x^2 -1}.\cos \phi} \\ &= \frac{1}{\pi} \int\limits_0^{\pi} \frac{d\phi}{1-z(x+\sqrt{ x^2 -1}.\cos \phi)}\\ &= \frac{1}{\pi} \int\limits_0^{\pi} [1+(x+ \sqrt{ x^2 -1}.\cos \phi)z+(x+ \sqrt{ x^2 -1}.\cos \phi)^2 z^2\\ &\qquad\qquad+ (x+ \sqrt{ x^2 -1}.\cos \phi)^3 z^3 + \cdots ]d\phi \end{align*} if $z$ is taken so small that the modulus of $z(x + \sqrt{ x^2 -1}.\cos \phi)$ is less than 1. But by Art.~77 (2) $P_m(x)$ is the coefficient of $z^m$ in the development of $\dfrac{1}{(1-2xz+z^2)^{\frac{1}{2}}}$, \begin{flalign*} &\text{hence} & & P_m(x)=\frac{1}{\pi} \int\limits_0^{\pi} [x+ \sqrt{ x^2 -1}.\cos\phi]^m d\phi. \tag{2}&&\hspace{1em} \end{flalign*} By replacing $\phi$ by $\pi - \phi$ in (2) we get \[ P_m(x)=\frac{1}{\pi} \int\limits_0^{\pi} [x- \sqrt{ x^2 -1}.\cos \phi]^m d\phi. \tag{3} \] $\dfrac{1}{(1-2xz+z^2)^{\frac{1}{2}}} = \dfrac{1}{z}\dfrac{1}{\Big(1-2x \dfrac{1}{z }+ \dfrac{1}{z^2}\Big)^{\frac{1}{2} }}$ and if mod $\dfrac{1}{z}<1$ or in other words if mod $z>1$\;\;$\dfrac{1}{\Big(1-2x \dfrac{1}{z} + \dfrac{1}{z^2} \Big)^{\frac{1}{2}}}$ can be developed into a convergent series involving powers of $\dfrac{1}{z}$, and the coefficient of $\Big(\dfrac{1}{z} \Big)^m$ will be $P_m(x)$; but this will be the coefficient of $z^{-m-1}$ in the development of $\dfrac{1}{(1-2xz+z^2)^{\frac{1}{2}}}$ according to descending powers of $z$, mod $z$ being greater than 1. If now we let $a = zx - 1$ and $b=z\sqrt{x^2 -1}$, $a^2 - b^2 = 1 - 2xz +z^2$ and $z$ may be taken so great that $a^2 - b^2 > 0$. Then by (1) \begin{align*} \frac{1}{(1-2xz+z^2)^{\frac{1}{2}}} &= \frac{1}{\pi} \int\limits_0^{\pi}\frac{ d\phi}{zx-1-z \sqrt{ x^2 -1}.\cos \phi}&\\ ={}&\frac{1}{\pi}\int\limits_0^{\pi}\frac{ d\phi}{z(x-\sqrt{x^2-1}.\cos \phi) \Big[1- \dfrac{1}{ z(x-\sqrt{x^2-1}.\cos \phi)} \Big]}&\\ ={}&\frac{1}{\pi}\int\limits_0^{\pi}\frac{ 1}{(x- \sqrt{ x^2 -1}.\cos \phi) }\left [z^{-1}+ \frac{1}{(x- \sqrt{ x^2 -1}.\cos \phi) } z^{-2} \right .&\\ & \left . \qquad \qquad \qquad \qquad + \frac{1}{(x- \sqrt{ x^2 -1}.\cos \phi)^2} z^{-3} + \cdots \right ] d\phi & \end{align*} % -----File: 177.png and the coefficient of $z^{-m-1}$ is $\displaystyle\frac{1}{\pi}\int\limits_0^\pi\frac{d\phi}{[x-\sqrt{x^2-1}.\cos\phi]^{m+1}}$. \begin{flalign*} &\text{Hence} & P_m(x)=\frac{1}{\pi}\int\limits_0^\pi\frac{d\phi}{[x-\sqrt{x^2-1}.\cos\phi]^{m+1}}. & \tag{4}& \end{flalign*} Replace $\phi$ by $\pi-\phi$ and we get \[ P_m(x) =\frac{1}{\pi}\int\limits_0^\pi\frac{d\phi}{[x+\sqrt{x^2-1}.\cos\phi]^{m+1}}. \tag{5} \] \mypara{88.} In the problems in which we have already used \emph{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 $\theta$ for some given value of $r$, that is on the surface of some sphere, we must develop the function of $\theta$ in terms of \emph{zonal harmonics} of $\cos\theta$ (v.\ Art.~10), and our problem becomes the following:---To develop a given function of $\cos\theta$ in terms of zonal harmonics of $\cos\theta$, or to develop a given function of $x$ in terms of the functions $P_m(x)$, $x$ 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. \begin{flalign*} &\indent\text{Let} & f(x) = A_0P_0(x) + A_1P_1(x) + A_2P_2(x) + A_3P_3(x) + \cdots & \tag{1}& \end{flalign*} 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 $\Delta x$ between $x=-1$ and $x=1$ so that $(n+2)\Delta x=2$; $f(-1+\Delta x)$, $f(-1+2\Delta x)$, $f(-1+3\Delta x)$, $\cdots$ $f[-1+(n+1)\Delta x]$ will be the corresponding values of $f(x)$. Substitute these values in (1) and we have \begin{equation*}\left. \begin{aligned} f&(-1+\Delta x) = A_0 P_0(-1+\Delta x)+A_1 P_1(-1+\Delta x)\\ &\hspace{22ex}+ A_2 P_2(-1+\Delta x)+ \cdots + A_n P_n(-1+\Delta x)\\ f&(-1+2\Delta x) = A_0 P_0(-1+2\Delta x)+A_1 P_1(-1+2\Delta x)\\ &\hspace{20ex}+ A_2 P_2(-1+2\Delta x)+ \cdots + A_n P_n(-1+2\Delta x)\\ &\vdots\hspace{8ex}\vdots\hspace{8ex}\vdots\hspace{8ex}\vdots\hspace{8ex}\vdots \hspace{8ex}\vdots\hspace{8ex}\vdots\hspace{8ex}\vdots\hspace{8ex}\\ f&(1-\Delta x) = A_0 P_0(1-\Delta x) + A_1 P_1(1-\Delta x) + A_2 P_2(1-\Delta x) + \cdots\\ &\hspace{48ex}+ A_n P_n(1-\Delta x), \end{aligned}\right\}\tag{2} \end{equation*} that is, $n+1$ equations from which in theory the $n+1$ coefficients $A_0$, $A_1$, $\cdots$ $A_n$ can be determined. % -----File: 178.png Following the analogy of Art.~24 let us multiply the first equation by $P_{m}(- 1 + \Delta x).\Delta x$, the second by $P_{m}(-1 + 2\Delta x).\Delta x$, the third by $P_{m}(- 1 +3\Delta x).\Delta x$, \&c., and add the equations. The first member of the resulting equation is \[ \sum_{k=1}^{k=n+1} f(-1+k\Delta x)P_m(-1+k\Delta x).\Delta x, \tag{3} \] and the coefficient of any $A$ as $A_l$ in the second member is \[ \sum_{k=1}^{k=n+1} P_m(-1+k\Delta x)P_l(-1+k\Delta x).\Delta x. \tag{4} \] If now $n$ is indefinitely increased (3) approaches as its limiting value \begin{flalign*} &&&\int\limits_{-1}^1 f(x) P_m(x) dx \tag{5}\\ &\text{and (4) approaches}&& \int\limits_{-1}^1 P_m(x) P_l(x)dx. && \phantom{and (4) approaches}\tag{6} \end{flalign*}\\ \indent We have now to find the value of the integral (6) or as we shall write it for the sake of greater convenience \[ \int\limits_{-1}^1 P_m(x) P_n(x) dx. \] \mypara{89.} $\displaystyle\int\limits_{-1}^1 P_m(x) P_n(x) dx = \frac{1}{2^{m+n} m!n!} \int\limits_{-1}^1 \frac{d^m(x^2-1)^m}{dx^m}.\frac{d^n(x^2-1)^n}{dx^n}dx$ by (1) Art.~83. \begin{flalign*} \int\limits_{-1}^1 \frac{d^m(x^2-1)^m}{dx^m}.\frac{d^n(x^2-1)^n}{dx^n}dx &= \bigg[\frac{d^m(x^2-1)^m}{dx^m}.\frac{d^{n-1}(x^2-1)^n}{dx^{n-1}} \mathop{\bigg]}_{-1}^1 \\ &-\int\limits_{-1}^1 \frac{d^{m+1}(x^2-1)^m}{dx^{m+1}}.\frac{d^{n-1}(x^2-1)^n}{dx^{n-1}}dx \tag{1} \end{flalign*} by \textit{integration by parts}. Now if $z = X(x^2 - 1)^n$ \begin{align*} % recast to fit line \frac{dz}{dx} &= 2nxX(x^2-1)^{n-1} + (x^2-1)^n \frac{dX}{dx} \\ &= (x^2-1)^{n-1} \left [2nxX+(x^2-1)\frac{dX}{dx} \right ]. \tag{2} \end{align*} Hence the $p$th derivative with respect to $x$ of any function of $x$ containing $(x^2- 1)^n$ as a factor will contain $(x^2- 1)^{n-p}$ as a factor if $p < n$. % -----File: 179.png $\dfrac{d^{n-1} (x^2 -1)^n }{ dx^{n-1}}$, then, contains $(x^2 - 1)$ as a factor and is zero when $x = 1$ and when $x = -1$, so that (1) reduces to \[ \int\limits^1_{-1}\dfrac{d^m (x^2 -1)^m }{ dx^m}.\dfrac{d^n (x^2 -1)^n}{dx^n} dx = - \int\limits^1_{-1} \dfrac{d^{m+1} (x^2 -1)^m }{dx^{m+1}}.\dfrac{ d^{n-1} (x^2 -1)^n }{ dx^{n-1}} dx. \] It follows that \begin{align*} % recast to fit line \int\limits^1_{-1} \dfrac{d^m (x^2 -1)^m}{dx^m}.\dfrac{d^n (x^2 -1)^n }{ dx^n} &dx \\ ={}&(-1)^p \int \limits^1_{-1} \dfrac{d^{m+p} (x^2 -1)^m }{ dx^{m+p}}.\dfrac{ d^{n-p} (x^2 -1)^n }{ dx^{n-p} }dx\\ ={}&(-1)^p \int \limits^1_{-1} \dfrac{d^{m-p} (x^2 -1)^m }{dx^{m-p}}.\dfrac{d^{n+p}(x^2 -1)^n}{dx^{n+p}} dx. \tag{3} \end{align*} If $m < n$ we get from (3) \begin{align*} \int\limits^1_{-1} \dfrac{d^m (x^2 -1)^m }{ dx^m}.\dfrac{d^n (x^2 -1)^n }{dx^n} dx &= (-1)^m \int\limits^1_{-1} \dfrac{d^{2m} (x^2 -1)^m }{ dx^{2m} }.\dfrac{d^{n-m} (x^2 -1)^n }{ dx^{n-m}} dx\\ &= (-1)^m (2m)! \bigg[ \dfrac{d^{n-m-1} (x^2 -1)^n}{ dx^{n-m-1} }\mathop{\bigg]}^1_{-1} = 0,\\[-6ex] \end{align*} \begin{flalign*} &\text{since}&& \dfrac{d^{2m} (x^2 -1)^m }{ dx^{2m}} = (2m)!.&\phantom{since} \end{flalign*} If $m > n$ \begin{align*} \int \limits^1_{-1} \dfrac{d^m (x^2 -1)^m }{dx^m }.\dfrac{d^n (x^2 -1)^n }{ dx^n}dx &= (-1)^n \int \limits^1_{-1} \dfrac{d^{m-n} (x^2 -1)^m }{ dx^{m-n}}.\dfrac{d^{2n} (x^2 -1)^n }{ dx^{2n}} dx\\ &= (-1)^n (2n)! \bigg[ \dfrac{d^{m-n-1} (x^2 -1)^m }{dx^{m-n-1}} \mathop{\bigg]}^1_{-1} = 0. \end{align*} If, then, $m$ is not equal to $n$ \[ \int\limits^1_{-1} P_m (x) P_n (x) dx = 0. \tag{4} \] If $m = n$ we have to find $\displaystyle \int\limits^1_{-1} [P_m (x)]^2 dx$. \[ \int\limits^1_{-1} [P_m (x)]^2 dx = \dfrac{1}{2^{2m} (m!)^2} \int\limits^1_{-1} \dfrac{ d^m (x^2 -1)^m }{ dx^m}.\dfrac{ d^m (x^2 -1)^m }{ dx^m} dx. \] \begin{flalign*} &&\int\limits^1_{-1} \dfrac{d^m (x^2 -1)^m }{dx^m }.\dfrac{d^m (x^2 -1)^m }{ dx^m} dx &= (-1)^m \int\limits^1_{-1} \dfrac{ d^{2m} (x^2 -1)^m }{ dx^{2m}}.(x^2 - 1)^m dx&&\\ &\multispan{2}{by (3),\hfill}&= (-1)^m (2m)! \int\limits^1_{-1} (x^2 - 1)^m dx.&& \end{flalign*} % -----File: 180.png \begin{align*} \int\limits_{-1}^1(x^2 -1)^m dx &= \int\limits_{-1}^1 (x-1)^m (x+1)^m dx = -\frac{m}{m+1} \int\limits_{-1}^1 (x-1)^{m-1} (x+1)^{m+1} dx\\ &= (-1)^m \frac{m!}{(m+1)(m+2) \cdots 2m} \int\limits_{-1}^1 (x+1)^{2m} dx\\ &= (-1)^m \frac{2^{2m+1} m!}{(m+1)(m+2) \cdots (2m+1)}. \end{align*} \begin{flalign*} &\text{Hence} &\int\limits_{-1}^1 [P_m (x)]^2 dx &= \frac{1}{2^{2m} (m!)^2}\frac{ (-1)^m (2m)!(-1)^m m!2^{2m+1}}{(m+1)(m+2) \cdots (2m+1)} &&\phantom{Hence}\\ &\text{or} && \int\limits_{-1}^1 [P_m(x)]^2 dx = \frac{2}{2m+1}. \tag{5}&& \end{flalign*} \mypara{90.} The solution of the problem in Art.~88 is now readily obtained, and we have \begin{flalign*} &&f(x) = &A_0P_0(x) + A_1P_1(x) + A_2P_2(x) + \cdots \tag{1} &&\\ &\text{where}&&A_m = \frac{2m+1}{2 }\int\limits_{-1}^1 f(x)P_m(x)dx. \tag{2} &&\hspace{1em} \end{flalign*} 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 \begin{flalign*} &&f(\cos \theta) &= A_0P_0(\cos \theta) +A_1P_1(\cos \theta) + A_2P_2(\cos \theta) + \cdots &&\\ &\text{where} && A_m = \frac{2m+1}{2} \int\limits_{-1}^1 f(\cos \theta)P_m(\cos \theta)d(\cos \theta) && \phantom{where} \intertext{or if $f(\cos \theta) = F(\theta)$} &&F(\theta) = &A_0P_0(\cos \theta) + A_1P_1(\cos \theta) + A_2P_2(\cos \theta) + \cdots \tag{3} \\ &\text{where} && A_m = \frac{2m+1}{2} \int\limits_{0}^{\pi} F(\theta)P_m(\cos \theta)\sin\theta.d\theta && \tag{4} \end{flalign*} and the development holds good from $ \theta = 0$ to $\theta = \pi$. 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 \begin{flalign*} &&\int\limits_{-1}^1 f(x)&P_{2k}(x)dx = 2 \int\limits_0^1 f(x)P_{2k}(x)dx,\\ % -----File: 181.png &\text{and that}& &\int\limits_{-1}^1 f(x)P_{2k+1} (x)dx = 0; && \phantom{and that} \intertext{so that if $f(-x) =f(x)$} &&f(x) = A_0P_0(x) &+A_2P_2(x) + A_4P_4(x) + A_6P_6(x) + \cdots \tag{5} \\ &\text{where}& A_{2k} &= (4k+1)\int\limits_0^1 f(x)P_{2k}(x)dx. && \tag{6} \end{flalign*} If $f(x)$ is an odd function, that is, if $f(-x) = -f(x)$ it can be shown in like manner that \begin{flalign*} &&f(x) = A_1P_1&(x) +A_3P_3(x) + A_5P_5(x) + A_7P_7(x) + \cdots &&\tag{7}\\ &\text{where}&& A_{2k+1} = (4k+3) \int\limits_0^1 f(x)P_{2k+1}(x)dx. && \phantom{where} \tag{8} \end{flalign*} 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. \mypara{91.} We can establish the fact that $\displaystyle\int\limits_{-1}^1 P_m(x)P_n(x)dx = 0$ by a more general method than that used in Art.~89. Let $X_m$ be any solution of Legendre's Equation \begin{flalign*} &\hspace{6em}&&\frac{d}{dx} \left [ (1-x^2)\frac{ dz}{dx} \right ] + m(m+1)z = 0 && \tag*{[v.\ (1) Art.~16].} \intertext{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.} &\text{\indent Then}&&\frac{d}{dx} \left [ (1-x^2) \frac{dX_m }{ dx} \right ] + m(m+1)X_m = 0 \tag{1}&&\\ &\text{and}&&\frac{d}{dx} \left [ (1-x^2) \frac{dX_n }{ dx} \right ] + n(n+1)X_n = 0 \tag{2}&&\phantom{and} \end{flalign*} Multiply (1) by $X_n$ and (2) by $X_m$ and subtract and integrate and we get \begin{multline*} [m(m+1)-n(n+1)] \int\limits_{-1}^1 X_mX_n dx = \int\limits_{-1}^1 X_m \frac{d}{dx} \left [ (1-x^2) \frac{dX_n }{ dx} \right ] dx\\ - \int\limits_{-1}^1 X_n \frac{d}{dx} \left [ (1-x^2)\frac{ dX_m }{ dx} \right ] dx. \end{multline*} % -----File: 182.png Integrate by parts, \begin{multline*} [m(m + 1) - n(n + 1)] \int\limits_{-1}^1 X_m X_n dx = \bigg[X_m(1-x^2)\frac{dX_n}{dx} - X_n(1-x^2)\frac{dX_m}{dx} \!\!\!\!\!\mathop{\bigg]}_{x=-1}^{x=1} \\ -\int\limits_{-1}^1 (1-x^2)\frac{dX_n}{dx}\frac{ dX_m}{dx} dx + \int\limits_{-1}^1 (1-x^2)\frac{dX_m}{dx} \frac{dX_n}{dx} dx. \tag{3}\\[-5ex] \end{multline*} \begin{flalign*} &\text{Whence}&&\int\limits_{-1}^1 X_m X_n dx = 0 && \phantom{Whence} \tag{4}\\[-5ex] \end{flalign*} unless $m = n$. (3) gives at once the important formula \[ \int\limits_x^1X_mX_ndx =\frac{ (1-x^2) \Big [X_n \dfrac{dX_m}{dx} - X_m\dfrac{dX_n}{dx} \Big ]}{m(m + 1) - n(n + 1) } \tag{5} \\ \] from which come as special cases \[ \int\limits_x^1P_m(x)P_n(x)dx = \frac{(1-x^2) \Big [P_n(x)\dfrac{dP_m(x)}{dx} - P_m(x)\dfrac{dP_n(x)}{dx} \Big ] }{ m(m + 1) - n(n + 1)} \tag{6} \] and since $P_0(x) = 1$ \[ \int\limits_x^1P_m(x)dx = \frac{(1-x^2)\dfrac{dP_m(x)}{dx}}{ m(m + 1)}, \tag{7}\\ \] unless $m = 0$. \EXAMPLE{S} 1.\quad Show that \begin{align*} \int\limits_0^1 P_m(x)dx &= 0 \quad\text{ if $m$ is even and is not zero.}\\ &=(-1)^{\frac{m-1}{2}}\frac{ 1}{m(m+1)}\frac{ 3.5.7.\cdots m }{ 2.4.6.\cdots (m-1) } \quad\text{if $m$ is odd. } \end{align*} v.\ Art.~91 (7) and Art.~77 (10).\\ 2.\quad Show that \begin{multline*} \int\limits_0^1P_m(x)P_n(x)dx = 0 \quad \text{if $m$ and $n$ are both even or both odd.}\\ = (-1)^{\frac{m+n+1}{2}} \frac{m!\;n!}{ 2^{m+n-1} (m-n)(m+n+1) \left (\dfrac{m}{2} ! \right )^2 \Big (\dfrac{n-1}{2} ! \Big )^2} \end{multline*} 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.\\ 3.\quad Show that \quad $\displaystyle\int\limits_0^1[P_m(x)]^2dx = \dfrac{1}{2m+1}$ \quad v.\ Art.~89 (5). % -----File: 183.png \mypara{92.} Formula (4) Art.~91 can be obtained directly from Laplace's Equation by the aid of \emph{Green's Theorem} (v.\ Peirce's Newt.\ Pot.\ Func.\ \S~48). Take the special form of Green's Theorem, [(148) \S~48 Peirce's Newt.\ Pot.\ Func.] \[ \iiint(U\nabla^2V - V\nabla^2U)dxdydz = \int (UD_n V - VD_n U)ds \tag{1} \] where $\nabla^2$ stands for $(D_x^2 + D_y^2 + D_2^2)$, $D_n$ 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 $\nabla^2 V = \nabla^2 U = 0$ and (1) reduces to \[ \int(UD_n V - V D_n U) ds = 0. \tag{2} \] Now $r^mX_m$ and $r^nX_n$ are solutions of Laplace's Equation if $x = \cos \theta$ (v.~Art.~16). If the unit sphere is taken as the bounding surface and $U = r^mX_m$ and $V = r^nX_n$ (1) and (2) will hold good. \begin{gather*} D_nU = D_r(r^mX_m) = mr^{m-1}X_m,\\ D_nV = nr^{n-1}X_n,\\ ds = \sin \theta.d\theta d\phi, \end{gather*} and (2) becomes $\displaystyle \int\limits^{2\pi}_0 d\phi \int\limits^{\pi}_0 (n X_m X_n - m X_m X_n) \sin \theta.d \theta = 0$ \begin{flalign*} &\text{or}& 2\pi(n - m) \int\limits^{\pi}_0 X_m X_n \sin \theta.d \theta &= 0. & \tag{3} \end{flalign*} Since $x = \cos\theta$, $\sin\theta.d\theta = - dx$ and (3) reduces to \[ \int\limits^1_{-1} X_m X_n dx = 0\tag{4}\footnotemark \] unless $m = n$. \footnotetext{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 $r^mX_m$ 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.\ \S~48.) } \mypara{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 $\theta$ in the form \[ A_0P_0(\cos\theta) + A_1P_1(\cos\theta) + A_2P_2(\cos\theta) + \cdots\\[-2ex] \] % -----File: 184.png \begin{flalign*} &\text{given that}& f(\theta) = 1 \quad\text{from}\quad\theta = 0\quad\text{to}\quad\theta = \dfrac{\pi}{2} &&\phantom{given that}\\ &\text{and} & f(\theta) = 0 \quad\text{from}\quad\theta = \dfrac{\pi}{2}\quad\text{to}\quad\theta = \pi. && \end{flalign*} This amounts to the same thing as developing $F(x)$ into the series \[ F(x) = A_0 P_0(x) + A_1 P_1(x) + A_2 P_2(x) + A_3 P_3(x) + \cdots\\[-2ex] \] \begin{flalign*} &\text{where} &&F(x) = 0\quad\text{from}\quad x = -1\quad\text{to}\quad x = 0&&\phantom{where}\\ &\text{and} &&F(x) = 1\quad\text{from}\quad x = 0\phantom{-}\quad\text{to}\quad x = 1.&& \end{flalign*} By Art.~90 (1) and (2) \[ A_0 = \dfrac{1}{2} \int\limits^1_0 P_0(x)dx = \dfrac{1}{2} \int\limits^1_0 dx = \dfrac{1}{2} , \] \begin{flalign*} &\text{and any coefficient}&&A_m = \dfrac{(2m + 1)}{2} \int\limits^1_0 P_m(x)dx. &&\phantom{and any coefficient} \end{flalign*} By Art.~91, Ex.~1 \[ \int\limits^1_0 P_m(x)dx = 0\quad \text{ if $m$ is even}\rule{14.7em}{0em}\\[-5ex] \] \begin{flalign*} && &= (-1)^{\tfrac{m-1}{2}} \dfrac{1}{m(m+1)}\dfrac{ 3.5.7.\cdots m}{2.4.6.\cdots (m-1)} \quad \text{if $m$ is odd.}&\\ &\text{\indent Hence}& A_m &= 0\quad \text{ if $m$ is even}&\\ && &= (-1)^{\tfrac{m-1}{2}} \dfrac{2m+1}{2m+2}.\dfrac{ 1.3.5.\cdots(m-2)}{2.4.6.\cdots (m-1)}\quad \text{ if $m$ is odd}.&\\ &\text{\indent Then}& F(x) &= \dfrac{1}{2} + \dfrac{3}{4} P_1 (x) - \dfrac{7}{8}.\dfrac{1}{2} P_3(x) + \dfrac{11}{12}.\dfrac{1.3}{2.4} P_5(x) -\cdots &\tag{1} \end{flalign*} and \quad $u = \dfrac{1}{2} + \dfrac{3}{4} r P_1(\cos\theta) - \dfrac{7}{8}.\dfrac{1}{2} r^3 P_3 (\cos\theta) + \dfrac{11}{12}.\dfrac{1.3}{2.4} r^5 P_5(\cos \theta) + \cdots$\hfill(2)\\[1ex] for any point within the sphere. \mypara{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 symmetry\footnote{See note on page \pageref{notep12}.} 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 \[ r D_r^2(rV) + \dfrac{1}{\sin\theta} D_\theta (\sin\theta D_\theta V) = 0 \tag{1} \] % -----File: 185.png subject to the conditions \begin{alignat*}{3} &V=f(\theta)\quad && \text{when}\quad &&r=a\\ &V=0 &&\quad \text{``} &&r=\infty. \end{alignat*} \begin{flalign*} &\text{We have}& f(\theta) &= A_0 P_0(\cos\theta) + A_1 P_1(\cos\theta) + A_2 P_2(\cos\theta) +\cdots&&\\ &\text{where}&& A_m = \frac{(2m+1)}{2} \int\limits_0^{\pi} f(\theta) P_m(\cos\theta) \sin\theta.d\theta. && \tag*{v.\ Art.~90 (4).} \end{flalign*} Hence \[ V=A_0 + A_1 \left (\frac{r}{a} \right )P_1(\cos\theta) + A_2 \left (\frac{r}{a} \right )^2 P_2(\cos\theta) + A_3 \left (\frac{r}{a} \right )^3 P_3(\cos\theta) + \cdots (2) \] is the required solution for a point within the sphere, and \[ V=A_0\left (\frac{a}{r} \right ) + A_1 \left (\frac{a}{r} \right ) ^2P_1(\cos\theta) + A_2 \left (\frac{a}{r} \right ) ^3 P_2(\cos\theta) + A_3 \left (\frac{a}{r} \right ) ^4 P_3(\cos\theta) + \cdots \tag{3} \] is the required solution for an external point. \EXAMPLE{S} 1.\quad If on the surface of a sphere of radius $c$ $V$ is constant and equal to a show that $V=a$ for any point within the sphere and $V=\dfrac{ac}{r}$ for any external point.\\ 2.\quad 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. \begin{multline*} V= \frac{a+b}{2} + (b-a) \left [\frac{3}{4} \frac{r}{c} P_1(\cos\theta)- \frac{7}{8}.\frac{ 1}{2}\frac{ r^3}{c^3} P_3(\cos\theta) \right.\\ \left . + \frac{11}{12}.\frac{1.3}{2.4}\frac{ r^5}{c^5} P_5(\cos\theta)- \cdots \right ] \end{multline*} for an internal point \begin{multline*} V= \frac{a+b}{2}.\frac{c}{r} + (b-a) \left [\frac{3}{4} \frac{c^2}{r^2} P_1(\cos\theta)- \frac{7}{8}.\frac{ 1}{2}\frac{ c^4}{r^4} P_3(\cos\theta) \right.\\ \left . + \frac{11}{12}.\frac{1.3}{2.4}\frac{ c^6}{r^6} P_5(\cos\theta)- \cdots \right ] \end{multline*} for an external point.\\ % -----File: 186.png 3.\quad If $V_1 =f(\cos\theta)$ when $r = a$ and $V_1 = 0$ when $r = b$ show that for $a < r < b$ \begin{flalign*} &&V_1 = \sum_{m=0}^{m=\infty} &A_m \left ( \frac{b^{m+1}}{r^{m+1}} - \frac{r^m}{b^m} \right ) \left (\frac{b^{m+1}}{a^{m+1}} -\frac{ a^m}{b^m} \right )^{-1} P_m(\cos\theta)\\ &\text{where} &&A_m = \frac{2m+1}{2} \int\limits_{-1}^1 f(x)P_m(x)dx.&& \phantom{where} \end{flalign*} 4.\quad If $V_2 =F(\cos\theta)$ when $r = b$ and $V_2 = 0$ when $r = a$ show that for $a < r < b$ \begin{flalign*} &&V_2 = \sum_{m=0}^{m=\infty} &B_m \left ( \frac{r^{m}}{a^{m}} - \frac{a^{m+1}}{r^{m+1}} \right ) \left (\frac{b^{m}}{a^{m}} -\frac{ a^{m+1}}{b^{m+1}} \right )^{-1} P_m(\cos\theta)\\ &\text{where} &&B_m = \frac{2m+1}{2} \int\limits_{-1}^1 F(x)P_m(x)dx.&& \phantom{where} \end{flalign*} 5.\quad If the value of the potential function is given arbitrarily on the surfaces of a spherical shell but has circular symmetry\footnote{See note on page 12} about a diameter $V= V_1 + V_2$ (v.\ Exs.~3 and 4).\\ 6.\quad 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. \begin{align*} V&=p \quad\text{if}\quad r < a,\\ V&= \frac{pa}{b-a}\bigg ( \frac{b}{r}-1 \bigg ) +\frac{ qb}{b-a} \bigg (1- \frac{a}{r} \bigg ) \quad\text{if} \quad ab. \end{align*} 7.\quad If $V=0$ on the base of a hemisphere and $V=f(\cos\theta)$ on the convex surface, show that for a point within the hemisphere \begin{flalign*} &&V={}&\sum_{k=0}^{k=\infty} A_{2k+1}\bigg (\frac{r}{a} \bigg )^{2k+1} P_{2k+1} (\cos\theta)\\ &\text{where}& A_{2k+1} &= (4k + 3) \int\limits_0^1 f(x)P_{2k+1}(x)dx \tag*{ [v.\ Art.~90 (8)].}&& \end{flalign*} 8.\quad 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 \[ u = \frac{3}{2} \frac{ r}{a} P_1(\cos\theta) - \frac{ 7}{4}.\frac{1}{2} \frac{ r^3}{a^3} P_3(\cos\theta) + \frac{11}{6}.\frac{ 1.3}{2.4 } \frac{r^5}{a^5} P_5(\cos\theta) - \cdots \] % -----File: 187.png 9.\quad 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 \textit{stationary temperature} of any internal point.\\ \textit{Suggestion:}\quad $D_ru + hu - Mf(\theta) = 0$ \quad when \quad $r = a$. \begin{flalign*} &\text{Let} &u= \sum A_m\frac{ r^m }{a^m} P_m(\cos\theta), \quad \text{and}\quad f(\theta) = \sum B_m P_m(\cos\theta).&& \end{flalign*} Then we have \begin{flalign*} &&\sum m \frac{A_m}{a} P_m(\cos\theta) + h \sum &A_m P_m (\cos\theta) - M \sum B_m P_m (\cos\theta) = 0,\\ &\multispan{2}{whence\hfill}&A_m = \frac{MB_m }{ h+ \dfrac{m}{a}}. &&\phantom{whence} \end{flalign*} Here $f(\theta) = \cos\theta$ if $0 < \theta <\dfrac{\pi}{2}$ and $f(\theta) = 0$ if $\dfrac{\pi}{2} < \theta < \pi$. \begin{multline*} f(\theta) = \frac{1}{4} + \frac{1}{2} P_1(\cos\theta) + \frac{5}{16} P_2(\cos\theta) - \frac{3}{32} P_4(\cos\theta) + \cdots\\ + (-1)^{k+1}\frac{ (4k+1)(2k)!}{(4k+4)(2k-1)2^{2k} (k!)^2} P_{2k} (\cos \theta) + \cdots \end{multline*} v.\ Art.~91 Exs.~(2) and (3). cf. J.~W. Strutt (Lord Rayleigh), Lond.\ Phil.\ Trans.\ vol.\ 160, page 587. \mypara{95.} The formulas of Art.~90 enable us to develop a given function of $x$ in terms of \textit{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 $x^n$ 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 $P_1(x)$, $P_2(x), \cdots P_n(x)$ (v.\ Art.~77) may be regarded as $n$ algebraic equations of the first degree in terms of $x$, $x^2$, $x^3, \cdots x^n$ and $P_1(x)$, $P_2(x), \cdots P_n(x)$. From these equations the $n -1$ quantities $x$, $x^2$, $x^3, .\cdots x^{n-1}$, can be eliminated, and there will result an equation of the first degree in $x^n$ and $P_1(x)$, $P_2(x), \cdots P_n(x)$, which will enable us to express $x^n$ in the form \[ A_0 +A_1 P_1(x) + A_2 P_2(x) + \cdots + A_n P_n(x), \] no matter what the value of $x$, and we shall have the same formula when $-1 < x < 1$ as when $x > 1$ or $x < -1$. % -----File: 188.png Let us obtain this development. By Art.~90 (1) and (2) \begin{flalign*} &&x^n &= A_0 P_0(x) + A_1 P_1(x) + A_2 P_2(x) + \cdots\tag{1}\\ &\text{where}& &\qquad A_m = \dfrac{2m + 1}{2}\int\limits^1_{-1} x^n P_m(x)dx. &\phantom{where}\tag{2}\\ &\text{Then}\hspace{2em}& A_m &= \dfrac{2m + 1}{2}\dfrac{ 1}{2^m m!} \int\limits^1_{-1} x^n \dfrac{d^m(x^2-1)^m }{dx^m} dx & \tag*{by (1) Art.~83.} \end{flalign*} By \emph{integration by parts} we get \begin{alignat*}{2} \int\limits^1_{-1} x^n \dfrac{d^m(x^2-1)^m}{dx^m} dx &\multispan{3}{$\displaystyle{}= n(n-1)(n-2) \cdots(n-m+1) \int\limits^1_{-1} x^{n-m} (1-x^2)^m dx,$} \\[-2ex] &&& \text{if}\quad m < n + 1,\tag{3}\\ &=0 \quad &&\text{if}\quad m > n. \end{alignat*} By \emph{integration by parts} we readily obtain the reduction formula \begin{flalign*} &&\int\limits^1_{-1}x^p(1 - x^2)^q dx &= \dfrac{2q}{p+1} \int\limits^1_{-1} x^{p+2} (1 -x^2)^{q-1} dx & \text{whence}\\[-5ex] \end{flalign*} \[ \int\limits^1_{-1} x^{n-m} (1- x^2)^m dx = \dfrac{2^m m!}{(n - m + 1)(n - m + 3)\cdots (n + m - 1)} \int\limits^1_{-1} x^{n+m} dx.\\[-5ex] \] \begin{flalign*} \int\limits^1_{-1} x^{n+m} dx &= \dfrac{2}{(n + m + 1)}\quad \text{if}\quad n + m \quad\text{is even,}\\ &= 0\quad\text{if}\quad n + m \quad\text{is odd.} \end{flalign*} \begin{flalign*} &\text{Hence} &A_m &= \dfrac{(2m + 1)n(n - 1)(n - 2)\cdots (n - m + 1) }{(n - m + 1)(n - m + 3) (n - m + 5) \cdots (n + m + 1)}&\phantom{hence}\\[1ex] &&& \hspace{6em}\text{if } m < n + 1 \text{ and } m + n \text{ is even,}\\[1ex] &&&= 0 \quad \text{if } m > n \text{ or if } m + n \text{ is odd.} \end{flalign*} Therefore \begin{align*} &\rlap{$\displaystyle x^n = \dfrac{n! }{1.3.5 \cdots (2n+1)}\left [(2n+1) P_n(x) + (2n-3) \dfrac{(2n+1)}{2} P_{n-2} (x)\right.$}\\ &&&+ (2n-7) \dfrac{(2n+1)(2n-1)}{2.4 } P_{n-4}(x)\\ &\hspace{7em}&&+ \left.(2n-11) \dfrac{(2n+1)(2n-1)(2n-3)}{2.4.6 } P_{n-6}(x) + \cdots \right] && \tag{4} \end{align*} the second member ending with the term $\dfrac{1}{n+1} P_0(x)$ if $n$ is even and with the term $\dfrac{3}{n+2} P_1(x)$ if $n$ is odd. % -----File: 189.png For convenience of reference we write out a few powers of $x$. \[ \left . \begin{aligned} x^0 &= 1 = P_0(x)\\ x &= P_1(x)\\ x^2&=\frac{2}{3} P_2(x)+ \frac{1}{3}P_0(x)\\ x^3&=\frac{2}{5} P_3(x) + \frac{3}{5} P_1(x)\\ x^4 &= \frac{8}{35} P_4(x) + \frac{4}{7 }P_2(x)+ \frac{1}{5} P_0(x)\\ x^5 &= \frac{8}{63} P_5(x) + \frac{4}{9} P_3(x) + \frac{3}{7}P_1(x)\\ x^6 &= \frac{16}{231} P_6(x) +\frac{24}{77} P_4(x) +\frac{10}{21 }P_2(x) +\frac{1}{7} P_0(x)\\ x^7 &= \frac{16}{429} P_7(x) +\frac{8}{39} P_5(x) +\frac{14}{33} P_3(x) +\frac{1}{3 }P_1 (x)\\ x^8 &= \frac{128}{6435} P_8(x) + \frac{64}{495} P_6(x) +\frac{48}{143} P_4(x) + \frac{40}{99 }P_2 (x) + \frac{1}{9} P_0(x).\\ \end{aligned} \right \} \tag{ 5} \] If a given function of $x$ can be expressed as a \textit{terminating power series} it can be developed into a Zonal Harmonic Series by the aid of (4). Given that \begin{flalign*} &&f(x) &= a_0 + a_1x + a_2x^2 + a_3x^3 + \cdots,\\[1ex] &\text{let} &f(x) = B_0 &+ B_1P_1(x) + B_2P_2(x) + B_3P_3(x) + \cdots ;&&\phantom{let} \end{flalign*} then picking out carefully the coefficient of $P_m(x)$ we have \begin{multline*} B_m = \frac{m! }{ 1.3.5.\cdots (2m - 1)} \left [ a_m + \frac{(m+1)(m+2)}{ 2.(2m+3)} a_{m+2} \right .\\ \left . + \frac{(m + 1)(m + 2)(m + 3)(m + 4) }{2.4.(2m + 3)(2m + 5)} a_{m+4} + \cdots \right ]. \tag{6} \end{multline*} \mypara{96.} The development of $\dfrac{dP_n(x) }{dx}$ is useful and is easily obtained. \begin{flalign*} &\text{\indent Let}&\frac{dP_n(x) }{dx} = &A_0P_0(x) + A_1P_1(x) + A_2P_2(x) + \cdots &&\phantom{\indent Let}\\ &\text{Then}& A_m &= \frac{2m+1 }{2} \int\limits_{-1}^1 P_m(x)\frac{ dP_n(x) }{dx} dx && \tag{1} \end{flalign*} by Art.~90 (2); \[ \int\limits_{-1}^1 P_m(x) \frac{dP_n(x)}{dx } dx = \Big[P_m(x)P_n(x)\Big]_{x=-1}^{x=1}-\int\limits_{-1}^1 P_n(x) \frac{dP_m(x) }{dx} dx. \tag{2} \] % -----File: 190.png \begin{align*} \Big[P_m(x)P_n(x)\Big]_{x=-1}^{x=1}& =0 \quad \text{if $m + n$ is even}\\ & =2 \quad \text{if $m + n$ is odd.} \end{align*} Since $P_n(x)$ is an algebraic polynomial of the $n$th degree in $x$, $\dfrac{dP_n(x)}{dx}$ is an algebraic polynomial of the $n-1$st degree in $x$. Therefore in (1) $m$ is less than $n$; consequently $\dfrac{dP_m(x)}{dx}$ is an algebraic polynomial in $x$ of lower degree than $n$ and \begin{flalign*} &\phantom{by Art. 95 (3).} &&\int\limits_{-1}^1 P_n(x) \frac{dP_m(x)}{dx} dx = 0 && \tag*{by Art.~95 (3).} \end{flalign*} \begin{flalign*} &\text{We get then}&A_m & = 2m+1 \quad \text{if $m+n$ is odd and $m n-1$;} &&\tag*{and} \end{flalign*} \[ \frac{dP_n(x)}{dx }= (2n-1)P_{n-1}(x) + (2n-5)P_{n-3}(x) + (2n-9)P_{n-5}(x) + \cdots \tag{3} \] the second member ending with the term $3P_1(x)$ if $n$ is even and with the term $P_0(x)$ if $n$ is odd. From (3) a number of simple formulas are readily obtained. For example \begin{gather*} \frac{dP_{n+1}(x)}{dx} - \frac{dP_{n-1}(x)}{dx} =(2n+1)P_n(x) \tag{4}\\ \int\limits_x^1P_n(x) dx = \frac{1}{2n+1} [P_{n-1}(x)-P_{n+1}(x)]. \tag{5}\\ (2n+1)x \frac{dP_n(x)}{dx} = n \frac{dP_{n+1}(x)}{dx} + (n+1) \frac{dP_{n-1}(x)}{dx } \tag{6} \intertext{[v.\ (4) and Article 77 (12)].} (x^2-1) \frac{dP_n(x)}{dx} = nxP_n(x)-nP_{n-1}(x) \tag{7} \end{gather*} [v.\ (5) and Article 91 (7).] \mypara{97.} By the aid of the formulas of Art.~96 a number of valuable developments can be obtained. Let us get $\cos n\theta$ and $\sin n\theta$, $n$ being any positive real. $z=\cos n\theta$ and $z=\sin n\theta$ are solutions of the equation \[ \frac{d^2z}{d\theta^2} +n^2z =0 \] % -----File: 191.png or if we let $x = \cos \theta$, of the equation \[ (1-x^2)\dfrac{d^2z}{dx^2} -x\dfrac{dz}{dx} +n^2z=0. \tag{1} \] \begin{flalign*} &\text{\indent Let}&& a_0P_0(x) + a_1P_1(x)+a_2P_2(x) + \cdots&&\phantom{\indent Let} \end{flalign*} be the required development of $\cos n\theta$ or of $\sin n\theta$. \begin{flalign*} &\text{\indent Then} &&\sum\limits^{m=\infty}_{m=0} a_m\left[(1-x^2)\dfrac{d^2P_m(x)}{dx^2} - x\dfrac{ dP_m(x)}{dx} + n^2P_m(x)\right] = 0 \;&&\text{ by (1).} \end{flalign*} $z = P_m(x)$ is a solution of Legendre's Equation (v.\ Art.~77). Hence \[ (1-x^2)\dfrac{d^2P_m(x)}{dx^2}-x\dfrac{dP_m(x)}{dx}=x\dfrac{dP_m(x)}{dx}-m(m+1)P_m(x), \] and (1) becomes \[ \sum\limits^{m=\infty}_{m=0} a_m\left[x\dfrac{dP_m(x)}{dx}+[n^2-m(m+1)]P_m(x)\right]=0. \tag{2} \] Formulas (4) and (6) of Art.~96 enable us to throw (2) into the form \[ \sum\limits^{m=\infty}_{m=0}a_m\left[\dfrac{n^2-m^2}{2m+1}\dfrac{dP_{m+1}(x)}{dx} -\dfrac{ n^2-(m+1)^2}{2m+1} \dfrac{dP_{m-1}(x)}{dx}\right]=0.\tag{3} \] (3) must be identically true. Therefore the coefficient of $\dfrac{dP_{m+1}(x)}{dx}$ must equal zero, and we have \[ a_{m+2}=\dfrac{2m+5}{2m+1}.\dfrac{n^2-m^2}{n^2-(m+3)^2} a_m. \tag{4} \] If we are developing $\cos n\theta$ \begin{flalign*} &&a_0&=\dfrac{1}{2}\int\limits^{\pi}_0 \cos n\theta \sin\theta.d\theta &&\tag*{ by Art.~90 (4),}\\ && &=\dfrac{1}{4}\int\limits^{\pi}_0 [\sin(n+1)\theta-\sin(n-1)\theta]d\theta,& \end{flalign*} \begin{flalign*} &&a_0&=-\dfrac{1}{2}.\dfrac{1+\cos n\pi}{n^2-1}; && \tag{5}\\ &\text{and}\hspace{6em}& a_1={}&\dfrac{3}{2}\int\limits^{\pi}_0 \cos n\theta\cos\theta\sin\theta.d\theta &&\tag*{by Art.~90 (4),}\\ &&a_1&=-\dfrac{3}{2}.\dfrac{1-\cos n\pi }{n^2-4}. && \tag{6} \end{flalign*} % -----File: 192.png (4), (5), and (6) give us \begin{align*} \cos n\theta=-\frac{1+\cos n\pi}{2(n^2-1)} \Big[ & P_0(\cos\theta)+5\frac{n^2}{n^2-3^2}P_2(\cos\theta)\\ & +9\frac{n^2(n^2-2^2)}{(n^2-3^2)(n^2-5^2)}P_4(\cos\theta)+\cdots\Big]\\ -\frac{1-\cos n\pi}{2(n^2-2^2)} \Big[ & 3P_1(\cos\theta)+7\frac{n^2-1^2}{n^2-4^2}P_3(\cos\theta)\\ & +11\frac{(n^2-1^2)(n^2-3^2)}{(n^2-4^2)(n^2-6^2)}P_5(\cos\theta)+\cdots\Big]. \tag{7} \end{align*} If $n$ is a whole number $1+\cos n\pi$ or $1-\cos n\pi$ will vanish and the series will end with the term involving $P_n(\cos\theta)$. For this case (7) may be rewritten \begin{align*} \cos n\theta={}&\frac{1}{2}.\frac{2.4.6.\cdots 2n}{3.5.7.\cdots(2n+1)} \Big[(2n+1)P_n(\cos\theta)\\ & +(2n-3)\frac{n^2-(n+1)^2}{n^2-(n-2)^2}P_{n-2}(\cos\theta)\\ & +(2n-7)\frac{[n^2-(n+1)^2][n^2-(n-1)^2]}{[n^2-(n-2)^2][n^2-(n-4)^2]}P_{n-4}(\cos\theta)+\cdots\Big]. &&\tag{8} \end{align*} If we are developing $\sin n\theta$ \begin{align*} a_0=\frac{1}{2}\int\limits_0^\pi\sin n\theta\sin\theta.d\theta =-\frac{1}{2}.\frac{\sin n\pi}{n^2-1},\\ a_1=\frac{3}{2}\int\limits_0^\pi\sin n\theta\cos\theta\sin\theta.d\theta =\frac{3}{2}.\frac{\sin n\pi}{n^2-2^2} \tag*{and}\\[-5ex] \end{align*} \begin{align*} \sin n\theta=-\frac{1}{2}.\frac{\sin n\pi}{n^2-1} \Big[ & P_0(\cos\theta)+5\frac{n^2}{n^2-3^2}P_2(\cos\theta)\\ & + 9\frac{n^2(n^2-2^2)}{(n^2-3^2)(n^2-5^2)}P_4(\cos\theta)+\cdots\Big]\\ +\frac{1}{2}.\frac{\sin n\pi}{n^2-2^2} \Big[ &3P_1(\cos\theta)+7\frac{n^2-1^2}{n^2-4^2}P_3(\cos\theta)\\ & + 11\frac{(n^2-1^2)(n^2-3^2)}{(n^2-4^2)(n^2-6^2)}P_5(\cos\theta)+\cdots\Big]. \tag{9} \end{align*} If $n$ is a whole number $\sin n\pi=0$, and all the terms of (9) vanish except those involving $P_{n-1}(\cos\theta)$, $P_{n+1}(\cos\theta)$, $P_{n+3}(\cos\theta)$ \&c., which become indeterminate. For this case it is necessary to compute $a_{n-1}$ independently. % -----File: 193.png We have \begin{align*} a_{n-1} & = \frac{2n-1}{2}\int\limits_0^\pi\sin n\theta P_{n-1}(\cos\theta)\sin\theta.d\theta\\ & = \frac{2n-1}{4}\int\limits_0^\pi[\cos(n-1)\theta-\cos(n+1)\theta]P_{n-1}(\cos\theta)d\theta. \end{align*} \begin{flalign*} &\text{Hence} & a_{n-1}=\frac{2n-1}{4}.\frac{1.3.5.\cdots(2n-3)}{2.4.6.\cdots(2n-2)}\pi& \tag*{\text{[v.\ Art.~82 (1)],}}& \end{flalign*} and \begin{align*} \sin n\theta= & \frac{\pi}{4}.\frac{1.3.\cdots(2n-3)}{2.4.\cdots(2n-2)}\Big[(2n-1)P_{n-1}(\cos\theta)\\ &\quad\quad +(2n+3)\frac{n^2-(n-1)^2}{n^2-(n+2)^2}P_{n+1}(\cos\theta)\\ &\quad\quad +(2n+7)\frac{[n^2-(n-1)^2][n^2-(n+1)^2]}{[n^2-(n+2)^2][n^2-(n+4)^2]}P_{n+3}(\cos\theta)+\cdots\Big]. \tag{10} \end{align*} \EXAMPLE{S} 1.\quad Show that \[ \csc\theta=\frac{\pi}{2}\Big[1 +5 \Big(\frac{1}{2}\Big)^2P_2(\cos\theta) +9 \Big(\frac{1.3}{2.4}\Big)^2P_4(\cos\theta) +13\Big(\frac{1.3.5}{2.4.6}\Big)^2P_6(\cos\theta)+\cdots\Big] \] whence \[ \frac{1}{\sqrt{1-x^2}}=\frac{\pi}{2}\Big[1 +5 \Big(\frac{1}{2}\Big)^2P_2(x) +9 \Big(\frac{1.3}{2.4}\Big)^2P_4(x) +13\Big(\frac{1.3.5}{2.4.6}\Big)^2P_6(x)+\cdots\Big] \] [v.\ Art.~90 (4) and Art.~82].\\ 2.\quad Show that \[ \ctn\theta=\frac{\pi}{2}\Big[3\Big(\frac{1}{2}\Big)P_1(\cos\theta) +7 \Big(\frac{3}{4}\Big)\Big(\frac{1}{2}\Big)^2P_3(\cos\theta) +11\Big(\frac{5}{6}\Big)\Big(\frac{1.3}{2.4}\Big)^2P_5(\cos\theta)+\cdots\Big] \] whence \[ \frac{1}{\sqrt{1-x^2}}=\frac{\pi}{2}\Big[3\Big(\frac{1}{2}\Big)P_1(x) +7 \Big(\frac{3}{4}\Big)\Big(\frac{1}{2}\Big)^2P_3(x) +11\Big(\frac{5}{6}\Big)\Big(\frac{1.3}{2.4}\Big)^2P_5(x)+\cdots\Big] \] [v.\ Art.~90 (4) and Art.~82].\\ 3.\quad By integrating the result of Ex.~1 and simplifying by the aid of Art.~96 (5), obtain the development \begin{align*} \sin^{-1}x=\frac{\pi}{2}\Big[3 & \Big(\frac{1}{2}\Big)^2P_1(x) +7 \Big(\frac{1}{2.4}\Big)^2P_3(x)\\ &+11\Big(\frac{1.3}{2.4.6}\Big)^2P_5(x) +15\Big(\frac{1.3.5}{2.4.6.8}\Big)^2P_7(x)+\cdots\Big] \end{align*} % -----File: 194.png whence \qquad $\displaystyle\theta=\frac{\pi}{2}\Big[P_0(\cos\theta) -3\Big(\frac{1}{2}\Big)^2P_1(\cos\theta) -7\Big(\frac{1}{2.4}\Big)^2P_3(\cos\theta)$ \[ -11\Big(\frac{1.3}{2.4.6}\Big)^2P_5(\cos\theta)-\cdots\Big]. \] 4.\quad By integrating the result of Ex.~2 and simplifying by the aid of Art.~96 (5) obtain \begin{gather*} \sqrt{1-x^2}=\frac{\pi}{2}\Big[\frac{1}{2} -5\Big(\frac{1}{4}\Big)\Big(\frac{1}{2}\Big)^2P_2(x) -9\Big(\frac{3}{6}\Big)\Big(\frac{1}{2.4}\Big)^2P_4(x) \\ -13\Big(\frac{5}{8}\Big)\Big(\frac{1.3}{2.4.6}\Big)^2P_6(x)+\cdots\Big] \end{gather*} whence \[ \sin\theta=\frac{\pi}{2}\Big[\frac{1}{2}P_0(\cos\theta) -5\Big(\frac{1}{4}\Big)\Big(\frac{1}{2}\Big)^2P_2(\cos\theta) -9\Big(\frac{3}{6}\Big)\Big(\frac{1}{2.4}\Big)^2P_4(\cos\theta) -\cdots\Big]. \] To make clearer the analogy of development in Zonal Harmonic Series with development in Fourier's Series we give on page \label{was184}\pageref{was185} a cut representing the first seven Surface Zonal Harmonics $P_1(\cos\theta)$, $P_2(\cos\theta)$, $\cdots$ $P_7(\cos\theta)$, which are of course somewhat complicated Trigonometric curves resembling roughly $\cos\theta$, $\cos2\theta$, $\cdots$ $\cos7\theta$; and on page \pageref{was186}, the first four successive approximations to the Zonal Harmonic Series \[ \frac{1}{2}+\frac{3}{4}P_1(\cos\theta) -\frac{7}{8}.\frac{1}{2}P_3(\cos\theta) +\frac{11}{12}.\frac{1.3}{2.4}P_5(\cos\theta) -\cdots \tag*{\smallrom{I}} \] [v.\ (1) Art.~93], and \begin{multline} \frac{\pi}{2}\Big[P_0(\cos\theta) -3\Big(\frac{1}{2}\Big)^2P_1(\cos\theta) -7\Big(\frac{1}{2.4}\Big)^2P_3(\cos\theta) \\ -11\Big(\frac{1.3}{2.4.6}\Big)^2P_5(\cos\theta) -\cdots\Big] \tag*{\smallrom{II}} \end{multline} (v.\ Ex.~3 Art.~97). \smallrom{I} is equal to 1 from $\theta=0$ to $\theta=\dfrac{\pi}{2}$, and to 0 from $\theta=\dfrac{\pi}{2}$ to $\theta=\pi$; and \smallrom{II} is equal to $\theta$ from $\theta=0$ to $\theta=\pi$. The figures on page \pageref{was186} are constructed on precisely the same principle as those on pages \pageref{was63} and \pageref{was64}, with which they should be carefully compared. \pngcent{013.png}{1720} \label{was185} \begin{center} {\scriptsize The curves $y=P_0(\cos\theta)$, $y=P_1(\cos\theta)$, \ldots $y=P_7(\cos\theta)$. \qquad (v.\ page \pageref{was184}.)} \end{center} \mypara{98.} By applying \emph{Gauss's Theorem} (B.\ O.\ Peirce, Newt.\ Pot.\ Func.\ §~31) or the special Form of \emph{Green's Theorem}, \[ \iiint\nabla^2Vdxdydz=\int D_nVds=-4\pi\iiint\rho dxdydz, \] % -----File: 195.png % -----File: 196.png % -----File: 197.png \noindent[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 \[ 4\pi\rho\kappa = D_nV_1-D_nV_2. \tag{1} \] where $\rho$ is the density and $\kappa$ the thickness of the shell, $V_1$ the value of the potential function due to the shell at an internal point and $V_2$ its value at an external point, and where $D_n$ 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 $\rho\kappa$ in (1) by $\sigma$ where $\sigma$ is the surface density, whence \[ 4\pi\sigma = D_nV_1-D_nV_2 \tag{2} \] (v.\ Peirce, N.~P.~F.\ §§~45, 46, and 47). \pngcent{014.png}{1321} \label{was186} %[Illustration: I.] %[Illustration: II.] 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 $\rho$ and radius $a$. Since $V$ must be a solution of Laplace's Equation and must be finite both when $r=0$ and $r=\infty$ we have \begin{align*} V_1&=\sum A_mr^mP_m(\cos\theta)\\ V_2&=\sum B_m \frac{1}{r^{m+1}}P_m(\cos\theta). \end{align*} $V_1$ and $V_2$ must approach the same limiting values as $r$ approaches $a$. Hence \begin{flalign*} &&\frac{B_m}{a^{m+1}}&=A_ma^m&\\ &\text{or} & B_m=A&_ma^{2m+1}.&\\ &&D_nV_1=D_rV_1=\sum &mr^{m-1}A_mP_m(\cos\theta),&\\ &&D_nV_2=D_rV_2=-\sum(&m+1)\frac{A_ma^{2m+1}}{r^{m+2}}P_m(\cos\theta).& \end{flalign*} Therefore by (1) \[ 4\pi\rho\kappa=\sum(2m+1)A_ma^{m-1}P_m(\cos\theta) \] if $\kappa$ is the thickness of the shell. % -----File: 198.png \begin{flalign*} &\indent\text{Let} & \rho=f(\cos\theta)=&\sum C_mP_m(\cos\theta) && \phantom{\indent Let}\\ &\text{where} & C_m=\frac{2m+1}{2}&\int\limits_{-1}^1 f(x)P_m(x)dx && \tag*{by Art.~90 (2).}\\ &\indent\text{Then} & 4\pi\kappa C_m=(2&m+1)A_ma^{m-1}, && \tag*{and} \\ &&A_m=\frac{4\pi\kappa C_m}{(2m+1)a^{m-1}},\quad &\text{and}\quad B_m=\frac{4\pi\kappa}{2m+1}C_ma^{m+2}, \\ &\text{and} & V_1=4\pi a\kappa\sum&\frac{C_m}{2m+1}\frac{r_m}{a_m}P_m(\cos\theta), & \tag{3}\\ &\text{and} & V_2=4\pi a\kappa\sum&\frac{C_m}{2m+1}\frac{a^{m+1}}{r^{m+1}}P_m(\cos\theta). & \tag{4} \end{flalign*} \mypara{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 $Cr^kP_m(\cos\theta)$. Let $a$ be the radius of the outer surface and $b$ be the radius of the inner surface of the shell. 1st.---Let $\rho=Cr^kP_m(\cos\theta)$. Then for the shell of radius $s$ and thickness $ds$ \begin{flalign*} &&V_1={}&4\pi sds\frac{Cs^k}{2m+1}\frac{r^m}{s^m}P_m(\cos\theta) && \tag*{by (3) Art.~98,}\\ &\text{and}\hspace{6em} &V_2=4&\pi sds\frac{Cs^k}{2m+1}\frac{s^{m+1}}{r^{m+1}}P_m(\cos\theta) && \tag*{by (4) Art.~98.} \end{flalign*} Then if $ra$ \[ V=\int\limits_b^a V_2=\frac{4\pi C}{(2m+1)}\frac{(a^{k+m+3}-b^{k+m+3})}{(k+m+3)}\frac{P_m(\cos\theta)}{r^{m+1}}, \tag{2} \] and if $ba,\\ V&=2\pi\rho\bigg[a^2-\frac{2b^3}{3r}-\frac{r^2}{3}\bigg]\quad \text{if}\quad ba$, and $V=$ a constant if $ra. \] 4.\quad If the density at any point of a solid sphere is proportional to its distance from a diametral plane \[ V=\frac{M}{a}\bigg[\frac{a}{r} +\frac{1}{6} \frac{a^3}{r^3}P_2(\cos\theta) -\frac{1.1}{6.8} \frac{a^5}{r^5}P_4(\cos\theta) +\frac{1.1.3}{6.8.10}\frac{a^7}{r^7}P_6(\cos\theta) -\cdots\bigg] \] if $r>a$. Compare Ex.~2 Art.~80. \mypara{100.} We have seen in Art.~18 (\emph{c}) (3) that \[ Q_m(x)=CP_m(x)\int\frac{dx}{(1-x^2)[P_m(x)]^2}, \tag{1} \] no constant term being understood with $\displaystyle\int\frac{dx}{(1-x^2)[P_m(x)]^2}$. $\dfrac{1}{(1-x^2)[P_m(x)]^2}$ is a rational fraction and becomes infinite only for $x=1$, $x=-1$, and for the roots of $P_m(x)=0$, all of which are real and lie between $-1$ and 1, as can be proved by the aid of the relation $P_m(x)=\dfrac{1}{2^mm!}\dfrac{d^m(x^2-1)^m}{dx^m}$. If $x^2>1$ $\displaystyle\int\limits_{x}^\infty\frac{dx}{(1-x^2)[P_m(x)]^2}$ is finite and determinate and contains no constant term. Hence if $x^2>1$ \[ Q_m(x)=-P_m(x)\int\limits_{x}^\infty\frac{dx}{(1-x^2)[P_m(x)]^2}=P_m(x)\int\limits_{x}^\infty\frac{dx}{(x^2-1)[P_m(x)]^2} \tag{2} \] for the constant factor of $Q_m(x)$ has been chosen so that $C=-1$. % -----File: 200.png If $x^2 < 1$ 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 \begin{flalign*} && &\quad Q_0(x) = \dfrac{1}{2}\log \dfrac{1+x}{1-x} &&\tag{3}\\ &&&Q_1(x) = -1+ \dfrac{x}{2} \log \dfrac{1+x}{ 1-x} &&\tag{4} \intertext{if $x^2<1$.}\\[-7ex] &\text{\indent (2) gives us}&&\quad Q_0(x) = \dfrac{1}{2}\log \dfrac{x+1}{x-1}&&\tag{5}\hspace{4em}\\ &&&Q_1(x) = -1 + \dfrac{x}{2}\log \dfrac{x+1}{ x-1}& &\tag{6} \end{flalign*} if $x^2>1$. From Art.~85 (10) it follows that \begin{align*} Q_m(x) &= C\dfrac{d^m}{dx^m}\bigg[(x^2-1)^m \int\limits^x_0\dfrac{dx}{(x^2-1)^{m+1}}\bigg]\quad\text{if}\quad x^2 < 1,\\ &= C\dfrac{d^m}{dx^m}\bigg[(x^2-1)^m \int\limits^{\infty}_x\dfrac{dx}{(x^2-1)^{m+1}}\bigg]\quad\text{if}\quad x^2 > 1. \end{align*} $C$ can be determined and is equal to $\dfrac{(-1)^{m+1}2^mm!}{(2m)!}$ if $x^2 < 1$, and is equal to $\dfrac{(-1)^m2^m m!}{(2m)!}$ if $x^2 > 1$. \begin{flalign*} &\text{Hence}& Q_m(x) &= \dfrac{(-1)^{m+1}2^mm!}{(2m)!}\dfrac{d^m}{dx^m}\bigg[(x^2-1)^m \int\limits^x_0\dfrac{dx}{(x^2-1)^{m+1}}\bigg]& (7)\\[-2ex] \intertext{if $x^2<1$,}\\[-6ex] &\text{and} & Q_m(x) &= \dfrac{(-1)^m2^m m!}{(2m)!}\dfrac{d^m}{dx^m}\bigg[(x^2-1)^m \int\limits^{\infty}_x\dfrac{dx}{(x^2-1)^{m+1}}\bigg]&(8)\\[-5ex] \end{flalign*} if $x^2>1$. (7) and (8) give us for $Q_0(x)$ and $Q_1(x)$ the values already written in (3), (4), (5), and (6). By the repeated application of the formula \[ (m+1)Q_{m+1}(x) - (2m+1)xQ_m(x) + mQ_{m-1}(x)=0, \tag{9} \] which may be obtained for the case where $x^2 < 1$ from Art.~16 (13) and (14), and for the case where $x^2 > 1$ from Art.~16 (9), any Surface Zonal Harmonic of the Second Kind can be obtained from $Q_0(x)$ and $Q_1(x)$ as given in (3), (4), (5), and (6). % -----File: 201.png Analogous formulas for $p_m(x)$ and $q_m(x)$ can be obtained without difficulty from Art.~16 (4) and (5). They are \begin{flalign*} &&(m + 1)^2q_{m+1}(x) &- (2m + 1)xp_m(x) - m^2 q_{m-1}(x) = 0 \tag{10}\\ &\text{and}\hspace{1em} & p_{m+1}(x) + &(2m + 1)xq_m(x) - p_{m-1}(x) = 0 &\,\tag{11} \end{flalign*} and they hold good for any value of $m$. \EXAMPLE{S} 1.\quad Confirm the values of $Q_0(x)$ and $Q_1(x)$ given in Art.~100 (3), (4), (5), and (6) by expanding them and comparing them with Art.~16 (13), (14), and (9).\\ 2.\quad 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. If $V=\displaystyle \sum\left(A_mr^m + \dfrac{B_m }{r^{m+1}}\right)$ when $\theta = \alpha$ then \[ V=\sum\left(A_mr^m + \dfrac{B_m }{r^{m+1}}\right)\dfrac{P_m(\cos \theta)}{P_m(\cos \alpha)}. \] 3.\quad If $V=\displaystyle \sum\left(A_mr^m + \dfrac{B_m }{r^{m+1}}\right)$ when $\theta = \alpha$, and $V = 0$ when $\theta = \beta$, \[ V=\sum\left(A_mr^m + \dfrac{B_m }{r^{m+1}}\right)\left[\dfrac{Q_m(\cos\beta)P_m(\cos\theta)-P_m(\cos\beta)Q_m(\cos\theta)}{P_m(\cos \alpha)Q_m(\cos\beta)-P_m(\cos\beta)Q_m(\cos \alpha)}\right]. \] 4.\quad Find $V$ for points corresponding to values of $\theta$ between $\alpha$ and $\beta$ when $V$ can be given in terms of whole powers of $r$ for $\theta = \alpha$ and for $\theta = \beta$.\\ 5.\quad Find by the method of Art.~16 solutions of Legendre's Equation of the form \begin{align*} z = {}_1P_m(x) &= 1 + \dfrac{m(m+1) }{ 2} (x - 1) + \dfrac{(m-1)m(m+1)(m+2) }{ 2^2 (2!)^2} (x - 1)^2\\ &\quad+ \dfrac{(m-2)(m-1)m(m+1)(m+2)(m+3) }{2^3 (3!)^2} (x-1)^3 + \cdots,\\ z = {}_{-1}P_m(x) &= 1 - \dfrac{m(m+1)}{2}(x + 1) + \dfrac{(m-1)m(m+1)(m+2) }{2^2(2!)^2} (x + 1)^2\\ &\quad+ \dfrac{(m-2)(m-1)m(m+1)(m+2)(m+3)}{2^3(3!)^2} (x + 1)^3 + \cdots. \end{align*} If $m$ is a whole number, ${}_1P_m(x) = P_m(x)$ and ${}_{-1}P _m(x) = (-1)^m P_m (x)$. No matter what the value of $m$, ${}_1P_m (x)$ is absolutely convergent for $-1 < x < 3$, and ${}_{-1}P_m(x)$ is absolutely convergent for $ -3 < x < 1$.\\ % -----File: 202.png 6.\quad By the aid of (7) Art.~16 show that \[ \arraycolsep=0.3em \begin{array}{c@{\quad} |@{\quad} c} V= \dfrac{1}{\sqrt{r}} \sin (n \log r)k_n(\cos \theta), & V= \dfrac{1}{\sqrt{r}} \sin (n \log r)l_n(\cos \theta),\\[1.3em] V= \dfrac{1}{\sqrt{r}} \cos (n \log r)k_n(\cos \theta), & V= \dfrac{1}{\sqrt{r}} \cos (n \log r)l_n(\cos \theta), \end{array} \] are solutions of Laplace's Equation \begin{flalign*} &&r&D_r^2(rV) + \dfrac{1}{\sin \theta} D_\theta (\sin \theta D_\theta V) = 0, &\text{if}\\ &&k_n(x) = p_{-\frac{1}{2}+ni} (x) &= 1 + \dfrac{n^2 + \Big(\dfrac{1}{2}\Big)^2}{ 2!} x^2 + \dfrac{\Big[n^2 + \Big(\dfrac{1}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{5}{2}\Big)^2\Big]}{ 4!} x^4\\[1ex] && &+ \dfrac{\Big[n^2 + \Big(\dfrac{1}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{5}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{9}{2}\Big)^2\Big]}{ 6!} x^6 + \cdots \end{flalign*} and\\[-5ex] \begin{flalign*} &&l_n(x) = -q_{-\frac{1}{2}+ni} (x) &= x + \dfrac{n^2 + \Big(\dfrac{3}{2}\Big)^2}{ 3!} x^3 + \dfrac{\Big[n^2 + \Big(\dfrac{3}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{7}{2}\Big)^2\Big]}{ 5!} x^5&&\\[1ex] && &+ \dfrac{\Big[n^2 + \Big(\dfrac{3}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{7}{2}\Big)^2\Big]\Big[n^2 + \Big(\dfrac{11}{2}\Big)^2\Big]}{ 7!} x^7 + \cdots&& \end{flalign*} $k_n (x)$ and $l_n (x)$ are convergent if $x^2 < 1$, but are divergent if $x^2 = 1$.\\ 7.\quad Show by the aid of Example 5 that \[ \arraycolsep=0.3em \begin{array}{c@{\quad} |@{\quad} c} V= \dfrac{1}{\sqrt{r}} \sin (n \log r)K_n(\cos \theta), & V= \dfrac{1}{\sqrt{r}} \sin (n \log r)K_n(-\cos \theta),\\[1em] V= \dfrac{1}{\sqrt{r}} \cos (n \log r)K_n(\cos \theta), & V= \dfrac{1}{\sqrt{r}} \cos (n \log r)K_n(-\cos \theta), \end{array} \] \begin{flalign*} &\text{are solutions of}& &rD_r^2(rV) + \dfrac{1}{\sin \theta} D_\theta (\sin \theta D_\theta V) = 0&\phantom{are solutions of}\\[-5ex] \end{flalign*} \begin{flalign*} &\text{if }& K_n(x) =& {}_1P_{-\frac{1}{2}+ni}(x)=1-\dfrac{n^2+\Big(\dfrac{1}{2}\Big)^2}{2}(x-1)&\\[0.5em] & & &+ \dfrac{\Big[n^2+\Big(\dfrac{1}{2}\Big)^2\Big]\Big[n^2+\Big(\dfrac{3}{2}\Big)^2\Big]}{2^2(2!)^2}(x-1)^2 &\\[0.5em] & & & - \dfrac{\Big[n^2+\Big(\dfrac{1}{2}\Big)^2\Big]\Big[n^2+\Big(\dfrac{3}{2}\Big)^2\Big]\Big[n^2+\Big(\dfrac{5}{2}\Big)^2\Big]}{2^2(3!)^2}(x-1)^3+\cdots& \end{flalign*} % -----File: 203.png and\\[-5ex] \begin{align*} K_n(-x) = &{}_{-1}P_{-\frac{1}{2} + ni}(x) = 1 + \dfrac{n^2+\Big(\dfrac{1}{2}\Big)^2}{2} (x + 1)\\ &\phantom{{}_-}+\dfrac{ \Big[n^2+\Big(\dfrac{1}{2}\Big)^2\Big] \Big[n^2+\Big(\dfrac{3}{2}\Big)^2\Big]}{2^2(2!)^2}(x+1)^2\\ &\phantom{{}_-}+\dfrac{ \Big[n^2+\Big(\dfrac{1}{2}\Big)^2\Big] \Big[n^2+\Big(\dfrac{3}{2}\Big)^2\Big] \Big[n^2+\Big(\dfrac{5}{2}\Big)^2\Big]}{{2^3(3!)^2}} (x+1)^3 +\cdots. \end{align*} $K_n(\cos \theta)$ is convergent except for $\theta =\pi$, and $K_n(-\cos\theta)$ is convergent except for $\theta = 0$. $k_n(x)$, $l_n(x)$, $K_n(x)$, and $K_n(-x)$ are sometimes called \emph{Conal Harmonics}. They are particular values of $z$ which satisfy Legendre's Equation written in the form \[ (1-x^2)\dfrac{d^2z}{dx^2}- 2x \dfrac{dz}{dx}-\Big(n^2+ \dfrac{1}{4}\Big)z=0. \] 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.\quad If $V=f(r)$ when $\theta=\beta,$ \[ V = \dfrac{1}{\pi\sqrt{r}} \int\limits_{-\infty}^\infty d\lambda \int\limits_0^\infty e^{\tfrac{\lambda}{2}} f(e^\lambda) \dfrac{K_\alpha(\cos\theta)}{K_\alpha(\cos\beta)} \cos[\alpha(\lambda-\log r)]d\alpha; \quad\text{if}\quad\theta<\beta. \] 9.\quad If $V=f(r)$ when $\theta = \beta$ and $r \beta$\;\;$\cos \theta$ must be replaced by $(- \cos \theta)$ in examples 8, 9, and 10.\\ 12.\quad If $V=f(r)$ when $\theta=\beta$, and $V=0$ when $\theta=\gamma$, \begin{multline*} % recast to fit line V = \dfrac{1}{\pi\sqrt{r}} \int\limits^{\infty}_{-\infty}d\lambda \int\limits^{\infty}_0 e^{\tfrac{\lambda}{2}} f(e^{\lambda})\\ \dfrac{k_\alpha(\cos \theta)l_\alpha(\cos \gamma) - k_\alpha(\cos \gamma)l_\alpha(\cos \theta) } {k_\alpha(\cos \beta)l_\alpha(\cos \gamma) - k_\alpha(\cos \gamma) l_\alpha(\cos \beta) } \cos[\alpha(\lambda - \log r)] d\alpha; \end{multline*} if $\beta<\theta<\gamma$.\\ 13.\quad If $V=f(r)$ when $\theta = \beta$ and $a4,\\ &&& = 720 \int\limits_{-1}^1 (\mu^2-1)^4 d\mu=\dfrac{4096}{7}\quad\text{if}\quad m=4,\qquad \\ &\text{and} &B_{2,4} = \dfrac{1}{2^44!}&.\dfrac{9}{4}. \dfrac{2!}{6!}.\dfrac{4096}{7}=\dfrac{1}{105}. \end{flalign*} By a like process we find \begin{flalign*} &&&\qquad\qquad B_{2,3} = 0\quad\text{and}\quad B_{2,2} = \dfrac{1}{42}. &&\tag*{Hence} \\[1ex] && \sin^2\theta \cos^2\theta \sin\phi &\cos\phi = \dfrac{1}{42} P_2^2(\mu)\sin 2\phi + \dfrac{1}{105} P_4^2(\mu)\sin 2\phi, \tag{1} \\[1ex] && &=\dfrac{1}{42}\sin 2\phi \sin^2\theta \dfrac{d^2P_2(\mu)}{d\mu^2} + \dfrac{1}{105}\sin 2\phi \sin^2\theta \dfrac{d^2P_4(\mu)}{d\mu^2},&\tag{2}\\[1ex] && &=\dfrac{1}{14}\sin^2\theta \sin 2\phi + \dfrac{1}{14} \sin^2\theta (7\mu^2-1)\sin 2\phi. \tag{3} \end{flalign*} The required expression might have been obtained without using the formulas of Art.~107, by a very simple device, as follows: \[ \sin^2\theta \cos^2\theta \sin\phi \cos\phi = \dfrac{1}{2}\mu^2\sin^2\theta \sin 2\phi.\tag{4} \] % -----File: 216.png If now we can express $\mu^2$ in the form $\displaystyle\sum\dfrac{ d^2P_m(\mu)}{d\mu^2}$ the work will be done. \begin{flalign*} && \mu^2&=\dfrac{1}{4.3}\dfrac{ d^2(\mu^4)}{d\mu^2}, &\\[0.5em] && \mu^4&=\dfrac{8}{35}P_4(\mu)+\dfrac{4}{7}P_2(\mu)+\dfrac{1}{5}P_0(\mu), &\text{(5) Art.~95.}\\[0.5em] && \dfrac{d^2(\mu^4)}{d\mu^2}&=\dfrac{8}{35}\dfrac{d^2P_4(\mu)}{d\mu^2}+\dfrac{4}{7}\dfrac{d^2P_2(\mu)}{d\mu^2}; &\\[0.5em] &\text{whence}& \mu^2&=\dfrac{2}{105}\dfrac{d^2P_4(\mu)}{d\mu^2}+\dfrac{1}{21}\dfrac{d^2P_2(\mu)}{d\mu^2},& \end{flalign*} and substituting this value in (4) we get (2). \EXAMPLE{S} 1.\quad Show that \begin{multline*} \cos^3\theta \sin^3\theta \sin\phi \cos^2\phi=\left[\dfrac{1}{6930}P_6^3(\mu)+\dfrac{1}{1540}P_4^3(\mu)\right]\sin 3\phi\\ -\left[\dfrac{2}{693}P_6^1(\mu)-\dfrac{1}{770}P_4^1(\mu)-\dfrac{1}{63}P_2^1(\mu)\right]\sin\phi. \end{multline*} 2.\quad Show that \[ \cos 2\phi=2 \cos 2\phi\left[\dfrac{5}{4!}P_2^2(\mu)+\dfrac{9.2!}{6!}P_4^2(\mu)+\dfrac{13.4!}{8!}P_6^2(\mu)+\cdots\right]. \] 3.\quad If in a problem on the Potential Function $V=f(\mu,\phi)$ when $r = a$, we shall obviously have \[ V= \sum\limits^{m=\infty}_{m=0}\dfrac{ r^m}{a^m}\bigg[A_{0,m} P_m(\mu) + \sum\limits^{n=m}_{n=1} (A_{n,m} \cos n\phi+B_{n,m} \sin n\phi)P_m^n(\mu)\bigg] \] at an internal point and \[ V= \sum\limits^{m=\infty}_{m=0}\dfrac{a^{m+1}}{r^{m+1}}\bigg[A_{0,m} P_m(\mu) + \sum\limits^{n=m}_{n=1} (A_{n,m} \cos n\phi+B_{n,m} \sin n\phi)P_m^n(\mu)\bigg] \] at an external point, where $A_{0,m}$, $A_{n,m}$, and $B_{n,m}$ have the values given in (1), (2), and (3) Art.~107.\\ 4.\quad Solve problems (3), (4), and (5) of Art.~94 for the case where $V$ is not symmetrical with respect to an axis. \mypara{109.} Any Solid Spherical Harmonic $r^mY_m(\mu,\phi)$ being a value of $V$ that satisfies Laplace's Equation in Spherical Coördinates will transform into a function of $x$, $y$, and $z$ satisfying $\nabla^2 V = 0$ if we change to a set of rectangular % -----File: 217.png 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 $m$th degree. For each term of $r^m \cos n\phi P_m^n(\mu)$ is of the form \begin{flalign*} &&&Cr^m \cos^{n-2k} \phi \sin^{2k} \phi \sin^n \theta \cos^{m-2l-n} \theta &&\\[1ex] &\text{where}&&2k < n + 1\quad\text{and}\quad 2l < m - n + 1.&&\phantom{where} \end{flalign*} This may be written\\[-5ex] \begin{flalign*} \intertext{\qquad$Cr^{2l}.r^{m-2l-n}\cos^{m-2l-n} \theta.r^{n-2k}\sin^{n-2k} \theta \cos^{n-2k} \phi.r^{2k}\sin^{2k} \theta \sin^{2k} \phi$} &\text{which becomes}&& C(x^2 + y^2 + z^2)^l x^{m-2l-n} y^{n-2k} z^{2k},&&\phantom{which becomes} \end{flalign*} and is a homogeneous rational integral Algebraic function of $x$, $y$, and $z$ of the $m$th degree. The same thing may be shown of each term of $r^m \sin n\phi P_m^n(\mu)$. Consequently $r^mY_m(\mu,\phi)$ is a homogeneous rational integral Algebraic function of the $m$th degree in $x$, $y$, and $z$. \mypara{110.} Any homogeneous rational integral Algebraic function $S_m(x, y, z)$ of the $m$th degree in $x$, $y$, and $z$, which is a value of $V$ satisfying $\nabla^2 V = 0$ contains $2m + 1$ arbitrary constant coefficients. For $S_m(x, y, z)$ will in general consist of $\dfrac{(m+1)(m+2)}{2}$ terms and will therefore contain $\dfrac{(m+1)(m+2)}{2}$ coefficients. $\nabla^2 S_m(x, y, z)$ will be homogeneous of the $(m - 2)$d degree and will contain $\dfrac{m(m-1)}{2}$ coefficients, which, of course, will be functions of the coefficients in $S_m(x, y, z)$. Since $\nabla^2 S_m(x, y, z) = 0$ independently of the numerical values of $x$, $y$, and $z$ the $\dfrac{m(m-1)}{2}$ coefficients in $\nabla^2 S_m(x, y, z)$ must be separately zero, and that fact will give us $\dfrac{m(m-1)}{2}$ equations of condition between the $\dfrac{(m+1)(m+2)}{2}$ original coefficients and will leave $\dfrac{(m+1)(m+2)}{2} -\dfrac{ m(m-1)}{2}$ or $2m + 1$ of them undetermined. $S_m(x, y, z)$ contains, then, the same number of arbitrary coefficients as $r^m Y_m(\mu,\phi)$. We can then choose the coefficients in $r^m Y_m(\mu,\phi)$ so that it will transform into any given $S_m(x, y, z)$. Consequently a Solid Spherical Harmonic of the $m$th degree might be defined as \textit{a homogeneous rational integral Algebraic function of x, y, and z,} $S_m(x, y, z)$, \textit{of the $m$th degree satisfying the equation} $\nabla^2 S_m(x, y, z) = 0$; and a Surface Spherical Harmonic of the $m$th degree as such a function divided by $(x^2 + y^2 + z^2)^{\frac{m}{2}}$, that is by $r^m$. % -----File: 218.png \EXAMPLE{S} 1.\quad Show that if $S_m(x,y,z)$ is a Solid Spherical Harmonic of the $m$th degree \[ \nabla^2[r^nS_m(x,y,z)]=n(2m+n+1)r^{n-2}S_m(x,y,z). \] \emph{Suggestion:} \[ \nabla^2S_m=0. \quad \nabla^2r=\dfrac{2}{r}.\quad D_rS_m=\dfrac{mS_m}{r}.\quad (D_xr)^2+(D_yr)^2+(D_zr)^2=1. \] 2.\quad Show that if $f_n(x,y,z)$ is a rational integral homogeneous function of $x$, $y$, and $z$ of the $n$th degree it can be expressed in the form \[ f_n(x,y,z)=S_n(x,y,z) +r^2S_{n-2}(x,y,z) + r^4S_{n-4}(x,y,z) + \cdots,\tag{1} \] terminating with $r^{n-1}S_1(x,y,z)$ if $n$ is odd, and with $r^nS_0(x,y,z)$ if $n$ is even.\\ \emph{Suggestion}: If a term $rS_{n-1}$ were present in the second member of (1), and we were to operate with $\nabla^2$ on both members we should by Ex.~1 have a term $\dfrac{2n}{r }S_{n-1}$ which would be irrational when all the other terms of the resulting equation were rational. No such term, then, could occur. In the same way it may be shown by operating twice on (1) with $\nabla^2$ that there can be no term $r^3S_{n-3}$ in (1); and thus step by step we can reach the result formulated in (1).\\ 3.\quad Express $x^2yz$ in the form $S_4+r^2S_2+r^4S_0$. \begin{flalign*} &\text{\indent \emph{Suggestion}: let}& x^2yz&=S_4+r^2S_2+r^4S_0 \rule{6em}{0em}&\ \intertext{and take $\nabla^2$ of both members we get}\\[-4ex] && 2yz&=14S_2+20r^2S_0.&\\[1ex] &\text{Operate again with $\nabla^2$.}& 0&=120S_0. &\text{Whence} \end{flalign*} \[ S_0=0, \quad S_2=\dfrac{1}{7}yz,\quad \text{and}\quad S_4=\dfrac{1}{7}(6x^2-y^2-z^2)yz. \] 4.\quad Express $\sin^2\theta \cos^2\theta \sin\phi \cos\phi$ in terms of Surface Spherical Harmonics. \begin{flalign*} &\text{\indent \emph{Suggestion:}}& &\sin^2\theta \cos^2\theta \sin\phi \cos\phi =\dfrac{x^2yz}{r^4}.\rule{6em}{0em}& \end{flalign*} For result v.\ Art.~108 (3). \mypara{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 $\nabla^2V=0$ 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 $x\cos\alpha_1+y\cos\alpha_2+z\cos\alpha_3$, $x\cos\beta_1+y\cos\beta_2+z\cos\beta_3$ and $x\cos\gamma_1+y\cos\gamma_2+z\cos\gamma_3$ respectively where the cosines are the \emph{direction cosines} of the new axes, and it will leave % -----File: 219.png the function a homogeneous function of the $m$th degree in the new variables, and on dividing this by the $m$th power of the unchanged radius vector we shall have a Surface Spherical Harmonic of the $m$th degree. \mypara{112.} We have seen in Art.~75 that if $(x_1, y_1, z_1)$ are the coördinates of a given point \[ V = \dfrac{1}{\sqrt{ (x-x_1)^2 + (y-y_1)^2 + (z-z_1)^2}}\tag{1} \] is a solution of Laplace's Equation $\nabla^2V=0$, and transforming to spherical coördinates that \[ V=\dfrac{1}{\sqrt{r^2-2rr_1[\cos\theta \cos\theta_1 + \sin\theta \sin\theta_1 \cos(\phi-\phi_1)] + r_1^2}}\tag{2} \] is a solution of \[ rD_r^2(rV) + \dfrac{1}{\sin\theta} D_\theta (\sin\theta D_\theta V) + \dfrac{1}{\sin^2 \theta} D_\phi^2 V = 0. \tag{3} \] If $\gamma$ is the angle between the radii vectores $r$ and $r_1$ of the points $(x, y, z)$ and $(x_1, y_1, z_1)$ (1) can be written \[ V = \dfrac{1}{\sqrt{ r^2-2rr_1 \cos\gamma + r_1^2}}\tag{4} \] which must be equivalent to (2), and hence \[ \cos\gamma = \cos\theta \cos\theta_1 + \sin\theta \sin\theta_1 \cos(\phi - \phi_1). \] (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 \[ V=P_m(\cos\gamma) \] is a solution of the equation \begin{flalign*} &&m(&m+1)V + \dfrac{1}{\sin\theta} D_\theta(\sin\theta D_\theta V) + \dfrac{1 }{ \sin^2\theta} D_\phi^2 V = 0 &&\tag{5} \\ &\text{and that}&& V= r^m P_m(\cos\gamma)\quad\text{and}\quad V=\dfrac{1}{r^{m+1} }P_m(\cos\gamma)&&\phantom{and that} \end{flalign*} are solutions of (3). If we transform our coördinates keeping the origin unchanged and taking as our new polar axis the radius vector of $(x_1, y_1, z_1)$ $\gamma$ becomes our new $\theta$ and $P_m(\cos\gamma)$ reduces to $P_m(\cos\theta)$, a Surface Zonal Harmonic, or a \emph{Legendrian},\footnote{v.\ Art.~74.} of the $m$th degree. It is then a Legendrian having for its axis not the original polar axis but the radius vector of $(x_1, y_1, z_1)$. Since a Legendrian is a Surface Spherical Harmonic, \[ P_m(\cos\gamma) = P_m[\cos\theta \cos\theta_1 + \sin\theta \sin\theta_1 \cos(\phi-\phi_1)] \] is a Surface Spherical Harmonic of the $m$th degree. % -----File: 220.png It is, however, of very special form, since being a determinate function of $\mu$, $\phi$, $\mu_1$, and $\phi_1$ it contains but two arbitrary constants if we regard it as a function of $\mu$ and $\phi$, instead of containing $2m + 1$. It is known as a \textit{Laplace's Coefficient}, or briefly as a Laplacian, of the $m$th degree. We shall soon express it in the regulation form of a Surface Spherical Harmonic. The radius vector of $(x_1, y_1, z_1)$ is called the axis of the Laplacian and the point where the axis cuts the surface of the unit sphere is the \textit{pole} of the Laplacian. We shall represent the Laplacian $P_m(\cos\gamma)$ by $L_m(\mu, \phi, \mu_1, \phi_1)$. Of course $L_m(\mu, \phi, 1, \phi_1) = P_m(\mu) = P_m(\cos\theta)$, and is really independent of $\phi$. \mypara{113.} \emph{If the product of a Surface Spherical Harmonic of the $m$th degree by a Laplacian of the same degree is integrated over the surface of the unit sphere, the result is equal to $\dfrac{4\pi }{ 2m+1}$ multiplied by the value of the Spherical Harmonic at the pole of the Laplacian.} That is, \[ \int\limits^{2\pi}_0 d\phi \int\limits^1_{-1} Y_m(\mu, \phi) L_m(\mu, \phi, \mu_1, \phi_1) d\mu = \dfrac{4\pi }{ 2m+1} Y_m(\mu_1, \phi_1).\tag{1} \] Transform to the axis of the Laplacian as a new polar axis, and let $Z_m(\mu, \phi)$ be the transformed Spherical Harmonic. $L_m(\mu, \phi, \mu_1, \phi_1)$ will become $P_m(\mu)$, and (1) will be proved if we can show that \begin{gather*} \int \limits^{2\pi}_0 d\phi \int \limits^1_{-1} Z_m(\mu,\phi) P_m(\mu)d\mu = \dfrac{4\pi }{ 2m+1} Z_m\text{(1, 0)}.\tag{2}\\ Z_m(\mu, \phi) P_m(\mu) = A_0[P_m(\mu)]^2 + \sum \limits^{n=m}_{n=1} (A_n \cos n\phi + B_n \sin n\phi) P_m^n(\mu) P_m(\mu) \end{gather*} \indent(v.\ (5) Art.~102). \begin{flalign*} &\phantom{(v.\ (5) Art.~89).}&&\int\limits^{2\pi}_0 Z_m(\mu,\phi) P_m(\mu)d\phi = 2\pi A_0[P_m(\mu)]^2 &\text{and}\\ &&&\int\limits^1_{-1} d\mu \int\limits^{2\pi}_0 Z_m(\mu,\phi) P_m(\mu)d\phi = \dfrac{4\pi}{2m+1} A_0& \text{(v.\ (5) Art.~89).} \end{flalign*} But $Z_m(1,0) = A_0$, since $P_m(1) = 1$ and $P_m^n(1)$ contains $(1 - 1)^{\frac{n}{2}}$ as a factor and is equal to zero. Hence (2) is proved. % -----File: 221.png \mypara{114.} We can now express a Laplacian in the regulation form as a Spherical Harmonic, by the formulas of Art.~107. \begin{flalign*} L_m(\mu,\phi,\mu_1,\phi_1) =& P_m(\cos \gamma) = P_m[\cos \theta \cos \theta_1 + \sin \theta \sin \theta_1 \cos(\phi - \phi_1)]&\\ =&\sum\limits_{k = 0}^{k = \infty}\bigg[A_{0,k} P_k(\mu) + \sum\limits_{n = 1}^{n = k}(A_{n,k} \cos n\phi + B_{n,k} \sin n\phi) P_k^n (\mu)\bigg]& \end{flalign*} \begin{flalign*} &\text{where}& A_{0,m} &= \dfrac{2m + 1}{4\pi} \int\limits_{0}^{2\pi}d\phi \int\limits_{-1}^{1} L_m(\mu,\phi,\mu_1,\phi_1)P_m(\mu)d\mu.&&\phantom{where}\\ &&&= \dfrac{2m + 1}{4\pi} \dfrac{4\pi}{2m + 1}P_m(\mu_1)=P_m(\mu_1) &\tag*{by (1) Art.~113,} \end{flalign*} \begin{flalign*} && A_{n,m} &= \dfrac{2m + 1}{2\pi}\dfrac{(m - n)!}{(m + n)!} \int\limits_{0}^{2\pi}d\phi \int\limits_{-1}^{1}L_m(\mu,\phi,\mu_1,\phi_1) \cos n\phi P_m^n(\mu)d\mu\\ && &= \dfrac{2(m-n)!}{(m+n)!}\cos n\phi_1 P_m^n(\mu_1) &\tag*{by (1) Art.~113, and}\\ && B_{n,m} &= \dfrac{2m+1}{2\pi} \dfrac{(m-n)!}{(m+n)!} \int\limits_{0}^{2\pi}d\phi \int\limits_{-1}^{1}L_m(\mu,\phi,\mu_1,\phi_1) \sin n\phi P_m^n(\mu)d\mu\\ && &= \dfrac{2(m-n)!}{(m+n)!}\sin n\phi_1 P_m^n(\mu_1) &\tag*{by (1) Art.~113,} \end{flalign*} and $A_{0,k} = A_{n,k} = B_{n,k} = 0$ by Art.~105 unless $k = m$. Hence \begin{multline*} L_m(\mu,\phi,\mu_1,\phi_1) =\\ P_m(\mu)P_m(\mu_1) + 2\sum\limits_{n=1}^{n=m} \left[\dfrac{(m-n)!}{(m+n)!}P_m^n(\mu)P_m^n(\mu_1)\cos n(\phi - \phi_1)\right]. \tag{1} \end{multline*} Each term of a Laplacian involves a numerical coefficient, a factor which is a function of $\mu$, a second factor which is the same function of $\mu_1$, and a third factor which is of the form $\cos k(\phi - \phi_1)$. 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 $\mu_1$. By the aid of (1) we can write (5) Art.~107 more compactly. It becomes \begin{flalign*} &&f(\mu,\phi) &= \dfrac{1}{4\pi} \sum\limits_{m=0}^{m=\infty} (2m + 1) \int\limits_{0}^{2\pi} d\phi_1 \int\limits_{-1}^{1} f(\mu_1,\phi_1)L_m(\mu,\phi,\mu_1,\phi_1)d\mu_1 \tag{2}\\ &\text{or}& F(\theta,\phi) &= \dfrac{1}{4\pi} \sum\limits_{m=0}^{m=\infty}(2m + 1) \int\limits_{0}^{2\pi}d\phi_1 \int\limits_{0}^{\pi}F(\theta_1,\phi_1)P_m(\cos \gamma) \sin \theta_1 d \theta_1. &&\tag{3} \end{flalign*} % -----File: 222.png \begin{center} LAPLACIANS. \end{center} \[ \arraycolsep=0.3em \begin{array}{|c|c|c|c|} \hline \rule{3.5em}{0em} &\text{coef.\ of }\cos 0(\phi-\phi_1) &\text{coef.\ of}\cos(\phi-\phi_1) &\text{coef.\ of}\cos 2(\phi-\phi_1)\rule{0em}{1.5em}\\[0.5em] \hline L_0& 1&&\rule{0em}{1.5em}\\[0.5em] L_1& \mu&(1 - \mu^2)^{\frac{1}{2}}&\rule{0em}{1.5em}\\[0.5em] L_2& \dfrac{1}{4}(3\mu^2 - 1) & 3\mu(1-\mu^2)^{\frac{1}{2}} & \dfrac{3}{4}(1 - \mu^2)\\[1em] L_3& \dfrac{1}{4}(5\mu^3 - 3\mu) & \dfrac{3}{8}(1 - \mu^2)^{\frac{1}{2}}(5\mu^2 - 1) & \dfrac{15}{4} \mu(1 -\mu^2)\\[1em] L_4& \dfrac{1}{64}(35\mu^4 - 30\mu^2 + 3) & \dfrac{5}{8}(1- \mu^2)^{\frac{1}{2}}(7\mu^3 - 3\mu) & \dfrac{5}{16}(1 - \mu^2)(7\mu^2 - 1)\\[1em] \hline \end{array} \] \vspace{\baselineskip} \[ \arraycolsep=0.3em \begin{array}{|c|c|c|} \hline \rule{3.5em}{0em} &\text{coef.\ of}\cos 3(\phi-\phi_1) &\text{coef.\ of}\cos 4(\phi-\phi_1)\rule{0em}{1.5em}\\[0.5em] \hline L_0&&\rule{0em}{1.5em}\\[0.5em] L_1&&\rule{0em}{1.5em}\\[0.5em] L_2&&\rule{0em}{1.5em}\\[1em] L_3& \dfrac{5}{8}(1 - \mu^2)^{\frac{3}{2}} & \rule{0em}{1.5em}\\[1em] L_4& \dfrac{35}{8} \mu(1 - \mu^2)^{\frac{3}{2}} & \dfrac{35}{64} (1 - \mu^2)^2\rule{0em}{1.5em}\\[1em] \hline \end{array} \] \EXAMPLE{} Work the problems of Art.~108 and Art.~108 Exs.~1 and 2 by the aid of (3) Art.~114. % -----File: 223.png \mypara{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. \begin{flalign*} &\text{\indent Let}& \rho = Cr_1^kY_m(\mu_1,\phi_1). &&\phantom{\indent Let} \end{flalign*} Then using our ordinary notation we have \[ V = \int\limits_0^a dr_1 \int\limits_0^{2\pi} d\phi_1 \int\limits_{-1}^1 \dfrac{Cr^k_1 Y_m(\mu_1,\phi_1)r_1^2d\mu_1 }{\sqrt{r^2 - 2rr_1 \cos\gamma + r_1^2}}. \] But by (3) Art.~77 \begin{flalign*} \dfrac{1 }{\sqrt{ r^2 - 2rr_1 \cos \gamma + r_1^2}} = \dfrac{1}{r} &\bigg [P_0(\cos\gamma) + \dfrac{r_1}{r }P_1(\cos\gamma) \\ & + \dfrac{r_1^2}{r^2}P_2(\cos\gamma) + \cdots + \dfrac{r_1^m }{r^m }P_m(\cos\gamma) + \cdots \bigg]\\[1ex] \text{if} \quad &r > r_1. \end{flalign*} Consequently since \[ \int\limits_0^{2\pi} d\phi_1 \int\limits_{-1}^1 Y_m(\mu_1,\phi_1)Y_n(\mu_1,\phi_1)d\mu_1 = 0, \] $V$ reduces to the single term \begin{align*} V &= \frac{C}{r^{m+1}} \int\limits_0^a r_1^{m+k+2} dr_1 \int\limits_0^{2\pi} d\phi_1 \int\limits_{-1}^1Y_m(\mu_1, \phi_1)P_m(\cos\gamma)d\mu_1\\ &= \frac{C}{r^{m+1}} \int\limits_0^a r_1^{m+k+2} \left (\frac{4\pi }{ 2m+1} Y_m(\mu,\phi) \right )dr_1 \tag*{by Art.~113.} \end{align*} \[ \therefore \:V = \frac{4\pi C }{ 2m+1}. \frac{a^{m + k + 3} }{m + k + 3}. \frac{Y_m(\mu,\phi) }{r^{m+1}}. \] % -----File: 224.png \EXAMPLE{S} 1.\quad Solve by direct integration the problems worked in Arts.~98 and 99 and Examples 1, 2, 3, and 4 of Art.~99.\\ 2.\quad 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. \begin{alignat*}{3} &\textit{\indent Ans.}\qquad &\rho&= kr_1^4 \cos^2\theta_1 \sin^2\theta_1 \cos^2\phi_1 && \\[1ex] &&&= kr_1^4 \bigg [\frac{1}{15} P_0(\mu_1) + \frac{1}{21} P_2(\mu_1) &&+ \frac{1}{42} \cos 2\phi_1 P_2^2(\mu_1) \\ &&&&& - \frac{4}{35} P_4(\mu_1) + \frac{1}{105} \cos 2\phi_1 P_4^2(\mu_1) \bigg ].\\[-8ex] \end{alignat*} \begin{multline*} V = \frac{M}{a} \left [ \frac{a}{r} + \frac{a^3}{r^3} \left ( \frac{1}{9}P_2(\mu) + \frac{1}{18} \cos 2\phi P_2^2(\mu) \right ) \right .\\ \left . - \frac{a^5}{r^5} \left ( \frac{4}{33} P_4(\mu) - \frac{1}{99} \cos 2\phi P_4^2(\mu) \right ) \right ]. \end{multline*} 3.\quad Solve Example 2 by an extension of the method of Arts.~98 and 99.\\ 4.\quad 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.\\ \textit{Suggestion:} Let $V_1$ be the potential function due to the point, and $V_2$ that due to the induced charge, and let $b$ be the distance of the point from the centre of the sphere. Then \begin{align*} V_1 &= \frac{m}{\sqrt{\strxx b^2-2br \cos\theta + r^2} } &&\\ &=\frac{m}{b} \left [ P_0(\cos\theta) +\frac{ r}{b}P_1(\cos\theta) + \frac{r^2}{b^2}P_2(\cos\theta) + \cdots \right ] \quad \text{if} \quad rb.&&\\ V_2 &= A_0P_0(\cos\theta) + A_1 \frac{r}{a }P_1(\cos\theta) + A_2 \frac{r^2}{a^2 }P_2(\cos\theta) + \cdots \quad \text{if} \quad ra.&& \end{align*} When $r = a$\;\;$V_1 + V_2 = 0$. Hence \[ A_0 = -\dfrac{m}{b}, \quad A_1 = -\dfrac{ma}{b^2}, \quad A_2 = -\dfrac{ma^2}{b^3}, \cdots \] % -----File: 225.png and \begin{align*} V_2&=-\dfrac{m}{b}\left[P_0(\cos\theta)+ \dfrac{r}{b} P_1(\cos\theta) + \dfrac{ r^2}{b^2} P_2(\cos\theta)+ \cdots\right] \quad \text{if}\quad ra. \end{align*} Hence the effect of the induced charge is precisely the same at an external point as if the sphere were replaced by $\dfrac{ma}{b}$ units of negative electricity concentrated at the point $r=\dfrac{a^2}{b}$, $\theta=0$. v.\ Peirce, Newt.\ Pot.\ Func., \S~66. \mypara{116.} If the two points $P$ and $P'$ are taken on the line $OH$ whose direction cosines are $\lambda$, $\mu$, and $\nu$, and if $u$ and $u'$ are the values at $P$ and $P'$ of any continuous function of the space coördinates, then $\displaystyle\limit_{PP'\doteq 0} \left[\dfrac{u'-u}{PP'}\right]$ is called the \emph{partial derivative} of $u$ along the line $OH$ and will be represented by $D_hu$. Let $x$, $y$, $z$ be the coördinates of $P$ and $x + \Delta x$, $y + \Delta y$, $z + \Delta z$ the coördinates of $P'$; then \[ u'-u=D_xu.\Delta x+D_yu.\Delta y+D_zu.\Delta z+\epsilon \] where $\epsilon$ is an infinitesimal of higher order than the first if $\Delta x$, $\Delta y$, and $\Delta z$ are infinitesimal (v.\ Dif.\ Cal.\ Art.~198). \begin{flalign*} &\text{\indent Hence}& \dfrac{u'-u}{PP'} &=D_x u.\dfrac{\Delta x}{PP'} + D_y u.\dfrac{\Delta y}{PP'} + D_z u.\dfrac{\Delta z}{PP'} + \dfrac{\epsilon}{PP'}.&\phantom{\indent Henc}\\ &\text{But}&& \dfrac{\Delta x}{PP'} = \lambda,\quad \dfrac{\Delta y}{PP'}=\mu,\quad\text{and}\quad\dfrac{ \Delta z}{PP'}=\nu.&\\ &\text{Therefore}& &\qquad D_hu=\lambda D_xu+\mu D_yu+ \nu D_zu. & \tag{1} \end{flalign*} If $\nabla^2u=0$, $D_x^pD_y^qD_z^r u$ is a solution of Laplace's Equation. \begin{flalign*} &\text{\indent For}& \nabla^2(D_x^pD_y^qD_z^ru) &= D_x^pD_y^qD_z^r(\nabla^2u) = 0.&\phantom{\indent For} \end{flalign*} Hence if $\nabla^2u=0$ $D_hu$ is a solution of Laplace's Equation, and if $OH_1$, $OH_2$, $OH_3$, $\cdots$ are a set of lines through the origin $D_{h_1}D_{h_2}D_{h_3}\cdots u$ is a solution of Laplace's Equation. \mypara{117.} If $H_k$ is a rational integral homogeneous Algebraic function of $x$, $y$, and $z$ of the $k$th degree \begin{align*} D_x\left(\dfrac{H_k}{r^l}\right) &= D_r\left(\dfrac{H_k}{r^l}\right)D_xr+\dfrac{1}{r^l}D_x(H_k)\\ &=-\dfrac{lxH_k}{r^{l+2}} + \dfrac{H_{k-1}}{r^l} = -\dfrac{lxH_k}{r^{l+2}} + \dfrac{r^2H_{k-1}}{r^{l+2}}, \end{align*} and is of the form $\dfrac{H_{k+1}}{r^{l+2}}$. % -----File: 226.png The same thing can be proved of $D_y\Big(\dfrac{H_k}{r^l}\Big)$ and $D_z\Big(\dfrac{H_k}{r^l}\Big)$ and therefore holds good of $D_h\Big(\dfrac{H_k}{r^l}\Big)$. If $u$ is a homogeneous function of $x$, $y$, and $z$ of the degree $-m-1$ and $\nabla^2u=0$ then $\nabla^2(r^{2m+1}u)=0$. \begin{flalign*} &&\llap{$\displaystyle \nabla^2(r^{2m+1}u) ={}$}&(2m+1)(2m+2)r^{2m-1}u\\ && &+2(2m+1)r^{2m-1}(xD_xu + yD_yu + zD_zu) + r^{2m+1}\nabla^2u\hspace{1.5em}\\ && ={}&0,\\ &\text{since} &&\;xD_xu+yD_yu+zD_zu=-(m+1)u \end{flalign*} by Euler's Theorem (v.\ Dif.\ Cal.\ Art.~220). \mypara{118.} $\dfrac{M}{r}=\dfrac{M}{ \sqrt{x^2+y^2+z^2 }}$ is a solution of Laplace's Equation (v.\ Art.~75) and is of the form $\dfrac{H_0}{r}$. $D_{h_1}D_{h_2}D_{h_3} \cdots D_{h_m} \Big (\dfrac{M}{r} \Big )$ is then a solution of Laplace's Equation by Art. 116; it is of the form $\dfrac{H_m}{r^{2m+1}}$ by Art.~117 and is a homogeneous function of the degree $ -m-1$. Therefore $r^{2m+1}D_{h_1}D_{h_2}D_{h_3} \cdots D_{h_m} \Big (\dfrac{M}{r} \Big )$ is a solution of Laplace's Equation, and is a rational integral homogeneous Algebraic function of $x$, $y$, and $z$ of the $m$th degree, and is consequently a Solid Spherical Harmonic of the $m$th degree (v.\ Art.~110); and $r^{m+1}D_{h_1}D_{h_2}D_{h_3} \cdots D_{h_m} \Big (\dfrac{M}{r} \Big )$ is a Surface Spherical Harmonic of the $m$th degree. Moreover since the direction of each of the lines $OH_1$, $OH_2$, $\cdots$ $OH_m$ depends upon two angles which may be taken at pleasure, these angles and $M$ are $2m+1$ arbitrary constants and may be so chosen that $r^{m+1}D_{h_1}D_{h_2}D_{h_3} \cdots$ $D_{h_m} \Big (\dfrac{M}{r} \Big )$ may be any given Surface Spherical Harmonic. Consequently any given Surface Spherical Harmonic may be regarded as formed by differentiating $\dfrac{M}{r}$ successively along $m$ determinate lines $OH_1$, $OH_2$ $\cdots$ $OH_m$, and is given except for the undetermined factor $M$ when these lines are given. The lines $OH_1$, $OH_2$, $OH_3, \cdots OH_m$ are called the \textit{axes} of the Harmonic, and the points where they meet the surface of the unit sphere the \textit{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). % -----File: 227.png \mypara{119.} Any \emph{Surface Zonal Harmonic} $P_m(\mu)$ is equal to zero for $m$ real and distinct values of $\mu$ which lie between $-1$ and 1; and any \emph{Associated Function} $P_m^n(\mu)$ is equal to zero for $m-n$ real and distinct values of $\mu$, which lie between $-1$ and 1. \begin{flalign*} &\phantom{v.\ Art.~83 (1).}&P_m(\mu) &= \dfrac{1}{ 2^m m!}.\dfrac{ d^m (\mu^2 - 1)^m }{ d \mu^m}. & \text{v.\ Art.~83 (1).} \end{flalign*} $\dfrac{d^k (\mu^2 - 1)^m }{ d \mu^k}$ contains $(\mu^2 - 1)^{m-k}$ as a factor. v.\ Art.~89. From Rolle's Theorem, ``If $f(x)$ is continuous and single-valued and is equal to zero for the real values $a$ and $b$ of $x$, $\dfrac{df(x)}{dx}$ is equal to zero for at least one real value of $x$ between $a$ and $b$,'' (v.\ Dif.\ Cal.\ Art.~126) it follows that since $(\mu^2 - 1)^m = 0$ when $\mu = -1$ and when $\mu = 1$ $\dfrac{d(\mu^2 - 1)^m }{ d\mu} = 0$ for at least one value of $\mu$ between $-1$ and 1. $\dfrac{d(\mu^2 - 1)^m }{ d\mu}$ cannot be equal to zero for more than one value of $\mu$ between $-1$ and 1, for it contains $(\mu^2 - 1)^{m-1}$ as a factor and is a rational Algebraic polynomial of the $2m - 1$st degree. In like manner we can show that $\dfrac{d^2 (\mu^2 - 1)^m }{d\mu^2} = 0$ has $m-2$ roots equal to $-1$, $m-2$ roots equal to 1 and two real roots between $-1$ and 1 which separate the three distinct roots of $\dfrac{d(\mu^2 - 1)^m }{ d\mu} = 0$; and in general if $k < m+1$ that $\dfrac{d^k(\mu^2 - 1)^m }{ d\mu^k} = 0$ has $m-k$ roots equal to $-1$, $m-k$ roots equal to 1, and $k$ real roots separating the $k+1$ distinct roots of $\dfrac{d^{k-1}(\mu^2 - 1)^m}{d\mu^{k-1}} = 0$. Hence $P_m(\mu) = 0$ or $\dfrac{1}{2^m m!}.\dfrac{d^m (\mu^2 - 1)^m }{ d\mu^m} = 0$ has $m$ real and distinct roots between $-1$ and 1, and it has no more since it is of the $m$th degree. The same reasoning shows that $\dfrac{d^{m+n}(\mu^2 - 1)^m }{ d\mu^{m+n}} = 0$ has $m-n$ distinct real roots between $-1$ and 1, and therefore that $P_m^n(\mu)$ is equal to zero for $m-n$ distinct real values of $\mu$ between $-1$ and 1. Since $P_m^n(\mu)$ contains $\sin^n \theta$ as a factor it is also equal to zero when $\mu = -1$ and when $\mu = 1$. $\cos n\phi$ is equal to zero for $2n$ equidistant values of $\phi$, and $\sin n\phi$ is equal to zero for $2n$ values of $\phi$. Hence any \emph{Tesseral Harmonic} $\sin n\phi P_m^n(\mu)$ or $\cos n\phi P_m^n(\mu)$ is equal to zero for $2n$ equidistant values of $\phi$, for $\mu = 1$, for $\mu = -1$, and for $m-n$ real and different values of $\mu$ between $-1$ and 1. It follows that the value of any Surface Zonal Harmonic $P_m(\mu)$ 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 \textit{zones} into which the surface of the sphere is divided by % -----File: 228.png the $m$ circles of latitude corresponding to the $m$ roots of $P_m(\mu) = 0$, and will change sign whenever the point passes from one of these zones into an adjoining one; and that the value of any Tesseral Harmonic $\sin n\phi P^n_m(\mu)$ 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 \textit{tesserae} into which the surface of the sphere is divided by the $m-n$ circles of latitude corresponding to the roots of $P_m^n(\mu)=0$ and the $2n$ meridians corresponding to the roots of $\sin n\phi = 0$, and will change sign whenever the point passes from one of these tesserae into an adjoining one. \label{ch6end} % -----File: 229.png \mychap{CHAPTER VII.\footnotemark}{CYLINDRICAL HARMONICS (BESSEL'S FUNCTIONS).} \label{ch7start} \footnotetext{The student should re-read carefully Arts.~11, 17, and 18(\emph{d}) before beginning this chapter.} \mypara{120.} In Arts.~11 and 17 we obtained \[ z = AJ_0(x) + BK_0(x) \tag{1} \] as the general solution of \textit{Fourier's Equation} \begin{flalign*} &&&\frac{d^2z}{dx^2} + \frac{1}{x} \frac{dz}{dx}+ z = 0, \tag{2} \\ &\text{where}& J_0(x) ={}&1 - \frac{x^2}{2^2} + \frac{x^4}{2^2.4^2 } - \frac{x^6}{2^2.4^2.6^2} + \cdots \tag{3} &&\phantom{where} \end{flalign*} and is called a \textit{Cylindrical Harmonic} or \textit{Bessel's Function} of the zeroth order; and where \[ K_0(x) = J_0(x)\log x + \frac{x^2}{2^2} - \frac{x^4}{2^2.4^2}\left ( \frac{1}{1} + \frac{1}{2} \right ) + \frac{x^6}{2^2.4^2.6^2 } \left ( \frac{1}{1} + \frac{1}{2} + \frac{1}{3 }\right ) - \cdots \tag{4} \] and is called a \textit{Cylindrical Harmonic} or \textit{Bessel's Function} of the Second Kind, and of the zeroth order. In Art.~17 we found that\qquad\quad $z = J_n(x)$ \noindent is a particular solution of \textit{Bessel's Equation} \[ \frac{d^2z}{dx^2} + \frac{1}{x}\frac{ dz}{dx} +\left (1 - \frac{n^2}{x^2} \right ) z = 0, \tag{5} \] where if $n$ is unrestricted in value \begin{multline*} J_n(x) = \frac{x^n}{2^n \Gamma(n+1)} \left [1 - \frac{x^2}{2^2(n+1)} + \frac{x^4}{2^4.2!(n+1)(n+2)} \right . \\ \left . - \frac{x^6}{2^6.3!(n+1)(n+2)(n+3)} + \cdots \right ]\tag{6} \end{multline*} and is called a \textit{Cylindrical Harmonic} or \textit{Bessel's Function} of the $n$th order; and that unless $n$ is an integer \[ z = AJ_n(x)+ BJ_{-n}(x) \] is the general solution of Bessel's Equation. % -----File: 230.png If $n$ is an integer it can be shown that \[ J_n (x) = (-1)^n J_{-n} (x), \] (v.\ Forsyth's Diff.\ Eq.\ Art.~102), and then \[ z = AJ_n(x) + B\{K_n(x)\} \] is the general solution of Bessel's Equation and \begin{multline*} \{K_n(x)\}= J_n(x)\log x - \dfrac{1}{2}\left( \dfrac{x}{2}\right)^{-n} \sum\limits^{k=n-1}_{k=0} \dfrac{(n-k-1)!}{k!} \left(\dfrac{x}{2}\right)^{2k}\\ - \dfrac{1}{2}\left(\dfrac{x}{2}\right)^n \sum\limits^{k=\infty}_{k=0} \dfrac{(-1)^k}{(n+k)!k!}\left[1 + \dfrac{1}{2}+\dfrac{1}{3}+\cdots + \dfrac{1}{k}\right.\\ \left.+1+\dfrac{1}{2} + \dfrac{1}{3}+\cdots + \dfrac{1}{n+k}\right]\left( \dfrac{x}{2}\right)^{2k}\tag{7} \end{multline*} v.\ M.\ Bôcher, Ann.\ Math.\ Vol.\ VI, No.\ 4. \markright{CYLINDRICAL HARMONICS.} \mypara{121.} A useful expression for $J_n (x)$ 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 $z = x^nv$ is a solution of Bessel's Equation if $v$ satisfies the equation \begin{flalign*} &&\dfrac{d^2v}{dx^2} &+\dfrac{2n+1}{x} \dfrac{dv}{dx} + v = 0. \tag{1} \\ &\text{\indent Assume} &&v = \int\limits^b_a T\cos (xt)dt &\phantom{\indent Assume}\tag{2} \end{flalign*} where $x$ and $t$ are independent, $T$ is an unknown function of $t$, and $a$ and $b$ are at present undetermined. \begin{flalign*} &\text{\indent Then}& \dfrac{dv}{dx} &= - \int\limits^b_a tT\sin(xt)dt&\phantom{\indent Then}\\ &\text{and}& \dfrac{d^2v}{dx^2}& = - \int\limits^b_at^2 T \cos(xt)dt. \end{flalign*} Substituting in (1) after multiplying through by $x$, we have \[ \int\limits^b_a(1 - t^2) Tx \cos(xt)dt - \int\limits^b_a(2n + 1)tT\sin(xt)dt = 0. \tag{3} \] % -----File: 231.png By \emph{integration by parts} we find that \begin{multline*} \int\limits^b_a(1 - t^2)Tx \cos(xt)dt = \Big[(1 - t^2) T\sin (xt)\mathop{\Big]}\limits^b_a\\ -\int\limits^b_a\Big[(1 - t^2)\dfrac{dT}{dt} - 2tT\Big]\sin(xt)dt, \end{multline*} and (3) reduces to \[ \Big[\vphantom{\dfrac{X}{Y}}(1 - t^2)T\sin(xt)\mathop{\Big]}\limits^b_a - \int\limits^b_a\Big[(1 - t^2)\dfrac{dT}{dt} + (2n - 1)tT\Big]\sin(xt)dt = 0. \tag{4} \] If we determine $T$ so that \begin{flalign*} &&&(1-t^2) \dfrac{dT}{dt} +(2n-1)tT=0, \tag{5} \\ &\text{and $a$ and $b$ so that}&& \Big[(1 - t^2)T\sin(xt)\mathop{\Big]}\limits^b_a =0 &\phantom{and a and b so that} \tag{6} \end{flalign*} (4) will be satisfied and our problem will be solved. (5) gives \[ T=C(1-t^2)^{n- \frac{1}{2}}, \tag{7} \] and (6) will obviously be satisfied if $a = -1$ and $b = 1$. \begin{flalign*} &\text{\indent Hence}\qquad\qquad\quad & v&=C \int\limits^1_{-1}\dfrac{ (1-t^2)^n \cos(xt)dt }{ \sqrt{1-t^2}}&\text{is a solution of (1),}\\ &\text{and} & z&=Cx^n \int\limits^1_{-1}\dfrac{(1-t^2)^n \cos(xt)dt}{ \sqrt{1-t^2}} & (8) \end{flalign*} is a solution of Bessel's Equation. If we let $t = \cos\phi$ in (8) we get \[ z=Cx^n \int\limits^{\pi}_0 \sin^{2n} \phi \cos(x\cos\phi)d\phi. \] Expand $\cos(x\cos\phi)$ into a series involving powers of $x\cos\phi$, integrate term by term by the aid of the formulas \begin{flalign*} &\hspace{8.5em}&\int\limits^{\tfrac{\pi}{2}}_0 \sin^n x.dx &= \dfrac{\sqrt{\pi}}{2}.\dfrac{ \Gamma\Big(\dfrac{n+1}{2}\Big)}{\Gamma\Big(\dfrac{n}{2}+1\Big)} &\text{ [Int.\ Cal.\ (1) Art.~99],} \end{flalign*} % -----File: 232.png \[ \int\limits_0^{\tfrac{\pi}{2}} \sin^n x \cos^m x.dx = \frac{\Gamma \Big (\dfrac{m+1}{2}\Big ) \Gamma \Big ( \dfrac{n+1}{2} \Big ) }{2\Gamma \Big (\dfrac{m+n}{2} + 1 \Big )} \] (Int.\ Cal.\ Art.~99 Ex.~2), and compare with (6) Art.~120, and we get \[ J_n(x) = \frac{x^n }{2^n \sqrt{\pi}\ \Gamma\Big(n + \dfrac{1}{2}\Big)} \int\limits_0^{\pi}\sin^{2n} \phi \cos(x \cos\phi)d\phi. \tag{9} \] If $n$ is a positive integer (9) reduces to \[ J_n(x) = \frac{1}{\pi}. \frac{x^n }{1.3.5.\cdots(2n-1)} \int\limits_0^{\pi} \sin^{2n} \phi \cos(x \cos\phi)d\phi. \tag{10} \] Let $n = 0$ in (9) or (10) and we get \[ J_0(x) = \frac{1}{\pi} \int\limits_0^{\pi} \cos(x \cos\phi)d\phi. \tag{11} \] \EXAMPLE{S} 1.\quad Obtain Formula (11) directly from Fourier's Equation, (2) Art.~120.\\ 2.\quad Prove by \textit{integration by parts} that if $n > -\dfrac{1}{2}$ \[ \int\limits_0^{\pi} \sin^{2n} \phi \cos\phi \sin(x \cos\phi)d\phi = \frac{x}{ 2n+1} \int\limits_0^{\pi}\sin^{2n+2} \phi \cos(x \cos\phi)d\phi. \] 3.\quad Prove by \textit{integration by parts} that if $n > \dfrac{1}{2}$ \begin{align*} \int\limits_0^{\pi} \sin^{2n} \phi \cos\phi &\sin(x \cos\phi)d\phi\\ &= \frac{1}{x} \int\limits_0^{\pi} [2n \sin^{2n} \phi - (2n-1)\sin^{2n-2} \phi] \cos(x \cos\phi)d\phi. \end{align*} \mypara{122.} We can now readily obtain a number of useful formulas. \noindent Differentiate (11) Art.~121 with respect to $x$ and we get \begin{flalign*} \hspace{6em} \dfrac{dJ_0(x) }{dx}& = -\dfrac{1}{\pi} \int\limits_0^{\pi} \cos\phi \sin(x \cos\phi)d\phi\\ &= -\dfrac{x}{\pi} \int\limits_0^{\pi} \sin^2 \phi \cos(x \cos\phi)d\phi \tag*{by Ex.~2 Art.~121.}\\[-6ex] \end{flalign*} % -----File: 233.png \begin{flalign*} &\text{\indent Hence by (10) Art.~121} \qquad \dfrac{ dJ_0(x)}{dx} = -J_1(x).\tag{1}&& \end{flalign*} In like manner by the aid of Exs.~3 and 2, Art.~121, we can obtain the relations \[ \frac{d[x^n J_n(x)] }{ dx} = x^n J_{n-1}(x)\tag{2} \] if $n > \dfrac{1}{2}$, \[ \frac{d[x^{-n} J_n (x)] }{ dx }= -x^{-n} J_{n+1} (x)\tag{3} \] if $n > -\dfrac{1}{2}$.\\ (2) can be written \[ \int\limits_0^x x^nJ_{n-1}(x)dx=x^nJ_n(x)\tag{4}\\[-4ex] \] if $n > \dfrac{1}{2}$.\\ (2) and (3) can be written \begin{flalign*} &&x^n \frac{dJ_n(x)}{dx} +{}&nx^{n-1}J_n(x) = x^nJ_{n-1}(x) \\[2ex] &\text{and} &x^{-n} \frac{dJ_n(x)}{dx} - n&x^{-n-1}J_n(x) = -x^{-n}J_{n+1}(x), && \\[2ex] &\text{or} & \frac{dJ_n(x)}{dx} &= J_{n-1}(x) - \frac{n}{x}J_n(x)\tag{5}&& \\[2ex] &\text{and} & \frac{dJ_n(x)}{dx} ={}&- J_{n+1}(x) + \frac{n}{x}J_n(x);\tag{6}&& \\[2ex] &\text{whence}& 2\frac{dJ_n(x)}{dx} &= J_{n-1}(x) - J_{n+1}(x)\tag{7}&& \\[2ex] &\text{and}& \frac{2n}{x}J_n(x) &= J_{n-1}(x) + J_{n+1}(x).&&\tag{8} \end{flalign*} The repeated use of formula (8) will enable us to get from $J_0(x)$ and $J_1(x)$ any of Bessel's Functions whose order is a positive integer. For example, we have \begin{gather*} J_2(x) = \frac{2}{x}J_1(x) - J_0(x)\\ J_3(x) = \Big ( \frac{8}{x^2}-1 \Big )J_1(x) - \frac{4}{x}J_0(x). \end{gather*} % -----File: 234.png From a table giving the values of $J_0(x)$ and $J_1(x)$, then, tables for the functions 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 $J_n(x)$ can be expressed in terms of $J_n(x)$ and $J_{n+1}(x)$. For example \[ \dfrac{d^2J_n(x)}{dx^2} = \bigg [ \dfrac{n(n-1)}{x^2} -1\bigg ] J_n(x) + \dfrac{1}{x } J_{n+1}(x). \] If we write $J_0(x)$ 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 \textit{integration by parts}, we get \begin{flalign*} &&x &\dfrac{dJ_0(x)}{dx} + \int\limits_0^x x J_0(x)dx = 0; \\ &\text{whence by (1)}&& \int\limits_0^x x J_0(x)dx = xJ_1(x). \tag{9} &&\phantom{whence by} \end{flalign*} If we write $J_0(x)$ for $z$ in Fourier's Equation, then multiply through by $x^2 \dfrac{dJ_0(x)}{dx}\,dx$ and integrate from zero to $x$, simplifying by \textit{integration by parts} we get \begin{flalign*} &&\dfrac{x^2}{2} &\bigg [ \left ( \dfrac{dJ_0(x)}{dx} \right )^2+(J_0(x))^2 \bigg ] - \int\limits_0^x x(J_0(x))^2 dx = 0;&&\qquad \\ &\text{whence by (1)}& &\int\limits_0^x x(J_0(x))^2 dx = \dfrac{x^2}{2} [(J_0(x))^2 + (J_1(x))^2]. \tag{10}&& \end{flalign*} In like manner we can get from Bessel's Equation [(5) Art.~120] the formula \[ \int\limits_0^x x(J_n(x))^2 dx= \dfrac{1}{2} \bigg [x^2 \left ( \dfrac{dJ_n(x)}{dx} \right )^2 + (x^2 - n^2)(J_n(x))^2 \bigg ] \tag{11} \] which (6) enables us to reduce to the form \[ \int\limits_0^x x(J_n(x))^2 dx= \dfrac{x^2}{2} [(J_n(x))^2 + (J_{n+1}(x))^2] - nxJ_n(x)J_{n+1}(x).\tag{12} \] Formulas (9), (10), (11), and (12) will prove useful when we attempt to develop in terms of \textit{Cylindrical Harmonics}. % -----File: 235.png Values of $J_n(x)$ for larger values of $x$ than those given in Table VI., Appendix, may be computed very easily from the formula \begin{flalign*} J_n(x)& = \sqrt{\frac{ 2}{\pi x} } \left [1-\frac{(1^2-4n^2)(3^2-4n^2)}{2!(8x)^2} \right .\\ & \quad \left . + \frac{(1^2-4n^2)(3^2-4n^2)(5^2-4n^2)(7^2-4n^2)}{4!(8x)^4} - \cdots \right ]\cos \left (x-\frac{\pi}{4}-n\frac{\pi}{2} \right ) \\ & + \sqrt{ \frac{2}{\pi x}} \left [\frac{1^2-4n^2}{1!8x} \right .\\ & \quad \left . - \frac{ (1^2-4n^2)(3^2-4n^2)(5^2-4n^2)}{3!(8x)^3} + \cdots \right ]\sin \left (x-\frac{\pi}{4}-n\frac{\pi}{2} \right ).\tag{13} \end{flalign*} 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 $J_n(x)$ by less than the last term included, and consequently the formula can safely be used for numerical computation. \EXAMPLE{S} 1.\quad Confirm (1), (2), and (3), Art.~122, by obtaining them from (3) and (6), Art.~120.\\ 2.\quad Confirm (1), Art.~122, by showing that Fourier's Equation will differentiate into the special form assumed by Bessel's Equation when $n=1$.\\ 3.\quad Show that (9), Art.~122, is a special case of (4), Art.~122.\\ 4.\quad Show that the limit approached by $J_n(x)$ as $ n$ increases indefinitely is zero, and by the aid of this fact and of (8), Art.~122, prove that \[ J_{n-1}(x) = \frac{2}{x} [nJ_n(x) - (n+2)J_{n+2}(x) + (n+4)J_{n+4}(x) + \cdots ]. \] 5.\quad Prove that \[ \frac{dJ_n(x)}{dx} = \frac{2}{x} [ \tfrac{1}{2}nJ_n(x) - (n+2)J_{n+2}(x) + (n+4)J_{n+4}(x) - \cdots ]. \] 6.\quad Show that the substitution of $\left (1-\dfrac{y^2}{n^2}\right )^{\frac{1}{2}}$ for $x$ in Legendre's Equation will reduce it to the form \[ \left (1-\frac{y^2}{n^2}\right ) \frac{d^2z}{dy^2} +\left ( \frac{1}{y} -\frac{2y}{n^2} \right ) \frac{dz}{dy }+ \left (1+ \frac{1}{n} \right )z=0, \] and that the limiting form approached by this equation as $n$ is indefinitely increased is Fourier's Equation, and hence that $J_0(x)$ can be regarded as some constant factor multiplied by the limiting value approached by $P_n \bigg (1-\dfrac{x^2}{n^2} \bigg )^{\frac{1}{2}}$ as $n$ is indefinitely increased. % -----File: 236.png \mypara{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 \[ f(r) = A_1 J_0(\mu_1 r) + A_2 J_0(\mu_2 r) + A_3 J_0(\mu_3 r) + \cdots \] where $\mu_1$, $\mu_2$, $\mu_3$, \&c., are the roots of the transcendental equation $J_0(\mu a) = 0$; and in Art.~11, Ex.\ the development of unity in a series of precisely the same form was needed.\\ (\emph{a})\quad 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 (\smallrom{XIV} Art.~1). \[ D_r^2 u + \frac{1}{r }D_r u + \frac{1}{r^2}D_{\phi}^2 u + D_z^2 u = 0 \tag{1} \] subject to the conditions \begin{alignat*}{3} &u = 0 &&\text{when}\quad &&z = 0\\ &u = 0 &&\quad\text{``} &&r = a\\ &u =f(r) &&\quad\text{``} &&z = b, \end{alignat*} and from the symmetry of the problem we know that $D_{\phi}^2u = 0$. Assuming as usual $u = R.Z$ we break (1) up into the equations \begin{flalign*} &&& \frac{d^2Z}{dz^2} - \mu^2 Z = 0\\[1ex] && \frac{d^2R}{dr^2} &+ \frac{1}{r} \frac{dR}{dr} + \mu^2 R =0, \\[1ex] &\text{whence}& u &= \sinh (\mu z)J_0(\mu r)\tag{2}&&\phantom{whence}\\[1ex] &\text{and}& u &= \cosh (\mu z)J_0(\mu r)\tag{3}&&\\[-1ex] \intertext{are particular solutions of (1).} \end{flalign*} \begin{flalign*} &\text{\indent If $\mu_k$ is a root of} & J_0(\mu a) = 0 &&\phantom{\indent If mu is a root of}\tag{4} \end{flalign*} \[ u = \sinh(\mu_k z)J_0(\mu_k r) \] satisfies (1) and two of the three equations of condition. \begin{flalign*} & \text{\indent If then}& f(r) = A_1 J_0(\mu_1 r) + A_2 J_0(\mu_2 r) + A_3 J_0(\mu_3 r) + \cdots &&\tag{5} \end{flalign*} $\mu_1$, $\mu_2$, $\mu_3$, \&c., being roots of (4), \[ u = A_1 \frac{ \sinh(\mu_1z)}{\sinh(\mu_1b)} J_0(\mu_1 r) + A_2 \frac{\sinh(\mu_2z)}{\sinh(\mu_2b)} J_0(\mu_2 r) + A_3\frac{ \sinh(\mu_3z)}{\sinh(\mu_3b) }J_0(\mu_3 r) + \cdots \tag{6} \] satisfies (1) and all of the equations of condition, and is the required solution.\\ % -----File: 237.png (\emph{b})\quad If instead of keeping the convex surface of the cylinder at the temperature zero we surround it by a jacket impervious to heat, the equation of condition $u = 0$ when $r = a$ will be replaced by $D_ru = 0$ when $r = a$, or if \begin{flalign*} &\qquad & u = \sinh(\mu z&)J_0(\mu r), &&\\ &\text{by}& \frac{dJ_0(\mu r)}{dr} &= 0&&\tag*{when $r = a$,}\\ &\text{that is by}& \mu J'_0(\mu a) &= 0\footnotemark&&\tag*{or (v.\ (1) Art.~122)}\\[1ex] &\text{by} & J_1(\mu a)&=0. && \tag{7} \end{flalign*} \footnotetext{We shall find it convenient to use the familiar notation of $f'(x) = \dfrac{df(x)}{dx}$ (v.\ Dif. Cal., p.\ 119).} If now in (5) and (6) $\mu_1$, $\mu_2$, $\mu_3$, \&c., are roots of (7), (6) will be the solution of our new problem.\\ (\emph{c})\quad If instead of keeping the convex surface of the cylinder at the temperature zero we allow it to cool in air at the temperature zero, the condition $u = 0$ when $r = a$ will be replaced by $D_ru + hu = 0$ when $r = a$, or if \begin{flalign*} &&u = \sinh (\mu z)J_0(&\mu r)&&\\[1ex] &\text{by}& \mu J'_0(\mu r) + hJ_0(\mu r) &= 0 && \tag*{when \quad $r = a$}\\[1ex] &\text{that is by}& \mu a J'_0(\mu a) + ahJ_0(\mu a) &=0 && \tag*{or (v.\ (1) Art.~122)}\\[1ex] &\text{by}& \mu a J_1(\mu a) - ahJ_0(\mu a) &= 0. && \tag{8} \end{flalign*} If now in (5) and (6) $\mu_1$, $\mu_2$, $\mu_3$, \&c., are roots of (8), (6) will be the solution of our present problem. \mypara{124.} It can be shown that\\[-7.1ex] \begin{flalign*} && J_0(x) &= 0 &&\tag{1}\\ && J_1(x) &= 0 && \tag{2}\\ &\text{and} & xJ'_0(x) + \lambda &J_0(x) = 0&& \tag{3}\\[-4ex] \end{flalign*} have each an infinite number of real positive roots (v.\ Riemann, Par.\ Dif.\ Gl., \S~97). The earlier roots of these equations can be computed without serious difficulty from the table for the values of $J_0(x)$ (Table VI., Appendix). The first twelve roots of $J_0(x) = 0$ and $J_1(x) =0$ are given in Table IV., Appendix, a table due to Stokes. Large roots of $J_0(x) = 0$ and of $J_1(x) = 0$ may be very easily computed from the formulas \begin{gather*} \frac{x_{\phantom{(}0}^{(s)}}{\pi} = s-.25 + \frac{.050661}{4s-1} - \frac{.053041}{(4s-1)^3 } + \frac{.262051}{(4s-1)^5 }- \cdots \tag{4}\\ \frac{x_{\phantom{(}1}^{(s)}}{\pi} = s+.25 - \frac{.151982}{4s+1} + \frac{.015399}{(4s+1)^3} - \frac{.245270}{(4s+1)^5} + \cdots \tag{5} \end{gather*} given by Stokes in Camb.\ Phil.\ Trans., Vol.\ IX., $x_{\phantom{(}0}^{(s)}$ representing the $s$th root of $J_0(x) = 0$, and $x_{\phantom{(}1}^{(s)}$ the $s$th root of $J_1(x) = 0$. % -----File: 238.png \mypara{125.} We have seen in Art.~123 that $U=\sinh(\mu_k z)J_0(\mu_k r)$ and $V=\sinh(\mu_l z)J_0(\mu_l r)$ are solutions of $\nabla^2U=0$ and $\nabla^2V=0$ if we express Laplace's Equation in terms of Cylindrical Coördinates (v.\ (1) Art.~123). Hence, if $\int dS$ represents the surface integral over any closed surface, we have \[ \int(UD_nV-VD_nU)dS=0 \] by Green's Theorem (v.\ Art.~92). If we take the cylinder of Art.~123 as our surface, and perform the integrations and simplify the resulting equation, we find \begin{multline*} \int\limits^a_0rJ_0(\mu_k r)J_0(\mu_l r)dr=\dfrac{-1}{{\mu_k}^2-{\mu_l}^2} [\mu_kaJ_0(\mu_l a)J'_0(\mu_k a) -\mu_laJ_0(\mu_k a)J'_0(\mu_l a)]\\ =\dfrac{-1}{{\mu_l}^2-{\mu_k}^2} [\mu_kaJ_0(\mu_l a)J_1(\mu_k a) -\mu_laJ_0(\mu_k a)J_1(\mu_l a)]. \tag{1} \end{multline*} Hence if $\mu_k$ and $\mu_l$ are different roots of \begin{flalign*} & & J_0(\mu a)&=0, &&\\ &\text{or of} & J_1(\mu a)&=0, &&\phantom{or of}\\ &\text{or of}& \mu aJ_1(\mu a)-\lambda &J_0(\mu a)=0, &&\\ &\text{then} & \int\limits^a_0 rJ_0(\mu_k r)J_0(&\mu_l r)dr=0. &&\tag{2} \end{flalign*} \EXAMPLE{} Obtain (1) Art.~125 directly from Fourier's Equation \[ \dfrac{d^2 J_0(\mu r)}{dr^2} + \dfrac{1}{r}\dfrac{ dJ_0(\mu r)}{dr} + \mu^2 J_0 (\mu r)=0. \] \mypara{126.} We are now able to obtain the developments called for in Art.~123. \begin{flalign*} &\text{\indent Let}& f(r)&=A_1J_0(\mu_1 r)+A_2J_0(\mu_2 r)+A_3J_0(\mu_3 r)+\cdots&\phantom{\indent Let} \tag{1} \end{flalign*} $\mu_1$, $\mu_2$, $\mu_3$, \&c., being roots of $J_0(\mu a)=0$, or of $J_1(\mu a)=0$, or of \[ \mu aJ_1(\mu a)-\lambda J_0(\mu a)=0. \] To determine any coefficient $A_k$ multiply (1) by $rJ_0(\mu_k r)dr$ and integrate from zero to $a$. The first member will become \[ \int\limits^a_0 rf(r)J_0(\mu_k r)dr. \] % -----File: 239.png Every term of the second member will vanish by (2) Art.~125 except the term \begin{gather*} A_k \int\limits_0^a r(J_0(\mu_k r))^2dr.\\ \int\limits_0^a r(J_0(\mu_k r))^2dr = \frac{1}{\mu_k^2} \int\limits_0^{\mu_ka} x(J_0(x))^2dx = \frac{a^2}{2}[(J_0(\mu_k a))^2+(J_1(\mu_k a))^2] \end{gather*} by (10) Art.~122. \begin{flalign*} &\text{\indent Hence}& A_k = \frac{2}{a^2[(J_0(\mu_k a))^2 + (J_1(\mu_k a))^2] } \int\limits_0^a rf(r)J_0(\mu_k r)dr. && \tag{2} \end{flalign*} The development (1) holds good from $r = 0$ to $r = a$ (v.\ Arts.~24, 25, and 88). If $\mu_1$, $\mu_2$, $\mu_3$, \&c., are roots of $J_0(\mu a)=0$, (2) reduces to \[ A_k = \frac{2}{a^2(J_1(\mu_k a))^2} \int\limits_0^a rf(r)J_0(\mu_k r)dr. \tag{3} \] If $\mu_1$, $\mu_2$, $\mu_3$, \&c., are roots of $J_1(\mu a)=0$, (2) reduces to \[ A_k = \frac{2}{a^2(J_0(\mu_k a))^2} \int\limits_0^a rf(r)J_0(\mu_k r)dr. \tag{4} \] If $\mu_1$, $\mu_2$, $\mu_3$, \&c., are roots of $\mu aJ_1(\mu a)-\lambda J_0(\mu a)=0$, (2) reduces to\\ \[ A_k = \frac{2\mu_k^2}{(\lambda^2+\mu_k^2a^2)(J_0(\mu_k a))^2} \int\limits_0^a rf(r)J_0(\mu_k r)dr. \tag{5} \] For the important case where $ f(r) = 1$ \[ \int\limits_0^a r f(r) J_0(\mu_k r) dr = \int\limits_0^a r J_0(\mu_k r) dr = \frac{1}{\mu_k^2} \int\limits_0^{\mu_k a}xJ_0(x) dx = \frac{a}{\mu_k} J_1(\mu_k a) \tag{6} \] by (9) Art.~122, and (3) reduces to\\ \begin{flalign*} && A_k&=\frac{2}{\mu_k aJ_1(\mu_k a)},\hspace{1.6em} && \tag{7} \end{flalign*} (4) reduces to $A_k = 0$ except for $k = 1$ when $\mu_k = 0$ and we have $A_1 = 1$, \begin{flalign*} &\indent\text{(5) reduces to} &A_k &= \frac{2\lambda }{(\lambda^2 + \mu_k^2 a^2)J_0(\mu_k a)}. && \phantom{(5) reduces to} \tag{8} \end{flalign*} % -----File: 240.png \EXAMPLE{S} 1.\quad Show that in (12) Art.~11 any coefficient $A_k$ has the value given in (3) Art.~126; and in the answer to Art.~11, Ex.~the value given in (7) Art.~126.\\ 2.\quad 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 \textit{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 $J_0(x) = 0$; that the possible nodes for such vibrations are concentric circles whose radii are proportional to the roots of $J_0(x) = 0$.\\ 3.\quad 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 \textit{stationary temperature} is set up. Find the temperature 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.\\ \phantom{kindly go to the right hand margin}\hfill \emph{Ans.}, 29°.6; 47°.6; 71°.2; 25°.8.\\ 4.\quad An iron cylinder one meter long and twenty centimeters in diameter has its convex surface covered with a so-called non-conducting cement one centimeter 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 temperatures 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.\\ \phantom{right margin}\hfill \emph{Ans.}, 15°.4; 40°.85; 72°.8; 15°.3; 40°.7; 72°.5; 0°.0; 0°.0; 1°.3. \mypara{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 \begin{alignat*}{6} &u = 0 &&\text{when}\quad &&z = 0, \qquad\qquad && u = 0\quad &&\text{when}\quad &&r = a,\\ &u = f(r) &&\quad\text{`` } &&z = b, && u = 0 &&\quad\text{`` } &&r = c. \end{alignat*} Then (v.\ Art.~123) \[ u = \sinh(\mu_k z) \left [J_0(\mu_k r) - \frac{J_0(\mu_kc)}{K_0(\mu_kc)}K_0(\mu_k r) \right ] \] % -----File: 241.png where $\mu_k$ is a root of the equation \[ J_0(\mu a)-\dfrac{J_0(\mu c)}{K_0(\mu c)} K_0(\mu a)=0 \tag{1} \] will satisfy Laplace's Equation [(1) Art.~123] and all of the equations of condition except the second. \begin{flalign*} &\text{\indent Hence} & u&= \sum\limits^{k=\infty}_{k=1} A_k \dfrac{\sinh(\mu_k z)}{\sinh(\mu_k b)}\left[J_0(\mu_k r)-\dfrac{J_0(\mu_kc)}{K_0(\mu_kc)} K_0(\mu_k r)\right]&\phantom{Hence}\tag{2} \intertext{is the required solution if} &&&f(r) = \sum\limits^{k=\infty}_{k=1} A_k \bigg[J_0(\mu_k r)-\dfrac{J_0(\mu_kc)}{K_0(\mu_kc)} K_0(\mu_k r)\bigg]. \tag{3} \end{flalign*} The development (3) is easily obtained. Call the parenthesis for the sake of brevity $B_0(\mu_kr)$. Then by the method of Art.~125 we get if we integrate over our cylindrical shell \[ \int\limits^c_a rB_0(\mu_kr)B_0(\mu_lr)dr=0 \tag{4} \] if $\mu_k$ and $\mu_l$ are roots of (1); and by an easy extension of (10) Art.~122 \[ \int\limits^c_ar[B_0(\mu_kr]^2dr = \tfrac{1}{2}\lbrace c^2[B'_0(\mu_kc)]^2-a^2[B'_0(\mu_ka)]^2\rbrace. \tag{5} \] Determining the coefficients in (3) as in Art.~124 and simplifying by the aid of (4) we have \noindent \[ A_k = \dfrac{\displaystyle 2\int\limits^c_a rf(r) \Big[J_0(\mu_k r) - \dfrac{J_0(\mu_kc)}{K_0(\mu_kc)}K_0(\mu_k r)\Big]dr} {c^2\Big[J'_0(\mu_k c) - \dfrac{J_0(\mu_kc)}{K_0(\mu_kc)} K'_0 (\mu_k c)\Big]^2-a^2\Big[{J_0}'(\mu_k a) - \dfrac{J_0(\mu_kc)}{K_0(\mu_kc)} K'_0(\mu_k a)\Big]^2}.\tag{6} \] \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 \[ y= \sum\limits^{k=\infty}_{k=1} A_k \cos(\mu_kct)\left[J_0(\mu_k r)-\dfrac{J_0(\mu_kb)}{K_0(\mu_kb)} K_0(\mu_k r)\right] \] where $A_k$ is obtained from (6) Art.~127 by replacing $c$ by $b$. % -----File: 242.png \mypara{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 $\mu$ is replaced by $\mu i$ we can readily obtain \begin{flalign*} & &z &= \sin (\mu z) J_0(\mu ri)&\\ &\text{and} & z& = \cos (\mu z) J_0(\mu ri)&\phantom{and} \end{flalign*} as particular solutions of Laplace's Equation [(1) Art.~123]; and \[ J_0(xi)=1 + \dfrac{x^2}{2^2} + \dfrac{x^4}{2^2.4^2} + \dfrac{x^6}{2^2.4^2.6^2} +\cdots \tag{1} \] and is real. \begin{flalign*} && &f(z)= \sum \limits^{k=\infty}_{k=1} A_k \sin \dfrac{k\pi z}{b} &&\\ &\text{where} &&A_k = \dfrac{2}{b}\int \limits^b_a f(z) \sin \dfrac{k\pi z}{b} dz &&\tag{2} \intertext{by Art.~31 (7) and (8).} &\text{\indent Hence}&u&=\sum \limits^{k=\infty}_{k=1} A_k \sin \dfrac{k\pi z}{b} \dfrac{ J_0\Big(\dfrac{k\pi ri}{b}\Big)} {J_0\Big(\dfrac{k\pi ai}{b}\Big)} && \tag{3} &\intertext{is our required solution.} \end{flalign*} \EXAMPLE{S} 1.\quad 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 % recast to fit page width \begin{multline*} u=\sum\limits^{k=\infty}_{k=1} A_k \sin \dfrac{k\pi z}{b} \left[ \dfrac{ J_0\Big(\dfrac{k\pi ri}{b}\Big)}{ J_0\Big(\dfrac{k\pi ci}{b}\Big)} -\dfrac{ \overline{K_0}\Big(\dfrac{k\pi ri}{b}\Big)}{ \overline{K_0}\Big(\dfrac{k\pi ci}{b}\Big)} \right]\\ \div \left [ \dfrac{ J_0\Big(\dfrac{k\pi ai}{b}\Big)}{ J_0\Big(\dfrac{k\pi ci}{b}\Big)} -\dfrac{ \overline{K_0}\Big(\dfrac{k\pi ai}{b}\Big)}{ \overline{K_0}\Big(\dfrac{k\pi ci}{b}\Big)} \right] \end{multline*} where $A_k$ has the value given in (2) Art.~128, and \begin{align*} \overline{K_0}(xi) &= K_0(xi) - J_0(xi) \log i\\ &= J_0(xi) \log x - \dfrac{x^2}{2^2} - \dfrac{x^4}{2^2.4^2}(\tfrac{1}{1}+\tfrac{1}{2}) - \dfrac{x^6}{2^2.4^2.6^2}(\tfrac{1}{1}+\tfrac{1}{2}+\tfrac{1}{3}) -\cdots \end{align*} [v.\ (4) Art.~120], and is real.\\ 2.\quad 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 % -----File: 243.png 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. \hfill \emph{Ans.}, 49°.6; 20°.2. \mypara{129.} If in the problem of Art.~123 the temperatures of the points of the upper base of the cylinder are unsymmetrical so that $u=f(r,\theta)$ when $z = b$, we have to get particular solutions of Laplace's Equation [(1) Art.~123] for the case where $D_\phi^2u$ is not equal to zero. We readily find that \begin{flalign*} && u&=\sinh\;(\mu z)[A \cos n\phi +B \sin n\phi]J_n(\mu r)&\\[1ex] &\text{and}&u&=\cosh\;(\mu z)[A \cos n\phi +B \sin n\phi]J_n(\mu r)&\phantom{and} \end{flalign*} are such solutions, and that \[ u= \sum\limits^{n=\infty}_{n=0} \sum\limits^{k=\infty}_{k=1} \dfrac{\sinh \mu_{k} z}{\sinh \mu_k b} [A_{n,k} \cos n\phi +B_{n,k} \sin n\phi]J_n(\mu_k r) \tag{1} \] is the solution of the given problem if \[ f(r,\phi) = \sum\limits^{n=\infty}_{n=0} \sum\limits^{k=\infty}_{k=1} (A_{n,k} \cos n\phi +B_{n,k} \sin n\phi)J_n(\mu_k r) \tag{2} \] where $\mu_k$ is a root of the equation \[ \dfrac{J_n(\mu a)}{\mu^n a^n} = 0. \tag{3} \] \EXAMPLE{S} 1.\quad Show that % recast to fit page width \begin{align*} \int\limits^a_0 rJ_n(\mu_k r)J_n(\mu_l r)&dr\\[-2ex] &=\dfrac{a}{{\mu_k}^2-{\mu_l}^2} [\mu_l J_n(\mu_k a)J'_n(\mu_l a)-\mu_k J_n(\mu_l a)J'_n(\mu_k a)]\\ &=\dfrac{a}{{\mu_k}^2-{\mu_l}^2} [\mu_k J_n(\mu_l a)J_{n+1}(\mu_k a)-\mu_l J_n(\mu_k a)J_{n+1}(\mu_l a)]. \end{align*} 2.\quad Show that % recast to fit page width \begin{align*} \int\limits^a_0 r [J_n(\mu_k r)]^2&dr\\[-2ex] &=\dfrac{1}{2} \Big[a^2(J'_n(\mu_k a))^2+\Big(a^2 - \dfrac{n^2}{{\mu_k}^2}\Big) (J_n(\mu_k a))^2\Big]\\ &=\dfrac{a^2}{2}[(J_n (\mu_k a))^2 + (J_{n+1} (\mu_k a))^2]- \dfrac{na}{\mu_k} J_n (\mu_k a)J_{n+1} (\mu_k a). \end{align*} % -----File: 244.png 3.\quad Show that in Art.~129 \begin{align*} A_{0,k}&=\dfrac{1}{\pi} \dfrac{\displaystyle \int\limits^{2\pi}_0d\phi\int\limits^a_0 rf(r,\phi)J_0(\mu_kr)dr} {a^2[J_1(\mu_ka)]^2},\\[0.5em] B_{0,k}&=0,\\ A_{n,k}&=\dfrac{2}{\pi} \dfrac{\displaystyle\int\limits^{2\pi}_0 d\phi \int\limits^a_0 rf(r,\phi)\cos n\phi J_n(\mu_k r)dr} {a^2[J_{n+1}(\mu_ka)]^2},\\ B_{n,k}&=\dfrac{2}{\pi} \dfrac{\displaystyle\int\limits^{2\pi}_0 d\phi \int\limits^a_0 rf(r,\phi)\sin n\phi J_n(\mu_kr)dr} {a^2[J_{n+1}(\mu_ka)]^2}. \end{align*} 4.\quad Obtain the coefficients for the case where the convex surface of the cylinder is impervious to heat.\\ 5.\quad Obtain the coefficients for the case where the convex surface of the cylinder is exposed to air at the temperature zero.\\ 6.\quad Show that if in a drumhead problem of Art.~11 the initial distortion is unsymmetrical, so that we have to solve the equation \smallrom{XI} Art.~1 subject to the conditions $ z = f(r,\phi)$ when $t = 0$, $D_t z = 0$ when $t = 0$, $z = 0$ when $r = a$, the solution is \[ z=\sum\limits^{n=\infty}_{n=0}\sum\limits^{k=\infty}_{k=1}\cos(\mu_k ct)(A_{n,k}\cos n\phi + B_{n,k}\sin n\phi) J_n(\mu_k r) \] where $A_{0,k}$, $B_{0,k}$, $A_{n,k}$, and $B_{n,k}$ have the values given in Ex.~3.\\ 7.\quad 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 circles and of radii whose outer extremities are equidistant. v.\ Rayleigh's Sound, Vol.\ I., Arts.~(202-207).\\ 8.\quad Solve the problem of Art.~127 and of Art.~127.\ Ex.~for the unsymmetrical case. \quad \emph{Suggestion:} $AJ_n(x) + BK_n(x)$ is a solution of Bessel's Equation.\\ 9.\quad Solve the problem of Art.~128 and of Art.~128.\ Ex.~1.\ for the case where $u = f(z,\phi)$ when $r = a$. \quad \emph{Suggestion:} $u = \sin \mu z (A \cos n\phi + B \sin n\phi) J_n(\mu ri)$ is a solution of Laplace's Equation, and $f(z,\phi)$ can be developed into a double Fourier's Series [v.\ (15) Art.~71].\\ % -----File: 245.png 10.\quad 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.\quad \textit{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 $D_t^2y=c^2(xD_x^2y+D_xy)$, subject to the conditions $D_ty=0$ 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. \begin{flalign*} &\text{Show that} &&y=\sum_{k=1}^{k=\infty}A_k \cos\mu_k ct\,B_0(\mu_k^2x),&&\\ &\text{where} & B_0(x)&=1-\frac{x}{1^2}+\frac{x^2}{1^2.2^2}-\frac{x^3}{1^2.2^2.3^2}+\cdots&&\phantom{where}\\[1ex] &&&=J_0(2\sqrt{x})&& \end{flalign*} and $\mu_k$ is a root of the equation \[B_0(\mu^2 a)=J_0(2\mu \sqrt{a})=0,\] \begin{flalign*} &\text{and}& A_k = \frac{\displaystyle\int\limits_0^af(x)B_0(\mu_k^2 x)dx}{\mu^2a^2[B'_0(\mu^2_ka)]^2}=\frac{\displaystyle\int\limits_0^af(x)J_0(2\mu_k\sqrt{x})dx}{a[J_1(2\mu_k\sqrt{a})]^2}.&&\\. \end{flalign*} 12.\quad 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 $\cos n\phi$ and by taking $n = \dfrac{m}{2}$. The required solution is \[ z=\sum_{m=1}^{m=\infty}\sum_{k=1}^{k=\infty}B_{m,k}\cos\mu_kct\sin\frac{m\phi}{2}J_{\frac{m}{2}}(\mu_kr) \] \begin{flalign*} &\text{where $\mu_k$ is a root of} &\frac{J_{\frac{m}{2}}(\mu a) }{\mu^{\frac{m}{2}}a^{\frac{m}{2}}}&=0&&\phantom{where mu is a root of} \end{flalign*} \begin{flalign*} &\text{and} &B_{m,k}&=\frac{2}{\pi}\frac{\displaystyle\int\limits_0^{2\pi}d\phi\int\limits_0^a rf(r,\phi)\sin\frac{m\phi}{2}J_{\frac{m}{2}}(\mu_k r)dr}{a^2[{J_{\frac{m}{2}}}'(\mu_ka)]^2}.&&\phantom{and} \end{flalign*} % -----File: 246.png For the terms in which $m$ is odd, $J_{\frac{m}{2}}(x)$ can be readily obtained from (13) Art.~122, which will become a finite sum. For example, (13) Art.~122 gives the values \begin{gather*} J_{\frac{1}{2}}(x)=\sqrt{\frac{2}{\pi x }}\sin x; \quad J_{\frac{3}{2}}(x) =\sqrt{\frac{2}{\pi x } }\bigg(\frac{1}{x} \sin x-\cos x \bigg);\\ J_{\frac{5}{2}}(x)=-\sqrt{\frac{2}{\pi x }}\bigg[\bigg(1+\frac{3}{x^2}\bigg)\sin x+\frac{3}{x} \cos x \bigg]; \quad \text{\&c}. \end{gather*} 13.\quad 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 \[ D_t u=\frac{a^2}{r^2}\biggr [D_r(r^2 D_r u) + \frac{1}{\sin\theta} D_{\theta}(\sin\theta D_{\theta} u) + \frac{1}{\sin^2\theta} D_{\phi}^2 u\biggr] \tag{1} \] (\smallrom{IV} Art.~1) subject to the conditions \begin{gather*} u = 0 \quad \text{when}\quad r = c,\\ u =f(r, \theta, \phi)\quad \text{when}\quad t = 0. \end{gather*} If we assume $u = T.R.V$ where $T$ is a function of $t$ only, $R$ of $ r$ only, and $V$ of $\theta$ and $\phi$ only, (1) can be broken up into \begin{gather*} \frac{dT}{dt} + a^2 \alpha^2 T = 0 \tag{2}\\ m(m+1)V + \frac{1}{\sin\theta }D_{\theta} (\sin\theta D_{\theta} V) + \frac{1}{\sin^2 \theta} D_{\phi}^2 V = 0 \tag{3} \end{gather*} \begin{flalign*} &\text{and}&\frac{ d^2R}{dr^2} + \frac{2}{r }\frac{dR}{dr }\ + \biggr[\alpha^2 - \frac{m(m+1)}{r^2} \biggr ]R=0. && \tag{4} \end{flalign*} Hence $T = e^{-a^2 \alpha^2 t}$, $V=Y_m (\mu, \phi)$ [v.\ Art.~102 (2)], and $R$ is still to be found. If in (4) we let $x=\alpha r$ and $z=R\sqrt{\alpha r}$ it becomes \[ \frac{d^2z}{dx^2} + \frac{1}{x} \frac{dz}{dx} + \biggr [ 1 - \frac{(m+\frac{1}{2})^2 }{x^2} \biggr ] z=0 \] which is satisfied by $z=J_{m+\frac{1}{2}} (x)$. (v.\ Art.~17.) \begin{flalign*} &\text{\indent Therefore}& R= \frac{1}{\sqrt{\alpha r }}J_{m+\frac{1}{2}} (\alpha r).&&\phantom{\indent Therefore} \end{flalign*} % -----File: 247.png \begin{flalign*} &&f(r,\theta,\phi)&=\dfrac{1}{4\pi}\sum\limits^{m=\infty}_{m=0} (2m+1) \int\limits^{2\pi}_0 d\phi_1 \int\limits^{\pi}_0 f(r,\theta_1,\phi_1)P_m(\cos \gamma)\sin\theta_1 d\theta_1&\tag*{by (3) Art.~114,}\\ &&&=\sum\limits^{m=\infty}_{m=0} \sum\limits^{n=m}_{n=0} [A_{m,n}f_{m,n}(r)\cos n\phi + B_{m,n}F_{m,n}(r)\sin n\phi]P_m^n(\mu).& \end{flalign*} \[ \sqrt{r}f_{m,n}(r)=\sum\limits^{k=\infty}_{k=0} C_{m,n,k}J_{m+\frac{1}{2}}(\alpha_kr) \] where $\alpha_k$ is a root of the equation \[ \dfrac{J_{m+\frac{1}{2}}(\alpha c)}{(\alpha c)^{m+\frac{1}{2}}}=0 \] \begin{flalign*} &\text{and} &&C_{m,n,k}= \dfrac{2\displaystyle \int\limits^c_0 r^{\frac{3}{2}}f_{m,n}(r)J_{m+\frac{1}{2}}(\alpha_kr)dr}{c^2[J'_{m+\frac{1}{2}}(\alpha_kc)]^2}.&&\\ & &&\sqrt{r}F_{m,n}(r)=\sum\limits^{m=\infty}_{m=0} D_{m,n,k}J_{m+\frac{1}{2}}(\alpha_kr)\\ &\text{where }&& D_{m,n,k}= \dfrac{2\displaystyle\int\limits^c_0 r^{\frac{3}{2}}F_{m,n}(r)J_{m+\frac{1}{2}}(\alpha_kr)dr}{c^2[J'_{m+\frac{1}{2}}(\alpha_kc)]^2}.&&\phantom{where} \end{flalign*} The final solution is \begin{multline*} u=\dfrac{1}{\sqrt{r}}\sum\limits^{m=\infty}_{m=0} \sum\limits^{n=m}_{n=0} \bigg[P_m^n(\mu)\sum\limits^{k=\infty}_{k=1} (A_{m,n}C_{m,n,k}\cos n\phi\\ \vphantom{ \sum\limits^{n=m}_{n=0}} +B_{m,n}D_{m,n,k} \sin n\phi)e^{-a^2\alpha_k^2 t} J_{m+\frac{1}{2}}(\alpha_kr)\bigg] \end{multline*} cf.\ Riemann, Par.\ Dif.\ Gl., \S\S~72 and 73. \label{ch7end} % -----File: 248.png \mychap{CHAPTER VIII.}{LAPLACE'S EQUATION IN CURVILINEAR COÖRDINATES. ELLIPSOIDAL HARMONICS.} \label{ch8start} \mypara{130.} \textit{Orthogonal Curvilinear Coördinates.} \begin{flalign*} &\raisebox{3.5ex}{If}& \left. \begin{gathered} F_1(x,y,z) =\rho_1\\ F_2(x,y,z) = \rho_2\\ F_3(x,y,z) = \rho_3 \end{gathered} \right\}&&\tag{1} \end{flalign*} are the equations in rectangular coördinates of three surfaces that are mutually perpendicular no matter what the values of $\rho_1$, $\rho_2$, and $\rho_3$, the parameters $\rho_1$, $\rho_2$, and $\rho_3$, may be regarded as a set of coördinates for a point of intersection of the three surfaces, in the sense that when $\rho_1$, $\rho_2$, $\rho_3$ are given the point in question is determined, and when the point is given the corresponding values of $\rho_1$, $\rho_2$, $\rho_3$, can be found. From equations (1) $x$, $y$, and $z$ can be expressed in terms of $\rho_1$, $\rho_2$, and $\rho_3$. Suppose this done. If now $x$, $y$, $z$ are the rectangular coördinates of the point $\rho_1=a$, $\rho_2=b$, $\rho_3=c$, the rectangular coördinates of the points $\rho_1=a+d\rho_1$, $\rho_2=b$, $\rho_3=c$, are obviously $x+D_{\rho_1}x.d\rho_1+\epsilon_1$, $y+D_{\rho_1} y.d\rho_1+\epsilon_2$, $z+D_{\rho_1}z.d\rho_1+\epsilon_3$, where $\epsilon_1$, $\epsilon_2$, and $\epsilon_3$ are infinitesimals of higher order than $d\rho_1$. Hence the square of the distance between the points will differ by an infinitesimal of higher order than that of $d\rho_1^2$ from $dn_1^2$ where \[ dn_1^2=[(D_{\rho_1}x)^2+(D_{\rho_1}y)^2+(D_{\rho_1}z)^2]d\rho_1 ^2. \] \begin{flalign*} &\raisebox{6.5ex}{\indent Let}& \left. \begin{aligned} \frac{1}{h_1^2}&=(D_{\rho_1}x)^2+(D_{\rho_1}y)^2+(D_{\rho_1}z)^2\\ \frac{1}{h_2^2}&=(D_{\rho_2}x)^2+(D_{\rho_2}y)^2+(D_{\rho_2}z)^2\\ \frac{1}{h_3^2}&=(D_{\rho_3}x)^2+(D_{\rho_3}y)^2+(D_{\rho_3}z)^2. \end{aligned} \right\}&&\tag{2} \end{flalign*} Then if $dn_1$ is the element of length normal to the surface $\rho_1=a$, $dn_2$ normal to $\rho_2=b$, and $dn_3$ normal to $\rho_3=c$ \[ dn_1=\frac{d\rho_1}{h_1},\quad dn_2=\frac{d\rho_2}{h_2},\quad dn_3=\frac{d\rho_3}{h_3}. \tag{3} \] % -----File: 249.png The element of surface $dS_1$ on the surface $\rho_1=a$ is easily seen to be \[ dS_1=\frac{d\rho_2d\rho_3}{h_2h_3}; \tag{4} \] and the element of volume $dv$ is \[ dv=\frac{d\rho_1d\rho_2d\rho_3}{h_1h_2h_3}. \tag{5} \] \EXAMPLE{} \begin{flalign*} &\text{\indent Show that} & h_1^2&=(D_x\rho_1)^2+(D_y\rho_1)^2+(D_z\rho_1)^2&\phantom{\indent Show that}\\ &&h_2^2&=(D_x\rho_2)^2+(D_y\rho_2)^2+(D_z\rho_2)^2&\\ &&h_3^2&=(D_x\rho_3)^2+(D_y\rho_3)^2+(D_z\rho_3)^2.& \end{flalign*} \textit{Suggestion:} If $h_1$ has the value just given $\dfrac{D_x\rho_1}{h_1}$, $\dfrac{D_y\rho_1}{h_1}$, $\dfrac{D_z\rho_1}{h_1}$ are the direction cosines of the normal at any given point of $\rho_1=a$. (v.\ Int.\ Cal.\ page 161.) Then \[ dn_1 = \frac{D_x\rho_1}{h_1}dx + \frac{D_y\rho_1}{h_1}dy + \frac{D_z\rho_1}{h_1}dz = \frac{1}{h_1}d\rho_1. \] \markright{ELLIPSOIDAL HARMONICS.} \mypara{131.} \textit{Laplace's Equation in orthogonal curvilinear coördinates.} If we apply the special form of Green's Theorem \[ \iiint\nabla^2Vdxdydz=\int D_nVdS \tag{v.\ Art.~98} \] to the space bounded by the surfaces $\rho_1=a$, $\rho_2=b$, $\rho_3=c$, $\rho_1=a+d\rho_1$, $\rho_2=b+d\rho_2$, $\rho_3=c+d\rho_3$, we have % recast to fit page width \begin{align*} \frac{\nabla^2Vd\rho_1d\rho_2d\rho_3}{h_1h_2h_3}&=\\ -h_1D_{\rho_1}&V\frac{d\rho_2d\rho_3}{h_2h_3}+h_1D_{\rho_1}V\frac{d\rho_2d\rho_3}{h_2h_3}+D_{\rho_1}\left(\frac{h_1}{h_2h_3}D_{\rho_1}V\right)d\rho_1d\rho_2d\rho_3\\ -h_2D_{\rho_2}&V\frac{d\rho_3d\rho_1}{h_3h_1}+h_2D_{\rho_2}V\frac{d\rho_3d\rho_1}{h_3h_1}+D_{\rho_2}\left(\frac{h_2}{h_3h_1}D_{\rho_2}V\right)d\rho_1d\rho_2d\rho_3\\ -h_3D_{\rho_3}&V\frac{d\rho_1d\rho_2}{h_1h_2}+h_3D_{\rho_3}V\frac{d\rho_1d\rho_2}{h_1h_2}+D_{\rho_3}\left(\frac{h_3}{h_1h_2}D_{\rho_3}V\right)d\rho_1d\rho_2d\rho_3; \end{align*} whence \[ \nabla^2V=h_1h_2h_3\bigg[ D_{\rho_1}\bigg(\frac{h_1}{h_2h_3}D_{\rho_1}V\bigg) + D_{\rho_2}\bigg(\frac{h_2}{h_3h_1}D_{\rho_2}V\bigg) + D_{\rho_3}\bigg(\frac{h_3}{h_1h_2}D_{\rho_3}V\bigg)\bigg], \tag{6} \] and Laplace's Equation in our curvilinear system is \[ h_1h_2h_3\bigg[ D_{\rho_1}\bigg(\frac{h_1}{h_2h_3}D_{\rho_1}V\bigg) + D_{\rho_2}\bigg(\frac{h_2}{h_3h_1}D_{\rho_2}V\bigg) + D_{\rho_3}\bigg(\frac{h_3}{h_1h_2}D_{\rho_3}V\bigg)\bigg]=0. \tag{7} \] % -----File: 250.png If it happens that $\nabla^2\rho_1=0$, $V=\rho_1$ will satisfy (7) and we shall have $h_1h_2h_3D_{\rho_1}\bigg(\dfrac{h_1}{h_2h_3}\bigg)=0$. In like manner if $\nabla^2\rho_2=0$ we have $D_{\rho_2}\bigg(\dfrac{h_2}{h_3h_1}\bigg)=0$, and if $\nabla^2\rho_3=0$ we have $D_{\rho_3}\bigg(\dfrac{h_3}{h_1h_2}\bigg)=0$; and therefore (7) reduces to \[ h_1^2 D_{\rho_1}^2V + h_2^2 D_{\rho_2}^2V + h_3^2 D_{\rho_3}^2V=0\tag{8} \label{err250} \] when $\nabla^2\rho_1=0$, $\nabla^2\rho_2=0$, and $\nabla^2\rho_3=0$. \mypara{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 \emph{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 \emph{spheroid} and the \emph{hyperboloid} orthogonally. The equations of the three surfaces can be written:--- \begin{gather*} F_1(x,y,z,\lambda)=\frac{x^2}{\lambda^2}+\frac{y^2}{\lambda^2-b^2}+\frac{z^2}{\lambda^2}-1=0 \tag{1}\\ F_2(x,y,z,\mu) =\frac{x^2}{\mu^2} +\frac{y^2}{\mu^2 -b^2}+\frac{z^2}{\mu^2} -1=0 \tag{2}\\ F_3(x,y,z,\nu) =z - \nu x = 0, \tag{3} \end{gather*} where $\lambda^2>b^2>\mu^2$, $2b$ being the distance between the foci. For all values of $\lambda$, $\mu$, and $\nu$ consistent with the inequality above written the surfaces (1), (2), (3) intersect in real points and cut orthogonally. $\lambda$, $\mu$, and $\nu$ 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 \smallrom{XV} Art.~1. From (1), (2), and (3) we readily get \begin{gather*}\left. \begin{split} x^2 &= \frac{\lambda^2\mu^2}{b^2(1+\nu^2)}\\ y^2 &= \frac{(\lambda^2-b^2)(b^2-\mu^2)}{b^2}\\ z^2 &= \frac{\lambda^2\mu^2\nu^2}{b^2(1+\nu^2)}; \end{split}\right\}\tag{4} \end{gather*} % -----File: 251.png whence \quad $D_\lambda x=\dfrac{\mu}{b\sqrt{1+\nu^2}}$, \quad $D_\lambda y=\dfrac{\lambda}{b}\sqrt{\dfrac{b^2-\mu^2}{\lambda^2-b^2}}$, \quad $D_\lambda z=\dfrac{\mu\nu}{b\sqrt{1+\nu^2}}$; \begin{flalign*} &\text{and} & \frac{1}{h_1^2} = \frac{\lambda^2-\mu^2}{\lambda^2-b^2}& \tag{5}& \end{flalign*} [v.~130~(2)]. In like manner we get \[ \frac{1}{h_2^2} = \frac{\lambda^2-\mu^2}{b^2-\mu^2} \tag{6} \] \begin{flalign*} &\text{and} & \frac{1}{h_3^2} = \frac{\lambda^2\mu^2}{b^2(1+\nu^2)^2}, &\tag{7}& \end{flalign*} and \smallrom{XV} Art.~1 becomes % recast to fit page width \begin{align*} &\frac{\mu}{b(1+\nu^2)\sqrt{b^2-\mu^2}}D_\lambda[\lambda\sqrt{\lambda^2-b^2}.D_\lambda V]\\ + &\frac{\lambda}{b(1+\nu^2)\sqrt{\lambda^2-b^2}}D_\mu[\mu\sqrt{b^2-\mu^2}.D_\mu V]\\ + &\frac{b(\lambda^2-\mu^2)}{\lambda\mu\sqrt{(\lambda^2-b^2)(b^2-\mu^2)}}D_\nu[(1+\nu^2)D_\nu V]=0, \tag{8} \end{align*} which is Laplace's Equation in terms of our \emph{Spheroidal Coördinates} $\lambda$, $\mu$, and $\nu$. If now in place of $\lambda$, $\mu$, and $\nu$ we can introduce some function of $\lambda$, some function of $\mu$, and some function of $\nu$ which, therefore, will represent the same set of orthogonal surfaces, and if we can choose these functions $\alpha$, $\beta$, and $\gamma$, which of course are functions of $x$, $y$, and $z$, so that $\nabla^2\alpha=0$, $\nabla^2\beta=0$, and $\nabla^2\gamma=0$, equation (8) must reduce to the simple and symmetrical form given in \smallrom{XVI} Art.~1. These functions $\alpha$, $\beta$, and $\gamma$ are easily found. Equation (8) is $\nabla^2V=0$ expressed in terms of $\lambda$, $\mu$, and $\nu$. Assume that $V$ is a function of $\lambda$ only; then $D_\mu V=0$, and $D_\nu V=0$, and (8) reduces to \begin{flalign*} &&D_\lambda[&\lambda\sqrt{\lambda^2-b^2}.D_\lambda V]=0&&\\ &\text{whence} && \lambda\sqrt{\lambda^2-b^2}\ \frac{dV}{d\lambda}=c_1,&&\phantom{whence}\\ &&&\hspace{0.5em}dV=\frac{c_1d\lambda}{\lambda\sqrt{\lambda^2-b^2}},&&\\ &\text{and} & &\hspace{1em}V=\frac{c_1}{b}\sec^{-1}\frac{\lambda}{b},&& \end{flalign*} and is a function of $\lambda$ which satisfies Laplace's Equation. % -----File: 252.png Take this as $\alpha$ leaving $c_1$ at present undetermined, so that \[ d\alpha=\frac{c_1d\lambda}{\lambda\sqrt{\lambda^2-b^2}}\quad \text{and}\quad \alpha=\frac{c_1}{b}\sec^{-1}\frac{\lambda}{b}. \] In the same way we get \[ d\beta=\frac{c_2d\mu}{\mu\sqrt{b^2-\mu^2}}\quad \text{and}\quad \beta=\frac{c_2}{b}\sech^{-1}\frac{\mu}{b}, \] (v.\ Int.\ Cal.\ Art.~46, Ex.) \[ d\gamma=\frac{c_3d\nu}{1+\nu^2},\quad \text{and}\quad \gamma=c_3\tan^{-1}\nu. \] Substituting these values in (8) and taking $c_1=-c_2=b$, and $c_3=1$, (8) reduces at once to \[ \frac{D_\alpha^2V}{\lambda^2}+\frac{D_\beta^2V}{\mu^2}+\frac{\lambda^2-\mu^2}{\lambda^2\mu^2}D_\gamma^2V=0, \tag{9}\\[-4ex] \] \begin{flalign*} &\text{or since} & \lambda&=b\sec\alpha,\quad \mu=b\sech\beta,\quad \text{and}\quad \nu=\tan\gamma, & \tag{10}&\\[1ex] &\text{to} & \cos^2\alpha &D_\alpha^2V+\cosh^2\beta D_\beta^2V+(\cosh^2\beta-\cos^2\alpha)D_\gamma^2V=0 & \tag{11}& \end{flalign*} which is Laplace's Equation in terms of what we may call \emph{Normal Oblate Spheroidal Coördinates}. In using (11) it is to be noted that the point whose coördinates are $(\alpha, \beta, \gamma)$ is the point of intersection of an oblate spheroid whose semi-axes are $b\sec\alpha$ and $b\tan\alpha$, an unparted hyperboloid of revolution whose semi-axes are $b\sech\beta$ and $b\tanh\beta$, and a plane containing the axis of the system and making the angle $\gamma$ 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 $(\alpha, \beta, \gamma)$ are \[ x=b\sec\alpha\sech\beta\cos\gamma,\quad y=b\tan\alpha\tanh\beta,\quad z=b\sec\alpha\sech\beta\sin\gamma \tag{12} \] [v.~(4)]. If now we let $\alpha$ range from 0 to $\dfrac{\pi}{2}$, $\beta$ from $-\infty$ to $\infty$, and $\gamma$ from 0 to $2\pi$, we shall be able to represent all points in space; and if we agree that negative values of $\beta$ shall belong to points below a plane through the origin and perpendicular to the axis of revolution and positive values of $\beta$ 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. % -----File: 253.png \EXAMPLE{S} 1.\quad If the spheroid is a \emph{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, \begin{align*} \frac{x^2}{\lambda^2}+\frac{y^2}{\lambda^2-b^2}+\frac{z^2}{\lambda^2-b^2}-1&=0\\ \frac{x^2}{\mu^2}+\frac{y^2}{\mu^2-b^2}+\frac{z^2}{\mu^2-b^2}-1&=0 \end{align*} \[ z-\nu y=0 \] \begin{flalign*} &\text{where} & \lambda^2>b^2>\mu^2. &&\phantom{where} \end{flalign*} \[ h_1^2=\frac{\lambda^2-b^2}{\lambda^2-\mu^2},\quad h_2^2=\frac{b^2-\mu^2}{\lambda^2-\mu^2},\quad h_3^2=\frac{b^2(1+\nu^2)^2}{(\lambda^2-b^2)(b^2-\mu^2)}. \] Laplace's Equation becomes \begin{align*} \frac{1}{b^2(1+\nu^2)}D_\lambda[(\lambda^2-b^2)D_\lambda V] &+\frac{1}{b^2(1+\nu^2)}D_\mu[(b^2-\mu^2)D_\mu V]\\ &+\frac{\lambda^2-\mu^2}{(\lambda^2-b^2)(b^2-\mu^2)}D_\nu[(1+\nu^2)D_\nu V]=0. \tag{1} \end{align*} \begin{flalign*} &\text{(1) reduces to} \quad \frac{D_\alpha^2V}{\lambda^2-b^2} + \frac{D_\beta^2V}{b^2-\mu^2} + \frac{\lambda^2-\mu^2}{(\lambda^2-b^2)(b^2-\mu^2)}D_\gamma^2V=0, &\hfill\tag{2}\\[-5ex] \end{flalign*} \begin{flalign*} &\text{where} && d\alpha=-\frac{bd\lambda}{\lambda^2-b^2},\quad d\beta=\frac{bd\mu}{b^2-\mu^2},\quad d\gamma=\frac{d\nu}{1+\nu^2},&&\phantom{where} \end{flalign*} \[ \alpha=\ctnh^{-1}\frac{\lambda}{b},\quad \beta =\tanh^{-1}\frac{\mu}{b},\quad \text{and}\quad \gamma=\tan^{-1}\nu. \] \begin{flalign*} &\indent\text{Since} & \lambda=b\ctnh\alpha,\quad \mu =b\tanh\beta,\quad \text{and}\quad \nu = \tan\gamma&&\phantom{\indent Since} \end{flalign*} (2) can be reduced to \[ \sinh^2\alpha D_\alpha^2V +\cosh^2\beta D_\beta^2 V + (\sinh^2\alpha + \cosh^2\beta)D_\gamma^2 V=0. \tag{3} \] In using (3) it is to be noted that the point $(\alpha, \beta, \gamma)$ is the point of intersection of a prolate spheroid whose semi-axes are $b\ctnh\alpha$ and $b\csch\alpha$, a biparted hyperboloid of revolution whose semi-axes are $b\tanh\beta$ and $b\sech\beta$, and a plane containing the axis of revolution and making the angle $\gamma$ with a fixed plane. % -----File: 254.png If the fixed plane is that of ($XY$) the rectangular coördinates of any point $(\alpha, \beta, \gamma)$ are \[ x = b \ctnh \alpha \tanh \beta, \quad y = b \csch \alpha \sech \beta \cos \gamma, \quad z = b \csch \alpha \sech \beta \sin\gamma, \] and $\alpha$ may range from $\infty$ to 0, $\beta$ from $-\infty$ to $\infty$, and $\gamma$ from 0 to $2\pi$. Negative values of $\beta$ are to be taken for points lying to the left of a plane through the origin perpendicular to the axis of revolution.\\ 2.\quad Transform Laplace's Equation in Spherical Coördinates \smallrom{XIII} Art.~1 to the symmetrical form \[ \alpha^2D_\alpha^2V + \cosh^2\beta D_\beta^2V + \cosh^2\beta D_\gamma^2V = 0 \] \begin{flalign*} &\text{where}&& \alpha = \dfrac{1}{r},\quad \beta =\log\tan \dfrac{\theta}{2},\quad \text{and}\quad \gamma = \phi.&&\phantom{where} \end{flalign*} 3.\quad Transform Laplace's Equation in Cylindrical Coördinates \smallrom{XIV} Art.~1 to the symmetrical form \[ D_\alpha^2V + D_\beta^2V + e^{2\alpha}D_\gamma^2V = 0 \] \begin{flalign*} &\text{where}& &\alpha = \log r, \quad \beta = \phi, \quad \text{and}\quad \gamma = z.&&\phantom{where} \end{flalign*} \mypara{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 \smallrom{XVI} Art.~1; and it is evident that our new coördinate $\alpha$ is a value of $V$ corresponding to such a distribution that the surfaces obtained by giving particular values to $\rho_1$ are \emph{equipotential} surfaces; that $\beta$ is a value of $V$ corresponding to such a distribution that the surfaces obtained by giving particular values to $\rho_2$ are equipotential surfaces; and that $\gamma$ is a value of $V$ corresponding to such a distribution that the surfaces obtained by giving particular values to $\rho_3$ are equipotential surfaces. $\alpha$, $\beta$, and $\gamma$ are called by Lamé ``\emph{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 $\alpha$. It is merely the condition that $V$ in Laplace's Equation may be a function of $\rho_1$ alone. If $V$ is a function of $\rho_1$ alone \[ D_xV = \dfrac{dV}{d\rho_1}D_x\rho_1, \quad D_yV = \dfrac{dV}{d\rho_1}D_y\rho_1,\quad D_zV = \dfrac{dV}{d\rho_1}D_z\rho_1,\\[-2ex] \] \begin{align*} D_x^2 V &= \dfrac{d^2 V}{{d \rho_1}^2} (D_x \rho_1)^2 + \dfrac{dV}{d \rho_1} D_x^2 \rho_1\\ % -----File: 255.png D_y^2 V &= \dfrac{d^2 V}{{d \rho_1}^2} (D_y \rho_1)^2 + \dfrac{dV}{d \rho_1} D_y^2 \rho_1\\ D_z^2 V &= \dfrac{d^2 V}{{d \rho_1}^2} (D_z \rho_1)^2 + \dfrac{dV}{d \rho_1} D_z^2 \rho_1. \end{align*} Therefore $[(D_x \rho_1)^2 + (D_y \rho_1)^2 + (D_z \rho_1)^2]\dfrac{d^2 V}{d {\rho_1}^2} + [D_x^2 \rho_1 + D_y^2 \rho_1 + D_z^2 \rho_1]\dfrac{ dV}{d \rho_1} = 0$ \begin{flalign*} &\text{whence}& &\dfrac{D_x^2 \rho_1 + D_y^2 \rho_1 + D_z^2 \rho_1}{(D_x \rho_1)^2 + (D_y \rho_1)^2 + (D_z \rho_1)^2} = - \dfrac{d^2V}{d{\rho_1}^2}\div \dfrac{ dV}{d \rho_1}.&&\phantom{whence} \end{flalign*} \begin{flalign*} &\text{or} & &\dfrac{\nabla^2\rho_1}{h_1^2} = F_1 (\rho_1)&&\phantom{or} \end{flalign*} where $F_1(\rho_1)$ may be any function of $\rho_1$ alone. Our required conditions are then \[ \left. \begin{aligned} \dfrac{\nabla^2 \rho_1}{h_1^2} &= F_1 (\rho_1)\\ \dfrac{\nabla^2 \rho_2}{h_2^2} &= F_2 (\rho_2)\\ \dfrac{\nabla^2 \rho_3}{h_3^2} &= F_3 (\rho_3) \end{aligned} \right\} \tag{1} \] and when they are fulfilled the original curvilinear coördinates $\rho_1$, $\rho_2$, $\rho_3$, correspond to possible \emph{equipotential} or \emph{isothermal} surfaces, \emph{thermometric parameters} $\alpha$, $\beta$, and $\gamma$ exist, and the reduction of Laplace's Equation to the symmetrical form \smallrom{XVI} Art.~1 is possible. \mypara{134.} Returning to our Oblate Spheroid problem of Art.~132 we can proceed as usual to break up our equation (\textbf{11}) Art.~132. Assume that $V = L.M.N$, where $L$ is a function of $\alpha$ only, $M$ of $\beta$ only, and $N$ of $\gamma$ only. (11) Art.~132 becomes \begin{flalign*} &&\dfrac{\cos^2 \alpha}{L}\dfrac{ d^2L}{d \alpha^2} + \dfrac{\cosh^2 \beta}{M}\dfrac{ d^2 M}{ d \beta^2} +\dfrac{[\cosh^2 \beta - \cos^2 \alpha]}{N}\dfrac{ d^2N}{d\gamma^2} &= 0\\ &\text{or }& \dfrac{1}{L}\dfrac{\cos^2 \alpha}{ \cosh^2 \beta - \cos^2 \alpha} \dfrac{d^2L}{d \alpha^2} + \dfrac{1}{M}\dfrac{\cosh^2 \beta}{ \cosh^2 \beta - \cos^2 \alpha} \dfrac{d^2M}{d \beta^2} = -&\dfrac{ 1}{N}\dfrac{ d^2 N}{d \gamma^2}.&&\phantom{or} \end{flalign*} The first member is independent of $\gamma$, and the second member is independent of $\alpha$ and $\beta$, and the two members are identically equal. The second member is then independent of $\alpha$, $\beta$, and $\gamma$ and must be constant; call it $n^2$. We have, then, \begin{flalign*} &\rule{14em}{0em} \dfrac{d^2N}{d \gamma^2} + n^2 N = 0 \tag{1}&\\ % -----File: 256.png &\text{and} \rule{4em}{0em} \dfrac{ \cos^2\alpha}{L}\dfrac{ d^2L}{d\alpha^2} + \dfrac{\cosh^2\beta}{M}\dfrac{ d^2M}{d\beta^2} - n^2(\cosh^2\beta - \cos^2\alpha) = 0.&&&(2)\\ &\text{(1) gives us} \rule{6.5em}{0em} N = A \cos n\gamma + B \sin n\gamma. && &\tag{3} \end{flalign*} (2) can be written \begin{flalign*} &&\dfrac{\cos^2\alpha}{L}\dfrac{ d^2L}{d\alpha^2} + n^2\cos^2&\alpha = n^2\cosh^2\beta - \dfrac{\cosh^2\beta}{M}\dfrac{d^2M}{d\beta^2} = m(m+1), &\quad\\ &\text{whence}&\phantom{c}\cos^2\alpha \dfrac{d^2L}{d\alpha^2}& + [n^2\cos^2\alpha - m(m+1)]L = 0&\tag{4}\\ &\text{and} &\cosh^2\beta \dfrac{d^2M}{d\beta^2} &+ [m(m+1) - n^2\cosh^2\beta]M = 0. &\tag{5} \end{flalign*} If we introduce $x = \tanh\beta$ in (5) it becomes \[ (1-x^2)\dfrac{d^2M}{dx^2} - 2x \dfrac{dM}{dx} + \left[m(m+1) - \dfrac{n^2}{1-x^2}\right]M = 0 \tag{6} \] where since $x = \tanh\beta$ and $\beta$ may have any value from $-\infty$ to $\infty$, $x$ may have any value between $-1$ and 1. (6) is a familiar equation having for a particular solution \[ M = (1-x^2)^{\frac{n}{2}}\dfrac{ d^nP_m(x)}{dx^n} = P_m^n(x) = P_m^n(\tanh\beta).\tag{7} \] (v.\ Arts.~101 and 102). If we introduce in (4) $x = \tan\alpha$ it reduces to \[ (1+x^2)\dfrac{d^2L}{dx^2} + 2x \dfrac{dL}{dx} + \left[\dfrac{n^2}{1+x^2} - m(m+1)\right]L = 0.\tag{8} \] (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. This labor can be saved, however, by noting that if we let $x = \dfrac{y}{i}$ (8) becomes \[ (1-y^2)\dfrac{ d^2L}{dy^2} - 2y\dfrac{ dL}{dy} + \left[m(m+1) - \dfrac{n^2}{1-y^2}\right]L = 0 \] and is identical in form with (6). Hence \begin{flalign*} &&\rule{4em}{0em}L &= P_m^n(y)\quad\text{and}\quad L = (1-y^2)^{\frac{n}{2}}\dfrac{d^nQ_m(y)}{dy^n}&\text{(v.\ Art.~101)}, \end{flalign*} where $y = i \tan\alpha$, are particular solutions of (4). We can avoid imaginaries if we use the values \[ L = (-i)^{m-n} P_m^n(y)\quad \text{and}\quad L = i^{m+n+1}(1-y^2)^{\frac{n}{2}}\dfrac{ d^nQ_m(y)}{dy^n}.\qquad \tag{9} \] % -----File: 257.png Since we assumed $V= L.M.N$ we have \begin{flalign*} &\raisebox{-1.7ex}{and}& \left. \begin{aligned} V &= (A \cos n \gamma + B \sin n \gamma) P_m^n(\tanh \beta)(-i)^{m-n} P_m^n (i \tan \alpha)\\ V &= (A \cos n \gamma + B \sin n \gamma) P_m^n(\tanh \beta) i^{m+n+1} \sec^n \alpha \dfrac{d^n Q_m (i \tan \alpha)}{ (d(i \tan \alpha))^n} \end{aligned} \right\} (10)\\[-6ex] \end{flalign*} as particular solutions of (\textbf{11}) Art.~132. If the problem is symmetrical with respect to the axis of the spheroid $D_\gamma^2 V = 0$, $n^2 = 0$ and our particular solutions (10) reduce to \begin{flalign*} &\raisebox{-2ex}{and}& \left. \begin{aligned} V& = (-i)^m P_m (i \tan \alpha) P_m (\tanh \beta)&\\ V&= i^{m+1} Q_m (i \tan \alpha) P_m (\tanh \beta). \end{aligned} \right\}&&\tag{11} \end{flalign*} If, then, $V$ is given on the surface of a spheroid as a function of $\beta$ and $\gamma$, we must express it as a function of $\tanh \beta$ and $\gamma$, and shall be obliged to develop it in terms of \emph{Spherical Harmonics} of $\tanh \beta$ and $\gamma$ 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 $\tanh \beta$ by the formulas of Chapter VI. A convenient form for $Q_m (i \tan \alpha)$ is obtained from (2) Art.~100; it is \begin{flalign*} &&Q_m (i \tan \alpha) = -i P_m &(i \tan \alpha) \int\limits^{\infty}_{\tan\alpha}\dfrac{ dx}{ (1 + x^2)[P_m(xi)]^2}.& \tag{12} \\ &\text{Hence} &Q_0 (i \tan \alpha) = -&i \int\limits^{\infty}_{\tan\alpha}\dfrac{dx}{ 1 + x^2} = -i \left(\dfrac{\pi}{ 2} - \alpha\right).& \tag{13} \end{flalign*} \EXAMPLE{S} 1.\quad A conductor in the form of an oblate spheroid whose semi-axes are $b \sec \alpha_0$ and $b \tan \alpha_0$ is charged with electricity and is found to be at potential $V_0$; find the value of the potential function at any internal or external point. Here $V_0 = V_0P_0(\tanh \beta)$. Hence at an internal point \[ V = V_0 \dfrac{P_0 (i \tan \alpha) }{ P_0 (i \tan \alpha_0)} P_0 (\tanh \beta) = V_0, \tag{1} \] and at an external point \[ V = V_0 \dfrac{Q_0 (i \tan \alpha) }{Q_0 (i \tan \alpha_0)} P_0 (\tanh \beta) =V_0 \dfrac{ \left(\dfrac{\pi}{2} - \alpha\right) }{\left(\dfrac{ \pi }{ 2} - \alpha_0\right)}. \tag{2} \] Since $V$ in (2) involves $\alpha$ only, the equipotential surfaces are all spheroids confocal with the conductor.\\ % -----File: 258.png 2.\quad The upper half of an oblate spheroid whose semi-axes are $b \sec \alpha_0$ and $b \tan \alpha_0$ is kept at the temperature unity, and the lower half at the temperature zero. Find the permanent temperature at any internal point. \begin{flushright} \textit{Ans.} \quad $u = \dfrac{1}{2} + \dfrac{3}{4} \dfrac{P_1(i \tan\alpha)}{P_1(i \tan\alpha_0)}P_1(\tanh\beta) - \dfrac{7}{8}.\dfrac{1}{2 }\dfrac{P_3(i \tan\alpha)}{P_3(i \tan\alpha_0)}P_3(\tanh\beta) + \cdots$ \end{flushright} (v.\ Art.~93). $u$ may be expressed in terms of $x$, $y$, and $z$ without serious difficulty [v.\ (12) Art.~132]. \[ u = \frac{1}{2 }+ \frac{3}{4} \frac{y}{c} - \frac{7}{8}.\frac{1}{2}.\frac{1}{2}\frac{ [25y^3-15y(x^2+y^2+z^2-b^2)-9b^2y]}{5c^3+3b^2c} + \cdots \] if $2c = 2b \tan \alpha_0 = $ minor axis of spheroid. \mypara{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 $\alpha = \phi$ and $\alpha = \phi + d\phi$. By (1) Art.~98 we have \[ 4\pi\rho\kappa = [D_nV_1 - D_nV_2]_{\alpha=\phi}, \tag{1} \] where $\rho$ is the density and $\kappa$ the thickness of the shell, $V_1$ the value of the potential function at an internal point, and $V_2$ the value of the potential function at an external point. \begin{flalign*} &\indent \text{Let }&V_1 &= \sum A_m(-i)^mP_m(i \tan\alpha)P_m(\tanh\beta)&\\ &\text{and}&V_2 &= \sum B_m i^{m+1}Q_m(i \tan\alpha)P_m(\tanh\beta) \tag*{[v.\ (11) Art.~134].}& \end{flalign*} Since $V_1$ and $V_2$ must have the same value when $\alpha = \phi$ \[ A_m = B_m i^{2m+1}\frac{ Q_m(i \tan\phi)}{P_m(i \tan\phi)} = (-1)^m B_m \int\limits_{\tan\phi}^{\infty}\frac{dx}{(1+x^2)[P_m(xi)]^2} \tag{2} \] [v.\ (12) Art.~134]. \begin{flalign*} \left.\!\! \begin{alignedat}{2} &\indent \text{Hence\quad} & V_1 &= \sum i^mB_mP_m(\tanh\beta)P_m(i \tan\alpha) \int\limits_{\tan\phi}^{\infty}\frac{dx}{(1+x^2)[P_m(xi)]^2}\\ &\text{and}& V_2 &= \sum i^mB_mP_m(\tanh\beta)P_m(i \tan\alpha) \int\limits_{\tan\alpha}^{\infty}\frac{dx}{(1+x^2)[P_m(xi)]^2} . \end{alignedat} \right \}\hspace{2.8pt}(3) \end{flalign*} \[ D_nV_1 = D_{\alpha}V_1.D_n\alpha. \qquad\qquad D_nV_2 = D_{\alpha}V_2.D_n\alpha \] % -----File: 259.png \begin{align*} [D_nV_1-D_nV_2]_{\alpha=\phi}&=[D_\alpha V_1-D_\alpha V_2]_{\alpha=\phi}(D_n\alpha)_{\alpha=\phi}\\ &\qquad =[D_\alpha(V_1-V_2)]_{\alpha=\phi}[D_n\alpha]_{\alpha=\phi}. \end{align*} \[ V_1-V_2= \sum i^mB_mP_m(\tanh\beta)P_m(i \tan\alpha) \int\limits^{\tan\alpha}_{\tan\phi}\dfrac{ dx}{(1+x^2)[P_m(xi)]^2}. \] \begin{multline*} D_\alpha(V_1-V_2)= \sum i^mB_mP_m(\tanh\beta)\bigg[P_m(i \tan\alpha) \dfrac{\sec^2\alpha}{(1+\tan^2\alpha)[P_m(i \tan\alpha)]^2}\\ + \dfrac{dP_m(i \tan\alpha)}{d\alpha} \int\limits^{\tan\alpha}_{\tan\phi}\dfrac{ dx}{(1+x^2)[P_m(xi)]^2}\bigg]. \end{multline*} \begin{gather*} D_\alpha[V_1-V_2]_{\alpha=\phi} = \sum i^mB_m \dfrac{P_m(\tanh\beta)}{P_m(i \tan\phi)}.\\ D_n\alpha = \dfrac{d\alpha}{dn} \end{gather*} \[ dn=\dfrac{d\rho_1}{h_1}=\dfrac{d\lambda}{h_1}= \dfrac{\sqrt{ \lambda^2-\mu^2}}{\sqrt{\lambda^2-b^2}}d\lambda=b \sec\alpha \sqrt{\tan^2\alpha+\tanh^2\beta}.d\alpha \tag{4} \] v.\ Art.~130 (3), and Art.~132 (5) and (10). \begin{flalign*} & &[D_n\alpha]_{\alpha=\phi} &=\dfrac{1}{b \sec\phi\sqrt{\tan^2\phi+\tanh^2\beta}}.& % recast to fit page width &\intertext{\indent Hence} && [D_nV_1-D_nV_2]_{\alpha=\phi}& = \dfrac{1}{b \sec\phi\sqrt{\tan^2\phi+\tanh^2\beta}}\sum i^mB_m \dfrac{P_m(\tanh\beta)}{P_m(i \tan\phi)}.&\\ & & \kappa = [dn]_{\alpha=\phi} &= b\sec\phi \sqrt{\tan^2\phi+\tanh^2\beta}.d\phi& \end{flalign*} by (4), and (1) may be written \[ 4\pi\rho b^2 \sec^2\phi(\tan^2\phi+\tanh^2\beta)d\phi = \sum i^mB_m \dfrac{P_m(\tanh\beta)}{P_m(i \tan\phi)}.\tag{5} \] \begin{flalign*} &\indent\text{Since}& \tanh^2\beta &= \tfrac{1}{3}P_0(\tanh\beta)+\tfrac{2}{3}P_2(\tanh\beta)&\phantom{\indent Since} \end{flalign*} by (5) Art.~95, to satisfy (5) we must give $m$ the values 0 and 2 and \begin{flalign*} & &B_0&=\tfrac{4}{3}\pi\rho b^2 \sec^2\phi(3\tan^2\phi+1)d\phi&\\ &\text{and}& B_2&=\tfrac{4}{3}\pi\rho b^2 \sec^2\phi(3\tan^2\phi+1)d\phi.& \end{flalign*} % -----File: 260.png So that by (3) \begin{flalign*} &&V_1&=\tfrac{4}{3} \pi \rho b^2 \sec^2 \phi (3 \tan^2 \phi + 1)d \phi \bigg[\int\limits^{\infty}_{\tan\phi}\dfrac{ dx}{1+x^2}\\ &&&\hspace{7em}-P_2(\tanh \beta)P_2(i \tan \alpha) \int\limits^{\infty}_{\tan\phi} \dfrac{dx}{(1+x^2)[P_2(xi)]^2}\bigg] \tag{6} \\ &\text{and}& V_2&=\tfrac{4}{3} \pi \rho b^2 \sec^2 \phi (3 \tan^2 \phi + 1)d \phi [i Q_0 (i \tan \alpha)&\\ &&&\hspace{16em}+ i^3 P_2 (\tanh \beta) Q_2 (i \tan \alpha)]. &&\tag{7} \end{flalign*} The potential function at an external point due to the solid spheroid for which $\alpha = \alpha_0$ is \[ V=\!\!\int\limits^{\phi=\alpha_0}_{\phi=0}\!\!V_2 = \tfrac{4}{3} \pi \rho b^2 \sec^2 \alpha_0 \tan \alpha_0 [i Q_0 (i \tan \alpha) + i^3 P_2 (\tanh \beta) Q_2 (i \tan \alpha)].\tag{8} \] If $2a$ is the major axis and $2c$ the minor axis of the spheroid \[ \tfrac{4}{3} \pi\rho b^2 \sec^2\alpha_0 \tan\alpha_0 =\tfrac{4}{3} \dfrac{ \pi\rho a^2c}{b}=\dfrac{M}{b} \] where $M$ is the mass of the spheroid. Therefore \[ V = \dfrac{M}{b} [i Q_0 (i \tan \alpha) + i^3 P_2 (\tanh \beta)Q_2 (i \tan \alpha)] \tag{9} \] is the required value. (9) can be reduced to\\[2ex] \phantom{.}$\displaystyle\;V = \dfrac{M}{b} \left\{\dfrac{\pi}{2} - \alpha + \dfrac{1}{4}\left[\left(\dfrac{\pi}{2} - \alpha\right)(3 \tan^2 \alpha+1) - 3 \tan \alpha\right][3 \tanh^2 \beta - 1]\right\}.$\hfill(10) \EXAMPLE{S} 1.\quad Break up the equation (3) Ex.~1, Art.~132, for the prolate spheroid, and obtain particular solutions of the term \begin{gather*} V= (A \cos n\gamma + B \sin n\gamma)P_m^n(\tanh\beta)P_m^n(\ctnh\alpha),\\ V= (A \cos n\gamma + B \sin n\gamma)P_m^n(\tanh\beta)(-1)^{\frac{n}{2}} \csch^n \alpha \dfrac{d^n Q_m (\ctnh\alpha)}{(d \ctnh \alpha)^n}. \end{gather*} 2.\quad Break up and solve the equations of Exs.\ 2 and 3, Art.~132, and show that they lead to familiar forms.\\ 3.\quad If in Ex.~1, Art.~132, the conductor is a prolate spheroid whose semi-axes are $b \ctnh\alpha_0$ and $b \csch\alpha_0$ show that \[ V= V_0\text{ at an internal point.}\qquad V = V_0\dfrac{ \alpha}{\alpha_0}\text{ at an external point}. \] % -----File: 261.png 4.\quad Show that the potential function at an external point due to the attraction of a homogeneous solid prolate spheroid is \[ V = \dfrac{M}{b} [Q_0(\ctnh \alpha) - P_2(\tanh \beta)Q_2(\ctnh \alpha)]. \] \vspace{1ex}\begin{center} \emph{Ellipsoidal Harmonics.} \end{center}\vspace{-1ex} \mypara{136.} If we are dealing with an \emph{ellipsoid} instead of a spheroid, we can take as our orthogonal system of surfaces a set of \emph{confocal quadrics}; \[ \left. \begin{aligned} \dfrac{x^2}{\lambda^2} + \dfrac{y^2}{\lambda^2 -b^2} + \dfrac{z^2}{\lambda^2 - c^2} - 1& = 0\;\\ \dfrac{x^2}{\mu^2} + \dfrac{y^2}{\mu^2 - b^2} + \dfrac{z^2}{\mu^2 - c^2} - 1 &= 0\\ \dfrac{x^2}{\nu^2} + \dfrac{y^2}{\nu^2 - b^2} + \dfrac{z^2}{\nu^2 - c^2} - 1 &= 0 \end{aligned} \right\} \tag{1} \] where $\lambda^2 > c^2 > \mu^2 > b^2 > \nu^2$. 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. $\lambda$, $\mu$, and $\nu$ will be our curvilinear coördinates, and are known as Ellipsoidal Coördinates. We find without difficulty that % recast to fit page width \begin{multline*} x^2 = \dfrac{\lambda^2 \mu^2 \nu^2}{b^2 c^2},\quad y^2=\dfrac{(\lambda^2 - b^2)(\mu^2 - b^2)(b^2 - \nu^2)}{ b^2(c^2 - b^2)}, \\ z^2 = \dfrac{(\lambda^2 - c^2)(c^2 - \mu^2)(c^2 - \nu^2)}{ c^2(c^2 - b^2)},\tag{2} \end{multline*} % recast to fit page width \begin{multline*} h_1^2 = \dfrac{(\lambda^2 - b^2)(\lambda^2 - c^2)}{(\lambda^2 - \mu^2)(\lambda^2 - \nu^2)}, \quad h_2^2 = \dfrac{(\mu^2 - b^2)(c^2 - \mu^2)}{(\mu^2 -\nu^2)(\lambda^2 - \mu^2)},\\ h_3^2 = \dfrac{(b^2 - \nu^2)(c^2 - \nu^2)}{(\lambda^2 - \nu^2)(\mu^2 - \nu^2)}.\tag{3} \end{multline*} To avoid ambiguity, we shall suppose that of the nine semi-axes in (1) $\sqrt{\strxx c^2 - \mu^2}$ 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; $\sqrt{\strxx b^2 - \nu^2}$ 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; $\nu$ 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 to know not merely the values of its coördinates $\lambda$, $\mu$, and $\nu$, but the signs of $\sqrt{\strxx c^2 - \mu^2}$, and $\sqrt{\strxx b^2 - \nu^2}$ as well. % -----File: 262.png We shall see later, Art.~139, when we come to introduce what we may call the \emph{Normal Ellipsoidal Coördinates} $\alpha$, $\beta$, and $\gamma$ that they are free from this disadvantage. It is to be observed that $\lambda$ may range from $c$ to $\infty$, $\mu$ from $b$ to $c$, and $\nu$ from $-b$ to $b$. The element of length perpendicular to the Ellipsoid is \[ dn = \dfrac{d \lambda}{h_1} = \sqrt{\dfrac{(\lambda^2 - \mu^2)(\lambda^2 - \nu^2)}{(\lambda^2 - b^2)(\lambda^2 - c^2)}}.d \lambda. \tag{4} \] The element of Ellipsoidal surface is \[ dS = \dfrac{d \mu d \nu }{ h_2 h_3} = (\mu^2 - \nu^2) \sqrt{\dfrac{ (\lambda^2 - \mu^2)(\lambda^2 - \nu^2)}{(\mu^2 - b^2)(c^2 - \mu^2)(b^2 - \nu^2)(c^2 - \nu^2)}}.d \mu d\nu, \tag{5} \] and the element of volume is % recast to fit page width \begin{align*} dv &=\dfrac{ d \lambda d \mu d \nu}{ h_1h_2h_3}\\ &=\dfrac{ (\lambda^2 - \mu^2)(\lambda^2 - \nu^2)(\mu^2 - \nu^2)}{\sqrt{(\lambda^2 - b^2)(\lambda^2 - c^2)(\mu^2 - b^2)(c^2 - \mu^2)(b^2 - \nu^2)(c^2 - \nu^2)}} d \lambda d \mu d \nu. \tag{6} \end{align*} The surface integral of any given function of $\mu$ and $\nu$ taken over the ellipsoid is \begin{multline*} \int f(\mu, \nu)dS =\int\limits^{b}_{-b} d \nu \int\limits^{c}_{b} [f_1 (\mu, \nu) + f_2 (\mu, \nu) + f_3 (\mu, \nu)\\ +f_4 (\mu, \nu)](\mu^2 - \nu^2) \sqrt{\dfrac{(\lambda^2 - \mu^2)(\lambda^2 - \nu^2)}{(\mu^2 - b^2)(c^2 - \mu^2)(b^2 - \nu^2)(c^2 - \nu^2)}}.d \mu, \tag{7} \end{multline*} where $f_1(\mu, \nu)$, $f_2(\mu, \nu)$, $f_3(\mu, \nu)$ and $f_4(\mu, \nu)$ 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 \[ (\mu^2 - \nu^2) D_{\alpha}^2 V + (\lambda^2 - \nu^2) D_{\beta}^2 V + (\lambda^2 - \mu^2) D_{\gamma}^2 V = 0 \tag{8} \] \begin{flalign*} &\text{where}& \alpha = c \int\limits^{\lambda}_c \dfrac{d\lambda}{\sqrt{(\lambda^2 - b^2)(\lambda^2 - c^2)}},\quad \beta &= c \int\limits^{\mu}_b \dfrac{d \mu}{\sqrt{(c^2 - \mu^2)(\mu^2 - b^2)}},&\\ & & \gamma &= c \int \limits^{\nu}_0 \dfrac{d \nu }{\sqrt{(b^2 - \nu^2)(c^2 - \nu^2)}}. &\tag{9} \end{flalign*} % -----File: 263.png $\alpha$, $\beta$, and $\gamma$ can be expressed as Elliptic Integrals of the first class and are \begin{multline*} \alpha=F\Big(\dfrac{b}{c}, \dfrac{\pi}{2}\Big)-F\Big(\dfrac{b}{c}, \sin^{-1} \dfrac{c}{\lambda}\Big),\quad \beta=F\Big(\sqrt{ 1-\dfrac{ b^2}{c^2}},\, \sin^{-1} \sqrt{\dfrac{1- \dfrac{b^2}{\mu^2}}{1- \dfrac{b^2}{c^2}}}\Big),\\ \gamma=F\Big(\dfrac{b}{c}, \sin^{-1} \dfrac{\nu}{b}\Big); \tag{10} \end{multline*} \begin{flalign*} &\text{whence} &\lambda = \frac{c}{\sn(K-\alpha)} &\Big(\bmod \frac{b}{c}\Big) = c\,\frac{ \dn \alpha}{\cn \alpha} \Big(\bmod \frac{b}{c}\Big),&& \\ && \mu = \frac{b}{\dn \beta } &\Big( \bmod\Big( 1-\dfrac{b^2}{c^2} \Big)^{\frac{1}{2}} \Big),\quad \nu = b \sn \gamma\Big( \bmod \frac{b}{c} \Big)&& \tag{11} \end{flalign*} (v.\ Int.\ Cal.\ Arts.~179, 192, and 196). \mypara{137.} If in (8) Art.~136 we assume $V=L.M.N$ where $L$ involves $\alpha$ only, $M$ involves $\beta$ only, and $N$ involves $\gamma$ only, (8) can be written \[ \dfrac{\mu^2-\nu^2}{L}\dfrac{ d^2 L}{d\alpha^2} + \dfrac{\lambda^2 - \nu^2}{M} \dfrac{d^2 M}{d\beta^2} + \dfrac{\lambda^2 - \mu^2}{N}\dfrac{ d^2 N}{d\gamma^2} =0. \tag{1} \] (1) is too complicated to be broken up by our usual method. If, however, we let \[ \dfrac{1}{L}\dfrac{d^2L}{d\alpha^2} ={\textstyle\sum} a_k \lambda^k,\quad\dfrac{1}{M}\dfrac{ d^2M}{d\beta^2} ={\textstyle\sum} b_k \mu^k,\quad \dfrac{1}{N}\dfrac{ d^2N}{d\gamma^2} ={\textstyle\sum} c_k \nu^k, \] 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 $a_0 = -b_0 = c_0$, and $a_2 = -b_2 = c_2$. Therefore (1) can be broken up into the three equations \begin{align*} \dfrac{d^2L}{d\alpha^2}&= (a_0 + a_2 \lambda^2)L \\ \dfrac{d^2M}{d\beta^2} &= -(a_0 + a_2 \mu^2)M \\ \dfrac{d^2N}{d\gamma^2}& = (a_0 + a_2 \nu^2)N. \end{align*} % -----File: 264.png We shall find it convenient to take $a_2$ as $m(m + 1)$ and $a_0$ as $-(b^2 + c^2)p$; whence \[\left .\begin{aligned} \frac{d^2 L }{d \alpha^2} &- [m(m + 1) \lambda^2 - (b^2 + c^2)p]L = 0\\[1ex] \frac{d^2 M }{ d \beta^2} &+ [m(m + 1) \mu^2 - (b^2 + c^2)p]M = 0\; \\[1ex] \frac{d^2 N }{d \gamma^2} &- [m(m + 1) \nu^2 - (b^2 + c^2)p]N = 0 . \end{aligned} \right \} \tag{2}\] If now in (2) we replace $\alpha$, $\beta$, and $\gamma$ by their values in terms of $\lambda$, $\mu$, and $\nu$, we get \begin{equation}\left . \begin{aligned} (\lambda^2 - b^2) (\lambda^2 - c^2) \frac{d^2 L }{d\lambda^2} + \lambda (\lambda^2 - b^2 + \lambda^2 - c^2) \frac{dL }{d\lambda}\hspace{8em}\\ - [m(m + 1) \lambda^2 - (b^2 + c^2)p] L = 0\phantom{.}\\ (\mu^2 - b^2) (\mu^2 - c^2) \frac{d^2 M }{ d\mu^2} + \mu (\mu^2 - b^2 + \mu^2 - c^2) \frac{dM }{d\mu}\hspace{7em}\\ - [m(m + 1) \mu^2 - (b^2 + c^2)p] M = 0\phantom{.}\\ (\nu^2 - b^2)(\nu^2 - c^2)\frac{ d^2 N }{d\nu^2} + \nu (\nu^2 - b^2 + \nu^2 - c^2) \frac{dN}{ d\nu}\hspace{7.5em}\\ - [m(m + 1) \nu^2 - (b^2 + c^2)p] N = 0 . \end{aligned} \right \} \tag{3}\end{equation} Whence if $L = E_m^p (\lambda)$, it follows that $M = E_m^p (\mu)$ and $N = E_m^p (\nu)$, and that \[ V = E_m^p (\lambda) E_m^p (\mu) E_m^p (\nu) \tag{4} \] is a solution of Laplace's Equation, (8) Art.~136. The equation \begin{multline*} (x^2 - b^2) (x^2 - c^2)\frac{ d^2 z }{ dx^2} + x(x^2 - b^2 + x^2 - c^2) \frac{dz }{dx}\\ - [m(m + 1) x^2 - (b^2 + c^2)p] z = 0 \tag{5} \end{multline*} is known as Lamé's Equation, and $E_m^p (x)$ as a \textit{Lamé's Function} or an \textit{Ellipsoidal} \textit{Harmonic}. We shall suppose $m$ a positive integer. To get a particular solution of (5) let $z = \sum a_k x^k$. Substitute in (5) and reduce and we get \begin{multline*} [k(k + 1) - m(m + 1)] a_k - (b^2 + c^2) [(k + 2)^2 - p] a_{k+2}\\ + b^2 c^2 (k + 3)(k + 4) a_{k+4} = 0. \tag{6} \end{multline*} We have now only to choose a sequence of coefficients satisfying (6), and we may take any two consecutive coefficients arbitrarily. % -----File: 265.png (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 $a_{m + 2} = 0$, $a_{m+4}$, $a_{m+6}$, \&c., will vanish. Let $a_m = 1$. If $m$ is even the coefficient of $a_0$ 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 $a_{m-2}$, $a_{m-4}$, \&c., by the aid of (6) we see that $a_{m-2}$ is of the first degree in $p$, $a_{m-4}$ of the second degree in $p$, \&c., and $a_{-2}$ of the degree $\dfrac{m}{2} +1$ in $p$. There are then $\dfrac{m}{2} +1$ values of $p$ which we shall call $p_1$, $p_2$, $p_3$, \&c., for which $a_{-2}$ will vanish, and for which our solutions will be of the form \[ E_m^p(x)=x^m+a_{m-2}x^{m-2}+a_{m-4}x^{m-4}+ \cdots+a_0 \] if $m$ is even. If $m$ is odd, the coefficient of $a_1$ 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. $a_{-1}$ is of the degree $\dfrac{m+1}{2}$ in $p$, and there will be $\dfrac{m+1}{2}$ values $p_1$, $p_2$, $p_3$, \&c., of $p$ for which \[ E_m^p(x)=x^m+a_{m-2}x^{m-2}+a_{m-4}x^{m-4}+ \cdots+a_1x. \] Following Heine we shall call the solution just obtained $K_m^p(x)$ so that \[ K_m^p(x)=x^m+a_{m-2}x^{m-2}+a_{m-4}x^{m-4}+ \cdots \tag{7} \] terminating with $ a_0$ if $m$ is even, and with $a_1x$ if $m$ is odd. If $m$ is even, there are $\dfrac{m}{2} +1$ of these functions $K_m^{p_1}(x)$, $K_m^{p_2}(x)$, \&c., and there are $\dfrac{m+1}{2}$ of them if $m$ is odd. The coefficients can be computed by the aid of (6). If in Lamé's Equation (5) we let $z=v\sqrt{\strxx x^2-b^2}$ we get the equation \begin{multline*} (x^2-b^2) (x^2-c^2)\frac{ d^2v}{dx^2} + x[x^2-b^2+3(x^2-c^2)]\frac{ dv}{dx}\\ -[(m+2)(m-1)x^2 + c^2 - (b^2+c^2)p] v = 0. \tag{8} \end{multline*} Letting $v=\sum a_kx^k$ we obtain the relation \begin{multline*} [k(k+3)-(m+2) (m-1)]a_k - \{(b^2+c^2) [(k+2)^2-p] + c^2(2k+5)\} a_{k+2}\\ + b^2c^2 (k+3) (k+4) a_{k+4} = 0. \tag{9} \end{multline*} % -----File: 266.png Proceeding exactly as before, we find that there are $\dfrac{m}{2}$ values $q_1$, $q_2$, $q_3$, \&c., of $ p$ for which $v=x^{m-1}+a_{m-3}x^{m-3} + \cdots + a_{1}x$ if $m$ is even, and $\dfrac{m+1}{2}$ values for which $v=x^{m-1}+a_{m-3}x^{m-3} + \cdots + a_{0}$ if $m$ is odd. Calling $v\sqrt{\strxx x^2-b^2}$ $L_{m}^{p}(x)$ so that \[ L_{m}^{p}(x) = \sqrt{ x^2-b^2}[x^{m-1}+a_{m-3}x^{m-3} + a_{m-5}x^{m-5} + \cdots], \tag{10} \] terminating with $a_1x$ if $m$ is even and with $a_0$ if $m$ is odd, we have $\dfrac{m}{2}$ values of $E_m^p(x)$, namely $L_m^{q_1}(x)$, $L_m^{q_2}(x)$, \&c., of the form (10) if $m$ is even and $\dfrac{m+1}{2}$ values if $m$ is odd. By interchanging $b$ and $c$ in (8), (9), and (10) we may show that if \[ M_{m}^{p}(x) = \sqrt{x^2-c^2}[x^{m-1}+a_{m-3}x^{m-3} + a_{m-5}x^{m-5} + \cdots ] \tag{11} \] there are $\dfrac{m}{2}$ values of $E_m^p(x)$, namely $M_m^{r_1}(x)$, $M_m^{r_2}(x)$, $M_m^{r_3}(x)$, \&c., of the form (11) if $m$ is even and $\dfrac{m+1}{ 2}$ values if $m$ is odd. Finally if in Lamé's Equation (5) we let $z=v \sqrt{\strxx(x^2-b^2)(x^2-c^2)}$ we get \begin{multline*} (x^2-b^2)(x^2-c^2)\frac{d^2v}{dx^2}+3x(x^2-b^2+x^2-c^2)\frac{dv}{dx}\\ -[(m+3)(m-2)x^2-(b^2+c^2)(p-1)]v=0. \tag{12} \end{multline*} If now we let $v= \sum a_kx^k$ we obtain the relation \begin{multline*} [k(k+5)-(m-2)(m+3)]a_k\\ -(b^2+c^2)[(k+2)(k+4)+1-p]a_{k+2}+b^2c^2(k+3)(k+4)a_{k+4}=0. \tag{13} \end{multline*} Proceeding as before we find that there are $\dfrac{m}{2}$ values $s_1$, $s_2$, $s_3$, \&c., of $ p$ for which $v=x^{m-2} + a_{m-4}x^{m-4} + a_{m-6}x^{m-6} + \cdots + a_0$ if $m$ is even, and $\dfrac{m+1}{2}$ values for which $ v=x^{m-2}+a_{m-4}x^{m-4} + \cdots + a_1x$ if $m$ is odd. Calling $v\sqrt{\strxx(x^2-b^2)(x^2-c^2) }$ $N_m^p(x)$ so that \[ N_m^p(x) = \sqrt{(x^2-b^2)(x^2-c^2) }[x^{m-2} + a_{m-4}x^{m-4} + a_{m-6}x^{m-6} + \cdots] \tag{14} \] terminating with $a_0$ if $m$ is even and with $a_1x$ if $m$ is odd, we have $\dfrac{m}{2 }$ values of $ E_m^p(x) $, namely $N_m^{s_1}(x)$, $N_m^{s_2}(x)$, $N_m^{s_3}(x)$, \&c., of the form (14) if $m$ is even and $\dfrac{m-1}{2}$ values if $m$ is odd. % -----File: 267.png Summing up our results we see that there are $2m+1$ Ellipsoidal Harmonics $E_m^p(x)$ each of which is a finite sum of the $m$th degree in $x$, or in $x$ and $\sqrt{\strxx x^2-b^2}$, or in $x$ and $\sqrt{\strxx x^2-c^2}$, or in $x$ and $\sqrt{\strxx x^2-b^2}$ and $\sqrt{\strxx x^2-c^2}$. It was proved by Lamé that the $2m + 1$ values of $p$, namely $p_1$, $p_2$, $p_3$, \&c., $q_1$, $q_2$, $q_3$, \&c., $r_1$, $r_2$, $r_3$, \&c., $s_1$, $s_2$, $s_3$, \&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). \rule{3em}{0em} \begin{minipage}{0.25\linewidth}{ \[ \arraycolsep=0.7em \begin{array}{|l@{}l|} \multicolumn{2}{c}{E_0(x)}\\[0.5em] \hline K_0(x)&=1\rule{0em}{1.3em}\\[0.5em] L_0(x)&=0\\[0.5em] M_0(x)&=0\\[0.5em] N_0(x)&=0\\[0.5em] \hline \end{array} \] } \end{minipage} \begin{minipage}{0.6\linewidth}{ \[ \begin{array}{|l@{}l|} \multicolumn{2}{c}{E_1(x)}\\[0.5em] \hline K_1(x)&=x\rule{0em}{1.3em}\\[0.5em] L_1(x)&=\sqrt{\strxx x^2-b^2}\\[0.5em] M_1(x)&=\sqrt{\strxx x^2-c^2}\\[0.5em] N_1(x)&=0\\[0.5em] \hline \end{array}\rule{3.7em}{0em} \] } \end{minipage} \[ \begin{array}{|l@{}l|} \multicolumn{2}{c}{E_2(x)}\\[0.5em] \hline K_2^{p_1}(x)&=x^2 - \frac{1}{3} [b^2 + c^2 - \sqrt{(b^2 + c^2)^2-3b^2c^2}]\rule{0em}{1.3em}\\[0.5em] K_2^{p_2}(x)&=x^2 - \frac{1}{3} [b^2 + c^2 + \sqrt{(b^2 + c^2)^2-3b^2c^2}]\\[0.5em] L_2(x)&=x \sqrt{\strxx x^2-b^2}\\[0.5em] M_2(x)&=x \sqrt{\strxx x^2-c^2}\\[0.5em] N_2(x)&=\sqrt{(x^2-b^2)(x^2-c^2)}\\[0.5em] \hline \end{array} \] \[ \begin{array}{|l@{}l|} \multicolumn{2}{c}{E_3(x)}\\[0.5em] \hline K_3^{p_1}(x)&=x^3 - \dfrac{x}{5} [2(b^2+c^2) - \sqrt{4(b^2+c^2)^2-15b^2c^2}] \rule{0em}{1.3em}\\[0.9em] K_3^{p_2}(x)&=x^3 -\dfrac{x}{5} [2(b^2+c^2) + \sqrt{4(b^2+c^2)^2-15b^2c^2}]\\[0.9em] L_3^{q_1}(x)&=\sqrt{\strxx x^2-b^2 }[x^2-\frac{1}{5} (b^2+2c^2 - \sqrt{(b^2+2c^2)^2-5b^2c^2})]\\[0.7em] L_3^{q_2}(x)&=\sqrt{\strxx x^2-b^2} [x^2-\frac{1}{5} (b^2+2c^2 + \sqrt{(b^2+2c^2)^2-5b^2c^2})]\\[0.7em] M_3^{r_1}(x)&=\sqrt{\strxx x^2-c^2} [x^2-\frac{1}{5} (2b^2+c^2 - \sqrt{(2b^2+c^2)^2-5b^2c^2})]\\[0.7em] M_3^{r_2}(x)&=\sqrt{\strxx x^2-c^2} [x^2-\frac{1}{5} (2b^2+c^2 + \sqrt{(2b^2+c^2)^2-5b^2c^2})]\\[0.7em] N_3(x)&=x \sqrt{(x^2-b^2)(x^2-c^2)}\\[0.5em] \hline \end{array} \] \vspace{\baselineskip} % -----File: 268.png It is to be noted that since in the solution (4) of Laplace's Equation, \[ V = E_m^p(\lambda) E_m^p(\mu) E_m^p(\nu), \] 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, \[ K_m^{p_k}(\lambda)K_m^{p_k}(\mu)K_m^{p_k}(\nu),\quad L_m^{q_k}(\lambda)L_m^{q_k}(\mu)L_m^{q_k}(\nu),\quad \text{\&c.;} \] and that in a solution of the form \[ V=\sum A_{m,p} E_m^p(\lambda) E_m^p(\mu) E_m^p(\nu) \] we shall have for a given $m$ just $2m+1$ terms. \mypara{138.} From the particular solution of Lamé's Equation [(5) Art.~137] $z=E_m^p(x)$, we can get by formula (5), Art.~18, the general solution. \begin{flalign*} &\indent\text{It is} & z=AE_m^p(x)+BE_m^p(x)\int\limits_x^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}[E_m^p(x)]^2}. & \tag{1}& \end{flalign*} Making $A=0$ and $B=2m+1$ we get a second form of particular solution of Lamé's Equation, $z=F_m^p(x)$ where \[ F_m^p(x)=(2m+1)E_m^p(x)\int\limits_x^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}[E_m^p(x)]^2}. \tag{2} \] We shall call $F_m^p(x)$ a \emph{Lamé's Function of the second kind}. It is easily seen to approach the value zero as $x$ is indefinitely increased. \EXAMPLE{S} 1.\quad If an ellipsoidal conductor is charged with electricity, and is found to be at potential $V_0$, show that since $V_0=V_0K_0(\lambda)$, \[ V=V_0K_0(\lambda)K_0(\mu)K_0(\nu)=V_0 \] at an internal point, and \begin{multline*} V=V_0K_0(\mu)K_0(\nu)\bigg[K_0(\lambda)\int\limits_\lambda^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}[K_0(x)]^2}\\ ÷ K_0(\lambda_0)\int\limits_{\lambda_0}^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}[K_0(x)]^2}\bigg] \end{multline*} \[ =V_0\bigg[\int\limits_\lambda^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}} ÷ \int\limits_{\lambda_0}^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}}\bigg] =V_0\frac{F\Big(\dfrac{b}{c},\sin^{-1}\dfrac{c}{\lambda}\Big)}{F\Big(\dfrac{b}{c},\sin^{-1}\dfrac{c}{\lambda_0}\Big)}, \] % -----File: 269.png \begin{flalign*} &\text{whence} & V=V_0\dfrac{\Big(F\Big(\dfrac{b}{c},\dfrac{\pi}{2}\Big)-\alpha\Big)}{F\Big(\dfrac{b}{c},\dfrac{\pi}{2}\Big)-\alpha_0} & \tag*{\text{v.~(10) Art.~136.}}& \end{flalign*} 2.\quad 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 \begin{align*} (l^2-\mu^2)(l^2&-\nu^2)=\tfrac{1}{3}[3l^4-2(b^2+c^2)l^2+b^2c^2]K_0(\mu)K_0(\nu)\\ &+\tfrac{1}{2}\bigg[1+\frac{b^2+c^2-3l^2}{\sqrt{(b^2+c^2)^2-3b^2c^2}}\bigg]K_2^{p_1}(\mu)K_2^{p_1}(\nu)\\ &+\tfrac{1}{2}\bigg[1-\frac{b^2+c^2-3l^2}{\sqrt{(b^2+c^2)^2-3b^2c^2}}\bigg]K_2^{p_2}(\mu)K_2^{p_2}(\nu); \end{align*} and that \[ \int\limits_0^{\lambda_0}\tfrac{4}{3}\pi\rho\frac{3l^4-2(b^2+c^2)l^2+b^2c^2}{\sqrt{(l^2-b^2)(l^2-c^2)}}dl =\tfrac{4}{3}\pi\rho\lambda_0\sqrt{(\lambda_0^2-b^2)(\lambda_0^2-c^2)}=M \] where $M$ is the mass of the ellipsoid. % recast to fit page width \begin{flalign*} &\textit{Ans.} & V=M\bigg\{&\int\limits_\lambda^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}}-\frac{3}{2\sqrt{(b^2+c^2)^2-3b^2c^2}}\\ &&&\bigg[K_2^{p_1}(\mu)K_2^{p_1}(\nu)K_2^{p_1}(\lambda) \int\limits_\lambda^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}.(K_2^{p_1}(x))^2}\\ &&&-K_2^{p_2}(\mu)K_2^{p_2}(\nu)K_2^{p_2}(\lambda) \int\limits_\lambda^\infty\frac{dx}{\sqrt{(x^2-b^2)(x^2-c^2)}.(K_2^{p_2}(x))^2}\bigg]\bigg\}. \end{flalign*} \mypara{139.} If for the sake of brevity we represent $\dfrac{b}{c}$ by $k$, and $\Big(1-\dfrac{b^2}{c^2}\Big)^\frac{1}{2}$ by $k'$ in the formulas (11) Art.~136 we have \[ \lambda=c\,\frac{\dn\alpha}{\cn\alpha}(\bmod k),\quad \mu=\frac{b}{\dn\beta(\bmod k')},\quad \nu=b\sn\gamma(\bmod k) \tag{1} \] and from these we get without difficulty (v.\ Int.\ Cal.\ Art.~192) \begin{equation*}\left. \begin{aligned} \sqrt{\lambda^2-b^2}&=\frac{ck'}{\cn\alpha(\bmod k)}, &\sqrt{\mu^2-b^2} &=\frac{bk'\sn\beta}{\dn\beta}(\bmod k'),\;\\ \sqrt{b^2-\nu^2} &=b\cn\gamma(\bmod k), &\sqrt{\lambda^2-c^2}&=\frac{ck'\sn\alpha}{\cn\alpha}(\bmod k),\\ \sqrt{c^2-\mu^2} &=\frac{ck'\cn\beta}{\dn\beta}(\bmod k'), &\sqrt{c^2-\nu^2} &=c\dn\gamma(\bmod k). \end{aligned}\right\} \tag{2} \end{equation*} % -----File: 270.png If we let $\alpha$ range from 0 to $K$, and $\beta$ from 0 to $2K'$, and $\gamma$ from 0 to $4K$, where $K$ and $K'$ are the complete Elliptic Integrals $F\Big(k,\dfrac{\pi}{2}\Big)$ and $F\Big(k',\dfrac{\pi}{2}\Big)$ respectively, $(\alpha, \beta, \gamma)$ may represent any point in space, and there will be no ambiguity in sign (v.\ Art.~136). We may note that if $0<\beta\mu^2>b^2>\nu^2$. Show that \[ x^2=\frac{r^2\mu^2\nu^2}{b^2c^2},\quad y^2=\frac{r^2(\mu^2-b^2)(\nu^2-b^2)}{b^2(b^2-c^2)},\quad z^2=\frac{r^2(\mu^2-c^2)(\nu^2-c^2)}{c^2(c^2-b^2)}; \] \[ h_1^2=\frac{(\mu^2-b^2)(c^2-\mu^2)}{r^2(\mu^2-\nu^2)},\quad h_2^2=\frac{(b^2-\nu^2)(c^2-\nu^2)}{r^2(\mu^2-\nu^2)},\quad h_3^2=1. \] Laplace's Equation is \[ D^2_\alpha V + D^2_\beta V +(\mu^2-\nu^2)D_r(r^2 D_r V) = 0 \tag{2} \] where \hspace{\stretch{1}}$\displaystyle\alpha=\int\limits_b^\mu\frac{d\mu}{\sqrt{(\mu^2-b^2)(c^2-\mu^2)}}\quad \text{and}\quad \beta = \int\limits_0^\nu\frac{d\nu}{\sqrt{(b^2-\nu^2)(c^2-\nu^2)}},$\hspace{\stretch{1}} If $V=U.R$ (2) breaks up into \begin{flalign*} &&\frac{d}{dr}\bigg(r^2&\frac{dR}{dr}\bigg)=m(m+1)R, \tag{3}\\ &&D^2_\alpha U + D^2_\beta U &+ m(m+1)(\mu^2-\nu^2)U=0. \tag{4}\\[1ex] &\text{(3) gives} & R&=Ar^m+Br^{-m-1}. &&\\[1ex] &\text{(4) gives} && U=E^p_m(\mu)E^p_m(\nu) & \tag*{(v.\ Art.~142).}& \end{flalign*} So that a solution of (2) is \[ V=Ar^m E^p_m(\mu)E^p_m(\nu). \] But since (2) is Laplace's Equation, $V=Ar^mY_m(\mu,\phi)$, if expressed in Conical Coördinates, must satisfy it, consequently $E^p_m(\mu)E^p_m(\nu)$ must be simply a Spherical Harmonic of the $m$th degree. % -----File: 274.png \vspace{1ex}\Needspace*{5\baselineskip} \begin{center} \emph{Toroidal Coördinates.} \end{center}\vspace{-1ex} \mypara{143.} Any pair of circles belonging to the orthogonal system obtained and figured in Art.~46 can be represented by the equations \[ \left. \begin{aligned} \dfrac{2ax}{\sinh \alpha} &= \dfrac{x^2 + y^2 + a^2}{ \cosh \alpha}\;\\ \dfrac{2ay }{\sin \beta} &= \dfrac{x^2 + y^2 - a^2 }{\cos \beta} \end{aligned} \right\} \tag{1} \] 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 \emph{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 \[ \left. \begin{aligned} \dfrac{4a^2(x^2+y^2) }{\sinh^2 \alpha} &= \dfrac{[x^2+y^2+z^2+a^2]^2}{\cosh^2 \alpha}\;\\ \dfrac{2az}{\sin \beta} &= \dfrac{x^2+y^2+z^2-a^2}{\cos \beta}\\ y &= x \tan\gamma \end{aligned} \right\} \tag{2} \] $\alpha$, $\beta$, and $\gamma$ being regarded as the coördinates of a point of intersection of the three surfaces. Finding Laplace's Equation in the usual manner we get \begin{gather*} x= \dfrac{a \sinh\alpha \cos\gamma}{\cosh\alpha \mp \cos\beta}, \quad y=\dfrac{a \sinh\alpha \sin\gamma}{\cosh\alpha \mp \cos\beta},\quad z=\dfrac{a \sin\beta}{\cosh\alpha \mp \cos\beta},\\[0.5em] r=\sqrt{x^2+y^2}=\dfrac{a \sinh\alpha}{\cosh\alpha \mp \cos\beta},\quad a+z \ctn\beta=\dfrac{a \cosh\alpha}{\cosh\alpha \mp \cos\beta};\\[0.5em] h_1=\dfrac{\cosh\alpha \mp \cos\beta}{a},\quad h_2=\dfrac{\cosh\alpha \mp \cos\beta}{a},\quad h_3=\dfrac{\cosh\alpha \mp \cos\beta}{a \sinh\alpha}; \end{gather*} and Laplace's Equation becomes \begin{align*} D_\alpha\left[\dfrac{a \sinh\alpha}{\cosh\alpha \mp \cos\beta} D_\alpha V\right]&+D_\beta\left[\dfrac{a \sinh\alpha}{\cosh\alpha \mp \cos\beta} D_\beta V\right]\\[0.5em] &+D_\gamma\left[\dfrac{a}{\sinh\alpha(\cosh\alpha \mp \cos\beta)} D_\gamma V\right] = 0, \tag{1} \end{align*} % -----File: 275.png \begin{flalign*} &\text{or}& D_\alpha(rD_\alpha V) + D_\beta(rD_\beta V) + \dfrac{1}{\sinh^2\alpha} rD_\gamma^2 V &= 0.& \tag{2} \end{flalign*} We cannot proceed further by our usual method, for the assumption that $V$ is a function of $\alpha$ alone, or that $V$ is a function of $\beta$ alone, proves to be inadmissible. Indeed, not only are $\alpha$, $\beta$, and $\gamma$ not \textit{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 \[ D_\alpha^2(V\sqrt{r}) + D_\beta^2(V\sqrt{r})+ \dfrac{1}{\sinh^2\alpha} D_\gamma^2(V\sqrt{r})-V(D_\alpha^2 \sqrt{r}+D_\beta^2\sqrt{r})=0. \tag{3} \] $D_\alpha^2\sqrt{r}+D_\beta^2\sqrt{r}$ proves equal to $-\dfrac{\sqrt{r}}{4\sinh^2\alpha}$; hence if $U = V\sqrt{r}$ (3) becomes \[ \sinh^2\alpha(D_\alpha^2U+D_\beta^2U) + D_\gamma^2 U + \tfrac{1}{4} U=0, \tag{4} \] for which particular solutions can readily be found by our usual process. (4) can be broken up into the three equations \begin{gather*} \dfrac{d^2N}{d\gamma^2} + (m+\tfrac{1}{2})^2N = 0 \tag{5}\\ \dfrac{d^2M}{d\beta^2} + n^2M = 0 \tag{6}\\ \sinh^2\alpha \dfrac{d^2L}{d\alpha^2} - [m(m+1)+n^2\sinh^2\alpha]L=0. \tag{7}\\ N = A \cos(m+\tfrac{1}{2})\gamma + B \sin(m+\tfrac{1}{2})\gamma\\[1ex] M = A_1 \cos n\beta + B_1 \sin n\beta. \end{gather*} If we introduce into (7) $x = \ctnh \alpha$ it becomes \[ (1-x^2) \dfrac{d^2L}{dx^2} - 2x\dfrac{ dL}{dx} + \left[m(m+1) - \dfrac{n^2}{1-x^2}\right]L=0, \] a solution of which is \begin{flalign*} &&L &= P_m^n(x) = (1-x^2)^{\frac{n}{2}} \dfrac{d^nP_m(x)}{dx^n} &\text{(v.\ Art.~102).} \end{flalign*} It is to be noted that since $\ctnh \alpha$ is greater than 1 \[ P_m^n(\ctnh \alpha) = i^{\frac{n}{2}}\csch^n \alpha \dfrac{d^n P_m(\ctnh \alpha)}{(d \ctnh \alpha)^n}. \] % -----File: 276.png The constant coefficient $i^{\frac{n}{2}}$ can be rejected and we get \[ U=[A \cos(m+\tfrac{1}{2})\gamma + B \sin(m+\tfrac{1}{2})\gamma](A_1 \cos n\beta + B_1 \sin n\beta)\csch^n \alpha \dfrac{d^nP_m(\ctnh \alpha)}{(d \ctnh \alpha)^n} \] as a particular solution of (4). \[ \dfrac{1}{i^{\frac{n}{2}}} P_m^n(\ctnh \alpha)=\csch^n \alpha \dfrac{d^nP_m(\ctnh \alpha)}{(d \ctnh\alpha)^n} \] has been called a Toroidal Harmonic. \EXAMPLE{S} 1.\quad 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. \emph{Suggestion}: If $V = f(\beta,\gamma)$ when $\alpha = \alpha_0$, the function to be developed is \[ \sqrt{r}.f(\beta,\gamma)\quad \text{\textit{i.e}.}\quad \left[\dfrac{ a \sinh \alpha_0}{ \cosh \alpha_0\mp \cos \beta} \right]^{\frac{1}{2}} f(\beta,\gamma) \] and the development will be in a double Fourier's Series (v.\ Art.~71).\\ 2.\quad Show that if we let $\alpha$ range from 0 to $\infty$, $\beta$ from $-\pi$ to $\pi$, and $\gamma$ from 0 to $2\pi$, each of the double signs on page 264 may be replaced by the minus sign without loss of generality. \label{ch8end} % -----File: 277.png \mychap{CHAPTER IX.\footnotemark}{HISTORICAL SUMMARY.} \label{ch9start} \footnotetext{See preface.}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 1753\footnote{See two articles by Bernoulli and one by Euler in the Memoirs of the Academy of Berlin for this year.} that Daniel Bernoulli enunciated the principle of the coexistence of small oscillations, which, in connection with Taylor'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.\footnote{See the Transactions of the Academy of St.\ Petersburg for 1732-33, 1734 and 1781.} 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 vibration of rectangular and circular membranes, which were discussed by Euler\footnote{Transactions of the Academy of St.\ Petersburg.} 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 successful 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 $P_m$ 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 written\footnote{``Théorie des attractions des sphéroïdes et de la figure des Planètes'' Mémoires de l'académie des sciences 1782. This article, although bearing an earlier date than that of Legendre, was really inspired by it. It is here that ``Laplace's equation'' first appears, occurring, however, only in polar coördinates.} determined the potential of a solid differing but little from a sphere by means of the development according to the spherical harmonics $Y_m$. % -----File: 278.png Very closely related to this problem is Gauss's celebrated treatment of the theory of terrestrial magnetism,\footnote{Resultate aus den Beobachtungen des magnetischen Vereins im Jahre 1838. Leipzig, 1839. Reprinted in Gauss's collected works, Vol.\ V., p.\ 121.} 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: \textit{Théorie du mouvement de la chaleur dans les corps solides} was crowned by the French Academy. Although not printed until the years 1824-26,\footnote{Mémoires de l'académie des sciences for 1819-20 and 1821-22.} 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 \textit{Théorie analytique de la chaleur} (1822), contains a treatment of the following problems and of practically all of their special cases: (\emph{a}) The one dimensional flow of heat. (\emph{b}) The two dimensional flow of heat in a rectangle. (\emph{c}) The three dimensional flow of heat in a rectangular parallelopiped. (\emph{d}) The flow of heat in a sphere when the temperature depends only on the distance from the centre. (\emph{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." \markright{HISTORICAL SUMMARY.} 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 posthumous paper Euler had given the formulae for determining the coefficients,\footnote{Lagrange had practically determined these coefficients long before but failed to notice what he had got.} 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 1820\footnote{Connaissance des Temps pour l'an 1823.} to the general (unsymmetrical) case of the flow of heat in a sphere, and by Poisson\footnote{Journal de l'École Polytechnique, 19\textsuperscript{e} Cahier. Although the final forms to which Poisson reduces his results are similar to Fourier's, his methods are very different.} (1821) to the unsymmetrical flow of heat in a cylinder. % -----File: 279.png In 1835 Green published a paper\footnote{``On the determination of the exterior and interior attraction of ellipsoids of variable densities.'' Transactions of the Cambridge Philosophical Society.} 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 these\footnote{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.} the equation $\nabla^2V=0$ is transformed to ellipsoidal coördinates and is then broken up into three ordinary differential equations. The rest of the solution, however, is hardly touched upon. Lamé's most important work on this subject\footnote{ ``Sur l'équilibre des Températures dans un ellipsoïde à trois axes inégaux.'' An article by the same author on the two dimensional potential will be found in Vol.\ I. of this Journal.} 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 $\theta$ and $\phi$, 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 dissertation\footnote{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.} (1842) determined 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 simplified 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 afterwards\footnote{\textit{Heine:} Crelle's Journal, Vol.\ 29, 1845. \textit{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 of a development of $\dfrac{1}{r}$ in terms of Lamé's functions, see a paper by Heine in Crelle's Journal, Vol.\ 42, 1851.} Heine and Liouville % -----File: 280.png 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 \emph{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 memoirs\footnote{See especially the one in the Mémoires de l'académie des sciences, Vol.\ VIII., 1829.} 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. He was, nevertheless, able to solve the problem of the \emph{symmetrical} vibration of a free circular plate. The complete theory of the vibration of a free circular plate was first given by Kirchhoff.\footnote{Crelle's Journal, Vol.\ 40, 1850.} 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 investigations 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)\footnote{Crelle's Journal, Vol.\ 88.} and Lamb (1882).\footnote{Proc.\ Lond.\ Math.\ Soc.} 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,\footnote{First edition, 1867. This appendix was evidently written as early as 1862, as Thomson refers to it in the memoir quoted above.} to which we have already referred, deserves to be regarded as one of the most important % -----File: 281.png contributions to the general theory. The way in which spherical harmonics are introduced (as homogeneous functions of the rectangular coördinates) was then new\footnote{The same method was used at about the same time by Clebsch.} 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.\footnote{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.} Returning now to the Continent, we find as the next important question taken up the problem of the potential of an anchor ring. The first publication on this subject is a monograph by C. Neumann\footnote{``Theorie der Elektricitäts- und Wärme-Vertheilung in einem Ringe.'' Halle.} (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 (\emph{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 papers\footnote{Crelle's Journal, Vol.\ 68, 1868.} (1868) deals with the solid bounded by two intersecting spheres, while in the second\footnote{Jahresbericht des Gymnasiums zu Elbing.} (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 % -----File: 282.png 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 Mathieu\footnote{Liouville's Journal, Vol.\ XIII.} 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,\footnote{``Ueber die Integration der partiellen Differentialgleichung $\frac{\delta^2 u}{\delta x^2} + \frac{\delta^2 u}{\delta y^2} + k^2 u=0.$'' Math.\ Ann., Vol.\ I\@. No \emph{physical} problem is mentioned in this paper.} 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 ellipsoids 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 of problems here considered by treating in succession the potential problem for the paraboloid of revolution,\footnote{``Ueber das Gleichgewicht und die Bewegung der Wärme in einem Rotationsparaboloid.'' Dissertation, Halle, 1881.} the parabolic cylinder\footnote{``Die Funktion des parabolischen Cylinders,'' Gymnasialprogramm Cüstrin, 1883.} and the general paraboloid.\footnote{``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.} 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 % -----File: 283.png degree. Wangerin\footnote{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.} (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, \textit{cyclidic} coördinates.\footnote{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.} He showed, however, merely how to break up Laplace's equation into three ordinary differential equations.\footnote{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.} 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.\footnote{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.} 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''\footnote{Math.\ Ann., 18.} (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.\footnote{For an exposition of this theory see the treatise: Ueber die Reihenentwickelungen der Potentialtheorie, Leipsic, Teubner, 1894, by the writer of the present chapter.} Of late years, especially since the year 1880, the younger English mathematicians 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 \emph{development} of the method; it is rather an application of it to 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. \textit{Niven:} On the Conduction of Heat in Ellipsoids of Revolution. Phil.\ Trans., 1880. \textit{Niven:} On the Induction of Electric Currents in Infinite Plates and Spherical Shells.\ Phil.\ Trans., 1881. % -----File: 284.png \emph{Hicks:} On Toroidal Functions. Phil.\ Trans., 1881. \emph{Hicks:} On the Steady Motion and Small Vibrations of a Hollow Vortex. Phil.\ Trans., 1884, 1885. \emph{Lamb:} On Ellipsoidal Current Sheets. Phil.\ Trans., 1887. \emph{Chree:} The Equations of an Isotropic Elastic Solid in Polar and Cylindrical Coördinates, their Solution and Application. Camb.\ Phil.\ Soc.\ Trans., XIV., 1889. \emph{Hobson:} On a Class of Spherical Harmonics of Complex Degree with Applications to Physical Problems. Camb.\ Phil.\ Soc.\ Trans., XIV., 1889. \emph{Chree:} On some Compound Vibrating Systems. Camb.\ Phil.\ Soc.\ Trans., XV., 1891. \emph{Niven:} On Ellipsoidal Harmonics. Phil.\ Trans., 1892. The historical sketch we have just given would naturally require as a supplement 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, however, 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: \emph{Fourier:} Théorie Analytique de la Chaleur, 1822. \emph{Lamé:} Leçons sur les Functions inverses des Transcendantes et les Surfaces isothermes, 1857. \emph{Lamé:} Leçons sur les Coordonnées Curvilignes et leurs diverses Applications, 1859. \emph{Mathieu:} Cours de Physique Mathématique, 1873. \emph{Riemann:} Partielle Differentialgleichungen, und deren Anwendung auf phys\-ikalische Fragen (edited by Hattendorf), third edition. 1882. \emph{F.~Neumann:} Theorie des Potentials und der Kugelfunktionen (edited by C. Neumann), 1887. \emph{Thomson and Tait:} Natural Philosophy, second edition, 1879. \emph{Rayleigh:} Theory of Sound, 1877. \emph{Basset:} Hydrodynamics, 1888. \emph{Love:} Theory of Elasticity, 1892. % -----File: 285.png \emph{Heine:} Handbuch der Kugelfunktionen (second edition), 1878-81. \emph{Ferrers:} Spherical Harmonics, 1881. \emph{Haentzschel:} Reduction der Potentialgleichung auf gewöhnliche Differentialgleichungen, 1893. These last three books would also belong in the following list of books relating to the theory of the various functions we use: \emph{Todhunter:} The Functions of Laplace, Lamé and Bessel, 1875. \emph{Lommel:} Studien über die Bessel'schen Funktionen, 1868. \emph{F.~Neumann:} Beiträge zur Theorie der Kugelfunktionen, 1878. And finally concerning the question of convergence: \emph{C.~Neumann:} Über die nach Kreis-, Kugel- und Cylinder-Functionen fort\-schreitenden Entwickelungen, 1881. \label{ch9end} % -----File: 286.png %[Blank Page] % -----File: 287.png \label{tablestart} \setcounter{footnote}{0} \newpage \medskip\begin{center}{\Large APPENDIX.}\par\rule{5em}{0.5pt}\par\bigskip{TABLES.}\end{center} \markright{} Table I., a table of Surface Zonal Harmonics (Legendrians), gives the values of the first seven Harmonics $P_1(\cos \theta)$, $P_2(\cos \theta)$, $\cdots P_7(\cos \theta)$ for the argument $\theta$ 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 $P_1(x)$, $P_2(x)$, $\cdots P_7(x)$ 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 $e^x$, $e^{-x}$, $\sinh x$, $\cosh x$, 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 \[ \log \sinh x = \log \tan (\gd x)\qquad \log \cosh x = \log \sec (\gd x). \] Table IV.\ gives the first twelve roots of $J_0(x) = 0$ and $J_1(x) = 0$ each divided by $\pi$. 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 $J_0(x) = 0$, $J_1(x) = 0$, $\cdots J_5(x)=0$. 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 $J_0(x)$ and $J_1(x)$ 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. % -----File: 288.png \newpage \markright{APPENDIX} \label{tableI} \begin{center} TABLE I. --- \textsc{Surface Zonal Harmonics} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|c|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut\theta}& \multicolumn{1}{c|}{P_1(\cos\theta)}& \multicolumn{1}{c|}{P_2(\cos\theta)}& \multicolumn{1}{c|}{P_3(\cos\theta)}& \multicolumn{1}{c|}{P_4(\cos\theta)}& \multicolumn{1}{c|}{P_5(\cos\theta)}& \multicolumn{1}{c|}{P_6(\cos\theta)}& \multicolumn{1}{c|}{P_7(\cos\theta)}\\ \hline \vrule width0ptdepth0ptheight3.5ex \phantom{\text{°}0}0\text{°} & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000 \\ \phantom{0}1 & .9998 & .9995 & .9991 & .9985 & .9977 & .9967 & .9955 \\ \phantom{0}2 & .9994 & .9982 & .9963 & .9939 & .9909 & .9872 & .9829 \\ \phantom{0}3 & .9986 & .9959 & .9918 & .9863 & .9795 & .9713 & .9617 \\ \phantom{0}4 & .9976 & .9927 & .9854 & .9758 & .9638 & .9495 & .9329 \\[2ex] \phantom{0}5 & .9962 & .9886 & .9773 & .9623 & .9437 & .9216 & .8961 \\ \phantom{0}6 & .9945 & .9836 & .9674 & .9459 & .9194 & .8881 & .8522 \\ \phantom{0}7 & .9925 & .9777 & .9557 & .9267 & .8911 & .8476 & .7986 \\ \phantom{0}8 & .9903 & .9709 & .9423 & .9048 & .8589 & .8053 & .7448 \\ \phantom{0}9 & .9877 & .9633 & .9273 & .8803 & .8232 & .7571 & .6831 \\[2ex] 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 \\[2ex] 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 \\[2ex] 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 \\[2ex] 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 \\[2ex] 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 \\[2ex] 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 \\[2ex] 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 \\[2ex] \phantom{\text{°}}45\text{°} & .7071 & .2500 & -.1768 & -.4062 & -.3757 & -.1485 & .1270 \\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 289.png \begin{center} TABLE I. --- \textsc{Surface Zonal Harmonics} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|c|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut\theta}& \multicolumn{1}{c|}{P_1(\cos\theta)}& \multicolumn{1}{c|}{P_2(\cos\theta)}& \multicolumn{1}{c|}{P_3(\cos\theta)}& \multicolumn{1}{c|}{P_4(\cos\theta)}& \multicolumn{1}{c|}{P_5(\cos\theta)}& \multicolumn{1}{c|}{P_6(\cos\theta)}& \multicolumn{1}{c|}{P_7(\cos\theta)}\\ \hline \vrule width0ptdepth0ptheight3.5ex \phantom{\text{°}}45\text{°} & .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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] \phantom{\text{°}}90\text{°} & .0000 & -.5000 & .0000 & .3750 & .0000 & -.3125 & .0000\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 290.png \label{tableII} \begin{center} TABLE II.---\textsc{Surface Zonal Harmonics.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& \multicolumn{1}{c|}{P_1(x)}& \multicolumn{1}{c|}{P_2(x)}& \multicolumn{1}{c|}{P_3(x)}& \multicolumn{1}{c|}{P_4(x)}& \multicolumn{1}{c|}{P_5(x)}& \multicolumn{1}{c|}{P_6(x)}& \multicolumn{1}{c|}{P_7(x)}\\ \hline \vrule width0ptdepth0ptheight3.5ex 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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .50 & .5000 & -.1250 & -.4375 & -.2891 & .0898 & .3232 & .2231\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 291.png \begin{center} TABLE II.---\textsc{Surface Zonal Harmonics.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& \multicolumn{1}{c|}{P_1(x)}& \multicolumn{1}{c|}{P_2(x)}& \multicolumn{1}{c|}{P_3(x)}& \multicolumn{1}{c|}{P_4(x)}& \multicolumn{1}{c|}{P_5(x)}& \multicolumn{1}{c|}{P_6(x)}& \multicolumn{1}{c|}{P_7(x)}\\ \hline \vrule width0ptdepth0ptheight3.5ex .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] 1.00 & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000 & 1.0000\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 292.png \label{tableIII} \begin{center} TABLE III.---\textsc{Hyperbolic Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& \multicolumn{1}{c|}{e^x}& \multicolumn{1}{c|}{e^{-x}}& \multicolumn{1}{c|}{\sinh x}& \multicolumn{1}{c|}{\cosh x}& \multicolumn{1}{c|}{\gd x}\\ \hline \vrule width0ptdepth0ptheight3.5ex \;0.00 & \;1.0000 & \;1.0000 & \;0.0000 & \;1.0000 & \;0\raisebox{0.8ex}{\text{°}}\!\!\!.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\\[2ex] .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\\[2ex] .10 & 1.1052 & .9048 & .1002 & 1.0050 & 5.720\phantom{0}\\ .11 & 1.1163 & .8958 & .1102 & 1.0061 & 6.290\phantom{0}\\ .12 & 1.1275 & .8869 & .1203 & 1.0072 & 6.859\phantom{0}\\ .13 & 1.1388 & .8781 & .1304 & 1.0085 & 7.428\phantom{0}\\ .14 & 1.1503 & .8694 & .1405 & 1.0098 & 7.995\phantom{0}\\[2ex] .15 & 1.1618 & .8607 & .1506 & 1.0113 & 8.562\phantom{0}\\ .16 & 1.1735 & .8521 & .1607 & 1.0128 & 9.128\phantom{0}\\ .17 & 1.1853 & .8437 & .1708 & 1.0145 & 9.694\phantom{0}\\ .18 & 1.1972 & .8353 & .1810 & 1.0162 & 10.258\phantom{0}\\ .19 & 1.2092 & .8270 & .1911 & 1.0181 & 10.821\phantom{0}\\[2ex] .20 & 1.2214 & .8187 & .2013 & 1.0201 & 11.384\phantom{0}\\ .21 & 1.2337 & .8106 & .2115 & 1.0221 & 11.945\phantom{0}\\ .22 & 1.2461 & .8025 & .2218 & 1.0243 & 12.505\phantom{0}\\ .23 & 1.2586 & .7945 & .2320 & 1.0266 & 13.063\phantom{0}\\ .24 & 1.2712 & .7866 & .2423 & 1.0289 & 13.621\phantom{0}\\[2ex] .25 & 1.2840 & .7788 & .2526 & 1.0314 & 14.177\phantom{0}\\ .26 & 1.2969 & .7711 & .2629 & 1.0340 & 14.732\phantom{0}\\ .27 & 1.3100 & .7634 & .2733 & 1.0367 & 15.285\phantom{0}\\ .28 & 1.3231 & .7558 & .2837 & 1.0395 & 15.837\phantom{0}\\ .29 & 1.3364 & .7483 & .2941 & 1.0423 & 16.388\phantom{0}\\[2ex] .30 & 1.3499 & .7408 & .3045 & 1.0453 & 16.937\phantom{0}\\ .31 & 1.3634 & .7334 & .3150 & 1.0484 & 17.484\phantom{0}\\ .32 & 1.3771 & .7261 & .3255 & 1.0516 & 18.030\phantom{0}\\ .33 & 1.3910 & .7189 & .3360 & 1.0549 & 18.573\phantom{0}\\ .34 & 1.4049 & .7118 & .3466 & 1.0584 & 19.116\phantom{0}\\[2ex] .35 & 1.4191 & .7047 & .3572 & 1.0619 & 19.656\phantom{0}\\ .36 & 1.4333 & .6977 & .3678 & 1.0655 & 20.195\phantom{0}\\ .37 & 1.4477 & .6907 & .3785 & 1.0692 & 20.732\phantom{0}\\ .38 & 1.4623 & .6839 & .3892 & 1.0731 & 21.267\phantom{0}\\ .39 & 1.4770 & .6771 & .4000 & 1.0770 & 21.800\phantom{0}\\[2ex] .40 & 1.4918 & .6703 & .4108 & 1.0811 & 22.331\phantom{0}\\ .41 & 1.5068 & .6636 & .4216 & 1.0852 & 22.859\phantom{0}\\ .42 & 1.5220 & .6570 & .4325 & 1.0895 & 23.386\phantom{0}\\ .43 & 1.5373 & .6505 & .4434 & 1.0939 & 23.911\phantom{0}\\ .44 & 1.5527 & .6440 & .4543 & 1.0984 & 24.434\phantom{0}\\[2ex] .45 & 1.5683 & .6376 & .4653 & 1.1030 & 24.955\phantom{0}\\ .46 & 1.5841 & .6313 & .4764 & 1.1077 & 25.473\phantom{0}\\ .47 & 1.6000 & .6250 & .4875 & 1.1125 & 25.989\phantom{0}\\ .48 & 1.6161 & .6188 & .4986 & 1.1174 & 26.503\phantom{0}\\ .49 & 1.6323 & .6126 & .5098 & 1.1225 & 27.015\phantom{0}\\[2ex] 0.50 & 1.6487 & 0.6065 & 0.5211 & 1.1276 & 27\raisebox{0.8ex}{\text{°}}\!\!\!.524\phantom{0}\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 293.png \begin{center} TABLE III.---\textsc{Hyperbolic Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& \multicolumn{1}{c|}{e^x}& \multicolumn{1}{c|}{e^{-x}}& \multicolumn{1}{c|}{\sinh x}& \multicolumn{1}{c|}{\cosh x}& \multicolumn{1}{c|}{\gd x}\\ \hline \vrule width0ptdepth0ptheight3.5ex \;0.50 & \;1.6487 & \;0.6065 & \;0.5211 & \;1.1276 & \;27\raisebox{0.8ex}{\text{°}}\!\!\!.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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] .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\\[2ex] 1.00 & 2.7183 & 0.3679 & 1.1752 & 1.5431 & 49\raisebox{0.8ex}{\text{°}}\!\!\!.605\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 294.png \begin{center} TABLE III.---\textsc{Hyperbolic Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& l\sinh x& l\cosh x& \multicolumn{1}{c|}{x}& l\sinh x& l\cosh x& \multicolumn{1}{c|}{x}& l\sinh x& l\cosh x\\ \hline \vrule width0ptdepth0ptheight3.5ex \;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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 1.50 & 0.3282 & 0.3715 & 2.00 & 0.5595 & 0.5754 & 2.50 & 0.7818 & 0.7876\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 295.png \begin{center} TABLE III.---\textsc{Hyperbolic Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|}{\tablestrut x}& l\sinh x& l\cosh x& \multicolumn{1}{c|}{x}& l\sinh x& l\cosh x& \multicolumn{1}{c|}{x}& l\sinh x& l\cosh x\\ \hline \vrule width0ptdepth0ptheight3.5ex \;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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 2.75 & 0.8915 & 0.8951 & 3.00 & 1.0008 & 1.0029 & 10.0 & 4.0419 & 4.0419\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 296.png \begin{center} \label{tableIV}TABLE IV.---\textsc{Roots of Bessel's Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|c|c|r@{\quad}|c|c|} \hline \tablestrut{}& \dfrac{x}{\pi}$ for $J_0(x)=0& \dfrac{x}{\pi}$ for $J_1(x)=0& \qquad& \dfrac{x}{\pi}$ for $J_0(x)=0& \dfrac{x}{\pi}$ for $J_1(x)=0\\ \hline \vrule width0ptdepth0ptheight3.5ex 1 & 0.7655 & 1.2197 & 7 & \phantom{0}6.7519 & \phantom{0}7.2448\\ 2 & 1.7571 & 2.2330 & 8 & \phantom{0}7.7516 & \phantom{0}8.2454\\ 3 & 2.7546 & 3.2383 & 9 & \phantom{0}8.7514 & \phantom{0}9.2459\\ 4 & 3.7534 & 4.2411 & 10 & \phantom{0}9.7513 & 10.2463\\ 5 & 4.7527 & 5.2428 & 11 & 10.7512 & 11.2466\\ \;6 & 5.7522 & 6.2439 & \;12 & 11.7511 & 12.2469\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \bigskip \begin{center} \label{tableV}TABLE V.---\textsc{Roots of $J_n(x)=0$.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|c|c|c|c|c|c|} \hline \tablestrut{}& \;n=0\;& \;n=1\;& \;n=2\;& \;n=3\;& \;n=4\;& \;n=5\;\\ \hline \vrule width0ptdepth0ptheight3.5ex 1 & \phantom{0}2.405 & \phantom{0}3.832 & \phantom{0}5.135 & \phantom{0}6.379 & \phantom{0}7.586 & \phantom{0}8.780\\ 2 & \phantom{0}5.520 & \phantom{0}7.016 & \phantom{0}8.417 & \phantom{0}9.760 & 11.064 & 12.339\\ 3 & \phantom{0}8.654 & 10.173 & 11.620 & 13.017 & 14.373 & 15.700\\ 4 & 11.792 & 13.323 & 14.796 & 16.224 & 17.616 & 18.982\\ 5 & 14.931 & 16.470 & 17.960 & 19.410 & 20.827 & 22.220\\ 6 & 18.071 & 19.616 & 21.117 & 22.583 & 24.018 & 25.431\\ 7 & 21.212 & 22.760 & 24.270 & 25.749 & 27.200 & 28.628\\ 8 & 24.353 & 25.903 & 27.421 & 28.909 & 30.371 & 31.813\\ \;9 & 27.494 & 29.047 & 30.571 & 32.050 & 33.512 & 34.983\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \newpage % -----File: 297.png \label{tableVI} \begin{center} TABLE VI.--- \textsc{Bessel's Functions.} \end{center} \begin{scriptsize} \[ \arraycolsep=0.2em \begin{array}{|c|} \hline\\[-2.1ex] \arraycolsep=0.5em \begin{array}{|r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|||r@{\quad}|r@{\quad}|r@{\quad}|} \hline \multicolumn{1}{|c|} {\tablestrut x} & \multicolumn{1}{c|} {J_0(x)} & \multicolumn{1}{c|||}{J_1(x)} & \multicolumn{1}{c|} {x } & \multicolumn{1}{c|} {J_0(x)} & \multicolumn{1}{c|||}{J_1(x)} & \multicolumn{1}{c|} {x} & \multicolumn{1}{c|} {J_0(x)} & \multicolumn{1}{c|} {J_1(x)}\\ \hline \vrule width0ptdepth0ptheight3.5ex \;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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 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\\[2ex] 5.0 & -.1776 & -.3276 & 10.0 & -.2459 & .0435 & 15.0 & -.0142 & .2051\\[1ex] \hline \end{array}\\[-2.1ex] {}\\ \hline \end{array} \] \end{scriptsize} \label{tableend} \newpage \pagestyle{empty} % -----File: 298.png %[Blank Page] % -----File: 299.png \markright{}\vspace*{\fill} \begin{center} ANNOUNCEMENTS \end{center} \vspace*{\fill} \pagebreak % -----File: 300.png %[Blank Page] % -----File: 301.png \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \begin{LARGE} APPLICATIONS OF THE CALCULUS\\ TO MECHANICS\par \end{LARGE} \vspace{\baselineskip} By \textsc{E. R. Hedrick}, Professor of Mathematics in the University of Missouri,\\ and \textsc{O. D. Kellogg}, Assistant Professor of Mathematics\\ in the University of Missouri \rule{0.25\textwidth}{1pt} 8vo, cloth, 116 pages, with diagrams, \$1.25 \rule{0.25\textwidth}{1pt} \end{center} \noindent{\Huge 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 previously 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. \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \vspace{\baselineskip} \begin{LARGE} GINN~~AND~~COMPANY~~~\textsc{Publishers} \end{LARGE} \end{center} \newpage % -----File: 302.png \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \begin{LARGE} ELECTRICAL PROBLEMS\par \end{LARGE} \vspace{\baselineskip} By \textsc{William L. Hooper}, Professor of Electrical Engineering,\\ Tufts College, Mass., and \textsc{Roy T. Wells}, in Charge of\\ Electrical Engineering, State University of Iowa \rule{0.25\textwidth}{1pt} 8vo, cloth, 170 pages, with diagrams, \$1.25 \rule{0.25\textwidth}{1pt} \end{center} \noindent{\Huge 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:\medskip\par \begin{small} 1.\quad Twelve sets of problems and calculations on combinations of electro-motive forces and resistances in series and multiple grouping; distribution and fall of potential in railway and lighting circuits, inductance of coils, capacity of condensers, thermo-electricity, electro-chemistry; and output and efficiency of batteries, generators, motors, etc. 2.\quad Four sets of problems on combinations of alternating electromotive forces and currents and the impedance of circuits with constant and with varying values of resistance, inductance, capacity, and frequency. 3.\quad Five sets of problems on calculating and making winding tables and drawings for direct and alternating current armatures, armature reactions, field windings, etc. 4.\quad Problems on winding and operation of transformers, rotary converters, and induction motors, and on testing of dynamos and transmission of power. \end{small}\medskip\par 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. \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \vspace{\baselineskip} \begin{LARGE} GINN~~\&~~COMPANY~~~\textsc{Publishers} \end{LARGE} \end{center} \newpage % -----File: 303.png \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \begin{LARGE} THEORETICAL MECHANICS \par \end{LARGE} \vspace{\baselineskip} By \textsc{Percey F. Smith}, Professor of Mathematics in the Sheffield Scientific\\ School, Yale University, and \textsc{William R. Longley}, Assistant Professor of\\ Mathematics in the Sheffield Scientific School, Yale University \rule{0.25\textwidth}{1pt} 8vo, semiflexible cloth, 288 pages, with diagrams, \$2.50 \rule{0.25\textwidth}{1pt} \end{center} \noindent{\Huge 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 impact), 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 coplanar 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 cognizant 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 equation. The problems are carefully selected, and numerous illustrative examples are worked out in the text. \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \vspace{\baselineskip} \begin{LARGE} GINN~~AND~~COMPANY~~~\textsc{Publishers} \end{LARGE} \end{center} \newpage % -----File: 304.png \newenvironment{hang}{% \begin{list}{}{% \setlength{\leftmargin}{1cm}% \setlength{\itemindent}{-1cm}% }% \item[]}{\end{list}} \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \begin{LARGE} BOOKS ON HIGHER MATHEMATICS \par \end{LARGE} \rule{0.25\textwidth}{1pt} \end{center} \begin{hang} \spreadout{BYERLY: ELEMENTARY TREATISE ON FOURIER'S SERIES, AND} \\ \spreadout{SPHERICAL, CYLINDRICAL, AND ELLIPSOIDAL HARMONICS,} \\ \spreadout{WITH APPLICATIONS TO PROBLEMS IN MATHEMATICAL} \\ PHYSICS\@. \$3.00. \end{hang} \textsc{An} introduction to the treatment of some of the important linear partial differential equations which lie at the foundation of modern theories in physics. \begin{hang} HEDRICK: GOURSAT'S COURSE IN MATHEMATICAL ANALYSIS, VOLUME I\@. \$4.00. \end{hang} \textsc{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. \begin{hang} HEDRICK AND KELLOGG: APPLICATIONS OF THE CALCULUS TO MECHANICS\@. \$1.25. \end{hang} \textsc{A completed} summary of those parts of mechanics which occur as applications of the calculus. \begin{hang} PEIRCE, B. O.: ELEMENTS OF THE THEORY OF THE NEWTONIAN POTENTIAL FUNCTION (Third, Revised, and Enlarged Edition). \$2.50. \end{hang} \textsc{Intended} for general students who wish to acquire a sound knowledge of the properties of the Newtonian Potential Function. Nearly four hundred miscellaneous problems have been added in this new edition. \begin{hang} PEIRCE, J. M.: ELEMENTS OF LOGARITHMS\@. 50 cents. \end{hang} \textsc{Designed} to give the student a complete and accurate knowledge of the nature and use of logarithms. \begin{hang} TAYLOR, J. M.: ELEMENTS OF THE DIFFERENTIAL AND INTEGRAL CALCULUS (Revised Edition, Enlarged and Entirely Rewritten). With Examples and Applications. \$2.00. \end{hang} \textsc{Presents} clearly, scientifically, and in their true relations the three common methods in the calculus. \newpage \begin{hang} MATHEMATICAL TEXTS. Edited by Percey F. Smith, of the Sheffield Scientific School of Yale University. \end{hang} \begin{footnotesize} \textsc{Eisenhart: Differential Geometry of Curves and Surfaces.} \$4.50. \textsc{Granville: Plane Trigonometry and Tables.} \$1.00. \textsc{Granville: Plane and Spherical Trigonometry, without Tables.} \$1.00. \textsc{Granville: Plane and Spherical Trigonometry, and Tables.} \$1.25. \textsc{Granville: Elements of Differential and Integral Calculus.} \$2.50. \textsc{Granville: Logarithmic Tables.} 50 cents. \textsc{Hawkes: Advanced Algebra.} \$1.40. \textsc{Smith and Gale: Elements of Analytic Geometry.} \$2.00. \textsc{Smith and Gale: Introduction to Analytic Geometry.} \$1.25. \end{footnotesize} \begin{center} \rule{\textwidth}{1pt}\\[-2ex] \rule{\textwidth}{1pt} \vspace{\baselineskip} \LARGE GINN~~AND~~COMPANY~~~\textsc{Publishers} \end{center} \newpage \begin{center}\textsc{Typographical Errors corrected in Project Gutenberg edition}\end{center} p.~\pageref{err053}~Equation (3), the second $\displaystyle \int\limits_0^\pi$ was not printed.\\ p.~\pageref{err111}~Example~1, the = sign on the second line (beginning $\dfrac{2}{\pi}$) was not printed.\\ p.~\pageref{err149}~Example~6, second sin term $\sin \dfrac{m\pi y}{b}$ in original, amended to $\sin \dfrac{n\pi y}{b}$.\\ p.~\pageref{err250}~Equation (8), $D_{\rho_3}^2V$ was printed as $D_{\rho_3}V$. \newpage \small \pagenumbering{Roman} \begin{verbatim} 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 business@pglaf.org. 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 gbnewby@pglaf.org 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. \end{verbatim} % Syntax is control word, 1=Mandatory/0=Optional, 1=Keep/0=Delete, opening replacement, closing replacement, ... % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % % 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-t.tex or 29779-t.zip ***** % % This and all associated files of various formats will be found in: % % http://www.gutenberg.org/2/9/7/7/29779/ % % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \end{document} ### lprep configuration @ControlwordReplace = ( ['\Example',"Example"], ['\EXAMPLE',"EXAMPLE"] ); @ControlwordArguments = ( ['\\pngright',1,0,"\n[Illustration","]\n",1,0,'','',1,0,'','',1,0,'',''], ['\\pngrightl',1,0,"\n[Illustration","]\n",1,0,'','',1,0,'','',1,0,'',''], ['\\pngcent',1,0,"\n[Illustration","]\n",1,0,'','',1,0,'',''], ['\\mychap',1,1,'',"\n",1,1,'',''], ['\\smallrom',1,1,'[',']'], ['\\smallromr',1,1,'(',')'] ); ### This is pdfTeXk, Version 3.141592-1.40.3 (Web2C 7.5.6) (format=pdflatex 2009.6.29) 24 AUG 2009 04:54 entering extended mode %&-line parsing enabled. **29779-t.tex (./29779-t.tex LaTeX2e <2005/12/01> Babel and hyphenation patterns for english, usenglishmax, dumylang, noh yphenation, loaded. (/usr/share/texmf-texlive/tex/latex/base/book.cls Document Class: book 2005/09/16 v1.4f Standard LaTeX document class (/usr/share/texmf-texlive/tex/latex/base/bk10.clo File: bk10.clo 2005/09/16 v1.4f Standard LaTeX file (size option) ) \c@part=\count79 \c@chapter=\count80 \c@section=\count81 \c@subsection=\count82 \c@subsubsection=\count83 \c@paragraph=\count84 \c@subparagraph=\count85 \c@figure=\count86 \c@table=\count87 \abovecaptionskip=\skip41 \belowcaptionskip=\skip42 \bibindent=\dimen102 ) (/usr/share/texmf-texlive/tex/latex/amsmath/amsmath.sty Package: amsmath 2000/07/18 v2.13 AMS math features \@mathmargin=\skip43 For additional information on amsmath, use the `?' option. (/usr/share/texmf-texlive/tex/latex/amsmath/amstext.sty Package: amstext 2000/06/29 v2.01 (/usr/share/texmf-texlive/tex/latex/amsmath/amsgen.sty File: amsgen.sty 1999/11/30 v2.0 \@emptytoks=\toks14 \ex@=\dimen103 )) (/usr/share/texmf-texlive/tex/latex/amsmath/amsbsy.sty Package: amsbsy 1999/11/29 v1.2d \pmbraise@=\dimen104 ) (/usr/share/texmf-texlive/tex/latex/amsmath/amsopn.sty Package: amsopn 1999/12/14 v2.01 operator names ) \inf@bad=\count88 LaTeX Info: Redefining \frac on input line 211. \uproot@=\count89 \leftroot@=\count90 LaTeX Info: Redefining \overline on input line 307. \classnum@=\count91 \DOTSCASE@=\count92 LaTeX Info: Redefining \ldots on input line 379. LaTeX Info: Redefining \dots on input line 382. LaTeX Info: Redefining \cdots on input line 467. \Mathstrutbox@=\box26 \strutbox@=\box27 \big@size=\dimen105 LaTeX Font Info: Redeclaring font encoding OML on input line 567. LaTeX Font Info: Redeclaring font encoding OMS on input line 568. \macc@depth=\count93 \c@MaxMatrixCols=\count94 \dotsspace@=\muskip10 \c@parentequation=\count95 \dspbrk@lvl=\count96 \tag@help=\toks15 \row@=\count97 \column@=\count98 \maxfields@=\count99 \andhelp@=\toks16 \eqnshift@=\dimen106 \alignsep@=\dimen107 \tagshift@=\dimen108 \tagwidth@=\dimen109 \totwidth@=\dimen110 \lineht@=\dimen111 \@envbody=\toks17 \multlinegap=\skip44 \multlinetaggap=\skip45 \mathdisplay@stack=\toks18 LaTeX Info: Redefining \[ on input line 2666. LaTeX Info: Redefining \] on input line 2667. ) (/usr/share/texmf-texlive/tex/latex/amsfonts/amssymb.sty Package: amssymb 2002/01/22 v2.2d (/usr/share/texmf-texlive/tex/latex/amsfonts/amsfonts.sty Package: amsfonts 2001/10/25 v2.2f \symAMSa=\mathgroup4 \symAMSb=\mathgroup5 LaTeX Font Info: Overwriting math alphabet `\mathfrak' in version `bold' (Font) U/euf/m/n --> U/euf/b/n on input line 132. )) (/usr/share/texmf-texlive/tex/latex/tools/tabularx.sty Package: tabularx 1999/01/07 v2.07 `tabularx' package (DPC) (/usr/share/texmf-texlive/tex/latex/tools/array.sty Package: array 2005/08/23 v2.4b Tabular extension package (FMi) \col@sep=\dimen112 \extrarowheight=\dimen113 \NC@list=\toks19 \extratabsurround=\skip46 \backup@length=\skip47 ) \TX@col@width=\dimen114 \TX@old@table=\dimen115 \TX@old@col=\dimen116 \TX@target=\dimen117 \TX@delta=\dimen118 \TX@cols=\count100 \TX@ftn=\toks20 ) (/usr/share/texmf-texlive/tex/latex/graphics/graphicx.sty Package: graphicx 1999/02/16 v1.0f Enhanced LaTeX Graphics (DPC,SPQR) (/usr/share/texmf-texlive/tex/latex/graphics/keyval.sty Package: keyval 1999/03/16 v1.13 key=value parser (DPC) \KV@toks@=\toks21 ) (/usr/share/texmf-texlive/tex/latex/graphics/graphics.sty Package: graphics 2006/02/20 v1.0o Standard LaTeX Graphics (DPC,SPQR) (/usr/share/texmf-texlive/tex/latex/graphics/trig.sty Package: trig 1999/03/16 v1.09 sin cos tan (DPC) ) (/etc/texmf/tex/latex/config/graphics.cfg File: graphics.cfg 2007/01/18 v1.5 graphics configuration of teTeX/TeXLive ) Package graphics Info: Driver file: pdftex.def on input line 90. (/usr/share/texmf-texlive/tex/latex/pdftex-def/pdftex.def File: pdftex.def 2007/01/08 v0.04d Graphics/color for pdfTeX \Gread@gobject=\count101 )) \Gin@req@height=\dimen119 \Gin@req@width=\dimen120 ) (/usr/share/texmf-texlive/tex/latex/base/inputenc.sty Package: inputenc 2006/05/05 v1.1b Input encoding file \inpenc@prehook=\toks22 \inpenc@posthook=\toks23 (/usr/share/texmf-texlive/tex/latex/base/latin1.def File: latin1.def 2006/05/05 v1.1b Input encoding file )) (/usr/share/texmf-texlive/tex/latex/ltxmisc/needspace.sty Package: needspace 2003/02/18 v1.3a reserve vertical space ) (/usr/share/texmf-texlive/tex/latex/wrapfig/wrapfig.sty \wrapoverhang=\dimen121 \WF@size=\dimen122 \c@WF@wrappedlines=\count102 \WF@box=\box28 \WF@everypar=\toks24 Package: wrapfig 2003/01/31 v 3.6 ) (/usr/share/texmf-texlive/tex/latex/tools/verbatim.sty Package: verbatim 2003/08/22 v1.5q LaTeX2e package for verbatim enhancements \every@verbatim=\toks25 \verbatim@line=\toks26 \verbatim@in@stream=\read1 ) (/usr/share/texmf-texlive/tex/latex/wasysym/wasysym.sty Package: wasysym 2003/10/30 v2.0 Wasy-2 symbol support package \symwasy=\mathgroup6 LaTeX Font Info: Overwriting symbol font `wasy' in version `bold' (Font) U/wasy/m/n --> U/wasy/b/n on input line 90. ) \c@wrapwidth=\count103 \Zw=\count104 \Zh=\count105 (./29779-t.aux) \openout1 = `29779-t.aux'. LaTeX Font Info: Checking defaults for OML/cmm/m/it on input line 170. LaTeX Font Info: ... okay on input line 170. LaTeX Font Info: Checking defaults for T1/cmr/m/n on input line 170. LaTeX Font Info: ... okay on input line 170. LaTeX Font Info: Checking defaults for OT1/cmr/m/n on input line 170. LaTeX Font Info: ... okay on input line 170. LaTeX Font Info: Checking defaults for OMS/cmsy/m/n on input line 170. LaTeX Font Info: ... okay on input line 170. LaTeX Font Info: Checking defaults for OMX/cmex/m/n on input line 170. LaTeX Font Info: ... okay on input line 170. LaTeX Font Info: Checking defaults for U/cmr/m/n on input line 170. LaTeX Font Info: ... okay on input line 170. [1 {/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}] LaTeX Font Info: Try loading font information for OMS+cmr on input line 235. (/usr/share/texmf-texlive/tex/latex/base/omscmr.fd File: omscmr.fd 1999/05/25 v2.5h Standard LaTeX font definitions ) LaTeX Font Info: Font shape `OMS/cmr/m/n' in size <10> not available (Font) Font shape `OMS/cmsy/m/n' tried instead on input line 235. [2] LaTeX Font Info: Try loading font information for U+msa on input line 287. (/usr/share/texmf-texlive/tex/latex/amsfonts/umsa.fd File: umsa.fd 2002/01/19 v2.2g AMS font definitions ) LaTeX Font Info: Try loading font information for U+msb on input line 287. (/usr/share/texmf-texlive/tex/latex/amsfonts/umsb.fd File: umsb.fd 2002/01/19 v2.2g AMS font definitions ) LaTeX Font Info: Try loading font information for U+wasy on input line 287. (/usr/share/texmf-texlive/tex/latex/wasysym/uwasy.fd File: uwasy.fd 2003/10/30 v2.0 Wasy-2 symbol font definitions ) [1 ] Underfull \hbox (badness 10000) in paragraph at lines 317--318 [] Underfull \hbox (badness 10000) in paragraph at lines 319--346 [] Underfull \hbox (badness 10000) in paragraph at lines 349--350 [] Underfull \hbox (badness 10000) in paragraph at lines 351--365 [] [2] Underfull \hbox (badness 10000) in paragraph at lines 369--370 [] Underfull \hbox (badness 10000) in paragraph at lines 371--379 [] Underfull \hbox (badness 10000) in paragraph at lines 383--384 [] Underfull \hbox (badness 10000) in paragraph at lines 385--417 [] [3] Underfull \hbox (badness 10000) in paragraph at lines 418--419 [] Underfull \hbox (badness 10000) in paragraph at lines 420--422 [] Underfull \hbox (badness 10000) in paragraph at lines 425--426 [] Underfull \hbox (badness 10000) in paragraph at lines 427--454 [] [4] Underfull \hbox (badness 10000) in paragraph at lines 457--458 [] Underfull \hbox (badness 10000) in paragraph at lines 459--478 [] Underfull \hbox (badness 10000) in paragraph at lines 483--484 [] Underfull \hbox (badness 10000) in paragraph at lines 485--504 [] [5] Underfull \hbox (badness 10000) in paragraph at lines 508--510 [] Underfull \hbox (badness 10000) in paragraph at lines 511--530 [] Underfull \hbox (badness 10000) in paragraph at lines 535--536 [] [6] [7] [1 ] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] Underfull \hbox (badness 10000) in paragraph at lines 1853--1854 [] Underfull \hbox (badness 10000) in paragraph at lines 1857--1858 [] [23] [24] [25] [26] Underfull \hbox (badness 10000) in paragraph at lines 2022--2025 [] Underfull \hbox (badness 10000) in paragraph at lines 2037--2039 [] [27] Underfull \hbox (badness 10000) in paragraph at lines 2053--2055 [] Underfull \hbox (badness 10000) in paragraph at lines 2071--2074 [] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] Underfull \hbox (badness 10000) in paragraph at lines 2658--2660 [] File: images/001.png Graphic file (type png) [39 <./images/001.png (PNG copy)>] File: images/002.png Graphic file (type png) [40 <./images/002.png (PNG copy)>] [41] [42] [43] [44] [45 ] File: images/003.png Graphic file (type png) [46 <./images/003.png (PNG copy)>] File: images/004.png Graphic file (type png) Underfull \hbox (badness 10000) in paragraph at lines 3063--3064 [] Underfull \hbox (badness 10000) in paragraph at lines 3067--3072 [] Underfull \hbox (badness 10000) in paragraph at lines 3074--3079 [] [47 <./images/004.png (PNG copy)>] Underfull \hbox (badness 10000) in paragraph at lines 3079--3080 [] [48] [49] [50] Underfull \hbox (badness 10000) in paragraph at lines 3295--3299 [] [51] [52] [53] [54] [55] [56] File: images/005.png Graphic file (type png) [57] [58 <./images/005.png (PNG copy)>] [59] [60] [61] [62 ] File: images/006.png Graphic file (type png) [63] File: images/007.png Graphic file (type png) [64 <./images/006.png (PNG copy)>] [65 <./images/007.png ( PNG copy)>] [66] [67] [68] [69] [70] Underfull \hbox (badness 10000) in paragraph at lines 4226--4227 [] [71] [72] File: images/008.png Graphic file (type png) [73 <./images/008.png (PNG copy)>] [74] [75] File: images/009.png Graphic file (type png) [76 <./images/009.png (PNG copy)>] Underfull \hbox (badness 10000) in paragraph at lines 4520--4529 [] [77] [78] [79] Underfull \hbox (badness 10000) in paragraph at lines 4675--4677 [] [80] Underfull \hbox (badness 10000) in paragraph at lines 4718--4720 [] Underfull \hbox (badness 10000) in paragraph at lines 4727--4729 [] [81] [82] [83] Underfull \hbox (badness 10000) in paragraph at lines 4873--4875 [] Underfull \hbox (badness 10000) in paragraph at lines 4888--4890 [] Underfull \hbox (badness 10000) in paragraph at lines 4915--4921 [] Underfull \hbox (badness 10000) in paragraph at lines 4922--4926 [] [84] [85] Underfull \hbox (badness 10000) in paragraph at lines 4998--5000 [] Underfull \hbox (badness 10000) in paragraph at lines 5023--5026 [] [86] Underfull \hbox (badness 10000) in paragraph at lines 5032--5036 [] [87] [88] Underfull \hbox (badness 10000) in paragraph at lines 5157--5163 [] Underfull \hbox (badness 10000) in paragraph at lines 5168--5170 [] [89] Underfull \hbox (badness 10000) in paragraph at lines 5205--5207 [] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] Underfull \hbox (badness 10000) in paragraph at lines 5720--5722 [] Underfull \hbox (badness 10000) in paragraph at lines 5751--5753 [] Underfull \hbox (badness 10000) in paragraph at lines 5758--5759 [] [101] Underfull \hbox (badness 10000) in paragraph at lines 5763--5766 [] Underfull \hbox (badness 10000) in paragraph at lines 5769--5771 [] [102] [103] [104] [105] Underfull \hbox (badness 10000) in paragraph at lines 5992--5998 [] [106] Underfull \hbox (badness 10000) in paragraph at lines 5999--6006 [] [107] Underfull \hbox (badness 10000) in paragraph at lines 6044--6048 [] Underfull \hbox (badness 10000) in paragraph at lines 6049--6053 [] Underfull \hbox (badness 10000) in paragraph at lines 6054--6059 [] [108] [109] [110] [111] [112] Underfull \hbox (badness 10000) in paragraph at lines 6297--6303 [] [113] [114] [115] Underfull \hbox (badness 10000) in paragraph at lines 6438--6439 [] [116] [117] [118] [119] [120] Underfull \hbox (badness 10000) in paragraph at lines 6708--6710 [] [121] [122] Underfull \hbox (badness 10000) in paragraph at lines 6755--6762 [] Underfull \hbox (badness 10000) in paragraph at lines 6769--6771 [] Underfull \hbox (badness 10000) in paragraph at lines 6779--6781 [] [123] [124] Underfull \hbox (badness 10000) in paragraph at lines 6847--6849 [] [125] Underfull \hbox (badness 10000) in paragraph at lines 6890--6892 [] [126] [127] [128] [129] [130] [131] [132] File: images/010.png Graphic file (type png) File: images/011.png Graphic file (type png) Underfull \hbox (badness 10000) in paragraph at lines 7208--7209 [] [133 <./images/010.png (PNG copy)> <./images/011.png (PNG copy)>] File: images/012.png Graphic file (type png) [134 <./images/012.png (PNG copy)>] Underfull \hbox (badness 10000) in paragraph at lines 7309--7311 [] [135] Underfull \hbox (badness 10000) in paragraph at lines 7342--7344 [] [136] [137] Underfull \hbox (badness 10000) in paragraph at lines 7445--7447 [] [138] Underfull \hbox (badness 10000) in paragraph at lines 7489--7491 [] [139] [140] Underfull \hbox (badness 10000) in paragraph at lines 7600--7604 [] [141] Underfull \hbox (badness 10000) in paragraph at lines 7613--7615 [] Underfull \hbox (badness 10000) in paragraph at lines 7636--7638 [] Underfull \hbox (badness 10000) in paragraph at lines 7643--7645 [] [142] Underfull \hbox (badness 10000) in paragraph at lines 7671--7675 [] [143] [144] [145] [146] Underfull \hbox (badness 10000) in paragraph at lines 7831--7836 [] [147] [148] [149] [150] [151] [152] [153] [154] [155] Underfull \hbox (badness 10000) in paragraph at lines 8295--8297 [] [156] Underfull \hbox (badness 10000) in paragraph at lines 8321--8323 [] [157] [158] [159] [160] [161] [162] [163] [164] Underfull \hbox (badness 10000) in paragraph at lines 8679--8681 [] [165] [166] [167] [168] [169] [170] [171] [172] Underfull \hbox (badness 10000) in paragraph at lines 8975--8977 [] Underfull \hbox (badness 10000) in paragraph at lines 8982--8985 [] [173] LaTeX Font Info: Font shape `OMS/cmr/m/n' in size <8> not available (Font) Font shape `OMS/cmsy/m/n' tried instead on input line 9031. [174] [175] Underfull \hbox (badness 10000) in paragraph at lines 9103--9106 [] Underfull \hbox (badness 10000) in paragraph at lines 9120--9122 [] [176] Underfull \hbox (badness 10000) in paragraph at lines 9138--9141 [] Underfull \hbox (badness 10000) in paragraph at lines 9167--9170 [] [177] [178] [179] [180] [181] [182] [183] Underfull \hbox (badness 10000) in paragraph at lines 9446--9448 [] Underfull \hbox (badness 10000) in paragraph at lines 9460--9462 [] [184] File: images/013.png Graphic file (type png) [185] File: images/014.png Graphic file (type png) [186 <./images/013.png (PNG copy)>] [187 <./images/014.png (PNG copy)>] [188] Underfull \hbox (badness 10000) in paragraph at lines 9644--9646 [] [189] [190] Underfull \hbox (badness 10000) in paragraph at lines 9733--9735 [] Underfull \hbox (badness 10000) in paragraph at lines 9750--9752 [] [191] Underfull \hbox (badness 10000) in paragraph at lines 9760--9765 [] Underfull \hbox (badness 10000) in paragraph at lines 9779--9781 [] Underfull \hbox (badness 10000) in paragraph at lines 9784--9786 [] [192] Underfull \hbox (badness 10000) in paragraph at lines 9802--9805 [] Underfull \hbox (badness 10000) in paragraph at lines 9825--9828 [] [193] Underfull \hbox (badness 10000) in paragraph at lines 9886--9887 [] Underfull \hbox (badness 10000) in paragraph at lines 9897--9899 [] Underfull \hbox (badness 10000) in paragraph at lines 9929--9931 [] [194] [195] Underfull \hbox (badness 10000) in paragraph at lines 10001--10004 [] [196] [197] [198] [199] Underfull \hbox (badness 10000) in paragraph at lines 10258--10261 [] [200] [201] Underfull \hbox (badness 10000) in paragraph at lines 10309--10313 [] [202] [203] Underfull \hbox (badness 10000) in paragraph at lines 10415--10417 [] Underfull \hbox (badness 10000) in paragraph at lines 10418--10420 [] [204] [205] [206] [207] Underfull \hbox (badness 10000) in paragraph at lines 10618--10621 [] Underfull \hbox (badness 10000) in paragraph at lines 10639--10640 [] [208] Underfull \hbox (badness 10000) in paragraph at lines 10694--10696 [] Underfull \hbox (badness 10000) in paragraph at lines 10697--10703 [] [209] [210] [211] [212] [213] [214] Underfull \hbox (badness 10000) in paragraph at lines 11002--11004 [] Underfull \hbox (badness 10000) in paragraph at lines 11019--11020 [] Underfull \hbox (badness 10000) in paragraph at lines 11021--11025 [] [215] [216] [217] [218] [219] [220] [221] [222] Underfull \hbox (badness 10000) in paragraph at lines 11371--11372 [] Underfull \hbox (badness 10000) in paragraph at lines 11409--11411 [] [223] Underfull \hbox (badness 10000) in paragraph at lines 11415--11417 [] [224] Underfull \hbox (badness 10000) in paragraph at lines 11512--11514 [] [225] Underfull \hbox (badness 10000) in paragraph at lines 11522--11524 [] Underfull \hbox (badness 10000) in paragraph at lines 11525--11527 [] Underfull \hbox (badness 10000) in paragraph at lines 11528--11529 [] Underfull \hbox (badness 10000) in paragraph at lines 11570--11574 [] [226] Underfull \hbox (badness 10000) in paragraph at lines 11629--11632 [] [227] Underfull \hbox (badness 10000) in paragraph at lines 11648--11650 [] Underfull \hbox (badness 10000) in paragraph at lines 11664--11665 [] [228] [229] Underfull \hbox (badness 10000) in paragraph at lines 11774--11775 [] Underfull \hbox (badness 10000) in paragraph at lines 11787--11789 [] [230] Underfull \hbox (badness 10000) in paragraph at lines 11803--11805 [] Underfull \hbox (badness 10000) in paragraph at lines 11806--11813 [] Underfull \hbox (badness 10000) in paragraph at lines 11814--11821 [] [231] [232] Underfull \hbox (badness 10000) in paragraph at lines 11956--11958 [] [233] [234] Underfull \hbox (badness 10000) in paragraph at lines 12030--12032 [] Underfull \hbox (badness 10000) in paragraph at lines 12033--12035 [] Underfull \hbox (badness 10000) in paragraph at lines 12042--12044 [] Underfull \hbox (badness 10000) in paragraph at lines 12048--12051 [] Underfull \hbox (badness 10000) in paragraph at lines 12052--12055 [] [235] Underfull \hbox (badness 10000) in paragraph at lines 12056--12061 [] Underfull \hbox (badness 10000) in paragraph at lines 12062--12066 [] [236] [237] [238] [239] [240] [241] [242] [243] Underfull \hbox (badness 10000) in paragraph at lines 12477--12481 [] [244] [245] [246] [247] Underfull \hbox (badness 10000) in paragraph at lines 12664--12667 [] [248] [249] [250] Underfull \hbox (badness 10000) in paragraph at lines 12791--12793 [] [251] [252] [253] [254] [255] [256] [257] [258] [259] [260] Underfull \hbox (badness 10000) in paragraph at lines 13329--13331 [] [261] [262] Underfull \hbox (badness 10000) in paragraph at lines 13378--13381 [] Underfull \hbox (badness 10000) in paragraph at lines 13396--13398 [] Underfull \hbox (badness 10000) in paragraph at lines 13399--13400 [] [263] [264] [265] Underfull \hbox (badness 10000) in paragraph at lines 13563--13565 [] [266] [267] [268] [269] [270] [271] [272] [273] [274] [275] [276] [277] [278] [ 279] [280] [281] [282] [283] [284] [285] [286] [287] [288] [289] [290] [291] Underfull \hbox (badness 10000) in paragraph at lines 15113--15114 [] Underfull \hbox (badness 10000) in paragraph at lines 15115--15116 [] Underfull \hbox (badness 10000) in paragraph at lines 15117--15118 [] [292] [1] [2] [3] [4] [5] [6] [7] [8] (./29779-t.aux) *File List* book.cls 2005/09/16 v1.4f Standard LaTeX document class bk10.clo 2005/09/16 v1.4f Standard LaTeX file (size option) amsmath.sty 2000/07/18 v2.13 AMS math features amstext.sty 2000/06/29 v2.01 amsgen.sty 1999/11/30 v2.0 amsbsy.sty 1999/11/29 v1.2d amsopn.sty 1999/12/14 v2.01 operator names amssymb.sty 2002/01/22 v2.2d amsfonts.sty 2001/10/25 v2.2f tabularx.sty 1999/01/07 v2.07 `tabularx' package (DPC) array.sty 2005/08/23 v2.4b Tabular extension package (FMi) graphicx.sty 1999/02/16 v1.0f Enhanced LaTeX Graphics (DPC,SPQR) keyval.sty 1999/03/16 v1.13 key=value parser (DPC) graphics.sty 2006/02/20 v1.0o Standard LaTeX Graphics (DPC,SPQR) trig.sty 1999/03/16 v1.09 sin cos tan (DPC) graphics.cfg 2007/01/18 v1.5 graphics configuration of teTeX/TeXLive pdftex.def 2007/01/08 v0.04d Graphics/color for pdfTeX inputenc.sty 2006/05/05 v1.1b Input encoding file latin1.def 2006/05/05 v1.1b Input encoding file needspace.sty 2003/02/18 v1.3a reserve vertical space wrapfig.sty 2003/01/31 v 3.6 verbatim.sty 2003/08/22 v1.5q LaTeX2e package for verbatim enhancements wasysym.sty 2003/10/30 v2.0 Wasy-2 symbol support package omscmr.fd 1999/05/25 v2.5h Standard LaTeX font definitions umsa.fd 2002/01/19 v2.2g AMS font definitions umsb.fd 2002/01/19 v2.2g AMS font definitions uwasy.fd 2003/10/30 v2.0 Wasy-2 symbol font definitions images/001.png images/002.png images/003.png images/004.png images/005.png images/006.png images/007.png images/008.png images/009.png images/010.png images/011.png images/012.png images/013.png images/014.png *********** ) Here is how much of TeX's memory you used: 2133 strings out of 95086 24700 string characters out of 1183255 95415 words of memory out of 1500000 5258 multiletter control sequences out of 10000+50000 16199 words of font info for 62 fonts, out of 1200000 for 2000 28 hyphenation exceptions out of 8191 26i,21n,24p,386b,285s stack positions out of 5000i,500n,6000p,200000b,5000s < /usr/share/texmf-texlive/fonts/type1/bluesky/cm/cmsy7.pfb>< /usr/share/texmf-texlive/fonts/type1/bluesky/cm/cmti8.pfb> Output written on 29779-t.pdf (309 pages, 1393827 bytes). PDF statistics: 1173 PDF objects out of 1200 (max. 8388607) 0 named destinations out of 1000 (max. 131072) 155 words of extra memory for PDF output out of 10000 (max. 10000000)